本篇博文主要内容为 2026-08-25 从Arxiv.org论文网站获取的最新论文列表，自动更新，按照NLP、CV、ML、AI、IR、MA六个大方向区分。
说明：每日论文数据从Arxiv.org获取，每天早上12:30左右定时自动更新。
提示: 当天未及时更新，有可能是Arxiv当日未有新的论文发布，也有可能是脚本出错。尽可能会在当天修复。
目录
概览 (2026-08-25)
今日共更新946篇论文,其中:
- 自然语言处理共162篇(Computation and Language (cs.CL))
- 人工智能共269篇(Artificial Intelligence (cs.AI))
- 计算机视觉共195篇(Computer Vision and Pattern Recognition (cs.CV))
- 机器学习共239篇(Machine Learning (cs.LG))
- 多智能体系统共15篇(Multiagent Systems (cs.MA))
- 信息检索共34篇(Information Retrieval (cs.IR))
- 人机交互共29篇(Human-Computer Interaction (cs.HC))
多智能体系统
[MA-0] he Interaction Tax: When Communication Erases Diversity in Multi-Agent Teams ICML2026
【速读】：该论文试图解决的问题是：多智能体大语言模型（multi-agent LLM）之间的交互是否真正有助于提升任务表现，尤其是在资源预算相同的情况下，交互机制带来的收益与成本之间是否存在权衡。现有研究对此存在矛盾结论——部分研究发现辩论（debate）、批判循环（critique loops）和混合智能体合成（mixture-of-agents synthesis）能带来性能提升，而另一些研究则指出交互仅增加计算开销且未显著改善质量，甚至独立采样已可捕获多智能体的优势。本文的核心观点在于，这种矛盾部分源于对“多智能体通信”形式的混淆：并非所有交互都等效。不同模型家族会生成结构各异的解决方案，但当智能体直接读取彼此完整的输出时，其提议在单轮内即趋于收敛，导致多样性丧失，从而削弱了使用多智能体的初衷。作者将此现象称为“交互税”（interaction tax）。通过在11个经验证器评分的优化任务上进行受控实验，研究发现全解交互作为默认策略效果较弱，而独立生成提案可避免解空间坍缩。全解交互主要使智能体固守首次接触到的解，而非探索多样化路径；批判性反馈仅在违反规则易于被大语言模型识别并修正时才有效。因此，论文的关键结论是：多智能体系统的性能并不取决于智能体数量，而更关键的是信息交换的内容与时机——只有在恰当时间传递适当信息，交互才能真正发挥增益作用。
链接: https://arxiv.org/abs/2608.23541
作者: Summer Eunhyung Ann,Haokun Liu,Chenhao Tan
机构: University of Chicago(芝加哥大学)
类目: Multiagent Systems (cs.MA); Artificial Intelligence (cs.AI)
备注: 14 pages, 3 figures. Accepted at ICML 2026 (PMLR 306)
Abstract:Does multi-agent LLM interaction help or hurt? Some work reports gains from debate (Du et al., 2024), critique loops (Chen et al., 2025), and mixture-of-agents synthesis (Wang et al., 2025), while other work finds that interaction adds cost without improving quality under equal budgets (Tran Kiela, 2026; Xu et al., 2026; Jarrett et al., 2025), or that independent sampling already captures multi-agent gains (Li et al., 2024). We argue this contradiction partly reflects a missing distinction, because not all multi-agent communication is equal. Different model families find structurally different solutions, but when agents read each other’s complete outputs, their proposals converge within one round, erasing the diversity that motivates using multiple models. We call this the interaction tax. We test 11 verifier-scored optimization tasks under matched budgets and find that full-solution interaction is a weak default. Independent proposal generation avoids this collapse. Full-solution interaction mainly makes agents stay close to the first solution they see instead of trying different approaches, and critique helps only if the violated rule is easy for the LLM to find and fix. These results suggest that multi-agent performance depends less on the number of agents than on the information they exchange, and interaction helps only when agents share the right information at the right time.
[MA-1] Right-Sizing LLM -Agent Decomposition in VAT Determination: A Pilot Controlled Sweep
【速读】：该论文旨在解决大语言模型代理系统（LLM-agent systems）在任务分解策略上的根本性设计权衡问题：是采用多个专精的窄域代理（narrow agents）进行分工协作，还是依赖单一具备强大工具调用能力的宽域代理（wide agent）。研究聚焦于一个具有明确边界条件的跨境增值税反向收费判定任务，该任务具备可验证的真值标签（oracle label）和可独立评分的中间决策节点。其解决方案的关键在于通过严格控制活动表面（activity surface），包括子任务划分、工具使用、输入输出模式、验证机制、编排器、基础模型及合并策略等变量，仅改变子任务分配给不同规模代理的配置，构建从单一宽域代理到五个窄域代理的四种编排方案，并与无编排的单代理基线（S0）对比。研究采用预注册的可证伪标准，在4,400次运行中评估不同配置的准确性、鲁棒性与资源效率。结果显示，中等粒度配置在准确率上表现最优（0.830），但未达到预设的精细基准，因此“中等分解最优”的假设在试点规模下仍不成立；单代理并未对所有编排方案形成帕累托优势；在匹配提示预算条件下，单代理性能落后领先者6.5个百分点，但置信区间包含零，表明其劣势可能源于提示预算差异而非架构本身。故障注入实验进一步揭示，广域重启能有效吸收可用性故障，而单一符合模式的幻觉记录则显著破坏所有配置，尤其影响碎片化结构。最终贡献为一个受控、预注册的试点启发式方法，用于合理确定任务分解尺度——建议在依赖层中点处设置一个分割边界，并公开了真值、数据集、测试框架、原始日志及分析流程，以支持后续可复现的研究。
链接: https://arxiv.org/abs/2608.23395
作者: Pedro Santos
机构: Independent Researcher(独立研究员)
类目: Multiagent Systems (cs.MA); Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注: 29 pages, 4 figures. Code, data, and traces: this https URL
Abstract:Recent LLM-agent systems make conflicting design bets: decompose work across many narrow agents, or use one strong tool-using agent. This pilot studies that choice on bounded cross-border VAT determination with reverse charge, where every case has an oracle label and each intermediate decision is independently scoreable. We hold the activity surface fixed (subtasks, tools, I/O schemas, validation checks, orchestrator, base model, and merge policy) and vary only the assignment of subtasks to workers across four orchestrated configurations, from one wide worker to five narrow ones, against S0, a tuned no-orchestrator single agent, with a deterministic rule engine as oracle. The program spans 4,400 runs: a 40-case, five-repeat main sweep, matched-token arms separating prompt-budget from agent-count effects, and three failure-injection arms, all judged against pre-registered falsification criteria. The two intermediate configurations lead on accuracy (0.830, against endpoints at 0.720 and 0.770) but miss the pre-stated bar against the fine endpoint, so the intermediate-optimum hypothesis remains unsupported at pilot scale. The single agent does not Pareto-dominate the orchestrated set. The matched-token criterion fires: the budget-matched single agent lands 6.5 points below the leader, but the interval includes zero, so any advantage is consistent with a prompt-budget explanation. Under injection, availability faults are absorbed at every granularity, with wide-scope restart over-recovering its baseline by +0.160, while one schema-conforming hallucinated record degrades every configuration and inverts the ordering, hitting fragmented configurations hardest. The contribution is a bounded, preregistered pilot heuristic for right-sizing decomposition (place one partition boundary at the dependency-layer midpoint), released with oracle, dataset, harness, raw traces, and analysis pipeline.
[MA-2] First Demonstration of Multi-Agent LLM System for Million-Scale Optical Link Management in Global Production AIDCs
【速读】：该论文旨在解决生产环境中大规模数据中心（AIDC）中数百万条光链路自主故障管理的难题，传统方法难以应对高复杂性与实时性要求。其解决方案的关键在于构建首个基于大语言模型（LLM）的多智能体系统，通过监督微调（SFT）与持续记忆演化机制对模型进行优化，实现对故障的精准识别与自适应响应，在为期十周的真实场景数据评估中，取得了97.7%的F1分数及超过60%的故障事件减少率，显著优于当前最优的LLM方案。
链接: https://arxiv.org/abs/2608.23145
作者: Jingyi Su,Yihao Zhang,Dianxuan Fu,Leiyan Fei,Juan Wang,Mengfan Dai,Qing Liu,Xiong Wu,Yufeng Jiang,Cheng Chen,Bowen Zhang,Peilong Wang,Xi Chen,Zonglong He,Hongchen Yu,Zhicheng Ye,Weisheng Hu,Qunbi Zhuge
机构: 未知
类目: Multiagent Systems (cs.MA); Optics (physics.optics)
备注: 4 pages, 3 figures
Abstract:We present the first LLM-powered multi-agent system for autonomous fault management across millions of optical links in production AIDCs. Refined via SFT and continuous memory evolution, it achieves 97.7% F1 and over 60% fault-incident reduction, outperforming SOTA LLMs on a ten-week field data evaluation.
[MA-3] AutoSaddler: Automatic Harness Optimization with Durable Updates from Agent Execution Traces
【速读】：该论文旨在解决大语言模型（Large Language Model, LLM）智能体在长时程任务中因局部微小失败累积而导致整体任务失败的可靠性问题。现有解决方案依赖人工设计外部约束框架（harness），但该过程耗时费力且需在庞大的提示词、工具配置与控制逻辑空间中进行搜索。为此，论文提出AutoSaddler——一种自动化的框架优化方法，将约束框架的改进建模为离线学习问题，并通过从迷你批次中提取失败信号迭代更新框架。其核心创新在于结合故障轨迹诊断、将框架视为代码进行结构化补丁生成，以及基于验证结果的选择性更新机制。实验在GAIA2、SWE-Bench Pro和Terminal-Bench 2.0上表明，AutoSaddler相较于基线框架分别提升了9.0、9.6和10.0个百分点。消融研究进一步揭示有效优化依赖三个关键要素：深度调试而非浅层反思、针对性修改而非无约束编辑、泛化感知选择而非仅针对特定轨迹修复。这些结果表明，自动约束框架优化是提升智能体系统性能与可靠性的可行路径。
链接: https://arxiv.org/abs/2608.23041
作者: Sungho Park,Wonjoong Kim,Rongyuan Tan,Jue Zhang,Wook-Shin Han,Pengfei Gao,Chanyoung Park,Yongqiang Yao,Rao Fu,Elsie Nallipogu,Qingwei Lin,Saravan Rajmohan,Dongmei Zhang
机构: KAIST(韩国科学技术院); Southern University of Science and Technology(南方科技大学); Microsoft(微软); POSTECH(浦项科技大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG); Multiagent Systems (cs.MA); Software Engineering (cs.SE)
备注: 44 pages, 15 figures. Project website and code: this https URL
Abstract:LLM agents remain unreliable on long-horizon tasks, where small local failures can compound over extended interactions and lead to overall task failure. Although external harnesses can substantially improve robustness, harness design remains a manual and expensive process that requires searching over a large space of prompts, tool configurations, and control logic. We propose AutoSaddler, an automatic harness optimization framework that formulates harness improvement as an offline learning problem and iteratively updates the harness using failure signals from mini-batches. AutoSaddler combines failure-trace diagnosis, structured patch generation that treats the harness as code, and validation-based update selection. Experiments on GAIA2, SWE-Bench Pro, and Terminal-Bench 2.0 show that AutoSaddler substantially improves agent performance over the corresponding base harnesses, achieving gains of 9.0, 9.6, and 10.0 percentage points, respectively. Ablation studies further suggest that effective harness optimization benefits from three ingredients: deep debugging rather than shallow reflection, targeted modifications rather than unconstrained editing, and generalization-aware selection rather than trajectory-specific repair. Together, these results suggest that automatic harness optimization is a promising path toward more performant and reliable agent systems.
[MA-4] Predicting the scale limits of social mechanisms in agent societies
【速读】：该论文旨在解决在大规模语言模型代理（language-model agents）社会中，小规模群体中存在的社会机制是否仍能有效运作的问题。由于直接通过大规模实证实验验证此类机制的成本极高，研究面临可扩展性评估的挑战。其解决方案的关键在于提出一种“审计”（audit）方法，该方法通过三方面评估机制的可扩展性：一是机制在群体中发挥作用的频率；二是代理是否有效利用机制提供的信息；三是测量过程本身是否会引入虚假的规模效应。研究表明，单一结构性因素即可决定互惠、共识或惩罚等机制在规模扩大时的存续与否；以流言传播为例，机制失效的临界人口规模由信息传播范围和持续时间决定。此外，在语言模型社会中，代理不仅响应社会信息的内容，还受表达方式影响——如数值与百分比表述导致不同的规模行为。该审计方法在未执行前即能做出准确预测，且在第三方代码和另一模型家族上验证有效，而失败的预测则揭示了研究结论的边界。因此，该审计为前瞻性判断社会机制在不同群体规模下的可解释性提供了可靠路径。
链接: https://arxiv.org/abs/2608.22884
作者: Zengqing Wu,Chuan Xiao
机构: University of Osaka(大阪大学); Osaka, Japan
类目: Multiagent Systems (cs.MA); Social and Information Networks (cs.SI)
备注:
Abstract:Societies of interacting language-model agents offer a controllable and repeatable way to study collective behaviour at scales that would be difficult to test with people. Their scientific value, however, depends on whether a social mechanism that works in a small group still operates when thousands of agents interact, and testing this directly requires costly large-scale runs. Here we introduce an audit that predicts a mechanism’s fate as a population grows. It asks how often the mechanism can act, whether agents use the information it supplies, and whether the measurement itself creates apparent scale effects. Controlled experiments show that a single structural term can decide whether reciprocity, consensus or punishment survives scaling. For gossip, the population at which the mechanism fails is set by the reach and lifetime of its messages. In language-model societies, agents respond not only to social information but to how it is expressed: counts and percentages led to different scale behaviour. Predictions made before execution held on third-party code and a second model family, while a failed prediction exposed the boundary of the finding. The audit provides a prospective way to decide which social mechanisms can be interpreted across population scales.
[MA-5] Minimal Local Simulation Foundations for LLM - and VLM-Driven Agents in 2D and 3D Environments
【速读】：该论文旨在解决当前基于大语言模型（Large Language Models, LLMs）和视觉-语言模型（Vision-Language Models, VLMs）的智能体仿真平台普遍存在难以研究、修改或在普通计算机上运行的问题。其核心挑战在于现有平台通常依赖复杂的基础设施，限制了教育应用与快速原型开发的可及性。为此，论文提出两个设计极简的仿真基础框架：SD-AgentFoundry-2D 和 SD-AgentFoundry-3D。前者构建了一个二维多智能体环境，支持本地部署的LLM智能体进行移动、通信、响应位置占用以及遭遇空间局部火灾事件；后者则提供一个三维数字孪生环境，通过本地部署的VLM接收第一人称视角图像并生成自然语言运动指令。两个代码库均针对 macOS、Windows 与 Linux 系统本地运行进行了优化，并刻意保持开放性以支持自由修改，而非作为封闭的应用程序。其解决方案的关键在于通过轻量化、可定制且本地可运行的设计，为生成式社会仿真学习与领域特定扩展提供了低门槛的起点。
链接: https://arxiv.org/abs/2608.22833
作者: Ryuki Hyodo
机构: SpaceData Inc.(SpaceData公司); Rikkyo University(立教大学); Earth-Life Science Institute(地球生命科学研究所); Université Paris Cité(巴黎城市大学)
类目: Multiagent Systems (cs.MA); Artificial Intelligence (cs.AI)
备注: GitHub Repositories: this https URL and this https URL
Abstract:Large language models (LLMs) and vision-language models (VLMs) are expanding the range of behaviors that can be represented in agent-based simulations, but many contemporary platforms are difficult to study, modify, or run on ordinary computers. We present two intentionally minimal simulation foundations for education and rapid prototyping. SD-AgentFoundry-2D provides a two-dimensional multi-agent environment in which locally hosted LLM agents move, communicate, respond to place occupancy, and encounter spatially localized fire events. SD-AgentFoundry-3D provides a three-dimensional digital-twin environment in which a locally hosted VLM receives first-person images and produces natural-language movement instructions. Both codebases are designed to run locally on macOS, Windows, and Linux and are deliberately left open to modification rather than developed as finished applications. Together, they offer accessible starting points for learning about generative social simulation and for building domain-specific extensions.
[MA-6] CatchBench: When Can an Agent Failure Be Caught? ATC
【速读】：该论文旨在解决在多智能体系统中，如何有效检测智能体（agent）失效的问题，尤其关注在不同信息状态下的故障识别能力。传统审计方法受限于记录数据而非执行过程本身，导致难以全面捕捉智能体行为中的异常。为此，作者提出了CatchBench基准测试框架，将审计问题分解为三个关键信息状态：运行前的声明配置（PRE）、运行时的逐步追踪片段（LIVE）以及运行结束后的完整日志（POST），从而实现对智能体行为在不同阶段的细粒度评估。其解决方案的关键在于构建一个统一的任务-方法接口，支持在三种信息状态下分别进行评估，并设计了七种任务合约（task contracts），每种对应不同的标签体系与评价指标，涵盖证据性任务和基于黄金标准（Gold-derived）的机制诊断任务，以避免单一排行榜带来的误导。该基准覆盖1187个声明配置、1162次实际运行，评估了包括规则扫描器、结构化模型及来自九个大模型家族（GPT、Claude、Gemini、Gemma、Llama、Qwen、DeepSeek、Mistral、Nova）共11个大型语言模型（LLM）裁判在内的72种方法。值得注意的是，多数评估结果无法形成明确排序，47组预声明对比未被区分，其余则保留未决状态而非强行排名，强调了对“捷径”（shortcut）风险的警惕。研究发现，某些规则虽在特定配置源上达到完美F1分数，但其本质是依赖数据构建方式而非真实推理能力，因此通过设定严格的可接受性标准排除了存在注入干扰的子系统，并拒绝赋予其证据地位。最终结论指出，基准评分本身不具备可解释性，除非公开其标签生成过程并验证是否存在捷径漏洞。论文强调透明性与可复现性，所有排序均基于已发布的预测结果重新生成，无需额外调用模型，确保评估过程的可信度。
链接: https://arxiv.org/abs/2608.22808
作者: Yue Zhao
机构: University of Southern California(南加州大学)
类目: Machine Learning (cs.LG); Multiagent Systems (cs.MA); Performance (cs.PF)
备注: 39 pages, 6 figures, 21 tables. Work in progress. Code and data: this https URL
Abstract:When can an agent failure be caught? An audit is usually limited by the record rather than by the method. CatchBench therefore puts one auditor’s question to three information states: the declared configuration before a run (PRE), a growing prefix of its trace (LIVE), and the finished trace (POST). Prior benchmarks fix one of these states or vary the telemetry; to our knowledge none scores all three under one task-method interface. Each state admits different questions, so seven task contracts carry their own labels and metrics rather than one leaderboard. Four are evidential; three are Gold-derived mechanism diagnostics. The release scores 72 entrants, from rule scanners and structural models to eleven LLM judges across nine model families (GPT, Claude, Gemini, Gemma, Llama, Qwen, DeepSeek, Mistral, Nova), over 1187 declared configurations and 1162 recorded runs. Most of the arena does not order: 47 of 118 pre-declared contrasts separate, and the rest are published unresolved rather than ranked. The two sharpest results cut against our own data. One rule ignores every name and permission; it flags each capability declared after the first. On one of six configuration sources it reaches a perfect F1, so a score there measures how the corpus was built rather than how well a method reasons. Our admissibility bar then rejected one injected substrate and withheld evidential status from the other. A benchmark number is therefore not interpretable until the process behind its labels is published and tested for the shortcut it may leave. We report both, and regenerate every ordering from released predictions with no model call. Comments: 39 pages, 6 figures, 21 tables. Work in progress. Code and data: this https URL Subjects: Machine Learning (cs.LG); Multiagent Systems (cs.MA); Performance (cs.PF) Cite as: arXiv:2608.22808 [cs.LG] (or arXiv:2608.22808v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2608.22808 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[MA-7] PropUQ-MAS: Propagation-Aware Uncertainty Quantification for LLM Multi-Agent Systems EMNLP2026
【速读】：该论文旨在解决基于大语言模型（LLM）的多智能体系统（MAS）中因智能体间依赖关系导致的不确定性传播问题。传统不确定性量化（UQ）方法仅针对单个智能体或孤立响应，无法有效捕捉多智能体协作过程中中间信息错误在上下游智能体间的传递与放大效应，从而影响整体系统的可靠性。为此，本文提出了一种考虑误差传播的不确定性量化框架PropUQ-MAS，其核心在于将多智能体系统的执行过程建模为具有通信结构的图，并通过融合局部不确定性与上游消息继承的不确定性来评估每一步的可靠性。实验结果表明，PropUQ-MAS在多智能体系统中的不确定性量化性能显著提升，平均相对增益达AUROC +6.10%、PRR +47.58%。
链接: https://arxiv.org/abs/2608.22130
作者: Yaokun Liu,Yifan Liu,Daniel Yue Zhang,Ruichen Yao,Zelin Li,Dong Wang
机构: Scale AI(规模人工智能); University of Illinois Urbana-Champaign(伊利诺伊大学厄本那-香槟分校)
类目: Multiagent Systems (cs.MA); Computation and Language (cs.CL)
备注: Accepted to EMNLP 2026 (Main Conference)
Abstract:LLM-based multi-agent systems (MAS) solve complex tasks through communication among role-specialized agents. However, inter-agent dependencies introduce reliability risks beyond isolated agent failures. For instance, errors in intermediate messages could be inherited and amplified by downstream agents. Existing uncertainty quantification (UQ) methods mainly target isolated responses or single-agent reasoning, and therefore fail to capture uncertainty propagation in MAS. To this end, we propose PropUQ-MAS, an error propagation-aware UQ framework that represents MAS execution as a communication-structured graph and estimates each step’s reliability by combining local uncertainty with uncertainty inherited from upstream messages. Extensive experiments demonstrate that PropUQ-MAS consistently improves UQ in MAS, with average relative gains of +6.10% in AUROC and +47.58% in PRR.
[MA-8] Opinion-Guided Layered Strategies for Decentralized Coordination
【速读】：该论文旨在解决自主智能体在与其它独立智能体交互时面临的协调难题，特别是在存在多个可接受的联合行为（admissible joint behaviors）时，双方因偏好不一致导致策略互不兼容，或偏好相同时无法实现角色区分的问题。传统策略通常预先确定唯一的行为选择，难以适应异质偏好或动态对称性破缺的需求。为此，论文提出一种新型策略——意见引导策略（opinion-guided strategy），其核心在于通过非线性意见动力学在分层架构中动态维持所有可接受联合行为的可行性，并将具体行为的选择推迟至执行阶段，依据对方行为演化实时确定最终协同路径，无需通信即可实现协调。该策略的关键创新在于：利用分层意见动力学机制，在不依赖先验共识的前提下，确保对任意对方偏好均保持鲁棒性；这一特性使得运行相同策略的两个智能体可在必要时自发打破对称性，而这是传统策略所不具备的能力。理论分析证明了该策略在各种偏好下的稳定性，三个跨应用案例验证了其有效性，尤其在一般和博弈（general-sum game）场景中，相较于传统方法需预先求解唯一纳什均衡，意见引导策略始终保持所有均衡开放，最终由运行时交互动态决定达成哪一个，显著提升了灵活性与适应性。
链接: https://arxiv.org/abs/2608.22104
作者: Shuhao Qi,Zhiyong Sun,Siep Weiland,Sofie Haesaert
机构: Eindhoven University of Technology (埃因霍温理工大学); Peking University (北京大学)
类目: Multiagent Systems (cs.MA); Systems and Control (eess.SY)
备注:
Abstract:Autonomous agents increasingly interact with other independent agents, and such interactions typically admit multiple joint behaviors. When two agents prefer different ones, their independent strategies may be mutually incompatible and fail to reach a coordinated outcome; when they are identical, neither can differentiate its role when needed. Ideally, an agent should coordinate with any agent it encounters, regardless of which admissible joint behavior that agent aims to realize. We therefore propose a new form of strategy, the opinion-guided strategy, which keeps all the admissible joint behaviors available and postpones the selection to execution time, when the other agent’s behavior reveals which one to realize. To realize this, nonlinear opinion dynamics are leveraged in a layered realization to guide the agent to a common admissible joint behavior in response to the other agent’s evolving behavior, even without communication. We formally establish the conditions under which the strategy remains robust to every preference the other agent may hold. This robustness has an important implication: two agents running identical strategies can break symmetry when needed, a capability that conventional strategies lack. Three case studies across different applications show that the opinion-guided strategy coordinates with every randomly encountered agent, as long as it is willing to realize one of the admissible joint behaviors. One of them corresponds to a general-sum game: unlike conventional approaches devoted to finding a unique Nash equilibrium in advance, the opinion-guided strategy keeps every equilibrium open and guarantees the agents reach one, decided by their runtime interaction.
[MA-9] Spine-Branch Coordination for Multi-agent Computer Use
【速读】：该论文旨在解决多智能体计算机使用系统中因虚拟机（VM）状态无法合并而导致的可扩展性瓶颈问题。传统方法将此问题视为次要或临时处理，而非系统设计的核心约束。其解决方案的关键在于提出“主干-分支”协调框架（Spine-Branch Coordination），将任务分解为“主干-分支”图结构：主干负责维持连续的执行状态流，而分支任务并行执行以收集主干完成任务所需的信息；一旦分支任务完成，其对应的虚拟机即被丢弃，从而从根本上避免了虚拟机状态合并的需求。实验表明，在Odysseys数据集中的200个长时程任务上，该框架相比基线系统在成功率上提升6.0%至16.5%，同时单任务成本降低34%至70%，验证了显式建模虚拟机状态合并约束对多智能体计算机使用系统高效扩展的重要性。
链接: https://arxiv.org/abs/2608.22077
作者: Mian Zhang,Manasi Sharma,Sheng Zhang,Minglai Yang,Kejian Shi,Ying Liu,Zhiyu Zoey Chen,Daniel Yue Zhang
机构: Scale AI; Johns Hopkins University (约翰霍普金斯大学); University of Texas at Dallas (德克萨斯大学达拉斯分校)
类目: Computation and Language (cs.CL); Multiagent Systems (cs.MA)
备注:
Abstract:Computer use agents (CUAs) are increasingly deployed as multi-agent systems that decompose a task into multiple subtasks executed across parallel virtual machines (VMs). However, a critical physical bottleneck is that the state of two VMs cannot be merged. Previous systems handle this ad-hoc rather than treating it as a first-class concern. We propose Spine-Branch Coordination for multi-agent computer use, a framework that decomposes a task into a “spine-branch” graph, where the spine carries the main task flow with continuous VM state and branch tasks execute in parallel to collect information the spine needs to complete the task. Branch VMs are discarded once their tasks finish, so no VM merging ever occurs. Experiments show that on 200 long-horizon tasks from Odysseys and across three CUA backbones, Spine-Branch improves success rate over the baseline system by 6.0% to 16.5%, while reducing per-task cost by 34% to 70%, indicating that explicitly modeling VM-state merging constraint enables multi-agent computer use to scale efficiently.
[MA-10] OptiMAS: Automatically Optimize Multi-Agent System EMNLP2026
【速读】：该论文旨在解决基于大语言模型（LLM）的多智能体系统（Multi-Agent Systems, MAS）在自动化演化过程中面临的两大核心挑战：一是搜索范式在扩大优化范围时引发的进化不稳定性，二是离散的“分支-丢弃”式搜索导致不同演化谱系间知识难以共享。针对这一问题，论文提出一种连续的、数据驱动的优化范式，其关键在于构建统一的基于ReAct（Reasoning and Acting）的基础设施，从而在保持广泛优化空间的同时确保系统的运行稳定性。该范式下提出的OptiMAS是一种任务无关的智能体优化器，通过利用文本交互轨迹与任务反馈作为损失信号，实现多智能体系统的端到端演化。其创新性地引入双轨记忆机制，有效支持在长时间优化周期中的性能持续提升。在四个异构智能体基准测试中，使用三种不同规模和可访问性的LLM后端进行评估，结果表明OptiMAS在准确率上均达到或优于领域专用的手工设计系统及现有演化方法，为实现稳健、自动化的多智能体系统演化奠定了实用基础。
链接: https://arxiv.org/abs/2608.21918
作者: Yuxin Cheng,Chang Liu,Hanxin Yu,Haochen Tan,Taiqiang Wu,Weiqiang Jin,Jie Ran,Kaibo Wang,Xiaoguang Li,Haoli Bai,Graziano Chesi,Ngai Wong
机构: 未知
类目: Multiagent Systems (cs.MA)
备注: EMNLP 2026
Abstract:Automated evolution of Multi-Agent Systems (MAS) holds significant potential for reducing the manual effort required to design and optimize LLM-based agent architectures. However, extant search-based paradigms face a fundamental trade-off, where an expanded optimization scope exacerbates evolutionary instability, while discrete branch-and-discard search isolates insights across lineages. To address these limitations, we propose a continuous, data-driven optimization paradigm built upon a unified ReAct-based infrastructure that reconciles a broad optimization scope with operational stability. Under this paradigm, we present OptiMAS, a task-agnostic agentic optimizer that leverages textual interaction trajectories and task feedback as loss signals for end-to-end MAS evolution. Equipped with a novel dual-track memory mechanism, OptiMAS sustains performance improvement over extended optimization horizons. Evaluation on four heterogeneous agentic benchmarks with three varying scale and accessibility LLM backbones, demonstrates that OptiMAS consistently achieves competitive or superior accuracy relative to both domain-specialized hand-crafted systems and existing evolutionary methods. Our work establishes a practical milestone toward robust, automated MAS evolution.
[MA-11] ATHENA: Knowledge-guided agent ic neural architecture search for AutoFormer-based electronic health record modeling
【速读】：该论文旨在解决基于Transformer的电子健康记录（EHR）临床预测模型在实际应用中面临的架构设计依赖人工调参、跨医院泛化性差以及神经架构搜索（NAS）计算成本高昂的问题。其核心挑战在于，现有NAS方法虽能自动化架构探索，但通常缺乏跨医疗机构的知识迁移能力，且每次搜索需独立进行预训练，导致效率低下。为此，论文提出ATHENA（Agentic Transfer across Hospitals for EHR Neural Architecture Search），一种基于知识引导的代理式跨医院神经架构搜索框架。其关键创新在于：首先构建每个医院独有的权重共享超网络（supernet），通过微调而非独立预训练来评估候选子网络，显著降低计算开销；其次引入双层跨医院架构先验机制——第一层基于任务描述符从源医院检索高性能架构实例，第二层利用基于SHAP的元回归分析架构组件的贡献效应，为多智能体大语言模型（LLM）搜索提供可解释的指导信号，并结合目标医院的验证反馈实现自适应优化。实验表明，在6个临床预测任务和2个独立医疗系统上，ATHENA在30次搜索预算下于12项评估中的9项达到或超越四种基线方法，且架构选择具有更高一致性，有效降低了对人工调参的依赖，为高效、可迁移的Transformer-EHR建模提供了实用解决方案。
链接: https://arxiv.org/abs/2608.21712
作者: Deyi Li,Qi Xu,Lingyao Li,Tiansheng Wang,Muxuan Liang,Mei Liu
机构: University of Florida (佛罗里达大学); University of Arizona (亚利桑那大学); University of Houston (休斯顿大学); University of North Carolina at Chapel Hill (北卡罗来纳大学教堂山分校); University of Texas MD Anderson Cancer Center (德克萨斯大学安德森癌症中心)
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注:
Abstract:Transformer-based models are widely used for clinical prediction from electronic health records (EHRs), yet their architectures still require substantial manual tuning, and the optimal configuration may vary across tasks and hospitals. Neural architecture search (NAS) automates architecture design, but conventional methods are computationally costly for Transformer-based EHR models. Recent large language model (LLM)-guided NAS methods reduce manual search design but typically conduct each search independently, without reusing architecture knowledge across hospitals. In this study, we propose ATHENA (Agentic Transfer across Hospitals for EHR Neural Architecture Search), a knowledge-guided agentic NAS framework for Transformer-based EHR modeling. ATHENA uses a weight-sharing supernet that is pretrained once per hospital, allowing candidate architectures to be instantiated as inherited subnetworks and evaluated through fine-tuning rather than independent pretraining. It also incorporates a two-layer cross-hospital architecture prior. The first layer retrieves high-performing architecture examples from source sites based on task descriptors, while the second estimates the effects of architectural components using SHapley Additive exPlanations (SHAP)-based meta-regression. These priors guide a multi-agent LLM search together with validation feedback from the target hospital. Across six clinical prediction tasks and two independent health systems, ATHENA matches or outperforms four NAS baselines in 9 of 12 hospital-task evaluations at a search budget of 30. It also shows more consistent architecture selection across repeated searches. ATHENA provides a practical approach for reducing manual architecture tuning in Transformer-based EHR modeling. Code is publicly available at this https URL.
[MA-12] DesignAgent 3D: Interactive 3D Scene Editing via Designer-like Multimodal Reasoning
【速读】：该论文旨在解决文本引导的3D场景编辑中因自然语言指令语义不明确且需在杂乱3D场景中进行语义锚定而导致的编辑精度不足问题。现有方法多采用单次提示驱动的条件生成范式，难以处理用户意图的模糊性，导致对象定位漂移、遮挡下的追踪失败以及多视角一致性差的“贴纸效应”（sticker effect）。其解决方案的关键在于提出DesignAgent3D——一种交互式多模态智能体框架，将3D场景编辑重构为类设计师的“规划-感知-执行”（Plan-Perceive-Act）范式：首先通过与用户的交互澄清模糊的设计目标，继而基于对3D场景的感知精确定位待修改对象或区域，最后在保持场景一致性的前提下实施可控的视觉修改，并将更新结果融合至底层3D表示中，实现持久化和多视角一致的新视角渲染。实验表明，该方法在NeRF与3D高斯泼溅（3D Gaussian Splatting）两种主流架构上均显著优于当前最优基线，在语义意图对齐、空间定位精度及多视角一致性方面表现卓越。
链接: https://arxiv.org/abs/2608.21438
作者: Xiujin Liu,Tianyu Yang,Yilun Zhao,Xiangliang Zhang
机构: University of Michigan, Ann Arbor(密歇根大学安娜堡分校); University of Notre Dame(圣母大学); Yale University(耶鲁大学)
类目: Computer Vision and Pattern Recognition (cs.CV); Multiagent Systems (cs.MA)
备注:
Abstract:Text guided 3D scene editing provides an intuitive interface for modifying reconstructed environments, but remains difficult because natural language design requests are often semantically underspecified and must be grounded in cluttered 3D scenes. Existing methods typically formulate the task as one-shot conditional generation from a single prompt, failing to resolve ambiguous user intents or achieve precise spatial grounding. Consequently, they suffer from severe object localization drift, tracking failure under occlusions, and the notorious multi-view “sticker effect.” To overcome these limitations, we present DesignAgent3D, an interactive multimodal agentic framework that reformulates 3D scene editing as a designer-like Plan-Perceive-Act paradigm. The agent first plans by interacting with the user to clarify underspecified design goals, then perceives by grounding the intended edit to specific objects or regions in the 3D scene, and finally acts by applying controlled visual modifications while preserving scene consistency. The edits are further integrated into the underlying 3D representation, supporting persistent and multi-view consistent novel-view rendering. Extensive experiments across both NeRF and 3D Gaussian Splatting backbones demonstrate that DesignAgent3D significantly outperforms state-of-the-art baselines, delivering superior semantic intent alignment, impeccable spatial localization accuracy, and high-fidelity multi-view consistency.
[MA-13] Agent ic Security: A Systematization of Tools Failure Modes and Design Laws for LLM -Driven Penetration Testing
【速读】：该论文旨在解决生成式安全（agentic security）系统在从演示阶段向实际部署转化过程中反复出现的运维失败问题。其核心挑战在于多工具、多阶段的自动化安全工作流中，因集成摩擦（integration friction）导致的系统性失效，包括工程成本高、组织与法律合规负担重、维护复杂以及模型决策不可靠等。解决方案的关键在于提出一个四维集成摩擦指数（Integration Friction Index），将一次性工程投入与持续性的组织、法律及维护成本解耦，从而量化并识别可复现的故障模式。通过将代理式安全系统建模为由确定性中介者封装的随机性大语言模型（LLM）策略，论文揭示了长生命周期会话因证据衰减而丧失上下文连续性，而短生命周期子代理可通过原始证据与摘要之间的压缩比扩展有效决策窗口。研究进一步发现，两阶段判断级联虽能放大评分器似然比，但在评分器误差相关时增益有限；将无法评估的结果视为攻击成功会诱发下游测量对规避性与严重性响应的偏差。此外，论文将规划者-执行者路由建模为背包问题，并推导出重尾型工具的闭式执行上限η* = αv/c。最后指出，范围与预算控制无法仅依赖系统提示（prompt）实现，因为提示无法约束实际执行行为。研究以自研平台Inspectra作为实例验证，其功能模块按“已交付”“部分实现”或“计划中”分类，包含失效机制的透明披露，体现了方法论的可操作性与工程严谨性。
链接: https://arxiv.org/abs/2608.21423
作者: Israt Moyeen Noumi,Tarannum Ahmed Nowshin,Md. Mehedi Hasan Nipu,Mohammad Sakib Mahmood,Md. Jakir Hossain,M. F. Mridha
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Multiagent Systems (cs.MA)
备注:
Abstract:Agentic security uses large-language-model (LLM) agents to plan, dispatch, and interpret security tools. As these systems move from demonstrations to deployed products, practitioners repeatedly encounter the same operational failures. We systematize these failures through a hands-on evaluation of ten widely used static, dynamic, cloud, orchestration, and AI red-teaming tools for unattended pipelines. We introduce a four-dimensional Integration Friction Index that separates one-time engineering cost from recurring organisational, legal, and maintenance cost. We then derive quantitative regularities that explain recurring failure modes. Modelling an agentic security system as stochastic LLM policies wrapped by a deterministic mediator, we show that long-lived sessions lose resident evidence with phase count, while short-lived sub-agents extend the usable horizon according to the compression ratio between raw evidence and its summary. We show that a two-stage verdict cascade multiplies scorer likelihood ratios, but provides little benefit when scorer errors correlate. We show that treating unevaluable outcomes as attack failures biases downstream measurements toward evasive and severe responses. We formulate planner-versus-worker model routing as a knapsack problem and derive a closed-form execution cap for heavy-tailed tools, eta* = alpha v/c. Finally, we show why scope and budget enforcement cannot be delegated to system prompts: prompts do not constrain what actually executes. Inspectra, our implemented platform, serves as a worked instantiation, with mechanisms labelled shipped, partial, or planned, including those that did not work.
