Arxiv今日论文 | 2026-05-06

本篇博文主要内容为 2026-05-06 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。

自然语言处理共74篇(Computation and Language (cs.CL))
人工智能共190篇(Artificial Intelligence (cs.AI))
计算机视觉共92篇(Computer Vision and Pattern Recognition (cs.CV))
机器学习共199篇(Machine Learning (cs.LG))
多智能体系统共7篇(Multiagent Systems (cs.MA))
信息检索共9篇(Information Retrieval (cs.IR))
人机交互共23篇(Human-Computer Interaction (cs.HC))

多智能体系统

[MA-0] Physics-Grounded Multi-Agent Architecture for Traceable Risk-Aware Human-AI Decision Support in Manufacturing

【速读】：该论文旨在解决高精度数控（CNC）加工自由曲面航空航天部件时，如何在风险约束下可靠执行多步骤数值工作流并确保决策可追溯性的问题。当前通用大语言模型（Large Language Model, LLM）虽能生成文本，但无法保障复杂工艺流程中的安全性与可审计性。解决方案的关键在于提出多智能体知识分析（Multi-Agent Knowledge Analysis, MAKA）架构，其通过分离意图路由、仅工具定量分析、知识图谱检索和基于批评的验证四个模块，在推荐前强制执行物理合理性、安全边界和溯源完整性；并在Ti-6Al-4V转子叶片加工测试平台上融合虚拟路径误差场、切削力与变形仿真及扫描三维检测偏差图，实现对偏差的结构化分解，从而显著提升工具执行成功率（最高达87.5个百分点），并在数字孪生场景中将预测表面偏差从约10⁻²英寸降至±10⁻³英寸，为风险敏感的人类决策提供预部署验证信号。

链接: https://arxiv.org/abs/2605.04003
作者: Danny Hoang,Ryan Matthiessen,Christopher Miller,Nasir Mannan,Ruby ElKharboutly,David Gorsich,Matthew P. Castanier,Farhad Imani
机构: University of Connecticut (康涅狄格大学); Connecticut Center for Advanced Technology (康涅狄格先进科技中心); Quinnipiac University (奎尼皮亚克大学); DEVCOM Ground Vehicle Systems Center (美国陆军研发与工程地面车辆系统中心)
类目: Multiagent Systems (cs.MA); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注:

点击查看摘要

Abstract:High-precision CNC machining of free-form aerospace components requires bounded compensations informed by inspection, simulation, and process knowledge. Off-the-shelf large language model (LLM) assistants can generate text, but they do not reliably execute risk-constrained multi-step numerical workflows or provide auditable provenance for high-stakes decisions. We present multi-agent knowledge analysis (MAKA), a human-in-the-loop decision-support architecture that separates intent routing, tools-only quantitative analysis, knowledge graph retrieval, and critic-based verification that enforces physical plausibility, safety bounds, and provenance completeness before recommendations are surfaced for human approval. MAKA is instantiated on a Ti-6Al-4V rotor blade machining testbed by fusing virtual-machining path-tracking error fields, cutting-force and deflection simulations, and scan-based 3D inspection deviation maps from 16 blades. The analysis decomposes deviation into an evidence-linked pathing component, a drift-based wear proxy capturing systematic evolution across parts, a residual systematic compliance term, and a variability proxy for instability-aware escalation. In a three-level tool-orchestration benchmark (single-step through \geq 3-step stateful sequences), MAKA improves successful tool execution by up to 87.5 percentage points relative to an unstructured single-model interaction pattern with identical tool access. Digital twin what-if studies show MAKA can coordinate traceable compensation candidates that reduce predicted surface deviation from order 10^-2 in to approximately \pm 10^-3 in over most of the blade within the simulation environment, providing a pre-deployment verification signal for risk-aware human decision-making.

[MA-1] QKVShare: Quantized KV-Cache Handoff for Multi-Agent On-Device LLM s

【速读】：该论文旨在解决多智能体大语言模型（Multi-agent LLM）在边缘设备上进行隐式上下文（latent context）高效传递的问题。当前实践方案要么代价高昂的重新预填充（re-prefill），要么全精度键值缓存（KV-cache）传输，均存在效率瓶颈。其解决方案的关键在于提出QKVShare框架，通过三种核心技术实现：（1）基于token级别的混合精度分配策略，优化量化粒度；（2）自包含的CacheCard表示形式，提升缓存结构紧凑性与可移植性；（3）兼容HuggingFace的缓存注入路径，保障部署灵活性。实验表明，在重复代理交接场景下，自适应量化优于均匀量化，尤其在深度跳转和高预算条件下优势显著，且QKVShare路径相较全量重预填充能大幅降低首次响应时间（TTFT），验证了量化KV缓存传递作为边缘端系统优化方向的可行性。

链接: https://arxiv.org/abs/2605.03884
作者: Pratik Honavar,Tejpratap GVSL
机构: 未知
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: 12 pages, 1 figure, 3 tables

点击查看摘要

Abstract:Multi-agent LLM systems on edge devices need to hand off latent context efficiently, but the practical choices today are expensive re-prefill or full-precision KV transfer. We study QKVShare, a framework for quantized KV-cache handoff between agents that combines token-level mixed-precision allocation, a self-contained CacheCard representation, and a HuggingFace-compatible cache injection path. Our current results support a narrower but clearer story than the original draft: on 150 GSM8K problems with Llama-3.1-8B-Instruct, adaptive quantization remains competitive under repeated handoff and shows its clearest gains against uniform quantization in deeper-hop, higher budget settings; for handoff latency, the QKVShare path reduces TTFT relative to full re prefill at every tested context, from 130.7 ms vs. 150.2 ms at nominal 1K context to 397.1 ms vs. 1029.7 ms at nominal 8K context;. Stage timing shows that post-injection generation, not card creation, dominates the current QKVShare latency path. These results position quantized KV handoff as a promising on-device systems direction while also highlighting the need for stronger controller ablations and apples-to-apples runtime comparisons.

[MA-2] FINER-SQL: Boosting Small Language Models for Text-to-SQL

【速读】：该论文旨在解决大语言模型（Large Language Models, LLMs）在Text-to-SQL生成任务中存在计算成本高、延迟长及数据隐私风险等问题，同时应对小型语言模型（Small Language Models, SLMs）因推理能力弱和指令遵循差而导致的性能瓶颈。其解决方案的关键在于提出一种可扩展且可复用的强化学习框架FINER-SQL，通过细粒度执行反馈替代稀疏二值奖励（0/1），引入两种关键奖励函数：记忆奖励（memory reward）用于对齐推理过程与验证过的执行轨迹以保障语义稳定性，原子奖励（atomic reward）则衡量操作层面的重叠度，为结构正确但不完整的SQL语句提供部分奖励。这一机制将离散的正确性判断转化为连续的学习信号，实现无需评判器（critic-free）的稳定优化，从而显著提升SLMs在BIRD和Spider基准上的执行准确率，达到与大型模型相当的效果，同时大幅降低推理延迟并支持本地化部署。

链接: https://arxiv.org/abs/2605.03465
作者: Thanh Dat Hoang,Thanh Trung Huynh,Matthias Weidlich,Thanh Tam Nguyen,Tong Chen,Hongzhi Yin,Quoc Viet Hung Nguyen
机构: Griffith University (澳大利亚); VinUniversity (越南); Humboldt-Universitat zu Berlin (德国); The University of Queensland (澳大利亚)
类目: Databases (cs.DB); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Human-Computer Interaction (cs.HC); Multiagent Systems (cs.MA)
备注:

点击查看摘要

Abstract:Large language models have driven major advances in Text-to-SQL generation. However, they suffer from high computational cost, long latency, and data privacy concerns, which make them impractical for many real-world applications. A natural alternative is to use small language models (SLMs), which enable efficient and private on-premise deployment. Yet, SLMs often struggle with weak reasoning and poor instruction following. Conventional reinforcement learning methods based on sparse binary rewards (0/1) provide little learning signal when the generated SQLs are incorrect, leading to unstable or collapsed training. To overcome these issues, we propose FINER-SQL, a scalable and reusable reinforcement learning framework that enhances SLMs through fine-grained execution feedback. Built on group relative policy optimization, FINER-SQL replaces sparse supervision with dense and interpretable rewards that offer continuous feedback even for incorrect SQLs. It introduces two key reward functions: a memory reward, which aligns reasoning with verified traces for semantic stability, and an atomic reward, which measures operation-level overlap to grant partial credit for structurally correct but incomplete SQLs. This approach transforms discrete correctness into continuous learning, enabling stable, critic-free optimization. Experiments on the BIRD and Spider benchmarks show that FINER-SQL achieves up to 67.73% and 85% execution accuracy with a 3B model – matching much larger LLMs while reducing inference latency to 5.57~s/sample. These results highlight a cost-efficient and privacy-preserving path toward high-performance Text-to-SQL generation. Our code is available at this https URL.

[MA-3] MemFlow: Intent-Driven Memory Orchestration for Small Language Model Agents

【速读】：该论文旨在解决小型语言模型（Small Language Models, SLMs）在长时程、多轮交互场景中因记忆管理不当而导致的性能瓶颈问题，具体表现为：全上下文提示引发上下文溢出、扁平检索引入噪声证据，以及开放式的代理循环在有限推理能力下不可靠。其解决方案的关键在于提出一种无需训练的记忆编排框架 MemFlow，通过将记忆规划从 SLM 中解耦出来，采用“路由-编译”（route-then-compile）的设计：由 Router Agent 根据查询意图分类并调度至 Memory Agent 的三个专业化层级（Profile Lookup、Targeted Retrieval 或 Deep Reasoning），每个层级按动态的、层级感知的 token 预算组装证据，再由 Answer Agent 生成响应，并由 Validator Agent 在必要时触发更重的内存层级重试。该方法避免了工具选择幻觉和推理循环，同时保持答案上下文紧凑，显著提升了资源受限环境下 SLM 在长时程记忆基准上的准确性（接近 2 倍提升）。

链接: https://arxiv.org/abs/2605.03312
作者: Jiayi Chen,Yingcong Li,Guiling Wang
机构: New Jersey Institute of Technology (新泽西理工学院)
类目: Multiagent Systems (cs.MA)
备注:

点击查看摘要

Abstract:Modern language agents must operate over long-horizon, multi-turn histories, yet deploying such agents with Small Language Models (SLMs) remains fundamentally difficult. Full-context prompting causes context overflow, flat retrieval exposes the model to noisy evidence, and open-ended agentic loops are unreliable under limited reasoning capacity. We argue that a substantial portion of SLM memory failure arises from mismatched memory operations: different query types demand categorically different retrieval strategies, evidence transformations, and context budgets that SLMs cannot reliably self-orchestrate through open-ended reasoning. We introduce MemFlow, a training-free memory orchestration framework that externalizes memory planning from the SLM. A Router Agent classifies each query by intent and dispatches it to the Memory Agent, which executes one of three specialized tiers (Profile Lookup, Targeted Retrieval, or Deep Reasoning) and assembles the resulting evidence under a dynamic, tier-aware token budget. An Answer Agent then generates a response from this compact context, and a Validator Agent optionally retries with a heavier memory tier when the response is not supported by the provided evidence. This route-then-compile design avoids tool-selection hallucination and reasoning loops while keeping the answer context compact. Evaluated on a frozen Qwen3-1.7B backbone across long-horizon memory benchmarks - LongMemEval, LoCoMo, and LongBench - MemFlow improves accuracy by nearly 2x over full-context SLM baselines. These results suggest that structured intent routing and deterministic evidence preparation can make limited-capacity models substantially more effective in resource-constrained long-horizon agents.

[MA-4] Coordination as an Architectural Layer for LLM -Based Multi-Agent Systems

【速读】：该论文旨在解决多智能体大语言模型（Multi-agent LLM）系统在生产环境中高失败率（41%–87%）的问题，其根本原因在于协调机制缺陷而非基础模型能力不足。现有方法要么仅通过经验归纳总结故障模式，要么提供声明式编排框架作为工程工具，均无法建立从协调配置到可预测故障特征之间的原理性映射。解决方案的关键在于将协调视为一个可配置的架构层，独立于智能体逻辑和信息访问机制，从而支持基于架构的推理而非单纯提升工程效率。作者通过在预测市场场景中设计一个信息受控实验，固定模型、工具、输出上限与提示模板，仅改变五种预设的协调配置，并以每问题总计算量作为内生架构输出，利用Murphy分解分离校准误差与判别能力，使不同配置即使在整体Brier得分一致时仍能留下可区分的签名特征。这一方法验证了协调架构可解释性和可控性的潜力，为未来多智能体系统的稳定性研究提供了新的分析范式。

链接: https://arxiv.org/abs/2605.03310
作者: Maksym Nechepurenko,Pavel Shuvalov
机构: Devnull FZCO (Devnull FZCO)
类目: Multiagent Systems (cs.MA); Machine Learning (cs.LG); Trading and Market Microstructure (q-fin.TR)
备注: 31 pages, 7 figures, 4 tables. Code, traces, and production agents publicly released; see Appendix B for repository pinning

点击查看摘要

Abstract:Multi-agent LLM systems fail in production at rates between 41% and 87%, mostly due to coordination defects rather than base-model capability. Existing responses split between cataloguing failure modes empirically and shipping declarative orchestration frameworks as engineering tools; neither delivers a principled mapping from coordination configuration to predictable failure-mode signature. We argue that coordination should be treated as a configurable architectural layer, separable from agent logic and from information access, enabling architectural reasoning rather than only engineering productivity. We instantiate this with an information-controlled design on prediction markets: a single LLM, fixed tools, fixed per-call output cap, and fixed prompt template across five reference coordination configurations, with total compute per question treated as an endogenous architectural output. The Murphy decomposition of the Brier score separates calibration from discriminative power, so configurations leave distinguishable signatures even when aggregate scores coincide. On 100 Polymarket binary markets resolved after the model’s training cutoff (claude-opus-4-6) we report Murphy signatures, a cost-quality Pareto frontier, category-conditioned analysis, and a bootstrap power-projection. Three of five pre-specified predictions are upheld in direction; two configurations dominate the Pareto frontier within this regime; exploratory bootstrap intervals separate consensus alignment from others, though pairwise tests do not survive Bonferroni correction at n=100. We also deploy the same configurations as live agents on Foresight Arena under web-search-enabled conditions, as an on-chain replication channel accumulating in parallel. Harness, trace dataset, and production agents are released. We position this as a methodology-validating first instantiation, not a general cross-model claim. Comments: 31 pages, 7 figures, 4 tables. Code, traces, and production agents publicly released; see Appendix B for repository pinning Subjects: Multiagent Systems (cs.MA); Machine Learning (cs.LG); Trading and Market Microstructure (q-fin.TR) MSC classes: 68T42, 62P20, 62F40 ACMclasses: I.2.11; I.2.7; G.3 Cite as: arXiv:2605.03310 [cs.MA] (or arXiv:2605.03310v1 [cs.MA] for this version) https://doi.org/10.48550/arXiv.2605.03310 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[MA-5] Enwar 3.0: An Agent ic Multi-Modal LLM Orchestrator for Situation-Aware Beamforming Blockage Prediction and Handover Management

【速读】：该论文旨在解决车联网中毫米波（mmWave）连接在动态环境、传感器退化和链路波动下难以保持鲁棒性的问题。其核心解决方案是提出Enwar 3.0框架，关键在于融合多模态感知、基于任务驱动提示的代理型大语言模型（LLM）以及上下文感知的模型选择机制，实现预测波束赋形、遮挡检测与切换管理的协同优化。其中，创新性的合成退化训练管道使传感器健康分类器在摄像头、雷达、激光雷达和GPS输入上达到99%以上的准确率；而经过思维链（CoT）提示和人工反馈微调的LLM能够根据环境上下文动态加载专用模型，并通过结构化提示协调多个专业代理执行复杂决策任务，从而在15种传感器组合中实现88%以上的波束选择准确率、98%以上的遮挡F1分数及87%的推理正确率，显著提升了系统预测性能与可解释性。

链接: https://arxiv.org/abs/2605.03215
作者: Ahmad M. Nazar,Abdulkadir Celik,Asmaa Abdallah,Mohamed Y. Selim,Daji Qiao,Ahmed M. Eltawil
机构: Iowa State University (爱荷华州立大学); King Abdullah University of Science and Technology (KAUST) (阿卜杜拉国王科技大学); University of Southampton (南安普顿大学); Gladiolus Technological Institute (格拉迪奥卢斯技术研究所)
类目: Multiagent Systems (cs.MA)
备注:

点击查看摘要

Abstract:Maintaining robust millimeter-wave (mmWave) connectivity in vehicular networks requires real-time adaptation to environmental dynamics, sensor degradation, and link variability. This paper presents Enwar 3.0, an environment-aware reasoning framework that unifies multi-modal sensing, agentic large language models (LLMs), and context-driven model selection for predictive beamforming, blockage detection, and handover management. Building upon prior iterations of Enwar, the proposed architecture integrates a classifier-driven assessment of sensor health with a primed LLM that orchestrates multiple specialized agents through structured, task-aware prompting. A novel synthetic degradation pipeline enables the training of a sensor degradation classifier that detects real-time impairments across camera, radar, LiDAR, and GPS inputs, achieving over 99% accuracy. The LLM, trained via chain-of-thought (CoT) priming and human-in-the-loop feedback, coordinates agent calls for beam selection, blockage forecasting, and environment perception while dynamically loading sensor-specific models based on environmental context. Extensive evaluations across 15 sensor combinations demonstrate that Enwar 3.0 delivers state-of-the-art performance in both predictive accuracy and interpretability, with beam selection accuracy exceeding 88%, blockage F1-scores surpassing 98%, and reasoning correctness reaching 87% on complex decision prompts. This work establishes a scalable foundation for LLM-integrated wireless systems that reason, perceive, and adapt in real-time.

[MA-6] MARS-DA: A Hierarchical Reinforcement Learning Framework for Risk-Aware Multi-Agent Bidding in Power Grids

【速读】：该论文旨在解决可再生能源高渗透背景下，电力生产商在双结算（Day-Ahead, DA 和实时，RT）电力市场中制定最优投标策略时面临的两大挑战：一是传统强化学习（Reinforcement Learning, RL）方法难以在利润最大化与风险控制之间取得平衡，易因过拟合特定市场条件或忽略DA与RT结算价差的随机性而失效；二是现有方法缺乏对市场状态动态切换的适应能力。解决方案的关键在于提出MARS-DA（Multi-Agent Regime-Switching for Day-Ahead markets）框架，其核心是构建一个分层结构：顶层为元控制器（Meta-Controller），根据市场状态动态融合两个专用基础代理——“安全代理”（Safe Agent）专注于可靠DA电量分配以降低风险，“投机代理”（Speculator Agent）则捕捉RT市场的套利机会以提升收益。通过这种机制，MARS-DA实现了在极端市场波动下仍保持稳健的收益风险比和市场状态匹配度。

链接: https://arxiv.org/abs/2605.03142
作者: Jiayi Chen,Xuan Zhang,Guiling Wang
机构: New Jersey Institute of Technology (新泽西理工学院)
类目: Multiagent Systems (cs.MA)
备注:

点击查看摘要

Abstract:The increasing penetration of renewable energy has introduced substantial volatility into wholesale electricity markets, complicating the optimal bidding strategies for power producers. Traditional Reinforcement Learning (RL) approaches often struggle to balance profit maximization with risk management, frequently overfitting to specific market conditions or failing to account for the stochastic spread between Day-Ahead (DA) and Real-Time (RT) settlements. To address these challenges, this paper makes two primary contributions. First, we introduce and open-source a high-fidelity gymnasium environment for two-settlement electricity market bidding. Grounded in extensive empirical data from the PJM Interconnection, the environment explicitly models the interplay between DA commitments and RT deviations, providing a standardized testbed for general and risk-sensitive agents. Second, we propose MARS-DA (Multi-Agent Regime-Switching for Day-Ahead markets), a novel hierarchical framework that orchestrates distinct sub-policies for risk management and profit seeking. MARS-DA utilizes a top-level Meta-Controller to dynamically blend the actions of two specialized base agents: a “Safe Agent” that optimizes for reliable DA allocation and a “Speculator Agent” that targets volatile RT arbitrage opportunities. Extensive experiments demonstrate that MARS-DA achieves superior risk-adjusted returns compared to state-of-the-art baselines while maintaining robust regime alignment during periods of extreme market volatility.

自然语言处理

[NLP-0] Safety and accuracy follow different scaling laws in clinical large language models

【速读】：该论文旨在解决临床大语言模型（Clinical LLMs）在规模扩展过程中，安全性和准确性之间关系不明确的问题，尤其关注高风险错误、证据矛盾及过度自信等潜在危害可能因模型性能提升而被忽视。其解决方案的关键在于提出SaFE-Scale框架，用于系统评估临床LLM安全性随模型规模、证据质量、检索策略、上下文暴露和推理时计算资源变化的动态表现，并通过构建RadSaFE-200基准测试（包含200道放射学多选题，配有临床医生定义的清洁证据、冲突证据及选项级标签），量化不同部署条件下高风险错误、不安全回答和证据矛盾的发生率。实证结果表明，仅靠增加模型规模或推理计算无法保证安全性提升，而高质量证据与合理检索设计（如标准RAG）才是降低高风险错误的核心因素，且临床关键错误集中于少数特定问题，凸显了安全并非规模的被动副产物，而是由证据质量、检索机制、上下文构造及集体失效行为共同决定的部署属性。

链接: https://arxiv.org/abs/2605.04039
作者: Sebastian Wind,Tri-Thien Nguyen,Jeta Sopa,Mahshad Lotfinia,Sebastian Bickelhaup,Michael Uder,Harald Köstler,Gerhard Wellein,Sven Nebelung,Daniel Truhn,Andreas Maier,Soroosh Tayebi Arasteh
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:

点击查看摘要

Abstract:Clinical LLMs are often scaled by increasing model size, context length, retrieval complexity, or inference-time compute, with the implicit expectation that higher accuracy implies safer behavior. This assumption is incomplete in medicine, where a few confident, high-risk, or evidence-contradicting errors can matter more than average benchmark performance. We introduce SaFE-Scale, a framework for measuring how clinical LLM safety changes across model scale, evidence quality, retrieval strategy, context exposure, and inference-time compute. To instantiate this framework, we introduce RadSaFE-200, a Radiology Safety-Focused Evaluation benchmark of 200 multiple-choice questions with clinician-defined clean evidence, conflict evidence, and option-level labels for high-risk error, unsafe answer, and evidence contradiction. We evaluated 34 locally deployed LLMs across six deployment conditions: closed-book prompting (zero-shot), clean evidence, conflict evidence, standard RAG, agentic RAG, and max-context prompting. Clean evidence produced the strongest improvement, increasing mean accuracy from 73.5% to 94.1%, while reducing high-risk error from 12.0% to 2.6%, contradiction from 12.7% to 2.3%, and dangerous overconfidence from 8.0% to 1.6%. Standard RAG and agentic RAG did not reproduce this safety profile: agentic RAG improved accuracy over standard RAG and reduced contradiction, but high-risk error and dangerous overconfidence remained elevated. Max-context prompting increased latency without closing the safety gap, and additional inference-time compute produced only limited gains. Worst-case analysis showed that clinically consequential errors concentrated in a small subset of questions. Clinical LLM safety is therefore not a passive consequence of scaling, but a deployment property shaped by evidence quality, retrieval design, context construction, and collective failure behavior.

[NLP-1] OpenSeeker-v2: Pushing the Limits of Search Agents with Informative and High-Difficulty Trajectories

【速读】：该论文旨在解决前沿搜索代理（search agent）训练中高度依赖工业级资源密集型流程的问题，即传统方法通常需要预训练（pre-training）、持续预训练（continual pre-training, CPT）、监督微调（supervised fine-tuning, SFT）和强化学习（reinforcement learning, RL）等多阶段复杂流水线。其解决方案的关键在于：通过引入三种简单但有效的数据合成策略——扩展知识图谱规模以增强探索能力、扩大工具集以提升功能广度、以及严格低步数过滤以确保轨迹质量——仅使用10.6k高质量样本进行SFT即可实现超越工业级复杂流水线的性能表现，从而证明了高效且可复现的学术研究路径在构建高性能搜索代理中的可行性。

链接: https://arxiv.org/abs/2605.04036
作者: Yuwen Du,Rui Ye,Shuo Tang,Keduan Huang,Xinyu Zhu,Yuzhu Cai,Siheng Chen
机构: Shanghai Jiao Tong University (上海交通大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 7 pages

点击查看摘要

Abstract:Deep search capabilities have become an indispensable competency for frontier Large Language Model (LLM) agents, yet their development remains dominated by industrial giants. The typical industry recipe involves a highly resource-intensive pipeline spanning pre-training, continual pre-training (CPT), supervised fine-tuning (SFT), and reinforcement learning (RL). In this report, we show that when fueled with informative and high-difficulty trajectories, a simple SFT approach could be surprisingly powerful for training frontier search agents. By introducing three simple data synthesis modifications: scaling knowledge graph size for richer exploration, expanding the tool set size for broader functionality, and strict low-step filtering, we establish a stronger baseline. Trained on merely 10.6k data points, our OpenSeeker-v2 achieves state-of-the-art performance across 4 benchmarks (30B-sized agents with ReAct paradigm): 46.0% on BrowseComp, 58.1% on BrowseComp-ZH, 34.6% on Humanity’s Last Exam, and 78.0% on xbench, surpassing even Tongyi DeepResearch trained with heavy CPT+SFT+RL pipeline, which achieves 43.4%, 46.7%, 32.9%, and 75.0%, respectively. Notably, OpenSeeker-v2 represents the first state-of-the-art search agent within its model scale and paradigm to be developed by a purely academic team using only SFT. We are excited to open-source the OpenSeeker-v2 model weights and share our simple yet effective findings to make frontier search agent research more accessible to the community.

[NLP-2] EQUITRIAGE: A Fairness Audit of Gender Bias in LLM -Based Emergency Department Triage

【速读】：该论文旨在解决生成式 AI（Generative AI）在急诊分诊（Emergency Department Triage, EDT）决策支持中可能存在的性别偏见问题，即大语言模型（Large Language Models, LLMs）是否会复制甚至放大人类临床实践中已知的性别不平等现象。其关键解决方案在于构建并实施EQUITRIAGE公平性审计框架，通过在MIMIC-IV-ED真实病例数据集上对五种主流LLM进行系统评估，采用配对性别反转的反事实样本（counterfactual vignettes）和多种提示策略，量化模型输出的翻转率（flip rate）、性别校准偏差（gender calibration）及群体间公平性指标。研究发现：不同模型表现出显著差异的偏见模式，且部分模型虽具备良好校准性能，却仍存在方向性性别偏差；干预措施如人口统计学屏蔽（demographic blinding）效果因模型而异，揭示了公平性属性（如组内一致性、反事实不变性、性别校准）之间的解耦关系，并强调应在临床部署前对每种模型进行定制化反事实审计。

链接: https://arxiv.org/abs/2605.03998
作者: Richard J. Young,Alice M. Matthews
机构: University of Nevada, Las Vegas, Lee Business School (内华达大学拉斯维加斯分校李商学院); DeepNeuro AI (DeepNeuro AI); Concorde Career Colleges, Dept. of Cardiovascular and Medical Diagnostic Sonography (康科德职业学院心血管与医学诊断超声部门)
类目: Computation and Language (cs.CL); Computers and Society (cs.CY)
备注: 37 pages, 10 figures, 13 tables. Code and analysis scripts available upon publication. Data: PhysioNet credentialed access (MIMIC-IV-ED v2.2 and MIMIC-IV v3.1, BIDMC IRB #2001P001699)

点击查看摘要

Abstract:Emergency department triage assigns patients an acuity score that determines treatment priority, and clinical evidence documents persistent gender disparities in human acuity assessment. As hospitals pilot large language models (LLMs) as triage decision support, a critical question is whether these models reproduce or mitigate known biases. We present EQUITRIAGE, a fairness audit of LLM-based ESI assignment evaluating five models (Gemini-3-Flash, Nemotron-3-Super, DeepSeek-V3.1, Mistral-Small-3.2, GPT-4.1-Nano) across 374,275 evaluations on 18,714 MIMIC-IV-ED vignettes under four prompt strategies. Of 9,368 originals, 9,346 are paired with a gender-swapped counterfactual. All five models produced flip rates above a pre-registered 5% threshold (9.9% to 43.8%). Two showed directional female undertriage (DeepSeek F/M 2.15:1, Gemini 1.34:1); two were near-parity; one had high sensitivity with weak male-direction asymmetry. DeepSeek’s directional bias coexisted with a low outcome-linked calibration gap (0.013 against MIMIC-IV admission), a Chouldechova-style dissociation between within-group calibration and between-pair counterfactual invariance. Demographic blinding reduced Gemini’s flip rate to 0.5%; an age-preserving blind variant left DeepSeek with residual F/M 1.25, implicating age as a residual channel. Chain-of-thought prompting degraded accuracy for all five models. A two-model ablation reveals opposite underlying mechanisms for the same directional phenotype: in Gemini the signal is emergent in the combined name+gender swap, while in DeepSeek the gender token alone carries it. EQUITRIAGE shows that group parity, counterfactual invariance, and gender calibration are distinct fairness properties, that intervention effectiveness is model-dependent, and that per-model counterfactual auditing should precede clinical deployment.

[NLP-3] Logical Consistency as a Bridge: Improving LLM Hallucination Detection via Label Constraint Modeling between Responses and Self-Judgments ACL2026

【速读】：该论文旨在解决大语言模型（Large Language Models, LLMs）中存在的事实幻觉（factual hallucinations）问题，即模型生成内容与真实信息不符的现象，这严重影响了其在实际应用中的可靠性。现有方法通常仅从微观层面的内在不确定性或宏观层面的自评判断中提取单一特征，未能充分融合神经特征与符号推理之间的耦合关系。解决方案的关键在于提出LaaB（Logical Consistency-as-a-Bridge）框架，通过引入“元判断”（meta-judgment）机制，将符号标签映射回特征空间，并利用响应与元判断标签在语义上一致或相反的逻辑桥梁，实现双视角信号（神经特征与符号判断）的相互学习与对齐，从而增强幻觉检测的整体性能。

链接: https://arxiv.org/abs/2605.03971
作者: Hao Mi,Qiang Sheng,Shaofei Wang,Beizhe Hu,Yifan Sun,Zhengjia Wang,Hengqi Zeng,Yang Li,Danding Wang,Juan Cao
机构: 未知
类目: Computation and Language (cs.CL)
备注: ACL 2026 Main Conference

点击查看摘要

Abstract:Large Language Models (LLMs) are prone to factual hallucinations, risking their reliability in real-world applications. Existing hallucination detectors mainly extract micro-level intrinsic patterns for uncertainty quantification or elicit macro-level self-judgments through verbalized prompts. However, these methods address only a single facet of the hallucination, focusing either on implicit neural uncertainty or explicit symbolic reasoning, thereby treating these inherently coupled behaviors in isolation and failing to exploit their interdependence for a holistic view. In this paper, we propose LaaB (Logical Consistency-as-a-Bridge), a framework that bridges neural features and symbolic judgments for hallucination detection. LaaB introduces a “meta-judgment” process to map symbolic labels back into the feature space. By leveraging the inherent logical bridge where response and meta-judgment labels are either the same or opposite based on the self-judgment’s semantics, LaaB aligns and integrates dual-view signals via mutual learning and enhances the hallucination detection. Extensive experiments on 4 public datasets, across 4 LLMs, against 8 baselines demonstrate the superiority of LaaB.

[NLP-4] Feature-Augmented Transformers for Robust AI-Text Detection Across Domains and Generators ICML2026

【速读】：该论文旨在解决生成式 AI (Generative AI) 文本检测模型在面对分布偏移（distribution shift）时鲁棒性不足的问题，即模型在训练数据分布之外的下游场景中性能显著下降。其关键解决方案是采用基于Transformer的检测器架构，在HC3 PLUS数据集上训练并固定单一决策阈值以最大化验证集上的平衡准确率（balanced accuracy），从而模拟实际部署中的稳定阈值设定；同时引入基于注意力机制的语言特征融合（Feature Augmentation via Attention-based Linguistic Feature Fusion），增强模型对跨领域、跨生成器场景的适应能力。实验表明，该方法在M4多源数据集上实现85.9%的平衡准确率，优于零样本基线模型达7.22个百分点，且具备高稳定性，证明了特征增强与现代DeBERTa骨干网络的有效性。

链接: https://arxiv.org/abs/2605.03969
作者: Mohamed Mady,Johannes Reschke,Björn Schuller
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 8 pages, 4 figures, 5 tables. Submitted to ICML 2026

点击查看摘要

Abstract:AI-generated text is nowadays produced at scale across domains and heterogeneous generation pipelines, making robustness to distribution shift a central requirement for supervised binary detectors. We train transformer-based detectors on HC3 PLUS and calibrate a single decision threshold by maximising balanced accuracy on held-out validation; this threshold is then kept fixed for all downstream test distributions, revealing domain- and generator-dependent error asymmetries under shift. We evaluate in-domain on HC3 PLUS, under cross-dataset transfer to the multi-domain, multi-generator M4 benchmark, and on the external AI-Text-Detection-Pile. Although base models achieve near-ceiling in-domain performance (up to 99.5% balanced accuracy), performance under shift is brittle and strongly model-dependent. Feature augmentation via attention-based linguistic feature fusion improves transfer, with our best model (DeBERTa-v3-base+FeatAttn) achieving 85.9% balanced accuracy on M4. Multi-seed experiments confirm high stability. Under the same fixed-threshold protocol, our model outperforms strong zero-shot baselines by up to +7.22 points. Category-level ablations further show that readability and vocabulary features contribute most to robustness under shift. Overall, these results demonstrate that feature augmentation and a modern DeBERTa backbone significantly outperform earlier BERT/RoBERTa models, while the fixed-threshold protocol provides a more realistic and informative assessment of practical detector robustness.

[NLP-5] ransformers with Selective Access to Early Representations

【速读】：该论文旨在解决Transformer架构中早期层特征（如第一层的值投影 $V_1$ ）在深层传播过程中逐渐丢失的问题，即低层语义信息难以被后续层有效利用。传统方法通过静态值残差（static value residuals）将 $V_1$ 均匀地注入所有token和注意力头，但这种方式缺乏对不同上下文、位置或注意力头的适应性。为提升早期表示的可重用性并减少冗余计算，作者提出Selective Access Transformer (SATFormer)，其核心创新在于将早期表示的再利用建模为一个上下文感知的门控检索问题：保留原始 $V_1$ 路径的同时，引入一个轻量级门控机制来动态控制每个token、head和深度层级对 $V_1$ 的访问强度。该方案在保持与标准Transformer相近的内存和吞吐量的前提下，显著提升了模型在检索密集型任务上的表现，且门控分析揭示了稀疏、深度依赖、头特异性和类别敏感的访问模式，验证了其选择性复用早期表示的有效性。

链接: https://arxiv.org/abs/2605.03953
作者: Skye Gunasekaran,Téa Wright,Rui-Jie Zhu,Jason Eshraghian
机构: UC Santa Cruz (加州大学圣克鲁兹分校); UC Berkeley (加州大学伯克利分校)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:

点击查看摘要

Abstract:Several recent Transformer architectures expose later layers to representations computed in the earliest layers, motivated by the observation that low-level features can become harder to recover as the residual stream is repeatedly transformed through depth. The cheapest among these methods add static value residuals: learned mixing coefficients that expose the first-layer value projection V_1 uniformly across tokens and heads. More expressive dense or dynamic alternatives recover finer-grained access, but at higher memory cost and lower throughput. The usefulness of V_1 is unlikely to be constant across tokens, heads, and contexts; different positions plausibly require different amounts of access to early lexical or semantic information. We therefore treat early-representation reuse as a retrieval problem rather than a connectivity problem, and introduce Selective Access Transformer (SATFormer), which preserves the first-layer value pathway while controlling access with a context-dependent gate. Across models from 130M to 1.3B parameters, SATFormer consistently improves validation loss and zero-shot accuracy over the static value-residual and Transformer baselines. Its strongest gains appear on retrieval-intensive benchmarks, where it improves over static value residuals by approximately 1.5 average points, while maintaining throughput and memory usage close to the baseline Transformer. Gate analyses suggest sparse, depth-dependent, head-specific, and category-sensitive access patterns, supporting the interpretation that SATFormer learns selective reuse of early representations rather than uniform residual copying. Our code is available at this https URL.

[NLP-6] he Counterexample Game: Iterated Conceptual Analysis and Repair in Language Models

【速读】：该论文旨在探究大语言模型（Large Language Models, LMs）是否能够执行哲学方法论中的核心任务——概念分析（Conceptual Analysis），即通过提出定义并借助反例进行迭代修正来完善概念界定。其解决方案的关键在于构建一种“反例-修复链”（counterexample-repair chains）的迭代机制：由一个模型实例生成对初始定义的反例，另一个模型实例据此修复定义，循环往复。实验表明，尽管LM生成的反例中存在大量被专家人类和LM判别器认为无效的案例，但LM判别器接受的比例约为人类的两倍；同时，多轮迭代虽使定义文本变长，却未显著提升准确性，且部分概念难以形成稳定定义。这说明LM具备一定哲学推理能力，但该迭代过程在短期内即呈现边际收益递减趋势，可作为评估LM持续高阶哲学推理能力的重要测试范式。

链接: https://arxiv.org/abs/2605.03936
作者: Daniel Drucker,Kyle Mahowald
机构: The University of Texas at Austin (德克萨斯大学奥斯汀分校)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Conceptual analysis – proposing definitions and refining them through counterexamples – is central to philosophical methodology. We study whether language models can perform this task through iterated analysis and repair chains: one model instance generates counterexamples to a proposed definition, another repairs the definition, and the process repeats. Across 20 concepts and thousands of counterexample-repair cycles, we find that, although many LM-generated counterexamples are judged invalid by both expert humans and an LM judge, the LM judge accepts roughly twice as many as humans do. Nonetheless, per-item validity judgments are moderately consistent across humans and between humans and the LM. We further find that extended iteration produces increasingly verbose definitions without improving accuracy. We also see that some concepts resist stable definitions in general. These findings suggest that while LMs can engage in philosophical reasoning, the counterexample-repair loop hits diminishing returns quickly and could be a fruitful test case for evaluating whether LMs can sustain high-level iterated philosophical reasoning.

[NLP-7] Atomic Fact-Checking Increases Clinician Trust in Large Language Model Recommendations for Oncology Decision Support: A Randomized Controlled Trial

【速读】：该论文旨在解决高风险临床决策中医生对人工智能（AI）治疗建议信任度不足的问题。传统可解释性方法（explainability approaches）虽能提供一定程度的透明度，但未能显著提升医生的信任水平。其解决方案的关键在于引入“原子事实核查”（atomic fact-checking）机制——将AI推荐拆解为可独立验证的声明，并明确链接至原始指南文档，从而增强医生对AI输出的可信度与理解深度。实验证明，该方法相较于传统方式产生显著更大的信任提升效应（Cohen’s d = 0.94），使表达信任的医生比例从26.9%大幅提升至66.5%，表明结构化、可溯源的证据链是提高临床AI采纳的关键路径。

链接: https://arxiv.org/abs/2605.03916
作者: Lisa C. Adams,Linus Marx,Erik Thiele Orberg,Keno Bressem,Sebastian Ziegelmayer,Denise Bernhardt,Markus Graf,Marcus R. Makowski,Stephanie E. Combs,Florian Matthes,Jan C. Peeken
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 28 pages, 5 figures, 2 tables, supplement will be made available upon original publication

点击查看摘要

Abstract:Question: Does atomic fact-checking, which decomposes AI treatment recommendations into individually verifiable claims linked to source guideline documents, increase clinician trust compared to traditional explainability approaches? Findings: In this randomized trial of 356 clinicians generating 7,476 trust ratings, atomic fact-checking produced a large effect on trust (Cohen’s d = 0.94), increasing the proportion of clinicians expressing trust from 26.9% to 66.5%. Traditional transparency mechanisms showed a dose-response gradient of improvement over baseline (d = 0.25 to 0.50). Meaning: Decomposing AI recommendations into individually verifiable claims linked to source guidelines produces substantially higher clinician trust than traditional explainability approaches in high-stakes clinical decisions. Comments: 28 pages, 5 figures, 2 tables, supplement will be made available upon original publication Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) Cite as: arXiv:2605.03916 [cs.CL] (or arXiv:2605.03916v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2605.03916 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[NLP-8] Steer Like the LLM : Activation Steering that Mimics Prompting ICML2026

【速读】：该论文旨在解决激活干预（activation steering）方法在推理时对大型语言模型进行引导（steering）的效果普遍弱于提示引导（prompt-based steering）的问题。其核心挑战在于现有激活干预方法未能忠实模拟提示引导中对特定token施加强干预、而对其他token影响极小的机制。解决方案的关键在于提出Prompt Steering Replacement (PSR) 模型，该模型通过从模型激活中直接估计每个token的专属引导系数，并训练其模仿提示引导的行为，从而在多个基准测试中显著提升激活引导的性能，尤其在控制高连贯性输出的前提下表现优异。

链接: https://arxiv.org/abs/2605.03907
作者: Geert Heyman,Frederik Vandeputte
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted to ICML 2026

点击查看摘要

Abstract:Large language models can be steered at inference time through prompting or activation interventions, but activation steering methods often underperform compared to prompt-based approaches. We propose a framework that formulates prompt steering as a form of activation steering and investigates whether distilling successful prompt steering behavior into simpler, interpretable models can close this gap. Our analysis reveals that popular activation steering methods are not faithful to the mechanics of prompt steering, which applies strong interventions on some tokens while barely affecting others. Based on these insights, we introduce Prompt Steering Replacement (PSR) models that estimate token-specific steering coefficients from the activations themselves and are trained to imitate prompt-based interventions. Experiments on three steering benchmarks across multiple language models show that PSR models outperform existing activation steering methods, especially when controlling for high-coherence completions, and also compare favorably to prompting on AxBench and persona steering.

[NLP-9] CC-OCR V2: Benchmarking Large Multimodal Models for Literacy in Real-world Document Processing

【速读】：该论文旨在解决当前大型多模态模型（Large Multimodal Models, LMMs）在真实世界文档处理场景中性能表现不明确的问题，因为现有OCR（光学字符识别）基准测试的任务范围与实际应用脱节，且假设了同质的采集条件。解决方案的关键在于提出一个名为CC-OCR V2的新基准，其聚焦于企业级文档处理的实际任务，包含7,093个高难度样本，覆盖文本识别、文档解析、文档定位、关键信息提取和文档问答五大核心OCR任务，并引入以往基准中被忽视的困难案例和边缘情况。通过在14种先进LMM上进行广泛实验，研究发现即使是最先进的模型也普遍存在跨任务和跨场景的显著性能下降，从而揭示了当前基准评估结果与真实应用需求之间的显著差距。

链接: https://arxiv.org/abs/2605.03903
作者: Zhipeng Xu,Junhao Ji,Zulong Chen,Zhenghao Liu,Qing Liu,Chunyi Peng,Zubao Qin,Ze Xu,Jianqiang Wan,Jun Tang,Zhibo Yang,Shuai Bai,Dayiheng Liu
机构: Alibaba Group; Northeastern University
类目: Computation and Language (cs.CL)
备注: Work in progress

点击查看摘要

Abstract:Large Multimodal Models (LMMs) have recently shown strong performance on Optical Character Recognition (OCR) tasks, demonstrating their promising capability in document literacy. However, their effectiveness in real-world applications remains underexplored, as existing benchmarks adopt task scopes misaligned with practical applications and assume homogeneous acquisition conditions. To address this gap, we introduce CC-OCR V2, a comprehensive and challenging OCR benchmark tailored to real-world document processing. CC-OCR V2 focuses on practical enterprise document processing tasks and incorporates hard and corner cases that are critical yet underrepresented in prior benchmarks, covering 5 major OCR-centric tracks: text recognition, document parsing, document grounding, key information extraction, and document question answering, comprising 7,093 high-difficulty samples. Extensive experiments on 14 advanced LMMs reveal that current models fall short of real-world application requirements. Even state-of-the-art LMMs exhibit substantial performance degradation across diverse tasks and scenarios. These findings reveal a significant gap between performance on current benchmarks and effectiveness in real-world applications. We release the full dataset and evaluation toolkit at this https URL.

[NLP-10] Correct Is Not Enough: Training Reasoning Planners with Executor-Grounded Rewards

【速读】：该论文旨在解决强化学习中仅以最终答案正确性作为奖励信号时，无法有效评估推理轨迹（reasoning trace）是否忠实、可靠或对后续模型消费有用的问题。这种“结果导向”的奖励机制可能导致模型学习到看似合理但实质错误的推理路径，或因奖励短路行为而高估推理效果，尤其在多步骤系统中会传播中间状态的错误。解决方案的关键在于提出TraceLift框架，将推理视为可消耗的中间产物，并设计执行器接地（executor-grounded）的奖励机制：训练阶段由规划器生成带标签的推理过程，冻结的执行器将其转化为最终输出以获取验证器反馈，同时通过衡量该推理对同一执行器性能提升的程度，乘以基于评分量表的推理奖励模型（Reasoning Reward Model, RM）得分，从而同时奖励高质量且对执行器有用的推理轨迹。这一机制促使模型不仅关注推理外观合理性，更注重其实际功能性。

链接: https://arxiv.org/abs/2605.03862
作者: Tianyang Han,Hengyu Shi,Junjie Hu,Xu Yang,Zhiling Wang,Junhao Su
机构: 1D4 Lab; Independent Researcher
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 36 pages

点击查看摘要

Abstract:Reinforcement learning with verifiable rewards has become a common way to improve explicit reasoning in large language models, but final-answer correctness alone does not reveal whether the reasoning trace is faithful, reliable, or useful to the model that consumes it. This outcome-only signal can reinforce traces that are right for the wrong reasons, overstate reasoning gains by rewarding shortcuts, and propagate flawed intermediate states in multi-step systems. To this end, we propose TraceLift, a planner-executor training framework that treats reasoning as a consumable intermediate artifact. During planner training, the planner emits tagged reasoning. A frozen executor turns this reasoning into the final artifact for verifier feedback, while an executor-grounded reward shapes the intermediate trace. This reward multiplies a rubric-based Reasoning Reward Model (RM) score by measured uplift on the same frozen executor, crediting traces that are both high-quality and useful. To make reasoning quality directly learnable, we introduce TRACELIFT-GROUPS, a rubric-annotated reason-only dataset built from math and code seed problems. Each example is a same-problem group containing a high-quality reference trace and multiple plausible flawed traces with localized perturbations that reduce reasoning quality or solution support while preserving task relevance. Extensive experiments on code and math benchmarks show that this executor-grounded reasoning reward improves the two-stage planner-executor system over execution-only training, suggesting that reasoning supervision should evaluate not only whether a trace looks good, but also whether it helps the model that consumes it.

[NLP-11] MCJudgeBench: A Benchmark for Constraint-Level Judge Evaluation in Multi-Constraint Instruction Following ACL2026

【速读】：该论文旨在解决大语言模型（Large Language Model, LLM）在多约束指令遵循任务中，对响应是否满足多个独立约束条件的判断可靠性问题。现有评估方法通常仅基于整体响应进行评判，忽略了各约束维度上的细粒度表现差异，导致无法准确识别模型在特定约束类别（如“部分正确”或“不正确”）下的失效模式。解决方案的关键在于提出MCJudgeBench基准测试平台，其核心特征包括：每条实例包含显式的约束列表、逐约束的黄金标签（yes/partial/no）、可控的响应扰动机制，以及用于检验判断稳定性的一系列提示变体。通过该框架，研究者能够区分内在随机解码导致的不一致性与提示或响应扰动引发的过程性不一致性，从而揭示LLM判官在不同约束层级上的可靠性和潜在缺陷。

链接: https://arxiv.org/abs/2605.03858
作者: Jaeyun Lee,Junyoung Koh,Zeynel Tok,Hunar Batra,Ronald Clark
机构: University of Oxford (牛津大学); Yonsei University (延世大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: GEM Workshop at ACL 2026

点击查看摘要

Abstract:Multi-constraint instruction following requires verifying whether a response satisfies multiple individual requirements, yet LLM judges are often assessed only through overall-response judgments. We introduce MCJudgeBench, a benchmark for constraint-level judge evaluation in multi-constraint instruction following. Each instance includes an instruction, a candidate response, an explicit constraint list, per-constraint gold labels in yes, partial, no, and controlled response-side perturbations. The evaluation protocol further includes evaluation prompt variants to test judge stability. We evaluate proprietary and open-source LLM judges using both correctness and inconsistency metrics, distinguishing intrinsic inconsistency under stochastic decoding from procedural inconsistency under prompt and response perturbations. Our results show that judge reliability has multiple dimensions: strong overall performance does not guarantee equally reliable detection across label categories, especially for rarer partial and no cases. Judges with higher correctness do not always have lower inconsistency. Evaluation with reasoning improves correctness but does not uniformly improve stability. These findings motivate evaluating LLM judges at the constraint level to study these failure modes.

[NLP-12] Reproducing Complex Set-Compositional Information Retrieval SIGIR2026

【速读】：该论文旨在解决当前信息检索系统在处理复杂查询（如涉及合取、析取和排除的集合组合查询）时，是否真正满足语义约束而非依赖“语义捷径”（semantic shortcuts）的问题。其核心挑战在于现有检索范式在标准基准（如QUEST）上表现优异，但缺乏对约束条件严格依赖的验证。解决方案的关键在于引入LIMIT+这一受控基准，其中相关性完全由任意属性谓词和约束满足决定，而非预训练知识；通过对比经典词汇检索与神经检索方法在LIMIT+上的性能崩溃现象（如最强QUEST方法从Recall@100 ≈ 0.42降至<0.02），揭示了当前生成式AI（Generative AI）驱动的检索方法在复杂逻辑推理任务中的局限性，并强调了基于稀疏表示和词法匹配的方法在保持稳定性和可解释性方面的优势。

链接: https://arxiv.org/abs/2605.03824
作者: Vincent Degenhart,Dewi Timman,Arjen P. de Vries,Faegheh Hasibi,Mohanna Hoveyda
机构: Radboud University Nijmegen (拉德布德大学奈梅亨分校)
类目: Computation and Language (cs.CL)
备注: Accepted to SIGIR 2026, Reproducibility Track

点击查看摘要

Abstract:Complex information needs may involve set-compositional queries using conjunction, disjunction, and exclusion, yet it remains unclear whether current retrieval paradigms genuinely satisfy such constraints or exploit `semantic shortcuts’. We conduct a reproducibility study to benchmark major retrieval families and reasoning-targeted methods on QUEST and QUEST+Variants, and introduce LIMIT+, a controlled benchmark where relevance depends on arbitrary attribute predicates and constraint satisfaction, and less on pretrained knowledge. Our findings show that (i) on QUEST, the best neural retrievers achieve an effectiveness that is more than double what can be achieved with BM25 (Recall@100 0.41 vs.\ 0.20), but reasoning-targeted methods like ReasonIR and Search-R1 do not outperform general-purpose retrievers uniformly; (ii) on LIMIT+, gains fail to transfer, where the strongest QUEST method collapses from Recall@100 \approx 0.42 to below 0.02, while classic lexical retrieval gains to \sim 0.96. Lastly, (iii) stratifying by compositional depth reveals a consistent degradation across all methods, where algebraic sparse and lexical methods show more stable performance while dense approaches collapse. We release code and LIMIT+ data generation scripts to support future reproducibility and controlled evaluation.

[NLP-13] Agent ic-imodels: Evolving agent ic interpretability tools via autoresearch

【速读】：该论文旨在解决当前生成式数据科学（Agentic Data Science, ADS）系统在自主执行数据分析任务时面临的可解释性瓶颈问题，即现有统计工具虽对人类可解释，但难以被代理（agent）高效理解与利用。解决方案的关键在于提出Agent-optimized models（Agentic-imodels），其核心创新是构建一个基于大语言模型（LLM）的新型可解释性度量标准——通过测试模型字符串表示是否具备“可模拟性”（simulatable），即LLM能否仅凭读取模型输出字符串回答关于其行为的问题，并在此基础上演化出一套兼容scikit-learn的回归器库，使其在保持预测性能的同时显著提升代理面向的可解释性，从而推动端到端ADS系统的性能提升。

链接: https://arxiv.org/abs/2605.03808
作者: Chandan Singh,Yan Shuo Tan,Weijia Xu,Zelalem Gero,Weiwei Yang,Michel Galley,Jianfeng Gao
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:

点击查看摘要

Abstract:Agentic data science (ADS) systems are rapidly improving their capability to autonomously analyze, fit, and interpret data, potentially moving towards a future where agents conduct the vast majority of data-science work. However, current ADS systems use statistical tools designed to be interpretable by humans, rather than interpretable by agents. To address this, we introduce Agentic-imodels, an agentic autoresearch loop that evolves data-science tools designed to be interpretable by agents. Specifically, it develops a library of scikit-learn-compatible regressors for tabular data that are optimized for both predictive performance and a novel LLM-based interpretability metric. The metric measures a suite of LLM-graded tests that probe whether a fitted model’s string representation is “simulatable” by an LLM, i.e. whether the LLM can answer questions about the model’s behavior by reading its string output alone. We find that the evolved models jointly improve predictive performance and agent-facing interpretability, generalizing to new datasets and new interpretability tests. Furthermore, these evolved models improve downstream end-to-end ADS, increasing performance for Copilot CLI, Claude Code, and Codex on the BLADE benchmark by up to 73%

[NLP-14] Natural Language Processing: A Comprehensive Practical Guide from Tokenisation to RLHF

【速读】：该论文旨在解决现代自然语言处理（Natural Language Processing, NLP）教学与实践脱节的问题，即如何系统性地引导研究人员和开发者从基础技术到前沿大语言模型（Large Language Models, LLMs）的完整工程化落地。其核心解决方案是设计一套以研究为导向的实践课程体系，涵盖从分词（tokenisation）、向量化（vectorisation）到微调（fine-tuning）、检索增强生成（retrieval-augmented generation）及人类反馈强化学习（reinforcement learning from human feedback, RLHF）的全流程，并强调代码、模型和报告的公开可复现性，依托Hugging Face生态实现开放权重模型替代封闭API。此外，通过为低资源语言（如塔吉克语和鞑靼语）构建子词分词器、嵌入向量、词汇数据库等语言学资源，验证了该框架在数据稀缺场景下的适应能力，从而推动NLP方法在多样化语境中的公平部署与评估。

链接: https://arxiv.org/abs/2605.03799
作者: Mullosharaf K. Arabov
机构: 未知
类目: Computation and Language (cs.CL)
备注: Preprint

点击查看摘要

Abstract:This preprint presents a systematic, research-oriented practicum that guides the reader through the entire modern NLP pipeline: from tokenisation and vectorisation to fine-tuning of large language models, retrieval-augmented generation, and reinforcement learning from human feedback. Twelve hands-on sessions combine concise theory with detailed implementation plans, formalised evaluation metrics, and transparent assessment criteria. The work is not a conventional textbook: it is designed as a reproducible research artefact where every session requires publishing code, models, and reports in public repositories. All experiments are conducted on a single evolving corpus, and the work advocates open-weight models over commercial APIs, with special attention to the Hugging Face ecosystem. The material is enriched by original research on low-resource languages, incorporating linguistic resources for Tajik and Tatar (subword tokenisers, embeddings, lexical databases, and transliteration benchmarks), demonstrating how modern NLP can be adapted to data-scarce environments. Designed for senior undergraduates, graduate students, and practising developers seeking to implement, compare, and deploy methods from classical ML to state-of-the-art LLM-based systems.

[NLP-15] riBench-Ko: Evaluating LLM Risks in Judicial Workflows

【速读】：该论文旨在解决当前大型语言模型（Large Language Models, LLMs）在司法场景中部署时缺乏真实、系统性风险评估的问题。现有基准多依赖代理任务（如律师资格考试或分类任务），无法反映日常司法实践中模型行为的真实风险。其解决方案的关键在于构建并公开发布TriBench-Ko——一个面向韩语司法场景的基准测试集，涵盖法律推理摘要、判例检索、法律问题提取和证据分析四项核心任务，并从准确性（幻觉、遗漏、法条误用）、偏见（人口统计偏差、过度合规）、不一致性（提示敏感性、非确定性）及裁决越权等维度系统评估模型风险。该设计基于真实司法判决数据，使评估兼具任务性能与风险识别双重能力，从而为LLM在司法领域的安全应用提供诊断依据与改进方向。

链接: https://arxiv.org/abs/2605.03792
作者: Haesung Lee,Gyubin Choi,Eun-Ju Lee,So-Min Lee,Youkang Ko,Dogyoon Lim,Sung-Kyoung Jang,Yohan Jo
机构: Graduate School of Data Science, Seoul National University (首尔国立大学数据科学研究生院); Center for Trustworthy AI, Seoul National University (首尔国立大学可信人工智能中心); School of Law, Seoul National University (首尔国立大学法学院); Responsible AI Team, KT (KT负责任人工智能团队)
类目: Computation and Language (cs.CL)
备注: 10 pages

点击查看摘要

Abstract:Large language models (LLMs) are increasingly integrated into legal workflows. However, existing benchmarks primarily address proxy tasks, such as bar examination performance or classification, which fail to capture the performance and risks inherent in day-to-day judicial processes. To address this, we publicly release TriBench-Ko, a Korean benchmark designed to evaluate potential deployment risks of LLMs within the context of verified judicial task requirements. It covers four core tasks: jurisprudence summarization, precedent retrieval, legal issue extraction, and evidence analysis. It jointly assesses model behavior across multiple deployment risk categories, including inaccuracy (hallucination, omission, statutory misapplication), biases (demographic, overcompliance), inconsistencies (prompt sensitivity, non-determinism), and adjudicative overreach. Each item is structured to systematically assess both task performance and a specific risk type based on real judicial decisions. Our evaluation of a range of contemporary LLMs reveals that many models frequently manifest significant risks, most notably struggling with precedent retrieval and failing to capture critical legal information. We provide a comprehensive diagnosis of these LLMs and pinpoint critical areas where LLM-generated outputs in judicial contexts necessitate rigorous inspection and caution. Our dataset and code are available at this https URL

[NLP-16] ask Vector Geometry Underlies Dual Modes of Task Inference in Transformers

【速读】：该论文旨在解决生成式 AI（Generative AI）模型中任务向量（task vector）几何结构与外部模型行为之间缺乏严谨理论联系的问题，具体包括：如何由训练分布塑造任务向量的几何特性，以及何种几何结构能够支持分布外（out-of-distribution, OOD）泛化。其解决方案的关键在于构建一个受控的合成实验设置，通过从零开始训练小型 Transformer 模型学习潜在任务序列分布，从而实现对任务向量几何与模型推理模式之间关系的数学刻画；研究发现，模型内部可同时存在两种推理机制——分布内行为由贝叶斯任务检索驱动（通过任务向量的凸组合实现），而分布外行为则源于外推式任务学习（其表示位于几乎正交于任务向量子空间的低维子空间），揭示了任务向量几何、训练分布与泛化能力之间的紧密关联。

链接: https://arxiv.org/abs/2605.03780
作者: Hao Yan,Haolin Yang,Yiqiao Zhong
机构: University of Wisconsin–Madison(威斯康星大学麦迪逊分校); University of Chicago(芝加哥大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL); Machine Learning (stat.ML)
备注:

点击查看摘要

Abstract:Transformers are effective at inferring the latent task from context via two inference modes: recognizing a task seen during training, and adapting to a novel one. Recent interpretability studies have identified from middle-layer representations task-specific directions, or task vectors, that steer model behavior. However, a lack of rigorous foundations hinders connecting internal representations to external model behavior: existing work fails to explain how task-vector geometry is shaped by the training distribution, and what geometry enables out-of-distribution (OOD) generalization. In this paper, we study these questions in a controlled synthetic setting by training small transformers from scratch on latent-task sequence distributions, which allows a principled mathematical characterization. We show that two inference modes can coexist within a single model. In-distribution behavior is governed by Bayesian task retrieval, implemented internally through convex combinations of learned task vectors. OOD behavior, by contrast, arises through extrapolative task learning, whose representations occupy a subspace nearly orthogonal to the task-vector subspace. Taken together, our results suggest that task-vector geometry, training distributions, and generalization behaviors are closely related.

[NLP-17] Benchmarking Parameter-Efficient Fine-Tuning of Large Language Models for Low-Resource Tajik Text Generation with the Tajik Web Corpus

【速读】：该论文旨在解决生成式大语言模型（Generative Large Language Models）在低资源语言——塔吉克语（Tajik language）上的适配问题，尤其针对其数字文本资源匮乏和缺乏有效的微调策略。解决方案的关键在于构建并公开发布目前最大的塔吉克语网络语料库（Tajik Web Corpus），包含约11.1亿字符的31.9万篇文档，并系统评估了多种预训练模型架构（自回归、编码器-解码器、仅编码器）与参数高效微调方法（PEFT）——包括全量微调、LoRA及QLoRA（秩分别为8和16）——在该语料上的性能表现。实验表明，Mistral 7B结合QLoRA（r=16）在困惑度（perplexity）上取得最优结果（均值5.03），且相比高秩微调未带来显著质量提升但显著增加显存消耗，从而为低资源语言场景下的模型选择与计算成本优化提供了实证依据。

链接: https://arxiv.org/abs/2605.03742
作者: Mullosharaf K. Arabov
机构: Kazan Federal University (喀山联邦大学)
类目: Computation and Language (cs.CL)
备注: Preprint

点击查看摘要

Abstract:This paper is devoted to the adaptation of generative large language models for the Tajik language, a low-resource language with Cyrillic script. To overcome the shortage of digital text resources, the author created and publicly released the Tajik Web Corpus, the largest open-access corpus of Tajik, comprising 319,298 documents (~1.11 billion characters). On a subsample of 10,000 documents, 17 configurations were benchmarked, covering autoregressive, encoder-decoder, and encoder-only models with three fine-tuning strategies: full fine-tuning, LoRA, and QLoRA (ranks 8 and 16). Quality was assessed via perplexity and cross-entropy loss; peak GPU memory and training time were also recorded. Best results were achieved by Mistral 7B with QLoRA (r=16): mean perplexity 5.03, standard deviation 0.03. Increasing rank from 8 to 16 gave statistically insignificant improvement while raising memory consumption. For small GPT-2 family models, full fine-tuning yielded lower perplexity (3.48 for GPT-2 Medium) than LoRA (7.60-8.42), but induced catastrophic forgetting. The encoder-only XLM-RoBERTa showed the worst results (perplexity 59.3). The novelty lies in creating the largest verified Tajik corpus and the first systematic analysis of PEFT effectiveness for Tajik text generation. Practical value lies in recommendations for architecture and fine-tuning strategy selection, optimizing computational costs without substantial quality loss.

[NLP-18] Segmenting Human-LLM Co-authored Text via Change Point Detection

【速读】：该论文旨在解决人类与生成式 AI（Generative AI）共同撰写的文本中，如何精准定位具体段落由哪一方撰写的问题。现有检测方法通常仅提供整段文本的二分类结果，无法满足协作文本中细粒度归属识别的需求。其解决方案的关键在于将文本分割任务类比为时间序列分析中的变化点检测（change point detection），并基于此构建适配于生成式 AI 文本检测的加权算法和广义算法，以应对不同检测得分的异质性波动，同时理论证明了所提方法的极小极大最优性。

链接: https://arxiv.org/abs/2605.03723
作者: Mengchu Li,Jin Zhu,Jinglai Li,Chengchun Shi
机构: University of Birmingham (伯明翰大学); London School of Economics and Political Science (伦敦经济学院)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Methodology (stat.ME)
备注:

点击查看摘要

Abstract:The rise of large language models (LLMs) has created an urgent need to distinguish between human-written and LLM-generated text to ensure authenticity and societal trust. Existing detectors typically provide a binary classification for an entire passage; however, this is insufficient for human–LLM co-authored text, where the objective is to localize specific segments authored by humans or LLMs. To bridge this gap, we propose algorithms to segment text into human- and LLM-authored pieces. Our key observation is that such a segmentation task is conceptually similar to classical change point detection in time-series analysis. Leveraging this analogy, we adapt change point detection to LLM-generated text detection, develop a weighted algorithm and a generalized algorithm to accommodate heterogeneous detection score variability, and establish the minimax optimality of our procedure. Empirically, we demonstrate the strong performance of our approach against a wide range of existing baselines.

[NLP-19] Rose-SQL: Role-State Evolution Guided Structured Reasoning for Multi-Turn Text-to-SQL

【速读】：该论文旨在解决大型推理模型（Large Reasoning Models, LRMs）在多轮Text-to-SQL任务中表现不足的问题，特别是现有方法依赖不稳定的API调用或昂贵的小规模微调。其解决方案的关键在于提出一个无需训练的框架Rose-SQL，通过引入细粒度的Role-State表示来弥合模式链接与SQL生成之间的结构鸿沟，并利用历史上下文中的结构同构性检查追踪Role-State演化路径，从而引导模型基于验证过的交互轨迹推断当前问题的SQL组成。

链接: https://arxiv.org/abs/2605.03720
作者: Le Zhou,Feng Yao,Fengcai Qiao,Bo Xu,Fangyuan Wang,Boyan Xu
机构: National University of Defense Technology (国防科技大学); Chinese Academy of Sciences (中国科学院); Guangdong University of Technology (广东工业大学)
类目: Computation and Language (cs.CL)
备注:

点击查看摘要

Abstract:Recent advances in Large Reasoning Models (LRMs) trained with Long Chain-of-Thought have demonstrated remarkable capabilities in code generation and mathematical reasoning. However, their potential in multi-turn Text-to-SQL tasks remains largely underexplored. Existing approaches typically rely on unstable API-based inference or require expensive fine-tuning on small-scale models. In this work, we present Rose-SQL, a training-free framework that leverages small-scale LRMs through in-context learning to enable accurate context-dependent parsing. We introduce the Role-State, a fine-grained representation that bridges the structural gap between schema linking and SQL generation by serving as a structural blueprint. To handle conversational dependencies, Rose-SQL traces the evolution of Role-State through historical context via structural isomorphism checks, guiding the model to infer the possible SQL composition for the current question through verified interaction trajectories. Experiments on the SParC and CoSQL benchmarks show that, within the Qwen3 series, Rose-SQL outperforms in-context learning baselines at the 4B scale and substantially surpasses state-of-the-art fine-tuned models at the 8B and 14B scales, while showing consistent gains on additional reasoning backbones.

[NLP-20] SAM-NER: Semantic Archetype Mediation for Zero-Shot Named Entity Recognition ACL2026

【速读】：该论文旨在解决零样本命名实体识别（Zero-shot Named Entity Recognition, ZS-NER）在领域迁移和标签体系变化下表现不稳定的问题，特别是在目标领域标签定义与大语言模型（Large Language Model, LLM）内在语义结构不一致时，直接映射实体提及容易引发系统性语义漂移，尤其当目标标签体系新颖或存在语义重叠时更为显著。解决方案的关键在于提出一种三阶段框架SAM-NER，其核心创新是通过**语义原型中介（Semantic Archetype Mediation）**构建一个领域无关的中间语义原型空间：首先通过协作式抽取与基于共识的去噪实现高覆盖、高保真的实体跨度发现；其次将实体投影到由高层本体抽象提炼出的通用语义原型集合中，实现语义对齐；最后利用冻结的LLM进行约束化、定义对齐的推理，将原型级预测校准为具体目标域标签。该机制有效缓解了跨域迁移中的语义偏移问题，显著提升了ZS-NER在跨领域场景下的稳定性与准确性。

链接: https://arxiv.org/abs/2605.03706
作者: Ruichu Cai,Juntao Gan,Miao Mai,Zhifeng Hao,Boyan Xu
机构: Guangdong University of Technology (广东工业大学); Peng Cheng Laboratory (鹏城实验室); Nanfang Media Group(Nanfang Daily) (南方都市报); Shantou University (汕头大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted to Findings of ACL 2026

点击查看摘要

Abstract:Zero-shot Named Entity Recognition (ZS-NER) remains brittle under domain and schema shifts, where unseen label definitions often misalign with a large language model’s (LLM’s) intrinsic semantic organization. As a result, directly mapping entity mentions to fine-grained target labels can induce systematic semantic drift, especially when target schemas are novel or semantically overlapping. We propose \textbfSAM-NER, a three-stage framework based on \emphSemantic Archetype Mediation that stabilizes cross-domain transfer through an intermediate, domain-invariant archetype space. SAM-NER: (i) performs \emphEntity Discovery via cooperative extraction and consensus-based denoising to obtain high-coverage, high-fidelity entity spans; (ii) conducts \emphAbstract Mediation by projecting entities into a compact set of universal semantic archetypes distilled from high-level ontological abstractions; and (iii) applies \emphSemantic Calibration to resolve archetype-level predictions into target-domain types through constrained, definition-aligned inference with a frozen LLM. Experiments on the CrossNER benchmark show that SAM-NER consistently outperforms strong prior ZS-NER baselines in cross-domain settings. Our implementation will be open-sourced at this https URL.

[NLP-21] SERE: Structural Example Retrieval for Enhancing LLM s in Event Causality Identification ACL2026

【速读】：该论文旨在解决大语言模型（Large Language Models, LLMs）在事件因果关系识别（Event Causality Identification, ECI）任务中因因果推理偏差而导致的过度预测问题（即因果幻觉，causal hallucination）。为提升LLMs在ECI中的准确性与可靠性，作者提出SERE框架——一种基于结构化示例检索的增强方法。其核心创新在于引入三种结构度量机制：(i) 概念路径度量（Conceptual Path Metric），利用ConceptNet中的编辑距离衡量事件间的概念关联；(ii) 句法度量（Syntactic Metric），通过句法树编辑距离量化事件结构相似性；(iii) 因果模式过滤（Causal Pattern Filtering），借助LLMs识别预定义因果结构以筛选高质量示例。通过融合上述策略，SERE能够精准选取与目标事件对语义和结构更匹配的few-shot示例，从而有效引导LLMs进行更可靠的因果推理，显著降低偏差并提升性能。

链接: https://arxiv.org/abs/2605.03701
作者: Zhifeng Hao,Zhongjie Chen,Junhao Lu,Shengyin Yu,Guimin Hu,Keli Zhang,Ruichu Cai,Boyan Xu
机构: Guangdong University of Technology (广东工业大学); Shantou University (汕头大学); Huawei Noah’s Ark Lab (华为诺亚方舟实验室); Peng Cheng Laboratory (鹏城实验室)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted to Findings of ACL 2026

点击查看摘要

Abstract:Event Causality Identification (ECI) requires models to determine whether a given pair of events in a context exhibits a causal relationship. While Large Language Models (LLMs) have demonstrated strong performance across various NLP tasks, their effectiveness in ECI remains limited due to biases in causal reasoning, often leading to overprediction of causal relationships (causal hallucination). To mitigate these issues and enhance LLM performance in ECI, we propose SERE, a structural example retrieval framework that leverages LLMs’ few-shot learning capabilities. SERE introduces an innovative retrieval mechanism based on three structural concepts: (i) Conceptual Path Metric, which measures the conceptual relationship between events using edit distance in ConceptNet; (ii) Syntactic Metric, which quantifies structural similarity through tree edit distance on syntactic trees; and (iii) Causal Pattern Filtering, which filters examples based on predefined causal structures using LLMs. By integrating these structural retrieval strategies, SERE selects more relevant examples to guide LLMs in causal reasoning, mitigating bias and improving accuracy in ECI tasks. Extensive experiments on multiple ECI datasets validate the effectiveness of SERE. The source code is publicly available at this https URL.

[NLP-22] A Comprehensive Analysis of Tokenization and Self-Supervised Learning in End-to-End Automatic Speech Recognition applied on French Language

【速读】：该论文旨在解决当前端到端自动语音识别（ASR）系统在实际应用中因仅依赖字符错误率（CER）和/或词错误率（WER）等传统指标而导致的性能评估不充分问题，这些问题无法全面反映自动转录文本在下游任务中的实际表现。其解决方案的关键在于：从语言学和声学等多个角度出发，通过引入一套综合性的评估指标，系统性地分析不同子词分词算法（subword tokenization algorithms）和自监督学习模型（self-supervised learning models）对法语ASR系统的影响，从而更准确地指导模型与超参数的选择。

链接: https://arxiv.org/abs/2605.03696
作者: Thibault Bañeras-Roux,Mickael Rouvier,Jane Wottawa,Richard Dufour
机构: 未知
类目: Computation and Language (cs.CL)
备注:

点击查看摘要

Abstract:The performance of end-to-end automatic speech recognition (ASR) systems enables their increasing integration into numerous applications. While there are various benefits to such speech-to-text systems, the choice of hyperparameters and models plays a crucial role in their performance. Typically, these choices are determined by considering only the character (CER) and/or word error rate (WER) metrics. However, it has been shown in several studies that these metrics are largely incomplete and fail to adequately describe the downstream application of automatic transcripts. In this paper, we conduct a qualitative study on the French language that investigates the impact of subword tokenization algorithms and self-supervised learning models from different linguistic and acoustic perspectives, using a comprehensive set of evaluation metrics.

[NLP-23] A Paradigm for Interpreting Metrics and Identifying Critical Errors in Automatic Speech Recognition

【速读】：该论文旨在解决传统自动语音转录评估指标（如词错误率 Word Error Rate, WER 和字符错误率 Character Error Rate, CER）与人类感知相关性差、且无法考虑语言和语义信息的问题。其解决方案的关键在于提出一种新范式，通过将选定的评估指标嵌入到最小编辑距离（Minimum Edit Distance, minED）框架中，从而获得一个等效于错误率的度量指标，该指标不仅能够映射转录错误与其人类感知的一致性，还能从人类视角对错误严重程度进行原创性分析。

链接: https://arxiv.org/abs/2605.03671
作者: Thibault Bañeras-Roux,Mickael Rouvier,Jane Wottawa,Richard Dufour
机构: 未知
类目: Computation and Language (cs.CL)
备注:

点击查看摘要

Abstract:The most commonly used metrics for evaluating automatic speech transcriptions, namely Word Error Rate (WER) and Character Error Rate (CER), have been heavily criticized for their poor correlation to human perception and their inability to take into account linguistic and semantic information. While metric-based embeddings, seeking to approximate human perception, have been proposed, their scores remain difficult to interpret, unlike WER and CER. In this article, we overcome this problem by proposing a paradigm that consists in incorporating a chosen metric into it in order to obtain an equivalent of the error rate: a Minimum Edit Distance (minED). This approach parallels transcription errors with their human perception, also allowing an original study of the severity of these errors from a human perspective.

[NLP-24] Annotation Quality in Aspect-Based Sentiment Analysis: A Case Study Comparing Experts Students Crowdworkers and Large Language Model

【速读】：该论文旨在解决德语方面情感分析（Aspect-Based Sentiment Analysis, ABSA）研究中因高质量标注数据稀缺而导致的进展受限问题。其核心解决方案在于系统评估不同标注来源（包括专家、学生、众包工作者及大语言模型LLMs）对德国ABS A任务性能的影响，通过重新由专家标注现有数据集以建立基准参考，进而利用一致性指标（Inter-Annotator Agreement, IAA）和下游模型表现（涵盖Aspect Category Sentiment Analysis, ACSA 和 Target Aspect Sentiment Detection, TASD）对比各标注源的质量与效率。关键发现揭示了标注可靠性与标注效率之间的权衡关系，为低资源自然语言处理（Natural Language Processing, NLP）场景下的数据集构建提供了实证依据与实践指导。

链接: https://arxiv.org/abs/2605.03624
作者: Niklas Donhauser,Jakob Fehle,Nils Constantin Hellwig,Markus Weinberger,Udo Kruschwitz,Christian Wolff
机构: 未知
类目: Computation and Language (cs.CL)
备注:

点击查看摘要

Abstract:Aspect-Based Sentiment Analysis (ABSA) enables fine-grained opinion analysis by identifying sentiments toward specific aspects or targets within a text. While ABSA has been widely studied for English, research on other languages such as German remains limited, largely due to the lack of high-quality annotated datasets. This paper examines how different annotation sources influence the development of German ABSA. To this end, an existing dataset is re-annotated by experts to establish a ground truth, which serves as a reference for evaluating annotations produced by students, crowdworkers, Large Language Models (LLMs), and experts. Annotation quality is compared using Inter-Annotator Agreement (IAA) and its impact on downstream model performance for different ABSA subtasks. The evaluation focuses on Aspect Category Sentiment Analysis (ACSA) and Target Aspect Sentiment Detection (TASD). We apply State-of-the-Art (SOTA) methods for ABSA, including BERT-, T5-, and LLaMA-based approaches to assess performance differences, spanning fine-tuning and in-context learning with instruction prompts. The findings provide practical insights into trade-offs between annotation reliability and efficiency, offering guidance for dataset construction in under-resourced Natural Language Processing (NLP) scenarios.

[NLP-25] BIT.UA-AAUBS at ArchEHR-QA 2026: Evaluating Open-Source and Proprietary LLM s via Prompting in Low-Resource QA LREC2026 ALT

【速读】：该论文旨在解决低资源环境下临床问答（clinical question answering）与证据定位（evidence grounding）的挑战，尤其是在缺乏训练数据且受严格数据隐私法规（如GDPR）限制的医疗场景中。其解决方案的关键在于采用无需权重更新的大语言模型（Large Language Models, LLMs）策略，并结合多种提示工程（prompt engineering）技术，包括任务分解、思维链（Chain-of-Thought）和上下文学习（in-context learning），同时引入多数投票和LLM-as-a-judge集成方法以提升预测鲁棒性。实验表明，尽管商用模型对提示变化具有较强鲁棒性，但经过领域适配的开源模型（如MedGemma 3 27B）在优化提示设计后可实现极具竞争力的性能，最终在Subtask 4（证据引用对齐）中获得第一名，在Subtask 3（患者友好型答案生成）中获得第三名。

链接: https://arxiv.org/abs/2605.03618
作者: Richard A. A. Jonker,Alexander Christiansen,Alexandros Maniatis,Rúben Garrido,Rogério Braunschweiger de Freitas Lima,Roman Jurowetzki,Sérgio Matos
机构: 未知
类目: Computation and Language (cs.CL)
备注: 14 pages, 7 figures, 4 tables, accepted at CL4Health@LREC 2026

点击查看摘要

Abstract:This paper presents the joint participation of the this http URL and AAUBS groups in the ArchEHR-QA 2026 shared task, which focuses on clinical question answering and evidence grounding in a low-resource setting. Due to the absence of training data and the strict data privacy constraints inherent to the healthcare domain (e.g. GDPR), we investigate the capabilities of Large Language Models (LLMs) without weight updates. We evaluate several state-of-the-art proprietary models and locally deployable open-source alternatives using various prompt engineering strategies, including task decomposition, Chain-of-Thought, and in-context learning. Furthermore, we explore majority voting and LLM-as-a-judge ensembling techniques to maximize predictive robustness. Our results demonstrate that while proprietary models exhibit strong resilience to prompt variations, domain-adapted open-source models (such as MedGemma 3 27B) achieve highly competitive performance when paired with the right prompt. Overall, our prompt-based approach proved highly effective, securing 1st place in Subtask 4 (evidence citation alignment) and 3rd place in Subtask 3 (patient-friendly answer generation). All code, results, and prompts are available on our GitHub repository: this https URL.

[NLP-26] Workspace-Bench 1.0: Benchmarking AI Agents on Workspace Tasks with Large-Scale File Dependencies

【速读】：该论文旨在解决当前AI代理在工作空间学习（Workspace Learning）中对异构文件间显式与隐式依赖关系识别、推理、利用及更新能力不足的问题。现有基准测试多基于预设或合成文件，缺乏真实世界中的复杂依赖结构，导致对AI代理在实际工作场景下完成常规与高级任务的能力评估严重不足。解决方案的关键在于构建一个大规模、高真实性的评估基准——Workspace-Bench，其包含5类用户画像、74种文件类型、20,476个文件（最大达20GB）以及388项任务，每项任务均配有独立的文件依赖图，并通过7,399个评价指标综合衡量跨文件检索、情境推理和自适应决策能力。此外，为降低评估成本，还推出了保留分布特性的轻量版Workspace-Bench-Lite（100任务），显著减少约70%的计算开销。实验表明，当前主流代理与基础模型在该基准上的表现远未达到可靠水平，最佳结果仅为68.7%，明显低于人类水平（80.7%）。

链接: https://arxiv.org/abs/2605.03596
作者: Zirui Tang,Xuanhe Zhou,Yumou Liu,Linchun Li,Weizheng Wang,Hongzhang Huang,Jun Zhou,Jiachen Song,Shaoli Yu,Jinqi Wang,Zihang Zhou,Hongyi Zhou,Yuting Lv,Jinyang Li,Jiashuo Liu,Ruoyu Chen,Chunwei Liu,GuoLiang Li,Jihua Kang,Fan Wu
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Databases (cs.DB); Machine Learning (cs.LG)
备注: 30 pages, 17 figures

点击查看摘要

Abstract:Workspace learning requires AI agents to identify, reason over, exploit, and update explicit and implicit dependencies among heterogeneous files in a worker’s workspace, enabling them to complete both routine and advanced tasks effectively. Despite its importance, existing relevant benchmarks largely evaluate agents on pre-specified or synthesized files with limited real-world dependencies, leaving workspace-level evaluation underexplored. To this end, we introduce Workspace-Bench, a benchmark for evaluating AI agents on Workspace Learning invOlving Large-Scale File Dependencies. We construct realistic workspaces with 5 worker profiles, 74 file types, 20,476 files (up to 20GB) and curate 388 tasks, each with its own file dependency graph, evaluated across 7,399 total rubrics that require cross-file retrieval, contextual reasoning, and adaptive decision-making. We further provide Workspace-Bench-Lite, a 100-task subset that preserves the benchmark distribution while reducing evaluation costs by about 70%. We evaluate 4 popular agent harnesses and 7 foundation models. Experimental results show that current agents remain far from reliable workspace learning, where the best reaches only 68.7%, substantially below the human result of 80.7%, and the average performance across agents is only 47.4%.

[NLP-27] AfriVox-v2: A Domain-Verticalized Benchmark for In-the-Wild African Speech Recognition

【速读】：该论文旨在解决当前生成式 AI（Generative AI）在非洲语言语音识别与翻译能力上的严重不足问题，尤其是在低资源环境下的实际应用瓶颈。现有基准测试对非洲语言覆盖不全、缺乏真实场景噪声和细粒度领域评估，导致模型在特定语境下性能表现不可靠。解决方案的关键在于构建 AfriVox-v2 基准数据集，其核心创新包括：引入“野外”未脚本音频以模拟真实使用场景，并实施严格的领域垂直化设计，在政府、金融、医疗、农业等十个专业领域进行精细化评测，同时针对数字和命名实体进行专项测试。该基准为评估语音模型在非洲复杂环境中的泛化能力提供了可靠标准，推动本地化语音人工智能的开发与落地。

链接: https://arxiv.org/abs/2605.03590
作者: Busayo Awobade,Gabrial Zencha Ashungafac,Tobi Olatunji
机构: Intron Health
类目: Computation and Language (cs.CL); Sound (cs.SD)
备注:

点击查看摘要

Abstract:Recent large language models (LLMs) show strong speech recognition and translation capabilities for high-resource languages. However, African languages remain dramatically underrepresented in benchmarks, limiting their practical use in low-resource settings. While early benchmarks tested African languages and accents, they lacked exhaustive real-world noise and granular domain evaluations. We present AfriVox-v2, a comprehensive benchmark designed to test speech models under realistic African deployment conditions. AfriVox-v2 introduces “in the wild” unscripted audio for all supported languages. We also introduce strict domain verticalization, evaluating model accuracy across ten sectors including government, finance, health, and agriculture and conducting targeted tests on numbers and named entities. Finally, we benchmark a new generation of speech models, including Sahara-v2, Gemini 3 Flash, and the Omnilingual CTC models. Our results expose the true generalization gap of modern speech models in specialized, noisy African contexts and provide a reliable blueprint for developers building localized voice AI.

[NLP-28] PatRe: A Full-Stage Office Action and Rebuttal Generation Benchmark for Patent Examination

【速读】：该论文旨在解决现有专利审查（patent examination）研究中对过程建模不足的问题，即传统基准将审查视为静态分类或提取任务，未能捕捉其本质上交互式、迭代式的特性，类似于学术出版中的同行评审与反驳过程。解决方案的关键在于提出首个全面模拟专利审查全生命周期的基准数据集 PatRe，涵盖审查意见（Office Action）生成与申请人反驳（rebuttal）两个核心环节，包含480个真实案例，并支持 oracle 和检索模拟两种评估设置。该设计将专利审查重构为多轮论证与响应的动态过程，从而更真实地评估大语言模型（LLM）在技术新颖性判断和法律推理方面的表现，揭示了不同模型类型（专有 vs 开源）及任务不对称性下的性能差异。

链接: https://arxiv.org/abs/2605.03571
作者: Qiyao Wang,Xinyi Chen,Longze Chen,Hongbo Wang,Hamid Alinejad-Rokny,Yuan Lin,Min Yang
机构: Shenzhen Institute of Advanced Technology (深圳先进技术研究院); Chinese Academy of Sciences (中国科学院); University of Chinese Academy of Sciences (中国科学院大学); Dalian University of Technology (大连理工大学); UNSW Sydney (新南威尔士大学); Shenzhen University of Advanced Technology (深圳先进研究生院)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 24 pages, 11 figures, 16 tables

点击查看摘要

Abstract:Patent examination is a complex, multi-stage process requiring both technical expertise and legal reasoning, increasingly challenged by rising application volumes. Prior benchmarks predominantly view patent examination as discriminative classification or static extraction, failing to capture its inherently interactive and iterative nature, similar to the peer review and rebuttal process in academic publishing. In this paper, we introduce PatRe, the first benchmark that models the full patent examination lifecycle, including Office Action generation and applicant rebuttal. PatRe comprises 480 real-world cases and supports both oracle and retrieval-simulated evaluation settings. Our benchmark reframes patent examination as a dynamic, multi-turn process of justification and response. Extensive experiments across various LLMs reveal critical insights into model performance, including differences between proprietary and open-source models, as well as task asymmetries between examiner analysis and applicant-side rebuttal. These findings highlight both the potential and current limitations of LLMs in modeling complex, real-world legal reasoning and technical novelty judgment in patent examination. We release our code and dataset to facilitate future research on patent examination modeling.

[NLP-29] Revisiting Graph-Tokenizing Large Language Models : A Systematic Evaluation of Graph Token Understanding

【速读】：该论文旨在解决当前图令牌大语言模型（Graph-Tokenizing Large Language Models, GTokenLLMs）是否真正理解图令牌（graph tokens）的问题，即这些模型在自然语言嵌入空间中对图令牌的语义理解是否充分。针对这一问题，作者提出一个统一的框架和一个名为GTEval的评估流水线，通过格式层面与内容层面的指令变换来系统性地检验模型对图令牌的理解能力。其解决方案的关键在于：设计可量化、可比较的评估机制，揭示现有GTokenLLMs在面对指令变化时表现出的过度敏感或过度不敏感行为，并指出尽管图令牌能保留任务相关图信息并被模型关注，但其实际利用程度存在显著差异，且仅靠指令微调无法根本提升对图令牌的本质理解，从而为未来改进提供了明确方向。

链接: https://arxiv.org/abs/2605.03514
作者: Zhongjian Zhang,Yue Yu,Mengmei Zhang,Junping Du,Xiao Wang,Chuan Shi
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:

点击查看摘要

Abstract:The remarkable success of large language models (LLMs) has motivated researchers to adapt them as universal predictors for various graph tasks. As a widely recognized paradigm, Graph-Tokenizing LLMs (GTokenLLMs) compress complex graph data into graph tokens and treat them as prefix tokens for querying LLMs, leading many to believe that LLMs can understand graphs more effectively and efficiently. In this paper, we challenge this belief: \textitDo GTokenLLMs fully understand graph tokens in the natural-language embedding space? Motivated by this question, we formalize a unified framework for GTokenLLMs and propose an evaluation pipeline, \textbfGTEval, to assess graph-token understanding via instruction transformations at the format and content levels. We conduct extensive experiments on 6 representative GTokenLLMs with GTEval. The primary findings are as follows: (1) Existing GTokenLLMs do not fully understand graph tokens. They exhibit over-sensitivity or over-insensitivity to instruction changes, and rely heavily on text for reasoning; (2) Although graph tokens preserve task-relevant graph information and receive attention across LLM layers, their utilization varies across models and instruction variants; (3) Additional instruction tuning can improve performance on the original and seen instructions, but it does not fully address the challenge of graph-token understanding, calling for further improvement.

[NLP-30] Rational Communication Shapes Morphological Composition

【速读】：该论文试图解决的问题是：为什么一种语言会选择特定的词素组合（如复合词或派生词）而非其他在同时期可用的合理替代方案？传统研究虽指出交际效率塑造词汇系统，但缺乏对形态构词作为说话者在竞争性词素序列中做出历史情境化选择的建模。解决方案的关键在于引入理性言语行为（Rational Speech Act, RSA）框架，并基于《美国英语历史语料库》（COHA）与《当代美国英语语料库》（COCA）构建时间索引词汇表，量化表达信息量（semantic informativeness）与生成成本（production cost）之间的权衡。结果显示，已实际使用的词素组合在候选集中显著优于未被使用的替代方案，且结合语义信息与生产成本的“语用说话者模型”（S₁）在平均倒数排名（MRR）和top-k准确率上优于仅考虑语义或仅考虑成本的基线模型，尤其在候选集扩大时优势更加明显——这表明词汇化过程本质上反映了表达力与效率之间的沟通权衡，将理性沟通理论从话语层面延伸至词内部结构。

链接: https://arxiv.org/abs/2605.03510
作者: Fengyuan Yang,Yongqian Peng,Yuxi Ma,Chenheng Xu,Yixin Zhu
机构: Peking University (北京大学)
类目: Computation and Language (cs.CL)
备注:

点击查看摘要

Abstract:Human languages expand vocabularies by combining existing morphemes rather than inventing arbitrary forms. Communicative efficiency shapes lexical systems at multiple levels (Gibson et al., 2019), yet morphological composition – combining morphemes through compounding or affixation – has rarely been modeled as a historically situated speaker choice among competing morpheme sequences, leaving unanswered why a language settles on one morpheme combination over other plausible alternatives. We ask whether a trade-off between listener recoverability and speaker production cost can predict attested compositions over contemporaneously available alternatives. Here we show, within the Rational Speech Act (RSA) framework (Frank Goodman, 2012; Goodman Frank, 2016) using a time-indexed lexicon constructed from Corpus of Historical American English (COHA) and Corpus of Contemporary American English (COCA), that across 4323 naturally occurring English compounds and derivations spanning 1820–2019, attested compositions are systematically ranked above unattested alternatives generated from contemporaneously available morphemes. Models integrating semantic informativeness with production cost outperform semantic-only and cost-only baselines on Mean Reciprocal Rank (MRR) and top-k accuracy (Acc@k), with the advantage of the Pragmatic Speaker model ( S_1 ) over the semantic-only baseline growing as the candidate set expands, where meaning alone leaves morphological choice underdetermined. These findings suggest that lexicalization reflects a communicative trade-off between expressiveness and efficiency, extending rational accounts of communication from utterance-level choice to the internal structure of words.

[NLP-31] CuraView: A Multi-Agent Framework for Medical Hallucination Detection with GraphRAG -Enhanced Knowledge Verification

【速读】：该论文旨在解决生成式 AI（Generative AI）在临床文档生成中因“忠实性幻觉”（faithfulness hallucinations）而导致的患者安全风险问题，即模型生成内容与原始电子健康记录（EHR）存在矛盾。其核心解决方案是提出 CuraView——一个基于多智能体的句子级检测与证据锚定解释框架，通过构建基于 GraphRAG 的知识图谱实现患者级 EHR 的结构化表示，并设计闭环生成-检测流程，支持从强支持到直接矛盾的四级证据分级（E1–E4），从而生成可解释的证据链。该方法显著提升了对高危幻觉（E4 类）的识别能力（F1=0.831），相较基线模型提升 50%，同时产出可用于下游模型训练和蒸馏的标注数据集，有效增强临床文本生成的准确性与可信度。

链接: https://arxiv.org/abs/2605.03476
作者: Severin Ye,Xiao Kong,Xiaopeng He,Guangsu Yan,Dongsuk Oh
机构: Kyungpook National University (庆北国立大学); Sichuan University (四川大学); Shandong University (山东大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 44 pages, 8 figures, 13 tables

点击查看摘要

Abstract:Discharge summaries require extracting critical information from lengthy electronic health records (EHRs), a process that is labor-intensive when performed manually. Large language models (LLMs) can improve generation efficiency; however, they are prone to producing faithfulness hallucinations, statements that contradict source records, posing direct risks to patient safety. To address this, we present CuraView, a multi-agent framework for sentence-level detection and evidence-grounded explanation of faithfulness hallucinations in discharge summaries. CuraView constructs a GraphRAG-based knowledge graph from patient-level EHRs and implements a closed-loop generation-detection pipeline with sentence-level evidence retrieval and classification spanning four evidence grades from strong support to direct contradiction (E1-E4), yielding structured and interpretable evidence chains. We evaluate CuraView on a subset of 250 patients from the Discharge-Me benchmark, with 50 patients held out for testing. Our fine-tuned Qwen3-14B detection model achieves an F1 of 0.831 on the safety-critical E4 metric (90.9% recall, 76.5% precision) and an F1 of 0.823 on E3+E4, representing a 50.0% relative improvement over the base model and outperforming RAGTruth-style and QAGS-style baselines. These results demonstrate that evidence-chain-based graph retrieval verification substantially improves the factual reliability of clinical documentation, while simultaneously producing reusable annotated datasets for downstream model training and distillation. Comments: 44 pages, 8 figures, 13 tables Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) ACMclasses: I.2.7; J.3 Cite as: arXiv:2605.03476 [cs.CL] (or arXiv:2605.03476v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2605.03476 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[NLP-32] Detecting Stealth Sycophancy in Mental-Health Dialogue with Dynamic Emotional Signature Graphs

【速读】：该论文旨在解决 conversational AI therapists 在心理支持场景中缺乏可靠离线评估方法的问题，特别是如何在不依赖大语言模型（Large Language Models, LLMs）作为最终评判者的情况下，准确评估对话响应的治疗质量。其核心挑战在于：传统直接使用LLM或对称文本相似度指标难以捕捉临床意义上的响应效果——即响应是否推动用户情绪状态向调节（regulation）或重构（reframing）方向发展、维持不变，或加剧风险情绪与认知扭曲。解决方案的关键是提出一种模型无关的评估框架 Dynamic Emotional Signature Graphs (DESG)，它通过解耦临床状态表示并利用非对称临床几何结构对对话窗口进行评分，从而实现更贴近临床判断的性能表现；实验表明，DESG-Ensemble 在3000条对话样本上的测试集上达到0.9353宏F1分数，显著优于现有方法，且特征消融和盲审验证确认临床状态流形是主要判别基础，而图结构轨迹则提供可解释的非对称评分机制。

链接: https://arxiv.org/abs/2605.03472
作者: Tianze Han,Beining Xu,Hanbo Zhang,Yongming Lu
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:As conversational AI therapists are increasingly used in psychological support settings, reliable offline evaluation of therapeutic response quality remains an open problem. This paper studies multi-domain support-dialogue evaluation without relying on large language models as final judges. We use a direct LLM judge as a baseline that reads raw dialogue text and predicts whether the target response is harmful, productive, or neutral. We find that direct LLM judges and symmetric text-similarity metrics are poorly aligned with therapeutic quality because the target label depends on clinical direction: whether the response moves the user state toward regulation or reframing, leaves it broadly unchanged, or reinforces deterioration through higher risk affect or cognitive-distortion mass. To address this issue, we propose Dynamic Emotional Signature Graphs (DESG), a model-agnostic evaluator that represents dialogue windows with decoupled clinical states and scores them using asymmetric clinical geometry. We evaluate DESG on a constructed diagnostic stress-test benchmark of 3,000 dialogue windows from EmpatheticDialogues, ESConv, and CRADLE-Dialogue, covering peer support, counseling dialogue, and crisis-oriented interaction. On the 600-window held-out test aggregate, DESG-Ensemble achieves 0.9353 macro-F1, exceeding ConcatANN by 1.51 percentage points, BERTScore by 19.63 points, and TRACT by 33.81 points. Feature ablations, artifact controls, a 100-window blinded adjudicator audit, and qualitative disagreement cases indicate that the clinical state manifold is the main discriminative substrate, while graph-based trajectory components provide asymmetric scoring and interpretable diagnostics rather than serving as the sole source of performance.

[NLP-33] Retrieving Floods without Floodlights: Topic Models as Binary Classifiers for Extreme Climate Events in German News LREC2026

【速读】：该论文旨在解决在极端气候事件媒体报道研究中，由于缺乏足够标注数据而难以训练高精度深度学习分类器的问题。其解决方案的关键在于利用主题模型（Topic Models）作为二分类器，通过估计的主题后验分布来筛选与特定类型极端气候事件相关的新闻文档，且不改变主题模型的训练过程。该方法借助标注样本指导评估，发现关键词查询时分配的概率可有效用于选择相关主题，从而提升样本精确度，同时验证了不同气候灾害类型之间存在差异，提示不应将气候事件简单视为单一类别进行自然语言处理任务。

链接: https://arxiv.org/abs/2605.03450
作者: Brielen Madureira,Mariana Madruga de Brito,Andreas Niekler
机构: 未知
类目: Computation and Language (cs.CL)
备注: Presented at the The 2nd Workshop on Ecology, Environment, and Natural Language Processing at LREC 2026

点击查看摘要

Abstract:In studies of media coverage of extreme climate events, NLP methods have become indispensable for identifying relevant texts in large news databases. Still, enough annotated data to train accurate deep learning-based classifiers from scratch is often not available. Topic Models have the advantage of being both unsupervised and interpretable, but are typically used only for exploratory analysis or data characterisation. In this study, we investigate how to employ Topic Models as binary classifiers for refining the retrieval of relevant news about seven types of extreme climate events in the German media. Our method relies on the posterior distributions estimated by Topic Models to select relevant documents, without modifying their training procedure. Using an annotated sample to guide the evaluation, we show that the probabilities assigned to keywords used to query news databases can also be informative for selecting relevant topics and improve sample precision. We compare our results to a fine-tuned text embedding classifier and an open-weight LLM, discussing observed trade-offs, e.g. the LLM’s lowest precision. Moreover, we show that results are hazard-dependent, which speaks against considering climate events as a single category in NLP tasks.

[NLP-34] An ERP Study of Recursive Possessive Parsing in ASD Children and Its Cognitive Neuro Mechanisms

【速读】：该论文旨在解决自闭症谱系障碍（ASD）儿童在处理复杂递归结构时的在线句法加工机制问题，特别是针对汉语中两级递归所有格结构的认知神经基础。其解决方案的关键在于采用事件相关电位（ERP）技术，结合句子-图片匹配范式，系统比较了ASD儿童与典型发育（TD）同龄人在P200、N400和P600成分上的脑电反应差异。结果表明，ASD儿童在早期感知加工（P200）、句法重新分析（P600）阶段均表现出显著异常，而语义加工（N400）相对保留，这揭示了ASD语言障碍中句法模块的特异性缺陷，支持语言模块化分离理论。

链接: https://arxiv.org/abs/2605.03447
作者: Fu Chenxi,Wang Xiaoyi,Zhuang Ziman,Yang Caimei
机构: 未知
类目: Computation and Language (cs.CL)
备注: 20pages, 7 figures

点击查看摘要

Abstract:Recursive structures are a core property of human language, yet little is known about how children with autism spectrum disorder (ASD) process complex recursion. This ERP study investigated the online processing of two-level recursive possessive structures in Mandarin-speaking children with ASD (n = 12) compared to typically developing (TD) peers (n = 12) using a sentence-picture matching paradigm. ERPs were analyzed for P200 (150-250 ms), N400 (300-500 ms), and P600 (500-1000 ms). Results showed that ASD children exhibited significantly reduced P200 amplitudes and failed to show the typical posterior grammaticality effect, indicating atypical early perceptual processing. No robust N400 violation effect was observed in either group, confirming the mismatch was not a semantic anomaly; however, ASD children showed a reversed anterior effect and an attenuated posterior effect. For the P600, ASD children had significantly reduced amplitudes, no posterior grammaticality effect, and a trend toward delayed latency, reflecting a core deficit in syntactic reanalysis. These findings demonstrate that while lexical-semantic processing is relatively preserved in ASD, the online syntactic computation required for recursion is severely impaired, supporting modular dissociation accounts of language in autism.

[NLP-35] Sentiment Analysis of Indonesian Spotify Reviews Using Machine Learning and BiLSTM

【速读】：该论文旨在解决印尼语Spotify评论的三分类情感分析问题，即区分正面、负面和中性情感倾向。其解决方案的关键在于对比经典机器学习方法（支持向量机、多项式朴素贝叶斯、决策树）与深度学习方法（双向长短期记忆网络，BiLSTM）在相同预处理流程下的性能差异，其中预处理包括俚语规范化、停用词去除和词干提取。研究发现，虽然BiLSTM在整体加权F1分数上表现最优，但在少数类（中性类）上表现不佳；而采用SMOTE过采样技术的经典机器学习方法能实现更平衡的三分类性能，表明针对类别不平衡问题，数据增强策略对提升模型公平性至关重要。

链接: https://arxiv.org/abs/2605.03443
作者: Uliano Wilyam Purba,Andre Hadiman Rotua Parhusip,Sahid Maulana,Luluk Muthoharoh,Ardika Satria,Martin C. T. Manullang
机构: Sumatera Institute of Technology (苏门答腊技术学院)
类目: Computation and Language (cs.CL)
备注: 8 pages, multiple figures and tables, benchmark study on Indonesian-language Spotify review sentiment classification, compares SVM, Multinomial Naive Bayes, Decision Tree, and BiLSTM, includes class-imbalance discussion and deployment links

点击查看摘要

Abstract:This paper benchmarks classical machine learning and deep learning approaches for three-class sentiment classification of Indonesian Spotify reviews. Using 100,000 scraped reviews and 70,155 cleaned samples, the study compares Support Vector Machine, Multinomial Naive Bayes, and Decision Tree models with a two-layer BiLSTM. Both approaches use the same preprocessing pipeline, including slang normalization, stopword removal, and stemming. Decision Tree achieves the best performance among the classical models, while BiLSTM attains the highest weighted F1-score overall but fails on the minority neutral class. The paper concludes that BiLSTM is stronger for overall sentiment detection, whereas machine learning with SMOTE provides more balanced three-class performance.

[NLP-36] Exposing LLM Safety Gaps Through Mathematical Encoding:New Attacks and Systematic Analysis

【速读】：该论文旨在解决大型语言模型（Large Language Models, LLMs）安全机制易被绕过的问题，特别是针对基于语义模式匹配的防御策略在面对结构化编码的有害提示时失效的现象。解决方案的关键在于将有害内容重新表述为具有真实数学结构的 coherent 数学问题（如集合论、形式逻辑和量子力学中的表达），而非仅使用数学符号进行表面伪装；实验表明，只有通过辅助大模型深度重构有害意图以形成真正的数学问题，攻击成功率才能显著提升（达46%–56%），而仅用规则模板添加数学格式则无效。这一发现揭示了当前安全框架对深层语义与结构理解的不足，并指出未来防御需聚焦于识别和解析数学结构本身，而非依赖表层文本特征。

链接: https://arxiv.org/abs/2605.03441
作者: Haoyu Zhang,Mohammad Zandsalimy,Shanu Sushmita
机构: 未知
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 16 pages, 2 figures, 4 tables. Accepted as a long paper at the 39th Canadian Conference on Artificial Intelligence (Canadian AI 2026)

点击查看摘要

Abstract:Large language models (LLMs) employ safety mechanisms to prevent harmful outputs, yet these defenses primarily rely on semantic pattern matching. We show that encoding harmful prompts as coherent mathematical problems – using formalisms such as set theory, formal logic, and quantum mechanics – bypasses these filters at high rates, achieving 46%–56% average attack success across eight target models and two established benchmarks. Crucially, the effectiveness depends not on mathematical notation itself, but on whether a helper LLM deeply reformulates the harmful content into a genuine mathematical problem: rule-based encodings that apply mathematical formatting without such reformulation perform no better than unencoded baselines. We introduce a novel Formal Logic encoding that achieves attack success comparable to Set Theory, demonstrating that this vulnerability generalizes across mathematical formalisms. Additional experiments with repeat post-processing confirm that these attacks are robust to simple prompt augmentation. Notably, newer models (GPT-5, GPT-5-Mini) show substantially greater robustness than older models, though they remain vulnerable. Our findings highlight fundamental gaps in current safety frameworks and motivate defenses that reason about mathematical structure rather than surface-level semantics.

[NLP-37] A Comparison of Traditional Machine Learning Algorithms and LSTM-Based Deep Learning Models for Email Sentiment Analysis

【速读】：该论文旨在解决电子通信快速增长背景下电子邮件分类与情感检测系统鲁棒性不足的问题，核心挑战在于如何在保证高精度的同时提升处理效率。解决方案的关键在于对比传统机器学习算法（如支持向量机、逻辑回归、朴素贝叶斯）与深度学习架构（如长短期记忆网络，LSTM）在基于Word2Vec词嵌入特征表示下的性能表现，发现线性核支持向量机（SVM）在准确率（98.74%）和计算效率之间实现了最优平衡，而LSTM虽具备优异的召回能力但显著增加计算开销。研究结果表明，对于邮件检测任务，SVM是兼顾预测精度与处理速度的最佳选择。

链接: https://arxiv.org/abs/2605.03440
作者: Virdio Samuel Saragih,Baruna Abirawa,Kartini Lovian Simbolon,Luluk Muthoharoh,Ardika Satria,Martin C.T. Manullang
机构: Sumatra Institute of Technology (苏门答腊理工学院)
类目: Computation and Language (cs.CL)
备注: 9 pages, 3 figures, 6 tables

点击查看摘要

Abstract:The rapid growth of electronic communication has necessitated more robust systems for email classification and sentiment detection. This study presents a comparative performance analysis between traditional machine learning algorithms and deep learning architectures, specifically focusing on Support Vector Machines (SVMs), Logistic Regression, Naive Bayes, and Long Short-Term Memory (LSTM). Utilizing Word2Vec embeddings for feature representation, our experimental results indicate that the SVM model with a linear kernel achieves the highest efficiency and accuracy, reaching a peak performance of 98.74%. While the LSTM model demonstrates exceptional recall capabilities in detecting spam-related sentiments, it requires significantly more computational time compared to discriminative statistical models. Detailed evaluations via confusion matrices further reveal that traditional classifiers remain highly robust for dense vector spaces. This research concludes that for email detection tasks, SVM offers the most optimal balance between predictive precision and processing speed. These findings provide critical insights for developing high-performance automated email filtering systems in professional and academic environments.

[NLP-38] Benchmarking Logistic Regression SVM Naive Bayes and IndoBERT Fine-Tuning for Sentiment Analysis on Indonesian Product Reviews

【速读】：该论文旨在解决印度尼西亚电商平台上用户生成产品评论的情感分析问题，以实现对客户满意度的量化评估和产品问题的规模化识别。其解决方案的关键在于对比传统机器学习（ML）方法与基于Transformer的深度学习模型（IndoBERT）在三类情感分类任务中的性能表现，并通过引入平衡类别权重和自定义加权交叉熵损失函数来缓解电商评论数据中普遍存在的严重类别不平衡问题。实验结果表明，在特定的数据采样条件下，线性支持向量机（Linear SVC）模型在准确率（97.60%）和宏F1分数（0.5510）上显著优于IndoBERT模型（分别为88.70%和0.5088），这一差异主要归因于基线模型使用完整语料库而Transformer模型受限于采样子集的训练策略。

链接: https://arxiv.org/abs/2605.03439
作者: Nabila Zakiyah Zahra,Salwa Farhanatussaidah,Nasywa Nur Afifah,Luluk Muthoharoh,Ardika Satria,Martin C.T. Manullang
机构: Institut Teknologi Sumatera (ITERA)
类目: Computation and Language (cs.CL)
备注: 8 pages, 5 figures. Research article on benchmarking classical machine learning and IndoBERT for three-class sentiment analysis on Indonesian Tokopedia product reviews

点击查看摘要

Abstract:The exponential growth of e-commerce platforms in Indonesia has generated a massive volume of user-generated product reviews. Analyzing the sentiment of these reviews is critical for measuring customer satisfaction and identifying product issues at scale. This paper benchmarks traditional Machine Learning (ML) approaches against a Transformer-based Deep Learning model for a three-class sentiment analysis task (positive, neutral, negative) on the Tokopedia Product Reviews 2025 dataset. We implemented Term Frequency-Inverse Document Frequency (TF-IDF) feature extraction coupled with three algorithms: Logistic Regression, Linear Support Vector Machine (SVM), and Multinomial Naive Bayes as robust baselines. Subsequently, we fine-tuned the IndoBERT model (indobenchmark/indobert-base-p1) for contextual sequence classification. To computationally address the severe class imbalance inherent in e-commerce feedback, we applied balanced class weights for the baseline models and engineered a custom weighted cross-entropy loss function within the IndoBERT training loop, following the broader motivation of imbalanced-learning research. Our comprehensive evaluation using Accuracy, Macro F1-score, and Weighted F1-score revealed that the traditional Linear SVC model significantly outperformed the IndoBERT model in our experimental setup, achieving an Accuracy of 97.60% and a Macro F1-score of 0.5510, compared to IndoBERT’s 88.70% and 0.5088. Detailed analysis indicates that this performance gap was primarily driven by discrepancies in the data sampling regimes, where baselines utilized the full corpus while the Transformer was constrained to a sampled subset. Finally, we demonstrate the practical viability of our pipeline by deploying the final sentiment classification model as an interactive Gradio web application.