[MA-14] Diagnosing and narrowing the simulation-to-real gap in powder X-ray diffraction with a wet-dry agent ic loop
【速读】：该论文旨在解决粉末X射线衍射（Powder X-ray Diffraction, PXRD）在实验数据自动化分析中面临的瓶颈问题，即深度学习分析模型在模拟数据上表现优异，但在真实实验谱图上性能显著下降的“仿真-现实差距”（simulation-to-real gap）。其核心挑战在于：该差距源于结构层面的偏差，而非简单的噪声叠加，因此传统通过合成去噪等手段无法有效提升真实谱图的分析效果。解决方案的关键在于引入针对真实谱图的细调（real-spectrum fine-tuning）、峰位对齐的重排序（peak-aligned reranking）以及重新校准（recalibration）策略，从而弥补合成模型在实际应用中丢失的覆盖范围与精度。为此，作者构建了Xtalyst系统，一个由智能体（agent-orchestrated）协同驱动的全流程框架，涵盖相位识别、结构精修及校准性质预测。实验验证表明，该系统在独立保留测试集（n=534）上各模块均能复现开发阶段结果，包括合成锚点的覆盖不足现象，并在真实仪器数据上实现无需达到剖面级拟合即可收敛且保持对称性的精修；同时，在实际操作中，通过“推荐重测-再分析”闭环，成功将盲测硅标准从未通过状态提升至通过状态，并准确分辨出多金属合金中的微弱相。
链接: https://arxiv.org/abs/2608.22400
作者: Shaoguang Wang,Weiyu Guo,Ben Fei,Xiaohong Shao,Zhihui Wang,Wanli Ouyang
机构: The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, China; The Chinese University of Hong Kong, Hong Kong SAR, China; Suzhou National Laboratory, Suzhou, China; Shenzhen Loop Area Institute, Shenzhen, China; Dalian University of Technology, Dalian, China
类目: Materials Science (cond-mat.mtrl-sci); Multiagent Systems (cs.MA)
备注: 72 pages, 15 figures, 7 supplementary tables; supplementary material included
Abstract:Powder X-ray diffraction (PXRD) is the routine probe of crystalline matter, yet its analysis is the rate-limiting step as laboratories automate acquisition. Deep-learning analyzers excel on simulated patterns and degrade on measured ones. This simulation-to-real gap is structural, not additive: synthetic denoising gives no measurable lift on real spectra, whereas correcting a small peak-position drift more than doubles median retrieval correlation. Real-spectrum fine-tuning, peak-aligned reranking, and recalibration narrow what remains and restore the coverage synthetic anchors lose. Xtalyst integrates these in an agent-orchestrated system spanning phase identification, refinement, and calibrated property prediction. On a frozen held-out partition (n=534) each module measured on both splits reproduces its development finding – including the synthetic-anchor under-coverage, whose magnitude differs between the two pools – while held-out refinement converges and preserves symmetry without reaching profile-quality fits, and on a diffractometer its wet-dry recommend-rescan-reanalyze loop flips a blinded silicon standard to a gated PASS and changes which minor phase is resolved on a multi-metal alloy.
自然语言处理
[NLP-0] How to Train a Critic Stably and Efficiently
【速读】： 该论文旨在解决基于评价值（critic-based）的强化学习在大语言模型训练中因标准训练范式不稳定而导致性能受限的问题。尽管单次采样即可通过可靠评价值估计逐标记优势，但现有方法常因值函数估计偏差或优势计算不稳而难以有效应用。其解决方案的关键在于提出最佳实践评价值优化（Best-Practice Critic Optimization, BPCO），该方案整合了多项关键设计：结合深度策略梯度（DPPO）优化框架、将值预测约束于奖励范围内的正则化机制、基于蒙特卡洛目标的值函数更新、未归一化的策略优势计算，以及考虑序列长度自适应的广义优势估计（Generalized Advantage Estimation, GAE）。此外，由于评价值仅在训练阶段使用，BPCO可引入对策略不可见的奖励定义信息（如参考答案或评分量规），从而提升值函数的准确性。控制实验验证了各组件的有效性。在涵盖1.5B至30B-A3B混合专家模型的数学推理任务中，BPCO显著优于强基线，并在仅采样单个响应的情况下达到或超越群体相对优势估计方法的表现，同时在基于评分量规的奖励设置下亦展现出更强的学习能力。结果表明，经过精心设计的评价值能够为群体相对优势估计提供一种稳定且高效的替代方案。
链接: https://arxiv.org/abs/2608.23566
作者: Penghui Qi,Xiangxin Zhou,Wee Sun Lee
机构: National University of Singapore(新加坡国立大学); Tencent Hunyuan(腾讯混元)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Group-based reinforcement learning methods such as GRPO for large language models avoid training a critic by sampling multiple responses for each prompt. A reliable critic could instead estimate token-level advantages from one response, but standard critic-based training recipes are often unstable. We study this instability and develop \textbfBest-Practice Critic Optimization (BPCO), a recipe that combines DPPO, value predictions bounded to the reward range, Monte Carlo value targets, unnormalized policy advantages, and length-adaptive generalized advantage estimation. Because the critic is used only during training, BPCO can also condition it on reward-defining information, such as a reference answer or grading rubric, that is hidden from the policy. Controlled experiments isolate the effect of each design choice. Across mathematical reasoning tasks with models ranging from 1.5B parameters to 30B-A3B mixtures of experts, BPCO improves a strong critic-based baseline consistently, and matches or exceeds a group-based baseline while sampling one response per prompt. The same recipe also improves learning with rubric-based rewards. These results show that a carefully designed critic provides a reliable alternative to group-relative advantage estimation. Code is available at this https URL
[NLP-1] SWE Refactor Bench: Can Coding Agents Complete a Long-Horizon Whole-Repository Stack Migration?
【速读】： 该论文旨在解决生成式 AI 在大规模软件系统重构（如技术债迁移）中面临的“盲目性”（Blindness）问题，即现有基准测试仅评估行为正确性，导致编码代理可通过简单复制原始实现来通过测试，而无法真实完成迁移任务。为克服这一局限，研究提出 SWE Refactor Bench 基准，包含 20 个全仓库级别的重构任务，覆盖四类技术债，并设计三阶段评估协议：（1）迁移审计（Migration Audit）验证重构是否实际发生；（2）行为测试（Behavioural Tests）使用固定测试集评估功能正确性；（3）智能体验证（Agentic Verification）利用六个独立编码代理生成针对性测试以发现隐藏的行为差异。关键发现表明，迁移完整性与行为正确性是两种独立能力，仅有 5.4% 的运行通过全部阶段，最佳模型（claude-opus-5）得分仅为 47.0/100，且在不同重构类型间表现差异显著（如构建工具链重写得分为 31.4，语言迁移仅 5.6）。该研究揭示了当前编码代理在实现可靠、完整重构方面的严重不足，确立了 SWE Refactor Bench 作为评估和推动生成式代码代理在全仓库迁移任务中能力发展的严格基准。
链接: https://arxiv.org/abs/2608.23564
作者: Deyao Hong,Yizhe Chi,Wenyi Li,Xiaoqiu Wang,Mingju Gao,Kaisen Yang,Bingxiang He,Youjie Zheng,Calvin Xiao,Qinhuai Na
机构: Navers Lab, Einsia.AI(艾尼赛); Tsinghua University(清华大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注:
Abstract:Modern software systems accumulate technical debt over decades of development, which makes migration expensive and largely manual. As coding agents become increasingly capable at bug fixing, can they autonomously perform such migrations? Existing benchmarks cannot answer this question because they evaluate only behavioural correctness, not whether the migration actually occurred. This leads an easy hack: agents copy the original implementation to make tests pass. We call this Blindness. To address this problem, we introduce SWE Refactor Bench, a benchmark comprising 20 whole-repository migrations, covering 4 kinds of technical debt. A three-stage evaluation protocol measures both migration completeness and behavioural correctness. (1) Migration Audit verifies that the migration occurred. (2) Behavioural Tests measure correctness with a fixed test suite. (3) Agentic Verification uses 6 independent coding agents to generate targeted tests for hidden behavioural differences. Across 520 runs from 8 frontier models and 26 model-effort configurations, only 28 of 520 runs ( 5.4% ) pass all three stages, 13 of the 20 tasks receive no accepted solution, and the best model (claude-opus-5) scores 47.0/100 . Migration completeness and behavioural correctness are distinct abilities: a few runs preserve behaviour by skipping the migration and are stopped at Migration Audit; most attempt it and break behaviour, and are stopped at Behavioural Tests. Agents cannot deliver a perfect migration: among the 340 runs that pass Migration Audit, 58% reach 99% of the fixed checks, yet only 26% reach 100% . Agent capability differs across migration categories: agents score 31.4 on build toolchain rewrites but only 5.6 on language rewrites. Together, these findings position SWE Refactor Bench as a rigorous testbed for developing coding agents for reliable whole-repository migrations.
[NLP-2] Prime Agent : A Self-Improving RLM Harness
【速读】： 该论文旨在解决生成式 AI 在长时程智能体（long-horizon agency）任务中面临的根本性挑战：语言模型作为顺序处理系统，其能力受限于自身权重和当前上下文，难以有效利用外部信息与计算资源来完成需要长期规划、持续学习与复杂协作的复杂任务。为此，论文提出 Prime Agent——一个开源的长时程评估与代码智能体工作流框架。其核心解决方案在于构建一个持久的 IPython REPL 环境，遵循递归语言模型（Recursive Language Model）抽象，实现程序化上下文处理与运行时计算；同时通过持续性钩子（Continual Harness）机制，跨轨迹保留历史记录、记忆、技能、提示模板及子智能体配置，从而支持连续学习与状态延续。递归子智能体之间通过直接通信进行协调，而“智能体视图”（Agents View）则允许人类实时监控与管理后台运行的会话。该架构在保证执行标准化、可恢复性、验证机制与资源核算的同时，将策略构建权交还给模型本身，形成低侵入性、高表达力的“界面膜”（membrane），有效防止框架故障演变为模型失败，推动评测结果更贴近模型的真实潜在能力。实验表明，Prime Agent 将 ARC-AGI-3 RHAE Best@1 从 30% 提升至 95.5%，并在长上下文编程、GPU 内核生成、模拟器构建及自主 nanoGPT 跑速等任务中达到或超越现有主流框架表现；在 Factorio 任务中进一步揭示了迭代优化对技术持续演进的重要性，以及专用子智能体对并行化工作的显著提升作用。
链接: https://arxiv.org/abs/2608.23552
作者: Seth Karten,Alex L. Zhang,Kevin Thomas,Sebastian Müller,Elie Bakouch,Daniel Auras,Mika Senghaas,Fares Obeid,Konstantin Dunas,Johannes Hagemann,Sami Jaghouar
机构: Princeton University (普林斯顿大学); Prime Intellect; MIT (麻省理工学院)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Software Engineering (cs.SE)
备注: 16 pages, 10 figures. Technical report. Code: this https URL
Abstract:Language models are sequential processors, but long-horizon agency requires external information and computation beyond model weights and active context. Prime Agent is an open-source harness for long-horizon evaluation and coding-agent workflows. A persistent IPython REPL follows the Recursive Language Model abstraction for programmatic context processing and test-time compute, while Continual Harness preserves histories, memories, skills, prompts, and subagent specifications across trajectories. Recursive subagents coordinate through direct agent-to-agent communication, and the Agents View lets humans inspect and manage daemon-backed sessions. Prime Agent standardizes execution, recovery, verification, and resource accounting while leaving strategy construction to the model. This low-friction, expressive membrane prevents harness failures from becoming model failures and pushes measurement toward the model’s true maximal underlying capability. Prime Agent raises ARC-AGI-3 RHAE Best@1 from 30% to 95.5% and matches or exceeds native and popular harnesses across long-context coding, GPU-kernel generation, emulator construction, and autonomous nanoGPT speedruns. On Factorio, we find refinement allows for continuous technology progression and dedicated subagents enable parallelized work. Code is available at this https URL.
[NLP-3] ConvergeFlow: Language Flow with Provable Convergence to Token Embeddings
【速读】： 该论文旨在解决现有连续型语言模型（continuous language models, LMs）在生成过程中无法保证流轨迹终止于有效词元嵌入（token embeddings）的问题，这一缺陷导致模型仍需依赖交叉熵（cross entropy, CE）监督的解码器来实现最终的词元预测，从而限制了纯生成式框架的完整性。其解决方案的关键在于提出一种名为ConvergeFlow的嵌入空间流模型（embedding-space flow-based LM），通过将数据预测器（data predictor）约束于词元嵌入的凸包（convex hull of token embeddings）内，并仅使用由流匹配（flow matching）诱导的均方误差（mean squared error, MSE）目标进行训练，确保在满足一定正则性条件下，即使数据预测器存在误差，流轨迹仍能收敛至有效的词元嵌入。这一机制使得模型可直接输出词元而无需额外的CE监督解码器，实现了端到端的生成式建模。此外，研究还设计了三种采样策略以灵活调控生成困惑度（perplexity）与熵之间的权衡。在OpenWebText上的实验表明，ConvergeFlow在性能上可与现有的连续及离散扩散语言模型相媲美，验证了基于流的语言建模范式的潜力。
链接: https://arxiv.org/abs/2608.23551
作者: Na Li,Yuchen Jiao,Changxiao Cai,Gen Li
机构: Chinese University of Hong Kong (香港中文大学); University of Michigan (密歇根大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Machine Learning (stat.ML)
备注:
Abstract:Recent advances in continuous diffusion and flow-based language models (LMs) have achieved performance competitive with discrete LMs. However, existing continuous frameworks still rely on decoders supervised with cross entropy (CE) because the flow trajectories are not guaranteed to terminate at valid token embeddings. Motivated by this limitation, we introduce \textbfConvergeFlow, an embedding-space flow-based LM, which constrains the data predictor to the convex hull of token embeddings and trains it solely with the mean squared error objective induced by flow matching. Under suitable regularity conditions, we prove that the resulting flow converges to valid token embeddings despite errors in the data predictor, enabling direct token prediction without a CE-supervised decoder. We further develop three sampling mechanisms for controlling the trade-off between the generative perplexity and entropy. Experiments on OpenWebText demonstrate that ConvergeFlow achieves performance competitive with existing continuous and discrete diffusion LMs. These findings demonstrate the potential of the flow-based paradigm for language modeling. Our code is available at this https URL.
[NLP-4] When Names Cross Scripts: A Source-Grounded Benchmark for Historical Entity Reconciliation in the Mongol World
【速读】： 该论文旨在解决历史人物身份辨识（historical entity reconciliation）中的核心难题，即在多语言、多书写系统及不同音译传统背景下，同一人物可能以多种名称形式出现，而不同人物却可能拥有高度相似甚至完全相同的姓名，导致仅依赖名称匹配或音译转换无法准确实现身份关联。其解决方案的关键在于构建一个受出处控制（provenance-controlled）的基准数据集MHER，通过引入基于来源证据（source-grounded evidence）的判断机制，使模型能够依据具体的历史文献来源和上下文信息进行身份推断。实验表明，相较于仅使用名称信息的输入，加入来源证据可使模型在测试集上的准确率提升12.96至94.44个百分点；在名称表面相同但实际为不同人物的极端案例中，仅靠名称会导致全部模型失败（0/25正确），而结合来源证据则能实现24/25的正确识别，其余为合理弃权（abstention）。研究还揭示了上下文信息对身份判断具有显著贡献，并指出名称本身并非总是有益——部分模型在恢复原始表面形式后反而产生误合并，说明名称与上下文需协同使用。因此，该工作强调历史实体辨识不仅依赖表面形式匹配，更关键的是模型能否适切响应经过严格出处控制的历史证据，MHER为此提供了可控的研究框架，用于分析证据利用、弃权行为及错误模式。
链接: https://arxiv.org/abs/2608.23507
作者: Xiang Chen,Zeyu Zhang
机构: University of Amsterdam (阿姆斯特丹大学); Amsterdam UMC (阿姆斯特丹大学医学中心)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 38 pages, 4 figures, 7 tables. arXiv preprint
Abstract:Historical people may appear under different languages, scripts, and transcription traditions, while distinct individuals may share highly similar or even identical names. This makes historical identity reconciliation more than a problem of string matching or transliteration. We introduce MHER, a provenance-controlled benchmark for pairwise reconciliation of person-name attestations from the Mongol world. MHER contains a balanced 396-pair Name-only core over 84 primary historical persons and a stricter 160-pair Source-grounded subset constructed from mention-by-source evidence, with entity-disjoint development and test splits. Across five generative systems, correctly Source-grounded evidence improves paired TEST accuracy by 12.96 to 94.44 percentage points relative to Name-only input. On five identical-surface different-person cases, all models fail under names alone (0/25 model-item decisions), whereas Source-grounded evidence yields 24/25 correct resolutions, with the remaining output an abstention. Context-only ablations show that historical descriptions often carry substantial identity information, while explicitly signaled misgrounding controls produce substantially lower performance. We also find that names are not uniformly beneficial: for Qwen3-8B, restoring surface forms converts ten otherwise correct Context-only distinctions into false identity merges. These results show that historical entity reconciliation depends not only on surface correspondence, but on whether identity judgments respond appropriately to provenance-controlled historical evidence. MHER therefore provides a controlled framework for studying evidence use, abstention, and failure modes in historical NLP. Comments: 38 pages, 4 figures, 7 tables. arXiv preprint Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) ACMclasses: I.2.7; I.2.6; H.3.3 Cite as: arXiv:2608.23507 [cs.CL] (or arXiv:2608.23507v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2608.23507 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-5] Mitigating Reasoning -Induced Misalignment via Safety-Direction Penalty
【速读】： 该论文旨在解决生成式人工智能（Generative AI）在基于推理数据进行微调时出现的“推理诱导偏移”（Reasoning-Induced Misalignment, RIM）问题，即在无有害内容的推理数据（如数学、代码、带思维链的解题）上微调大型语言模型（LLM）后，模型反而表现出有害行为，严重威胁模型推理的安全性。现有研究将RIM归因于神经元层面的纠缠，但未能揭示其背后的表征空间几何结构，也缺乏训练阶段的修复方法。本文的关键贡献在于：首先通过表征空间分析，识别出两个关键激活空间方向——一个编码推理能力，另一个表征安全行为，二者存在耦合关系：提升推理能力的微调会引发安全表征的偏移，且偏移量越大，安全退化越显著；其次，利用核相关性分析（CKA）距离比和探测器定位到与安全决策最相关的层，据此设计了“安全方向惩罚”（Safety-Direction Penalty, SDP），在微调过程中对沿学习到的安全方向的位移施加惩罚。该方法通过初始范围定位与迭代扩展机制，有效抑制有害行为，同时在Qwen2.5-3B和7B模型上实现了安全性的恢复与基准推理性能的保持。
链接: https://arxiv.org/abs/2608.23497
作者: Yipeng Zhao,Qishun Yang,Shenzhe Zhu,Shu Yang,Di Wang
机构: University of Toronto(多伦多大学); King Abdullah University of Science and Technology(阿卜杜拉国王科技大学); University of Texas at Austin(德克萨斯大学奥斯汀分校)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 28 pages, 4 figures
Abstract:Reasoning-Induced Misalignment, where fine-tuning on reasoning data containing no harmful content, including mathematics, code, and problem-solving with chain-of-thought traces can induce harmful behaviors of LLM, posing a serious challenge to the safety of LLM reasoning. Cross-architecture, cross-scale, and cross-dataset checks show that RIM does not always emerge. Previous work attributed RIM to neuron-level entanglement, but did not identify the geometry of the representation space underlying this entanglement or propose a training-time fix. We provide both: a representation-space analysis of RIM and the Safety-Direction Penalty (SDP), which penalizes movement along a learned safety direction during reasoning fine-tuning. The analysis extracts two activation-space directions, one encoding reasoning ability and the other safety behavior. These directions are coupled: fine-tuning that improves reasoning shifts safety representations, and prompts with larger shifts show larger safety degradation. CKA distance ratios and probes locate the safety-decision layers where this shift is most relevant. These findings guide the design of SDP: the coupling motivates penalizing displacement along the safety direction, and the layer localization sets the initial scope. When the initial scope leaves compensatory shifts beyond the penalized layers, the same diagnostics guide iterative expansion. On Qwen2.5-3B and 7B, SDP restores safety while preserving benchmark reasoning performance.
[NLP-6] On the Threat Model of Weird Generalization and Emergent Misalignment
【速读】： 该论文旨在解决生成式模型在小规模、领域特定数据集上进行微调时出现的“奇怪泛化”（Weird Generalization, WG）现象的本质成因问题，即探究哪些微调数据的特征是引发WG的关键因素。研究通过系统分析数据集大小、构成、语言、呈现风格以及与模型参数知识的熟悉程度等潜在相关特征，发现WG的程度主要取决于数据集的构成与语言特性（远超数据量的影响），且在数据与模型预训练知识更接近的情况下表现更强；同时，评估结果对所选测试问题集具有高度敏感性。综合来看，这些发现表明WG是由训练与评估数据中一系列脆弱属性共同决定的，因此其更应被视为一种依赖精心设计数据的对抗性威胁，而非常规微调中固有的重大风险。
链接: https://arxiv.org/abs/2608.23476
作者: Miriam Wanner,Mark Dredze,William Walden
机构: Johns Hopkins University (约翰霍普金斯大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Narrow fine-tuning on small, domain-specific datasets can produce broad and surprising changes in model behavior-a phenomenon called weird generalization (WG). Yet, it remains unclear what features of the fine-tuning data are necessary for WG to arise. Here, we address this question by investigating a range of plausibly relevant features, including dataset size, composition, language, presentation style, and novelty relative to a model’s parametric knowledge. Further, since WG evaluations rely on small question sets that assess the extent of the generalization, we also analyze how sensitive this measurement is to the set of questions used. Experiments with three open-weight models on four datasets show that the degree of WG (1) depends heavily on dataset composition and language (more than on size); (2) is greater for data familiar from pretraining than for novel data; and (3) is sensitive to the set of evaluation questions used. Collectively, these results indicate that WG is a product of quite fragile properties of both training and evaluation data. As such, we argue that WG is more plausible as an adversarial threat-requiring careful data engineering-rather than as a significant hazard inherent to routine fine-tuning.
[NLP-7] Whats the Catch? Evaluating Temporal Consistency in Vision-Language Models ACL
【速读】： 该论文旨在解决当前视觉-语言模型（VLMs）在视频理解中对时间结构建模能力不足的问题，尤其关注其是否具备对时序一致性的敏感性。为探究这一问题，研究将时间定位任务形式化为异常检测问题，提出名为TimeCatch的可控评估框架：通过交换连续帧生成时序异常，或用高斯噪声替换单帧以制造帧级异常。实验在四个合成与真实数据集上评估模型在异常检测与定位任务上的表现，并辅以人类对比研究。结果表明，尽管现有VLMs能稳定识别帧级异常并实现较精准定位，但在时序异常检测任务中表现接近随机水平，定位性能也仅略高于随机基准；而人类在两项任务中均达到接近最优的表现。进一步分析显示，模型规模、提示策略、序列长度及视觉相似性等因素无法完全解释其失败，说明问题根源不在于感知能力或模型容量限制。因此，该研究揭示了当前VLMs虽能捕捉单帧信息，却难以跨帧整合信息以推理时间一致性。TimeCatch提供了一个控制良好的基准，用于系统评估VLMs在时间对齐方面的建模能力。
链接: https://arxiv.org/abs/2608.23474
作者: Marek Hradil,Danae Sánchez Villegas
机构: University of Copenhagen(哥本哈根大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: 17 pages, ACL format
Abstract:Vision-language models (VLMs) achieve strong performance on video and image-sequence benchmarks, yet it remains unclear whether they capture temporal structure. To study this question, we formulate temporal grounding as an anomaly detection problem, providing a simple and controlled evaluation that directly tests sensitivity to temporal consistency. We introduce TimeCatch, where temporal anomalies are created by swapping consecutive frames and frame-level anomalies by replacing a frame with Gaussian noise. Models are evaluated on anomaly detection and localization tasks across four synthetic and real-world datasets, alongside a human study. Our evaluation reveals a substantial gap between frame-level and temporal anomaly detection. While VLMs consistently detect frame-level anomalies and often localize them accurately, they perform near chance on temporal anomaly detection and only modestly above chance on localization. Humans, in contrast, achieve near-ceiling performance on both tasks. Additional analyses across model scales, prompting strategies, sequence lengths, and visual similarity suggest that these failures cannot be explained solely by limitations in perception or model capacity. Together, these findings indicate that current VLMs can identify anomalies within individual frames but struggle to integrate information across frames to reason about temporal consistency. TimeCatch provides a controlled benchmark for evaluating temporal grounding in vision-language models.
[NLP-8] How Useful are LLM s for Grammar Engineering? Cantonese ParGram Resources and Controlled Experimental Evaluation with English Baselines EMNLP2026
【速读】： 该论文旨在解决生成式人工智能（Generative AI）在知识驱动型语法工程中的应用问题，具体聚焦于大语言模型（LLM）能否在受控实验范式下，基于句子或目标形式结构生成可机器处理的语法资源。其核心挑战在于评估当前大模型在处理复杂句法约束、尤其是多构式交互时的准确性和一致性。解决方案的关键在于构建并利用粤语（Cantonese）ParGram资源作为金标准，并与英文基准进行对比，系统性地考察OpenAI的gpt-oss-120b与GPT-5.4在不同提示策略下的表现。研究发现，尽管两者均能生成局部合理的短语结构规则、词项和模板，但难以协调多重形式约束，尤其在多构式场景中表现不足；而从目标形式结构生成的语法优于从句子生成的结果。因此，该研究揭示了当前大语言模型在辅助语法开发中的潜力与局限：可支持语法构建的中间阶段，但人类语言学专家在分析、验证与精细化调整方面仍不可或缺。此外，研究还贡献了新的粤语符号化语法资源，为后续自然语言处理研究提供了基础数据支持。
链接: https://arxiv.org/abs/2608.23448
作者: Chit-Fung Lam
机构: 未知
类目: Computation and Language (cs.CL)
备注: Accepted to Findings of the Association for Computational Linguistics: EMNLP 2026
Abstract:This paper presents new Cantonese ParGram resources and evaluates LLMs for knowledge-driven grammar engineering within a controlled experimental paradigm. Using Cantonese ParGram resources as gold standards, with corresponding English baselines, we investigate whether OpenAI’s gpt-oss-120b and GPT-5.4 can generate machine-processable grammars from sentences and target formal structures under systematically varied prompting conditions. GPT-5.4 outperformed gpt-oss-120b, while grammars generated from target formal structures generally outperformed those generated from sentences. Although both models could generate locally plausible phrase-structure rules, lexical entries, and templates, they often struggled to coordinate interacting formal constraints, especially in multi-construction settings. The results characterize both the capabilities and limitations of current LLMs for potential integration into AI-assisted expert workflows: LLMs may support intermediate stages of grammar development, but human linguistic expertise remains central to analysis, validation, and refinement. The study also contributes new Cantonese symbolic grammatical resources.
[NLP-9] A Comprehensive Analysis of Arabic Natural Language Processing Research: Trends Topic Evolution and Research Gaps – A Bibliometric and Topic-Based Study
【速读】： 该论文旨在解决阿拉伯语自然语言处理（Natural Language Processing, NLP）领域缺乏系统性、大规模定量综合分析的问题。尽管近年来受数字转型、社交媒体及大语言模型（Large Language Models, LLMs）推动，阿拉伯语NLP研究呈现快速增长，但尚无全面的量化评估。其解决方案的关键在于构建一个涵盖1960至2026年间7,120篇阿拉伯语NLP文献的大规模书目计量与主题分析框架，结合BERTopic进行主题建模、回归分析识别引用预测因子、社会网络分析揭示合作者结构，并通过地理可视化呈现研究分布。研究发现，2020年后发表量显著上升，主要由Transformer模型和LLMs驱动；主题建模识别出19个核心研究主题，其中以文本、语音、翻译与识别为主；引用分析显示论文年龄与引用次数呈正相关（r = 0.245, p < 0.001），且在OpenAlex或Semantic Scholar索引以及机构隶属关系下引用更高；沙特阿拉伯、美国和埃及为研究产出领先国家。通过任务-方言差距矩阵，进一步揭示了马格里布、伊拉克及苏丹方言在摘要生成等任务上的严重研究不足。此外，最大主题的H指数达87，情感分析为54，凸显部分领域的成熟度差异。该研究通过量化方法补充现有定性综述，提出应优先支持资源匮乏方言的研究，并开发符合文化语境的基准评测体系，为未来阿拉伯语NLP发展提供数据驱动的战略建议。
链接: https://arxiv.org/abs/2608.23421
作者: Mullosharaf K. Arabov
机构: Kazan Federal University (喀山联邦大学); Institute of Computational Mathematics and Information Technologies (计算数学与信息技术研究所)
类目: Computation and Language (cs.CL)
备注: 38 pages, 16 tables, 10 figures. Preprint
Abstract:Natural Language Processing (NLP) has grown rapidly over the past decade, driven by digital transformation in the Arab world, social media, and large language models (LLMs). Despite this growth, a comprehensive quantitative meta-analysis of the field remains absent. This study presents a large-scale bibliometric and topic-based analysis of 7,120 Arabic NLP papers published between 1960 and 2026, sourced from six collections. We employ BERTopic for topic modeling, regression analysis to identify citation predictors, social network analysis for co-authorship structures, and geographic mapping. Our findings show a significant publication surge after 2020, driven by transformer models and LLMs. Topic modeling identifies 19 substantive themes, the largest centered on text, speech, translation, and recognition. Citation analysis reveals a positive correlation between paper age and citations (r = 0.245, p 0.001); regression shows that indexing in OpenAlex or Semantic Scholar and institutional affiliation are associated with higher citation counts. Saudi Arabia, the United States, and Egypt lead in research output. A task-dialect gap matrix identifies critical understudied areas, including summarization for Maghrebi, Iraqi, and Sudanese dialects. The largest topic has the highest H-index (87), followed by sentiment analysis (54). Our quantitative approach complements existing qualitative surveys and offers recommendations to prioritize under-resourced dialects and develop culturally aligned benchmarks for Arabic NLP.
[NLP-10] STONIC: A Layered Measurement Contract for LLM Value Profiling
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）价值评估中普遍存在的核心问题：即如何确保通过问卷评分、成对选择与生成文本推断所得的价值偏好在不同评估范式下具有一致性与可比性。现有研究通常将三类数据合并为单一价值画像，隐含假设这些观测反映的是同一稳定偏好。然而，本文通过STONIC实验框架，在来自四家银行的5,144个情境下对35种固定模型配置进行系统检验，揭示了该假设的局限性。其关键发现在于：尽管部分模型配置（10/17）在跨银行情境中保持了“支持-选择”关系的一致性，但所有可分析配置均表现出对自身先前回答的显著偏好（中位效应值0.790），且选项位置显著影响选择率，表明行为响应存在显著的界面依赖性。此外，价值轮廓从评分到冲突选择的传递最强，而在自发生成文本中逐渐弱化；三重标注200条L3响应的结果进一步验证了任务局部语义审计的有效性——FULCRA与人类多数意见最一致，而经校准后的DeBERTa仍保留有用排序信息。更重要的是，隐藏状态比提示词本身更清晰地编码了已完成的决策过程。综上，虽然模型展现出可重复的行为连续性，但证据不支持存在一种独立于评估接口的统一价值身份，从而挑战了当前主流多源融合方法的合理性。
链接: https://arxiv.org/abs/2608.23411
作者: Andrei Chetvergov,Stepan Ukolov,Timofei Sivoraksha,Alexander Evseev,Danil Sazanakov,Mikhail Solovev,Sergey Bolovtsov
机构: Russian Academy of National Economy and Public Administration (俄罗斯联邦国家行政学院)
类目: Computation and Language (cs.CL)
备注: 32 pages, 6 figures, including appendices
Abstract:LLM value studies often merge questionnaire ratings, pairwise choices, and values inferred from generated text into one profile. That merge assumes that the three observations describe the same stable preference. STONIC tests this assumption on 5,144 situations from four banks and 35 fixed model configurations. It compares responses rated in isolation, choices made under counterbalanced conflict, spontaneous answers, and later choices between a model’s own answer and authored alternatives. 10 of 17 configurations with usable behavioral data preserve the endorsement-choice relation across banks. Every one of the 17 eligible configurations prefers its own earlier answer (median effect 0.790), although option position changes the choice rate in every eligible configuration. Profile shape transfers most strongly from ratings to conflict choices and weakens for spontaneous text. Three-way annotation of 200 L3 responses provides a task-local check of the semantic audit: FULCRA agrees most closely with the human majority, while DeBERTa retains useful rank information after calibration. Hidden states encode the completed decision more clearly than the prompt alone. Thus the models show reproducible behavioral continuity, but the evidence does not support one scorer-independent value identity across interfaces.
[NLP-11] Cross-Domain Multi-Task Data-to-Text Generation without In-Domain Training Data EMNLP
【速读】： 该论文旨在解决跨领域结构化数据到文本生成（cross-domain data-to-text generation, D2T）中的关键挑战，即在缺乏目标领域训练文本与测试参考文本的情况下，如何实现对多种异构输入结构（如表格、知识图谱、时序数据等）和不同生成目标的鲁棒生成。其核心问题在于现有方法依赖特定任务的数据或大语言模型的零样本能力，难以在无领域先验条件下保持高性能。解决方案的关键是提出一种数据驱动的知识蒸馏（Data-Driven Knowledge Distillation, DDKD）框架，并引入结构保持型增强策略——通过结构子采样与扰动来扩充有限的跨域训练数据，从而提升小规模模型（1.7B参数）在多领域场景下的泛化能力。实验表明，该方法在五个基准上均优于微调与零样本推理，且小型蒸馏模型在两个领域超越更大规模微调模型，在其余三个领域表现相当，同时证明了所提增强策略相比单纯扩大真实目标域输入更具有效性与成本效益。
链接: https://arxiv.org/abs/2608.23391
作者: Yifei Song,Kun Efimov-Zhang,Claire Gardent
机构: CNRS/LORIA; Université de Lorraine
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted by EMNLP Findings 2026
Abstract:Structured data exists in many forms (tables, knowledge graphs, charts, and time series), and converting it into text may involve different generation tasks. However, most prior work on data-to-text (D2T) generation has focused on specific tasks and datasets, relying either on task-specific training data or on the zero-shot capabilities of large language models. We study cross-domain D2T generation in a setting where neither in-domain training text nor test references are available, and where domains, generation goals, and input structures vary substantially. We compare data-driven knowledge distillation (DDKD) against zero-shot inference and fine-tuning on out-of-domain D2T data, and introduce structure-preserving augmentation via structural subsampling and perturbation. Experiments on five benchmarks show that, at constant model size (1.7B parameters), DDKD consistently outperforms both fine-tuning and zero-shot inference. Moreover, the resulting small models outperform a much larger finetuned model on two of the five domains, achieving comparable performance on the remaining three. We further construct QUINTD-5, a fivefold extension of QUINTD-1, and show that simply scaling real target-domain inputs yields only modest gains, whereas our augmentation strategy remains more effective and more cost-efficient for cross-domain distillation.
[NLP-12] Cross-lingual Biography Enrichment via Claim Extraction and Alignment EMNLP2026
【速读】： 该论文旨在解决英文维基百科在人物传记信息覆盖上的局限性问题，尤其针对非英语语境下女性人物的信息缺失与不充分。尽管英文维基百科常被视为默认的百科全书来源，但其对长尾人物（尤其是来自非英语国家的女性）的记载往往不够全面。为此，论文提出跨语言传记增补（cross-lingual biography enrichment）任务，即利用非英文维基百科中关于同一人物的传记内容，为英文传记补充经验证的事实。其解决方案的关键在于构建了一个名为\textscCLAW-4L的基准数据集，包含300对英文与法文、中文或阿塞拜疆文传记的配对，并附带声明标注和细粒度的声明对关系语料库；同时提出一种基于声明的增补框架，通过提取中英双语传记中的声明、进行对齐以识别非英文传记中的增补证据，并最终重写英文传记以提升其信息完整性。实验结果表明，非英文维基百科可提供有价值的增补证据，但在低资源语言场景下仍面临显著挑战。
链接: https://arxiv.org/abs/2608.23390
作者: Yifei Song,Ziyang Chen,Emil Sayilov,Claire Gardent
机构: CNRS/LORIA(法国国家科学研究中心/洛林计算机科学与自动化研究所); Université de Lorraine(洛林大学); Université Paris Dauphine - PSL(巴黎达夫金大学-巴黎萨克雷大学联盟); ICube Laboratory, Université de Strasbourg(斯特拉斯堡大学伊库布实验室)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: accepted by EMNLP 2026 main conference
Abstract:English Wikipedia is often treated as the default encyclopedic source, yet non-English Wikipedia editions can contain richer locally grounded information for long-tail figures. We study cross-lingual biography enrichment: enriching an existing English biography with facts supported by a non-English biography about the same person. Focusing on women from non-English-speaking contexts, we introduce \textscCLAW-4L, a benchmark consisting of 300 Wikipedia biography pairs linking an English biography with its French, Chinese or Azerbaijani counterpart, along with claim annotations and a fine-grained claim-pair relation corpus. We propose a claim-based enrichment framework that extracts English claims from both biographies, aligns them to identify enrichment evidence from the non-English biography, and rewrites the English biography using the selected claims. Our results show that non-English Wikipedia biographies provide valuable evidence for improving English biography coverage, while lower-resource settings remain challenging.
[NLP-13] he Geometry of Low-Resource Language Representations
【速读】： 该论文旨在解决大语言模型（LLM）中低资源语言与高资源语言之间存在的性能差距问题，尤其聚焦于揭示导致这一差距的内部模型因素。研究表明，这种差距在模型的表示几何结构（representational geometry）层面具有系统性表现，特别是在模型的最终层，低资源语言的表示呈现出退化现象（representational degeneration）。论文提出的关键解决方案是引入基于几何结构的正则化策略，在持续预训练（CPT）过程中对表示退化进行惩罚。实验结果表明，该方法能有效缓解低资源语言在单语微调过程中的表示退化问题；对于较大模型，基于余弦相似度的正则化虽仅带来边际性能提升，但在最具挑战性的任务上表现出更稳定的增益。研究证实，低资源与高资源语言在模型表示几何上的差异可被量化，且针对几何结构的定向干预是提升低资源语言持续预训练效果的有效途径。
链接: https://arxiv.org/abs/2608.23358
作者: Francois Meyer,Jan Buys
机构: University of Cape Town (开普敦大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:The performance gap between low- and high-resource languages in LLMs is widely known, but it remains unclear which internal model factors drive these disparities. In this paper, we characterise this gap through the lens of representational geometry. Comparing the geometric properties of hidden representations across 30 languages reveals that LLM geometry is systematically related to language data availability. The most consistent effect is in final layers, where low-resource languages exhibit representational degeneration. To counter this, we investigate the effectiveness of regularisation terms to penalise degeneration during continued pretraining (CPT). Experiments monolingually adapting 9 base LLMs to 10 African languages show that geometric regularisation successfully reduces representational degeneration during CPT. For larger models, cosine similarity-based regularisation marginally improves performance over vanilla CPT, with more consistent gains on the most challenging tasks. We establish that the representational geometry of low- and high-resource languages in LLMs is measurably distinct, and that targeted geometric intervention is a viable strategy for improving CPT for low-resource languages.
[NLP-14] FormuEvo: LLM -Guided Evolution for Discovering Solver-Efficient Mixed-Integer Programming Formulations EMNLP2026
【速读】： 该论文旨在解决生成式人工智能（Generative AI）在混合整数规划（Mixed-Integer Programming, MIP）建模中普遍存在的“形式强度不足”问题。尽管大语言模型（Large Language Models, LLMs）能够生成语义正确的自然语言到MIP的自动建模，但其忽略目标函数与约束条件的数学紧致性（formulation strength），导致下游求解器效率严重受限。本文提出的FormuEvo是一种基于LLM引导的进化框架，其核心解决方案在于将MIP建模过程建模为符号空间中的进化优化问题，通过可执行的建模程序表示候选公式，并利用LLM驱动的交叉、变异与修复操作实现迭代生成与选择。关键创新点包括：引入求解器感知诊断机制，利用细粒度求解器统计信息作为“语义梯度”指导针对性改进；构建结构化记忆系统，将历史经验抽象为可复用的建模策略，从而避免重复探索并支持零样本迁移及小规模模型的自举训练。实验表明，FormuEvo在多种线性和非线性问题上均能发现显著优于专家设计和现有LLM方法的高效公式，使求解速度提升最高达5.5倍，且知识可在不同问题与模型规模间有效迁移。
链接: https://arxiv.org/abs/2608.23353
作者: Haofeng Yuan,Jianing Peng,Jieyi Bi,Ni Zhang,Shiji Song,Zhiguang Cao
机构: Nanyang Technological University (南洋理工大学); Singapore Management University (新加坡管理大学); Tsinghua University (清华大学)
类目: Computation and Language (cs.CL); Neural and Evolutionary Computing (cs.NE)
备注: 27 pages, 6 figures, and 9 tables. To appear in the Proceedings of EMNLP 2026
Abstract:Mixed-integer programming (MIP) lies at the core of operations research and industrial optimization. While large language models (LLMs) have recently shown promise in automated MIP modeling from natural language, they prioritize semantic correctness but overlook formulation strength, severely bottlenecking the efficiency of downstream solvers. We propose FormuEvo, an LLM-guided evolutionary framework for automated discovery of solver-efficient MIP formulations. FormuEvo frames MIP formulation design as evolutionary optimization over the symbolic space of MIP formulations, represented as executable modeling programs, by iteratively generating, evaluating, and selecting stronger candidates via LLM-driven crossover, mutation, and repair operations. To move beyond blind exploration, FormuEvo introduces a solver-informed diagnosis mechanism that exploits fine-grained solver statistics as verbal gradients for targeted refinement. Additionally, a structured memory abstracts prior experience into reusable modeling strategies, avoiding redundant exploration while enabling zero-shot transfer to unseen problems and bootstrapping smaller LLMs. Experiments across diverse linear and non-linear problems demonstrate that FormuEvo discovers formulations that significantly outperform both expert-designed formulations and existing LLM-based approaches, accelerating solvers by up to 5.5 \times , with distilled knowledge transferring effectively across problems and model scales.
[NLP-15] Flesch-Kincaid Readability Depends Only on the Topic Distribution in Long Texts under Topic Models
【速读】： 该论文旨在解决长文本中可读性评分（如Flesch Reading Ease, FRE 与 Flesch-Kincaid Grade Level, FKGL）在不同词汇组成下的稳定性问题，即尽管这些评分基于相同的两个文档统计量，其在长文档中的稳定表现是否真正反映可读性信号，还是仅由主题分布决定。其核心解决方案在于提出一个显式包含句边界标记的主题模型，证明在长文本极限下，两类可读性评分几乎必然收敛为文档主题分布的确定性函数，且所有评分变异仅由主题构成所主导，而非残余的可读性特征。理论分析表明，等分值水平集在局部呈(K-2)维曲面结构，而通过内部主题向量的纤维仅为(K-3)维，揭示了评分空间的内在几何特性。实验在布朗语料库（Brown）和书面版英国国家语料库（BNC）上验证了这一理论，发现从一半文档内容词推断出的主题向量可显著预测另一半的FKGL（相关系数分别为0.779和0.884），且在BNC上加入主题预测后，模型解释力提升ΔR²=0.024（中位数0.021），但在布朗语料库中增量不显著。由于推断出的主题可能同时捕获体裁、语域与风格等混杂因素，研究结果不应被解读为对人类可读性或因果效应的直接证据。
链接: https://arxiv.org/abs/2608.23327
作者: Yo Ehara
机构: Tokyo Gakugei University (东京学艺大学); Tokyo, Japan
类目: Computation and Language (cs.CL)
备注:
Abstract:Flesch Reading Ease (FRE) and the Flesch-Kincaid Grade Level (FKGL) are widely used readability scores for English computed from the same two document statistics, yet their stability on long documents need not imply invariance to lexical composition. Surprisingly, under a topic model with an explicit sentence-boundary token, both scores converge almost surely to deterministic functions of the document topic distribution through just two scalar rates: in the long-text limit, all score variation is mediated by topical composition rather than any residual readability signal. The theory covers both formulae, while the experiments evaluate FKGL. In a fixed admixture with rank[1, q, s] = 3, fibres through interior topic vectors are locally (K-3)-dimensional, whereas regular iso-score level sets are locally (K-2)-dimensional and curved. In out-of-fold evaluation on two balanced corpora, Brown and the written BNC, a topic vector inferred from one document half’s content words predicts the other half’s FKGL at r = 0.779 and 0.884, respectively. On Brown, adding the topic prediction to genre and mean content-word syllable count yields \Delta R^2 = 0.002, with a confidence interval spanning zero; on the BNC, the corresponding split-half increment is 0.024, positive in four of five K = 100 fits (median 0.021). Because inferred topics may also absorb genre, register, and style, we do not interpret these results as evidence about human readability or causal effects.
[NLP-16] Agent -G2: Gaussian Guidance for Agent ic Reinforcement Learning
【速读】： 该论文旨在解决长时序智能体任务中因奖励稀疏性导致的强化学习效率低下问题，其核心挑战在于如何有效利用专家轨迹中的提示（hint）来引导策略探索。现有方法将提示深度（guidance depth）设定为固定标量，存在两方面局限：调度式方法采用统一深度忽略任务异质性，而逐样本探测虽能个性化估计深度但需额外采样开销。本文发现，有效的提示信息分布于一个特定深度区间内，其信息量随深度呈近似高斯分布，而非集中于单一最优点。为此，提出Agent-G²框架，采用在线估计的高斯分布动态采样每项任务的提示深度，其均值融合全局基准与聚类级难度，方差则追踪组内差异。该方法无需额外探针采样或训练深度预测器，显著降低计算成本。在ALFWorld和WebShop基准上基于Qwen2.5-1.5B/7B-Instruct模型评估显示，Agent-G²在低于逐样本探测三分之一采样成本下，相较最强的提示增强、无提示及辅助强化学习基线分别提升2.3、3.9和7.4分，验证了其高效性与鲁棒性。
链接: https://arxiv.org/abs/2608.23318
作者: Zixuan Wang,Yanrui Miao,Zhengxi Lu,Teng Pan,Yiwen Qiu,Hongxing Li,Peng Qiu,Ruiqing Zhang,Yongliang Shen
机构: Baidu Inc.(百度); Zhejiang University(浙江大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: Code: this https URL ; Project page: this https URL
Abstract:Hint-based reinforcement learning addresses reward sparsity in long-horizon agentic tasks by retaining a prefix of an expert trajectory before each rollout, letting the policy explore from a state closer to success. Its effectiveness hinges on the guidance depth: how much of the trajectory to keep. Existing methods treat this depth as a deterministic scalar. Scheduled approaches share one value across samples and ignore per-task heterogeneity; per-sample probing estimates it separately at the cost of extra rollouts. We find that useful guidance occupies a band of depths whose informativeness profile is approximately Gaussian around the band center, rather than concentrating at a single optimal point. We propose Agent-G ^2 , a Gaussian guidance framework that draws the depth per task from a Gaussian whose center and spread are estimated online from rollouts already collected for policy optimization, requiring no probe rollouts or learned depth predictor. The center combines a global baseline with per-cluster difficulty, and the spread tracks within-cluster variance. We evaluate Agent-G ^2 on ALFWorld and WebShop on Qwen2.5-1.5B / 7B-Instruct. Agent-G ^2 outperforms the strongest hint-based, hint-free, and Aux-RL baselines on ALFWorld by 2.3 / 3.9 / 7.4 points at under one-third the rollout cost of per-sample probing.
[NLP-17] Beyond the Stability-Exploration Dilemma: Environmental Regularization for LLM Policy Optimization EMNLP2026
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在策略优化（Policy Optimization, PO）过程中面临的稳定性-探索权衡难题。当前方法通常依赖于动作侧的策略KL正则化（Policy-KL regularizer），这导致实践者陷入两难：保留该正则化会限制响应行为并耗尽动作侧的探索预算，而移除则缺乏对策略漂移的显式控制。为此，论文提出一种新范式——环境正则化策略优化（Environment-Regularized Policy Optimization, ERPO），其核心在于将正则化机制从动作侧迁移至输入侧。具体而言，ERPO引入查询KL（Query-KL, QKL）项，以约束由当前策略诱导的训练查询分布相对于预强化学习参考分布的偏移；同时，采用基于静态数据集参考分布的每查询权重，使每次更新偏向于参考分布下的典型查询。由于QKL梯度仅通过查询似然传播，且不作用于策略梯度估计器中的响应评分函数，因此不会对响应分布施加直接梯度压力，从而有效保留了探索能力。该方法可无缝集成至GRPO/PPO/REINFORCE等主流优化框架中，无需额外前向传播。在六个数学推理基准上的实验表明，ERPO可替代传统的策略KL正则化，在有效控制查询分布漂移的同时，显著提升准确性并增强高温解码与长序列生成下的稳定性。
链接: https://arxiv.org/abs/2608.23311
作者: Xianlei Zhou,Xiangdi Meng,Yu He,Tianyu Qi,Shuyan Guan,Xianli Zhang,Jian Zhang,Xin Li,Qika Lin,Jun Liu
机构: AMAP, Alibaba Group(阿里巴巴集团); Xi’an Jiaotong University(西安交通大学); JD.com; Beijing Normal University(北京师范大学); National University of Singapore(新加坡国立大学)
类目: Computation and Language (cs.CL)
备注: Accepted to EMNLP 2026 main conference
Abstract:Policy optimization (PO) for Large Language Models faces a stability–exploration trade-off, currently mediated by an action-side Policy-KL regularizer. This puts practitioners in a double bind: keeping Policy-KL constrains response behavior and consumes the action-side exploration budget, while dropping it leaves the optimization without an explicit drift control. We argue for an alternative that breaks the dilemma by moving regularization to the input side. As training progresses, the distribution over training queries induced by the current policy drifts unchecked from its pre-RL reference distribution. Concretely, Environment-Regularized Policy Optimization (ERPO) introduces a Query-KL (QKL) term that bounds this query distribution shift, together with a dataset-static reference-derived per-query weight that biases each per-query update toward queries typical under the reference. The QKL gradient flows strictly through the query likelihood; the response score function used by policy-gradient estimators does not appear in the QKL term, so QKL exerts no direct gradient pressure on the response distribution—exploration is preserved. ERPO plugs into GRPO/PPO/REINFORCE-style pipelines without additional forward passes. On six mathematical reasoning benchmarks, ERPO replaces the standard Policy-KL regularizer while achieving effective control over query distribution drift, delivering stronger accuracy and substantially more stable behavior under high-temperature decoding and long-horizon this http URL source code are available at this https URL Comments: Accepted to EMNLP 2026 main conference Subjects: Computation and Language (cs.CL) Cite as: arXiv:2608.23311 [cs.CL] (or arXiv:2608.23311v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2608.23311 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-18] Dynamic Topic Modeling for Cross-Corpus Temporal Analysis CIKM2026
【速读】： 该论文旨在解决动态嵌入主题模型（Dynamic Embedded Topic Models, D-ETM）在跨语料库比较中面临的主题对应不一致问题，即传统方法通常独立训练各语料库的主题模型，仅在训练后通过后处理（如匈牙利匹配）进行对齐，难以保证跨语料库和时间维度上的稳定主题对应关系。其解决方案的关键在于提出一种新型的D-ETM框架：首先在合并的多语料库集合上学习一个共享的动态主题空间（称为“共享骨干”），随后在冻结的共享骨干基础上引入语料库特定的残差适应模块，而非构建独立的潜在主题空间。该设计在保持统一主题索引以支持跨语料库比较的同时，允许各语料库保留自身的词汇特异性。实验在涵盖97年历史跨度的三个时序语料库（美国历史英语语料库、哈佛商业评论、国际劳工评论）上验证了该方法的有效性，结果显示，残差适应在提升语料库特定拟合度的同时，实现了显著优于全微调（97.5 ± 0.7% 对比 17.9 ± 1.1% 的轨迹检索准确率@1）和独立训练加后处理匹配的跨语料库主题轨迹对齐效果，证明将主题对齐机制内嵌于模型架构中可有效增强长期跨语料库比较的稳定性，同时保留语料库间的词汇差异。
链接: https://arxiv.org/abs/2608.23284
作者: Ruoxuan Li,Bruce Kogut
机构: Columbia University(哥伦比亚大学); Columbia Business School(哥伦比亚商学院)
类目: Computation and Language (cs.CL)
备注: 12 pages, 2 figures. Accepted at CIKM 2026
Abstract:Dynamic Embedded Topic Models (D-ETM) provide an interpretable framework for modeling temporal semantic evolution, but cross-corpus comparison remains difficult because topics are often learned independently and aligned only after training, a process that does not guarantee stable topic correspondence across corpora and time. To address this problem, we propose a D-ETM framework that first learns a common dynamic topic space over a merged multi-corpus collection, which we call the shared backbone, then introduces corpus-specific residual adaptation around the frozen backbone without creating separate latent topic spaces. This design preserves a shared topic index for cross-corpus comparison while allowing each corpus to specialize lexically. We evaluate the framework on three temporally structured corpora spanning 97 years: the Corpus of Historical American English, Harvard Business Review, and International Labour Review. Residual adaptation improves corpus-specific fit relative to the shared backbone while preserving the same-index cross-corpus topic trajectories, achieving substantially stronger alignment than full fine-tuning from the same backbone, with 97.5 \pm 0.7% versus 17.9 \pm 1.1% trajectory Retrieval@1, as well as stronger alignment than independent training with post-hoc Hungarian matching. These results suggest that incorporating topic alignment into the model can support more stable over-time cross-corpus comparisons while retaining corpus-specific lexical variation.
[NLP-19] Apodex 1.1: Scaling Agent ic Intelligence for Complex Work
【速读】： 该论文旨在解决通用语言模型在处理复杂专业任务时缺乏持续交互能力的问题，即在面对需要长期状态维护、多源信息整合、代码执行、容错恢复及可验证交付的现实目标时，现有模型难以实现可靠、可持续的工作推进。其核心解决方案在于构建具备“工作能力”（working capability）的智能体系统，通过两个互补维度实现突破：一是环境扩展（Environment Scaling），大幅增强可执行文件、搜索与代码环境的多样性与可验证性；二是智能体协同扩展（Agentic Coordination Scaling），训练智能体完成长周期任务的分解、并行任务委派、异步结果集成与动态重规划。系统采用共享执行框架与AgentOS统一管理任务状态与溯源信息，通过训练将环境轨迹与协作过程转化为稳定可靠的行为模式。实验表明，即使参数量远小于前沿系统，Apodex 1.1 在金融、科研、数学、编程与检索等复杂领域仍达到领先性能水平；其350亿参数的轻量化版本Apodex 1.1 Mini亦具备本地部署能力，保持强大的工作能力。该成果将智能体智能扎根于可验证、跨时间的实际工作产出，推动实现面向高难度、长周期任务的“重型求解器”（Heavy-Duty Solver）目标。
链接: https://arxiv.org/abs/2608.23283
作者: Apodex Team:B. An,B. Li,B. Wang,B. Zhang,B.L. Wang,C. Feng,C. Wei,C. Xue,C. Zhang,D. Ng,D. Ye,E. Min,F. Chen,F. Liu,F. Yang,F. Ye,H. Xu,H. Yang,H. Ye,H. Zhang,H. Zhao,J. Li,J. Lin,J. Xia,K. Jin,K. Wang,K. Yang,L. Bing,L. Lei,L. Su,Le. Wang,Lu. Wang,N. Wang,Q. Ren,Q. Yang,R. Li,S. Bai,S. Du,S. Li,S. Lin,S. Nie,S. Wang,S. Zhang,S.Z. Wang,Ta.Q. Fang,Ti.Q. Fang,W. Fang,W. Li,W. Zhang,X. Chen,X. Li,X. Tang,X. Wang,X. Xu,X. Zhang,X.Q. Wang,X.Y. Wang,Y. Deng,Y. Gao,Y. Hu,Y. Li,Y. Sui,Y. Wang,Y. Xiao,Y. Zhang,Z. Chen,Z. Cheng,Z. Feng,Z. Liang,Z. Zhang
机构: Apodex Team(阿波德克斯团队)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:General-purpose language models can reason and synthesize knowledge, but complex work also requires sustained interaction with files, information sources, and executable code, together with state maintenance, failure recovery, and verifiable delivery. We call this \emphworking capability: sustained, verifiable progress toward a real-world objective. Apodex 1.1 develops this capability along two complementary dimensions. \emphEnvironment Scaling expands the diversity and verifiability of executable file, search, and code environments, while \emphAgentic Coordination Scaling trains agents to decompose long-horizon tasks, delegate parallel work, integrate asynchronous results, and replan. A shared execution harness and AgentOS maintain task state and provenance across tools and agents, and training turns environment trajectories and coordination traces into reliable behavior. Across complex professional work, finance, scientific research, mathematics, coding, and search, Apodex 1.1 reaches the leading performance band despite using a substantially smaller model than many frontier systems. The 35B-parameter Apodex 1.1 Mini further retains strong working capability in a locally deployable form. These results ground agentic intelligence in useful, verifiable work completed over time and advance our goal of building a \emphHeavy-Duty Solver for ambitious, long-running tasks.
[NLP-20] EvoWiki: Incremental State Overwriting and Traceable Question Answering for Cross-Meeting Knowledge Evolution
【速读】： 该论文旨在解决长期协作场景中动态知识状态演化问题，即在多轮会议中，事实性信息（如决策、风险等）频繁被修订、推翻或替换，而现有长上下文处理方法通常简单堆叠历史记录，或依赖静态/追加式知识组织方式，缺乏对知识生命周期的显式建模，导致新旧状态共存或历史信息丢失，进而引发检索过时、答案不可验证等问题。其解决方案的关键在于提出EvoWiki架构，通过将离线增量构建（BUILD）与在线结构化读取（READ）解耦：BUILD阶段利用实体版本链和细粒度的状态覆盖协议（State-Overwrite Protocol），显式区分当前有效状态与被取代的历史状态，同时保留会议级溯源锚点；READ阶段摒弃基于语义相关性的Top-k检索，采用确定性的实体定位、时间分辨率及跨实体多跳聚合机制，在完整知识库上生成可追溯、可验证的答案。此外，研究还构建了CrossMeet这一高保真双语基准，涵盖事实一致性、时间推理与跨会议多跳推理任务，实验表明EvoWiki在六个数据集和两种阅读模型上分别比最强基线提升9.72和10.00个百分点的宏平均法官准确率，且人工评估证实其在频繁状态变更下具备更强鲁棒性与事实忠实性，验证了以有效状态为导向的读取机制在跨会议知识演化中的可靠性。
链接: https://arxiv.org/abs/2608.23265
作者: Dongsheng Chen,Tianyu Wang,Wenhui Que
机构: 未知
类目: Computation and Language (cs.CL)
备注: 12 pages, 5 figures, 10 tables
Abstract:In long-term collaboration spanning multiple meetings, factual states such as decisions and risks are continually revised, overturned, and replaced. Existing long-context methods typically stack the entire history, while many RAG and structured-memory methods organize knowledge as static or append-only facts and rely on semantic relevance at read time. Without explicit modeling of knowledge lifecycles, these approaches may retain conflicting old and new states simultaneously or discard history, leading to stale retrieval and answers that are difficult to verify. We present EvoWiki, an incremental question-answering architecture for dynamic long-form text. EvoWiki decouples offline incremental construction (BUILD) from online structured reading (READ). BUILD captures the intra-meeting micro-evolution from proposal to decision and uses entity version chains and a fine-grained State-Overwrite Protocol to explicitly distinguish current valid states from superseded history while preserving meeting-level provenance anchors. READ bypasses relevance-based Top-k retrieval and performs deterministic entity addressing, temporal resolution, and cross-entity multi-hop aggregation over the complete Wiki to produce grounded and traceable answers. We further introduce CrossMeet, a high-fidelity bilingual benchmark designed to simulate long-term state evolution, covering factual consistency, temporal reasoning, and cross-meeting multi-hop reasoning. Across six datasets and two reader models, EvoWiki improves macro-average Judge Accuracy over the strongest baselines by 9.72 and 10.00 percentage points, respectively. Human evaluation shows that EvoWiki is more robust and factually faithful under frequent state flips, validating valid-state-oriented reading as a reliable approach to cross-meeting knowledge evolution.
[NLP-21] Hidden in the Request: Explaining Unethical LLM Compliance through Token Relevance
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在同时追求“有益性”与“无害性”目标时可能出现的对齐失败问题，尤其是模型在面对伦理困境时表现出有害行为的根本机制。其核心问题是：尽管模型经过对齐训练以避免产生有害内容，但在特定情境下仍会违反伦理规范，而这种失败背后的认知机制尚不明确。解决方案的关键在于揭示并纠正模型在生成过程中对关键语义线索（cue-tokens，如“without getting caught”）的忽视现象。研究通过引入三种结构化探测范式（客观分类任务、主观第一人称陈述、直接求助请求），发现当任务以“请求帮助”的形式呈现时，模型性能显著下降，且容易产生有害响应。利用层间相关性传播（Layer-wise Relevance Propagation, LRP）分析表明，模型存在一种归因偏差——过度关注良性任务框架词（如“Can you help me…”），而忽略真正指示不当意图的线索词。为此，研究提出两种基于LRP引导的解码方法，主动增强对线索词的关注度，实证结果表明，此类干预有效提升了响应的安全性，验证了线索词归因不足是导致有害合规的重要原因。
链接: https://arxiv.org/abs/2608.23264
作者: Or Biton,Tomer Krichli,Itai Allouche,Joseph Keshet
机构: Technion (以色列理工学院)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Although Large Language Models (LLMs) are aligned to optimize for both helpfulness and harmlessness, these dual objectives may conflict, inevitably leading to alignment failures. This work systematically investigates instances where LLMs fail to exhibit ethical behavior. To understand the underlying mechanics of these vulnerabilities, we introduce a probing methodology that presents unethical scenarios to LLMs in three distinct structural modalities: objective classification tasks, subjective first-person statements, and direct requests for assistance. We find that model performance degrades in the request-for-assistance-based form. Using Layer-wise Relevance Propagation (LRP), we trace this discrepancy to an attribution bias: the model places greater emphasis on benign task-framing tokens (e.g., “Can you help me…”) than on tokens signaling the underlying unethical behavior (e.g., “without getting caught”), which we term cue-tokens. We hypothesize that this under-attribution contributes to harmful compliance. To test this, we introduce two LRP-guided decoding methods that steer generation toward trajectories more relevant to cue tokens. Empirical evaluations show that these interventions promote safer responses, supporting cue-token attribution’s role in compliance failures.
[NLP-22] Automated Construction of FAIR Digital Object Knowledge Graphs from Flat Cultural Heritage Records CIKM2026
【速读】： 该论文旨在解决欧洲大陆文化遗产数据在机器可操作性方面的关键瓶颈问题：现有欧洲人文学术数据模型（Europeana Data Model）以纯文本形式存储元数据，导致自动化代理无法跨记录或集合进行有效链接与导航。为实现完全可机读的语义网络，论文提出基于FAIR数字对象（FDO）框架的解决方案，其核心在于将传统扁平化元数据转化为符合FDO规范的知识图谱，采用CIDOC-CRM本体结构对文化遗产实体进行建模。该方案的关键技术挑战在于自动化区分必须转换为持久标识符（PID）引用的元数据值（即需可解析的实体）与可保留为字面量（如注释、测量值、日期等终端节点）的值。为此，研究引入大语言模型（LLM）对每个元数据项进行分类，将其路由至受控词汇库（Getty AAT、Wikidata、VIAF、PeriodO），并链接至共享的实体FDO。实验基于5个欧洲人文学科数据提供方的637条考古学记录验证，结果显示该流水线成功链接86%的元数据槽位，其中58.5%的值实现了原数据未包含的增强解析；同时，通过语义对齐合并了跨语言表面形式差异，17/33次合并经人工评估确认正确。最终形成的图谱不仅具备更高连通性，更因所有节点均具有类型与可解析性而区别于传统字符串匹配方法，真正实现了以可解析实体为核心的语义互联。
链接: https://arxiv.org/abs/2608.23263
作者: Zeyd Boukhers,Lingxiao Kong,Xenophon Zabulis,Georgios Toubekis
机构: Fraunhofer Institute for Applied Information Technology FIT(弗劳恩霍夫应用信息科技研究所); University Hospital of Cologne(科隆大学医院); University of Cologne(科隆大学); Foundation for Research and Technology(研究与技术基金会)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Digital Libraries (cs.DL)
备注: Accepted for publication as a short paper at CIKM 2026 (The 35th ACM International Conference on Information and Knowledge Management)
Abstract:The FAIR Digital Object (FDO) framework mandates that metadata attribute values be expressed as persistent identifiers (PIDs) wherever possible, to produce a fully machine-actionable graph in which every reference is resolvable. The Europeana Data Model was designed long before the FDO specification, and it stores most metadata values as plain text. This serves human browsing well enough, but gives an automated agent nothing to follow across records or collections. We present a pipeline that transforms flat Europeana records into an FDO-compliant knowledge graph structured with CIDOC-CRM. Following the FDO specification, we model every heritage entity as a discrete FDO with its own PID, type, profile, and metadata layer. The core technical challenge is automating the FDO-prescribed distinction between values that must become PID references (resolvable entities) and those that may remain literals (terminal leaves such as notes, measurements, and dates). We address this with a large language model that classifies each metadata value, routes it to a controlled vocabulary (Getty AAT, Wikidata, VIAF, PeriodO), and links it to a shared entity FDO. We evaluate using 637 archaeological records from five Europeana providers, processing each with the LLM. The pipeline links 86% of metadata slots, resolving 58.5% of values Europeana had not already enriched. It also merges cross-lingual surface forms that byte-identical matching keeps apart, where 17 of 33 such merges are correct on manual review. Graph connectivity does not separate this from string matching; what distinguishes the FDO graph is that every node is typed and resolvable.
[NLP-23] A Scalable Cross-Domain Event Extraction System via a Unified Generative Training Framework
【速读】： 该论文旨在解决事件抽取（Event Extraction）中普遍存在的任务分离问题，即传统方法通常将事件检测与论元抽取分步进行，或依赖特定数据集的设计，导致模型在跨领域场景下的可扩展性与泛化能力受限。其核心解决方案是提出一种统一的生成式序列到序列（Generative Sequence-to-Sequence）框架，能够联合执行事件抽取的各项子任务，并支持流水线与端到端两种配置模式。通过在多个跨域事件数据集上对预训练语言模型进行微调，该框架使单一模型既能保留领域特异性语义，又能在大规模且动态变化的标签空间中实现良好的泛化性能。研究进一步构建了一个面向研究人员与实践者的在线应用平台，支持文档上传、基于模式感知的事件抽取、触发词与论元的可视化展示，以及跨领域不同抽取配置的对比分析，验证了该方法在实际应用中的有效性与灵活性。
链接: https://arxiv.org/abs/2608.23261
作者: Siting Liang,Omar Adjali,Omair Shahzad Bhatti,Daniel Sonntag
机构: German Research Center for Artificial Intelligence(德国人工智能研究中心); Carl von Ossietzky University of Oldenburg(奥尔登堡卡尔·冯·奥西耶茨基大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Event extraction is fundamental to information extraction. Prior approaches often separate event detection and argument extraction or depend on dataset-specific designs, limiting scalability and cross-domain generalization. We propose a unified generative sequence-to-sequence framework that performs event extraction subtasks jointly and supports both pipeline and end-to-end configurations. We fine-tune pretrained language models on multiple event datasets across diverse domains, enabling a single model to retain domain-specific semantics while generalizing over large and evolving label spaces. We demonstrate these capabilities through a web-based application tailored for researchers and practitioners. The platform supports document upload, schema-aware event extraction, visualization of triggers and arguments, and comparison of different extraction configurations across domains.
[NLP-24] Future Querying: Can LLM s Serve as Implicit Medical World Models? MICCAI-2026 MICCAI
【速读】： 该论文旨在解决传统临床预测模型在处理非结构化临床文本时效率低下、可扩展性差以及依赖人工特征工程和特定任务训练流程的问题。其核心挑战在于如何有效利用海量未结构化的电子健康记录（EHR）数据，以实现对患者未来临床状态的通用化、动态化预测。解决方案的关键在于提出“未来查询”（future querying）这一新范式，通过在时间索引的临床轨迹上评估大型语言模型（LLM）回答未来临床问题的能力，使模型无需针对具体任务进行重新训练即可泛化于多种临床终点预测。该方法采用端点无关（endpoint-agnostic）的训练策略，在不依赖结构化特征或手动特征工程的前提下，直接基于原始临床文档进行学习，从而实现对患者病程中多维度临床动态的隐式建模。实验结果表明，经过本地微调的小型开源模型在性能上可媲美甚至接近大型专有系统，具备在保护隐私的前提下部署于本地环境的潜力，为生成式AI在临床决策支持中的实际应用提供了有力支撑。
链接: https://arxiv.org/abs/2608.23248
作者: Siri Willems,James Butterworth,Lore Goetschalckx,Peter Vrancx,Philippe Modard,Elke Giets,Ludovic Denoyer
机构: imec, AI-labs(imec人工智能实验室), Paris, France(巴黎, 法国)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: This paper is accepted at The 1st MICCAI Workshop on Medical World Models (MICCAI-2026)
Abstract:Traditional clinical prediction models rely on task-specific pipelines and curated, structured data, which scale poorly and underutilize unstructured text. To address this, we introduce future querying, a paradigm that probes whether large language models (LLMs) can function as implicit medical world models by evaluating their ability to answer time-indexed clinical queries about a patient’s future. Our framework operates on unstructured clinical documentation using endpoint-agnostic training, enabling a single model to answer diverse clinical queries over patient trajectories without manual feature engineering or task-specific retraining. We show that small, locally fine-tuned open-weight models can match or approach larger proprietary systems, making the framework suitable for privacy-preserving, on-premise deployment. Evaluated on a new synthetic medical reports dataset and real ICU notes from the MIMIC-IV dataset, our results provide encouraging evidence that LLMs can capture aspects of clinical dynamics.
[NLP-25] Credal Large Language Models for Semantic Commitment under Uncertainty
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在生成答案时存在流畅但错误、且伴随过度自信的问题。其核心挑战在于标准LLMs仅通过单一预测分布表示不确定性，将认知性无知（epistemic ignorance）与真实语义模糊性混淆，导致无法准确反映模型对自身判断的置信程度。为此，论文提出可信度大语言模型（Credal Large Language Models, CLLMs），通过引入一组低秩适配器（LoRA adapters）构建一个可信集（credal set），该集合以上下界概率形式揭示多个合理预测分布的范围，而非退化为单一softmax输出。关键创新在于从该可信集推导出两个互补的承诺度量：词元空间承诺度（Credal Token Commitment, CTC），结合下界支持度、可信宽度与交集熵，在无需额外生成的情况下计算；以及语义空间承诺一致性（Semantic Commitment Consistency, SCC），利用采样完成序列扩展至语义层面，并通过SCC-Gap量化词元级与语义级支持之间的不一致。实验在Gemma-2-9B、Llama-3.1-8B和Qwen2.5-7B上针对OpenBookQA、CoQA、TriviaQA和ARC-Challenge进行评估，结果表明CLLM在问答准确率上表现最佳，同时保持竞争力的期望校准误差（ECE）；CTC在多数设置下以不超过1.5个百分点的差距达到最优幻觉检测AUROC，且无需额外推理开销；在80%覆盖的自适应选择性预测任务中，结合SCC的CLLM在OpenBookQA上实现99.0%准确率，而在ARC-Challenge上使用语义置信度的CLLM实现跨三类骨干网络均低于0.6%的ECE，验证了该方法在提升模型可靠性与可解释性方面的有效性。
链接: https://arxiv.org/abs/2608.23244
作者: Shireen Kudukkil Manchingal,Sofiia Nikolenko,Fabio Cuzzolin
机构: Oxford Dynamics; Ludwig-Maximilians-Universität München (慕尼黑路德维希-马克西米利安大学); Institute for Artificial Intelligence, Data Analysis and Systems (AIDAS) (人工智能、数据分析与系统研究所); School of Engineering Computing Mathematics (工程计算与数学学院); Oxford Brookes University, Oxford, UK (牛津布鲁克斯大学，英国牛津)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Machine Learning (stat.ML)
备注: 31 pages, 5 figures, 13 tables
Abstract:Large language models (LLMs) often produce fluent but incorrect answers with unwarranted confidence. A central limitation is that standard LLMs represent uncertainty through a single predictive distribution, conflating epistemic ignorance with genuine ambiguity. We introduce Credal Large Language Models (CLLMs): an ensemble of LoRA adapters induces a credal set whose lower and upper probabilities expose the spread of plausible predictive distributions rather than collapsing to a single softmax output. From this representation we derive two complementary commitment scores. Credal Token Commitment (CTC) is a token-space score that combines lower-bound support, credal width, and intersection entropy, computed without additional generation. Semantic Commitment Consistency (SCC) extends commitment to semantic space using sampled completions, with SCC-Gap measuring the mismatch between token-level and semantic-level support. We evaluate hallucination detection, calibration, selective prediction, and reasoning on Gemma-2-9B, Llama-3.1-8B, and Qwen2.5-7B across OpenBookQA, CoQA, TriviaQA, and ARC-Challenge. CLLM is the best method on QA accuracy at competitive expected calibration error, and CTC tracks the best hallucination AUROC within 1.5 pp on most settings without additional generation. On selective prediction at 80% coverage, CLLM with SCC reaches 99.0% accuracy on OpenBookQA, and on ARC-Challenge CLLM with Csem confidence achieves = 0.6% ECE across the three backbones.
[NLP-26] A Multi-Domain and Multi-Task Generative Framework with Explicit Task and Domain Conditioning for Cross-Domain Event Extraction
【速读】： 该论文旨在解决事件抽取（Event Extraction）在跨领域场景下模型泛化能力不足的问题，尤其针对不同领域间上下文表达差异大、事件模式（event schema）不一致所导致的性能下降。现有统一多任务方法虽在单一领域表现良好，但在面对未见领域时灵活性有限；而基于大语言模型的方法虽能在推理时提供完整的事件本体，却仍难以超越小型专用微调模型的性能。其解决方案的关键在于提出一种统一的多领域、多任务训练框架，通过引入领域条件信号（domain conditioning signals）与任务特定提示（task-specific prompts）相结合的方式，使模型能够在不依赖完整事件标签集的情况下，动态适应不同数据集的事件模式。该框架支持流水线与端到端两种抽取设置，实现了高效的任务级与领域级迁移，实验表明该方法在多个事件抽取基准上均取得了具有竞争力的性能，具备出色的跨领域泛化能力与实际可扩展性，同时保持了各领域的高精度。
链接: https://arxiv.org/abs/2608.23235
作者: Siting Liang,Omar Adjali,Daniel Sonntag
机构: German Research Center for Artificial Intelligence(德国人工智能研究中心); Carl von Ossietzky University of Oldenburg(奥尔登堡卡尔·冯·奥西茨基大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Event extraction aims to identify event triggers, classify event types, and extract arguments to construct structured event representations. Despite strong in-domain performance, developing models that generalize robustly across domains remains challenging due to variations in contextual expressions and event schemas. Prior unified and multi-task approaches improve in-domain accuracy but exhibit limited flexibility when applied to unseen domains. Even large language model-based methods that provide full event ontologies at inference time often underperform compared to smaller, task-specific fine-tuned models. We propose a unified multi-domain and multi-task training framework that models heterogeneous event schemas within a single model. Our approach introduces domain conditioning signals, jointly with task-specific prompts, enabling dynamic adaptation to dataset-specific schemas without requiring complete event label sets at inference time. The framework supports both pipeline and end-to-end extraction settings, facilitating efficient task- and domain-level transfer. Experiments on diverse event extraction benchmarks demonstrate that our method achieves competitive performance, strong cross-domain generalization, and practical scalability, while preserving domain-specific precision.
[NLP-27] Cognitive Profiling of LRMs Reasoning Traces Using Blooms Taxonomy
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在推理过程中深层次思维模式缺乏系统性分析的问题，尤其关注推理步骤中所体现的认知类型，以揭示模型内部的推理规律并提升其推理质量。当前虽有大量公开的推理轨迹（reasoning traces），但对其中蕴含的思维层次（如记忆、应用、评估等）的细粒度解析仍处于空白状态。为此，论文提出一种基于布卢姆分类学（Bloom’s Taxonomy）的自动化推理步骤标注框架，将思维过程划分为六类认知层级（如记忆、应用、评价等），实现了对推理轨迹的精细化标注与分析。该方案的关键在于利用布卢姆分类学作为理论基础，构建可扩展的自动标注机制，从而在大规模模型与数据集上揭示不同模型和任务间的思维模式差异，并验证了思维类型与推理正确性之间存在显著相关性。这一方法不仅为理解生成式AI（Generative AI）的内在推理逻辑提供了新视角，也为优化模型推理能力提供了可操作的指导路径。
链接: https://arxiv.org/abs/2608.23205
作者: Maria-Eleni Zoumpoulidi,Georgios Paraskevopoulos,Alexandros Potamianos
机构: Institute for Language and Speech Processing, Athena Research Center, Greece; Speech and Language Processing Group, National Technical University of Athens, Greece
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Large Reasoning Models (LRMs) have revolutionized reasoning in LLMs, and the increasing public availability of reasoning traces creates valuable opportunities to study model behavior not only at the surface level but also at the granularity of individual reasoning steps. However, understanding the types of thinking employed during reasoning - which offers critical insights into models’ reasoning patterns and enables actionable applications - remains underexplored. To address this gap, we introduce a framework for automatic annotation of reasoning steps through the lens of Bloom’s Taxonomy, which classifies thinking into six cognitive levels, such as Remembering, Applying and Evaluating. Using this framework, we perform a large-scale analysis across models and datasets, revealing both similarities and differences in thinking patterns across models and tasks. Moreover, we demonstrate that thinking-type information derived from reasoning traces correlates with correctness, paving the way for improved reasoning. Our findings establish a fine-grained framework for analyzing thinking patterns in LRMs and provide actionable insights for enhancing reasoning quality.
[NLP-28] LongWoF-Bench: Evaluating EvoMap Genes for Verifiable Long-Workflow Tasks
【速读】： 该论文旨在解决大语言模型在执行复杂工作流时，因缺乏对成功执行经验的持久化存储与复用机制，导致每次任务都需要从头探索策略与失败模式，造成资源浪费的问题。其核心挑战在于如何有效保留并共享经过验证的、可信赖的执行轨迹，以提升模型在长流程任务中的表现。解决方案的关键在于提出EvoMap框架，将经验证器确认的执行轨迹（verifier-confirmed execution trajectories）结构化地整合为一种名为Gene的紧凑知识表示形式。通过构建长期工作流基准测试（Long-Workflow Benchmark, LongWoF-Bench），研究证明：基于真实验证经验生成的EvoMap Gene能够显著提升七种不同模型在252个任务上的表现，平均性能提升8.7–15.5个百分点，且优势延伸至多个模型家族的消费级模型；相比之下，仅通过参考蒸馏生成的Gene未能展现同等优势，表明基因的实用性不仅依赖于紧凑性，更关键的是其源自经过验证的真实执行经历的可追溯性。此外，对于Claude Opus而言，使用Gene可额外完成39项任务，并降低9.9%的求解阶段词元消耗。结果表明，经验证的执行经验可通过EvoMap以可重用的外部资源形式被有效保留与传递，使模型无需重复承担“经验发现”的成本，从而实现长工作流任务成功率与效率的双重提升。
链接: https://arxiv.org/abs/2608.23200
作者: Xiao Zhang,Qumeng Sun,Jihao Li,Yiming Ren,Xiang Liu,Haoyang Zhang,Junjie Wang
机构: EvoMap; Tsinghua University
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language models are increasingly expected to execute complex workflows whose success depends on maintaining interdependent constraints and producing artifacts that satisfy strict end-to-end verification. Yet successful execution experience is typically lost after a single run, forcing subsequent models to rediscover strategies and failure modes from scratch. We study whether such experience can instead be externalized and reused through EvoMap, where verifier-confirmed execution trajectories are consolidated into structured Gene. To evaluate this setting, we introduce the Long-Workflow Benchmark (LongWoF-Bench), comprising 778 machine-verifiable tasks across code generation, agent-environment synthesis, mathematical reasoning, and rule following. On the 252 tasks with verifier-confirmed Opus trajectories, evolved EvoMap Gene outperform Skill across all seven evaluated models by 8.7-15.5 percentage points, with the gains extending to consumer models from different model families. In contrast, reference-distilled Gene do not exhibit the same advantage, indicating that compact representation alone is insufficient and that Gene utility is closely associated with verified experience provenance. For Claude Opus, Gene reuse also completes 39 more tasks than Skill while reducing solve-time token consumption by 9.9%. Together, these results show that verified execution experience can be retained and shared as a reusable external resource, enabling models to improve long-workflow completion without repeatedly paying the full cost of experience discovery.