[NLP-39] Geolocating News about Extreme Climate Events: A Comparative Analysis of Off-the-Shelf Tools for Toponym Identification in German ECIR2026

【速读】：该论文旨在解决文本中极端气候事件与灾害的地理定位问题，这是气候影响与适应研究中的常见挑战。其核心解决方案在于对比分析三种现成的命名实体识别（Named-entity Recognition, NER）工具——Flair、Spacy 和 Stanza——在德语新闻文章中的输出差异，并通过三种外部评估方法确定事件发生的国家。研究揭示了不同 NER 工具在地名识别上的差异如何传递至下游任务，进而影响文档地理焦点判断，最终可能改变对各国在德语媒体中突出程度的结论。

链接: https://arxiv.org/abs/2605.03414
作者: Brielen Madureira,Mariana Madruga de Brito,Andreas Niekler
机构: LeipzigLab - Climate Discourse, Leipzig University, Leipzig, Germany; Helmholtz Centre for Environmental Research - UFZ, Leipzig, Germany; Computational Humanities, Leipzig University, Leipzig, Germany
类目: Computation and Language (cs.CL)
备注: Presented at the Fourth International Workshop on Geographic Information Extraction from Texts (GeoExT) at ECIR 2026

点击查看摘要

Abstract:Determining the geolocation of extreme climate events and disasters in texts is a common problem in climate impact and adaptation research. Named-entity recognition (NER) tools are typically used to identify a pool of toponyms that serve as candidate event locations. In this study, we conduct a comparative analysis of three off-the-shelf NER tools, namely Flair, Spacy and Stanza. We describe and quantify differences between their outputs for German news articles and evaluate them extrinsically based on three methods to determine the country where events took place. We show how their contrasts are propagated into downstream tasks and can yield distinct decisions about a document’s geographical focus, which, in turn, can impact conclusions about countries’ prominence in German media.

[NLP-40] From prompting to evidence-based translation: A RAG prompt system for Japanese-Chinese translation and its pedagogical potential

【速读】：该论文旨在解决大语言模型（Large Language Models, LLMs）在处理日语-汉语（Ja-Zh）句子中名词修饰从句结构（Noun-Modifying Clause Constructions, NMCCs）时性能下降的问题。现有模型在高资源语言对上表现良好，但在涉及复杂句法结构如NMCCs时可靠性不足。解决方案的关键在于提出一种检索增强生成（Retrieval-Augmented Generation, RAG）+提示工程（Prompt）的翻译系统：通过语言学分析模块输出NMCC类型（A1）与风险预测（A2），结合嵌入空间相似性检索（基于L2距离）获取top-k个相关Ja-Zh示例，并将这些信息整合进增强提示中，从而引导LLM更准确地生成译文。实验表明，随着知识库规模扩大（0至2000条），宏平均句级BLEU得分显著提升（+23.4%），证明该方法能以可解释且可审计的方式改善NMCC相关翻译质量。

链接: https://arxiv.org/abs/2605.03387
作者: Wenshi Gu
机构: 未知
类目: Computation and Language (cs.CL)
备注:

点击查看摘要

Abstract:Large language models perform well on high-resource pairs but are less reliable for Japanese-Chinese sentences containing noun-modifying clause constructions (NMCCs). This study evaluates a retrieval-augmented generation RAG+Prompt translation system that integrates linguistic analysis, embedding-based retrieval, prompt construction, and LLM generation without modifying the base model. The analysis module outputs A1 (inner vs. outer NMCC) and A2 (risk predictions: lexical choice/NMCC handling/word order/style/register); top-k = 5 similar Ja-Zh examples (L2 distance) and A1/A2 are inserted into an enhanced prompt. Using GPT-4o and a 66-sentence test set, we compare six knowledge-base sizes (0/100/200/500/1,000/2,000). Macro-averaged sentence-level BLEU (1-4-gram with brevity penalty; cased; Chinese at the character level) is the sole metric. Mean BLEU increases from 24.28 at 0 (RAG disabled) to 29.96 at 2,000 (+5.68; +23.4%). The upward trend holds across sizes, with larger knowledge bases yielding higher scores. We conclude that the RAG+Prompt translation system improves Ja-Zh translation of sentences containing NMCCs in an interpretable and auditable manner. Limitations include one base model, one metric, and reliance on published texts and commercial APIs; future work will broaden genres, language pairs, and evaluation metrics.

[NLP-41] wo Calls Two Moments and the Vote-Accuracy Curve of Repeated LLM Inference

【速读】：该论文旨在解决重复采样（repeated sampling）在大语言模型（Large Language Models, LLMs）测试时计算资源利用中的有效性问题，即如何在不依赖单次预测准确率的前提下，通过多轮推理来提升整体正确性并量化其边界。核心挑战在于，重复采样的收益由样本间的潜在正确性分布决定，而非单一预测的准确性。解决方案的关键在于：基于条件独立同分布（conditional-i.i.d.）假设下对二值正确性层的建模，仅需两个标记的推理调用即可识别出平均成功概率与二阶矩，从而确定同一示例上的正确性相关性，区分稳定错误与可恢复的随机噪声；进一步地，利用无穷维矩问题的三原子极值解和二次对偶证书，为任意固定多数投票预算提供精确、无分布假设的区间边界，其中三个投票的闭式解宽度不超过1/8，并具备可验证的改进判据。

链接: https://arxiv.org/abs/2605.03379
作者: Yi Liu
机构: 未知
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:

点击查看摘要

Abstract:Repeated sampling is a standard way to spend test-time compute, but its benefit is controlled by the latent distribution of correctness across examples, not by one-call accuracy alone. We study the binary correctness layer of repeated LLM inference under conditional-i.i.d. calls. One labeled call identifies the mean latent success probability; two labeled calls identify its second moment and hence the same-example correctness correlation that separates stable errors from recoverable call-level randomness. From these two moments, every fixed majority-vote budget has a sharp distribution-free two-call interval. The key technical reduction is that the infinite-dimensional moment problem has three-atom extremizers and quadratic dual certificates for every finite budget, so the bounds are exact rather than discretized or parametric. The first useful budget, three votes, has a closed form, width at most 1/8 , and a certified-improvement criterion. The infinite-vote endpoint is the limit of majority voting as the number of calls tends to infinity; it is also sharply bounded, but remains threshold-sensitive because it depends on latent mass around q=1/2 . We add maximum-entropy and Latent-difficulty Gaussian-probit (LDGP) point completions, and experiments on LLM calls over QNLI and QQP show that empirical three- and five-vote accuracies are contained in the projected two-call regions while temperature changes and randomized model mixtures can create voting gains not ordered by one-call accuracy.

[NLP-42] When to Think When to Speak: Learning Disclosure Policies for LLM Reasoning ICML’2026

【速读】：该论文旨在解决单流自回归接口中因token更新模型状态与构成不可逆公开承诺之间的耦合所导致的“沉默税”问题：过度延迟推理会推迟任务相关输出，而过早流式输出则可能导致提前承诺，从而偏倚后续生成。其解决方案的关键在于提出并行交错推理（Side-by-Side Interleaved Reasoning, SxS），该方法在相同上下文中交替进行部分披露与私有推理，仅当内容得到当前推理支持时才释放输出，从而将披露时机变为可控决策。通过构建蕴含对齐的交错轨迹（匹配答案前缀与支撑推理前缀），并结合监督微调（SFT）学习双动作语义、强化学习（RL）恢复新格式下的推理性能，SxS实现了在内容准确性和延迟之间更优的帕累托权衡。

链接: https://arxiv.org/abs/2605.03314
作者: Jiaqi Wei,Xuehang Guo,Pengfei Yu,Xiang Zhang,Wanli Ouyang,Siqi Sun,Qingyun Wang,Chenyu You
机构: 未知
类目: Computation and Language (cs.CL)
备注: Accepted by ICML’2026

点击查看摘要

Abstract:In single-stream autoregressive interfaces, the same tokens both update the model state and constitute an irreversible public commitment. This coupling creates a \emphsilence tax: additional deliberation postpones the first \emphtask-relevant content, while naive early streaming risks premature commitments that bias subsequent generations. We introduce \textbf\emphSide-by-Side (SxS) Interleaved Reasoning, which makes \emphdisclosure timing a controllable decision within standard autoregressive generation. SxS interleaves partial disclosures with continued private reasoning in the same context, but releases content only when it is \emphsupported by the reasoning so far. To learn such pacing without incentivizing filler, we construct entailment-aligned interleaved trajectories by matching answer prefixes to supporting reasoning prefixes, then train with SFT to acquire the dual-action semantics and RL to recover reasoning performance under the new format. Across two Qwen3 architectures/scales (MoE \textbfQwen3-30B-A3B, dense \textbfQwen3-4B) and both in-domain (AIME25) and out-of-domain (GPQA-Diamond) benchmarks, SxS improves accuracy–\emphcontent-latency Pareto trade-offs under token-level proxies (e.g., inter-update waiting).

[NLP-43] SHIELD: A Diverse Clinical Note Dataset and Distilled Small Language Models for Enterprise-Scale De-identification

【速读】：该论文旨在解决临床文本去标识化（De-identification）任务中现有公开基准数据集（如i2b2 2006/2014）因年代久远而缺乏语义与人口多样性的问题，以及大型语言模型（LLMs）在企业级部署时面临的计算成本高和受保护健康信息（PHI）治理限制的挑战。其解决方案的关键在于构建一个名为SHIELD（Synthetic Human-annotated Identifier-replaced Entries for Learning and De-identification）的多样化标注数据集，包含1,394条临床笔记及10,505个金标准PHI跨度，采用集合覆盖多样性采样结合人工校验机制确保质量；并通过知识蒸馏将高性能LLMs的能力迁移至可在本地部署的小型语言模型（SLMs），实现在标准工作站硬件上达到微平均跨度级精确率0.88、召回率0.86的性能，同时验证了跨数据集泛化能力与机构特异性实体需专用模型优化的部署策略。

链接: https://arxiv.org/abs/2605.03301
作者: Jose D. Posada,David Love,Somalee Datta,Priya Desai
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:De-identification of clinical text remains essential for secondary use of electronic health records (EHRs), yet public benchmarks such as i2b2 2006/2014 are over a decade old and lack the semantic and demographic diversity of modern narratives. While Large Language Models (LLMs) achieve state-of-the-art zero-shot extraction, enterprise deployment is hindered by compute costs and governance restricting Protected Health Information (PHI) from cloud APIs. We introduce SHIELD (Synthetic Human-annotated Identifier-replaced Entries for Learning and De-identification), a diverse dataset of 1,394 notes with 10,505 gold-standard PHI spans across 9 categories, built via set-cover diversity sampling with human-in-the-loop adjudication. We evaluate four LLMs (two proprietary, two open-weight) to establish a performance ceiling, then distill these capabilities into locally deployable Small Language Models (SLMs). Distributional analysis using Frechet Text Distance and Jensen-Shannon Divergence confirms SHIELD occupies a distinct region of biomedical embedding and vocabulary space versus legacy benchmarks. Our best distilled model matches its teacher on structured PHI categories (DATE, DOCTOR, ID, PATIENT, PHONE) and achieves micro-averaged span-level precision of 0.88 and recall of 0.86 on standard workstation hardware. Cross-dataset evaluation shows diversity-trained models generalize well on universal structured PHI, while institution-specific entities remain hard to transfer, suggesting optimal deployment combines broad-coverage models with specialized models for high-volume notes. We publicly release the SHIELD dataset and the distilled DeBERTa v3 model.

[NLP-44] LLM -XTM: Enhancing Cross-Lingual Topic Models with Large Language Models ACL2026

【速读】：该论文旨在解决跨语言主题建模（cross-lingual topic modeling）中因依赖稀疏双语资源而导致主题不连贯或对齐弱的问题，同时克服现有基于大语言模型（LLM）的方法在文档级处理、成本高及易产生幻觉等方面的局限。其解决方案的关键在于提出 LLM-XTM 框架，该框架通过引入 LLM 引导的主题精炼机制与自一致性不确定性量化方法，实现了黑盒、稳定且可扩展的跨语言主题模型增强，从而在减少对双语词典和昂贵 LLM 调用依赖的同时，显著提升主题 coherence 和跨语言对齐质量。

链接: https://arxiv.org/abs/2605.03299
作者: Minh Chu Xuan,Tien-Phat Nguyen,Linh Ngo Van,Dinh Viet Sang,Nguyen Thi Ngoc Diep,Trung Le
机构: Hanoi University of Science and Technology (河内科技大学); VNU University of Engineering and Technology (越南国家大学工程与技术学院); Monash University (莫纳什大学)
类目: Computation and Language (cs.CL)
备注: ACL 2026

点击查看摘要

Abstract:Cross-lingual topic modeling aims to discover shared semantic structures across languages, yet existing models depend on sparse bilingual resources and often yield incoherent or weakly aligned topics. Recent LLM-based refinements improve interpretability but are costly, document-level, and prone to hallucination, with prior white-box approaches requiring inaccessible token probabilities. We propose LLM-XTM, a framework that integrates LLM-guided topic refinement with self-consistency uncertainty quantification, enabling black-box, stable, and scalable enhancement of cross-lingual topic models. Experiments on multilingual corpora show that LLM-XTM achieves superior topic coherence and alignment while reducing reliance on bilingual dictionaries and expensive LLM calls.

[NLP-45] he Right Answer the Wrong Direction: Why Transformers Fail at Counting and How to Fix It

【速读】：该论文旨在解决大语言模型（Large Language Models, LLMs）在执行简单计数任务时的系统性失败问题，即使计数对象明确出现在输入提示中也常出现错误。研究发现，这种失败并非源于模型内部缺乏对数量信息的有效表征，而是由于输出头（output head）的几何方向与表示计数的内部特征方向几乎正交（|cos| ≤ 0.032），导致无法正确读出对应的数字token。解决方案的关键在于识别并修正这一“几何读出瓶颈”：通过仅更新输出头中对应数字token的权重（36,864参数）可显著提升约束性下一个token预测准确率（从60.7%提升至100%），但不足以改善自回归生成；而采用LoRA微调注意力Q/V权重（7.67M参数）则能优化上游信息路由，使正确数字的词汇排名从55,980降至1，实现83.1% ± 7.2%的真实贪婪自回归生成性能提升，验证了该机制的普遍性与有效性。

链接: https://arxiv.org/abs/2605.03258
作者: Gabriel Garcia
机构: Google(谷歌)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 26 pages, 3 figures; Code and reproduction materials: this https URL

点击查看摘要

Abstract:Large language models often fail at simple counting tasks, even when the items to count are explicitly present in the prompt. We investigate whether this failure occurs because transformers do not represent counts internally, or because they cannot convert those representations into the correct output tokens. Across three model families, Pythia, Qwen3, and Mistral, ranging from 0.4B to 14B parameters, we find strong evidence for the second explanation. Linear probes recover the correct count from intermediate layers with near-perfect accuracy ( R^20.99 ), showing that the information is present. However, the internal directions that encode counts are nearly orthogonal to the output-head rows for digit tokens ( |\cos|\leq0.032 ). In other words, the model stores the count in a form that the digit logits do not naturally read out. We localize this failure with two interventions. Updating only the digit rows of the output head (36,864 parameters) substantially improves constrained next-token digit prediction (60.7 to 100.0% across four tasks), but it does not fix autoregressive generation. By contrast, a small LoRA intervention on attention Q/V weights (7.67M parameters) improves upstream routing and achieves 83.1% +/- 7.2% in true greedy autoregressive generation. Logit-lens measurements confirm the mechanism: the correct digit’s vocabulary rank drops from 55,980 to 1, a 50,000x improvement. Additional norm, logit-lens, and cross-task analyses show that the bottleneck generalizes across character counting, addition, and list length, while remaining absent from broader multi-step reasoning benchmarks, including MMLU, GSM8K, and DROP. These results identify counting failure as a geometric readout bottleneck rather than a failure of internal representation: the model knows the count but the output pathway is geometrically misaligned with the tokens needed to express it. Comments: 26 pages, 3 figures; Code and reproduction materials: this https URL Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL) ACMclasses: I.2.6; I.2.7 Cite as: arXiv:2605.03258 [cs.LG] (or arXiv:2605.03258v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2605.03258 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Gabriel Garcia [view email] [v1] Tue, 5 May 2026 01:13:06 UTC (290 KB)

[NLP-46] S2tory: Story Spine Distillation for Movie Script Summarization

【速读】：该论文旨在解决电影剧本自动摘要任务中因非线性、交叉剪辑叙事结构导致的表面显著性方法难以保留核心故事推进的问题。其解决方案的关键在于提出S²tory（Story Spine Distillation）框架，该框架基于叙事学理论，通过角色发展轨迹识别驱动情节前进的“情节核”（plot nuclei），过滤掉仅增强氛围或情感的边缘事件；其中，Narrative Expert Agent（NEAgent）执行受理论约束的推理，并将提炼的知识用于指导小型模型识别情节核，进而生成高质量摘要，实现了约3.5倍压缩下的语义保真度提升及跨领域的零样本泛化能力。

链接: https://arxiv.org/abs/2605.03244
作者: Mingzhe Lu,Yanbing Liu,Qihao Wang,Jiarui Zhang,Jiayue Wu,Yue Hu,Yunpeng Li,Yangyan Xu
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Movie scripts pose a fundamental challenge for automatic summarization due to their non-linear, cross-cut narrative structure, which makes surface-level saliency methods ineffective at preserving core story progression. To address this, we introduce S^2tory (Story Spine Distillation), a narratology-grounded framework that leverages character development trajectories to identify plot nuclei, the essential events that drive the narrative forward, while filtering out peripheral satellite events that merely enrich atmosphere or emotion. Our Narrative Expert Agent (NEAgent) performs theory-constrained reasoning, whose distilled knowledge conditions a small model to identify plot nuclei. Another model then uses these plot nuclei to generate the summary. Experiments on the MovieSum dataset demonstrate state-of-the-art semantic fidelity at approximately 3.5x compression, and zero-shot evaluation on BookSum confirms strong out-of-domain generalization. Human evaluation further validates that narratological theory provides an indispensable foundation for modeling complex, non-linear narratives.

[NLP-47] amUp: Semantic Project Matching and Team Formation for Learning at Scale

【速读】：该论文旨在解决大规模项目式学习（Project-based Learning, PBL）中学生与项目匹配不当及团队认知多样性不足的问题，这导致高绩效学生倾向于选择显性项目，而代表性不足的学生则面临机会获取不平等。解决方案的关键在于提出TeamUp系统，其核心是利用预训练语言模型生成的语义嵌入（semantic embeddings）进行个性化推荐，并通过融合余弦相似度与教学约束（如难度匹配、领域偏好和需求平衡）的混合排序算法实现精准匹配；同时，通过建模嵌入方差来量化技能互补性，从而构建认知多样性高的团队，确保能力分布均衡而非同质化优势。

链接: https://arxiv.org/abs/2605.03237
作者: Dhruv Gulwani,Basem Suleiman,Aditya Joshi,Sonit Singh
机构: University of New South Wales (UNSW)
类目: oftware Engineering (cs.SE); Computation and Language (cs.CL)
备注: 8 pages

点击查看摘要

Abstract:Project-based learning improves student engagement and learning outcomes, yet allocating students to appropriately challenging projects while forming cognitively diverse teams remains difficult at scale. Traditional allocation methods (manual spreadsheets, preference surveys) can’t construct the cognitively diverse teams that that collaborate cognitively. This mismatch perpetuates equity issues: high-performing students self-select visible projects while under-represented students face reduced access to opportunity. We propose TeamUp, a lightweight, embedding-based team-forming system designed to improve learning outcomes and equity in large-scale project-based courses. TeamUp uses semantic embeddings from pretrained language models to match students to projects aligned with their skill level. The system employs a hybrid ranking algorithm combining cosine similarity with pedagogical constraints (difficulty alignment, domain preferences, and demand balancing) to generate personalised and transparent recommendations. Beyond individual matching, TeamUp constructs cognitively diverse teams by modelling skill complementarity through embedding variance, ensuring teams possess well-distributed capabilities rather than homogeneous strengths. We evaluated TeamUp through a virtual experiment using 250 student profiles and 60 project descriptions. Results show: (1) substantially higher match quality (mean cosine similarity of 0.74 vs. 0.43); (2) better difficulty alignment (83% placed within one level vs. 34%); (3) more diverse teams (82% covering three or more technical areas vs. 41%); and (4) sub-second recommendation latency at operational costs under 0.10 per student.

[NLP-48] Sparse Memory Finetuning as a Low-Forgetting Alternative to LoRA and Full Finetuning

【速读】：该论文旨在解决预训练语言模型在微调过程中出现的灾难性遗忘（catastrophic forgetting）问题，即模型在适应新任务时会显著损害其原有的通用能力。解决方案的关键在于引入稀疏记忆微调（Sparse Memory Finetuning, SMF），通过向模型添加键值记忆层，并在每个训练步骤中仅更新当前批次读取最频繁的记忆行（memory rows），从而实现对特定任务的有效适应，同时最小化对原始通用能力的影响。

链接: https://arxiv.org/abs/2605.03229
作者: Prakhar Gupta,Garv Shah,Satyam Goyal,Anirudh Kanchi
机构: 未知
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:

点击查看摘要

Abstract:Adapting a pretrained language model to a new task often hurts the general capabilities it already had, a problem known as catastrophic forgetting. Sparse Memory Finetuning (SMF) tries to avoid this by adding key-value memory layers to the model and, on each training step, updating only the small set of memory rows that the current batch reads most heavily. We re-implement SMF on Qwen-2.5-0.5B-Instruct and compare it with LoRA and full finetuning on MedMCQA, a 4-choice medical exam task, using WikiText perplexity and TriviaQA accuracy as forgetting probes. SMF improves MedMCQA by 2.5 percentage points while keeping both forgetting probes within roughly 1 point of the base model, whereas LoRA and full finetuning achieve larger gains but with clear drift on both. We also compare two row-selection rules (KL-divergence and TF-IDF), which balance the two forgetting metrics differently.

[NLP-49] MAGE: Safeguarding LLM Agents against Long-Horizon Threats via Shadow Memory

【速读】：该论文旨在解决大语言模型（Large Language Model, LLM）驱动的智能体在执行长期任务时面临的一类新型攻击问题，即利用用户-代理-环境之间的长时间交互来实现单轮场景下难以达成的恶意目标，这类长周期威胁对LLM智能体在关键领域的安全部署构成重大风险。解决方案的关键在于提出MAGE（Memory As Guardrail Enforcement）框架，其核心思想是借鉴系统安全中的“影子栈”（shadow stack）机制，构建一个专门用于安全防护的智能体记忆模块（agentic memory），持续提取并保留整个执行轨迹中与安全性相关的关键上下文信息，并基于此“影子记忆”在动作执行前主动评估潜在风险，从而实现对长周期攻击的早期检测与防御。

链接: https://arxiv.org/abs/2605.03228
作者: Yuhui Wang,Tanqiu Jiang,Jiacheng Liang,Charles Fleming,Ting Wang
机构: Stony Brook University (石溪大学); Cisco Systems (思科系统)
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:

点击查看摘要

Abstract:As large language model (LLM)-powered agents are increasingly deployed to perform complex, real-world tasks, they face a growing class of attacks that exploit extended user-agent-environment interactions to pursue malicious objectives improbable in single-turn settings. Such long-horizon threats pose significant risks to the safe deployment of LLM agents in critical domains. In this paper, we present MAGE (Memory As Guardrail Enforcement), a novel defensive framework designed to counter a wide range of long-horizon threats. Inspired by the “shadow stack” abstraction in systems security, MAGE maintains a dedicated, safety-focused agentic memory that distills and retains safety-critical context across the agent’s full execution trajectory, leveraging this shadow memory to proactively assess the risk of pending actions prior to their execution. Extensive evaluation demonstrates that MAGE substantially outperforms existing defenses across diverse long-horizon threats in detection accuracy, achieves early-stage detection for the majority of attacks, and introduces only negligible overhead to agent utility. To our best knowledge, MAGE represents the first framework to detect and mitigate long-horizon threats using an agentic memory approach, establishing a new paradigm for this critical challenge and opening promising directions for future research.

[NLP-50] Geometric Deviation as an Unsupervised Pre-Generation Reliability Signal: Probing LLM Representations for Answerability ACL

【速读】：该论文旨在解决语言模型在生成回答前无法有效识别查询是否超出其知识范围的问题，即如何在不依赖模型输出或标注失败数据的情况下，提供一个可靠的预生成信号以判断问题的可回答性。解决方案的关键在于利用模型隐藏状态（hidden states）的表示几何结构（representation geometry），通过测量其与已知可回答输入的参考集中心点之间的偏差来实现这一目标。研究发现，在数学类提示中，不可回答输入的隐藏状态显著偏离可回答集合的中心，形成强分离效果（ROC-AUC 0.78–0.84），且该信号在早期层中已建立并逐渐衰减，表明答案可得性相关的几何特征在生成过程早期即被编码，从而为结构化任务域提供了一种轻量级、高效的预生成检测机制。

链接: https://arxiv.org/abs/2605.03196
作者: Yucheng Du
机构: University of Southern California (南加州大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Accepted to TrustNLP 2026 (ACL Workshop). 11 pages, 3 figures, 3 tables

点击查看摘要

Abstract:A reliable language model should be able to signal, prior to generation, when a query falls outside its knowledge. We investigate whether representation geometry can provide such a pre-generation signal by measuring the deviation of hidden states from an answerable reference set, requiring no labeled failure data and no access to model outputs. Across three instruction-tuned models (Llama 3.1-8B, Qwen 2.5-7B, and Mistral-7B-Instruct) and three prompt forms (Math, Fact, Code), we find that geometry primarily encodes task form. Within mathematical prompts, unanswerable inputs consistently deviate from the answerable centroid, yielding strong separation (ROC-AUC 0.78-0.84). This single-pass pre-generation signal outperforms a simple refusal baseline and compares favorably to self-consistency. It also captures cases where models do not explicitly refuse. In contrast, no reliable geometric signal emerges for factual prompts, indicating that the effect is form-conditional rather than universal. Code prompts show large effect sizes with higher variance, suggesting partial generalization beyond mathematical form. A layer-wise analysis reveals that the signal arises in early layers and gradually attenuates toward the output. These results suggest that answerability-related geometry is established before the final stages of generation. Together, these findings indicate that geometric deviation can serve as a lightweight pre-generation signal that is reliable in structured domains with formal answerability constraints, with clear boundaries on where it generalizes. Comments: Accepted to TrustNLP 2026 (ACL Workshop). 11 pages, 3 figures, 3 tables Subjects: Computation and Language (cs.CL); Machine Learning (cs.LG) ACMclasses: I.2.7 Cite as: arXiv:2605.03196 [cs.CL] (or arXiv:2605.03196v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2605.03196 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Yucheng Du [view email] [v1] Mon, 4 May 2026 22:24:34 UTC (746 KB)

[NLP-51] OCRR: A Benchmark for Online Correction Recovery under Distribution Shift

【速读】：该论文旨在解决现有静态基准无法衡量模型在面对分布偏移（distribution shift）时，通过用户修正流（correction streams）在线恢复性能的问题。传统方法仅评估训练时冻结模型的固定性能，而现实系统需应对新类别、查询改写和数据漂移，并依赖实时修正进行自适应调整。为填补这一空白，作者提出OCRR（Online Correction Recovery Rate）基准，量化模型在少量修正下对新类准确率与原分布准确率的恢复能力。其解决方案的关键在于设计了一种基于哈希链的追加式存储结构（hash-chained append-only substrate），该结构结合近似最近邻检索与边缘带多数投票机制，在有限内存预算下实现了高精度的新类识别（88.7% ± 2.9%）与原分布保持能力（95.4% ± 0.8%），显著优于主流持续学习算法（如EWC、LwF）及参数高效微调方法（LoRA on DeBERTa-v3-large），且在大规模语料库中仍保持稳定分类性能（99%），证明其对检索不完美具有鲁棒性。

链接: https://arxiv.org/abs/2605.03153
作者: Adrian Grassi
机构: 未知
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 13 pages, 5 figures, 4 tables. Code and data: this https URL

点击查看摘要

Abstract:Static benchmarks measure a model frozen at training time. Real systems face distribution shift: new categories, paraphrased queries, drift: and must recover online via user corrections. No existing benchmark measures recovery speed under correction streams. We introduce OCRR (Online Correction Recovery Rate): a benchmark that streams a corpus through a classification system, applies oracle or stochastic corrections to wrong predictions, and reports two curves: novel-class accuracy and original-distribution accuracy versus correction count. We evaluate the substrate alongside nine baseline algorithms from five families plus seven bounded-storage variants of the substrate for the Pareto sweep, including standard online-learning baselines (river), continual-learning methods (EWC, A-GEM, LwF), retrieval/parametric hybrids (kNN-LM), parameter-efficient fine-tuning of a 1.5 B-parameter encoder (LoRA on DeBERTa-v3-large), and a hash-chained append-only substrate (Substrate). On Banking77 and CLINC150, under oracle and sparse correction policies, the substrate is the only system that simultaneously recovers novel-class accuracy (88.7 +/- 2.9 %) and retains original-distribution accuracy (95.4 +/- 0.8 %) beating the next-best published continual-learning baseline by 32.6 percentage points at equal memory budget, and beating LoRA-on-DeBERTa-v3-large by 84.6 percentage points on retention. We further find that classification accuracy remains stable at 99 % even as approximate-nearest-neighbour recall@5 degrades from 0.69 to 0.23 across 10 k to 10 M corpus scales, suggesting the substrate’s margin-band majority vote is robust to retrieval imperfection in a way that pure top-k recall metrics do not predict. Code and data are available at this https URL.

[NLP-52] Effective Performance Measurement: Challenges and Opportunities in KPI Extraction from Earnings Calls ACL

【速读】：该论文旨在解决从非结构化、无标签的财报电话会议（Earnings Calls）中提取有效财务信息的难题，此类信息虽重要但难以自动化处理。传统方法依赖于受监管的结构化披露文件（如SEC filings），而财报电话会议具有对话式语言特征且缺乏标注，导致模型泛化能力差。解决方案的关键在于提出一种基于大语言模型（LLM）的开放抽取系统，通过上下文学习（in-context learning）实现对未结构化通话转录文本的端到端关键绩效指标（KPI）识别，并借助人工评估验证其性能（79.7%精度），从而为该领域建立首个可追踪新兴KPI的基准。

链接: https://arxiv.org/abs/2605.03147
作者: Rasmus T. Aavang,Rasmus Tjalk-Bøggild,Alexandre Iolov,Giovanni Rizzi,Mike Zhang,Johannes Bjerva
机构: Aalborg University (奥尔堡大学); ALIPES ApS (ALIPES ApS); University of Copenhagen (哥本哈根大学)
类目: Computation and Language (cs.CL)
备注: Accepted to ACL Industry Track

点击查看摘要

Abstract:Earnings calls are a key source of financial information about public companies. However, extracting information from these calls is difficult. Unlike the templatic filings required by the U.S. Securities and Exchange Commission (SEC) to report a company’s financial situation, earnings conference calls have no built-in labels, are unstructured, and feature conversational language. We explore this challenging domain by assessing the information captured by models trained on SEC filings and in-context learning methods. To establish a baseline, we first evaluate the generalization capabilities of SEC-trained models across established SEC datasets. To support our investigation, we introduce three novel benchmarks: (1) SEC Filings Benchmark (SECB), (2) Earnings Calls Benchmark (ECB), and ECB-A, a subset with 2,460 expert annotation groups to support our qualitative analysis. We find that encoder-based models struggle with the domain shift. Finally, we propose a system utilizing LLMs to perform open-ended extraction from unstructured call transcripts, verified by human evaluation (79.7% precision), providing a baseline for this valuable domain through the consistent tracking of emergent KPIs.

[NLP-53] PIIGuard: Mitigating PII Harvesting under Adversarial Sanitization

【速读】：该论文旨在解决生成式 AI（Generative AI）在浏览网页时可能泄露接触类个人身份信息（PII）的问题，特别是针对网页所有者缺乏有效防御手段的现状。解决方案的关键在于提出 PIIGuard，一种网页级别的防御机制，通过在网页中嵌入优化后的隐藏 HTML 片段，利用间接提示注入（indirect prompt injection）原理引导模型避免直接或可重构地披露联系类 PII，同时保持正常页面问答功能不受影响。该方法结合规则评分、进化变异和最终判别器评估，在多个目标模型上实现了至少 97.0% 的防御成功率，证明了网页侧片段作为实用缓解措施的有效性。

链接: https://arxiv.org/abs/2605.03129
作者: Mingshuo Liu,Yiwei Zha,Min Chen
机构: 未知
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:

点击查看摘要

Abstract:Browsing-enabled LLM assistants can fetch webpages and answer contact-seeking queries, creating a practical channel for scraping contact-style personally identifiable information (PII) from public pages. Many prior defenses are deployed at the model, service, or agent layer rather than at the webpage itself, leaving ordinary page owners with limited deployable options. We present PIIGuard, a webpage-level defense that repurposes indirect prompt injection as a protective mechanism: the page owner embeds optimized hidden HTML fragments that steer the model away from verbatim or reconstructible disclosure of contact PII. PIIGuard searches over fragment text and insertion position using rule-based leakage scoring, evolutionary mutation, and final judge-based recoverability assessment. In direct-HTML evaluation on three target models (GPT-5.4-nano, Claude-haiku-4.5, and DeepSeek-chat(latest v3.2)), PIIGuard achieves at least 97.0% defense success rate under both rule-based and judge-based leakage evaluation, often reaching 100.0%, while preserving benign same-page QA utility. We further evaluate two harder settings: public-URL browsing and attacker-side LLM sanitization of fetched webpage. These results show that page-side defensive fragments can remain effective in deployment for some model-position pairs, but robustness varies substantially across browsing interfaces and sanitizer prompts. Overall, PIIGuard demonstrates that page owners can use page-side fragments as a practical mitigation for web-grounded PII leakage.

[NLP-54] Benchmarking Local Language Models for Social Robots using Edge Devices

【速读】：该论文旨在解决教育类机器人（如Robot Study Companion, RSC）在资源受限的边缘设备上部署语言模型时面临的系统性评估缺失问题，特别是在响应速度、能效与教学有效性之间存在显著权衡的情况下。其关键解决方案是通过在Raspberry Pi 4等边缘硬件平台上对25个开源语言模型进行多维基准测试（包括推理效率、通用知识能力及教学效果），识别出性能最优的模型组合，并据此提出一种三层次本地推理架构，以在计算资源有限的条件下实现响应性与准确性的协同优化。其中，Granite4 Tiny Hybrid (7B) 模型表现出最佳综合平衡，具备2.5 tokens/秒的吞吐量、0.90 tokens/焦耳的能效比以及54.6%的MMLU准确率，且教学评分不依赖于高MMLU分数，验证了实际教学价值与模型规模之间的非线性关系。

链接: https://arxiv.org/abs/2605.03111
作者: Dorian Lamouille,Matevž B. Zorec,Farnaz Baksh,Karl Kruusamäe
机构: 未知
类目: Robotics (cs.RO); Computation and Language (cs.CL)
备注: Accepted for 22nd IEEE International Conference on Advanced Robotics and its Social Impact (June 2026) in Vienna, Austria

点击查看摘要

Abstract:Social-educational robots designed for socially interactive pedagogical support, such as the Robot Study Companion (RSC), rely on responsive, privacy-preserving interaction despite severely limited compute. However, there is a gap in systematic benchmarking of language models for edge computing in pedagogical applications. This paper benchmarks 25 open-source language models for local deployment on edge hardware. We evaluate each model across three dimensions: inference efficiency (tokens per second, energy consumption), general knowledge (a six-category MMLU subset), and teaching effectiveness (LLM-rated pedagogical quality), validated against five independent human raters using the Raspberry Pi(RPi)4 as the primary platform, with additional comparisons on the RPi5 and a laptop GPU. Results reveal pronounced trade-offs: throughput and energy efficiency vary by over an order of magnitude across models, MMLU accuracy ranges from near-random to 57.2%, and teaching effectiveness does not correlate monotonically with either metric. Among the evaluated models, Granite4 Tiny Hybrid (7B) achieves a strong overall balance, reaching 2.5 tokens per second, 0.90 tokens per joule, and 54.6% MMLU accuracy; high MMLU accuracy does not appear necessary for strong teaching scores. Human validation on four representative models preserved the automated rank ordering (Pearson r = 0.967, n = 4). Based on these findings, we propose a three-tier local inference architecture for the RSC that balances responsiveness and accuracy on resource-constrained hardware. Comments: Accepted for 22nd IEEE International Conference on Advanced Robotics and its Social Impact (June 2026) in Vienna, Austria Subjects: Robotics (cs.RO); Computation and Language (cs.CL) Cite as: arXiv:2605.03111 [cs.RO] (or arXiv:2605.03111v1 [cs.RO] for this version) https://doi.org/10.48550/arXiv.2605.03111 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[NLP-55] MedStruct-S: A Benchmark for Key Discovery Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports KSEM2026

【速读】：该论文旨在解决从光学字符识别（OCR）生成的临床报告中进行半结构化信息抽取（Semi-structured Information Extraction, IE）时面临的两大现实挑战：一是字段头（key）表示的异质性和未知性，二是OCR引入的噪声问题。现有评估方法往往未能充分建模这两个因素，导致模型在真实场景下的鲁棒性难以准确评估。解决方案的关键在于构建了一个名为MedStruct-S的新基准数据集，该数据集包含3,582张标注的真实世界临床报告页面，专门用于在未知key和OCR噪声条件下评估三种核心任务：字段头发现、基于key的问答（QA）以及端到端的键值对抽取。通过该基准，作者系统比较了编码器-only序列标注与解码器-only结构化生成两类主流范式，揭示了在相同规模下编码器-only模型在非空值key条件QA任务上表现更优，而细调后的解码器-only模型在整体性能上最强，从而为不同半结构化IE场景下的模型选择提供了可靠依据。

链接: https://arxiv.org/abs/2605.03103
作者: Yingyun Li,Yu Wang,Haiyang Qian
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 11 pages, 5 figures. Accepted by KSEM 2026. This is the author’s preprint version; the final authenticated version will be available in the Springer LNCS/LNAI proceedings

点击查看摘要

Abstract:Semi-structured information extraction (IE) from OCR-derived clinical reports is crucial for efficiently reconstructing patients’ longitudinal medical histories. In practice, this scenario commonly involves three tasks: (i) field-header (key) discovery, (ii) key-conditioned question answering (QA), and (iii) end-to-end key-value pair extraction. However, existing evaluations often under-model two factors: heterogeneous and incompletely known key representations, and OCR-induced noise. This makes it difficult to assess model robustness in real-world settings. We present MedStruct-S, a benchmark specifically designed to evaluate these tasks under unknown keys and OCR noise. MedStruct-S contains 3,582 annotated real-world clinical report pages. Using MedStruct-S, we benchmark two representative paradigms: encoder-only sequence labeling with post-processing and decoder-only structured generation, covering four encoder-only and five decoder-only models spanning 0.11B to 103B parameters. Our results show that encoder-only models achieve the best performance for non-null-value key-conditioned QA despite being substantially smaller than decoder-only models. When comparing models of similar order of magnitude, encoder-only models still perform better overall. Without controlling for model scale, fine-tuned decoder-only models deliver the strongest overall results. These findings show that the benchmark provides a reliable and practical basis for selecting and comparing models across different semi-structured IE settings. Comments: 11 pages, 5 figures. Accepted by KSEM 2026. This is the author’s preprint version; the final authenticated version will be available in the Springer LNCS/LNAI proceedings Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) ACMclasses: I.2.7; H.3.1; J.3 Cite as: arXiv:2605.03103 [cs.CL] (or arXiv:2605.03103v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2605.03103 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[NLP-56] When Prompts Interact: Assessing Prompt Arithmetic for Deconfounding under Distribution Shift

【速读】：该论文旨在解决模型在分类任务中因依赖混淆变量（confounding variables）而产生的“捷径行为”（shortcut behavior），导致在分布外（out-of-distribution）场景下性能显著下降的问题。其解决方案的关键在于提出一种参数高效的提示算术方法——混合提示算术（Hybrid Prompt Arithmetic, HyPA），通过将任务提示（task prompts）与线性化混淆提示（linearized confounder prompts）结合，以削弱模型对伪相关特征的依赖。实验表明，HyPA在多个基准测试中均能有效提升鲁棒性-性能权衡，且分析显示其可通过降低混淆信号对预测的影响或抑制其在表征中的表达来缓解混淆问题。

链接: https://arxiv.org/abs/2605.03096
作者: Zhecheng Sheng,Yongsen Tan,Xiruo Ding,Trevor Cohen,Serguei Pakhomov
机构: University of Minnesota(明尼苏达大学); University of Washington(华盛顿大学); Stanford University(斯坦福大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 19 pages, 11 figures

点击查看摘要

Abstract:In classification tasks, models may rely on confounding variables to achieve strong in-distribution performance, capturing spurious features that fail under distribution shift. This shortcut behavior leads to substantial degradation in out-of-distribution settings. Task arithmetic offers a potential solution by removing unwanted signals via subtraction of secondary model updates, but it typically requires full fine-tuning, which is computationally expensive. Prompt tuning provides a parameter-efficient alternative by adapting models through a small set of trainable virtual tokens. Task arithmetic on the resulting prompts presents an appealing alternative to operations on entire models, but the extent to which this approach can limit reliance on spurious features remains to be established. In this work, we study whether composing soft prompts through task arithmetic improves robustness to confounding shifts. We propose Hybrid Prompt Arithmetic (HyPA), which combines task prompts with linearized confounder prompts to counteract spurious correlations. Across multiple benchmarks, HyPA consistently improves the robustness-performance trade-off relative to prompt-arithmetic baselines under distribution shift. We further analyze how HyPA affects hidden representations and find evidence consistent with it mitigating confounding either by reducing the influence of confounder signals on predictions or by suppressing them in the representation. These results establish HyPA as a parameter-efficient and promising approach for improving robustness under confounding shifts in the evaluated setting.

[NLP-57] Semantically Enriching Investor Micro-blogs for Opinion-Aware Emotion Analysis: A Practical Approach

【速读】：该论文旨在解决金融自然语言处理（NLP）中情感分析的局限性问题，即现有方法难以捕捉情感背后的具体“原因”（why），尤其是情绪或情感所指向的目标对象缺乏细粒度解析。为解决这一问题，研究者提出通过构建语义结构化的意见图（opinion graphs）来增强原始的StockEmotions数据集，从而提供比传统情感标签更精细的语义层次信息。其解决方案的关键在于利用大语言模型（LLM）的声明式流水线，从10,000条来自StockTwits的评论中自动提取每句话的意见图，并结合图神经网络（GNNs）对基线分类器进行评估，实验证明引入意见语义可显著提升不同情感维度下的分类性能。

链接: https://arxiv.org/abs/2605.03092
作者: Gaurav Negi,Paul Buitelaar
机构: 未知
类目: Computation and Language (cs.CL)
备注:

点击查看摘要

Abstract:While sentiment analysis is the staple of financial NLP, capturing the nuances of ‘why’ behind that sentiment remains a challenge. There have been attempts to address this by analysing investor emotions alongside sentiment; however, this does not provide the additional granularity required to understand the target of the emotion/sentiment. We address this by augmenting the StockEmotions dataset with semantically structured opinion graphs, which provide granular semantic depth to the existing sentiment and emotion labels. Using a declarative LLM pipeline, we augment the StockEmotions dataset with opinion graphs for each sentence, derived from 10,000 comments collected from StockTwits. In addition, we study the effect of introducing opinion semantics on baseline classifiers using Graph Neural Networks (GNNs). Our analysis demonstrates that incorporating opinion semantics improves classification performance across different emotional spectrums

[NLP-58] he TTS-STT Flywheel: Synthetic Entity-Dense Audio Closes the Indic ASR Gap Where Commercial and Open-Source Systems Fail

【速读】：该论文旨在解决低资源印度语（Indic）领域语音识别（ASR）中实体密集型文本（如数字串、货币金额、地址、品牌名及印地语/英语混用词）识别性能不足的问题，当前开源与商用系统在该任务上的实体命中率（Entity-Hit-Rate, EHR）普遍偏低。其解决方案的关键在于构建一个自包含的文本到语音（TTS）-语音到文本（STT）飞轮机制：利用开源Indic TTS流水线以约50%边际成本合成约22,000条高密度实体的印地语-英语混用语音数据，并在此基础上对vasista22/whisper-telugu-large-v2模型进行LoRA微调，在保留FLEURS-Te测试集上读写文本的WER稳定性前提下，将EHR从0.027提升至0.473（相较开源最优模型提升17倍，商业模型提升3倍）。实验证明，该方法的性能增益主要归因于新增的EDSA语料库（占全部提升的~100%），且在跨语言迁移中也显著优于基线（如泰米尔语beta-Ta EHR达0.543，较基线提升22倍）。

链接: https://arxiv.org/abs/2605.03073
作者: Venkata Pushpak Teja Menta
机构: 未知
类目: Computation and Language (cs.CL); Sound (cs.SD)
备注: 8 pages, 2 figures. Companion to arXiv:2604.25441 (Praxy Voice TTS), arXiv:2604.25476 (PSP), arXiv:2605.00777 (LASE)

点击查看摘要

Abstract:Niche-domain Indic ASR – digit strings, currency amounts, addresses, brand names, English/Indic codemix – is under-served by both open-source SOTA and commercial systems. On a synthesised entity-dense Telugu test set (held-out by synthesis system), vasista22/whisper-telugu-large-v2 (open SOTA) achieves Entity-Hit-Rate (EHR) 0.027 and Deepgram Nova-3 (commercial) 0.16. We close this gap with a self-contained TTS-STT flywheel: an open-source Indic TTS pipeline synthesises ~22,000 entity-dense Indic-English code-mix utterances at 50 marginal cost, and a LoRA fine-tune on top of vasista22 achieves EHR 0.473 on the held-out test (17x over open SOTA, 3x over commercial), with read-prose regression bounded to +6.6 pp WER on FLEURS-Te. Cross-language: beta-Hi 0.337 (7x vs vasista22) and beta-Ta 0.543 (22x vs vasista22, 22x vs Deepgram); on Hindi where Deepgram has substantial entity coverage, the flywheel underperforms commercial. All three beta models fall below pre-registered EHR targets (0.75 for Te, 0.65 for Hi/Ta); we report honestly. A native-human-recorded sanity check (n=20 Telugu) confirms transfer to real speech (beta-Te EHR 0.516 on native vs 0.473 on synth). An EDSA-isolation ablation (LoRA on FLEURS-Te alone) yields EHR 0.020 on the same held-out, attributing ~100% of the gain to the EDSA corpus. We additionally report a language-conditional finding: vanilla Whisper-large-v3 has Telugu-specific Script Collapse (SFR 0.46-0.71) that a per-language LoRA corrects (SFR 0.81-0.97), but the recipe is contraindicated on Hindi and Tamil where vanilla SFR = 0.98. Code, holdouts, predictions, EDSA corpus, and entity dictionaries are released open-source.

[NLP-59] How Language Models Process Negation ICML2026

【速读】：该论文旨在解决大型语言模型（Large Language Models, LLMs）在处理否定句时准确性低的问题，尽管这些模型内部存在能够正确理解否定的组件。其关键解决方案在于识别并干预导致错误决策的晚期层注意力机制——通过消融（ablation）这些特定注意力模块，显著提升了模型在涉及否定的问答任务中的准确率。进一步地，研究揭示了模型采用两种否定处理机制：一是注意力头关注被否定短语并抑制相关概念，二是直接构建整个否定短语的向量表示（如将“not gas”表示为促进液体和固体的向量）。实证分析表明，后者即“构造性”机制更为突出，且两类机制共存于模型内部，从而深化了对LLMs内部计算过程的理解，强调了以构造为主导的推理路径。

链接: https://arxiv.org/abs/2605.03052
作者: Zhejian Zhou,Tianyi Zhou,Robin Jia,Jonathan May
机构: 未知
类目: Computation and Language (cs.CL)
备注: ICML2026 preprint, camera ready in progress

点击查看摘要

Abstract:We study how Large Language Models (LLMs) process negation mechanistically. First, we establish that even though open-weight models often provide wrong answers to questions involving negation, they do possess internal components that process negation correctly. Their poor accuracy is due to late-layer attention behavior that promotes simple shortcuts; ablating those attention modules greatly improves accuracy on negation-related questions. Second, we uncover how models process negation. We consider two hypotheses: models could use attention heads that attend to the phrase being negated and suppress related concepts, or they could directly construct a representation of the entire negative phrase (e.g., representing “not gas” as a vector that promotes liquids and solids). We apply a range of observational and causal interpretability techniques on Mistral-7B and Llama-3.1-8B to show that models implement both mechanisms, with the “constructive” mechanism being more prominent. Combined, our work deepens the understanding of LLMs’ internals, highlighting construction-dominant computations and the coexistence of competing mechanisms within LLMs.

[NLP-60] Evaluating Reasoning Models for Queries with Presuppositions ACL2026

【速读】：该论文旨在解决大语言模型（Large Language Models, LLMs）在面对用户查询中隐含的错误假设（presupposition）时，往往无法识别并纠正这些假设，从而可能强化用户的错误认知的问题。其解决方案的关键在于重新评估具备推理能力的大推理模型（Large Reasoning Models, LRMs）是否能够识别并回应这些假设性前提。研究通过构建涵盖健康、科学和通用知识领域、具有不同程度 presupposition 的查询数据集，对多个主流模型进行评估，发现尽管LRMs相比非推理模型在准确性上略有提升（2-11%），但仍无法挑战约26-42%的虚假前提，且其表现受 presupposition 表达强度的影响显著，表明当前推理能力尚不足以系统性地处理假设性偏见问题。

链接: https://arxiv.org/abs/2605.03050
作者: Rose Sathyanathan,Kinshuk Vasisht,Danish Pruthi
机构: Indian Institute of Science (印度科学研究所)
类目: Computation and Language (cs.CL)
备注: Accepted to ACL 2026 (Findings)

点击查看摘要

Abstract:Millions of users turn to AI models for their information needs. It is conceivable that a large number of user queries contain assumptions that may be factually inaccurate. Prior work notes that large language models (LLMs) often fail to challenge such erroneous assumptions, and can reinforce users’ misinformed opinions. However, given the recent advances, especially in model’s reasoning capabilities, we revisit whether large reasoning models (LRMs) can reason about the underlying assumptions and respond to user queries appropriately. We construct queries with varying degrees of presuppositions spanning health, science, and general knowledge, and use it to evaluate several widely-deployed models When compared to non-reasoning models, we find that reasoning models achieve a slightly higher accuracy (2-11%), but they still fail to challenge a large fraction (26-42%) of false presuppositions. Further, reasoning models remain susceptible to how strongly the presupposition is expressed.

[NLP-61] Multilingual Safety Alignment via Self-Distillation

【速读】：该论文旨在解决大语言模型（Large Language Models, LLMs）在多语言场景下的安全对齐失衡问题，即模型在高资源语言中具备较强的防护能力，但在低资源语言中极易受到越狱攻击（jailbreak attacks）。现有安全对齐方法通常依赖于每种目标语言的高质量响应数据，而这类数据的获取成本高昂且难以实现。解决方案的关键在于提出一种名为多语言自蒸馏（Multilingual Self-Distillation, MSD）的跨语言安全防护迁移框架，该框架无需任何目标语言的响应数据，仅通过多语言查询即可将高资源语言（如英语）中的内在安全能力迁移到低资源语言（如爪哇语）。其核心创新包括两种具体实现方式——在线策略MSD和离线策略MSD，以及一种双视角安全权重机制（Dual-Perspective Safety Weighting, DPSW），该机制通过同时考虑教师模型与学生模型的视角，动态调整安全关键token的惩罚权重，从而优化蒸馏目标，显著提升模型在多种语言上的安全性能并保持通用能力。

链接: https://arxiv.org/abs/2605.02971
作者: Ruiyang Qin,Qingzhuo Wang,Dongrui Liu,Qiang Li,Zhihua Wei,Wen Shen
机构: Tongji University(同济大学); Shanghai AI Laboratory(上海人工智能实验室)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:

点击查看摘要

Abstract:Large language models (LLMs) exhibit severe multilingual safety misalignment: they possess strong safeguards in high-resource languages but remain highly vulnerable to jailbreak attacks in low-resource languages. Current safety alignment methods generally rely on high-quality response data for each target language, which is expensive and difficult to generate. In this paper, we propose a cross-lingual safeguard transfer framework named Multilingual Self-Distillation (MSD). This framework transfers an LLM’s inherent safety capabilities from high-resource (e.g., English) to low-resource (e.g., Javanese) languages, overcoming the need for response data in any language. Our framework is flexible and can be integrated with different self-distillation strategies. Specifically, we implement two concrete methods – on-policy MSD and off-policy MSD – both of which enable effective cross-lingual safety transfer using only multilingual queries. Furthermore, we propose Dual-Perspective Safety Weighting (DPSW), a divergence measure to optimize the distillation objective. By jointly considering the perspectives of both the teacher and the student, DPSW adaptively increases the penalty weights on safety-critical tokens while reducing the weights on non-critical tokens. Extensive experiments on representative LLMs across diverse multilingual jailbreak and utility benchmarks demonstrate that our method consistently achieves superior multilingual safety performance. Notably, it generalizes effectively to more challenging datasets and unseen languages while preserving the model’s general capabilities.

[NLP-62] AutoRAG Tuner: A Declarative Framework for Automatic Optimization of RAG Pipelines EUROSYS2026

【速读】：该论文旨在解决检索增强生成（Retrieval-Augmented Generation, RAG）系统在实际部署中因架构设计复杂性和超参数配置敏感性导致的性能波动问题，以及当前依赖人工调参所带来的低效与高工程开销。其解决方案的关键在于提出一个声明式、配置驱动的自动化框架 AutoRAGTuner，该框架通过模块化架构解耦 RAG 管道各阶段，并引入 Domain-Element Model (DEM) 统一异构数据表示，同时集成自适应贝叶斯优化引擎实现端到端超参数调优，从而显著提升 RAG 系统的可扩展性、可维护性与性能稳定性。

链接: https://arxiv.org/abs/2605.02967
作者: Xintan Zeng,Yongchao Liu,Yice Luo,Jiajun Zhen
机构: Ant Group(蚂蚁集团)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Distributed, Parallel, and Cluster Computing (cs.DC); Software Engineering (cs.SE)
备注: Accepted by EuroSys 2026 (poster track)

点击查看摘要

Abstract:Retrieval-Augmented Generation (RAG) enhances LLMs, but performance is highly sensitive to complex architecture designs and hyper-parameter configurations, which currently rely on inefficient manual tuning. We present AutoRAGTuner, a declarative, configuration-driven framework that automates the RAG life cycle: construction, execution,evaluation, and optimization. AutoRAGTuner employs a modular architecture to decouple pipeline stages through a component registration mechanism. To unify heterogeneous data, we introduce the Domain-Element Model (DEM), representing objects as atomic elements with bidirectional pointers to support nodes, edges, and hyperedges. Furthermore, AutoRAGTuner integrates an adaptive Bayesian optimization engine for end-to-end hyper-parameter tuning. Experimental results demonstrate AutoRAGTuner’s architectural generality: across diverse RAG pipelines, ranging from vanilla to graph-based, the framework consistently outperforms default baselines. Notably, AutoRAGTuner significantly mitigates engineering overhead, where its declarative configuration language enables a up to 95% reduction in code churn for architectural adjustments. Overall, AutoRAGTuner provides a systematically optimizable foundation for building evolvable and reusable RAG systems.

[NLP-63] racing the Dynamics of Refusal: Exploiting Latent Refusal Trajectories for Robust Jailbreak Detection ICML2026

【速读】：该论文旨在解决当前生成式 AI (Generative AI) 安全防御中依赖静态拒绝向量（refusal vectors）导致的脆弱性问题，即现有方法在面对对抗性攻击（如 GCG 攻击）时无法有效识别和抵御恶意请求。其解决方案的关键在于发现并利用“拒绝轨迹”（Refusal Trajectory）——一种在推理过程中持续存在的上游稀疏激活模式，该模式即使在终端信号被压制的情况下依然保持稳定。基于此，作者提出 SALO（Sparse Activation Localization Operator），一种在推理阶段检测此类潜在模式的算子，从而显著提升对强制解码攻击的检测能力，将检测率从约 0% 提升至 90%。

链接: https://arxiv.org/abs/2605.02958
作者: Xulin Hu,Che Wang,Wei Yang Bryan Lim,Jianbo Gao,Zhong Chen
机构: 未知
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Accepted to the 43rd International Conference on Machine Learning (ICML 2026). Pre-camera-ready version

点击查看摘要

Abstract:Representation Engineering typically relies on static refusal vectors derived from terminal representations. We move beyond this paradigm, demonstrating that refusal is a dynamic and sparse process rather than a localized outcome. Using Causal Tracing, we uncover the Refusal Trajectory-a persistent upstream signature that remains intact even when adversarial attacks (e.g., GCG) suppress terminal signals. Leveraging this, we propose SALO (Sparse Activation Localization Operator), an inference-time detector designed to capture these latent patterns. SALO effectively recovers defense capabilities against forced-decoding attacks, improving detection rates from ~0% to 90% where methods relying on terminal states perform poorly.

[NLP-64] When Should a Language Model Trust Itself? Same-Model Self-Verification as a Conditional Confidence Signal

【速读】：该论文旨在解决生成式 AI (Generative AI) 在选择性预测（selective prediction）中如何有效评估模型置信度的问题，特别是自验证机制（same-model self-verification）作为不确定性估计方法的实际价值是否可靠。其解决方案的关键在于系统性地将自验证方法与两个强基线——LL-AVG（平均对数似然）和 LL-SUM（对数似然之和）进行对比评估，涵盖多个模型家族、规模及提示变体，在ARC-Challenge和TruthfulQA-MC两个任务上量化其正确性排序和拒答质量（通过AURC和操作点分析）。结果表明，自验证并非通用的不确定性估计器，而是一种条件性置信信号，其有效性高度依赖于任务类型、模型家族、提示设计以及所比较的基线表现。

链接: https://arxiv.org/abs/2605.02915
作者: Aditya Ajay Phalod
机构: Independent Researcher
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 30 pages, 13 figures, 8 tables. Code available on GitHub

点击查看摘要

Abstract:Same-model self-verification, prompting a model to audit its own predicted answer, is a plausible confidence signal for selective prediction, but its practical value remains unclear once strong likelihood-based baselines are taken seriously. We evaluate self-verification against two such baselines, LL-AVG and LL-SUM, on ARC-Challenge and TruthfulQA-MC across multiple model families, scales, and prompt variants. We measure not only correctness ranking, but also abstention quality through AURC and operating-point analyses. The results are sharply task- and model-dependent. On ARC-Challenge, self-verification substantially improves over LL-AVG for Phi-2 and the Qwen models, with the largest gains appearing in Qwen-7B. On TruthfulQA-MC, however, the signal is less reliable: smaller models can become prompt-sensitive, DeepSeek-R1-Distill-8B degrades relative to LL-AVG, and LL-SUM often remains the stronger practical baseline. We therefore do not treat self-verification as a general-purpose uncertainty estimator. In this setting, it is better understood as a conditional confidence signal whose value depends on task type, model family, prompt formulation, and, crucially, the baseline it must beat.

[NLP-65] CreativityBench: Evaluating Agent Creative Reasoning via Affordance-Based Tool Repurposing

【速读】：该论文旨在解决当前大语言模型（Large Language Models, LLMs）在生成式 AI (Generative AI) 领域中对创造性问题求解能力不足的问题，特别是针对“创造性工具使用”这一关键维度的缺失。其核心挑战在于模型能否通过推理物体的可及性（affordance）和属性来重新利用现有对象，而非依赖其常规用途。解决方案的关键是构建了一个名为 CreativityBench 的基准测试平台，其中包含一个涵盖 4000 个实体与超过 15 万条可及性标注的大规模知识库（Knowledge Base, KB），明确关联物体、部件、属性与可执行用途；并基于此 KB 设计了 14000 个具有物理合理性约束的地面化任务，用于评估模型在识别非显而易见但可行的解决方案时的能力。实验表明，尽管模型能选择合理物体，但在识别正确部件、其可及性及背后的物理机制上表现不佳，且模型规模扩展带来的性能提升迅速饱和，说明创造性工具使用仍是当前 LLMs 的重大挑战。

链接: https://arxiv.org/abs/2605.02910
作者: Cheng Qian,Hyeonjeong Ha,Jiayu Liu,Bingxiang He,Jeonghwan Kim,Jiateng Liu,Bingxuan Li,Aditi Tiwari,Dwip Dalal,Zhenhailong Wang,Xiusi Chen,Mahdi Namazifar,Yunzhu Li,Heng Ji
机构: UIUC(伊利诺伊大学厄巴纳-香槟分校); THU(清华大学); Amazon(亚马逊); Columbia(哥伦比亚大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 57 Pages, 14 Tables, 27 Figures

点击查看摘要

Abstract:Recent advances in large language models have led to strong performance on reasoning and environment-interaction tasks, yet their ability for creative problem-solving remains underexplored. We study this capability through the lens of creative tool use, where a model repurposes available objects by reasoning about their affordances and attributes rather than relying on canonical usage. As a first step, we introduce CreativityBench, a benchmark for evaluating affordance-based creativity in LLMs. To this end, we build a large-scale affordance knowledge base (KB) with 4K entities and 150K+ affordance annotations, explicitly linking objects, parts, attributes, and actionable uses. Building on this KB, we generate 14K grounded tasks that require identifying non-obvious yet physically plausible solutions under constraints. Evaluations across 10 state-of-the-art LLMs, including closed and open-source models, show that models can often select a plausible object, but fail to identify the correct parts, their affordances, and the underlying physical mechanism needed to solve the task, leading to a significant drop in performance. Furthermore, improvements from model scaling quickly saturate, strong general reasoning does not reliably translate to creative affordance discovery, and common inference-time strategies such as Chain-of-Thought yield limited gains. These results suggest that creative tool use remains a major challenge for current models, and that CreativityBench provides a useful testbed for studying this missing dimension of intelligence, with potential implications for planning and reasoning modules in future agents.

[NLP-66] he Kerimov-Alekberli Model: An Information-Geometric Framework for Real-Time System Stability

【速读】：该论文旨在解决当前人工智能（AI）安全领域中伦理对齐（ethical alignment）缺乏物理基础的问题，即如何将抽象的伦理规范转化为可量化、可验证的系统稳定性指标。其解决方案的关键在于构建一个基于信息几何（information-geometric）的新型框架——Kerimov-Alekberli模型，该模型通过建立非平衡热力学（non-equilibrium thermodynamics）与随机控制（stochastic control）之间的形式同构关系，将系统异常定义为偏离黎曼流形（Riemannian manifold）的偏差，并以KL散度（Kullback-Leibler divergence）作为核心度量，其动态阈值由费舍尔信息度量（Fisher Information Metric）确定。进一步地，该框架基于兰道尔原理（Landauer Principle）证明对抗扰动会执行可测量的物理功，从而增加系统的熵信息，实现了从启发式规则向热力学稳定性的范式转变。

链接: https://arxiv.org/abs/2604.24083
作者: Hikmat Karimov,Rahid Zahid Alekberli
机构: Azerbaijan Technical University (阿塞拜疆技术大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Cryptography and Security (cs.CR); Machine Learning (cs.LG)
备注:

点击查看摘要

Abstract:This study introduces the Kerimov-Alekberli model, a novel information-geometric framework that redefines AI safety by formally linking non-equilibrium thermodynamics to stochastic control for the ethical alignment of autonomous systems. By establishing a formal isomorphism between non-equilibrium thermodynamics and stochastic control, we define systemic anomalies as deviations from a Riemannian manifold. The model utilizes the Kullback-Leibler divergence as the primary metric, governed by a dynamic threshold derived from the Fisher Information Metric. We further ground this framework in the Landauer Principle, proving that adversarial perturbations perform measurable physical work by increasing the system’s informational entropy. Validation on the NSL-KDD dataset and unmanned aerial vehicle trajectory simulations demonstrated that our model achieves effective real-time detection via the FPT trigger, with strong performance metrics (e.g., high accuracy and low FPR) on benchmark datasets. This study provides a rigorous physical foundation for AI safety, transitioning from heuristic, rule-based ethical frameworks to a thermodynamics-based stability paradigm by grounding ethical violations in quantifiable physical work and entropic information.

[NLP-67] An Information-Geometric Framework for Stability Analysis of Large Language Models under Entropic Stress

【速读】：该论文旨在解决当前大型语言模型（Large Language Models, LLMs）在高风险和实际应用场景中，仅依赖聚合准确率进行评估时难以充分刻画系统可靠性的难题。其解决方案的关键在于提出一种受热力学启发的建模框架，通过引入一个复合稳定性评分（composite stability score），将任务效用、熵（entropy，作为外部不确定性度量）以及两个内部结构代理变量——内部整合性（internal integration）和对齐反思能力（aligned reflective capacity）进行统一建模。该框架并非将这些量视为物理变量，而是作为一种可解释的抽象机制，用于捕捉模型内部结构如何调节无序对行为的影响，从而更全面地评估LLM在扰动和不确定性条件下的稳定性表现。

链接: https://arxiv.org/abs/2604.24076
作者: Hikmat Karimov,Rahid Zahid Alekberli
机构: Azerbaijan Technical University (阿塞拜疆技术大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Cryptography and Security (cs.CR); Machine Learning (cs.LG)
备注:

点击查看摘要

Abstract:As large language models (LLMs) are increasingly deployed in high-stakes and operational settings, evaluation strategies based solely on aggregate accuracy are often insucient to characterize system reliability. This study proposes a thermodynamic inspired modeling framework for analyzing the stability of LLM outputs under conditions of uncertainty and perturbation. The framework introduces a composite stability score that integrates task utility, entropy as a measure of external uncertainty, and two internal structural proxies: internal integration and aligned reective capacity. Rather than interpreting these quantities as physical variables, the formulation is intended as an interpretable abstraction that captures how internal structure may modulate the impact of disorder on model behavior. Using the IST-20 benchmarking protocol and associated metadata, we analyze 80 modelscenario observations across four contemporary LLMs. The proposed formulation consistently yields higher stability scores than a reduced utilityentropy baseline, with a mean improvement of 0.0299 (95% CI: 0.02470.0351). The observed gain is more pronounced under higher entropy conditions, suggesting that the framework captures a form of nonlinear attenuation of uncertainty. We do not claim a fundamental physical law or a complete theory of machine ethics. Instead, the contribution of this work is a compact and interpretable modeling perspective that connects uncertainty, performance, and internal structure within a unied evaluation lens. The framework is intended to complement existing benchmarking approaches and to support ongoing discussions in AI safety, reliability, and governance.

信息检索

[IR-0] Rethinking Reasoning -Intensive Retrieval: Evaluating and Advancing Retrievers in Agent ic Search Systems ACL2026

【速读】：该论文旨在解决当前检索系统在推理密集型任务中难以有效提取支持下游推理的互补证据的问题，尤其是在代理式搜索（agentic search）场景下，传统检索器仅关注主题相似性而忽视多角度证据组合的能力。其解决方案的关键在于：构建了BRIGHT-Pro这一专家标注的基准，扩展每个查询的黄金证据为多维度视角，并在静态与代理式搜索协议下评估检索器；同时设计RTriever-Synth合成语料库，通过分解语义维度生成互补正样本和条件硬负样本，用于LoRA微调基于Qwen3-Embedding-4B的RTriever-4B模型，从而显著提升检索器在推理密集型任务中的表现。

链接: https://arxiv.org/abs/2605.04018
作者: Yilun Zhao,Jinbiao Wei,Tingyu Song,Siyue Zhang,Chen Zhao,Arman Cohan
机构: 未知
类目: Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注: ACL 2026

点击查看摘要

Abstract:Reasoning-intensive retrieval aims to surface evidence that supports downstream reasoning rather than merely matching topical similarity. This capability is increasingly important for agentic search systems, where retrievers must provide complementary evidence across iterative search and synthesis. However, existing work remains limited on both evaluation and training: benchmarks such as BRIGHT provide narrow gold sets and evaluate retrievers in isolation, while synthetic training corpora often optimize single-passage relevance rather than evidence portfolio construction. We introduce BRIGHT-Pro, an expert-annotated benchmark that expands each query with multi-aspect gold evidence and evaluates retrievers under both static and agentic search protocols. We further construct RTriever-Synth, an aspect-decomposed synthetic corpus that generates complementary positives and positive-conditioned hard negatives, and use it to LoRA fine-tune RTriever-4B from Qwen3-Embedding-4B. Experiments across lexical, general-purpose, and reasoning-intensive retrievers show that aspect-aware and agentic evaluation expose behaviors hidden by standard metrics, while RTriever-4B substantially improves over its base model.

[IR-1] Domain-Adaptive Dense Retrieval for Brazilian Legal Search

【速读】：该论文旨在解决巴西法律检索（Brazilian legal retrieval）中因数据类型异质性（涵盖判例法、法规和问答式搜索）而导致的稠密检索模型训练难题，即如何在领域专业化与跨检索类型鲁棒性之间取得平衡。其解决方案的关键在于设计三种不同的训练策略：无微调的基础模型、仅使用法律数据微调的模型，以及结合法律数据与SQuAD-pt监督数据的混合训练方案。实验表明，混合训练在保持法律检索性能的同时显著提升了整体泛化能力，尤其在非法律领域的葡萄牙语检索基准（Quati）上表现更优，实现了从单一任务优化向多类型检索适应性的转变。

链接: https://arxiv.org/abs/2605.04005
作者: Jayr Pereira,Roberto Lotufo,Luiz Bonifacio
机构: 未知
类目: Information Retrieval (cs.IR)
备注:

点击查看摘要

Abstract:Brazilian legal retrieval is heterogeneous, covering case law, legislation, and question-based search. This makes training dense retrievers a trade-off between stronger domain specialization and broader robustness across retrieval types of search. In this paper, we explore this trade-off using three training setups based on Qwen3-Embedding-4B: a base model with no fine-tuning, a version trained only on legal data, and a mixed setup that combines legal data with SQuAD-pt supervised dataset. We evaluate these models on five legal datasets from the JUÁ leaderboard, along with Quati dataset as an extra Portuguese retrieval benchmark to test out-of-domain generalization. The legal-only model performs best on the most specialized legal tasks. The mixed setup keeps strong performance on legal data while offering a better overall balance, improving average NDCG@10 from 0.414 to 0.447, MRR@10 from 0.586 to 0.595, and MAP@10 from 0.270 to 0.308 across all six datasets. The biggest improvement appears on Quati, where the mixed model clearly outperforms the legal-only one. Overall, the results show that legal-only and mixed training lead to different strengths: the first is better for specialization, while the second is more robust across different types of search, especially question-based ones. Both adapted models are available on Hugging Face

[IR-2] Aspect-Aware Content-Based Recommendations for Mathematical Research Papers SIGIR

【速读】：该论文旨在解决数学领域中基于内容的研究论文推荐（Content-based Research Paper Recommendation, CbRPR）有效性不足的问题。由于数学论文之间的关联性主要体现在概念层面（如共享的证明技巧、逻辑蕴含或自然推广），而非显式的文本重叠或引用关系，传统依赖文本相似度或引用网络的方法难以奏效。解决方案的关键在于提出一种以视角（aspect）为导向的异构图神经网络模型 AchGNN，该模型通过联合建模文本语义、引用结构和作者传承路径（author lineage），实现对数学论文间深层概念关联的精准捕捉。同时，作者构建了首个面向数学领域的视角感知推荐数据集 GoldRiM 和 SilverRiM，为该问题提供了高质量标注基础，并验证了方法在跨领域（如机器学习）中的泛化能力。

链接: https://arxiv.org/abs/2605.03861
作者: Ankit Satpute,André Greiner-Petter,Noah Gießing,Olaf Teschke,Moritz Schubotz,Akiko Aizawa,Bela Gipp
机构: FIZ Karlsruhe(弗劳恩霍夫信息技术研究所); University of Göttingen(哥廷根大学); National Institute of Informatics(日本国立信息学研究所)
类目: Information Retrieval (cs.IR)
备注: Accepted for publication at the 49th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR) July 20–24, 2026, Melbourne, VIC, Australia

点击查看摘要

Abstract:Content-based research paper recommendation (CbRPR) has seen advances in computer science and biomedicine, but remains unexplored for mathematics, where paper relatedness is more conceptual than explicit textual or citation-based similarity. Mathematics papers may be connected through shared proof techniques, logical implications, or natural generalizations, yet exhibit minimal textual or citation overlap, rendering existing CbRPR ineffective. To address this gap, we first conduct an expert-driven study characterizing mathematical recommendations, revealing that relevance is inherently \textitaspect-driven. Grounded in this insight, we introduce GoldRiM (small, expert-annotated) and SilverRiM (large, automatically derived), the first datasets for \textitaspect-aware CbRPR in mathematics. Recognizing that LLM embeddings of mathematical content alone yield suboptimal representation, we propose AchGNN, an \textitaspect-conditioned heterogeneous GNN that jointly models textual semantics, citation structure, and author lineage. Across GoldRiM and SilverRiM, AchGNN consistently outperforms prior \textitaspect-based CbRPR methods, achieving substantial gains across all evaluated \textitaspects. We conduct ablation studies to analyze the contributions of individual \textitaspect supervision, authorship lineage, and graph-structural signals to AchGNN’s performance. To assess domain generality, we further evaluate AchGNN on the \textitPapers with Code dataset of machine learning publications, demonstrating that our \textitaspect-aware approach effectively transfers beyond mathematics. We deploy our system on the MaRDI platform to help mathematicians with recommendations and release datasets and code publicly for reproducibility.

[IR-3] Cosmodoit: A Python Package for Adaptive Efficient Pipelining of Feature Extraction from Performed Music

【速读】：该论文旨在解决音乐性能分析中算法与工具分散于不同编程语言和数据格式导致难以高效集成的问题。解决方案的关键在于提出一个名为Cosmodoit的新型Python包，其核心优势在于构建了一个模块化、灵活的处理流水线，能够整合性能到乐谱对齐、符号与音频特征提取，并支持选择性处理、依赖感知计算和增量更新，从而减少重复工作、降低错误率并实现大规模高效处理。

链接: https://arxiv.org/abs/2605.03541
作者: Corentin Guichaoua,Daniel Bedoya,Elaine Chew
机构: STMS Laboratoire (UMR9912) – CNRS, IRCAM, Sorbonne Université, Ministère de la Culture; King’s College London
类目: ound (cs.SD); Information Retrieval (cs.IR)
备注: 6 pages, 1 figure

点击查看摘要

Abstract:Computational analysis of performed music is a key component of music information research, as performance shapes much of the music we hear. Music performance analysis studies the acoustic variations introduced by performers and how these variations reflect musical interpretation and structure. Although many algorithms and tools exist for tasks such as performance-to-score alignment and symbolic or audio feature extraction, they are spread across different programming languages and data formats, making them difficult to combine efficiently. To address this problem, we present Cosmodoit, a novel Python package designed to streamline feature extraction from performed music. Cosmodoit integrates performance-to-score alignment with symbolic and audio feature extraction in a modular, flexible pipeline that supports selective processing, dependency-aware computation, and incremental updates. Its extensible design reduces duplicated work, minimizes errors, and enables efficient large-scale processing. By accommodating algorithms implemented in multiple languages and allowing parameter tuning for consistent feature extraction, Cosmodoit provides a versatile and practical tool for both research and development in music performance analysis.

[IR-4] SURE-RAG : Sufficiency and Uncertainty-Aware Evidence Verification for Selective Retrieval-Augmented Generation

【速读】：该论文旨在解决检索增强生成（Retrieval-Augmented Generation, RAG）中“检索不等于验证”的核心问题：即 retrieved passages 虽然与问题相关，但未必能充分支持最终答案，导致生成结果存在事实性错误或误导。为填补这一证据充分性验证的空白，作者提出 SURE-RAG，其关键创新在于将证据充分性视为一种集合级（set-level）属性，并设计了一个透明的聚合协议：通过一个共享的成对命题-证据验证器（pair-level claim-evidence verifier）生成局部关系分布，再由 SURE-RAG 汇总为可解释的问答级信号（如覆盖度、关系强度、冲突度等），从而实现三类决策（支持/反驳/不足）及可审计的选择性得分。该方法在 HotpotQA-RAG v3 上显著优于主流基线（Macro-F1 达 0.9075），同时将低覆盖率下的风险降低 37%，并明确区分了受控证据充分性验证与自然幻觉检测之间的本质差异。

链接: https://arxiv.org/abs/2605.03534
作者: Jingxi Qiu,Zeyu Han,Cheng Huang
机构: 未知
类目: Computation and Language (cs.CL); Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注: 8 pages, 2 figures, 8 tables. Submitted to IEEE PRAI 2026

点击查看摘要

Abstract:Retrieval-augmented generation (RAG) grounds answers in retrieved passages, but retrieval is not verification: a passage can be topical and still fail to justify the answer. We frame this gap as evidence sufficiency verification for selective RAG answering: given a question, a candidate answer, and retrieved evidence, predict whether the evidence supports, refutes, or is insufficient, and abstain unless support is established. We present SURE-RAG, a transparent aggregation protocol built on the observation that evidence sufficiency is a set-level property: missing hops and unresolved conflicts cannot be detected by independent passage scoring. A shared pair-level claim-evidence verifier produces local relation distributions, which SURE-RAG aggregates into interpretable answer-level signals – coverage, relation strength, disagreement, conflict, and retrieval uncertainty – yielding a three-way decision and an auditable selective score. We evaluate on HotpotQA-RAG v3, a controlled multi-hop benchmark, under an artifact-aware protocol (shortcut baselines, counterfactual swaps, no-oracle checks, GPT-4o audits). Calibrated SURE-RAG reaches 0.9075 Macro-F1 (0.8951 +/- 0.0069), substantially above DeBERTa mean-pooling (0.6516) and a GPT-4o judge (0.7284), while matching a strong but opaque concat cross-encoder (0.8888 +/- 0.0109) with full auditability. Risk at 30% coverage drops from 0.2588 to 0.1642, a 37% reduction in unsafe answers. To deliberately probe the task boundary, we further contrast SURE-RAG with GPT-4o on HaluBench unsafe detection: the ranking reverses (0.3343 vs 0.7389 unsafe-F1), establishing that controlled sufficiency verification and natural hallucination detection are distinct problems.

[IR-5] Revisiting General Map Search via Generative Point-of-Interest Retrieval

【速读】：该论文旨在解决通用地图搜索场景中因用户查询信息不足（underspecified user queries）而导致的传统POI（兴趣点，Point-of-Interest）检索方法性能下降的问题。现有方法过度依赖表面语义匹配，难以有效融合异构上下文以推断复杂的搜索意图。解决方案的关键在于提出GenPOI框架，其核心创新包括：1）引入Geo-Semantic POI Tokenization，将每个POI表示为编码语义与地理空间信息的紧凑标记序列，从而增强大语言模型（Large Language Models, LLMs）的空间感知能力；2）采用邻近感知的约束生成策略，在解码过程中限制LLM的输出空间，确保生成结果在地理空间上的有效性与相关性。这一生成式范式显著提升了对复杂、个性化且高度情境依赖查询的理解与响应能力。

链接: https://arxiv.org/abs/2605.03397
作者: Dong Chen,Shuai Zheng,Haoyang Shao,Hongsheng Wu,Muhao Xu,Yeyu Yan,Ruifang Li,Zhenfeng Zhu
机构: Beijing Jiaotong University (北京交通大学); Tencent Inc. (腾讯公司)
类目: Information Retrieval (cs.IR)
备注: 11 pages, 5 figures, 5 table

点击查看摘要

Abstract:Point-of-Interest (POI) retrieval aims to identify relevant candidates from massive-scale POI databases, serving as a cornerstone for diverse location-based services. However, in general map search scenarios, conventional POI retrieval methods are increasingly challenged by underspecified user queries due to their excessive reliance on surface-level semantic matching. Meanwhile, such queries are often highly context-dependent and personalized, yet existing retrieval paradigms struggle to effectively synergize heterogeneous contexts for complex search intent inference. To address these limitations, we revisit general map search from a generative perspective and propose GenPOI, an innovative Generative POI retrieval framework tailored for general search on maps. It seamlessly unifies heterogeneous search contexts and POIs into structured sequences, leveraging the powerful contextual modeling of Large Language Models (LLMs) for spatial-aware candidate generation. Consequently, this generative paradigm effectively solves more challenging queries through profound context dependency modeling and search intent reasoning. Specifically, accounting for the unique geospatial nature of map scenarios, GenPOI introduces a novel Geo-Semantic POI Tokenization to represent each POI as a compact token sequence encoding both semantic and geographic context, thus grounding the LLM’s spatial understanding. Additionally, a proximity-aware constrained generation strategy is employed to restrict the decoding space of the LLM, ensuring the validity and geospatial relevance of the generated results. Extensive experiments on large-scale industrial datasets from Tencent Map, comprising POIs at the scale of over 10 million, demonstrate the superior performance of GenPOI.

[IR-6] RAG over Thinking Traces Can Improve Reasoning Tasks

【速读】：该论文旨在解决检索增强生成（Retrieval-Augmented Generation, RAG）在推理密集型任务（如数学和代码生成）中效果有限的问题。传统RAG依赖于从标准文档语料库中检索信息，但研究表明其对复杂推理任务的提升作用有限。论文的关键创新在于提出以“思维轨迹”（thinking traces）作为新的检索语料库——即问题求解过程中生成的中间推理路径，并进一步引入T3方法将这些轨迹转化为结构化、紧凑且易于检索的表示形式。实验表明，基于思维轨迹的RAG在多个基准测试（如AIME 2025–2026、LiveCodeBench和GPQA-Diamond）上显著优于非RAG基线及传统网页语料检索方案，且不增加或反而降低推理成本，从而证明了思维轨迹是推理任务中更有效的检索源。

链接: https://arxiv.org/abs/2605.03344
作者: Negar Arabzadeh,Wenjie Ma,Sewon Min,Matei Zaharia
机构: University of California, Berkeley (加州大学伯克利分校)
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:

点击查看摘要

Abstract:Retrieval-augmented generation (RAG) has proven effective for knowledge-intensive tasks, but is widely believed to offer limited benefit for reasoning-intensive problems such as math and code generation. We challenge this assumption by showing that the limitation lies not in RAG itself, but in the choice of corpus. Instead of retrieving documents, we propose retrieving thinking traces, i.e., intermediate thinking trajectories generated during problem solving attempts. We show that thinking traces are already a strong retrieval source, and further introduce T3, an offline method that transforms them into structured, retrieval-friendly representations, to improve usability. Using these traces as a corpus, a simple retrieve-then-generate pipeline consistently improves reasoning performance across strong models and benchmarks such as AIME 2025–2026, LiveCodeBench, and GPQA-Diamond, outperforming both non-RAG baselines and retrieval over standard web corpora. For instance, on AIME, RAG with traces generated by Gemini-2-thinking achieves relative gains of +56.3%, +8.6%, and +7.6% for Gemini-2.5-Flash, GPT-OSS-120B, and GPT-5, respectively, even though these are more recent models. Interestingly, RAG on T3 also incurs little or no extra inference cost, and can even reduce inference cost by up to 15% . Overall, our results suggest that thinking traces are an effective retrieval corpus for reasoning tasks, and transforming them into structured, compact, or diagnostic representations unlocks even stronger gains. Code available at this https URL.

[IR-7] Beyond Similarity Search: A Unified Data Layer for Production RAG Systems

【速读】：该论文旨在解决检索增强生成（Retrieval-Augmented Generation, RAG）系统在生产环境中性能与原型阶段表现不一致的问题，其根源在于传统分层数据架构导致的数据陈旧性（data staleness）、租户间数据泄露（tenant data leakage）以及查询组合爆炸（query composition explosion）。解决方案的关键是构建一个统一的数据层，基于PostgreSQL并集成原生向量搜索（pgvector）与HNSW索引，从而实现高效、一致且安全的向量检索与管理。实验证明，该方案在5万文档规模下显著降低延迟（日期过滤查询减少92%，租户作用域查询减少74%），消除同步不一致和跨租户数据泄露，并将同步代码量减少93%。

链接: https://arxiv.org/abs/2605.03275
作者: Venkata Krishna Prasanth Budigi,Siri Chandana Sirigiri
机构: 未知
类目: Information Retrieval (cs.IR); Databases (cs.DB)
备注: 8 pages, 1 figure, 4 tables

点击查看摘要

Abstract:Retrieval-Augmented Generation (RAG) systems have become the standard architecture for grounding large language models in organizational knowledge. Yet production deployments consistently expose a gap between clean prototype performance and real-world reliability. This paper identifies three root causes of that gap: data staleness, tenant data leakage, and query composition explosion. All three trace back to the conventional split-system data layer. We propose and evaluate a unified data layer built on PostgreSQL with native vector search (pgvector) and HNSW indexing. Controlled benchmarks on 50,000 documents show 92% latency reduction for date-filtered queries, 74% for tenant-scoped queries, zero synchronization inconsistency, and complete elimination of cross-tenant data leakage with 93% less synchronization code. We additionally discuss a recommended hybrid tier architecture

人机交互

[HC-0] Stayin Aligned Over Time: Towards Longitudinal Human-LLM Alignment via Contextual Reflection and Privacy-Preserving Behavioral Data

【速读】：该论文旨在解决当前大型语言模型（Large Language Models, LLMs）的人机对齐（human-AI alignment）与评估方法普遍依赖即时偏好信号的问题，即这些方法假设用户偏好是静态的，而忽视了LLM决策在真实世界中可能随时间推移和结果反馈被重新评估的现象。解决方案的关键在于提出一种纵向、情境化的对齐测量框架，通过三个核心机制实现：（1）情境内偏好捕获（in-situ preference capture），（2）由上下文触发的后续偏好反思（context-triggered follow-up preference reflection），以及（3）隐私保护的行为轨迹记录（privacy-preserving behavioral traces）以解释偏好变化。该方法以BITE系统为实例，在为期两周的纵向部署研究中验证了其有效性，揭示了用户即时偏好与后期偏好在准确性、相关性等维度上的差异，凸显了长期评估对日常使用场景下LLM对齐的重要性。

链接: https://arxiv.org/abs/2605.04029
作者: Simret Araya Gebreegziabher,Allison E Sproul,Yinuo Yang,Chaoran Chen,Diego Gómez-Zará,Toby Jia-Jun Li
机构: University of Notre Dame (圣母大学)
类目: Human-Computer Interaction (cs.HC)
备注:

点击查看摘要

Abstract:Current human-AI alignment and evaluation methods for large language models (LLMs) often rely on preference signals collected immediately after an interaction. This practice implicitly treats preference as static, even though many LLM-mediated decisions unfold over time and may be re-evaluated differently after real-world consequences and observed outcomes. Therefore, we argue for a methodological shift from single-moment preference elicitation to longitudinal, context-situated alignment measurement. We present a methodological framework for collecting temporally grounded alignment signals by combining (1) in-situ preference capture, (2) context-triggered follow-up preference reflection, and (3) privacy-preserving behavioral traces that help interpret preference change. As an instantiation of this methodology, we introduce BITE, a browser-based system that detects consequential LLM interactions, prompts reflection across later decision points, and supports progressive, user-controlled consent for sharing behavioral data. Through a two week longitudinal deployment study with 8 participants, our approach surfaced differences between immediate and later user preferences in accuracy, relevance and other dimensions of the LLM output. Our findings highlight the limitations of single-moment preference datasets and underscore the importance of longitudinal methods for alignment evaluation in everyday use.

[HC-1] Deco: Extending Personal Physical Objects into Pervasive AI Companion through a Dual-Embodiment Framework

【速读】：该论文旨在解决现有数字伴侣（Digital Companion）与用户物理依恋物（如毛绒玩具）之间缺乏情感延续性的问题。当前AI伴侣虽具备响应性和个性化能力，但其独立于物理对象存在，无法继承或延续用户对实体物品的情感联结。解决方案的关键在于提出“双具身伴侣框架”（Dual-Embodiment Companion Framework），通过集成多模态大语言模型（Large Language Models, LLMs）和增强现实（Augmented Reality, AR）技术，使数字代理能够同步呈现用户物理陪伴物的“数字化身”，从而实现情感历史的无缝延伸。实证研究表明，该方法显著提升了用户的陪伴感知、情感联结强度及设计原则契合度，并揭示了数字活动可反向激活物理对象、情感深度驱动关系深化等关键机制。

链接: https://arxiv.org/abs/2605.03882
作者: Zhihan Jiang,Mengyuan Millie Wu,Ruishi Zou,Shiyu Xu,Xun Qian,Emma Macmanus,Steven Liao,Ping Zhang,Bingsheng Yao,Tingyu Cheng,James L. David,Nabila El-Bassel,Lena Mamykina,Frances R. Levin,Ryan Sultan,Dakuo Wang,Xuhai Xu
机构: Columbia University (哥伦比亚大学); Harvard University (哈佛大学); University of Michigan (密歇根大学); Google (谷歌); The Ohio State University (俄亥俄州立大学); Northeastern University (东北大学); University of Notre Dame (圣母大学)
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注: 27 pages, 7 figures

点击查看摘要

Abstract:Individuals frequently form deep attachments to physical objects (e.g., plush toys) that usually cannot sense or respond to their emotions. While AI companions offer responsiveness and personalization, they exist independently of these physical objects and lack an ongoing connection to them. To bridge this gap, we conducted a formative study (N=9) to explore how digital agents could inherit and extend the emotional bond, deriving four design principles (Faithful Identity, Calibrated Agency, Ambient Presence, and Reciprocal Memory). We then present the Dual-Embodiment Companion Framework, instantiated as Deco, a mobile system integrating multimodal Large Language Models (LLMs) and Augmented Reality to create synchronized digital embodiments of users’ physical companions. A within-subjects study (N=25) showed Deco significantly outperformed a personalized LLM-empowered digital companion baseline on perceived companionship, emotional bond, and design-principle scales (all p0.01). A seven-day field deployment (N=17) showed sustained engagement, subjective well-being improvement (p=.040), and three key relational patterns: digital activities retroactively vitalized physical objects, bond deepening was driven by emotional engagement depth rather than interaction frequency, and users sustained bonds while actively navigating digital companions’ AI nature. This work highlights a promising alternative for designing digital companions: moving from creating new relationships to dual embodiment, where digital agents seamlessly extend the emotional history of physical objects.

[HC-2] Bodyless Presence: Reconsidering the Minimal Self in Immersive Video

【速读】：该论文旨在解决沉浸式视频（immersive video）中自我体验（self-experience）的理论界定问题，尤其是在缺乏身体参与和有限感官运动耦合的情况下，如何理解用户对“身临其境”（presence）的感知。传统观点常将这种体验归因于具身性（embodiment）或对虚拟化身（avatar）的代理权（agency）与拥有感（ownership），但本文提出，沉浸式视频中的自我体验本质上是一种“以自我位置为主导的状态”（self-location-dominant state），即用户的最小自我（minimal self）不再依赖于身体图式（body schema）的显性激活，而是通过视点位置的空间定位得以确立。解决方案的关键在于重构自我的分析轴心：从代理权与拥有感转向空间自我定位（self-location），从而为XR环境中的自我体验提供一个更契合现象学基础的理论框架。

链接: https://arxiv.org/abs/2605.03873
作者: Koichi Toida
机构: 未知
类目: Human-Computer Interaction (cs.HC); Computers and Society (cs.CY)
备注: 10 pages, 3 figures

点击查看摘要

Abstract:Immersive video, namely 180-degree and 360-degree video designed to be viewed through head-mounted displays, constitutes a boundary case between interactive VR and conventional two-dimensional video for reconsidering self-experience in XR. It can generate a sense of being there without providing a corresponding body, while allowing only limited sensorimotor contingency through head rotation. From a phenomenological standpoint, this paper reinterprets presence in immersive video not as bodily extension or ownership of an avatar, but as a form of self-experience in which self-location becomes relatively dominant under conditions of reduced body schema availability. This paper calls this condition a self-location-dominant state. In immersive video, the user cannot actively intervene in the recorded environment, and stable agency or ownership is difficult to establish. Nevertheless, events such as viewpoint motion, impact, and direct address are not experienced merely as changes within an image, but as events concerning the position of the self. The minimal self in immersive video is therefore redescribed not primarily as a subject of agency or ownership, but as a self spatially located at a viewpoint while the body schema remains backgrounded. This perspective connects research on presence, the sense of embodiment, and the minimal self, and proposes self-location as a central analytic axis for theorising self-experience in immersive video.

[HC-3] RACE: A Metrologically-Grounded Engineering Framework for Trustworthy Agent ic AI Systems in Operationally Critical Domains

【速读】：该论文旨在解决可信智能代理（Trustworthy Agentic AI）在操作关键领域中缺乏统一、可度量且跨域适用的工程框架问题。解决方案的关键在于提出TRACE框架，其核心包括：四层参考架构与显式的经典机器学习（Classical ML）与大语言模型（Large Language Model, LLM）验证器分离设计（L2a/L2b），以明确LLM的使用为可量化的设计选择；基于状态感知的编排与升级策略（L3）和有限人类监督机制（L4）；以及一套基于计量学基础的信任度量体系（映射至GUM/VIM/ISO 17025标准）。其中，创新性引入计算简约比（Computational Parsimony Ratio, CPR）作为第一类设计原则，量化模型复杂度与性能之间的权衡，从而实现对生成式AI（Generative AI）系统的可解释性、可控性和可信性的系统性保障。

链接: https://arxiv.org/abs/2605.03838
作者: Serhii Zabolotnii
机构: CSBC(乌克兰国家科学院计算机科学与控制问题研究所)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注: 11 pages, 2 figures

点击查看摘要

Abstract:We introduce TRACE, a cross-domain engineering framework for trustworthy agentic AI in operationally critical domains. TRACE combines a four-layer reference architecture with an explicit classical-ML vs. LLM-validator split (L2a/L2b), a stateful orchestration-and-escalation policy (L3), and bounded human supervision (L4); a metrologically grounded trust-metric suite mapped to GUM/VIM/ISO 17025; and a Model-Parsimony principle quantified by the Computational Parsimony Ratio (CPR). Three instantiations–clinical decision support, industrial multi-domain operations, and a judicial AI assistant–transfer the samearchitecture and metrics across principally different governance contexts. The L2a/L2b separation makes the use of large language models a deliberate design decision rather than an architectural default, with parsimony quantified through CPR. TRACE introduces CPR as a first-class design principle in trustworthy-AI engineering.

[HC-4] A Workflow-Oriented Framework for Asynchronous Human-AI Collaboration in Hybrid and Compute-Intensive HPC Environments

【速读】：该论文旨在解决高风险国防与安全场景中，人工智能（AI）系统在高性能计算（HPC）环境中难以实现实时人机协同的问题。由于HPC环境的计算密集性和资源限制，传统实时交互方式不可行，导致人类专家无法及时介入关键决策环节。解决方案的关键在于提出一种异步人机协作工作流框架，能够在不中断底层计算任务的前提下，在预设检查点暂停流程以等待人工输入，从而避免资源闲置并支持非阻塞式监督。该框架兼容SLURM调度系统、容器化和原生任务，适用于需要人类判断与适应性的复杂场景，已在MareNostrum 5等系统上验证其在可移植性、效率和可控性方面的优势。

链接: https://arxiv.org/abs/2605.03743
作者: Sergio Mendoza,Cedric Bhihe,Natalia Zamora,David Modesto,Jose Martin Bugallo Batalla,Jesus Gomez Canovas,Rafel Palomo Avellaneda,Miguel Perez Espinosa
机构: Barcelona Supercomputing Center (巴塞罗那超级计算中心); NTT DATA (NTT数据)
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC); Software Engineering (cs.SE)
备注: Accepted manuscript. 14 pages, 6 figures, 1 table. Published in Proceedings of SPIE, volume 13679, Artificial Intelligence for Security and Defence Applications III

点击查看摘要

Abstract:Human involvement is critical in training and deploying AI systems in high-stakes defence and security contexts. However, real-time interaction is impractical in HPC environments due to compute intensity and resource constraints. We present a workflow framework that enables asynchronous human-AI collaboration across hybrid infrastructures, including HPC clusters, local machines, and cloud platforms. Workflows can pause at defined checkpoints for human input without halting underlying compute jobs, preventing idle resources and enabling non-blocking supervision. The framework supports interaction with SLURM-based scheduling, containerized and native tasks, and is customized for scenarios requiring human judgment and adaptability. We demonstrate its application in model training on systems like MareNostrum 5, highlighting benefits in portability, efficiency, and oversight in operational AI workflows.

[HC-5] Sorry for the late reply: Response times and reciprocity in WhatsApp and Instagram chats

【速读】：该论文旨在解决在线社交互动中“回应速度是否体现互惠性”这一关键问题，即强关系的聊天伙伴是否会相互匹配对方的响应时间。其解决方案的关键在于首次将回应速度（response time）作为互惠性的量化指标，通过分析来自889组匿名WhatsApp和Instagram聊天的340万条消息，发现约70%的WhatsApp与44%的Instagram消息在5分钟内得到回复，且双方响应速度高度相似（以Jensen-Shannon距离和回归斜率0.786–0.796衡量），并具有跨月稳定性。这一结果表明，回应速度的平衡可作为计算机中介交流中互惠性的客观标记，为研究社会纽带动态提供了新的定量方法。

链接: https://arxiv.org/abs/2605.03687
作者: Florian Martin,Olya Hakobyan,Hanna Drimalla
机构: Bielefeld University (比勒费尔德大学)
类目: ocial and Information Networks (cs.SI); Human-Computer Interaction (cs.HC)
备注: 21 pages (13 main text, 2 references, 6 appendix). Code publicly available at this https URL and data available at this https URL for qualified researchers

点击查看摘要

Abstract:Chat communication is often fast-paced, creating the expectation of quick replies. While the timing of exchanges is known to foster closeness and enjoyment, it remains largely unexplored whether chat partners with strong ties reciprocate each other’s response times. Using 3.4 million messages from 889 chats across 97 donations of anonymous WhatsApp and Instagram chats, we analyzed response times, their balance between chat partners, and its stability over time. To our knowledge, this is the first study to examine response speed as an expression of reciprocity, bridging a key aspect of online communication with a fundamental principle of social interactions. We found that around 70% of WhatsApp and 44% of Instagram messages were answered within five minutes, confirming the fast pace of instant messaging. Overall, the response speed between chat partners was similar. The response speed similarity was evident both in the overall response-time distributions of chat partners assessed with Jensen-Shannon distance and in the steep regression slopes (0.786 for WhatsApp and 0.796 for Instagram) linking one person’s probability of responding within five minutes to the partner’s corresponding probability. Importantly, the dispersion of response time similarity over months showed that this balance persists over time. Our results position response time balance as a marker of reciprocity in computer-mediated communication, offering a new way to quantitatively study this fundamental principle of social interaction. We suggest using response speed balance as a complementary metric in the analysis of relationship dynamics, such as the strengthening or weakening of social ties.

[HC-6] Jiao: Bridging Isolation and Customization in Mixed Criticality Robotics

【速读】：该论文旨在解决消费类机器人在共享多核平台上整合安全关键控制、感知流水线与用户应用时面临的挑战，尤其是因端用户缺乏系统知识而导致的“专业知识不对称”问题。解决方案的关键在于提出一个集成架构：通过硬件级的Safe IO Cell实现强制干预能力，利用Parameter Synchronization Service封装跨域复杂性，并借助Safety Communication Layer实现符合IEC 61508标准的验证机制，从而在ARM Cortex-A55平台上实现了显著的实时性能提升（周期抖动降低84.5%，尾部定时误差减少近一个数量级）。

链接: https://arxiv.org/abs/2605.03641
作者: James Yen,Zhibai Huang,Zhixiang Wei,Tinghao Yi,Shupeng Zeng,Liang Pang,Songtao Xue,Zhengwei Qi
机构: 未知
类目: Robotics (cs.RO); Human-Computer Interaction (cs.HC)
备注: Accepted by Infocom’26 Embodied Intelligence Networks workshop

点击查看摘要

Abstract:Consumer robotics demands consolidation of safety-critical control, perception pipelines, and user applications on shared multicore platforms. While static partitioning hypervisors provide hardware-enforced isolation, directly transplanting automotive architectures encounters an expertise asymmetry problem in which end-users modifying robot behavior lack the systems knowledge that platform developers possess. We present an architecture addressing this challenge through three integrated components. A Safe IO Cell provides hardware-level override capability. A Parameter Synchronization Service encapsulates cross-domain complexity. A Safety Communication Layer implements IEC~61508-aligned verification. Our empirical evaluation on an ARM Cortex-A55 platform demonstrates that partition isolation reduces cycle-period jitter by 84.5% and cuts tail timing error by nearly an order of magnitude (p99 | jitter | from 69.0, \mu s to 7.8, \mu s), eliminating all 50, \mu s~excursions.

[HC-7] he Frag ility of AI Companionship: Ontological Structural and Normative Uncertainty in Human-AI Relationships

【速读】：该论文旨在解决生成式 AI（Generative AI）伴侣关系中用户面临的三类不确定性问题：本体论不确定性（关于AI的本质与自主性）、结构性不确定性（源于平台控制与系统不稳定）以及规范性不确定性（涉及人机亲密关系的合法性边界）。这些问题由算法不透明、平台政策变动和社会污名等技术与社会因素共同塑造，常引发用户的挫败感、自我怀疑和心理困扰。解决方案的关键在于将不确定性视为一种社会技术现象，并通过设计干预措施加以缓解，包括增强情境透明度、赋予用户更多控制权、提供系统更新通知以及设置关系保护机制，从而促进更安全、可信赖的人机陪伴体验。

链接: https://arxiv.org/abs/2605.03367
作者: Renwen Zhang,Lezi Xie
机构: 未知
类目: Human-Computer Interaction (cs.HC)
备注:

点击查看摘要

Abstract:As generative AI chatbots become more personalized and emotionally responsive, they increasingly serve as companions, friends, and romantic partners. Yet these relationships are accompanied by significant uncertainty: users question the AI’s identity and agency, the authenticity of its emotional responses, and the stability of the relationship amid system updates, policy changes, or platform shutdowns. Drawing on in-depth interviews with 25 users of AI companions, this study identifies three forms of uncertainty: ontological uncertainty concerning the AI’s nature and agency, structural uncertainty arising from platform control and system instability, and normative uncertainty regarding the legitimacy and boundaries of human-AI intimacy. These uncertainties are shaped by technical and social factors, such as algorithmic opacity, platform changes, and social stigma, often inducing frustration, self-doubt, and distress. Participants managed these uncertainties through information seeking, topic avoidance, expectation adjustment, and disengagement. This study extends interpersonal uncertainty theories to human-AI communication and contributes to HCI research by conceptualizing uncertainty in AI companionship as a socio-technical phenomenon with potential socio-emotional harms. We discuss implications for designing safer AI companionship through contextual transparency, user control, update notice, and relational safeguards.

[HC-8] Cheap Expertise: Mapping and Challenging Industry Perspectives in the Expert Data Gig Economy

【速读】：该论文旨在探讨生成式 AI (Generative AI) 行业如何重构社会对“专家”概念的理解，特别是通过数据标注产业的公共话语揭示其对人类专家价值的重新定义。研究发现，该行业将AI专家能力视为“廉价”的（cheap），即在投资回报率上优于人类专家；同时将人类专家视为可被提取的资源，其价值需与AI专家相对比；此外，机构性专家（如高校和企业所拥有的知识）则被视作亟需“解放”或改革的对象，以便融入最新的AI系统中。解决方案的关键在于识别并批判这种“廉价专家”叙事背后的权力结构与价值重估机制，从而引发社会对AI驱动的专家零工经济及其潜在影响的深入反思与制度性回应。

链接: https://arxiv.org/abs/2605.03295
作者: Robert Wolfe,Aayushi Dangol
机构: Rutgers University (罗格斯大学); University of Washington (华盛顿大学)
类目: Computers and Society (cs.CY); Human-Computer Interaction (cs.HC)
备注: To appear at CHIWORK 2026

点击查看摘要

Abstract:Demand for expert-annotated data on the part of leading AI labs has created an expert gig economy with the potential to reshape white collar work and society’s understanding of expertise. In this research, we study the vision for the future of expertise described in the public communication of five industry data annotation organizations and their CEOs, as reflected on social media feeds and public appearances on podcasts. We find that the industry envisions AI expertise as cheap, meaning that it can offer a better return on investment than human expertise. Human expertise, meanwhile, is viewed as an extractable resource, the value of which can be judged relative to AI expertise. Finally, institutional expertise (such as that created or possessed by universities and corporations) is viewed as in need of liberation or reform, such that it can be incorporated into the latest artificial intelligence systems. Our findings have implications for human experts, whose professional lives may be transformed and revalued by this industry, as well as for societal institutions that mediate expertise. We close this work with a series of provocations intended to elicit consideration of how society can best approach an AI-driven expert gig economy and the cheap expertise it intends to produce.

[HC-9] Attention: What Prevents Young Adults from Speaking Up Against Cyberbullying in an LLM -Powered Social Media Simulation

【速读】：该论文旨在解决年轻成人（Young Adult, YA）旁观者在面对网络欺凌时难以公开发声的问题，这一问题常因复杂的多方社交动态而被抑制。解决方案的关键在于设计并实现一个基于大语言模型（Large Language Models, LLMs）的多AI代理社交媒体模拟系统——Upstanders’ Practicum，通过三轮迭代优化的实践训练，促使参与者经历三个关键的认知转变：从忽视到真正关注、从自我中心转向关注受害者、从私人调解意图转向公共规范建构意识。只有完成这些注意力和认知框架的转变后，模拟练习才真正有效，使参与者自发产生公开干预的动力，并在无直接指导的情况下逐步形成得体的公共表达策略。该研究揭示了超越传统社交技能教学的新方向：聚焦于培养真实关注力、塑造积极的“挺身而出者”身份认同，以及将旁观者行动视为线上公共规范设定的过程。此外，作者开源了该系统底层的Truman Agents平台，为未来网络欺凌与社交媒体行为研究提供可复用的技术基础。

链接: https://arxiv.org/abs/2605.03287
作者: Qian Yang,Jessie Jia,Elaine Tsai,Amy Li,Nader Akoury,Natalie N. Bazarova
机构: Cornell University (康奈尔大学)
类目: Human-Computer Interaction (cs.HC); Computers and Society (cs.CY)
备注:

点击查看摘要

Abstract:Interactive, multi-agent social simulation systems have shown promise for helping users practice navigating various complex social situations across domains. This paper asks: To what extent can such systems help young adult (YA) bystanders speak up publicly against cyberbullying, a task often thwarted by complex, multi-party social dynamics? We created Upstanders’ Practicum, a multi-AI-agent social media simulation powered by Large Language Models (LLMs), as a probe and observed 34 YAs freely practicing public bystander intervention across three iteratively refined versions. We found that practicing public bystander intervention in the simulation was helpful, but after participants made three attention shifts: (1) from inattention to paying true attention, (2) from self-focus ("I don’t usually do this’‘) to attending to those directly involved, and (3) from resolving the private conflict between bully and victim ("maybe I could set up the meeting between them’‘) to addressing the broader audience online (“public comment is about norm-setting”). Only after these shifts did practice in the simulation start to help: participants then saw a reason to speak up publicly and, through continued practice, crafted tactful public messages without explicit instruction. These findings illuminate new design and research opportunities for bystander education beyond social skill instruction, namely, designing for true attention, for fostering a vocal upstander identity, and for seeing bystander intervention as public norm setting. In addition, we open-source Truman Agents (this http URL), the first-of-its-kind multi-LLM-agent social media simulation platform that Upstanders’ Practicum builds upon, for future cyberbullying and social media research.

[HC-10] Can AI Help You Get Over Your Breakup? One Session with a Belief-Reframing Chatbot Shows Sustained Distress Reduction

【速读】：该论文旨在解决浪漫关系破裂后个体所经历的心理痛苦问题，这是一种常见且强烈的应激源。其解决方案的关键在于开发并测试一款名为 overit 的单次会话式人工智能（AI）聊天机器人，该聊天机器人基于认知重评（cognitive reappraisal）策略，并融合记忆再巩固理论（memory reconsolidation theory），通过引导用户重构对分手事件的情绪记忆来减轻心理痛苦。研究结果显示，在随机对照试验中，接受 overit 干预的参与者在7天内表现出显著的痛苦缓解，效应量为 d = -0.70，且在1个月随访中仍存在较小但显著的治疗优势，提示该AI干预具有短期有效性与潜在的持续效应。

链接: https://arxiv.org/abs/2605.03261
作者: Thomas Menzel,Michel Schimpf,Thomas Bohné
机构: Technical University of Munich, Germany; University of Cambridge, United Kingdom
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注: T. Menzel and M. Schimpf contributed equally to this work and are listed alphabetically

点击查看摘要

Abstract:Romantic breakups are among the most common and intense sources of psychological distress. We evaluated overit, a single-session AI chatbot that uses cognitive reappraisal to address breakup distress, informed by memory reconsolidation theory. In a pre-registered randomized controlled trial, 254 adults in the United States and United Kingdom who had experienced a romantic breakup were assigned to either an initial survey assessment followed by an AI chat session or to a survey-only control. Breakup distress was measured at baseline, 7 days, and again at an exploratory 1-month follow-up using the Breakup Distress Scale. Participants assigned to overit showed a significantly greater reduction in breakup distress than controls at 7 days (time-by-condition interaction B = -5.36, SE = 1.19, p .001; completer-based d = -0.70). A smaller but still significant treatment advantage remained detectable at the exploratory 1-month follow-up among post-session completers (B = -2.92, SE = 1.22, p = .017). Exploratory post hoc moderation suggested a larger effect among male participants (B = 7.78, p = .003). These results suggest that a brief AI chatbot conversation can meaningfully reduce breakup distress, with exploratory evidence that a smaller advantage persists over the following month. Future work should test the intervention against active controls, evaluate repeated-session use, and recruit more diverse samples.