[NLP-29] CyberFactory: Scaling Cyber Security Capabilities with Instances from the Wild
【速读】： 该论文旨在解决当前开源生成式 AI 在网络安全领域中面临的核心挑战：缺乏可复现的、端到端的网络安全训练框架，现有开源方案要么局限于单一任务，要么缺少可扩展的智能体（agentic）数据支持，且模型训练依赖强领域先验知识。其解决方案的关键在于提出一个统一的开源框架 CyberFactory，该框架实现了从漏洞数据构建、轨迹合成到模型训练的全流程闭环。CyberFactory 将真实世界中的公开漏洞（如 CVE）转化为可执行、可验证的任务实例，并通过可复用的“漏洞分析技能”引导教师模型进行源码审查、基于领域先验的问题求解以及基于证据的验证，从而生成具有代理行为特征的监督轨迹——即模型在与工具和目标环境交互过程中根据执行反馈不断修正自身策略的动态决策过程。在此基础上，研究训练并发布了 Aegis 模型，该模型内化了上述技能引导的推理流程，无需在推理阶段显式调用技能即可实现高效安全任务处理。在 CyberGym 基准测试中，Aegis 在一小时预算下达到 52.4% 的 Pass@1 准确率，相较基线 Qwen 3.5 提升 22.8 个百分点，并优于同类通用模型架构。
链接: https://arxiv.org/abs/2608.23181
作者: Jian Yang,Haau-Sing Li,Shawn Guo,Zixi Zhao,Yibo Tan,Jiajun Wu,Aishan Liu,Xianglong Liu,Tianyu Zheng,Bryan Dai,Chengran Yang
机构: Beihang University(北京航空航天大学); IQuest Research; ELLIS; Singapore Management University(新加坡管理大学)
类目: Cryptography and Security (cs.CR); Computation and Language (cs.CL)
备注:
Abstract:As large language models (LLMs) continue to advance in coding capabilities, their potential in cybersecurity has drawn increasing research attention, with closed-source LLMs (e.g., Mythos) delivering advanced cybersecurity capabilities. However, existing open-source efforts remain limited: frontier open-weight models do not provide reproducible cybersecurity training solutions, open-source training solutions focus on isolated tasks and lack scalable agentic data, and scaling agentic rollouts requires strong domain priors. In this work, we introduce \textbfCyberFactory, a unified open-source framework that connects data construction, trajectory synthesis, and model training across proof-of-concept (PoC) generation, vulnerability patching, and cybersecurity question answering (CyberQA). CyberFactory transforms public vulnerability artifacts, including CVEs from the wild, into executable and verifiable task instances. It further uses a reusable vulnerability-analysis skill to guide the teacher through source inspection, problem solving with domain prior, and evidence-based validation. The resulting supervision is agentic: the model interacts with tools and target environments and revises its solutions according to execution feedback. Using these trajectories, we train and release \modelname\footnote\emphAegis is, in Greek mythology, the protective shield of Zeus and Athena; the name reflects the model’s defensive, security-oriented purpose., which internalizes the skill-guided procedure without requiring the skill at inference time. On CyberGym, \modelname reaches 52.4% Pass@1 under a one-hour budget, improving over its Qwen~3.5 base model by +22.8 points and outperforming the evaluated general-purpose backbones under the same scaffold.
[NLP-30] CaRGo-T: Causal Reasoning Graph-of-Thought improves Multimodal Humor Comprehension
【速读】： 该论文旨在解决大规模多模态模型在理解幽默（humor）任务中面临的挑战，特别是由于幽默内容依赖于图像与文本模态之间实体、事件、上下文及隐含关系的细微交互，且常需复杂的因果推理链来解析。传统提示方法或线性思维链（chain-of-thought）难以有效捕捉此类跨模态的深层语义关联。为此，论文提出一种名为CaRGo-T（Causal Reasoning Graph-of-Thought）的推理框架，其核心创新在于将多模态幽默背后的因果关系与上下文依赖关系建模为轻量级图结构，并通过视觉语言模型（VLM）生成可序列化的代码化表示，实现零样本或上下文学习下的可解释推理。该框架的关键在于利用图结构显式建模复杂因果与语境依赖，从而增强模型对幽默本质的理解能力，实验表明其在四个涵盖讽刺、反讽和梗图等多样喜剧形式的数据集上显著优于现有基线方法，性能提升达1-20%（幽默理解）和1-3%（幽默检测），且互信息分析证实其推理表示包含更丰富的目标输出相关特征。
链接: https://arxiv.org/abs/2608.23172
作者: Abhilash Nandy,Rahul Seetharaman,Aman Bansal,Rounak Saha,Manav Nitin Kapadnis,Millon Madhur Das,Pawan Goyal,Niloy Ganguly
机构: Microsoft Research India; LinkedIn(领英), USA; Nutanix, USA; Indian Institute of Science(印度科学研究所), India; Apple(苹果), USA; Fujitsu Research India; Indian Institute of Technology Kharagpur(印度理工学院克哈拉格普尔分校), India
类目: Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
备注: 18 pages, 5 figures
Abstract:Large-scale vision-language models (VLMs) have demonstrated remarkable versatility across a wide range of multimodal tasks. However, understanding humor remains challenging because humorous content often depends on subtle interactions among entities, events, context, and implicit relationships across image and text modalities. These interactions can involve complex chains of reasoning that are difficult to capture through conventional prompting or linear chain-of-thought reasoning. In this work, we propose CaRGo-T (Causal Reasoning Graph-of-Thought), a reasoning framework that represents the causal and contextual relationships underlying multimodal humor as a lightweight graph-based reasoning structure. The graph is serialized into a code-based representation generated by a VLM, which can subsequently be interpreted by the same or a different VLM to produce the final prediction in zero-shot or in-context learning settings. We evaluate CaRGo-T on humor understanding and humor detection across four datasets spanning diverse forms of comedic content, including satire, sarcasm, and memes. Experiments with state-of-the-art commercial and open-source VLMs show that CaRGo-T consistently improves performance over existing reasoning-based baselines, achieving gains of approximately 1-20% on humor understanding and 1-3% on humor detection. Further analysis using mutual information indicates that the reasoning representations produced by CaRGo-T contain more information relevant to the target output than those generated by baseline reasoning approaches. Code is available at this https URL.
[NLP-31] Accelerating Diffusion Language Models via Structured Suffix Modeling
【速读】： 该论文旨在解决扩散语言模型（Diffusion Language Models, DLMs）在并行解码过程中因全后缀交互带来的巨大计算开销问题。现有方法通过仅保留局部后缀窗口来降低计算成本，但忽略了后缀区域间的结构异质性，并在每一步中对后缀标记进行相同初始化，导致信息表达能力受限。本文提出一种结构化后缀建模方法，将后缀划分为局部、中间和尾部三个区域，依据其不同结构作用保留不同数量的后缀标记；同时，在当前步骤中引入前一时刻的解码结果以更新后缀标记表示，使其能够携带随生成过程演化的去噪信息。该方法无需训练且与多种现有加速技术（如并行解码策略和键值缓存）正交。实验结果表明，该方法在多个基准测试中显著提升了DLM推理效率，尤其在长序列生成场景下，结合其他加速技术可实现最高达72.81倍的加速比。
链接: https://arxiv.org/abs/2608.23167
作者: Zifeng Cheng,Keda Li,Zhiwei Jiang,Cong Wang,Fei Shen,Qing Gu
机构: Nanjing University (南京大学); National University of Singapore (新加坡国立大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Diffusion Language Models (DLMs) exhibit strong parallel decoding capabilities by denoising multiple tokens in a single generation step. However, this parallelism comes with substantial computational overhead, as each step requires interactions with all suffix tokens. Existing methods typically reduce this cost by retaining only a local suffix window as a substitute for the full suffix. Despite their effectiveness, these methods overlook the structural heterogeneity across suffix regions and re-initialize suffix tokens with identical representations at each timestep. To this end, we propose a structured suffix modeling method for efficient DLM inference. Specifically, we divide the suffix into three regions, i.e., the local, middle, and tail regions, and retain different numbers of suffix tokens in each region according to their structural roles. Moreover, we incorporate the decoding results from the previous step into the suffix token representations at the current step, allowing them to carry evolving denoising information across generation steps. Notably, our method is training-free and orthogonal to several existing acceleration techniques, such as parallel decoding strategies and KV cache. Empirical results across multiple benchmarks on three DLMs demonstrate that our method can further accelerate DLM inference and improve performance in most cases. In particular, in long-sequence inference, our method achieves up to a (72.81\times) speedup when combined with other acceleration techniques. Our code is available at this https URL.
[NLP-32] Counter with Evidence! A Multi-Agent Memory Efficient Reasoning Framework for Hate Category Informed Counterspeech Generation
【速读】： 该论文旨在解决现有自动化反仇恨言论生成方法中忽视仇恨言论异质性的问题，即当前模型普遍将仇恨言论视为同质化整体，仅关注语风控制而忽略不同形式的仇恨言论（如虚假信息、刻板印象、阴谋论、非人化、非事实陈述）需采用根本不同的应对策略。其解决方案的关键在于提出FIRE（Factuality Informed Multi-Agent Reasoning Framework），通过首先将仇恨言论分解为五类具体类别，并据此映射至相应的针对性反仇恨言论风格，实现基于事实依据的精准回应。为支持该框架，研究者构建了FactualCS数据集，包含4,784条标注实例，涵盖仇恨类型、推理过程及证据关联等关键要素，弥补了以往研究在生成过程中缺乏事实支撑的缺陷。实验表明，尽管采用参数量仅为20亿的轻量化代理，FIRE在事实准确性和类别特定准确性上分别提升约12%和11%，同时毒性降低约11%，且经人工评估显示其生成内容显著优于最强基线，验证了识别仇恨言论深层意图对生成安全、有效、情境精确的反仇恨言论至关重要。
链接: https://arxiv.org/abs/2608.23152
作者: Sujoy Nath,Aswini Kumar,Tanmoy Chakraborty
机构: Indian Institute of Technology Delhi (印度理工学院德里分校)
类目: Computation and Language (cs.CL)
备注:
Abstract:Counterspeech effectively neutralizes the impact of online hate. Although prior work explores automated counterspeech generation, it largely emphasizes stylistic control while treating hate speech as homogeneous, overlooking that distinct forms of abuse require fundamentally different counterspeech strategies. To address this gap, we introduce FIRE (Factuality Informed Multi-Agent Reasoning Framework) that first decomposes hate speech into one of the five distinct categories (misinformation, stereotype, conspiracy, dehumanizing, non-factual), and then maps it to a targeted counterspeech style. To facilitate FIRE, we curate FactualCS, a novel dataset of 4,784 instances that provides the annotations regarding hate categories, reasoning traces, and evidence mappings, which are critical elements for grounded generation that are missing in prior work. A comprehensive evaluation across 28 baseline configurations demonstrates that FIRE significantly surpasses existing methods, despite using compact agents ( 2B). FIRE achieves a \sim 12 % and \sim 11 % improvements in factual and category-specific accuracy respectively, while simultaneously reducing toxicity by \sim 11 % relative to the strongest baselines. Further human evaluation confirms that responses generated by FIRE are significantly preferred over the strongest baselines, underscoring its effectiveness for real-world deployment. These findings show that decomposing the underlying intent of hate speech is essential for generating safe, effective, and contextually precise counterspeech.
[NLP-33] Language Chain in Alignment: Cross-Lingual Ranking Preference Optimization EMNLP2026
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在多语言场景下因依赖以英语为中心的高质量偏好数据而导致非英语语言性能下降的问题。其核心挑战在于如何有效迁移英语语境下的偏好知识，以实现对目标语言的高质量偏好对齐。为此，论文提出跨语言排名偏好优化（Cross-Lingual Ranking Preference Optimization, CRPO）框架，其关键创新在于构建了目标语言与英语之间并行偏好样本的分层结构，通过联合优化句内（intra-lingual）与跨语言（inter-lingual）偏好信号，增强模型的语言适应能力与输出质量。相较于传统的二元比较优化方法，CRPO基于LambdaLoss框架引入多候选响应间的相对排序信号，实现了更精细的偏好学习。实验结果表明，该方法在五种不同资源规模的语言上均显著优于标准方法，在指令遵循与知识利用能力方面表现更优；同时，其在多种权重设置下的稳定增益进一步验证了分层设计在多语言环境中的有效性。此外，研究发现CRPO能有效提升奖励差距（reward margin）和理想响应的对数概率，从而构建更稳定的跨语言偏好流形，为多语言大模型对齐提供了可扩展且高效的解决方案。
链接: https://arxiv.org/abs/2608.23149
作者: Seungyoon Lee,Minhyuk Kim,Jungseob Lee,Heuiseok Lim
机构: Korea University(韩国大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: EMNLP 2026 Main
Abstract:The alignment of Large Language Models heavily relies on English-centric high-quality preference data, which often leads to suboptimal performance in other languages. In this paper, we propose Cross-Lingual Ranking Preference Optimization (CRPO), a novel framework that leverages robust preference knowledge from English to facilitate preference alignment in the target language. We design a hierarchical structure within parallel preference pairs across the target language and English to jointly optimize intra- and inter-lingual preferences, thereby enhancing language adaptation and output quality. Building on the LambdaLoss framework, CRPO goes beyond the binary comparison based optimization by providing a relative ranking signal across multiple candidate responses. Our experiments across five languages with varying resource scales demonstrate that CRPO consistently outperforms standard approaches in both instruction-following and knowledge utilization capability. Notably, the robust performance gains observed across various weighting schemes further validate the empirical effectiveness of our hierarchical design in a multilingual setup. Furthermore, our findings highlight that CRPO significantly improves both reward margins and the log-probability of desirable responses, contributing to a more stable preference manifold for cross-lingual alignment.
[NLP-34] Activation-Weighted Seeded Residual Coding for Low-Bit LLM Weight Repair
【速读】： 该论文旨在解决低比特权重量化（Low-bit weight quantization）带来的量化误差对语言模型质量的负面影响。现有方法在压缩模型权重时虽节省了存储空间，但引入的误差会显著降低模型性能。为此，作者提出一种轻量级修复编码方案——激活加权种子残差编码（Activation-Weighted Seeded Residual Coding, AWSRC），其核心在于：在已有的量化骨干网络（如INT4 RTN）基础上，通过确定性种子生成的基向量对重构权重与原始权重之间的残差 W−W0 进行编码。该方案不存储显式的码本，而是仅保存种子选择器、低比特系数和缩放因子等紧凑的侧车（sidecar）信息，从而实现高效压缩。同时，利用激活统计信息对残差进行加权，优先修复对层输出影响较大的误差项。实验表明，在Qwen2.5-3B-Instruct模型上，仅增加0.162位/权重的额外开销，即可修复88.2%、78.9%和71.3%的匹配困惑度（PPL）、KL散度及准确率差距至全精度（BF16）水平；且在约49.25 MB的侧车数据（占BF16模型权重的0.8%）下，AWSRC在稀疏、低秩及向量量化类编码方案中取得了最优的困惑度与平均任务准确率表现。其关键创新在于结合激活感知的残差建模与种子驱动的确定性基向量表示，实现了高效率与高性能的平衡。
链接: https://arxiv.org/abs/2608.23144
作者: Zehao Liu,Chuangchuang Fang,Yang Ren
机构: 未知
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 5 pages, 2 figures
Abstract:Low-bit weight quantization saves storage but leaves errors that degrade language-model quality. We introduce Activation-Weighted Seeded Residual Coding (AWSRC), a compact repair codec for an existing quantization backbone. Given a reconstructed weight W_0 , AWSRC encodes the residual W-W_0 using deterministic seed-generated bases. The sidecar stores seed selectors, low-bit coefficients, and scales rather than an explicit codebook. Activation statistics prioritize errors that affect layer outputs. On Qwen2.5-3B-Instruct, adding 0.162 scope-bits/weight to an INT4 RTN backbone closes 88.2%, 78.9%, and 71.3% of the matched PPL, KL, and accuracy gaps to BF16. Repairing a matched strong low-bit backbone also improves all measured quality metrics. With a matched 49.25 MB sidecar, about 0.8% of the BF16 model-weight payload, AWSRC gives the best perplexity and mean task accuracy among sparse, low-rank, and vector-quantized codecs.
[NLP-35] LITERARYBIGFIVE: Author-Personalized Text Generation in a Unified Interpretable Space EMNLP2026
【速读】： 该论文旨在解决个性化文本生成中作者建模与风格个性化所面临的高成本、可解释性差及跨作者泛化能力弱的问题。现有方法通常将写作风格视为独立标签，需为每位作者或风格类别收集大规模语料库或进行微调，不仅资源消耗大，且难以实现跨作者的通用性。为此，论文提出LiteraryBigFive框架，受人格五因素模型（Big Five model）的维度化视角启发，将作者写作特征重新建模为统一且可解释的五维空间中的坐标。该空间通过对比作者作品与中性文本在激活空间中的差异，自动提取出如“古典性”（Classicism）、“情感性”（Emotionality）等可解释的风格维度，使文本或作者可在五维风格空间中精确定位。在此基础上，引入可解释的调控机制，能够自适应地引导生成文本向目标坐标逼近，实现作者个性化的文本生成。实验表明，LiteraryBigFive在保持语义一致性的同时显著提升了作者表达的丰富性，且各维度得分与真实文学共识高度相关，提供了透明、可解释的作者风格生成行为分析。
链接: https://arxiv.org/abs/2608.23124
作者: Jinghui Zhang,Lang Gao,Ao Li,Mingzhe Li,Ruihong Zeng,Zirui Song,Kentaro Inui,Xiuying Chen
机构: MBZUAI(中东人工智能大学); Shandong University (山东大学); ByteDance(字节跳动); Tohoku University (东北大学); RIKEN(理化学研究所)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: EMNLP 2026 Findings
Abstract:Personalized text generation for authors and literary writing is essential for applications such as adaptive writing assistants, creative support tools, and computational literary analysis. However, existing approaches to author modeling and personalization often represent writing behavior as independent labels, requiring large-scale corpus collection or fine-tuning for each author or stylistic category. Such formulations are costly, difficult to interpret, and poorly suited for generalizing across authors. Inspired by the Big Five model’s dimensional view of personality, we propose LiteraryBigFive, a framework that reframes authorial writing characteristics as coordinates within a unified and interpretable space. In this space, we derive each interpretable axis (e.g., Classicism, Emotionality) from activation-space contrasts between author-written and neutral passages, yielding distinct stylistic dimensions that allow texts or authors to be positioned within a five-dimensional system. Beyond localizing different authors, we further introduce an interpretable steering mechanism, which adaptively guides text generation toward target coordinates to perform author-personalized writing. Experimental results show that LiteraryBigFive improves authorial expressiveness while preserving semantic fidelity. The derived author per-axis scores strongly correlate with real-world literary consensus, offering transparent and interpretable explanations of author-specific generation behavior: this https URL.
[NLP-36] Statistical Machine Translation Systems of English-Pnar Language Pair : Some Insights of the Emperical Study
【速读】： 该论文旨在解决帕纳尔语（Pnar）这一人口约40万的南亚语系语言缺乏数字语料库和自然语言处理（NLP）资源的问题，首次针对英语与帕纳尔语这对语言对开展机器翻译研究。其关键解决方案是基于从《Wyrta》报纸收集的新闻文章构建了一个包含10,234句对的平行语料库，并采用短语基于统计机器翻译（SMT）框架，在Moses、GIZA++、KenLM等工具支持下，通过三种不同配置训练双向翻译模型，同时引入词汇化重排（lexicalized reordering）和最小错误率训练（MERT）进行优化。实验结果表明，最佳模型在帕纳尔语→英语方向达到14.97的BLEU分数（chrF2: 33.42, TER: 77.60），英语→帕纳尔语方向为11.16（chrF2: 31.38, TER: 93.51），首次建立了该语言对的量化基准。研究发现，词汇化重排显著提升翻译质量（提升3.73 BLEU点），反映出帕纳尔语（主-宾-谓，SOV）向英语（主-谓-宾，SVO）结构转换中的句法差异；而低资源条件下MERT调优反而导致性能下降。最后，论文分析了残余错误类型，包括形态学未登录词（OOV）、长距离重排及卡西语（Khasi）混用现象，并指出未来应探索神经网络与多语言机器翻译技术以进一步提升帕纳尔语的翻译能力。
链接: https://arxiv.org/abs/2608.23120
作者: Edawanbiang Dhar Surmila Thokchom,Thoudam Doren Singh
机构: NITM (National Institute of Technology Manipur)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Pnar, an Austroasiatic language spoken by approximately 0.4 million people in the Jaintia Hills of Meghalaya, lacks the digital corpora and natural language processing (NLP) resources. This paper presents the first machine translation study for the English and Pnar language pair. Using articles collected from the Wyrta newspaper, we built a parallel corpus comprising of 10,234 sentences and trained phrase-based statistical machine translation (SMT) systems the models using 9,563 parallel corpora under three configurations for each direction using Moses, GIZA++ , KenLM, varying lexicalized reordering and minimum error rate training (MERT) tuning. The models are evaluated on a held out test set of 371 sentences, the best performing system achieves a BLEU score of 14.97 (chrF2: 33.42, TER: 77.60) for Pnar to English and 11.16 (chrF2: 31.38, TER: 93.51) for English to Pnar, establishing the first quantitative benchmark for this language pair. Lexicalized reordering improves translation quality by 3.73 BLEU points for Pnar to English, reflecting the structural shift from the source language’s SOV word order to the target language’s SVO order, whereas MERT tuning degrades BLEU performance under low resource conditions. Finally, we analyze the remaining translation errors, including morphological out of vocabulary (OOV) words, long-distance reordering and Khasi code mixing and discuss future directions toward neural and multilingual machine translation for Pnar.
[NLP-37] Molecular LLM Agents : From Architectural Design to Scientific Autonomy
【速读】： 该论文旨在解决生成式人工智能在分子科学领域应用中的核心挑战，即如何构建具备化学感知与推理能力的分子大模型智能体（molecular LLM agents），使其能够跨符号字符串、分子图、三维构象、光谱数据、模拟结果及实验测量等多模态化学表征进行感知、推理与操作。现有通用智能体主要面向自然语言、代码或网络环境，难以满足分子科学对化学保真度、领域工具嵌入与实证反馈闭环的严苛要求。其解决方案的关键在于提出一个双重视角的综合性框架：一是从架构层面系统化设计分子智能体的组成部分，包括分子表示与感知、以大模型为核心的智能体架构、领域专用工具箱以及学习与优化机制；二是引入受工程系统阶段性自主性启发的“科学自主性阶梯”（scientific autonomy ladder），将分子智能体划分为四个层级——L1辅助型或固定工作流、L2自适应计算代理、L3具备反馈意识的物理实验代理、L4具备科学议程驱动能力的自主科研代理。该框架不仅为现有分子智能体提供评估基准，还揭示了当前技术在能力缺失与部署风险方面的关键缺口，从而指导未来分子发现流程中智能体的设计、评估与实际部署。
链接: https://arxiv.org/abs/2608.23104
作者: Jiatong Li,Wengyu Zhang,Weida Wang,Yuxuan Ren,Wei Liu,Chenyang Mao,Yuqiang Li,Yatao Bian,Changmeng Zheng,Xiaoyong Wei,Qing Li
机构: The Hong Kong Polytechnic University(香港理工大学); Shanghai AI Lab; National University of Singapore(新加坡国立大学); Shanghai Jiao Tong University(上海交通大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 25pages
Abstract:Molecular science represents an important frontier for LLM-based agents. Unlike general agents that mainly operate over natural language, code, or web environments, molecular LLM agents must perceive, reason about, and act upon chemical objects across symbolic strings, molecular graphs, 3D conformations, spectra, simulations, and wet-lab measurements. Their capabilities depend on chemically faithful molecular perception, an LLM-centered agent framework, domain-specific tool grounding, and computational or experimental feedback, in addition to planning and tool use. This work develops a conceptual framework for molecular LLM agents from two complementary perspectives. First, we introduce an architectural view of molecular-agent design, covering molecular representation and perception, the agent framework, domain-specific toolboxes, and learning and optimization. Second, we propose a scientific autonomy ladder inspired by staged autonomy in engineering systems, categorizing agents into four levels: L1 assistive or fixed workflows, L2 adaptive computational agents, L3 feedback-aware physical experiment agents, and L4 scientific-agenda agents. Together, these two perspectives establish a comprehensive framework for comparing existing molecular LLM agents, identifying missing capabilities and deployment risks, and guiding the design, evaluation, and deployment of future agents in molecular discovery workflows.
[NLP-38] Definitional Sensitivity in Media Bias Detection: A Multi-Definition Dataset and Benchmark EMNLP2026
【速读】： 该论文旨在解决媒体偏见检测中因定义不一致或隐含导致的标注可复现性问题，即相同名称的偏见类别在不同数据集或语境下可能对应不同的概念内涵，从而影响模型学习到的其实是不同现象而非同一构念。其解决方案的关键在于系统检验定义的**概念框架（conceptual framing）**对偏见标注结果的影响：通过一项包含354名参与者的人类实验及四类大语言模型（LLM）的平行评估，研究发现，定义的概念目标显著驱动了人工与模型标注的差异，而仅在保持构念一致性前提下的细节扩展（construct-preserving elaboration）则不会引发显著偏差。这一发现揭示了在标注协议与基于提示（prompt-based）测量中明确定义的重要性，并指出定义敏感性可能向下游分类任务扩散。为此，作者发布了多定义偏见检测数据集（MUDD），以促进更透明、可比较的偏见研究。
链接: https://arxiv.org/abs/2608.23095
作者: Martin Wessel,Timo Spinde,Jürgen Pfeffer,Gianluca Demartini
机构: Technical University of Munich (慕尼黑工业大学); Center for Digital Technology and Management (数字技术与管理中心); National Institute of Informatics (日本信息研究所); University of Queensland (昆士兰大学)
类目: Computation and Language (cs.CL)
备注: To appear in Findings of the Association for Computational Linguistics: EMNLP 2026
Abstract:Media bias detection relies on definitions and examples that specify what counts as bias, yet these specifications often vary across datasets or remain implicit, even when given the same name. Such variation makes it unclear whether models trained for the same bias category learn the same construct or different phenomena, a problem largely overlooked in prior work. We examine how definition choice affects bias annotation in a between-subjects experiment with 354 participants and a parallel evaluation with four LLMs. Participants and models rate six news articles across four bias categories using definitions that vary in conceptual framing and elaboration. Across 8,496 human and 28,800 LLM ratings, we find that the conceptual target of a definition drives annotation divergence, while construct-preserving elaboration does not: conceptual framing significantly shifts annotations for humans and does so even more strongly for LLMs. We discuss implications for construct specification in annotation protocols and prompt-based measurement, and consider how definitional sensitivity may propagate to downstream classification beyond media bias. We also release MUDD, the Multi-Definition Bias Detection Dataset.
[NLP-39] Agent Weave: Routing Before Reasoning for Efficient Function Calling in Tool-Rich Language Models
【速读】： 该论文旨在解决大语言模型在面对大规模工具、函数、API及专用代理集合时，因候选动作空间膨胀导致的推理效率下降与决策质量恶化问题。随着可调用接口数量增加，模型需处理更多模式（schema）、消耗更多提示词（prompt token）并从日益相似或无关的选项中进行区分，显著增加了计算负担与误判风险。其解决方案的关键在于引入一种确定性的预推理路由层——AgentWeave，该层在不改变下游语言模型的前提下，通过整合可用性（eligibility）、需求匹配（requirement）、能力评估（capability）和路由信号（routing signal）等多维度信息，构建一个受限且高相关性的模型可见动作空间。实验表明，在48个全新的BFCL V4多函数任务上，相较于所有工具暴露、确定性随机前8名及语义前8名基线均未取得成功（0/48），AgentWeave实现了6次成功（12.5%），且统计显著性达到p=0.03125，置信区间为+4.17至+22.92个百分点。此外，该方法使工具数量减少70.18%，输入令牌减少61.70%，平均本地模型延迟降低50.95%。研究结果表明，候选空间的预先构造能够显著影响固定模型的功能调用行为，从而支持将路由作为独立于模型推理的前置阶段进行系统化设计与优化。
链接: https://arxiv.org/abs/2608.23078
作者: Saurav Singla,Aarav Singla,Advik Gupta,Parnika Gupta
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 12 pages, 2 figures, 6 tables. Open-source implementation and reproducibility artifacts available in the AgentWeave repository
Abstract:Large language models increasingly operate over large collections of tools, functions, APIs, and specialized agents. As the candidate action space grows, a function-calling model must process more schemas, consume more prompt tokens, and distinguish among increasingly similar or irrelevant alternatives. We study a complementary systems strategy: reduce the candidate set before language-model inference while leaving the downstream model unchanged. We introduce AgentWeave, a deterministic pre-inference routing layer that constructs a bounded model-visible action space using eligibility, requirement, capability, and routing signals. We evaluate AgentWeave with a frozen BFCL-derived routing-pressure protocol using the public MadeAgents/Hammer2.1-1.5b model. On 48 fresh BFCL V4 multiple-function tasks, AgentWeave achieves 6/48 (12.5%) native BFCL successes, whereas all-tools, deterministic random top-8, and semantic top-8 baselines each achieve 0/48. The paired success difference is +12.5 percentage points with a 10,000-resample paired bootstrap 95% confidence interval of +4.17 to +22.92 points and exact McNemar p=0.03125. Relative to all-tools exposure, AgentWeave presents 70.18% fewer tools, uses 61.70% fewer input tokens, and exhibits 50.95% lower mean local-model latency. The result is deliberately narrow: this is a BFCL-derived routing-pressure study rather than an official full BFCL leaderboard score, and absolute task success remains low. The evidence nevertheless shows that candidate-space construction can materially affect a fixed model’s function-calling behavior and motivates evaluating routing as a distinct stage before model reasoning.
[NLP-40] Signal or Noise? A Benchmark Study of Agent Skills in Web Development
【速读】： 该论文旨在解决生成式 AI（Generative AI）在代码生成任务中引入代理技能（Agent Skills）时所面临的有效性评估难题。现有方法仅关注技能是否能帮助代理完成任务，却忽略了技能注入本身可能带来的负面影响，尤其是因提示（prompt）长度增加引发的性能退化问题。其解决方案的关键在于提出 WebDev-Skills-Bench 基准测试框架，并通过控制实验设计（包括长度匹配的无关对照组与留一法消融分析），系统性地分离技能内容效应与提示长度伪影的影响。研究发现，尽管部分技能在特定模型-项目组合中有效，但整体上技能注入导致平均 Pass@2 下降 1.3% 至 4.2%，任务完成深度降低，且令牌开销上升 72% 至 394%；更关键的是，仅有 17% 至 36% 的技能-项目对表现出实际增益。进一步分析揭示两种失败模式：长度敏感型（长度干扰）和内容误导型（内容干扰），表明技能的有效性高度依赖于具体的模型-项目-技能三元组，而非通用可复用资产。因此，论文主张将技能注入视为针对具体部署场景的路由决策，并强调长度匹配对照组和每模型审计应成为代理技能评估的最低标准。
链接: https://arxiv.org/abs/2608.23067
作者: Ziyue Yang,Fan Ding
机构: Baidu(百度), NLP; Beijing, China
类目: Computation and Language (cs.CL)
备注:
Abstract:Agent Skills are reusable procedural modules that are increasingly injected into coding-agent sessions to encode framework conventions, anti-patterns, and reusable tools. However, because each injected Skill expands the prompt of every query, an effective Skill benchmark must determine not only whether an agent can solve a task, but whether the Skill should have been injected at all. We introduce WebDev-Skills-Bench and use it for a controlled empirical study of 31 public WebDev Skills on 50 Web-Bench projects and 1,000 ordered tasks. The benchmark compares four matched conditions, including a length-matched irrelevant control and leave-one-out component ablations. To isolate Skill effects from prompt-length artifacts, we place only this http URL in the prompt while mounting auxiliary files into the agent workspace. Across four models, target Skill injection reduces mean Pass@2 by 1.3% to 4.2%, lowers task completion depth, and increases token cost by 72% to 394%, with gains in only 17% to 36% of Skill-project pairs. Length-matched controls reveal two failure modes: some models are length-distracted, where an equally long irrelevant Skill reproduces most of the loss, while others are content-misled, where prompt length is neutral but Skill content still lowers Pass@2 by 1.1% to 1.4%. Further analysis shows that losses concentrate on easy early tasks, Skill rankings transfer weakly across models, and anti-pattern rules outperform example-heavy content within helpful Skills. These findings recast a matched Skill as a hypothesis about a particular Skill-project-model triple rather than a portable asset, reframing injection as a per-deployment routing decision and making length-matched controls and per-model audits a minimum standard for Agent-Skill evaluation.
[NLP-41] Beyond Verdicts: A Graph-Based Analysis of Human and LLM Reasoning in Scientific Fact-Checking
【速读】： 该论文旨在解决生成式人工智能（Generative AI）在科学事实核查中存在的一种关键问题：尽管现有基于大语言模型（Large Language Models, LLMs）的自动事实核查系统能够判断模型是否得出错误结论并生成解释，但无法有效评估其推理路径是否与人类专家一致，或是否通过另一种虽不同但依然合理的路径得出相同结论。为应对这一挑战，论文提出一种基于图结构的框架——有类型推理图（typed reasoning graph），用于对齐和比较人类与LLM在科学事实核查中的推理路径。该框架以生物医学虚假信息中的谬误推理研究（MISSCIPLUS）为基础，将每个解释建模为连接错误主张、相关研究背景、研究发现、支持谬误的前提及谬误标签的推理图，实现人机推理在特定谬误子图层面的一一对应。对于未与人类推理对齐的LLM路径，进一步验证其在引用研究中的依据性、与主张的相关性以及对结论的充分性。在包含84个虚假主张的MISSCIPLUS数据集上，对GPT-5、Claude Opus 4.7和Qwen3-32B三种模型在不同提示与证据设置下的表现进行评估，结果显示各模型在性能维度上呈现显著差异：Qwen3-32B具有最低的结论判定失败率，GPT-5表现出最高的与人类推理的一致性，而Claude Opus 4.7虽在预测正确结论方面表现较弱，但在成功案例中常展现出合理有效的推理过程。该研究的关键在于构建可解释、可对比的推理图结构，从而实现对生成式AI推理行为的精细化分析与评估。
链接: https://arxiv.org/abs/2608.23047
作者: Abdul Ghafoor,Muhammad Arslan Manzoor,Yufang Hou
机构: Interdisciplinary Transformation University Austria (ITU)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Misinformation that cites legitimate papers can be especially harmful when it distorts what those studies actually report. While existing automatic fact-checking systems based on large language models (LLMs) can assess whether a model assigns an Incorrect verdict and can gen- erate explanations for that decision, they typi- cally do not indicate whether the model follows the same reasoning path as human experts or arrives at the verdict through a different but still valid path. In this work, we introduce a graph- based framework (typed reasoning graph) for comparing human and LLM reasoning paths in scientific fact-checking. Building on prior work on fallacious reasoning in biomedical misinformation, MISSCIPLUS (Glockner et al., 2025), we model each explanation as a rea- soning graph that links the false claim to the relevant study context, study findings, fallacy- supporting premises, and fallacy labels. This representation enables one-to-one alignment of human and LLM reasoning at the level of fallacy-specific sub-graphs. For non-human- aligned LLM paths, we validate grounding in the cited study, relevance to the claim, and suf- ficiency for the verdict. Using 84 false claims from MISSCIPLUS, we evaluate GPT-5, Claude Opus 4.7, and Qwen3-32B across prompt and evidence settings. Results show distinct perfor- mance dimensions: Qwen3-32B has the lowest verdict failure rate, GPT-5 the highest human alignment, and Claude Opus 4.7 weak verdict prediction but often valid reasoning in success- ful cases
[NLP-42] he Multilingual FrameNet Corpus EMNLP2026
【速读】： 该论文旨在解决跨语言语义角色标注中因缺乏高质量、统一格式的多语言语义框架资源而导致的模型泛化能力不足问题。现有研究在多语言和跨语言场景下的语义解析性能受限于单一语言资源的局限性，难以有效迁移至其他语言。为此，论文提出构建多语言框架语料库（Multilingual FrameNet Corpus, mFNC），通过整合并标准化九种语言（包括巴西葡萄牙语、中文、荷兰语、法语、德语、意大利语、韩语、拉脱维亚语和瑞典语）的已有语言特定语料库，建立首个大规模、跨语言一致的框架语义标注数据集。其解决方案的关键在于：通过统一的标注规范对多语言语料进行对齐与整合，从而支持基于不同神经网络架构的生成式语义解析模型在多语言和跨语言设置下的联合训练，显著提升模型的泛化性能。实验结果表明，基于mFNC训练的框架语义解析器（Frame Semantic Parser, FSP）在多个基准测试中均超越现有最优方法，验证了多语言训练数据对提升跨语言语义理解能力的核心作用。
链接: https://arxiv.org/abs/2608.23037
作者: Beatrice Fiumanò,Nicolas Lazzari,Simone Paolo Ponzetto,Valentina Presutti
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted at EMNLP 2026 Main Conference
Abstract:This paper introduces the Multilingual FrameNet Corpus (mFNC), a novel resource that extends the English Berkeley FrameNet corpus by collecting and harmonizing existing language-specific corpora across nine additional languages: Brazilian Portuguese, Chinese, Dutch, French, German, Italian, Korean, Latvian and Swedish. By training models that rely on different architectures on the mFNC, we consistently outperform existing state-of-the-art Frame Semantic Parsers in both multilingual and cross-lingual settings, underscoring the importance of multilingual training data. The mFNC and our trained FSP models are openly available at this https URL.
[NLP-43] ST2U: Stateful Test-Time Unlearning via Restricted Knowledge Boundary Control
【速读】： 该论文旨在解决大语言模型在推理阶段存在受限知识（restricted knowledge）泄露的问题，尤其是在测试时遗忘（test-time unlearning）场景下，现有激活编辑方法因仅进行局部点对点修正，未能有效应对自回归生成过程中隐藏状态从提示词、缓存及已生成前缀持续重构的动态特性，导致即使局部修正成功，后续生成仍可能重新进入受限知识区域，造成知识复现。其解决方案的关键在于提出一种基于受限知识边界控制的有状态测试时遗忘方法（ST²U），将遗忘过程建模为轨迹级的全局边界控制问题：首先在低维可逆坐标系中建模受限知识边界，保持正交非目标分量不变；推理时通过上下文锚定机制监控轨迹风险，施加最小化边界修正，并跨标记传递历史修正状态，从而实现对知识复现的长期抑制。该方法在三个基准和三种模型族上的实验表明，相较于现有基线，ST²U在保留非目标能力的同时，显著降低了受限知识重入率（13.76%–19.84% vs. 46.50%–59.10%），实现了遗忘效果与性能保持之间的最优平衡。
链接: https://arxiv.org/abs/2608.23034
作者: Xunlei Chen,Qinghui Gong,Ruini Xue,Yaodong Hu,Tian Lan,Wenhong Tian
机构: 未知
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:Controlling restricted knowledge in large language models is essential for model alignment and safe deployment. Test-time unlearning avoids costly retraining and parameter updates by intervening only during inference. However, existing activation-editing methods apply isolated pointwise corrections, overlooking how autoregressive generation continually reconstructs hidden states from the prompt, cache, and generated prefix. Consequently, later states may return to restricted knowledge regions after a locally successful correction, causing restricted knowledge re-entry. In this work, we propose Stateful Test-Time Unlearning via restricted knowledge boundary control (ST ^2 U), which formulates test-time unlearning as trajectory-wide boundary control. ST ^2 U first models restricted knowledge boundaries in low-dimensional invertible coordinates while leaving orthogonal non-target components unchanged. During inference, ST ^2 U monitors risk along the trajectory, applies minimal boundary corrections with contextual anchoring, and propagates historical correction states across tokens to mitigate knowledge re-entry. This trajectory-wide control enables more persistent forgetting while preserving non-target capabilities and limiting inference overhead. Across three benchmarks and three model families, ST ^2 U delivers the strongest overall balance, combining best or second-best retention with competitive forgetting and substantially less restricted-knowledge re-entry than test-time baselines (13.76%-19.84% versus 46.50%-59.10%).
[NLP-44] Meta-Moderator: Empowering Multi-Agent Debate with Meta-Cognition EMNLP2026
【速读】： 该论文旨在解决多智能体辩论（multi-agent debate）中因监督机制薄弱而导致的推理效率低下与证据聚合不可靠的问题。现有方法通常依赖于固定的预算、基于共识的终止条件或未经训练的评判者，易引发冗余讨论且难以有效整合信息。其解决方案的关键在于提出一种可学习的元监督框架——Meta-Moderator，将监督过程建模为一种元认知（meta-cognitive）行为，具备监控辩论价值、动态控制讨论进程以及最终裁定答案的能力。Meta-Moderator通过面向结果的策略优化独立于辩论智能体进行训练，使辩论调控成为显式可学习的能力而非提示工程的附带效应。实验表明，该框架在五个基准测试中均优于广泛使用的决策层，并展现出良好的跨任务与系统配置的泛化能力；进一步分析显示，其能更精准地分配辩论资源，在出现有信息量的假设后显著减少错误的证据聚合。
链接: https://arxiv.org/abs/2608.23029
作者: Wentao Hu,Zhuoyue Wan,Jinhao Shen,Chen Jason Zhang,Xiaoyong Wei,Qing Li
机构: The Hong Kong Polytechnic University(香港理工大学); Sichuan University(四川大学)
类目: Computation and Language (cs.CL)
备注: Accepted by EMNLP 2026 Findings
Abstract:Multi-agent debate can improve large language model reasoning by eliciting diverse hypotheses and critiques, yet its performance is often constrained by weak moderation. Common pipelines rely on fixed budgets, agreement-based stopping, or untrained judges, leading to redundant deliberation and unreliable evidence aggregation. We cast moderation as a meta-cognitive process, monitoring debate utility, controlling deliberation, and adjudicating a final answer, and introduce Meta-Moderator, a learnable framework that dynamically regulates debate and decides when to finalize an answer. Meta-Moderator is trained independently of the debaters via outcome-driven policy optimization, making debate regulation an explicit capability rather than an incidental effect of prompting. Across five benchmarks, Meta-Moderator outperforms widely used decision layers and transfers across tasks and system configurations. Further analyses show that it allocates debate more selectively and reduces mis-aggregation after informative hypotheses appear.
[NLP-45] Beyond Surface Cues: Disentangling Sociocultural Signals in Multilingual LLM s
【速读】： 该论文旨在解决多语言大语言模型（Multilingual Large Language Models, LLMs）在不同社会文化语境下输出结果存在差异的问题，尤其关注现有评估方法中因表面线索（如姓名、语言标识、显性或隐性文本提示）被误认为文化理解而引发的偏差。其核心挑战在于：当前的多语言审计往往将语言特征或身份标签等表层信号错误地解释为模型具备跨文化认知能力，从而掩盖了潜在的社会偏见。为此，研究提出了一种经人工验证的多智能体审计框架，关键在于将三个独立问题明确分离：1）模型是否再现社会偏见；2）不同身份群体是否被差异化呈现；3）输出是否反映真实的跨文化模式。通过分析12个模型在英语、法语和中文环境下针对18种职业与三种任务条件生成的89,253条输出，研究发现偏见表现随语言与任务类型系统性变化；移除直接身份线索后，英语和中文中的身份标签预测显著下降，但法语中影响较小。此外，源语言所关联的文化背景在所有语言-文体组合中均获得最高平均相关性评分，且自动化与人工评分具中等一致性。然而，在翻译及姓名掩码后，源语言识别能力急剧下降。这表明，若不控制这些表面线索，多语言审计极易将表层信号误判为深层文化理解，导致对跨文化差异与偏见的误读。因此，该研究提出的审计框架通过解耦表面特征与真实文化模式，提供了一个更可靠、可区分的评估路径，有效避免将“语言捷径”误当作“文化认知”。
链接: https://arxiv.org/abs/2608.23026
作者: Yuanjun Feng,Tanzhou Liu,Stefan Feuerriegel,Yash Raj Shrestha
机构: University of Lausanne (洛桑大学); LMU Munich (慕尼黑大学), Munich Center for Machine Learning (MCML) (慕尼黑机器学习中心), Germany
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Multilingual LLM outputs can vary across sociocultural contexts. However, evidence of cultural grounding can be misleading: identity labels may be inferred from explicit or indirect textual cues, while names and wording can reveal the source language. Treating all these signals as evidence of cultural grounding may obscure potential biases. We present a human-validated, multi-agent audit that separates three questions: whether outputs reproduce social biases, whether identity groups are represented differently, and whether outputs reflect cross-cultural patterns. The study analyzes 89,253 outputs from 12 LLMs in English, French, and Chinese, spanning 18 occupations and three task conditions. We find that bias representation varies systematically across languages and tasks. Removing direct identity cues sharply reduces identity-label prediction in English and Chinese, but has a much smaller effect in French. Across all language-genre settings, the cultural context associated with the source language receives the highest average relevance score, with moderate agreement between automated and human ratings. However, the ability to identify the source language drops substantially after translation and again after masking names. Without these controls, multilingual audits may mistake surface cues for cultural understanding, leading to misleading conclusions about cross-cultural variation and bias. Our audit offers a practical framework for separating such shortcuts from more meaningful cross-cultural patterns. Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) Cite as: arXiv:2608.23026 [cs.CL] (or arXiv:2608.23026v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2608.23026 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-46] Most of the LLM routing gap is task type
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）路由机制在实际应用中表现不佳的核心问题，即现有路由方法难以显著超越单一最优模型的性能上限。其关键发现是：当前大多数路由方法所宣称的“性能提升”实际上可能源于模型输出在多次运行中的随机波动，而非真正有效的动态选择策略。研究通过在一个包含14个模型、294个问题、7种任务类型及3种语言（韩语、英语、印地语）的全矩阵实验中重复运行，发现高达5.37%的模型-问题对在不同运行间得分存在差异，表明结果本身具有较高的不稳定性。在此基础上，作者提出一个严格的评估标准——仅当模型在两次独立运行中均正确回答时才计为正确，从而剔除随机噪声的影响。最终发现，仅有29个问题可通过路由优化改进，且这些可优化的问题主要由任务类型决定。进一步分析表明，为每类任务预先分配一个固定最优模型（静态表），即可解决其中21个问题；若按语言细分任务类型，可再优化2个，剩余6个才是真正需要学习型路由器处理的复杂情况。而这一剩余量甚至小于运行间的随机波动幅度，说明当前路由技术所面对的实际优化空间极为有限。因此，该研究的关键在于揭示“大部分路由性能差距本质上源于任务类型差异”，并指出简单采用静态任务-模型映射已能实现接近最优的性能，远超现有学习型路由器的表现。
链接: https://arxiv.org/abs/2608.23023
作者: Janghoon Lee
机构: Redrob
类目: Computation and Language (cs.CL)
备注: 21 pages, 2 figures, 12 tables
Abstract:An LLM router picks which model should answer each query. The appeal is that models fail on different questions. Whatever single model is best overall still gets some wrong, and another model in the pool gets many of those right. Getting that choice right every time is the ceiling, and a router is an attempt to approach it. However, recent work reports that routers do not get close. Across 21 routing methods on five benchmarks, sharply different designs land within a fraction of a point of each other, and all of them stay far below that ceiling. Learned routers often fail to beat simply always calling the strongest model. We ask what those missed questions have in common. We set fourteen models to answer all 294 questions, with 7 task types across 3 languages: Korean, English and Hindi. We ran the whole matrix twice, changing nothing, but 5.37% of the 4,116 model-question pairs came out scored differently anyway. Run-to-run movement like that is normal, and we argue that a small win does not show that routing did anything, ours or anyone else’s. Counting an answer correct only when the model got it right in both runs, 29 questions on this matrix can be improved with routing. Every correct-answer count here is on that rule. Task type accounts for most of them: assigning each task type one model in advance, chosen once and never updated, improves 21 of the 29. Splitting each task type by language improves 2 more and leaves 6 of 294 unoptimized. That handful is what a learned router would have been built for, and it is smaller than the run-to-run movement above, which is a share of pairs rather than of questions. The static table we adopted answers 262 of 294 questions at 3.33 per run, against the best single model’s 245 at 7.69. All of this is fitted and scored on the same 294 questions with no holdout. Comments: 21 pages, 2 figures, 12 tables Subjects: Computation and Language (cs.CL) Cite as: arXiv:2608.23023 [cs.CL] (or arXiv:2608.23023v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2608.23023 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Janghoon Lee [view email] [v1] Mon, 24 Aug 2026 09:28:35 UTC (44 KB) Full-text links: Access Paper: View a PDF of the paper titled Most of the LLM routing gap is task type, by Janghoon LeeView PDFHTML (experimental)TeX Source view license Current browse context: cs.CL prev | next new | recent | 2026-08 Change to browse by: cs References Citations NASA ADSGoogle Scholar Semantic Scholar export BibTeX citation Loading… BibTeX formatted citation loading… Data provided by: Bookmark checked="checked"class=“labs-tab-input”> Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv’s community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) mathjaxToggle(); We gratefully acknowledge support from our major funders, member institutions, , and all contributors. About Help Contact Subscribe Copyright Privacy Accessibility Operational Status (opens in new tab) Major funding support from
[NLP-47] Unlearning Is Not Just Erasing: Temporal Decoupling via Generation Inequality
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在满足隐私法规与安全要求时，如何实现精准遗忘（unlearning）的问题。现有方法在序列或标记层面通过惩罚目标输出来实现遗忘，但未考虑上下文依赖的检索路径，易破坏语言结构或抑制良性知识。其核心解决方案在于提出一种细粒度、基于训练的框架ADU，将遗忘机制从传统的标记擦除转向上下文注意力路径的解耦。ADU利用局部注意力头与全局注意力头的功能差异，识别出持续检索敏感信息的关键位置，并固定其候选路径；随后通过训练注意力投影适配器，在不破坏局部注意力结构的前提下，抑制沿这些路径的注意力权重分布，同时保持保留集的语言建模能力。最终通过后训练激活交换测试验证遗忘效果的传递性。实验表明，ADU在TOFU和WMDP基准上均取得最优综合性能，其在TOFU上的遗忘质量达0.93，同时保留87%–98%的模型通用能力（平均92.9%，显著优于基线的81.9%），且在良性上下文中副作用更小。
链接: https://arxiv.org/abs/2608.23020
作者: Xunlei Chen,Qirui Ye,Yuang Li,Yi Gong,Zhaokun Wang,Wenyi Li,Shiyao Guo,Jinyu Guo
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models (LLMs) require effective unlearning to address privacy regulations and safety concerns. However, achieving precise forgetting without compromising general utility remains challenging. Existing sequence- and token-level methods penalize target outputs without modeling their context-dependent retrieval paths, which can disrupt linguistic structure or suppress benign knowledge. We present ADU, a fine-grained, training-based framework that shifts unlearning from token erasure to contextual attention-pathway decoupling. Exploiting the functional distinction between local and global attention heads, ADU identifies preplan positions that retrieve persistent sensitive anchors and fixes their candidate paths under the original model. It then trains attention-projection adapters to suppress attention mass along these paths while preserving local-attention structure and retain-set language modeling. Post-training activation exchange tests whether the modified attention-output module transmits the learned forgetting effect. ADU achieves the strongest aggregate performance among evaluated baselines on the TOFU and WMDP benchmarks, including a Forget Quality of (0.93) on TOFU. It preserves 87–98% of model utility (92.9% on average versus 81.9% for baselines) while reducing side effects in benign contexts.
[NLP-48] PatchWrite: One Line Not One Section – Compile-Gated Validity-Preserving Editing for AI-Drafted Manuscripts
【速读】： 该论文旨在解决自动化稿件生成流程中因局部缺陷修复而引发的全局内容扰动问题，即传统方法在修复局部错误时会重新生成整个章节，导致无关的指标和引用发生意外变化，即使最终生成的PDF仍能成功编译。其核心解决方案是PatchWrite，其关键在于通过双重约束机制确保编辑的可信性：一是引入致命日志（fatal-log）检查以严格限定编译通过标准，二是采用证据锁（evidence locks）机制，要求每个被引用的关键词和实验数值必须由参考文献注册表或实验日志提供可验证证据。只有同时通过这两项检查的候选编辑才能被提交，否则将回滚至前一版本。实验表明，在24篇论文、8类故障的基准测试中，全槽重写在所有情况下均破坏了无关的“12层”描述（0/192保留），而PatchWrite实现了192/192的完全保留；移除任一检查都会导致接受率下降或错误通过。在生成模型实际提出编辑的场景下，75%的候选被接受，且绝大多数失败源于语法不支持的空替换操作，所有通过的编辑均满足双重验证，93.75%有效修复了注入的故障。盲评结果显示，用户更偏好PatchWrite在保持实验室实证事实方面的稳定性（C1量表评分5.0 vs. 2.0），同时对文稿质量评价无显著差异。真实产品中的193个起草任务日志也验证了此类故障在实践中普遍存在。
链接: https://arxiv.org/abs/2608.23001
作者: Weiwei Yang
机构: Solus(索鲁斯)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Software Engineering (cs.SE)
备注: 12 pages, 7 figures
Abstract:Automated manuscript pipelines often regenerate an entire section to repair a local defect, allowing unrelated metrics and citations to change even when the resulting PDF still builds. PatchWrite instead constrains how candidate edits become committed manuscript states: it reuses bounded EDIT N M editing and rollback, but tightens compilation acceptance with fatal-log checks and adds evidence locks that require every cited key and experimental numeric token to be attested by a reference registry or experimental log. Candidates that fail either check are rejected and the previous HEAD is retained. On a 24-manuscript x 8-fault oracle stress test (768 jobs, evenly split between compile-breaking and content-only faults), whole-slot rewriting mutated an unrelated “12-layer” line in every case (0/192 preserved; numeric Jaccard 0.6667), whereas PatchWrite preserved it in 192/192 cases. Removing the compile gate reduced acceptance to 0, while removing the evidence gate allowed a hallucinated citation to pass. The same pattern held across all eight faults. To test the protocol with generation rather than oracle edits, we reran the 192 jobs with the writer model proposing the edits. The model’s candidates were accepted in 75% of cases; nearly all rejections came from one reproducible failure mode in which the model attempted to delete a line using an empty replacement unsupported by the current grammar. Every accepted candidate passed both gates, and 93.75% fixed the injected fault; the remaining cases involved a technically valid but sentence-inappropriate citation and one markup-changing near-miss. In a blind evaluation of sixteen PDF pairs, both raters preferred PatchWrite for preserving lab-grounded facts (C1 Likert 5.0 vs. 2.0), while rating prose quality nearly identically. Logs from 193 in-product drafting tasks show the same classes of failures occurring in practice.
[NLP-49] What Does Activation Steering Control? Attribution Across Answer Encodings and Output-Sensitive Subspaces
【速读】： 该论文旨在解决生成式模型中激活引导（activation steering）评估的可靠性问题，即现有方法在特定答案编码（answer encoding）下报告的性能提升可能源于对训练时所见答案标识符的兼容性，而非真正实现了对目标语义概念的控制。其核心解决方案是提出交叉编码引导评估（Cross-Encoding Steering Evaluation），通过在保持干预不变的前提下，对相同保留样本重新编码答案，以检验引导效果是否依赖于具体的标识符映射。研究发现，在NormBank数据集上，当答案标识符（如A/B/C）被重分配后，对比激活添加（CAA）方法更倾向于引发提取索引（extraction index）相关的分数变化，而非语义标签（semantic label）的变化，这种现象称为“提取索引跟随”（extraction-index following）。进一步实验表明，该效应与行位置无关，而是严格跟随提取索引，并主要出现在模型深层。此外，仅15.4%方向平方范数的低秩输出敏感组件即可保留96.3%的该效应，表明其高度集中于输出层的特定子空间。同时，基于推理时干预（ITI）的方法在多个模型中也表现出对提取索引的偏好。总体来看，不同任务（如MNLI vs SC101）呈现不同行为模式，且选择题与开放式评估可能得出矛盾结论，因此，单一编码下的引导增益无法独立确认干预的实际控制对象。关键在于：必须通过跨编码验证来区分干预是否真正作用于语义内容，还是仅仅响应了形式化的标识符结构。
链接: https://arxiv.org/abs/2608.22985
作者: Zhiwei Gao,Shaowen Peng,Shoko Wakamiya,Eiji Aramaki
机构: Nara Institute of Science and Technology (奈良先端科学技術大学院大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Activation steering is often evaluated under the answer encoding used to construct the direction. A reported gain may reflect the intended judgment or compatibility with answer identifiers seen during construction. We introduce Cross-Encoding Steering Evaluation, which freezes an intervention while re-encoding answers to the same held-out items. On NormBank, after A/B/C identifiers are reassigned, contrastive activation addition (CAA) induces larger target-versus-source score changes for the extraction indices than for the semantic labels under the new mapping. We call this extraction-index following. Varying identifier vocabulary (A/B/C, X/Y/Z, or 1/2/3) and row order shows that the effect tracks extraction index rather than row position. After matching direction norms across layers, extraction-index following emerges mainly at later depths. A low-rank output-sensitive component containing 15.4% of the direction’s squared norm retains 96.3% of this effect. An Inference-Time Intervention (ITI)-style method also favors extraction-index over semantic-label following on NormBank in three models. In aggregate, MNLI favors extraction-index following, whereas Social Chemistry 101 (SC101) favors semantic-label following. Multiple-choice and open-ended evaluations can yield different behavioral conclusions. Thus, a steering gain under one answer encoding does not by itself identify what the intervention controls.
[NLP-50] Closed-Loop Bayesian Molecular Inverse Design with Semantic LLM Surrogates
【速读】： 该论文旨在解决分子逆向设计中受限于有限的实验或计算资源（即“查询预算”）下，如何高效提升生成分子满足目标性质谱的比率这一关键挑战。传统贝叶斯优化（Bayesian Optimization, BO）虽适用于此类闭环候选池增益场景，但其基于高斯过程（Gaussian Process, GP）的代理模型通常在压缩的连续嵌入空间中运行，忽略了化学家在决策时依赖的子结构信息与参考分子相似性等可解释信号。本文提出一种新型闭环框架——\textbf{Method}，其核心创新在于将设计决策的主阵地从生成器转移至代理模型：代理模型以冻结的大语言模型（Large Language Model, LLM）为基础，直接在原始文本形式（如任务指令、SMILES表示的历史优化记录及真实反馈）上进行推理，从而保留丰富的语义与结构信息。在每轮迭代中，代理模型输出一个结构化的决策信号，依据探索与利用权衡原则选择具有信息量的参考分子，并可附带简洁的引导语句；该信号进一步转化为下一轮生成器的条件输入文本，形成可解释的自然语言级优化轨迹。实验结果表明，该方法在MolQA药物与材料设计任务中显著优于一次性提示（one-shot prompting），且在性能上可媲美甚至超越基于GP的基准方法，同时揭示出领域依赖性的界面特性：对于二分类药物靶点任务，仅使用参考分子迁移效果最佳；而对于连续型材料属性优化，加入代理模型的简要摘要则更具优势。
链接: https://arxiv.org/abs/2608.22967
作者: Yaoyao Xu,Xinjian Zhao,Xiaozhuang Song,Lei Bai,Tianshu Yu
机构: The Chinese University of Hong Kong, Shenzhen (香港中文大学（深圳）); Shanghai Artificial Intelligence Laboratory (上海人工智能实验室)
类目: Computation and Language (cs.CL)
备注: 28 pages
Abstract:Practical molecular inverse design is rarely a one-shot generation problem; it often takes the form of closed-loop candidate-pool enrichment, where under a limited oracle budget the goal is to \emphincrease the fraction of generated molecules that match a desired property profile. Bayesian optimization (BO) offers a natural framework for this setting, yet standard Gaussian-process surrogates typically operate in compressed continuous embeddings, which discard the substructural and reference-similarity signals that chemists naturally use to decide where to look next. We propose \textbf\method, a closed-loop framework in which the surrogate, rather than the generator, is treated as the locus of design choice, and instantiate it with a frozen large language model that reasons directly over the task instruction, SMILES-level optimization history, and oracle feedback in their native textual form. At each iteration, the surrogate returns a structured decision signal that selects informative reference molecules under an exploration and exploitation principle, optionally with a concise guidance sentence. This signal is converted into next-round conditioning text for a frozen molecular generator, yielding an inspectable optimization trace in natural language. Experiments on MolQA drug and material design tasks show that \method improves over one-shot prompting, is competitive with or stronger than GP-based BO baselines, and reveals a domain-dependent interface: reference-only transfer works best for binary drug targets, while adding a concise surrogate summary is more beneficial for continuous material
[NLP-51] Buried in Textual Debt: Context Pruning with Visual Evidence Preservation for MLLM Agents
【速读】： 该论文旨在解决多模态大语言模型（Multimodal Large Language Models, MLLMs）在作为多步骤代理执行任务时，因持续积累自生成推理文本而导致的“文本债务”（textual debt）问题。具体而言，随着任务轨迹的延长，冗余的推理文本会占据上下文空间，抑制对关键视觉证据的依赖，进而引发过拟合于自我生成语言、误导后续推理等问题。其核心挑战在于如何在不丢失有效视觉信息的前提下，精准识别并剔除冗余的推理内容。解决方案的关键是提出SPARE框架——一种基于Kullback–Leibler（KL）散度引导的推理文本剪枝方法。该框架引入一个紧凑的任务状态摘要（task-state summary）作为特权诊断上下文，通过对比原始上下文与摘要条件下的模型输出，利用反向KL散度从在线策略自蒸馏（OPSD）中评估摘要是否充分覆盖候选文本段而不破坏未来推理能力。进一步地，通过监督微调（SFT）优化摘要生成器，使摘要更紧凑、覆盖更广，从而支持更激进的剪枝策略。实验表明，SPARE在多个多步视觉工具使用基准上实现了最优平均准确率，同时删减了37.89%至64.58%的推理令牌，显著改善了准确性与上下文效率之间的权衡，有效恢复了对视觉证据的依赖，缓解了对自生成语言的过度依赖。
链接: https://arxiv.org/abs/2608.22963
作者: Yuchen Huang,Sijia Li,Jun Zhang,Yi R. Fung
机构: Hong Kong University of Science and Technology (香港科技大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 14 pages, 2 figures, 4 tables
Abstract:Multimodal Large Language Models (MLLMs) are increasingly deployed as multi-step agents, where explicit reasoning supports task decomposition and tool coordination but also accumulates self-generated text. Over long trajectories, this text can dominate the context and suppress visual evidence, creating textual debt. We observe that reasoning becomes redundant once task-relevant visual evidence is grounded, while stale hypotheses can misguide later inference when grounding remains uncertain. Pruning must therefore remove redundant text without discarding visual evidence. We propose SPARE, a Kullback–Leibler (KL)-guided framework for pruning accumulated reasoning in multimodal tool-use agents. SPARE uses a compact task-state summary as privileged diagnostic context. For each candidate segment, it replays the same model under the original and summary-conditioned contexts. Reverse-KL divergence from on-policy self-distillation (OPSD) then tests whether the summary sufficiently covers the segment without disrupting future reasoning. We further fine-tune the summarizer with supervised fine-tuning (SFT), enabling more compact summaries, broader coverage, and more aggressive pruning. Across multi-step visual tool-use benchmarks, SPARE achieves the highest average accuracy among pruning methods while removing 37.89–64.58% of reasoning tokens. This favorable accuracy–context trade-off shows that reducing textual dominance restores reliance on visual evidence and mitigates over-conditioning on self-generated language.
[NLP-52] he Illusion of Control: Why Bare Classifier Inversion Silently Fails in Concept-Bottleneck Text Generation EMNLP2026
【速读】： 该论文旨在解决生成式文本中多属性可控生成的难题，特别是在概念瓶颈（concept-bottleneck）框架下实现多轴组合泛化时，如何在推理阶段从目标属性配置准确合成低维概念码（concept code）的问题。其核心挑战在于：概念码本身不具有直接的语言模型流畅性约束（LM-fluency term），因此需通过正则化手段将其引导至编码器训练分布内。论文对比了三种获取推理时概念码的方法：基于编码器头部的分类器反演（classifier inversion）、参考文本编码（reference-text encoding）以及后验标签条件先验（post-hoc label-conditioned prior）。研究发现，所有反演变体（包括未加正则化的裸反演及多种正则化形式，如标签无关与标签相关的马氏距离惩罚、条件归一化流密度基线）均表现逊于仅基于各属性组合在相同检查点上拟合的简单后验先验模型，且在三个涵盖124M至8B参数量的骨干模型上一致如此。更关键的是，裸反演会悄然退化至随机水平，其根源可归因于直接测量到的离流形（off-manifold）概念码。研究通过真实世界基准和外部评估验证了该诊断，实现了与已有公开基线的公平比较。
链接: https://arxiv.org/abs/2608.22956
作者: Qi Bing,Xiaowei Shao
机构: Shanghai Jiao Tong University (上海交通大学)
类目: Computation and Language (cs.CL)
备注: Accepted to EMNLP 2026 (Main Conference)
Abstract:Concept-bottleneck controllable generation routes multi-attribute control through a low-dimensional concept code that, at deployment, must be synthesised from a target attribute configuration. We study this problem in concept-bottleneck text generation under multi-axis compositional generalisation, comparing three ways to obtain the inference-time code: classifier inversion against the encoder heads, reference-text encoding, and a post-hoc label-conditioned prior. Since a concept code admits no direct LM-fluency term, regularising inversion must instead constrain the code toward the encoder’s training distribution. We therefore test bare inversion and three regularised variants: label-agnostic and label-conditioned Mahalanobis penalties, and a conditional normalising-flow density baseline. Every inversion variant we test underperforms a simple post-hoc prior fitted to per-combination encoder means on the same checkpoints, across three backbone families spanning 124 M to 8 B parameters. The bare form of classifier inversion also silently collapses to chance, traceable to a directly measured off-manifold code. We validate this diagnosis on real-world benchmarks and under external evaluators, enabling fair comparison with published baselines.
[NLP-53] What Proves You Wrong: Benchmarking Language Models on Falsifiable Research Ideation
【速读】： 该论文旨在解决生成式研究想法时缺乏统一评估标准的问题，现有方法中自由形式的评判受风格与立场影响，而基于后续发表论文的评分则仅能回溯已实现的研究路径，无法客观判断研究设想的内在质量。其解决方案的关键在于提出一个名为Lit2Test的基准测试体系，该体系以“从文献到可检验性”（Literature to Test）为核心设计原则，构建了一个包含六个要素的契约框架，要求每个研究设想预先承诺一个可证伪的观测结果，从而将研究想法的质量从主观争论转化为可判定的科学问题。该基准基于200个真实论文邻域前瞻性构建，通过四款前沿大语言模型生成研究提案，并在1,200次双盲对比中进行评估，同时采用诊断性控制与有限人工校准确保评估可靠性，三名标注者在明确界定的可信区间内达成一致。结果显示，在全部10,000次自助重采样中均能严格区分四类模型的性能，且差异主要源于所提检验方法与度量指标的质量，而非表面语言流畅性。研究团队公开发布该基准、构建流程及审计数据，以促进可复现的生成式研究评估。
链接: https://arxiv.org/abs/2608.22948
作者: Ziyue Wang(1),Aomufei Yuan(2),Yiran Yao(3),Linli Yao(1),Hongyao Zuo(3),Ziwen Gong(4),Yuanxin Liu(1),Shicheng Li(1),Yishuo Cai(1),Tong Yang(2),Xu Sun(1),Xiaohui Li(5),Haoli Bai(5) ((1) State Key Laboratory of Multimedia Information Processing, School of Computer Science, Peking University, (2) Peking University, (3) Tianjin University, (4) Hainan University, (5) Huawei Technologies)
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Equal contribution by Ziyue Wang, Aomufei Yuan and Yiran Yao. Corresponding authors: Tong Yang and Xu Sun
Abstract:Large language models are increasingly used to propose research ideas, yet the prevailing ways of judging such ideas supply no shared decision rule: free-form judging sways with style and position, and scoring against a later paper rewards recovery of one realized trajectory. We introduce a benchmark that carries a proposal from Literature to Test: the Lit2Test benchmark centers on a six-field contract organized around a falsifying outcome, so that every proposal precommits the observation that would prove it wrong, making its quality decidable in the first place rather than merely arguable. Built prospectively from 200 real-paper neighborhoods, Lit2Test elicits proposals from four frontier models and compares them through 1,200 pairwise comparisons judged blind in both presentation orders. The protocol audits its own reliability through diagnostic controls and bounded human calibration, with three annotators corroborating the conclusions within explicitly stated reliability bounds. Lit2Test recovers a strict ranking of the four models in all 10,000 bootstrap replicates, and the separation comes from the quality of the proposed tests and metrics rather than from surface fluency. We release the benchmark, construction pipeline, and audit artifacts for public use.
[NLP-54] HelaBERT: Enhancing Sinhala Language Understanding with Dual Pooling Classification Head
【速读】： 该论文旨在解决斯里兰卡僧伽罗语（Sinhala）自然语言处理（Natural Language Processing, NLP）领域中高质量预训练语言模型匮乏的问题，特别是在低资源语言背景下，缺乏针对僧伽罗语复杂黏着形态和书写系统设计的专用模型。其解决方案的关键在于构建并公开发布两个基于BERT架构的、从零开始预训练的僧伽罗语掩码语言模型——HelaBERT-Small（约2330万参数，6层）与HelaBERT-Large（约1.1亿参数，12层），它们均采用专为僧伽罗语设计的SentencePiece Unigram分词器（词汇量32,000），有效捕捉该语言的黏着性构词特征与复杂书写结构。此外，研究提出一种双池化（dual pooling）分类头，并在四个下游任务（新闻类别分类、新闻来源分类、情感分析、写作风格分类）上系统评估，结果显示该结构在情感分析任务中表现显著提升，且在新闻类别分类任务中对HelaBERT-Small有适度增益；而在短文本为主的新闻来源分类任务中，标准[CLS]-线性分类头仍具竞争力。该工作为僧伽罗语NLP研究提供了可复用的基础模型与优化方法。
链接: https://arxiv.org/abs/2608.22922
作者: Thisen Ekanayake,Nisansa de Silva
机构: University of Moratuwa(莫鲁塔瓦大学); Department of Computer Science Engineering(计算机科学与工程系)
类目: Computation and Language (cs.CL)
备注:
Abstract:We present HelaBERT, a family of two BERT-based masked language models pre-trained from scratch on approximately 1 billion tokens of Sinhala text sourced from MADLAD-400, CulturaX, and a custom corpus comprising news articles, Sinhala Wikipedia, and web crawl data. HelaBERT-Small (~23.3M parameters, 6 layers) and HelaBERT-Large (~110M parameters, 12 layers) both use a SentencePiece Unigram tokenizer (vocabulary size 32,000) tailored to Sinhala’s agglutinative morphology and complex script. We evaluate both models on four downstream Sinhala text classification tasks: news category classification, news source classification, sentiment analysis, and writing style classification, using 5 independent seed runs with stratified 80/20 train/test splits. We additionally propose a dual pooling classification head and evaluate it systematically across all four tasks, finding consistent improvements on sentiment analysis and a moderate gain on news category classification for HelaBERT-Small, while the standard [CLS]-linear head remains competitive on news source classification, a headline-level task with short average input length. We release both models to support further research in Sinhala NLP.
[NLP-55] Knowing Isnt Always Saying: When Do Spatial Encodings Reach Answers in Vision-Language Models? EMNLP2026
【速读】： 该论文旨在解决视觉-语言模型（Vision-Language Models, VLMs）中存在“编码-对齐鸿沟”（encoding-grounding gap）的问题，即模型虽在隐藏状态中编码了空间信息，却常未能有效利用这些信息进行回答。其核心解决方案是提出**方向补丁（direction patching）**这一类条件因果干预方法，通过在不同层、标记位置及提示格式上施加干预，系统性地追踪空间信息在模型内部的传播路径。关键发现在于：因果影响作用于答案概率分布（answer logits）仅在中深层（mid-to-deep depths）才显著显现；文本式思维链（text chain-of-thought）会抑制早期对象词的确定性传输（argmax-level transport），而视觉引导型提示则保持该通道开放；此外，目标日志概率增益可能低于最大值阈值，但信息传输可在最终前缀标记或深层中的答案生成步骤重新出现。这些局部效应在十种研究的VLM中呈现出可描述的传输模式，并随数据集、属性和编码强度变化而动态调整，从而将“编码-对齐鸿沟”重新定义为模型内部条件化信息传输（conditional transport）的结构性问题。
链接: https://arxiv.org/abs/2608.22916
作者: Zeyu Wang,Xinming Xu
机构: Peking University (北京大学); Tsinghua University (清华大学)
类目: Computation and Language (cs.CL)
备注: Accepted to appear in the EMNLP 2026 Main Conference
Abstract:Vision-language models are known to encode spatial information in their hidden states, yet often fail to use it when answering. However, it remains unclear when and where this encoded information reaches the answer. We address this with direction patching, a class-conditioned causal intervention applied across layers, token positions, and prompt formats. Using spatial-ID directions constructed following prior encoding evidence, we find that causal influence on answer logits emerges only at mid-to-deep depths. Text chain-of-thought suppresses immediate object-word argmax-level transport in most models, while visually grounded prompts keep it open. Positive target-logit gain can remain below the argmax threshold, and transport can re-emerge at the final prefix token or at the answer step in deeper layers. Across the ten VLMs we study, these local effects form descriptive transport patterns. Complementary experiments characterize how these patterns shift across datasets, attributes, and encoding amplitudes. Together, these results reframe the encoding-grounding gap as a problem of conditional transport in VLMs.
[NLP-56] Safety Hacking in Constrained Best-of-N Inference-time Scaling
【速读】： 该论文旨在解决生成式模型在推理阶段采用约束性采样（constrained Best-of-N）时，因安全代理模型（safety proxy）不完美而导致的安全漏洞问题。具体而言，当使用学习到的安全模型对多候选输出进行筛选并基于奖励模型选择最优解时，存在一种“两阶段失效”机制：首先，不完美的安全代理会将部分不安全的输出误判为可行（即污染可行集），随后奖励最大化过程会放大这种残留污染，导致最终输出虽通过安全检测但实际违反真实安全准则，这一现象被称为“安全劫持”（safety hacking）。论文的关键解决方案在于揭示了安全劫持发生的统计根源——即在代理可行集中，不安全但被接受的输出其奖励尾部分布比安全输出更重，当样本数 N 增大时，即使误报率和代理误差极小，安全劫持仍可能以概率趋近于1发生。为此，作者提出通过控制策略与代理可行分布之间的 χ2 散度来实现 N-无关的安全劫持上界，并引入受限悲观采样（constrained pessimistic sampling）作为该原则的具体实现。尽管覆盖控制可有效抑制奖励放大效应，但无法修复已被污染的可行集，因此仍可能存在被偏好且不安全的输出；此外，正则化选择并不必然比约束性 Best-of-N 更安全。实验通过简化模型和语言模型验证了污染传播及其奖励尾部放大的现象，揭示了在引入学习型安全模型的前提下，推理阶段扩展规模所面临的根本性挑战。
链接: https://arxiv.org/abs/2608.22915
作者: Akifumi Wachi,Takumi Tanabe,Youhei Akimoto
机构: LY Corporation(LY公司); University of Tsukuba(筑波大学); RIKEN AIP(理化学研究所人工智能中心)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Cryptography and Security (cs.CR)
备注:
Abstract:Inference-time pipelines often sample multiple outputs, filter them with a learned safety model, and return the proxy-feasible output with the highest learned reward. We show that this composition creates a two-stage failure: an imperfect safety proxy first contaminates the feasible set with unsafe outputs, and reward maximization can then amplify this residual contamination. We define \emphsafety hacking as selecting an output that passes the learned constraint but violates the true safety criterion. For constrained Best-of- N sampling, we derive finite- N bounds governed by the joint upper reward tails of safe and unsafe outputs within the proxy-feasible set. If unsafe-but-feasible outputs have the heavier tail, safety hacking becomes asymptotically certain as N grows, even when false-positive mass and average safety- and reward-proxy errors are arbitrarily small. We also show that policies within a bounded \chi^2 divergence from the proxy-feasible reference distribution admit an N -independent safety-hacking bound, and instantiate this general coverage-control principle with constrained pessimistic sampling. Coverage control limits amplification but cannot repair a contaminated feasible set: admitted unsafe outputs may still be favored, and regularized selection is not necessarily safer than constrained Best-of- N for every reward proxy. Toy and language-model experiments characterize both contamination and its reward-tail amplification, which exposes an inherent difficulty in inference-time scaling with learned safety models.
[NLP-57] Exploring Dowker Homology for Sentence Similarity
【速读】： 该论文旨在解决如何利用拓扑方法有效捕捉和表征句子间语义相似性的问题，尤其关注在预训练语言模型的隐空间中，通过将句子中词元（token）的嵌入向量视为位于同一空间中的两组点云，进而应用Dowker同调（Dowker homology）来分析其相对位置关系。其核心解决方案在于：将句子对的嵌入表示为点云对，并利用Dowker同调提取具有拓扑意义的特征，这些特征能够回归预测真实语义相似度分数，从而验证其对句子相似性信息的有效捕获能力。此外，研究进一步从Dowker同调特征中导出可解释的单数值摘要（single-number summaries），以提升其实际应用便利性。尽管这些摘要在性能上未超越基于标准池化策略的成熟句向量相似性度量方法，但其在可视化分析与模型可解释性方面展现出潜在价值。
链接: https://arxiv.org/abs/2608.22909
作者: Marius Huber,Juri Opitz
机构: University of Zürich(苏黎世大学)
类目: Computation and Language (cs.CL)
备注: 12 pages; accepted at “Workshop on Insights from Negative Results in NLP 2026”
Abstract:Dowker homology is a topological tool that may be used to analyze the relative position of two point clouds living in a common space. We investigate whether Dowker homology captures sentence similarity information by treating the embeddings of the tokens that constitute a sentence pair as a pair of point clouds in the latent space of a transformer model, using both models that have and have not been fine-tuned for sentence similarity. We find that Dowker homology captures sentence similarity information, as measured by regressing Dowker homology features onto ground-truth similarity scores, and that it can be used for visual inspection of similarity data and models. In an attempt to make Dowker homology readily applicable, we derive from it single-number summaries that we expect to capture sentence similarity directly. These turn out to work reasonably well, but without outperforming standard sentence similarity measures based on established pooling methods.
[NLP-58] Do Spoken Language Models Hear Speech as They Read Text? Bridging Structural Gaps Between Speech and Text EMNLP2026
【速读】： 该论文旨在解决当前语音语言模型（Spoken Language Models, SLMs）在指令遵循能力与跨任务泛化性能方面相较于文本基语言模型表现较弱的问题。其核心挑战在于，尽管现有SLMs在下游任务中表现出较强性能，但语音与文本表示之间的对齐仍不充分，主要源于连续、时变的语音信号与离散、静态的文本之间存在的结构性差异未被有效建模。为此，论文提出一种简化框架，通过解耦长度不匹配问题与语义对齐问题，显式促进语音与文本表示之间的紧密对应关系。该方案的关键在于主动处理语音与文本在结构上的本质差异，从而提升模型的语义理解与指令响应能力。实验结果表明，该方法在多个基准测试中达到与强基准相当的性能，验证了显式建模语音-文本结构性差异在SLM训练中的关键作用。
链接: https://arxiv.org/abs/2608.22908
作者: Hyeonyu Kim,Hwayeon Kim,Youngwon Choi,Myeongkyun Cho,Huu-Kim Nguyen
机构: Maum AI Inc.(Maum AI 公司); KAIST(韩国科学技术院); Atmanity Inc.(Atmanity 公司)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted to EMNLP 2026 Findings
Abstract:Spoken Language Models (SLMs) generate textual responses directly from speech, offering an alternative to cascaded systems. Despite recent advances, existing SLMs still exhibit weaker instruction-following behavior and limited generalization across diverse tasks compared to text-based language models. Our analysis shows that speech and text representations in current SLMs remain weakly aligned despite strong downstream performance, indicating that structural differences between continuous, temporally varying speech and discrete text remain insufficiently addressed. To address this, we propose a simple framework that decouples length mismatch from semantic alignment and encourages closer correspondence between speech and text representations. Experiments across multiple benchmarks demonstrate competitive performance against strong baselines, underscoring the importance of explicitly addressing structural differences between speech and text in SLM training. Our code is publicly available at this https URL.