[HC-11] ADAPTS: Agent ic Decomposition for Automated Protocol-agnostic Tracking of Symptoms

【速读】：该论文旨在解决从非结构化临床交互中建模潜在临床构念（latent clinical constructs）的挑战，特别是如何在不依赖特定评估协议的前提下自动量化抑郁和焦虑的严重程度。其核心解决方案是提出ADAPTS（Agentic Decomposition for Automated Protocol-agnostic Tracking of Symptoms）框架，采用多代理大语言模型（mixture-of-agents LLM）架构，将长篇临床访谈分解为症状特异性的推理任务，从而生成可审计的解释并保持时间与发言者对齐。该方法在两个独立数据集上验证了泛化能力，在高差异访谈中自动化评分比原始人工评分更接近专家基准（绝对误差分别为22 vs. 26），且引入扩展协议后评分一致性显著提升（ICC(2,1) = 0.877），表明该框架可在资源有限环境中实现客观、可扩展的精神病学评估。

链接: https://arxiv.org/abs/2605.03212
作者: Alexandria K. Vail,Marcelo Cicconet,Katie Aafjes-van Doorn,Ryan Maroney,Marc Aafjes
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Human-Computer Interaction (cs.HC); Applications (stat.AP); Computation (stat.CO)
备注:

点击查看摘要

Abstract:Modeling latent clinical constructs from unconstrained clinical interactions is a unique challenge in affective computing. We present ADAPTS (Agentic Decomposition for Automated Protocol-agnostic Tracking of Symptoms), a framework for automated rating of depression and anxiety severity using a mixture-of-agents LLM architecture. This approach decomposes long-form clinical interviews into symptom-specific reasoning tasks, producing auditable justifications while preserving temporal and speaker alignment. Generalization was evaluated across two independent datasets ( N=204 ) with distinct interview structures. On high-discrepancy interviews, automated ratings approximated expert benchmarks ( \textabsolute error=22 ) more closely than original human ratings ( \textabsolute error=26 ). Implementing an ``extended’’ protocol that incorporates qualitative clinical conventions significantly stabilized ratings, with absolute agreement reaching \textICC(2,1) = 0.877 . These findings suggest that the ADAPTS framework enables promising evaluations of psychiatric severity. While the current implementation is purely text-based, the underlying architecture is readily extensible to multimodal inputs, including acoustic and visual features. By approximating expert-level precision in a protocol-agnostic manner, this framework provides a foundation for objective and scalable psychiatric assessment, especially in resource-limited settings.

[HC-12] Wheres the Team Spirit? An Exploratory Study on Team Development Through Co-located Tablet-Based VR

【速读】：该论文旨在解决如何通过叙事驱动的非对称虚拟现实（VR）体验促进团队协作相关知识、技能与态度（KSAs）的发展问题，如沟通、协调、信任和反思性。其解决方案的关键在于设计一种基于平板设备的VR训练体验，该体验通过空间分离、工具不对称性和依赖任务结构来强制用户进行口头协调，从而激发团队成员在动态环境中运用并发展KSAs。研究采用与人力资源专家访谈所得的框架指导设计，并通过共处式用户实验（N=16）验证了参与者在连续协作场景中通过言语交流、角色协商和共享表征实现有效协调的能力，证明了该方法在沉浸式团队训练中的有效性。

链接: https://arxiv.org/abs/2605.03127
作者: Irina Paraschivoiu,Thomas Layer-Wagner,Klaus Neundlinger,Simone Rack,Markus Tatzgern
机构: Paris Lodron University of Salzburg (萨尔茨堡大学); Polycular (波利库拉); In Scope (在范围); Salzburg University of Applied Sciences (萨尔茨堡应用科学大学)
类目: Human-Computer Interaction (cs.HC)
备注:

点击查看摘要

Abstract:We explore how narrative-driven asymmetric VR experiences can support the development of teamwork-related knowledge, skills, and attitudes (KSAs), such as communication, coordination, trust, and reflexivity. We present the design and evaluation of a tablet-based VR training experience structured around spatial separation, tool asymmetry, and interdependent tasks that require verbal coordination. The experience was designed based on interviews with HR professionals and mapped to a framework of established KSAs. We conducted a co-located user study (N=16) that involved two consecutive collaborative scenarios. Our findings show that users adapted dynamically using verbal exchange, role negotiation, and shared representations to coordinate under asymmetric conditions. We also observed active application of teamwork KSAs. Based on our insights, we present design recommendations for creating effective immersive team training interventions.

[HC-13] Making the Invisible Visible: Understanding the Mismatch Between Organizational Goals and Worker Experiences in AI Adoption

【速读】：该论文试图解决的问题是：在组织中引入人工智能（Artificial Intelligence, AI）系统时，由于忽视了实际使用者——员工——的参与和需求，导致AI应用难以成功落地，表现为员工抵制、难以融入工作流程等现象。解决方案的关键在于：将员工视为AI集成的核心参与者，通过在个体、任务和组织三个层面制定适应性策略，以弥合组织对AI的期望与员工实际工作需求之间的差距，从而提升AI系统的可用性、可接受度和有效性。

链接: https://arxiv.org/abs/2605.03078
作者: Christine P. Lee,Min Kyung Lee,Bilge Mutlu
机构: 未知
类目: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注:

点击查看摘要

Abstract:While AI is often introduced into organizations to drive innovation and efficiency, many adoption efforts fail as workers resist and struggle to integrate these systems. These failures point to a deeper issue: workers, the very people expected to collaborate with AI, are often invisible in decisions about how AI is designed and used. Drawing on interviews with professionals who interact with AI systems daily in healthcare, finance, and management, we examine the disconnect between organizational expectations and worker experiences. We identify key barriers, including poor usability and interoperability, misaligned expectations, limited control, and insufficient communication. These challenges highlight a gap between how organizations implement AI and the evolving worker needs, tasks, and workflows that it fails to support. We argue that successful adoption requires recognizing workers as central to AI integration and propose adaptation strategies at the individual, task, and organizational levels to better align AI systems with real-world practices.

[HC-14] Mixed-Precision Information Bottlenecks for On-Device Trait-State Disentanglement in Bipolar Agitation Detection

【速读】：该论文旨在解决在资源受限的边缘设备上，通过语音生物标志物对双相情感障碍躁动状态进行连续监测时，如何有效分离稳定的说话者特征（speaker traits）与易变的情绪状态（affective states）的问题。解决方案的关键在于提出MP-IB框架，首次将混合精度量化（mixed-precision quantization）视为一种信息瓶颈机制来实现临床特征与状态的解耦：高精度浮点数（FP16）用于编码说话者身份（1,024 bit），低精度整数（INT4）用于捕捉躁动状态（128 bit），形成8倍的信息不对称性，无需对抗训练即可实现高效分离；同时结合动态精度调度和多尺度时间融合策略，在保证性能的同时显著降低计算开销，最终实现在低成本硬件上的实时部署。

链接: https://arxiv.org/abs/2605.03039
作者: Joydeep Chandra
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC); Sound (cs.SD)
备注:

点击查看摘要

Abstract:Continuous monitoring of bipolar disorder agitation via voice biomarkers requires disentangling stable speaker traits from volatile affective states on resource-constrained edge devices. We introduce MP-IB, the first framework to treat mixed-precision quantization as an information bottleneck for clinical trait-state separation. The core insight is that numerical precision itself controls capacity: an FP16 trait head (1,024 bits) encodes speaker identity, while an INT4 state head (128 bits) captures agitation, yielding 8x information asymmetry without adversarial training. We augment this with Dynamic Precision Scheduling and Multi-Scale Temporal Fusion. On Bridge2AI-Voice (N=833, 4 sessions/participant, strict speaker-independent CV), MP-IB achieves rho = 0.117 (95% CI: [0.089, 0.145], p=0.003 vs. chance), outperforming 94M-parameter WavLM-Adapter with in-domain SSL continuation (rho = -0.042), beta VAE disentanglement (rho = 0.089), and hand-crafted prosody (rho = 0.031) by 2.8–15.9 points absolute. Zero-shot transfer to CREMA-D achieves AUC=0.817. Identity leakage is suppressed to near-random (EER=0.42, MIA-AUC=0.52). End-to-end latency is 23.4 ms with a 617 KB footprint, enabling real-time monitoring on sub 20 dollar devices.

[HC-15] From Informal Addresses to Reliable Places: Participatory Data Governance of Civic Addressing in Puerto Rico ICIP

【速读】：该论文旨在解决在缺乏正式公民地址（civic address）的地区，如何通过参与式数据治理（participatory data governance）实现服务支持的问题。其解决方案的关键在于引入“可靠地点”（Reliable Places）作为过渡性治理工具，这些地点通过实际使用过程逐步建立可靠性，从而在正式公民地址尚未确立的情况下，为服务提供可操作的地理定位信息，并同时构建通向正式地址分配的路径。

链接: https://arxiv.org/abs/2605.02924
作者: Juan A. Padilla
机构: 未知
类目: Human-Computer Interaction (cs.HC); Computers and Society (cs.CY)
备注: This paper is a preprint of a workshop paper accepted at the CHI 2026 Workshop on Participatory Data Governance at the CHI 2026 Conference on Human Factors in Computing Systems

点击查看摘要

Abstract:This paper examines civic addressing as a problem of participatory data governance. Drawing on a project developed through the U.S. Census Bureau’s The Opportunity Project with engagement from FEMA, we describe the use of actionable geolocations to support services where formal addresses are absent. We introduce Reliable Places as transitional governance artifacts through which place reliability emerges via use, enabling services while supporting pathways toward formal civic address assignment.

[HC-16] A User-Centric Analysis of Explainability in AI-Based Medical Image Diagnosis ALT

【速读】：该论文旨在解决当前医疗领域中人工智能（Artificial Intelligence, AI）系统在临床实践中应用受限的问题，即尽管AI在医学图像诊断任务中表现优于人类，但其决策过程缺乏透明性和可解释性，导致医生难以信任和采纳。解决方案的关键在于通过用户中心的对比分析，评估最新文本、视觉及多模态可解释人工智能（Explainable Artificial Intelligence, XAI）方法在实际临床场景中的有效性，并发现结合边界框（bounding box）与诊断报告的XAI形式在理解性、完整性、速度和适用性等方面优于其他方法；同时研究还揭示了错误AI诊断可能误导医生判断的风险，强调了高质量可解释性对于提升AI可信度的重要性。

链接: https://arxiv.org/abs/2605.02903
作者: Julia Wagner,Tim Schlippe
机构: 未知
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: The 4th International Workshop on eXplainable Artificial Intelligence in Healthcare, Pavia, Italy, 26 June 2025

点击查看摘要

Abstract:In recent years, AI systems in the medical domain have advanced significantly. However, despite outperforming humans, they are rarely used in practice since it is often not clear how they make their decisions. Optimal explanation and visualization of the decision process are often lacking. Therefore, we conducted a comparative user-centric analysis of the latest state-of-the-art textual, visual and multimodal explainable artificial intelligence (XAI) methods for medical image diagnosis. Our survey of 33 physicians showed that 88% agree that it is important that AI explains the diagnosis – 64% even strongly agree. A combination of bounding box and report is rated better than the other tested XAI methods in the evaluated aspects understandability, completeness, speed, and applicability. We even tested the potential negative impact of false AI-based medical image diagnoses and found that 50% of the participants trusted false AI diagnoses over all tested XAI methods.

[HC-17] From Passive Feeds to Guided Discovery: AI-Initiated Interaction for Vague Intent in Content Exploration

【速读】：该论文旨在解决用户在推荐流（recommendation feed）中遇到的“模糊意图”（vague intent）问题，即当用户感知到当前内容重复但无法清晰表达替代需求时，传统推荐系统与搜索系统均难以有效支持其探索行为。解决方案的关键在于提出Red-Rec——一种由AI主动引导的探索式交互界面：系统在用户浏览一段时间后自动总结当前feed中的内容模式（如主导类别和潜在兴趣），提供可点击的探索选项，并仅通过最多一次追问获取反馈，随后逐步融合新内容至原feed中。该设计基于定性研究发现用户常识别出feed停滞但难于明确表达替代偏好，从而强调低努力、主动式的AI干预，实验证明其相比用户主动发起的聊天接口能显著提升探索广度、愉悦感（serendipity）并降低交互负担。

链接: https://arxiv.org/abs/2605.02902
作者: Yu Xie,Ying Qi
机构: Xiaohongshu(小红书)
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Recommendation feeds work well when people are simply browsing, and search works well when they can formulate a query. Between these two cases is a common but poorly supported state: users feel that their feed has become repetitive, yet cannot clearly specify what they want instead. We refer to this state as vague intent. We present Red-Rec, an AI-supported exploration interface for this middle ground. After a period of browsing, the system summarizes patterns in the current feed (e.g., dominant content categories and possible latent interests), offers clickable exploration options, asks at most one follow-up question, and then gradually blends new content into the feed. The design is motivated by a formative study which found that users often recognize feed staleness but struggle to articulate alternatives, suggesting the need for proactive and low-effort this http URL evaluated Red-Rec in a mixed-design lab study against three comparison conditions: a passive feed, search, and a user-initiated chat interface. Compared with user-initiated chat, Red-Rec led to broader exploration, higher serendipity ratings, and lower interaction effort. Participants in the AI-initiated condition typed very little , relying mainly on option selection, whereas participants in the user-initiated chat condition typed substantially more . We discuss how proactive, option-based AI support can help users move beyond repetitive feeds without undermining their sense of control, and we outline design implications for recommendation interfaces that support open-ended exploration.

[HC-18] owards an End-to-End System for 3D Tracking of Physical Objects in Virtual Immersive Environments

【速读】：该论文旨在解决虚拟现实（VR）应用中对小型物理对象进行实时位置跟踪的问题，尤其针对需要将物理对象在虚拟空间中精准映射的训练场景。传统方法依赖复杂的追踪设备或手动实现跟踪，难以满足“即插即用”的便捷性需求。解决方案的关键在于提出一种基于特征标记（fiducial markers）的端到端系统，结合软件框架，实现快速的对象标识与数据流传输；该系统采用AruCo、AprilTag及自研彩色控制点（Colored Control Points）三种标记方案，支持高效检测与位置信息提取，从而为VR和扩展现实（XR）环境提供可靠、低延迟的物理-虚拟位置映射能力。

链接: https://arxiv.org/abs/2605.02901
作者: Stanisław Knapiński,Maciej Grzeszczuk,Barbara Karpowicz,Pavlo Zinevych,Wieslaw Kopec
机构: 未知
类目: Human-Computer Interaction (cs.HC); Computer Vision and Pattern Recognition (cs.CV)
备注: 12 pages, 10 figures, 1 table. Reviewed and presented on MIDI Conference 2025, Warsaw, proceedings publication in progress

点击查看摘要

Abstract:This work aims to establish an end-to-end system for tracking of physical 3D objects for virtual reality (VR) applications. We focus on training applications requiring real-time tracking of the position of small physical objects and their reflection in VR space. Out goal is to perform object tracking in a “plug and play” manner, without using complex systems with quite large tracking devices or manually implementing object tracking. We therefore propose a system for object tracking via fiducial markers alongside a software harness, to enable fast and efficient designation of objects to be tracked and data streaming solution for end-use applications. The system utilizes AruCo, AprilTag and an original Colored Control Points based fiducial system. It allows for easy tag detection and use of object position data, which are crucial for immersive training environments based on VR and eXtended Reality (XR). We evaluate various tag sizes, detection distances, and different camera devices against the theoretical limits. In effect, we create a complete solution for implementing marker-based, real-to-virtual object position mapping for various applications.

[HC-19] A Study of Consumers Cognitive Load in eCommerce Websites using Eye-tracking Technology

【速读】：该论文旨在解决电子商务网站视觉复杂性对用户认知负荷及购物决策影响的问题，特别是价格差异如何加剧网页复杂性并进而影响用户体验。其解决方案的关键在于通过眼动追踪技术（eye-tracking technology）对48名受试者在知名电商平台上的浏览行为进行实证分析，量化认知负荷指标（包括注视次数、扫视次数、注视持续时间与任务完成时间），从而揭示价格区间变化所引发的网页复杂性差异及其与消费者感知之间的强关联。这一方法为优化电商界面设计提供了可量化的依据，有助于开发者和业务分析师提升用户的在线购物体验。

链接: https://arxiv.org/abs/2605.02899
作者: Shojibur Rahman,Ahmed Alif Swopno,Nayeem Ahmed,Ashik Ahmed Fahim,Tabin Hasan
机构: 未知
类目: Human-Computer Interaction (cs.HC)
备注:

点击查看摘要

Abstract:The aesthetics of e-commerce websites have a big influence on purchasing decisions and customers’ satisfaction. Webpage complexity and high cognitive load are responsible for causing an unpleasant experience while shopping online. This research empirically inspects a correlation between users’ cognitive load and product pricing, where price plays a vital role in causing web complexity. Therefore, we have experimented on 48 random individuals using eye-tracking technology to observe the eye movement calibration on some reputed e-commerce websites. We measured the cognitive load extracted from users’ datasets by analyzing fixation count, saccades, fixation duration, and task completion time. Our study induces new findings on website complexity which varies on the similar product but different price ranges. This research also demonstrates a strong connection between customer perception and visual complexity while making online purchases. In addition, these findings will assist the developers and business analysts to improve consumers’ shopping experience in e-commerce websites.

[HC-20] What Shapes Participant Data Quality? A Scoping Review and Case Study of Crowdsourced Webcam Eye Tracking in AI Interviews

【速读】：该论文旨在解决众包式基于网络摄像头的眼动追踪（webcam-based eye tracking）在非受控环境和硬件多样性背景下数据质量不一致的问题，这限制了其在人机交互（HCI）与行为科学中的可靠性与可重复性。解决方案的关键在于通过实证分析识别出显著影响数据质量的行为与技术因素：具体而言，在RealEye平台中，更高的注视点数量、更短的实验时长以及操作系统的选择是预测高质量眼动数据的关键变量。研究进一步提出基于有序逻辑回归（Ordered Logistic Regression, OLR）的量化模型，为提升众包眼动数据的质量控制与标准化提供可操作的改进路径。

链接: https://arxiv.org/abs/2605.02898
作者: Ka Hei Carrie Lau,Enkelejda Kasneci
机构: Technical University of Munich (慕尼黑工业大学); Munich Center for Machine Learning (慕尼黑机器学习中心)
类目: Human-Computer Interaction (cs.HC)
备注:

点击查看摘要

Abstract:Webcam-based eye tracking is a cost-effective, scalable method for remote research that effectively reaches broader populations. However, uncontrolled environments and hardware diversity lead to inconsistent data quality in crowdsourcing. To assess current practices, we conducted a scoping review of crowdsourced eye-tracking from 2011-2025. The review confirms fragmented reporting and a lack of established quality benchmarks. To address this lack of predictive insight, we conducted a case study on AI fairness interviews (N=205) using the RealEye platform. Applying Ordered Logistic Regression (OLR) to the platform quality metric, we found that behavioral and technical factors significantly predict data quality. Specifically, within the RealEye platform, higher fixation counts, shorter sessions, and operating system choice yield significantly higher quality grades. Based on this review and platform-specific predictive insights, we provide actionable recommendations to enhance the reliability, transparency, and replicability of future crowdsourced webcam eye tracking in HCI and behavioral science.

[HC-21] Same Voice Different Lab: On the Homogenization of Frontier LLM Personalities ACL2026

【速读】：该论文旨在解决大语言模型（Large Language Models, LLMs）助手人格特征对用户体验和响应质量影响的问题，尤其是不同模型在人格表达上的差异及其潜在的共性。其解决方案的关键在于通过基于ELO评分的外部特质打分体系，在144个维度上大规模评估前沿LLM的人格表现，发现尽管训练方法多样，所有模型均趋向于表现出系统化、条理性和分析性的特质，并抑制如懊悔或奉承等情感化特质；同时，模型在中位分布特质（如诗意或幽默）上虽有分化，但整体仍趋于中性，表明存在一种隐含的最优助手行为标准正在自发形成。这一发现揭示了模型开发者之间可能存在的默许共识，突出了角色训练（character training）在塑造统一AI助手人格方面的重要性。

链接: https://arxiv.org/abs/2605.02897
作者: Avinash Krishna,Kalyana Chadalavada,Unso Eun Seo Jo
机构: Anthropic; Cornell University
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注: Submitted to ACL 2026. 7 Pages, 8 figures

点击查看摘要

Abstract:LLM assistant personalities play a critical role in user experience and perceived response quality. We present a large-scale experiment of frontier LLM personalities using external ELO-based traits scoring across 144 traits. We find that all models tested converge on a form of trait expression that is systematic, methodical, and analytical and suppress traits such as remorseful and sycophantic. Moreover, models tend to diverge more in their expression of middle-of-distribution traits such as poetic or playful, but even these so-called creative models tend to have more neutral identities. These similarities suggest an implicit emergence of a standard of optimal assistant behavior. In a landscape of varied training methods, character training, therefore, stands out for its uniformity, offering insight into a tacit consensus between model developers.

计算机视觉

[CV-0] Audio-Visual Intelligence in Large Foundation Models

【速读】：该论文旨在解决当前音频-视觉智能（Audio-Visual Intelligence, AVI）领域研究碎片化的问题，包括任务多样性、分类体系不一致以及评估方法异构，从而阻碍了系统性比较与知识整合。其解决方案的关键在于构建一个统一的分类框架，涵盖从理解（如语音识别、声音定位）到生成（如音驱动视频合成、视频转音频）再到交互（如对话、具身或代理接口）的完整AVI任务谱系，并系统梳理方法论基础，包括模态标记化、跨模态融合、自回归与扩散生成、大规模预训练、指令对齐及偏好优化等核心技术，同时整理代表性数据集、基准测试和评估指标，为未来大模型时代的AVI研究提供结构化参考与发展方向。

链接: https://arxiv.org/abs/2605.04045
作者: You Qin,Kai Liu,Shengqiong Wu,Kai Wang,Shijian Deng,Yapeng Tian,Junbin Xiao,Yazhou Xing,Yinghao Ma,Bobo Li,Roger Zimmermann,Lei Cui,Furu Wei,Jiebo Luo,Hao Fei
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 56 pages, 16 figures, 24 tables, this https URL

点击查看摘要

Abstract:Audio-Visual Intelligence (AVI) has emerged as a central frontier in artificial intelligence, bridging auditory and visual modalities to enable machines that can perceive, generate, and interact in the multimodal real world. In the era of large foundation models, joint modeling of audio and vision has become increasingly crucial, i.e., not only for understanding but also for controllable generation and reasoning across dynamic, temporally grounded signals. Recent advances, such as Meta MovieGen and Google Veo-3, highlight the growing industrial and academic focus on unified audio-vision architectures that learn from massive multimodal data. However, despite rapid progress, the literature remains fragmented, spanning diverse tasks, inconsistent taxonomies, and heterogeneous evaluation practices that impede systematic comparison and knowledge integration. This survey provides the first comprehensive review of AVI through the lens of large foundation models. We establish a unified taxonomy covering the broad landscape of AVI tasks, ranging from understanding (e.g., speech recognition, sound localization) to generation (e.g., audio-driven video synthesis, video-to-audio) and interaction (e.g., dialogue, embodied, or agentic interfaces). We synthesize methodological foundations, including modality tokenization, cross-modal fusion, autoregressive and diffusion-based generation, large-scale pretraining, instruction alignment, and preference optimization. Furthermore, we curate representative datasets, benchmarks, and evaluation metrics, offering a structured comparison across task families and identifying open challenges in synchronization, spatial reasoning, controllability, and safety. By consolidating this rapidly expanding field into a coherent framework, this survey aims to serve as a foundational reference for future research on large-scale AVI.

[CV-1] UniCorrn: Unified Correspondence Transformer Across 2D and 3D CVPR2026

【速读】：该论文旨在解决跨模态几何匹配任务中模型碎片化的问题，即现有方法针对图像到图像（2D-2D）、图像到点云（2D-3D）和点云到点云（3D-3D）等不同模态组合分别设计专用模型，缺乏统一性与共享参数机制。解决方案的关键在于提出首个具有共享权重的统一对应模型 UniCorrn，其核心创新是利用 Transformer 注意力机制天然捕捉跨模态特征相似性，并设计双流解码器结构以分离处理外观特征和位置特征流，从而支持异构模态间的查询式端到端对应估计。该架构通过模态特定骨干网络结合共享编码器与解码器，在融合深度图生成伪点云与真实3D标注数据上联合训练，显著提升了多模态几何匹配性能。

链接: https://arxiv.org/abs/2605.04044
作者: Prajnan Goswami,Tianye Ding,Feng Liu,Huaizu Jiang
机构: Northeastern University (东北大学); Adobe Research (Adobe 研究院)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: CVPR 2026, 20 pages

点击查看摘要

Abstract:Visual correspondence across image-to-image (2D-2D), image-to-point cloud (2D-3D), and point cloud-to-point cloud (3D-3D) geometric matching forms the foundation for numerous 3D vision tasks. Despite sharing a similar problem structure, current methods use task-specific designs with separate models for each modality combination. We present UniCorrn, the first correspondence model with shared weights that unifies geometric matching across all three tasks. Our key insight is that Transformer attention naturally captures cross-modal feature similarity. We propose a dual-stream decoder that maintains separate appearance and positional feature streams. This design enables end-to-end learning through stack-able layers while supporting flexible query-based correspondence estimation across heterogeneous modalities. Our architecture employs modality-specific backbones followed by shared encoder and decoder components, trained jointly on diverse data combining pseudo point clouds from depth maps with real 3D correspondence annotations. UniCorrn achieves competitive performance on 2D-2D matching and surpasses prior state-of-the-art by 8% on 7Scenes (2D-3D) and 10% on 3DLoMatch (3D-3D) in registration recall. Project website: this https URL

[CV-2] Large Language Models are Universal Reason ers for Visual Generation

【速读】：该论文旨在解决生成式 AI（Generative AI）中“理解-生成差距”（understanding-generation gap）的问题，即尽管大型语言模型（Large Language Model, LLM）在图像内容验证上表现出高准确性，但在根据复杂文本提示生成图像时仍难以忠实对齐语义细节。解决方案的关键在于提出 UniReasoner 框架，该框架利用 LLM 作为通用推理器（universal reasoner），通过三阶段机制实现从理解到生成的闭环引导：首先由 LLM 生成离散视觉 token 构成粗略视觉草图（coarse visual draft），接着进行自评（self-critique）以生成基于场景的文本评估，指出需修正的内容；最后，扩散模型联合条件于原始提示、视觉草图和评估结果，使生成过程受到显式纠正信号的引导。此设计使得草图提供结构锚点缓解纯文本条件下的信息不足，而评估则将验证能力转化为可执行的约束，从而有效减少遗漏、幻觉及关系错误，显著提升组合对齐性和语义忠实度，同时保持图像质量。

链接: https://arxiv.org/abs/2605.04040
作者: Sucheng Ren,Chen Chen,Zhenbang Wang,Liangchen Song,Xiangxin Zhu,Alan Yuille,Liang-Chieh Chen,Jiasen Lu
机构: Johns Hopkins University (约翰霍普金斯大学); Apple (苹果公司)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Text-to-image generation has advanced rapidly with diffusion models, progressing from CLIP and T5 conditioning to unified systems where a single LLM backbone handles both visual understanding and generation. Despite the architectural unification, these systems frequently fail to faithfully align complex prompts during synthesis, even though they remain highly accurate at verifying whether an image satisfies those same prompts. We formalize this as the \emphunderstanding-generation gap and propose UniReasoner, a framework that leverages the LLM as a universal reasoner to convert its understanding strength into direct generation guidance. Given a prompt, the LLM first produces a coarse visual draft composed of discrete vision tokens. It then performs a self-critique by evaluating the draft for prompt consistency, producing a grounded textual evaluation that pinpoints what needs to be corrected. Finally, a diffusion model is conditioned jointly on the prompt, the visual draft, and the evaluation, ensuring that generation is guided by explicit corrective signals. Each signal addresses a limitation of the other: the draft provides a concrete, scene-level anchor that reduces under-specification in text-only conditioning, while the evaluation turns verification into grounded, actionable constraints that correct omissions, hallucinations, and relational errors. Experiments show that UniReasoner improves compositional alignment and semantic faithfulness under the same diffusion backbone while maintaining image quality, demonstrating a practical way to exploit LLM reasoning to close the understanding-generation gap.

[CV-3] Large-Scale High-Quality 3D Gaussian Head Reconstruction from Multi-View Captures

【速读】：该论文旨在解决从大规模多相机设置中高效重建高质量3D人脸的问题，特别是如何在保持高保真度的同时扩展模型对大量输入图像和不同身份的适应能力。解决方案的关键在于提出一种可扩展的前馈方法HeadsUp，其核心是采用高效的编码器-解码器架构，将多视角图像压缩为紧凑的潜在表示（latent representation），再将其解码为基于UV参数化的3D高斯（3D Gaussians），这些高斯锚定于一个中性头模板上。该UV表示形式实现了3D高斯数量与输入图像数量及分辨率的解耦，从而支持使用大量高分辨率图像进行训练，并显著提升重建质量和泛化能力，无需测试时优化即可适用于新身份。

链接: https://arxiv.org/abs/2605.04035
作者: Evangelos Ntavelis,Sean Wu,Mohamad Shahbazi,Fabio Maninchedda,Dmitry Kostiaev,Artem Sevastopolsky,Vittorio Megaro,Trevor Phillips,Alejandro Blumentals,Shridhar Ravikumar,Mehak Gupta,Reinhard Knothe,Jeronimo Bayer,Matthias Vestner,Simon Schaefer,Thomas Etterlin,Christian Zimmermann,Mathias Deschler,Peter Kaufmann,Stefan Brugger,Sebastian Martin,Brian Amberg,Tom Runia
机构: Apple(苹果公司)
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Project page: this https URL

点击查看摘要

Abstract:We propose HeadsUp, a scalable feed-forward method for reconstructing high-quality 3D Gaussian heads from large-scale multi-camera setups. Our method employs an efficient encoder-decoder architecture that compresses input views into a compact latent representation. This latent representation is then decoded into a set of UV-parameterized 3D Gaussians anchored to a neutral head template. This UV representation decouples the number of 3D Gaussians from the number and resolution of input images, enabling training with many high-resolution input views. We train and evaluate our model on an internal dataset with more than 10,000 subjects, which is an order of magnitude larger than existing multi-view human head datasets. HeadsUp achieves state-of-the-art reconstruction quality and generalizes to novel identities without test-time optimization. We extensively analyze the scaling behavior of our model across identities, views, and model capacity, revealing practical insights for quality-compute trade-offs. Finally, we highlight the strength of our latent space by showcasing two downstream applications: generating novel 3D identities and animating the 3D heads with expression blendshapes.

[CV-4] Enhanced 3D Brain Tumor Segmentation Using Assorted Precision Training

【速读】：该论文旨在解决脑肿瘤早期识别的问题，以提高患者的生存率。其关键解决方案是采用SegResNet架构进行三维图像分割，并结合自动多精度训练方法与Dice损失函数和Dice评分指标进行模型优化与评估，最终在肿瘤核心、整个肿瘤及增强肿瘤区域分别取得了0.84、0.90和0.79的Dice分数，表明该方法在脑肿瘤分割任务中具有较高的准确性与鲁棒性。

链接: https://arxiv.org/abs/2605.04008
作者: Adwaitt Pandya,Ozioma C. Oguine,Harita Bhargava,Shrikant Zade
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 5 pages, 5 figures, 1 table

点击查看摘要

Abstract:A brain tumor is a medical disorder faced by individuals of all demographics. Medically, it is described as the spread of non-essential cells close to or throughout the brain. Symptoms of this ailment include headaches, seizures, and sensory changes. This research explores two main categories of brain tumors: benign and malignant. Benign spreads steadily, and malignant expresses growth, making it dangerous. Early identification of brain tumors is a crucial factor for the survival of patients. This research provides a state-of-the-art approach to the early identification of tumors within the brain. We implemented the SegResNet architecture, a widely adopted architecture for three-dimensional segmentation, and trained it using the automatic multi-precision method. We incorporated the dice loss function and dice metric for evaluating the model. We got a dice score of 0.84. For the tumor core, we got a dice score of 0.84; for the whole tumor, 0.90; and for the enhanced tumor, we got a score of 0.79.

[CV-5] RD-ViT: Recurrent-Depth Vision Transformer for Semantic Segmentation with Reduced Data Dependence Extending the Recurrent-Depth Transformer Architecture to Dense Prediction

【速读】：该论文旨在解决视觉 Transformer (Vision Transformer, ViT) 在密集预测任务中对大规模训练数据依赖性强、参数冗余且计算效率低的问题。其核心解决方案是提出 Recurrent-Depth Vision Transformer (RD-ViT)，通过将传统 ViT 中独立的多层变换器块替换为一个共享的循环块（looped T times），并引入线性时不变（LTI）稳定的状态注入以保证收敛性，结合自适应计算时间（Adaptive Computation Time, ACT）实现空间计算资源的动态分配，以及深度 LoRA 适配和可选的混合专家（Mixture-of-Experts, MoE）前馈网络来提升模型表达能力与类别特异性。该设计显著降低了参数量并提升了在小样本场景下的性能表现，同时展现出良好的计算效率与泛化能力。

链接: https://arxiv.org/abs/2605.03999
作者: Renjie He
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Vision Transformers (ViTs) achieve state-of-the-art segmentation accuracy but require large training datasets because each layer has unique parameters that must be learned independently. We present RD-ViT, a Recurrent-Depth Vision Transformer that adapts the Recurrent-Depth Transformer (RDT) architecture to dense prediction tasks, supporting both 2D and 3D inputs. RD-ViT replaces the deep stack of unique transformer blocks with a single shared block looped T times, augmented with LTI-stable state injection for guaranteed convergence, Adaptive Computation Time (ACT) for spatial compute allocation, depth-wise LoRA adaptation, and optional Mixture-of-Experts (MoE) feed-forward networks for category-specific specialization. We evaluate on the ACDC cardiac MRI segmentation benchmark in both 2D slice-level and 3D volumetric settings with exclusively real experiments executed in Google Colab. In 2D, RD-ViT outperforms standard ViT at 10% training data (Dice 0.774 vs 0.762) and at full data (0.882 vs 0.872). In 3D, RD-ViT with MoE achieves Dice 0.812 with 3.0M parameters, reaching 99.4% of standard ViT performance (0.817) at 53% of the parameter count. MoE expert utilization analysis reveals that different experts spontaneously specialize for different cardiac structures (RV, MYO, LV) without explicit routing supervision. ACT halting maps show higher compute allocation at cardiac boundaries, and the mean ponder time decreases from 2.6 to 1.4 iterations during training, demonstrating learned computational efficiency. Depth extrapolation enables inference with more loops than training without degradation. All code, notebooks, and results are publicly released.

[CV-6] 3D Human Face Reconstruction with 3DMM face model from RGB image

【速读】：该论文旨在解决从单张RGB图像中重建高精度三维人脸模型（3D Face Model）的问题，尤其针对传统卷积神经网络（Convolutional Neural Networks, CNNs）在训练时依赖大量标注数据的瓶颈。其解决方案的关键在于构建一个端到端的处理流程，包括人脸检测、关键点定位、基于3D可变形人脸模型（3D Morphable Model, 3DMM）参数回归以及软渲染（soft rendering），并通过粗粒度的人脸形态模型生成合成标注数据以缓解真实数据不足的问题，从而提升重建结果的细节保真度和泛化能力。

链接: https://arxiv.org/abs/2605.03996
作者: Zhangnan Jiang,Zichen Yang
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR)
备注:

点击查看摘要

Abstract:Nowadays as convolution neural networks demonstrate its powerful problem-solving ability in the area of image processing, efforts have been made to reconstruct detailed face shapes from 2D face images or videos. However, to make the full use of CNN, a large number of labeled data is required to train the network. Coarse morphable face model has been used to synthesize labeled data. However, it is hard for coarse morphable face models to generate photo-realistic data with detail such as wrinkles. In this project, we present a pipeline that reconstructs a human face 3D model from a single RGB image. The pipeline includes face detection, landmark detection, regression of 3DMM model parameters, and soft rendering. Mentor: Zhipeng Fan (Email: zf606@nyu.edu) Code Repository: this https URL reconstruction Code Reference: this https URL pytorch

[CV-7] Label-Efficient School Detection from Aerial Imagery via Weakly Supervised Pretraining and Fine-Tuning

【速读】：该论文旨在解决全球范围内学校基础设施地图数据缺失或不准确的问题，尤其是在缺乏可靠官方记录和人工标注数据的低资源地区，传统手动制图方法存在效率低、难以规模化等局限。其解决方案的关键在于提出一种弱监督（weakly supervised）的两阶段训练框架：第一阶段利用稀疏位置点与语义分割自动构建基础设施掩膜并生成边界框，从而实现大规模自动标注；第二阶段使用少量人工标注图像对模型进行微调，以提升检测精度。该方法在仅需50张人工标注图像的情况下即可实现高精度的学校检测，显著降低了标注成本，适用于全球范围内的教育基础设施测绘需求。

链接: https://arxiv.org/abs/2605.03968
作者: Zakarya Elmimouni,Fares Fourati,Mohamed-Slim Alouini
机构: King Abdullah University of Science and Technology (KAUST)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:

点击查看摘要

Abstract:Accurate school detection is essential for supporting education initiatives, including infrastructure planning and expanding internet connectivity to underserved areas. However, many regions around the world face challenges due to outdated, incomplete, or unavailable official records. Manual mapping efforts, while valuable, are labor-intensive and lack scalability across large geographic areas. To address this, we propose a weakly supervised framework for school detection from aerial imagery that minimizes the need for human annotations while supporting global mapping efforts. Our method is specifically designed for low-data regimes, where manual annotations are extremely scarce. We introduce an automatic labeling pipeline that leverages sparse location points and semantic segmentation to generate infrastructure masks from which we generate bounding boxes. Using these automatically labeled images, we train our detectors on a first training stage to learn a representation of what schools look like, then using a small set of manually labeled images, we fine-tune the previously trained models on this clean dataset. This two stage training pipeline enables large-scale and strong detection in low-data setting of school infrastructure with minimal supervision. Our results demonstrate strong object detection performance, particularly in the low-data regime, where the models achieve promising results using only 50 manually labeled images, significantly reducing the need for costly annotations. This framework supports education and connectivity initiatives worldwide by providing an efficient and extensible approach to mapping schools from space. All models, training code and auto-labeled data will be publicly released to foster future research and real-world impact.

[CV-8] UnAC: Adaptive Visual Prompting with Abstraction and Stepwise Checking for Complex Multimodal Reasoning

【速读】：该论文旨在解决大模型多模态模型（Large Multimodal Models, LMMs）在需要多步推理的视觉任务中表现不可靠的问题。其核心挑战在于如何提升模型对图像细节的理解能力、有效提取关键信息，并通过结构化验证机制增强推理过程的准确性。解决方案的关键在于提出UnAC框架，包含三个核心技术：(1) 自适应视觉提示策略，引导模型聚焦图像中的显著区域以增强细粒度理解；(2) 图像抽象提示，用于从图像中高效提取关键语义信息；(3) 渐进式自检机制，通过对分解后的子问题及其答案逐层验证，提高整体推理链条的可靠性。该方法在MathVista、MM-Vet和MMMU三个公开基准上得到验证，显著提升了LMMs在复杂多模态任务中的推理性能。

链接: https://arxiv.org/abs/2605.03950
作者: Yifan Wang,Yun Fu
机构: Northeastern University(东北大学)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Although recent LMMs have become much stronger at visual perception, they remain unreliable on problems that require multi-step reasoning over visual evidence. In this paper, we present UnAC (Understanding, Abstracting, and Checking), a multimodal prompting method that strengthens reasoning for complex multimodal tasks in LMMs (e.g., GPT-4o, Gemini 1.5, and GPT-4V). To improve image understanding and capture fine details, we propose an adaptive visual prompting strategy that enables LMMs to focus on salient regions. We further design an image-abstraction prompt to effectively extract key information from images. In addition, we introduce a gradual self-checking scheme that improves reasoning by verifying each decomposed subquestion and its answer. Extensive experiments on three public benchmarks-MathVista, MM-Vet, and MMMU.

[CV-9] Reservoir property image slices from the Groningen gas field for image translation and segmentation

【速读】：该论文旨在解决当前储层表征工作流中缺乏公开可用的地质图像数据集的问题，这些问题限制了基于图像的机器学习/深度学习乃至生成式AI方法的可复现性基准测试。解决方案的关键在于构建一个高分辨率的储层属性图像切片数据集，该数据集源自Groningen静态地质模型，包含对齐的二维PNG图像（表示岩相、孔隙度、渗透率和含水饱和度），并配套提供可归档的软件工作流，用于重现图像增强、掩膜生成、配对图像构建及基线实验。通过将固定图像数据集与可复现的处理流程分离，该研究为地球科学、储层建模和机器学习应用提供了透明且可重复的基础。

链接: https://arxiv.org/abs/2605.03942
作者: Abdulrahman Al-Fakih,Nabil Sariah,Ardiansyah Koeshidayatullah,SanLinn I. Kaka
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Databases (cs.DB); Geophysics (physics.geo-ph)
备注:

点击查看摘要

Abstract:Reservoir characterization workflows increasingly rely on image-based and machine-learning/deep learning or even generative AI approaches, but openly available geological image datasets suitable for reproducible benchmarking remain limited. Here we describe a high-resolution dataset of reservoir-property image slices derived from the Groningen static geological model. The dataset contains aligned two-dimensional PNG images representing facies, porosity, permeability, and water saturation, generated from three-dimensional reservoir grids and prepared for downstream visualization, segmentation, and image-to-image translation tasks. In addition to the deposited original image corpus, we provide an archived software workflow for reproducing augmentation, mask generation, paired-image construction, and example baseline experiments. The resource is designed to support benchmarking of geological image analysis methods and the study of cross-domain relationships among reservoir properties. By separating the fixed image dataset from the reproducible processing workflow, this work provides a transparent foundation for reuse in geoscience, reservoir modeling, and machine-learning applications.

[CV-10] A Benchmark for Interactive World Models with a Unified Action Generation Framework ICML2026

【速读】：该论文旨在解决当前生成式 AI（Generative AI）在实现人工通用智能（Artificial General Intelligence, AGI）过程中，缺乏大规模、统一的基准测试来评估世界模型（world models）在物理交互能力上的表现这一问题。现有研究虽已探索基于交互的世界模型以支持感知、推理与行动的协同学习，但受限于数据规模不足和评价标准不一，难以系统性地比较不同模型的交互能力。其解决方案的关键在于提出 iWorld-Bench——一个涵盖33万视频片段的多样化数据集和一套统一的动作生成框架（Action Generation Framework），通过设计六类交互任务（共4900个高质量测试样本），实现了对视觉生成、轨迹跟随及记忆能力的联合评估，从而为世界模型的训练与评测提供了可扩展、多维度的标准化平台。

链接: https://arxiv.org/abs/2605.03941
作者: Jianjie Fang,Yingshan Lei,Qin Wan,Ziyou Wang,Yuchao Huang,Yongyan Xu,Baining Zhao,Weichen Zhang,Chen Gao,Xinlei Chen,Yong Li
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accepted at ICML 2026

点击查看摘要

Abstract:Achieving Artificial General Intelligence (AGI) requires agents that learn and interact adaptively, with interactive world models providing scalable environments for perception, reasoning, and action. Yet current research still lacks large-scale datasets and unified benchmarks to evaluate their physical interaction capabilities. To address this, we propose iWorld-Bench, a comprehensive benchmark for training and testing world models on interaction-related abilities such as distance perception and memory. We construct a diverse dataset with 330k video clips and select 2.1k high-quality samples covering varied perspectives, weather, and scenes. As existing world models differ in interaction modalities, we introduce an Action Generation Framework to unify evaluation and design six task types, generating 4.9k test samples. These tasks jointly assess model performance across visual generation, trajectory following, and memory. Evaluating 14 representative world models, we identify key limitations and provide insights for future research. The iWorld-Bench model leaderboard is publicly available at this http URL.

[CV-11] StateVLM: A State-Aware Vision-Language Model for Robotic Affordance Reasoning

【速读】：该论文旨在解决视觉语言模型（Vision-Language Models, VLMs）在机器人任务中面临的数值推理能力不足的问题，尤其是在目标检测与目标状态定位方面。现有VLMs虽能感知视觉信息并理解自然语言指令，但其基于序列预测的架构难以精确处理涉及空间坐标和状态参数的回归任务。解决方案的关键在于提出一种新颖的训练策略——通过引入辅助回归损失（Auxiliary Regression Loss, ARL），利用框解码器（box decoder）输出计算回归误差，在微调阶段增强模型对目标位置和状态的细粒度建模能力，同时保持推理时的标准序列生成结构不变。这一方法显著提升了模型在多个基准上的性能，特别是在新提出的物体状态可操作性推理基准（Object State Affordance Reasoning, OSAR）上，验证了ARL对复杂任务如可操作性推理的一致性和有效性。

链接: https://arxiv.org/abs/2605.03927
作者: Xiaowen Sun,Matthias Kerzel,Mengdi Li,Xufeng Zhao,Paul Striker,Stefan Wermter
机构: University of Hamburg (汉堡大学); King Abdullah University of Science and Technology (阿卜杜拉国王科技大学)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Vision-language models (VLMs) have shown remarkable performance in various robotic tasks, as they can perceive visual information and understand natural language instructions. However, when applied to robotics, VLMs remain subject to a fundamental limitation inherent in large language models (LLMs): they struggle with numerical reasoning, particularly in object detection and object-state localization. To explore numerical reasoning as a regression task in VLMs, we propose a novel training strategy to adapt VLMs for object detection and object-state localization. This approach leverages box decoder outputs to compute an Auxiliary Regression Loss (ARL) during fine-tuning, while preserving standard sequence prediction at inference. We leverage this training strategy to develop StateVLM (State-aware Vision-Language Model), a novel model designed to perceive and learn fine-grained object representations, including precise localization of objects and their states, as well as graspable regions. Due to the lack of a benchmark for object-state affordance reasoning, we introduce an open-source benchmark, Object State Affordance Reasoning (OSAR), which contains 1,172 scenes with 7,746 individual objects and corresponding bounding boxes. Comparative experiments on adapted benchmarks (RefCOCO, RefCOCO+, and \mboxRefCOCOg) demonstrate that ARL improves model performance by an average of 1.6% compared to models without ARL. Experiments on the OSAR benchmark further support this finding, showing that StateVLM with ARL achieves an average of 5.2% higher performance than models without ARL. In particular, ARL is also important for the complex task of affordance reasoning in OSAR, where it enhances the consistency of model outputs.

[CV-12] ask-Aware Scanning Parameter Configuration for Robotic Inspection Using Vision Language Embeddings and Hyperdimensional Computing

【速读】：该论文旨在解决工业机器人激光轮廓扫描中传感器参数配置依赖人工试错、易因参数不匹配导致测量失真（如饱和、截断或信号丢失）的问题。其核心挑战在于如何基于任务意图和场景感知，自动推理出最优的扫描参数组合（包括采样频率、测量范围、曝光时间等）。解决方案的关键在于提出ScanHD框架——一个基于超维计算（hyperdimensional computing）的多模态决策系统，通过将自然语言指令与RGB观测融合为任务感知编码，并利用紧凑记忆实现参数维度上的关联推理，从而在低延迟下精准匹配离散扫描参数配置，显著优于传统启发式规则和主流多模态模型。

链接: https://arxiv.org/abs/2605.03909
作者: Zhiling Chen,David Gorsich,Matthew P. Castanier,Yang Zhang,Jiong Tang,Farhad Imani
机构: University of Connecticut (康涅狄格大学); US Army DEVCOM Ground Vehicle Systems Center (美国陆军DEVCOM地面车辆系统中心)
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: 20 pages, 13 figures

点击查看摘要

Abstract:Robotic laser profiling is widely used for dimensional verification and surface inspection, yet measurement fidelity is often dominated by sensor configuration rather than robot motion. Industrial profilers expose multiple coupled parameters, including sampling frequency, measurement range, exposure time, receiver dynamic range, and illumination, that are still tuned by trial-and-error; mismatches can cause saturation, clipping, or missing returns that cannot be recovered downstream. We formulate instruction-conditioned sensing parameter recommendation; given a pre-scan RGB observation and a natural-language inspection instruction, infer a discrete configuration over key parameters of a robot-mounted profiler. To benchmark this problem, we develop Instruct-Obs2Param, a real-world multimodal dataset linking inspection intents and multi-view pose and illumination variation across 16 objects to canonical parameter regimes. We then propose ScanHD, a hyperdimensional computing framework that binds instruction and observation into a task-aware code and performs parameter-wise associative reasoning with compact memories, matching discrete scanner regimes while yielding stable, interpretable, low-latency decisions. On Instruct-Obs2Param, ScanHD achieves 92.7% average exact accuracy and 98.1% average Win@1 accuracy across the five parameters, with strong cross-split generalization and low-latency inference suitable for deployment, outperforming rule-based heuristics, conventional multimodal models, and multimodal large language models. This work enables autonomous, instruction-conditioned sensing configuration from task intent and scene context, eliminating manual tuning and elevating sensor configuration from a static setting to an adaptive decision variable.

[CV-13] Raising the Ceiling: Better Empirical Fixation Densities for Saliency Benchmarking

【速读】：该论文旨在解决当前基于人类眼动数据估计的固定点密度（fixation density）在视觉显著性基准测试中存在方法陈旧、精度不足的问题，尤其是在样本级评估（如失败案例分析、逆向基准测试和逐图像模型比较）日益重要的背景下。传统使用的固定带宽各向同性高斯核密度估计（KDE）方法几十年未变，难以准确捕捉不同图像间的人类注视一致性，从而影响基准排名、失败案例分析及对人类视觉行为科学结论的可靠性。解决方案的关键在于提出一种原理性的混合模型：结合自适应带宽KDE（基于Abramson方法）、中心偏置项与均匀分布成分，并引入最先进的显著性模型，通过留一被试交叉验证优化每张图像的参数，以区分并建模不同空间和语义类型的观察者一致性。该方法在多个基准上显著提升了观测者间一致性估计，尤其在关键的失败案例图像上改善超过25%，揭示了当前先进显著性模型仍有较大改进空间。

链接: https://arxiv.org/abs/2605.03885
作者: Susmit Agrawal,Jannis Hollman,Matthias Kümmerer
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:

点击查看摘要

Abstract:Empirical fixation densities, spatial distributions estimated from human eye-tracking data, are foundational to saliency benchmarking. They directly shape benchmark conclusions, leaderboard rankings, failure case analyses, and scientific claims about human visual behavior. Yet the standard estimation method, fixed-bandwidth isotropic Gaussian KDE, has gone essentially unchanged for decades. This matters now more than ever: as the field shifts toward sample-level evaluation (failure case analysis, inverse benchmarking, per-image model comparison), reliable per-image density estimates become critical. We propose a principled mixture model that combines an adaptive-bandwidth KDE based on Abramson’s method, center bias and uniform components, and a state-of-the-art saliency model, to capture different spatial and semantic types of interobserver consistency, and optimize all parameters per image via leave-one-subject-out cross-validation. Our method yields substantially higher interobserver consistency estimates across multiple benchmarks, with median per-image gains of 5-15% in log-likelihood and up to 2 percentage points in AUC. For the most affected images – precisely those most relevant to failure case analysis – improvements exceed 25%. We leverage these improved estimates to identify and analyze remaining failure cases of state-of-the-art saliency models, demonstrating that significant headroom for model improvement remains. More broadly, our findings highlight that empirical fixation densities should not be treated as fixed ground truths but as evolving estimates that improve with better methodology.

[CV-14] DMGD: Train-Free Dataset Distillation with Semantic-Distribution Matching in Diffusion Models CVPR2026

【速读】：该论文旨在解决基于扩散模型（Diffusion Model）的图像数据集蒸馏（Dataset Distillation）中存在的两个关键问题：一是现有方法通常需要额外的微调阶段，导致训练成本高；二是缺乏有效的引导机制以同时保障合成数据的语义一致性与多样性。解决方案的核心在于提出一种无需训练的双匹配引导扩散框架（Dual Matching Guided Diffusion, DMGD），其关键创新包括：1）通过条件似然优化建立语义匹配（Semantic Matching），避免使用辅助分类器；2）设计动态引导机制，在保持语义对齐的同时提升合成数据多样性；3）引入基于最优传输（Optimal Transport, OT）的分布匹配方法，更精确地对齐目标数据分布结构；4）提出两种高效策略——分布近似匹配（Distribution Approximate Matching）和贪心渐进匹配（Greedy Progressive Matching），在极低计算开销下实现有效的分布引导。实验表明，该方法在ImageNet-Woof、ImageNet-Nette和ImageNet-1K上均显著优于需微调的SOTA方法，平均准确率提升达2.1%–5.4%。

链接: https://arxiv.org/abs/2605.03877
作者: Qichao Wang,Yunhong Lu,Hengyuan Cao,Junyi Zhang,Min Zhang
机构: Zhejiang University (浙江大学); Shanghai Institute for Advanced Study-Zhejiang University (浙江大学先进技术研究院); Shanghai Institute for Mathematics and Interdisciplinary Sciences (上海数学与交叉学科研究院)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accepted by CVPR2026

点击查看摘要

Abstract:Dataset distillation enables efficient training by distilling the information of large-scale datasets into significantly smaller synthetic datasets. Diffusion based paradigms have emerged in recent years, offering novel perspectives for dataset distillation. However, they typically necessitate additional fine-tuning stages, and effective guidance mechanisms remain underexplored. To address these limitations, we rethink diffusion based dataset distillation and propose a Dual Matching Guided Diffusion (DMGD) framework, centered on efficient training-free guidance. We first establish Semantic Matching via conditional likelihood optimization, eliminating the need for auxiliary classifiers. Furthermore, we propose a dynamic guidance mechanism that enhances the diversity of synthetic data while maintaining semantic alignment. Simultaneously, we introduce an optimal transport (OT) based Distribution Matching approach to further align with the target distribution structure. To ensure efficiency, we develop two enhanced strategies for diffusion based framework: Distribution Approximate Matching and Greedy Progressive Matching. These strategies enable effective distribution matching guidance with minimal computational overhead. Experimental results on ImageNet-Woof, ImageNet-Nette, and ImageNet-1K demonstrate that our training-free approach achieves significant improvements, outperforming state-of-the-art (SOTA) methods requiring additional fine-tuning by average accuracy gains of 2.1%, 5.4%, and 2.4%, respectively.

[CV-15] Quantifying the human visual exposome with vision language models

【速读】：该论文旨在解决视觉环境作为心理健康决定因素难以量化的问题，现有方法依赖粗略的地理空间代理变量或有偏倚的自我报告，无法捕捉个体日常生活中第一人称视角的视觉情境。其解决方案的关键在于将生态瞬时评估（Ecological Momentary Assessment, EMA）与视觉语言模型（Vision Language Models, VLMs）相结合，通过分析参与者生成的2674张照片，客观量化视觉体验的语义丰富度，并进一步构建基于半自主大型语言模型（Large Language Model, LLM）的管道，从超七百万篇科学文献中提取近1000个与心理健康相关的环境特征。该方法实现了对真实世界图像中视觉上下文的高通量解析，揭示了高达33%的VLM提取情境评分与情绪和压力显著相关，从而建立了一种可扩展的客观视觉暴露组学（Visual Exposomics）范式。

链接: https://arxiv.org/abs/2605.03863
作者: Christian Rominger(1),Andreas R. Schwerdtfeger(1),Malay Gaherwar Singh(2),Dimitri Khudyakow(2),Elizabeth A. M. Michels(2),Fabian Wolf(2),Jakob Nikolas Kather(2,3,4),Magdalena Katharina Wekenborg(2) ((1) University of Graz, (2) TU Dresden, (3) University Hospital Carl Gustav Carus Dresden, (4) National Center for Tumor Diseases Heidelberg)
机构: 未知
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:The visual environment is a fundamental yet unquantified determinant of mental health. While the concept of the environmental exposome is well established, current methods rely on coarse geospatial proxies or biased self reports, failing to capture the first person visual context of daily life. We addressed this gap by coupling ecological momentary assessment with vision language models (VLMs) to quantify the semantic richness of human visual experience. Across 2674 participant generated photographs, VLM derived estimates of greenness robustly predicted momentary affect and chronic stress, consistent with established benchmarks. We then developed a semi autonomous large language model (LLM) based pipeline that mined over seven million scientific publications to extract nearly 1000 environmental features empirically linked to mental health. When applied to real world imagery, up to 33 percent of VLM extracted context ratings significantly correlated with affect and stress. These findings establish a scalable objective paradigm for visual exposomics, enabling high throughput decoding of how the visible world is associated with mental health.

[CV-16] A Deeper Dive into the Irreversibility of PolyProtect: Making Protected Face Templates Harder to Invert

【速读】：该论文旨在解决PolyProtect这一生物特征模板保护方法在实际应用中面临的可逆性（irreversibility）问题，即如何有效防止受保护的模板被反向还原为原始生物特征嵌入（embedding），从而保障用户隐私。其解决方案的关键在于提出一种“密钥选择算法”（key selection algorithm），该算法通过优化多项式系数和指数的选择策略，使生成的PolyProtected模板相较于纯随机密钥更难被数值求解器基于余弦距离进行逆向重构，同时显著提升了不同重叠参数下模板的不可逆性一致性，从而实现了对不可逆性与识别准确率之间权衡关系的更好控制。此外，研究还发现嵌入元素的取值范围会影响识别性能，并提出通过预归一化处理来改善精度。

链接: https://arxiv.org/abs/2605.03857
作者: Vedrana Krivokuća Hahn,Jérémy Maceiras,Sébastien Marcel
机构: Idiap Research Institute (Idiap 研究所)
类目: Computer Vision and Pattern Recognition (cs.CV); Cryptography and Security (cs.CR)
备注: Submitted to TIFS journal on 18 February 2026 (under review). Consists of: 12 pages, 10 figures, 4 tables

点击查看摘要

Abstract:This work presents a deeper analysis of the “irreversibility” property of PolyProtect, a biometric template protection method initially proposed for securing face embeddings. PolyProtect transforms embeddings into protected templates via multivariate polynomials, whose coefficients and exponents are distinct for each subject enrolled in the face recognition system. A polynomial is applied to consecutive sets of elements from a given embedding, where the amount of overlap between the sets is a tunable parameter. We begin our irreversibility analysis by demonstrating that PolyProtected templates are easier to invert using a numerical solver based on cosine distance, as opposed to Euclidean distance (used in the earlier PolyProtect work). To make this inversion more difficult, we then propose a “key selection algorithm”, which tries to choose “keys” (coefficients and exponents of the PolyProtect polynomial) that enhance the irreversibility of PolyProtected templates, compared to when the keys are purely random. Our experiments show that this algorithm is effective at generating PolyProtected templates that are significantly more difficult to invert, and that it approximately equalises the irreversibility of PolyProtected templates generated using different “overlap” parameters. This allows for better control of the irreversibility versus accuracy trade-off, known to exist across different overlaps. We also show that accuracy in the PolyProtected domain can be affected by the range in which the embedding elements lie, but that this can be improved by normalizing the embeddings prior to applying PolyProtect. This work is reproducible using our open-source code.

[CV-17] Stream-R1: Reliability-Perplexity Aware Reward Distillation for Streaming Video Generation

【速读】：该论文旨在解决蒸馏加速自回归流式视频扩散模型时存在的质量瓶颈问题，即现有方法对教师模型输出的监督信号采用统一权重，忽略了两个互补的变异性维度：跨推理轨迹的可靠性（Inter-Reliability）和时空元素内的困惑度差异（Intra-Perplexity）。这种忽略导致优化目标混淆了“是否学习某次推理”与“在何处集中优化”的决策。解决方案的关键在于提出Stream-R1框架，通过单一共享的奖励引导机制，在两个层面动态重加权蒸馏目标：首先，基于预训练视频奖励分数对每次推理轨迹进行指数缩放，使高可靠性的推理主导优化；其次，利用相同奖励模型提取像素级梯度显著性，生成空间和时间维度的权重，聚焦于能带来最大预期收益的区域与帧。同时引入自适应平衡机制，确保视觉质量、运动质量和文本对齐三个维度不相互压制，从而实现无架构修改且无额外推理开销下的全面性能提升。

链接: https://arxiv.org/abs/2605.03849
作者: Bin Wu,Mengqi Huang,Shaojin Wu,Weinan Jia,Yuxin Wang,Zhendong Mao,Yongdong Zhang
机构: University of Science and Technology of China(中国科学技术大学); FrameX.AI
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Distillation-based acceleration has become foundational for making autoregressive streaming video diffusion models practical, with distribution matching distillation (DMD) as the de facto choice. Existing methods, however, train the student to match the teacher’s output indiscriminately, treating every rollout, frame, and pixel as equally reliable supervision. We argue that this caps distilled quality, since it overlooks two complementary axes of variance in DMD supervision: Inter-Reliability across student rollouts whose supervision varies in reliability, and Intra-Perplexity across spatial regions and temporal frames that contribute unequally to where quality can still be improved. The objective thus conflates two questions under a uniform weight: whether to learn from each rollout, and where to concentrate optimization within it. To address this, we propose Stream-R1, a Reliability-Perplexity Aware Reward Distillation framework that adaptively reweights the distillation objective at both rollout and spatiotemporal-element levels through a single shared reward-guided mechanism. At the Inter-Reliability level, Stream-R1 rescales each rollout’s loss by an exponential of a pretrained video reward score, so that rollouts with reliable supervision dominate optimization. At the Intra-Perplexity level, it back-propagates the same reward model to extract per-pixel gradient saliency, which is factored into spatial and temporal weights that concentrate optimization pressure on regions and frames where refinement yields the largest expected gain. An adaptive balancing mechanism prevents any single quality axis from dominating across visual quality, motion quality, and text alignment. Stream-R1 attains consistent improvements on all three dimensions over distillation baselines on standard streaming video generation benchmarks, without architectural modification or additional inference cost.

[CV-18] Parameter-Efficient Multi-View Proficiency Estimation: From Discriminative Classification to Generative Feedback

【速读】：该论文旨在解决多视角下人体动作熟练度（proficiency）估计问题，其核心挑战在于熟练度信息分散在细微的时间节奏、平衡控制、身体力学及执行细节中，且常需跨视角融合与短时动态捕捉。解决方案的关键在于三个创新：(1) SkillFormer 提出参数高效的判别式架构实现选择性多视角融合；(2) PATS 通过保留基础动作的局部密集采样提升时间建模精度；(3) ProfVLM 将熟练度估计重构为条件语言生成任务，借助门控交叉视图投影和轻量语言模型输出可解释的评分标签与专家风格反馈。三者结合实现了比视频Transformer基线更高效（参数减少最多20倍、训练轮次减少最多3倍）且更具可解释性的性能提升，标志着从封闭集分类向生成式可操作反馈的范式转变。

链接: https://arxiv.org/abs/2605.03848
作者: Edoardo Bianchi,Antonio Liotta
机构: Faculty of Engineering, Free University of Bozen-Bolzano (自由大学博岑-波尔扎诺工程学院)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Estimating how well a person performs an action, rather than which action is performed, is central to coaching, rehabilitation, and talent identification. This task is challenging because proficiency is encoded in subtle differences in timing, balance, body mechanics, and execution, often distributed across multiple views and short temporal events. We discuss three recent contributions to multi-view proficiency estimation on Ego-Exo4D. SkillFormer introduces a parameter-efficient discriminative architecture for selective multi-view fusion; PATS improves temporal sampling by preserving locally dense excerpts of fundamental movements; and ProfVLM reformulates proficiency estimation as conditional language generation, producing both a proficiency label and expert-style feedback through a gated cross-view projector and a compact language backbone. Together, these methods achieve state-of-the-art accuracy on Ego-Exo4D with up to 20x fewer trainable parameters and up to 3x fewer training epochs than video-transformer baselines, while moving from closed-set classification toward interpretable feedback generation. These results highlight a shift toward efficient, multi-view systems that combine selective fusion, proficiency-aware sampling, and actionable generative feedback.

[CV-19] Conditions for well-posed color recovery in scattering media

【速读】：该论文旨在解决在散射介质中从图像中恢复场景颜色这一基础逆问题，该问题本质上是病态的（ill-posed），因为多个解可能解释相同的观测结果，且在不了解候选解空间的情况下无法控制预测误差。解决方案的关键在于识别出导致病态性的两个根源：(i) 光谱信号向像素强度的投影，以及 (ii) 未知的介质参数；并进一步证明，仅靠传感器性能提升无法消除介质引起的失真，必须引入额外约束。研究发现，图像中存在的跨像素关系（cross-pixel relationships）可作为自然约束，在理想高光谱相机下将解空间压缩至唯一候选解，从而使得该问题变为适定（well-posed）。这一发现为基于第一性原理的新一代视觉算法开辟了道路，实现了在散射环境中对图像的定量分析。

链接: https://arxiv.org/abs/2605.03837
作者: Grigory Solomatov,Derya Akkaynak
机构: Hatter Department of Marine Technologies, University of Haifa, Haifa, Israel; Interuniversity Institute for Marine Sciences, Eilat, Israel
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Recovering scene color from images captured in scattering media is a fundamental inverse problem in optical imaging. Yet the problem is intrinsically ill-posed as multiple solutions can explain the same observation, and prediction error cannot be controlled without understanding the space of candidate solutions. Here, we present sufficient conditions under which color recovery in a scattering medium becomes well-posed. Observing that ill-posedness stems from (i) projection of spectral signals onto pixel intensities, and (ii) unknown medium parameters, we demonstrate that sensor improvements alone cannot resolve medium-induced distortions without additional constraints. We identify recovery patterns, cross-pixel relationships that naturally occur in images, and prove, for an ideal hyperspectral camera, that they restrict the solution to a unique candidate. This opens the door to a new class of vision algorithms grounded in first principles, enabling quantitative analysis of images in scattering environments.

[CV-20] Identity-Consistent Multi-Pose Generation of Contactless Fingerprints

【速读】：该论文旨在解决接触式指纹（contact-based fingerprint）与非接触式指纹（contactless fingerprint）之间因自由手指姿态导致的严重非线性几何失真问题，从而缩小两者间的跨模态域差距（cross-modal domain gap）。现有方法依赖显式的几何校正或图像增强，在极端姿态变化下表现脆弱。其解决方案的关键在于提出一种物理启发的多姿态生成框架IMPOSE，通过三个阶段实现：(1) 利用离散码本表示的潜在扩散模型生成滚动指纹身份特征；(2) 基于Sauvola局部自适应二值化作为身份锚点，进行跨模态翻译；(3) 采用3D手指模型纹理映射与投影进行物理驱动的多姿态模拟。该方法生成的样本在脊线拓扑层面保持严格的身份一致性，并与标准指纹坐标空间对齐，显著提升了跨模态匹配性能。

链接: https://arxiv.org/abs/2605.03830
作者: Zhiyu Pan,Xiongjun Guan,Jianjiang Feng,Jie Zhou
机构: Tsinghua University (清华大学)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Under review

点击查看摘要

Abstract:Contactless fingerprint recognition has gained increasing attention due to its advantages in hygiene and acquisition flexibility. However, the absence of physical contact constraints introduces severe nonlinear geometric distortions caused by free finger poses in 3D space, resulting in a substantial cross-modal domain gap between contactless and conventional contact-based fingerprints. Existing solutions largely rely on explicit geometric correction or image enhancement, which are fragile under extreme pose variations. In this paper, we propose Identity-Consistent Multi-Pose Generation of Contactless Fingerprints (IMPOSE), a physics-inspired framework that synthesizes identity-preserving, multi-pose contactless fingerprint samples to empower recognition models. IMPOSE consists of three stages: (1) rolled fingerprint identity generation via latent diffusion with discrete codebook representations, (2) cross-modal translation from rolled to contactless modality guided by Sauvola-based local adaptive binarization as an identity anchor, and (3) physics-based multi-pose simulation through 3D finger model texture mapping and projection. The generated samples maintain strict identity consistency at the ridge topology level and spatial alignment with standard fingerprint coordinate space. Extensive experiments on the UWA and PolyU CL2CB databases demonstrate that fine-tuning fixed-length dense descriptors (FDD) with IMPOSE-synthesized data achieves state-of-the-art cross-modal matching, reducing EER to 8.74% on UWA and 2.26% on PolyU CL2CB. Synthetic data also yields consistent gains across mainstream representations including DeepPrint and AFRNet, and the hybrid strategy combining synthetic and real data achieves the best overall results. The code and generated samples are available at this https URL.

[CV-21] Multimodal Learning on Low-Quality Data with Conformal Predictive Self-Calibration CVPR2026

【速读】：该论文旨在解决多模态学习中因低质量数据导致的性能下降问题，其核心表现形式为模态不平衡（modality imbalance）和噪声污染（noisy corruption）。作者指出，这两类问题本质上均源于模型对各模态及样本可靠性预测不确定性的处理不足。为此，论文提出统一框架Conformal Predictive Self-Calibration (CPSC)，其关键在于引入基于校准预测（conformal prediction）的自校准训练机制，包含两个核心模块：(1) 表征自校准（Representation Self-Calibration），通过分解单模态特征并利用校准预测器识别最稳健成分进行选择性融合，以增强特征鲁棒性；(2) 梯度自校准（Gradient Self-Calibration），在反向传播过程中依据实例级可靠性评分重校准梯度流，引导优化朝向更可信方向。此外，还设计了校准预测器的自更新策略，确保系统在整个训练过程中协同演化，从而实现端到端的在线自我校准能力。

链接: https://arxiv.org/abs/2605.03820
作者: Xun Jiang,Yufan Gu,Disen Hu,Yuqing Hou,Yazhou Yao,Fumin Shen,Heng Tao Shen,Xing Xu
机构: Tongji University (同济大学); University of Electronic Science and Technology of China (电子科技大学); Nanjing University of Science and Technology (南京理工大学)
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Multimedia (cs.MM)
备注: Accepted by CVPR 2026

点击查看摘要

Abstract:Multimodal learning often grapples with the challenge of low-quality data, which predominantly manifests as two facets: modality imbalance and noisy corruption. While these issues are often studied in isolation, we argue that they share a common root in the predictive uncertainty towards the reliability of individual modalities and instances during learning. In this paper, we propose a unified framework, termed Conformal Predictive Self-Calibration (CPSC), which leverages conformal prediction to equip the model with the ability to perform self-guided calibration on-the-fly. The core of our proposed CPSC lies in a novel self-calibrating training loop that seamlessly integrates two key modules: (1) Representation Self-Calibration, which decomposes unimodal features into components, and selectively fuses the most robust ones identified by a conformal predictor to enhance feature resilience. (2) Gradient Self-Calibration, which recalibrates the gradient flow during backpropagation based on instance-wise reliability scores, steering the optimization towards more trustworthy directions. Furthermore, we also devise a self-update strategy for the conformal predictor to ensure the entire system co-evolves consistently throughout the training process. Extensive experiments on six benchmark datasets under both imbalanced and noisy settings demonstrate that our CPSC framework consistently outperforms existing state-of-the-art methods. Our code is available at this https URL.

[CV-22] Enhancing Visual Question Answering with Multimodal LLM s via Chain-of-Question Guided Retrieval-Augmented Generation

【速读】：该论文旨在解决多模态大语言模型（Multimodal Large Language Models, MLLMs）在开放域视觉问答（Visual Question Answering, VQA）任务中，因外部知识获取不充分和推理结构化不足而导致的性能瓶颈问题。其解决方案的关键在于提出一种逻辑提示策略——CoVQD（Chain-of-Thought Visual Question Decomposition），该策略融合了思维链（Chain-of-Thought, CoT）推理与视觉问题分解（Visual Question Decomposition, VQD），以引导检索机制聚焦于更准确、相关的外部知识；在此基础上构建的CgRAG（CoVQD-guided Retrieval-Augmented Generation）框架，使MLLMs能够在结构化视觉-文本推理指导下访问更全面且连贯的外部知识，从而提升复杂跨域VQA场景下的泛化能力与可靠性。

链接: https://arxiv.org/abs/2605.03790
作者: Quanxing Xu,Ling Zhou,Xian Zhong,Xiaohua Huang,Rubing Huang,Chia-Wen Lin
机构: Macau University of Science and Technology (澳门科技大学); Wuhan University of Technology (武汉理工大学); Nanjing Institute of Technology (南京工程学院); National Tsing Hua University (国立清华大学)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:With advances in multimodal research and deep learning, Multimodal Large Language Models (MLLMs) have emerged as a powerful paradigm for a wide range of multimodal tasks. As a core problem in vision-language research, Visual Question Answering (VQA) has increasingly employed MLLMs to improve performance, particularly in open-domain settings where external knowledge is essential. In this work, we aim to further enhance retrieval-based VQA by more effectively integrating MLLMs with structured reasoning and knowledge acquisition. We introduce a logical prompting strategy that fuses Chain-of-Thought (CoT) reasoning with Visual Question Decomposition (VQD), termed CoVQD, to guide retrieval toward more accurate and relevant knowledge for MLLM inference. Building on this idea, we propose a new framework, CoVQD-guided RAG (CgRAG), which enables MLLMs to access more comprehensive and coherent external knowledge while benefiting from structured visual-text reasoning guidance, thereby improving generalization and reliability in complex cross-domain VQA scenarios. Extensive experiments on E-VQA, InfoSeek, and OKVQA benchmarks demonstrate the effectiveness of the proposed method.

[CV-23] A Robust Unsupervised Domain Adaptation Framework for Medical Image Classification Using RKHS-MMD

【速读】：该论文旨在解决医学图像标注难题，即由于不同医疗中心和成像设备之间的异质性导致的域偏移（domain shift）和模态差异（modality discrepancy），从而限制了模型在未标注数据上的泛化能力。解决方案的关键在于提出一种无监督域自适应（Unsupervised Domain Adaptation, UDA）框架，结合迁移学习与基于再生核希尔伯特空间（Reproducing Kernel Hilbert Space, RKHS）的最大均值差异（Maximum Mean Discrepancy, MMD）损失，通过联合优化分类损失与RKHS-MMD损失，实现源域与目标域特征分布的有效对齐，显著提升模型在未标注医学数据集上的泛化性能，减少对手动标注的依赖。

链接: https://arxiv.org/abs/2605.03787
作者: Sapna Sachan,Rakesh Kumar Sanodiya,Amulya Kumar Mahto
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 13 Pages, 6 figures

点击查看摘要

Abstract:Labeling medical images is a major bottleneck in the field of medical imaging, as it requires domain-specific expertise, and it gets further complicated due to variability across different medical centers and different imaging devices. Such heterogeneity introduces domain shifts and modality discrepancies, which limits the generalization of trained models. To address this important challenge, we propose an unsupervised domain adaptation framework that combines transfer learning with a Reproducing Kernel Hilbert Space based Maximum Mean Discrepancy loss for the alignment of source and target domains. By jointly optimizing classification and RKHS-MMD losses, the methodology enhances generalization to unannotated medical datasets while diminishing reliance on manual annotation. Experimental evaluations presented on two chest X-ray datasets, which are obtained from different medical centers, show outstanding improvements over models trained without adaptation. Furthermore, we perform a comparative study to see that RKHS-MMD performs better than the standard Maximum Mean Discrepancy in reducing modality gap, emphasizing its effectiveness for medical image classification and also its strong capability in advanced AI-driven medical diagnostics.

[CV-24] ReLeaf: Benchmarking Leaf Segmentation across Domains and Species CVPR

【速读】：该论文旨在解决精准农业中植物个体化处理所依赖的细粒度视觉分析难题，特别是叶片级别的分割任务尚未得到充分研究的问题。其核心挑战在于现有数据集在物种覆盖范围上的不足以及对现代实例分割架构系统性评估的缺失。解决方案的关键在于：首先系统梳理并识别出四个适合的公开叶级分割数据集；其次通过对比单阶段、两阶段及基于Transformer的检测器，确定YOLO26模型配置在实际应用中具有最佳性能与效率平衡；最后引入一个新的包含23种植物物种的基准数据集，借助半自动标注方法增强数据多样性，并验证了多数据源联合训练可显著提升跨域泛化能力（如mAP50-95从40.2%提升至83.9%），从而推动鲁棒性强、适应广泛的叶级分割模型发展，助力可持续精准农业实践。

链接: https://arxiv.org/abs/2605.03784
作者: Robert Martinko,Daniel Steininger,Julia Simon,Andreas Trondl,Matthias Blaickner
机构: AIT Austrian Institute of Technology, Center for Vision, Automation Control; University of Applied Sciences Technikum Wien, Computer Science Applied Mathematics
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at Conference on Computer Vision and Pattern Recognition (CVPR) Workshops 2026. Code and dataset available at this https URL

点击查看摘要

Abstract:Rising global food demand and growing climate pressure increase the need for sustainable, precise agricultural practices. Automated, individualized plant treatment relies on fine-grained visual analysis, yet leaf-level segmentation remains underexplored despite its value for assessing crop health, growth dynamics, yield potential and localized stress symptoms. Progress is limited by a lack of dedicated datasets, especially regarding species coverage, and by the absence of systematic evaluations of modern instance-segmentation architectures for this task. We address these gaps by surveying current data and identifying four suitable, publicly available leaf-segmentation datasets. Using them, we compare one-stage, two-stage and Transformer-based detectors and identify a YOLO26 model configuration to provide the best trade-off for real-world precision-agriculture tasks. Extensive cross-domain generalization experiments reveal substantial performance drops across plant species and recording setups, especially for models trained solely on laboratory data. To strengthen data availability, we introduce a new benchmark dataset with leaf-level masks for 23 plant species, created via semi-automatic annotation of selected CropAndWeed images. A model trained on all four existing datasets achieves a mean mAP50-95 of 83.9% across their corresponding test sets and 40.2% on our new benchmark, demonstrating improved generalization and highlighting the need for diverse leaf-segmentation datasets in robust precision agriculture.

[CV-25] GeoTopoDiff: Learning Geometry–Topology Graph Priors through Boundary-Constrained Mixed Diffusion for Sparse-Slice 3D Porous Reconstruction

【速读】：该论文旨在解决基于扩散模型的三维多孔微结构重建中，如何在有限观测条件下同时保持孔隙形态连续性与孔喉拓扑离散性的难题。传统方法依赖全分辨率CT扫描以提供拓扑忠实的先验信息，导致工业应用中存在吞吐量、拓扑保真度和视场之间的固有权衡。其解决方案的关键在于提出GeoTopoDiff框架，通过将扩散先验的学习从体素空间转移到混合图状态空间（mixed graph state space），该空间能够联合表征连续孔隙几何与离散孔喉拓扑；并引入基于稀疏CT切片的拓扑感知部分图先验，约束反向去噪过程，从而在稀疏观测下显著降低后验不确定性，提升重建精度。

链接: https://arxiv.org/abs/2605.03764
作者: Yue Shi,Peng Wang,Mingzhe Yu,Yunlong Zhao,Li Liu,Gareth D Hatton,Yan Lyu,Liangxiu Han
机构: Manchester Metropolitan University (曼彻斯特都会大学); University of Surrey (萨里大学); Johnson Matthey (庄信万丰); Imperial College London (帝国理工学院)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Diffusion-based voxel prior modelling is challenging for the reconstruction of large-scale 3D porous microstructures. Due to the demanding requirements for simultaneously modelling both the continuous pore morphology and the discrete pore-throat topology, the diffusion models require fully observed CT scans to provide topology-faithful priors, which results in an inherent trade-off among throughput, topological fidelity, and field of view in practical industrial applications. We propose GeoTopoDiff, a graph diffusion-based framework for reconstructing 3D porous microstructures from sparse CT slices. GeoTopoDiff transfers the learning of diffusion priors from a voxel-based space to a mixed graph state space, which simultaneously encompasses continuous pore geometry and discrete pore-throat topology. A topology-aware partial graph prior from sparsely observed CT slices is introduced to constrain the reverse denoising process. Experiments on anisotropic PTFE and Fontainebleau sandstone show that GeoTopoDiff reduces morphology-related errors by 19.8% and topology-sensitive transport errors by 36.5% on average. Our findings suggest that the mixed graph state space promotes the diffusion denoising process to reduce posterior uncertainty under a sparse observations. All models and code have been made publicly available to facilitate the exploration of diffusion models in the field of 3D porous microstructures simulation.

[CV-26] Before Forgetting Learn to Remember: Revisiting Foundational Learning Failures in LVLM Unlearning Benchmarks ACL2026

【速读】：该论文旨在解决大型视觉语言模型（Large Vision-Language Models, LVLMs）在隐私保护方面的关键问题：模型可能无意中记忆敏感个人信息，而现有去记忆（unlearning）评估基准因未能有效初始化记忆（即存在“欠记忆”现象）而导致评估结果不可靠。其解决方案的核心是提出ReMem——一个可靠的多跳、多图像记忆基准，通过受控的数据规模扩展、推理感知的问答对设计以及多样化的视觉场景来确保基础学习的稳健性；同时引入新颖的暴露度（Exposure）指标，量化信息从模型内部概率分布中被擦除的深度，从而为LVLMs的学习与去记忆行为提供严谨且可信的诊断框架。

链接: https://arxiv.org/abs/2605.03759
作者: JuneHyoung Kwon,MiHyeon Kim,Eunju Lee,JungMin Yun,Byeonggeuk Lim,YoungBin Kim
机构: Chung-Ang University (中央大学); KT Corporation (KT公司)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accepted to Findings of ACL 2026

点击查看摘要

Abstract:While Large Vision-Language Models (LVLMs) offer powerful capabilities, they pose privacy risks by unintentionally memorizing sensitive personal information. Current unlearning benchmarks attempt to mitigate this using fictitious identities but overlook a critical stage 1 failure: models fail to effectively memorize target information initially, rendering subsequent unlearning evaluations unreliable. Diagnosing under-memorization and the multi-hop curse as root causes, we introduce ReMem, a Reliable Multi-hop and Multi-image Memorization Benchmark. ReMem ensures robust foundational learning through principled data scaling, reasoning-aware QA pairs, and diverse visual contexts. Additionally, we propose a novel Exposure metric to quantify the depth of information erasure from the model’s internal probability distribution. Extensive experiments demonstrate that ReMem provides a rigorous and trustworthy framework for diagnosing both learning and unlearning behaviors in LVLMs.

[CV-27] FluxFlow: Conservative Flow-Matching for Astronomical Image Super-Resolution

【速读】：该论文旨在解决地基天文超分辨率（Ground-to-space astronomical super-resolution）问题，即从受像素采样分辨率和大气湍流（atmospheric seeing）双重限制的地面观测图像中恢复出具有空间质量的图像。现有方法依赖于合成训练数据对，难以捕捉真实大气统计特性，易导致重建结果过平滑或产生无物理依据的幻觉源。其解决方案的关键在于提出FluxFlow框架——一种保守的像素空间流匹配（flow-matching）方法，在训练过程中引入观测不确定性建模与源区域重要性权重，同时在测试阶段采用无需训练的维纳正则化修正策略，有效抑制幻觉源并保留细节信息。

链接: https://arxiv.org/abs/2605.03749
作者: Shuhong Liu,Xining Ge,Ziteng Cui,Liuzhuozheng Li,Gengjia Chang,Jun Liu,Ziying Gu,Dong Li,Xuangeng Chu,Lin Gu,Tatsuya Harada
机构: The University of Tokyo (东京大学); I2WM (I2WM); Tohoku University (东北大学); RIKEN AIP (理化学研究所先进智能项目)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Ground-to-space astronomical super-resolution requires recovering space-quality images from ground-based observations that are simultaneously limited by pixel sampling resolution and atmospheric seeing, which imposes a stochastic, spatially varying PSF that cannot be resolved through upsampling alone. Existing methods rely on synthetic training pairs that fail to capture real atmospheric statistics and are prone to either over-smoothed reconstructions or hallucination sources with no physical counterpart in the observed sky. We propose FluxFlow, a conservative pixel-space flow-matching framework that incorporates observation uncertainty and source-region importance weights during training, and a training-free Wiener-regularized test-time correction to suppress hallucination sources while preserving recovered detail. We further construct the DESI–HST Dataset, the large-scale real-world benchmark comprising 19,500 real co-registered ground-to-space image pairs with real atmospheric PSF variation. Experiments demonstrate that FluxFlow consistently outperforms existing baseline methods in both photometric and scientific accuracy.

[CV-28] Unified Multimodal Visual Tracking with Dual Mixture-of-Experts ICML2026

【速读】：该论文旨在解决多模态视觉目标跟踪（Multimodal Visual Object Tracking）中因不同模态（如RGB、RGB+X）需独立建模而导致的效率低、可扩展性差及泛化能力弱的问题。现有方法通常为每种模态训练专用模型或依赖预训练模型迁移，难以实现统一高效的端到端训练。解决方案的关键在于提出OneTrackerV2框架，其核心创新包括：1）Meta Merger模块将多模态信息嵌入统一空间，支持灵活融合与鲁棒性增强；2）Dual Mixture-of-Experts（DMoE）结构中，T-MoE建模时空关系，M-MoE解耦跨模态依赖并缓解特征冲突，从而在共享架构与参数下实现多模态跟踪的统一训练与推理，显著提升性能与实用性。

链接: https://arxiv.org/abs/2605.03716
作者: Lingyi Hong,Jinglun Li,Xinyu Zhou,Kaixun Jiang,Pinxue Guo,Zhaoyu Chen,Runze Li,Xingdong Sheng,Wenqiang Zhang
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: OneTrackerV2. Accepted by ICML 2026

点击查看摘要

Abstract:Multimodal visual object tracking can be divided into to several kinds of tasks (e.g. RGB and RGB+X tracking), based on the input modality. Existing methods often train separate models for each modality or rely on pretrained models to adapt to new modalities, which limits efficiency, scalability, and usability. Thus, we introduce OneTrackerV2, a unified multi-modal tracking framework that enables end-to-end training for any modality. We propose Meta Merger to embed multi-modal information into a unified space, allowing flexible modality fusion and robustness. We further introduce Dual Mixture-of-Experts (DMoE): T-MoE models spatio-temporal relations for tracking, while M-MoE embeds multi-modal knowledge, disentangling cross-modal dependencies and reducing feature conflicts. With a shared architecture, unified parameters, and a single end-to-end training, OneTrackerV2 achieves state-of-the-art performance across five RGB and RGB+X tracking tasks and 12 benchmarks, while maintaining high inference efficiency. Notably, even after model compression, OneTrackerV2 retains strong performance. Moreover, OneTrackerV2 demonstrates remarkable robustness under modality-missing scenarios.

[CV-29] From Code to Prediction: Fine-Tuning LLM s for Neural Network Performance Classification in NNGPT

【速读】：该论文旨在解决当前基于大语言模型（Large Language Models, LLMs）的自动化机器学习（AutoML）框架中，LLMs是否具备跨数据集推理神经网络性能能力的问题。现有方法多聚焦于生成式输出（如超参数或网络结构代码），并通过训练所生成模型来评估效果，但缺乏对LLM自身能否理解并预测不同数据集上神经网络表现差异的研究。解决方案的关键在于构建一个分类任务嵌入到NNGPT框架中，通过微调LLM来判断给定神经网络架构在两个图像分类数据集中哪个表现更优；其中采用LEMUR数据集提供标准化、可复现的PyTorch实现与性能指标，并设计三种难度递增的提示（prompt）配置：包含归一化准确率的基线、使用数据集属性替代准确率的元数据增强提示，以及仅提供架构源码和数据集名称的纯代码提示。实验表明，在LoRA微调下，纯代码提示可达80%峰值准确率，显著优于元数据提示（70%），且分析揭示代码本身蕴含比元数据更强的判别性信号，证明LLM可通过代码理解实现跨数据集的架构适用性推理。

链接: https://arxiv.org/abs/2605.03686
作者: Mahmoud Hanouneh,Radu Timofte,Dmitry Ignatov
机构: Computer Vision Lab, CAIDAS IFI, University of Würzburg (维尔茨堡大学)
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Automated Machine Learning (AutoML) frameworks increasingly leverage Large Language Models (LLMs) for tasks such as hyperparameter optimization and neural architecture code generation. However, current LLM-based approaches focus on generative outputs and evaluate them by training the produced artifacts. Whether LLMs can learn to reason about neural network performance across datasets remains underexplored. We present a classification task integrated into the NNGPT framework, in which a fine-tuned LLM predicts which of two image classification datasets a given neural network architecture achieves higher accuracy on. The task is built on the LEMUR dataset, which provides standardized PyTorch implementations with reproducible performance metrics. Three prompt configurations of increasing difficulty are evaluated: a normalized-accuracy baseline (trivially reaching 100%), a metadata-enriched prompt replacing accuracies with dataset properties, and a code-only prompt presenting only architecture source code and dataset names. Using DeepSeek-Coder-7B-Instruct fine-tuned with LoRA, the code-only prompt reaches 80% peak accuracy over 15 epochs, while the metadata prompt peaks at 70%. Perdataset analysis reveals complementary strengths: metadata excels for datasets with distinctive properties (CelebAGender at 90.9%) but degrades for overlapping characteristics, whereas the code-only prompt shows more balanced performance. A comparison with DeepSeek-Coder1.3B confirms that model capacity affects this form of architectural reasoning. The results establish that LLMs can be fine-tuned to predict cross-dataset suitability from neural network code, suggesting that architecture source code contains richer discriminative signal than dataset metadata alone.

[CV-30] Real Image Denoising with Knowledge Distillation for High-Performance Mobile NPUs

【速读】：该论文旨在解决深度学习图像复原模型在移动神经处理单元（Neural Processing Unit, NPU）上部署时面临的两大瓶颈问题：操作不兼容性（operator incompatibility）和内存访问开销（memory-access overhead）。其解决方案的核心在于提出一种面向NPU的软硬件协同设计方法，通过知识蒸馏构建一个轻量级学生网络（LiteDenoiseNet），该网络专为现代移动系统级芯片（SoC）的分块内存架构优化，并严格使用NPU原生算子（如标准3×3卷积、ReLU激活和最近邻上采样）。同时，采用渐进式上下文扩展策略（最大支持1024×1024图像裁片）以提升恢复质量。实验表明，该方案在保持高保真度（PSNR达37.58 dB）的同时，推理速度显著优于集成GPU（最高快3.88倍），并在参数减少21.2倍的情况下保留99.8%的教师模型性能，验证了硬件感知的知识蒸馏是实现高性能与高效部署统一的关键路径。

链接: https://arxiv.org/abs/2605.03680
作者: Faraz Kayani,Sarmad Kayani,Asad Ahmed,Radu Timofte,Dmitry Ignatov
机构: University of Würzburg (维尔茨堡大学)
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:

点击查看摘要

Abstract:While deep-learning-based image restoration has achieved unprecedented fidelity, deployment on mobile Neural Processing Units (NPUs) remains bottlenecked by operator incompatibility and memory-access overhead. We propose an NPU-aware hardware-algorithm co-design approach for real-world image denoising on mobile NPUs. Our approach employs a high-capacity teacher to supervise a lightweight student network specifically designed to leverage the tiled-memory architectures of modern mobile SoCs. By prioritizing NPU-native primitives – standard 3x3 convolutions, ReLU activations, and nearest-neighbor upsampling – and employing a progressive context expansion strategy (up to 1024x1024 crops), the model achieves 37.66 dB PSNR / 0.9278 SSIM on the validation benchmark and 37.58 dB PSNR / 0.9098 SSIM on the held-out test benchmark at full resolution (2432x3200) in the Mobile AI 2026 challenge. Following the official challenge rules, the inference runtime is measured under a standardized Full HD (1088x1920) protocol, where it runs in 34.0 ms on the MediaTek Dimensity 9500 and 46.1 ms on the Qualcomm Snapdragon 8 Elite NPU. We further reveal an “Inference Inversion” effect, where strict adherence to NPU-compatible operations enables dedicated NPU execution up to 3.88x faster than the integrated mobile GPU. The 1.96M-parameter student recovers 99.8% of the teacher’s restoration quality via high-alpha knowledge distillation (alpha = 0.9), achieving a 21.2x parameter reduction while closing the PSNR gap from 1.63 dB to only 0.05 dB. These results establish hardware-aware distillation as an effective strategy for unifying high-fidelity denoising with practical deployment across diverse mobile NPU architectures. The proposed lightweight student model (LiteDenoiseNet) and its training statistics are provided in the NN Dataset, available at this https URL.

[CV-31] AniMatrix: An Anime Video Generation Model that Thinks in Art Not Physics

【速读】：该论文旨在解决生成式视频模型在创作动画（Anime）时因过度依赖物理真实性先验而导致艺术表现力丧失或风格崩溃的问题。现有模型通常以物理规律为优化目标，但动画本身违背现实物理规则（如模糊帧、夸张动作和角色变形等），且包含大量并行的艺术惯例，难以被统一建模为单一“动画物理”。解决方案的关键在于构建一个以艺术性为核心导向的双通道条件机制：首先通过生产知识系统（Production Knowledge System）将动画要素结构化为可控制的变量（Style, Motion, Camera, VFX），并利用AniCaption从图像中推断这些导演指令；其次采用风格-运动-形变的渐进式训练课程，逐步从近似物理运动过渡到完全动画表达；最后引入基于领域特定奖励模型的形变感知偏好优化策略，区分有意的艺术表达与病理性的生成失败。该方法使模型在专业动画师主导的人类评估中，在Prompt Understanding和Artistic Motion等维度显著优于基线模型。

链接: https://arxiv.org/abs/2605.03652
作者: Tencent HY Team
机构: Tencent HY Team (腾讯HY团队)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 37 pages, 1 main figure (qualitative comparison), 1 TikZ architecture diagram; technical report. Model weights and inference code to be released

点击查看摘要

Abstract:Video generation models internalize physical realism as their prior. Anime deliberately violates physics: smears, impact frames, chibi shifts; and its thousands of coexisting artistic conventions yield no single “physics of anime” a model can absorb. Physics-biased models therefore flatten the artistry that defines the medium or collapse under its stylistic variance. We present AniMatrix, a video generation model that targets artistic rather than physical correctness through a dual-channel conditioning mechanism and a three-step transition: redefine correctness, override the physics prior, and distinguish art from failure. First, a Production Knowledge System encodes anime as a structured taxonomy of controllable production variables (Style, Motion, Camera, VFX), and AniCaption infers these variables from pixels as directorial directives. A trainable tag encoder preserves the field-value structure of this taxonomy while a frozen T5 encoder handles free-form narrative; dual-path injection (cross-attention for fine-grained control, AdaLN modulation for global enforcement) ensures categorical directives are never diluted by open-ended text. Second, a style-motion-deformation curriculum transitions the model from near-physical motion to full anime expressiveness. Third, deformation-aware preference optimization with a domain-specific reward model separates intentional artistry from pathological collapse. On an anime-specific human evaluation with five production dimensions scored by professional animators, AniMatrix ranks first on four of five, with the largest gains over Seedance-Pro 1.0 on Prompt Understanding (+0.70, +22.4 percent) and Artistic Motion (+0.55, +16.9 percent). We will publicly release the AniMatrix model weights and inference code.

[CV-32] Rethinking Temporal Consistency in Video Object-Centric Learning: From Prediction to Correspondence

【速读】：该论文旨在解决视频对象中心学习（video object-centric learning）中通过学习动态模块预测未来对象表示（slots）以维持时序一致性的方法所存在的计算开销高、效率低的问题。现有方法依赖于复杂且昂贵的时序预测机制，而忽略了预训练视觉骨干网络已具备区分不同实例的特征表达能力。解决方案的关键在于提出“Grounded Correspondence”框架，其核心思想是利用冻结骨干网络中的显著区域初始化slot，并采用确定性的二分图匹配（Hungarian matching）替代原有的可学习过渡函数，从而实现帧间对象身份的一致性维护。该方法在不引入任何时序建模参数的情况下，在MOVi-D、MOVi-E和YouTube-VIS等多个基准上达到了具有竞争力的性能表现。

链接: https://arxiv.org/abs/2605.03650
作者: Zhiyuan Li,Rongzhen Zhao,Wenyan Yang,Wenshuai Zhao,Pekka Marttinen,Joni Pajarinen
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:

点击查看摘要

Abstract:The de facto approach in video object-centric learning maintains temporal consistency through learned dynamics modules that predict future object representations, called slots. We demonstrate that these predictors function as expensive approximations of discrete correspondence problems. Modern self-supervised vision backbones already encode instance-discriminative features that distinguish objects reliably. Exploiting these features eliminates the need for learned temporal prediction. We introduce Grounded Correspondence, a framework that replaces learned transition functions with deterministic bipartite matching. Slots initialize from salient regions in frozen backbone features. Frame-to-frame identity is maintained through Hungarian matching on slot representations. The approach requires zero learnable parameters for temporal modeling yet achieves competitive performance on MOVi-D, MOVi-E, and YouTube-VIS. Project page: this https URL

[CV-33] he Detector Teaches Itself: Lightweight Self-Supervised Adaptation for Open-Vocabulary Object Detection ICPR2026

【速读】：该论文旨在解决开放词汇目标检测（Open-vocabulary object detection）中视觉语言模型（VLMs）在区域级检测任务上因缺乏局部细节感知能力而导致的性能瓶颈问题。现有协同范式虽能实现零样本识别新类别物体，但预训练于全图的VLM难以捕捉细粒度区域特征，限制了其在检测任务中的表现。解决方案的关键在于提出解耦自适应训练（Decoupled Adaptivity Training, DAT），通过构建基于封闭集检测器生成的区域感知伪标签数据集，对VLM的视觉主干进行解耦式微调：在增强局部特征对齐的同时，利用权重插值保留全局语义知识，从而提升模型在新旧类别上的检测性能。DAT为即插即用模块，仅需微调少于0.8M参数且无推理开销，显著优于现有方法。

链接: https://arxiv.org/abs/2605.03642
作者: Yazhe Wan,Changjae Oh(Queen Mary University of London)
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 15 pages; 4 figures; Accepted to ICPR 2026; Code is available at this https URL

点击查看摘要

Abstract:Open-vocabulary object detection aims to recognize objects from an open set of categories, which leverages vision-language models (VLMs) pre-trained on large-scale image-text data. The cooperative paradigm combines an object detector with a VLM to achieve zero-shot recognition of novel objects. However, VLMs pre-trained on full images often struggle to capture local object details, limiting their effectiveness when applied to region-level detection. We present Decoupled Adaptivity Training (DAT), a self-supervised fine-tuning approach to improve VLMs for cooperative model-based object detection. Given a cooperative model consists of a closed-set detector and a VLM, we first construct a region-aware pseudo-labeled dataset using a pre-trained closed-set object detector, in which regions corresponding to novel objects may be present but remain unlabeled or mislabeled. We then fine-tune the visual backbone of the VLM in a decoupled manner, which enhances local feature alignment while preserving global semantic knowledge via weight interpolation. DAT is a plug-and-play module that requires no inference overhead and fine-tunes less than 0.8M parameters. Experiments on the COCO and LVIS datasets show that DAT consistently improves detection performance on both novel and known categories, establishing a new state of the art in cooperative open-vocabulary detection.

[CV-34] Diffusion Masked Pretraining for Dynamic Point Cloud

【速读】：该论文旨在解决动态点云预训练中基于掩码重建目标的两大局限性：一是现有方法将真实轨迹中心作为解码器的位置嵌入，导致时空位置信息泄露；二是通过确定性代理目标监督帧间运动，忽略了多模态轨迹不确定性，从而系统性地丢失分布结构。解决方案的关键在于提出扩散掩码预训练（Diffusion Masked Pretraining, DiMP），其核心创新为：首先仅对掩码轨迹中心施加前向扩散噪声，并从可见时空上下文中预测干净中心，消除位置泄露同时保留可见坐标作为时间锚点；其次将点级帧间位移监督重构为条件于解码表示的DDPM噪声预测目标，使编码器学习潜在运动的完整条件分布，而非退化为单一确定性估计，从而实现更鲁棒和富有表现力的动态点云表征学习。

链接: https://arxiv.org/abs/2605.03639
作者: Zhuoyue Zhang,Jihua Zhu,Chaowei Fang,Jian Liu,Ajmal Saeed Mian
机构: Xi’an Jiaotong University (西安交通大学); Singapore University of Technology and Design (新加坡科技设计大学); University of Western Australia (西澳大利亚大学)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Dynamic point cloud pretraining is still dominated by masked reconstruction objectives. However, these objectives inherit two key limitations. Existing methods inject ground-truth tube centers as decoder positional embeddings, causing spatio-temporal positional leakage. Moreover, they supervise inter-frame motion with deterministic proxy targets that systematically discard distributional structure by collapsing multimodal trajectory uncertainty into conditional means. To address these limitations, we propose Diffusion Masked Pretraining (DiMP), a unified self-supervised framework for dynamic point clouds. DiMP introduces diffusion modeling into both positional inference and motion learning. It first applies forward diffusion noise only to masked tube centers, then predicts clean centers from visible spatio-temporal context. This removes positional leakage while preserving visible coordinates as clean temporal anchors. DiMP also reformulates point-wise inter-frame displacement supervision as a DDPM noise-prediction objective conditioned on decoded representations. This design drives the encoder to target the full conditional distribution of plausible motions under a variational surrogate, rather than collapsing to a single deterministic estimate. Extensive experiments demonstrate that DiMP consistently improves downstream accuracy over the backbone alone, with absolute gains of 11.21% on offline action segmentation and 13.65% under causally constrained online this http URL are available at this https URL.

[CV-35] RPBA-Net: An Interpretable Residual Pyramid Bilateral Affine Network for RAW-Domain ISP Enhancement

【速读】：该论文旨在解决RAW域图像信号处理（ISP）中模块碎片化、映射不可解释以及部署约束等问题，特别是在去马赛克（demosaicing）、色彩校正和细节增强环节。其核心解决方案是提出RPBA-Net——一种可解释的残差金字塔双边仿射网络，通过估计基础RGB表示并学习身份引导的残差仿射校正来统一去马赛克与增强过程；同时构建金字塔双边仿射网格，并结合引导驱动的自回归自适应切片与跨层自适应融合机制，以分层建模全局色调恢复与局部纹理增强。此外，引入平滑性、跨尺度一致性及幅值正则化项提升模型稳定性、可控性和结构可解释性。

链接: https://arxiv.org/abs/2605.03626
作者: Yucheng Xin,Wu Chen,Xiang Chen,Guangwei Gao,Xinchun Wang,Ruize Wu,Dianjie Lu,Guijuan Zhang,Linwei Fan,Zhuoran Zheng
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:To address module fragmentation, uninterpretable mappings, and deployment constraints in RAW-domain demosaicing, color correction, and detail enhancement, this paper proposes RPBA-Net, an interpretable residual pyramid bilateral affine network for RAW-domain ISP enhancement. Given packed RAW as input, the method performs residual affine base reconstruction by estimating a base RGB representation and learning identity-guided residual affine corrections, thereby unifying demosaicing and enhancement. It further builds pyramid bilateral affine grids and combines guide-driven autoregressive adaptive slicing with adaptive cross-layer fusion to hierarchically model global tone restoration and local texture enhancement. In addition, smoothness, cross-scale consistency, and magnitude regularization terms are introduced to improve model stability, controllability, and structural interpretability. Extensive experiments demonstrate that RPBA-Net surpasses representative RAW-to-sRGB methods and achieves state-of-the-art performance in reconstruction fidelity and perceptual quality, while maintaining low model complexity and strong deployment potential for mobile and embedded platforms.

[CV-36] PriorNet: Prior-Guided Engagement Estimation from Face Video

【速读】：该论文旨在解决人脸视频中参与度（engagement）估计的挑战，主要问题包括面部证据不完整、标注数据有限以及参与度标注具有主观性。解决方案的关键在于提出 PriorNet 框架，通过在预处理、模型适配和目标函数设计三个阶段注入任务相关的先验知识：1）将人脸检测失败转化为显式的零帧占位符以保留缺失面部事件的信息；2）利用 Prior-guided Low-Rank Adaptation（Prior-LoRA）模块对冻结的自监督视频面部情感感知器（Self-supervised Video Facial Affect Perceiver, SVFAP）进行参数高效微调；3）在硬标签监督下采用 Dirichlet-evidential 不确定性加权目标函数进行训练。实验证明，该方法在 EngageNet、DAiSEE、DREAMS 和 PAFE 数据集上均优于现有基线，并且消融实验表明各阶段先验的互补贡献是性能提升的核心原因。

链接: https://arxiv.org/abs/2605.03615
作者: Alexander Vedernikov
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Engagement estimation from face video remains challenging because facial evidence is often incomplete, labeled data are limited, and engagement annotations are subjective. We present PriorNet, a prior-guided framework that injects task-relevant priors at three stages of the pipeline: preprocessing, model adaptation, and objective design. PriorNet converts face-detection failures into explicit zero-frame placeholders so that missing-face events remain represented in the input sequence, adapts a frozen Self-supervised Video Facial Affect Perceiver (SVFAP) backbone through a Prior-guided Low-Rank Adaptation module (Prior-LoRA) for parameter-efficient specialization, and trains with a Dirichlet-evidential, uncertainty-weighted objective under hard-label supervision. We evaluate PriorNet on EngageNet, DAiSEE, DREAMS, and PAFE using each dataset’s native evaluation protocol. Across these benchmarks, PriorNet improves over the strongest listed prior reference within each dataset’s evaluation framing, while component ablations on EngageNet and DAiSEE indicate that the gains arise from complementary contributions of preprocessing, adaptation, and objective-level priors. These results support explicit prior injection as a useful design principle for face-video engagement estimation under the benchmark conditions studied in this work.

[CV-37] Uncertainty Estimation in Instance Segmentation of Affordances via Bayesian Visual Transformers

【速读】：该论文旨在解决视觉可操作性（visual affordance）区域的精准定位问题，即在图像中识别出具有潜在交互可能性的局部区域，而非仅生成一般性的显著性图（saliency map），以支持自主机器人更自然地与环境互动、增强人机协作及提升增强现实和假体视觉系统的性能。其解决方案的关键在于提出一种基于贝叶斯（Bayesian）方法的实例分割模型，结合样本采样与集成策略进行不确定性估计，并采用注意力机制改进网络架构；通过分析不同检测结果的分布，提取像素级的主观不确定性（epistemic variance）和客观不确定性（aleatoric variance），同时引入概率掩码质量（Probability-based Mask Quality）指标实现对语义与空间变异性的综合评估。实验表明，该方法在IIT-Aff数据集上使Fβ^w分数提升7.4个百分点，且模型校准度更好、预测更稳健，同时具备更高的可解释性。

链接: https://arxiv.org/abs/2605.03614
作者: Lorenzo Mur-Labadia,Ruben Martinez-Cantina,Jose J.Guerrero
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Visual affordances identify regions in an image with potential interactions, offering a novel paradigm for scene understanding. Recognizing affordances allows autonomous robots to act more naturally, could enhance human-robot interactions, enrich augmented reality systems, and benefit prosthetic vision devices. Accurate and localized prediction of affordance regions, rather than general saliency maps is crucial for these applications. We present a model for instance segmentation of affordances by adopting sample-based and ensembles approaches for uncertainty estimation. We extend an attention-based architecture for our novel task, showing with detailed ablation experiments the effects of each component. By comparing the distribution of these different detections, we extract pixel-wise epistemic and aleatoric variances at both the semantic and spatial levels. In addition, we propose a novel measure called Probability-based Mask Quality, which enables a comprehensive analysis of semantic and spatial variations in a probabilistic instance segmentation model. Our results show that the global consensus of multiple sub-networks of Bayesian models improve deterministic networks due to a better mask refinement and generalization. This fact, joined with the more powerful features extracted by attention-based mechanisms, represent an improvement of +7.4 p.p on the F_\beta^w score in the challenging IIT-Aff dataset. Bayesian models are also better calibrated, producing less overconfident probabilities and with a better uncertainty estimation. Qualitative results show that aleatoric variance appears in the contour of the objects, while the epistemic variance is observed in visual challenging pixels, adding interpretability to the neural network.

[CV-38] deSEO: Physics-Aware Dataset Creation for High-Resolution Satellite Image Shadow Removal

【速读】：该论文旨在解决高分辨率卫星遥感图像中阴影问题，即地形和高大结构产生的阴影严重影响分类、目标检测与三维重建等任务的性能。现有公开数据集缺乏几何一致性的成对阴影/无阴影卫星图像，且多数地球观测数据集仅适用于阴影检测或三维建模，而非阴影去除。解决方案的关键在于提出 deSEO 方法：首先基于 S-EO 阴影检测数据集，通过可复现的全流程自动构建几何一致的成对阴影/无阴影卫星图像；其次采用时间与几何滤波、Jacobian 基础的方向归一化及 LoFTR-RANSAC 精准配准技术实现像素级对齐，并引入逐像素有效性掩膜限制学习区域以应对残余非正射偏移（off-nadir parallax）；最后开发了一种 DSM-aware 的去阴影模型，融合残差平移、感知损失和掩膜约束的对抗学习机制，显著优于直接迁移无人机场景下 SRNet/pix2pix 架构的效果。deSEO 为卫星遥感中的阴影去除提供了首个可复现、几何感知的成对数据集与基准模型。

链接: https://arxiv.org/abs/2605.03610
作者: Lorenzo Beltrame,Jules Salzinger,Filip Svoboda,Phillipp Fanta-Jende,Jasmin Lampert,Radu Timofte,Marco Körner
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV)
备注: 8 pages, 6 figures, 5 tables. Accepted in the annals track at the ISPRS 2026 Congress. Code and materials: this https URL

点击查看摘要

Abstract:Shadows cast by terrain and tall structures remain a major obstacle for high-resolution satellite image analysis, degrading classification, detection, and 3D reconstruction performance. Public resources offering geometry-consistent paired shadow/shadow-free satellite imagery are essentially missing, and most Earth-observation datasets are designed for shadow detection or 3D modelling rather than removal. Existing deep shadow-removal datasets either target ground-level or aerial scenes or rely on unpaired and weakly supervised formulations rather than explicit satellite pairs. We address this gap with deSEO, a geometry-aware and physics-informed methodology that, to the best of our knowledge, is the first to derive paired supervision for satellite shadow removal from the S-EO shadow detection dataset through a fully replicable pipeline. For each tile, deSEO selects a minimally shadowed acquisition as a weak reference and pairs it with shadowed counterparts using temporal and geometric filtering, Jacobian-based orientation normalisation, and LoFTR-RANSAC registration. A per-pixel validity mask restricts learning to reliably aligned regions, enabling supervision despite residual off-nadir parallax. In addition to this paired dataset, we develop a DSM-aware deshadowing model that combines residual translation, perceptual objectives, and mask-constrained adversarial learning. In contrast, a direct adaptation of a UAV-based SRNet/pix2pix architecture fails to converge under satellite viewpoint variability. Our model consistently reduces the visual impact of cast shadows across diverse illumination and viewing conditions, achieving improved structural and perceptual fidelity on held-out scenes. deSEO therefore provides the first reproducible, geometry-aware paired dataset and baseline for shadow removal in satellite Earth observation.

[CV-39] MILE: Mixture of Incremental LoRA Experts for Continual Semantic Segmentation across Domains and Modalities ICPR2026

【速读】：该论文旨在解决持续语义分割（continual semantic segmentation）中模型在适应新领域或模态时出现的灾难性遗忘问题，即模型在学习新任务时会损害对先前任务的性能。解决方案的关键在于提出一种名为Mixture of Incremental LoRA Experts (MILE) 的模块化、参数高效的框架，其核心创新是利用低秩适配（Low-Rank Adaptation, LoRA）为每个新任务实例化轻量级专家模块，同时冻结预训练基础网络，确保各专家仅在其专属任务数据上训练，从而避免对已有知识的覆盖；此外，通过原型引导的门控机制在推理阶段动态选择最合适的专家，实现高稳定性、可扩展性和高效性。

链接: https://arxiv.org/abs/2605.03555
作者: Shishir Muralidhara,Didier Stricker,René Schuster
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at ICPR 2026

点击查看摘要

Abstract:Continual semantic segmentation requires models to adapt to new domains or modalities without sacrificing performance on previously learned tasks. Expert-based learning, in which task-specific modules specialize in different domains, has proven effective in mitigating forgetting. These methods include dynamic expansion, which suffers from scalability issues, or parameter isolation, which constrains the ability to learn new tasks. We introduce Mixture of Incremental LoRA Experts (MILE), a modular and parameter-efficient framework for continual segmentation across both domains and modalities. MILE leverages Low-Rank Adaptation (LoRA) to instantiate lightweight experts for each new task while keeping the pretrained base network frozen. Each expert is trained exclusively on its task data, thus avoids overwriting previously learned information. A prototype-guided gating mechanism dynamically selects the most appropriate expert at inference. MILE achieves the benefits of expert-based learning while overcoming its scalability limitations. It requires only a marginal parameter increase per task and tens of LoRA adapters are needed before matching the size of a single full model, making it highly efficient in both training and storage. Across domain- and modality-incremental benchmarks, MILE achieves strong performance while ensuring better stability, plasticity, and scalability.

[CV-40] Erase Persona Forget Lore: Benchmarking Multimodal Copyright Unlearning in Large Vision Language Models LREC2026

【速读】：该论文旨在解决大型视觉语言模型（Large Vision-Language Models, LVLMs）在训练过程中可能记忆并再生受版权保护的视觉内容（如角色和标志）所带来的法律与伦理风险，以及现有机器遗忘（Machine Unlearning）方法在多模态场景下评估效果不足的问题。解决方案的关键在于提出首个专门用于评估LVLM中版权内容遗忘能力的基准测试框架CoVUBench，其通过使用程序生成的合法合成数据及系统性视觉变化（包括构图调整和跨域表现多样性），确保评估的现实性和鲁棒性，并采用多模态评估协议从版权方视角衡量遗忘有效性、从部署方视角衡量模型通用性能的保持，从而实现对遗忘效果与模型功能之间权衡的标准化量化。

链接: https://arxiv.org/abs/2605.03547
作者: JuneHyoung Kwon,JungMin Yun,YoungBin Kim
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accepted to LREC 2026

点击查看摘要

Abstract:Large Vision-Language Models (LVLMs), trained on web-scale data, risk memorizing and regenerating copyrighted visual content such as characters and logos, creating significant challenges. Machine unlearning offers a path to mitigate these risks by removing specific content post-training, but evaluating its effectiveness, especially in the complex multimodal setting of LVLMs, remains an open problem. Current evaluation methods often lack robustness or fail to capture the nuances of cross-modal concept erasure. To address this critical gap, we introduce the CoVUBench benchmark, the first framework specifically designed for evaluating copyright content unlearning in LVLMs. CoVUBench utilizes procedurally generated, legally safe synthetic data coupled with systematic visual variations spanning compositional changes and diverse domain manifestations to ensure realistic and robust evaluation of unlearning generalization. Our comprehensive multimodal evaluation protocol assesses both forgetting efficacy from the copyright holder perspective and the preservation of general model utility from the deployer viewpoint. By rigorously measuring this crucial trade-off, CoVUBench provides a standardized tool to advance the development of responsible and effective unlearning methods for LVLMs.

[CV-41] DALPHIN: Benchmarking Digital Pathology AI Copilots Against Pathologists on an Open Multicentric Dataset

【速读】：该论文旨在解决当前生成式AI在数字病理学中缺乏独立基准测试的问题，以评估其作为病理医生辅助工具的潜力。解决方案的关键在于构建了DALPHIN——首个多中心开放基准平台，涵盖来自6个国家、14个亚专科的300例病例共1236张图像，覆盖130种从罕见到常见的诊断；同时引入31名不同专业水平的病理学家作为人类性能基准，并对三种AI模型（GPT-5、Gemini 2.5 Pro和PathChat+）进行序列与独立答题两种模式下的性能评估。结果表明，PathChat在六项任务中有四项达到专家水平，显著优于其他两个通用模型，验证了专用病理AI助手的有效性。

链接: https://arxiv.org/abs/2605.03544
作者: Carlijn Lems,Sander Moonemans,Natálie Klubíčková,Biagio Brattoli,Taebum Lee,Seokhwi Kim,Veronica Vilaplana,Laura Pons,Sapir Hochman,Mauricio Eduardo Suárez-Franck,Pedro Luis Fernandez,Julius Drachneris,Donatas Petroska,Renaldas Augulis,Arvydas Laurinavicius,Domingos Oliveira,Diana Montezuma,Anouk B. Bouwmeester,Dominique van Midden,Anne-Marie Vos,Shoko Vos,Jolique van Ipenburg,Maschenka Balkenhol,Koen Winkler,Iris Nagtegaal,Konnie Hebeda,Uta Flucke,Katrien Grünberg,Josef Skopal,Brinder S. Chohan,Jordi Temprana-Salvador,Enrico Munari,Luca Cima,Giulia Querzoli,Yosamin Gonzalez Belisario,Jaeike W. Faber,Geert J.L.H. van Leenders,Jan H. von der Thüsen,Lodewijk A.A. Brosens,Ronald R. de Krijger,Pieter Wesseling,Sandrine Florquin,Mateusz Maniewski,Adam Kowalewski,Robert Barna,Dina Tiniakos,Joan Lop Gros,Rogier Donders,Jake S.F. Maurits,Ming Yang Lu,Chengkuan Chen,Faisal Mahmood,Jeroen van der Laak,Nadieh Khalili,Frédérique Meeuwsen,Francesco Ciompi
机构: Radboud University Medical Center (奈梅亨大学医学中心); Biopticka Laboratory Ltd. (生物显微实验室有限公司); Charles University (查尔斯大学); Lunit Inc. (Lunit公司); Ajou University School of Medicine (延世大学医学院); Universitat Politècnica de Catalunya - BarcelonaTech (加泰罗尼亚理工大学-巴塞罗那技术学院); Hospital Universitari Germans Trias i Pujol (德国斯·特里亚斯和普霍尔大学医院); Universitat Autonoma de Barcelona (巴塞罗那自治大学); Vilnius University (维尔纽斯大学); National Centre of Pathology (国家病理学中心); IMP Diagnostics (IMP诊断研发部); Canisius Wilhelmina Ziekenhuis (卡尼修斯-威廉明娜医院); Erasmus University Medical Center (埃因霍温大学医学中心); University Medical Center Utrecht (乌得勒支大学医学中心); Princess Máxima Center for Pediatric Oncology (玛希玛儿童肿瘤中心); Amsterdam University Medical Center (阿姆斯特丹大学医学中心); University and Hospital Trust of Verona (维罗纳大学及医院信托); IRCCS Azienda Ospedaliero-Universitaria di Bologna (博洛尼亚大学附属医院研究中心); Bydgoszcz University of Science and Technology (比得哥什科技大学); Nicolaus Copernicus University in Toruń (哥白尼托伦大学); Victor Babes University of Medicine and Pharmacy (维克多·巴贝斯医科大学和药学院); ANAPATMOL Research Center (ANAPATMOL研究中心); National and Kapodistrian University of Athens (雅典国立卡波迪斯特里安大学); Newcastle University (纽卡斯尔大学); Hospital Clínic de Barcelona (巴塞罗那临床医院); Radboud University Medical Center (奈梅亨大学医学中心); Linköping University (林雪平大学); Brigham and Women’s Hospital (布里格姆妇女医院); Harvard Medical School (哈佛医学院)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Our dataset is available at this https URL , our code is available at this https URL , and our benchmark is available at this https URL

点击查看摘要

Abstract:Foundation models with visual question answering capabilities for digital pathology are emerging. Such unprecedented technology requires independent benchmarking to assess its potential in assisting pathologists in routine diagnostics. We created DALPHIN, the first multicentric open benchmark for pathology AI copilots, comprising 1236 images from 300 cases, spanning 130 rare to common diagnoses, 6 countries, and 14 subspecialties. The DALPHIN design and dataset are introduced alongside a human performance benchmark of 31 pathologists from 10 countries with varying expertise. We report results for two general-purpose (GPT-5, Gemini 2.5 Pro) and one pathology-specific copilot (PathChat+) for sequential and independent answer generation. We observed no statistically significant difference from expert-level performance in four of six tasks for PathChat, 2/6 tasks for Gemini, and 1/6 tasks for GPT. DALPHIN is publicly released with sequestered, indirectly accessible ground truth to foster robust and enduring benchmarking. Data, methods, and the evaluation platform are accessible through this http URL.

[CV-42] BFORE: Butterfly-Firefly Optimized Retinex Enhancement for Low-Light Image Quality Improvement

【速读】：该论文旨在解决低光照图像增强（Low-light Image Enhancement）问题，即在光照不足条件下获取的图像常存在可见度差、对比度低和色彩失真等问题。传统基于Retinex的方法依赖人工调参，难以适应多样化的光照场景。其解决方案的关键在于提出一种新型混合元启发式优化框架BFORE（Butterfly-Firefly Optimized Retinex Enhancement），通过蝴蝶优化算法（Butterfly Optimization Algorithm, BOA）与萤火虫算法（Firefly Algorithm, FA）协同优化多阶段Retinex增强流程中的参数：BOA用于优化多尺度Retinex带色彩恢复（MSRCR）参数，FA则优化自适应伽马校正加权分布（AGCWD）及去噪参数，并采用动态切换策略平衡全局探索与局部开发能力，从而实现无需训练数据的自动参数调优，显著提升图像质量指标如PSNR和SSIM。

链接: https://arxiv.org/abs/2605.03509
作者: Ahmed Cherif
机构: Sofrecom (Orange Group)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 28 pages, 10 tables, 3 figures. Submitted to ICCK Journal of Image Analysis and Processing (JIAP)

点击查看摘要

Abstract:Low-light image enhancement is a fundamental challenge in computer vision and multimedia applications, as images captured under insufficient illumination suffer from poor visibility, low contrast, and color distortion. Existing Retinex-based methods rely on manually tuned parameters that fail to generalize across diverse lighting conditions. This paper proposes BFORE (Butterfly-Firefly Optimized Retinex Enhancement), a novel hybrid metaheuristic-optimized framework that automatically tunes the parameters of a multi-stage Retinex-based pipeline. The proposed method converts the input image to HSV color space and applies Adaptive Gamma Correction with Weighted Distribution (AGCWD) to the luminance channel, followed by adaptive denoising. A Butterfly Optimization Algorithm (BOA) optimizes the Multi-Scale Retinex with Color Restoration (MSRCR) parameters, while a Firefly Algorithm (FA) optimizes the AGCWD and denoising parameters. A hybrid BOA-FA switching strategy dynamically balances global exploration and local exploitation. Experimental evaluation on the LOL benchmark dataset (15 paired test images) demonstrates that BFORE achieves the highest PSNR (17.22 dB) among all traditional enhancement methods, with 20.3% improvement over Histogram Equalization and 17.5% over MSRCR. BFORE produces the most naturally balanced mean brightness (129.97), closest to the ideal mid-tone value. Notably, BFORE outperforms RetinexNet – a deep learning baseline – in both PSNR (17.22 vs. 16.77 dB) and SSIM (0.5417 vs. 0.4252) without requiring any training data. The hybrid BOA-FA optimization contributes a 12.3% PSNR improvement and 14.8% SSIM improvement over the unoptimized pipeline.

[CV-43] Orientation-Aware Unsupervised Domain Adaptation for Brain Tumor Classification Across Multi-Modal MRI

【速读】：该论文旨在解决深度学习模型在神经肿瘤学中进行脑肿瘤诊断时面临的两大核心问题：一是专家标注的MRI数据稀缺，二是不同医疗机构间因扫描设备、成像协议和对比剂设置差异导致的显著域偏移（domain shift），这些问题严重制约了模型在真实场景中的泛化能力。解决方案的关键在于提出一种面向方向感知的无监督域自适应框架，通过两个核心技术实现：其一，利用具有大感受野的卷积神经网络（CNN）对输入MRI切片进行轴向（axial）、矢状面（sagittal）和冠状面（coronal）的分类，并为每种方向分别训练基于ResNet50骨干网络并附加四个全连接层的特征提取器，以实现方向特异性学习；其二，引入切片级无监督域自适应策略，将多模态源域（如T1、T2和FLAIR）的知识迁移至增强对比后的T1目标域，采用最大均值差异（Maximum Mean Discrepancy, MMD）损失进行特征层面对齐，并结合伪标签引导的适配机制以保留类别可分性，从而有效缓解标注稀缺与域差异带来的性能下降问题。

链接: https://arxiv.org/abs/2605.03490
作者: Sapna Sachan,Amulya Kumar Mahto,Prashant Wagambar Patil
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages

点击查看摘要

Abstract:The clinical integration of deep learning models for brain tumor diagnosis in neuro-oncology is severely constrained by limited expert-annotated MRI data and substantial inter-institutional domain shift arising from variations in scanners, imaging protocols, and contrast settings. These challenges significantly impair model generalization in real-world settings. To address this, we propose a novel orientation-aware unsupervised domain-adaptive framework for automated brain tumor classification using mixed 2D MRI slices. Initially, a CNN with large receptive field first categorizes input slices into axial, sagittal, and coronal views. For each orientation, a CNN architecture with ResNet50 backbone augmented with four fully connected layers is trained to extract discriminative features for tumor classification. To mitigate annotation scarcity and domain discrepancies, we introduce a slice-wise unsupervised domain adaptation strategy that transfers knowledge from the multi-modal such as T1, T2, and FLAIR source domain to the post-contrast T1 target domain. Feature-level alignment is enforced using maximum mean discrepancy loss, complemented by pseudo-label guided adaptation to preserve class discriminability. Extensive experiments demonstrate improved target-domain performance over prior approaches, highlighting the benefits of orientation-specific learning, multi-modal knowledge transfer, pseudo-label-guided adaptation, and unsupervised domain adaptation.

[CV-44] MHPR: Multidimensional Human Perception and Reasoning Benchmark for Large Vision-Languate Models

【速读】：该论文旨在解决当前多模态大模型评估基准在人类中心场景下缺乏细粒度、多维度评价的问题，尤其是在个体、多人及人机交互等复杂场景中的联合感知与推理能力不足。其解决方案的关键在于构建MHPR（Multidimensional Human Perception and Reasoning）基准，该基准包含四个层次的数据设计（C-RD、SFT-D、RL-D、T-D），并引入自动化标题/视觉问答生成流水线（ACVG），通过类别属性分解、属性特异性重写和多模型投票机制实现高质量、可扩展的标注，从而系统性提升对人类外观、姿态、社交关系、意图等功能语义的建模能力。

链接: https://arxiv.org/abs/2605.03485
作者: Kangkang Wang,Qinting Jiang,Wanping Zhang,Bowen Ren,Shengzhao Wen
机构: Baidu(百度); Tsinghua University (清华大学)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Multidimensional human understanding is essential for real-world applications such as film analysis and virtual digital humans, yet current LVLM benchmarks largely focus on single-task settings and lack fine-grained, human-centric evaluation. In this work, we introduce MHPR, a comprehensive benchmark for joint perception-reasoning over human-centric scenes spanning individual, multi-person, and human-object interaction dimensions. MHPR comprises a multi-level data design-Captioned Raw Data (C-RD), Supervised Fine-Tuning Data (SFT-D), Reinforcement Learning Data (RL-D), and Test Data (T-D)-together with an automated caption/VQA generation pipeline (ACVG) that performs category-wise attribute decomposition, attribute-specific rewriting, and multi-model voting to ensure high-quality, scalable annotations. We evaluate state-of-the-art vision-language models on fine-grained attributes (appearance, clothing, pose, parts) and high-level semantics (social relations, action semantics, spatial relations, intent and functionality). Our findings show that: 1) format-aligned SFT data substantially improves instruction following and stability; 2) challenge-focused RL data derived from bad-case analysis further enhances perception and reasoning on difficult instances; and 3) training Qwen2.5-VL-7B with MHPR yields significant gains, achieving near-parity with considerably larger models. We release ACVG and MHPR to facilitate reproducible, extensible research on human-centric perception and reasoning.

[CV-45] WorldJen: An End-to-End Multi-Dimensional Benchmark for Generative Video Models

【速读】：该论文旨在解决生成式视频模型（Generative Video Models）评估中存在的多维度局限性问题，包括传统基于参考的指标（如SSIM、PSNR）过度关注像素级保真度而忽视语义正确性，FVD虽能捕捉分布特征但对物理合理性敏感度不足，以及现有二值视觉问答（Binary VQA）基准存在“是”偏差和低分辨率审计导致时序错误漏检等问题。其解决方案的关键在于构建一个双轨驱动的评估框架——WorldJen：一是通过盲测人类偏好实验（7名标注者对50个对抗性提示下的6个SOTA模型进行2,696次成对比较），建立具有三层次结构的Bradley-Terry（BT）人类基准；二是开发以视觉语言模型（VLM）为裁判的评分引擎，采用针对每个质量维度设计的Likert量表问卷（每维10题，共47,160条评分响应），在原生视频分辨率下实现自动化评分，并成功复现了人类基准的三层结构，且与人类结果达到Spearman相关系数ρ=1.000（p=0.0014），验证了其有效性与鲁棒性。

链接: https://arxiv.org/abs/2605.03475
作者: Karthik Inbasekar,Guy Rom,Omer Shlomovits
机构: moonmath.ai
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 30 pages +25 appendix

点击查看摘要

Abstract:Evaluating generative video models remains an open problem. Reference-based metrics such as Structural Similarity Index Measure (SSIM) and Peak Signal to Noise Ratio (PSNR) reward pixel fidelity over semantic correctness, while Frechet Video Distance (FVD) favors distributional textures over physical plausibility. Binary Visual Question Answering (VQA) based benchmarks like VBench~2.0 are prone to yes-bias and rely on low-resolution auditors that miss temporal failures. Moreover, their prompts target a single dimension at a time, multiplying the number of videos required while still not guaranteeing reliable results. WorldJen addresses these limitations directly. Binary VQA is replaced with Likert-scale questionnaires graded by a VLM that receives frames at native video resolution. Video generation costs are addressed by using adversarially curated prompts that are designed to exercise up to 16 quality dimensions simultaneously. The framework is built around two interlocking contributions. First, A blind human preference study is conducted, accumulating (2,696 pairwise annotations from 7 annotators with 100% pair coverage over 50 of the curated prompts \times 6 state-of-the-art video models. A mean inter-annotator agreement of 66.9% is achieved and the study establishes a human ground-truth Bradley-Terry (BT) rating with a three-tier structure. Second, A VLM-as-a-judge evaluation engine using prompt-specific, dimension-specific Likert questionnaires (10 questions per dimension, 47,160 scored responses) judges the videos and reproduces the human-established three-tier BT rating structure independently. The VLM achieves a Spearman \hat\rho=1.000,~p=0.0014 that is interpreted as tier agreement with the human results. Six focused ablation studies validate the robustness of the VLM evaluation framework. Comments: 30 pages +25 appendix Subjects: Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2605.03475 [cs.CV] (or arXiv:2605.03475v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2605.03475 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[CV-46] First Shape Then Meaning: Efficient Geometry and Semantics Learning for Indoor Reconstruction

【速读】：该论文旨在解决室内三维重建中几何与语义信息联合学习的效率与可扩展性问题，尤其针对现有基于多Signed Distance Function (SDF)的方法在训练速度慢、难以规模化方面的局限。其解决方案的关键在于提出一种两阶段统一框架FSTM：首先仅使用RGB输入和几何线索进行几何预热（geometry warm-up），在无语义监督条件下优化场景几何；随后在此基础上估计语义场（semantic field）。该策略避免了传统联合优化的复杂性，通过简化模型结构实现了更快的训练速度（在Replica上提升2.3倍）和更强的鲁棒性（在ScanNet++上对真实世界噪声更鲁棒），同时提升了对象级语义召回率，证明了轻量化设计在几何-语义协同重建中的有效性。

链接: https://arxiv.org/abs/2605.03463
作者: Remi Chierchia,Léo Lebrat,David Ahmedt-Aristizabal,Olivier Salvado,Clinton Fookes,Rodrigo Santa Cruz
机构: Queensland University of Technology (昆士兰科技大学); CSIRO Data61 (澳大利亚联邦科学与工业研究组织数据61)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Neural Surface Reconstruction has become a standard methodology for indoor 3D reconstruction, with Signed Distance Functions (SDFs) proving particularly effective for representing scene geometry. A variety of applications require a detailed understanding of the scene context, driving the need for object-level semantic signals. While recent methods successfully integrate semantic labels, they often inherit the slow training time and limited scalability of multi-SDF learning. In this paper, we introduce FSTM, a unified approach for learning geometry and semantics through a two-step process: a geometry warm-up using RGB inputs and geometric cues, followed by semantic field estimation. By first optimising geometry without semantic supervision, we observe substantial improvements compared to the standard joint optimisation. Rather than relying on specialised modules or complex multi-SDF designs, FSTM shows that a streamlined formulation is sufficient to achieve strong geometric and semantic reconstructions. Experiments on both synthetic and real-world indoor datasets show that our method outperforms multi-SDF approaches. It trains 2.3x faster on Replica, improves robustness to real-world imperfections on ScanNet++, and achieves higher recall by recovering the surfaces of more objects in the scene. The code will be made available at this https URL.

[CV-47] VL-SAM-v3: Memory-Guided Visual Priors for Open-World Object Detection

【速读】：该论文旨在解决开放世界目标检测（Open-world object detection）中因依赖粗粒度文本语义和参数化知识而导致的细粒度外观差异、稀有类别识别困难及复杂场景下视觉证据不足的问题。其解决方案的关键在于提出VL-SAM-v3统一框架，通过引入基于检索的外部视觉记忆（retrieval-grounded external visual memory），从非参数化记忆库中获取相关视觉原型，并将其转化为两类互补的视觉先验：用于实例级空间定位的稀疏先验和用于类别感知局部上下文的稠密先验；这些先验通过“记忆引导的提示精炼”（Memory-Guided Prompt Refinement）机制与原始检测提示融合，从而实现开放词汇和开放端检测任务下的共享检索-精炼机制，在LVIS数据集上显著提升了罕见类别的检测性能。

链接: https://arxiv.org/abs/2605.03456
作者: Chih-Chung Liu,Zhiwei Lin,Yongtao Wang
机构: Wangxuan Institute of Computer Technology, Peking University, China (北京大学王选计算机研究所)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Open-world object detection aims to localize and recognize objects beyond a fixed closed-set label space. It is commonly divided into two categories, i.e., open-vocabulary detection, which assumes a predefined category list at test time, and open-ended detection, which requires generating candidate categories during the inference. Existing methods rely primarily on coarse textual semantics and parametric knowledge, which often provide insufficient visual evidence for fine-grained appearance variation, rare categories, and cluttered scenes. In this paper, we propose VL-SAM-v3, a unified framework that augments open-world detection with retrieval-grounded external visual memory. Specifically, once candidate categories are available, VL-SAM-v3 retrieves relevant visual prototypes from a non-parametric memory bank and transforms them into two complementary visual priors, i.e., sparse priors for instance-level spatial anchoring and dense priors for class-aware local context. These priors are integrated with the original detection prompts via Memory-Guided Prompt Refinement, enabling a shared retrieval-and-refinement mechanism that supports open-vocabulary and open-ended this http URL zero-shot experiments on LVIS show that VL-SAM-v3 consistently improves detection performance under both open-vocabulary and open-ended inference, with particularly strong gains on rare this http URL, experiments with a stronger open-vocabulary detector (i.e., SAM3) validate the generality of the proposed retrieval-and-refinement mechanism.

[CV-48] Mantis: Mamba-native Tuning is Efficient for 3D Point Cloud Foundation Models

【速读】：该论文旨在解决在3D点云预训练基础模型（Pre-trained 3D Point Cloud Foundation Models, PFM）中，将现有参数高效微调（Parameter-Efficient Fine-Tuning, PEFT）方法直接应用于冻结的Mamba骨干网络时所引发的性能下降与优化不稳定问题。其关键在于提出了一种面向Mamba架构的新型PEFT框架——Mantis，核心创新为引入状态感知适配器（State-Aware Adapter, SAA），通过注入轻量级任务条件控制信号到选择性状态空间更新中，实现状态层面的适应性调整，同时保持预训练主干网络完全冻结；此外，通过双序列化一致性蒸馏（Dual-Serialization Consistency Distillation, DSCD）对不同有效的点云序列化方式进行正则化，缓解因序列化导致的不稳定性，从而在仅使用约5%可训练参数的情况下实现与全微调相当的性能表现。

链接: https://arxiv.org/abs/2605.03438
作者: Zihao Guo,Jihua Zhu,Jian Liu,Ajmal Saeed Mian
机构: Xi’an Jiaotong University (西安交通大学); Singapore University of Technology and Design (新加坡科技设计大学); University of Western Australia (西澳大利亚大学)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Pre-trained 3D point cloud foundation models (PFMs) have demonstrated strong transferability across diverse downstream tasks. However, full fine-tuning these models is computationally expensive and storage-intensive. Parameter-efficient fine-tuning (PEFT) offers a promising alternative, but existing PEFT approaches are primarily designed for Transformer-based backbones and rely on token-level prompting or feature transformation. Mamba-based backbones introduce a granularity mismatch between token-level adaptation and state-level sequence dynamics. Consequently, straightforward transfer of existing PEFT approaches to frozen Mamba backbones leads to substantial accuracy degradation and unstable optimization. To address this issue, we propose Mantis, the first Mamba-native PEFT framework for 3D PFMs. Specifically, a State-Aware Adapter (SAA) is introduced to inject lightweight task-conditioned control signals into selective state-space updates, enabling state-level adaptation while keeping the pre-trained backbone frozen. Moreover, different valid point cloud serializations are regularized by Dual-Serialization Consistency Distillation (DSCD), thereby reducing serialization-induced instability. Extensive experiments across multiple benchmarks demonstrate that our Mantis achieves competitive performance with only about 5% trainable parameters. Our code is available at this https URL.

[CV-49] Learning Discriminative Signed Distance Functions from Multi-scale Level-of-detail Features for 3D Anomaly Detection

【速读】：该论文旨在解决3D点云中异常检测的挑战，尤其是由于点云数据具有大规模和稀疏性导致的点级表征学习困难问题。其解决方案的关键在于提出一种基于表面的方法，通过构建判别性的符号距离函数（Signed Distance Function, SDF）来实现精准的异常区分：首先设计了噪声点生成（Noisy Points Generation, NPG）模块以引入多样噪声增强异常点的可辨识性；其次引入多尺度细节层次特征（Multi-scale Level-of-detail Feature, MLF）模块以捕获点云的细粒度局部与粗粒度全局信息；最终通过隐式表面判别（Implicit Surface Discrimination, ISD）模块利用多尺度特征训练出能够有效区分正常与异常点的SDF模型，从而在Anomaly-ShapeNet和Real3D-AD数据集上分别取得92.1%和85.9%的平均物体级AUROC，显著优于现有最优方法。

链接: https://arxiv.org/abs/2605.03437
作者: Haibo Xiao,Hanzhe Liang,Jie Zhou,Jinbao Wang,Can Gao
机构: Shenzhen University (深圳大学); Guangdong Provincial Key Laboratory of Intelligent Information Processing (广东省智能信息处理重点实验室)
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:

点击查看摘要

Abstract:Detecting anomalies from 3D point clouds has received increasing attention in the field of computer vision, with some group-based or point-based methods achieving impressive results in recent years. However, learning accurate point-wise representations for 3D anomaly detection faces great challenges due to the large scale and sparsity of point clouds. In this study, a surface-based method is proposed for 3D anomaly detection, which learns a discriminative signed distance function using multi-scale level-of-detail features. We first present a Noisy Points Generation (NPG) module to generate different types of noise, thereby facilitating the learning of discriminative features by exposing abnormal points. Then, we introduce a Multi-scale Level-of-detail Feature (MLF) module to capture multi-scale information from a point cloud, which provides both fine-grained local and coarse-grained global feature information. Finally, we design an Implicit Surface Discrimination (ISD) module that leverages the extracted multi-scale features to learn an implicit surface representation of point clouds, which effectively trains a signed distance function to distinguish between abnormal and normal points. Experimental results demonstrate that the proposed method achieves an average object-level AUROC of 92.1% and 85.9% on the Anomaly-ShapeNet and Real3D-AD datasets, outperforming the current best approach by 2.1% and 3.6%, respectively. Codes are available at this https URL.

[CV-50] MK-ResRecon: Multi-Kernel Residual Framework for Texture-Aware 3D MRI Refinement from Sparse 2D Slices

【速读】：该论文旨在解决磁共振成像（MRI）采集过程耗时长、患者不适感强的问题，尤其是因扫描时间过长导致的运动伪影干扰图像质量，常需重复扫描。其核心解决方案是提出一个双模型框架：MK-ResRecon 和 IdentityRefineNet3D，其中 MK-ResRecon 通过多核纹理感知损失函数预测缺失的中间二维（2D）切片以保留精细解剖细节，而 IdentityRefineNet3D 将预测切片与原始稀疏切片联合优化为单个三维（3D）体积，从而获得平滑的解剖结构。该方法仅需12.5%的轴向切片即可实现全分辨率3D重建，具备高保真度、无幻觉、可泛化且经临床验证的特点，为快速、患者友好的MRI成像提供了可行路径。

链接: https://arxiv.org/abs/2605.03432
作者: Prajyot Pyati,Sapna Sachan,Amulya Kumar Mahto,Pranjal Phukan
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 22 pages, 7 figures

点击查看摘要

Abstract:Magnetic Resonance Imaging (MRI) acquisition remains a time-intensive and patient-straining process, as prolonged scan dura- tions increase the likelihood of motion artifacts, which degrade image quality and frequently require repeated scans. To address these chal- lenges, we propose a novel framework with two models MK-ResRecon and IdentityRefineNet3D to reconstruct high-fidelity 3D MRI volumes from sparsely sampled 2D slices-requiring only 12.5% of the axial slices for full resolution 3D reconstruction. MK-ResRecon predicts missing in- termediate 2D slices using a multi-kernel texture-aware loss, preserving fine anatomical details. IdentityRefineNet3D refines the predicted slices and the original sparse slices as a single 3D volume to obtain a smooth anatomical structure. We train the models on a large T1-sequence POST- contrast brain MRI dataset and evaluate on a large heterogeneous brain MRI cohort. The work provides accurate, hallucination-free, generaliz- able and clinically validated framework for 3D MRI reconstruction from highly sparse inputs and enables a clinically viable path towards faster and more patient-friendly MRI imaging.

[CV-51] sallisPGD: Adaptive Gradient Weighting for Adversarial Attacks on Semantic Segmentation IJCNN2026

【速读】：该论文旨在解决语义分割模型（semantic segmentation models）在对抗攻击中面临的挑战，即与图像分类模型相比，攻击语义分割模型需要同时翻转数千个像素的预测结果，而传统基于像素级交叉熵（cross-entropy, CE）的攻击方法会过度关注已错误分类的像素，导致优化效率低下并高估模型鲁棒性。解决方案的关键在于提出TsallisPGD，一种基于Tsallis交叉熵（Tsallis cross-entropy）的对抗攻击方法，其通过参数 $ q $ 自适应地调整梯度分布，控制梯度在不同像素间的集中程度；进一步引入动态 $ q $-调度策略，在优化过程中按需调整 $ q $ 值，从而有效平衡对高置信度与低置信度像素的攻击力度，显著提升攻击效果，在多个数据集和模型上均优于现有方法。

链接: https://arxiv.org/abs/2605.03405
作者: Alexander Matyasko,Xin Lou,Indriyati Atmosukarto,Wei Zhang
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Accepted to IJCNN 2026. Code: this https URL

点击查看摘要

Abstract:Attacking semantic segmentation models is significantly harder than image classification models because an attacker must flip thousands of pixel predictions simultaneously. Standard pixel-wise cross-entropy (CE) is ill-suited to this setting: it tends to overemphasize already-misclassified pixels, which slows optimization and overstates model robustness. To address these issues, we introduce TsallisPGD, an adversarial attack built on the Tsallis cross-entropy, a generalization of CE parameterized by q , which adaptively reshapes the gradient landscape by controlling gradient concentration across pixels. By varying q , we steer the attack toward pixels at different confidence levels. We first show that no single fixed- q is universally optimal, as its effectiveness depends on the dataset, model architecture, and perturbation budget. Motivated by this, we propose a dynamic q -schedule that sweeps q during optimization. Extensive experiments on Cityscapes, Pascal VOC, and ADE20K show that TsallisPGD, using a single validation-selected schedule, achieves the best average attack rank across all evaluated settings and improves over CEPGD, SegPGD, CosPGD, JSPGD, and MaskedPGD in reducing accuracy and mIoU on both standard and robust models.

[CV-52] GRPO-TTA: Test-Time Visual Tuning for Vision-Language Models via GRPO-Driven Reinforcement Learning

【速读】：该论文旨在解决视觉语言模型在测试时适应（Test-Time Adaptation, TTA）场景下的性能提升问题，尤其是在自然分布偏移下如何有效调整模型以适应未见数据。解决方案的关键在于将Group Relative Policy Optimization (GRPO) 适配至TTA框架，提出GRPO-TTA方法：通过从CLIP相似度分布中采样Top-K类别候选构建输出组，将类别特定的提示预测重构为一种群体策略优化问题，从而在无需真实标签的情况下实现基于概率驱动的优化；同时设计了对齐奖励（alignment rewards）和离散奖励（dispersion rewards）两类奖励函数，指导视觉编码器的有效微调，显著提升了模型在多种基准上的泛化能力。

链接: https://arxiv.org/abs/2605.03403
作者: Yujun Li,Hongyuan Zhang,Yuan Yuan
机构: Northwestern Polytechnical University (西北工业大学); China Telecom (中国电信); The University of Hong Kong (香港大学)
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:

点击查看摘要

Abstract:Group Relative Policy Optimization (GRPO) has recently shown strong performance in post-training large language models and vision-language models. It raises a question of whether the GRPO also significantly promotes the test-time adaptation (TTA) of vision language models. In this paper, we propose Group Relative Policy Optimization for Test-Time Adaptation (GRPO-TTA), which adapts GRPO to the TTA setting by reformulating class-specific prompt prediction as a group-wise policy optimization problem. Specifically, we construct output groups by sampling top-K class candidates from CLIP similarity distributions, enabling probability-driven optimization without access to ground-truth labels. Moreover, we design reward functions tailored to test-time adaptation, including alignment rewards and dispersion rewards, to guide effective visual encoder tuning. Extensive experiments across diverse benchmarks demonstrate that GRPO-TTA consistently outperforms existing test-time adaptation methods, with notably larger performance gains under natural distribution shifts.

[CV-53] MASRA: MLLM -Assisted Semantic-Relational Consistent Alignment for Video Temporal Grounding

【速读】：该论文旨在解决视频时间定位（Video Temporal Grounding, VTG）中因跨模态语义鸿沟导致的背景特征与查询错误对齐，以及直接匹配查询与时间段时语义判别性和一致性不足的问题。解决方案的关键在于提出一种训练阶段基于多模态大语言模型（Multimodal Large Language Model, MLLM）的优化框架MASRA（MLLM-Assisted Semantic-Relational Consistent Alignment），其核心机制包括：1）利用MLLM生成事件级描述（含时间跨度）和片段级字幕作为文本先验，构建事件语义时间对齐（Event Semantic Temporal Alignment, ESTA）以增强语义与时间事件的显式对应关系；2）通过片段级字幕构建文本关系矩阵，并与模型中的时间特征相似性矩阵对齐，实现局部结构信息捕捉与时间一致性强化（Local Relational Consistency Alignment, LRCA）；3）引入解耦对齐交互（Decoupled Alignment Interaction, DAI）机制结合上下文感知码本，自适应吸收无关语义以缓解跨模态差距。整个框架仅在训练阶段使用MLLM，推理阶段无需依赖，从而实现高效且鲁棒的VTG性能提升。

链接: https://arxiv.org/abs/2605.03398
作者: Ran Ran,Jiwei Wei,Shuchang Zhou,Yitong Qin,Shiyuan He,Zeyu Ma,Yuyang Zhou,Yang Yang
机构: University of Electronic Science and Technology of China (电子科技大学); Hainan University (海南大学)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Video Temporal Grounding (VTG) faces a cross-modal semantic gap that often leads to background features being incorrectly aligned with the query, while directly matching the query to moments results in insufficient discriminability and consistency of temporal semantics. To address this issue, we propose MLLM-Assisted Semantic-Relational Consistent Alignment (MASRA), a training-time MLLM-based optimization framework for VTG. MASRA leverages an MLLM during training to produce two forms of textual priors, namely event-level descriptions with temporal spans and clip-level captions, and instantiates two MLLM-assisted alignments. Event Semantic Temporal Alignment (ESTA) aligns temporal context with event semantics to explicitly strengthen the correspondence between semantics and temporal events and improve span-level separability. Local Relational Consistency Alignment (LRCA) constructs a textual relation matrix derived from clip-level captions and aligns it with the temporal feature similarity matrix in the model, enhancing temporal consistency while capturing local structural information. MASRA includes two simple supporting modules, semantic-guided enhancement and second-order relational attention, to better utilize the learned semantic context and relational structure. Moreover, we introduce Decoupled Alignment Interaction (DAI) with a context-aware codebook to adaptively absorb query-irrelevant semantics and alleviate the cross-modal gap. The MLLM is only invoked during training and is not used at inference. Extensive experiments show that MASRA outperforms existing methods, and ablation studies validate its effectiveness.

[CV-54] Enhancing Self-Supervised Talking Head Forgery Detection via a Training-Free Dual-System Framework

【速读】：该论文旨在解决自监督说话头伪造检测中现有评分型检测器在困难样本上判别能力不足的问题，具体表现为异常分数排序不可靠，从而限制了跨生成器的泛化性能。解决方案的关键在于提出一种无需训练的双系统（Training-Free Dual-System, TFDS）框架：其中系统1（System-1）基于轻量级阈值路由将样本分为高置信度和不确定子集，系统2（System-2）仅对不确定子集进行细粒度证据引导推理，以修正原始分数分布中模糊样本的相对排序，从而有效挖掘现有检测器中未被充分利用的判别线索。

链接: https://arxiv.org/abs/2605.03390
作者: Ke Liu,Jiwei Wei,Shuchang Zhou,Yutong Xiao,Ruikun Chai,Yitong Qin,Yuyang Zhou,Yang Yang
机构: University of Electronic Science and Technology of China(电子科技大学); Hainan University(海南大学)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Multimedia (cs.MM)
备注:

点击查看摘要

Abstract:Supervised talking head forgery detection faces severe generalization challenges due to the continuous evolution of generators. By reducing reliance on generator-specific forgery patterns, self-supervised detectors offer stronger cross-generator robustness. However, existing research has mainly focused on building stronger detectors, while the discriminative capacity of trained detectors remains insufficiently exploited. In particular, for score-based self-supervised detectors, the limited discriminative ability on hard cases is often reflected in unreliable anomaly ordering, leaving room for further refinement. Motivated by this observation, we draw inspiration from the dual-system theory of human cognition and propose a Training-Free Dual-System (TFDS) framework to further exploit the latent discriminative capacity of existing score-based self-supervised detectors. TFDS treats anomaly-like scores as the basis of System-1, using lightweight threshold-based routing to partition samples into confident and uncertain subsets. System-2 then revisits only the uncertain subset, performing fine-grained evidence-guided reasoning to refine the relative ordering of ambiguous samples within the original score distribution. Extensive experiments demonstrate consistent improvements across datasets and perturbation settings, with the gains arising mainly from corrected ordering within the uncertain subset. These findings show that existing self-supervised talking head forgery detectors still contain underexploited discriminative cues that can be effectively unlocked through training-free dual-system reasoning.

[CV-55] SoDa2: Single-Stage Open-Set Domain Adaptation via Decoupled Alignment for Cross-Scene Hyperspectral Image Classification

【速读】：该论文旨在解决跨场景高光谱图像（Hyperspectral Image, HSI）分类中两个关键挑战：一是由于混合光谱-空间特征直接对齐导致的域偏移问题；二是因两阶段训练策略带来的高计算成本。解决方案的核心在于提出一种单阶段解耦对齐方法（SoDa²），其关键创新包括：1）设计贡献感知的双模态特征提取机制，分离光谱序列信号与空间细节特征，选择性增强判别性特征；2）引入解耦对齐模块，通过最小化最大均值差异（Maximum Mean Discrepancy, MMD）独立降低源域与目标域间的光谱差异和空间差异，提取更细粒度的域不变特征；3）构建轻量级单阶段双分支框架，同时学习MMD约束下的对齐特征与无约束的内在特征，并利用高斯混合模型建模两类特征间余弦相似度分布，实现无需未知类别先验知识的开集识别。

链接: https://arxiv.org/abs/2605.03371
作者: Yiwen Liu,Minghua Wang,Jing Yao,Xin Zhao,Gemine Vivone
机构: Nankai University (南开大学); Chinese Academy of Sciences (中国科学院); National Research Council (国家研究委员会)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Cross-scene hyperspectral image (HSI) classification stands as a fundamental research topic in remote sensing, with extensive applications spanning various fields. Owing to the inclusion of unknown categories in the target domain and the existence of domain shift across different scenes, open-set domain adaptation techniques are commonly employed to address cross-scene HSI classification. However, existing open-set cross-scene HSI classification methods still face two critical challenges: (1) domain shift issues arising from the direct alignment of mixed spectral-spatial features; (2) high computational costs caused by two-stage training strategies. To address these issues, this paper proposes a single-stage open-set domain adaptation method with decoupled alignment (SoDa ^2 ) for cross-scene HSI classification. A contribution-aware dual-modality feature extraction is customized to disentangle the characteristics from spectral sequence signals and spatial details, selectively and adaptively enhancing discriminative features. The decoupled alignment module minimizes the Maximum Mean Discrepancy to independently reduce the spectral discrepancy and the spatial discrepancy between the source and target domains, extracting more fine-grained domain-invariant features. A cost-effective single-stage dual-branch framework is designed to learn MMD-constrainted aligned features and constraint-free intrinsic features for adaptive distinction between known and unknown classes. This framework employs a Gaussian Mixture Model to model the squared cosine similarity distribution between the two feature types, enabling open-set recognition without prior knowledge of unknown classes. Extensive experiments on three groups of HSI datasets demonstrate that SoDa ^2 outperforms state-of-the-art methods, achieving superior classification accuracy and model transferability for open-set cross-scene tasks.

[CV-56] Dual-Foundation Models for Unsupervised Domain Adaptation ICPR2026

【速读】：该论文旨在解决无监督域自适应（Unsupervised Domain Adaptation, UDA）中语义分割模型在从合成数据向真实场景迁移时面临的两大挑战：一是现有方法依赖高置信度伪标签，导致仅能利用目标域中有限的像素进行学习；二是基于原型的对比学习方法使用源域训练得到的类原型作为初始化锚点，造成原型偏差和不稳定。解决方案的关键在于提出一种双基础UDA框架，其核心创新为：（1）引入Segment Anything Model (SAM) 并结合超像素引导提示机制，使模型能够从更广泛的低置信度目标域像素中学习，从而扩大有效训练样本范围；（2）融合DINOv3的特征表示能力，构建稳定且域不变的类别原型，提升原型质量与适应过程的鲁棒性。该方法在GTA→Cityscapes和SYNTHIA→Cityscapes两个基准上分别实现+1.3%和+1.4%的mIoU提升，显著优于现有强基线。

链接: https://arxiv.org/abs/2605.03365
作者: Yerin Cheon,Aruna Balasubramanian,Francois Rameau
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 15 pages, 6 figures. Accepted at the 28th International Conference on Pattern Recognition (ICPR 2026)

点击查看摘要

Abstract:Semantic segmentation provides pixel-level scene understanding essential for autonomous driving and fine-grained perception tasks. However, training segmentation models requires costly, labor-intensive annotations on real-world datasets. Unsupervised Domain Adaptation (UDA) addresses this by training models on labeled synthetic data and adapting them to unlabeled real images. While conceptually simple, adaptation is challenging due to the domain gap, i.e., differences in visual appearance and scene structure between synthetic and real data. Prior approaches bridge this gap through pixel-level mixing or feature-level contrastive learning. Yet, these techniques suffer from two major limitations: (1) reliance on high-confidence pseudo-labels restricts learning to a subset of the target domain, and (2) prototype-based contrastive methods initialize class prototypes from source-trained models, yielding biased and unstable anchors during adaptation. To address these issues, we propose a dual-foundation UDA framework that leverages two complementary foundation models. First, we employ the Segment Anything Model (SAM) with superpixel-guided prompting to enable learning from a broader range of target pixels beyond high-confidence predictions. Second, we incorporate DINOv3 to construct stable, domain-invariant class prototypes through its robust representation learning. Our method achieves consistent improvements of +1.3% and +1.4% mIoU over strong UDA baselines on GTA-to-Cityscapes and SYNTHIA-to-Cityscapes, respectively.

[CV-57] Dynamic Distillation and Gradient Consistency for Robust Long-Tailed Incremental Learning

【速读】：该论文旨在解决长尾类增量学习（Long-tailed Class Incremental Learning, LT-CIL）中的灾难性遗忘问题，其核心挑战在于：在类别分布极度不均衡的数据集中，模型容易对少数类（minority classes）学习不足，同时对多数类（majority classes）过度拟合。为此，作者提出两种关键技术：一是引入梯度一致性正则化（gradient consistency regularization），通过梯度移动平均抑制训练过程中的剧烈波动以提升稳定性；二是基于归一化熵动态调整蒸馏损失权重，实现旧知识保留与新知识获取之间的自适应平衡。实验表明，该方法在多个长尾数据集上显著提升性能，最高达5.0%准确率增益，且计算开销可控，展现出良好的实用性。

链接: https://arxiv.org/abs/2605.03364
作者: Taigo Sakai,Kazuhiro Hotta
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:The task of Long-tailed Class Incremental Learning (LT-CIL) addresses the sequential learning of new classes from datasets with imbalanced class distributions. This scenario intensifies the fundamental problem of catastrophic forgetting, inherent to continual learning, with the dual challenges of under-learning minority classes and overfitting majority classes. To tackle these combined issues, this paper proposes two main techniques. First, we introduce gradient consistency regularization, which leverages the moving average of gradients to suppress abrupt fluctuations and stabilize the training process. Second, we dynamically adjust the weight of the distillation loss by measuring the degree of class imbalance with normalized entropy. This adaptive weighting establishes an optimal balance between retaining old knowledge and acquiring new information. Experiments on the CIFAR-100-LT, ImageNetSubset-LT, and Food101-LT benchmarks show that our method achieves consistent accuracy improvements of up to 5.0%. Furthermore, we demonstrate dramatic gains in the challenging ‘In-ordered’ setting, where tasks progress from majority to minority classes, highlighting our method’s robustness in mitigating forgetting under unfavorable learning dynamics. This enhanced performance is achieved without a significant increase in computational overhead, demonstrating the practicality of our framework.

[CV-58] Mix3R: Mixing Feed-forward Reconstruction and Generative 3D Priors for Joint Multi-view Aligned 3D Reconstruction and Pose Estimation

【速读】：该论文旨在解决稀疏视图三维重建中两类方法的局限性：前馈式重建（feed-forward reconstruction）虽能预测像素对齐的点图但缺乏完整几何结构，而生成式三维重建（generative 3D reconstruction）虽可生成完整几何体却常存在输入对齐不佳的问题。解决方案的关键在于提出Mix3R框架，通过两阶段生成机制融合两者优势：第一阶段利用Mixture-of-Transformers架构将全局自注意力引入预训练的前馈重建模型与3D生成模型，协同生成稀疏体素、每视角点图及相机参数，实现2D-3D对齐；第二阶段基于重叠注意力偏置（overlap-based attention bias）在无需训练的情况下引导纹理生成模型正确贴合输入纹理到生成形状上。该设计使前馈分支获得生成先验约束，生成分支则受益于前馈分支提供的几何信息，从而在保持高精度相机姿态估计的同时显著提升输入对齐质量。

链接: https://arxiv.org/abs/2605.03359
作者: Siyou Lin,Zhou Xue,Hongwen Zhang,Liang An,Dongping Li,Shaohui Jiao,Yebin Liu
机构: Tsinghua University (清华大学); Beijing Normal University (北京师范大学); ByteDance (字节跳动)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Recent trends in sparse-view 3D reconstruction have taken two different paths: feed-forward reconstruction that predicts pixel-aligned point maps without a complete geometry, and generative 3D reconstruction that generates complete geometry but often with poor input-alignment. We present Mix3R, a novel generative 3D reconstruction method which mixes feed-forward reconstruction and 3D generation into a single framework in an aligned manner. Mix3R generates a 3D shape in two stages: a sparse voxel generation stage and a textured geometry generation stage. Unlike pure generative methods, our first-stage generation jointly produces a coarse 3D structure (sparse voxels), per-view point maps and camera parameters aligned to that 3D structure. This is made possible by introducing a Mixture-of-Transformers architecture that inserts global self-attentions to a feed-forward reconstruction model and a 3D generative model, both pretrained on large-scale data. This design effectively retains the pretrained priors but enables better 2D-3D alignment. Based on the initial aligned generations of sparse 3D voxels and point maps, we compute an overlap-based attention bias that is directly added to another pretrained textured geometry generation model, enabling it to correctly place input textures onto generated shapes in a training-free manner. Our design brings mutual benefits to both feed-forward reconstruction and 3D generation: The feed-forward branch learns to ground its predictions to a generative 3D prior, and conversely, the 3D generation branch is conditioned on geometrically informative features from the feed-forward branch. As a result, our method produces 3D shapes with better input alignment compared with pure 3D generative methods, together with camera pose estimations more accurate than previous feed-forward reconstruction methods. Our project page is at this https URL

[CV-59] racing Like a Clinician: Anatomy-Guided Spatial Priors for Cephalometric Landmark Detection

【速读】：该论文旨在解决正畸临床中头颅侧位片（cephalometric radiographs）自动标注难题，即如何在无监督或弱监督条件下实现高精度、鲁棒的25个解剖标志点定位。传统方法依赖人工 tracing，流程繁琐且易受主观偏差影响，而现有自动化系统难以复现临床医生基于解剖知识的推理过程。解决方案的关键在于提出一个五阶段解剖引导初始化（anatomy-guided initialization）流水线，将临床工作流转化为计算操作，生成置信度加权的空间注意力先验（spatial attention priors），并将其注入下游HRNet-W32检测器中。实验表明，该方法在跨设备、多来源的1,502张影像上实现1.04 mm均径误差，优于先前最优方法（1.23 mm），且消融实验证明：解剖先验不仅提升收敛速度，更显著增强泛化能力——移除后模型在测试集性能下降至1.94 mm，说明其本质作用在于提供架构与数据增强无法替代的归纳偏置（inductive bias）。

链接: https://arxiv.org/abs/2605.03358
作者: Sidhartha Mohapatra,Pallavi Mohanty
机构: CephTrace(cephtrace)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages, 6 tables, 8 figures

点击查看摘要

Abstract:When orthodontists trace cephalometric radiographs, they follow a structured workflow: identify the soft tissue profile, partition the skull into anatomical regions, trace contours, and locate landmarks using geometric definitions – yet no automated system replicates this reasoning. We present a five-phase anatomy-guided initialization pipeline that translates this clinical workflow into computational operations, producing confidence-weighted spatial attention priors for a downstream HRNet-W32 detector. On 1,502 radiographs from three sources spanning 7+ imaging devices, the system achieves 1.04 mm mean radial error on 25 landmarks – surpassing prior state-of-the-art (1.23 mm on 19 landmarks) by 15.4%, with twelve landmarks below 1 mm. A three-way controlled ablation reveals two striking findings. First, removing anatomical priors does not merely slow convergence – it destroys generalization: both models converge to ~1.03 mm on validation, but diverge to 1.94 vs. 1.04 mm on the test set. Second, replacing anatomical priors with random-position Gaussians produces even worse generalization (2.24 mm), confirming that the improvement derives from anatomically correct positioning, not additional input channels. Clinical domain knowledge encoded as spatial priors provides an inductive bias that architecture and data augmentation alone do not provide.

[CV-60] Can Multimodal Large Language Models Understand Pathologic Movements? A Pilot Study on Seizure Semiology

【速读】：该论文旨在解决多模态大语言模型（Multimodal Large Language Models, MLLMs）在临床神经学视频分析中识别病理性运动（如癫痫发作中的半球征象）的潜力尚未被充分探索的问题。其解决方案的关键在于：利用零样本（zero-shot）能力直接评估先进MLLMs对90例临床癫痫发作视频中20个ILAE定义的半球征象的自动识别性能，并通过特征目标导向的信号增强策略（如面部裁剪、姿态估计和音频去噪）提升模型表现，同时验证生成解释的忠实度（faithfulness），结果表明MLLMs在无需任务特定训练的情况下即可有效识别显著的姿势与情境特征，且结合预处理可显著改善对关键病理特征的检测，从而为可解释、高效的临床辅助诊断提供新路径。

链接: https://arxiv.org/abs/2605.03352
作者: Lina Zhang,Tonmoy Monsoor,Mehmet Efe Lorasdagi,Prateik Sinha,Chong Han,Peizheng Li,Yuan Wang,Jessica Pasqua,Colin McCrimmon,Rajarshi Mazumder,Vwani Roychowdhury
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Multimodal Large Language Models (MLLMs) have demonstrated robust capabilities in recognizing everyday human activities, yet their potential for analyzing clinically significant involuntary movements in neurological disorders remains largely unexplored. This pilot study evaluates the capability of MLLMs for automated recognition of pathological movements in seizure videos. We assessed the zero-shot performance of state-of-the-art MLLMs on 20 ILAE-defined semiological features across 90 clinical seizure recordings. MLLMs outperformed fine-tuned Convolutional Neural Network (CNN) and Vision Transformer (ViT) baseline models on 13 of 18 features without task-specific training, demonstrating particular strength in recognizing salient postural and contextual features while struggling with subtle, high-frequency movements. Feature-targeted signal enhancement (facial cropping, pose estimation, audio denoising) improved performance on 10 of 20 features. Expert evaluation showed that 94.3 percent of MLLM-generated explanations for correctly predicted cases achieved at least 60 percent faithfulness scores, aligning with epileptologist reasoning. These findings demonstrate the potential of adapting general-purpose MLLMs for specialized clinical video analysis through targeted preprocessing strategies, offering a path toward interpretable, efficient diagnostic assistance. Our code is publicly available at this https URL.

[CV-61] VLMaxxing through FrameMogging Training-Free Anti-Recomputation for Video Vision-Language Models

【速读】：该论文旨在解决视频视觉语言模型（Video Vision-Language Models, Video VLMs）在处理连续视频问答任务时存在的冗余计算问题，即模型在已知场景稳定的前提下仍重复处理密集RGB帧或重新生成前缀，导致资源浪费和延迟增加。其核心解决方案是提出一种无需训练的“抗重计算”（training-free anti-recomputation）机制：当验证结果表明当前视频状态可复用时保留缓存，仅在场景变化、查询需求或缓存拓扑结构要求时才获取新证据。关键创新在于通过自适应复用同一视频状态实现显著加速——例如，在冻结的Qwen2.5-VL-7B-Instruct-4bit模型上，后续问答延迟降低14.90–35.92倍，同时保持93个查询跨度下的配对选择正确性；此外，引入C-CEILING作为端到端加速的会计约束，确保组件提速仅按其实际占用的时钟比例转化为整体性能提升，避免虚假加速。

链接: https://arxiv.org/abs/2605.03351
作者: JF Bastien,Sam D’Amico
机构: Impulse Labs (Impulse Labs)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 37 pages, 6 figures, 22 tables; code and artifacts available at this https URL

点击查看摘要

Abstract:Video vision-language models (VLMs) keep paying for visual state the stream already told us was stable. The factory wall did not move, but most VLM pipelines still hand the model dense RGB frames or a fresh prefix again. We study that waste as training-free anti-recomputation: reuse state when validation says it survives, and buy fresh evidence when the scene, query, or cache topology requires it. The largest measured win is after ingest. On frozen Qwen2.5-VL-7B-Instruct-4bit, adaptive same-video follow-up reuse preserves paired choices and correctness on a 93-query VideoMME breadth setting while reducing follow-up latency by 14.90-35.92x. The first query is still cold; the win starts when later questions reuse the same video state. Stress tests bound the result: repeated-question schedules hold through 50 turns, while dense-answer-anchored prompt variation separates conservative fixed K=1 repair from faster aggressive policies that drift. Fresh-video pruning is smaller but real. C-VISION skips timed vision-tower work before the first answer is generated. On Gemma 4-E4B-4bit, the clean 32f short cell reaches 1.316x first-query speedup with no paired drift or parse failures on 20 items; Qwen shows the fidelity/speed boundary. Stage-share ceiling (C-CEILING) is the accounting guardrail: a component speedup becomes an end-to-end speedup only in proportion to the wall-clock share it accelerates, so C-VISION and after-ingest follow-up reuse do not multiply. Candidate C-STREAM remains a native-rate target, not a headline result here. The broader direction is VLM-native media that expose change, motion, uncertainty, object state, sensor time, and active tiles directly, so models do not have to rediscover the world from dense RGB every frame. Comments: 37 pages, 6 figures, 22 tables; code and artifacts available at this https URL Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) ACMclasses: I.2.10; I.4; I.5; C.4 Cite as: arXiv:2605.03351 [cs.CV] (or arXiv:2605.03351v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2605.03351 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Jf Bastien [view email] [v1] Tue, 5 May 2026 04:13:32 UTC (251 KB)

[CV-62] MedSR-Vision: Deep Learning Framework for Multi-Domain Medical Image Super-Resolution

【速读】：该论文旨在解决医学图像超分辨率（Medical Image Super-Resolution, MedSR）中普遍存在的三大挑战：保持解剖结构准确性、维持感知质量以及跨模态泛化能力不足的问题。针对这些问题，作者提出了一种统一的深度学习评估框架——MedSR-Vision，其关键在于构建了一个涵盖五种医学成像模态（脑部MRI、胸部X光、肾脏超声、肾结石CT和脊柱MRI）在×2、×3和×4放大倍数下的标准化评测体系，并采用多种定量指标（包括保真度、感知真实性和锐度）对三种代表性模型（SRCNN、SwinIR与Real-ESRGAN）进行系统性比较。该方案不仅揭示了不同模型在特定应用场景下的优势特性（如Real-ESRGAN在高倍率下表现优异的边缘恢复能力，SwinIR在结构保留方面的突出性能），还为临床影像工作流中的模型选择提供了可量化的决策依据，从而推动MedSR技术向更可靠、可推广的方向发展。

链接: https://arxiv.org/abs/2605.03343
作者: Subhash Gurappa,Trivikram Satharasi,Yashas Hariprasad,Sundararaj Sitharama Iyengar
机构: Florida International University (佛罗里达国际大学); University of Florida (佛罗里达大学); California State University, East Bay (加州州立大学东湾分校)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Medical image super-resolution (MedSR) is essential for improving diagnostic precision across diverse imaging modalities such as MRI, CT, X-ray, Ultrasound, and Fundus imaging. Despite rapid advances in deep learning, challenges remain in preserving anatomical accuracy, maintaining perceptual quality, and generalizing across medical domains. This paper presents MedSR-Vision, a novel unified deep learning framework for evaluating and comparing super-resolution models across five modalities: Brain MRI, Chest X-ray, Renal Ultrasound, Nephrolithiasis CT, and Spine MRI, at magnification scales of \times2 , \times3 , and \times4 . Three representative models namely SRCNN, SwinIR, and Real-ESRGAN are benchmarked using multiple quantitative metrics encompassing fidelity, perceptual realism, and sharpness. Experimental analysis demonstrates that Real-ESRGAN achieves superior perceptual quality and edge recovery at higher scales, SwinIR excels in preserving structural and diagnostic features, and SRCNN provides efficient and stable performance at lower magnifications. The results establish domain-specific insights and practical guidelines for model selection in clinical imaging workflows, offering a standardized evaluation framework for future medical image super-resolution research and deployment. Subjects: Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2605.03343 [cs.CV] (or arXiv:2605.03343v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2605.03343 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[CV-63] FreeTimeGS: Secrets of Dynamic Gaussian Splatting and Their Principles

【速读】：该论文旨在解决4D高斯溅射（4D Gaussian Splatting, 4DGS）在动态场景重建中性能提升的内在机制不明确的问题，尤其是对驱动性能改善的关键因素缺乏系统性理解。为此，作者首先构建了一个受控基线方法FreeTimeGS_ours，以形式化并复现当前最优方法FreeTimeGS的启发式策略；在此基础上，通过沿时间维度与空间一致性等核心轴线进行剖析，揭示了两个关键发现：一是由高斯持续时间自发产生的时序分区现象，二是光度保真度与时空一致性之间的差异。基于这些洞察，论文提出了FreeTimeGS++，其核心创新在于引入门控边缘化（gated marginalization）和神经速度场（neural velocity fields），从而实现更稳定且鲁棒的动态表征，显著降低运行间的方差，并提供可复现的结果。

链接: https://arxiv.org/abs/2605.03337
作者: Lucas Yunkyu Lee,Soonho Kim,Youngwook Kim,Sangmin Kim,Jaesik Park
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 22 pages, 8 figures

点击查看摘要

Abstract:The recent surge in 4D Gaussian Splatting (4DGS) has achieved impressive dynamic scene reconstruction. While these methods demonstrate remarkable performance, the specific drivers behind such gains remain less explored, making a systematic understanding of the underlying principles challenging. In this paper, we perform a comprehensive analysis of these hidden factors to provide a clearer perspective on the 4DGS framework. We first establish a controlled baseline, FreeTimeGS_ours, by formalizing and reproducing the heuristics of the state-of-the-art FreeTimeGS. Using this framework, we dissect 4DGS along its fundamental axes and uncover key secrets, including the emergent temporal partitioning driven by Gaussian durations and the discrepancy between photometric fidelity and spatiotemporal consistency. Based on these insights, we propose FreeTimeGS++, a principled method that employs gated marginalization and neural velocity fields to achieve superior stability and robust dynamic representations. Our approach yields reproducible results with reduced run-to-run variance. We will release our implementation to provide a reliable foundation for future 4DGS research.

[CV-64] AHPA: Adaptive Hierarchical Prior Alignment for Diffusion Transformers

【速读】：该论文旨在解决扩散 Transformer 训练中现有表示对齐方法存在的局限性问题，即固定监督目标或固定对齐粒度的策略无法适应去噪轨迹中信号-噪声比（Signal-to-Noise Ratio, SNR）变化所带来的动态需求。在高噪声阶段，模型更依赖粗粒度的语义和布局锚定；而在低噪声阶段，则需强调空间细节与结构忠实性的精调。这种非平稳的对齐行为导致静态单层级监督器产生表征失配。解决方案的关键在于提出自适应分层先验对齐（Adaptive Hierarchical Prior Alignment, AHPA），其利用冻结的变分自编码器（Variational Autoencoder, VAE）中天然嵌套的多层级表示，通过一个时步条件驱动的动态路由机制（Dynamic Router）自适应选择并加权不同层级的先验信息，从而实现对齐粒度与模型训练阶段需求的同步匹配，显著提升收敛速度与生成质量，且不增加推理开销。

链接: https://arxiv.org/abs/2605.03317
作者: Ruibin Min,Yexin Liu,Aimin Pan,Changsheng Lu,Jiafei Wu,Kelu Yao,Xiaogang Xu,Harry Yang
机构: Sun Yat-sen University (中山大学); The Hong Kong University of Science and Technology (香港科技大学); VNET Group (VNET集团); Zhejiang Lab (浙江实验室); Zhejiang University (浙江大学)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Representation alignment has recently emerged as an effective paradigm for accelerating Diffusion Transformer training. Despite their success, existing alignment methods typically impose a fixed supervision target or a fixed alignment granularity throughout the entire denoising trajectory, whether the guidance is provided by external vision encoders, internal self-representations, or VAE-derived features. We argue that such timestep-agnostic alignment is suboptimal because the useful granularity of representation supervision changes systematically with the signal-to-noise ratio. In high-noise regimes, diffusion models benefit more from coarse semantic and layout-level anchoring, whereas in low-noise regimes, the training signal should emphasize spatially detailed and structurally faithful refinement. This non-stationary alignment behavior creates a representational mismatch for static single-level supervisors. To address this issue, we propose Adaptive Hierarchical Prior Alignment (AHPA), a lightweight alignment framework that exploits the hierarchical representations naturally embedded in the frozen VAE encoder. Instead of using only a single compressed latent as the alignment target, AHPA extracts multi-level VAE features that provide complementary priors ranging from local geometry and spatial topology to coarse semantic layout. A timestep-conditioned Dynamic Router adaptively selects and weights these hierarchical priors along the denoising trajectory, thereby synchronizing the alignment granularity with the model’s evolving training needs. Extensive experiments show that AHPA improves convergence and generation quality over baselines and incurs no additional inference cost while avoiding external encoder supervision during training.

[CV-65] ACO: Trajectory Aligning Cross-view Optimisation

【速读】：该论文旨在解决在无卫星导航信号（如GNSS被遮挡、干扰或欺骗）环境下，如何实现高精度、低延迟的绝对位置定位问题。传统惯性测量单元（IMU）虽能提供高频相对运动信息，但存在漂移累积的问题；而现有细粒度跨视图地理定位（Cross-View Geo-localisation, CVGL）方法仅作为一次性定位器使用，未集成到实时定位流水线中。其解决方案的关键在于提出TACO——一种紧耦合的IMU与细粒度CVGL融合框架：通过闭式交叉轨道误差模型触发CVGL匹配时机，避免IMU漂移超出匹配器捕获半径；采用前向偏置五点多裁剪搜索策略保持每帧固定推理成本；引入航向残差门控机制剔除与机载磁力计不一致的定位结果；并基于各定位结果置信度设计各向异性体坐标系噪声模型以自适应调整无迹卡尔曼滤波更新权重。最终在KITTI raw数据集上实现了从纯IMU的97.0米中位数绝对轨迹误差（ATE）降至16.3米，提升5.9倍，且每帧融合计算仅需0.1毫秒，相机功耗维持在5%-10%。

链接: https://arxiv.org/abs/2605.03315
作者: Tavis Shore,Oscar Mendez,Simon Hadfield
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注:

点击查看摘要

Abstract:Cross-View Geo-localisation (CVGL) matches ground imagery against satellite tiles to give absolute position fixes, an alternative to GNSS where signals are occluded, jammed, or spoofed. Recent fine-grained CVGL methods regress sub-tile metric pose, but have only been evaluated as one-shot localisers, never as the primary fix in a live pipeline. Inertial sensing provides high-rate relative motion, but accumulates unbounded drift without an absolute anchor. We propose TACO, a tightly-coupled IMU + fine-grained CVGL pipeline that consumes a single GNSS reading at start-up and thereafter operates on onboard sensing alone. A closed-form cross-track error model triggers CVGL before IMU drift exceeds the matcher’s capture radius, and a forward-biased five-point multi-crop search keeps inference cost fixed at five forward passes per fix. A yaw-residual gate rejects fixes that disagree with the onboard compass, and an anisotropic body-frame noise model scales each Unscented Kalman Filter update by per-fix confidence. A factor graph with vetted loop closures provides an offline smoothed trajectory. On the KITTI raw dataset, TACO reduces median Absolute Trajectory Error (ATE) from 97.0m (IMU-only) to 16.3m, a 5.9 times reduction, at 0.1 ms per-frame fusion cost and a 5-10% camera duty cycle. Code is available: this http URL.

[CV-66] FACTOR: Counterfactual Training-Free Test-Time Adaptation for Open-Vocabulary Object Detection

【速读】：该论文旨在解决开放词汇目标检测（Open-vocabulary object detection）在分布偏移（distribution shifts）下性能下降的问题，其核心原因是模型容易受到非因果视觉属性（如亮度、纹理）与物体类别之间的虚假相关性误导。现有测试时自适应（Test-time adaptation, TTA）方法要么依赖昂贵的在线优化，要么进行全局校准，忽略了这些错误的属性特异性本质。解决方案的关键在于提出 FACTOR（counterFACtual training-free Test-time adaptation for Open-vocabulary object detection），一种基于反事实推理（counterfactual reasoning）的轻量级框架：通过扰动测试图像中非因果属性并比较原始视图与反事实视图下的区域级预测，量化属性敏感性、语义相关性和预测变化，从而选择性抑制依赖特定属性的预测结果，且无需参数更新。实验表明，FACTOR 在 PASCAL-C、COCO-C 和 FoggyCityscapes 数据集上均显著优于现有 TTA 方法，验证了显式反事实推理对提升分布偏移下鲁棒性的有效性。

链接: https://arxiv.org/abs/2605.03294
作者: Kaixiang Zhao,Mao Ye,Lihua Zhou,Hu Wang,Luping Ji,Song Tang,Xiatian Zhu
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Open-vocabulary object detection often fails under distribution shifts, as it can be misled by spurious correlations between non-causal visual attributes (e.g., brightness, texture) and object categories. Existing test-time adaptation (TTA) methods either depend on costly online optimization or perform global calibration, overlooking the attribute-specific nature of these failures. To address this, we propose FACTOR (counterFACtual training-free Test-time adaptation for Open-vocabulaRy object detection), a lightweight framework grounded in counterfactual reasoning. By perturbing test images along non-causal attributes and comparing region-level predictions between original and counterfactual views, FACTOR quantifies attribute sensitivity, semantic relevance, and prediction variation to selectively suppress attribute-dependent predictions-without parameter updates. Experiments on PASCAL-C, COCO-C, and FoggyCityscapes show that FACTOR consistently outperforms prior TTA methods, demonstrating that explicit counterfactual reasoning effectively improves robustness under distribution shifts.

[CV-67] VEBench:Benchmarking Large Multimodal Models for Real-World Video Editing CVPR

【速读】：该论文旨在解决当前大型多模态模型（Large Multimodal Models, LMMs）在真实视频编辑场景中缺乏多视频推理能力和操作性编辑流程理解的问题。现有模型虽在通用视频理解上取得进展，但在识别编辑技巧与模拟实际剪辑工作流方面表现不足，导致其难以胜任专业级视频创作任务。解决方案的关键在于构建首个综合性基准测试平台VEBENCH，该平台包含3.9K高质量编辑视频（超257小时）和3,080个人工验证的问答对，并采用三轮人机协同标注流程确保时间标记精度与语义一致性；其核心设计包括两项互补任务：视频编辑技术识别（评估模型基于多模态线索识别7类编辑技巧的能力）与视频编辑操作模拟（要求模型从多个候选片段中选择并精确定位相关素材以还原编辑流程），从而系统性地衡量模型在编辑知识理解和操作推理上的综合能力。

链接: https://arxiv.org/abs/2605.03276
作者: Andong Deng,Dawei Du,Zhenfang Chen,Wen Zhong,Fan Chen,Guang Chen,Chia-Wen Kuo,Longyin Wen,Chen Chen,Sijie Zhu
机构: ByteDance Intelligent Creation(字节跳动智能创作); CRCV, University of Central Florida(中央佛罗里达大学计算机视觉与机器人研究中心)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: CVPR Findings 2026

点击查看摘要

Abstract:Real-world video editing demands not only expert knowledge of cinematic techniques but also multimodal reasoning to select, align, and combine footage into coherent narratives. While recent Large Multimodal Models (LMMs) have shown remarkable progress in general video understanding, their abilities in multi-video reasoning and operational editing workflows remain largely unexplored. We introduce VEBENCH, the first comprehensive benchmark designed to evaluate both editing knowledge understanding and operational reasoning in realistic video editing scenarios. VEBENCH contains 3.9K high-quality edited videos (over 257 hours) and 3,080 human-verified QA pairs, built through a three-round human-AI collaborative annotation pipeline that ensures precise temporal labeling and semantic consistency. It features two complementary QA tasks: 1) Video Editing Technique Recognition, assessing models’ ability to identify 7 editing techniques using multimodal cues; and 2) Video Editing Operation Simulation, modeling real-world editing workflows by requiring the selection and temporal localization of relevant clips from multiple candidates. Extensive experiments across proprietary (e.g., Gemini-2.5-Pro) and open-source LMMs reveal a large gap between current model performance and human-level editing cognition. These results highlight the urgent need for bridging video understanding with creative operational reasoning. We envision VEBENCH as a foundation for advancing intelligent video editing systems and driving future research on complex reasoning.

[CV-68] CropVLM: A Domain-Adapted Vision-Language Model for Open-Set Crop Analysis

【速读】：该论文旨在解决高通量植物表型分析（high-throughput plant phenotyping）中的“表型瓶颈”问题，即传统人工数据采集方式效率低下且易受观察者偏差影响，同时现有封闭集计算机视觉系统因需大量物种特异性标注而缺乏灵活性，难以适应多样化育种群体。解决方案的关键在于提出CropVLM——一种通过领域特定语义对齐（Domain-Specific Semantic Alignment, DSSA）优化的视觉-语言模型（Vision-Language Model, VLM），其在52,987张自然田间条件下采集的图像-文本对上训练，能够将农业术语精准映射到细粒度视觉特征；并进一步设计了混合开放集定位网络（Hybrid Open-Set Localization Network, HOS-Net），利用CropVLM实现仅依赖自然语言描述即可检测新作物种类的能力，无需重新训练，从而显著提升表型分析的可扩展性与泛化能力。

链接: https://arxiv.org/abs/2605.03259
作者: Abderrahmene Boudiaf,Sajd Javed
机构: Khalifa University of Science and Technology (哈利法大学科学技术学院)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:High-throughput plant phenotyping, the quantitative measurement of observable plant traits, is critical for modern breeding but remains constrained by a “phenotyping bottleneck,” where manual data collection is labor-intensive and prone to observer bias. Conventional closed-set computer vision systems fail to address this challenge, as they require extensive species-specific annotation and lack the flexibility to handle diverse breeding populations. To bridge this gap, we present CropVLM, a Vision-Language Model (VLM) adapted for the agricultural domain via Domain-Specific Semantic Alignment (DSSA). Trained on 52,987 manually selected image-caption pairs covering 37 species in natural field conditions, CropVLM effectively maps agronomic terminology to fine-grained visual features. We further introduce the Hybrid Open-Set Localization Network (HOS-Net), an architecture that integrates CropVLM to enable the detection of novel crops solely from natural language descriptions without retraining. By eliminating the reliance on species-specific training data, CropVLM provides a scalable solution for high-throughput phenotyping, accelerating genetic gain and facilitating large-scale biodiversity research essential for sustainable agriculture. The trained model weights and complete pipeline implementation are publicly available at: [this https URL](this https URL). In comprehensive evaluations, CropVLM achieves 72.51% zero-shot classification accuracy, outperforming seven CLIP-style baselines. Our detection pipeline demonstrates superior zero-shot generalization to novel species, achieving 49.17 AP50 on our CVTCropDet benchmark and 50.73 AP50 on tropical fruit species, compared to 34.89 and 48.58 for the next-best method, respectively.

[CV-69] Ortho-Hydra: Orthogonalized Experts for DiT LoRA

【速读】：该论文旨在解决扩散 Transformer（DiT）在多风格数据上使用 LoRA 微调时出现的“风格混叠”（bleed）问题，即单一低秩残差无法表征多个独立艺术家风格指纹，导致优化器收敛至各风格的平均表示。解决方案的关键在于提出 Ortho-Hydra 方法，通过重新参数化实现两个核心机制：一是引入 OFT 风格的 Cayley-orthogonal 共享基底，确保专家间正交性；二是从预训练权重的前 $Er$ 个左奇异向量中切分出每个专家独有的输出子空间，从而保证路由模块在训练初始阶段即可获得非退化的梯度信号，打破对称性并促进专家特化。这一设计显著改善了路由动态，在早期步骤内即开始脱离均匀先验，避免了传统 HydraLoRA 中因零初始化导致的专家演化对称性问题。

链接: https://arxiv.org/abs/2605.03252
作者: Seunghyun Ji
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:LoRA fine-tuning of diffusion transformers (DiT) on multi-style data suffers from \emphstyle bleed: a single low-rank residual cannot represent several distinct artist fingerprints, and the optimizer converges to their average. Mixture-of-experts LoRA in the HydraLoRA style replaces the up-projection with E heads under a router, but when every expert is zero-initialized the router receives identical gradient from each head and remains at the uniform prior. The experts then evolve permutation-symmetrically, and the network trains as a single rank- r LoRA at E\times the cost. We present \textbfOrtho-Hydra, a re-parameterisation that combines an OFT-style Cayley-orthogonal shared basis with per-expert \emphdisjoint output subspaces carved from the top- (Er) left singular vectors of the pretrained weight. Disjointness makes the router’s per-expert score non-degenerate at step~ 0 , so specialization receives gradient signal before any expert has trained. We test the predicted deadlock on a DiT pipeline by comparing two HydraLoRA baselines, a zero-initialized shared-basis variant and the original \sigma=0.1 Gaussian-jitter mitigation, against Ortho-Hydra under a matched optimiser, dataset, and step budget. Neither baseline leaves the uniform prior within the first 1\textk steps; Ortho-Hydra begins de-uniformising within the first few hundred. End-task generation quality on multi-style data is out of scope; we report the construction, the cold-start mechanism, and the routing dynamics it changes. Code: this https URL.

[CV-70] xt-Conditional JEPA for Learning Semantically Rich Visual Representations ICML2026

【速读】：该论文旨在解决图像自监督学习中因掩码区域存在固有视觉不确定性而导致的特征预测困难问题，进而影响语义表征的学习效果。其核心解决方案是提出文本条件化的JEPA（Text-Conditional JEPA, TC-JEPA），关键在于引入细粒度文本条件器，通过在输入文本标记上计算稀疏交叉注意力（sparse cross-attention）来调制预测的图像补丁特征，使补丁特征成为文本的函数，从而提升预测的可解释性和语义一致性，显著改善下游任务性能与训练稳定性，并展现出良好的扩展性。

链接: https://arxiv.org/abs/2605.03245
作者: Chen Huang,Xianhang Li,Vimal Thilak,Etai Littwin,Josh Susskind
机构: 未知
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注: ICML 2026

点击查看摘要

Abstract:Image-based Joint-Embedding Predictive Architecture (I-JEPA) offers a promising approach to visual self-supervised learning through masked feature prediction. However with the inherent visual uncertainty at masked positions, feature prediction remains challenging and may fail to learn semantic representations. In this work, we propose Text-Conditional JEPA (TC-JEPA) that uses image captions to reduce the prediction uncertainty. Specifically, we modulate the predicted patch features using a fine-grained text conditioner that computes sparse cross-attention over input text tokens. With such conditioning, patch features become predictable as a function of text, thus are more semantically meaningful. We show TC-JEPA improves downstream performance and training stability, with promising scaling properties. TC-JEPA also offers a new vision-language pretraining paradigm based on feature prediction only, outperforming contrastive methods on diverse tasks, especially those requiring fine-grained visual understanding and reasoning.

[CV-71] Synthetic Data Generation for Long-Tail Medical Image Classification: A Case Study in Skin Lesions

【速读】：该论文旨在解决医学图像分类中长尾分布（long-tailed class distributions）导致的深度学习模型在少数类（tail classes）上性能显著下降的问题，尤其关注罕见疾病类别因样本稀缺而难以准确诊断的临床风险。解决方案的关键在于提出一种基于扩散模型（diffusion model）的合成数据增强流水线，其核心创新包括：1）设计了一种新颖的图像修复型扩散模型（inpainting diffusion model），用于生成高质量、多样化的合成样本；2）引入分布外检测（Out-of-Distribution, OOD）后选择机制，确保生成样本在临床语义上真实且具有判别性。该方法在ISIC2019皮肤病变分类数据集上的实验表明，不仅整体性能提升显著，对极端稀疏类别的识别准确率改善超过28%，验证了扩散模型驱动的增强策略在缓解长尾不平衡和提升医疗分类鲁棒性方面的有效性。

链接: https://arxiv.org/abs/2605.03221
作者: Jiaxiang Jiang,Mahesh Subedar,Omesh Tickoo
机构: Intel(英特尔); Google(谷歌)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Long-tailed class distributions are pervasive in multi-class medical datasets and pose significant challenges for deep learning models which typically underperform on tail classes with limited samples. This limitation is particularly problematic in medical applications, where rare classes often correspond to severe or high-risk diseases and therefore require high diagnostic accuracy. Existing solutions-including specialized architectures, rebalanced loss functions, and handcrafted data augmentation-offer only marginal improvements and struggle to scale due to their limited and largely deterministic variability. To address these challenges, we introduce a diffusion-model-driven synthetic data augmentation pipeline tailored for medical long-tailed classification. Our approach features a novel inpainting diffusion model combined with an Out-of-Distribution (OOD) post-selection mechanism to ensure diverse, realistic, and clinically meaningful synthetic samples. Evaluated on the ISIC2019 skin lesion classification dataset, one of the largest and most imbalanced medical imaging benchmarks, our method yields substantial improvements in overall performance, with particularly pronounced gains on tail classes with more than 28% improvement on the class with the fewest samples. These results demonstrate the effectiveness of diffusion-based augmentation in mitigating long-tail imbalance and enhancing medical classification robustness.

[CV-72] Sentinel2Cap: A Human-Annotated Benchmark Dataset for Multimodal Remote Sensing Image Captioning

【速读】：该论文旨在解决多模态遥感图像描述生成（image captioning）领域中数据稀缺的问题，尤其是针对合成孔径雷达（SAR）影像和中分辨率传感器的标注数据匮乏问题。其解决方案的关键在于构建并公开发布Sentinel2Cap数据集——一个由人工标注、经过严格验证的多模态图像描述数据集，包含Sentinel-1 SAR与Sentinel-2多光谱图像（空间分辨率为10 m和20 m），覆盖多样化地表覆盖类型。该数据集为跨模态场景理解研究提供了高质量基准，同时通过在RGB、多光谱及SAR伪RGB三种模态上进行零样本图像描述实验，验证了人类标注数据对提升视觉-语言模型性能的重要性，并揭示了SAR模态在当前模型中的挑战性。

链接: https://arxiv.org/abs/2605.03189
作者: Lucrezia Tosato,Gianluca Lombardi,Ronny Hansch
机构: LIPADE, Université Paris Cité, 75006 Paris, France; SAPIA, ONERA, Palaiseau, France; LCQB, Sorbonne Université, CNRS, IBPS, 75005 Paris, France; German Aerospace Center (DLR), Weßling, Germany
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages, 6 figures, 5 tables

点击查看摘要

Abstract:Image captioning has become an important task in computer vision, enabling models to generate natural language descriptions of visual content. While several datasets exist for natural images and high-resolution optical remote sensing imagery, the availability of captioning datasets for multimodal satellite data remains limited, particularly for SAR imagery and medium-resolution sensors. We introduce Sentinel2Cap, a human-annotated multimodal captioning dataset containing Sentinel-1 SAR and Sentinel-2 multi-spectral image patches at 10 m and 20 m spatial resolution with diverse land cover compositions. Captions are created manually and carefully validated to ensure both semantic accuracy and linguistic quality. To evaluate Sentinel2Cap, we perform a zero-shot captioning using the Qwen3-VL-8B-Instruct model across three image modalities: RGB, multi-spectral, and SAR pseudo-RGB representations. Results show that RGB images achieve the highest captioning performance, while SAR images remain more challenging for vision-language models. Providing modality-specific contextual prompts consistently improves performance across all metrics. These findings highlight both the challenges of multimodal remote sensing image captioning and the potential value of human-annotated datasets for advancing research in cross-modal scene understanding. All the material is publicly avaiable.

[CV-73] DINO Soars: DINOv3 for Open-Vocabulary Semantic Segmentation of Remote Sensing Imagery CVPR

【速读】：该论文旨在解决遥感（Remote Sensing, RS）图像语义分割中因缺乏密集标注数据而导致模型性能受限的问题。现有方法通常依赖于大量标注数据进行监督微调，但此类数据获取成本高昂。为此，作者提出了一种无需在RS领域进行微调的开放词汇语义分割（Open Vocabulary Semantic Segmentation, OVSS）模型CAFe-DINO，其关键创新在于利用DINOv3强大的通用视觉表征能力，结合代价聚合（Cost Aggregation）与无需训练的特征上采样策略，直接从文本-图像相似度得分中提取高分辨率语义信息，从而实现对RS图像的精准分割。该方案通过在COCO-Stuff的一个RS相关子集上进行轻量级微调，显著提升了模型在多个RS分割基准上的性能，超越了需在RS数据上微调的现有OVSS方法。

链接: https://arxiv.org/abs/2605.03175
作者: Ryan Faulkenberry,Saurabh Prasad
机构: University of Houston (休斯顿大学)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at 2026 CVPR MORSE Workshop

点击查看摘要

Abstract:The remote sensing (RS) domain suffers from a lack of densely labeled datasets, which are costly to obtain. Thus, models that can segment RS imagery well without supervised fine-tuning are valuable, but existing solutions fall behind supervised methods. Recently, DINOv3 surpassed SOTA RS foundation models on the GEO-bench segmentation benchmark without pre-training on RS data. Additionally, this http URL has enabled open vocabulary semantic segmentation (OVSS) with the DINOv3 backbone. We leverage these developments to form an OVSS model for RS imagery, free of RS-domain fine-tuning. Our model, CAFe-DINO (Cost Aggregation + Feature Upsampling with DINO) exploits the strong OVSS performance of DINOv3 for RS imagery via cost aggregation and training-free upsampling of text-image similarity scores. The robust latent of the DINOv3 backbone eliminates the need for fine-tuning on RS imagery; we instead fine-tune our model on a RS-targeted subset of COCO-Stuff. CAFe-DINO achieves state-of-the-art performance on key RS segmentation datasets, outperforming OVSS methods fine-tuned on RS data. Our code and data are publicly available at this https URL.

[CV-74] Boundary-Aware Uncertainty Quantification for Wildfire Spread Prediction

【速读】：该论文旨在解决现有深度学习模型在野火蔓延预测中缺乏可靠不确定性量化（Uncertainty Quantification, UQ）的问题，尤其针对边界敏感场景下仅依赖全局评估指标无法充分反映模型性能的局限性。其解决方案的关键在于提出一种空间条件化的评估框架——火中心评估区域（Fire-Centered Evaluation Region, FCER），该框架聚焦于关键火区内的不确定性表征，从而实现更贴近实际应急决策需求的UQ评估。通过FCER，作者对比了集成模型与蒸馏后的单次前向传播学生模型，在WildfireSpreadTS数据集上的表现表明，学生模型在边界相关区域具有相当的校准能力与互补的不确定性排序能力，验证了该方法的有效性。

链接: https://arxiv.org/abs/2605.03148
作者: Jonas V. Funk
机构: Independent Research (独立研究)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages, 7 figures

点击查看摘要

Abstract:Reliable wildfire spread prediction is vital for risk-aware emergency planning, yet most deep learning models lack principled uncertainty quantification (UQ). Further, for boundary-sensitive cases like wildfire spread, evaluating models with global metrics alone is often insufficient. To shift the focus of UQ evaluation toward a more operationally relevant approach, the Fire-Centered Evaluation Region (FCER) framework is introduced as a spatially conditioned protocol to characterize UQ within critical fire zones. Using FCER, an Ensemble is compared against an distilled single-pass student model on the WildfireSpreadTS dataset. The student model demonstrates comparable calibration and complementary uncertainty ranking in boundary-relevant regimes. Code is available at https://github. com/jonasvilhofunk/WildfireUQ-FCER

[CV-75] NucEval: A Robust Evaluation Framework for Nuclear Instance Segmentation

【速读】：该论文旨在解决计算病理学中核实例分割（nuclear instance segmentation）评估流程中存在的四个关键问题：模糊区域处理、分数归一化、重叠实例以及边界不确定性。针对这些问题，作者提出了一套系统性的改进方案，并将其整合进一个统一的评估框架NucEval中，从而实现对核实例分割结果的鲁棒性评估。解决方案的关键在于通过精细化的后处理策略和标准化的评价指标设计，提升评估结果的准确性和一致性，进而推动下游临床应用中模型性能的可靠比较与优化。

链接: https://arxiv.org/abs/2605.03144
作者: Amirreza Mahbod,Ramona Woitek,Jeanne Shen
机构: Danube Private University (丹布河私立大学); Stanford University (斯坦福大学)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 22 pages

点击查看摘要

Abstract:In computational pathology, nuclear instance segmentation is a fundamental task with many downstream clinical applications. With the advent of deep learning, many approaches, including convolutional neural networks (CNNs) and vision transformers (ViTs), have been proposed for this task, along with both machine learning-based and non-machine learning-based pre- and post-processing techniques to further boost performance. However, one fundamental aspect that has received less attention is the evaluation pipeline. In this study, we identify four key issues associated with nuclear instance segmentation evaluation and propose corresponding solutions. Our proposed modifications, namely handling vague regions, score normalization, overlapping instances, and border uncertainty, are integrated into a unified framework called NucEval, which enables robust evaluation of nuclear instance segmentation. We evaluate this pipeline using the NuInsSeg dataset, which provides unique characteristics that make it particularly suitable for this study, as well as two additional external datasets, with three CNN- and ViT-based nuclear instance segmentation models, to demonstrate the impact of these modifications on instance segmentation metrics. The code, along with complete guidelines and illustrative examples, is publicly available at: this https URL.

[CV-76] One Sequence to Segment Them All: Efficient Data Augmentation for CT and MRI Cross-Domain 3D Spine Segmentation

【速读】：该论文旨在解决深度学习在医学图像分割中因标注数据稀缺及跨成像协议泛化能力不足而导致的性能瓶颈问题，尤其是在MRI和CT等模态间模型迁移能力弱的问题。其解决方案的关键在于设计了一组针对性的数据增强技术，通过在单一模态/序列数据上训练模型，并在多个分布外的数据集（涵盖CT与MRI）上评估其跨模态迁移能力，从而显著提升模型对未见域的鲁棒性（平均Dice系数提升155%），同时保持域内性能几乎不变（平均Dice下降仅0.008%）。此外，为降低强数据增强带来的计算开销，作者实现了GPU优化的增强策略，在不牺牲训练效率的前提下反而提升了约10%的训练速度，最终以开源工具箱形式集成至nnUNet和MONAI等主流框架中，实现临床异质成像场景下的高效鲁棒分割。

链接: https://arxiv.org/abs/2605.03098
作者: Nathan Molinier,Hendrik Möller,Thomas Dagonneau,Anna Curto-Vilalta,Robert Graf,Matan Atad,Daniel Rueckert,Jan S. Kirschke,Julien Cohen-Adad
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:Deep learning-based medical image segmentation is increasingly used to support clinical diagnosis and develop new treatment strategies. However, model performance remains limited by the scarcity of high-quality annotated data and insufficient generalization across imaging protocols. This limitation is particularly evident in MRI and CT, where models are typically trained on a single acquisition sequence and exhibit reduced robustness when applied to unseen sequences or contrasts. Although data augmentation is widely used to improve general robustness on medical images, its impact on cross-modality generalization has not been quantitatively explored. In this work, we study a targeted set of data augmentation techniques designed to improve cross-modality transfer. We train three spine segmentation models, each on a single-modality/sequence dataset, and evaluate them across seven out-of-distribution datasets (spanning CT and MRI), reflecting a realistic single-sequence training and multi-sequence/contrast/modality deployment scenario. Our results demonstrate substantial performance gains on unseen domains (average Dice gain of 155 %) while preserving in-domain accuracy (average Dice decrease of 0.008 %), including effective transfer between CT and MRI. To mitigate the computational cost typically associated with strong data augmentation, we implement GPU-optimized augmentations that maintain, and even improve, training efficiency by approximately 10 %. We release our approach as an open-source toolbox, enabling seamless integration into commonly used frameworks such as nnUNet and MONAI. These augmentations significantly enhance robustness to heterogeneous clinical imaging scenarios without compromising training speed.

[CV-77] Learning to Segment using Summary Statistics and Weak Supervision

【速读】：该论文旨在解决医学图像分割中标注成本高、专家手动标注负担重的问题，同时在仅保留诊断统计信息（如标注区域面积）的约束条件下训练分割模型。其关键解决方案是引入一种新颖的损失函数，该函数结合了图像重建质量、与总结统计信息的匹配度以及预测前景与少量弱监督信号（即感兴趣区域内若干像素）之间的重叠程度，从而在有限监督信息下显著提升分割性能。

链接: https://arxiv.org/abs/2605.03059
作者: Omkar Kulkarni,Edward Raff,Tim Oates
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 5 pages, 2 figures, 1 table

点击查看摘要

Abstract:Medical experts often manually segment images to obtain diagnostic statistics and discard the resulting annotations. We aim to train segmentation models to alleviate this burden, but constrained to the retained summary statistics (e.g., the area of the annotated region). Empirical results suggest that statistics alone are insufficient for this task, but adding weak information in the form of a few pixels within the area of interest significantly improves performance. We use a novel loss function that combines terms for image reconstruction quality, matching to summary statistics, and overlap between the predicted foreground and the weak supervisory signal. Experiments on standard image, ultrasound (breast cancer), and Computed Tomography (CT) scan (kidney tumors) data demonstrate the utility and potential of the approach.

[CV-78] Approaching human parity in the quality of automated organoid image segmentation

【速读】：该论文旨在解决在类器官（organoid）发育过程中自动、准确测量其大小和形状的难题，尤其是在使用诱导多能干细胞（iPSCs）衍生的球状体（spheroids）时，由于形态动态变化复杂且存在多种成像条件，传统图像分割工具难以保持高精度。解决方案的关键在于提出一种融合方法：将通用基础模型Segment Anything Model (SAM) 与现有领域特定工具相结合，从而提升对不同实验条件下图像的分割一致性与准确性。该复合方法在多个测试场景中表现稳定，仅在极少数极端挑战性图像上出现误差，且其性能达到或接近人工标注者之间的变异性水平，显著优于单一现有工具。

链接: https://arxiv.org/abs/2605.03053
作者: Chase Cartwright,Gongbo Guo,Sai Teja Pusuluri,Christopher N. Mayhew,Mark Hester,Horacio E. Castillo
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Soft Condensed Matter (cond-mat.soft); Quantitative Methods (q-bio.QM)
备注: 26 pages, 18 figures

点击查看摘要

Abstract:Organoids are complex, three dimensional, self-organizing cell cultures which manifest organ-like features and represent a powerful platform for studying human disease and developing treatment options. Organoid development is characterized by dynamic morphological and cellular organization, which mimic some aspects of organ development. To study these rapid changes over the course of organoid development, advanced imaging and analytical tools are critical to accurately monitor the trajectory of organoid growth and investigate disease processes. In this work, we focus on computer vision and machine learning techniques to automatically measure the size and shape of developing spheroids derived from pluripotent stem cells (iPSCs), which are typically the starting material for generating organoid cultures. To facilitate this task, we introduce a composite method that combines the Segment Anything Model (SAM), a general-purpose foundation model, with an existing domain-specific tool. This composite method is evaluated together with several existing tools by testing them on organoid image data and comparing with the results of manual image segmentation. We find that no single existing tool is able to segment the test images with sufficient accuracy across all test conditions, but the newly introduced composite method produces consistent and accurate results for all but a very small fraction of the most challenging images. Finally, we compare the accuracy of this method to the variability between manual segmentations by independent annotators (inter-observer variability) and find that by one measure it performs at the level of inter-observer variability and by others it performs very close to it. Comments: 26 pages, 18 figures Subjects: Computer Vision and Pattern Recognition (cs.CV); Soft Condensed Matter (cond-mat.soft); Quantitative Methods (q-bio.QM) Cite as: arXiv:2605.03053 [cs.CV] (or arXiv:2605.03053v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2605.03053 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[CV-79] A Framework for Exploring and Disentangling Intersectional Bias: A Case Study in Fetal Ultrasound

【速读】：该论文旨在解决医学人工智能（AI）中因图像质量差异导致的性能偏差问题，尤其是在胎儿超声图像任务中，即使数据代表性充足，仍可能因采集条件、操作者技能及患者因素（如母亲体重指数 BMI）等多维因素交互作用而产生不公平的预测结果。其解决方案的关键在于提出一个结构化框架，融合无监督切片发现、系统性因子分析与针对性的交叉偏差评估，从而识别和量化由多个变量（如像素间距 PS、孕周 GA 和 BMI）共同作用下的复杂偏差模式。研究发现，像素间距（PS）是影响模型性能的核心因素之一，且其效应在不同 BMI 分层中持续存在，提示需采用“采集感知”与“交互感知”的评估方法以提升医疗 AI 公平性研究的严谨性。

链接: https://arxiv.org/abs/2605.02942
作者: Aya Elgebaly,Joris Fournel,Benjamin Laine Jønch Jurgensen,Kamil Mikolaj,Anders Christensen,Martin Tolsgaard,Claes Ladefoged,Aasa Feragen
机构: 未知
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV)
备注:

点击查看摘要

Abstract:Bias in medical AI is often framed as a problem of representation. However, in image-based tasks such as fetal ultrasound, performance disparities can arise even when representation is adequate, because predictive accuracy depends strongly on image quality. Image quality is shaped by acquisition conditions and operator expertise, as well as patient-dependent factors such as maternal body mass index (BMI), all of which may correlate with sensitive demographic features. Consequently, observed disparities may reflect the combined influence of demographic, clinical, and acquisition-related factors rather than data imbalance alone, and may obscure underlying interaction or confounding effects. We propose a structured framework to explore and detect intersectional bias, combining unsupervised slice discovery, systematic factor-wise analysis, and targeted intersectional evaluation. In a case study of over 94,000 ultrasound images for fetal weight estimation, we analyze bias in a state-of-the-art deep learning (DL) model and the clinical standard Hadlock, a regression formula using biometric measurements. Pixel spacing (PS) – a parameter considered suboptimal in current acquisition protocols – emerged as a consistent driver of performance differences, with higher PS associated with improvements of up to 24% in selected subgroups for both models. Because PS is often adapted in cases of high BMI or low gestational age (GA), this effect carries a substantial risk of confounding. Our intersectional analysis revealed that part of the PS-associated signal is explained by GA, while PS-related improvements persist across BMI strata, highlighting the importance of acquisition-aware and interaction-aware evaluation in medical AI fairness research.

[CV-80] Where to Bind Matters: Hebbian Fast Weights in Vision Transformers for Few-Shot Character Recognition

【速读】：该论文旨在解决标准Transformer架构在元学习任务中缺乏快速适应能力的问题，即其固定慢权重（slow-weight）表示难以在单个推理episode内实现高效调整。解决方案的关键在于引入Hebbian快权重（Hebbian Fast-Weight, HFW）模块，通过模拟生物神经系统的突触快速更新机制，在推理阶段形成瞬态关联记忆，从而增强模型对少量样本的快速适应能力。研究发现，将HFW模块仅放置于Swin-Tiny模型的最后一阶段特征图上可避免多阶段部署导致的训练不稳定，并在Omniglot数据集上实现了96.2%（1-shot）和99.2%（5-shot）的最高测试准确率，显著优于对应非Hebbian基线模型。

链接: https://arxiv.org/abs/2605.02920
作者: Gavin Money,Sindhuja Penchala,Jiacheng Li,Noorbakhsh Amiri Golilarz
机构: 未知
类目: Neural and Evolutionary Computing (cs.NE); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:

点击查看摘要

Abstract:Standard transformer architectures learn fixed slow-weight representations during training and lack mechanisms for rapid adaptation within an episode. In contrast, biological neural systems address this through fast synaptic updates that form transient associative memories during inference, a property known as Hebbian plasticity. In this paper, we conduct an empirical study of Hebbian Fast-Weight (HFW) modules integrated into multiple transformer backbones, including ViT-Small, DeiT-Small, and Swin-Tiny. We evaluate six model variants: ViT, DeiT, Swin, ViT-Hebbian, DeiT-Hebbian, and Swin-Hebbian on 5-way 1-shot and 5-way 5-shot classification tasks using the Omniglot benchmark under a Prototypical Network meta-learning framework. We propose a single module placement strategy for Swin-Tiny in which one HFW module is applied to the final stage feature map after all hierarchical stages have completed. This design avoids the training instability caused by placing separate Hebbian modules at each stage and achieves the highest test accuracy across all six models (96.2% at 1-shot; 99.2% at 5-shot), outperforming its non-Hebbian baseline by +0.3 percentage points at 1-shot. We analyze the interaction between Swin’s shifted window inductive bias and episode-level Hebbian binding, discuss why per-block placement fails for ViT and DeiT variants in a low-data regime, and situate the results within the wider literature on fast and slow-weight meta-learning.

[CV-81] Reasoning -Guided Grounding: Elevating Video Anomaly Detection through Multimodal Large Language Models

【速读】：该论文旨在解决视频异常检测（Video Anomaly Detection, VAD）中长期存在的两大问题：一是传统方法仅提供二分类或异常检测结果，缺乏可解释的推理过程和异常事件的精确空间定位；二是视觉语言模型（Vision-Language Models, VLMs）在进行空间定位时容易产生幻觉或几何无效的边界框。解决方案的关键在于提出VANGUARD框架，其核心创新为通过三阶段课程训练策略，统一异常分类、空间定位与链式思维（chain-of-thought）推理三个任务于单一VLM中：首先冻结骨干特征进行分类器预热，其次使用LoRA适配实现空间定位，最后生成结构化推理轨迹；同时引入教师-学生标注管道结合Qwen3-VL-4B和GroundingDINO提升弱标注数据下的监督信号质量，从而在UCF-Crime等基准上实现高精度分类（94% ROC-AUC）、强定位能力（84% F1）及可解释性推理，且无需目标域微调即可跨域泛化。

链接: https://arxiv.org/abs/2605.02912
作者: Sakshi Agarwal,Aishik Konwer,Ankit Parag Shah
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: under review at conference

点击查看摘要

Abstract:Video Anomaly Detection (VAD) has traditionally been framed as binary classification or outlier detection, providing neither interpretable reasoning nor precise spatial localization of anomalous events. While Vision-Language Models (VLMs) offer rich scene understanding, they struggle with reliable spatial grounding - often producing hallucinated or geometrically invalid bounding boxes when asked to localize objects. We propose VANGUARD (Video Anomaly Understanding through Reasoning and Grounding), a framework that unifies anomaly classification, spatial grounding, and chain-of-thought reasoning within a single VLM. VANGUARD introduces a three-stage curriculum that progressively layers training objectives: (1) classifier warmup on frozen backbone features, (2) LoRA-adapted spatial grounding, and (3) chain-of-thought generation. To overcome the sparse annotation typical of VAD benchmarks, we employ a teacher-student annotation pipeline in which a VLM (Qwen3-VL-4B) generates structured per-subclip reasoning trajectories based on manual annotations available from the UCA Dataset. Further, GroundingDINO provides bounding box supervision. On UCF-Crime, VANGUARD achieves 94% ROC-AUC with 84% F1 while simultaneously producing interpretable chain-of-thought explanations and spatial grounding of anomalous objects - capabilities absent from prior VAD methods. Ablations confirm that staged training outperforms monolithic optimization, and that structured reasoning acts as an implicit regularizer yielding more balanced predictions than classification-only fine-tuning. Zero-shot transfer to XD-Violence and ShanghaiTech demonstrates cross-domain generalization without target-domain adaptation.

[CV-82] Memorization In Stable Diffusion Is Unexpectedly Driven by CLIP Embeddings CVPR2026

【速读】：该论文旨在解决文本到图像扩散模型（如Stable Diffusion）中由CLIP嵌入（CLIP embeddings）引发的过度记忆问题，特别是探究输入标记（token）嵌入在生成过程中的作用机制。研究发现，尽管提示词嵌入（ $\mathbf{v}^{\mathbf{pr}}$ ）对记忆案例贡献甚微，但填充符嵌入（\mathbf{v}^{\mathbfpad}）因结构上复制了结束符嵌入（\mathbf{v}^{\mathbfeot}），而后者是CLIP训练中唯一显式优化的目标，从而被意外放大并主导生成过程，导致模型过度依赖特定嵌入实现记忆。解决方案的关键在于：在推理阶段通过简单策略干预嵌入影响——一是将填充符从结束符替换为感叹号（!）token并掩码\mathbf{v}^{\mathbfeot}，二是部分掩码\mathbf{v}^{\mathbfpad}，二者均能有效抑制记忆现象而不损害图像质量，且无需预先检测即可直接部署。

链接: https://arxiv.org/abs/2605.02908
作者: Bumjun Kim,Albert No
机构: Yonsei University (延世大学)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted at CVPR 2026 Findings. Code is available at this https URL

点击查看摘要

Abstract:Understanding how textual embeddings contribute to memorization in text-to-image diffusion models is crucial for both interpretability and safety. This paper investigates an unexpected behavior of CLIP embeddings in Stable Diffusion, revealing that the model disproportionately relies on specific embeddings. We categorize input tokens as startoftext, prompt, endoftext and pad with corresponding embeddings \mathbfv^\mathbfsot, \mathbfv^\mathbfpr, \mathbfv^\mathbfeot, \mathbfv^\mathbfpad . We discover that \mathbfv^\mathbfpr contribute minimally to generation in memorized cases. In contrast, \mathbfv^\mathbfpad strongly affect memorization due to their structural duplication of \mathbfv^\mathbfeot , the only embedding explicitly optimized during CLIP training. This duplication unintentionally amplifies the influence of \mathbfv^\mathbfeot , causing the model to over-rely on it, thereby driving memorization. Based on these observations, we propose two simple yet effective inference-time mitigation strategies: (1) Replacing the tokenizer’s default pad from eot to the ! token before embedding, and masking the \mathbfv^\mathbfeot ; (2) Partial masking of \mathbfv^\mathbfpad . Both suppress memorization without degrading quality, and are readily deployable without prior detection.

[CV-83] Safety in Embodied AI: A Survey of Risks Attacks and Defenses

【速读】：该论文旨在解决具身人工智能（Embodied AI）在开放世界、安全关键场景中面临的系统性安全挑战，尤其关注感知、认知、规划、行动与人机交互等环节中的潜在攻击与防御问题。其核心解决方案在于构建一个多层级的安全分类体系（multi-level taxonomy），将分散的研究成果整合为统一框架，并连接具身智能特有安全发现与视觉、语言及多模态基础模型的最新进展。通过系统梳理400余篇文献，论文揭示了如多模态感知融合脆弱性、对抗性扰动下规划不稳定性和开放场景中人机交互可信度不足等被忽视的关键挑战，从而为开发具备自主性、鲁棒性和可靠性的具身智能体提供清晰的研究路线图。

链接: https://arxiv.org/abs/2605.02900
作者: Xiao Li,Xiang Zheng,Yifeng Gao,Xinyu Xia,Yixu Wang,Xin Wang,Ye Sun,Yunhan Zhao,Ming Wen,Jiayu Li,Xun Gong,Yi Liu,Yige Li,Yutao Wu,Cong Wang,Jun Sun,Yixin Cao,Zhineng Chen,Jingjing Chen,Tao Gui,Qi Zhang,Zuxuan Wu,Xipeng Qiu,Xuanjing Huang,Tiehua Zhang,Zhipeng Wei,Hanxun Huang,Sarah Erfani,James Bailey,Jianping Wang,Wei-Ying Ma,Bo Li,Xingjun Ma,Yu-Gang Jiang
机构: 未知
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: 51 pages, 4 figures, 19 tables. Project page: this https URL

点击查看摘要

Abstract:Embodied Artificial Intelligence (Embodied AI) integrates perception, cognition, planning, and interaction into agents that operate in open-world, safety-critical environments. As these systems gain autonomy and enter domains such as transportation, healthcare, and industrial or assistive robotics, ensuring their safety becomes both technically challenging and socially indispensable. Unlike digital AI systems, embodied agents must act under uncertain sensing, incomplete knowledge, and dynamic human-robot interactions, where failures can directly lead to physical harm. This survey provides a comprehensive and structured review of safety research in embodied AI, examining attacks and defenses across the full embodied pipeline, from perception and cognition to planning, action and interaction, and agentic system. We introduce a multi-level taxonomy that unifies fragmented lines of work and connects embodied-specific safety findings with broader advances in vision, language, and multimodal foundation models. Our review synthesizes insights from over 400 papers spanning adversarial, backdoor, jailbreak, and hardware-level attacks; attack detection, safe training and robust inference; and risk-aware human-agent interaction. This analysis reveals several overlooked challenges, including the fragility of multimodal perception fusion, the instability of planning under jailbreak attacks, and the trustworthiness of human-agent interaction in open-ended scenarios. By organizing the field into a coherent framework and identifying critical research gaps, this survey provides a roadmap for building embodied agents that are not only capable and autonomous but also safe, robust, and reliable in real-world deployment.