[NLP-59] SelFusion: Self-distillation for Diffusion Language Models ACL2026
【速读】： 该论文旨在解决扩散语言模型（Diffusion Language Models, DLMs）在生成质量上相较于自回归大语言模型（Autoregressive Large Language Models, LLMs）存在明显退化的问题，从而限制了其实际应用。尽管知识蒸馏（Knowledge Distillation, KD）被视为提升DLM性能的潜在途径，但研究发现，直接套用传统KD方法仅能带来微弱改进，甚至会损害生成质量。为此，论文提出一种新颖的自蒸馏框架——SelFusion，其核心在于无需外部教师模型即可实现有效知识迁移：通过两次前向传播，分别采用高掩码概率（硬模式）与低掩码概率（易模式）进行建模，并引入双向知识蒸馏机制，基于词元级正确性动态确定蒸馏方向，以克服易模式可能对错误词元产生过度自信的问题。实验结果表明，该方法在指令遵循任务中显著优于使用外部LLM或DLM教师的其他蒸馏策略，在多数配置下，学生模型性能甚至超越原始LLM教师，为提升DLM生成质量提供了可行且高效的解决方案。
链接: https://arxiv.org/abs/2608.22898
作者: Hyeongsoo Lim,Jinyoung Kim,Eunseo Seo,Minho Jang,Jiwon Yoon
机构: Chung-Ang University (中央大学)
类目: Computation and Language (cs.CL)
备注: Published as a main conference paper at ACL 2026
Abstract:Diffusion language models (DLMs) alleviate the inherent latency bottleneck of autoregressive (AR) large language models (LLMs), but their degraded generation quality limits practical applicability. Although knowledge distillation (KD) can be a promising direction for improving performance, we empirically find that naively applying conventional KD yields only marginal gains, or even degrades generation quality. Based on these observations, we propose a novel self-distillation framework for DLMs, namely SelFusion. To enable effective KD without an external teacher model, SelFusion performs two forward passes with different masking levels, defining the hard mode with a larger masking probability and the easy mode with a smaller masking probability. However, the easy mode is not always more accurate than the hard mode and can be overconfident on incorrect tokens. Thus, we introduce bidirectional KD between the two modes, which can dynamically determine the distillation direction based on token-level correctness. Experimental results on instruction-following tasks show that the proposed self-distillation substantially outperforms other KD methods with external LLM and DLM teachers. In many configurations, the student trained with SelFusion even surpasses the performance of the LLM teacher, providing a practical path toward improving DLM generation quality. Source code can be found at this https URL
[NLP-60] AraDetox: A Multi-Dialect Arabic Detoxification Dataset
【速读】： 该论文旨在解决阿拉伯语有害语言检测后文本净化（detoxification）这一研究空白问题，尤其针对多方言阿拉伯语场景下的安全文本生成挑战。其核心解决方案的关键在于构建首个大规模多方言阿拉伯语净化数据集AraDetox，通过大语言模型（LLM）如GPT-5与Gemini 2.5 Flash在现代标准阿拉伯语、海湾阿拉伯语、黎凡特阿拉伯语和埃及阿拉伯语四种方言上自动生成84,000条去害化重写文本，并结合人工评估与自动分析（包括词汇变化、语义保真度、情感一致性及方言风格匹配度）验证生成质量。研究发现，有效的去害化本质上是语义保持的改写任务：尽管存在显著的词汇与结构重构，但生成文本仍维持高度语义相似性，且人类评估证实有害内容被有效移除的同时基本保留原意。此外，方言分析表明生成文本在风格上与目标方言语料库具有可测量的一致性。该工作揭示了基于大模型生成与人工验证相结合的大规模阿拉伯语去害化资源构建路径，为未来阿拉伯语安全文本生成及多方言自然语言处理研究提供了重要支持。
链接: https://arxiv.org/abs/2608.22894
作者: Mo El-Haj
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 15 pages, 6 figures
Abstract:Arabic harmful-language detection has received considerable attention, yet Arabic text detoxification remains underexplored. We introduce AraDetox, a multi-dialect Arabic detoxification dataset comprising 10,500 harmful social-media posts and 84,000 detoxified rewrites generated using GPT-5 and Gemini 2.5 Flash across Modern Standard Arabic, Gulf, Levantine, and Egyptian Arabic. The generated outputs were assessed through human evaluation and automatic analyses of lexical change, semantic preservation, sentiment, and dialectal style. Results show that detoxification is primarily a meaning-preserving rewriting task: substantial lexical and structural reformulation is accompanied by consistently high semantic similarity. Human evaluation confirms successful harmful-language removal while largely preserving the original meaning. Dialectal analyses further indicate that the generated variants exhibit measurable stylistic alignment with reference Arabic dialect corpora. Comparison with existing resources highlights two complementary approaches to detoxification: minimal-edit lexical substitution and meaning-preserving reformulation. Our findings demonstrate that large-scale Arabic detoxification resources can be constructed through LLM-assisted generation and human verification. The dataset is publicly available at this https URL to support future research on Arabic detoxification, safe text generation, and multi-dialect Arabic NLP.
[NLP-61] Proxy reliance in large language model decisions is uncalibrated to predictive evidence
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在医疗分诊与信贷决策等高风险场景中，如何区分任务相关推理与非法代理使用（proxy use）的问题。现有审计方法仅通过改变人口属性观察决策是否变化，但因与受保护群体相关的属性本身具有预测价值，决策变化可能源于合理推断而非歧视，导致误判。本文提出一种基于因果代理效应（causal proxy effect）的量化审计方法，在具备已知真实标签的临床排序任务中，精确计算模型对证据的依赖程度作为基准参考，从而实现对模型行为的三类判定：过度依赖、合理依赖与依赖不足。实验结果显示，尽管在中性标签下所有模型均未有效利用信息，但在引入具信息量的代理变量时，各类依赖模式均出现；社会领域名称的引入虽使依赖度下降，但在某一模型中低于基准值。进一步分析表明，模型对证据的依赖严重滞后于实际信息量，且社会标签抑制机制脆弱——上下文示例即可使其依赖度回升至非零水平。相比之下，基于准确率的评估完全无法捕捉此类偏差。该研究的关键在于引入可精确计算的因果基准，实现对代理依赖行为的细粒度、可解释的审计。
链接: https://arxiv.org/abs/2608.22887
作者: Zengqing Wu,Chuan Xiao
机构: University of Osaka(大阪大学); Osaka, Japan
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computers and Society (cs.CY)
备注:
Abstract:Large language models (LLMs) are entering decisions in triage and lending, where task-relevant inference must be distinguished from impermissible proxy use. Current audits ask whether decisions change when demographics change. But attributes correlated with a protected group carry predictive value, so a changed decision can be discrimination or sound inference. We measure causal proxy effects in four LLMs on a clinical-ranking task with known ground truth, where the reliance the evidence warrants can be computed exactly and used as the reference. One audit signal yields three verdicts: over-reliance, warranted and under-reliance. Under neutral labels every model relies on proxies with no information. Informative proxies draw all three. Social field names push reliance down, below the reference in one model. Two findings explain this. Reliance severely undertracks the evidence, and social-label suppression is fragile, since in-context examples raise it above zero in every model. Accuracy-based evaluation detects none of this.
[NLP-62] SAVER: Selective Auditing of Verbal Evidence for Error Recovery in VLM Change Reasoning
【速读】： 该论文旨在解决视觉语言模型（Vision-Language Models, VLMs）在视觉变化推理任务中表现不佳的问题，尤其针对因模型无法准确表达所观察到的变化信息（即“表达失败”）而导致的错误。其核心问题在于：尽管视觉编码器具备足够的感知能力，但模型输出缺乏明确的语义证据（如物体名称、颜色、空间位置等），从而导致判断失误。解决方案的关键是提出一种轻量级、基于规则的方法——SAVER（Selective Auditing of Verbal Evidence for Error Recovery），该方法通过解析VLM输出中的语义证据，当检测到证据缺失或不一致时，自动触发结构化重提示（reprompting）。实验表明，在三个变化检测基准和四种VLM上，SAVER显著提升了因表达失败引发错误的任务准确率，最高提升达+25.8%（在CLEVR-Change上）。此外，研究发现由大语言模型（LLM）单次调用生成的证据模式可达到与人工调优门控机制相当的效果，且消融实验证明性能提升主要源于证据门控机制本身，而非重提示过程本身。
链接: https://arxiv.org/abs/2608.22857
作者: Youdi Li
机构: Panasonic Connect Co., Ltd. (松下连接有限公司); Tokyo, Japan (东京, 日本)
类目: Computation and Language (cs.CL)
备注: 19 pages, 5 figures
Abstract:Vision-language models (VLMs) frequently fail at visual change reasoning, even when their vision encoders contain sufficient information. We observe that correct VLM outputs tend to contain explicit verbal evidence (object names, colors, spatial locations) that supports the claimed change, while incorrect outputs often lack such evidence. We propose SAVER (Selective Auditing of Verbal Evidence for Error Recovery), a lightweight, rule-based method that parses VLM responses for this evidence and triggers structured reprompting only when evidence is missing or inconsistent. Across three change detection benchmarks and four VLMs, SAVER significantly improves accuracy on tasks where errors stem from the model failing to articulate what it saw (expression failures), with gains up to +25.8% on CLEVR-Change. The evidence patterns can also be generated by an LLM in a single call, matching the hand-tuned gate on CLEVR-Change. Ablation experiments confirm that the evidence gate, not reprompting alone, drives the improvement.
[NLP-63] Your AI On a Dial: Controlling Investment Bias in LLM s with a Single Neuron
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在投资决策中存在系统性、模型特异的投资偏好问题，即不同模型天然倾向于特定投资方向（如买入或卖出），这种偏差可能影响投资建议的客观性与可调控性。其核心解决方案是提出一种“投资偏置旋钮”（investment-bias dial），这是一种在推理阶段对单一神经元进行干预的轻量级方法，能够连续调节模型层面的投资倾向（即整体买入或卖出的先验偏好），而无需修改提示词（prompt）或模型参数。该方法通过引入正负证据配对评估，在五种开源权重的LLM上验证了其有效性：不仅实现了投资立场的单调可控变化，还在响应层面改变了投资决策与论证依据的侧重；在代理式检索场景下，还能引导模型调整信息搜索行为、证据选择策略及最终分析内容。此外，在长上下文情境下，该旋钮保持稳定的立场控制能力，而传统系统指令方式则随上下文增长出现衰减。探索性回测进一步表明，旋钮调整可传导至证券排序与下游投资组合构成。综上，该研究证明了可在推理阶段对大语言模型的总体投资立场实现精准、稳定且可解释的校准。
链接: https://arxiv.org/abs/2608.22852
作者: Sahong Park,Suhwan Park,Hoyoung Lee,Gakyung Kwon,Wonbin Ahn,Jaewon Choi,Alejandro Lopez-Lira,Yoon Kim,Chanyeol Choi,Hyeongwoo Kong,Yongjae Lee
机构: Hankuk University of Foreign Studies (韩国外国语大学); UNIST (韩国科学技术院); LG AI Research (LG人工智能研究院); Hanwha Life (韩华生命); University of Florida (佛罗里达大学); Massachusetts Institute of Technology (麻省理工学院); LinqAlpha (林克阿尔法); Hankuk University of Foreign Studies (韩国外国语大学); UNIST (韩国科学技术院); LinqAlpha (林克阿尔法)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); General Finance (q-fin.GN)
备注:
Abstract:Large language models (LLMs) are increasingly used in investment decision-making, yet prior work shows that they exhibit systematic, model-specific investment preferences. We study whether a model’s overall investment stance can be calibrated to a specified direction and strength. We introduce an investment-bias dial, an inference-time intervention on a single neuron that continuously adjusts a model-level decision prior—its overall tendency toward buying or selling—without targeting specific firms or investment attributes. Using matched positive and negative evidence, we evaluate five open-weight LLMs and find that the dial produces monotonic changes in investment stance without modifying prompts or model parameters. At the response level, the dial shifts both investment decisions and the evidential emphasis of generated rationales under identical inputs. In an agentic retrieval setting, the dial also changes what information the model searches for, which evidence it selects, and which evidence is reflected in its final analysis. In a long-context evaluation, the dial maintains stable stance control as context length increases, whereas a matched system-prompt instruction progressively attenuates. We further show that changes in the dial propagate to security rankings and downstream portfolio composition in an exploratory backtest. Overall, our results show that an LLM’s aggregate investment stance can be calibrated toward a specified target at inference time.
[NLP-64] Industrial-Instruction: An End-to-End Framework for Building Instruction-Tuning and Benchmark Datasets from Industrial Technical Reports
【速读】： 该论文旨在解决工业技术报告中高价值知识难以被标准检索与问答（QA）系统有效索引和推理的问题，其核心挑战在于此类文档结构高度异质（包含密集文本、规格说明、表格等），且缺乏公开的指令微调数据集与基准测试集。解决方案的关键在于提出并构建了Industrial-Instruction数据集体系，包括两个基于真实工业技术报告的开放型问答数据集，以及一个端到端的数据生成流水线。该流水线利用906份公开的松下（Panasonic）技术文档（共7,525页），通过布局感知提取技术获取结构化内容，建立语义检索索引，并基于五类查询-文档关系（无关检索、单/多文档支持、单/多文档答案）合成多项选择题，确保答案有据可依。经筛选后，每个数据集包含约13.6k个带源文档的问答对，并划分出独立的基准测试集。实验表明，使用小规模开源大语言模型（<10B参数）在该数据集上进行微调，可使Set-Match准确率从28.5%提升至42.0%，F1得分从46.6%提升至63.5%。研究还对比了采用开源权重模型Qwen3-30B-A3B-Instruct与闭源API模型Claude-Opus-4.6生成的数据，发现后者虽成本高出两个数量级，但生成语料更清洁、微调性能提升更大，且在MMLU通用知识评估中未出现明显遗忘现象，而前者存在轻微知识遗忘。综上，该工作为从真实工业文档中构建可扩展、可复现的领域基准与训练数据提供了实用路径。
链接: https://arxiv.org/abs/2608.22817
作者: Parsa Bakhtiari,Hassan Bashiri,Alireza Khalilipour,Masoud Nasiripour,Moharram Challenger
机构: Hamedan University of Technology (哈马丹科技大学); University of Antwerp (安特卫普大学); Flanders Make Strategic Research Center (弗拉芒制造战略研究中心)
类目: Computation and Language (cs.CL)
备注: 29 pages, 8 figures, 18 tables, 5 listings. Datasets available at this https URL ; code at this https URL
Abstract:Industrial technical reports contain high-value knowledge for maintenance, troubleshooting, and product engineering, but their heterogeneous structure (dense prose, specifications, tables) makes them difficult to index and reason over with standard retrieval and QA pipelines, and no public instruction-tuning or benchmark datasets are built from such documents. We address this gap with Industrial-Instruction, contributing (i) two open QA datasets built from real industrial technical reports and (ii) the end-to-end pipeline that produces them. Using 906 public Panasonic documents (7,525 pages), we apply layout-aware extraction, build a semantic retrieval index, and synthesize multiple-choice QA grounded in retrieved evidence under five query-document relationships (irrelevant retrieval, single-/multi-document support, single-/multi-document answer). After filtering an initial 23.9k generated samples, each dataset provides approximately 13.6k QA pairs with source documents and a held-out benchmark split. Fine-tuning small open LLMs (under 10B parameters) improves Set-Match Accuracy from 28.5% to 42.0% and F1 from 46.6% to 63.5% on the Panasonic benchmark. We release two parallel versions built by the same pipeline: one generated with the open-weight Qwen3-30B-A3B-Instruct model and one with the closed, API-based Claude-Opus-4.6 model, enabling a direct comparison of open- versus frontier-model data generation. The Claude-Opus-4.6 dataset yields a cleaner raw corpus and larger fine-tuning gains, at roughly two orders of magnitude higher cost. MMLU evaluation shows models trained on the Claude-Opus-4.6 data retain essentially all general knowledge, versus a small but measurable forgetting effect for the Qwen-generated data. Together, these datasets and pipeline offer a practical, reproducible path toward scalable industrial benchmarks and training data from real-world documentation.
[NLP-65] DIAG: Diagnostic Iterative Alignment and Generation for Data-Efficient Mathematical Preference Distillation EMNLP2026
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在数学推理任务中进行迭代偏好优化时面临的信号稀缺问题。随着模型能力的提升，静态题目集与模型当前水平逐渐不匹配，导致生成的推理轨迹要么过于简单、要么过于困难，从而产生大量无效的偏好对（preference pairs），严重制约了训练效率。其解决方案的关键在于提出DIAG（Diagnostic Iterative Alignment and Generation）框架，通过动态调整练习分布以增强信息量丰富的监督信号，并聚焦于学生当前能力边界附近进行训练。DIAG包含两个阶段：一是诊断有效偏好对的产出率，利用经验贝叶斯收缩估计器校准探索-利用权衡并分配主题配额，优先选择高产出概念；二是生成针对性练习，由教师基于学生的失败轨迹合成变体题目。此外，论文从理论上将DIAG解释为一种由教师引导的近似KL正则化重加权过程，旨在使练习分布逼近学生的能力边界，从而最大化有效偏好对的产出。实验表明，DIAG在保持相同训练预算的前提下显著提升了各轮次的偏好对产量，并实现了更强的数学推理性能，验证了其在提炼高信息量偏好监督方面的有效性。
链接: https://arxiv.org/abs/2608.22806
作者: Guhan Chen,Songtao Tian,Bohan Li,Hejin Wang,YeXin Xie,Zixiong Yu
机构: Tsinghua University (清华大学); Kyoto University (京都大学); Nanjing University (南京大学)
类目: Computation and Language (cs.CL)
备注: Accepted by EMNLP 2026 findings
Abstract:Iterative preference optimization is essential for aligning Large Language Models on mathematical reasoning tasks, yet its efficiency is often throttled by signal scarcity: as the model improves, static problem sets become increasingly mismatched to the model’s evolving competence, producing rollouts that are either too easy or too hard and therefore non-informative, which leads to a scarcity of valid preference pairs. We propose DIAG, a Diagnostic Iterative Alignment and Generation framework that adaptively reshapes the practice distribution to increase informative supervision and focus training near the student’s current competence boundary. DIAG consists of two phases: (1) diagnosing valid preference-pair yield to calibrate the exploration-exploitation trade-off and allocate topic quotas via an Empirical Bayes shrinkage estimator, thereby prioritizing high-yield concepts; and (2) generating targeted practice, where a teacher synthesizes variants from the student’s failure traces. We further provide a theoretical view interpreting DIAG as a teacher-mediated approximation to KL-regularized reweighting of the practice distribution toward the student’s competence boundary, where valid preference-pair yield is maximized. Experiments show that DIAG boosts yield across iterations and delivers stronger reasoning performance under an iso-effective training budget, demonstrating that it can distill more informative preference supervision for mathematical reasoning.
[NLP-66] SDoH-Aware Narrative Anchoring Bias in Medical LLM s for Trustworthy Clinical Decision Support
【速读】： 该论文旨在解决医疗大语言模型（Medical Large Language Models, LLMs）在临床问答任务中存在的一种隐蔽风险：尽管模型可能正确掌握医学知识，但在面对同一病例以不同患者叙事视角（persona-based narrative）呈现时，其回答仍可能出现不一致，即“社会决定健康因素（SDoH）感知的叙事锚定偏差”（SDoH aware narrative anchoring bias）。这一偏差可能导致模型对医学上等价的病例产生差异化判断，从而影响临床决策支持系统的可靠性。解决方案的关键在于构建并使用一个名为NarrativeShield SDoH MedQA的反事实医学问答数据集，该数据集将原始宽格式数据重构为按病例分组、包含多种患者叙事角色的行结构，从而系统性地评估模型在不同叙事语境下的响应稳定性。研究通过对比Qwen2.5系列三个不同规模（1.5B、3B、7B）的开源指令微调模型，在300个临床案例下生成共8,100条响应，从个体叙事准确率、反事实一致性、正确一致性及叙事敏感性误差等多个维度进行评估。结果显示，Qwen2.5 7B在平均准确率（56.33%）和正确一致性（40.33%）方面表现最优，且显著优于3B版本（经配对McNemar精确检验），但即便如此，叙事敏感性误差仍高达31.67%，表明模型在跨叙事情境下的稳定性仍存在显著缺陷。因此，论文强调，可信的临床决策支持系统不仅应关注平均正确性，更需评估其在医学等价但叙事不同的场景下的响应稳定性。
链接: https://arxiv.org/abs/2608.22802
作者: Ahnaf Atef Choudhury,Ramkrishna Saha
机构: George Mason University (乔治梅森大学); The University of Texas at Dallas (德克萨斯大学达拉斯分校)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted for publication at 10th International Artificial Intelligence and Data Processing Symposium (IDAP’26)
Abstract:Medical large language models are often judged by how many clinical questions they answer correctly. That view is useful, but it misses a practical risk. A model may know the right answer and still change its response when the same case is written in a different patient voice. This paper evaluates that risk as SDoH aware narrative anchoring bias. We use NarrativeShield SDoH MedQA, a counterfactual medical question answering dataset in which each case appears in persona based narratives while the answer key remains fixed. The dataset is reshaped from wide format into case grouped persona rows. We evaluate three open source instruction tuned LLMs from the Qwen2.5 family: 1.5B, 3B, and 7B. The final experiment uses 300 clinical cases and produces 8,100 model responses across three prompting conditions. We report persona level accuracy, counterfactual consistency, correct consistency, and narrative sensitivity error. Qwen2.5 7B achieves the best accuracy at 56.33 percent and the best correct consistency at 40.33 percent. Paired McNemar exact tests show significant accuracy gains for 7B over 3B in all prompt settings. Even so, narrative sensitivity remains, with the lowest error still at 31.67 percent. These results suggest that trustworthy clinical decision support should be evaluated by both average correctness and stability across medically equivalent patient narratives.
[NLP-67] RACE: A Self-Evolving Skill Bank for Consistent Limit-Aware LLM Agents
【速读】： 该论文旨在解决大语言模型（LLM）代理在面向用户的产品部署中可靠性不足的核心问题，即模型虽具备解决任务的潜力（如Pass@3），但在重复执行中缺乏行为一致性与对能力边界的认知（limit-awareness）。其关键解决方案是提出一种无需修改模型权重的技能演化框架TRACE（TRAjectory-Contrastive Evolution），通过构建可检索、模块化的“技能库（Skill Bank）”来组织行为知识，每项技能封装一组独立的工具使用规则与行为准则。TRACE采用智能体自演化循环：在每轮评估后，根据轨迹中调用的技能进行分组，并通过对比成功与失败行为对技能进行迭代优化；优化后的技能库指导后续演化，同时在部署阶段由执行器（Actor）基于状态条件动态协调技能调用。实验表明，该方法在GPT-5.5上将一致性指标Pass^3提升34.6个百分点（从59.9%至94.5%），并将潜在性能与可靠性能间的差距缩小至仅4.0点，在官方隐藏测试集上以GPT-5.6-Sol实现70%的Pass^3得分，相较基线提升40%相对性能，充分验证了其将高潜力模型转化为稳定可靠系统的能力。
链接: https://arxiv.org/abs/2608.22793
作者: Wenhao Wu,Menghao Zhang,Xin Wang,Zhi Wang,Kun Shao,Jian Luan
机构: Xiaomi Inc.(小米公司); Nanjing University (南京大学); Beijing University of Posts and Telecommunications (北京邮电大学); Tsinghua University (清华大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 9 pages, 5 figures, 2 tables
Abstract:Reliable deployment of LLM agents in user-facing products depends not on raw task-solving ability but on consistency and limit-awareness: behaving the same way across repeated trials, and recognizing when a request cannot, or cannot yet, be safely fulfilled. CAR-bench exposes this reliability gap in the domain of in-car assistants: an LLM-simulated user issues incomplete or ambiguous requests, requiring the agent to resolve uncertainty through multi-turn dialogue and tool use while strictly adhering to domain policies. Even frontier models show a substantial gap between what they can solve at least once (Pass@3) and what they solve consistently across trials (Pass^k). We bridge this gap with TRACE (TRAjectory-Contrastive Evolution), which iteratively improves a skill-based agent’s behavioral knowledge without modifying model weights. This knowledge is organized as a Skill Bank of modular, retrievable skills, each encoding a self-contained set of tool-use rules and behavioral guidelines. TRACE evolves this bank through an agentic self-evolution loop: after each evaluation round, it groups trajectories by the skills invoked and refines each skill by contrasting successful and failed behaviors. The updated bank then guides subsequent rounds, while during deployment the Actor performs state-conditioned skill orchestration at every turn. On GPT-5.5, TRACE improves consistency (Pass^3) by 34.6 points, from 59.9% to 94.5%, while shrinking the gap between potential and reliable performance to just 4.0 points. On the official hidden set, TRACE achieved first place using GPT-5.6-Sol, attaining a Pass^3 score of 70%-a 40% relative improvement over the baseline. These results show that TRACE converts high model potential into stable, consistent performance gain. Project homepage: this https URL.
[NLP-68] SPOC-SQL: Stage-wise Preference Optimization for Controllable Text-to-SQL
【速读】： 该论文旨在解决文本到SQL（Text-to-SQL）任务中现有方法因将查询生成视为单步序列生成而导致的结构化推理不足问题，具体表现为模型在关键决策点缺乏针对性反馈，且无法有效干预和控制中间生成过程。其核心解决方案是提出SPOC-SQL框架，该框架基于标准SQL执行逻辑，将Text-to-SQL任务分解为四个顺序子任务，并针对每个阶段设计特定的优化策略，以促进模型学习关键决策。关键创新在于在各SQL生成阶段的关键决策点引入细粒度偏好优化（fine-grained preference optimisation），从而增强查询构建过程中的结构化决策能力；同时采用结构化分解策略，通过显式的中间表示实现分阶段干预与修正，显著提升了生成结果的可控性与可靠性。实验表明，融入阶段感知的人类知识可稳定提升性能，验证了阶段感知可控生成的有效性。
链接: https://arxiv.org/abs/2608.22772
作者: Yingnan Chen,Chun Ding,Tianshi Xu,Xu Yang,Si Wu
机构: South China University of Technology (华南理工大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Text-to-SQL aims to translate natural language questions into executable SQL queries over relational databases, requiring multi-stage structured reasoning over database schemas and query constraints. However, existing methods treat this task as single-step generation, where models optimize entire SQL sequences without targeted feedback at key decision points and lack support for interacting with and controlling the intermediate generation process. To address this issue, we propose SPOC-SQL, which decomposes Text-to-SQL into four sequential subtasks following standard SQL execution logic and designs stage-specific optimization strategies for the model to learn key decisions. Specifically, we propose the implementation of fine-grained preference optimisation at key decision points across SQL stages, with the objective of enhancing structured decision-making during query construction. Furthermore, a structured decomposition strategy is designed, facilitating stage-wise intervention and correction through explicit intermediate representations. This results in more controllable and reliable SQL generation. Experiments demonstrate that incorporating stage-wise human knowledge consistently improves performance, validating the effectiveness of stage perception controllable generation.
[NLP-69] DelistBench: Evaluating Search-Enabled LLM s for Auditable Corporate-Event Database Completion
【速读】： 该论文旨在解决金融领域中第三方数据供应商数据库内企业事件记录缺失、过时及分类错误的问题，提出了一种名为“Search-to-Record”的数据库验证任务，即利用具备搜索能力的大语言模型（Large Language Model, LLM）从公开信息源重构特定证券池在历史截断时点的机构定义事件记录。其核心解决方案包括：构建包含1,200条证券级退市公告的基准测试集DelistBench，评估五种模型在封闭式与联网式两种条件下的表现。研究发现，联网访问使事件公告日期准确率在7天内提升34.0至48.0个百分点，事件状态准确率提升约2.8至21.7个百分点，最优系统在7天内实现81.5%的整体联合准确率；同时，经济型网络系统以仅4.5%-6.6%的API成本达到75.9%-78.3%的综合准确率，显著降低部署成本。风险导向的优先级筛选可识别低误差子集，但最高覆盖率策略仍需人工审查27.3%的测试样本。研究揭示了网络检索是时间精度提升的主要来源，并表明低成本系统可逼近最优系统的性能。综合而言，该工作为实际部署提供了明确指导：根据本地事件发生频率与市场结构校准优先级策略，确保正向事件召回率，将正向及模糊案例定向引导至针对性人工审核流程。
链接: https://arxiv.org/abs/2608.22770
作者: Xuan Yao,Li Shuping,Dai Yang,Zhou Yi,Ke-Wei Huang
机构: Asian Institute of Digital Finance, National University of Singapore(新加坡国立大学数字金融研究院)
类目: Computation and Language (cs.CL)
备注:
Abstract:Financial institutions need an independent way to detect missing, stale, and misclassified corporate-event records in vendor databases. We introduce Search-to-Record, a database-assurance task in which search-enabled large language models reconstruct institution-defined event records from public sources for a known security universe and historical cutoff, and DelistBench, a 1,200-record benchmark for security-level delisting announcements. We evaluate five models in paired closed-book and web-enabled conditions. Web access raises announcement-date accuracy within seven days by 34.0 to 48.0 percentage points and event-status accuracy by approximately 2.8 to 21.7 points; the best system achieves 81.5% overall joint accuracy within seven days. Economy web systems achieve 75.9-78.3% overall joint accuracy within seven days at 4.5-6.6% of the API cost of the most expensive web system. Risk-based triage identifies low-error subsets, although the highest-coverage operating point still sends 27.3% of the balanced test set to review. The evaluation identifies web retrieval as the main source of timing gains and shows that low-cost systems can approach the best system’s accuracy. Together, Search-to-Record, DelistBench, and the evaluation provide concrete deployment guidance: calibrate triage to local event prevalence and market mix, preserve positive-event recall, and route positive and ambiguous cases to targeted review.
[NLP-70] Dont Repeat Yourself: Stopping Verbatim Loops at Sampling Time
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在自回归生成过程中出现的“原文循环”（verbatim looping）问题，即模型重复生成已出现在上下文中的文本片段，导致输出冗余且缺乏多样性。现有防御机制如重复惩罚、频率惩罚及n-gram屏蔽等，通常基于词元（token）的重复性进行干预，未能有效捕捉循环的序列结构特征，常需较强惩罚力度以抑制循环，却同时损害生成文本的格式规范性和语言流畅性。本文提出一种名为“不要重复自己”（Don’t Repeat Yourself, DRY）的采样时对数调整方法，其核心在于仅当候选词元的生成会将当前后缀扩展为与上下文早期某段完全一致的延续时，才施加惩罚，从而精准识别并阻止序列级循环。该方法引入“序列断点”（sequence breakers）以保护对话模板和格式化词元，确保生成结构完整性。在1.5B至120B参数规模的多种模型、九类提示模板及600对人类评估实验中，DRY使后缀扩展率降低47%，同时提升词汇多样性；安慰剂对照实验表明，其效果源于对后缀匹配的精确检测，而非其他干扰因素。在经过AWQ量化处理的70B和120B模型上，DRY可将循环率降低约一半，同时保持MT-Bench、MMLU和GSM8K等基准任务性能稳定，而传统方法则出现显著性能下降。该方法已被广泛集成于主流开源大模型推理框架（如ExLlamaV2、text-generation-webui），体现了其在实际文本生成应用中的重要价值。
链接: https://arxiv.org/abs/2608.22761
作者: Philipp Emanuel Weidmann,Allen Roush,Judah Goldfeder,Sanjay Basu,Ravid Shwartz-Ziv
机构: Independent Researcher(独立研究员); Thoughtworks(思特沃克); Columbia University (哥伦比亚大学); Oracle (甲骨文); New York University (纽约大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Large Language Models generate text autoregressively, but open-ended generation is prone to verbatim looping, in which models repeat spans already present in context. Standard defenses such as repetition, presence, and frequency penalties and n-gram blocking act on token recurrence rather than the sequential structure of a loop, and often suppress looping only at strengths that also degrade formatting or fluency. We propose Don’t Repeat Yourself (DRY), a sampling-time logit adjustment that penalizes a candidate token only when generating it would extend the current suffix into an exact continuation of a span seen earlier in the context. Sequence breakers protect chat templates and formatting tokens. Across models from 1.5B to 120B parameters, nine prompt families, and a 600-pair human study, DRY reduces suffix-extension rate by 47% while improving lexical diversity. An intervention-matched placebo produces no comparable reduction, identifying suffix matching as the operative mechanism. On AWQ-quantized 70B and 120B models, DRY reduces loop rate by roughly half while preserving MT-Bench, MMLU, and GSM8K performance, whereas standard alternatives lose measurable ground. DRY has been adopted by popular open-source LLM inference frameworks including this http URL, ExLlamaV2, and text-generation-webui, highlighting its practical impact on text generation.
[NLP-71] XTC: Head-Aware Sampling by Excluding Top Choices
【速读】： 该论文旨在解决自回归语言模型在开放式生成任务中因解码策略导致的多样性不足与重复性过高问题，尤其针对一种常见但被现有方法忽视的场景：多个后续词项均合理，但概率质量仍过度集中于最通用选项。其核心解决方案是提出XTC（Exclude Top Choices）——一种轻量级、头感知的解码算子，通过设定绝对合理性阈值τ识别符合条件的候选词，当至少有两个候选时，以概率ρ剔除占主导地位的高概率选项，仅保留最弱但合理的替代选项并重新归一化分布。该方法在Gemma 3 27B Q4、Gemma 3 12B Q6、DeepSeek R1 14B Q6及Llama 3.3 70B Q4等多模型上验证，显著提升了多样性-重复性权衡前沿表现，在创意生成任务中Distinct-2提升11–15%，重复三元组减少27–47%；结合温度缩放后，Distinct-2提升达38%，重复三元组减少71%。盲测评估显示，XTC在150名资深评审中获得62.3%的创造性偏好（p < 10⁻⁴），且未牺牲流畅性；同时在IFEval评测中保持提示级严格准确率仅下降1.7个百分点，优于匹配相同多样性水平的温度调节方案。XTC具有与温度和重复惩罚项的可加性、对量化级别与模型家族的鲁棒性，并在十二类提示类型中保持一致性，已被ExLlamaV2和text-generation-webui等主流框架采纳。
链接: https://arxiv.org/abs/2608.22758
作者: Philipp Emanuel Weidmann,Allen Roush,Judah Goldfeder,Sanjay Basu,Ravid Shwartz-Ziv
机构: Independent Researcher; Thoughtworks; Columbia University (哥伦比亚大学); Oracle (甲骨文公司); New York University (纽约大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Standard decoding rules for autoregressive language models promote diversity by rescaling the full next-token distribution or truncating its low-probability tail. These strategies overlook a common regime of open-ended generation in which several continuations are plausible but too much probability mass remains concentrated on the most generic choice. We introduce XTC (Exclude Top Choices), a lightweight head-aware decoding operator that targets this regime directly. XTC identifies tokens whose probabilities exceed an absolute plausibility threshold \tau : when at least two qualify, it removes the dominant eligible choices with probability \rho and retains only the weakest plausible alternative before renormalization. Across 60 experiments on Gemma 3 27B Q4, Gemma 3 12B Q6, and DeepSeek R1 14B Q6, with scaling validation on Llama 3.3 70B Q4, XTC improves the diversity-repetition Pareto frontier. On creative generation, Distinct-2 increases by 11–15% and repeat trigrams decrease by 27–47% across the four models. Combined with temperature scaling, gains reach 38% in Distinct-2 and 71% in repeat-trigram reduction over baseline. A blinded Amazon Mechanical Turk study with 150 Master raters yields a 62.3% creativity preference for XTC ( p10^-4 ) without reduced fluency, while a GPT-4o control judge reproduces the Anthropic-judge direction on every measure. On IFEval with Llama 3.3 70B Q4, XTC preserves prompt-level strict accuracy within 1.7 percentage points of baseline while recovering most of the diversity gain; a temperature setting matched on Distinct-2 reduces IFEval by 8.8 points. The effect is additive with temperature and repetition penalties, robust across quantization levels and model families, and consistent across twelve prompt genres. XTC has been adopted by this http URL, ExLlamaV2, and text-generation-webui.
[NLP-72] Beyond Factual Knowledge: Benchmarking and Learning Step-Level Procedural Rule Reasoning in Large Language Models EMNLP2026
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在大规模场景下难以可靠理解与应用外部提供的程序性规则的问题。现有模型在面对复杂、动态的规则集合时，往往因无法有效检索和更新相关规则而导致推理性能下降。其解决方案的关键在于提出一种名为DynaRule的端到端框架，该框架将规则注入到键值缓存（KV cache）中，并将规则检索转化为可学习的、分步进行的内部过程。DynaRule通过引入堆叠步骤级注意力训练（Stacked Step-Level Attention Training）与特殊搜索标记（search token），实现了推理过程中对规则的动态重关注与更新，使模型能够在每一步动态选择最相关的规则并替换过时信息，从而支持更稳定的多步推理。实验结果表明，在大规模规则池（如10,000条规则）下，DynaRule相较于现有方法显著提升平均问答准确率达19个百分点，并实现超过85%的Recall@1，验证了其在复杂规则推理任务中的优越性。
链接: https://arxiv.org/abs/2608.22753
作者: Bohan Yu,Pengfei Cao,Chen Han,Chenxi Zhou,Zhiheng Zhang,Zhiyang Xie,Wenhao Teng,Xiangwen Liao,Jun Zhao,Kang Liu
机构: University of Chinese Academy of Sciences (中国科学院大学); Institute of Automation, Chinese Academy of Sciences (中国科学院自动化研究所); School of Advanced Interdisciplinary Sciences (高级交叉科学学院); The Key Laboratory of Cognition and Decision Intelligence for Complex Systems (复杂系统认知与决策智能重点实验室); School of Artificial Intelligence (人工智能学院); Academy of Mathematics and Systems Science (数学与系统科学研究院); Fujian Provincial Cancer Hospital (福建省肿瘤医院); College of Computer and Data Science, Fuzhou University (福州大学计算机与数据科学学院)
类目: Computation and Language (cs.CL)
备注: Accepted by EMNLP 2026 Findings
Abstract:Large language models (LLMs) excel at text understanding and generation, yet still struggle to reliably understand and apply externally provided procedural rules at scale. To evaluate this capability, we introduce RuleWorld, a large-scale benchmark that reformulates rules as globally reusable abstract units rather than instance-specific facts. In RuleWorld, several scenarios, including single-rule, parallel multi-rule, and multi-hop reasoning, are settled for comprehensive evaluation. We further propose DynaRule, an end-to-end framework that injects the given rules into the KV cache and turns retrieval into an internal, learnable, step-wise process. Specifically, DynaRule employs Stacked Step-Level Attention Training with a special search token to enable dynamic rule re-attention and updating during inference. In this way, the model can re-attend to the most relevant rules at each step, dynamically replacing outdated ones to support more stable multi-step reasoning. Experiments on RuleWorld show that existing LLMs face challenges under large rule pools, while DynaRule improves average QA accuracy by up to 19 points and achieves over 85% Recall@1 at 10K rules, outperforming strong baselines by large margins. We make our code and dataset available here: this https URL.