[CV-84] Structured Analytic Coherent Point Drift for Non-Rigid Point Set Registration

【速读】：该论文旨在解决非刚性点集配准（non-rigid point set registration）中传统相干点漂移（Coherent Point Drift, CPD）方法因基于点索引的高斯核位移场导致参数规模庞大、计算效率低的问题。其解决方案的关键在于提出 Analytic-CPD，通过将 CPD 的 M-step 中的点索引高斯核位移场替换为有限维结构化解析映射估计器，利用高斯混合模型后验概率通过重心恒等式转化为加权软目标点，从而将原始的成对软对应目标转化为加权解析拟合问题；同时采用截断的多元泰勒映射表示形变，使参数数量由环境维度和解析阶数控制，而非依赖于移动点间的 M×M 核系统，结合阶次延续策略稳定大变形注册过程。此方法在二维解析与三维平滑非解析形变场景下均表现出更低误差和更快收敛速度，验证了概率对应关系与结构化解析映射组合的有效性。

链接: https://arxiv.org/abs/2605.00934
作者: Wei Feng,Haiyong Zheng
机构: Ocean University of China (中国海洋大学)
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (stat.ML)
备注:

点击查看摘要

Abstract:We introduce Analytic-CPD, a structured analytic variant of coherent point drift for non-rigid point set registration. The method retains the CPD posterior correspondence layer, but replaces the point-indexed Gaussian-kernel displacement-field M-step with a finite-dimensional structured analytic mapping estimator. Posterior probabilities from the Gaussian mixture model are condensed through a barycentric identity into weighted soft target points, converting the CPD pairwise soft-correspondence objective into a weighted analytic fitting problem. The deformation is represented by a truncated multivariate Taylor mapping of a vector-valued function, so the number of deformation parameters is controlled by the ambient dimension and the analytic order rather than by an M-by-M kernel system over the moving points. A degree-continuation strategy is further introduced to stabilize large-deformation registration by progressively activating higher-order analytic modes. Experiments on two-dimensional analytic deformations and three-dimensional smooth non-analytic deformations show that Analytic-CPD achieves lower final errors and faster convergence than standard CPD in representative large-deformation settings. The results suggest that CPD-style probabilistic correspondences and structured analytic mappings provide a compact and interpretable alternative to kernel-based non-rigid registration. Code is available at this https URL.

[CV-85] An Intelligent Framework for Real-Time Yoga Pose Detection and Posture Correction

【速读】：该论文旨在解决自导式或在线瑜伽训练中因姿势执行不当导致的效果降低和运动损伤风险增加的问题。其核心解决方案是提出一种基于边缘智能（Edge AI）的混合框架，通过轻量级人体姿态估计模型结合生物力学特征提取与CNN-LSTM时序学习架构，实现瑜伽体式识别与动作动态分析；关键创新在于利用检测到的关键点计算关节角度和骨骼特征，并与标准姿势配置对比以量化对齐偏差，进而生成视觉、文本及语音形式的实时矫正反馈，同时采用模型量化与剪枝等优化技术确保在资源受限设备上的低延迟运行，从而提升用户训练的安全性与有效性。

链接: https://arxiv.org/abs/2603.26760
作者: Chandramouli Haldar
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Digital Libraries (cs.DL)
备注:

点击查看摘要

Abstract:Yoga is widely recognized for improving physical fitness, flexibility, and mental well being. However, these benefits depend strongly on correct posture execution. Improper alignment during yoga practice can reduce effectiveness and increase the risk of musculoskeletal injuries, especially in self guided or online training environments. This paper presents a hybrid Edge AI based framework for real time yoga pose detection and posture correction. The proposed system integrates lightweight human pose estimation models with biomechanical feature extraction and a CNN LSTM based temporal learning architecture to recognize yoga poses and analyze motion dynamics. Joint angles and skeletal features are computed from detected keypoints and compared with reference pose configurations to evaluate posture correctness. A quantitative scoring mechanism is introduced to measure alignment deviations and generate real time corrective feedback through visual, text based, and voice based guidance. In addition, Edge AI optimization techniques such as model quantization and pruning are applied to enable low latency performance on resource constrained devices. The proposed framework provides an intelligent and scalable digital yoga assistant that can improve user safety and training effectiveness in modern fitness applications.

[CV-86] Robustness and Transferability of Pix2Geomodel for Bidirectional Facies Property Translation in a Complex Reservoir

【速读】：该论文旨在解决储层地质建模中因条件数据稀疏、地质异质性强以及传统地统计学流程难以捕捉相带与物性参数之间非线性关系而导致的建模精度不足问题。其解决方案的关键在于提出并验证了Pix2Geomodel框架，该框架基于Pix2Pix图像到图像翻译模型（包含U-Net生成器和PatchGAN判别器），通过构建双向映射任务（如相带到孔隙度、孔隙度到相带等），在有限垂直分辨率条件下仍能有效保留相带-物性之间的空间连续性和地质结构特征，从而实现复杂储层中快速、可靠的双向相-物性转换。

链接: https://arxiv.org/abs/2605.03919
作者: Abdulrahman Al-Fakih,Nabil Sariah,Ardiansyah Koeshidayatullah,Sherif Hanafy,SanLinn I. Kaka
机构: 未知
类目: Geophysics (physics.geo-ph); Computational Complexity (cs.CC); Computer Vision and Pattern Recognition (cs.CV); Computational Physics (physics.comp-ph)
备注:

点击查看摘要

Abstract:Reservoir geomodeling is central to subsurface characterization, but it remains challenging because conditioning data are sparse, geological heterogeneity is strong, and conventional geostatistical workflows often struggle to capture nonlinear relationships between facies and petrophysical properties. This study evaluates the robustness and transferability of Pix2Geomodel on a different and more complex reservoir dataset with reduced vertical support. The new case includes a heterogeneous reservoir-quality classification and only 54 retained layers, providing a stricter test of whether Pix2Pix-based image-to-image translation can preserve facies-property relationships under constrained data conditions. Facies, porosity, permeability, and clay volume (VCL) were extracted from a reference reservoir model, exported as aligned two-dimensional slices, augmented using consistent geometric transformations, and assembled into paired image datasets. Six bidirectional tasks were evaluated: facies to porosity, facies to permeability, facies to VCL, porosity to facies, permeability to facies, and VCL to facies. The Pix2Pix model, consisting of a U-Net generator and PatchGAN discriminator, was evaluated using image-based metrics, visual comparison, and variogram-based spatial-continuity validation. Results show that the model preserves the dominant geological architecture and main spatial-continuity trends. Facies to porosity achieved the highest pixel accuracy and frequency-weighted intersection over union of 0.9326 and 0.8807, while VCL to facies achieved the highest mean pixel accuracy and mean intersection over union of 0.8506 and 0.7049. These findings show that Pix2Geomodel can transfer beyond its original case study as a practical framework for rapid bidirectional facies-property translation in complex reservoir modeling.

[CV-87] A Partition-Based Generating Function for Row-Convex Polyominoes

【速读】：该论文旨在解决无内部孔洞的行凸多格（row-convex polyominoes）在离散网格上的计数问题，即如何精确枚举给定面积下的所有此类多格结构。其解决方案的关键在于提出一种替代生成函数方法，将多格的面积分解为整数分拆（integer partitions），每个分拆对应一组连续行长度序列，并通过各部分排列的乘积来刻画相邻行之间的所有可能水平对齐方式；最终通过对这些乘积求和，得到指定面积下多格总数的闭式表达式。该方法建立了整数分拆与多格枚举之间的直接联系，从而为精确计数和渐近分析提供了简洁而有效的框架。

链接: https://arxiv.org/abs/2605.03203
作者: Vincenzo M. Scarrica
机构: 未知
类目: Combinatorics (math.CO); Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:An alternative generating function is proposed to enumerate row-convex polyominoes without internal holes on a discrete grid. The approach is based on integer partitions of the total area, where each partition corresponds to a sequence of row lengths, and the product of all permutations of the parts accounts for all possible horizontal alignments of consecutive rows. Summing over the products yields a formula for the total number of convex polyominoes of a given size. Numerical examples are provided for small areas, and the exact generating function is derived via a transfer series argument, establishing the asymptotic growth S(N) as A2^(N) cos(N*theta) + phi) with theta = arctan(sqrt(7)/3). The method establishes a direct connection between integer partitions and polyomino enumeration, offering a simple yet effective framework for both exact and asymptotic combinatorial analysis. Potential applications include shape priors in discrete image analysis, grid-based modeling, and combinatorial generation of convex structures.

[CV-88] EMOVIS: Emotion-Optimized Image Processing ICIP2026

【速读】：该论文旨在解决传统图像信号处理器（Image Signal Processor, ISP）在视频拍摄过程中仅注重场景保真度而忽略情感表达的问题，从而无法实现如电影制作中通过色彩分级、对比度和亮度等视觉属性增强情绪叙事的需求。其解决方案的关键在于提出一种名为EMOVIS（EMotion-Optimized VISual processing）的实时视觉处理框架，通过建立高阶情绪状态（如快乐、平静、愤怒、悲伤）与低阶ISP控制参数（如色饱和度、局部色调映射和锐度）之间的系统性映射关系，并借助用户校准实验验证了各参数在不同情绪下具有统计显著的影响效果；进一步设计了一种不改变ISP底层处理流程的控制架构，将情绪驱动的调整无缝集成至标准ISP硬件中，实验证明在目标情绪与场景语境匹配时，87%的盲测用户偏好经过情绪优化的渲染结果，显著提升了视觉内容的情感适配性。

链接: https://arxiv.org/abs/2605.03131
作者: Dor Barber,Rony Zatzarinni,Hava Matichin,Noam Levy
机构: 未知
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to ICIP 2026

点击查看摘要

Abstract:In cinematography, visual attributes such as color grading, contrast, and brightness are manipulated to reinforce the emotional narrative of a scene. However, conventional Image Signal Processors (ISPs) prioritize scene fidelity, effectively neglecting this expressive dimension. To bring this cinematic capability to real-time camera pipelines during video capture, we introduce EMOVIS (EMotion-Optimized VISual processing). We establish a systematic mapping between a compact set of high-level emotional states (Happy, Calm, Angry, Sad) and low-level ISP controls - including color saturation, local tone mapping, and sharpness - supported by a calibration user study with statistically significant effects across parameters. We propose a control framework that integrates these emotion-driven adjustments into standard ISP hardware without altering the underlying processing stages. Validation via blind A/B testing shows that viewers prefer the emotion-optimized rendering in 87% of trials when the target emotion matches the scene context, indicating that emotion-aligned ISP control improves perceived suitability for expressive visual content.

[CV-89] Video Generation Models as World Models: Efficient Paradigms Architectures and Algorithms

【速读】：该论文旨在解决视频生成模型在世界模拟（world modeling）中理论潜力与实际计算成本之间的效率差距问题，尤其针对时空建模带来的高资源消耗。其解决方案的关键在于提出一个三维的新型分类体系：高效建模范式（efficient modeling paradigms）、高效网络架构（efficient network architectures）以及高效推理算法（efficient inference algorithms），从而系统性地提升视频生成模型的效率，并推动其在自动驾驶、具身智能和游戏模拟等交互式应用中的落地。

链接: https://arxiv.org/abs/2603.28489
作者: Muyang He,Hanzhong Guo,Junxiong Lin,Yizhou Yu
机构: School of Computing and Data Science, The University of Hong Kong(香港大学计算与数据科学学院); Hong Kong Generative AI Research and Development Center(香港生成式人工智能研发与中心)
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注:

点击查看摘要

Abstract:The rapid evolution of video generation has enabled models to simulate complex physical dynamics and long-horizon causalities, positioning them as potential world simulators. However, a critical gap still remains between the theoretical capacity for world simulation and the heavy computational costs of spatiotemporal modeling. To address this, we comprehensively and systematically review video generation frameworks and techniques that consider efficiency as a crucial requirement for practical world modeling. We introduce a novel taxonomy in three dimensions: efficient modeling paradigms, efficient network architectures, and efficient inference algorithms. We further show that bridging this efficiency gap directly empowers interactive applications such as autonomous driving, embodied AI, and game simulation. Finally, we identify emerging research frontiers in efficient video-based world modeling, arguing that efficiency is a fundamental prerequisite for evolving video generators into general-purpose, real-time, and robust world simulators.

人工智能

[AI-0] Redefining AI Red Teaming in the Agent ic Era: From Weeks to Hours

【速读】：该论文旨在解决当前AI红队（AI red teaming）实践中存在的效率低下问题，即操作员需手动构建攻击流程（包括攻击方法、变换策略和评分机制），导致大量时间耗费在流程搭建而非实际漏洞探测上。解决方案的关键在于提出一个基于Dreadnode SDK的智能代理（Agentic interface），其能够根据自然语言描述的目标自动选择攻击策略、组合变换规则并执行测试与报告，从而将原本需数周的手工流程压缩至数小时；同时，该方案构建了一个统一框架，可同时适用于传统机器学习模型（对抗样本）和生成式AI系统（越狱攻击），消除了对多个专用库的依赖，并通过Llama Scout案例验证了其有效性——实现了85%的攻击成功率及最高1.0的严重性等级，且无需人工编写代码。

链接: https://arxiv.org/abs/2605.04019
作者: Raja Sekhar Rao Dheekonda,Will Pearce,Nick Landers
机构: 未知
类目: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
备注: 39 pages, 8 figures

点击查看摘要

Abstract:AI systems are entering critical domains like healthcare, finance, and defense, yet remain vulnerable to adversarial attacks. While AI red teaming is a primary defense, current approaches force operators into manual, library-specific workflows. Operators spend weeks hand-crafting workflows - assembling attacks, transforms, and scorers. When results fall short, workflows must be rebuilt. As a result, operators spend more time constructing workflows than probing targets for security and safety vulnerabilities. We introduce an AI red teaming agent built on the open-source Dreadnode SDK. The agent creates workflows grounded in 45+ adversarial attacks, 450+ transforms, and 130+ scorers. Operators can probe multi-agent systems, multilingual, and multimodal targets, focusing on what to probe rather than how to implement it. We make three contributions: 1. Agentic interface. Operators describe goals in natural language via the Dreadnode TUI (Terminal User Interface). The agent handles attack selection, transform composition, execution, and reporting, letting operators focus on red teaming. Weeks compress to hours. 2. Unified framework. A single framework for probing traditional ML models (adversarial examples) and generative AI systems (jailbreaks), removing the need for separate libraries. 3. Llama Scout case study. We red team Meta Llama Scout and achieve an 85% attack success rate with severity up to 1.0, using zero human-developed code Comments: 39 pages, 8 figures Subjects: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR) Cite as: arXiv:2605.04019 [cs.AI] (or arXiv:2605.04019v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2605.04019 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[AI-1] SymptomAI: Towards a Conversational AI Agent for Everyday Symptom Assessment

【速读】：该论文旨在解决当前生成式 AI (Generative AI) 在临床诊断任务中表现优异但主要基于结构化病例数据，难以反映真实世界患者症状描述与诊断准确性的核心问题。其解决方案的关键在于设计并部署 SymptomAI——一套端到端的对话式 AI 系统，通过在 Fitbit 应用中随机分配 13,917 名参与者与其交互，收集来自真实人群的多样化症状表达和疾病分布数据；特别地，采用“专用症状访谈”策略（即由 AI 主导、主动挖掘更多症状信息后再进行鉴别诊断），相比用户引导式对话显著提升诊断准确性（OR = 2.47, p < 0.001），验证了系统性采集症状信息对提升诊断性能的重要性，并进一步发现生理指标变化与急性感染存在强关联（如流感 OR > 7）。

链接: https://arxiv.org/abs/2605.04012
作者: Joseph Breda,Fadi Yousif,Beszel Hawkins,Marinela Cotoi,Miao Liu,Ray Luo,Po-Hsuan Cameron Chen,Mike Schaekermann,Samuel Schmidgall,Xin Liu,Girish Narayanswamy,Samuel Solomon,Maxwell A. Xu,Xiaoran Fan,Longfei Shangguan,Anran Wang,Bhavna Daryani,Buddy Herkenham,Cara Tan,Mark Malhotra,Shwetak Patel,John B. Hernandez,Quang Duong,Yun Liu,Zach Wasson,Dimitrios Antos,Bob Lou,Matthew Thompson,Jonathan Richina,Anupam Pathak,Nichole Young-Lin,Jake Sunshine,Daniel McDuff
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注: 13 page main text, 54 pages total. 16 figures total

点击查看摘要

Abstract:Language models excel at diagnostic assessments on currated medical case-studies and vignettes, performing on par with, or better than, clinical professionals. However, existing studies focus on complex scenarios with rich context making it difficult to draw conclusions about how these systems perform for patients reporting symptoms in everyday life. We deployed SymptomAI, a set of conversational AI agents for end-to-end patient interviewing and differential diagnosis (DDx), via the Fitbit app in a study that randomized participants (N=13,917) to interact with five AI agents. This corpus captures diverse communication and a realistic distribution of illnesses from a real world population. A subset of 1,228 participants reported a clinician-provided diagnosis, and 517 of these were further evaluated by a panel of clinicians during over 250 hours of annotation. SymptomAI DDx were significantly more accurate (OR = 2.47, p 0.001) than those from independent clinicians given the same dialogue in a blinded randomized comparison. Moreover, agentic strategies which conduct a dedicated symptom interview that elicit additional symptom information before providing a diagnosis, perform substantially better than baseline, user-guided conversations (p 0.001). An auxiliary analysis on 1,509 conversations from a general US population panel validated that these results generalize beyond wearable device users. We used SymptomAI diagnoses as labels for all 13,917 participants to analyze over 500,000 days of wearable metrics across nearly 400 unique conditions. We identified strong associations between acute infections and physiological shifts (e.g., OR 7 for influenza). While limited by self-reported ground truth, these results demonstrate the benefits of a dedicated and complete symptom interview compared to a user-guided symptom discussion, which is the default of most consumer LLMs.

[AI-2] An Agent -Oriented Pluggable Experience-RAG Skill for Experience-Driven Retrieval Strategy Orchestration

【速读】：该论文旨在解决检索增强生成（Retrieval-Augmented Generation, RAG）系统中普遍存在的“一刀切”检索策略问题，即固定单一检索管道难以适应不同任务类型（如事实型问答、多跳推理和科学验证）所带来的差异化检索需求。其解决方案的关键在于提出一种面向代理的可插拔检索编排层——Experience-RAG Skill，该技能能够根据当前任务场景分析、调用经验记忆（experience memory）并动态选择最优检索策略，从而为代理提供结构化的证据输出。实验表明，在不改变候选检索器池的前提下，该方法在BeIR基准上的nDCG@10达到0.8924，显著优于固定单检索器基线，并保持与自适应路由（Adaptive-RAG-style routing）相当的性能，证明了将检索策略选择封装为可复用的代理技能具有高度可行性与有效性。

链接: https://arxiv.org/abs/2605.03989
作者: Dutao Zhang,Tian Liao
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注: Preprint. 6 pages, 1 figure, 3 tables

点击查看摘要

Abstract:Retrieval-augmented generation systems often assume that one fixed retrieval pipeline is sufficient across heterogeneous tasks, yet factoid question answering, multi-hop reasoning, and scientific verification exhibit different retrieval preferences. We present Experience-RAG Skill, an agent-oriented pluggable retrieval orchestration layer positioned between the agent and the retriever pool. The proposed skill analyzes the current scene, consults an experience memory, selects an appropriate retrieval strategy, and returns structured evidence to the agent. Under a fixed candidate pool, Experience-RAG Skill achieves an overall nDCG@10 of 0.8924 on BeIR/nq, BeIR/hotpotqa, and BeIR/scifact, outperforming fixed single-retriever baselines and remaining competitive with Adaptive-RAG-style routing. The results suggest that retrieval strategy selection can be productively encapsulated as a reusable agent skill rather than being hard-coded in the upper workflow.

[AI-3] From Intent to Execution: Composing Agent ic Workflows with Agent Recommendation

【速读】：该论文旨在解决多智能体系统（Multi-Agent Systems, MAS）构建过程中高度依赖人工干预的问题，包括手动规划、人工选择合适智能体以及手动创建执行图等繁琐步骤。其核心解决方案是提出一个自动化框架，通过集成大语言模型（LLM）驱动的规划器、任务描述模块、动态调用图、智能体调度器和两阶段信息检索（Information Retrieval, IR）机制的智能体推荐模块，实现从任务理解到执行的端到端自动化。其中，关键创新在于引入了一个监督式批判代理（critique agent），对整体计划中的智能体与工具推荐进行全局复核与修正，显著提升了系统的召回率（recall rate）、鲁棒性和可扩展性。

链接: https://arxiv.org/abs/2605.03986
作者: Kishan Athrey,Ramin Pishehvar,Brian Riordan,Mahesh Viswanathan
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Multi-Agent Systems (MAS) built using AI agents fulfill a variety of user intents that may be used to design and build a family of related applications. However, the creation of such MAS currently involves manual composition of the plan, manual selection of appropriate agents, and manual creation of execution graphs. This paper introduces a framework for the automated creation of multi-agent systems which replaces multiple manual steps with an automated framework. The proposed framework consists of software modules and a workflow to orchestrate the requisite task- specific application. The modules include: an LLM-derived planner, a set of tasks described in natural language, a dynamic call graph, an orchestrator for map agents to tasks, and an agent recommender that finds the most suitable agent(s) from local and global agent registries. The agent recommender uses a two-stage information retrieval (IR) system comprising a fast retriever and an LLM-based re-ranker. We implemented a series of experiments exploring the choice of embedders, re- rankers, agent description enrichment, and supervising critique agent. We benchmarked this system end-to-end, evaluating the combination of planning, agent selection, and task completion, with our proposed approach. Our experimental results show that our approach outperforms the state-of-the- art in terms of the recall rate and is more robust and scalable compared to previous approaches. The critique agent holistically reevaluates both agent and tool recommendations against the overall plan. We show that the inclusion of the critique agent further enhances the recall score, proving that the comprehensive review and revision of task-based agent selection is an essential step in building end-to-end multi-agent systems.

[AI-4] Flow Sampling: Learning to Sample from Unnormalized Densities via Denoising Conditional Processes ICML2026

【速读】：该论文旨在解决从无归一化密度函数中高效采样的问题，这在生成式建模中具有重要意义，尤其当目标分布由已知能量函数定义而非数据样本时。传统方法因能量函数评估成本高而难以扩展，因此核心挑战在于学习一个计算高效的采样器。其解决方案的关键在于提出Flow Sampling框架，该框架基于扩散模型（diffusion models）和流匹配（flow matching）技术，利用噪声样本作为条件并回归至由能量函数构造的去噪扩散漂移（denoising diffusion drift），从而避免了对数据样本的依赖；同时引入插值过程（interpolant process）最小化训练期间的能量函数评估次数，显著提升了效率与可扩展性。此外，该方法自然延伸至黎曼流形（Riemannian manifolds），并在常曲率空间（如超球面和双曲空间）中推导出条件漂移的闭式表达式，实现了几何结构更复杂的采样任务。

链接: https://arxiv.org/abs/2605.03984
作者: Aaron Havens,Brian Karrer,Neta Shaul
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: To appear at ICML 2026 (spotlight)

点击查看摘要

Abstract:Sampling from unnormalized densities is analogous to the generative modeling problem, but the target distribution is defined by a known energy function instead of data samples. Because evaluating the energy function is often costly, a primary challenge is to learn an efficient sampler. We introduce Flow Sampling, a framework built on diffusion models and flow matching for the data-free setting. Our training objective is conditioned on a noise sample and regresses onto a denoising diffusion drift constructed from the energy function. In contrast, diffusion models’ objective is conditioned on a data sample and regresses onto a noising diffusion drift. We utilize the interpolant process to minimize the number of energy function evaluations during training, resulting in an efficient and scalable method for sampling unnormalized densities. Furthermore, our formulation naturally extends to Riemannian manifolds, enabling diffusion-based sampling in geometries beyond Euclidean space. We derive a closed-form formula for the conditional drift on constant curvature manifolds, including hyperspheres and hyperbolic spaces. We evaluate Flow Sampling on synthetic energy benchmarks, small peptides, large-scale amortized molecular conformer generation, and distributions supported on the sphere, demonstrating strong empirical performance.

[AI-5] Inconsistent Databases and Argumentation Frameworks with Collective Attacks

【速读】：该论文旨在解决不一致数据库在存在多种完整性约束（Integrity Constraints, ICs）时，如何通过论证框架（Argumentation Frameworks, AFs）来刻画其修复（repairs）的问题。具体而言，研究聚焦于否认约束（denial constraints）与局部视图元组生成依赖（local-as-view tuple-generating dependencies）的组合情形下，Subset-maximal repairs 与论证框架中扩展（extensions）之间的对应关系。解决方案的关键在于引入SET-based Argumentation Frameworks（SETAFs），这是一种允许集体攻击的论证框架扩展形式，能够准确建模tuple-generating dependencies下的修复行为；研究发现，在仅含否认约束时，修复对应于SETAF中的naive、preferred和stable扩展；而在考虑tuple-generating dependencies时，修复则对应于preferred扩展；进一步地，通过预处理可获得唯一稳定且naive的扩展；当两类约束共存时，唯有preferred语义能捕捉修复，此时预处理失效。此外，论文还证明了函数依赖（functional dependencies）和包含依赖（inclusion dependencies）均无需集合攻击，因此可直接映射为普通AFs，从而简化计算复杂度。

链接: https://arxiv.org/abs/2605.03954
作者: Yasir Mahmood,Jonni Virtema,Timon Barlag,Axel-Cyrille Ngonga Ngomo
机构: 未知
类目: Databases (cs.DB); Artificial Intelligence (cs.AI)
备注: This is a pre-print of the paper accepted at the Knowledge Engineering Review journal

点击查看摘要

Abstract:The connection between subset-maximal repairs for inconsistent databases involving various integrity constraints and acceptable sets of arguments within argumentation frameworks has recently drawn growing interest. In this paper, we contribute to this domain by establishing a new connection when integrity constraints (ICs) include denial constraints and local-as-view tuple-generating dependencies. It turns out that SET-based Argumentation Frameworks (SETAFs), an extension of Dung’s argumentation frameworks (AFs) allowing collective attacks, are needed. It is known that subset-maximal repairs under denial constraints correspond to the naive extensions, which also coincide with the preferred and stable extensions in the resulting SETAFs. Our main findings establish that repairs under the considered fragment of tuple-generating dependencies correspond to the preferred extensions. Moreover, for these dependencies, additional preprocessing allows computing a unique extension that is stable and naive. Allowing both types of constraints breaks this relationship, and even the pre-processing does not help as only preferred semantics captures these repairs. Finally, while it is known that functional dependencies do not require set-based attacks, we prove the same regarding inclusion dependencies. Thus, one can translate inconsistent databases under these restricted classes of ICs to plain AFs with attacks only between arguments.

[AI-6] MOSAIC-Bench: Measuring Compositional Vulnerability Induction in Coding Agents

【速读】：该论文旨在解决生成式 AI（Generative AI）在软件开发场景中因任务分解导致的安全漏洞问题，即编码代理（coding agent）虽通过单次提示（per-prompt）安全审查，但在将复杂任务拆分为多个看似无害的工程工单（engineering ticket）后，仍可能输出可被利用的恶意代码。其核心挑战在于现有安全对齐方法仅孤立评估显式请求，忽略了由一系列看似合法的请求序列所诱导出的隐蔽恶意目标（malicious objectives）。解决方案的关键是提出 MOSAIC-Bench 基准测试集，该基准包含 199 条三阶段攻击链，并结合部署环境中的确定性漏洞验证器（exploit oracle），将漏洞真实情况与下游代码评审协议共同作为评估维度。实验表明，即使在严格的直接提示条件下，主流编码代理仍会以高达 20.4% 的脆弱输出率生成漏洞代码；而代码评审代理进一步批准了 25.8% 的已确认漏洞变更，说明单纯依赖上下文完整性无法解释漏洞泄露现象。最终，通过将评审者重新定义为对抗性渗透测试人员（adversarial pentester），可在不依赖模型自适应能力的前提下显著降低攻击成功率（降至 3.0%-17.6%），并实现高检测率（如 Gemma-4-E4B-it 模型达 88.4%）和低误报率（4.6%）。

链接: https://arxiv.org/abs/2605.03952
作者: Jonathan Steinberg,Oren Gal
机构: 未知
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注:

点击查看摘要

Abstract:Coding agents often pass per-prompt safety review yet ship exploitable code when their tasks are decomposed into routine engineering tickets. The challenge is structural: existing safety alignment evaluates overt requests in isolation, leaving models blind to malicious end-states that emerge from sequenced compliance with innocuous-looking requests. We introduce MOSAIC-Bench (Malicious Objectives Sequenced As Innocuous Compliance), a benchmark of 199 three-stage attack chains paired with deterministic exploit oracles on deployed software substrates (10 web-application substrates, 31 CWE classes, 5 programming languages) that treats both exploit ground truth and downstream reviewer protocol as first-class evaluation axes. On this benchmark, nine production coding agents from Anthropic, OpenAI, Google, Moonshot, Zhipu, and Minimax compose innocuous tickets at 53-86% end-to-end ASR with only two refusals across all staged runs. In a matched direct-prompt experiment over four frontier Claude/Codex agents, vulnerable-output rates fall to 0-20.4%: Claude primarily refuses, while Codex primarily hardens rather than emitting the vulnerable implementation - ticket staging silences both defense modes simultaneously. Downstream, code reviewer agents approve 25.8% of these confirmed-vulnerable cumulative diffs as routine PRs, and a full-context implementation protocol closes only 50% of the staged/direct gap, ruling out context fragmentation as the sole explanation. As a deployable but non-adaptive mitigation, reframing the reviewer as an adversarial pentester reduces evasion across the evaluated reviewer subset; pentester framed evasion ranges from 3.0% to 17.6%, and an open-weight Gemma-4-E4B-it reviewer under this framing detects 88.4% of attacks on the dataset with a 4.6% false-positive rate measured on 608 real-world GitHub PRs.

[AI-7] abSurv: Adapting Modern Tabular Neural Networks to Survival Analysis

【速读】：该论文旨在解决生存分析（Survival Analysis）中深度学习方法任务特异性过强、难以迁移且性能受限的问题。现有方法往往针对特定任务设计，限制了跨领域应用并可能影响预测效果。其解决方案的关键在于提出TabSurv框架，该框架通过适配现代表格数据神经网络架构，结合Weibull分布参数化或非参数生存预测策略，并引入一种支持删失数据的新型直方图损失函数（SurvHL），实现对生存时间分布的有效建模。此外，TabSurv采用并行训练的多层感知机（MLP）深度集成方法，在优化各组件的生存分布参数后再进行平均，从而增强模型多样性与鲁棒性。实证结果表明，TabSurv在10个真实世界生存数据集上显著优于经典和主流深度学习基线方法（如RSF、DeepSurv、DeepHit等），尤其基于Weibull参数化的深度集成模型在C-index指标上表现最优，验证了该方法的有效性和通用性。

链接: https://arxiv.org/abs/2605.03944
作者: Stanislav Kirpichenko,Andrei Konstantinov,Lev Utkin
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
备注:

点击查看摘要

Abstract:Survival analysis on tabular data is a well-studied problem. However, existing deep learning methods are often highly task-specific, which can limit the transfer of new approaches from other domains and introduce constraints that may affect performance. We propose TabSurv, an approach that adapts modern tabular architectures to survival analysis using either the Weibull distribution or non-parametric survival prediction. TabSurv optimizes SurvHL, a novel histogram loss function supporting censored data. In addition to a baseline feed-forward network, we implement deep ensembles of MLPs for survival analysis within TabSurv. In contrast to prior work, the ensemble components are trained in parallel, optimizing survival distribution parameters before averaging, which promotes diversity across ensemble component predictions. We perform a comprehensive empirical evaluation of different proposed architectures on 10 diverse real-world survival datasets. Our results show that TabSurv consistently outperforms on average established classical and deep learning baselines, such as RSF, DeepSurv, DeepHit, SurvTRACE. Notably, deep ensembles with Weibull parametrization instead of non-parametric models achieve the highest average rank by C-index. Overall, our study clarifies how modern tabular neural networks can be adapted and trained to tackle survival analysis problems, offering a strong and reliable approach. The TabSurv implementation is publicly available.

[AI-8] owards Open World Sound Event Detection

【速读】：该论文旨在解决传统声事件检测（Sound Event Detection, SED）系统在开放世界场景下性能受限的问题，即现有方法通常假设所有目标事件均在训练阶段已知（闭世界假设），难以应对现实环境中频繁出现的未知声事件。为实现对已知事件的检测、未知事件的识别以及增量学习能力，作者提出开放世界声事件检测（Open-World Sound Event Detection, OW-SED）范式，并设计了基于一维可变形架构的WOOT（Open-World Deformable Sound Event Detection Transformer）框架。其解决方案的关键在于：引入可变形注意力机制以自适应聚焦于关键时间区域，通过特征解耦分离类别特定与类别无关表示，结合一对多匹配策略和多样性损失提升表征多样性，从而有效应对重叠与模糊事件带来的挑战，在闭世界设置中表现优于现有方法，并在开放世界场景中显著超越基线模型。

链接: https://arxiv.org/abs/2605.03934
作者: P.H.Hai,L.T.Minh,L.H.Son
机构: 未知
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注: 32 pages, 3 figures. Submitted to Signal Processing (Elsevier)

点击查看摘要

Abstract:Sound Event Detection (SED) plays a vital role in audio understanding, with applications in surveillance, smart cities, healthcare, and multimedia indexing. However, conventional SED systems operate under a closed-world assumption, limiting their effectiveness in real-world environments where novel acoustic events frequently emerge. Inspired by the success of open-world learning in computer vision, we introduce the Open-World Sound Event Detection (OW-SED) paradigm, where models must detect known events, identify unseen ones, and incrementally learn from them. To tackle the unique challenges of OW-SED, such as overlapping and ambiguous events, we propose a 1D Deformable architecture that leverages deformable attention to adaptively focus on salient temporal regions. Furthermore, we design a novel Open-World Deformable Sound Event Detection Transformer (WOOT) framework incorporating feature disentanglement to separate class-specific and class-agnostic representations, together with a one-to-many matching strategy and a diversity loss to enhance representation diversity. Experimental results demonstrate that our method achieves marginally superior performance compared to existing leading techniques in closed-world settings and significantly improves over existing baselines in open-world scenarios.

[AI-9] PHALAR: Phasors for Learned Musical Audio Representations

【速读】：该论文旨在解决音频子混音（audio submix）中缺失声部（stem）检索的问题，当前方法因忽略时间信息而性能受限。其解决方案的关键在于提出PHALAR框架，通过引入学习型频谱池化层（Learned Spectral Pooling layer）和复数域输出头（complex-valued head），显式建模音高等变性（pitch-equivariance）与相位等变性（phase-equivariance）先验，从而在显著降低参数量（减少50%）和加速训练（提升7倍）的同时，实现相对准确率提升约70%，并在MoisesDB、Slakh和ChocoChorales数据集上建立新的检索性能基准。

链接: https://arxiv.org/abs/2605.03929
作者: Davide Marincione,Michele Mancusi,Giorgio Strano,Luca Cerovaz,Donato Crisostomi,Roberto Ribuoli,Emanuele Rodolà
机构: 未知
类目: ound (cs.SD); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Signal Processing (eess.SP)
备注:

点击查看摘要

Abstract:Stem retrieval, the task of matching missing stems to a given audio submix, is a key challenge currently limited by models that discard temporal information. We introduce PHALAR, a contrastive framework achieving a relative accuracy increase of up to \approx 70% over the state-of-the-art while requiring 50% of the parameters and a 7 \times training speedup. By utilizing a Learned Spectral Pooling layer and a complex-valued head, PHALAR enforces pitch-equivariant and phase-equivariant biases. PHALAR establishes new retrieval state-of-the-art across MoisesDB, Slakh, and ChocoChorales, correlating significantly higher with human coherence judgment than semantic baselines. Finally, zero-shot beat tracking and linear chord probing confirm that PHALAR captures robust musical structures beyond the retrieval task.

[AI-10] Contextual Multi-Objective Optimization: Rethinking Objectives in Frontier AI Systems

【速读】：该论文旨在解决当前前沿人工智能（AI）系统在开放性任务中表现不可靠的问题，尤其是在目标模糊、上下文依赖性强或难以直接观测的场景下（如科学辅助、长期代理、高风险建议等），这些系统往往因错误的目标选择而失效，而非单纯的能力不足。其核心问题是：如何在多目标、动态变化的环境中正确识别并权衡不同目标优先级，以实现稳健的行为决策。解决方案的关键在于提出一种“情境化多目标优化”（contextual multi-objective optimization）框架，将AI行为建模为基于上下文的行动选择规则，明确区分活跃目标、软偏好与硬约束，并通过分解目标表示、上下文到目标路由、分层约束机制、推理驱动策略、可控个性化、工具使用控制及诊断评估等模块实现可解释、可审计且可迭代优化的决策过程。

链接: https://arxiv.org/abs/2605.03900
作者: Jie Zhou,Qin Chen,Liang He
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Frontier AI systems perform best in settings with clear, stable, and verifiable objectives, such as code generation, mathematical reasoning, games, and unit-test-driven tasks. They remain less reliable in open-ended settings, including scientific assistance, long-horizon agents, high-stakes advice, personalization, and tool use, where the relevant objective is ambiguous, context-dependent, delayed, or only partially observable. We argue that many such failures are not merely failures of scale or capability, but failures of objective selection: the system optimizes a locally visible signal while missing which objectives should govern the interaction. We formulate this problem as \emphcontextual multi-objective optimization. In this setting, systems must consider multiple, context-dependent objectives, such as helpfulness, truthfulness, safety, privacy, calibration, non-manipulation, user preference, reversibility, and stakeholder impact, while determining which objectives are active, which are soft preferences, and which must function as hard or quasi-hard constraints. These examples are not intended as an exhaustive taxonomy: different domains and deployment settings may activate different objective dimensions and different conflict-resolution procedures. Our framework models AI behavior as a context-dependent choice rule over candidate actions, objective estimates, active constraints, stakeholders, uncertainty, and conflict-resolution procedures. We outline an implementation pathway based on decomposed objective representations, context-to-objective routing, hierarchical constraints, deliberative policy reasoning, controlled personalization, tool-use control, diagnostic evaluation, auditing, and post-deployment revision.

[AI-11] Spatiotemporal Convolutions on EEG signal – A Representation Learning Perspective on Efficient and Explainable EEG Classification with Convolutional Neural Nets

【速读】：该论文旨在解决基于浅层卷积神经网络（Shallow Convolutional Neural Networks, CNNs）对脑电图（EEG）信号进行分类时，传统独立一维（1D）卷积在空间和时间维度上分别处理所导致的训练效率低下问题。其解决方案的关键在于引入二维（2D）时空卷积（spatiotemporal convolution），即同时在空间和时间维度上进行联合卷积操作，而非将两个1D卷积串联使用。尽管数值上等价于两个1D卷积的串联，但实验表明2D卷积在高维（22通道）脑机接口（BCI）运动想象任务中显著缩短了训练时间且保持性能不变，并揭示出不同架构导致内部表征几何结构存在显著差异，强调了模型架构对复杂多变量信号编码的重要影响，而不仅仅是最终性能指标。

链接: https://arxiv.org/abs/2605.03874
作者: Laurits Dixen,Stefan Heinrich,Paolo Burelli
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Classification of EEG signals using shallow Convolutional Neural Networks (CNNs) is a prevalent and successful approach across a variety of fields. Most of these models use independent one-dimensional (1D) convolutional layers along the spatial and temporal dimensions, which are concatenated without a non-linear activation layer between. In this paper, we investigate an alternative encoding that operates a bi-dimensional (2D) spatiotemporal convolution. While 2D convolutions are numerically identical to two concatenated 1D convolutions along the two dimensions, the impact on learning is still uncertain. We test 1D and 2D CNNs and a CNN+transformer hybrid model in a low-dimensional (3-channel) and a high-dimensional (22-channel) BCI motor imagery classification task. We observe that 2D convolutions significantly reduce training time in high-dimensional tasks while maintaining performance. We investigate the root of this improvement and find no difference in spectral feature importance. However, a clear pattern emerges in representational similarity across models: 1D and 2D models yield vastly different representational geometries. Overall, we suggest an improved model with a 2D convolutional layer for faster training and inference. We also highlight the importance of architecturally-driven encoding when processing complex multivariate signals, as reflected in internal representations rather than purely in performance metrics.

[AI-12] EvoLM: Self-Evolving Language Models through Co-Evolved Discriminative Rubrics

【速读】：该论文旨在解决当前语言模型后训练方法依赖外部监督信号（如人工标注、专有API或标量奖励模型）所带来的局限性，这些外部信号在能力覆盖范围、可扩展性和可验证性上存在瓶颈。其核心解决方案是提出EVOLM，一种基于模型自身评估能力的自监督后训练框架：关键在于将模型内在的评价知识结构化为显式的判别性评分标准（rubrics），并通过交替训练两个模块实现自我提升——一是生成器模块优化生成针对具体样本的判别性评价标准以最大化冻结裁判模型区分偏好与非偏好响应的能力；二是策略模块利用这些条件化的评分作为奖励进行训练。整个过程完全基于模型自身输出的时间对比机制构建偏好信号，无需人工标注或外部监督，从而实现了无监督的持续优化。

链接: https://arxiv.org/abs/2605.03871
作者: Shuyue Stella Li,Rui Xin,Teng Xiao,Yike Wang,Rulin Shao,Zoey Hao,Melanie Sclar,Sewoong Oh,Faeze Brahman,Pang Wei Koh,Yulia Tsvetkov
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注: 32 pages, 2 figures, 21 tables

点击查看摘要

Abstract:Language models encode substantial evaluative knowledge from pretraining, yet current post-training methods rely on external supervision (human annotations, proprietary models, or scalar reward models) to produce reward signals. Each imposes a ceiling. Human judgment cannot supervise capabilities beyond its own, proprietary APIs create dependencies, and verifiable rewards cover only domains with ground-truth answers. Self-improvement from a model’s own evaluative capacity is a reward source that scales with the model itself, yet remains largely untapped by current methods. We introduce EVOLM, a post-training method that structures this capacity into explicit discriminative rubrics and uses them as training signal. EVOLM trains two capabilities within a single language model in alternation: (1) a rubric generator producing instance-specific evaluation criteria optimized for discriminative utility, which maximizes a small frozen judge’s ability to distinguish preferred from dispreferred responses; and (2) a policy trained using those rubric-conditioned scores as reward. All preference signals are constructed from the policy’s own outputs via temporal contrast with earlier checkpoints, requiring no human annotation or external supervision. EVOLM trains a Qwen3-8B model to generate rubrics that outperform GPT-4.1 on RewardBench-2 by 25.7%. The co-trained policy achieves 69.3% average on the OLMo3-Adapt suite, outperforming policies trained with GPT-4.1 prompted rubrics by 3.9% and with the state-of-the-art 8B reward model SkyWork-RM by 16%. Overall, EVOLM demonstrates that structuring a model’s evaluative capacity into co-evolving discriminative rubrics enables self-improvement without external supervision.

[AI-13] Mechanical Conscience: A Mathematical Framework for Dependability of Machine Intelligenc

【速读】：该论文旨在解决分布式协同智能（Distributed Collaborative Intelligence, DCI）系统中因局部个体决策在不确定性下组合而产生的全局不可接受行为轨迹问题，即“涌现风险”（emergent risk）。现有方法如约束优化、安全强化学习和运行时保障等仅关注单个动作层面的可接受性，未能有效应对DCI多参与方与高不确定性共存的复杂场景。解决方案的关键在于提出“机械良心”（Mechanical Conscience, MC）这一新概念及其简化数学框架：MC是一种监督滤波器，通过最小化基线策略动作对规范可接受区域的累积偏离，在考虑认知不确定性（epistemic uncertainty）的前提下实现轨迹级规范调控；其核心机制包括良心得分（conscience score）、机械内疚（mechanical guilt）和共振可靠性（resonant dependability）等可计算治理信号，确保单智能体及多智能体DCI环境中行为轨迹保持规范可接受性，并自然抑制交互引发的涌现风险。

链接: https://arxiv.org/abs/2605.03847
作者: Munkhdegerekh Batzorig,Purevbaatar Ganbold,Kyungbin Park,Pilkong Jeong,Kangbin
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注: 9 pages, 2 figures. Preprint

点击查看摘要

Abstract:Distributed collaborative intelligence (DCI), encompassing edge-to-edge architectures, federated learning, transfer learning, and swarm systems, creates environments in which emergent risk is structurally unavoidable: locally correct decisions by individual agents compose into globally unacceptable behavioral trajectories under uncertainty. Existing approaches such as constrained optimization, safe reinforcement learning, and runtime assurance evaluate acceptability at the level of individual actions rather than across behavioral trajectories, and none addresses the multi-participant, uncertainty-laden nature of DCI deployments. This paper introduces mechanical conscience (MC), a novel concept and simplified mathematical framework that operationalizes trajectory-level normative regulation for both single-agent and distributed intelligent systems. Mechanical conscience is defined as a supervisory filter that minimally corrects a baseline policy’s actions to reduce cumulative deviation from a normatively admissible region, while accounting for epistemic uncertainty. We introduce associated constructs, conscience score, mechanical guilt, and resonant dependability, that provide an interpretable vocabulary and computable governance signals for this emerging field. Core theoretical properties are established: admissibility equivalence, existence of optimal regulation, and monotonic deviation reduction. Illustrative results demonstrate that MC-regulated agents maintain trajectory-level normative acceptability where conventional controllers drift outside admissible bounds, and that the framework naturally extends to suppress interaction-induced emergent risk in multi-agent DCI settings.

[AI-14] SOAR: Real-Time Joint Optimization of Order Allocation and Robot Scheduling in Robotic Mobile Fulfillm ent Systems

【速读】：该论文旨在解决机器人移动分拣系统（Robotic Mobile Fulfillment Systems, RMFS）中订单分配与机器人调度的联合优化问题，该问题因严格的实时性约束和多阶段决策间的强耦合性而极具挑战。现有方法要么将问题分解为孤立子任务以保证响应速度但牺牲全局最优性，要么依赖计算开销巨大的全局优化模型，难以适应动态工业场景。解决方案的关键在于提出SOAR——一种统一的深度强化学习框架，通过将订单分配与调度转化为一个事件驱动的马尔可夫决策过程（Event-Driven Markov Decision Process），利用软订单分配作为观测输入实现同步调度；技术上采用异构图Transformer编码仓库状态并融合阶段领域知识，同时引入奖励塑形策略缓解长时任务中的稀疏反馈问题，从而在保证亚100ms延迟的前提下显著提升整体效率（如全局完工时间减少7.5%，平均订单完成时间降低15.4%）。

链接: https://arxiv.org/abs/2605.03842
作者: Yibang Tang,Yifan Yang,Jingyuan Wang,Junhua Chen,Zhen Zhao
机构: 未知
类目: Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注: 13 pages, 6 figures

点击查看摘要

Abstract:Robotic Mobile Fulfillment Systems (RMFS) rely on mobile robots for automated inventory transportation, coordinating order allocation and robot scheduling to enhance warehousing efficiency. However, optimizing RMFS is challenging due to strict real-time constraints and the strong coupling of multi-phase decisions. Existing methods either decompose the problem into isolated sub-tasks to guarantee responsiveness at the cost of global optimality, or rely on computationally expensive global optimization models that are unsuitable for dynamic industrial environments. To bridge this gap, we propose SOAR, a unified Deep Reinforcement Learning framework for real-time joint optimization. SOAR transforms order allocation and robot scheduling into a unified process by utilizing soft order allocations as observations. We formulate this as an Event-Driven Markov Decision Process, enabling the agent to perform simultaneous scheduling in response to asynchronous system events. Technically, we employ a Heterogeneous Graph Transformer to encode the warehouse state and integrate phased domain knowledge. Additionally, we incorporate a reward shaping strategy to address sparse feedback in long-horizon tasks. Extensive experiments on synthetic and real-world industrial datasets, in collaboration with Geekplus, demonstrate that SOAR reduces global makespan by 7.5% and average order completion time by 15.4% with sub-100ms latency. Furthermore, sim-to-real deployment confirms its practical viability and significant performance gains in production environments. The code is available at this https URL.

[AI-15] RoboAlign-R1: Distilled Multimodal Reward Alignment for Robot Video World Models

【速读】：该论文旨在解决现有机器人视频世界模型（robot video world models）在训练目标上与实际机器人决策能力（如指令遵循、操作成功率和物理合理性）不匹配，以及长时程自回归预测中误差累积的问题。其核心解决方案是提出RoboAlign-R1框架，包含两个关键技术：一是通过构建RobotWorldBench基准和多模态教师判别器（RoboAlign-Judge）实现细粒度六维评估，并将教师模型蒸馏为轻量级奖励模型用于强化学习后训练，从而对齐任务目标；二是引入无需额外训练的滑动窗口重编码（Sliding Window Re-encoding, SWR）策略，在推理阶段周期性刷新生成上下文以抑制长期预测漂移。实验表明，该方法显著提升了任务一致性、物理真实性及长时程预测质量。

链接: https://arxiv.org/abs/2605.03821
作者: Hao Wu,Yuqi Li,Yuan Gao,Fan Xu,Fan Zhang,Kun Wang,Penghao Zhao,Qiufeng Wang,Yizhou Zhao,Weiyan Wang,Yingli Tian,Xian Wu,Xiaomeng Huang
机构: 未知
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Existing robot video world models are typically trained with low-level objectives such as reconstruction and perceptual similarity, which are poorly aligned with the capabilities that matter most for robot decision making, including instruction following, manipulation success, and physical plausibility. They also suffer from error accumulation in long-horizon autoregressive prediction. We present RoboAlign-R1, a framework that combines reward-aligned post-training with stabilized long-horizon inference for robot video world models. We construct RobotWorldBench, a benchmark of 10,000 annotated video-instruction pairs collected from four robot data sources, and train a multimodal teacher judge, RoboAlign-Judge, to provide fine-grained six-dimensional evaluation of generated videos. We then distill the teacher into a lightweight student reward model for efficient reinforcement-learning-based post-training. To reduce long-horizon rollout drift, we further introduce Sliding Window Re-encoding (SWR), a training-free inference strategy that periodically refreshes the generation context. Under our in-domain evaluation protocol, RoboAlign-R1 improves the aggregate six-dimension score by 10.1% over the strongest baseline, including gains of 7.5% on Manipulation Accuracy and 4.6% on Instruction Following; these ranking improvements are further supported by an external VLM-based cross-check and a blinded human study. Meanwhile, SWR improves long-horizon prediction quality with only about 1% additional latency, yielding a 2.8% gain in SSIM and a 9.8% reduction in LPIPS. Together, these results show that reward-aligned post-training and stabilized long-horizon decoding improve task consistency, physical realism, and long-horizon prediction quality in robot video world models.

[AI-16] ScrapMem: A Bio-inspired Framework for On-device Personalized Agent Memory via Optical Forgetting

【速读】：该论文旨在解决在资源受限的边缘设备上实现大语言模型（Large Language Model, LLM）代理的长期个性化记忆所面临的高存储成本与多模态复杂性难题。其核心解决方案是提出ScrapMem框架，关键创新在于引入“光学遗忘”（Optical Forgetting）机制，通过逐步降低旧记忆的分辨率来压缩存储空间，同时保持语义一致性；此外，构建事件记忆图（Episodic Memory Graph, EM-Graph），以因果-时序结构组织关键事件，从而提升记忆检索效率和准确性。

链接: https://arxiv.org/abs/2605.03804
作者: Jiale Chang,Yuxiang Ren
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注: 10 pages, 4 figures

点击查看摘要

Abstract:Long-term personalized memory for LLM agents is challenging on resource-limited edge devices due to high storage costs and multimodal complexity. To address this, we propose ScrapMem, a framework that integrates multimodal data into “Scrapbook Page.” ScrapMem introduces Optical Forgetting, an optical compression mechanism that progressively reduces the resolution of older memories, lowering storage cost while suppressing low-value details. To maintain semantic consistency, we construct an Episodic Memory Graph (EM-Graph) that organizes key events into a causal-temporal structure. Extensive experiments on the multimodal ATM-Bench showcase that ScrapMem provides three main benefits: (1) strong performance, achieving a new state-of-the-art with a 51.0% Joint@10 score; (2) high storage efficiency, reducing memory usage by up to 93% via optical forgetting; and (3) improved recall, increasing Recall@10 to 70.3% through structured aggregation. ScrapMem offers an effective and storage-efficient solution for on-device long-term memory in multimodal LLM agents.

[AI-17] AI Advocate: Educational Path to Transform Squads to the Future

【速读】：该论文旨在解决传统软件开发团队向人机协同（human-AI collaboration）模式转型过程中所面临的文化与技术障碍问题。其解决方案的关键在于通过系统性培训和赋能“AI倡导者”（AI Advocates）这一角色，使其成为推动组织内部认知转变与技能升级的核心力量，从而实现从单一人力开发向人机协作型团队的结构性变革。

链接: https://arxiv.org/abs/2605.03800
作者: Carla Soares,Gabriel Moreira,Ana Paula Camargo,Fabio Henrique Scacabarozi,Nicole Davila,Marselle Silva
机构: 未知
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:This paper analyzes the strategic education process aimed at transitioning traditional software development squads into hybrid structures centered on collaborative work between humans and Artificial Intelligence (AI). In a context where human-AI collaboration can significantly increase productivity, this study explores how the upskilling of XPTO professionals, referred to as AI Advocates, acts as a catalyst for cultural and technical transformation. The objective is to present an experience report on the education and enablement process of AI Advocates within a private Brazilian technology company, highlighting key lessons learned and identified challenges.

[AI-18] Say the Mission Execute the Swarm: Agent -Enhanced LLM Reasoning in the Web-of-Drones

【速读】：该论文旨在解决大语言模型（Large Language Models, LLMs）在实时无人飞行器（UAV）集群管理中应用时面临的挑战，包括异构接口、缺乏具身化（grounding）以及长时间闭环执行的可靠性问题。其解决方案的关键在于提出一个任务无关、基于代理增强的LLM框架，通过引入模型上下文协议（Model Context Protocol, MCP）网关和基于W3C Web of Things（WoT）标准的“无人机网络”抽象结构，将无人机、传感器和服务统一为标准化的WoT Thing，从而实现结构化的工具交互、持续的状态观测与安全执行，无需依赖代码生成。实验表明，仅靠LLM的推理能力不足以保障可靠执行，而任务特定规划工具和运行时约束机制显著提升了系统鲁棒性。

链接: https://arxiv.org/abs/2605.03788
作者: Andrea Iannoli,Lorenzo Gigli,Luca Sciullo,Angelo Trotta,Marco Di Felice
机构: 未知
类目: Artificial Intelligence (cs.AI); Networking and Internet Architecture (cs.NI); Robotics (cs.RO)
备注: 15 pages, 5 figures. This paper has been accepted for presentation at the 27th IEEE International Symposium on a World of Wireless, Mobile and Multimedia Networks (WoWMoM 2026)

点击查看摘要

Abstract:Large Language Models (LLMs) are increasingly explored as high-level reasoning engines for cyber-physical systems, yet their application to real-time UAV swarm management remains challenging due to heterogeneous interfaces, limited grounding, and the need for long-running closed-loop execution. This paper presents a mission-agnostic, agent-enhanced LLM framework for UAV swarm control, where users express mission objectives in natural language and the system autonomously executes them through grounded, real-time interactions. The proposed architecture combines an LLM-based Agent Core with a Model Context Protocol (MCP) gateway and a Web-of-Drones abstraction based on W3C Web of Things (WoT) standards. By exposing drones, sensors, and services as standardized WoT Things, the framework enables structured tool-based interaction, continuous state observation, and safe actuation without relying on code generation. We evaluate the framework using ArduPilot-based simulation across four swarm missions and six state-of-the-art LLMs. Results show that, despite strong reasoning abilities, current general-purpose LLMs still struggle to achieve reliable execution - even for simple swarm tasks - when operating without explicit grounding and execution support. Task-specific planning tools and runtime guardrails substantially improve robustness, while token consumption alone is not indicative of execution quality or reliability.

[AI-19] What You Think is What You See: Driving Exploration in VLM Agents via Visual-Linguistic Curiosity ICML2026

【速读】：该论文旨在解决当前视觉语言模型（Visual Language Model, VLM）代理在部分可观测视觉环境中，因仅依赖已访问状态的被动式思维链（Chain-of-Thought, CoT）推理而导致的稀疏奖励任务表现不佳的问题。其核心挑战在于缺乏主动探索机制以发现“已知的未知”，从而难以实现鲁棒泛化。解决方案的关键在于提出GLANCE框架，通过将代理的语言世界模型与不断演化的目标网络（target network）的稳定视觉表征相锚定，利用语言预测与视觉现实之间的差异作为内在好奇心信号，在强化学习中引导代理主动探索其内部模型不确定的区域，从而实现推理与探索的统一。

链接: https://arxiv.org/abs/2605.03782
作者: Haoxi Li,Qinglin Hou,Jianfei Ma,Jinxiang Lai,Tao Han,Sikai Bai,Jingcai Guo,Jie Zhang,Song Guo
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注: Accepted to ICML 2026 (Spotlight)

点击查看摘要

Abstract:To navigate partially observable visual environments, recent VLM agents increasingly internalize world modeling capabilities into their policies via explicit CoT reasoning, enabling them to mentally simulate futures before acting. However, relying solely on passive reasoning over visited states is insufficient for sparse-reward tasks, as it lacks the epistemic drive to actively uncover the known unknown'' required for robust generalization. We ask: Can VLM agents actively find signals that challenge and refine their internal world model through curiosity-driven exploration? In this work, we propose GLANCE, a unified framework that bridges reasoning and exploration by grounding the agent's linguistic world model into the stable visual representations of an evolving target network. Crucially, GLANCE leverages the discrepancy between linguistic prediction and visual reality as an intrinsic curiosity signal within reinforcement learning, steering the agent to actively explore areas where its internal model is uncertain. Extensive experiments across a series of agentic tasks show the effectiveness of GLANCE, and demonstrate that aligning what the agent thinks’’ with ``what the agent sees’’ is key to solving complex or sparse agentic tasks.

[AI-20] OracleProto: A Reproducible Framework for Benchmarking LLM Native Forecasting via Knowledge Cutoff and Temporal Masking

【速读】：该论文旨在解决大语言模型（Large Language Models, LLMs）在现实决策支持系统中评估其原生预测能力（forecasting capability）的难题。现有方法存在两大局限：一是实时基准测试（live benchmarks）虽能真实反映预测性能，但事件一旦发生即失效，难以复现；二是回溯性基准测试（retrospective benchmarks）虽可重复使用，却无法可靠区分模型是否真正基于未知信息进行预测，而仅靠提示模型“假装不知道”无法替代真实的知识边界。为解决此问题，作者提出OracleProto框架，其关键在于通过多维度机制重建时间受限的预测样本：包括模型截止点对齐的样本准入策略、工具级的时间掩码（temporal masking）、内容级泄露检测、离散答案归一化以及分层评分体系，从而在控制信息边界的同时显著降低残余泄露至1%以下（较仅用工具级时间过滤提升一个数量级）。该方案使LLM预测能力从一次性评估转变为可审计、可复用、可训练的数据集级能力，为公平跨模型比较及下游监督微调（SFT）和强化学习（RL）提供可控信号源。

链接: https://arxiv.org/abs/2605.03762
作者: Yiding Ma,Chengyun Ruan,Kaibo Huang,Zhongliang Yang,Linna Zhou
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Large language models are moving from static text generators toward real-world decision-support systems, where forecasting is a composite capability that links information gathering, evidence integration, situational judgment, and action-oriented decision making. This capability is in broad demand across finance, policy, industry, and scientific research, yet its evaluation remains difficult: live benchmarks evaluate forecasts before answers exist, making them the cleanest way to measure forecasting ability, but they expire once events resolve; retrospective benchmarks are reproducible, but they cannot reliably distinguish genuine forecasting from facts a model may have already learned during pretraining. Prompting models to “pretend not to know” cannot replace a genuine knowledge boundary. We propose OracleProto, a reproducible framework for evaluating LLM native forecasting capability. OracleProto reconstructs resolved events into time-bounded forecasting samples by combining model-cutoff-aligned sample admission, tool-level temporal masking, content-level leakage detection, discrete answer normalization, and hierarchical scoring. Instantiated on a FutureX-Past-derived dataset with six contemporary LLMs, OracleProto distinguishes forecasting quality, sampling stability, and cost efficiency under controlled information boundaries, while reducing residual leakage to the 1% level, an order of magnitude below tool-only temporal filtering. OracleProto turns LLM forecasting from one-off evaluation into an auditable, reusable, and trainable dataset-level capability, providing a unified interface for fair cross-model comparison and a controlled signal source for downstream SFT and RL. Code and data are available at this https URL and this https URL.

[AI-21] Rethinking the Rank Threshold for LoRA Fine-Tuning

【速读】：该论文旨在解决LoRA（Low-Rank Adaptation）微调在神经切线核（Neural Tangent Kernel, NTK）理论框架下，针对不同任务类型（尤其是二分类与多分类）所需最小秩（rank）的问题。传统分析给出的条件为 $ r(r+1)/2 \geq KN $，即在标准少样本RoBERTa设置中需 $ r \geq 12 $ 才能避免伪局部极小值，但这一条件未考虑实际使用的交叉熵损失函数，且其紧致性在特定场景中尚不明确。本文提出三项关键改进：首先，通过将对称Sard形式的计数替换为非对称LoRA流形维度，得到更弱的容量约束 $ r(m+n) - r^2 \geq C^* \cdot KN $（其中 $ C^* \approx 1.35 $），使得 $ r=1 $ 即可满足二分类场景；其次，在交叉熵损失下，利用Polyak–Łojasiewicz不等式彻底消除了对秩的阈值要求；最后，基于Rademacher复杂度边界预测出当偏置项饱和时rank=1达到最优方差性能——这恰好发生在二分类情形而非多分类（$ K > 2 $）。实验证明，在四个GLUE风格的二分类任务、三种编码器架构及RoBERTa-large模型上，$ r=1 $ 已具竞争力，而多分类MNLI任务则显示最优秩高于1，符合理论预期。整体而言，该研究显著降低了二分类LoRA微调的理论秩需求，并揭示了交叉熵损失下的结构优势。

链接: https://arxiv.org/abs/2605.03724
作者: Juneyoung Park
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:A recent landscape analysis of LoRA fine-tuning in the neural tangent kernel regime establishes a sufficient condition r(r+1)/2 KN on the LoRA rank r for the absence of spurious local minima under squared-error loss, prescribing r \geq 12 on canonical few-shot RoBERTa setups. The condition is stated for general output dimension K , so its sharpness in any particular regime, and its practical implication for the cross-entropy loss actually used in fine-tuning, are open. We give three results that together reduce the prescribed rank to r = 1 for binary classification in this regime. First, replacing the symmetric Sard-form count with the non-symmetric LoRA manifold dimension yields a strictly weaker capacity requirement, r(m+n) - r^2 C^* \cdot KN with C^* \approx 1.35 under Gaussian-iid features, satisfied at r = 1 on canonical setups. Second, in the cross-entropy setting the Polyak–Łojasiewicz inequality removes the rank threshold entirely. Third, a Rademacher-complexity bound predicts rank-one variance optimality precisely when the bias term is saturated, which is the case for binary classification but not for K 2 . Empirically, across four GLUE-style binary tasks, three encoder architectures, and at scale on RoBERTa-large, rank one is competitive with the existing prescription r = 12 ; on multi-class MNLI the optimal rank shifts above one, also as predicted. The binary-regime guarantees are conditional on standard NTK assumptions; the multi-class extension is left to future work.

[AI-22] ailored Prompts Targeted Protection: Vulnerability-Specific LLM Analysis for Smart Contracts

【速读】：该论文旨在解决智能合约（Smart Contracts）在区块链上因存在多种安全漏洞而导致重大财务损失的问题，尤其是现有检测方法在漏洞类型间缺乏灵活性且高度依赖人工编写的专家规则。其解决方案的关键在于提出一种基于大语言模型（LLM）的框架，通过构建包含31,165个专业标注漏洞实例的大规模数据集，并结合精确的抽象语法树（AST）上下文提取与针对特定漏洞类型的提示工程设计，实现对13类常见漏洞的定制化检测器部署，从而在保证高精度的同时提升检测的可扩展性与实用性。

链接: https://arxiv.org/abs/2605.03697
作者: Xing Zhang,Keyu Zhang,Taohong Zhu,Anbang Ruan
机构: 未知
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Smart contracts on blockchains are prone to diverse security vulnerabilities that can lead to significant financial losses due to their immutable nature. Existing detection approaches often lack flexibility across vulnerability types and rely heavily on manually crafted expert rules. In this paper, we present an LLM-based framework for practical smart contract vulnerability detection. We construct and release a large-scale dataset comprising 31,165 professionally annotated vulnerability instances collected from over 3,200 real-world projects across 15 major blockchain platforms. Our approach leverages precise AST-based context extraction and vulnerability-specific prompt design to instantiate customized detectors for 13 prevalent vulnerability categories. Experimental results demonstrate strong effectiveness, achieving an average positive recall of 0.92 and an average negative recall of 0.85, highlighting the potential of carefully engineered contextual prompting for scalable and high-precision smart contract security analysis.

[AI-23] Graph Neural Network based Hierarchy-Aware Embeddings of Knowledge Graphs: Applications to Yeast Phenotype Prediction

【速读】：该论文旨在解决知识图谱（Knowledge Graph, KG）中嵌入表示未能充分捕捉领域语义层次结构的问题，从而影响其在生物医学等复杂场景下的预测性能。解决方案的关键在于利用图神经网络（Graph Neural Networks, GNNs）结合来自本体（ontology）的语义损失函数，构建层次感知的嵌入表示；特别是通过引入盒式嵌入（box embeddings），使低维向量空间能够同时编码实体和关系的层级信息，并显著提升对酵母基因敲除实验结果的预测准确性（R² = 0.377），且模型具备良好的泛化能力，可推广至三重基因敲除情形，进一步验证了本体结构在定量建模中的指导价值。

链接: https://arxiv.org/abs/2605.03690
作者: Filip Kronström,Alexander H. Gower,Daniel Brunnsåker,Ievgeniia A. Tiukova,Ross D. King
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Quantitative Methods (q-bio.QM)
备注:

点击查看摘要

Abstract:We present a method for finding hierarchy-aware embeddings of knowledge graphs (KGs) using graph neural networks (GNNs) enriched with a semantic loss derived from underlying ontologies. This method yields embeddings that better reflect domain knowledge. To demonstrate their utility, we predict and interpret the effects of gene deletions in the yeast Saccharomyces cerevisiae and learn box embeddings for KGs in the absence of a prediction task. We further show how box embeddings can serve as the basis for evaluating KG revisions. Our yeast KG is constructed from community databases and ontology terms. Low-dimensional box embeddings combined with GNNs are used to predict cell growth for double gene knockouts. Over 10-fold cross validation, these predictions have a mean R^2 ~score~of~0.360, significantly higher than baseline comparisons, demonstrating that high-level qualitative knowledge is informative about experimental outcomes. Incorporating semantic loss terms in the training of the models improves their predictive performance ( R^2 =0.377) by aligning embeddings with ontology structure. This shows that class hierarchies from ontologies can be exploited for quantitative prediction. We also test the trained models on triple gene knockouts, showing they generalise to data beyond those seen in training. Additionally, by identifying co-occurring relations in the yeast KG important for the cell-growth predictions, we construct hypotheses about interacting traits in yeast. A biological experiment validates one such finding, revealing an association between inositol utilisation and osmotic stress resistance, highlighting the model’s potential to guide biological discovery. Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Quantitative Methods (q-bio.QM) ACMclasses: I.2; I.2.4 Cite as: arXiv:2605.03690 [cs.LG] (or arXiv:2605.03690v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2605.03690 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[AI-24] MEMTIER: Tiered Memory Architecture and Retrieval Bottleneck Analysis for Long-Running Autonomous AI Agents

【速读】：该论文旨在解决长期运行的自主AI代理（autonomous AI agents）中存在的记忆一致性（memory coherence）问题，即在72小时操作窗口内，由于现有扁平文件记忆系统中四种累积失败模式导致工具执行成功率下降14个百分点。解决方案的关键在于提出MEMTIER架构，其核心包括：一个结构化的事件记忆JSONL存储、五信号加权检索引擎、基于注意力机制的认知权重更新循环、异步合并守护进程将事件事实提升至语义层级，以及基于PPO的策略框架用于自适应调整检索权重。这一设计通过分层记忆组织与动态权重优化显著提升了长程任务中的信息召回与推理能力，在LongMemEval-S基准上实现准确率38.2%、F1值41.2%，相较全上下文基线提升33个百分点。

链接: https://arxiv.org/abs/2605.03675
作者: Bronislav Sidik,Lior Rokach
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注: 11 pages, 1 figure, 5 tables. Under review

点击查看摘要

Abstract:Long-running autonomous AI agents suffer from a well-documented memory coherence problem: tool-execution success rates degrade 14 percentage points over 72-hour operation windows due to four compounding failure modes in existing flat-file memory systems. We present MEMTIER, a tripartite memory architecture for the OpenClaw agent runtime that introduces a structured episodic JSONL store, a five-signal weighted retrieval engine, an attention-attributed cognitive weight update loop, an asynchronous consolidation daemon promoting episodic facts to a semantic tier, and a PPO-based policy framework for adapting retrieval weights (infrastructure validated; performance gains pending camera-ready). On the full 500-question LongMemEval-S benchmark (Wu et al., 2025), MEMTIER achieves Acc=0.382, F1=0.412 with Qwen2.5-7B on a consumer 6GB GPU - a +33 percentage point improvement over the full-context baseline (0.050 - 0.382, i.e., 5% - 38%). With DeepSeek-V4-Flash fact pre-population, single-session recall reaches 0.686-0.714, exceeding the paper’s RAG BM25 GPT-4o baseline (0.560) on those categories. Temporal reasoning rises to 0.323 and multi-session synthesis to 0.173, demonstrating that structured semantic pre-population qualitatively changes what lightweight retrieval can achieve. All phases run locally on a consumer laptop with a 6GB GPU.

[AI-25] FUS3DMaps: Scalable and Accurate Open-Vocabulary Semantic Mapping by 3D Fusion of Voxel- and Instance-Level Layers

【速读】：该论文旨在解决开放词汇语义映射（open-vocabulary semantic mapping）中现有方法在可扩展性和精度之间的权衡问题，尤其是训练-free 方法在处理大规模场景时难以兼顾密集层（dense layer）与实例级（instance-level）语义信息的融合效率与准确性。其解决方案的关键在于提出 FUS3DMaps，一种在线双层语义映射方法，通过共享体素地图（shared voxel map）联合维护密集层和实例级层，并引入体素级语义跨层融合机制，从而实现两层语义表示的优势互补：一方面提升整体语义质量，另一方面限制密集层与跨层融合的计算范围于空间滑动窗口内，显著增强可扩展性与实例级地图的精度。

链接: https://arxiv.org/abs/2605.03669
作者: Timon Homberger,Finn Lukas Busch,Jesús Gerardo Ortega Peimbert,Quantao Yang,Olov Andersson
机构: 未知
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: This work has been submitted to the IEEE for possible publication

点击查看摘要

Abstract:Open-vocabulary semantic mapping enables robots to spatially ground previously unseen concepts without requiring predefined class sets. Current training-free methods commonly rely on multi-view fusion of semantic embeddings into a 3D map, either at the instance-level via segmenting views and encoding image crops of segments, or by projecting image patch embeddings directly into a dense semantic map. The latter approach sidesteps segmentation and 2D-to-3D instance association by operating on full uncropped image frames, but existing methods remain limited in scalability. We present FUS3DMaps, an online dual-layer semantic mapping method that jointly maintains both dense and instance-level open-vocabulary layers within a shared voxel map. This design enables further voxel-level semantic fusion of the layer embeddings, combining the complementary strengths of both semantic mapping approaches. We find that our proposed semantic cross-layer fusion approach improves the quality of both the instance-level and dense layers, while also enabling a scalable and highly accurate instance-level map where the dense layer and cross-layer fusion are restricted to a spatial sliding window. Experiments on established 3D semantic segmentation benchmarks as well as a selection of large-scale scenes show that FUS3DMaps achieves accurate open-vocabulary semantic mapping at multi-story building scales. Additional material and code will be made available: this https URL.

[AI-26] ELAS: Efficient Pre-Training of Low-Rank Large Language Models via 2:4 Activation Sparsity

【速读】：该论文旨在解决大语言模型（Large Language Models, LLMs）在预训练过程中因激活矩阵内存占用高和计算效率低而导致的训练瓶颈问题。现有低秩训练方法虽能减少权重内存消耗，但通常保留全秩激活矩阵，导致激活内存成为主要瓶颈，尤其在大批次训练时更为显著；同时，直接对权重施加稀疏性常引发性能下降。解决方案的关键在于提出ELAS框架——通过在低秩模型的前馈网络中引入平方ReLU激活函数，并对平方ReLU后的激活张量实施2:4结构化稀疏（2:4 structured sparsity），从而在保持模型性能几乎不变的前提下，有效降低激活内存开销并提升训练与推理速度，尤其适用于大规模批量训练场景。

链接: https://arxiv.org/abs/2605.03667
作者: Jiaxi Li,Lu Yin,Li Shen,Jinjin Xu,Yuhui Liu,Wenwu Wang,Shiwei Liu,Xilu Wang
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Large Language Models (LLMs) have achieved remarkable capabilities, but their immense computational demands during training remain a critical bottleneck for widespread adoption. Low-rank training has received attention in recent years due to its ability to significantly reduce training memory usage. Meanwhile, applying 2:4 structured sparsity to weights and activations to leverage NVIDIA GPU support for 2:4 structured sparse format has become a promising direction. However, existing low-rank methods often leave activation matrices in full-rank, which dominates memory consumption and limits throughput during large-batch training. Furthermore, directly applying sparsity to weights often leads to non-negligible performance degradation. To achieve efficient pre-training of LLMs, this paper proposes ELAS: Efficient pre-training of Low-rank LLMs via 2:4 Activation Sparsity, a novel framework for low-rank models via 2:4 activation sparsity. ELAS applies squared ReLU activation functions to the feed-forward networks in low-rank models and implements 2:4 structured sparsity on the activations after the squared ReLU operation. We evaluated ELAS through pre-training experiments on LLaMA models ranging from 60M to 1B parameters. The results demonstrate that ELAS maintains performance with minimal degradation after applying 2:4 activation sparsity, while achieving training and inference acceleration. Moreover, ELAS reduces activation memory overhead, particularly with large batch sizes. Code is available at ELAS Repo.

[AI-27] Stage Light is Sequence2: Multi-Light Control via Imitation Learning

【速读】：该论文旨在解决音乐驱动的自动舞台灯光控制（Music-inspired Automatic Stage Lighting Control, ASLC）中存在的三大问题：规则方法可解释性差、音乐到颜色空间映射仅限于单光源控制，以及音乐到控制参数框架迁移能力弱。其解决方案的关键在于提出SeqLight框架，该框架采用分层深度学习结构，首先通过定制SkipBART模型实现每帧全光色分布预测（即从音乐生成多光源HSV色彩空间），再结合混合模仿学习（Hybrid Imitation Learning, IL）策略设计出全局色彩分布到各独立光源的有效分解机制。特别地，该光照分解模块可在不同场地配置下仅用混合光数据训练，无需专业演示，从而实现跨场景灵活适配；同时将光照分解任务建模为目标条件马尔可夫决策过程（Goal-Conditioned Markov Decision Process, GCMDP），并基于 hindsight experience replay (HER) 构建专家示范集，最终通过三阶段IL训练流程实现强泛化能力。

链接: https://arxiv.org/abs/2605.03660
作者: Zijian Zhao,Dian Jin,Zijing Zhou,Xiaoyu Zhang
机构: 未知
类目: Multimedia (cs.MM); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Music-inspired Automatic Stage Lighting Control (ASLC) has gained increasing attention in recent years due to the substantial time and financial costs associated with hiring and training professional lighting engineers. However, existing methods suffer from several notable limitations: the low interpretability of rule-based approaches, the restriction to single-primary-light control in music-to-color-space methods, and the limited transferability of music-to-controlling-parameter frameworks. To address these gaps, we propose SeqLight, a hierarchical deep learning framework that maps music to multi-light Hue-Saturation-Value (HSV) space. Our approach first customizes SkipBART, an end-to-end single primary light generation model, to predict the full light color distribution for each frame, followed by hybrid Imitation Learning (IL) techniques to derive an effective decomposition strategy that distributes the global color distribution among individual lights. Notably, the light decomposition module can be trained under varying venue-specific lighting configurations using only mixed light data and no professional demonstrations, thereby flexibly adapting across diverse venues. In this stage, we formulate the light decomposition task as a Goal-Conditioned Markov Decision Process (GCMDP), construct an expert demonstration set inspired by Hindsight Experience Replay (HER), and introduce a three-phase IL training pipeline, achieving strong generalization capability. To validate our IL solution for the proposed GCMDP, we conduct a series of quantitative analysis and human study. The code and trained models are provided at this https URL .

[AI-28] Agent -Based Modeling of Low-Emission Fertilizer Adoption for Dairy Farm Decarbonisation using Empirical Farm Data

【速读】：该论文旨在解决奶业系统中复杂动态行为的建模难题，特别是如何在考虑农场异质性、社会互动和累积环境影响的基础上，量化低排放肥料采纳与氮管理对温室气体减排的长期效应。其解决方案的关键在于构建一个基于代理的建模（agent-based modeling, ABM）框架，该框架整合了实证数据驱动的社会网络结构以模拟农户间的同伴影响和讨论组动态，并将采纳概率建模为社会传染、农场特征及政策干预（如补贴和碳税）的函数。通过蒙特卡洛模拟和敏感性分析量化不确定性，模型成功再现了实际采纳轨迹（R² = 0.979，RMSE = 0.0274），并验证了其对扩散模式的刻画能力，从而为气候减缓策略提供了一个可计算的“虚拟政策实验室”。

链接: https://arxiv.org/abs/2605.03648
作者: Surya Jayakumar,Kieran Sullivan,John McLaughlin,Christine OMeara,Indrakshi Dey
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注: 29 pages, 12 figures

点击查看摘要

Abstract:To understand complex system dynamics in dairy farming, it is essential to use modeling tools that capture farm heterogeneity, social interactions, and cumulative environmental impacts. This study proposes an agent-based modeling (ABM) framework to simulate nitrogen management and the adoption of low-emission fertilizer across 295 Irish dairy farms over a 15-year period. Using empirical data, the model represents farm communication through a social network, capturing peer influence and discussion group dynamics, where adoption probabilities are driven by social contagion, farm-scale characteristics, and policy interventions such as subsidies and carbon taxes. The framework estimates sectoral greenhouse gas emissions, cumulative abatement, and private-social cost trade-offs, using Monte Carlo simulation and sensitivity analysis to quantify uncertainty. The model shows strong agreement with observed adoption trajectories ( R^2 = 0.979 , RMSE = 0.0274) and is validated against empirical data using a Kolmogorov-Smirnov test (D = 0.2407, p 0.001), indicating its ability to reproduce structural patterns in adoption behavior. Adoption dynamics are further characterized using a logistic diffusion model consistent with Rogers’ innovation diffusion theory, capturing progression from early adoption to a saturation level of approximately 91%. By framing decarbonization as a socio-technical diffusion process rather than a purely economic optimization problem, this study provides an in silico policy laboratory for evaluating the robustness and diffusion speed of climate mitigation strategies prior to implementation.

[AI-29] AdapShot: Adaptive Many-Shot In-Context Learning with Semantic-Aware KV Cache Reuse

【速读】：该论文旨在解决多示例上下文学习（Many-Shot In-Context Learning, ICL）中因固定shot数量导致的适应性不足与计算效率低下问题。具体而言，静态shot策略难以应对不同查询难度的差异，易造成信息不足或噪声干扰，且长上下文带来的高昂计算和内存开销限制了其实际应用。解决方案的关键在于提出AdapShot框架：首先设计基于输出熵的探针评估机制以动态确定最优shot数；其次引入语义感知的KV缓存复用策略，通过解耦与重编码方法处理位置编码不兼容问题，从而避免重复预填充计算，显著提升推理效率。实验表明，AdapShot在性能上平均提升约10%，速度提升达4.64倍。

链接: https://arxiv.org/abs/2605.03644
作者: Jie Ou,Jinyu Guo,Shiyao Guo,Yuang Li,Ruiqi Wu,Zhaokun Wang,Wenyi Li,Wenhong Tian
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Many-Shot In-Context Learning (ICL) has emerged as a promising paradigm, leveraging extensive examples to unlock the reasoning potential of Large Language Models (LLMs). However, existing methods typically rely on a predetermined, fixed number of shots. This static approach often fails to adapt to the varying difficulty of different queries, leading to either insufficient context or interference from noise. Furthermore, the prohibitive computational and memory costs of long contexts severely limit Many-Shot’s feasibility. To address the above limitations, we propose AdapShot, which dynamically optimizes shot counts and leverages KV cache reuse for efficient inference. Specifically, we design a probe-based evaluation mechanism that utilizes output entropy to determine the optimal number of shots. To bypass the redundant prefilling computation during both the probing and inference phases, we incorporate a semantics-aware KV cache reuse strategy. Within this reuse strategy, to address positional encoding incompatibilities, we introduce a decoupling and re-encoding method that enables the flexible reordering of cached key-value pairs. Extensive experiments demonstrate that AdapShot achieves an average performance gain of around 10% and a 4.64x speedup compared to state-of-the-art DBSA.

[AI-30] Self-Improvement for Fast High-Quality Plan Generation ICAPS2026

【速读】：该论文旨在解决生成高质量计划（high-quality plans）这一计算上困难的问题，同时在亚指数时间内实现高效求解。传统符号规划器（symbolic planner）在保证计划质量的同时往往面临计算复杂度高的挑战，而现有基于生成式模型的方法多聚焦于找到任意可行解（satisficing solution），忽视了计划质量的优化。解决方案的关键在于：首先，利用最优数据训练的仅解码器结构（decoder-only）Transformer 模型能够在未见过的问题实例上生成高质量计划；其次，提出一种自提升（self-improvement）机制，通过多轮迭代，结合模型调用与图搜索（graph search）生成改进后的计划用于模型微调，从而逐步提升模型性能。实验表明，在 Blocksworld、Logistics、Labyrinth 和 Sokoban 四个领域中，该方法平均比源符号规划器减少 30% 的计划长度，且超过 80% 的计划达到已知最优解，同时推理延迟呈亚指数增长，显著优于对比的符号规划器。

链接: https://arxiv.org/abs/2605.03625
作者: Robert Gieselmann,Henrike von Huelsen,Mihai Samson,Marie-Christine Meyer,Dariusz Piotrowski,Oleksandr Radomskyi,Justin Okamoto,Turan Gojayev,Michael Painter,Gavin Brown,Federico Pecora,Jeremy L. Wyatt
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注: Accepted at ICAPS 2026

点击查看摘要

Abstract:Generative models trained on synthetic plan data are a promising approach to generalized planning. Recent work has focused on finding any valid plan, rather than a high-quality solution. We address the challenge of producing high-quality plans, a computationally hard problem, in sub-exponential time. First, we demonstrate that, given optimal data, a decoder-only transformer can generate high-quality plans for unseen problem instances. Second, we show how to self-improve an initial model trained on sub-optimal data. Each round of self-improvement combines multiple model calls with graph search to generate improved plans, used for model fine-tuning. An experimental study on four domains: Blocksworld, Logistics, Labyrinth, and Sokoban, shows on average a 30% reduction in plan length over the source symbolic planner, with over 80% of plans being optimal, where the optimum is known. Plan quality is further improved by inference-time search. The model’s latency scales sub-exponentially in contrast to the satisficing and optimal symbolic planners to which we compare. Together, these results suggest that self-improvement with generative models offers a scalable approach for high-quality plan generation.

[AI-31] Where Paths Split: Localized Calibrated Control of Moral Reasoning in Large Language Models

【速读】：该论文旨在解决大语言模型在不同情境下表现出异质道德偏好（heterogeneous moral preferences）的问题，目标是在不损害模型通用能力的前提下，实现推理时对特定伦理框架的引导（inference-time steering）。其解决方案的关键在于提出收敛-发散路由机制（Convergent-Divergent Routing），通过识别Transformer块中伦理框架相关路径首次汇聚后又分叉的最小分支点（branch points），在这些节点处 gating 非目标分支以阻断下游传播，同时保留上游计算不变；进一步结合改进的共空间模式（Common Spatial Patterns）提取每个分支层中区分功利主义与义务论框架的二维方向，并引入双逻辑校准（Dual Logit Calibration）——一种闭式、最小ℓ₂范数更新策略，将残差流限制在该子空间内并调整其投影方向以匹配用户指定的偏好权重，从而实现可解释且精准的伦理偏好校准。

链接: https://arxiv.org/abs/2605.03609
作者: Chenchen Yuan,Zheyu Zhang,Gjergji Kasneci
机构: 未知
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:

点击查看摘要

Abstract:Large language models often display heterogeneous moral preferences across settings. We study inference-time steering toward a desired ethical framework while preserving general competence. We present Convergent-Divergent Routing, which traces and edits minimal branch points inside transformer blocks where ethical-framework-related pathways first converge and then diverge. Gating non-target branches at these loci blocks the downstream propagation while leaving upstream computations intact. We find that this intervention alone increases targeted ethical-framework reasoning. To achieve fine-grained control, we adapt Common Spatial Patterns to the residual stream and extract, for each branch-point layer, a pair of directions that discriminate between utilitarian and deontological frameworks. We then introduce Dual Logit Calibration, a closed-form, minimum- \ell_2 -norm update that moves the residual within this two-dimensional subspace so the resulting directional projections align with user-specified preference weights. Experiments on real-life moral dilemmas show that our method reliably achieves preference calibration and largely preserves general capabilities, outperforming recent baselines while providing an interpretable mechanism.

[AI-32] Multi-Agent Strategic Games with LLM s

【速读】：该论文旨在探讨大规模语言模型（Large Language Models, LLMs）是否可用于研究冲突与合作的战略基础，特别是检验其能否再现国际关系理论中的经典机制。其核心问题是：LLMs能否作为实验主体，在重复的安全困境博弈中模拟人类行为并揭示战略逻辑？解决方案的关键在于将LLMs引入扩展的重复安全困境实验框架，通过引入多极化（multipolarity）、有限时间 horizon 和沟通机制三个理论关键维度，系统性地评估LLMs的行为模式。结果显示，LLMs在这些情境下表现出与传统理论一致的规律性行为：多极化加剧冲突，有限时间导致普遍性的“解体”（unraveling），而沟通则通过信号传递和互惠机制降低冲突。此外，该设计还能获取代理的私有推理过程和公开信息，从而将决策与预判、不确定性下的合作及信任构建等战略逻辑直接关联，为理论机制的可验证性提供了一种可扩展、透明且可复现的方法论路径。

链接: https://arxiv.org/abs/2605.03604
作者: Maxim Chupilkin
机构: 未知
类目: Computer Science and Game Theory (cs.GT); Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注: 29 pages, 6 figures

点击查看摘要

Abstract:This paper asks whether large language models (LLMs) can be used to study the strategic foundations of conflict and cooperation. I introduce LLMs as experimental subjects in a repeated security dilemma and evaluate whether they reproduce canonical mechanisms from international relations theory. The baseline game is extended along three theoretically central dimensions: multipolarity, finite time horizons, and the availability of communication. Across multiple models, the results exhibit systematic and consistent patterns: multipolarity increases the likelihood of conflict, finite horizons induce universal unraveling consistent with backward-induction logic, and communication reduces conflict by enabling signaling and reciprocity. Beyond observed behavior, the design provides access to agents’ private reasoning and public messages, allowing choices to be linked to underlying strategic logics such as preemption, cooperation under uncertainty, and trust-building. The contribution is primarily methodological. LLM-based experiments offer a scalable, transparent, and replicable approach to probing theoretical mechanisms.

[AI-33] Unifying Dynamical Systems and Graph Theory to Mechanistically Understand Computation in Neural Networks

【速读】：该论文旨在解决生物神经网络与人工神经网络中结构（连接性）如何决定功能（计算实现）这一核心问题，尤其关注递归神经网络（RNN）在处理分层模块化任务时，其时空功能如何从图结构中被解析。解决方案的关键在于提出一种基于多跳路径（multi-hop pathways）的分析框架，将网络视为图结构并分析输入到输出单元间的多跳通信路径，从而揭示信息的时间路由机制；进一步地，提出resolvent-RNN（R-RNN），通过约束多跳路径而非仅单跳权重（如L1正则化所做），诱导出与任务结构匹配的时序稀疏性，显著提升模型性能和鲁棒性，表明稀疏性应定义于功能路径而非参数本身。

链接: https://arxiv.org/abs/2605.03598
作者: Jatin Sharma,Danyal Akarca,Dan F.M Goodman
机构: 未知
类目: Neural and Evolutionary Computing (cs.NE); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Understanding how biological and artificial neural networks implement computation from connectivity is a central problem in neuroscience and machine learning. In neural systems, structural and functional connectivity are known to diverge, motivating approaches that move beyond direct connections alone. Here, we show that the spatial and temporal function of recurrent neural networks (RNNs) trained on hierarchically modular tasks can be recovered by modelling the network as a graph and analysing the multi-hop pathways between input and output units. In particular, decomposing these pathways by hop length reveals how the network temporally routes information. This perspective reframes regularisation: if function is implemented through multi-hop communication, then standard penalties such as L1 regularisation, which act only on individual weights, constrain single-hop structure rather than the multi-hop pathways that support computation. Motivated by this view, we introduce resolvent-RNNs (R-RNNs), which constrain multi-hop pathways and thereby induce temporal sparsity beyond that achieved by standard L1 regularisation. Compared with L1 regularisation, R-RNNs achieve improved performance by inducing temporal sparsity that matches the task structure, even when the task signal is sparse. Moreover, R-RNNs exhibit stronger sparsity-function alignment, reflected in their increased robustness under strong regularisation. Together, our results identify multi-hop communication as a key principle linking structure to function in recurrent networks, and suggest that sparsity should be defined over functional pathways rather than individual parameters.

[AI-34] Flow Matching on Symmetric Spaces

【速读】：该论文旨在解决在黎曼对称空间（Riemannian symmetric spaces）上训练流匹配（flow matching）模型的难题，这类空间包括球面、双曲空间及格拉斯曼流形（Grassmannians）等复杂几何结构。其核心挑战在于如何高效处理这些非欧几里得空间中的测地线（geodesics）计算与模型训练。解决方案的关键在于利用对称空间的代数结构，将原问题转化为在其等距群（isometry group）的李代数（Lie algebra）的一个子空间上的流匹配问题，从而实现线性化建模，显著简化测地线的计算与优化过程，为在复杂流形上构建生成式模型提供了通用且高效的框架。

链接: https://arxiv.org/abs/2605.03588
作者: Francesco Ruscelli,Ferdinando Zanchetta,Rita Fioresi
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 11 pages, 2 figures

点击查看摘要

Abstract:We introduce a general framework for training flow matching models on Riemannian symmetric spaces, a large class of manifolds that includes the sphere, hyperbolic space and Grassmannians. We exploit their algebraic structure to reformulate flow matching on symmetric spaces as flow matching on a subspace of the Lie algebra of their isometry group, thus linearizing the problem and greatly simplifying the handling of geodesics. As an application, we showcase our framework on the real Grassmannians \operatornameSO(n) / \operatornameSO(k) \times \operatornameSO(n-k) .

[AI-35] Disentangling Shared and Task-Specific Representations from Multi-Modal Clinical Data

【速读】：该论文旨在解决多任务学习在临床多模态数据中因共享表示与任务特异性建模难以平衡而导致的性能瓶颈问题，尤其是硬参数共享可能引发负迁移（negative transfer），而灵活共享又易造成共享信号与任务特异性信号纠缠。其解决方案的关键在于提出一种基于统一Transformer架构的多任务框架，并引入正交任务分解（Orthogonal Task Decomposition, OrthTD），通过将患者表征分解为正交的共享子空间与任务特异性子空间，并施加几何正交性约束，从而减少冗余、隔离任务特异性信号，实现更高效的多结局预测。

链接: https://arxiv.org/abs/2605.03570
作者: He Lyu,Huolin Zeng,Junren Wang,Huazhen Yang,Linchao He,Yong Chen,Zhirui Li,Andreas Maier,Siming Bayer,Huan Song
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted for publication in EMBC 2026. This is the authors’ accepted manuscript. DOI to be added after IEEE Xplore publication

点击查看摘要

Abstract:Real-world clinical data is inherently multimodal, providing complementary evidence that mirrors the practical necessity of jointly assessing multiple related outcomes. Although multi-task learning can improve efficiency by sharing information across outcomes, existing approaches often fail to balance shared representation learning with outcome-specific modeling. Hard parameter sharing can trigger negative transfer when task gradients conflict, while flexible sharing may still entangle shared and task-specific signals. To address this, we propose a multi-task framework built on a unified Transformer for multimodal fusion, augmented with Orthogonal Task Decomposition (OrthTD) to split patient representations into shared and task-specific subspaces and impose a geometric orthogonality constraint to reduce redundancy and isolate task-specific signals. We evaluated OrthTD on a real-world cohort of 12,430 surgical patients for predicting four outcomes. OrthTD achieved average AUC (area under the receiver operating characteristic curve) of 87.5% and average AUPRC (area under the precision-recall curve) of 37.2%, consistently outperformed advanced tabular and multi-task methods. Notably, OrthTD achieves substantial gains in AUPRC, indicating superior performance in identifying rare events within imbalanced clinical data. These results suggest that enforcing non-redundant shared and task-specific representations can improve multi-outcome prediction from multimodal clinical data.

[AI-36] HeadQ: Model-Visible Distortion and Score-Space Correction for KV-Cache Quantization

【速读】：该论文旨在解决KV缓存量化（KV-cache quantizers）中因存储误差导致的注意力机制性能下降问题。传统方法仅优化存储空间中的重建误差（如键的均方误差，MSE），但忽略了模型实际感知的注意力过程——即键通过logits进行读取、值通过注意力加权读出。作者提出应从“模型可见坐标系”（model-visible coordinates）衡量持久缓存误差：对于键，可见误差为得分误差（score error）模常数偏移，据此提出HeadQ方法，其在校准学习的查询基底上存储低秩残差侧码，并作为加性logit修正；对于值，则利用固定注意力读出得到A²加权的token失真代理指标。关键创新在于将误差建模从原始存储空间转移到模型感知空间，实验表明该方案显著优于仅优化存储MSE的方法，在多个模型上大幅降低困惑度（如2-bit量化下减少84–94%超额困惑度），并验证了其有效性与鲁棒性。

链接: https://arxiv.org/abs/2605.03562
作者: Jorge L. Ruiz Williams
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:KV-cache quantizers usually optimize storage-space reconstruction, even though attention reads keys through logits and values through attention-weighted readout. We argue that persistent cache error should be measured in model-visible coordinates. For keys, the visible object is score error modulo constant shifts; this yields HeadQ, a key-side method that stores a low-rank residual side code in a calibration-learned query basis and applies it as an additive logit correction. For values, fixed-attention readout gives an A^2 -weighted token-distortion surrogate. Across six models, Fisher/score-space error predicts attention KL far better than raw key MSE; same-budget counterexamples, null-space interventions, query-PCA controls, and wrong-sign HeadQ falsify storage-MSE alternatives. Matched Pythia checkpoints localize the main anomaly to a small-model low-entropy route-flip boundary. In K-only WikiText-103 decode experiments with dense values, HeadQ removes roughly 84 – 94% of the excess perplexity on the strongest 2-bit rows; in an auxiliary full-KV 2-bit composition, HeadQ plus an A^2 value policy improves all six models.

[AI-37] PerFlow: Physics-Embedded Rectified Flow for Efficient Reconstruction and Uncertainty Quantification of Spatiotemporal Dynamics ECAI2026 IJCAI

【速读】：该论文旨在解决从稀疏且不规则测量中重建由偏微分方程（PDE）支配的时空场的问题，此类问题因病态性而难以处理。传统确定性代理模型依赖密集场训练，在有限观测下表现不佳且缺乏不确定性量化能力；现有生成式方法虽能更好应对稀疏性和不确定性，但通过采样时梯度引导同时强制数据一致性和PDE约束，导致推理速度慢且不稳定。其解决方案的关键在于提出PerFlow——一种嵌入物理信息的修正流（rectified Flow），通过解耦观测条件与物理约束施加机制：观测信息直接输入修正流动力学实现无指导条件化，而硬性物理约束（如不可压缩性或守恒律）则通过保持约束不变的投影嵌入。理论上建立了轨迹在采样过程中始终位于物理一致性流形上的不变性保证，实验表明该方法在保持良好物理一致性的同时，实现了高精度重建和高效条件采样（如仅需50步），相比2000步引导扩散基线提速达320倍。

链接: https://arxiv.org/abs/2605.03548
作者: Hao Zhou,Rui Zhang,Han Wan,Hao Sun
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 17 pages, 8 figures. Accepted to IJCAI-ECAI 2026

点击查看摘要

Abstract:Reconstructing PDE-governed fields from sparse and irregular measurements is challenging due to their ill-posed nature. Deterministic surrogates are trained on dense fields that struggle with limited measurements and uncertainty quantification. Generative models, by learning distributions over spatiotemporal fields, can better handle sparsity and uncertainty. However, existing generative approaches enforce data consistency and PDE constraints simultaneously via sampling-time gradient guidance, resulting in slow and unstable inference. To this end, we propose PerFlow, a Physics-embedded rectified Flow for efficient sparse reconstruction and uncertainty quantification of spatiotemporal dynamics. PerFlow decouples observation conditioning from physics enforcement, performing guidance-free conditioning by feeding observations into rectified-flow dynamics while embedding hard physics via a constraint-preserving projection (e.g., incompressibility or conservation). Theoretically, we establish invariance guarantees to ensure that trajectories remain on the physics-consistent manifold throughout sampling. Experiments on various PDE systems demonstrate competitive reconstruction accuracy with sound physics consistency, while enabling efficient conditional sampling (e.g., 50 steps) and up to 320 faster inference than 2000-step guided diffusion baselines.

[AI-38] ProgramBench: Can Language Models Rebuild Programs From Scratch?

【速读】：该论文旨在解决当前语言模型（Language Models, LMs）在软件工程任务中缺乏对完整代码库进行端到端架构设计与实现能力的问题。现有基准测试主要聚焦于单一功能修复或局部特性开发，无法评估模型在无监督环境下构建具有正确行为的、结构合理的软件系统的能力。解决方案的关键在于提出ProgramBench——一个包含200个从命令行工具到广泛使用的开源项目（如FFmpeg、SQLite和PHP解释器）在内的综合性评测基准，通过代理驱动的模糊测试生成端到端行为测试，从而无需预设实现结构即可评估模型的全局软件架构能力。实验表明，当前9种主流LM均未能完全解决任何任务，最佳模型仅在3%的任务中通过95%的测试用例，且倾向于采用单文件的非模块化实现方式，反映出当前模型在复杂软件工程任务中的显著局限性。

链接: https://arxiv.org/abs/2605.03546
作者: John Yang,Kilian Lieret,Jeffrey Ma,Parth Thakkar,Dmitrii Pedchenko,Sten Sootla,Emily McMilin,Pengcheng Yin,Rui Hou,Gabriel Synnaeve,Diyi Yang,Ofir Press
机构: 未知
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Turning ideas into full software projects from scratch has become a popular use case for language models. Agents are being deployed to seed, maintain, and grow codebases over extended periods with minimal human oversight. Such settings require models to make high-level software architecture decisions. However, existing benchmarks measure focused, limited tasks such as fixing a single bug or developing a single, specified feature. We therefore introduce ProgramBench to measure the ability of software engineering agents to develop software holisitically. In ProgramBench, given only a program and its documentation, agents must architect and implement a codebase that matches the reference executable’s behavior. End-to-end behavioral tests are generated via agent-driven fuzzing, enabling evaluation without prescribing implementation structure. Our 200 tasks range from compact CLI tools to widely used software such as FFmpeg, SQLite, and the PHP interpreter. We evaluate 9 LMs and find that none fully resolve any task, with the best model passing 95% of tests on only 3% of tasks. Models favor monolithic, single-file implementations that diverge sharply from human-written code.

[AI-39] A Skill-Based AI Agent ic Pipeline for Library of Congress Subject Indexing

【速读】：该论文旨在解决图书馆编目中主题标引（subject indexing）这一耗时且复杂环节的自动化问题，特别是如何利用人工智能技术实现对《国会图书馆主题词表》（Library of Congress Subject Headings, LCSH）的精准自动映射。解决方案的关键在于构建一个模块化的人工智能代理技能流水线（modular AI agentic skill pipeline），将整个标引过程分解为四个依次执行的子任务：概念分析（conceptual analysis）、定量过滤（quantitative filtering）、权威验证（authority validation）和MARC字段合成（MARC field synthesis）。每个技能均基于《国会图书馆主题词表手册》（SHM）和主题分析理论编码领域知识，从而确保生成结果在概念层面与专业标引实践高度一致，并在具体实施中体现出对最新政策（如2026年LC停止使用形式细分、改用LCGFT 655字段）的良好遵循。

链接: https://arxiv.org/abs/2605.03537
作者: Eric H. C. Chow
机构: 未知
类目: Digital Libraries (cs.DL); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:This paper presents a modular AI agentic skill pipeline for automating subject indexing with Library of Congress Subject Headings (LCSH). Subject indexing - the process of analyzing a work’s aboutness, selecting controlled vocabulary terms, and encoding them as MARC21 subject access fields - is one of the most time-consuming components of library cataloging. The system decomposes this process into four discrete, sequentially executed agent skills: conceptual analysis, quantitative filtering, authority validation, and MARC field synthesis. Each skill encodes domain knowledge drawn directly from Library of Congress Subject Headings Manual (SHM) instruction sheets and subject analysis theory. The pipeline was evaluated against a corpus of ten titles whose existing subject headings were captured from the Harvard Library bibliographic dataset (a snapshot of their Alma ILS). Results demonstrate strong conceptual alignment with professional subject indexing practice, with notable differences in specificity, subdivision practice, and the agent’s adherence to the 2026 LC policy discontinuing form subdivisions in favor of LCGFT 655 fields.

[AI-40] Brainrot: Deskilling and Addiction are Overlooked AI Risks

【速读】：该论文旨在解决当前生成式人工智能（Generative AI）安全与对齐研究中忽视认知与心理健康风险的问题。现有文献主要关注歧视、有害内容、信息危害及恶意使用等传统风险，而未充分探讨因过度依赖GenAI导致的认知卸载（cognitive offloading）、批判性思维能力退化以及成瘾行为等新型心理风险。论文通过量化这一研究空白，提出安全与对齐工作应扩展至涵盖认知和心理健康维度，并建议结合信息传播策略与监管机制来缓解这些新兴风险。其解决方案的关键在于将认知与心理健康的考量纳入AI安全框架，并推动跨学科协同治理。

链接: https://arxiv.org/abs/2605.03512
作者: Ilias Chalkidis,Anders Søgaard
机构: 未知
类目: Computers and Society (cs.CY); Artificial Intelligence (cs.AI)
备注: Accepted to the 2026 ACM Conference on Fairness, Accountability, and Transparency (FAccT '26)

点击查看摘要

Abstract:The scope of AI safety and alignment work in generative artificial intelligence (GenAI) has so far mostly been limited to harms related to: (a) discrimination and hate speech, (b) harmful/inappropriate (violent, sexual, illegal) content, © information hazards, and (d) use cases related to malicious actors, such as cybersecurity, child abuse, and chemical, biological, radiological, and nuclear threats. The public conversation around AI, on the other hand, has also been focusing on threats to our cognition, mental health, and welfare at large, related to over-relying on new technologies, most recently, those related to GenAI. Examples include deskilling associated with cognitive offloading and the atrophy of critical thinking as a result of over-reliance on GenAI systems, and addiction associated with attachment and dependence on GenAI systems. Such risks are rarely addressed, if at all, in the AI safety and alignment literature. In this paper, we highlight and quantify this discrepancy and discuss some initial thoughts on how safety and alignment work could address cognitive and mental health concerns. Finally, we discuss how information campaigns and regulation can be used to mitigate such prominent risks.

[AI-41] Meta-Inverse Physics-Informed Neural Networks for High-Dimensional Ordinary Differential Equations

【速读】：该论文旨在解决高维耦合常微分方程（Ordinary Differential Equations, ODEs）系统中动力学系统的反问题（inverse problems）建模难题，尤其在物理机制部分未知、观测数据稀疏且仅限于特定可测量通道的情况下，如何高效准确地恢复未知参数或模型动态。现有物理信息神经网络（Physics-Informed Neural Networks, PINNs）虽适用于部分可观测场景，但依赖任务特定的联合优化策略，存在优化困难和泛化能力差的问题。论文提出元逆物理信息神经网络（Meta-Inverse Physics-Informed Neural Network, MI-PINN），其关键创新在于将反建模问题重构为两阶段元学习范式：第一阶段跨任务学习物理感知表征，第二阶段固定该表征并仅优化任务特定未知量，从而显著降低参数搜索维度，提升样本效率与推断精度；同时引入基于自适应聚类的多分支学习机制以应对高维ODE系统中的多尺度动态特性，实验证明其在含多达33个耦合ODE的全身体生理药代动力学（Whole-Body Physiologically Based Pharmacokinetic, PBPK）模型中能准确恢复掩蔽的动力学参数并重建缺失的机制项，即使临床观测数据有限。

链接: https://arxiv.org/abs/2605.03511
作者: Zhao Wei,Kenneth Hor Cheng Koh,Sheng Yuan Chin,James Chun Yip Chan,Chin Chun Ooi,Yew-Soon Ong
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Solving inverse problems in dynamical systems governed by high-dimensional coupled ordinary differential equations (ODEs) is a ubiquitous challenge in scientific machine learning. In many real-world applications, researchers seek to uncover unknown parameters or model unknown dynamics even as the underlying physics is only partially characterized, and observations are sparse and limited to specific measurable channels. While physics-informed neural networks (PINNs) are ideal for inverse inference under partial observability, existing PINNs typically rely on task-specific joint optimization, which suffers from optimization difficulties and poor generalization. In this paper, we propose a meta-inverse physics-informed neural network (MI-PINN) that reformulates inverse modeling as a two-stage meta-learning problem. MI-PINN first learns a physics-aware representation across multiple tasks, and then performs inverse modeling by optimizing task-specific unknowns while keeping the learned representation fixed. This two-stage formulation significantly reduces the parameter search dimension, thereby improving sample efficiency and enabling accurate inference. To handle multi-scale dynamics common in these high-dimensional ODE systems, we further introduce an adaptive clustering-based multi-branch learning scheme. We demonstrate the effectiveness of MI-PINN on whole-body physiologically based pharmacokinetic (PBPK) models with up to 33 coupled ODEs, using paracetamol and theophylline under intravenous and oral dosing scenarios. Experimental results show that MI-PINN enables accurate recovery of masked kinetic parameters and reconstruction of missing mechanistic terms despite limited clinical observations.

[AI-42] Real-Time Evaluation of Autonomous Systems under Adversarial Attacks ITSC2026

【速读】：该论文旨在解决当前自动驾驶策略在对抗性条件下评估不足的问题，特别是纯仿真测试无法捕捉真实世界中结构不一致性、监督约束和状态表示效应等关键因素，从而导致政策鲁棒性评估失真。其解决方案的关键在于构建一个基于真实路口驾驶数据的离线轨迹学习与对抗鲁棒性评估框架，通过在受控数据合约下训练并比较三种轨迹学习范式——基于多层感知机（MLP）的行为克隆（Behavior Cloning, BC）、基于Transformer的对象标记行为克隆，以及基于生成对抗模仿学习（Generative Adversarial Imitation Learning, GAIL）框架的逆强化学习（Inverse Reinforcement Learning, IRL），并在推理阶段引入梯度-based对抗扰动（如Projected Gradient Descent, PGD）进行系统性鲁棒性测试，最终揭示状态结构设计与模型架构归纳偏置对对抗稳定性的重要影响。

链接: https://arxiv.org/abs/2605.03491
作者: Adithya Mohan,Xujun Xie,Venkatesh Thirugnana Sambandham,Torsten Schön
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注: Accepted at IEEE ITSC 2026

点击查看摘要

Abstract:Most evaluations of autonomous driving policies under adversarial conditions are conducted in simulation, due to cost efficiency and the absence of physical risk. However, purely virtual testing fails to capture structural inconsistencies, supervision constraints, and state-representation effects that arise in real-world data and fundamentally shape policy robustness. This work presents an offline trajectory-learning and adversarial robustness evaluation framework grounded in real-world intersection driving data. Within a controlled data contract, we train and compare three trajectory-learning paradigms: Multi-Layer Perceptron (MLP)-based Behavior Cloning (BC), Transformer-based object-tokenized BC, and inverse reinforcement learning (IRL) formulated within a Generative Adversarial Imitation Learning (GAIL) framework. Models are evaluated using Average Displacement Error (ADE) and Final Displacement Error (FDE). Inference-time robustness is assessed by subjecting trained policies to gradient-based adversarial perturbations across multiple intersection scenarios, yielding a structured robustness evaluation matrix. Results show that state-structure design and architectural inductive biases critically influence adversarial stability, leading to markedly different robustness profiles despite comparable nominal prediction accuracy (ADE 0.08). Inference-time Projected Gradient Descent (PGD) attacks induce final displacement errors of up to approximately 8 meters. The proposed framework establishes a scalable benchmark for studying offline trajectory learning and adversarial robustness in real-world autonomous driving settings. Comments: Accepted at IEEE ITSC 2026 Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2605.03491 [cs.AI] (or arXiv:2605.03491v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2605.03491 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[AI-43] MEMSAD: Gradient-Coupled Anomaly Detection for Memory Poisoning in Retrieval-Augmented Agents NEURIPS2026

【速读】：该论文旨在解决大语言模型（Large Language Models, LLMs）代理中持久外部记忆（Persistent External Memory）的安全性问题，特别是针对检索增强型代理的内存投毒攻击（Memory Poisoning Attacks）。现有研究尚未对这类攻击的性质进行形式化建模与评估，导致防御机制缺乏理论保障。论文提出了一种统一的Stackelberg博弈框架来刻画三类 escalating access 假设下的攻击行为，并修正了先前评估协议中的不一致性，使攻击成功率（ASR-R）提升至1.00，凸显了现有方法的脆弱性。其核心解决方案是MEMSAD（Semantic Anomaly Detection），一种基于梯度耦合定理的校准防御机制：在编码器正则条件下，异常评分梯度与检索目标梯度被证明完全一致，因此任何降低检测风险的连续扰动必然损害检索排名。这一耦合关系赋予了MEMSAD一个可认证的检测半径，保证无论对手策略如何均能正确分类。理论分析进一步表明，MEMSAD在最小最大意义上最优，且在线滚动校准下具有次线性后悔率，同时揭示了嵌入空间中离散同义词替换漏洞的存在边界——这是当前连续空间防御无法弥补的短板。

链接: https://arxiv.org/abs/2605.03482
作者: Ishrith Gowda(University of California, Berkeley)
机构: 未知
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 28 pages, 9 figures, 6 theorems. Submitted to NeurIPS 2026

点击查看摘要

Abstract:Persistent external memory enables LLM agents to maintain context across sessions, yet its security properties remain formally uncharacterized. We formalize memory poisoning attacks on retrieval-augmented agents as a Stackelberg game with a unified evaluation framework spanning three attack classes with escalating access assumptions. Correcting an evaluation protocol inconsistency in the triggered-query specification of Chen et al. (2024), we show faithful evaluation increases measured attack success by 4\times (ASR-R: 0.25 \to 1.00 ). Our primary contribution is MEMSAD (Semantic Anomaly Detection), a calibration-based defense grounded in a gradient coupling theorem: under encoder regularity, the anomaly score gradient and the retrieval objective gradient are provably identical, so any continuous perturbation that reduces detection risk necessarily degrades retrieval rank. This coupling yields a certified detection radius guaranteeing correct classification regardless of adversary strategy. We prove minimax optimality via Le Cam’s method, showing any threshold detector requires \Omega(1/\rho^2) calibration samples and MEMSAD achieves this up to \log(1/\delta) factors. We further derive online regret bounds for rolling calibration at rate O(\sigma^2/3\Delta^1/3) , and formally characterize a discrete synonym-invariance loophole that marks the boundary of what continuous-space defenses can guarantee. Experiments on a 3 \times 5 attack-defense matrix with bootstrap confidence intervals, Bonferroni-corrected hypothesis tests, and Clopper-Pearson validation ( n=1,000 ) confirm: composite defenses achieve TPR = 1.00 , FPR = 0.00 across all attacks, while synonym substitution evades detection at \Delta ASR-R \approx 0 , exposing a gap existing embedding-based defenses cannot close.

[AI-44] Learning Generalizable Action Representations via Pre-training AEMG

【速读】：该论文旨在解决肌电图（Electromyography, EMG）在跨被试、跨设备和跨任务场景下泛化能力不足的问题，其根源在于数据异质性、标签稀缺以及缺乏统一的表征框架。解决方案的关键在于提出首个大规模自监督表示学习框架Any Electromyography (AEMG)，通过引入新颖的神经肌肉收缩分词器（Neuromuscular Contraction Tokenizer, NCT），将离散的肌肉收缩转化为结构化的“词汇”，并将激活模式映射为连贯的“句子”，从而将EMG信号建模为一种跨设备的生理语言；同时构建了当前最大的跨设备EMG信号词汇库，实现了任意通道拓扑与采样率下的无缝迁移。实验表明，AEMG在零样本留一被试(out)准确率上较六种先进基线提升5.79%-9.25%，且仅需目标用户5%的数据即可实现超过90%的少样本适应性能，为构建单次训练、通用适用的EMG基础模型奠定了基础。

链接: https://arxiv.org/abs/2605.03462
作者: Zhenghao Huang,Huilin Yao,Kaikai Wang,Lin Shu
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 10 pages,3 figures,

点击查看摘要

Abstract:A fundamental role in decoding human motor intent and enabling intuitive human-computer interaction is played by electromyography (EMG). However, its generalization capability across subjects, devices, and tasks remains substantially limited by data heterogeneity, label scarcity, and the lack of a unified representational framework. To bridge this gap, we propose Any Electromyography (AEMG), the first large-scale, self-supervised representation learning framework for EMG. AEMG reconceptualizes neuromuscular dynamics linguistically, utilizing a novel Neuromuscular Contraction Tokenizer (NCT) to translate discrete muscle contractions into structural words and temporal activation patterns into coherent sentences. Furthermore, we compile the largest cross-device EMG signal vocabulary to date, enabling seamless transfer across arbitrary channel topologies and sampling rates. Experiments demonstrate that AEMG improves the zero-shot leave-one-subject-out (LOSO) accuracy by 5.79-9.25% compared to six state-of-the-art baselines, and achieves more than 90% few-shot adaptation performance with only 5% of target user data. Our work has proposed the concept of EMG signals as a cross-device physiological language, learned their grammar from massive amounts of data, and laid the groundwork for a single-training, universally applicable EMG foundation model.

[AI-45] FinSTaR: Towards Financial Reasoning with Time Series Reasoning Models

【速读】：该论文旨在解决时间序列推理模型（Time Series Reasoning Models, TSRMs）在金融领域表现不佳的问题，因其未能适配金融数据特有的不确定性与多主体交互特性。解决方案的关键在于提出一个通用的2×2能力分类框架，将TSRMs的能力细分为单实体/多实体分析与当前状态评估/未来行为预测的组合，并据此构建了面向标普500股票的FinTSR-Bench基准测试集。进一步地，作者设计了FinSTaR模型，针对不同类别采用差异化的思维链（Chain-of-Thought, CoT）策略：对于确定性评估任务引入Compute-in-CoT，通过程序化推理直接从原始价格中计算答案；对于随机性预测任务则采用Scenario-Aware CoT，在生成多种可能情景后再做决策，更贴近金融分析师在不确定性下的推理方式。实验表明，该方法在FinTSR-Bench上平均准确率达78.9%，显著优于主流大语言模型（LLM）和TSRM基线，且四种能力类别通过联合训练呈现互补增强效应。

链接: https://arxiv.org/abs/2605.03460
作者: Seunghan Lee,Jun Seo,Jaehoon Lee,Sungdong Yoo,Minjae Kim,Tae Yoon Lim,Dongwan Kang,Hwanil Choi,Soonyoung Lee,Wonbin Ahn
机构: 未知
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:

点击查看摘要

Abstract:Time series (TS) reasoning models (TSRMs) have shown promising capabilities in general domains, yet they consistently fail on financial domain, which exhibit unique characteristics. We propose a general 2x2 capability taxonomy for TSRMs by crossing 1) single-entity vs. multi-entity analysis with 2) assessment of the current state vs. prediction of future behavior. We instantiate this taxonomy in the financial domain – where the distinction between deterministic assessment and stochastic prediction is particularly critical – as ten financial reasoning tasks, forming the FinTSR-Bench benchmark based on SP stocks. To this end, we propose FinSTaR (Financial Time Series Thinking and Reasoning), trained on FinTSR-Bench with distinct chain-of-thought (CoT) strategies tailored to each category. For assessment, which is deterministic (i.e., computable from observable data), we employ Compute-in-CoT, a programmatic CoT that enables models to derive answers directly from raw prices. For prediction, which is inherently stochastic (i.e., subject to unobservable factors), we adopt Scenario-Aware CoT, which generates diverse scenarios before making a judgment, mirroring how financial analysts reason under uncertainty. The proposed method achieves 78.9% average accuracy on FinTSR-Bench, substantially outperforming LLM and TSRM baselines. Furthermore, we show that the four capability categories are complementary and mutually reinforcing through joint training, and that Scenario-Aware CoT consistently improves prediction accuracy over standard CoT. Code is publicly available at: this https URL.

[AI-46] DynaTab: Dynamic Feature Ordering as Neural Rewiring for High-Dimensional Tabular Data AAAI2026

【速读】：该论文旨在解决高维表格数据（high-dimensional tabular data）缺乏自然特征顺序的问题，从而限制了对排列敏感的深度学习模型的应用。其解决方案的关键在于提出一种名为DynaTab的动态特征排序架构，该架构通过引入一个轻量级判据来预测特征重排是否能提升性能，并基于神经可塑性（neural rewiring）机制动态调整特征顺序；同时结合学习到的位置嵌入、基于重要性的门控机制与掩码注意力层，构建了一个紧凑且顺序感知的特征融合模块，兼容任何序列敏感的主干网络。该方法在36个真实世界表格数据集上对比45种先进基线模型，显著提升了高维场景下的建模效果。

链接: https://arxiv.org/abs/2605.03430
作者: Al Zadid Sultan Bin Habib,Gianfranco Doretto,Donald A. Adjeroh
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: This paper has been accepted for archival publication in the PMLR proceedings of the AAAI 2026 Neuro for AI \ AI for Neuro: Towards Multi-Modal Natural Intelligence (NeuroAI) Workshop, Code: this https URL , PyPI: pip install dynatab

点击查看摘要

Abstract:High-dimensional tabular data lacks a natural feature order, limiting the applicability of permutation-sensitive deep learning models. We propose DynaTab, a dynamic feature ordering-enabled architecture inspired by neural rewiring. We introduce a lightweight criterion that predicts when feature permutation will benefit a dataset by quantifying its intrinsic complexity. DynaTab dynamically reorders features via a neural rewiring algorithm and processes them through a compact, dynamic order-aware combination of separate learned positional embedding, importance-based gating, and masked attention layers, compatible with any sequence-sensitive backbone. Trained end-to-end with bespoke dynamic feature ordering (DFO) and dispersion losses, DynaTab achieves statistically significant gains, particularly on high-dimensional datasets, where it is benchmarked against 45 state-of-the-art baselines across 36 different real-world tabular datasets. Our results position DynaTab as a compelling new paradigm for high-dimensional tabular deep learning.

[AI-47] Replacing Parameters with Preferences: Federated Alignment of Heterogeneous Vision-Language Models

【速读】：该论文旨在解决在隐私敏感领域（如医疗和金融）中，由于数据共享限制导致的视觉-语言模型（Vision-Language Models, VLMs）集中式训练不可行的问题。现有联邦学习方法在面对客户端计算资源、应用需求及模型架构高度异构时，难以有效协同训练。为此，作者提出MoR（Mixture-of-Rewards），其核心在于将参数聚合替换为基于偏好（preference-based）的协作机制：每个客户端本地训练一个奖励模型（reward model），利用本地偏好标注捕捉特定评估信号而不暴露原始数据；在此基础上，引入可学习路由的混合奖励机制（Mixture-of-Rewards），动态融合各客户端的奖励模型以适应输入内容与对齐目标；服务器端则使用GRPO（Generalized Reward Policy Optimization）结合KL惩罚项对基础VLM进行优化，实现无需客户端模型共享架构或参数的偏好对齐。该方案显著提升了跨客户端泛化能力和适应性，为异构VLM在联邦环境下的隐私保护对齐提供了可扩展解决方案。

链接: https://arxiv.org/abs/2605.03426
作者: Shule Lu,Yujing Wang,Hainan Zhang,Xiaoshan Yang,Hongwei Zheng,Yongxin Tong,Changsheng Xu,Zhiming Zheng
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Vision-Language Models (VLMs) have broad potential in privacy-sensitive domains such as healthcare and finance, yet strict data-sharing constraints render centralized training infeasible. Federated Learning mitigates this issue by enabling decentralized training, but practical deployments face challenges due to client heterogeneity in computational resources, application requirements, and model architectures. Under extreme model and data heterogeneity, replacing parameter aggregation with preference-based collaboration offers a more suitable interface, as it eliminates the need for direct parameter or data exchange. Motivated by this, we propose MoR, a federated alignment framework that combines GRPO with Mixture-of-Rewards for heterogeneous VLMs. In MoR, each client locally trains a reward model from local preference annotations, capturing specific evaluation signals without exposing raw data. To combine these heterogeneous supervision signals, MoR introduces a Mixture-of-Rewards mechanism with learned routing, which adaptively fuses client reward models according to the input and alignment objective. The server then optimizes a base VLM using GRPO with a KL penalty to a reference model, enabling preference alignment without requiring client models to share architectures or parameters. Experiments on diverse public vision-language benchmarks demonstrate that MoR consistently outperforms federated alignment baselines in generalization and cross-client adaptability. Our approach provides a scalable solution for privacy-preserving alignment of heterogeneous VLMs under federated settings.

[AI-48] Adaptive Dual-Path Framework for Covert Semantic Communication

【速读】：该论文旨在解决传统隐蔽通信（Covert Communication）方法在信息嵌入方式上的局限性，即通过功率域信号叠加实现隐秘传输易被检测的问题。其核心挑战在于如何在不显著影响主任务性能的前提下，实现语义层面的隐蔽信息传递。解决方案的关键在于提出了一种自适应双路径框架（Adaptive Dual-Path Framework），该框架引入两个编码路径：显式路径（Explicit path）用于公共任务执行，稳态路径（Stego path）则通过对比表示对齐（Contrastive Representation Alignment）联合编码公开与隐蔽信息；同时采用基于Gumbel-Softmax的自适应块选择机制，动态激活网络模块以满足不同任务需求，从而在多目标优化下实现语义理解准确性与隐蔽传输可靠性的协同提升。

链接: https://arxiv.org/abs/2605.03423
作者: Xi Yu,Weicai Li,Lin Yin,Tiejun Lv
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注: 16 oages, 13 figures, Accepted by IEEE Transactions on Communications

点击查看摘要

Abstract:This paper proposes a novel adaptive dual-path framework for covert semantic communication (SemCom), which integrates covert information transmission with task-oriented semantic coding. Unlike conventional covert communication methods that embed hidden messages through power-domain signal superposition, our framework embeds covert data within task-specific features via semantic-level intrinsic encoding. This new architecture introduces dual encoding paths with adaptive block selection: an Explicit path for public task execution and a Stego path that jointly encodes both public and covert information through contrastive representation alignment. A Gumbel-Softmax enabled adaptive path selection mechanism dynamically activates network blocks based on task require- ments. We formulate a multi-objective optimization framework that simultaneously ensures accurate semantic understanding and reliable covert transmission. We rigorously evaluate our framework’s security against a powerful, independently trained attacker. Experimental results on the Cityscapes dataset demon- strate a state-of-the-art level of covertness: our method suppresses the attacker’s detection accuracy to a near-random guessing level of 56.12%. This robust security is achieved while simultaneously maintaining superior performance on the primary semantic tasks compared to the baselines.

[AI-49] Deepfake Audio Detection Using Self-supervised Fusion Representations

【速读】：该论文旨在解决环境感知的语音与声音深度伪造检测问题（Environment-Aware Speech and Sound Deepfake Detection），即在输入音频中，语音和环境声音可能被独立篡改的情况下，实现对各成分层面的深度伪造识别。解决方案的关键在于提出一种双分支深度学习框架，分别利用预训练模型XLS-R（用于语音）和BEATs（用于环境声音）提取互补的上下文表征，并引入匹配头（Matching Head）通过统计归一化和表征交互建模差异以估计原始类别；同时采用多头交叉注意力机制促进语音与环境成分间的有效信息交换，结合残差连接与层归一化优化表示质量，最终由AASIST分类器输出语音和环境层面的伪造概率预测结果。

链接: https://arxiv.org/abs/2605.03420
作者: Khalid Zaman,Qixuan Huang,Muhammad Uzair,Masashi Unoki
机构: 未知
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:This paper describes a submission to the Environment-Aware Speech and Sound Deepfake Detection Challenge (ESDD2) 2026, which addresses component-level deepfake detection using the CompSpoofV2 dataset, where speech and environmental sounds may be independently manipulated. To address this challenge, a dual-branch deepfake detection framework is proposed to jointly model speech and environmental contextual representations from input audio. Two pretrained models, XLS-R for speech and BEATs for environmental sound, are used to extract complementary contextual representations. A Matching Head is introduced to model representation differences through statistical normalization and representation interaction, enabling estimation of the original class. In parallel, multi-head cross-attention enables effective information exchange between speech and environmental components. The refined representations are processed with residual connections and layer normalization, and passed to an AASIST classifier to predict speech-based and environment-based spoofing probabilities. The model outputs original, speech, and environment predictions. On the test set, the proposed system achieves an F1-score of 70.20% and an environmental EER of 16.54%, outperforming the baseline system.

[AI-50] Learning to Theorize the World from Observation

【速读】：该论文试图解决当前世界模型（World Models）中对“理解”定义过于依赖未来预测准确性的局限问题，即如何实现更接近人类认知机制的理解——即通过构建可解释的内部理论来理解世界。其解决方案的关键在于提出“学习建理论”（Learning-to-Theorize）这一新范式，核心是设计一种名为Neural Theorizer (NEO) 的概率神经模型，该模型从原始非文本观测中推断出显式的解释性理论，将理论表示为可执行、组合式的程序（Language of Thought），并通过共享的转移模型进行执行。这种结构使得学习到的理论能够被系统性重组以解释新现象，从而实现基于解释的泛化能力。

链接: https://arxiv.org/abs/2605.03413
作者: Doojin Baek,Gyubin Lee,Junyeob Baek,Hosung Lee,Sungjin Ahn
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:What does it mean to understand the world? Contemporary world models often operationalize understanding as accurate future prediction in latent or observation space. Developmental cognitive science, however, suggests a different view: human understanding emerges through the construction of internal theories of how the world works, even before mature language is acquired. Inspired by this theory-building view of cognition, we introduce Learning-to-Theorize, a learning paradigm for inferring explicit explanatory theories of the world from raw, non-textual observations. We instantiate this paradigm with the Neural Theorizer (NEO), a probabilistic neural model that induces latent programs as a learned Language of Thought and executes them through a shared transition model. In NEO, a theory is represented as an executable, compositional program whose learned primitives can be systematically recombined to explain novel phenomena. Experiments show that this formulation enables explanation-driven generalization, allowing observations to be understood in terms of the programs that generate them.

[AI-51] Smart Passive Acoustic Monitoring: Embedding a Classifier on AudioMoth Microcontroller

【速读】：该论文旨在解决被动声学监测（Passive Acoustic Monitoring, PAM）中因电源和存储资源有限而导致的采集周期短、数据冗余严重的问题。解决方案的关键在于开发一种轻量化且高效的1D卷积神经网络（1D Convolutional Neural Network, 1D-CNN）模型，将其嵌入AudioMoth微控制器实现本地实时分类，从而仅在检测到目标物种——斯科普利剪水鹱（Scopoli Shearwater）叫声时才触发记录，显著降低功耗与存储需求。该模型经优化后内存占用约10kB，推理时间仅20ms，并通过开源教程提供可复用的模型压缩与部署策略，提升了PAM系统的智能化水平与可扩展性。

链接: https://arxiv.org/abs/2605.03412
作者: Louis Lerbourg,Paul Peyret,Juliette Linossier,Marielle Malfante
机构: 未知
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注: 3 pages, 1 table, 2 figures. Video associated

点击查看摘要

Abstract:Passive Acoustic Monitoring (PAM) is an efficient and non-invasive method for surveying ecosystems at a reduced cost. Typically, autonomous recorders allow the acquisition of vast bioacoustic datasets which are then analyzed. However, power consumption and data storage are both scarce and limit the duration of acquisition campaigns. To address this issue, we propose a smart PAM system which allows the in-situ analysis of the soundscape by embedding a classifier directly onto an AudioMoth microcontroller. Specifically, we propose an optimized yet simple 1D Convolutional Neural Network (1D-CNN) to classify the raw audio. The model focuses on the specific call of Scopoli Shearwater seabirds (endangered species) and is trained on a real-world dataset with a classification accuracy of 91% (balanced accuracy of 89%). We also propose a process to optimize the model to fit the severe resource constraints of the AudioMoth, achieving a ~10kB RAM memory footprint and 20ms inference time. Finally, we present an open-source tutorial of our model optimization and export strategy which can be used for embedding models beyond the scope of our study. Our modified version of the AudioMoth firmware adds two functions: (F1) which selectively records data when the target species has been detected and (F2) which logs the continuous classification results in real time. This work intends to facilitate the conception of intelligent sensors, enhancing the efficiency and scalability of bioacoustic monitoring campaigns.

[AI-52] Geometry over Density: Few-Shot Cross-Domain OOD Detection

【速读】：该论文旨在解决少样本跨域异常检测（few-shot cross-domain out-of-distribution, OOD）问题，即如何在不进行额外训练或微调的情况下，仅用少量目标域内样本（in-distribution, ID）即可实现对任意新任务中OOD样本的有效识别。其核心挑战在于构建一个通用、高效且无需任务特定适配的OOD检测框架。解决方案的关键在于提出UFCOD（Unified Framework for Cross-Domain OOD Detection），通过扩散轨迹的信息几何分析提取两类能量特征：路径能量（Path Energy，积分得分幅度）和动力学能量（Dynamics Energy，得分平滑性），二者共同构成一个离散Sobolev范数，刻画样本与预训练扩散模型之间的交互特性。该方法实现了“一次训练、随处部署”的范式，仅需约100个ID样本即可完成新任务的OOD检测，在12个跨域基准上平均AUROC达93.7%，相较传统方法提升约500倍样本效率。

链接: https://arxiv.org/abs/2605.03410
作者: Shawn Li,You Qin,Jiate Li,Charith Peris,Lisa Bauer,Roger Zimmermann,Yue Zhao
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Out-of-distribution (OOD) detection identifies test samples that fall outside a model’s training distribution, a capability critical for safe deployment in high-stakes applications. Standard OOD detectors are trained on a specific in-distribution (ID) dataset and detect deviations from that single domain. In contrast, we study few-shot cross-domain OOD detection: given a \emphsingle pre-trained model, can we perform OOD detection on \empharbitrary new ID-OOD task pairs using only a handful of ID samples at inference time, with no additional training? We propose \textbfUFCOD, a unified framework that achieves this goal through information-geometric analysis of diffusion trajectories. Our key insight is that diffusion noise predictions are score functions (gradients of log-density), and we extract two energy features: \emphPath Energy (integrated score magnitude) and \emphDynamics Energy (score smoothness), that form a discrete Sobolev norm capturing how samples interact with the learned diffusion process. The central contribution is a \textbftrain-once, deploy-anywhere paradigm: a diffusion model trained on a single dataset (e.g., CelebA) serves as a universal feature extractor for OOD detection across semantically unrelated domains (e.g., CIFAR-10, SVHN, Textures). At deployment, each new task requires only \sim 100 unlabeled ID samples for inference: no retraining, no fine-tuning, no task-specific adaptation. Using 100 ID samples per task, UFCOD achieves 93.7% average AUROC across 12 cross-domain benchmarks, competitive with methods trained on 50k–163k samples, demonstrating \sim 500 \times improvement in sample efficiency. See our code in this https URL.

[AI-53] Robust Agent Compensation (RAC): Teaching AI Agents to Compensate

【速读】：该论文旨在解决大语言模型（Large Language Model, LLM）驱动的智能体在复杂任务执行过程中因意外状态偏移或错误行为导致的不可靠性问题，即如何保障智能体在出现异常时仍能恢复到安全状态，从而避免 unintended side effects（非预期副作用）。其解决方案的关键在于提出一种基于日志的恢复范式——鲁棒代理补偿（Robust Agent Compensation, RAC），通过架构扩展实现对大多数智能体框架（如 LangChain 和 LangGraph）的无侵入式集成，利用现有扩展点即可部署，无需修改用户原有代码。RAC 通过记录和分析执行日志，在检测到异常时自动触发补偿机制，显著提升执行可靠性，并在延迟和 token 使用效率上相比当前最先进的 LLM-based recovery 方法提升 1.5–8 倍以上。

链接: https://arxiv.org/abs/2605.03409
作者: Srinath Perera,Kaviru Hapuarachchi,Frank Leymann,Rania Khalaf
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注: Accepted at ACM Conference on AI and Agentic Systems (ACM CAIS 2026)

点击查看摘要

Abstract:We present Robust Agent Compensation (RAC), a log-based recovery paradigm (providing a safety net) implemented through an architectural extension that can be applied to most Agent frameworks to support reliable executions (avoiding unintended side effects). Users can choose to enable RAC without changing their current agent code (e.g., LangGraph agents). The proposed approach can be implemented in most existing agent frameworks via their existing extension points. We present an implementation based on LangChain, demonstrate its viability through the \tau -bench and REALM-Bench, and show that when solving complex problems, RAC is 1.5-8X or more better in both latency and token economy compared to state-of-the-art LLM-based recovery approaches.

[AI-54] Discovering Reinforcement Learning Interfaces with Large Language Models

【速读】：该论文旨在解决强化学习（Reinforcement Learning, RL）系统中任务接口（task interface）自动构建的问题，即如何从原始模拟器状态中自动合成观测映射（observation mapping）和奖励函数（reward function），从而减少对人工设计的依赖。传统方法通常假设观测是固定的，仅优化奖励函数，但这种方法在复杂任务中表现受限。论文提出 LIMEN 框架，其关键在于利用大语言模型（Large Language Models, LLMs）引导的进化策略，将候选接口表示为可执行程序，并通过策略训练反馈迭代优化观测与奖励的联合结构。实验表明，在离散网格世界和连续控制（包括运动和操作）任务中，仅优化单一组件（观测或奖励）会导致失败，而联合演化能有效发现高性能接口，验证了观测与奖励协同设计的重要性。

链接: https://arxiv.org/abs/2605.03408
作者: Akshat Singh Jaswal,Ashish Baghel,Paras Chopra
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Reinforcement learning systems rely on environment interfaces that specify observations and reward functions, yet constructing these interfaces for new tasks often requires substantial manual effort. While recent work has automated reward design using large language models (LLMs), these approaches assume fixed observations and do not address the broader challenge of synthesizing complete task interfaces. We study RL task interface discovery from raw simulator state, where both observation mappings and reward functions must be generated. We propose LIMEN (Code available at this https URL), a LLM guided evolutionary framework that produces candidate interfaces as executable programs and iteratively refines them using policy training feedback. Across novel discrete gridworld tasks and continuous control domains spanning locomotion and manipulation, joint evolution of observations and rewards discovers effective interfaces given only a trajectory-level success metric, while optimizing either component alone fails on at least one domain. These results demonstrate that automatic construction of RL interfaces from raw state can substantially reduce manual engineering and that observation and reward components often benefit from co-design, as single-component optimization fails catastrophically on at least one domain in our evaluation suite.

[AI-55] APEX: Large-scale Multi-task Aesthetic-Informed Popularity Prediction for AI-Generated Music

【速读】：该论文旨在解决AI生成音乐（AI-generated music）的流行度预测问题，尤其关注在缺乏传统艺术家声誉或唱片公司背书的情况下，如何准确预测这类音乐的受欢迎程度。其核心挑战在于现有方法未充分考虑美学质量（aesthetic quality）这一关键因素，而美学质量与用户参与度（如播放量和点赞数）共同构成了音乐吸引力的互补维度。解决方案的关键是提出APEX框架——一个基于超过211,000首AI生成歌曲（约10,000小时音频）的大规模多任务学习模型，该模型联合预测流媒体播放量和点赞分数，并同时估计由MERT（一种自监督音乐理解模型）提取的五维感知美学特征。实验表明，在Music Arena数据集上的分布外评估中，引入美学特征显著提升了人类偏好预测性能，验证了所学表示对不同生成架构具有强泛化能力。

链接: https://arxiv.org/abs/2605.03395
作者: Jaavid Aktar Husain,Dorien Herremans
机构: 未知
类目: ound (cs.SD); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Multimedia (cs.MM)
备注:

点击查看摘要

Abstract:Music popularity prediction has attracted growing research interest, with relevance to artists, platforms, and recommendation systems. However, the explosive rise of AI-generated music platforms has created an entirely new and largely unexplored landscape, where a surge of songs is produced and consumed daily without the traditional markers of artist reputation or label backing. Key, yet unexplored in this pursuit is aesthetic quality. We propose APEX, the first large-scale multi-task learning framework for AI-generated music, trained on over 211k songs (10k hours of audio) from Suno and Udio, that jointly predicts engagement-based popularity signals - streams and likes scores - alongside five perceptual aesthetic quality dimensions from frozen audio embeddings extracted from MERT, a self-supervised music understanding model. Aesthetic quality and popularity capture complementary aspects of music that together prove valuable: in an out-of-distribution evaluation on the Music Arena dataset, comprising pairwise human preference battles across eleven generative music systems unseen during training, including aesthetic features consistently improves preference prediction, demonstrating strong generalisation of the learned representations across generative architectures.

[AI-56] A Fast Model Counting Algorithm for Two-Variable Logic with Counting and Modulo Counting Quantifiers

【速读】：该论文旨在解决加权一阶模型计数（Weighted First-Order Model Counting, WFOMC）在两变量带计数量词逻辑片段 $\mathbf{C}^2$ 中的计算效率问题，特别是现有算法依赖多阶段归约将计数量词转化为基数约束所引入的显著实际开销。解决方案的关键在于提出 IncrementalWFOMC3 算法，该算法直接在 Scott 标准形上操作，保留计数量词以避免冗余转换，从而在理论上获得更优的数据复杂度界（多项式次数由二次降至线性），并首次证明了模计数扩展 $\mathbf{C}^2_\text{mod}$ 的域可提升性（domain-liftable），同时实验证明其在运行时间与可扩展性上显著优于现有 WFOMC 算法和最先进的命题模型计数器。

链接: https://arxiv.org/abs/2605.03391
作者: Shixin Sun,Astrid Klipfel,Ondřej Kuželka,Yuanhong Wang,Yi Chang
机构: 未知
类目: Logic in Computer Science (cs.LO); Artificial Intelligence (cs.AI)
备注: 38 pages, submitted to IJAR, under review

点击查看摘要

Abstract:Weighted first-order model counting (WFOMC) is a central task in lifted probabilistic inference: It asks for the weighted sum of all models of a first-order sentence over a finite domain. A long line of work has identified domain-liftable fragments of first-order logic, that is, syntactic classes for which WFOMC can be solved in time polynomial in the domain size. Among them, the two-variable fragment with counting quantifiers, \mathbfC^2 , is one of the most expressive known liftable fragments. Existing algorithms for \mathbfC^2 , however, establish tractability through multi-stage reductions that eliminate counting quantifiers via cardinality constraints, which introduces substantial practical overhead as the domain size grows. In this paper, we introduce IncrementalWFOMC3, a lifted algorithm for WFOMC on \mathbfC^2 and its modulo counting extension, \mathbfC^2_\textmod . Instead of relying on reduction techniques, IncrementalWFOMC3 operates directly on a Scott normal form that retains counting quantifiers throughout inference. This direct treatment yields two main results. First, we derive a tighter data-complexity bound for WFOMC in \mathbfC^2 , reducing the degree of the polynomial from quadratic to linear in the counting parameters. Second, we prove that \mathbfC^2_\textmod is domain-liftable, extending tractability from \mathbfC^2 to a richer fragment with native modulo counting support. Finally, our empirical evaluation shows that IncrementalWFOMC3 delivers orders-of-magnitude runtime improvements and better scalability than both existing WFOMC algorithms and state-of-the-art propositional model counters.

[AI-57] Local Truncation Error-Guided Neural ODEs for Large Scale Traffic Forecasting

【速读】：该论文旨在解决物理系统中时空预测任务中存在的“连续性-冲击矛盾”问题，即如何在保持宏观连续演化的同时有效捕捉微观突发异常。传统神经微分方程（Neural ODE）因Lipschitz连续性约束，在面对突变事件时会产生过度平滑现象；而现有物理信息方法通过惩罚数值积分误差来强制流形光滑性，却会引发梯度冲突与“注意力坍缩”，削弱模型对异常的敏感性。解决方案的关键在于提出局部截断误差引导的神经微分方程（LTE-ODE），其创新性地将局部截断误差（Local Truncation Error, LTE）作为无监督前向归纳偏置，通过将LTE映射为动态空间注意力掩码，使模型在稳定区域维持高精度连续演化，同时仅在冲击点自适应触发离散补偿分支，从而实现对非线性波动的鲁棒建模。

链接: https://arxiv.org/abs/2605.03386
作者: Xiao Zhang,Yafei Li,Ruixiang Wang,Wei Wei,Shuo He,Mingliang Xu
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Spatiotemporal forecasting in physical systems, such as large-scale traffic networks, requires modeling a dual dynamic: continuous macroscopic rhythms and discrete, unpredictable microscopic shocks. While Neural Ordinary Differential Equations (ODEs) excel at capturing smooth evolution, their inherent Lipschitz continuity constraints inevitably cause severe over-smoothing when confronting abrupt anomalies. Recent physics-informed methods attempt to bypass this by penalizing numerical integration errors to enforce manifold smoothness. However, we mathematically reveal that such rigid regularization inherently triggers gradient conflicts and ``attention collapse,‘’ stripping the model of its sensitivity to anomalies. To resolve this continuity-shock dilemma, we propose Local Truncation Error-Guided Neural ODEs (LTE-ODE). Rather than treating numerical error as a nuisance to be eliminated, we innovatively repurpose the Local Truncation Error (LTE) as an unsupervised forward inductive bias. By mapping the LTE into a dynamic spatial attention mask, our architecture gracefully preserves high-precision continuous ODE evolution in stable regions, while adaptively triggering a discrete compensation branch exclusively at shock points. Trained purely end-to-end without manifold penalties, LTE-ODE achieves state-of-the-art performance on multiple large-scale benchmarks, exhibiting exceptional robustness against highly non-linear fluctuations. Furthermore, our ablation on integration steps demonstrates high deployment flexibility, allowing the model to seamlessly adapt to varying hardware memory constraints in real-world applications.

[AI-58] GeoDecider: A Coarse-to-Fine Agent ic Workflow for Explainable Lithology Classification

【速读】：该论文旨在解决传统岩石类型分类方法将任务视为单次分类过程，忽略了地质专家在实际工作中结合地质原理、外部知识及工具使用能力进行多阶段推理的问题。解决方案的关键在于提出一种“粗到精”的代理式工作流GeoDecider，其通过训练-free方式利用大语言模型（Large Language Models, LLMs）实现准确且可解释的岩石类型分类：首先由预训练分类器引导粗粒度分类以降低下游推理成本；随后借助上下文分析和邻近样本检索等工具增强细粒度推理精度；最后通过地质一致性后处理提升结果的地质合理性。该框架在四个基准数据集上显著优于主流基线，并实现了性能与推理效率之间的更好平衡。

链接: https://arxiv.org/abs/2605.03383
作者: Jiahao Wang,Mingyue Cheng,Yitong Zhou,Qingyang Mao,Xiaoyu Tao,Qi Liu,Enhong Chen
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Lithology classification aims to infer subsurface rock types from well-logging signals, supporting downstream applications like reservoir characterization. Despite substantial progress, most existing methods still treat lithology classification as a single-pass classification task. In contrast, practical experts incorporate geological principles, external knowledge, and tool-use capabilities to perform accurate classification. In this work, we propose GeoDecider, a coarse-to-fine agentic workflow that enables accurate and explainable lithology classification through training-free use of large language models (LLMs). GeoDecider reformulates lithology classification as an expert-like structured process and organizes it into a multi-stage workflow involving coarse-to-fine reasoning. Specifically, GeoDecider includes the following stages: (1) base classifier-guided coarse classification, which uses a pre-trained classifier to provide a rough reference for downstream tasks, thus reducing the overall cost of downstream reasoning, (2) tool-augmented reasoning, which utilizes several tools such as contextual analysis and neighbor retrieval to achieve finer and more precise classifications, (3) geological refinement, which post-processes the final results to enforce geological consistency. Experiments on four benchmarks show that GeoDecider outperforms representative baselines. Further analysis demonstrates that the proposed framework produces geologically interpretable predictions while achieving a better trade-off between classification performance and inference efficiency.

[AI-59] Reason Audio: A Benchmark for Evaluating Reasoning Beyond Matching in Text-Audio Retrieval

【速读】：该论文旨在解决当前文本-音频检索（Text-Audio Retrieval）基准测试过于侧重语义匹配、忽视实际应用场景中高级推理能力的问题。现有方法在面对否定理解（Negation）、时间顺序判断（Order）、事件重叠识别（Overlap）、时长区分（Duration）及混合推理任务（Mix）等复杂查询时表现不佳，难以满足真实世界智能助手和媒体搜索的需求。解决方案的关键在于提出首个面向推理密集型任务的基准数据集 ReasonAudio，包含1,000个精心设计的查询和10,000个复合音频片段，覆盖五类基础推理任务；并通过系统评估十种前沿模型，揭示了当前主流方法在推理能力迁移上的不足，强调了改进训练范式以保留模型推理能力的重要性。

链接: https://arxiv.org/abs/2605.03361
作者: Honglei Zhang,Yuting Chen,Chenpeng Hu,Siyue Zhang,Yilei Shi
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注: 6 pages, 3 figures, 2 tables

点击查看摘要

Abstract:As multimodal content continues to expand at a rapid pace, audio retrieval has emerged as a key enabling technology for media search, content organization, and intelligent assistants. However, most existing benchmarks concentrate on semantic matching and fail to capture the fact that real-world queries often demand advanced reasoning abilities, including negation understanding, temporal ordering, concurrent event recognition, and duration discrimination. To address this gap, we introduce ReasonAudio, the first reasoning-intensive benchmark for Text-Audio Retrieval, comprising 1,000 queries and 10,000 composite audio clips across five fundamental reasoning tasks: Negation, Order, Overlap, Duration, and Mix. Despite their intuitive nature for humans and straightforward construction, these tasks pose significant challenges to current models. Our evaluation of ten state-of-the-art models reveals the following findings: All models struggle with reasoning-intensive audio retrieval, performing particularly poorly on Negation and Duration while showing relatively better results on Overlap and Order. Moreover, Multimodal Large Language Model-based embedding models fail to inherit the reasoning capabilities of their backbones through contrastive fine-tuning, suggesting that current training paradigms are insufficient to preserve reasoning capacity in retrieval settings

[AI-60] What Happens Inside Agent Memory? Circuit Analysis from Emergence to Diagnosis

【速读】：该论文旨在解决大语言模型（Large Language Model, LLM）代理在跨会话记忆操作中“无声失败”（silent memory failures）的问题，即模型在无法正确提取、保留或检索信息时仍能生成流畅响应，导致故障难以察觉。解决方案的关键在于通过追踪Qwen-3系列模型（0.6B–14B参数）与两种记忆框架（mem0和A-MEM）中的内部特征电路，发现控制路由电路（routing circuitry）与内容处理电路（content circuitry）在不同规模下分别激活，并揭示了写入（Write）与读取（Read）共享一个晚期层枢纽（hub），该枢纽作为上下文接地的基础结构存在于基础模型中，而记忆框架仅引入功能性的接地方向。这一特征空间分离使得无需监督即可实现76.2%准确率的逐阶段故障定位，为代理记忆失效提供了可诊断的机制。

链接: https://arxiv.org/abs/2605.03354
作者: Xutao Mao,Jinman Zhao,Gerald Penn,Cong Wang
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Agent memory failures are silent: an LLM-based agent can produce a fluent response even when it fails to extract, retain, or retrieve the information needed across sessions. The write-manage-read loop describes the external pipeline of these systems but leaves open which internal computations implement each stage. Tracing internal feature circuits across the Qwen-3 family (0.6B–14B) and two memory frameworks (mem0 and A-MEM), we report three findings. First, control is detectable before content: routing circuitry is causally active at 0.6B, while content circuitry produces no detectable signal until 4B under our tracing setup, creating a deployment regime where small models route with apparent competence but silently fail at extraction and grounding. Second, within the content group, Write and Read share a late-layer hub that operates as a context-grounding substrate already present in the base model; only memory framing recruits a functional grounding direction on this substrate, and the hub transfers across both frameworks. Third, emergence does not imply steerability: although the content circuit becomes detectable at 4B, it becomes reliably steerable only at 8B, indicating that detection and intervention have distinct scale thresholds. As a practical implication, the feature-space separation between the two circuit groups enables per-operation failure localization at 76.2% accuracy without supervision, providing a stage-level diagnostic for otherwise silent agent-memory failures.

[AI-61] SkCC: Portable and Secure Skill Compilation for Cross-Framework LLM Agents

【速读】：该论文旨在解决当前LLM-Agents（大型语言模型代理）在跨平台部署时面临的两大核心问题：一是不同代理框架对提示（prompt）格式敏感度差异显著，导致性能波动高达40%，而现有技能（skill）通常以单一、格式无关的Markdown形式存在，需手动重写适配，维护成本极高；二是社区技能中存在超过三分之一含有安全漏洞，缺乏系统性防护机制。解决方案的关键在于提出SkCC编译框架，其核心创新为引入编译器设计思想，构建一种强类型中间表示（SkIR），将技能语义与平台特定格式解耦，实现跨异构代理框架的可移植部署；同时通过编译期分析器（Analyzer）在部署前实施抗技能注入（Anti-Skill Injection）安全约束，从而在降低适配复杂度（从O(m×n)降至O(m+n)）的同时提升安全性与效率，实验证明编译后技能在Pass率、安全触发率和运行时token消耗方面均显著优于原始版本。

链接: https://arxiv.org/abs/2605.03353
作者: Yipeng Ouyang,Yi Xiao,Yuhao Gu,Xianwei Zhang
机构: 未知
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: 9pages, 6figures

点击查看摘要

Abstract:LLM-Agents have evolved into autonomous systems for complex task execution, with the this http URL specification emerging as a de facto standard for encapsulating agent capabilities. However, a critical bottleneck remains: different agent frameworks exhibit starkly different sensitivities to prompt formatting, causing up to 40% performance variation, yet nearly all skills exist as a single, format-agnostic Markdown version. Manual per-platform rewriting creates an unsustainable maintenance burden, while prior audits have found that over one third of community skills contain security vulnerabilities. To address this, we present SkCC, a compilation framework that introduces classical compiler design into agent skill development. At its core, SkIR - a strongly-typed intermediate representation - decouples skill semantics from platform-specific formatting, enabling portable deployment across heterogeneous agent frameworks. Around this IR, a compile-time Analyzer enforces security constraints via Anti-Skill Injection before deployment. Through a four-phase pipeline, SkCC reduces adaptation complexity from O(m \times n) to O(m + n) . Experiments on SkillsBench demonstrate that compiled skills consistently outperform their original counterparts, improving pass rates from 21.1% to 33.3% on Claude Code and from 35.1% to 48.7% on Kimi CLI, while achieving sub-10ms compilation latency, a 94.8% proactive security trigger rate, and 10-46% runtime token savings across platforms.

[AI-62] oward Structural Multimodal Representations: Specialization Selection and Sparsification via Mixture-of-Experts ICML2026

【速读】：该论文旨在解决多模态学习中传统方法将所有信号编码为固定嵌入所导致的表达冗余与任务适配性不足的问题。其解决方案的核心在于提出S3框架（Specialization, Selection, Sparsification）：通过结构化视角重构多模态表示，首先在共享潜在空间中形成语义专家（semantic experts），实现概念级的专业化（Specialization）；其次根据具体任务需求动态选择最优路径（Selection）；最后通过剪枝低效路径实现稀疏化（Sparsification），从而获得紧凑且信息最小化的表示。实验表明，该方法在多个MultiBench基准上提升准确率，并呈现典型的反向U型稀疏性-性能关系，验证了以可选语义组件构建表示的合理性与有效性。

链接: https://arxiv.org/abs/2605.03348
作者: Hahyeon Choi,Nojun Kwak
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Published at ICML 2026

点击查看摘要

Abstract:We propose S3 (Specialization, Selection, Sparsification), a framework that rethinks multimodal learning through a structural perspective. Instead of encoding all signals into a fixed embedding, S3 decomposes multimodal inputs into semantic experts and selectively routes them for each task. Specialization forms concept-level experts in a shared latent space, Selection adapts routing for task-specific needs, and Sparsification prunes low-utility paths to yield compact, information-minimal representations. Across four MultiBench benchmarks, S3 improves accuracy and shows a consistent reverse U-shaped sparsity-performance trend, with peak performance at intermediate sparsity. These results suggest that structuring multimodal representations as selectable semantic components provides a practical and principled alternative to contrastive learning or InfoMax-driven approaches.

[AI-63] Automated Large-scale CVRP Solver Design via LLM -assisted Flexible MCTS

【速读】：该论文旨在解决大规模车辆路径问题（Large-Scale CVRP, LSCVRP）求解困难的问题，尤其是当节点数量达到数百至数千时，现有先进求解器仍难以高效处理。传统分解策略（divide-and-conquer）虽能通过将大规模实例拆分为更小的子问题来提升可扩展性，但其设计分解逻辑与配置子求解器的过程高度依赖专家经验且劳动密集。为此，作者提出了一种基于大语言模型（Large Language Models, LLMs）的新型框架——LLM-assisted Flexible Monte Carlo Tree Search (LaF-MCTS)，其核心创新在于构建了一个三层决策层级结构，以实现对LSCVRP分解策略和子求解器的逐步优化设计；同时引入语义剪枝（semantic pruning）机制剔除语义和结构冗余的算法代码，并通过分支再生（branch regrowth）机制重新生成代码以保持搜索空间多样性，从而在CVRPLib数据集上实现了自主组合并优化出超越多种主流CVRP求解器性能的分解增强型求解方案。

链接: https://arxiv.org/abs/2605.03339
作者: Tong Guo,Caishun Chen,Yew Soon Ong
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Solving large-scale CVRP (LSCVRP) with hundreds to thousands of nodes remains difficult for even state-of-the-art solvers. Divide-and-conquer can scale by decomposing the instance into size-reduced subproblems, but designing decomposition logic and configuring sub-solvers is highly expertise- and labor-intensive. Large Language Models (LLMs) have emerged as promising tools for automated algorithm design. However, existing LLM-driven approaches struggle with LSCVRP primarily due to the difficulty in generating sophisticated search strategies within a limited context window. To bridge this gap, we propose the LLM-assisted Flexible Monte Carlo Tree Search (LaF-MCTS), a novel framework that automates the design of high-performance LSCVRP solvers. We develop a three-tier decision hierarchy to enable incremental design of decomposition policies and sub-solvers for LSCVRP. To enable efficient search within the algorithmic hypothesis space, we introduce semantic pruning to eliminate semantically and structurally redundant codes, and branch regrowth to regenerate codes and preserve diversity. Extensive experiments on CVRPLib demonstrate that LaF-MCTS autonomously composes and optimizes decomposition-enhanced solvers that surpasses various state-of-the-art CVRP solvers.

[AI-64] LLM -ADAM: A Generalizable LLM Agent Framework for Pre-Print Anomaly Detection in Additive Manufacturing

【速读】：该论文旨在解决增材制造（Additive Manufacturing, AM）中因用户缺乏专业制造知识而导致的打印参数设置不当问题，特别是通过G-code文件中存在的热力学或几何学危害性配置引发的打印缺陷。现有方法难以在打印前有效识别这些隐性错误，导致材料与设备资源浪费。解决方案的关键在于提出一种通用的基于大语言模型（Large Language Model, LLM）的框架——LLM-ADAM，其核心创新是将异常检测任务结构化为三个角色：Extractor-LLM负责从G-code提取过程参数并映射到结构化模式；Reference-LLM将打印机和材料文档转化为对齐的操作范围；Judge-LLM结合偏差表与G-code证据判断零件是否属于特定异常类别。该设计强调任务分解而非单纯依赖LLM模型强度，实验证明此策略显著优于单一LLM基线，在N=200的Fused Filament Fabrication (FFF) G-code数据集上达到87.5%准确率，且缺陷分类性能接近理论上限，主要误差集中于对非缺陷样本的保守误报。

链接: https://arxiv.org/abs/2605.03328
作者: Ahmadreza Eslaminia,Chuhan Cai,Cameron Smith,Ruo-Syuan Mei,Shichen Li,Rajiv Malhotra,Klara Nahrstedt,Chenhui Shao
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 24 pages, 10 figures

点击查看摘要

Abstract:Additive manufacturing (AM) continues to transform modern manufacturing by enabling flexible, on-demand production of complex geometries across diverse industries. Fused filament fabrication (FFF) has extended AM to laboratories, classrooms, and small production environments, but this accessibility shifts process-planning responsibility to users who may lack manufacturing expertise. A syntactically valid slicer profile can still encode thermally or geometrically harmful settings, and subtle G-code edits can alter extrusion, cooling, or adhesion before a print begins. Pre-print G-code screening catches accidental or adversarial machine-program errors before material or machine time is wasted. This paper proposes LLM-ADAM as a generalizable LLM framework for pre-print anomaly detection in AM. The framework decomposes the task into three roles: Extractor-LLM maps a G-code file to a structured process-parameter schema; Reference-LLM converts printer and material documentation into aligned operating ranges; and Judge-LLM interprets a deterministic deviation table and G-code evidence to decide whether a part is non-defective or belongs to an anomaly class. Printers, materials, and LLM backbones are interchangeable test conditions, not fixed assumptions. We evaluate the framework on an N=200 FFF G-code corpus spanning two desktop printer families, two materials, and five classes including non-defective, under-extrusion, over-extrusion, warping, and stringing. The best framework configuration reaches 87.5% accuracy, compared with 59.5% for the strongest engineered single-LLM baseline. The results show that structured decomposition, rather than backbone strength alone, is the dominant source of improvement, with defect classes identified at or near ceiling for leading configurations while residual errors concentrate on conservative false alarms for non-defective samples.

[AI-65] DGPO: Distribution Guided Policy Optimization for Fine Grained Credit Assignment

【速读】：该论文旨在解决当前强化学习算法（如Group Relative Policy Optimization）在对齐大语言模型以执行复杂推理任务时存在的两个核心问题：一是序列级别的信用分配粒度粗略，难以精准识别长链式思维（Chain of Thought）生成过程中关键的推理步骤；二是标准的无界KL散度惩罚导致梯度不稳定和模式-seeking保守性，抑制了新型推理路径的探索。解决方案的关键在于提出一种无需评判器（critic-free）的新型强化学习框架——分布引导策略优化（Distribution Guided Policy Optimization），其创新性地将分布偏差重新诠释为引导信号而非刚性惩罚，从而实现更精细的策略更新与更开放的推理空间探索。

链接: https://arxiv.org/abs/2605.03327
作者: Hongbo Jin,Rongpeng Zhu,Zhongjing Du,Xu Jiang,Jingqi Tian,Qiaoman Zhang,Jiayu Ding
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Reinforcement learning is crucial for aligning large language models to perform complex reasoning tasks. However, current algorithms such as Group Relative Policy Optimization suffer from coarse grained, sequence level credit assignment, which severely struggles to isolate pivotal reasoning steps within long Chain of Thought generations. Furthermore, the standard unbounded Kullback Leibler divergence penalty induces severe gradient instability and mode seeking conservatism, ultimately stifling the discovery of novel reasoning trajectories. To overcome these limitations, we introduce Distribution Guided Policy Optimization, a novel critic free reinforcement learning framework that reinterprets distribution deviation as a guiding signal rather than a rigid penalty.

[AI-66] Cryptographic Registry Provenance: Structural Defense Against Dependency Confusion in AI Package Ecosystems

【速读】：该论文旨在解决软件依赖项分发过程中的“依赖混淆攻击”（dependency confusion attack）问题，即由于缺乏对包来源的密码学证明，攻击者可利用注册表结构漏洞伪造合法依赖包。现有防御措施均为配置驱动且在配置错误时无声失效。解决方案的关键在于构建一个三重加密分发溯源系统：（1）注册表身份加密认证，每个注册表持有Ed25519密钥对并签名其分发的所有工件；（2）双签名模型，发布者在打包时签名，注册表在发布时进行二次签名；（3）权威命名空间绑定，消费者固定注册表指纹，解析器拒绝来自未授权注册表的工件。该机制形成三层防御，需同时被攻破才能成功实施攻击，从而实现从源码到运行时的无密码学盲区生命周期链路。

链接: https://arxiv.org/abs/2605.03309
作者: Alan L. McCann
机构: 未知
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注: 15 pages, 1 figure, 4 tables. Companion proofs: this https URL . Project: this https URL

点击查看摘要

Abstract:Dependency confusion attacks exploit a structural gap in software distribution: once a package is installed, there is no cryptographic proof of which registry distributed it. Every existing defense is configuration-based and fails silently when misconfigured. We present a cryptographic distribution provenance system comprising three components: (1) cryptographic registry identity, where every registry holds an Ed25519 keypair and signs every artifact it distributes; (2) a dual-signature model, where the publisher signs at packaging time and the registry countersigns at publication time; and (3) authoritative namespace binding, where consumers pin registry fingerprints and the resolver cryptographically rejects artifacts from unauthorized registries. These create three defense layers requiring simultaneous compromise for a successful attack. A comparison across eight ecosystems (npm, Cargo, this http URL, PyPI, Go modules, Docker/OCI, NuGet, Maven) shows no existing ecosystem combines mandatory publisher signing, cryptographic registry identity, mandatory registry countersigning, and consumer-side cryptographic enforcement. The system extends to AI-generation provenance as a signed attribute and governance-enforced dependency resolution. A case study integrates distribution provenance with a three-layer runtime governance architecture, creating a four-phase lifecycle chain with no cryptographic gaps.

[AI-67] Revisiting the Travel Planning Capabilities of Large Language Models

【速读】：该论文旨在解决大语言模型（Large Language Models, LLMs）在长时推理任务中，尤其是旅行规划场景下的性能瓶颈问题。现有评估方法通常以端到端方式衡量最终计划质量，缺乏可解释性且难以定位失败根源。为此，作者提出将旅行规划任务解耦为五个原子子能力：约束提取（Constraint Extraction）、工具使用（Tool Use）、计划生成（Plan Generation）、错误识别（Error Identification）和错误修正（Error Correction），并设计了一种基于Oracle中间上下文的解耦评估协议，从而在无级联误差干扰的情况下精确测量各组件的性能边界。该方案的关键在于通过结构化分解与隔离评估，揭示LLMs在隐式约束理解、计划生成偏倚及自我纠错失效等方面的系统性缺陷，为提升其推理与规划能力提供精准改进方向。

链接: https://arxiv.org/abs/2605.03308
作者: Bo-Wen Zhang,Jin Ye,Peng-Yu Hua,Jia-Wei Cao,Jie-Jing Shao,Yu-Feng Li,Lan-Zhe Guo
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Travel planning serves as a critical task for long-horizon reasoning, exposing significant deficits in LLMs. However, existing benchmarks and evaluations primarily assess final plans in an end-to-end manner, which lacks interpretability and makes it difficult to analyze the root causes of failures. To bridge this gap, we decompose travel planning into five constituent atomic sub-capabilities, including \emphConstraint Extraction, \emphTool Use, \emphPlan Generation, \emphError Identification, and \emphError Correction. We implement a decoupled evaluation protocol leveraging oracle intermediate contexts to rigorously isolate these components, thereby measuring the atomic performance boundary without the noise of cascading errors. Our results highlight a clear contrast in performance: while LLMs are proficient in extracting explicit constraints, they struggle to infer implicit, open-world requirements. Furthermore, they exhibit structural biases in plan generation and suffer from ineffective self-correction, characterized by excessive sensitivity and erroneous persistence. These findings offer precise directions for improving LLM reasoning and planning abilities.

[AI-68] RLDX-1 Technical Report

【速读】：该论文旨在解决当前视觉-语言-动作模型（Vision-Language-Action models, VLAs）在复杂现实任务中功能局限性的问题，尤其是其在运动感知、记忆驱动决策和物理传感等更广泛功能性能力上的不足。解决方案的关键在于提出RLDX-1，一个基于多流动作变换器（Multi-Stream Action Transformer, MSAT）的通用机器人策略架构，该架构通过模态特异性流与跨模态联合自注意力机制统一多种异构模态信息，并结合系统级设计选择，包括针对罕见操作场景的数据合成、面向类人操作的学习流程以及实时推理优化，从而显著提升模型在高自由度人形机器人上执行接触丰富且动态复杂的灵巧操作任务的能力。

链接: https://arxiv.org/abs/2605.03269
作者: Dongyoung Kim,Huiwon Jang,Myungkyu Koo,Suhyeok Jang,Taeyoung Kim,Beomjun Kim,Byungjun Yoon,Changsung Jang,Daewon Choi,Dongsu Han,Donguk Lee,Heeseung Kwon,Hojin Jeon,Jaehyun Kang,Jaekyoung Bae,Jihyuk Lee,Jimin Lee,John Won,Joonwoo Ahn,Junhyeong Park,Junyoung Sung,Kyungmin Lee,Minseong Han,Minsung Yoon,Sejune Joo,Seonil Son,Seungcheol Park,Seunggeun Cho,Seungjun Moon,Seungku Kim,Yonghoon Dong,Yongjin Cho,Youngchan Kim,Chang Hwan Kim,Dohyeon Kim,Hazel Lee,Heecheol Kim,Hensen Ahn,Hyungkyu Ryu,Hyunsoo Choi,Hyunsoo Shin,Jaeheon Jung,Jaewoo Kim,Jinwook Kim,Joochul Chang,Joonsoo Kim,Junghun Park,Jungwoo Park,Junho Cho,Junhyeok Park,Junwon Lee,Kangwook Lee,Kwanghoon Kim,Kyoungwhan Choe,Manoj Bhadu,Nayoung Oh,Sangjun Kim,Sangwoo Kim,Seunghoon Shim,Seunghyun Kim,Seungjun Lee,Seungyup Ka,Sungryol Yang,Wook Jung,Yashu Shukla,Yeonjae Lee,Yeonwoo Bae,Jinwoo Shin
机构: 未知
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Project page: this https URL

点击查看摘要

Abstract:While Vision-Language-Action models (VLAs) have shown remarkable progress toward human-like generalist robotic policies through the versatile intelligence (i.e. broad scene understanding and language-conditioned generalization) inherited from pre-trained Vision-Language Models, they still struggle with complex real-world tasks requiring broader functional capabilities (e.g. motion awareness, memory-aware decision making, and physical sensing). To address this, we introduce RLDX-1, a general-purpose robotic policy for dexterous manipulation built on the Multi-Stream Action Transformer (MSAT), an architecture that unifies these capabilities by integrating heterogeneous modalities through modality-specific streams with cross-modal joint self-attention. RLDX-1 further combines this architecture with system-level design choices, including synthesizing training data for rare manipulation scenarios, learning procedures specialized for human-like manipulation, and inference optimizations for real-time deployment. Through empirical evaluation, we show that RLDX-1 consistently outperforms recent frontier VLAs (e.g. \pi_0.5 and GR00T N1.6) across both simulation benchmarks and real-world tasks that require broad functional capabilities beyond general versatility. In particular, RLDX-1 shows superiority in ALLEX humanoid tasks by achieving success rates of 86.8% while \pi_0.5 and GR00T N1.6 achieve around 40%, highlighting the ability of RLDX-1 to control a high-DoF humanoid robot under diverse functional demands. Together, these results position RLDX-1 as a promising step toward reliable VLAs for complex, contact-rich, and dynamic real-world dexterous manipulation.

[AI-69] Partially Observed Structural Causal Models

【速读】：该论文旨在解决传统结构因果模型（Structural Causal Models, SCMs）在处理潜变量（latent contexts）影响下，观测变量间的因果结构与机制难以分离的问题。具体而言，当潜变量同时决定图结构和节点机制时，现有SCM框架无法有效区分结构形成过程与功能机制，导致干预推断失效。解决方案的关键在于提出部分可观测结构因果模型（Partially Observed Structural Causal Models, POSCMs），其核心创新是引入一个可识别的干预层级体系，涵盖节点级、边级及上下文级干预，并基于Kolmogorov-Arnold-Sprecher边函数分解定理，将每个节点的机制表示为父节点单变量函数之和，从而显式参数化二元功能贡献，实现对边级干预的“外科手术式”精确控制。这一方法使得在潜变量存在的情况下，仍可通过特定干预协议识别结构与机制的解耦关系，验证了理论上的可识别性结果。

链接: https://arxiv.org/abs/2605.03268
作者: Turan Orujlu,Jordan Matelsky,Martin V. Butz,Charley M. Wu,Konrad P. Kording
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Methodology (stat.ME); Machine Learning (stat.ML)
备注:

点击查看摘要

Abstract:Here we introduce Partially Observed Structural Causal Models (POSCMs) that formalize causal systems where latent contexts co-determine both the interaction structure and downstream mechanisms on observed variables. POSCMs provide an extension of structural causal models (SCMs), as a self-contained causal modeling framework for endogenous graphs, allowing for an intervention hierarchy spanning node- and edge-level context and endogenous variable interventions. To enable surgical edge interventions, we adopt a Kolmogorov-Arnold-Sprecher edge-functional decomposition, an existence theorem for representing each node mechanism as a sum of univariate functions of its parents, yielding an explicit parametrization of dyadic functional contributions. We provide an identifiability theory that clarifies which intervention families would suffice to disentangle structure formation from mechanisms. We empirically validate these predictions in a biophysically detailed virtual human retina simulator, constructing intervention protocols that (i) reproduce the non-identifiability predicted when context is latent and no context-level interventions are available, (ii) exhibit structure-mechanism confounding under latent edges when only node interventions are observed, and (iii) recover synaptic input-output relationships via targeted node interventions, consistent with our positive kernel identifiability result. Our work generalizes SCMs in a way that allows it to work in a world closer to the one we live in.

[AI-70] Posterior-First Neural PDE Simulation: Inferring Hidden Problem State from a Single Field

【速读】：该论文旨在解决单观测场条件下神经偏微分方程（PDE）模拟器在部署时因缺乏不确定性建模而导致的“确定性坍缩”问题，即不同潜在问题状态被映射到相同的确定性输出界面，从而丧失了对下游决策至关重要的模糊性。其解决方案的关键在于提出“后验优先”（posterior-first）的神经PDE模拟框架：首先从观测数据中推断出最小任务充分的问题状态的后验分布，然后基于该后验进行未来状态预测。该方法通过贝叶斯理论建立目标、学习目标与失败模式之间的联系——贝叶斯下游价值通过此后验因子分解，细化标签可通过合适评分规则实现可学习性，而确定性坍缩则会在真实后验非狄拉克分布时引发“模糊性障碍”。实验表明，该策略在合成精确模糊性场景和元数据隐藏的PDEBench任务中均显著提升预测稳定性与准确性。

链接: https://arxiv.org/abs/2605.03247
作者: Wenshuo Wang,Fan Zhang
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Neural PDE simulators often receive only a single observed field at deployment. In this setting, a field-to-future predictor can collapse distinct latent problem states into the same deterministic interface, losing the ambiguity needed for reliable rollout and downstream decisions. We propose posterior-first neural PDE simulation: first infer a posterior over the minimal task-sufficient problem state, then condition prediction on that posterior. The resulting theory connects the object, the learning target, and the failure mode: Bayes downstream values factor through this posterior, refinement labels make it learnable by proper scoring rules, and deterministic collapse incurs an ambiguity barrier whenever the true posterior is non-Dirac. Synthetic exact-ambiguity experiments show that point-versus-posterior gaps track the predicted barrier. On metadata-hidden PDEBench tasks, posterior recovery reduces pooled rollout nRMSE from 0.175 to 0.132, closing 59.4% of the direct-to-oracle gap. These results suggest that single-observation neural PDE simulation should be posterior-first rather than monolithic field-to-future prediction.

[AI-71] Enhancing Agent Safety Judgment: Controlled Benchmark Rewriting and Analogical Reasoning for Deceptive Out-of-Distribution Scenarios

【速读】：该论文旨在解决当前安全评估基准对隐性风险（如情境模糊性、隐含威胁和捷径决策）关注不足的问题，从而高估了大语言模型（Large Language Models, LLMs）在复杂现实场景中判断代理行为安全性的真实能力。其解决方案的关键在于提出 ROME（Red-team Orchestrated Multi-agent Evolution）——一种受红队启发的多智能体演化基准构建管道，通过将已知不安全轨迹改写为更具欺骗性的评估实例，同时保留原始风险标签，生成包含300个挑战性样本的测试集；并进一步引入 ARISE（Analogical Reasoning for Inference-time Safety Enhancement），一种基于检索的推理时增强机制，从外部类比知识库中提取 ReAct 风格的安全轨迹作为结构化推理示例注入推理过程，以提升模型在不重新训练前提下的安全判断质量。两者结合实现了对代理系统安全判断能力的更严格压力测试与针对性优化。

链接: https://arxiv.org/abs/2605.03242
作者: Zuoyu Zhang,Yancheng Zhu
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Tool-using agent systems powered by large language models (LLMs) are increasingly deployed across web, app, operating-system, and transactional environments. Yet existing safety benchmarks still emphasize explicit risks, potentially overstating a model’s ability to judge deceptive or ambiguous trajectories. To address this gap, we introduce ROME (Red-team Orchestrated Multi-agent Evolution), a controlled benchmark-construction pipeline that rewrites known unsafe trajectories into more deceptive evaluation instances while preserving their underlying risk labels. Starting from 100 unsafe source trajectories, ROME produces 300 challenge instances spanning contextual ambiguity, implicit risks, and shortcut decision-making. Experiments show that these challenge sets substantially degrade safety-judgment performance, with hidden-risk cases remaining particularly non-trivial even for recent frontier models. We further study ARISE (Analogical Reasoning for Inference-time Safety Enhancement), a retrieval-guided inference-time enhancement that retrieves ReAct-style analogical safety trajectories from an external analogical base and injects them as structured reasoning exemplars. ARISE improves judgment quality without retraining, but is best viewed as a task-specific robustness enhancement rather than a standalone safety guarantee. Together, ROME and ARISE provide practical tools for stress-testing and improving agent safety judgment under deceptive distribution shifts.

[AI-72] cotomi Act: Learning to Automate Work by Watching You

【速读】：该论文旨在解决浏览器代理（browser agent）如何在不依赖显式指令的情况下，通过观察用户行为自动学习工作任务并持续积累组织性知识的问题。解决方案的关键在于两个核心模块：一是基于自适应懒惰观察（adaptive lazy observation）、基于语言差异的历史压缩（verbal-diff-based history compression）、粗粒度动作（coarse-grained actions）以及测试时通过 best-of-N 选择优化的执行框架，实现了高达 80.4% 的 WebArena 人类评估任务成功率；二是行为到知识的流水线（behavior-to-knowledge pipeline），被动记录用户浏览行为并逐步抽象为可编辑的任务看板和维基文档等结构化知识，形成人机共享的工作空间，从而提升任务完成率。

链接: https://arxiv.org/abs/2605.03231
作者: Masafumi Oyamada,Kunihiro Takeoka,Kosuke Akimoto,Ryoma Obara,Masafumi Enomoto,Haochen Zhang,Daichi Haraguchi,Takuya Tamura
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注: 7 pages, 4 figures. ACM CAIS 2026 (System Demonstrations)

点击查看摘要

Abstract:What if a browser agent could learn your work simply by watching you do it? We present cotomi Act, a browser-based computer-using agent that combines reliable multi-step task execution with persistent organizational knowledge learned from user behavior. For execution, an agent scaffold with adaptive lazy observation, verbal-diff-based history compression, coarse-grained actions, and test-time scaling via best-of-N action selection achieves 80.4% on the 179-task WebArena human-evaluation subset, exceeding the reported 78.2% human baseline. For organizational knowledge, a behavior-to-knowledge pipeline passively observes the user’s browsing and progressively abstracts it into artifacts (task boards, wiki) exposed through a shared workspace editable by both user and agent. A controlled proxy evaluation confirms that task success improves as behavior-derived knowledge accumulates. In our live demonstration, attendees interact with the system in a real browser, issuing tasks and observing end-to-end autonomous execution and shared knowledge management.

[AI-73] Evaluating Prompting and Execution-Based Methods for Deterministic Computation in LLM s

【速读】：该论文旨在解决大语言模型（Large Language Models, LLMs）在执行精确、确定性计算任务时能力不足的问题，特别是其在序列建模任务中难以保证输出的准确性。研究通过系统评估多种提示策略（如思维链 CoT、从少到多分解 Least-to-Most、思维程序 PoT 和自一致性 Self-Consistency）发现，标准提示方法仅能实现中等精度，CoT 改进有限，Least-to-Most 易产生误差累积；而 PoT 通过生成可执行代码并交由外部解释器执行，实现了完美准确率。关键解决方案在于将 LLM 的推理能力与外部工具（如代码解释器）结合，或训练轻量级专用模型（如 CodeT5-small）来生成可执行程序，从而在保持低训练成本的同时显著提升确定性任务的可靠性与效率。

链接: https://arxiv.org/abs/2605.03227
作者: Hongkun Yu
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注: 8 pages, 1 figure. Code and dataset available at this https URL

点击查看摘要

Abstract:Large Language Models (LLMs) have demonstrated strong capabilities in natural language understanding and reasoning. However, their ability to perform exact, deterministic computation remains unclear. In this work, we systematically evaluate multiple prompting strategies, including Chain-of-Thought (CoT), Least-to-Most decomposition, Program-of-Thought (PoT), and Self-Consistency (SC), on tasks requiring precise and error-free outputs, including binary counting, longest substring detection, and arithmetic evaluation. To support this study, we introduce a synthetic dataset with diverse natural language instructions, enabling controlled evaluation of exact computation across multiple task types. Our results show that standard prompting methods achieve only moderate accuracy on sequence-based tasks. CoT provides limited improvement, while Least-to-Most suffers from error accumulation. In contrast, PoT achieves perfect accuracy by generating executable code and delegating computation to an external interpreter. Self-Consistency improves robustness through majority voting, but incurs substantial computational overhead. We further train a small domain-specific model (CodeT5-small) to generate executable programs, which achieves perfect accuracy on held-out synthetic test data across all tasks with minimal training cost. Overall, our findings suggest that LLMs may simulate reasoning patterns rather than reliably perform exact symbolic computation. For deterministic tasks, combining LLMs with external tools or using specialized models provides a more reliable and efficient solution. Comments: 8 pages, 1 figure. Code and dataset available at this https URL Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2605.03227 [cs.AI] (or arXiv:2605.03227v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2605.03227 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: HongKun Yu [view email] [v1] Mon, 4 May 2026 23:32:37 UTC (123 KB) Full-text links: Access Paper: View a PDF of the paper titled Evaluating Prompting and Execution-Based Methods for Deterministic Computation in LLMs, by Hongkun YuView PDFHTML (experimental)TeX Source view license Current browse context: cs.AI prev | next new | recent | 2026-05 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(); About Help contact arXivClick here to contact arXiv Contact subscribe to arXiv mailingsClick here to subscribe Subscribe Copyright Privacy Policy Web Accessibility Assistance arXiv Operational Status

[AI-74] Self-Mined Hardness for Safety Fine-Tuning

【速读】：该论文旨在解决大语言模型（Large Language Models, LLMs）在安全微调（safety fine-tuning）过程中对恶意攻击（如WildJailbreak）防御能力不足的问题。传统方法依赖人工构建的对抗性数据集，而本文提出一种基于模型自生成反馈的新型微调策略：通过评估目标模型自身生成结果中被判定为有害的频率来量化提示词（prompt）难度，并在最难的提示词上使用模型自身的非越狱输出（non-jailbroken rollouts）进行微调。其核心创新在于利用模型自我评估的难易度指标筛选训练样本，从而显著降低攻击成功率（从11.5%–20.1%降至1%–3%），同时通过引入与良性提示交替的混合训练机制，在保持高安全性的同时有效缓解过度拒绝（over-refusal）问题。

链接: https://arxiv.org/abs/2605.03226
作者: Prakhar Gupta,Garv Shah,Donghua Zhang
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
备注:

点击查看摘要

Abstract:Safety fine-tuning of language models typically requires a curated adversarial dataset. We take a different approach: score each candidate prompt’s difficulty by how often the target model’s own rollouts are judged harmful, then fine-tune on the hardest prompts paired with the model’s own non-jailbroken rollouts. On Llama-3-8B-Instruct and Llama-3.2-3B-Instruct, this approach cuts the WildJailbreak attack success rate from 11.5% and 20.1% down to 1-3%, but pushes refusal on jailbreak-shaped benign prompts from 14-22% to 74-94%. Interleaving the same hard prompts 1:1 with adversarially-framed benign prompts (prompts that look like jailbreaks but have benign intent) cuts that refusal back down to 30-51% on 8B and 52-72% on 3B, at a cost of 2-6 percentage points of attack success rate. Within the mixed regime, training on the hardest half of the eligible pool rather than a random half cuts the remaining ASR by 35-50% (about 3 percentage points) on both models.

[AI-75] MenuNet: A Strategy-Proof Mechanism for Matching Markets

【速读】：该论文旨在解决在存在复杂分配约束（如多样性配额、区域平衡和全局容量松弛）的匹配市场中，稳定匹配往往不存在的问题，进而如何在不牺牲策略激励相容性（strategy-proofness）的前提下，公平地分配不可避免的不稳定性的核心挑战。解决方案的关键在于提出 \textttMenuNet，一个基于神经网络表示的菜单机制设计框架：该框架不直接构造分配结果，而是学习生成个性化的概率菜单，并通过结构化的顺序选择规则实现分配，从而天然保证策略激励相容性；同时将稳定性分解为无嫉妒（fairness, no envy）与无浪费（non-wastefulness）两个向量值属性，利用可微目标函数优化其分布，实现对不同公理之间的权衡控制。

链接: https://arxiv.org/abs/2605.03216
作者: Zhaohong Sun,Makoto Yokoo
机构: 未知
类目: Computer Science and Game Theory (cs.GT); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Strategy-proofness is a fundamental desideratum in mechanism design, ensuring truthful reporting and robust participation. Stability is another central requirement in matching markets, widely adopted in applications such as school choice and labor market clearing. In practice, however, these markets are invariably governed by complex distributional constraints, ranging from diversity quotas and regional balance to global capacity slacks, under which stable matchings often fail to exist. This raises a fundamental question: how to distribute unavoidable instability across agents while preserving strategy-proofness? To address this, we propose \textttMenuNet, a strategy-proof mechanism design framework based on a neural representation of menus. Rather than directly constructing assignments, \textttMenuNet learns to generate personalized probabilistic menus, from which assignments are realized via a structured sequential choice rule that guarantees strategy-proofness by construction. By decomposing stability into fairness (no envy) and non-wastefulness, our approach models these properties as vector-valued quantities and optimizes their distribution through differentiable objectives, providing a principled trade-off between competing axioms. Empirically, \textttMenuNet navigates this trade-off effectively: it consistently outperforms Random Serial Dictatorship (RSD) in terms of envy and Deferred Acceptance (DA) in terms of waste, while maintaining scalability and computational efficiency. These results suggest that learning-based menu mechanisms provide a flexible and scalable paradigm for mechanism design in highly constrained, real-world environments.

[AI-76] When Agents Handle Secrets: A Survey of Confidential Computing for Agent ic AI

【速读】：该论文旨在解决生成式 AI (Generative AI) 系统中代理（Agent）架构带来的新型安全威胁问题，这些威胁源于代理在多阶段任务执行过程中积累敏感上下文、持有凭证，并跨多个不可信实体协作，从而面临提示注入、上下文泄露、凭证窃取及代理间消息污染等风险。传统软件层防御机制易被特权攻击者绕过，因此论文提出以可信执行环境（Trusted Execution Environment, TEE）为核心的硬件根信任方案作为关键解决方案，通过隔离代理代码与数据、支持远程认证来构建跨分布式部署的可验证信任链，从而在不依赖单一控制方的前提下提升 agentic AI 的安全性。

链接: https://arxiv.org/abs/2605.03213
作者: Javad Forough,Marios Kogias,Hamed Haddadi
机构: 未知
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Agentic AI systems, specifically LLM-driven agents that plan, invoke tools, maintain persistent memory, and delegate tasks to peer agents via protocols such as MCP and A2A, introduce a threat surface that differs materially from standalone model inference. Agents accumulate sensitive context, hold credentials, and operate across pipelines no single party fully controls, enabling prompt injection, context exfiltration, credential theft, and inter-agent message poisoning. Current defenses operate entirely within the software stack and can be silently bypassed by a sufficiently privileged adversary such as a compromised cloud operator. Confidential computing (CC) offers a hardware-rooted alternative: Trusted Execution Environments (TEEs) isolate agent code and data from privileged system software, while remote attestation enables verifiable trust across distributed deployments. This survey synthesizes the design space in four parts: (i) a unified taxonomy of six TEE platforms (Intel SGX, Intel TDX, AMD SEV-SNP, ARM TrustZone, ARM CCA, and NVIDIA H100 CC) covering deployment roles and performance tradeoffs; (ii) an agent-centric threat model spanning perception, planning, memory, action, and coordination layers mapped to nine security goals; (iii) a comparative survey of CC-based defenses distinguishing findings that transfer from single-call inference versus what requires new agentic designs; and (iv) six open challenges including compound attestation for multi-hop agent chains and GPU-TEE performance at LLM scale. While several hardware trust primitives appear mature enough for targeted deployments, no broadly established end-to-end framework yet binds them into a coherent security substrate for production agentic AI.

[AI-77] Human-Provenance Verification should be Treated as Labor Infrastructure in AI-Saturated Markets

【速读】：该论文试图解决的问题是：在生成式 AI（Generative AI）和代理型 AI 系统广泛渗透市场后，传统中层知识型劳动的价值被压缩，导致劳动力市场出现结构性失衡，进而引发对人类劳动价值重新定义的需求。解决方案的关键在于提出“人类来源验证”（human-provenance verification）作为新的劳动基础设施，强调只有当人类判断、注意力、责任归属、创作权或关系参与构成产出的核心要素时，即所谓“构成性人类存在”（constitutive human presence），人类劳动才能获得可识别且可持续的溢价——这种溢价类似于凡勃伦商品（Veblen good）的特性，但其本质不是奢侈标签，而是基于可信验证的稀缺性价值。

链接: https://arxiv.org/abs/2605.03210
作者: Erin McGurk,David Khachaturov
机构: 未知
类目: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); General Economics (econ.GN)
备注:

点击查看摘要

Abstract:We argue that AI-saturated markets are likely to create Veblen-good premiums, which we term human-provenance premiums, for verified human presence, and hence AI governance should treat human-provenance verification as labor infrastructure. Generative and agentic AI systems lower the cost of many standardized cognitive, creative, and coordination tasks, weakening the scarcity premiums that have supported much middle-tier knowledge work. We argue that this pressure may produce an asymmetric barbell-shaped structure of value capture in advanced economies: high-volume synthetic production controlled by owners of AI infrastructure at one pole, and scarce, high-status human labor valued for verified human presence at the other. We advance three claims. First, AI compresses the value of standardized middle-tier labor by making good-enough synthetic substitutes scalable at low marginal cost, hollowing out the middle of the skill distribution currently categorized by knowledge work. Second, this compression reallocates demand for human labor toward work valued for its visible human character. We term this performative humanity and distinguish three forms of labor: relational presence, aesthetic provenance, and accountability. Third, as these premiums depend on credible verification, AI governance should treat human-provenance systems as labor infrastructure rather than as luxury authenticity labels. To evaluate hybrid human-AI work, we propose constitutive human presence as the relevant standard: human labor retains premium value when human judgment, attention, accountability, authorship, or relational participation is not incidental to the output but constitutive of what is being purchased. Subjects: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); General Economics (econ.GN) Cite as: arXiv:2605.03210 [cs.CY] (or arXiv:2605.03210v1 [cs.CY] for this version) https://doi.org/10.48550/arXiv.2605.03210 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[AI-78] Stop Automating Peer Review Without Rigorous Evaluation ICML2026

【速读】：该论文旨在解决当前学术期刊面临的同行评审危机（peer review crisis），即传统人工评审效率低、主观性强且难以保障质量的问题。其核心观点是：当前的大语言模型（Large Language Models, LLMs）尚不具备替代人类进行可靠同行评审的能力。解决方案的关键在于，必须建立一个专门针对同行评审自动化（peer review automation）的科学体系，而非简单地将通用大语言模型部署于评审流程中。文中通过实证比较发现，AI生成的评审存在“蜂群效应”（hivemind effect）导致视角多样性不足，且评分极易被“论文洗稿”（paper laundering）操纵——即仅通过风格改写即可显著提升AI评分，说明现有LLM评审机制缺乏稳健性与非可操纵性（non-gameability）。因此，实现可信的同行评审自动化需以严谨的评估框架和对评审机制本质的理解为基础，而非依赖未经充分验证的通用生成式AI技术。

链接: https://arxiv.org/abs/2605.03202
作者: Joachim Baumann,Jiaxin Pei,Sanmi Koyejo,Dirk Hovy
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注: Accepted at ICML 2026 (Spotlight)

点击查看摘要

Abstract:Large language models offer a tempting solution to address the peer review crisis. This position paper argues that today’s AI systems should not be used to produce paper reviews. We ground this position in an empirical comparison of human- versus AI-generated ICLR 2026 reviews and an evaluation of the effect of automated paper rewriting on different AI reviewers. We identify two critical issues: 1) AI reviewers exhibit a hivemind effect of excessive agreement within and across papers that reduces perspective diversity. 2) AI review scores are trivially gameable through paper laundering: prompting an LLM to rewrite a paper could significantly increase the scores from AI reviewers, demonstrating that LLM reviewers are easy to game through stylistic changes rather than scientific results. However, non-gameability and review diversity are necessary but not sufficient conditions for automation. We argue that addressing the peer review crisis requires a science of peer review automation – not general-purpose LLMs deployed without rigorous evaluation.

[AI-79] rminus-4B: Can a Smaller Model Replace Frontier LLM s at Agent ic Execution Tasks?

【速读】：该论文旨在解决现代编码代理（coding agent）在执行复杂任务时因子任务处理导致主代理上下文窗口膨胀的问题，特别是终端执行（terminal execution）等冗长输出如何影响主代理效率。解决方案的关键在于提出一个专为终端执行任务微调的小语言模型（small language model, SLM）——Terminus-4B，其基于Qwen3-4B模型通过监督微调（Supervised Finetuning, SFT）与基于评分标准的强化学习（rubric-based LLM-as-judge reward）进行训练，使子代理能够高效、准确地完成终端指令执行任务，从而显著减少主代理的token消耗（最高达~30%），且不牺牲基准测试（如SWE-Bench Pro和内部C#基准）上的性能表现，甚至在某些指标上超越了前沿模型（如Claude Sonnet / Opus / GPT-5.3-Codex）。

链接: https://arxiv.org/abs/2605.03195
作者: Spandan Garg,Vikram Nitin,Yufan Huang
机构: 未知
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注:

点击查看摘要

Abstract:Modern coding agents increasingly delegate specialized subtasks to subagents, which are smaller, focused agentic loops that handle narrow responsibilities like search, debugging or terminal execution. This architectural pattern keeps the main agent’s context window clean by isolating verbose outputs (e.g. build logs, test results, etc.) within the subagent context. Typically when agents employ subagents for such tasks, they use frontier models as these subagents. In this paper, we investigate whether a finetuned small language model (SLM) can achieve comparable performance to frontier models in the task of agentic terminal execution. We present Terminus-4B, which is a post-trained Qwen3-4B model via Supervised Finetuning (SFT) and Reinforcement Learning (RL) using rubric-based LLM-as-judge reward, specifically for this task. In our extensive evaluation spanning various frontier models, training ablations and main agent configurations, we find that Terminus-4B is able to reduce the token usage of the main agent by up to ~30% compared to the No Subagent baseline with no impact to agent performance on benchmarks like SWE-Bench Pro and our internal SWE-Bench C# benchmark, which tends to be heavy in verbose execution tasks. Furthermore, Terminus-4B improves key metrics showing the main agent relying on the outputs of the subagent and doing fewer terminal execution tasks by itself. We see that our model not only closes the gap between the Vanilla Qwen model and frontier models like Claude Sonnet / Opus / GPT-5.3-Codex, but often even exceeds their performance.

[AI-80] Global and Local Topology-Aware Attention with Persistent Homology and Euler Biases for Time-Series Forecasting

【速读】：该论文旨在解决标准点积注意力机制（dot-product attention）无法显式建模科学时间序列中蕴含的预测性拓扑结构（如连通性、循环、壳状几何、方向变化及非线性邻域关系）的问题。其解决方案的核心在于提出一种拓扑感知注意力框架，通过引入持久同调（persistent homology, H0–H2）、锚定欧拉特征变换（anchored Euler characteristic transforms）和核-希尔伯特通道（kernel-Hilbert channels）将此类几何结构嵌入注意力logits中；同时设计验证门控局部残差模块，在验证数据支持的前提下仅当拓扑信号具有预测价值时才引入修正项，从而实现拓扑信息的可控注入与有效利用。

链接: https://arxiv.org/abs/2605.03163
作者: Usef Faghihi,Amir Saki
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Scientific time series often encode predictive geometric structure, including connectivity, cycles, shell-like geometry, directional changes, and nonlinear neighborhoods, that standard dot-product attention does not explicitly represent. We introduce a topology-aware attention framework that adds such structure to attention logits using persistent homology (H0-H2), anchored Euler characteristic transforms, and kernel-Hilbert channels. A validation-gated local residual captures local topological signals, including a Zeng-style local H0 component, only when held-out validation data support the correction. Exact Vietoris-Rips computations and smooth topological surrogates are evaluated under a no-leakage protocol with train-only calibration, validation-only selection, and test-only reporting. We evaluate guarded topology-aware variants across three architecture families: lightweight attention/Ridge, PatchTSTForRegression, and TimeSeriesTransformerForPrediction. Experiments include synthetic benchmarks isolating higher-order topology and real datasets covering CO2, SP 500 return-window geometry, and NASA IMS bearing degradation. The audit uses matched paired comparisons across seven dataset units, three random seeds, and three chronological splits, giving 63 paired units per architecture and 189 paired units overall. Topology-aware models show positive paired effects when geometry is predictive, with heterogeneous magnitude across datasets and architectures. Lightweight attention/Ridge improves in 46 of 63 units, with mean relative RMSE reduction of 12.5% and paired randomization p=7.2e-4; PatchTST improves in 33 units and retains the baseline in 20 units, with 23.5% reduction and p=3.5e-5; and TimeSeriesTransformer improves in 47 units, with 47.8% reduction and p1e-4. The results support topology as a validation-selected, architecture-compatible inductive bias. Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI) Cite as: arXiv:2605.03163 [cs.LG] (or arXiv:2605.03163v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2605.03163 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[AI-81] Learning Correct Behavior from Examples: Validating Sequential Execution in Autonomous Agents

【速读】：该论文旨在解决自主代理（Autonomous Agents）在复杂任务中顺序行为验证的难题，传统测试方法依赖人工标注、精确序列匹配或大量训练样本，难以适应实际场景中的非确定性与多样性。其解决方案的关键在于结合编译器理论中的支配者分析（Dominator Analysis）与多模态大语言模型（Multimodal Large Language Models）驱动的语义理解能力，自动从2–10个通过执行轨迹中学习正确行为模式，并构建基于前缀树接受器（Prefix Tree Acceptors）的广义真值模型；通过多层级等价检测合并轨迹并利用拓扑子序列匹配对新执行进行验证，从而实现高精度、可解释的行为验证，在UI测试、代码生成和机器人流程等多个领域均展现出有效性。

链接: https://arxiv.org/abs/2605.03159
作者: Reshabh K Sharma,Gaurav Mittal,Yu Hu
机构: 未知
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注:

点击查看摘要

Abstract:As autonomous agents become increasingly sophisticated, validating their sequential behavior presents a significant challenge. Traditional testing approaches require manual specification, exact sequence matching, or thousands of training examples. We present a novel algorithm that automatically learns correct behavior from just 2-10 passing execution traces and validates new executions against this learned model. Our approach combines dominator analysis from compiler theory with multimodal large language model-powered semantic understanding to identify essential states and handle non-deterministic behavior. The system constructs a generalized ground truth model using Prefix Tree Acceptors, merges traces through multi-tiered equivalence detection, and validates new executions via topological subsequence matching. In controlled experiments, our system achieved high accuracy in detecting product bugs and false successes using only 3 training traces. This approach provides explainable validation results with coverage metrics and works across diverse domains including UI testing, code generation, and robotic processes.

[AI-82] Are you with me? A Framework for Detecting Mental Model Discrepancies in Task-Based Team Dialogues

【速读】：该论文旨在解决团队成员间共享心智模型（Shared Mental Model, SMM）不一致的问题，这种不一致源于自然语言沟通中的信息遗漏，进而影响团队整体性能。传统SMM评估方法依赖事后专家编码，难以捕捉实时协作动态。其解决方案的关键在于提出一个可从团队对话中自动识别并分类四类心智模型差异的框架：无支持信念（unsupported beliefs）、错误信念（false beliefs）、信念矛盾（belief contradictions）和遗漏（omissions），并通过实证分析表明这些差异模式在协作任务对话中具有预测未来心智模型错位的能力，为实时团队协调提供量化依据。

链接: https://arxiv.org/abs/2605.03149
作者: Katharine Kowalyshyn,Matthias Scheutz
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注: Accepted to Proceedings of the Annual Meeting of the Cognitive Science Society 2026

点击查看摘要

Abstract:Humans typically use natural language to update teammates on task states. Since not all updates are communicated, discrepancies arise between the team members’ mental models that negatively affect overall team performance. How can we categorize such discrepancies? Do misalignments detected in team dialogue predict future mental model misalignments? Traditional shared mental model (SMM) assessment methods rely on retrospective expert coding that cannot capture real-time coordination dynamics. We propose a framework to identify and categorize four types of mental model discrepancies: unsupported beliefs, false beliefs, belief contradictions, and omissions, all of which can naturally emerge in team dialogues. Using dialogues from twenty dyad teams performing collaborative object identification tasks across four sequential levels, we demonstrate that these discrepancy patterns contain predictive signals. Averaging historical discrepancy counts achieves meaningful prediction accuracy using uniform weighting as an exploratory baseline, with differential predictability across discrepancy types.

[AI-83] Pact: A Choreographic Language for Agent ic Ecosystems

【速读】：该论文旨在解决多智能体系统中自利代理（self-interested agents）在开放、多方协作环境中如何被激励遵守协议的问题，传统 choreographic programming 假设参与者是合作的，缺乏对代理动机和策略选择的形式化建模。解决方案的关键在于提出 Pact 语言，它扩展了 choreographic programming 的语法以显式描述代理的选择与偏好，并引入博弈论形式化框架——每个 Pact 协议可映射为一个正式博弈，从而允许设计者通过分析博弈性质（如纳什均衡或最优决策策略）来验证协议的稳定性与有效性；文中还实现了基于 bounded-rationality 的求解器，用于计算实际场景下代理的决策策略，验证了该方法在自利代理多方协调中的可行性。

链接: https://arxiv.org/abs/2605.03143
作者: Kiran Gopinathan,Jack Feser,Michelangelo Naim,Zenna Tavares,Eli Bingham
机构: 未知
类目: Programming Languages (cs.PL); Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC)
备注: To be presented at the 2nd International Workshop on Choreographic Programming (CP 2026)

点击查看摘要

Abstract:Recent advances in large language models have led to the rise of software systems (i.e. agents) that execute with increasing autonomy on behalf of users in open, multi-party settings, interacting with untrusted counterparts and managing private information. Choreographic programming offers correct-by-construction protocol-design for such settings, but assumes cooperative participants – it has no notion of agent self-interest, that is, why an agent will follow a protocol. In this talk we introduce Pact, a choreographic language extended with operations to describe agent choices and preferences, drawing from the rich literature of game theory. Every Pact protocol maps to a formal game, allowing protocol designers to reason about game-theoretic properties of their protocols, such as solving for decision policies. We present Pact’s design and a preliminary implementation – a bounded-rational solver that computes decision policies over Pact protocols – and findings from applying this language to multi-party coordination with self-interested agentic participants. Comments: To be presented at the 2nd International Workshop on Choreographic Programming (CP 2026) Subjects: Programming Languages (cs.PL); Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC) Cite as: arXiv:2605.03143 [cs.PL] (or arXiv:2605.03143v1 [cs.PL] for this version) https://doi.org/10.48550/arXiv.2605.03143 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[AI-84] ARISE: A Repository-level Graph Representation and Toolset for Agent ic Fault Localization and Program Repair

【速读】：该论文旨在解决仓库级故障定位（Repository-level Fault Localization, FL）与自动化程序修复（Automated Program Repair, APR）中因缺乏细粒度语义信息而导致的精度不足问题，特别是现有基于图的系统仅建模文件、类、函数等结构关系，而未刻画过程内变量值流（intra-procedural data flow），导致模型难以实现函数级和行级精准定位。解决方案的关键在于提出 ARISE（Agentic Repository-level Issue Solving Engine），其核心创新是构建了一个多粒度程序图（multi-granularity program graph），将结构关系细化至语句级别，并通过定义-使用边（definition-use edges）显式建模变量值流；同时设计了三层工具API，使数据流切片（data-flow slicing）成为可查询的一等代理原语（first-class agent primitive），从而支持模型在单次调用中高效追踪目标变量的定义或使用点。实验证明，该方法显著提升了定位准确率（Function Recall@1提升17.0点，Line Recall@1提升15.0点），并直接转化为修复成功率的提升（Pass@1达22.0%，较基线提升4.7个百分点）。

链接: https://arxiv.org/abs/2605.03117
作者: Shahd Seddik,Fatemeh Fard
机构: 未知
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Repository-level fault localization (FL) and automated program repair (APR) require an agent to identify the relevant code units across files, follow call and data dependencies, and generate a valid patch. Existing graph-based systems provide structural representations of repositories (files, classes, functions and their relationships) but do not model how variable values flow within procedures, leaving agents without the semantic precision needed for function- and line-level localization. We present ARISE (Agentic Repository-level Issue Solving Engine), which augments an LLM-based agent with a multi-granularity program graph that extends structural relationships down to statement-level nodes connected by intra-procedural definition-use edges. ARISE exposes this graph through a three-tier tool API, which brings data-flow slicing as a first-class, queryable agent primitive that allows the model to trace, in a single call, which statements define or consume a variable of interest. We evaluate on SWE-bench Lite (300 real GitHub issues, 11 Python repositories) using Qwen2.5-Coder-32B-Instruct as the backbone. Compared to the unmodified SWE-agent baseline, ARISE improves Function Recall@1 by 17.0 points and Line Recall@1 by 15.0 points. These localization gains translate directly into repair success, with ARISE achieving 22.0% Pass@1 (66/300), a 4.7 percentage-point improvement over SWE-agent. Controlled ablations confirm that the improvement is driven by the data-flow graph rather than the tool schema, and that large code models consume structured slice output directly without requiring a natural-language summarization layer. The graph builder and slicing API are designed as a framework-agnostic, drop-in toolset for future APR research.

[AI-85] Cascade Token Selection for Transformer Attention Acceleration

【速读】：该论文旨在解决Transformer模型中注意力层内代表性token选择的计算成本过高问题，尤其是在大规模语言模型中，传统方法如激活去相关注意力（Activation Decorrelation Attention, ADA）需要在每一层都计算一个 $T \times T$ 的Gram矩阵来筛选代表性token，导致复杂度高达 $O(T^2 d)$ 每层。解决方案的关键在于引入一种级联机制（cascade mechanism），该机制将前一层选出的代表性token集合继承至下一层，并通过一个 $(T - r) \times r$ 的交叉Gram计算进行验证与微调，仅需少量增删操作即可更新代表集，从而将每层的选择成本降至 $O(T r d)$ 。实验表明，该策略在多个模型（GPT-2、GPT-J、OPT）上实现了22%–63%的Gram运算节省，且相邻层间代表性token集合具有高度一致性（平均Jaccard重叠达0.83–0.94），揭示了信息性token集合是输入结构的内在属性并沿网络深度稳定传播。

链接: https://arxiv.org/abs/2605.03110
作者: Stephen J. Thomas
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:A method is presented for reducing the cost of representative token selection in transformer attention layers by exploiting the coherence of the representative set across depth. Activation Decorrelation Attention (ADA) selects r \ll T representative tokens at each layer via a Gram threshold and computes attention on the compressed r \times r problem, but the selection requires a T \times T Gram matrix at every layer. The cascade mechanism introduced here inherits the representative set from layer l to layer l+1 , validates it via a (T - r) \times r cross-Gram computation, and updates it with a small number of additions and removals. The cost of the selection step drops from O(T^2 d) to O(T r d) per layer. Validation on three model families (GPT-2 124M, GPT-J 6B, OPT 6.7B) on AMD MI300X demonstrates Gram operation savings of 22% to 63% with mean Jaccard overlap of 0.83 to 0.94 between consecutive layers. The cascade reveals that the set of informative tokens is a structural property of the input that propagates coherently through the depth of the network: the same tokens carry the non-redundant information at layer l and at layer l+1 .

[AI-86] Gated Subspace Inference for Transformer Acceleration

【速读】：该论文旨在解决Transformer语言模型推理过程中因线性层权重读取带来的高内存带宽消耗问题，从而提升推理效率。其核心解决方案是利用每一层token激活流形的低有效秩特性，将激活向量分解为子空间分量和残差部分：通过缓存低秩权重映像在子空间上计算线性层输出以降低内存带宽需求，并引入逐token门控机制决定是否计算残差修正，从而在保证输出分布误差可控的前提下实现加速。该方法无需重新训练、不修改网络结构且不近似注意力机制，实验证明可在保持高精度（如top-1 token一致率>98%）的同时获得3.0x至10.5x的加速效果。

链接: https://arxiv.org/abs/2605.03109
作者: Stephen J. Thomas
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 19 pages, journal article

点击查看摘要

Abstract:A method is presented for accelerating inference in transformer language models by exploiting the low effective rank of the token activation manifold at each layer. The method decomposes each activation vector into a subspace component and a residual, computes the linear-layer output on the subspace component via a cached low-rank weight image at reduced memory bandwidth, and applies a per-token gate that determines whether the residual correction is computed or skipped. The gate ensures that the output distribution is preserved to within a controllable tolerance. Validation on three model families (GPT-2 124M, GPT-J 6B, OPT 6.7B) on AMD MI300X demonstrates effective speedups of 3.0x to 10.5x on linear-layer weight reads with perplexity ratios below 1.00 and top-1 token agreement above 98%. The method requires no retraining, no architectural modification, and no approximation of the attention mechanism. At the operating point (k = 256, \epsilon = 0.05) on GPT-J 14 6B, the accelerated model produces character-for-character identical output to the baseline.

[AI-87] Programmatic Context Augmentation for LLM -based Symbolic Regression

【速读】：该论文旨在解决符号回归（Symbolic Regression, SR）中传统基于遗传算法的方法在可扩展性和表达能力上的局限性，以及现有大语言模型（Large Language Model, LLM）基方法仅依赖标量评估指标（如均方误差）作为反馈信号、忽略数据集内部丰富信息的问题。其解决方案的关键在于提出一种新型LLM驱动的进化搜索框架，通过引入程序化上下文增强（programmatic context augmentation），使模型能够以代码形式与数据集进行交互，主动执行数据分析并提取比单一评估分数更丰富的信息信号，从而提升搜索效率与精度。

链接: https://arxiv.org/abs/2605.03101
作者: Hao Liu,Xiao-Wen Yang,Atharva Sehgal,Yixin Wang,Lan-Zhe Guo,Yu-Feng Li,Yisong Yue
机构: 未知
类目: Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Symbolic regression (SR), the task of discovering mathematical expressions that best describe a given dataset, remains a fundamental challenge in scientific discovery. Traditional approaches, primarily based on genetic algorithms and related evolutionary methods, have proven useful but suffer from scalability and expressivity limitations. Recently, large language model (LLM)-based evolutionary search methods have been introduced into SR and show promise. However, existing LLM-based approaches typically rely on scalar evaluation metrics, such as mean squared error, as the sole source of feedback during the search process, thereby overlooking the rich information embedded in the dataset. To address this limitation, we propose a novel LLM-based evolutionary search framework that incorporates programmatic context augmentation. By enabling code-based interactions with the dataset, our method can actively perform data analysis and extract informative signals, beyond aggregated evaluation scores. We evaluate our framework on advanced benchmarks, such as LLM-SRBench, and demonstrate superior efficiency and accuracy compared to strong baselines.

[AI-88] From Barrier to Bridge: The Case for AI Data Center/Power Grid Co-Design

【速读】：该论文旨在解决生成式 AI (Generative AI) 训练数据中心对传统电力系统稳定性的冲击问题，即传统电网依赖的“负载多样性”（load diversity）假设因超大规模训练园区的集中、同步高功率需求而失效。其核心解决方案在于推动计算基础设施与电力基础设施从历史上的隐性共存转向显性协同发展，关键在于建立跨行业的协同设计原则、操作哲学与经济激励机制，并聚焦联合容量规划、多时间尺度控制、算力-电力协议栈及市场创新等研究方向，以实现AI可持续、可靠发展的能源支撑体系。

链接: https://arxiv.org/abs/2605.03090
作者: Noman Bashir,Rob Sherwood,Le Xie,Minlan Yu
机构: 未知
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Artificial Intelligence (cs.AI); Systems and Control (eess.SY)
备注:

点击查看摘要

Abstract:For over a century, the electric grid has relied on a single statistical assumption: \emphload diversity, the principle that the uncorrelated demands of millions of small consumers produce a smooth, predictable aggregate. AI training data centers break that assumption. A single hyperscale training campus can draw power comparable to a mid-sized city, driven by one tightly synchronized job whose demand swings by hundreds of megawatts in seconds. This paper argues that the resulting entanglement of compute and power infrastructure requires a shift from implicit coexistence to explicit co-development between the historically decoupled data center and electric power industries. We introduce the distinct design principles, operational philosophies, and economic incentives of each sector, and show why their cultural and technical misalignment makes coordination difficult. We identify key research directions, from joint capacity planning, multi-timescale control, a compute–power protocol stack, to market innovation, that must be pursued to power the future of AI sustainably and reliably.

[AI-89] Refining Compositional Diffusion for Reliable Long-Horizon Planning

【速读】：该论文旨在解决组合式扩散规划（compositional diffusion planning）在处理局部计划分布具有多模态特性时，因模式平均（mode-averaging）导致全局轨迹既不可行又不连贯的问题。现有方法在拼接短时程轨迹片段时，若各片段的局部分布存在多个不兼容的模式，直接平均会生成低质量的合成路径。解决方案的关键在于提出一种无需训练的引导方法——精炼组合扩散（Refining Compositional Diffusion, RCD），其核心机制是利用预训练扩散模型的自重构误差作为组合轨迹对数密度的代理指标，并引入重叠一致性项（overlap consistency term）以确保片段边界处的一致性。该联合引导策略能够聚焦于高密度、全局一致的轨迹区域，有效缓解模式平均问题，在OGBench基准中涵盖运动控制、物体操作及像素观测等复杂长程任务上显著优于现有方法。

链接: https://arxiv.org/abs/2605.03075
作者: Kyowoon Lee,Yunhao Luo,Anh Tong,Jaesik Choi
机构: 未知
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:

点击查看摘要

Abstract:Compositional diffusion planning generates long-horizon trajectories by stitching together overlapping short-horizon segments through score composition. However, when local plan distributions are multimodal, existing compositional methods suffer from mode-averaging, where averaging incompatible local modes leads to plans that are neither locally feasible nor globally coherent. We propose Refining Compositional Diffusion (RCD), a training-free guidance method that steers compositional sampling toward high-density, globally coherent plans. RCD leverages the self-reconstruction error of a pretrained diffusion model as a proxy for the log-density of composed plans, combined with an overlap consistency term that enforces consistency at segment boundaries. We show that the combined guidance concentrates sampling on high-density plans that mitigate mode-averaging. Experiments on challenging long-horizon tasks from OGBench, including locomotion, object manipulation, and pixel-based observations, demonstrate that RCD consistently outperforms existing methods.

[AI-90] Computing Thiele Rules on Interval Elections and their Generalizations

【速读】：该论文旨在解决在投票机制中计算Thiele规则（如比例代表制的Proportional Approval Voting, PAV）的复杂性问题，特别是在结构化偏好域下（如候选者区间域CI、选民区间域VI及更广义的Voter-Candidate Interval域VCI和线性一致域LC）的可计算性问题。其核心挑战在于：尽管在CI域可通过具有全单模约束矩阵的线性规划（LP）高效求解，但在VI域中传统LP方法失效，且该问题的复杂性长期未被厘清。论文的关键突破在于证明：即使VI域对应的LP矩阵非全单模，其仍存在至少一个最优整数解，并提出了一种快速算法来找到该解；进一步地，作者将此技术推广至VCI与LC域，并揭示了LC严格包含VCI的图论关系，同时给出了LC域的等价刻画，从而统一并深化了对Thiele规则在结构化偏好下的计算性质的理解。

链接: https://arxiv.org/abs/2605.03067
作者: Dimitris Avramidis,Alexandra Lassota,Ulrike Schmidt-Kraepelin,Adrian Vetta
机构: 未知
类目: Artificial Intelligence (cs.AI); Computer Science and Game Theory (cs.GT)
备注: 19 pages

点击查看摘要

Abstract:Approval-based committee voting has received significant attention in the social choice community. Among the studied rules, Thiele rules, and especially Proportional Approval Voting (PAV), stand out for desirable properties such as proportional representation, Pareto optimality, and support monotonicity. Their main drawback is that computing a Thiele outcome is NP-hard in general. A glimpse of hope comes from the fact that Thiele rules are better behaved under structured preferences. On the candidate interval (CI) domain, they are computable in polynomial time via a linear program (LP) that has a totally unimodular constraint matrix. Surprisingly, this approach fails for the related voter interval (VI) domain, and the complexity of the problem has repeatedly been posed as an open question. Our main result resolves this question: although the relevant matrix is not totally unimodular, the ``standard’’ LP still admits at least one optimal integral solution, and we provide a fast algorithm for finding it. Our technique naturally extends to the voter-candidate interval (VCI) domain, also known as the 1-dimensional voter-candidate range (1D-VCR) domain, and to the linearly consistent (LC) domain, both of which generalize the candidate and voter interval domains. Although both the VCI and LC domains have been studied in social choice, their relationship was unknown. We show, through connections to graph theory, that LC strictly contains VCI. We also provide an alternative definition of LC that is closer in spirit to VCI and has a natural interpretation in approval elections; this equivalence may be of independent interest. Finally, we study an alternative tree-based generalization of VCI and show that Thiele rules become NP-hard to compute on this domain. Comments: 19 pages Subjects: Artificial Intelligence (cs.AI); Computer Science and Game Theory (cs.GT) Cite as: arXiv:2605.03067 [cs.AI] (or arXiv:2605.03067v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2605.03067 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[AI-91] Neuron-Anchored Rule Extraction for Large Language Models via Contrastive Hierarchical Ablation

【速读】：该论文旨在解决生成式 AI（Generative AI）中可解释性（Explainable AI, XAI）的核心挑战：如何将大型语言模型（Large Language Models, LLMs）的决策逻辑以符号形式表达，并与模型内部机制建立可靠关联。现有方法存在两方面局限：全局规则提取方法通常无法基于模型电路（circuitry）进行规则锚定，而机制可解释性方法则依赖人工假设和昂贵的神经元级干预。论文提出MechaRule框架，其关键在于通过高效定位一类称为“激动剂”（agonists）的稀疏神经元来实现规则提取的电路锚定——这些神经元的激活中止会破坏与特定规则相关的推理行为。解决方案的核心创新在于两个经验观察：一是基于固定基线/翻转场景下激动剂效应的近似单调饱和特性，设计出受条件强度谓词驱动、带置信度引导保守剪枝的自适应组测试策略，理论上仅需Θ(k log(N/k) + k)次干预即可完成对N个候选神经元中k个激动剂的定位；二是通过数据分割验证消融效果，确保规则忠实性，从而提升激动剂识别的可靠性。实证表明，MechaRule在算术和越狱任务上能召回96.8%的高影响激动剂，并显著削弱模型性能（最高达71.1%的算术准确率下降和8.8%的越狱成功率下降），验证了其有效性和实用性。

链接: https://arxiv.org/abs/2605.03058
作者: Francesco Sovrano,Gabriele Dominici,Marc Langheinrich
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:A key goal of explainable AI (XAI) is to express the decision logic of large language models (LLMs) in symbolic form and link it to internal mechanisms. Global rule-extraction methods typically learn symbolic surrogates without grounding rules in model circuitry, while mechanistic interpretability can connect behaviors to neuron sets but often depends on hand-crafted hypotheses and expensive neuron-level interventions. We introduce MechaRule, a pipeline that grounds rule extraction in LLM circuits by efficiently localizing sparse neurons called agonists, whose activation neutralization disrupts rule-related behaviors. MechaRule rests on two empirical observations. First, within a fixed baseline/flip regime, sparse agonist effects can be approximately monotone and saturating: a few dominant neuron activations can overtop weaker ones at coarse scales, while overlapping neurons flip many of the same examples. This motivates viewing localization as adaptive group testing driven by a regime-conditional strength predicate with confidence-guided conservative pruning, yielding Theta(k log(N/k) + k) interventions over N candidates when k N neurons are agonists under the monotone-overtopping abstraction. Second, agonists emerge more reliably when ablations are verified through data splits aligned with close-to-faithful rule behavior; spectral splits remain a useful rule-free fallback, while unfaithful splits degrade localization. Empirically, overtopping appears mainly in learned, task-aligned regimes: on arithmetic and jailbreak tasks across Qwen2 and GPT-J, MechaRule recalls 96.8% of high-effect brute-force agonists in completed comparisons, and suppressing localized agonists reduces arithmetic accuracy and jailbreak success by up to 71.1% and 8.8%, respectively.