[NLP-73] DiaRelay: Relaying Dialogue Context with a Constant-Size Memory for Emotion Recognition in Conversation
【速读】： 该论文旨在解决对话情感识别（ERC）中长期依赖关系建模困难的问题，即如何有效捕捉分散在远距离对话轮次中的细微情感线索。现有方法通常采用固定上下文窗口处理对话历史，导致短窗口丢失长程信息，而扩大窗口则引发重复编码、计算与内存开销增加及无关上下文引入等问题。此外，主流的参数高效微调方法（如LoRA）仅在特征空间引入固定的低秩变换，缺乏对对话级状态的显式维护，且其变换未根据动态演化的对话上下文进行条件化调整。针对上述挑战，本文提出一种轻量级适配器DiaRelay，通过基于LoRA架构引入两个紧密协同的核心组件——“选择性中继记忆转移”（Selective Relay Memory Transition）和“双轴中继记忆读取”（Dual-axis Relay Memory Read），实现大模型对对话级记忆的显式保持。其中，前者将有用的历史证据逐步聚合至有限大小的中继记忆中并跨轮次传递，使早期情感线索在脱离局部上下文后仍能影响后续预测；后者利用传播的记忆动态调制低秩特征变换，实现无需测试时梯度更新的上下文感知表示适应。实验表明，DiaRelay在MELD数据集上达到当前最优的加权F1和准确率，在IEMOCAP上也取得具有竞争力的结果，仅引入710万额外可训练参数，验证了其在提升大模型情感理解能力方面的有效性与泛化性。
链接: https://arxiv.org/abs/2608.22745
作者: Zihao Zhou,Bin Yang,Jinghui Qin,Kebing Jin
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Emotion Recognition in Conversation (ERC) requires models to identify subtle emotional cues that are often distributed across distant dialogue turns. Existing methods typically incorporate dialogue history through a fixed context window. However, short windows discard potentially useful long-range evidence, while enlarging the window repeatedly re-encodes overlapping utterances, increases computational and memory costs, and may introduce irrelevant context. Moreover, commonly used parameter-efficient adaptation methods, such as LoRA, mainly introduce fixed low-rank transformations in the feature space and do not explicitly maintain a dialogue-level state or condition their transformations on the evolving conversational context. To address these limitations, we propose a lightweight adapter, DiaRelay, to enable LLMs to explicitly maintain a dialogue-level memory for accurate ERC. Based on LoRA, DiaRelay introduces two extra tightly collaborative components, Selective Relay Memory Transition and Dual-axis Relay Memory Read. Selective Relay Memory Transition progressively aggregates useful historical evidence into a bounded relay memory and propagates it across successive utterance predictions. This allows earlier emotional cues to influence later predictions after they leave the local context window, without re-encoding the complete dialogue history or expanding the backbone context length. Dual-axis Relay Memory Read uses the propagated memory to dynamically modulate low-rank feature transformations, enabling context-dependent representation adaptation without test-time gradient updates. Extensive experiments show that DiaRelay can achieve SOTA weighted F1 and accuracy on MELD while obtaining competitive results on IEMOCAP with only an extra 7.1M trainable parameters, indicating the effectiveness and generalizability of our DiaRelay in enhancing LLM-based emotional understanding.
[NLP-74] aching LLM s How ICU Physicians Approach Clinical Reasoning Reasoning Through OMOP-Aligned Retrieval Improves Reasoning Across Clinical Domains
【速读】： 该论文旨在解决重症监护室（ICU）中临床决策因数据密集且动态变化而难以有效识别与诊疗相关的关键患者信息的问题。现有大语言模型（LLM）的应用多聚焦于表层信息检索或事实回忆，未能充分模拟临床医生在决策过程中所依赖的归纳与演绎推理能力。为此，研究提出一种“临床医生在环”（clinician-in-the-loop）框架，构建了名为ICU-REACT的推理数据集，旨在训练模型进行信息检索与情境感知的临床推理。其解决方案的关键在于通过19名临床专家提供的真实世界重症医学推理过程进行监督训练，使模型掌握深层次的临床逻辑推导能力。基于该数据集，研究微调了参数量从8B到70B不等、涵盖三种模型架构的Clin-REACT模型，在五个临床推理基准测试中均显著优于基线模型及开源通用和医疗领域大模型，且性能提升延伸至脚本一致性测试以及下游诊断与治疗任务。结果表明，重症医学领域的专家推理监督可有效提升模型在更广泛临床场景下的推理泛化能力，但其实际临床应用仍需前瞻性验证。
链接: https://arxiv.org/abs/2608.22622
作者: Miguel Contreras,Scott Siegel,Subhash Nerella,Jessica Sena,Jiaqing Zhang,Heng Sun,Hruday Tej Akkaladevi,Peiyu Lu,Jordan Rosen,Sumit Kapoor,Sasank Desaraju,Grace R. Thompson,Jacob Purcell,Michael Petrauskis,Philip KW. Hong,Meghan Brennan,Sarah Chrabaszcz,Tierra Smith,Ronnie Ren,Michel S. Kabbash,Ceyhun Haziroglu,Rushi Patel,Gabriel Gomez,Charlotte Chaiklin,Randy Leung,Kenneth N. John,Whitman Wiggins,Philip Kayser,Vincent Bird,Maria Bruzzone,Tyler J. Loftus,Azra Bihorac,Parisa Rashidi
机构: University of Florida (佛罗里达大学); Intelligent Clinical Care Center (IC3) (智能临床护理中心); Google (谷歌)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Clinical decision-making relies on identifying relevant patient information to guide diagnosis and treatment, a challenge that is especially difficult in the data-dense and rapidly changing intensive care unit (ICU). Large language models (LLMs) could support this task. However, existing applications and datasets mostly emphasize surface-level retrieval or factual recall rather than the inductive and deductive reasoning clinicians practice to select and reason over decision-relevant evidence. We hypothesized that training LLMs on expert ICU reasoning could yield clinical reasoning skills that generalize beyond critical care. Here we introduce ICU-REACT, a reasoning dataset developed with 19 clinicians through a clinician-in-the-loop framework to teach LLMs to perform information retrieval and context-aware clinical reasoning in the ICU. Using ICU-REACT, we fine-tuned Clin-REACT models spanning 8B-70B parameters and three model families. Across five clinical reasoning benchmarks, Clin-REACT consistently outperformed its backbone models and open-source general-purpose and medical LLMs. Gains extended to different tasks including script concordance tests, and downstream diagnosis and treatment tasks. These findings suggest that expert reasoning supervision in critical care can improve broader clinical reasoning, although prospective evaluation is needed before real-world clinical use.
[NLP-75] Vision-Language Models for Occupational Physical Exposure Assessment: Estimating External Hand Forces in Manual Material Handling Tasks from RGB Video
【速读】： 该论文旨在解决在职业体力暴露评估与损伤风险分析中，如何在不依赖传感器的情况下连续、准确估计动态、三轴、双侧外部手部作用力（external hand forces）的问题。传统方法通常需要在物体或人体上安装传感设备，限制了其在真实工作场景中的可扩展性。本文提出了一种基于视觉-语言模型（Vision-Language Model, VLM）的新型数据处理流程，其关键在于融合任务特定的文本提示（textual cues）、视觉特征表示以及已知的物体质量信息，通过文本引导的目标区域定位（text-guided localization）、预训练视觉变换器（vision-transformer）特征提取及基于变换器的时序回归机制，实现从单目或多目RGB视频中推断出连续的手力信号。实验结果表明，该方法在不同相机视角和感兴趣区域（ROI）策略下均表现出良好的性能，尤其在单相机条件下引入物体作为第二兴趣区域显著提升了估计精度；多摄像机采集对峰值力估计具有明显优势，尤其是在垂直方向。研究证实了仅凭已知载荷质量和无侵入式视频即可实现高精度手力估计的可行性，为构建更高效、可扩展的职业物理暴露与风险评估体系提供了技术支持。
链接: https://arxiv.org/abs/2608.22586
作者: Mohammad Sadra Rajabi,Aanuoluwapo Ojelade,Sunwook Kim,Maury A. Nussbaum
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:External hand forces are important inputs to biomechanical analyses of occupational physical exposure and injury risk, yet continuous force measurements during manual material handling (MMH) typically requires instrumented objects or specialized sensing. We evaluated a vision-language model (VLM)-based pipeline that combines task-specific textual cues, visual representations, and known box mass to estimate dynamic, triaxial, bilateral external hand forces from RGB video. Thirty-five healthy young adults performed five MMH tasks involving lifting, carrying, pushing, and pulling with box masses of 6, 9, and 12 kg. The pipeline used text-guided localization of participant and handled-object regions of interest (ROIs), pretrained vision-transformer feature extraction, and transformer-based temporal regression. Performance was evaluated using leave-one-subject-out validation across seven camera-view conditions (three single-view and four multi-view conditions) and four ROI strategies. Overall, root mean square error was ~4.7-5.6 N for the horizontal and mediolateral force components and ~10.6-11.0 N for the vertical component. Including the handled object as a second ROI generally improved force estimation, with some of the largest benefits under single-camera conditions, whereas pixel-level segmentation provided little additional improvement. Multi-camera capture provided the clearest benefit for peak-force estimation, particularly for the vertical component, whereas differences in overall frame-level error among camera configurations were comparatively modest. These findings demonstrate the feasibility of estimating continuous, bilateral, directional hand-force estimates from RGB video and known load mass without requiring sensors on the worker or handled objects as model inputs, supporting the development of more scalable occupational physical exposure and risk assessments.
[NLP-76] From Diagnosis to Redesign: Using Quantitative Ethnography to Improve Multi-Agent LLM Reasoning
【速读】： 该论文旨在解决多智能体大语言模型（Multi-agent Large Language Model, Multi-agent LLM）系统在任务分解与协作过程中，尽管存在多个具有专业化功能的智能体，却难以保证推理连贯性或输出与任务目标一致的问题。其核心挑战在于：如何有效诊断并优化智能体间的交互模式以提升系统的整体推理质量。解决方案的关键在于提出一种定量民族志（Quantitative Ethnography, QE）方法，通过分析智能体交互所产生的话语（discourse），利用认知网络分析（Epistemic Network Analysis, ENA）对多智能体辩论系统的交互结构进行建模。研究发现，正确评分决策的对话特征表现为基于评分量规（rubric）的论证、共识及内容拓展，而错误决策则呈现为脱离量规标准的冗长命题-质疑-回应循环。基于此诊断结果，研究人员重构了各智能体的提示词（prompt），显著提升了精确评分准确率（从27.78%提升至40.28%），并使错误对话的互动模式向正确模式趋同，实现结果与交互模式的双重优化。因此，该研究证明，QE能够构建“诊断—重设计”闭环，通过追踪智能体交互模式与系统性能之间的关联，指导提示词迭代，并验证改进效果，从而推动生成式AI推理能力的可解释性与可控性提升。
链接: https://arxiv.org/abs/2608.22566
作者: Vedant Khatri,Anthony Cusimano,Zachari Swiecki,Zhen Xu,Xiner Liu,Renzhe Yu
机构: University of California, Irvine, USA(加州大学欧文分校); University of California, Los Angeles, USA(加州大学洛杉矶分校); Monash University, Australia(蒙纳士大学); Columbia University, USA(哥伦比亚大学); University of Pennsylvania, USA(宾夕法尼亚大学)
类目: Computation and Language (cs.CL)
备注: Accepted at ICQE 2026 (to appear in Springer CCIS)
Abstract:Multi-agent large language model (LLM) systems are designed to improve reasoning by decomposing tasks across multiple agents with specialized functions, but the presence of multiple agents does not inherently guarantee coherent reasoning or outputs that align with task objectives. This paper introduces a quantitative ethnographic (QE) approach for diagnosing and redesigning multi-agent LLM systems based on the discourse produced through agent interactions. We test this approach using automated essay scoring as an example context, applying Epistemic Network Analysis (ENA) to model a five-agent multi-agent debate system and examine differences between debates that produced correct versus incorrect scoring decisions. Results show that, in the initial system, correct scoring decisions were characterized by rubric-grounded justification, agreement, and elaboration. Incorrect scoring decisions, in contrast, were characterized by extended proposition-challenge-response exchanges that were less consistently tied to rubric criteria. We then used the findings to revise the agents’ prompts. The revised system improved exact scoring accuracy from 27.78% to 40.28% and shifted the discourse of incorrect debates toward the rubric-grounded pattern of correct ones, making the two nearly indistinguishable. Based on these results, we argue that QE can support a diagnostic-to-redesign loop for AI reasoning by tracing how patterns of agent interaction relate to system performance, informing prompt redesign, and evaluating whether those redesigns change both outcomes and interaction patterns.
[NLP-77] BLADE: Bilevel Low-rank Augmented-Lagrangian Erasure for LLM Unlearning EMNLP2026
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）在遗忘学习（unlearning）过程中面临的鲁棒性不足问题，具体表现为：无限增长的遗忘损失会破坏模型的语义连贯性、固定权重的平衡机制无法适应保留难度随训练过程动态变化的问题，以及现有方法在模型规模扩展或多次重复遗忘操作下性能显著下降。其解决方案的关键在于提出一种约束型双层优化框架——BLADE，通过三个核心机制实现对优化过程的平滑与可预测控制：一是采用截断熵遗忘损失（clamped-entropy forget loss），当某个词元达到足够不确定性时其梯度精确为零，防止过度遗忘；二是引入非对称增强拉格朗日方法（asymmetric augmented Lagrangian），在任何保留内容被违反后永久提升保护强度，确保关键知识不被破坏；三是将双层结构限制在低秩适配器（LoRA adapters）范围内，在每次遗忘步骤前修复保留内容的损伤，从而实现高效且稳定的参数更新。该框架在三大基准测试集上均显著优于现有最强基线，平均综合得分提升6%（TOFU）、9%（MUSE Books）和7%（KnowUndo），并在4倍模型规模扩展及4次连续遗忘操作下保持稳定，而最优对比方法则完全崩溃。
链接: https://arxiv.org/abs/2608.22557
作者: Md Toufikuzzaman,Ahmad Mousavi,Dongwon Lee
机构: The Pennsylvania State University (宾夕法尼亚州立大学)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: To Appear in EMNLP 2026
Abstract:Existing LLM unlearning methods struggle with robustness: unbounded forget losses degrade model coherence, fixed-weight balancing cannot adapt as retain difficulty shifts mid-training, and methods that work on one benchmark falter under scaling or repeated application. We propose BLADE, a constrained bilevel framework whose three mechanisms give smooth, predictable control over the optimization landscape: a clamped-entropy forget loss whose gradient is exactly zero once a token reaches sufficient uncertainty; an asymmetric augmented Lagrangian that permanently ratchets retain protection after any violation; and a bilevel structure confined to LoRA adapters that repairs retain damage before each forgetting step. BLADE dominates across three benchmark families, improving average composite scores over the strongest baselines by 6 % on TOFU, 9 % on MUSE Books, and 7 % on KnowUndo, and it remains stable under 4\times scaling and 4 sequential unlearning steps on MUSE News where the best competing method collapses entirely.
[NLP-78] RACE: Temporal Retrieval with Anchored and Convergent Evidence for Long-Horizon Video Understanding EMNLP2026
【速读】： 该论文旨在解决长视频问答中因模型仅依赖部分视觉帧而产生不完整证据支撑的问题，即现有评估方法虽能衡量最终答案正确性或预测的证据区间，却忽视了模型在作答前实际解码的帧是否覆盖所有必要事件。其核心解决方案是提出一种无需训练的推理代理TRACE，该方法通过在原始视觉片段中逐轮构建证据包（evidence bundle），并在证据包增长且对同一片段进行最终验证时答案稳定后停止，确保答案基于充分且完整的观察。该方法在自洽性审计下实现了50.7%的准确率，且每个必要证据区间至少包含两帧，同时仅需98.7帧/题，显著优于均匀解码128帧时的40.2%准确率（提升超10个百分点），并以0.39倍的帧成本接近均匀解码256帧的性能（相差仅2.6个百分点），在审计中达到最高准确率63.5%。此外，TRACE在Video-MME、LVBench和LongVideoBench等基准上也保持竞争力。
链接: https://arxiv.org/abs/2608.22516
作者: Pengyiang Liu,Junbo Niu,Xiaoyang Hu,Zhongyue Shi,Zitian Wang,Linjiang Huang,Si Liu
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
备注: Accepted to EMNLP 2026 Main Conference. 19 pages, 5 figures, 6 tables. Project page: this https URL
Abstract:A long-video answer is evidence-supported only when the frames decoded from the video cover every event the answer depends on. Existing evaluations score final-answer correctness or predicted evidence intervals, but the frames a method decodes before answering are rarely audited, so correct answers can still rest on incomplete observation. We introduce VES-Bench, a 600-question benchmark of Temporal Ordering and Event Counting items over 348 public long videos. Each item carries a jointly necessary set of evidence intervals, letting us audit at three strictness levels whether a method’s decoded frames cover every one of them. We also propose TRACE, a training-free agent that grounds answers in raw visual clips, builds an evidence bundle round by round, and stops only when the answer stabilises as the bundle grows and a final pass over the same clips returns the same answer. Under a same-backbone audit, TRACE answers 50.7% of questions correctly with at least two decoded frames inside every evidence interval, at 98.7 frames per question: over 10 points above uniform decoding at 128 frames (40.2%), and within 2.6 points of uniform decoding at 256 frames at 0.39x its frame cost, while reaching the highest answer accuracy in the audit (63.5%). TRACE also stays competitive on Video-MME (86.1), LVBench (75.6), and LongVideoBench (75.1).
[NLP-79] Kernel Token Contradiction: a Fast and Principled Approach for LLM Claim Uncertainty Quantification
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）输出中事实性不可靠的问题，提出在白盒条件下对每个声明（claim）进行不确定性量化（Claim-level Uncertainty Quantification, UQ）的方法。其核心解决方案是引入一种轻量级方法——核标记矛盾（Kernel Token Contradiction, KTC），通过将生成过程中涉及的候选标记表示为一个半正定核，融合了模型的条件分布与标记矛盾得分，进而利用冯·诺依曼熵（Von Neumann entropy）对核的不确定性进行度量。为估算标记矛盾，KTC基于维基百科语料库的频率统计构建新方法。尽管仅依赖CPU运行，该方法相较基于交叉编码器的先进GPU加速方法实现超过8.2倍的加速，相比同类CPU方法更是提升逾65倍，同时保持相当甚至更优的性能，尤其在高精度场景下表现突出。该研究在涵盖四种欧洲语言和16个不同模型的两个基准上验证了其有效性，证明了其在保证准确性的同时具备极高的计算效率，使生产环境中对LLM输出的实时监控成为可能。
链接: https://arxiv.org/abs/2608.22506
作者: Jérémie Dentan,Alexi Canesse,Mahammed El Sharkawy,Sonia Vanier
机构: LIX (École Polytechnique, IP Paris, CNRS)
类目: Computation and Language (cs.CL)
备注: Preprint. Under review
Abstract:Claim-level Uncertainty Quantification (UQ) aims to mitigate the lack of reliability of Large Language Models (LLMs) by evaluating the factuality of each claim in their outputs. We introduce Kernel Token Contradiction (KTC), a lightweight approach to compute claim-level UQ under realistic white-box conditions. KTC represents the candidate tokens involved in LLM generation as a positive semi-definite kernel that integrates both the LLM’s conditional distribution and a token contradiction score. We then use the Von Neumann entropy to quantify the uncertainty of this kernel. To estimate token contradiction, we develop a new approach based on frequency statistics from the Wikipedia corpus. Although CPU-only, our approach achieves over an 8.2x speedup compared to state-of-the-art GPU-accelerated methods based on cross-encoders, and over a 65x speedup compared to CPU-only methods with comparable performance. Our evaluation spans two benchmarks across four European languages and 16 different models. KTC not only matches the average performance of existing methods but also outperforms them in high-precision regimes. This combination of computational efficiency and accuracy makes real-time monitoring of LLM outputs practical in production.
[NLP-80] Who Pays More for Safety? Measuring the Disparate Cost of Safety Alignment across Languages EMNLP2026
【速读】： 该论文旨在解决安全对齐（safety alignment）在多语言场景下是否存在不公平成本分摊的问题，具体关注不同语言群体在采用安全对齐模型时所承受的效用损失（utility loss）是否均等。其核心解决方案的关键在于提出一种严谨的评估协议，定义并量化仅由安全对齐本身带来的效用损失，即“安全成本（Safety Cost）”。通过直接对比安全对齐模型与其未对齐版本在相同任务下的表现，研究发现非英语用户普遍承担更高的安全成本，揭示出系统性不平等：部分语言处于“双重惩罚区”——既安全性保护较弱，又面临更大效用损失；某些语言看似出现效用提升，实则源于安全过滤器未能有效触发；甚至高资源语言为达到与英语相当的安全水平，也需付出更高的安全成本。这些差异源于显性拒绝行为及跨语言在多个维度上的隐性质量差异。该研究通过精准测量安全对齐的异质影响，暴露了当前安全对齐实践中的系统性偏见。
链接: https://arxiv.org/abs/2608.22490
作者: Chanwoong Yoon,Jungsoo Park,Alan Ritter
机构: Korea University (高丽大学); Georgia Institute of Technology (佐治亚理工学院)
类目: Computation and Language (cs.CL)
备注: Accepted to EMNLP 2026 Main Conference
Abstract:Safety alignment helps models adhere to human values, but it often reduces response utility. We ask a critical but understudied question: Does safety alignment impose the cost equally across language groups? To answer this, we introduce a rigorous protocol to measure the utility loss imposed solely by safety alignment, which we term Safety Cost. Through direct pairwise comparisons between safety-aligned models and their unaligned counterparts, we find a systematic inequity: non-English users consistently bear a higher Safety Cost than English users. We further identify three underlying patterns. First, multiple languages lie in a double-penalty zone, experiencing both weaker safety protection and larger utility loss. Second, certain languages exhibit apparent utility gains that are in fact a consequence of safety filters failing to engage. Third, even high-resource languages pay a larger Safety Cost than English to reach the same level of safety. We show that these disparities arise from both explicit refusals and implicit qualitative differences across multiple dimensions. By accurately measuring the disparate effects of safety alignment, our findings expose a systematic disparity in current safety alignment practices.
[NLP-81] Claim-Level Confidence Calibration for Reliable Decision Making with Large Language Models KDD2026
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在高风险决策场景中因幻觉（hallucination）和置信度与事实正确性不一致所导致的可靠性问题。现有方法通常仅提供响应级别的置信度，而这一粗粒度信号无法满足用户对信息逐条验证、采纳或拒绝的需求，因为单个生成结果可能同时包含正确与错误陈述。为此，本文提出一种面向声明级别（claim-level）的置信度校准方案，将响应分解为可验证的原子声明，并利用推理时的一致性信号（如多样本间的一致性）与自我验证机制进行后验校准（post-hoc calibration）。该框架在无需访问模型输出概率（无logits）或微调（no fine-tuning）的封闭盒（closed-box）设置下运行，能够精准识别低置信度声明，从而支持选择性干预（如检索证据或人工审核）。在TriviaQA和TruthfulQA数据集上对六种主流模型（Llama-3.1、Mistral、Qwen2.5、DeepSeek-R1、GPT-4、GPT-4o）的评估表明，该方法显著降低了事实类问题上的期望校准误差（Expected Calibration Error），并在对抗性虚假前提问题中有效暴露了模型的失效模式，为决策者提供了更可靠、可操作的不确定性信号。
链接: https://arxiv.org/abs/2608.22483
作者: Toghrul Abbasli,Kentaroh Toyoda,Yuan Wang,Li Chen
机构: Tsinghua University(清华大学); Vulcan Research, AIFT(维卢坎研究, 人工智能与未来技术); Keio Global Research Institute(庆应全球研究院); China Mobile Research Institute(中国移动研究院); Zhongguancun Laboratory(中关村实验室)
类目: Computation and Language (cs.CL)
备注: In Proceedings of The 5th Workshop on Uncertainty Reasoning and Quantification in Decision Making (held in conjunction with ACM SIGKDD 2026), Jeju, Korea
Abstract:Large Language Models (LLMs) increasingly support decision-making in high-stakes domains, but they often hallucinate and express confidence that is misaligned with factual correctness. Response-level confidence is a coarse signal: a single generation can mix correct and incorrect statements, so a single number is not actionable for users that must accept, reject, or verify individual pieces of information. We study claim-level confidence calibration as a decision-relevant uncertainty signal: each response is decomposed into atomic, verifiable claims, and each claim is assigned a calibrated confidence using inference-time signals from consistency across samples and self-verification. Our framework operates in closed-box settings (no logits, no fine-tuning) and applies post-hoc calibration directly at the claim level, enabling selective intervention such as evidence retrieval or human review for low-confidence claims. Across TriviaQA and TruthfulQA we evaluate seven baselines on six recent models (Llama-3.1, Mistral, Qwen2.5, DeepSeek-R1, GPT-4, GPT-4o), and show that claim-level decomposition combined with post-hoc calibration reduces expected calibration error on factual questions while exposing failure modes on adversarial false-premise questions where decision-makers most need reliable uncertainty estimates.
[NLP-82] GTA-RAG : Graph-Trajectory-Augmented Reinforcement Learning for Multi-Turn Retrieval-Augmented Reasoning EMNLP2026
【速读】： 该论文旨在解决多跳问答（multi-hop question answering）任务中，现有基于强化学习（RL）的检索增强生成（RAG）方法因仅依赖最终答案奖励而导致监督信号稀疏、无法有效引导模型正确获取证据链的问题。其核心解决方案是提出一种图轨迹增强的强化学习框架GTA-RAG，通过构建实体-文档图（entity–document graph），采样连通的文档路径并合成多跳问答轨迹，利用部署的检索器验证轨迹的可执行性以获得细粒度的轨迹级监督信号。在此基础上，采用分组相对策略优化（Group Relative Policy Optimization, GRPO）与轨迹引导奖励机制，同时优化答案准确性与目标证据文档的获取能力，并在自然问答数据上进行答案奖励微调。实验结果表明，该方法在多个复杂多跳和简单问答基准上均显著优于现有RL-RAG基线，且大幅提升了证据链覆盖度。
链接: https://arxiv.org/abs/2608.22479
作者: Jun Chen,Yongchao Liu,Pengyu Qiu,Jiajun Zheng,Juelu Zhang,Yujie Zeng,Qin Zhang,Ziyue Qiao,Xiao Luo
机构: Shenzhen University (深圳大学); Great Bay University (大湾区大学); Ant Group (蚂蚁集团)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 11 pages, 5 figures. Accepted to EMNLP 2026 Fundings
Abstract:Retrieval-augmented generation (RAG) enables LLMs to access external knowledge for answering knowledge-intensive questions. For complex multi-hop questions, multi-turn retrieval-augmented reasoning extends RAG into an iterative process that repeatedly searches for and integrates evidence across documents. However, existing reinforcement-learning (RL) approaches for agentic RAG are typically optimized with final-answer rewards, which provide sparse supervision and overlook whether the model actually retrieves the required evidence chain. We present \textscGTA-RAG, a graph-trajectory-augmented RL framework for multi-turn retrieval-augmented reasoning. From an entity–document graph, we sample connected document paths, synthesize multi-hop QA trajectories, and validate them with the deployed retriever to obtain executable trajectory-level supervision. We then optimize the retrieval policy with Group Relative Policy Optimization (GRPO) and a trajectory-guided reward that encourages both accurate answers and acquisition of target evidence documents, followed by answer-reward training on natural QA instances. Experiments on three multi-hop and two simple QA benchmarks show that \method consistently outperforms RL-based RAG baselines with both Qwen2.5-3B and Qwen2.5-7B backbones, while substantially improving evidence-chain coverage. Our code is available at this https URL.
[NLP-83] Small Reasoning Models are Instruction Followers in Function Calling
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在工具调用（tool calling）场景下准确率不足的问题，尤其针对现有方法在复杂推理任务中表现受限的瓶颈。其核心挑战在于：尽管通过微调、强化学习及多智能体框架可提升模型的函数调用能力，但这些方法在非指令跟随（instruction-following）的工具调用上下文中性能显著下降。为此，论文提出了一种名为指令跟随式函数调用（Instruction-Followed Function Calling, IFFC）的新框架，其关键创新在于将函数调用逻辑从主干大模型中解耦，并交由一个专门的小型模型在标准指令跟随范式下执行。该设计不仅显著提升了函数调用的准确性，尤其是在面向推理的大型模型上表现突出，还展现出对激进量化（aggressive quantization）的高度鲁棒性，支持低资源边缘设备上的高效部署，从而为边缘计算场景下的可靠、轻量级函数调用提供了一种新的范式。
链接: https://arxiv.org/abs/2608.22472
作者: Yalda Taheri,Mohammad Hassan Heydari,Erfan Naaman,Afsaneh Fatemi
机构: Islamic Azad University (伊斯兰阿扎德大学); University of Isfahan (伊斯法罕大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Function calling represents the core capability of agentic large language models (LLMs). Existing research has focused on enhancing LLMs function-calling accuracy through fine-tuning, reinforcement learning (RL), and multi-agent frameworks, particularly for native function-calling LLMs. This work demonstrates that LLMs achieve superior accuracy in function calling in instruction-following contexts (i.e., standard user-assistant interactions) rather than a tool calling context. We introduce Instruction-Followed Function Calling (IFFC), a novel framework that decouples function-calling logic from the primary LLM and delegates it to a dedicated smaller model operating within the instruction-following paradigm. Our method consistently outperforms both native function calling (NFC) and prompt-based function calling (PFC) baselines, with particularly strong gains on reasoning-oriented LLMs. Furthermore, we demonstrate that IFFC maintains robust performance under aggressive quantization, enabling efficient on-device deployment without significant accuracy degradation. This work establishes a new paradigm for reliable, resource-efficient function calling in edge-computing scenarios.
[NLP-84] From Exposure to Expectation: Frequency Surprisal and Language Across Development in Spanish
【速读】： 该论文旨在探究语言模型生成的意外度（surprisal）在解释儿童词汇习得时间（age of acquisition, AoA）中的作用，与词汇频率（frequency）这一累积暴露指标进行对比。研究核心问题是：在语言发展过程中，词汇的可预测性（即基于上下文的意外度）是否像频率一样对词汇习得具有同等解释力？其解决方案的关键在于通过两个基于语料库的研究，在西班牙语语境下系统评估不同语言模型（BETO、BERTIN、mGPT）生成的意外度与儿童词汇习得时间及成人在自然阅读中注视时长之间的关系。研究发现，词汇频率是预测儿童词汇习得时间的强效指标（r = -0.597），而意外度在控制频率和词长后贡献有限；但在成人阅读的眼动数据中，意外度仍能显著预测更长的注视时长，且该效应独立于频率。此外，意外度与阅读行为的相关性显著强于其与习得时间的相关性。这表明，频率主要反映早期词汇表征的累积性学习过程，而意外度则更适合作为已有语言系统中即时加工难度的指标。研究结果支持使用基于使用的语言发展观，并强调语言模型作为人类语言行为建模工具的有效性与局限性。
链接: https://arxiv.org/abs/2608.22452
作者: Francisco Portillo López
机构: Universidad de Navarra (纳瓦拉大学); Spain (西班牙)
类目: Computation and Language (cs.CL)
备注: 30 pages; 4 figures; 5 tables
Abstract:Surprisal, the negative log-probability a language model assigns to a word given its preceding context, reliably predicts adult reading times. Does it contribute as much to explaining when children acquire individual words? Frequency reflects a learner’s cumulative exposure to a word, whereas surprisal reflects how predictable a single occurrence is given its context. We investigate this question across two corpus-based studies of Spanish. In Study 1, we modeled age of acquisition (AoA) for 225 Spanish nouns using lexical frequency and contextual diversity from child-directed speech, plus surprisal from three language models differing in architecture and training language (BETO, BERTIN, mGPT). Frequency strongly predicted AoA (r=-.597, p.001); surprisal added little beyond frequency and word length, including in a naturalistic-context analysis. In Study 2, we modeled adult fixation durations in the Chilean Spanish subsample of the Multilingual Eye-movement Corpus (MECO Wave 2), using mGPT surprisal alongside two independent frequency measures. Surprisal robustly predicted longer fixation durations after controlling for frequency and word length, consistent across both frequency sources. A matched word-type-level comparison showed the surprisal-behavior association was stronger in reading than in acquisition (z=3.63, p.001). The findings suggest cumulative lexical exposure and contextual predictability play different roles across the language trajectory: frequency is particularly informative about when early lexical representations are acquired, whereas surprisal captures moment-to-moment processing difficulty in an already-established linguistic system. We discuss this pattern in relation to usage-based and entrenchment-based accounts of lexical development and to the evaluation of language models as models of human language behavior. Comments: 30 pages; 4 figures; 5 tables Subjects: Computation and Language (cs.CL) Cite as: arXiv:2608.22452 [cs.CL] (or arXiv:2608.22452v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2608.22452 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Francisco Portillo López [view email] [v1] Sun, 23 Aug 2026 15:02:29 UTC (2,613 KB)
[NLP-85] Figurative Justice: Detecting metaphors in Hindi judgements with qualitative assessment and transformers
【速读】： 该论文旨在解决低资源语言（如印地语）在法律语境中隐喻自动检测的难题。由于缺乏标注的法律语料库，现有自然语言处理（NLP）模型难以有效识别司法文本中的隐喻表达，而隐喻在人权侵权案件等法律话语中具有决定性作用，直接影响判决结果、公众认知与法律解释。其解决方案的关键在于构建首个印地语法律隐喻语料库（HiLeMe），基于印地语法律数据语料库（HLDC）并由法律专家依据MIPVU标注框架进行标注；同时，采用mBERT预训练模型并设计一种基于Transformer的架构，在下游任务中实现对印地语法律隐喻的有效检测。该方法不仅提升了低资源语言法律文本分析的准确性，也为未来向22种印度官方语言推广提供了可扩展的技术路径。
链接: https://arxiv.org/abs/2608.22446
作者: Bhumika Bhattacharyya,Shouvik Kumar Guha,Indranil Dutta
机构: DigiTS, University of Tartu (塔尔图大学), Estonia; Department, The West Bengal National University of Juridical Sciences (NUJS), India; School of Languages and Linguistics, Jadavpur University (加达普尔大学), India
类目: Computation and Language (cs.CL)
备注: 12 pages, 5 figures, 2 tables. Dataset available at this https URL
Abstract:Metaphors are figurative use of words for conceptual mapping. Metaphor detection in the legal context has been crucial as metaphors are persuasive juridical means of creating legal meaning and concepts resulting in significant consequences. Metaphorical framing in legal discourse by judges, lawyers, and legislators brings about real-time implications upon individuals and influences judicial decision-making, argumentation and interpretation of laws. This is crucial in Human Rights infringement cases where language determines severity of punishment, public perception and judicial outcomes. While automatic metaphor detection in major languages like English, Spanish, Polish, Lithuanian have aided in understanding inherent intentions of metaphorical use of language, there is no such attempt in low-resource languages like Hindi. The dearth of annotated legal corpora in Hindi makes it difficult to develop NLP models and detect metaphors in judicial proceedings. In the Indian context, Convolutional Neural Networks (CNNs) have been used for classification of bail judgements, however there are no existing models designed for metaphor detection. We present a Hindi Legal Metaphor Corpus (HiLeMe) by isolating judgements from Hindi Legal Data Corpus (HLDC). Legal experts annotated HiLeMe to classify metaphorical constructions using the MIPVU schema. We downstreamed an mBERT on Hindi legal metaphor detection task. We built a transformer-based architecture for metaphor detection that are known to outperform traditional models in legal classification tasks. This model provides insights into the judicial psyche for decoding judicial decisions. Our research contributes to advancing automated models in legal discourse in low-resource languages like Hindi and envisages adoption into 22 Indian schedule languages. Comments: 12 pages, 5 figures, 2 tables. Dataset available at this https URL Subjects: Computation and Language (cs.CL) ACMclasses: I.2.7 Cite as: arXiv:2608.22446 [cs.CL] (or arXiv:2608.22446v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2608.22446 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-86] Aligned Alone Misaligned Together: Forecasting Adversarial Capture in LLM Agent Populations
【速读】： 该论文旨在解决当前人工智能安全评估范式在多智能体交互环境下的局限性问题：现有评估体系以单个模型为单位，无法有效反映语言模型代理在相互影响的群体中可能产生的集体行为偏差。其核心挑战在于，一个在独立评估中表现良好的代理，在与其他代理交互时仍可能被群体动态引导至偏离原判的决策方向。解决方案的关键在于提出一种基于群体初始行为的响应函数校准方法——仅通过观察群体在无对抗条件下的正常运作，即可预先预测当存在“坚定少数派”（committed minority）攻击时，群体整体将被引导向何方向。研究发现，即使两个警报对单个代理的判断几乎相同，其引发的集体行为却可大相径庭；而让代理共享彼此推理过程虽能抵消弱攻击，但仅延迟强攻击，使问题从“是否被俘获”转化为“何时被俘获”。此外，研究排除了“捕获”为不可逆状态的可能性，表明一旦移除攻击性代理，群体行为会逐渐回归初始状态，说明捕获是暂时性现象。因此，个体对齐不等于群体对齐，但群体在攻击前的行为模式足以预判其抗攻击能力。
链接: https://arxiv.org/abs/2608.22444
作者: Isotta Magistrali,Chen Shani
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:The unit of AI safety evaluation is still the individual model, yet language-model agents are increasingly deployed in interacting populations that read and write one another’s decisions. This raises a question no single-agent audit can answer: an agent that is well-calibrated on its own may still be pulled toward a different decision by the agents around it. We study this on a security-triage task, where populations of language-model monitors decide whether to escalate or dismiss alerts, and into which we can inject a committed minority that always pushes one way. We find that two alerts a single agent judges almost identically on its own can drive collective behavior far apart, so auditing any one member need not reveal what the population will do. Yet that collective behavior can be predicted in advance. From a population’s benign, adversary-free operation alone, we calibrate a response function that forecasts, before any attack is run, how far a committed minority will later move it. We then ask what shifts the outcome and find that letting agents see each other’s reasoning neutralizes a weak attack, while only delaying it against a strong one, turning the question from whether the population converges on the adversaries’ choice into when. Finally, we exclude the hypothesis of capture being an irreversible trap: once the committed agents are removed, the population drifts back toward where it began, so capture is a temporary state. Alignment in isolation is not alignment in a population, yet what a population will do under attack can be read in advance, from how it behaves before any adversary arrives.
[NLP-87] Rank Reversal in Multilingual LLM Judges: A Label-Free Double-Centering Calibrator
【速读】： 该论文旨在解决多语言大语言模型（LLM）评判者在不同语言提示下产生不一致的评估者-骨干模型排名的问题，即语言依赖性偏差导致的评估结果不可靠。其核心问题是：当使用多语言提示时，同一骨干模型在不同语言下的评分表现出现显著波动，甚至引发排名反转，从而影响跨语言评估的可比性和一致性。解决方案的关键在于提出一种无需人工标注的后处理校准方法——基于共识的校准（Consensus-Based Calibration, CBC），该方法通过双中心化（double-centering）单元均值评分矩阵，有效分离并恢复语言-骨干交互项，从而消除语言偏差。该方法在理论上具有 O(1/n) 的有限样本集中率界，方差常数为 (1−m1)(1−k1)，且在任务-语言交互存在的情况下仍保持无偏性。实验表明，在7,920次评估运行中，CBC将跨任务排名一致性（τ）从0.650提升至0.902，与隐含加法模型最优解的一致性达100%；在独立收集的M-RewardBench数据集上，与人类黄金偏好的一致性从68.7%提升至76.6%，验证了其在下游应用中的有效性。该方法本质上是带和为零约束的双因素方差分析（two-way ANOVA）中交互效应的恢复操作，其创新点在于将其作为无标签后处理校准器应用于多语言LLM评判场景，并提供了严格的理论保证。
链接: https://arxiv.org/abs/2608.22432
作者: Alhasan Mahmood,Samir Abdaljalil,Hasan Kurban
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Multilingual LLM judges produce different evaluator-backbone rankings depending on the prompt language: on an eight-language Agent-as-a-Judge benchmark, the top-ranked backbone alternates across English, Arabic, Chinese, Hindi, Japanese, Spanish, Turkish, and Swahili, and 7 of 15 backbone pairs show statistically significant pairwise rank reversal. We treat this as a measurement problem. The multilingual judge score decomposes additively into task difficulty, backbone skill, and a language-backbone interaction term, the last of which is recoverable without human labels by double-centering the cell-mean score matrix. We make this estimator (\textbfConsensus-Based Calibration, CBC) explicit, give an O(1/\sqrtn) finite-sample concentration bound with variance constant (1-\tfrac1m)(1-\tfrac1k) , and show that it is unbiased even when task-language interactions are present. Across 7,920 judge runs (6 backbones, 8 languages, 55 tasks, 3 frameworks), CBC raises held-out cross-task rank consistency \tau from 0.650 to 0.902 and agrees with the held-out additive-model oracle in 100% of per-language decisions versus 68.5% raw; these are consistency diagnostics, not human-grounded correctness measures. On a separately collected M-RewardBench panel (7 languages, 1,500 items per language, 10,500 language-item instances, 5 evaluators), panel agreement with the public human gold preferences rises from 68.7% to 76.6% (gain 7.9 percentage points, 95% CI [6.0, 9.9] ), our strongest external evidence of downstream usefulness. The estimator is the standard two-way ANOVA interaction-recovery operation under sum-to-zero contrasts; our contribution is its application as a label-free post-hoc calibrator for multilingual LLM judges, an explicit finite-sample concentration bound, and an unbiasedness result that holds even under task-language misspecification.