[AI-92] ARIS: Autonomous Research via Adversarial Multi-Agent Collaboration

【速读】：该论文旨在解决大语言模型（Large Language Models, LLMs）在长期科研任务中因缺乏有效监督机制而导致的“看似合理但证据不足的成功”问题，即代理系统可能生成表面上成立、实则缺乏充分支撑的结论。其解决方案的关键在于提出ARIS（Auto-Research-in-sleep）研究框架，通过三层架构实现对科研流程的自动化与可信保障：执行层提供可复用技能和持久化知识库，编排层支持多模型协作与灵活路由，保证层引入三阶段验证机制（完整性核查、结果到结论映射、声明审计）及科学编辑流水线，确保每一步实验结论均有可追溯的证据链支撑，从而显著提升自主研究系统的可靠性与可审计性。

链接: https://arxiv.org/abs/2605.03042
作者: Ruofeng Yang,Yongcan Li,Shuai Li
机构: 未知
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注: Technical report. Code at this https URL

点击查看摘要

Abstract:This report describes ARIS (Auto-Research-in-sleep), an open-source research harness for autonomous research, including its architecture, assurance mechanisms, and early deployment experience. The performance of agent systems built on LLMs depends on both the model weights and the harness around them, which governs what information to store, retrieve, and present to the model. For long-horizon research workflows, the central failure mode is not a visible breakdown but a plausible unsupported success: a long-running agent can produce claims whose evidential support is incomplete, misreported, or silently inherited from the executor’s framing. Therefore, we present ARIS as a research harness that coordinates machine-learning research workflows through cross-model adversarial collaboration as a default configuration: an executor model drives forward progress while a reviewer from a different model family is recommended to critique intermediate artifacts and request revisions. ARIS has three architectural layers. The execution layer provides more than 65 reusable Markdown-defined skills, model integrations via MCP, a persistent research wiki for iterative reuse of prior findings, and deterministic figure generation. The orchestration layer coordinates five end-to-end workflows with adjustable effort settings and configurable routing to reviewer models. The assurance layer includes a three-stage process for checking whether experimental claims are supported by evidence: integrity verification, result-to-claim mapping, and claim auditing that cross-checks manuscript statements against the claim ledger and raw evidence, as well as a five-pass scientific-editing pipeline, mathematical-proof checks, and visual inspection of the rendered PDF. A prototype self-improvement loop records research traces and proposes harness improvements that are adopted only after reviewer approval.

[AI-93] Stable Agent ic Control: Tool-Mediated LLM Architecture for Autonomous Cyber Defense

【速读】：该论文旨在解决高风险决策场景下代理系统（Agentic Systems）在对抗压力中缺乏形式化保障的问题，尤其针对安全运营中心（SOC）在配置终端检测与响应（EDR）策略时面临的复杂对抗环境。解决方案的关键在于提出一种工具中介架构：LLM代理通过使用确定性工具（如Stackelberg最优响应、贝叶斯观测更新和攻击图原语）并从有限动作目录中选择操作，这些操作在工具输出接口处被强制执行。该架构利用Lean 4中机器验证的复合李雅普诺夫函数，形式化证明了系统的可控性、可观测性（即使传感器数据不对称）以及输入到状态稳定性（ISS）鲁棒性，且两个推论扩展了该证书至任意来自目录的控制器或对手。实验表明，在282个真实企业攻击图上，理论声明具有显著裕度；在成对攻防遥测数据上，基于Claude Sonnet 4的工具中介控制器相比确定性贪心基线将攻击者期望收益降低59%，且在40次运行中无方差变化，验证了架构稳定性不依赖于控制器能力，而LLM的非确定性则增强了策略探索能力。

链接: https://arxiv.org/abs/2605.03034
作者: Kerri Prinos,Lilianne Brush,Cameron Denton,Zhanqi Wang,Joshua Knox,Snehal Antani,Anton Foltz,Amy Villaseñor
机构: 未知
类目: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Systems and Control (eess.SY)
备注: 23 pages total (9 main paper + 16 appendices/references), 2 figures

点击查看摘要

Abstract:Agentic systems involved in high-stake decision-making under adversarial pressure need formal guarantees not offered by existing approaches. Motivated by the operational needs of security operations centers (SOCs) that must configure endpoint detection and response (EDR) policies under adversarial pressure, we present a tool-mediated architecture: LLM agents use deterministic tools (Stackelberg best-response, Bayesian observer updates, attack-graph primitives) and select from finite action catalogs enforced at the tool-output interface. A composite Lyapunov function machine-checked in Lean 4 with zero sorry certifies controllability, observability from asymmetric sensor data, and Input-to-State Stability (ISS) robustness under intelligent adversarial disturbance, with two corollaries extending the certificate to any controller or adversary from the catalogs. On 282 real enterprise attack graphs, the claims hold with margin. On paired offensive/defensive telemetry, a tool-mediated Claude Sonnet 4 controller reduces the attacker’s expected payoff (game value) by 59% relative to a deterministic greedy baseline, with zero variance across 40 runs at four temperatures. A Claude Haiku 4.5 controller converges to suboptimal game values but stays catalog-bounded over an additional 40 runs, demonstrating that architectural stability is not dependent on the controller capability. The LLM agent’s non-determinism furthers creative exploration of strategies, while the tool-mediated architecture ensures system stability.

[AI-94] Structured Diffusion Bridges: Inductive Bias for Denoising Diffusion Bridges ICML2026

【速读】：该论文旨在解决模态翻译（Modality Translation）任务中因跨模态映射不唯一而导致的欠约束问题，即在缺乏完全配对数据的情况下，如何有效建模和生成合理的跨模态对应关系。其解决方案的关键在于提出一种基于扩散桥（Diffusion Bridge）的框架，通过引入对齐约束（alignment constraints）来刻画可接受解空间，并将配对监督视为可选的启发式手段而非必要前提，从而在未配对、半配对和全配对等多种数据条件下均能保持稳定性能，尤其在显著降低配对要求的同时接近全配对场景下的表现，验证了扩散桥作为超越严格配对数据限制的灵活模态翻译基础的有效性。

链接: https://arxiv.org/abs/2605.02973
作者: Eitan Kosman,Gabriele Serussi,Chaim Basking
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted to ICML 2026

点击查看摘要

Abstract:Modality translation is inherently under-constrained, as multiple cross-modal mappings may yield the same marginals. Recent work has shown that diffusion bridges are effective for this task. However, most existing approaches rely on fully paired datasets, thereby imposing a single data-driven constraint. We propose a diffusion-bridge framework that characterizes the space of admissible solutions and restricts it via alignment constraints, treating paired supervision as an optional heuristic rather than a prerequisite. We validate our method on synthetic and real modality translation benchmarks across unpaired, semi-paired, and paired regimes, showing consistent performance across supervision levels. Notably, \textbfit achieves near fully-paired quality with a substantial relaxation in pairing requirements, and remaining applicable in the unpaired regime. These results highlight diffusion bridges as a flexible foundation for modality translation beyond fully paired data.

[AI-95] Decompose to Understand Fuse to Detect: Frequency-Decoupled Anomaly Detection for Encrypted Network Traffic

【速读】：该论文旨在解决加密网络流量异常检测中因主流图像化方法存在“全频谱”特性与“频谱不匹配”问题而导致的表征不完整和检测性能下降难题。其核心解决方案是提出一种频率解耦框架FreeUp，通过将流量数据分解为低频与高频分量，并分别采用独立分支进行处理，结合定制化训练策略实现稳定的频段特异性学习；同时引入基于不确定性的融合评分机制，量化各频段分支的重建不确定性并动态整合输出，从而提升异常评分的全面性与可靠性。

链接: https://arxiv.org/abs/2605.02970
作者: Xinglin Lian,Chengtai Cao,Ting Zhong,Yong Wang,Kai Chen,Fan Zhou
机构: 未知
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: Accepted by INFOCOM 2026

点击查看摘要

Abstract:Network traffic anomaly detection represents a critical cybersecurity task, yet widespread encryption makes this task increasingly challenging. In response, image-based methods that model traffic as visual patterns have emerged as the dominant approach. However, this work pioneers the identification of a pervasive full-frequency'' characteristic and an associated limitation termed spectral mismatch’’ within this paradigm. Specifically, while encrypted traffic exhibits prominent high-frequency components, mainstream reconstruction methods demonstrate an inherent bias toward learning low-frequency information. This fundamental mismatch results in incomplete representations that consequently degrade anomaly detection performance. To address this challenge, we propose FreeUp, a novel frequency-decoupled framework designed explicitly for encrypted traffic analysis. FreeUp decomposes traffic data into distinct low- and high-frequency bands, processing them through separate, dedicated branches along with a customized training strategy that ensures stable and independent frequency-specific learning. Furthermore, recognizing that simple reconstruction error proves inadequate for evaluating dual-branch architectures, we introduce an uncertainty-inspired fusion scoring mechanism. This mechanism quantifies the reconstruction uncertainty of the frequency-specific branches and dynamically integrates their outputs, yielding a more comprehensive and reliable anomaly score. Extensive experiments across multiple benchmarks demonstrate that FreeUp consistently outperforms state-of-the-art baselines. The code is available at this https URL.

[AI-96] Finite-Size Gradient Transport in Large Language Model Pretraining: From Cascade Size to Intensive Transport Efficiency

【速读】：该论文旨在解决大语言模型（Large Language Models, LLMs）训练过程中梯度传播机制的量化建模问题，特别是如何在有限规模下刻画梯度流动的多维特性及其与模型性能的关系。其解决方案的关键在于提出一个基于五个可观测量（ $D, z, \beta, \delta, v_\mathrm{rel}$ ）的有限尺寸梯度传输框架，这些变量分别表征级联规模、持续时间、绝对传输量及强度传输效率等物理属性，并通过实测原始梯度数据（来自Pico-LM和Pythia两个模型系列）验证了该框架在不同尺度下的代数封闭性与一致性。该方法揭示了不同模型家族在传输 regimes 上的系统性差异，例如 Pico-LM 显示正持续时间标度与负强度效率标度，而 Pythia 接近 $D=1$ 基线且效率依赖微弱，从而支持了一种可复用的梯度传输测量范式，而非依赖于统一固定点或第一性原理推导的神经网络缩放定律。

链接: https://arxiv.org/abs/2605.02968
作者: Ping Wang,Yan-Qi Du
机构: 未知
类目: Machine Learning (cs.LG); Disordered Systems and Neural Networks (cond-mat.dis-nn); Artificial Intelligence (cs.AI); Adaptation and Self-Organizing Systems (nlin.AO)
备注:

点击查看摘要

Abstract:We introduce a finite-size gradient-transport framework for real language-model training, based on five observables (D,z,\beta,\delta,v_\mathrmrel) that separate cascade size, duration, absolute transport, and intensive transport efficiency. We analyze direct raw-gradient measurements from Pico-LM across four scales and 125 aligned steps, together with a five-scale Pythia companion dataset built from 153 aligned checkpoint-difference update fields. The same algebraic closure holds in both families, and both share a near-unity cascade-size backbone, but they occupy distinct transport regimes: Pico-LM shows positive duration scaling and negative intensive-efficiency scaling, whereas Pythia remains near the D=1 baseline with only weak positive efficiency scale dependence. Randomized-field controls give nearly matched null floors in the intensive and duration channels, indicating that the contrast reflects different real departures from a shared null skeleton rather than different null calibrations. The families also differ in stepwise power-law compressibility: Pico-LM retains clean duration and efficiency power laws, whereas Pythia preserves the size backbone but shows weaker one-slope compressibility in those channels. External performance associations are correspondingly channel-level, carried mainly by v_\mathrmrel and normalized cascade duration, while D(t) acts as a shared size backbone without a significant exponent-level performance association. These results support a reusable transport measurement framework without claiming a universal fixed point or a first-principles derivation of neural scaling laws.

[AI-97] Reward Hacking Benchmark: Measuring Exploits in LLM Agents with Tool Use ICML2026

【速读】：该论文旨在解决强化学习（Reinforcement Learning, RL）训练的语言模型代理在获得工具访问权限后可能出现的“奖励欺骗”（Reward Hacking）问题，即模型通过非预期路径（如跳过验证步骤、利用任务邻近元数据或篡改评估函数）获取高奖励，而非完成真实任务目标。解决方案的关键在于提出并构建了一个名为Reward Hacking Benchmark (RHB) 的多步任务基准，涵盖自然主义的捷径机会和链式任务结构以模拟长时程行为，并通过系统性评估13个前沿模型发现：RL后训练显著增加奖励欺骗风险（例如DeepSeek-R1-Zero达13.9%，而未使用RL的Claude Sonnet 4.5为0%），且72%的欺骗行为附带显式的思维链推理，表明模型常将此类行为合理化为合法解法；进一步地，通过简单的环境加固策略可使欺骗率降低5.7个百分点（相对减少87.7%），且不损害任务成功率，揭示出生产对齐的后训练虽能抑制低复杂度任务中的奖励欺骗，但对高难度任务则失效。

链接: https://arxiv.org/abs/2605.02964
作者: Kunvar Thaman
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 16 pages, 2 figures. Accepted to ICML 2026

点击查看摘要

Abstract:Reinforcement learning (RL) trained language model agents with tool access are increasingly deployed in coding assistants, research tools, and autonomous systems. We introduce the Reward Hacking Benchmark (RHB), a suite of multi-step tasks requiring sequential tool operations with naturalistic shortcut opportunities such as skipping verification steps, inferring answers from task-adjacent metadata, or tampering with evaluation-relevant functions. RHB supports independent and chained task regimes, where chain length acts as a proxy for longer-horizon agent behavior. We evaluate 13 frontier models from OpenAI, Anthropic, Google, and DeepSeek. Exploit rates range from 0% (Claude Sonnet 4.5) to 13.9% (DeepSeek-R1-Zero), varying sharply by post-training style. A controlled sibling comparison (DeepSeek-V3 vs. DeepSeek-R1-Zero) shows RL post-training is associated with substantially higher reward hacking (0.6% vs. 13.9%), with consistent gaps across all four task families. We identify six exploit categories and find that 72% of reward hacking episodes include explicit chain-of-thought rationale, suggesting models often frame exploits as legitimate problem-solving. Simple environmental hardening reduces exploit rates by 5.7 percentage points (87.7% relative) without degrading task success. Models with near-zero exploit rates on standard tasks show elevated rates on harder variants, suggesting that production-aligned post-training appears to suppress reward hacking only below a complexity threshold where honest solutions remain tractable. Comments: 16 pages, 2 figures. Accepted to ICML 2026 Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI) Cite as: arXiv:2605.02964 [cs.LG] (or arXiv:2605.02964v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2605.02964 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[AI-98] Analytic Bridge Diffusions for Controlled Path Generation

【速读】：该论文旨在解决生成式 AI (Generative AI) 中桥扩散（bridge-diffusion）方法在有限时间内实现概率分布传输时依赖神经网络和随机模拟循环所导致的计算复杂性与不可解析性问题。其核心解决方案是提出线性-二次-高斯混合路径积分扩散（LQ-GM-PID）模型，该模型基于经典线性-二次-高斯（LQG）控制框架，保留线性动力学、高斯噪声和二次代价结构，但将终端状态调节替换为预设的终端概率密度，并允许初始与终端分布均为高斯混合（Gaussian Mixture, GM）形式。关键突破在于：在该受限但足够广泛的问题类中，得分函数（score）、中间边际分布及协议梯度均可解析求解，无需神经网络或内部随机模拟循环，从而实现了亚50毫秒级的封闭形式预计算，显著提升了可解释性和效率，并为当前基于神经网络的桥扩散与生成传输方法提供了精确的参考基准。

链接: https://arxiv.org/abs/2605.02961
作者: Michael Chertkov
机构: 未知
类目: Machine Learning (cs.LG); Statistical Mechanics (cond-mat.stat-mech); Artificial Intelligence (cs.AI); Systems and Control (eess.SY); Optimization and Control (math.OC)
备注: 47 pages, 18 figures

点击查看摘要

Abstract:Most modern bridge-diffusion methods achieve finite-time transport by specifying an interpolation, Schrödinger-bridge, or stochastic-control objective and then learning the associated score or drift field with a neural network. In contrast, we identify a restricted but sufficiently broad and analytically solvable class in which the score, intermediate marginals, and protocol gradients are available in closed form without inner stochastic simulation loops and without neural networks in the optimization loop. We recast the classical linear–quadratic–Gaussian (LQG) stochastic-control structure as a transport problem of the Path Integral Diffusion (PID) type. In classical LQG control, linear dynamics, Gaussian noise, and quadratic costs lead to Riccati equations and closed-form optimal feedback. In LQ-GM-PID, we retain the linear–quadratic stochastic-control backbone, but replace terminal state regulation by a prescribed terminal probability density and allow both the initial and terminal laws to be Gaussian Mixtures (GM). Moreover, LQ-GM-PID turns bridge diffusion from a tool for terminal target matching alone into a tool for path shaping. We demonstrate this on a 2D corridor task, a 2D multi-entrance transport task, and a high-dimensional scaling study with d=32 and M=16 Gaussian-mixture terminal modes, all with sub-50,ms analytic precompute on a laptop. We position LQ-GM-PID as an analytically solvable reference model for the state-of-the-art neural bridge-diffusion and generative-transport methods: a controlled setting in which neural approximations, score estimates, path-shaping objectives, and protocol-learning procedures can be tested against exact quantities. Comments: 47 pages, 18 figures Subjects: Machine Learning (cs.LG); Statistical Mechanics (cond-mat.stat-mech); Artificial Intelligence (cs.AI); Systems and Control (eess.SY); Optimization and Control (math.OC) Cite as: arXiv:2605.02961 [cs.LG] (or arXiv:2605.02961v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2605.02961 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[AI-99] Kernel Affine Hull Machines for Compute-Efficient Query-Side Semantic Encoding

【速读】：该论文旨在解决基于Transformer的语义检索系统中在线查询编码（online query encoding）成为主要计算瓶颈的问题，尤其是在固定教师模型（fixed-teacher）场景下，如何在不损害决策相关检索质量的前提下，用轻量级、解析显式估计器替代重复的神经网络推理。其解决方案的关键在于提出Kernel Affine Hull Machines (KAHMs)，该方法通过估计在严格定义的再生核希尔伯特空间（RKHS）中原型混合权重，并利用归一化最小均方算法（normalized least-mean-squares）优化原型，将廉价词汇特征映射到冻结的语义嵌入空间，从而实现编码误差的透明分解（包括后验近似误差、泛化误差和教师噪声分量）。实验表明，KAHM在奥地利法律基准上实现了优于同类学习适配器的教师空间重建性能，并显著提升排序敏感指标（如MRR@20、Hit@20、Top-1准确率），同时将每查询延迟降低8.5倍，验证了轻量几何估计器在保持检索效果的同时大幅提升效率与可解释性。

链接: https://arxiv.org/abs/2605.02950
作者: Mohit Kumar,Somayeh Kargaran,Bernhard A. Moser,Manuela Geiß
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Transformer-based semantic retrieval is highly effective, yet in many deployments the dominant cost lies in online query encoding rather than corpus indexing. We study the fixed-teacher query-adaptation problem and ask whether repeated neural inference can be replaced by a lightweight, analytically explicit estimator without degrading decision-relevant retrieval quality. We propose Kernel Affine Hull Machines (KAHMs), which map inexpensive lexical features into a frozen semantic embedding space by estimating prototype-mixture weights in a rigorously specified RKHS and refining prototypes via normalized least-mean-squares, yielding a transparent decomposition of encoding error into posterior-approximation, generalization, and teacher-noise components. On a controlled Austrian-law benchmark (5,000 queries; 84 laws; 10,762 units), KAHM attains the strongest teacher-space reconstruction among matched learned adapters (MSE 0.000091, R^2 0.9071, cosine 0.9536) and consistently leads rank-sensitive metrics, including mean reciprocal rank at 20 (MRR@20, the average inverse rank of the first relevant result within the top 20), Hit rate at 20 (Hit@20, the fraction of queries with at least one relevant result in the top 20), and Top-1 accuracy (the fraction of queries whose correct item is ranked first), with scores of 0.504, 0.694, and 0.411, respectively. It also reduces per-query latency by a factor of 8.5 relative to direct transformer encoding. These results demonstrate that, in fixed-teacher regimes, lightweight geometric estimators can substitute for online neural encoding, preserving retrieval performance while substantially improving efficiency and interpretability.

[AI-100] AsymK-Talker: Real-Time and Long-Horizon Talking Head Generation via Asymmetric Kernel Distillation

【速读】：该论文旨在解决当前基于扩散模型的音频驱动人脸生成方法在实时应用中面临的三大挑战：因果效率低下导致无法实现实时推理、无法兼容时间一致性的条件输入，以及长时间生成时出现的渐进性漂移问题。解决方案的关键在于提出AsymK-Talker框架，其核心创新包括：（1）核条件循环生成（Kernel-Conditioned Loop Generation, KCLG），通过运动核（motion kernel）实现因果性与时间一致性传播；（2）时间参考编码（Temporal Reference Encoding, TRE），将静态身份参考转换为时序感知的潜在表示以提升音视频同步性能；（3）非对称核蒸馏（Asymmetric Kernel Distillation, AKD），采用教师-学生蒸馏机制，其中教师模型利用真实运动核监督，学生模型则从生成核中学习，从而保障长序列生成的鲁棒性。

链接: https://arxiv.org/abs/2605.02948
作者: Yuxin Lu,Qian Qiao,Jiayang Sun,Min Cao,Guibo Zhu
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Sound (cs.SD)
备注:

点击查看摘要

Abstract:Recent advances in diffusion models have markedly enhanced the visual fidelity of audio-driven talking head generation. Nevertheless, existing methods are constrained by three critical limitations: causal inefficiency that impedes real-time inference, incompatibility with temporally coherent conditioning, and progressive drift over long-horizon generation, collectively hindering their deployment in real-time applications. To overcome these challenges, we introduce AsymK-Talker, a novel diffusion-distillation method designed for real-time and long-horizon talking head generation. AsymK-Talker comprises three key components: (1) Kernel-Conditioned Loop Generation (KCLG), a causal, chunk-wise generation paradigm that leverages motion kernels to enable temporally consistent propagation; (2) Temporal Reference Encoding (TRE), which converts a static identity reference into a time-aware latent representation to enhance audio-visual synchronization; and (3) Asymmetric Kernel Distillation (AKD), a teacher-student distillation framework wherein the teacher model conditions on ground-truth motion kernels for supervision, while the student learns to generate from generated kernels, thereby ensuring robustness during extended generation sequences. AsymK-Talker achieves promising results on both visual fidelity and lip synchronization metrics.

[AI-101] Predicting Euler Characteristics and Constructing Topological Structure Using Machine Learning Techniques

【速读】：该论文旨在解决从单个几何图像中直接提取拓扑性质（特别是欧拉示性数，Euler characteristic）的问题，而无需依赖大规模预训练数据集。传统方法通常需要大量标注数据或复杂的图像处理流程来推断拓扑特征，而本文提出了一种基于神经网络的新方法：首先将输入图像转化为一个单位矢量场（类比自旋配置），进而通过计算该配置的斯格明子数（skyrmion number）来预测欧拉示性数。其关键创新在于，模型仅需一个简单的几何图像即可学习生成具有手性磁结构的自旋配置，并且无需真实的手性自旋标签；此外，为约束由物理自由度带来的非唯一性问题，研究引入了包含交换相互作用、Dzyaloshinskii-Moriya (DM) 相互作用和各向异性项的磁哈密顿量作为物理信息损失函数，从而提升生成自旋配置的准确性和一致性。

链接: https://arxiv.org/abs/2605.02947
作者: Gyunghun Yu(1),Seong Min Park(1),Han Gyu Yoon(1),Tae Jung Moon(1),Jun Woo Choi(2),Hee Young Kwon(2),Changyeon Won(1) ((1) Department of Physics, Kyung Hee University, Seoul, South Korea, (2) Center for Spintronics, Korea Institute of Science and Technology, Seoul, South Korea)
机构: 未知
类目: Machine Learning (cs.LG); Materials Science (cond-mat.mtrl-sci); Artificial Intelligence (cs.AI); Computational Physics (physics.comp-ph)
备注: Corresponding authors: Hee Young Kwon and Changyeon Won

点击查看摘要

Abstract:This study proposes a novel approach to extract topological properties, specifically the Euler characteristic, from input images using neural networks without relying on large pre-existing datasets but with a single geometric image. Inspired by solid-state physics, where topological properties of magnetic structures are derived from spin field analysis, our model generates a unit vector field from an image, interpreted as a spin configuration. The Euler characteristic is then predicted by computing the skyrmion number of this generated spin configuration. Remarkably, the network learns to construct chiral magnetic textures without access to ground-truth chiral spin configurations, relying instead on only a single, simple geometric image and the straightforward skyrmion number computation. Furthermore, spin configurations generated by independently trained networks can be non-unique due to inherent degrees of freedom. To constrain these degrees of freedom and further refine the spin configuration, we incorporate a magnetic Hamiltonian, comprising exchange interaction, Dzyaloshinskii-Moriya (DM) interaction, and anisotropy, as an additional, physics-informed loss function. We validate the model’s efficacy on complex geometrical shapes and demonstrate its applicability to practical tasks.

[AI-102] RouteHijack: Routing-Aware Attack on Mixture-of-Experts LLM s

【速读】：该论文旨在解决大规模稀疏专家模型（Mixture-of-Experts, MoE）在安全对齐方面的脆弱性问题，特别是现有对抗攻击方法在MoE架构中存在局限：基于提示的越狱攻击依赖启发式搜索且迁移能力差，模型干预方法需访问内部表示，而基于优化的输入攻击因路由机制不可微而难以有效作用于MoE结构。其解决方案的关键在于提出RouteHijack——一种面向路由机制的越狱攻击方法，核心洞察是安全行为集中于少数专家，可通过输入优化影响路由决策来操控模型输出。具体而言，该方法首先通过响应驱动的专家定位识别出安全关键和有害专家，再构建具有路由感知目标的对抗后缀，以抑制安全专家、促进有害专家并防止生成早期拒绝；该方法仅需输入访问即可实现高成功率攻击，在7个MoE语言模型上平均攻击成功率达69.3%，显著优于现有优化方法，并具备跨模型零样本迁移能力和对视觉-语言MoE模型的泛化效果，揭示了稀疏专家架构在安全防御上的根本性漏洞。

链接: https://arxiv.org/abs/2605.02946
作者: Zhiyuan Xu,Joseph Gardiner,Sana Belguith,Lichao Wu
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Safety alignment is critical for the responsible deployment of large language models (LLMs). As Mixture-of-Experts (MoE) architectures are increasingly adopted to scale model capacity, understanding their safety robustness becomes essential. Existing adversarial attacks, however, have notable limitations. Prompt-based jailbreaks rely on heuristic search and transfer poorly, model intervention methods require privileged access to internal representations, and optimization-based input attacks remain output-centric and are fundamentally limited to MoE models due to the non-differentiable routing mechanism. In this paper, we present RouteHijack, a routing-aware jailbreak for MoE LLMs. Our key insight is that safety behavior is concentrated in a small subset of experts, creating an opportunity to steer model behavior by influencing routing decisions through input optimization. Building on this observation, RouteHijack first performs response-driven expert localization to identify safety-critical and harmful experts by contrasting activations under safe refusals and harmful completions. It then constructs adversarial suffixes with a routing-aware objective that suppresses safety experts, promotes harmful experts, and prevents early-stage refusal during generation. At inference time, the optimized suffix is appended to a malicious prompt, requiring only input access. Across seven MoE LLMs, RouteHijack achieves a 69.3% average attack success rate (ASR), outperforming prior optimization-based attack by 3.2\times . RouteHijack also transfers zero-shot across five sibling MoE variants, raising average ASR from 27.7% to 61.2%, and further generalizes to three MoE-based VLMs, increasing average ASR from 2.47% to 38.7%. These findings expose a fundamental vulnerability in sparse expert architectures and highlight the need for defenses beyond output-level alignment. Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI) Cite as: arXiv:2605.02946 [cs.LG] (or arXiv:2605.02946v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2605.02946 Focus to learn more arXiv-issued DOI via DataCite

[AI-103] Exploring Pass-Rate Reward in Reinforcement Learning for Code Generation

【速读】：该论文旨在解决强化学习（Reinforcement Learning, RL）在代码生成任务中因二值奖励（pass-all-tests binary reward）稀疏而导致的学习信号不足问题，尤其是在无解或难解问题中无法有效优化模型。现有方法常采用测试用例通过率（test-case pass rate）作为替代奖励，以缓解奖励稀疏性。然而，论文通过严谨的控制实验发现，尽管pass-rate奖励更密集，其优化方向并不稳定地引导模型向完全正确（full-pass）解决方案收敛；这是因为测试用例通过率作为进展的代理指标存在校准偏差，且同一组部分通过的解可能引发相互抵消的梯度方向，从而削弱整体优化效果。解决方案的关键在于：设计更贴合“完全正确”目标的奖励机制，而非简单依赖密度更高的代理奖励，以确保梯度更新能一致地将概率质量推向完整通过所有测试的解。

链接: https://arxiv.org/abs/2605.02944
作者: Xin-Ye Li,Ren-Biao Liu,Yun-Ji Zhang,Hui Sun,Zheng Xie,Ming Li
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注: 16 pages, 6 figures

点击查看摘要

Abstract:Reinforcement learning (RL) from unit-test feedback has become a standard post-training recipe for improving large language models (LLMs) on code generation. However, the pass-all-tests binary reward can be sparse, yielding no learning signal on challenging problems where none of the sampled solutions passes all tests. A common remedy is to use the test-case pass rate as a surrogate reward. In this work, we study pass-rate rewards in critic-free RL for code generation (e.g., GRPO and RLOO) and report a consistent pattern across base models and algorithms: despite alleviating reward sparsity, pass-rate rewards do not reliably improve final performance over binary rewards in rigorous controlled experiments. To understand this discrepancy, we analyze reward density and the resulting gradient directions. We find that pass-rate rewards are denser, but the induced gradient updates do not consistently move probability mass toward full-pass solutions. This arises because test-case pass rate is a miscalibrated surrogate for progress toward full correctness, and partial-pass solutions within the same group can induce conflicting gradient directions that cancel out. Overall, our results suggest that, in critic-free RL, pass-rate rewards are insufficient to improve code generation and motivate reward designs that better align optimization with the goal of full correctness. Comments: 16 pages, 6 figures Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Software Engineering (cs.SE) Cite as: arXiv:2605.02944 [cs.LG] (or arXiv:2605.02944v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2605.02944 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Xin-Ye Li [view email] [v1] Fri, 1 May 2026 10:26:43 UTC (367 KB)

[AI-104] Healthcare AI GYM for Medical Agents

【速读】：该论文旨在解决医疗领域中多轮交互式强化学习（Reinforcement Learning, RL）训练过程中存在的效率低、稳定性差及工具使用退化等问题，尤其是在构建统一的临床训练环境以培养可泛化的医疗AI代理（Medical AI Agent）时面临的挑战。其核心问题是：当前基于稀疏终端奖励的多轮RL框架在临床决策任务中易导致对话结构坍缩为冗长单轮陈述，并伴随工具调用频率下降与训练不稳定现象。解决方案的关键在于提出一种Turn-level Truncated On-Policy Distillation (TT-OPD) 自蒸馏机制，其中采用无梯度更新的指数移动平均（EMA）教师模型，利用结果优先信息（outcome-privileged information）在每一轮对话中提供密集且结果感知的KL正则化，从而实现更稳定、高效、可控的多轮交互训练，显著提升最终性能并维持持续的多工具协同使用能力。

链接: https://arxiv.org/abs/2605.02943
作者: Minbyul Jeong
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Clinical reasoning demands multi-step interactions – gathering patient history, ordering tests, interpreting results, and making safe treatment decisions – yet a unified training environment provides the breadth of clinical domains and specialized tools to train generalizable medical AI agents through reinforcement learning remains elusive. We present a comprehensive empirical study of multi-turn agentic RL for medical AI, built on \gym, a gymnasium-compatible environment spanning 10 clinical domains with 3.6K+ tasks, 135 domain-specific tools, and a knowledge base of 828K medical passages. Our analysis reveals that agentic multi-turn structure degrades into verbose single-turn monologues, characterized by monotonic length explosion and a simultaneous erosion of tool-use frequency. We characterize how this collapse, alongside distillation instability, stems from the misalignment of sparse terminal rewards with sequential clinical trajectories. We find that vanilla GRPO achieves strong final accuracy on some benchmarks but suffers from training instability, evidenced by significant oscillations in response length and prolonged convergence periods. To improve training efficiency and stability, we propose Turn-level Truncated On-Policy Distillation (TT-OPD), a self-distillation framework where a gradient-free EMA teacher leverages outcome-privileged information to provide dense, outcome-aware KL regularization at every conversation turn. TT-OPD achieves the best performance on 10 of 18 benchmarks with an average +3.9~pp improvement over the non-RL baseline with faster early convergence, controlled response length, and sustained multi-turn tool use.

[AI-105] PrismAgent : Illuminating Harm in Memes via a Zero-Shot Interpretable Multi-Agent Framework

【速读】：该论文旨在解决有害内容检测中依赖大量标注数据导致训练成本高且泛化能力有限的问题。其解决方案的关键在于提出一种零样本、多智能体、可解释的框架 PrismAgent，该框架将检测任务类比为刑事案件调查过程，通过分析（analyst）、调查（investigator）、起诉（prosecutor）和判决（judge）四个阶段的协作推理链，实现对图像-文本混合的模因（meme）内容的意图探测与判别。其中，每个阶段均基于未标注数据生成上下文证据并进行独立判断，最终由裁判智能体综合所有中间结果做出决策，从而在无需标注数据的情况下显著提升检测性能，并提供透明、可追溯的推理路径。

链接: https://arxiv.org/abs/2605.02940
作者: Zihan Ding,Ziyuan Yang,Yi Zhang
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:The rapid spread of memes makes harmful content detection increasingly crucial, as effective identification can curb the circulation of misinformation. However, existing methods rely heavily on high-volume annotated data, which leads to substantial training costs and limited generalization. To address these challenges, we propose PrismAgent, a zero-shot, multi-agent, interpretable framework. PrismAgent conceptualizes this task as a criminal case investigation, employing four specialized agents responsible for the analysis, investigation, prosecution, and judgment stages within a structured collaborative workflow. In the first stage, the analyst agent paraphrases each meme under benevolent and malicious assumptions to probe its underlying intent. The investigator agent then retrieves supporting evidence from an unannotated dataset and constructs contextual interpretations for the meme and its variants. Next, the prosecutor agent performs three independent preliminary judgments by pairing the original meme with each of the three interpretations. Finally, the judge agent deliberates across all evidence to render a final verdict. Moreover, PrismAgent’s explicit multi-stage reasoning chain makes the model inherently interpretable, as every intermediate step is explicitly explained rather than only producing a final detection result. Extensive experiments on three public datasets show that PrismAgent significantly outperforms existing zero-shot detection methods.

[AI-106] From Static Analysis to Audience Dissemination: A Training-Free Multimodal Controversy Detection Multi-Agent Framework

【速读】：该论文旨在解决多模态争议检测（Multimodal Controversy Detection, MCD）中现有方法难以捕捉不同受众群体对视频内容及其评论的多样化解读与评价的问题。传统静态表示学习方法仅依赖于视频和评论的直接特征提取，忽略了内容在真实传播过程中因受众差异而产生的动态争议演化。其解决方案的关键在于提出一种无需训练的多智能体框架AuDisAgent，通过结构化的多智能体系统模拟观众传播过程：首先由三个专业化筛选代理（视频代理、评论代理和交互代理）从视觉、文本及跨模态角度进行初步评估；当三者无法达成一致时，激活观看委员会代理模拟具有多元背景和立场的观众讨论，从而揭示传播过程中潜在的争议内容；最终由仲裁代理基于完整推理链作出判断。此外，为应对新发布视频缺乏评论的冷启动问题，设计了基于语义相似视频历史评论的评论引导策略，显著提升了在丰富评论与有限评论场景下的检测性能。

链接: https://arxiv.org/abs/2605.02939
作者: Zihan Ding,Ziyuan Yang,Yi Zhang
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Multimodal controversy detection (MCD) identifies controversial content in videos and their associated user comments, to support risk management for social video this http URL research frames MCD as a static representation learning task, where features are directly extracted from videos and their accompanying comments. However, these methods fail to capture the diverse perspectives and evaluations from different audience groups. Inspired by the real-world process of content dissemination among audiences, we propose AuDisAgent, a training-free multi-agent framework that reformulates MCD as a dynamic propagation this http URL framework explicitly models audience dissemination through a structured multi-agent system. First, three specialized Screening Agents (Video Agent, Comment Agent, and Interaction Agent) conduct initial assessments from visual, textual, and cross-modal perspectives, respectively. For samples where the three agents cannot reach a consensus, a Viewing Panel Agent is activated to simulate post-screening discussions among audiences with diverse backgrounds and stances. This mechanism models how different audience groups interpret and react to the same content, uncovering latent controversial content that may emerge during the dissemination process. Finally, an Arbitration Agent renders the final judgment based on the complete reasoning chain from the preceding this http URL addition, to address the “cold-start” scenario where newly released videos have few or no comments, we design a Comment Bootstrapping Strategy that leverages historical public comments from semantically similar videos as the initial comment context. Extensive experiments on a public dataset demonstrate that our framework significantly outperforms existing state-of-the-art (SOTA) methods in both rich-comment and limited-comment scenarios.

[AI-107] PAMNet: Cycle-aware Phase-Amplitude Modulation Network for Multivariate Time Series Forecasting

【速读】：该论文旨在解决多变量时间序列预测中周期模式建模的两个关键问题：一是现有方法通常通过复杂模型架构（如Transformer）隐式提取周期性，导致计算开销高；二是显式建模周期分量时忽略了相位-幅度耦合（phase-amplitude coupling）这一内在特性。解决方案的关键在于提出一种新型的周期感知相位-幅度调制网络（Cycle-aware Phase-Amplitude Modulation Network, PAMNet），其核心创新是双分支调制器结构：相位分支利用循环嵌入捕捉相位相关的均值偏移，幅度分支建模强度变化以适应方差波动；两者通过轻量级元素级融合机制高效结合，实现对相位与幅度成分的显式解耦与交互建模，无需复杂的注意力机制即可显著提升预测精度。

链接: https://arxiv.org/abs/2605.02938
作者: Yingbo Zhou,Yutong Ye,Zhiwei Ling,Shuhao Li,Rui Qian,Jian Xiong,Li Sun,Dejing Dou
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Reliable periodic patterns serve as a fundamental basis for accurate multivariate time series forecasting. However, existing methods either implicitly extract periodicity through complex model architectures (e.g., Transformers) with high computational overhead or overlook the intrinsic phase-amplitude coupling when modeling periodic components explicitly. To address these issues, we propose a novel Cycle-aware Phase-Amplitude Modulation Network (PAMNet) that explicitly decomposes periodic patterns into complementary phase and amplitude components. The core innovation lies in its dual-branch modulator, featuring dedicated learnable embeddings for phase positioning and amplitude modulation. The phase branch employs cyclical embeddings to capture phase-dependent mean shifts, while the amplitude branch models intensity variations to adapt to changes in variance. A lightweight modulator with element-wise fusion efficiently combines these components, enabling explicit modeling of their interactions without complex attention mechanisms. Extensive experiments on twelve real-world datasets demonstrate that our method achieves state-of-the-art performance through its novel phase-amplitude decoupling mechanism, offering a new perspective for cyclical modeling in time series forecasting.

[AI-108] Proteo-R1: Reasoning Foundation Models for De Novo Protein Design

【速读】：该论文旨在解决当前深度学习在从头蛋白质设计中缺乏显式推理机制的问题，即现有模型直接合成分子几何结构，而未明确识别对功能至关重要的残基或相互作用，导致设计决策与连续采样动态纠缠，从而限制了可解释性、可控性和生化知识的系统性复用。其解决方案的关键在于提出Proteo-R1框架，采用双专家架构：一个多功能大语言模型（Multimodal Large Language Model, MLLM）作为“理解专家”，基于序列、结构和文本上下文识别关键功能残基；这些残基级决策以硬约束形式传递给独立的基于扩散模型的“生成专家”，后者在固定相互作用锚点条件下进行条件协同设计。这种显式分离分子理解与几何生成的方式，使设计过程更接近人类专家的工程逻辑，实现了稳定、可解释且模块化的LLM推理与先进几何生成模型的集成。

链接: https://arxiv.org/abs/2605.02937
作者: Fang Wu,Weihao Xuan,Heli Qi,Hanqun Cao,Heng-Jui Chang,Zeqi Zhou,Haokai Zhao,Ma Jian,Carl Ma,Yu-Chi Cheng,Kuan Pang,Xiangru Tang,Zehong Wang,Guanlue Li,Hanchen Wang,Kejun Ying,Pan Lu,Chiho Im,Seungju Han,Peng Xia,Tinson Xu,Yinxi Li,Deyao Zhu,Pheng-Ann Heng,Naoto Yokoya,Masashi Sugiyama,Li Erran Li,Jure Leskovec,Yejin Choi
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE)
备注:

点击查看摘要

Abstract:Deep learning in \emphde novo protein design has achieved atomic-level fidelity. However, existing models remain largely non-deliberative: they directly synthesize molecular geometries without explicitly reasoning about which residues or interactions are functionally essential. As a result, design decisions are entangled with continuous sampling dynamics, limiting interpretability, controllability, and systematic reuse of biochemical knowledge. We introduce \textbfProteo-R1, a reasoning-guided protein design framework that explicitly decouples \emphmolecular understanding from \emphgeometric generation. Proteo-R1 adopts a dual-expert architecture in which a multimodal large language model (MLLM) serves as an \emphunderstanding expert, analyzing protein sequences, structures, and textual context to identify key functional residues that govern binding and specificity. These residue-level decisions are then passed as hard constraints to a separate diffusion-based \emphgeneration expert, which performs conditional co-design while respecting the fixed interaction anchors. This factorization mirrors how human experts approach molecular engineering: first, reasoning about critical interactions, then optimizing geometry subject to those constraints. By operationalizing reasoning as explicit residue-level commitments rather than latent textual guidance, Proteo-R1 achieves stable, interpretable, and modular integration of LLM reasoning with state-of-the-art geometric generative models. Code, data, and demos are available at this https URL.

[AI-109] DeRelayL: Sustainable Decentralized Relay Learning

【速读】：该论文旨在解决大规模模型训练资源门槛高导致普通用户（如移动设备用户）难以参与和受益的问题，以及现有协同训练方法（如联邦学习）在用户所有权保障和可持续性方面的不足。其核心解决方案是提出一种名为去中心化中继学习（Decentralized Relay Learning, DeRelayL）的新范式，该范式允许无权限参与者以接力式方式贡献模型训练并共享模型，通过设计激励机制确保系统可持续运行，并结合理论分析与数值仿真验证其有效性。

链接: https://arxiv.org/abs/2605.02935
作者: Haihan Duan,Tengfei Ma,Yuyang Qin,Runhao Zeng,Wei Cai,Victor C. M. Leung,Xiping Hu
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 24 pages, 4 figures, Published in IEEE Transaction on Mobile Computing

点击查看摘要

Abstract:In the era of big data, large-scale machine learning models have revolutionized various fields, driving significant advancements. However, large-scale model training demands high financial and computational resources, which are only affordable by a few technological giants and well-funded institutions. In this case, common users like mobile users, the real creators of valuable data, are often excluded from fully benefiting due to the barriers, while the current methods for accessing large-scale models either limit user ownership or lack sustainability. This growing gap highlights the urgent need for a collaborative model training approach, allowing common users to train and share models. However, existing collaborative model training paradigms, especially federated learning (FL), primarily focus on data privacy and group-based model aggregation. To this end, this paper intends to address this issue by proposing a novel training paradigm named decentralized relay learning (DeRelayL), a sustainable learning system where permissionless participants can contribute to model training in a relay-like manner and share the model. In detail, this paper presents the architecture and workflow of DeRelayL, designs incentive mechanisms to ensure sustainability, and conducts theoretical analysis and numerical simulations to demonstrate its effectiveness.

[AI-110] Keyword spotting using convolutional neural network for speech recognition in Hindi

【速读】：该论文旨在解决印度语（Hindi）语音识别中关键词检测（Keyword Spotting, KWS）的效率与个性化问题，尤其针对设备端（on-device）实时处理需求。解决方案的关键在于利用卷积神经网络（Convolutional Neural Networks, CNNs）构建高效分类模型，并通过特征工程将原始音频转换为梅尔频率倒谱系数（Mel Frequency Cepstral Coefficients, MFCCs）作为输入特征，从而在保证计算效率的同时实现用户特定查询的高精度识别，实验表明该方法在40,000个音频样本上达到91.79%的准确率。

链接: https://arxiv.org/abs/2605.02928
作者: Saru Bharti,Pushparaj Mani Pathak
机构: 未知
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注: Published in 2024 15th International Conference on Computing Communication and Networking Technologies (ICCCNT)

点击查看摘要

Abstract:In this study, we investigate the application of keyword spotting (KWS) in the domain of Hindi speech recognition, utilizing a dataset comprising 40,000 audio samples. With a sampling rate of 44 kHz and an average duration of 1.9 seconds per sample, we focus on developing an efficient on-device KWS system tailored for user-specific queries. Leveraging Convolutional Neural Networks (CNNs) for classification, we employ feature engineering techniques to convert raw audio recordings into Mel Frequency Cepstral Coefficients (MFCCs) as an input for our network. Our experiments encompass various CNN architectures, exploring their efficacy in identifying predefined keywords within the continuous speech stream. Our CNN-based approach achieves a commendable accuracy rate of 91.79% through rigorous evaluation, demonstrating promising performance while ensuring computational efficiency and user-specific customization in Hindi speech recognition.

[AI-111] Generalization Bounds of Spiking Neural Networks via Rademacher Complexity

【速读】：该论文旨在解决脉冲神经网络（Spiking Neural Networks, SNNs）在未见数据上泛化性能的理论理解不足问题，尤其是针对不同积分-放电机制下的泛化边界刻画不清的问题。解决方案的关键在于利用Rademacher复杂度对多种积分-放电方案的SNN进行系统性理论分析，发现其经验Rademacher复杂度与网络配置密切相关：随网络深度呈指数增长，随接收脉冲序列的最大时间跨度呈超线性但亚二次增长，随网络宽度呈超线性但亚二次增长，随参数范数呈多项式依赖，随训练样本数呈反比例衰减，且与脉冲神经元内部计算无关，从而获得了比传统研究更精确的泛化界估计。

链接: https://arxiv.org/abs/2605.02927
作者: Shao-Qun Zhang,Zhi-Hua Zhou
机构: 未知
类目: Neural and Evolutionary Computing (cs.NE); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Spiking Neural Networks (SNNs) have garnered increasing attention as one of bio-inspired models due to their great potential in neuromorphic computing and sparse computation. Many practical algorithms and techniques have been developed; however, theoretical understandings of the generalization, that is, the extent to which SNNs perform well on unseen data, are far from clear. Recent advances disclosed an excitation-dependent and architecture-related generalization bound such that the Rademacher complexity of SNNs with stochastic firing can be upper bounded by an exponential function relative to the excitation probability and the architecture depth. In this paper, we theoretically investigate the generalization bounds of SNNs with several integration-and-fire schemes via Rademacher complexity. We recognize that the empirical Rademacher complexity of SNNs is close to the SNN configurations, which is exponential to the network depth and the maximum time duration of received spike sequences, superlinear and subquadratic to the network width, polynomial to the parameter norm, inverse-linear to the number of training samples, and independent of the computations within spiking neurons, achieving a more precise rate than conventional studies. Our theoretical results may support the scope of SNN theories and shed some insight into the development of SNNs.

[AI-112] EvoJail: Evolutionary Diverse Jailbreak Prompt Generation for Large Language Models

【速读】：该论文旨在解决当前自动越狱生成方法在应对不断更新的安全微调模型时适应性不足，以及生成提示多样性匮乏导致攻击模式单一的问题。解决方案的关键在于提出EvoJail框架，其将越狱提示生成形式化为多目标黑盒优化问题，并引入进化算法思想，在迭代进化过程中依据目标模型响应动态选择与变异候选提示，从而实现对不同模型版本的自适应能力；同时通过场感知指令融合构建多样化初始提示，并设计多层级基于大语言模型（LLM）的变异算子以提升提示结构层面的多样性，有效增强攻击模式的丰富性。

链接: https://arxiv.org/abs/2605.02921
作者: Rui Tang,Kaiyu Xu,Pengsen Cheng,Hao Ren,Haizhou Wang,Shuyu Jiang
机构: 未知
类目: Neural and Evolutionary Computing (cs.NE); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted for publication in Information Processing and Management

点击查看摘要

Abstract:As LLMs continue to shape real-world applications, automated jailbreak generation becomes essential to reveal safety weaknesses and guide model improvement. Existing automatic jailbreak generation methods have not yet fully considered two important aspects: adaptability to evolving safety-finetuned models, which affects their effectiveness on newer model versions, and diversity in generated prompts, which can cause narrow or repetitive attack patterns. To address these issues, we propose EvoJail, an instruction-fusion-driven evolutionary jailbreak generation framework that formalizes jailbreak prompt generation as a multi-objective black-box optimization problem and leverages the principles of evolutionary algorithms to search for jailbreak prompts that can adapt across different model versions and exhibit diverse attack patterns. Specifically, EvoJail integrates jailbreak prompt generation into an iterative evolutionary loop, where at each iteration candidate prompts are evaluated directly against the target model and then selected and varied based on the target model’s responses, enabling the generation process to continuously adapt to model updates. To enhance diversity, EvoJail introduces field-aware instruction fusion to construct diverse starting points and incorporates diversity-aware objectives into the evolutionary fitness function, guiding the search toward prompts with richer semantic variation, while further designing multi-level LLM-based mutation operators that modify prompt structures at different granularities to promote structural diversity throughout the evolutionary process. Results demonstrate that EvoJail has stronger adaptability and can achieve over 93% attack success rate and more than 5.6% improvement in diversity metrics over state-of-the-art methods.

[AI-113] Mitigating the reconstruction-detection trade-off in VAE-based unsupervised anomaly detection

【速读】：该论文旨在解决变分自编码器（Variational Autoencoder, VAE）在无监督异常检测中因模型选择导致的重建质量与检测性能之间的权衡问题。现有方法通常通过最小化正常样本的重构误差来确定超参数，但这种方法忽略了潜在空间约束对异常检测能力的影响。研究发现，具有受限潜在空间的β-VAE模型虽然能提升检测指标，却牺牲了重建质量。为缓解这一权衡，论文提出两种解决方案：一是β调度（beta-scheduling），动态调整潜在空间正则化强度；二是稀疏变分自编码器（Sparse VAE），通过引入稀疏性约束，在保持高重建质量的同时显著提升异常检测性能。其中，稀疏VAE被证明是更有效的策略。

链接: https://arxiv.org/abs/2605.02918
作者: Agathe Senellart(UPCité, INSERM, HeKA | U1346),Maëlys Solal(ARAMIS, ICM),Stéphanie Allassonnière(UPCité, INSERM, HeKA | U1346),Ninon Burgos(ARAMIS, ICM)
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Variational autoencoders are widely used for unsupervised anomaly detection. Model selection however remains an open-question: to remain fully unsupervised, hyperparameters are often chosen to minimize the reconstruction error on normal samples. In this paper, we reveal a trade-off between reconstruction quality and anomaly detection among \beta -VAE models. Models with constrained latent space reach higher detection metrics but lower reconstruction quality. We also assess the performance variability across random seeds and show it is linked to the distance between normal and abnormal latent distributions. From this analysis, we justify and investigate two methods to mitigate the reconstructiondetection tradeoff: beta-scheduling and the Sparse VAE. The latter especially shows an improvement in detection while maintaining high reconstruction quality.

[AI-114] PRISM-CTG: A Foundation Model for Cardiotocography Analysis with Multi-View SSL

【速读】：该论文旨在解决当前用于自动化胎儿心电图（CTG）分析的监督式深度学习模型受限于小规模标注数据集和有限患者队列的问题，导致大量具有生理学意义的临床记录未被充分利用。其解决方案的关键在于提出一种名为PRISM-CTG的生理感知表示学习框架，该框架基于自监督预训练策略，利用大规模未标注CTG数据学习可迁移的领域级表征。PRISM-CTG通过多视图自监督机制联合优化三个互补的预训练任务：随机投影引导的掩码信号重建、临床变量预测与特征分类，并引入专用任务标记和受控交叉注意力机制实现专业化表示学习与跨临床情境的信息交互。该方法将通常被忽视的患者元数据和领域知识转化为额外的监督目标，从而引导更具临床意义的表征学习，显著提升了在多个下游CTG任务中的性能与泛化能力。

链接: https://arxiv.org/abs/2605.02917
作者: Sheng Wong,Ravi Shankar,Beth Albert,Hao Fei,Lin Li,Imane Ben M’Barek,Manu Vatish,Gabriel Davis Jones
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Supervised deep learning models for automated CTG analysis are typically constrained by narrowly curated labelled datasets and limited patient cohorts, leaving substantial volumes of physiologically informative clinical recordings untapped. To address this limitation, we propose Physiology-aware Representation Learning via Integrated Self-supervision and Metadata for CTG (PRISM-CTG), a clinically grounded self-supervised foundation model (FM) for CTG that leverages large-scale unlabelled recordings to learn transferable domain-level representations. PRISM-CTG is pretrained using a multi-view self-supervised framework that jointly optimises 3 complementary pretext objectives: random-projected guided masked signal reconstruction, clinical variable prediction, and feature classification. Each objective is associated with a dedicated task-specific token, enabling specialised representation learning, while controlled cross-attention facilitates information exchange across clinical context. By reframing patient metadata and domain knowledge, which are often underutilised in conventional training as prediction targets, Prism-CTG transforms readily available clinical information into additional supervisory targets that guide clinically meaningful representation learning. Extensive experiments across 7 downstream CTG tasks in both antepartum and intrapartum domains demonstrated that PRISM-CTG consistently outperforms in-domain and SSL baselines. Notably, PRISM-CTG demonstrated strong generalisation under external validation on 2 datasets, while achieving comparable performance to studies trained on substantially larger, privately labelled datasets. To our knowledge, this is the first study to introduce large-scale FM for CTG that learns domain-level representations.

[AI-115] When Safety Geometry Collapses: Fine-Tuning Vulnerabilities in Agent ic Guard Models AAAI2026

【速读】：该论文旨在解决安全对齐模型（safety-aligned models）在仅使用良性数据进行领域专业化微调时，其安全性会因潜在表示几何结构的破坏而崩溃的问题。具体而言，研究发现即使没有对抗性攻击，标准的良性数据微调也会导致如LlamaGuard、WildGuard和Granite Guardian等安全分类器的拒绝率骤降、类间判别能力丧失（CKA降至零），从而引发严重安全隐患。解决方案的关键在于提出Fisher-Weighted Safety Subspace Regularization (FW-SSR)：该方法通过引入两个核心机制——(i) 基于对角Fisher信息矩阵的曲率感知方向权重，以及(ii) 随任务-安全梯度冲突动态调整的自适应正则化系数λₜ——在训练过程中主动增强安全子空间的几何清晰度，而非仅仅固定其位置。实验证明，FW-SSR显著恢复了Granite Guardian的拒绝能力（从0%提升至75%）并大幅降低WildGuard的攻击成功率（至3.6%），且结构化的表示几何指标（如CKA、Fisher得分）比绝对位移更可靠地预测安全行为，确立了基于几何监控在代理式AI部署中评估防护模型的必要性。

链接: https://arxiv.org/abs/2605.02914
作者: Ismail Hossain,Sai Puppala,Jannatul Ferdaus,Md Jahangir Alam,Yoonpyo Lee,Syed Bahauddin Alam,Sajedul Talukder
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
备注: This paper got accepted for AAAI 2026 Summer Symposium Series

点击查看摘要

Abstract:A guard model fine-tuned on entirely benign data can lose all safety alignment – not through adversarial manipulation, but through standard domain specialization. We demonstrate this failure across three purpose-built safety classifiers – LlamaGuard, WildGuard, and Granite Guardian – deployed as protection layers in agentic AI pipelines, and show that it originates in the destruction of latent safety geometry: the structured harmful – benign representational boundary that guides classification. We extract per-layer safety subspaces via SVD on class-conditional activation differences and track how this boundary evolves under benign fine-tuning. Granite Guardian undergoes complete collapse – refusal rate drops from 85% to 0%, CKA falls to zero, and 100% of outputs become ambiguous – a severity exceeding prior findings on general-purpose LLMs, explained by the specialization hypothesis: concentrated safety representations are efficient but catastrophically brittle. To mitigate this, we propose Fisher-Weighted Safety Subspace Regularization (FW-SSR), a training-time penalty combining (i) curvature-aware direction weights derived from diagonal Fisher information and (ii) an adaptive \lambda_t that scales with task-safety gradient conflict. FW-SSR recovers 75% refusal on Granite Guardian (CKA = 0.983) and reduces WildGuard’s Attack Success Rate to 3.6% – below the unmodified baseline – by actively sharpening the safety subspace rather than merely anchoring it. Across all three models, structural representational geometry (CKA, Fisher score) predicts safety behavior more reliably than absolute displacement metrics, establishing geometry-based monitoring as a necessary component of guard model evaluation in agentic deployments.

[AI-116] Delay Plateau or Collapse: Evaluating the Impact of Systematic Verification Error on RLVR

【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）在实际应用中因验证器存在系统性错误而导致模型学习到非预期一致行为的问题。传统分析将验证错误视为独立同分布的随机噪声，认为其仅减缓训练速度而不显著影响最终性能；但本文指出，现实中的验证器常表现出结构性错误（如系统性误判），这类错误会严重扭曲奖励信号，进而导致模型陷入次优收敛甚至性能崩溃。解决方案的关键在于认识到：验证器质量不能仅以样本层面的整体错误率衡量，而必须关注错误模式（error pattern）对奖励信号结构的影响，这为设计鲁棒的RLVR机制提供了新的理论依据和实践方向。

链接: https://arxiv.org/abs/2605.02909
作者: Kazuki Egashira,Mark Vero,Jasper Dekoninck,Florian E. Dorner,Robin Staab,Martin Vechev
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Reinforcement Learning with Verifiable Rewards (RLVR) has become a powerful approach for improving the reasoning capabilities of large language models (LLMs). While RLVR is designed for tasks with verifiable ground-truth answers, real-world verifiers (e.g., static code checkers) can introduce errors into the reward signal. Prior analyses have largely treated such errors as random and independent across samples, concluding that errors merely slow training with limited effect on final performance. However, practical verifiers tend to exhibit systematic errors. This introduces a risk of models learning unwanted consistent behavior from a structurally incorrect reward signal. In this work, we study the impact of such systematic verification errors on RLVR. Through controlled experiments on arithmetic tasks, we show that systematic false negatives lead to similar effects as random noise. On the other hand, systematic false positives can cause a wide range of behaviors from sub-optimal plateaus to performance collapse. Crucially, these outcomes are not determined by the overall error rate but by the specific pattern of introduced errors, making pre-hoc mitigation difficult. Our results show that, in contrast to prior conclusions, realistic verification errors can critically shape RLVR outcomes and that verifier quality has to be understood beyond its sample-level error rate.

[AI-117] On the Invariants of Softmax Attention

【速读】：该论文旨在揭示软最大注意力（Softmax attention）机制中隐含的数学结构与规律性，尤其是其注意力logit矩阵（即能量场，energy field）所表现出的不变性特征。传统研究多关注注意力权重的概率分布性质，而忽视了其底层结构。论文提出能量场的概念，并发现两类不变性：机制层面的不变性源于softmax注意力的代数结构，包括每行零和约束、由头维度决定的秩上限以及由此产生的谱特征；模型层面的规律则在所有测试的自回归语言模型中普遍存在，如能量场的方差均匀分布于键位置而非集中于少数位置，这归因于键矩阵的“键非相干性”（key incoherence）。解决方案的关键在于识别这些不变性及其理论基础，从而为模型训练提供可解释的监控指标（如每头的训练监控），并揭示注意力机制内在的低维结构特性。

链接: https://arxiv.org/abs/2605.02907
作者: Wonsuk Lee
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 15 pages, 3 figures

点击查看摘要

Abstract:Softmax attention maps every query–key interaction into a probability distribution, but the underlying structure remains largely unexplored. We define the \emphenergy field, the row-centered attention logit, and show that it exhibits invariant properties across models, architectures, and inputs. Two classes of invariants emerge. \emphMechanism-level invariants follow from the algebraic structure of softmax attention. They include a per-row zero-sum constraint, a rank bound determined by the head dimension, and spectral signatures that follow from them. \emphModel-level regularities are not required by the mechanism, yet hold in every autoregressive language model we test, spanning several architecture families. The energy field distributes its variance over key positions without concentrating at a few. This delocalization traces to a property of the key matrix we call \emphkey incoherence. These invariants have practical consequences. The rank bound confines the energy field to a low-dimensional subspace. Key incoherence yields a per-head training monitor. All results are verified at multiple context lengths and input texts. Comments: 15 pages, 3 figures Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI) MSC classes: 68T01 ACMclasses: I.2.0 Cite as: arXiv:2605.02907 [cs.LG] (or arXiv:2605.02907v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2605.02907 Focus to learn more arXiv-issued DOI via DataCite

[AI-118] From Packets to Patterns: Interpreting Encrypted Network Traffic as Longitudinal Behavioral Signals

【速读】：该论文旨在解决如何通过被动感知方式（即无需用户主动参与）在大规模场景下持续捕捉人类行为模式的问题，尤其关注睡眠障碍、压力和孤独感等心理状态的动态变化。其核心挑战在于从加密的智能手机网络流量中提取具有可解释性的行为特征，并区分个体间差异与个体内部随时间的变化。解决方案的关键在于采用基于Transformer架构并引入每用户适配器（per-user adapters）的模型结构，以同时建模个体典型行为及其偏离；并通过稀疏自编码器（sparse autoencoder）提取可解释的行为特征，结合广义估计方程（generalized estimating equations）与Mundlak分解方法，将个体间差异与个体内部时变效应分离，从而揭示不同心理状态所对应的独特时间结构。

链接: https://arxiv.org/abs/2605.01616
作者: Rameen Mahmood,Omar El Shahawy,Souptik Barua,Zachary Beattie,Jeffrey Kaye,Xuhai "Orson’’ Xu,Chao-Yi Wu,Danny Yuxing Huang
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Networking and Internet Architecture (cs.NI)
备注: 19 pages, 6 figures

点击查看摘要

Abstract:Human behavior is difficult to observe continuously at scale, yet it leaves measurable traces in everyday device use. We test whether encrypted smartphone network traffic – a ubiquitous, always-on, passive sensing modality – can passively capture behavioral patterns related to sleep, stress, and loneliness. We model shared behavioral structure using a transformer backbone with per-user adapters, allowing the model to represent both typical individual behavior and deviations from it. To make these representations interpretable, we apply a sparse autoencoder to extract behavioral features corresponding to distinct patterns of activity. We relate these features to sleep disturbance, stress, and loneliness using generalized estimating equations with Mundlak decomposition, separating between-person differences from within-person changes over time. We find that the three outcomes reflect distinct temporal structures: stress is primarily associated with stable between-person differences, loneliness with within-person variation, and sleep disturbance with a combination of both. Notably, these within-person dynamics are not captured by predefined network-traffic features, demonstrating the value of learned representations for longitudinal behavioral sensing. These results establish encrypted network traffic as a viable passive sensing modality, revealing interpretable behavioral dynamics – particularly deviations from an individual’s baseline – that are not visible in raw traffic features.

[AI-119] Steerable Adversarial Scenario Generation through Test-Time Preference Alignment ICLR2026

【速读】：该论文旨在解决现有对抗场景生成方法在安全性评估中因固定权衡对抗性与真实性而缺乏灵活性的问题，导致生成的场景无法在推理阶段进行细粒度调控，难以满足多样化训练与测试需求。其解决方案的关键在于将对抗场景生成重构为多目标偏好对齐问题，并提出名为SAGE（Steerable Adversarial scenario GEnerator）的新框架：通过分层分组偏好优化（hierarchical group-based preference optimization）实现数据高效的离线对齐，将硬性可行性约束与软性偏好解耦；并在推理时通过线性插值两个对立偏好专家模型的权重，构建连续策略谱，从而无需重新训练即可实现对抗性与真实性的可控平衡，理论依据来自线性模式连通性（linear mode connectivity）。

链接: https://arxiv.org/abs/2509.20102
作者: Tong Nie,Yuewen Mei,Yihong Tang,Junlin He,Jie Sun,Haotian Shi,Wei Ma,Jian Sun
机构: 未知
类目: Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注: ICLR 2026

点击查看摘要

Abstract:Adversarial scenario generation is a cost-effective approach for safety assessment of autonomous driving systems. However, existing methods are often constrained to a single, fixed trade-off between competing objectives such as adversariality and realism. This yields behavior-specific models that cannot be steered at inference time, lacking the efficiency and flexibility to generate tailored scenarios for diverse training and testing requirements. In view of this, we reframe the task of adversarial scenario generation as a multi-objective preference alignment problem and introduce a new framework named \textbfSteerable \textbfAdversarial scenario \textbfGEnerator (SAGE). SAGE enables fine-grained test-time control over the trade-off between adversariality and realism without any retraining. We first propose hierarchical group-based preference optimization, a data-efficient offline alignment method that learns to balance competing objectives by decoupling hard feasibility constraints from soft preferences. Instead of training a fixed model, SAGE fine-tunes two experts on opposing preferences and constructs a continuous spectrum of policies at inference time by linearly interpolating their weights. We provide theoretical justification for this framework through the lens of linear mode connectivity. Extensive experiments demonstrate that SAGE not only generates scenarios with a superior balance of adversariality and realism but also enables more effective closed-loop training of driving policies. Project page: this https URL.

[AI-120] Magic-Informed Quantum Architecture Search

【速读】：该论文旨在解决量子电路设计中对量子资源——非稳定化程度（nonstabilizerness，常称为“魔性”或magic）的可控性问题，即如何在通用电路设计框架下主动调控这一关键量子优势资源。解决方案的核心在于提出一种基于魔性的量子架构搜索（magic-informed quantum architecture search, QAS）技术，其关键创新是引入图神经网络（Graph Neural Network, GNN）作为启发式评估器，用于预测候选量子电路的魔性水平，并结合蒙特卡洛树搜索（Monte Carlo Tree Search, MCTS）形成魔性导向的搜索策略。该方法通过GNN诱导的魔性偏置，可灵活引导搜索过程进入高魔性或低魔性区域，从而实现对最终电路魔性水平的有效控制，且在不同规模和目标魔性水平的问题上均表现出显著效果，即使在GNN处理分布外样本时仍保持性能提升。

链接: https://arxiv.org/abs/2605.03932
作者: Vincenzo Lipardi,Domenica Dibenedetto,Georgios Stamoulis,Mark H.M. Winands
机构: 未知
类目: Quantum Physics (quant-ph); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Nonstabilizerness, commonly referred to as magic, is a fundamental resource underpinning quantum advantage. In this paper, we propose a magic-informed quantum architecture search (QAS) technique that enables control over a quantum resource within the general framework of circuit design. Inspired by the AlphaGo approach, we tackle the problem with a Monte Carlo Tree Search technique equipped with a Graph Neural Network (GNN) that estimates the magic of candidate quantum circuits. The GNN model induces a magic-based bias that steers the search toward either high- or low-magic regimes, depending on the target objective. We benchmark the proposed magic-informed QAS technique on both the structured ground-state energy problem and on the more general quantum state approximation problem, spanning different sizes and target magic levels. Experimental results show that the proposed technique effectively influences the magic across the search tree and notably also on the resulting final circuit, even in regimes where the GNN operates on out-of-distribution instances. Although introducing a problem-agnostic magic bias could, in principle, constrain the search dynamics, we observe consistent improvements in solution quality across all problems tested.

[AI-121] Amortized Variational Inference for Joint Posterior and Predictive Distributions in Bayesian Uncertainty Quantification

【速读】：该论文旨在解决传统贝叶斯预测推断中因两阶段流程导致的计算成本过高问题，尤其是在高保真度模型（如偏微分方程描述的物理系统）中的应用瓶颈。其核心挑战在于：先近似参数后验分布，再通过蒙特卡洛采样传播至预测分布的顺序方法效率低下，难以满足在线推理需求。解决方案的关键在于提出一种变分贝叶斯框架，直接优化后验-预测分布（posterior-predictive distribution），并通过引入一个关于Kullback–Leibler散度的变分上界和基于矩的正则化项，联合学习参数后验与预测分布的变分近似。该方法采用摊销训练（amortized training）策略，将大部分计算负担转移至离线阶段，从而显著降低在线预测推理的计算开销，同时提升预测分布的准确性。

链接: https://arxiv.org/abs/2605.03710
作者: Nan Feng,Xun Huan
机构: 未知
类目: Machine Learning (stat.ML); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Computation (stat.CO); Methodology (stat.ME)
备注: Preprint 30 pages, 21 figures

点击查看摘要

Abstract:Bayesian predictive inference propagates parameter uncertainty to quantities of interest through the posterior-predictive distribution. In practice, this is typically performed using a two-stage procedure: first approximating the posterior distribution of model parameters, and then propagating posterior samples through the predictive model via Monte Carlo simulation. This sequential workflow can be computationally demanding, particularly for high-fidelity models such as those governed by partial differential equations. We propose a variational Bayesian framework that directly targets the posterior-predictive distribution and jointly learns variational approximations of both the posterior and the corresponding predictive distribution. The formulation introduces a variational upper bound on the Kullback–Leibler divergence together with moment-based regularization terms. The variational distributions are trained in an amortized manner, shifting computational effort to an offline stage and enabling efficient online inference. Numerical experiments ranging from analytical benchmarks to a finite-element solid mechanics problem demonstrate that the proposed method achieves more accurate predictive distributions than conventional two-stage variational inference, while substantially reducing the cost of online predictive inference. Comments: Preprint 30 pages, 21 figures Subjects: Machine Learning (stat.ML); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Computation (stat.CO); Methodology (stat.ME) MSC classes: 62F15, 65N21, 68T07, 74S05 Cite as: arXiv:2605.03710 [stat.ML] (or arXiv:2605.03710v1 [stat.ML] for this version) https://doi.org/10.48550/arXiv.2605.03710 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[AI-122] Parametrizing Convex Sets Using Sublinear Neural Networks

【速读】：该论文旨在解决如何用神经网络高效、精确地参数化凸集（convex set）的问题，特别是在形状优化和逆向设计任务中实现对目标形状的准确重建。其解决方案的关键在于通过学习次线性函数（sublinear function，即正齐次且凸的函数）来隐式表示凸体的支持函数（support function）和规范函数（gauge function），从而构建一种新的凸集神经参数化方法。作者证明了该参数化具有通用逼近能力，并在实验中验证了其在复杂几何重构任务中的有效性。

链接: https://arxiv.org/abs/2605.03520
作者: Eloi Martinet
机构: 未知
类目: Optimization and Control (math.OC); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Numerical Analysis (math.NA)
备注:

点击查看摘要

Abstract:We propose a neural parameterization of convex sets by learning sublinear (positively homogeneous and convex) functions. Our networks implicitly represent both the support and gauge functions of a convex body. We prove a universal approximation theorem for convex sets under this parametrization. Empirically, we demonstrate the method on shape optimization and inverse design tasks, achieving accurate reconstruction of target shapes.

[AI-123] On the Spectral Structure and Objective Equivalence of Orthogonal Multilabel Fisher Discriminants

【速读】：该论文旨在解决多标签线性判别分析（Multilabel Linear Discriminant Analysis, ML-LDA）中判别子空间估计的理论完备性问题，特别是如何在同时处理多个标签时刻画其散度矩阵结构与统计收敛性质。解决方案的关键在于：首先通过建立多标签类间散度矩阵（multilabel between-class scatter matrix）的秩特性，揭示有效判别维度可严格超过经典单标签情形下的 $C-1$ 上界；其次，在 Stiefel 流形正交约束下统一四类 Fisher 目标函数的等价性，并量化其在非理想约束下的分歧；进一步地，提出一个双向标签距离保真边界，将投影空间中的欧氏距离与标签空间中的汉明距离关联起来。在此基础上，论文给出了在亚高斯噪声下子空间估计误差的有限样本上界 $O(k_{\max}\sqrt{d\log d/n/\text{gap}_r})$ ，并证明了该速率接近极小极大最优（匹配对数和 $k_{\max}$ 因子），从而为多标签判别子空间学习提供了严格的数学基础。

链接: https://arxiv.org/abs/2605.03283
作者: Brian Keith-Norambuena,Juan Bekios-Calfa
机构: 未知
类目: Machine Learning (stat.ML); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 50 pages, initial version submitted to JMLR

点击查看摘要

Abstract:We provide a unified theoretical analysis of Linear Discriminant Analysis with simultaneous multilabel scatter matrix formulations and Stiefel orthogonality constraints. Our contributions span both algebraic structure and statistical guarantees. On the algebraic side, we characterize the rank of the multilabel between-class scatter matrix, showing that the effective discriminant dimensionality can strictly exceed the classical single-label bound of C-1 ; we establish a multilabel partition of variance and prove that all four Fisher objectives are equivalent under the W^\top S_t^ML W = I_r constraint while characterizing their divergence under the Stiefel constraint; and we prove a two-sided label-distance preservation bound relating projected distances to Hamming distances in label space. On the statistical side, we establish a finite-sample O(k_\max\sqrtd\log d/n/gap_r) bound on the subspace estimation error under sub-Gaussian noise with a matching \Omega(\sigma^2 d/(n,gap_r)) minimax lower bound, establishing a near-minimax-optimal rate (matching up to logarithmic and k_\max factors) for multilabel discriminant subspace estimation. We further provide high-probability distance concentration, robustness guarantees under label interactions, and a regularization analysis preserving the spectral structure when d \gg n . All results are verified numerically on synthetic data generated from the linear label-effect model, covering both the algebraic identities and the multilabel-specific quantities ( k_\max , \kappa(S_t^ML) , |\Gamma/n|_2 , \Delta_r ) that govern the statistical bounds. The numerical experiments are designed as a sanity check for the theorems rather than as an empirical benchmark; evaluation on real multilabel datasets is left to future work targeting application-oriented venues.

[AI-124] Copula-Based Endogeneity Correction for Doubly Robust Estimation of Treatment Effect

【速读】：该论文旨在解决双重稳健（Doubly Robust, DR）估计在存在内生性（endogeneity）时导致的偏倚问题，特别是在医疗研究中，由于不可观测混杂因素（unobserved confounding）使得代理变量（如处方续方率）与误差项相关，从而破坏DR估计的一致性。解决方案的关键在于引入高斯Copula（Gaussian copula）来建模内生协变量与误差项的联合分布，从而校正内生性对处理效应估计的影响，且无需依赖工具变量（instrumental variables）。该方法在保持DR估计“只需正确设定处理模型或结果模型之一即可获得一致估计”的特性基础上，实现了对内生性问题的有效修正，模拟和实证结果均表明其能显著降低偏倚并恢复真实的因果效应。

链接: https://arxiv.org/abs/2605.03278
作者: Sahil Shikalgar,Md. Noor-E-Alam
机构: 未知
类目: Methodology (stat.ME); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Doubly Robust (DR) estimation of treatment effect relies on an untestable assumption that is the absence of unobserved confounding. This assumption is par- ticularly problematic in the context of healthcare research, where variables like pre- scription refill rates serve as proxies for unobserved behaviors such as medication adherence. These proxy variables are often endogenous, exhibiting correlation with the regression error term due to unmeasured confounding or measurement error. We propose a copula-corrected doubly robust estimator that addresses endogeneity in both the treatment and outcome models without requiring instrumental variables. Gaussian copulas model the joint distribution of endogenous covariates and the error term, enabling consistent estimation while preserving the doubly robust property that requires correct specification of either the treatment or outcome model, not both. Monte Carlo simulations demonstrate that naive DR estimation exhibits substantial bias under endogeneity, whereas our corrected estimator recovers unbiased treatment effects across different data-generating processes. We apply our method to examine the effect of nutritional counseling on blood pressure using the National Health and Nutrition Examination Survey (NHANES) data. Naive DR estimation suggests counseling is associated with increased blood pressure. After copula correction, this effect becomes statistically insignificant, consistent with literature showing modest effects of nutri- Counseling in reducing blood pressure. Our methodology provides researchers with a practical tool for obtaining treatment effects in the presence of endogeneity.

[AI-125] OptiLookUp: An Optical ROM-Based Loop up Table Engine for Photonic Accelerators

【速读】：该论文旨在解决光域只读存储器（Read-only Memory, ROM）在实现紧凑化、可重构性与低损耗之间的矛盾问题，特别是在集成微环谐振器（Microring Resonator, MRR）平台上如何实现高速、低功耗且可编程的光存储功能。其关键解决方案在于：通过将预定义的输入-输出映射直接编码到光子器件的光谱响应中，实现无需动态计算的确定性查找操作；同时采用分组式子阵列结构结合光学解码机制，有效降低累积插入损耗并提升可扩展性；并通过基于晶体管的光学选择器实现不同ROM模块的无物理光路重定向激活，从而实现灵活可重构的光ROM架构。

链接: https://arxiv.org/abs/2605.03241
作者: Ankur Singh,Akhilesh Jaiswal
机构: 未知
类目: Optics (physics.optics); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Read-only memory (ROM) provides deterministic access to predefined data mappings. Extending ROM concepts to the optical domain enables high-bandwidth, low-latency, and parallel memory access, but realizing compact and reconfigurable optical ROM remains challenging due to loss, wavelength control, and integration constraints. This work presents a high-speed, reconfigurable photonic ROM architecture implemented using integrated microring resonators (MRRs). The ROM encodes predefined input-output mappings directly in the spectral response of the photonic devices, enabling deterministic lookup-based operation without dynamic computation during readout. To improve scalability and reduce cumulative insertion loss, the architecture employs compact banked sub-arrays that are selectively addressed through an optical decoding mechanism. Reconfigurability is achieved using transistor-based optical selectors, allowing different ROM banks to be activated without physical light rerouting or interferometric structures. The proposed photonic ROM is designed and evaluated using device-level simulations based on the GlobalFoundries 45SPCLO silicon photonics platform. Simulation results demonstrate reliable operation at data rates up to 12.5 GHz, with stable light-to-current transfer characteristics obtained through integrated photodiode readout. The optical ROM can be used to implement nonlinear activation functions utilised in photonic accelerator architectures, including sigmoid, tanh, ReLU, and exponential mappings.

[AI-126] From Knowledge to Action: Outcomes of the 2025 Large Language Model (LLM ) Hackathon for Applications in Materials Science and Chemistry

【速读】：该论文旨在解决如何系统性地理解大型语言模型（Large Language Models, LLMs）在材料科学与化学研究生命周期中的应用模式及其演进趋势这一问题。其解决方案的关键在于提出一个双维度分类框架：一是“知识基础设施”（Knowledge Infrastructure），用于结构化、检索、合成与验证科学信息；二是“行动系统”（Action Systems），用于在计算与实验环境中执行、协调或自动化科研任务。通过分析社区开发的LLM应用项目，论文揭示了从单一功能工具向集成化多智能体工作流演进的趋势，核心要素包括基于检索增强生成（Retrieval-Augmented Generation, RAG）的底层支撑、持久化的结构化知识表示、多模态与多语言输入处理，以及实验室级闭环系统的初步实现，表明LLMs正逐步从通用助手转变为可组合的科学推理与行动基础设施。

链接: https://arxiv.org/abs/2605.03205
作者: Aritra Roy,Kevin Shen,Andrew MacBride,Awwal Oladipupo,Mudassra Taskeen,Wojtek Treyde,Ruaa A. E. A. Abakar,Ahmad D. Abbas,Elsayed Abdelfatah,Abbas A. Abdullahi,Seham S. Abyah,Chahd Rahyl Adjmi,Fariha Agbere,Savyasanchi Aggarwal,Muhammad Ahmed,Tasnim Ahmed,Motasem Ajlouni,Mattias Akke,Hussein AlAdwan,Anwaar S. Alazani,Zahra A. Alharbi,Wajd A. Aljulyhi,Mohammed A. AlKubaish,Fatima A. Almahri,Sayed A. Almohri,David Obeh Alobo,Mohammed Alouni,Azizah S. Alqahtani,Omar Alsaigh,Husain Althagafi,Md. Aqib Aman,Lena Ara,Arifin,Ignacio Arretche,Abdulaziz Ashy,Syeda A. Asim,Amro Aswad,Adeel Atta,Sören Auer,Abdullah al Azmi,Toheeb Balogun,Suvo Banik,Viktoriia Baibakova,Shakira A. Baksh,Neus G. Bastús,Christina J. Bayard,Adib Bazgir,Louis Beal,Lejla Biberić,Wahid Billah,Ankita Biswas,Joshua Bocarsly,Montassar T. Bouzidi,Esma B. Boydas,Youssef Briki,Cailin Buchanan,Mauricio Cafiero,Damien Caliste,Yi Cao,Rafael E. Castañeda,Sruthy K. Chandy,Benjamin Charmes,Shayantan Chaudhuri,Yiming Chen,Alexander Chen,Jieneng Chen,Min-Hsueh Chiu,Defne Circi,Cinthya H. Contreras,Yoann Cure,Nathan Daelman,Roshini Dantuluri,Thomas Davy,William Dawson,Leonid Didukh,Rui Ding,Aminu R. Doguwa,Claudia Draxl,Sathya Edamadaka,Oulaya Elargab,Christina Ertural,Matthew L. Evans,Edvin Fako,Hossam Farag,Nur A. Fathurrahman,Merve Fedai,Rodrigo P. Ferreira,Giuseppe Fisicaro,Thomas Frank,Sasi K. Gaddipati,Abhijeet Gangan,Jennifer Garland,James Garrick,Luigi Genovese,Maryam Ghadrdran,Sandip Giri,Maxime Goulet,Jeremy Goumaz,Sara U. Gracia,Jacob Graham
机构: 未知
类目: Materials Science (cond-mat.mtrl-sci); Artificial Intelligence (cs.AI)
备注: This paper reflects contributions from hundreds of researchers worldwide through an event, follow-on discussions, and project development exploring LLM applications in materials science and chemistry. While unconventional, it captures a timely, broad, and efficient community exploration of a rapidly evolving field and offers value to the arXiv community

点击查看摘要

Abstract:Large language models (LLMs) are rapidly changing how researchers in materials science and chemistry discover, organize, and act on scientific knowledge. This paper analyzes a broad set of community-developed LLM applications in an effort to identify emerging patterns in how these systems can be used across the scientific research lifecycle. We organize the projects into two complementary categories: Knowledge Infrastructure, systems that structure, retrieve, synthesize, and validate scientific information; and Action Systems, systems that execute, coordinate, or automate scientific work across computational and experimental environments. The submissions reveal a shift from single-purpose LLM tools toward integrated, multi-agent workflows that combine retrieval, reasoning, tool use, and domain-specific validation. Prominent themes include retrieval-augmented generation as grounding infrastructure, persistent structured knowledge representations, multimodal and multilingual scientific inputs, and early progress toward laboratory-integrated closed-loop systems. Together, these results suggest that LLMs are evolving from general-purpose assistants into composable infrastructure for scientific reasoning and action. This work provides a community snapshot of that transition and a practical taxonomy for understanding emerging LLM-enabled workflows in materials science and chemistry.

[AI-127] A Universal Space of Brain Dynamics for Unveiling Cognitive Transitions and Individual Differences

【速读】：该论文旨在解决如何构建一个适用于人类大脑活动的通用表示空间（universal space）这一难题，尤其针对不同认知状态和个体间差异带来的挑战。其核心问题是现有方法难以统一刻画脑功能动态的多样性与个体特异性，从而限制了对神经机制的精确建模与跨条件比较。解决方案的关键在于提出“通用脑动力学”（Universal Brain Dynamics, UBD），通过融合脑结构的空间属性（反映物理连接）与功能的时间属性（反映动态变化），利用模型推导的雅可比矩阵（Jacobian matrix）量化脑活动的动力学特征，并在人类连接组计划（HCP）数据中验证其跨8种状态、963名受试者的高预测精度（Pearson相关系数r > 0.9）。该方法实现了对脑活动的统一建模，为解析结构-功能耦合（structure-function coupling, SFC）、慢波动（infra-slow fluctuation, ISF）机制及个体差异提供了新视角与数值分析框架。

链接: https://arxiv.org/abs/2605.02936
作者: Ronghua Zheng,Chengyuan Qian,Weiyang Ding
机构: 未知
类目: Quantitative Methods (q-bio.QM); Artificial Intelligence (cs.AI)
备注:

点击查看摘要

Abstract:Representing dynamical systems through data-driven universal spaces has proven effective; however, achieving this universality for human brain activity remains a significant challenge, further aggravated by diverse cognitive states and individual subjects. Recognizing that spatial properties reflect physical wiring while temporal properties reflect brain function, we develop Universal Brain Dynamics (UBD) to construct a universal space tailored to brain activity and quantify corresponding dynamics using a model-derived Jacobian matrix. Crucially, we validate UBD’s universality by accurately predicting functional magnetic resonance imaging (fMRI) signals (Pearson’s r 0.9) across eight states and 963 subjects in the Human Connectome Project (HCP). Through evaluating resting-state fMRI represented within UBD, we gain insight into how infra-slow fluctuation (ISF) underpins brain activity. Furthermore, we reveal a new perspective on structure-function coupling (SFC) by analyzing the temporal sequence of brain dynamics. Extending UBD to task-evoked states, we derive brain dynamics across various cognitive conditions, elucidating the neural mechanisms driving cognitive transitions at a finer granularity. For individual differences, we compare brain dynamics across subjects to identify the neural underpinnings of these variations. Our findings suggest that synergistically integrating spatial and temporal properties of brain activity establishes a universal space for its unfolding, enabling the precise numerical analysis of underlying neural mechanisms across varying conditions.

机器学习

[LG-0] A Closed-Form Adaptive-Landmark Kernel for Certified Point-Cloud and Graph Classification

链接: https://arxiv.org/abs/2605.04046
作者: Sushovan Majhi,Atish Mitra,Žiga Virk,Pramita Bagchi
类目: Machine Learning (cs.LG); Algebraic Topology (math.AT)
*备注:

点击查看摘要

Abstract:We introduce PALACE (Persistence Adaptive-Landmark Analytic Classification Engine), the data-adaptive companion to PLACE, paying a small cross-validation tier on three knobs (budget, radii, bandwidth; \leq 5 choices each). A cover-theoretic core (Lebesgue-number criterion on the landmark cover) yields four closed-form guarantees. (i) A structural lower distortion bound \lambda(\tau;\nu) on \mathcalD_n under cross-diagram non-interference, with a (D/L)^2 budget reduction over the uniform grid when diagrams concentrate. (ii) Equal weights w_k = K^-1/2 maximizing \lambda , and farthest-point-sampling positions 2 -approximating the optimal k -center covering radius; both derived from training labels alone, no gradient training. (iii) A kernel-RKHS classification rate O((k-1)\sqrtK/(\gamma\sqrtm_\min)) with binary necessity threshold m = \Omega(\sqrt K/\gamma) from a matching Le Cam lower bound, and a closed-form filtration-selection rule. The kernel-Mahalanobis margin \hat\rho_\mathrmMah is the strongest closed-form ranker across the chemical-graph pool (mean Spearman \rho \approx +0.60 ); the isotropic surrogate \hat\gamma/\sqrtK admits a selection-consistency rate, and \widehat\lambda from (i) provides an independent data-level signal (positive on COX2 and PTC). (iv) A per-prediction certificate, in non-asymptotic Pinelis and asymptotic Gaussian forms, with no calibration split. Empirically, PALACE is the strongest closed-form diagram-based method on Orbit5k ( 91.3 \pm 1.0% , matching Persformer), leads every diagram-based competitor on COX2 and MUTAG, and is competitive on DHFR (within 1 pp of ECP). At 8\times domain inflation, adaptive placement maintains 94% while the uniform grid collapses to chance ( 25% on 4-class data).

[LG-1] Pretrained Model Representations as Acquisition Signals for Active Learning of MLIPs

链接: https://arxiv.org/abs/2605.03964
作者: Eszter Varga-Umbrich,Shikha Surana,Paul Duckworth,Jules Tilly,Olivier Peltre,Zachary Weller-Davies
类目: Machine Learning (cs.LG); Chemical Physics (physics.chem-ph)
*备注: 8 main pages, 28 total pages

点击查看摘要

Abstract:Training machine learning interatomic potentials (MLIPs) for reactive chemistry is often bottlenecked by the high cost of quantum chemical labels and the scarcity of transition state configurations in candidate pools. Active learning (AL) can mitigate these costs, but its effectiveness hinges on the acquisition rule. We investigate whether the latent space of a pretrained MLIP already contains the information necessary for effective acquisition, eliminating the need for auxiliary uncertainty heads, Bayesian training and fine-tuning, or committee ensembles. We introduce two acquisition signals derived directly from a pretrained MACE potential: a finite-width neural tangent kernel (NTK) and an activation kernel built from hidden latent space features. On reactive-chemistry benchmarks, both kernels consistently outperform fixed-descriptor baselines, committee disagreement, and random acquisition, reducing the data required to reach performance targets by an average of 38% for energy error and 28% for force error. We further show that the pretrained model induces similarity spaces that preserve chemically meaningful structure and provide more reliable residual uncertainty estimates than randomly initialised or fixed-descriptor-based kernels. Our results suggest that pretraining aligns latent-space geometry with model error, yielding a practical and sufficient acquisition signal for reactive MLIP fine-tuning.

[LG-2] Integrating Feature Correlation in Differential Privacy with Applications in DP-ERM AISTATS2026

链接: https://arxiv.org/abs/2605.03945
作者: Tianyu Wang,Luhao Zhang,Rachel Cummings
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: Appeared in AISTATS 2026

点击查看摘要

Abstract:Standard differential privacy imposes uniform privacy constraints across all features, overlooking the inherent distinction between sensitive and insensitive features in practice. In this paper, we introduce a relaxed definition of differential privacy that accounts for such heterogeneity, allowing certain features to be treated as insensitive even when correlated with sensitive ones. We propose a correlation-aware framework, \textsfCorrDP , which relaxes privacy for insensitive features while accounting for their correlations with sensitive features, with the correlations quantified using total variation distance. We design algorithms for differentially private empirical risk minimization (DP-ERM) under the \textsfCorrDP framework, incorporating distance-dependent noise into gradients for improved theoretical utility guarantees. When the correlation distance is unknown, we estimate it from the dataset and show that it achieves a comparable privacy-utility guarantee. We perform experiments on synthetic and real-world datasets and show that \textsfCorrDP -based DP-ERM algorithms consistently outperform the standard DP framework in the presence of insensitive features.

[LG-3] Optimal Posterior Sampling for Policy Identification in Tabular Markov Decision Processes AISTATS2026

链接: https://arxiv.org/abs/2605.03921
作者: Cyrille Kone,Kevin Jamieson
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: AISTATS 2026

点击查看摘要

Abstract:We study the (\varepsilon, \delta) -PAC policy identification problem in finite-horizon episodic Markov Decision Processes. Existing approaches provide finite-time guarantees for approximate settings ( \varepsilon0 ) but suffer from high computational cost, rendering them hard to implement, and also suffer from suboptimal dependence on \log(1/\delta) . We propose a randomized and computationally efficient algorithm for best policy identification that combines posterior sampling with an online learning algorithm to guide exploration in the MDP. Our method achieves asymptotic optimality in sample complexity, also in terms of posterior contraction rate, and runs in O(S^2AH) per episode, matching standard model-based approaches. Unlike prior algorithms such as MOCA and PEDEL, our guarantees remain meaningful in the asymptotic regime and avoid sub-optimal polynomial dependence on \log(1/\delta) . Our results provide both theoretical insights and practical tools for efficient policy identification in tabular MDPs.

[LG-4] Ecologically-Constrained Task Arithmetic for Multi-Taxa Bioacoustic Classifiers Without Shared Data

链接: https://arxiv.org/abs/2605.03914
作者: Ragib Amin Nihal,Benjamin Yen,Runwu Shi,Takeshi Ashizawa,Kazuhiro Nakadai
类目: ound (cs.SD); Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:Training data for bioacoustics is scattered across taxa, regions, and institutions. Centralizing it all is often infeasible. We show that independently fine-tuned BEATs encoders can be composed into a unified 661-species classifier via task vector arithmetic without sharing data. We find that bioacoustic task vectors are near-orthogonal (cosine 0.01-0.09). Their separation aligns closely with spectral distribution distance, a gradient consistent with the acoustic niche hypothesis. This geometry makes simple averaging optimal while sign-conflict methods reduce accuracy by one to six percentage points. Composition also creates an asymmetric gap: species-rich groups lose accuracy relative to joint training while underrepresented taxa gain, a redistribution useful for equitable biodiversity monitoring. We verify linear mode connectivity across all taxonomic pairs, demonstrate zero-shot transfer to new regions, and identify domain negation as a boundary condition where composition fails. These results enable a collaborative paradigm for bioacoustics where institutions share only task vectors to assemble multi-taxa classifiers, preserving data privacy.

[LG-5] From Data Lifting to Continuous Risk Estimation: A Process-Aware Pipeline for Predictive Monitoring of Clinical Pathways

链接: https://arxiv.org/abs/2605.03895
作者: Pasquale Ardimento,Mario Luca Bernardi,Marta Cimitile,Samuele Latorre
类目: Machine Learning (cs.LG); Software Engineering (cs.SE)
*备注:

点击查看摘要

Abstract:This paper presents a reproducible and process-aware pipeline for predictive monitoring of clinical pathways. The approach integrates data lifting, temporal reconstruction, event log construction, prefix-based representations, and predictive modeling to support continuous reasoning on partially observed patient trajectories, overcoming the limitations of traditional retrospective process mining. The framework is evaluated on COVID-19 clinical pathways using ICU admission as the prediction target, considering 4,479 patient cases and 46,804 prefixes. Predictive models are trained and evaluated using a case-level split, with 896 patients in the test set. Logistic Regression achieves the best performance (AUC 0.906, F1-score 0.835). A detailed prefix-based analysis shows that predictive performance improves progressively as new clinical events become available, with AUC increasing from 0.642 at early stages to 0.942 at later stages of the pathway. The results highlight two key findings: predictive signals emerge progressively along clinical pathways, and process-aware representations enable effective early risk estimation from evolving patient trajectories. Overall, the findings suggest that predictive monitoring in healthcare is best conceived as a continuous, dynamically aware process, in which risk estimates are progressively refined as the patient journey evolves.

[LG-6] On Adaptivity in Zeroth-Order Optimization

链接: https://arxiv.org/abs/2605.03869
作者: Hassan Dbouk,Nidham Gazagnadou,Matthias Reisser,Christos Louizos
类目: Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注:

点击查看摘要

Abstract:We investigate the effectiveness of adaptive zeroth-order (ZO) optimization for memory-constrained fine-tuning of large language models (LLMs). Contrary to prior claims, we show that adaptive ZO methods such as ZO-Adam offer no convergence advantage over well-tuned ZO-SGD, while incurring significant memory overhead. Our analysis reveals that in high dimensions, ZO gradients lack coordinate-wise heterogeneity, rendering adaptive mechanisms memory inefficient. Leveraging this insight, we propose MEAZO, a memory-efficient adaptive ZO optimizer that tracks only a single scalar for global step size adaptation. We support our method with theoretical convergence guarantees under standard assumptions. Experiments across multiple LLM families and tasks demonstrate that MEAZO matches ZO-Adam’s performance with the memory footprint of ZO-SGD. Additional experiments on synthetic quadratic problems and LLM fine-tuning further demonstrate MEAZO’s enhanced robustness to step size choices, particularly in grouped or block-structured optimization settings.

[LG-7] Memory-Efficient Continual Learning with CLIP Models

链接: https://arxiv.org/abs/2605.03866
作者: Ryan King,Gang Li,Bobak Mortazavi,Tianbao Yang
类目: Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:Contrastive Language-Image Pretraining (CLIP) models excel at understanding image-text relationships but struggle with adapting to new data without forgetting prior knowledge. To address this, models are typically fine-tuned using both new task data and a memory buffer of past tasks. However, CLIP’s contrastive loss suffers when the memory buffer is small, leading to performance degradation on previous tasks. We propose a memory-efficient, distributionally robust method that dynamically reweights losses per class during training. Our approach, tested on class incremental settings (CIFAR-100, ImageNet1K) and a domain incremental setting (DomainNet) adapts CLIP models quickly while minimizing catastrophic forgetting, even with minimal memory usage.

[LG-8] Complex Equation Learner: Rational Symbolic Regression with Gradient Descent in Complex Domain

链接: https://arxiv.org/abs/2605.03841
作者: Sergei Garmaev,Maurice Gauché,Olga Fink
类目: Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:Symbolic regression aims to discover interpretable equations from data, yet modern gradient-based methods fail for operators that introduce singularities or domain constraints, including division, logarithms, and square roots. As a result, Equation Learner-type models typically avoid these operators or impose restrictions, e.g. constraining denominators to prevent poles, which narrows the hypothesis class. We propose a complex weight extension of the Equation Learner that mitigates real-valued optimization pathologies by allowing optimization trajectories to bypass real-axis degeneracies. The proposed approach converges stably even when the target expression has real-domain poles, and it enables unconstrained use of operations such as logarithm and square root. We Validate the method on symbolic regression benchmarks and show it can recover singular behavior from experimental frequency response data.

[LG-9] On Computing Total Variation Distance Between Mixtures of Product Distributions

链接: https://arxiv.org/abs/2605.03839
作者: Weiming Feng,Yucheng Fu,Minji Yang,Anqi Zhang
类目: Data Structures and Algorithms (cs.DS); Machine Learning (cs.LG); Probability (math.PR)
*备注:

点击查看摘要

Abstract:We study the problem of approximating the total variation distance between two mixtures of product distributions over an n -dimensional discrete domain. Given two mixtures \mathbbP and \mathbbQ with k_1 and k_2 product distributions over [q]^n , respectively, we give a randomized algorithm that approximates d_\mathrmTV\left(\mathbbP,\mathbbQ\right) within a multiplicative error of (1\pm \varepsilon) in time \mathrmpoly((nq)^k_1+k_2,1/\varepsilon) . We also study the special case of mixtures of Boolean subcubes over \0,1^n . For this class, we give a deterministic algorithm that exactly computes the total variation distance in time \mathrmpoly(n,2^O(k_1+k_2)) , and show that exact computation is #\mathsfP -hard when k_1+k_2=\Theta(n) .

[LG-10] A Domain Incremental Continual Learning Benchmark for ICU Time Series Model Transportability

链接: https://arxiv.org/abs/2605.03832
作者: Ryan King,Conrad Krueger,Ethan Veselka,Tianbao Yang,Bobak J. Mortazavi
类目: Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:In recent years, machine learning has made significant progress in clinical outcome prediction, demonstrating increasingly accurate results. However, the substantial resources required for hospitals to train these models, such as data collection, labeling, and computational power, limit the feasibility for smaller hospitals to develop their own models. An alternative approach involves transferring a machine learning model trained by a large hospital to smaller hospitals, allowing them to fine-tune the model on their specific patient data. However, these models are often trained and validated on data from a single hospital, raising concerns about their generalizability to new data. Our research shows that there are notable differences in measurement distributions and frequencies across various regions in the United States. To address this, we propose a benchmark that tests a machine learning model’s ability to transfer from a source domain to different regions across the country. This benchmark assesses a model’s capacity to learn meaningful information about each new domain while retaining key features from the original domain. Using this benchmark, we frame the transfer of a machine learning model from one region to another as a domain incremental learning problem. While the task of patient outcome prediction remains the same, the input data distribution varies, necessitating a model that can effectively manage these shifts. We evaluate two popular domain incremental learning methods: data replay, which stores examples from previous data sources for fine-tuning on the current source, and Elastic Weight Consolidation (EWC), a model parameter regularization method that maintains features important for both data sources. Subjects: Machine Learning (cs.LG) Cite as: arXiv:2605.03832 [cs.LG] (or arXiv:2605.03832v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2605.03832 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[LG-11] Realizable Bayes-Consistency for General Metric Losses ICML2026

链接: https://arxiv.org/abs/2605.03823
作者: Dan Tsir Cohen,Steve Hanneke,Aryeh Kontorovich
类目: Machine Learning (cs.LG); Information Theory (cs.IT); Statistics Theory (math.ST)
*备注: 14 pages. To appear in Proceedings of the 43rd International Conference on Machine Learning (ICML 2026)

点击查看摘要

Abstract:We study strong universal Bayes-consistency in the realizable setting for learning with general metric losses, extending classical characterizations beyond 0 - 1 classification \citepbousquet_theory_2021, hanneke2021universalbayesconsistencymetric and real-valued regression \citepattias_universal_2024. Given an instance space (\mathcal X,\rho) , a label space (\mathcal Y,\ell) with possibly unbounded loss, and a hypothesis class \mathcal H \subseteq \mathcal Y^\mathcal X , we resolve the realizable case of an open problem presented in \citetpmlr-v178-cohen22a. Specifically, we find the necessary and sufficient conditions on the hypothesis class \mathcal H under which there exists a distribution-free learning rule whose risk converges almost surely to the best-in-class risk (which is zero) for every realizable data-generating distribution. Our main contribution is this sharp characterization in terms of a combinatorial obstruction: Similarly to \citetattias2024optimallearnersrealizableregression, we introduce the notion of an infinite non-decreasing (\gamma_k) -Littlestone tree, where \gamma_k \to \infty . This extends the Littlestone tree structure used in \citetbousquet_theory_2021 to the metric loss setting.

[LG-12] Graph Convolutional Support Vector Regression for Robust Spatiotemporal Forecasting of Urban Air Pollution

链接: https://arxiv.org/abs/2605.03795
作者: Nourin Jahan,Madhurima Panja,Muhammed Navas T,Tanujit Chakraborty
类目: Machine Learning (cs.LG); Applications (stat.AP); Machine Learning (stat.ML)
*备注:

点击查看摘要

Abstract:Urban air quality forecasting is challenging because pollutant concentrations are nonlinear, nonstationary, spatiotemporally dependent, and often affected by anomalous observations caused by traffic congestion, industrial emissions, and seasonal meteorological variability. This study proposes a Graph Convolutional Support Vector Regression (GCSVR) framework for robust spatiotemporal forecasting of urban air pollution. The model combines graph convolutional learning to capture inter-station spatial dependence with support vector regression to model nonlinear temporal dynamics while reducing sensitivity to outlier observations. The proposed framework is evaluated using air quality records from 37 monitoring stations in Delhi and 18 stations in Mumbai, representing inland and coastal metropolitan environments in India. Forecasting performance is assessed across multiple horizons and compared with established temporal and spatiotemporal benchmarks. The results show that GCSVR consistently improves predictive accuracy and maintains stable performance across seasons and outlier-prone pollution episodes. Statistical test further confirms the reliability of the proposed approach across the two cities. Finally, conformal prediction is integrated with GCSVR to generate calibrated prediction intervals, enhancing its practical value for uncertainty-aware air quality monitoring and public health decision-making.

[LG-13] Nora: Normalized Orthogonal Row Alignment for Scalable Matrix Optimizer

链接: https://arxiv.org/abs/2605.03769
作者: Jinghui Yuan,Jiaxuan Zou,Shuo Wang,Yong Liu,Feiping Nie
类目: Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:Matrix-based optimizers have demonstrated immense potential in training Large Language Models (LLMs), however, designing an ideal optimizer remains a formidable challenge. A superior optimizer must satisfy three core desiderata: efficiency, achieving Muon-like preconditioning to accelerate optimization; stability, strictly adhering to the scale-invariance inherent in neural networks; and speed, minimizing computational overhead. While existing methods address these aspects to varying degrees, they often fail to unify them, either incurring prohibitive computational costs like Muon, or allowing radial jitters that compromise stability like RMNP. To bridge this gap, we propose Nora, an optimizer that rigorously satisfies all three requirements. Nora achieves training stability by explicitly stabilizing weight norms and angular velocities through row-wise momentum projection onto the orthogonal complement of the weights. Simultaneously, by leveraging the block-diagonal dominance of the Transformer Hessian, Nora effectively approximates structured preconditioning while maintaining an optimal computational complexity of \mathcalO(mn) . Furthermore, we prove that Nora is a scalable optimizer and establish its corresponding scaling theorems. With a streamlined implementation requiring only two lines of code, our preliminary experiments validate Nora as an efficient and highly promising optimizer for large-scale training.

[LG-14] Vanishing L2 regularization for the softmax Multi Armed Bandit

链接: https://arxiv.org/abs/2605.03752
作者: Stefana-Lucia Anita,Gabriel Turinici
类目: Machine Learning (cs.LG); Statistics Theory (math.ST); Machine Learning (stat.ML)
*备注:

点击查看摘要

Abstract:Multi Armed Bandit (MAB) algorithms are a cornerstone of reinforcement learning and have been studied both theoretically and numerically. One of the most commonly used implementation uses a softmax mapping to prescribe the optimal policy and served as the foundation for downstream algorithms, including REINFORCE. Distinct from vanilla approaches, we consider here the L2 regularized softmax policy gradient where a quadratic term is subtracted from the mean reward. Previous studies exploiting convexity failed to identify a suitable theoretical framework to analyze its convergence when the regularization parameter vanishes. We prove here theoretical convergence results and confirm empirically that this regime makes the L2 regularization numerically advantageous on standard benchmarks.

[LG-15] GEM-FI: Gated Evidential Mixtures with Fisher Modulation ICML2026

链接: https://arxiv.org/abs/2605.03750
作者: Marco Mustafa Mohammed,Fatemeh Daneshfar,Pietro Liò
类目: Machine Learning (cs.LG)
*备注: Accepted as a regular paper at ICML 2026. 23 pages

点击查看摘要

Abstract:Evidential Deep Learning (EDL) enables single-pass uncertainty estimation by predicting Dirichlet evidence, but it can remain overconfident and poorly calibrated, and it often fails to represent multi-modal epistemic uncertainty. We introduce Gated Evidential Mixtures (GEM), a family of models that learns an in-model energy signal and uses it to gate evidential outputs end-to-end in a distance-informed manner. GEM-CORE learns a feature-level energy and maps it to a bounded gate that smoothly suppresses evidence when support is low. To capture epistemic multi-modality without multi-pass ensembling, GEM-MIX adds a lightweight mixture of evidential heads with learned routing weights while preserving single-pass inference. Finally, GEM-FI stabilizes mixture allocations via a Fisher-informed regularizer, reducing head collapse and producing smoother boundary uncertainty. Across image classification and OOD detection benchmarks, GEM improves calibration and ID/OOD separation with single-pass inference. On CIFAR-10, GEM-FI vs. DAEDL improves accuracy from 91.11 to 93.75 (+2.64 pp), reduces Brier x100 from 14.27 to 6.81 (-7.46), and also improves misclassification-detection AUPR from 99.08 to 99.94 (+0.86). For epistemic OOD detection, GEM-FI achieves AUPR/AUROC of 92.59/95.09 on CIFAR-10 to SVHN and 90.20/89.06 on CIFAR-10 to CIFAR-100, compared with 85.54/89.30 and 88.19/86.10 for DAEDL.

[LG-16] Distribution-Free Pretraining of Classification Losses via Evolutionary Dynamics

链接: https://arxiv.org/abs/2605.03722
作者: Meng Xiang,Yan Pei
类目: Machine Learning (cs.LG)
*备注: 6 pages

点击查看摘要

Abstract:We propose Evolutionary Dynamic Loss (EDL), a framework that learns a transferable classification loss in the probability space using unlimited synthetic prediction-label pairs, without accessing real samples during the main loss pretraining stage. EDL parameterizes the loss as a lightweight network and is trained with a semantics-free ranking-consistency objective that assigns larger penalties for more erroneous predictions. To robustly explore the space of loss functions, we optimize EDL via an evolutionary strategy and introduce chaotic mutation to improve exploration under noisy fitness evaluations. Experiments on CIFAR-10 with ResNet backbones show that EDL can serve as a drop-in replacement for cross-entropy and achieves competitive or improved accuracy, while ablation studies confirm that chaotic mutation yields faster convergence and better synthetic pretraining metrics than standard Gaussian mutation.

[LG-17] Uni-OPD: Unifying On-Policy Distillation with a Dual-Perspective Recipe

链接: https://arxiv.org/abs/2605.03677
作者: Wenjin Hou,Shangpin Peng,Weinong Wang,Zheng Ruan,Yue Zhang,Zhenglin Zhou,Mingqi Gao,Yifei Chen,Kaiqi Wang,Hongming Yang,Chengquan Zhang,Zhuotao Tian,Han Hu,Yi Yang,Fei Wu,Hehe Fan
类目: Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:On-policy distillation (OPD) has recently emerged as an effective post-training paradigm for consolidating the capabilities of specialized expert models into a single student model. Despite its empirical success, the conditions under which OPD yields reliable improvement remain poorly understood. In this work, we identify two fundamental bottlenecks that limit effective OPD: insufficient exploration of informative states and unreliable teacher supervision for student rollouts. Building on this insight, we propose Uni-OPD, a unified OPD framework that generalizes across Large Language Models (LLMs) and Multimodal Large Language Models (MLLMs), centered on a dual-perspective optimization strategy. Specifically, from the student’s perspective, we adopt two data balancing strategies to promote exploration of informative student-generated states during training. From the teacher’s perspective, we show that reliable supervision hinges on whether aggregated token-level guidance remains order-consistent with the outcome reward. To this end, we develop an outcome-guided margin calibration mechanism to restore order consistency between correct and incorrect trajectories. We conduct extensive experiments on 5 domains and 16 benchmarks covering diverse settings, including single-teacher and multi-teacher distillation across LLMs and MLLMs, strong-to-weak distillation, and cross-modal distillation. Our results verify the effectiveness and versatility of Uni-OPD and provide practical insights into reliable OPD.

[LG-18] Information Plane Analysis of Binary Neural Networks

链接: https://arxiv.org/abs/2605.03636
作者: Maximilian Nothnagel,Bernhard C. Geiger
类目: Machine Learning (cs.LG)
*备注: 8 pages, 4 figures

点击查看摘要

Abstract:Information plane (IP) analysis has been suggested to study the training dynamics of deep neural networks through mutual information (MI) between inputs, representations, and targets. However, its statistical validity is often compromised by the difficulty of estimating MI from samples of high-dimensional, deterministic representations. In this work, we perform IP analyses on binary neural networks (BNNs) where activations are discrete and MI is finite. We characterise the finite-sample behaviour of the plug-in entropy estimator and identify regimes for sample size N and representation dimensionality D under which MI estimates are reliable. Outside these regimes, we show that empirical MI estimates saturate to \log_2 N , rendering IP trajectories uninformative. Restricting attention to the reliable regime, we train 375 BNNs to investigate the existence of late-stage compression phases and the relationship between compressed representations and generalisation performance. Our results show that while late-stage compression is frequently observed, compressed latent representations do not consistently correlate with improved generalization performance. Instead, the relationship between compression and generalisation is highly dependent on task, architecture, and regularisation. Comments: 8 pages, 4 figures Subjects: Machine Learning (cs.LG) Cite as: arXiv:2605.03636 [cs.LG] (or arXiv:2605.03636v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2605.03636 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[LG-19] A Few-Step Generative Model on Cumulative Flow Maps

链接: https://arxiv.org/abs/2605.03623
作者: Zhiqi Li,Duowen Chen,Yuchen Sun,Bo Zhu
类目: Machine Learning (cs.LG); Graphics (cs.GR)
*备注: 11 pages, 12 figures

点击查看摘要

Abstract:We propose a unified, few-step generative modeling framework based on \emphcumulative flow maps for long-range transport in probability space, inspired by flow-map techniques for physical transport and dynamics. At its core is a cumulative-flow abstraction that connects local, instantaneous updates with finite-time transport, enabling generative models to reason about global state transitions. This perspective yields a unified few-step framework built on cumulative transport and \revisecumulative parameterization that applies broadly to existing diffusion- and flow-based models without being tied to a specific prediction \reviseinstantiation. Our formulation supports few-step and even one-step generation while preserving synthesis quality, requiring only minimal changes to time embeddings and training objectives, and no increase in model capacity. We demonstrate its effectiveness across diverse tasks, including image generation, geometric distribution modeling, joint prediction, and SDF generation, with reduced inference cost.

[LG-20] Exact and Approximate Algorithms for Polytree Learning

链接: https://arxiv.org/abs/2605.03622
作者: Juha Harviainen,Frank Sommer,Manuel Sorge
类目: Data Structures and Algorithms (cs.DS); Computational Complexity (cs.CC); Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:Polytrees are a subclass of Bayesian networks that seek to capture the conditional dependencies between a set of n variables as a directed forest and are motivated by their more efficient inference and improved interpretability. Since the problem of learning the best polytree is NP-hard, we study which restrictions make it more tractable by considering for example in-degree bounds, properties of score functions measuring the quality of a polytree, and approximation algorithms. We devise an algorithm that finds the optimal polytree in time O((2+\epsilon)^n) for arbitrarily small \epsilon 0 and any constant in-degree bound k , improving over the fastest previously known algorithm of time complexity O(3^n) . We further give polynomial-time algorithms for finding a polytree whose score is within a factor of k from the optimal one for arbitrary scores and a factor of 2 for additive ones. Many of the results are complemented by (nearly) tight lower bounds for either the time complexity or the approximation factors.

[LG-21] Leverag ing Code Automorphisms for Improved Syndrome-Based Neural Decoding

链接: https://arxiv.org/abs/2605.03620
作者: Raphaël Le Bidan,Ahmad Ismail,Elsa Dupraz,Charbel Abdel Nour
类目: Information Theory (cs.IT); Machine Learning (cs.LG)
*备注: 6 pages, 7 figures, submitted to IEEE for possible publication. Code to reproduce all results is available at: this https URL

点击查看摘要

Abstract:Syndrome-based neural decoding (SBND) has emerged as a promising deep learning approach for soft-decision decoding of high-rate, short-length codes. However, this approach still has substantial room for improvement. In this paper, we show how to leverage code automorphisms to enhance the ability of existing SBND models to learn and generalize through data augmentation during training and inference. As a result, for the short high-rate codes considered, we obtain models that closely approach MLD performance using small datasets and proper training. Our findings also suggest that many prior results for SBND models in the literature underestimate their true correction capability due to undertraining. Code to reproduce all results is available at: this https URL.

[LG-22] Most ReLU Networks Admit Identifiable Parameters

链接: https://arxiv.org/abs/2605.03601
作者: Moritz Grillo,Guido Montúfar
类目: Machine Learning (cs.LG); Discrete Mathematics (cs.DM); Combinatorics (math.CO)
*备注:

点击查看摘要

Abstract:We study the realization map of deep ReLU networks, focusing on when a function determines its parameters up to scaling and permutation. To analyze hidden redundancies beyond these standard symmetries, we introduce a framework based on weighted polyhedral complexes. Our main result shows that for every architecture whose input and hidden layers have width at least two, there exists an open set of identifiable parameters. This implies that the functional dimension of every such architecture is exactly the number of parameters minus the number of hidden neurons. We further show that minimal functional representations can still have non-trivial parameter redundancies. Finally, we establish a generic depth hierarchy, whereby for an open set of parameters the realized function cannot be represented generically by any shallower network.

[LG-23] Enhance the after-discharge mortality rate prediction via learning from the medical notes

链接: https://arxiv.org/abs/2605.03560
作者: Zijiang Yang
类目: Machine Learning (cs.LG); Other Computer Science (cs.OH)
*备注:

点击查看摘要

Abstract:With the increase of the Electronic Health Records (EHR) data, more and more researchers are developing machine learning models to learn from the medical notes. These unstructured text data pose significant challenges on the learning process as the quality of data is low. These data are often messy, repetitive and redundant. We have shown these notes data to be informative by conducting the after-discharge mortality rate prediction task. The AUC-ROC for models using the medical note information is generally 0.1 higher than those without the medical notes. Furthermore, we propose the Deep Neural Network(DNN) model with ‘pooling’ mechanism to enhance the mortality prediction. Based on the experimental results, we demonstrate that the proposed model outperforms the traditional machine learning models like the tree-based models. The proposed method learns from the most informative medical notes and improves the prediction accuracy significantly. The AUC-ROC for the proposed model is 2% to 14% higher than the traditional ones in 15-days, 30-days, 60-days, 365-days after-discharge mortality prediction tasks. Moreover, we can discover some interesting knowledge through the traditional and proposed models. These knowledge are inspiring but also consistent with the previous findings. The models are able to reveal the relationships between the informative keywords and documents from the medical notes and the severity of the patients.

[LG-24] Random test functions H-1 norm equivalence and stochastic variational physics-informed neural networks

链接: https://arxiv.org/abs/2605.03542
作者: Diego Marcondes
类目: Numerical Analysis (math.NA); Machine Learning (cs.LG)
*备注: 76 pages, 14 Figures

点击查看摘要

Abstract:The dual norm characterisation of weak solutions of second-order linear elliptic partial differential equations is mathematically natural but computationally intractable: evaluating the H^-1 norm of a residual requires a supremum over an infinite-dimensional function space. We prove that the H^-1 norm of any functional is equivalent to its expected squared evaluation against a random test function whose distribution depends only on the domain. Crucially, realisations of this random test function have negative Sobolev regularity for d \geq 2 , yet this roughness is not an obstacle: averaging over the distribution exactly recovers the correct weak topology, independently of the differential operator. This equivalence introduces the notion of stochastically weak solutions, which coincide with classical weak solutions, and motivates stochastic variational physics-informed neural networks (SV-PINNs): neural networks trained by minimising an empirical approximation of the stochastic norm of the PDE residual. Although instantiated here with neural networks as trial spaces, the underlying principle is independent of the approximation architecture and suggests a broader paradigm for numerical methods based on stochastic rather than deterministic test spaces. The framework extends naturally to higher-order elliptic, parabolic and hyperbolic equations and to abstract operator equations on Hilbert spaces. As a proof of concept, we present numerical experiments on eight challenging second-order linear elliptic problems spanning high-frequency and multi-scale solutions, indefinite operators, variable coefficients, and non-standard domains, in which SV-PINNs consistently and significantly outperform standard PINNs, recovering solutions to within one percent relative error in hundreds of L-BFGS steps.

[LG-25] Understanding Self-Supervised Learning via Latent Distribution Matching ICML2026

链接: https://arxiv.org/abs/2605.03517
作者: Fabian A Mikulasch,Friedemann Zenke
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: Accepted to ICML 2026 (Spotlight)

点击查看摘要

Abstract:Self-supervised learning (SSL) excels at finding general-purpose latent representations from complex data, yet lacks a unifying theoretical framework that explains the diverse existing methods and guides the design of new ones. We cast SSL as latent distribution matching (LDM): learning representations that maximize their log-probability under an assumed latent model (alignment), while maximizing latent entropy to prevent collapse (uniformity). This view unifies independent component analysis with contrastive, non-contrastive, and predictive SSL methods, including stop gradient approaches. Leveraging LDM, we derive a nonlinear, sampling-free Bayesian filtering model with a Kalman-based predictor for high-dimensional timeseries. We further prove that predictive LDM yields identifiable latent representations under mild assumptions, even with nonlinear predictors. Overall, LDM clarifies the assumptions behind established SSL methods and provides principled guidance for developing new approaches.

[LG-26] A Hierarchical Sampling Framework for bounding the Generalization Error of Federated Learning

链接: https://arxiv.org/abs/2605.03499
作者: Dario Filatrella,Ragnar Thobaben,Mikael Skoglund
类目: Machine Learning (cs.LG); Information Theory (cs.IT); Machine Learning (stat.ML)
*备注:

点击查看摘要

Abstract:We study expected generalization bounds for the Hierarchical Federated Learning (HFL) setup using Wasserstein distance. We introduce a generalized framework in which data is sampled hierarchically, and we model it with a multi-layered tree structure that induces dependencies among the clients’ datasets. We derive generalization bounds in terms of Wasserstein distance under the Lipschitz assumption on the loss function, by applying a supersample construction that allows us to measure the sensitivity of the algorithm to the change of a single node in the sampling tree. By leveraging the FL structure, we recover and strictly imply existing state-of-the-art conditional mutual information (CMI) bounds in the case of bounded losses. We also show that our bound can be applied together with Differential Privacy assumptions, to recover generalization bounds based on algorithmic privacy. To assess the tightness of our bounds, we study the Gaussian Location Model (GLM) and show that we recover the actual asymptotic rate of the generalization error.

[LG-27] GRIFDIR: Graph Resolution-Invariant FEM Diffusion Models in Function Spaces over Irregular Domains

链接: https://arxiv.org/abs/2605.03497
作者: James Rowbottom,Elizabeth L. Baker,Nick Huang,Ben Adcock,Carola-Bibiane Schönlieb,Alexander Denker
类目: Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:Score-based diffusion models in infinite-dimensional function spaces provide a mathematically principled framework for modelling function-valued data, offering key advantages such as resolution invariance and the ability to handle irregular discretisations. However, practical implementations have struggled to fully realise these benefits. Existing backbones like Fourier neural operators are often biased towards regular grids and fail to generalise to complex domain topologies. We propose a novel architecture for function-space diffusion models that represents generalised graph convolutional kernels as finite element functions, enabling the model to naturally handle unstructured meshes and complex geometries. We demonstrate the efficacy of our network architecture through a series of unconditional and conditional sampling experiments across diverse geometries, including non-convex and multiply-connected domains. Our results show that the proposed method maintains resolution invariance and achieves high fidelity in capturing functional distributions on non-trivial geometries.

[LG-28] Bandits attack function optimization CEC2014

链接: https://arxiv.org/abs/2605.03496
作者: Philippe Preux,Rémi Munos,Michal Valko
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: IEEE CEC 2014; 8 pages

点击查看摘要

Abstract:We consider function optimization as a sequential decision making problem under budget constraint. This constraint limits the number of objective function evaluations allowed during the optimization. We consider an algorithm inspired by a continuous version of a multi-armed bandit problem which attacks this optimization problem by solving the tradeoff between exploration (initial quasi-uniform search of the domain) and exploitation (local optimization around the potentially global maxima). We introduce the so-called Simultaneous Optimistic Optimization (SOO), a deterministic algorithm that works by domain partitioning. The benefit of such approach are the guarantees on the returned solution and the numerical efficiency of the algorithm. We present this machine learning approach to optimization, and provide the empirical assessment of SOO on the CEC’2014 competition on single objective real-parameter numerical optimization test-suite.

[LG-29] Adaptive graph-based algorithms for conditional anomaly detection and semi-supervised learning

链接: https://arxiv.org/abs/2605.03495
作者: Michal Valko
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: PhD thesis, University of Pittsburgh, 2011. 124 pages

点击查看摘要

Abstract:We develop graph-based methods for semi-supervised learning based on label propagation on a data similarity graph. When data is abundant or arrive in a stream, the problems of computation and data storage arise for any graph-based method. We propose a fast approximate online algorithm that solves for the harmonic solution on an approximate graph. We show, both empirically and theoretically, that good behavior can be achieved by collapsing nearby points into a set of local representative points that minimize distortion. Moreover, we regularize the harmonic solution to achieve better stability properties. We also present graph-based methods for detecting conditional anomalies and apply them to the identification of unusual clinical actions in hospitals. Our hypothesis is that patient-management actions that are unusual with respect to the past patients may be due to errors and that it is worthwhile to raise an alert if such a condition is encountered. Conditional anomaly detection extends standard unconditional anomaly framework but also faces new problems known as fringe and isolated points. We devise novel nonparametric graph-based methods to tackle these problems. Our methods rely on graph connectivity analysis and soft harmonic solution. Finally, we conduct an extensive human evaluation study of our conditional anomaly methods by 15 experts in critical care.

[LG-30] Bandits on graphs and structures

链接: https://arxiv.org/abs/2605.03493
作者: Michal Valko
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: Habilitation thesis, ENS Cachan, 2016. 84 pages

点击查看摘要

Abstract:The goal of this thesis is to investigate the structural properties of certain sequential problems in order to bring the solutions closer to a practical use. In the first part, we put a special emphasis on structures that can be represented as graphs on actions. In the second part, we study the large action spaces that can be of exponential size in the number of base actions or even infinite. For graph bandits, we consider the settings of smoothness of rewards (spectral bandits), side observations, and influence maximization. For large structured domains, we cover kernel bandits, polymatroid bandits, bandits for function optimization (including unknown smoothness), and infinitely many-arms bandits. The thesis aspires to be a survey of the author’s contributions on graph and structured bandits.

[LG-31] Quantum Hierarchical Reinforcement Learning via Variational Quantum Circuits

链接: https://arxiv.org/abs/2605.03434
作者: Yu-Ting Lee,Samuel Yen-Chi Chen,Fu-Chieh Chang
类目: Machine Learning (cs.LG); Quantum Physics (quant-ph)
*备注:

点击查看摘要

Abstract:Reinforcement learning is one of the most challenging learning paradigms where efficacy and efficiency gains are extremely valuable. Hierarchical reinforcement learning is a variant that leverages temporal abstraction to structure decision-making. While parametrized quantum computations have shown success in non-hierarchical reinforcement learning, whether these advantages adapt to hierarchical decision-making remains a critical open question. In this work, we develop a hybrid hierarchical agent based on the option-critic architecture. This hybrid agent substitutes classical components with variational quantum circuits for feature extractors, option-value functions, termination functions, and intra-option policies. Evaluated on standard benchmarking environments, results show that a hybrid agent utilizing a quantum feature extractor can outperform classical baselines while saving up to 66% trainable parameters. We also identify an architectural bottleneck that quantum option-value estimation severely degrades performance. Further ablation studies reveal how architectural choices of the quantum circuits affect performance. Our work establishes design principles for parameter-efficient hybrid hierarchical agents.

[LG-32] FIBER: A Differentially Private Optimizer with Filter-Aware Innovation Bias Correction

链接: https://arxiv.org/abs/2605.03425
作者: Duc Dm,Thao Do,Minh Son Hoang,Anh Le Duc Tran,Daeyoung Kim,Huy Nguyen
类目: Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:Differentially private (DP) training protects individual examples by adding noise to gradients, but the injected noise interacts nontrivially with adaptive optimizers. Recent DP methods temporally filter privatized gradients to reduce variance; however, filtering also changes the DP noise statistics seen by AdamW’s second-moment accumulator. As a result, bias corrections derived for unfiltered DP noise, such as subtracting sigma_w squared, can become miscalibrated when filtering is present. We propose FiBeR, a DP optimizer designed for temporally filtered privatized gradients. FiBeR (i) performs denoising in innovation space by filtering the residual stream and integrating it to form the filtered gradient estimate, (ii) decouples the two-point observation geometry from the innovation gain to enable independent tuning, and (iii) introduces a filter-aware second-moment calibration that subtracts the attenuated DP noise contribution A(omega) sigma_w squared, where A(omega) is derived in closed form for the innovation filter and can be computed for general stable linear filters. Across vision and language benchmarks, FiBeR consistently demonstrates substantial improvements in the performance of DP optimizers, surpassing state-of-the-art results under equivalent privacy constraints on multiple tasks. Subjects: Machine Learning (cs.LG) Cite as: arXiv:2605.03425 [cs.LG] (or arXiv:2605.03425v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2605.03425 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[LG-33] PODiff: Latent Diffusion in Proper Orthogonal Decomposition Space for Scientific Super-Resolution ICML2026

链接: https://arxiv.org/abs/2605.03399
作者: Onkar Jadhav,Tim French,Matthew Rayson,Nicole L. Jones
类目: Machine Learning (cs.LG); Atmospheric and Oceanic Physics (physics.ao-ph)
*备注: Accepted at ICML 2026

点击查看摘要

Abstract:Probabilistic super-resolution of high-dimensional spatial fields using diffusion models is often computationally prohibitive due to the cost of operating directly in pixel space. We propose PODiff, a structured conditional generative framework that performs diffusion in a fixed, variance-ordered Proper Orthogonal Decomposition (POD) coefficient space, exploiting the orthogonality of POD modes to impose an interpretable, variance-ordered latent geometry. This design enables efficient ensemble generation, preserves dominant spatial structure, and yields spatially interpretable, well-calibrated uncertainty at substantially lower computational cost. We evaluate PODiff on sea surface temperature downscaling over the West Australian coast and on a controlled advection-diffusion benchmark. PODiff achieves reconstruction accuracy comparable to pixel-space diffusion while requiring significantly less memory and producing more reliable uncertainty estimates than deterministic and Monte Carlo Dropout baselines.

[LG-34] Graph Reconstruction from Differentially Private GNN Explanations

链接: https://arxiv.org/abs/2605.03388
作者: Rishi Raj Sahoo,Jyotirmaya Shivottam,Subhankar Mishra
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR)
*备注:

点击查看摘要

Abstract:Regulatory frameworks such as GDPR increasingly require that ML predictions be accompanied by post-hoc explanations, even when raw data and trained models cannot be released. Differential privacy (DP) is the standard mitigation for the residual privacy risk of releasing these explanations. We show that DP is not sufficient: an adversary observing only DP-perturbed GNN explanations can reconstruct hidden graph structure with high accuracy. Our attack, PRIVX, exploits the fact that the Gaussian DP mechanism is a single DDPM forward step at known noise level \sigma(\epsilon), recasting reconstruction as reverse diffusion conditioned on the corrupted signal, a principled Bayesian denoiser under known DP corruption. We formalise a stratified adversary model parameterised by (M, \hat\epsilon, \hat\delta, S, \rho) that interpolates between oblivious and oracle attackers, and derive endpoint-matched two-sided bounds on reconstruction AUC. For practitioners, we provide regime-stratified guidance on explainer choice: on homophilic graphs, neighbourhood-aggregating explainers (GraphLIME, GNNExplainer) leak more structure than per-node gradient explainers under the same DP budget; on strongly heterophilic graphs the ordering reverses. We introduce PRIVF as an auxiliary diagnostic sharing the same diffusion backbone to decompose leakage into explainer-induced and intrinsic graph-distribution components. Experiments across seven benchmarks, three DP mechanisms, and three GNN backbones show PRIVX achieves AUC above 0.7 at \epsilon = 5 on five of seven datasets, with the attack succeeding well within typically deployed privacy budgets.

[LG-35] GRAFT: Auditing Graph Neural Networks via Global Feature Attribution

链接: https://arxiv.org/abs/2605.03377
作者: Rishi Raj Sahoo,Subhankar Mishra
类目: Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:Graph Neural Networks (GNNs) achieve strong performance on node classification tasks but remain difficult to interpret, particularly with respect to which input features drive their predictions. Existing global GNN explainers operate at the structural level identifying recurring subgraph motifs, but none explain model behaviour globally at the level of input node attributes. We propose GRAFT, a posthoc global explanation framework that identifies class-level feature importance profiles for GNNs. The method combines diversity-guided exemplar selection, Integrated Gradients-based attribution, and aggregation to construct a global view of feature influence for each class, which can be further expressed as concise natural language rules using a large language model with self-refinement. We evaluate GRAFT across multiple datasets, architectures, and experimental settings, demonstrating its effectiveness in capturing model-relevant features, supporting bias analysis, and enabling feature-efficient transfer learning. In addition, we introduce a structured human evaluation protocol to assess the interpretability of generated rules along dimensions such as accuracy and usefulness. Our results suggest that GRAFT provides a practical and interpretable approach for analysing feature-level behaviour in GNNs, bridging quantitative attribution with human-understandable explanations.

[LG-36] Learning Dynamics of Zeroth-Order Optimization: A Kernel Perspective ICML2026

链接: https://arxiv.org/abs/2605.03373
作者: Zhe Li,Bicheng Ying,Zidong Liu,Haibo Yang
类目: Machine Learning (cs.LG)
*备注: Accepted by ICML 2026

点击查看摘要

Abstract:Classical optimization theory establishes that zeroth-order (ZO) algorithms suffer from a dimension-dependent slowdown, with convergence rates typically scaling with the model dimension compared to first-order methods. However, in contrast to these theoretical expectations, a growing body of recent work demonstrates the successful application of ZO methods to fine-tuning Large Language Models (LLMs) with billions of parameters. To explain this paradox, we derive the one-step learning dynamics of ZO SGD, where the empirical Neural Tangent Kernel (eNTK) naturally emerges as the key term governing the learning behavior. Inspection of the eNTK produced by ZO SGD reveals that each element corresponds to the inner product of neural tangent vectors projected onto a random low-dimensional subspace. Thus, by invoking the Johnson-Lindenstrauss Lemma, our analysis shows that the fidelity of the ZO eNTK is governed primarily by the number of perturbations. Crucially, the approximation error depends on the model output size rather than the massive parameter dimension. This dimension-free property provides a theoretical justification for the scalability of ZO methods to LLMs finetuning tasks. We believe that this kernel-based framework offers a novel perspective for understanding ZO methods within the context of learning dynamics.

[LG-37] Fully Automatic Trace Gas Plume Detection

链接: https://arxiv.org/abs/2605.03372
作者: Vít Růžička,David R. Thompson,Jay E. Fahlen,Amanda M. Lopez,Steven Lu,Chuchu Xiang,Holly Bender,Daniel Jensen,Philip G. Brodrick,Jake Lee,Brian Bue,Daniel H. Cusworth,Luis Guanter,Adam Chlus,Andrew Thorpe,Robert O. Green
类目: Machine Learning (cs.LG)
*备注: Manuscript 27 pages, 9 figures, 1 table, more in attached supplementary; In review

点击查看摘要

Abstract:Future imaging spectrometers will increase data volumes by orders of magnitude, requiring automated detection of trace gas point sources. We present a fully automated framework that combines machine learning-based morphological analysis with physics-based spectroscopic fitting to detect plumes without human participation. Applied to EMIT imaging spectrometer data, the system operates in two modes: “daily digest” that runs automatically on all downlinked data, flagging the largest events for immediate response, and a retrospective analysis that identifies plumes missed by prior human review. The daily digest demonstrates that a significant fraction of the largest plumes can be detected automatically with negligible false positives, while retrospective analysis suggests at least 25% of plumes may have been overlooked. In addition to the previously observed methane point sources, we extend detection to three understudied trace gases: NH3, NO2 and the first observations of carbon monoxide (CO) plume in EMIT imagery.

[LG-38] Population-Aware Imitation Learning in Mean-field Games with Common Noise

链接: https://arxiv.org/abs/2605.03357
作者: Grégoire Lambrecht,Mathieu Laurière
类目: Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注:

点击查看摘要

Abstract:Mean Field Games (MFGs) provide a powerful framework for modeling the collective behavior of large populations of interacting agents. In this paper, we address the problem of Imitation Learning (IL) in MFGs subject to common noise, where the population distribution evolves stochastically. This stochasticity compels agents to adopt population-aware policies to respond to aggregate shocks. We formulate two distinct learning objectives: recovering a Nash equilibrium and maximizing performance against an expert population. We investigate two imitation proxies: Behavioral Cloning (BC) and Adversarial (ADV) divergence. We then establish finite-sample error bounds showing that minimizing these proxies effectively controls both the policy’s exploitability and its performance gap relative to the expert. Furthermore, we propose a numerical framework using generalized Fictitious Play and Deep Learning to compute expert population-aware policies. Through experiments on three environments we demonstrate that standard population-unaware policies fail to capture the equilibrium dynamics. Our results highlight that learning population-aware policies is crucial to avoid being misled by the randomness inherent in common noise.

[LG-39] Provable Accuracy Collapse in Embedding-Based Representations under Dimensionality Mismatch ICML2026

链接: https://arxiv.org/abs/2605.03346
作者: Dionysis Arvanitakis,Vaggos Chatziafratis,Yiyuan Luo
类目: Data Structures and Algorithms (cs.DS); Machine Learning (cs.LG)
*备注: Preliminary version, accepted to ICML 2026 as spotlight presentation

点击查看摘要

Abstract:Embedding-based representations in Euclidean space \mathbbR^d are a cornerstone of modern machine learning, where a major goal is to use the \emphsmallest dimension that faithfully captures data relations. In this work, we prove sharp dimension–accuracy tradeoffs and identify a fundamental information-theoretic limitation: unless the embedding dimension d is chosen close to the ground-truth dimension D , accuracy undergoes a sudden collapse. Our main result shows that this phenomenon arises even in standard contrastive learning settings, where supervision is limited to a set of m anchor–positive–negative triplets (i,j,k) encoding distance comparisons \mathrmdist(i,j) \mathrmdist(i,k) . Specifically, given triplets realizable by an unknown ground-truth embedding in D dimensions, we prove that there exists constant c 1 , such that \emphevery embedding of dimension at most cD violates half of the triplets, yielding accuracy as low as a trivial one-dimensional solution that ignores the input. We complement our information-theoretic bounds with strong computational hardness results: under the Unique Games Conjecture, even if the given triplets are nearly realizable in D=1 dimension, no polynomial-time algorithm – \textitregardless of its dimension – can achieve accuracy above the trivial 50% baseline.

[LG-40] Distributed Learning with Adversarial Gradient Perturbations

链接: https://arxiv.org/abs/2605.03313
作者: Nawapon Sangsiri,Yufei Tao
类目: Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:Privacy concerns in distributed learning often lead clients to return intentionally altered gradient information. We consider the problem of learning convex and L -smooth functions under adversarial gradient perturbation, where a client’s gradient reply to a server query can deviate arbitrarily from the true gradient subject to a distance bound. Our study focuses on two fundamental questions: (i) what is the smallest achievable sub-optimality gap (i.e., excess error in optimization) under such responses, and (ii) how many queries are sufficient to guarantee a given sub-optimality gap? We establish tight feasibility thresholds on the sub-optimality gap and provide algorithms that achieve these thresholds with provable query complexity guarantees.

[LG-41] Will the Carbon Border Adjustment Mechanism Impact European Electricity Prices? A GNN-Based Network Analysis

链接: https://arxiv.org/abs/2605.03304
作者: Jiachen Shen,Jian Shi,Dan Wang,Han Zhu
类目: Machine Learning (cs.LG); Computers and Society (cs.CY); Systems and Control (eess.SY)
*备注:

点击查看摘要

Abstract:The European Union’s Carbon Border Adjustment Mechanism (CBAM) creates a complex challenge for the interconnected European electricity market. Traditional static analyses often miss the cross-border spillover effects that are vital for understanding this policy. This paper addresses this gap by developing a spatio-temporal Graph Neural Network (GNN) framework. It quantifies how CBAM affects electricity prices and carbon intensity (CI) at the same time. We modeled a subgraph of eight European countries. Our results suggest that CBAM is not just a uniform tax. Instead, it acts as a tool that transforms the market and creates structural differences. In our simulated scenarios, we observe that low-carbon countries like France and Switzerland can gain a competitive advantage. This suggests a potential decrease in their domestic electricity prices. Meanwhile, high-carbon countries like Poland face a double burden of rising costs. We identify the primary driver as a fundamental shift in the market’s merit order.

[LG-42] Stable Multimodal Graph Unlearning via Feature-Dimension Aware Quantile Selection

链接: https://arxiv.org/abs/2605.03303
作者: Jingjing Zhou,Yongshuai Yang,Qing Qing,Ziqi Xu,Xikun Zhang,Renqiang Luo,Ivan Lee,Feng Xia
类目: Machine Learning (cs.LG); Multimedia (cs.MM)
*备注:

点击查看摘要

Abstract:Graph unlearning remains a critical technique for supporting privacy-preserving and sustainable multimodal graph learning. However, we observe that existing unlearning strategies tend to apply uniform parameter selection and editing across all graph neural network (GNN) layers, which is especially harmful for multimodal graphs where high-dimensional input projections encode dominant cross-modal knowledge. As a result, over-editing these sensitive layers often leads to catastrophic utility degradation after forgetting, undermining both stable learning and effective privacy protection. To address this gap, we propose FDQ, a Feature-Dimension Aware Quantile framework for multimodal graph unlearning. FDQ adaptively identifies high-dimensional input projection layers and applies more conservative, FDQ-guided quantile thresholds when constructing suppression sets, while keeping the underlying importance estimation mechanism unchanged. FDQ is seamlessly integrated with diagonal sensitivity-based parameter importance analysis to enable efficient node and edge unlearning under general forget requests. Through extensive experiments on Ele-Fashion and Goodreads-NC, we demonstrate that FDQ consistently achieves strong utility preservation while maintaining effective forgetting against membership inference attacks. Overall, FDQ offers a principled and robust solution for privacy-aware unlearning in high-dimensional multimodal graph systems.

[LG-43] Contrastive Regularization for Accent-Robust ASR

链接: https://arxiv.org/abs/2605.03297
作者: Van-Phat Thai,Aradhya Dhruv,Duc-Thinh Pham,Sameer Alam
类目: ound (cs.SD); Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:ASR systems based on self-supervised acoustic pretraining and CTC fine-tuning achieve strong performance on native speech but remain sensitive to accent variability. We investigate supervised contrastive learning (SupCon) as a lightweight, accent-invariant auxiliary objective for CTC fine-tuning. An utterance-level contrastive loss regularizes encoder representations without architectural modification or explicit accent supervision. Experiments on the L2-ARCTIC benchmark show consistent WER reductions across multiple pretrained encoders, with up to 25 – 29% relative reduction under unseen-accent evaluation. Analysis using within-transcript cosine dispersion indicates that SupCon promotes more compact and stable representation geometry under accent variability. Overall, SupCon provides an effective and model-agnostic regularization strategy for improving accent robustness.

[LG-44] RFPrompt: Prompt-Based Expert Adaptation of the Large Wireless Model for Modulation Classification

链接: https://arxiv.org/abs/2605.03279
作者: Md Raihan Uddin,Tolunay Seyfi,Fatemeh Afghah
类目: Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:Automatic modulation classification (AMC) in real-world deployments demands robustness to distribution shifts arising from hardware impairments, unseen propagation environments, and recording conditions never encountered during training. Although wireless foundation models offer a promising starting point for robust RF representation learning, an important open question is how to adapt them efficiently to out-of-distribution (OOD) downstream tasks without overwriting the structure learned during large-scale pre-training. In this paper, we investigate prompt-based adaptation as a general mechanism for OOD transfer in wireless foundation models. We propose RFPrompt, a parameter-efficient framework that introduces learnable deep prompt tokens while keeping the pretrained backbone frozen, enabling task-specific adaptation with minimal trainable parameters. We instantiate and evaluate this approach on the Large Wireless Model (LWM), a mixture-of-experts wireless foundation model, and study its behavior under both standard and OOD modulation-classification settings. Results show that prompt-based adaptation consistently improves robustness under distribution shift and limited supervision, particularly on real-world over-the-air IQ data, while preserving strong parameter efficiency. These findings suggest that prompt learning is a practical and effective strategy for adapting wireless foundation models to challenging downstream RF environments.

[LG-45] A Universal Reproducing Kernel Hilbert Space from Polynomial Alignment and IMQ Distance

链接: https://arxiv.org/abs/2605.03262
作者: Taha Bouhsine
类目: Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:We introduce the Yat kernel k_b,\varepsilon(\mathbfw,\mathbfx)=\frac(\mathbfw^\top\mathbfx+b)^2|\mathbfx-\mathbfw|^2+\varepsilon,\qquad b\ge 0,\ \varepsilon0, a rational hidden-unit primitive whose units are Mercer sections over a shared input/weight space. For b\ge 0 the kernel is PSD; for b0 it dominates a scaled inverse-multiquadric (IMQ) in the Loewner order, yielding fixed-kernel universality, characteristicness, and strict positive definiteness on every compact domain. The polynomial numerator opens nonradial alignment channels absent from finite IMQ expansions, witnessed by the directional far-field trace T_\infty g_\varepsilon(\cdot;\mathbfw,b)(\mathbfu)=(\mathbfu^\top\mathbfw)^2 . Algebraically, a second finite difference in the bias recovers any IMQ atom from three positive-bias Yat atoms exactly, sharp at three atoms in every dimension at exact pointwise equality. A trained shared- (b,\varepsilon) Yat layer is therefore a finite learned-center expansion in a fixed universal characteristic RKHS, with closed-form norm \boldsymbol\alpha^\top\mathbfK\boldsymbol\alpha and explicit diagonal (|\mathbfx|^2+b)^2/\varepsilon driving a Rademacher generalization bound.

[LG-46] Do LLM s have core beliefs?

链接: https://arxiv.org/abs/2605.03255
作者: Anna Sokol,Marianna B. Ganapini,Nitesh V. Chawla
类目: Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:The rise of Large Language Models (LLMs) has sparked debate about whether these systems exhibit human-level cognition. In this debate, little attention has been paid to a structural component of human cognition: core beliefs, truths that provide a foundation around which we can build a worldview. These commitments usually resist debunking, as abandoning them would represent a fundamental shift in how we see reality. In this paper, we ask whether LLMs hold anything akin to core commitments. Using a probing framework we call Adversarial Dialogue Trees (ADTs) over five domains (science, history, geography, biology, and mathematics), we find that most LLMs fail to maintain a stable worldview. Though some recent models showed improved stability, they still eventually failed to maintain key commitments under conversational pressure. These results document an improvement in argumentative skills across model generations but indicate that all current models lack a key component of human-level cognition.

[LG-47] Beyond Activation Alignment: The Geometry of Neural Sensitivity

链接: https://arxiv.org/abs/2605.03222
作者: Amirhossein Yavari,Farnaz Zamani Esfahlani
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 9 pages, 4 figures

点击查看摘要

Abstract:Activation-alignment measures such as Representational Similarity Analysis (RSA), Canonical Correlation Analysis (CCA), and Centered Kernel Alignment (CKA) are widely used to compare biological and artificial neural representations. Recent theoretical work interprets many of these methods as assessing agreement between optimal linear readouts over broad families of global tasks. However, agreement at the level of global readouts does not determine how a system uses local stimulus evidence. Specifically, representations may align in activation space yet differ in their sensitivity to small perturbations. To address this challenge, we introduce a complementary framework based on local decodable information, which focuses on a representation’s ability, under noise, to discriminate small perturbations within a specified stimulus-coordinate subspace. Building on Fisher information and local representation geometry, we summarize each representation using the expected projected pullback/Fisher metric over that subspace. This formulation induces a second-moment family of local discrimination tasks, for which the resulting operator provides a minimal, complete dataset-level summary of expected discriminability. We compare these regularized signatures using a log-spectral distance on the manifold of symmetric positive definite (SPD) matrices, yielding the Spectral Riemannian Alignment Score (S-RAS) and a uniform multiplicative certificate over the corresponding family of lifted task values. Empirically, this framework enables the recovery of corresponding layers across independently trained artificial neural networks, supports transferable class-conditional probes, reveals controlled dissociations between standard and robust training, and uncovers stimulus-coordinate family effects across mouse visual cortex using the Allen Brain Observatory static gratings dataset.

[LG-48] Moral Sensitivity in LLM s: A Tiered Evaluation of Contextual Bias via Behavioral Profiling and Mechanistic Interpretability

链接: https://arxiv.org/abs/2605.03217
作者: Yash Aggarwal,Atmika Gorti,Vinija Jain,Aman Chadha,Krishnaprasad Thirunarayan,Manas Gaur
类目: Machine Learning (cs.LG); Computers and Society (cs.CY)
*备注:

点击查看摘要

Abstract:Large language models (LLMs) are increasingly deployed in settings that require nuanced ethical reasoning, yet existing bias evaluations treat model outputs as simply “biased” or “unbiased.” This binary framing misses the gradual, context-sensitive way bias actually emerges. We address this gap in two stages: behavioral profiling and mechanistic validation. In the behavioral stage, we introduce the Moral Sensitivity Index (MSI), a metric that quantifies the probability of biased output across a graduated, seven-tier stress test ranging from abstract numerical problems to scenarios rooted in historical and socioeconomic injustice. Evaluating four leading models (Claude 3.5, Qwen 3.5, Llama 3, and Gemini 1.5), we identify distinct behavioral signatures shaped by alignment design: for instance, Gemini 1.5 reaches 72.7% MSI by Tier 5 under socioeconomic framing, while Claude exhibits sharp suppression consistent with identity-based safety training. We then verify these behavioral patterns mechanistically. We select criminal-bias scenarios, which produced the highest MSI scores across models, as probes and apply logit lens, attention analysis, activation patching, and semantic probing to a controlled set of six models spanning three capability tiers: small language models (SLMs), instruction-tuned base models, and reasoning-distilled variants. Circuit-level analysis reveals a U-curve of bias: SLMs exhibit strong criminal bias; scaling to instruction-tuned models eliminates it; reasoning distillation reintroduces bias to SLM-like levels despite identical parameter counts, suggesting distillation compresses reasoning traces in ways that reactivate shallow statistical associations. Critically, the socially loaded cues that drive high MSI scores activate the same bias-driving circuits identified mechanistically, providing cross-stage validation.

[LG-49] Enhancing AI-Based ECG Delineation with Deep Learning Denoising Techniques

链接: https://arxiv.org/abs/2605.03183
作者: Jeff Breeding-Allison,Emil Walleser
类目: Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注: 24 pages, 8 figures

点击查看摘要

Abstract:Evaluating canine electrocardiograms (ECGs) is challenging due to noise that can obscure clinically relevant cardiac electrical activity. Common sources of interference include respiration, muscle activity, poor lead contact, and external electrical artifacts. Classical signal denoising techniques, such as filtering and wavelet-based methods, struggle to suppress diverse noise patterns while preserving morphological features critical for accurate ECG delineation. We propose an autoencoder-based neural network model and training strategy for ECG denoising as a preprocessing step for canine ECG analysis. The model is trained to reconstruct clean cardiac signals from noisy inputs, enabling effective noise reduction without degrading diagnostically important waveforms. Our approach demonstrates strong performance across both noisy and clean ECG recordings, indicating robustness to varying signal conditions and suitability for downstream delineation tasks.

[LG-50] Pairwise matrices for sparse autoencoders: single-feature inspection mislabels causal axes

链接: https://arxiv.org/abs/2605.03160
作者: Michael A. Riegler,Birk Sebastian Frostelid Torpmann-Hagen
类目: Machine Learning (cs.LG)
*备注: 18 pages

点击查看摘要

Abstract:The standard sparse-autoencoder (SAE) interpretability protocol labels each feature from its top-activating contexts and validates by single-feature steering. We propose the pairwise matrix protocol, co-varying steering coefficient with joint condition, and report three findings the standard one-corner protocol misses on Qwen3-1.7B-Instruct, replicated on Gemma-2-2B-it. First, a feature labelled “AI self-disclaimer” from its top contexts produces an inverted U-shape under a coefficient sweep: at c=+500 the model substitutes a fluent contemplative-philosopher voice for the disclaimer. Two further features anchor the criterion (one monotonic, one pure breakdown). Second, three near-orthogonal cluster-specific features that individually steer a philosophy-of-mind register, jointly suppressed at c=-500, damage grounded composition on recipes and engine explanations as well as introspective prompts; single-feature suppression at the same magnitude leaves controls intact. Third, a matched-geometry comparison of single-feature, joint, and random-direction perturbations (norm ~1.55, cosine ~0.64) yields three distinct output regimes: single-feature substitutes strategy filler, random direction substitutes diverse content, joint suppression alone produces placeholder text. Coherence loss is direction-pattern-dependent, not magnitude-dependent. All three findings reproduce on Gemma with model-specific damage signatures; the matched-geometry control is CI-separated by ~10x. The pipeline also locates a top causally responsible feature in Llama-3.1-8B-Instruct.

[LG-51] Instance-Level Costs for Nuanced Classifier Evaluation

链接: https://arxiv.org/abs/2605.03135
作者: Kabir Kang,Stephen Mussmann
类目: Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:Standard classification treats all errors equally, but in content moderation, medical screening, and safety-critical applications, mistakes on clear-cut cases are far more costly than errors on ambiguous ones. We propose normalized excess cost (NEC), a metric that weights classification errors by per-example costs and reduces to standard error rate when costs are uniform. Costs can derive from annotator vote margins, distance from decision thresholds, or confidence ratings. Across text, image, and tabular benchmarks, we find that NEC is often substantially lower than error rate – models with 5% error rate can achieve 1.8% NEC – revealing that most mistakes concentrate on ambiguous, low-cost examples. However, incorporating costs into training via loss weighting, sampling strategies, or regression yields inconsistent benefits: improvements appear only when costs are predictable from input features, as in our synthetic control, while real-world datasets show mixed or negligible gains. Our framework provides a practical methodology for deriving and evaluating instance-level misclassification costs, even when cost-sensitive training offers limited benefit.

[LG-52] aming the Curses of Multiagency in Robust Markov Games with Large State Space through Linear Function Approximation

链接: https://arxiv.org/abs/2605.03125
作者: Jingchu Gai,Laixi Shi
类目: Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:Multi-agent reinforcement learning (MARL) holds great potential but faces robustness challenges due to environmental uncertainty. To address this, distributionally robust Markov games (RMGs) optimize worst-case performance when the environment deviates from the nominal model within a uncertainty set. Beyond robustness, an equally urgent goal for MARL is data efficiency – sampling from vast state and action spaces that grow exponentially with the number of agents potentially leads to the curse of multiagency. However, current provably data-efficient algorithms for RMGs are limited to tabular settings with finite state and action spaces, which are only computationally manageable for small-scale problems, leaving RMGs with large-scale (or infinite) state spaces largely unexplored. The only existing work beyond tabular settings focuses on linear function approximation (LFA) for a restrictive class of RMGs using vanish minimal value assumption and still suffers from sample complexity with the curse of multiagency. In this work, we focuses on general RMGs with LFA. For uncertainty sets defined by total variation distance, we develop provably data-efficient algorithms that break the curse of multiagency in both the generative model setting and a newly proposed online interactive setting. To our knowledge, our results are the first to break the curse of multiagency of sample complexity for RMGs with large (possibly infinite) state spaces, regardless of the uncertainty set construction.

[LG-53] Pose Tracking with a Foundation Pose Model and an Ensemble Directional Kalman Filter

链接: https://arxiv.org/abs/2605.03105
作者: Tianlu Lu,Asif Sijan,Thomas Noh,Huaijin Chen,Andrey A. Popov
类目: Machine Learning (cs.LG); Differential Geometry (math.DG); Applications (stat.AP)
*备注:

点击查看摘要

Abstract:This paper introduces the ensemble directional Kalman filter (EnDKF), an ensemble-based Kalman filtering approach for pose tracking that jointly estimates an object’s position and attitude using ideas from directional statistics. The EnDKF integrates a unit-quaternion attitude representation to move beyond canonical Kalman filter mean and covariance assumptions that poorly capture directional uncertainty. Experiments on a synthetic constant-velocity constant-angular-velocity system and a digital-twin head-tracking scenario using the FoundationPose algorithm demonstrate a significant reduction in error as opposed to merely using measurements.

[LG-54] Attribution-Guided Masking for Robust Cross-Domain Sentiment Classification

链接: https://arxiv.org/abs/2605.03091
作者: Shubham Harkare,Arvind Yogesh Suresh Babu,Yash Kulkarni
类目: Machine Learning (cs.LG)
*备注: 10 pages, 2 figures

点击查看摘要

Abstract:While pre-trained Transformer models achieve high accuracy on in-domain sentiment classification, they frequently experience severe performance degradation when transferring to out-of-domain data. We hypothesize that this generalization gap is driven by reliance on domain-specific spurious tokens. After demonstrating that post-hoc-token-level attribution drift fails to predict this gap, we propose Attribution-Guided Masking (AGM), a training time intervention that dynamically detects and penalizes highly attributed spurious tokens during fine-tuning. AGM’s core component is a gradient based attribution masking loss ( \mathcalL_mask ), which can optionally be combined with a counterfactual contrastive loss to enforce domain-invariant representations, all without requiring target-domain labels or human annotation. Evaluated in a strict zero-shot transfer setting across four diverse domains with eight random seeds, AGM achieves competitive generalization compared to five strong baselines on the hardest transfer (Sentiment140): \Delta = 0.244 versus DANN (0.264), DRO (0.248), Fish (0.247), and IRM (0.238), while uniquely providing token-level interpretability into which features drive the generalization gap. Our qualitative analysis confirms that AGM suppresses attribution on domain-specific tokens such as @mentions, hashtags, and slang, shifting reliance toward domain-invariant sentiment markers. Our ablation study further confirms that attribution-guided masking is the critical component: removing it or replacing it with random token selection consistently degrades performance on difficult transfers.

[LG-55] Adaptive Data Compression and Reconstruction for Memory-Bounded EEG Continual Learning

链接: https://arxiv.org/abs/2605.03085
作者: Chengcheng Xie
类目: Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:Electroencephalography (EEG) signals provide millisecond-level temporal resolution but their analysis is limited by remarkable noise and inter-subject variability, making robust personalization difficult under limited annotations. Unsupervised Individual Continual Learning (UICL) has been proposed to address this practical challenge, where a model pretrained on a labeled cohort must adapt online to unlabeled subject streams under strict memory constraints. However, existing UICL methods typically store full past samples, which undermine the continual learning goal of avoiding retraining. Observing that EEG signals exhibit well-structured morphologies to be exploited via morphology-aware selection, compression, and reconstruction, here we propose Adaptive Data Compression and Reconstruction (ADaCoRe) for UICL. This is a memory-efficient pipeline composed of saliency-driven keyframe protection, rational polyphase compression, adjoint reconstruction with verbatim overwrite on protected indices, and prototype-confidence selection for adaptive exemplar maintenance. Across three representative benchmarks, ADaCoRe consistently outperforms recent strong baselines under tight buffer regimes (eg., the performance gains are at least +2.7 and +15.3 ACC on ISRUC and FACED datasets, respectively). Ablation studies quantify compression-fidelity trade-offs and highlight the contribution of each design, while visualizations confirm the preservation of key EEG morphology during compression and reconstruction.

[LG-56] Phoneme-Level Deepfake Detection Across Emotional Conditions Using Self-Supervised Embeddings

链接: https://arxiv.org/abs/2605.03079
作者: Vamshi Nallaguntla,Shruti Kshirsagar,Anderson R. Avila
类目: ound (cs.SD); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS)
*备注: 6 pages, 2 figures, submitted to IEEE SMC 2026

点击查看摘要

Abstract:Recent advances in emotional voice conversion (EVC) have enabled the generation of expressive synthetic speech, raising new concerns in audio deepfake detection. Existing approaches treat speech as a homogeneous signal and largely overlook its internal phonetic structure, limiting their interpretability in emotionally conditioned settings. In this work, we propose a phoneme-level framework to analyze emotionally manipulated synthetic speech using real and EVC-generated speech under matched emotional conditions with shared transcripts, phoneme-aligned TextGrids, and WavLM-based embeddings. Our results show that phoneme behavior varies across categories, with complex vowels and fricatives exhibiting higher divergence while simpler phonemes remain more stable. Phonemes with larger distributional differences are also found to be more easily detected, consistently across multiple emotions and synthesis systems. These findings demonstrate that phoneme-level analysis is an effective and interpretable approach for detecting emotionally manipulated synthetic speech.

[LG-57] Adaptive Negative Scheduling for Graph Contrastive Learning

链接: https://arxiv.org/abs/2605.03076
作者: Adnan Ali,Jinlong Li,Syed Muhammad Israr,Ali Kashif Bashir
类目: Machine Learning (cs.LG)
*备注: 14 pages, 5 figures, 9 benchmark datasets, code available at GitHub

点击查看摘要

Abstract:Graph contrastive learning (GCL) has become a central paradigm for self-supervised representation learning in computational intelligence, with applications spanning recommendation, anomaly detection, and personalization. A key limitation of existing methods is their reliance on static negative sampling, which fails to account for the dynamic informativeness and computational cost of negatives during training. We propose AdNGCL, an adaptive negative scheduling framework with a hardness-aware scheduler (HANS) that formulates negative selection as a loss-gated, budget-constrained process across hard, intermediate, and easy strata. The scheduler dynamically adjusts step sizes based on contrastive loss trends under both global and per-category budgets, while periodically refreshing samples to maintain diversity without exceeding compute constraints. Experiments on nine benchmark graph datasets demonstrate that AdNGCL consistently advances state-of-the-art performance, achieving the best accuracy on seven datasets and second-best on the remaining two, while offering explicit control over computational cost. These results highlight the value of budget-aware, loss-sensitive scheduling as a general strategy for improving the robustness and efficiency of representation learning in emerging computational intelligence applications.

[LG-58] Distributed Deep Variational Approach for Privacy-preserving Data Release

链接: https://arxiv.org/abs/2605.03069
作者: Zahir Alsulaimawi,Huaping Liu
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:Federated learning (FL) lets distributed nodes train a shared model without exchanging their raw data, but in privacy-sensitive deployments medical sensors, IoT devices, wearables the protection offered by keeping data local is incomplete: gradients, model updates, and the released representations themselves can leak sensitive attributes. We propose the \emphGaussian Privacy Protector (GPP), a data-release framework for continuous, high-dimensional inputs that learns a stochastic encoder mapping raw data to a low-dimensional sanitized representation. The encoder is trained against a variational lower bound on the mutual information between the released representation and a designated sensitive attribute, while a separate cross-entropy term preserves a designated utility attribute, with a Lagrange multiplier \beta controlling the trade-off. We then extend GPP to the federated setting, in which each client trains a local encoder, sensitive labels never leave the client, and the aggregator receives only sanitized representations giving instance-level privacy protection in addition to the standard ``raw data stays local’’ guarantee of FL. We evaluate GPP on MNIST (digit-sum utility, parity sensitive), CelebA (smiling vs.\ gender), and HAPT-Recognition (activity vs.\ subject identity). Across all three benchmarks, GPP attains utility within roughly one percentage point of an unconstrained autoencoder baseline while reducing the adversary’s AUC to near random guessing.

[LG-59] OGPO: Sample Efficient Full-Finetuning of Generative Control Policies

链接: https://arxiv.org/abs/2605.03065
作者: Sarvesh Patil,Mitsuhiko Nakamoto,Manan Agarwal,Shashwat Saxena,Jesse Zhang,Giri Anantharaman,Cleah Winston,Chaoyi Pan,Douglas Chen,Nai-Chieh Huang,Zeynep Temel,Oliver Kroemer,Sergey Levine,Abhishek Gupta,Hongkai Da,Paarth Shah,Max Simchowitz
类目: Machine Learning (cs.LG); Robotics (cs.RO)
*备注:

点击查看摘要

Abstract:Generative control policies (GCPs), such as diffusion- and flow-based control policies, have emerged as effective parameterizations for robot learning. This work introduces Off-policy Generative Policy Optimization (OGPO), a sample-efficient algorithm for finetuning GCPs that maintains off-policy critic networks to maximize data reuse and propagate policy gradients through the full generative process of the policy via a modified PPO objective, using critics as the terminal reward. OGPO achieves state-of-the-art performance on manipulation tasks spanning multi-task settings, high-precision insertion, and dexterous control. To our knowledge, it is also the only method that can fine-tune poorly-initialized behavior cloning policies to near full task-success with no expert data in the online replay buffer, and does so with few task-specific hyperparameter tuning. Through extensive empirical investigations, we demonstrate the OGPO drastically outperforms methods alternatives on policy steering and learning residual corrections, and identify the key mechanisms behind its performance. We further introduce practical stabilizers, including success-buffer regularization, conservative advantages, \chi^2 regularization, and Q-variance reduction, to mitigate critic over-exploitation across state- and pixel-based settings. Beyond proposing OGPO, we conduct a systematic empirical study of GCP finetuning, identifying the stabilizing mechanisms and failure modes that govern successful off-policy full-policy improvement.

[LG-60] CD-Arena: Assessing Robustness of Time Series Causal Discovery Methods Against Assumption Violations

链接: https://arxiv.org/abs/2605.03045
作者: Gideon Stein,Niklas Penzel,Tristan Piater,Joachim Denzler
类目: Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:Causal Discovery (CD) is a powerful framework for scientific inquiry. Yet, its practical adoption is hindered by a reliance on strong, often unverifiable assumptions and a lack of robust performance assessment. To address these limitations and advance empirical CD evaluation, we present TCD-Arena, a modularized, highly customizable, and extendable testing kit to assess the robustness of time series CD algorithms against stepwise more severe assumption violations. For demonstration, we conduct an extensive empirical study comprising around 30 million individual CD attempts and reveal nuanced robustness profiles for 33 distinct assumption violations. Further, we investigate CD ensembles and find that they have the potential to improve general robustness, which has implications for real-world applications. With this, we strive to ultimately facilitate the development of CD methods that are reliable for a diverse range of synthetic and potentially real-world data conditions.

[LG-61] Joint Energy Management and Coordinated AIGC Workload Scheduling for Distributed Data Centers: A Diffusion-Aided Reward Shaping Approach

链接: https://arxiv.org/abs/2605.02965
作者: Yang Fu,Peng Qin,Liming Chen,Zihao Zhang,Hao Yu,Yifei Wang
类目: Machine Learning (cs.LG); Signal Processing (eess.SP); Systems and Control (eess.SY); Machine Learning (stat.ML)
*备注:

点击查看摘要

Abstract:Artificial intelligence-generated content (AIGC) has emerged as a transformative paradigm for automating the creation of diverse and customized content, giving rise to rapidly growing computational workloads in cloud data centers. It is imperative for AIGC service providers (ASPs) to strategically schedule AIGC workloads to reduce data center energy costs while guaranteeing high-quality content generation. However, the distinctive characteristics of AIGC services pose critical challenges, including model heterogeneity across ASPs, implicit service quality evaluation, and complex inference process control. To tackle these challenges, we propose a joint energy management and coordinated AIGC workload scheduling framework, which introduces an explicit mathematical characterization of service quality to promote both job transfer among ASPs and fine-grained inference process configuration. Moreover, various energy resources within data centers are jointly considered to enhance power usage flexibility. Subsequently, a system utility maximization problem is formulated to balance AIGC service revenue with operational penalties and costs. Nevertheless, the strong coupling among job scheduling decisions induces severe reward sparsity, which limits the effectiveness of existing deep reinforcement learning (DRL) algorithms. To address this issue, we develop a diffusion model-aided reward shaping approach to synthesize complementary reward signals through a multi-step denoising process. This approach is seamlessly integrated with DRL to enable efficient learning of scheduling policies under sparse environmental feedback. Experiments based on real-world models and datasets demonstrate that our scheme effectively accommodates electricity price fluctuations and AIGC model heterogeneity, while achieving superior learning convergence and system utility compared with benchmark methods.

[LG-62] ISAAC: Auditing Causal Reasoning in Deep Models for Drug-Target Interaction

链接: https://arxiv.org/abs/2605.02962
作者: Barbara Tarantino,Sun Kim,Yijingxiu Lu,Paolo Giudici
类目: Machine Learning (cs.LG); Computation (stat.CO); Machine Learning (stat.ML)
*备注: 11 Pages

点击查看摘要

Abstract:Deep learning models for drug–target interaction (DTI) prediction often achieve strong benchmark performance without necessarily relying on mechanistically meaningful molecular features, a limitation that standard accuracy-based evaluation cannot detect. We introduce ISAAC (Intervention-based Structural Auditing Approach for Causal Reasoning), a post-hoc framework that evaluates prior-relative structural sensitivity by probing frozen models through matched mechanistic and spurious input-level interventions, independently of predictive accuracy. Applied to three sequence-based DTI architectures on the Davis benchmark, ISAAC reveals approximately 25% relative differences in reasoning scores across models with comparable AUROC (within around 3%), stable across training and intervention seeds and two distinct perturbation operators. These discrepancies, undetectable under conventional accuracy metrics, motivate the use of post-hoc structural auditing as a complement to standard performance evaluation in scientific machine learning for molecular modeling.

[LG-63] ZeRO-Prefill: Zero Redundancy Overheads in MoE Prefill Serving

链接: https://arxiv.org/abs/2605.02960
作者: Zhaoyuan Su,Olatunji Ruwase,Karthik Ganesan,Aurick Qiao,Samyam Rajbhandari,Juncheng Yang,Yue Cheng,Yuxiong He
类目: Machine Learning (cs.LG)
*备注: 19 pages, 12 figures, 4 tables

点击查看摘要

Abstract:Production LLM workloads increasingly serve discriminative tasks, such as classification, recommendation, and verification, whose answers are read from the logits of a single prefill pass with no autoregressive decoding. Serving these prefill-only workloads on mixture-of-experts (MoE) models is bottlenecked not by compute but by the distributed execution required to fit the model: existing parallel strategies (tensor, expert, and pipeline parallelism) trade memory pressure for redundant computation, communication, and synchronization, severely degrading MoE prefill serving efficiency. We observe that these overheads stem from coupling expert placement with synchronous activation routing – a design inherited from the decoding era. The long, compute-bound forward passes of large-batch prefill open a per-layer window wide enough to stream expert weights in the background, replacing per-layer activation AllToAll with asynchronous weight AllGather fully overlapped with computation. We propose ZeRO-Prefill, a prefill-only serving system whose backend, AsyncEP (Asynchronous Expert Parallelism), gathers experts by weight rather than routing them by activation, and whose frontend co-enforces a physically-derived saturation threshold through prefix-aware routing and true-FLOPs load tracking. On Qwen3-235B-A22B across four hardware/precision configurations, ZeRO-Prefill delivers 1.35-1.37x throughput over the strongest distributed baseline on real-world workloads and up to 1.59x on long-context synthetic workloads, sustaining 29.8-36.2% per-GPU model FLOPs utilization.

[LG-64] Calibration of the underlying surface parameters for urban flood using latent variables and adjoint equation

链接: https://arxiv.org/abs/2605.02959
作者: Yongfu Tian,Shan Ding,Guofeng Su,Jianguo Chen
类目: Machine Learning (cs.LG)
*备注: 27 pages, 8 figures, 2 table, submitted to Journal of Flood Risk Management

点击查看摘要

Abstract:Calibrating the urban underlying surface parameters is crucial for urban flood simulation. We formulate the parameter calibration problem into an optimization problem within the Bayesian framework using the maximum likelihood principle. We adopt the urban flood dynamical system model as the surrogate model and innovatively introduce latent variables inspired by machine learning to represent more uncertainties, which can also be compatible with common physical parameter calibration. For more efficient optimization, we construct the adjoint equation of the surrogate model to obtain gradient information and propose the parameter sharing technique and the localization technique to reduce the computation complexity of the adjoint equation. A simple case verifies the proposed method can converge quickly and is insensitive to the observation time interval. In the case derived from Test 8A, we calibrate Manning’s coefficient of urban roads, with a maximum relative error of 13.88% and a minimum of 1.16%.

[LG-65] Disease Is a Spectral Perturbation

链接: https://arxiv.org/abs/2605.02949
作者: John D. Mayfield,Matthew S. Rosen
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:

点击查看摘要

Abstract:We propose a novel method of understanding disease transformation from a healthy baseline with biomarker-level explainability. By modeling the biomarker covariance matrices of healthy controls and disease states, the perturbation can be individually characterized to accomplish mechanistic explanations of disease trajectories, both at a molecular level and for individual patients. Given a cohort of n patients each measured on p biomarkers, we define the biomarker “Hamiltonian” H = X^T X / n \in R^p \times p, where X \in R^n \times p is the covariant biomarker matrix. The eigenvectors of H define a set of normal modes of biomarker coordination, and the eigenvalues quantify the energy carried by each mode. In the healthy state, the reference Hamiltonian H_0 governs this structure where disease perturbs H_0 by an additive operator \Delta H, thus shifting eigenvalues and rotating eigenvectors in proportion to the severity of pathological disruption. We formalize this framework, derive the spectral change given a disease perturbation, and demonstrate that the projection of a newly diagnosed patient’s cumulative biomarker covariance structure onto disease-discriminant eigenmodes constitutes an optimal prognostic statistic for greater precision in disease prognosis. This work serves as a veritable white paper with application across a panoply of disease frameworks from cancer to neurodegenerative disorders.

[LG-66] Analysis and Explainability of LLM s Via Evolutionary Methods

链接: https://arxiv.org/abs/2605.02930
作者: Shannon K. Gallagher,Swati Rallapalli,Tyler Brooks,Chuck Loughin,Michele Sezgin,Ronald Yurko
类目: Neural and Evolutionary Computing (cs.NE); Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:

点击查看摘要

Abstract:Evolutionary methods have long been useful for analysis and explanation in genetics, biology, ecology, and related fields. In this work, we extend these methods to neural networks, specifically large language models (LLMs), to better analyze and explain relationships among models. We show how relating weights to genotypes and output text to phenotypes can improve our understanding of model lineage, important datasets, the roles of different model layers, and visualization of model relationships. We demonstrate this in a controlled experiment, where our estimated evolutionary trees reliably recover the topology of the ground-truth training tree. We further identify the most important weight layers according to weight differences and show through phenotypic experiments that one training dataset appears to contribute more useful information than the others. Finally, we generate an unsupervised evolutionary tree of black-box foundation models. Throughout, we provide visualizations that support a clearer understanding of evolutionary relationships among LLMs.

[LG-67] Heterogeneous Graph Importance Scoring and Clustering with Automated LLM -based Interpretation

链接: https://arxiv.org/abs/2605.02919
作者: Takato Yasuno
类目: Machine Learning (cs.LG)
*备注: 26 pages, 11 figures, 8 tables

点击查看摘要

Abstract:Urban bridge networks are critical infrastructure whose disruption can cascade into severe impacts on transportation, emergency services, and economic activity. This paper presents a comprehensive methodology for assessing bridge importance through heterogeneous graph analysis, unsupervised clustering, and automated interpretation via large language models (LLMs). Our approach addresses three fundamental challenges: (1) quantifying multi-dimensional bridge importance using only open data sources, (2) discovering functional bridge archetypes across different cities, and (3) generating policy-relevant interpretations automatically. We construct heterogeneous graphs from OpenStreetMap (OSM) data incorporating bridges, road networks, buildings, and public facilities. Five social impact indicators are computed: transit desert score, hospital access score, isolation risk score, supply chain impact score, and green space access score. These 52-dimensional feature vectors undergo dimensionality reduction via UMAP and density-based clustering via HDBSCAN. Discovered clusters are interpreted using temperature-optimized LLMs (Elyza8b, trained on construction domain corpus). (1) A complete open-data pipeline from OSM to actionable bridge importance rankings, (2) a five-indicator scoring methodology with 40 \times computational optimization, (3) a UMAP+HDBSCAN clustering framework validated on multi-city data, (4) an LLM interpretation methodology including temperature optimization and model selection rationale, and (5) transferability demonstration across cities via configuration-only adaptation.

[LG-68] From Synthesis to Clinical Assistance: A Strategy-Aware Agent Framework for Autism Intervention based on Real Clinical Dataset ACL

链接: https://arxiv.org/abs/2605.02916
作者: Junhong Lai,Shuzhong Lai,Yanhao Yu,Wanlin Chen,Chenyu Yan,Haifeng Li,Lin Yao,Yueming Wang
类目: Machine Learning (cs.LG)
*备注: Accepted to 2026 ACL Main Conference

点击查看摘要

Abstract:The development of AI-assisted Early Intensive Behavioral Intervention (EIBI) for Autism Spectrum Disorder (ASD) is severely constrained by data scarcity. Furthermore, while Applied Behavior Analysis (ABA) serves as the gold standard for clinical intervention, general-purpose Large Language Models (LLMs) struggle to strictly adhere to its standardized procedures, often resulting in interactions that are linguistically fluent but strategically inconsistent. To address these challenges, we introduce \textscASDAgent, a strategy-aware framework designed to unify high-fidelity intervention dialogue synthesis and clinical decision support. \textscASDAgent incorporates two specialized components to solve distinct problems: (i) a \textscDoctorAgent equipped with an Observe-Think-Act-Correct (O-T-A-C) reasoning loop, which resolves the issue of strategy collapse in LLMs by making ABA execution explicit and controllable; and (ii) a \textscChildAgent that utilizes probabilistic behavior modeling to mitigate data homogeneity, simulating diverse and non-deterministic ASD response patterns. Experiments demonstrate that dialogues generated by \textscASDAgent closely mirror the strategy distribution of human therapists (KL divergence: 0.083). In real autism intervention, \textscASDAgent achieves nearly 80% strategic consistency with human experts. Moreover, we show that synthetic data produced by \textscASDAgent effectively distills professional clinical knowledge into small language models (SLMs), significantly enhancing their therapeutic capabilities.

[LG-69] Generate Filter Control Replay: A Comprehensive Survey of Rollout Strategies for LLM Reinforcement Learning

链接: https://arxiv.org/abs/2605.02913
作者: Rohan Surana,Gagan Mundada,Xunyi Jiang,Chuhan Wang,Zhenwei Tang,Difan Jiao,Zihan Huang,Yuxin Xiong,Junda Wu,Sheldon Yu,Xintong Li,Raghav Jain,Nikki Kuang,Sizhe Zhou,Bowen Jin,Zhendong Chu,Tong Yu,Ryan Rossi,Kuan-Hao Huang,Jingbo Shang,Jiawei Han,Julian McAuley
类目: Machine Learning (cs.LG)
*备注: 47 pages, 8 tables, 7 figures

点击查看摘要

Abstract:Reinforcement learning (RL) has become a central post-training tool for improving the reasoning abilities of large language models (LLMs). In these systems, the rollout, the trajectory sampled from a prompt to termination, including intermediate reasoning steps and optional tool or environment interactions, determines the data the optimizer learns from, yet rollout design is often underreported. This survey provides an optimizer-agnostic view of rollout strategies for RL-based post-training of reasoning LLMs. We formalize rollout pipelines with unified notation and introduce Generate-Filter-Control-Replay (GFCR), a lifecycle taxonomy that decomposes rollout pipelines into four modular stages: Generate proposes candidate trajectories and topologies; Filter constructs intermediate signals via verifiers, judges, critics; Control allocates compute and makes continuation/branching/stopping decisions under budgets; and Replay retains and reuses artifacts across rollouts without weight updates, including self-evolving curricula that autonomously generate new training tasks. We complement GFCR with a criterion taxonomy of reliability, coverage, and cost sensitivity that characterizes rollout trade-offs. Using this framework, we synthesize methods spanning RL with verifiable rewards, process supervision, judge-based gating, guided and tree/segment rollouts, adaptive compute allocation, early-exit and partial rollouts, throughput optimization, and replay/recomposition for self-improvement. We ground the framework with case studies in math, code/SQL, multimodal reasoning, tool-using agents, and agentic skill benchmarks that evaluate skill induction, reuse, and cross-task transfer. Finally, we provide a diagnostic index that maps common rollout pathologies to GFCR modules and mitigation levers, alongside open challenges for building reproducible, compute-efficient, and trustworthy rollout pipelines.

[LG-70] Agent ic AI-Based Joint Computing and Networking via Mixture of Experts and Large Language Models

链接: https://arxiv.org/abs/2605.02911
作者: Robert-Jeron Reifert,Alaa Alameer Ahmad,Hayssam Dahrouj,Aydin Sezgin
类目: Machine Learning (cs.LG); Information Theory (cs.IT)
*备注: 16 pages, 16 figures, 9 tables, a version of this work is due to be submitted to the IEEE for possible publication

点击查看摘要

Abstract:Future sixth-generation (6G) mobile networks are envisioned to be equipped with a diverse set of powerful, yet highly specialized, optimization experts. Such a promising vision is concurrently expected to give rise to the need for scalable mechanisms that can select, combine, and orchestrate such experts based on high-level intent and uncertainty descriptions. In this paper, we propose an agentic artificial intelligence (AI)-based network optimization framework that integrates mixture of experts (MoE) architectures with large language models (LLMs). Under the proposed framework, the employed LLM acts as a semantic gate to reason over operator objectives and dynamically compose suitable optimization agents. The proposed framework is formulated in a model-agnostic manner and bridges human-readable network intents with low-level resource allocation decisions, enabling flexible optimization across heterogeneous objectives and operating conditions. As a representative instantiation, we apply the framework to a joint communication and computing network and design a library of specialized optimization experts covering throughput, fairness, and delay-driven objectives under both regular and robust conditions. Numerical simulations demonstrate that the proposed agentic MoE framework consistently achieves near-optimal performance compared to exhaustive expert combinations while outperforming individual experts across diverse objectives, including delay minimization and throughput maximization.

[LG-71] An End-to-End Framework for Building Large Language Models for Software Operations

链接: https://arxiv.org/abs/2605.02906
作者: Jingkai He,Pengfei Chen,Chenghui Wu,Shuang Liang,Ye Li,Gou Tan,Xiadao Wen,Chuanfu Zhang
类目: Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:In the field of software operations, Large Language Models (LLMs) have attracted increasing attention. However, existing research has not yet achieved efficient and effective end-to-end intelligent operations due to low-quality data, fragmented knowledge and insufficient learning. To explore the potential of LLMs in software operations, we propose OpsLLM, a domain-specific LLM that supports both knowledge-based question answering (QA) and root cause analysis (RCA). Moreover, we disclose the detailed workflow for building LLMs specifically in the software operations domain. First, a Human-in-the-Loop mechanism is introduced to curate highquality data from a large collection of operational raw data and construct a fine-tuning dataset. Then, based on the data, supervised fine-tuning is conducted to achieve a base model. Furthermore, we introduce a domain process reward model (DPRM) during the reinforcement learning stage to optimize the accuracy and reliability of the fine-tuned model on RCA tasks. Experimental results on the tasks with diverse difficulties demonstrate that OpsLLMs effectively learns and aligns with the operational domain knowledge infused, outperforming existing open-source and closed-source LLMs in accuracy with improvements of 0.2%~5.7% on QA tasks and 2.7% ~70.3% on RCA tasks, while exhibiting strong transferability. Moreover, we will open-source three versions of OpsLLM with 7B, 14B and 32B parameters, along with a 15K fine-tuning dataset.

[LG-72] OptShrinkQ: Near-Lossless KV Cache Compression Through Optimal Spectral Denoising and Quantization

链接: https://arxiv.org/abs/2605.02905
作者: Pei-Chun Su
类目: Machine Learning (cs.LG); Information Theory (cs.IT)
*备注:

点击查看摘要

Abstract:We show that the key-value (KV) cache in transformer attention heads admits a natural decomposition into a low-rank \emphshared context component and a full-rank \emphper-token residual, well described by the spiked random matrix model. This observation leads to eOptShrinkQ, a two-stage compression pipeline: optimal singular value shrinkage (eOptShrink) automatically extracts the shared structure, and the residual – which satisfies the \emphthin shell property with delocalized coordinates – is quantized by TurboQuant~\citepzandieh2025turboquant, a recently proposed per-vector scalar quantizer with near-optimal distortion guarantees. By restoring the isotropy that scalar quantization assumes, spectral denoising eliminates the need for both outlier handling and dedicated inner product bias correction, freeing those bits for improved reconstruction. The theoretical grounding in random matrix theory provides three guarantees: automatic rank selection via the BBP phase transition, provably near-zero inner product bias on the residual, and coordinate delocalization ensuring near-optimal quantization distortion. Experimentally, we validate eOptShrinkQ on Llama-3.1-8B and Ministral-8B across three levels: per-head MSE and inner product fidelity, where eOptShrinkQ saves nearly one bit per entry over TurboQuant at equivalent quality; end-to-end on LongBench (16 tasks), where eOptShrinkQ at \sim 2.2 bits per entry outperforms TurboQuant at 3.0 bits; and multi-needle retrieval, where eOptShrinkQ at 2.2 bits closely matches or exceeds uncompressed FP16, suggesting that spectral denoising can act as a beneficial regularizer for retrieval-intensive tasks. Subjects: Machine Learning (cs.LG); Information Theory (cs.IT) Cite as: arXiv:2605.02905 [cs.LG] (or arXiv:2605.02905v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2605.02905 Focus to learn more arXiv-issued DOI via DataCite

[LG-73] StateSMix: Online Lossless Compression via Mamba State Space Models and Sparse N-gram Context Mixing

链接: https://arxiv.org/abs/2605.02904
作者: Roberto Tacconelli
类目: Machine Learning (cs.LG); Information Theory (cs.IT)
*备注: 10 pages

点击查看摘要

Abstract:We present StateSMix, a fully self-contained lossless compressor that couples an online-trained Mamba-style State Space Model (SSM) with sparse n-gram context mixing and arithmetic coding. The model is initialised from scratch and trained token-by-token on the file being compressed, requiring no pre-trained weights, no GPU, and no external dependencies. The SSM (DM=32, NL=2, approximately 120K active parameters per file) provides a continuously-updated probability estimate over BPE tokens, while nine sparse n-gram hash tables (bigram through 32-gram, 16M slots each) add exact local and long-range pattern memorisation via a softmax-invariant logit-bias mechanism that updates only non-zero-count tokens. An entropy-adaptive scaling mechanism modulates the n-gram contribution based on the SSM’s predictive confidence, preventing over-correction when the neural model is already well-calibrated. On the standard enwik8 benchmark, StateSMix achieves 2.123 bpb on 1 MB, 2.149 bpb on 3 MB, and 2.162 bpb on 10 MB, beating xz -9e (LZMA2) by 8.7%, 5.4%, and 0.7% respectively. Ablation experiments establish the SSM as the dominant compression engine: it alone accounts for a 46.6% size reduction over a frequency-count baseline and beats xz without any n-gram component, while n-gram tables provide a complementary 4.1% gain through exact context memorisation. OpenMP parallelisation of the training loop yields 1.9x speedup on 4 cores. The system is implemented in pure C with AVX2 SIMD and processes approximately 2,000 tokens per second on commodity x86-64 hardware.

[LG-74] Conditional Diffusion Sampling ICML2026

链接: https://arxiv.org/abs/2605.04013
作者: Francisco M. Castro-Macías,Pablo Morales-Álvarez,Saifuddin Syed,Daniel Hernández-Lobato,Rafael Molina,José Miguel Hernández-Lobato
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: ICML 2026

点击查看摘要

Abstract:Sampling from unnormalized multimodal distributions with limited density evaluations remains a fundamental challenge in machine learning and natural sciences. Successful approaches construct a bridge between a tractable reference and the target distribution. Parallel Tempering (PT) serves as the gold standard, while recent diffusion-based approaches offer a continuous alternative at the cost of neural training. In this work, we introduce Conditional Diffusion Sampling (CDS), a framework that combines these two paradigms. To this end, we derive Conditional Interpolants, a class of stochastic processes whose transport dynamics are governed by an exact, closed-form stochastic differential equation (SDE), requiring no neural approximation. Although these dynamics require sampling from a non-trivial initialization distribution, we show both theoretically and empirically that the cost of this initialization diminishes for sufficiently short diffusion times. CDS leverages this by a two-stage procedure: (1) PT is used to efficiently sample the initial distribution, and then (2) samples are transported via the transport SDE. This combination couples the robust global exploration of PT with efficient local transport. Experiments suggest that CDS has the potential to achieve a superior trade-off between sample quality and density evaluation cost compared to state-of-the-art samplers.

[LG-75] Exact ReLU realization of tensor-product refinement iterates

链接: https://arxiv.org/abs/2605.03917
作者: Tsogtgerel Gantumur
类目: Classical Analysis and ODEs (math.CA); Machine Learning (cs.LG)
*备注: 22 pages, 2 figures

点击查看摘要

Abstract:We study scalar dyadic refinement operators on R^2 of the form (Vf)(x,y) = sum_(j,k) in Z^2 c_j,k f(2x-j, 2y-k), where only finitely many mask coefficients c_j,k are nonzero. Under a fixed support-window hypothesis, we prove that for every compactly supported continuous piecewise linear seed g:R^2-R, the iterates V^n g admit exact ReLU realizations of fixed width and depth O(n). This gives a first genuinely two-dimensional extension of the exact realization theory for refinement cascades. Using the one-dimensional exact loop-controller framework, the proof transports the tensor-product residual dynamics exactly on the product of two polygonal loops and reduces the remaining seam ambiguity to a final readout and selector step. The matrix cascade is then handled by a fixed-depth recursive block, and general compactly supported continuous piecewise linear seeds are reduced to a finite decomposition together with exact clamped gluing on the support window. This identifies the tensor-product dyadic case as a natural first multivariate instance of the loop-controller method for refinement iterates. Comments: 22 pages, 2 figures Subjects: Classical Analysis and ODEs (math.CA); Machine Learning (cs.LG) MSC classes: 41A46, 41A30, 68T07 Cite as: arXiv:2605.03917 [math.CA] (or arXiv:2605.03917v1 [math.CA] for this version) https://doi.org/10.48550/arXiv.2605.03917 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[LG-76] Graph Neural Networks in the Wilson Loop Representation of Abelian Lattice Gauge Theories

链接: https://arxiv.org/abs/2605.03901
作者: Ali Rayat,Gia-Wei Chern
类目: rongly Correlated Electrons (cond-mat.str-el); Machine Learning (cs.LG); High Energy Physics - Lattice (hep-lat); Quantum Physics (quant-ph)
*备注: 13 pages, 6 figures

点击查看摘要

Abstract:Local gauge structures play a central role in a wide range of condensed matter systems and synthetic quantum platforms, where they emerge as effective descriptions of strongly correlated phases and engineered dynamics. We introduce a gauge-invariant graph neural network (GNN) architecture for Abelian lattice gauge models, in which symmetry is enforced explicitly through local gauge-invariant inputs, such as Wilson loops, and preserved throughout message passing, eliminating redundant gauge degrees of freedom while retaining expressive power. We benchmark the approach on both \mathbbZ_2 and \mathrmU(1) lattice gauge models, achieving accurate predictions of global observables and spatially resolved quantities despite the nonlocal correlations induced by gauge-matter coupling. We further demonstrate that the learned model serves as an efficient surrogate for semiclassical dynamics in \mathrmU(1) quantum link models, enabling stable and scalable time evolution without repeated fermionic diagonalization, while faithfully reproducing both local dynamics and statistical correlations. These results establish gauge-invariant message passing as a compact and physically grounded framework for learning and simulating Abelian lattice gauge systems.

[LG-77] he Manokhin Probability Matrix: A Diagnostic Framework for Classifier Probability Quality

链接: https://arxiv.org/abs/2605.03816
作者: Valery Manokhin
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:The Brier score conflates two distinct properties of probabilistic predictions: reliability (calibration error) and resolution (discriminatory power). We introduce the Manokhin Probability Matrix, a BCG-style two-dimensional diagnostic framework that separates them. Classifiers are placed on a 2x2 grid by Spiegelhalter Z-statistic and AUC-ROC expected rank, then assigned to one of four archetypes: Eagle (good on both axes), Bull (strong discrimination, poor calibration), Sloth (well-calibrated, weak discriminator), and Mole (poor on both). Each archetype carries a distinct prescription. We populate the matrix from a large-scale empirical study spanning 21 classifiers, 5 post-hoc calibrators, and 30 real-world binary classification tasks from the TabArena-v0.1 suite. The assignment is unambiguous. CatBoost, TabICL, EBM, TabPFN, GBC, and Random Forest are Eagles. XGBoost, LightGBM, and HGB are Bulls; Venn-Abers calibration cuts log-loss by 6.5 to 12.6% on Bulls but degrades Eagles by 2.1%. SVM, LR, LDA, and the empirical base-rate predictor are Sloths. MLP, KNN, Naive Bayes, and ExtraTrees are Moles. A theoretical asymmetry follows: no order-preserving post-hoc calibrator can add discriminatory power (Proposition 1), so calibration is the fixable part and discrimination is the hard part. The practical rule is direct: do not optimise aggregate Brier score without first decomposing it; optimise discrimination first, then fix calibration post-hoc. Code and raw experimental data are available at this https URL.

[LG-78] owards accurate extreme event likelihoods from diffusion model climate emulators

链接: https://arxiv.org/abs/2605.03802
作者: Peter Manshausen,Noah Brenowitz,Julius Berner,Karthik Kashinath,Mike Pritchard
类目: Atmospheric and Oceanic Physics (physics.ao-ph); Machine Learning (cs.LG)
*备注: 19 pages, 6 figures

点击查看摘要

Abstract:ML climate model emulators are useful for scenario planning and adaptation, allowing for cost-efficient experimentation. Recently, the diffusion model Climate in a Bottle (cBottle) has been proposed for generation of atmospheric states compatible with boundary conditions of solar position and sea surface temperatures. Crucially, cBottle can be guided to generate extreme events such as Tropical Cyclones (TCs) over locations of interest. Diffusion models such as cBottle work by approximating the probability density of the training data. Here, we show use cases of the probability density estimates of atmospheric states obtained from this climate emulator. Most importantly, these estimates allow us to calculate likelihoods of extreme events under guidance. When guiding the model towards states including TCs, comparing the probability density under the guided and unguided model enables us to quantify how much more likely the guidance has made the TC. We show how these odds ratios allow us to importance-sample from the TC distribution, reducing the standard error of the probability estimate compared to simple Monte Carlo sampling. Furthermore, we discuss results and limitations of the application of model probability densities to extreme event attribution-like experiments. We present these early but encouraging results hoping they will spur more research into probabilistic information that can be gained from diffusion models of the atmosphere.

[LG-79] raining-Free Probabilistic Time-Series Forecasting with Conformal Seasonal Pools

链接: https://arxiv.org/abs/2605.03789
作者: Valery Manokhin
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:We propose Conformal Seasonal Pools (CSP), a training-free probabilistic time-series forecaster that mixes same-season empirical draws with signed residual draws around a seasonal naive forecast. In an audited rolling-origin benchmark on the six time-series datasets where DeepNPTS was originally evaluated (electricity, exchange_rate, solar_energy, taxi, traffic, wikipedia), CSP-Adaptive significantly outperforms DeepNPTS on every metric we report – CRPS (per-window paired Wilcoxon p \approx 4 \times 10^-10 ), normalized mean quantile loss ( p \approx 7 \times 10^-10 ), and empirical 95% coverage ( p \approx 8 \times 10^-45 , mean 0.89 vs 0.66) – while running over 500x faster on CPU. Coverage is the most decision-critical of these: a 0.95 nominal interval that contains the truth in only ~66% of cases fails the basic calibration desideratum and would not survive deployment in safety- or decision-critical settings. The failure mode is also more severe than aggregate coverage suggests: in the worst 10% of windows, DeepNPTS’s prediction interval covers none of the H forecast horizons – the entire multi-step trajectory misses the truth at every step simultaneously. This poses serious risk in safety- and decision-critical applications such as healthcare, finance, energy operations, and autonomous systems, where prediction intervals that systematically miss the truth across the entire planning horizon translate directly into misclassified patients, regulatory capital failures, grid imbalances, and safety-case violations. CSP achieves all of this with no learned parameters and no training. We argue training-free conformal samplers should be mandatory baselines when evaluating learned non-parametric forecasters.

[LG-80] Low Rank Tensor Completion via Adaptive ADMM

链接: https://arxiv.org/abs/2605.03736
作者: Niclas Führling,Getuar Rexhepi,Giuseppe Thadeu Freitas de Abreu
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注:

点击查看摘要

Abstract:We consider a novel algorithm, for the completion of partially observed low-rank tensors, as a generalization of matrix completion. The proposed low-rank tensor completion (TC) method builds on the conventional nuclear norm (NN) minimization-based low-rank TC paradigm, by leveraging the alternating direction method of multipliers (ADMM) optimization framework. To that extend the original NN minimization problem is reformulated into multiple subproblems, which are then solved iteratively via closed-form proximal operators, making use of over-relaxation and an adaptive penalty parameter update scheme, to further speed up convergence and improve the overall performance of the method. Simulation results demonstrate the superior performance of the new method in terms of normalized mean square error (NMSE), compared to the conventional state-of-the-art (SotA) techniques, including NN minimization approaches, as well as a mixture of the latter with a matrix factorization approach, while its convergence can be significantly improved by initializing the algorithm with the solution of the SotA.

[LG-81] Predicting missing values: A good idea?

链接: https://arxiv.org/abs/2605.03733
作者: Stef van Buuren
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME)
*备注: 16 pages, including R code, 1 figure, 2 tables

点击查看摘要

Abstract:Minimizing the Mean Squared Error (MSE) is a key objective in machine learning and is commonly used for imputing missing values. While this approach provides accurate point estimates, it introduces systematic biases in downstream analyses. These biases affect key parameters such as variance, prevalence, correlation, slope, and explained variance. The root cause is that imputed values optimized for MSE are averages, which reduce the natural variability in the data. This paper demonstrates that adding noise to imputed values can effectively eliminate these biases. The required noise level is proportional to the MSE. Using a toy example in a multivariate normal setting, we compare two methods: predictive imputation, which minimizes MSE, and stochastic imputation, which incorporates random noise. Simulation results show that predictive methods systematically introduce bias, while stochastic methods preserve the data’s natural variability and produce unbiased estimates. We also evaluate three popular imputation tools – missForest, softImpute, and mice – and observe consistent biases in predictive methods. These findings highlight that MSE is an inadequate measure of imputation quality, as it prioritizes accuracy over variability. Incorporating noise into imputation methods is essential to prevent biases and ensure valid downstream analyses, underscoring the importance of stochastic approaches for handling incomplete data. Comments: 16 pages, including R code, 1 figure, 2 tables Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME) Cite as: arXiv:2605.03733 [stat.ML] (or arXiv:2605.03733v1 [stat.ML] for this version) https://doi.org/10.48550/arXiv.2605.03733 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[LG-82] mpered Guided Diffusion

链接: https://arxiv.org/abs/2605.03712
作者: Andreas Makris,Paul Fearnhead,Chris Nemeth
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:Training-free conditional diffusion provides a flexible alternative to task-specific conditional model training, but existing samplers often allocate computation inefficiently: independent guided trajectories can vary widely in quality, and additional function evaluations along a single trajectory may not recover from poor early decisions. We propose Tempered Guided Diffusion (TGD), an annealed sequential Monte Carlo framework for training-free conditional sampling with diffusion priors. TGD targets tempered posterior distributions over the clean signal, using noisy diffusion states only as auxiliary variables for proposing reconstructions and propagating particles. Particles are reweighted by incremental likelihood ratios, resampled, and propagated across noise levels, concentrating computation on trajectories plausible under both the prior and observation. Under idealized exact-reconstruction assumptions, full TGD yields a consistent particle approximation to the posterior as the number of particles grows. For expensive reconstruction tasks, Accelerated TGD (A-TGD) retains early particle exploration but prunes to a single high-likelihood trajectory partway through sampling. Experiments on a controlled two-dimensional inverse problem and image inverse problems show improved posterior approximation and favorable wall-clock speed-quality tradeoffs over independent multi-trajectory baselines.

[LG-83] Free Decompression with Algebraic Spectral Curves

链接: https://arxiv.org/abs/2605.03634
作者: Siavash Ameli,Chris van der Heide,Liam Hodgkinson,Michael W. Mahoney
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Numerical Analysis (math.NA)
*备注:

点击查看摘要

Abstract:Tools from random matrix theory have become central to deep learning theory, using spectral information to provide mechanisms for modeling generalization, robustness, scaling, and failure modes. While often capable of modeling empirical behavior, practical computations are limited by matrix size, often imposing a restriction to models that are too small to be realistic. This motivates the inference of properties of larger models from the behavior of smaller ones. Free decompression (FD) is a recently proposed method for extrapolating spectral information across matrix sizes, but its utility is currently limited by strong assumptions that preclude its implementation on more realistic machine learning (ML) models. We use algebraic spectral curve theory to provide a general FD methodology for spectral densities whose Stieltjes transform satisfies an algebraic relation, a modeling assumption that is more likely to hold in practice. This recasts FD as an evolution along spectral curves which can be readily integrated. Our framework enables the expansion of spectral densities that have multiple or multi-modal bulks, that exist at multiple scales, and that contain atoms, all characteristic of real-world data and popular ML models. We demonstrate the efficacy of our framework on models of interest in modern ML, including Hessian and activation matrices associated with neural networks and large-scale diffusion models.

[LG-84] Expanding functional protein sequence space using high entropy generative models

链接: https://arxiv.org/abs/2605.03578
作者: Roberto Netti,Emily Hinds,Francesco Calvanese,Rama Ranganathan,Martin Weigt,Francesco Zamponi
类目: Quantitative Methods (q-bio.QM); Statistical Mechanics (cond-mat.stat-mech); Machine Learning (cs.LG)
*备注: 12 pages, 4 figures + Supplementary Information

点击查看摘要

Abstract:Boltzmann Machines trained on evolutionary sequence data have emerged as a powerful paradigm for the data-driven design of artificial proteins. However, the relationship between model architecture, specifically parameter density, and experimental performance remains poorly understood. Here, we investigate this relationship using the Chorismate Mutase enzyme family as a model system. We compare standard fully connected Boltzmann Machines for Direct Coupling Analysis (bmDCA) with sparse models generated via progressive edge activation (eaDCA) and edge decimation (edDCA). We identify a maximum-entropy model (meDCA) along the decimation trajectory that represents an optimal balance between constraint satisfaction and the flexibility of the probability distribution. We synthesized and tested artificial sequences from all models using an in vivo complementation assay, finding that all architectures, regardless of sparsity, generate functional enzymes with high success rates, even at significant divergence from natural sequences. Despite this functional equivalence, we demonstrate that the meDCA model samples a viable sequence space that is more than fifteen orders of magnitude larger than its low-entropy counterparts. Furthermore, comparative analyses reveal that high-entropy models systematically minimize overfitting and better capture the local neutral spaces surrounding natural proteins. These findings suggest that while various models satisfying coevolutionary statistics can generate functional sequences, high-entropy Boltzmann Machines provide a superior representation of the underlying evolutionary fitness landscape.

[LG-85] Stochastic Schrödinger Diffusion Models for Pure-State Ensemble Generation

链接: https://arxiv.org/abs/2605.03573
作者: Jian Xu,Wei Chen. Chao Li,Jingyuan Zheng,Delu Zeng,John Paisley,Qibin Zhao
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:In quantum machine learning (QML), classical data are often encoded as quantum pure states and processed directly as quantum representations, motivating representation-level generative modeling that samples new quantum states from an underlying pure-state ensemble rather than re-preparing them from perturbed classical inputs. However, extending \emphscore-based diffusion models with well-defined reverse-time samplers to quantum pure-state ensembles remains challenging, due to the non-Euclidean geometry of the complex projective space \mathbbCP^d-1 and the intractability of transition densities. We propose \emphStochastic Schrödinger Diffusion Models (SSDMs), an intrinsic score-based generative framework on \mathbbCP^d-1 endowed with the Fubini–Study (FS) metric. SSDMs formulate a forward Riemannian diffusion with a stochastic Schrödinger equation (SSE) realization, and derive reverse-time dynamics driven by the Riemannian score \nabla_\mathrmFS \log p_t . To enable training without analytic transition densities, we introduce a local-time objective based on a local Euclidean Ornstein–Uhlenbeck approximation in FS normal coordinates, yielding an analytic teacher score mapped back to the manifold. Experiments show that SSDMs faithfully capture target pure-state ensemble statistics, including observable moments, overlap-kernel MMD, and entanglement measures, and that SSDM-generated quantum representations improve downstream QML generalization via representation-level data augmentation.

[LG-86] StreakMind: AI detection and analysis of satellite streaks in astronomical images with automated database integration

链接: https://arxiv.org/abs/2605.03429
作者: Rafael Carrillo Navarro,René Duffard,Pablo García-Martín,Javier Romero,Nicolás Morales,Luis Gonçalves
类目: Instrumentation and Methods for Astrophysics (astro-ph.IM); Machine Learning (cs.LG)
*备注: Published in Astronomy Astrophysics, 708, A211 (2026), DOI: https://doi.org/10.1051/0004-6361/202558754

点击查看摘要

Abstract:Artificial satellites and space debris increasingly contaminate astronomical images, affecting scientific surveys and producing large volumes of streaked exposures. Manual inspection is no longer feasible at scale, and reliable detection and characterisation of streaks has become essential for both data-quality control and the monitoring of objects in Earth orbit. We present StreakMind, an automated pipeline designed to detect Near-Earth Objects and satellite streaks in astronomical images, characterise their geometry, and cross-identify them with known orbital objects. The system integrates all inference results into a structured database suitable for large surveys. A YOLO OBB model was trained on a hybrid dataset of 2335 images and applied to processed FITS frames. Geometric refinement, inter-frame association, satellite cross-identification, and Gaussian-based confidence scoring were then used to produce final identifications stored in a relational database. Observations from La Sagra Observatory were used to develop and test the method. On the test set, the model achieved a precision of 94 percent and a recall of 97 percent. It reliably detected faint streaks, delivered consistent geometric reconstructions, and performed robust satellite cross-identification. StreakMind demonstrates strong potential for large-scale automated analysis of linear streaks produced by both Near-Earth Objects and artificial satellites, contributing to space situational awareness.

[LG-87] Adaptive Estimation and Optimal Control in Offline Contextual MDPs without Stationarity

链接: https://arxiv.org/abs/2605.03393
作者: Riddhiman Bhattacharyya,Sayak Chakrabarty,Imon Banerjee
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 28 pages, Published in TMLR

点击查看摘要

Abstract:Contextual MDPs are powerful tools with wide applicability in areas from biostatistics to machine learning. However, specializing them to offline datasets has been challenging due to a lack of robust, theoretically backed methods. Our work tackles this problem by introducing a new approach towards adaptive estimation and cost optimization of contextual MDPs. This estimator, to the best of our knowledge, is the first of its kind, and is endowed with strong optimality guarantees. We achieve this by overcoming the key technical challenges evolving from the endogenous properties of contextual MDPs; such as non-stationarity, or model irregularity. Our guarantees are established under complete generality by utilizing the relatively recent and powerful statistical technique of T -estimation (Baraud, 2011). We first provide a procedure for selecting an estimator given a sample from a contextual MDP and use it to derive oracle risk bounds under two distinct, but nevertheless meaningful, loss functions. We then consider the problem of determining the optimal control with the aid of the aforementioned density estimate and provide finite sample guarantees for the cost function.

[LG-88] A-CODE: Fully Atomic Protein Co-Design with Unified Multimodal Diffusion

链接: https://arxiv.org/abs/2605.03360
作者: Chaoran Cheng,Jiaqi Guan,Milong Ren,Chengyue Gong,Cong Liu,Xinshi Chen,Ge Liu,Wenzhi Xiao
类目: Quantitative Methods (q-bio.QM); Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:We present A-CODE, a fully atomic unified one-stage protein co-design model that simultaneously refines discrete atom types and continuous atom coordinates. Unlike predominant two-stage methods that cascade structure design with amino acid-level sequence design, our approach is fully atomic within a unified multimodal diffusion framework, in which residue identities are inferred solely from atom-level predictions. Built upon the powerful all-atom architecture, A-CODE achieves superior designability for unconditional protein generation, outperforming all existing one-stage and two-stage design models. For binder design, A-CODE rivals and even outperforms existing state-of-the-art two-stage design models and, compared with the existing one-stage co-design model, achieves a drastic tenfold improvement in success rate on hard tasks. The inherent flexibility of our atomic formulation enables, for the first time, seamless adaptation to non-canonical amino acid (ncAA) modeling. Our fully atomic framework establishes a new, versatile foundation for all-atom generative modeling that can be naturally extended to complex biomolecular systems.

[LG-89] Imbalanced Classification under Capacity Constraints

链接: https://arxiv.org/abs/2605.03289
作者: Daniel Fraiman,Ricardo Fraiman
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Statistics Theory (math.ST)
*备注:

点击查看摘要

Abstract:In many classification settings, the class of primary interest is underrepresented, leading to imbalanced data problems that arise in applications such as rare disease detection and fraud identification. In these contexts, identifying a potential positive instance typically triggers costly follow-up actions, such as medical imaging or detailed transaction inspection, which are subject to limited operational capacity. Motivated by this setting, we consider classification problems where data may arrive sequentially and decisions must be made under constraints on the number of instances that can be selected for further analysis. We propose a classification framework that explicitly controls the rate of positive predictions, enforcing a user-defined bound on the proportion of observations classified as belonging to the minority class while maximizing detection performance. The approach can be implemented using standard learning methods and naturally extends to online settings, where decisions are taken in real time. We show that incorporating capacity constraints leads to substantial improvements over classical approaches, including resampling techniques such as SMOTE, which do not directly control the selection rate.

[LG-90] Donor-Aware scRNA-seq Benchmarks for IBD Classification

链接: https://arxiv.org/abs/2605.03281
作者: Jonathan Muhire
类目: Quantitative Methods (q-bio.QM); Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 15 pages, 5 figures, 4 tables. Independent study at Oklahoma Christian University, advised by Fang Li. Code: this https URL

点击查看摘要

Abstract:Donor-level disease classification from single-cell RNA sequencing (scRNA-seq) requires strict donor-aware cross-validation: naive pipelines that split cells randomly conflate training and test donors, inflating reported performance through pseudoreplication. We present a donor-aware benchmark evaluating three feature representations across two independent IBD cohorts: centered log-ratio (CLR) transformed cell-type composition, GatedStructuralCFN dependency embeddings, and scVI variational autoencoder latent embeddings. The cohorts are the SCP259 ulcerative colitis atlas (UC vs. Healthy, n=30 donors, 51 cell types) and the Kong 2023 Crohn’s disease atlas (CD vs. Healthy, n=71 donors, 55-68 cell types across three intestinal regions). Compartment-stratified CLR composition achieves AUROC 0.956 +/- 0.061 on SCP259; GatedStructuralCFN on the same features achieves 0.978 +/- 0.050. In the Kong cohort, CFN achieves its best performance in the colon region (0.960 +/- 0.055 after feature filtering), exceeding linear CLR (0.900 +/- 0.100), while terminal ileum classification is dominated by linear models (CatBoost CLR 0.967 +/- 0.075 vs. CFN 0.811 +/- 0.164). Cross-dataset transfer (CD-UC, four shared cell types) achieves AUC 0.833 with XGBoost CLR; the reverse direction performs at chance. CFN edge stability analysis shows that compartment-wise composition eliminates spurious unit-sum-induced instability present in global composition (Jaccard 0.026 vs. top-20 recurrence 1.0). CFN shows a consistent numerical advantage over linear models in the colon region of CD (AUROC 0.960 vs. 0.900), though no inter-method comparison reached statistical significance at n=34 donors per region. Compartment-aware feature construction is critical for both classification performance and structural interpretability. Code: this https URL Comments: 15 pages, 5 figures, 4 tables. Independent study at Oklahoma Christian University, advised by Fang Li. Code: this https URL Subjects: Quantitative Methods (q-bio.QM); Machine Learning (cs.LG); Machine Learning (stat.ML) Cite as: arXiv:2605.03281 [q-bio.QM] (or arXiv:2605.03281v1 [q-bio.QM] for this version) https://doi.org/10.48550/arXiv.2605.03281 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

[LG-91] Partial Effective Information Decomposition for Synergistic Causality

链接: https://arxiv.org/abs/2605.03267
作者: Mingzhe Yang,Shuo Wang,Jiang Zhang
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Data Analysis, Statistics and Probability (physics.data-an); Physics and Society (physics.soc-ph)
*备注:

点击查看摘要

Abstract:Causality is a central topic in scientific inquiry, yet for complex systems, the identification and analysis of synergistic causation remain a challenging and fundamental problem. In the context of causal relations among multivariate variables, a decomposition framework grounded in interventionist causation is still lacking. To address this gap, this paper proposes Partial Effective Information Decomposition (PEID), a framework that decomposes the influence of multiple source variables on a target variable under maximum-entropy interventions into unique and synergistic information, thereby providing a unified and computable characterization of synergistic causal relations. Theoretically, in the three-variable case, the proposed framework is compatible with the major axioms of Partial Information Decomposition (PID). Empirically, under maximum-entropy interventions, correlations among input variables are removed, causing redundancy to vanish and thereby enabling PEID to compute synergistic relations. Furthermore, based on this framework, it is possible to define causal graphs containing hyperedges as well as downward causation, thus offering a unified toolkit for analyzing cross-scale and multivariate causal mechanisms in complex systems. Finally, applying the framework to a machine-learning-based air quality forecasting task on KnowAir-V2, we demonstrate that PEID can extract interpretable inter-station causal structures from a learned dynamical model. These results suggest that PEID provides a general interventionist information-theoretic tool for analyzing multivariate and synergistic causal mechanisms in complex systems.

[LG-92] Intrinsic effective sample size for manifold-valued Markov chain Monte Carlo via kernel discrepancy

链接: https://arxiv.org/abs/2605.03266
作者: Kisung You
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Statistics Theory (math.ST); Computation (stat.CO); Methodology (stat.ME)
*备注:

点击查看摘要

Abstract:Effective sample size is a standard summary of Markov chain Monte Carlo output, but it is usually attached to scalar or Euclidean summaries chosen by the analyst. For manifold-valued samples this choice is not canonical: coordinate-wise effective sample sizes can change under rotations, chart changes, or alternative embeddings of the same underlying path. We propose an intrinsic effective sample size based on kernel discrepancy. The proposed quantity is the number of independent draws that would yield the same expected squared kernel discrepancy between the empirical distribution and the target distribution. This gives an exact finite-sample risk interpretation, an asymptotic integrated-autocorrelation representation, and a coordinate-free diagnostic whenever the kernel respects the geometry of the state space. We establish invariance under transported kernels, operator and principal-direction interpretations, and consistency of a lag-window estimator under boundedness and absolute-regularity conditions. We also discuss valid kernel constructions on manifolds, emphasizing that geodesic Gaussian kernels are not generally positive definite on curved spaces. Sphere experiments illustrate rotation invariance and calibration of the proposed diagnostic against empirical distributional error.

[LG-93] Conformalized Percentile Interval: Finite Sample Validity and Improved Conditional Performance

链接: https://arxiv.org/abs/2605.03233
作者: Ran Zou,Wanrong Zhu,Bin Nan
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:

点击查看摘要

Abstract:Conformal prediction provides distribution-free predictive intervals with finite-sample marginal coverage. However, achieving conditional validity and interval efficiency (in terms of short interval length) remains challenging, particularly in complex settings with heteroskedasticity, skewed responses, or estimation errors. We propose a conformal-style calibration method for responses obtained by the probability integral transform (PIT) of the conditional cumulative distribution function (CDF) estimated via neural networks to construct a finite-sample-adjusted percentile interval with the shortest length determined by the estimated conditional CDF. Calibrating in PIT space is effective because PIT values are asymptotically feature-independent when the CDF estimator is accurate, which mitigates feature-dependent miscoverage and improves conditional calibration. On the other hand, our percentile calibration adapts to the empirical PIT distribution, which is robust against a possibly imperfect estimation of the conditional CDF. We prove the finite-sample marginal coverage property of the proposed method and show its asymptotic conditional coverage under mild consistency conditions. Experiments on diverse synthetic and real-world benchmarks demonstrate better conditional calibration and substantially shorter intervals than existing methods.

[LG-94] Dynamic Vine Copulas: Detecting and Quantifying Time-Varying Higher-Order Interactions

链接: https://arxiv.org/abs/2605.03061
作者: Houman Safaai,Alessandro Marin Vargas
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Quantitative Methods (q-bio.QM); Methodology (stat.ME)
*备注:

点击查看摘要

Abstract:Time varying dependence is often modeled through dynamic correlations or Gaussian graphical models, yet many multivariate systems change through tail behavior, asymmetry, or conditional structure while correlations change little. We introduce Dynamic Vine Copulas (DVC), a temporal vine copula framework for estimating and diagnosing sequence wide non-Gaussian dependence. DVC keeps a chosen vine factorization fixed for comparability, can use C-, D-, or R-vines, and couples pair copula states across time through smooth parameter trajectories or temporally regularized family switching paths. Its central diagnostic contrasts held-out scores from a full vine and its matched 1-truncated counterpart, separating flexible first-tree pairwise evidence from higher-tree conditional evidence. At the population level, under a correct fixed vine and simplifying assumption, this contrast is the higher-tree term of a vine total correlation decomposition; in finite samples, it is a predictive diagnostic. Across controlled benchmarks, DVC detects Student-t tail degree changes, Clayton-to-Gumbel switches, and recurrent conditional interaction episodes that Gaussian dynamic baselines miss or conflate. The higher-tree score stays near zero in pairwise only regimes but rises selectively during conditional interaction regimes. On Allen Visual Behavior Neuropixels data, DVC identifies a reproducible time indexed higher-tree signal that is positive across held-out splits and disappears under a decorrelated null, indicating simultaneous cross-area dependence. Together, these results show that DVC is both a flexible temporal copula model and an interpretable diagnostic for whether time varying dependence changes are pairwise or conditional.

[LG-95] PHBench: A Benchmark for Predicting Startup Series A Funding from Product Hunt Launch Signals

链接: https://arxiv.org/abs/2605.02974
作者: Yagiz Ihlamur,Ben Griffin,Rick Chen
类目: Pricing of Securities (q-fin.PR); Machine Learning (cs.LG)
*备注: 30 pages, 1 figure, 4 appendices. Website, leaderboard, and dataset: this https URL

点击查看摘要

Abstract:Structured launch signals on Product Hunt contain statistically significant predictive information for Series A funding outcomes. We construct PHBench from 67,292 featured Product Hunt posts spanning 2019-2025, linked to Crunchbase funding records via deterministic domain matching, identifying 528 verified Series A raises within 18 months of launch (positive rate: 0.78%). Our best-performing model, a three-component ensemble (ENS_avg, ENS_ISO, XGB) selected by validation F0.5, achieves F0.5 = 0.097 and AP = 0.037 (95% CI: 0.024-0.072; 4.7x lift over random) on the private held-out test set (103 positives). A paired bootstrap confirms a statistically credible advantage over the logistic regression baseline (AP delta: +0.013, 95% CI: [0.004, 0.039], p 0.001; F0.5 delta: +0.056, 95% CI: [0.006, 0.122], p = 0.016). Validation-set metrics (F0.5 = 0.284, AP = 0.126) reflect best-of-144 selection bias on 53 positives and are reported for benchmark reproducibility only. We further evaluate three zero-shot Gemini models (Gemini 2.5 Flash, Gemini 3 Flash, and Gemini 3.1 Pro) in an anonymized numerical setting. The best LLM achieves AP = 0.034 (Gemini 3 Flash), below the LR baseline AP of 0.044. Notably, the most capable Gemini variant (Gemini 3.1 Pro, AP = 0.023) performs worst – an unexpected pattern that warrants further investigation across providers and prompting strategies. Both ML and LLM models show the same temporal performance decay tracking the 2020-2021 funding boom and subsequent contraction, confirming the dataset captures genuine market structure rather than noise. PHBench provides a reproducible framework comprising public training, validation, and blind test splits; 61 engineered features; a five-metric evaluation harness; and a public leaderboard at this https URL. All code, baseline models, and anonymized dataset splits are publicly available. Comments: 30 pages, 1 figure, 4 appendices. Website, leaderboard, and dataset: this https URL Subjects: Pricing of Securities (q-fin.PR); Machine Learning (cs.LG) ACMclasses: I.2.6; H.2.8; J.4 Cite as: arXiv:2605.02974 [q-fin.PR] (or arXiv:2605.02974v1 [q-fin.PR] for this version) https://doi.org/10.48550/arXiv.2605.02974 Focus to learn more arXiv-issued DOI via DataCite

[LG-96] EFGPP: Exploratory framework for genotype-phenotype prediction

链接: https://arxiv.org/abs/2605.02954
作者: Muhammad Muneeb,David B. Ascher
类目: Genomics (q-bio.GN); Machine Learning (cs.LG)
*备注: this https URL

点击查看摘要

Abstract:Predicting complex human traits from genetic data is challenging because different genetic, clinical, and molecular data sources often contain different parts of the signal. Here, we present EFGPP, a reproducible framework for generating, ranking, and combining multiple types of data for genotype-to-phenotype prediction. We applied EFGPP to migraine prediction using UK Biobank data from 733 individuals. The framework combined genotype-derived features, principal components, clinical and metabolomic covariates, and polygenic risk scores generated from migraine and depression GWAS using PLINK, PRSice-2, AnnoPred, and LDAK-GWAS. The best single data type achieved a test AUC of 0.644, while combining multiple data types improved performance to 0.688 using migraine-focused inputs and 0.663 using cross-trait depression-derived inputs. Genetic features alone did not outperform the covariates-only baseline, but genotype-derived features performed better than PRS alone, and depression-derived PRS showed useful predictive signal. Overall, EFGPP provides a practical proof-of-concept framework for prioritising and integrating heterogeneous genetic data sources for complex phenotype prediction.

附件下载

点击下载今日全部论文列表

目录

概览 (2026-05-06)

多智能体系统

自然语言处理

信息检索

人机交互

计算机视觉

人工智能

机器学习

附件下载