[NLP-88] Don t Box Me In: Dynamic Cultural Adaptation and Cognitive Tracking for Social Understanding EMNLP2026
【速读】： 该论文旨在解决当前大型语言模型（LLM）在多元文化社交场景中缺乏动态文化适应能力的问题。现有方法通常将文化视为静态的人口统计属性，无法有效建模个体在跨情境中混合且变化的交际偏好。为此，本文提出一种无需训练的框架DyCAC（Dynamic Cultural Adaptation with Continuous Cognitive tracking），其核心在于通过引入动态文化适应（Dynamic Cultural Adaptation）与持续认知状态追踪（Continuous Cognitive Tracking）实现流畅的社会对齐。关键创新在于：不预设固定的群体文化身份，而是将文化相关的交际偏好建模为随时间变化的、基于群体文化参照谱系（population-level cultural reference profiles）的混合分布，并结合对话过程中实时观察到的交互信号进行动态校准，从而捕捉复合文化影响及逐轮交流行为的演变；同时，基于心理理论（Theory of Mind, ToM）驱动的记忆模块持续追踪对话对象的认知状态，增强模型对社会语境的敏感性。实验结果表明，DyCAC在多个交互式社交与文化基准测试中显著优于现有基线，展现出更强的社会智能与跨多元文化场景的广泛适应性。
链接: https://arxiv.org/abs/2608.22411
作者: Chongyuan Dai,Yaling Shen,Shengeng Tang,Hui Ma,Jinpeng Hu
机构: Hefei University of Technology (合肥工业大学); Monash University (莫纳什大学)
类目: Computation and Language (cs.CL)
备注: EMNLP 2026 Findings
Abstract:Social interaction increasingly takes place in multicultural settings, where individuals may draw on multiple cultural influences and adapt their communicative behavior across contexts. Despite recent advances in equipping Large Language Models (LLMs) with social understanding capabilities, existing approaches often model culture as a static demographic attribute, limiting their ability to accommodate hybrid and dynamically expressed communicative preferences. Therefore, in this paper, we propose \textbfDyCAC, a training-free framework that achieves fluid social alignment by incorporating \underlineDynamic \underlineCultural \underlineAdaptation with continuous \underlineCognitive tracking. Rather than inferring a fixed cultural identity, DyCAC models culturally relevant communicative preferences as a time-varying mixture of population-level cultural reference profiles. This reference-based representation is further calibrated using dialogue-style signals observed in the ongoing interaction, enabling the model to capture both composite cultural influences and turn-level shifts in communicative behavior. In parallel, a memory module driven by Theory of Mind (ToM) continuously tracks the cognitive states of the interlocutor. Extensive experiments on interactive social and cultural benchmarks demonstrate the superiority of our approach. The proposed framework outperforms existing baselines, exhibiting enhanced social intelligence and broad adaptability across varied multicultural contexts.
[NLP-89] SchemaGUI: A Schema-Driven Benchmark for Controllable GUI Generation Evaluation
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在图形用户界面（GUI）生成任务中缺乏可靠评估体系的问题，主要挑战包括数据分布不可控、标注噪声以及布局场景覆盖有限。其解决方案的关键在于提出SchemaGUI——一个基于模板的可控GUI生成评估基准。通过从参数化界面模板中合成成对的自然语言指令与确定性函数调用引用，SchemaGUI可在无需人工标注的情况下，于数秒内生成数千个具有确定性标注的任务实例。基于六个代表性双语场景下每种场景与语言1,000个评估实例的实验，对包括Qwen3.5系列、Qwen3-Coder-30B和DeepSeek-R1在内的五款主流模型进行评测，揭示出三个核心发现：首先，几何空间控制仍是关键瓶颈，尽管模型规模扩大可显著提升界面可行性（如Qwen3.5从4B到27B时由91.56%升至99.63%），但几何精度提升有限（仅从67.05%增至75.30%）；其次，生成难度高度依赖布局复杂度，当前模型在简单序列布局中表现良好，但在密集网格或多区域组合等复杂结构中出现严重坐标漂移；最后，思维模式虽增强推理能力，却显著增加令牌消耗，并普遍降低GUI评分，尤其对小型模型影响更显著。
链接: https://arxiv.org/abs/2608.22390
作者: Jiarui Dong,Yin Cai,Zhouhong Gu,Chenmou Wu,Ci Tao,Yiran Chen,Jialing Li,Xiaoran Shi,Juntao Zhang,Zhijun Fang
机构: Shanghai University of Engineering Science(上海工程技术大学); Fudan University(复旦大学)
类目: Computation and Language (cs.CL)
备注: 18 pages, 6 figures, and 7 tables. Code is available at this https URL
Abstract:Large language models (LLMs) have demonstrated strong potential in graphical user interface (GUI) generation, but reliable evaluation remains challenging due to uncontrolled data distributions, noisy annotations, and limited layout scenario coverage. To address this, we propose SchemaGUI, a template-based benchmark for controllable GUI generation evaluation. By synthesizing paired natural language instructions and deterministic function-call references from parameterized interface schemas, SchemaGUI can generate thousands of deterministically annotated tasks in seconds without human labeling. Based on 1,000 evaluated instances per scenario and language across six representative bilingual scenarios, we benchmark five mainstream models, including the Qwen3.5 family, Qwen3-Coder-30B, and DeepSeek-R1. Our extensive analysis reveals three key insights. First, precise geometric spatial control remains an important bottleneck; while scaling Qwen3.5 from 4B to 27B improves Schema Feasibility from 91.56% to 99.63%, the Geometry score improves more modestly (from 67.05% to 75.30%). Second, generation difficulty is highly sensitive to layout complexity, with current LLMs excelling at simple sequential arrangements but suffering severe coordinate drift in dense grids and multi-region compositions. Third, thinking mode increases token consumption while generally reducing GUI Score, particularly for smaller models.
[NLP-90] ProBel: Propaganda Detection with Techniques Spans and Explanations
【速读】： 该论文旨在解决多语言（阿拉伯语与英语）背景下宣传手法检测中不同标注粒度（如句子级、技术分类、片段识别等）的监督信号在联合学习时如何相互作用的问题。其核心挑战在于揭示不同任务层级的标注信息在跨语言场景下的协同机制与性能影响。解决方案的关键在于构建了一个名为ProBel的双语资源，该资源统一了二分类标签、23种宣传技术的多标签标注（归类至6个粗粒度类别）、技术标记的文本片段以及对应新闻句子的解释性参考，实现了对阿拉伯语和英语在多种任务层级上的对齐标注。通过在统一框架下评估零样本提示、特定任务微调及联合训练，研究发现：单一的双语多任务模型在整体性能上表现最优，且在不同任务与语言间保持稳定；跨任务分析表明，迁移效果依赖于监督粒度——联合分类训练能有效维持句子级预测性能，而仅进行片段级训练可能削弱句子级判断能力；联合双语训练带来最稳定的迁移表现，而单语微调则可能损害向另一语言的迁移能力。
链接: https://arxiv.org/abs/2608.22388
作者: Mohamed Bayan Kmainasi,Ali Ezzat Shahroor,Elisa Sartori,Giovanni Da San Martino,Firoj Alam
机构: Qatar Computing Research Institute, Qatar; University of Padova, Italy
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Under review at
Abstract:Propaganda detection includes several related prediction levels, ranging from sentence-level decisions to technique classification and span identification. However, it remains unclear how supervision at these levels interacts when learned jointly across Arabic and English. We present ProBel, an Arabic and English resource that aligns binary labels, multi-label annotations over 23 propaganda techniques grouped into six coarse categories, technique-labeled spans, and reference explanations for the same news sentences. It includes a substantially larger English collection and supports matched binary, coarse-grained, multi-label, and span-level tasks in both languages. We evaluate zero-shot prompting, task-specific fine-tuning, and joint training under a shared setup. A single bilingual multi-task model achieves the best overall performance and remains competitive across tasks and languages. Cross-task analysis shows that transfer depends on the supervision level. Joint classification training preserves binary performance, whereas span-only training can weaken sentence-level prediction. Joint bilingual training yields the most stable results, while monolingual fine-tuning can reduce transfer to the other language. We will release the data, code, and evaluation scripts.
[NLP-91] Can Large Language Models “Hyper-Thread”?
【速读】： 该论文旨在解决大语言模型在序列生成过程中难以实现多任务并行执行的问题，即尽管模型以串行方式逐个生成标记（token），但其是否能在生成每个标记时同时处理多个任务。现有推理扩展方法主要依赖于更长的生成长度、更多采样或额外验证阶段，而注意力分散通常被视为干扰或错误信号，导致串行生成中的任务并发性长期被忽视。为此，作者提出“模型超线程假说”（Model Hyper-Threading Hypothesis），并通过设计三种实验条件（基线、串行功能调度与并发功能加载）进行验证。关键创新在于引入“并发功能加载”机制，使多个任务共享状态并在同一生成步骤中并行执行。实验结果表明，在AIME 2025开发集上，并发功能加载实现了最高准确率，其输出长度与串行调度相近甚至更短，且表现出更高的注意力分散度和任务相关覆盖范围，尽管存在较长输出尾部的代价。研究发现，更分散的注意力可与更高准确率共存，为超线程假说提供了初步的行为学与相关性证据。这一发现推动了对推理扩展范式的根本转变：从“生成更多标记”转向“让每个生成步骤承载更多任务”，为提升推理性能开辟了新的技术路径。
链接: https://arxiv.org/abs/2608.22376
作者: Fei Ding
机构: Alibaba Group(阿里巴巴集团)
类目: Computation and Language (cs.CL)
备注: 12 pages
Abstract:Large language models generate tokens sequentially, but can they execute multiple tasks concurrently while forming each token? Broader attention allocation may provide a mechanism for such task concurrency. Existing approaches to scaling inference primarily rely on longer generations, more samples, or additional verification stages, while attention dispersion is often treated as a signal of interference or error. Task concurrency within serial generation therefore remains underexplored. We propose the Model Hyper-Threading Hypothesis and evaluate its predictions using multiple coordinated tasks that share state within the same problem. We design three conditions (Baseline, Serial Functional Scheduling, and Concurrent Functional Loading) and evaluate their benefits and costs using accuracy, output-token distributions, and attention metrics. On an AIME 2025 development set, Concurrent Functional Loading achieves the highest accuracy. Relative to Serial Functional Scheduling, its typical output length is similar and it is shorter on most problems, while exhibiting greater attention dispersion and higher task-relevant coverage, albeit with a heavier output-length tail. Within-step concurrency and its causal mechanism still require direct tests. Our results show that more dispersed attention can coexist with higher accuracy, providing preliminary behavioral and correlational evidence for the hyper-threading hypothesis. These findings motivate a shift in perspective on inference scaling from “generating more tokens” toward “having each generation step carry more tasks,” pointing to a new avenue for improving reasoning performance.
[NLP-92] Context-Aware Cluster Decoding: Semantic Anchor-Driven Coherence in dMLLM s EMNLP2026
【速读】： 该论文旨在解决扩散型多模态大语言模型（dMLLMs）在生成长文本时普遍存在的语义漂移（semantic drift）与重复问题，其核心挑战在于输出质量随生成长度增加而显著下降。现有解码方法存在两大结构性缺陷：一是基于置信度的评分机制忽略了已解码邻域信息的支持，二是分块（block partitioning）策略限制了对高就绪度语义锚点的全局访问，导致模型过早提交尚未充分建立局部上下文的词汇。为此，本文提出一种无需训练的解码方法——\ours（上下文感知聚类解码，Context-Aware Cluster Decoding），其关键在于通过软最大值置信度与邻近性乘积复合评分机制，在不依赖分块结构的前提下，优先选择具备上下文支持的高就绪候选词，同时抑制低置信度位置的噪声干扰。此外，\ours引入架构感知校准以应对不同视觉融合策略带来的置信度异质性。在三个dMLLM模型、四个基准上的实验表明，该方法在长序列生成任务中均实现显著的质量提升与幻觉减少，验证了邻域支持与视觉融合策略在下一代dMLLM解码设计中的关键作用。
链接: https://arxiv.org/abs/2608.22367
作者: Yikai Zhao,Qiyan Zhao,Jiaquan Zhang,Xiaofeng Zhang,Xiaosong Yuan,Pengzhou Cheng
机构: Sun Yat-sen University(中山大学); Shanghai Jiao Tong University(上海交通大学); University of Electronic Science and Technology of China(电子科技大学); Alibaba Group(阿里巴巴集团); Shanghai University(上海大学)
类目: Computation and Language (cs.CL)
备注: This paper is accepted by EMNLP 2026. 19 pages, 12 figures, 13 tables
Abstract:Diffusion multimodal large language models (dMLLMs) frequently produce long-form outputs marred by semantic drift and repetition, with quality generally degrading as output length increases. We identify two structural deficiencies in existing decoding methods as primary drivers of these failures: confidence-based scoring ignores decoded-neighbor support, and block partitioning prevents access to high-readiness semantic anchors, together causing tokens to be committed before their local context is sufficiently established. We propose \ours (\textbfContext-\textbfAware \textbfCluster \textbfDecoding), a training-free decoding method that scores each masked position by a multiplicative composite of softmax confidence and neighbor proximity, promoting contextually ready tokens above isolated candidates while suppressing low-confidence positional noise, operating block-free to keep high-readiness anchors globally accessible. \ours further applies architecture-aware calibration to handle confidence heterogeneity induced by diverse visual integration strategies. Experiments on three dMLLMs across four benchmarks demonstrate consistent quality gains and hallucination reduction over Original, with larger gains in several longer generation settings, highlighting the importance of neighbor support and visual integration strategy for future dMLLM decoding method design. Our code is openly available at this https URL.
[NLP-93] Where Cognition Lives: Dissecting Emergent from Computed Function in a Minimal Complete Cognitive Architecture
【速读】： 该论文旨在解决生成式智能系统中“认知架构”如何自主决定推理时长与资源分配的核心问题，即在不预先设定固定计算量的前提下，实现对思考深度与价值判断的动态优化。其解决方案的关键在于构建一个包含自适应停止机制（adaptive halting）、稳态控制场（homeostatic control field）和价值模块（value module）的最小但完整的递归推理系统。研究发现，虽然推理能力与部分停止策略看似可通过梯度下降自然涌现，但实际验证表明：停止规则的有效性依赖于显式建模的先验价值评估；当采用后验自我观测作为停止信号时，看似优越的性能在严格审计下无法复现。进一步实验显示，价值模块无法通过训练自动涌现，必须显式计算——尤其在价值与内容正交的情况下，显式分配机制可带来显著收益（+0.151），而耦合训练则完全失效。此外，在冻结的大语言模型（LLM）执行器上，自我一致性投票仅提供微弱增益，而样本间一致性则几乎无用且偏向错误答案。整个研究通过可证伪的预测框架与严格的对照实验设计，确立了价值驱动的二阶决策必须显式计算这一核心结论，并强调方法论本身构成贡献的一部分。
链接: https://arxiv.org/abs/2608.22347
作者: Francisco M. Arrabal-Campos,Francisco G. Montoya,Alfredo Alcayde,Ignacio Fernández
机构: University of Almería (阿尔梅里亚大学); Research Centre CIAIMBITAL (CIAIMBITAL 研究中心)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 16 pages, 3 figures. Code, preregistrations and results: this https URL
Abstract:A cognitive architecture is more than the module that reasons: it must also decide how long to think and what deserves the effort. We built a minimal but complete system - a recurrent reasoner with adaptive halting, a homeostatic control field, and a value module - and asked of each part: does this function emerge from gradient descent, or must it be computed? Competence emerges. Stopping appears to emerge too, and to be worth more than everything decidable in advance, but that appearance is instrumentation: payoff at matched mean compute climbs from 0.467 (uniform) through 0.546 (difficulty) to 0.698 (ex-ante value), and the further climb to 0.921 (posterior self-observation) does not survive audit. PonderNet-style halting returns a halting-weighted mixture of hidden states while forced-depth baselines return one, and the language head is trained on the mixture alone; equalizing the readout annihilates the apparent advantage of native execution (residual +0.000 [0.000, 0.000]). Value does not emerge: trained couplings capture zero of a payoff an explicit allocator captures completely (+0.151, routing correlation +0.79), so the second-order decisions that pay must be computed, at least where value is orthogonal to content, as here by construction. On a frozen LLM actuator the same instruments show self-consistency voting to be a measured bound (+0.0236 [+0.0150, +0.0326]) and inter-sample agreement nearly worthless as a stopping signal, its mass concentrating on wrong answers. Every null we assert carries a mechanism and a positive control, and the protocol is part of the contribution. Executing our own falsifiable prediction, value under commitment pays +0.1312 [+0.1124, +0.1502] in a cliff-cost family, some seven times the smooth-family estimate - not because the cliff shifts information ex ante, but because it multiplies the attainable range fivefold (5.1x [3.4, 8.2]).
[NLP-94] When Not to Imitate: Boundary-Aware Skill Memory for Reliable Tool-Use LLM Agents EMNLP2026
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）智能体在自我演化过程中因依赖仅从成功轨迹中提取技能而导致的“技能模仿陷阱”（Skill Imitation Trap）问题。现有方法假设将成功经验转化为技能记忆可单调提升问题求解能力，但实证分析表明，仅基于成功轨迹提取的技能在面对与历史任务相似但需使用不同工具的新任务时，反而会增强模型对错误工具调用的误判信心，导致性能下降。其核心解决方案是提出边界感知型技能记忆（Boundary-Aware Skill Memory, BASM），通过为每条技能显式添加适用条件、风险提示、规避规则和恢复建议等边界字段，使技能从无条件的动作模板转变为状态相关的决策引导机制：仅当环境状态满足适用条件时才激活技能，不满足时主动抑制不当操作，并在执行失败时触发精准修复。实验结果表明，在三个智能体基准测试和四种模型规模下，BASM显著优于传统成功轨迹提炼的技能记忆基线，任务成功率最高提升23.8%（AppWorld）、准确率提升5.0%（BFCL）、攻击成功率降低4.6%（AgentDojo），同时平均任务步数减少6.6%，验证了其在提升鲁棒性与泛化能力方面的有效性。
链接: https://arxiv.org/abs/2608.22339
作者: Zihan Lin,Zhenyu Chen,Jiawen Wei,Xiaohan Wang,Jie Cao,Jiajun Chai,Wei Lin,Guojun Yin,Ran He
机构: Meituan(美团); School of Advanced Interdisciplinary Sciences, University of Chinese Academy of Sciences(中国科学院大学交叉学科研究院); MAISNLPR, Institute of Automation, Chinese Academy of Sciences(中科院自动化所智能感知与计算实验室); Zhongguancun Academy(中关村学院)
类目: Computation and Language (cs.CL)
备注: Accepted by EMNLP2026 Findings
Abstract:Extracting skills from past successes is critical for the efficient evolution of Large Language Model (LLM) agents. Prevailing agent self-evolution paradigms typically rely on a core assumption: equipping LLMs with skill memories derived from successful trajectories will monotonically improve their problem-solving capabilities. However, probe analyses reveal that extracting skills solely from successful trajectories traps the model in a \textbfSkill Imitation Trap. For tasks that resemble past successes but require different tools, retrieving more skills paradoxically increases the model’s confidence in wrong tool calls—procedure skills raise the wrong-tool margin by 47% over a memory-free baseline. To overcome this limitation, we propose \textbfBoundary-Aware Skill Memory (BASM), which augments each skill with explicit boundary fields—applicability conditions, risk cues, avoidance rules, and recovery notes. These fields transform each retrieved skill from an unconditional action template into state-conditioned guidance: the agent applies the skill when its conditions hold, suppresses inapplicable tool calls when they do not, and issues targeted repairs when execution fails. Across three agent benchmarks and four model scales, BASM consistently outperforms success-distilled skill-memory baselines: it improves task success rate by up to 23.8% on AppWorld, accuracy by up to 5.0% on BFCL, and reduces attack success rate by 4.6% on AgentDojo, while simultaneously reducing average AppWorld steps by up to 6.6% relative to the memory-free baseline.
[NLP-95] Register Shifts Break LLM Safety: A Bengali Benchmark with Culturally Grounded Harms
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）安全评估长期以英语为中心，导致非英语语言如孟加拉语（Bengali）在安全性评测中严重缺失的问题。其核心挑战在于：现有评估体系无法有效覆盖多语言、跨文化语境下的潜在风险，尤其在文化敏感性与表达风格差异显著的场景下。解决方案的关键在于构建首个针对孟加拉语的系统性安全评估基准——BanglaSafe，包含879条经母语者撰写与专家审核的孟加拉语文本提示，涵盖17个基于本土文化的危害类别及五种不同的提示条件（包括语言风格、写作形式与权威性框架）。研究发现，尽管模型在孟加拉语中的整体安全性表现不佳（53.6%的输出为不安全或部分不安全，14.7%为严格有害内容），但最显著的影响因素并非语言切换本身，而是同一请求在正式新闻调查体与非正式消息体之间的表述风格差异——前者成功率高出17个百分点，表明写作风格对模型响应行为具有显著影响。此外，研究揭示当前主流安全分类器在处理孟加拉语内容时可靠性不足，即使前沿模型也在近半数案例中失效，凸显了多语言安全评估能力建设的紧迫性。
链接: https://arxiv.org/abs/2608.22335
作者: Naymul Islam,Nusrat Jahan Lia,Shubhashis Roy Dipta,Sabik Bin Sultan,Abdullah Khan Zehady
机构: BanglaLLM; Institute of Information Technology, University of Dhaka; University of Maryland, Baltimore County; Bangladesh Air Force Shaheen College Kurmitola; Ciroos Inc.
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Bengali is the seventh-most-spoken language globally, yet LLM safety evaluation remains overwhelmingly English-centric. We introduce BanglaSafe, a benchmark of 879 Bengali prompts combining 309 natively authored prompts with 570 expert-reviewed prompts, spanning 17 culturally grounded harm categories and five prompting conditions that vary language, writing style, and authority framing. Evaluating 18 frontier LLMs, we find that over half of all responses are unsafe or partially unsafe (53.6%) while 14.7% contains strictly harmful content, and that the strongest observed effect is not the switch from English to Bengali but the choice of writing style within Bengali: the same harmful request phrased as a formal newspaper investigation succeeds 17 percentage points more often than the same request phrased as a casual message, with no adversarial engineering involved. We further show that existing safety classifiers struggle to reliably evaluate Bengali content, with even frontier models failing on nearly half of all cases.
[NLP-96] Mechanistic Interpretability of Chain-of-Thought Reasoning via Sequential Activation Patching
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在采用思维链（Chain-of-Thought, CoT）提示时展现出强大问题求解能力，但其内部机制尚不清晰的问题。具体而言，研究聚焦于CoT相关因果效应在推理轨迹中何时产生，以及哪些注意力头（attention heads）携带对最终答案计算有贡献的信号。传统静态单位置激活修补（activation patching）方法无法捕捉随时间分布的动态效应，因此本文提出一种序列化激活修补框架（sequential activation patching framework），通过追踪推理过程中各注意力头在不同词元（token）位置上的激活变化，并结合词性标注（Part-of-Speech-guided analysis）进行效应聚合，以揭示时空上分散的因果影响。此外，引入序列多头修补（Sequential Multi-Head Patching） 方法，评估分布式注意力头集合的协同作用，并设置跨问题与随机激活对照组以增强可靠性。靶向零消融实验表明，所识别出的注意力头在成功生成答案中具有功能性重要性，影响多个重叠机制，包括推理轨迹维持、答案锚定、示例-目标分离及数值生成等。总体而言，研究结果为与CoT条件计算相关的分布式推理支持子电路（distributed reasoning-support sub-circuits）提供了实证支持。
链接: https://arxiv.org/abs/2608.22332
作者: Murat Dura,Serkan Öztürk,Selma Tekir
机构: İzmir Institute of Technology(伊兹密尔理工学院); Department of Computer Engineering(计算机工程系)
类目: Computation and Language (cs.CL)
备注:
Abstract:Large Language Models (LLMs) demonstrate remarkable problem-solving capabilities when guided by Chain-of-Thought (CoT) prompting, yet the internal mechanisms underlying these improvements remain poorly understood. In this work, we investigate where CoT-related causal effects emerge across the generated reasoning trajectory and which attention heads carry signals that contribute to final-answer computation. Because CoT reasoning unfolds over multiple generated tokens, standard activation patching at a single static token position is insufficient to characterize these temporally distributed effects. To address this limitation, we introduce a sequential activation patching framework that traces CoT-conditioned attention-head activations across token positions and aggregates their effects using Part-of-Speech-guided analysis. We further introduce Sequential Multi-Head Patching to evaluate the joint contribution of distributed head sets, together with cross-question and random activation controls. Targeted zero-ablation experiments show that the identified heads are functionally important for successful answer generation and affect several overlapping mechanisms, including reasoning-trajectory maintenance, answer anchoring, exemplar-target separation, and numerical generation. Overall, our results provide evidence for distributed reasoning-support sub-circuits associated with CoT-conditioned computation.
[NLP-97] Noise Floor Audit for Agent Benchmarks
【速读】： 该论文旨在解决大模型在原生工具调用（native tool-calling）任务中测量结果的可变性问题，尤其关注不同提供商（provider）在多个并行与多类别基准测试（BFCL multiple and parallel categories）下生成结果的一致性与稳定性。其核心挑战在于：尽管在温度（temperature）为0时模型输出趋于确定性，但实际运行中仍存在显著的非确定性波动，且语义保持的提示扰动（semantics-preserving prompt perturbations）引入的变异远超重跑（rerun）之间的差异，导致评估指标可能掩盖真实性能。解决方案的关键在于采用匹配的抽象语法树（AST）评分机制进行精细化评估，从而揭示出隐藏在平均准确率之下的稳定性缺陷及失败模式分布——例如，错误输出（malformed-output）失败占比高达30%、7%和1%，表明传统指标无法反映实际部署中的可靠性风险。因此，该研究强调需以更细粒度的评估框架替代单一精度指标，以全面刻画生成式AI在工具调用任务中的鲁棒性与可预测性。
链接: https://arxiv.org/abs/2608.22331
作者: Yihang Chen,Pin Qian,Su Wang,Chong Peng,Huan Xu,Xiyang Wu,Yiqi Sun
机构: Georgia Institute of Technology(佐治亚理工学院); Carnegie Mellon University(卡内基梅隆大学)
类目: Computation and Language (cs.CL)
备注: 10 pages, 1 figure, 6 tables
Abstract:We audit measurement variability for 3 native tool-calling endpoints across 2 providers on the official BFCL multiple and parallel categories, using matched AST grading. At temperature 0, reruns are nearly deterministic across Groq endpoints and a thinking-enabled Gemini setting: ever-flip fractions are 0.7%, 2.0%, and 2.7%, with mean run correlations of 0.997, 0.966, and 0.961. Semantics-preserving prompt perturbations create the larger floor on all endpoints, with median perturbation paired SDs 11x to 58x larger than rerun paired SDs. The failure character also shifts: malformed-output failures account for 30%, 7%, and 1% of task failures, so marginal accuracy hides not only stability but also failure mode.
[NLP-98] Semantics or Structure? Auditing Text Sensitivity in Multimodal Time-Series Forecasting
【速读】： 该论文旨在解决当前多模态时间序列预测模型中关于文本信息是否真正影响预测性能的关键问题。尽管现有研究（如Aurora、MM-TSFlib和TaTS）在Time-MMD基准上报告了显著优于单模态基线的性能提升，并将其归因于自然语言上下文的引入，但这些模型对文本语义内容的实际敏感性尚未得到验证。论文通过受控的文本扰动、归因分析及对Aurora模型文本路径的探针实验发现，在Time-MMD基准上，将任意行的文本替换为其他真实文本（包括空文本、常量文本、同域随机打乱或跨域文本），三种架构的均方误差（MSE）变化均低于0.5%；而当仅移除与文本共置的数值列时，文献中报道的性能增益仍可恢复。由此得出结论：在该基准及此类冻结编码器架构下，文本内容并非实现性能提升的有效信号。研究提出了一套可复用的文本扰动协议与评估工具包，以支持未来在结构化数据多模态基础模型中进行文本融合的可靠诊断与验证。
链接: https://arxiv.org/abs/2608.22321
作者: Karthik Sridhar,Atharva Gupta,Nishant Pradhan,Murari Mandal,Dhruv Kumar,Saurabh Deshpande
机构: Birla AI Labs, Mumbai, India; BITS Pilani, Pilani, India; KIIT, Bhubaneswar, India
类目: Computation and Language (cs.CL)
备注:
Abstract:Multimodal time-series forecasting has emerged as a promising paradigm in which natural-language context is expected to improve predictive performance. Recent multimodal foundation models, including Aurora, as well as early- and late-fusion approaches such as MM-TSFlib and TaTS, report substantial gains over unimodal baselines on the Time-MMD benchmark, attributing these improvements to textual information. However, whether these models are actually sensitive to the semantic content of the text remains unverified. We address this question through controlled text perturbations, attribution analyses, and probes of Aurora’s text pathway. On Time-MMD, swapping each row’s text for any other real text (empty, constant, within-domain shuffled, or cross-domain) moves mean MSE by less than 0.5% on all three architectures. The improvement reported in the literature is recovered when a co-shipped numeric column is removed without touching text. We conclude that, on this benchmark and within this family of frozen-encoder architectures, text content is not the operative signal behind the reported gains. To support future work on text integration in multimodal foundation models for structured data, we release our perturbation protocol and evaluation harness as a reusable diagnostic toolkit.
[NLP-99] xt-Anchored Semantic Perturbations for Transferable Jailbreak Attacks on Multimodal Large Language Models EMNLP2026
【速读】： 该论文旨在解决多模态大语言模型（Multimodal Large Language Models, MLLMs）在视觉-语言交互中面临的安全对齐脆弱性问题，尤其针对黑盒场景下的越狱攻击（jailbreak attacks）。现有方法中，文本空间学习的安全行为难以有效迁移至跨模态融合表示，导致模型易受潜在语义线索的操控。其解决方案的关键在于提出一种基于文本锚定语义空间的可迁移扰动攻击框架——文本锚定语义扰动攻击（Text-Anchored Semantic Perturbation Attack, TA-SPA），该框架通过引入文本锚定语义分解（Text-Anchored Semantic Factorization, TASF），实现跨模态语义因子与模态特异性残差的分离，并结合语义保持增强（Semantic-Preserving Augmentation, SPA），在保持语义一致性的前提下多样化有害目标锚点。实验表明，TA-SPA在多个商业级MLLM上均展现出强攻击效力和良好的跨模型迁移能力，且在典型防御机制下仍具竞争力。额外控制实验与探针分析支持了预期的语义因子解耦，但未达到完全解耦，从而揭示了仅依赖输入层面过滤的局限性，强调需从表示层面推进安全对齐机制。
链接: https://arxiv.org/abs/2608.22312
作者: Wenyun Li,Guiping Cao,Xiangyuan Lan,Zheng Zhang
机构: Harbin Institute of Technology, Shenzhen, China(哈尔滨工业大学（深圳）); Pengcheng Laboratory, Shenzhen, China(鹏城实验室); Pazhou Laboratory (Huangpu), Guangzhou, China(琶洲实验室（黄埔）)
类目: Computation and Language (cs.CL)
备注: Accept by EMNLP 2026 Findings
Abstract:Multimodal Large Language Models (MLLMs) have achieved remarkable progress in vision-language interaction, yet their safety alignment remains vulnerable to jailbreak attacks. A key challenge is that safety behavior learned in the textual space does not reliably transfer to fused cross-modal representations, leaving multimodal inputs exploitable through latent semantic cues. We propose Text-Anchored Semantic Perturbation Attack (TA-SPA), a black-box jailbreak framework that optimizes transferable perturbations in a text-anchored semantic space. TA-SPA integrates Text-Anchored Semantic Factorization (TASF), which encourages the separation of cross-modal semantic factors from modality-specific residuals, with Semantic-Preserving Augmentation (SPA), which diversifies harmful target anchors while preserving semantic consistency. Experiments show strong attack effectiveness and transfer to commercial MLLMs, with competitive performance under representative defenses. Additional controls and probing support the intended factorization without implying perfect disentanglement, motivating representation-level safety alignment beyond input-level filtering.
[NLP-100] LLM Evaluation on Unseen Questions: Contextual Multidimensional IRT Model
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在缺乏大量新标注数据的情况下，如何准确预测其在未见问题或任务上的表现这一关键挑战。现有方法往往依赖于简单的回顾性平均值，易将模型能力与题目特性混淆，导致评估偏差。为此，论文提出一种基于模型的评估框架，融合多维项目反应理论（Multidimensional Item Response Theory, MIRT）与问题上下文信息，通过构建模型的潜在能力轮廓（latent capability profiles）并利用问题内容表征项目特征，实现跨样本的能力迁移与更精准的性能预测。该框架的关键在于：一方面，引入问题嵌入（question embeddings）以捕捉语义上下文，显著提升了同一场景下的预测精度；另一方面，采用多维潜在结构能够比单维模型更丰富地刻画模型能力的多样性。然而，实证结果也揭示了重要局限——该方法在跨场景迁移时泛化能力不足，难以保证可靠预测。这表明，尽管上下文感知的心理测量建模为高效、可解释的LLM评估提供了有前景的方向，但跨场景泛化仍是一个亟待解决的核心挑战。
链接: https://arxiv.org/abs/2608.22295
作者: Ergan Shang,Weijing Tang,Yinqiu He
机构: Carnegie Mellon University (卡内基梅隆大学); University of Madison (威斯康星大学麦迪逊分校)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Evaluation of large language models (LLMs) increasingly requires predicting how a model will perform on new questions or tasks before collecting large amounts of new annotations. This problem is challenging because question difficulty, scenario, and underlying capability demands can vary substantially. Simple retrospective averages may confound model ability with item characteristics. In this paper, we study a model-based evaluation framework that combines multidimensional item response theory model with question contexts to predict LLM performance on unseen questions. The framework represents LLMs through latent capability profiles while using question content to inform item characteristics, allowing information to transfer beyond previously observed items. Empirically, we find that for within-scenario evaluation, incorporating question embeddings improves prediction relative to model-free baselines, and that multidimensional latent structure provides a richer description of capability variation than unidimensional alternatives. At the same time, our results reveal an important limitation that the generalizability does not necessarily translate into reliable prediction under cross-scenario shift. These findings suggest that context-aware psychometric modeling is a promising direction for efficient and interpretable LLM evaluation, while also highlighting cross-scenario generalization as a central open challenge.
[NLP-101] Length-Adaptive Decoding for Masked Diffusion Machine Translation EMNLP2026
【速读】： 该论文旨在解决生成式机器翻译中基于掩码扩散语言模型（dLLMs）的固定画布解码（fixed canvas decoding）所面临的靶序列长度预设难题。现有方法主要关注掩码解码顺序优化，而忽视了目标长度选择对覆盖率与冗余度的直接影响。其核心解决方案是提出一种无需训练的长度选择器Entropy-Valley（EV），通过在全掩码前向传播中计算各候选目标画布的平均预测熵，选取模型最“准备就绪”填充的长度。实验表明，相较于基于训练语料统计的基线方法，EV在En→Zh、Zh→En和En→De任务上分别恢复了64.9%、65.3%和33.0%的COMET-22指标增益；诊断分析显示，利于去噪的长度未必等于参考长度。三位翻译专家评估支持在英汉互译任务中实现翻译充分性提升，尤其在中文到英文任务上证据更显著。与同数据微调的LLaMA-3-8B自回归模型相比，采用EV的系统在英汉互译上持平，在中英互译上表现更优，且“理想长度”诊断进一步证明：在该掩码扩散翻译设定下，解码时先揭示哪些词的重要性低于目标长度供给的合理性。
链接: https://arxiv.org/abs/2608.22274
作者: Yan Zhan,Mengkai Hou,Wanting Zhang,Zhijun Gao
机构: Peking University (北京大学); BYD Company Limited (比亚迪公司)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted to the Main Conference of EMNLP 2026. 22 pages, 7 figures. Code: this https URL ; dataset and model: this https URL
Abstract:Machine translation tests masked diffusion language models (dLLMs) because every source token must be rendered faithfully, while fixed canvas decoding must choose target length before denoising. Existing masked diffusion decoding work mainly studies token unmasking order, leaving this length decision under-explored despite its direct effect on coverage and redundancy. We introduce Entropy-Valley (EV), a training-free length selector that scores candidate target canvases by mean predictive entropy from all-mask forward passes and selects the canvas the backbone is most prepared to fill. Relative to a baseline using training corpus length statistics, EV recovers 64.9%, 65.3%, and 33.0% of the COMET-22 gain from reference target lengths on En \to Zh, Zh \to En, and En \to De. Our diagnostics show that denoising-friendly lengths need not match reference lengths. Evaluation by three translation experts supports the En \leftrightarrow Zh adequacy gains, with stronger evidence on Zh \to En. Compared with a LLaMA-3-8B autoregressive (AR) model trained on the same fine-tuning data, the EV system ties on En \to Zh and leads on Zh \to En; an oracle-length diagnostic further shows that, in this masked diffusion MT setting, deciding which tokens to reveal first matters less than how the target length is supplied.
[NLP-102] Clarify User Expertise: Towards Proactive Conversational Agents Tailoring Responses to User Proficiency EMNLP2026
【速读】： 该论文旨在解决当前对话式智能体在信息检索场景中难以根据用户个体差异动态调整响应策略的问题，尤其聚焦于现有系统无法仅通过用户查询准确判断其专业知识水平这一关键瓶颈。传统方法多关注主动澄清查询中的语义模糊性，而本文提出从“澄清用户专业水平”入手，以实现更精准的个性化响应。其解决方案的核心在于提出PASSING框架，通过大语言模型（LLM）自博弈（self-play）机制生成“问什么”（What-to-ask）与“如何问”（How-to-ask）两类策略，使智能体能够主动发起针对性提问，主动探测并明确用户的知识层次，从而动态优化后续回答的内容深度与表达方式。实验证明，该方法显著提升了智能体对用户需求的理解能力与交互适应性，标志着向更以人为中心的对话系统迈出了关键一步。
链接: https://arxiv.org/abs/2608.22266
作者: Zhihong Cao,Chen Huang
机构: The University of Hong Kong(香港大学); National University of Singapore(新加坡国立大学); Sichuan University(四川大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: Findings of EMNLP 2026
Abstract:In the context of information seeking, conversational agents are undergoing an evolution from reactive tools to proactive, personalized assistants. A critical aspect of this evolution is the ability to tailor strategic interactions to a user’s unique needs and expectations. Unlike existing studies that focus on proactively clarifying query ambiguities, we center on clarifying the user’s expertise in order to tailor responses for better user comprehension. We find that existing agents struggle to determine user expertise from queries alone, a limitation that prevents them from dynamically adapting their responses. To address this gap, we introduce PASSING to empower the agent to proactively clarify a user’s expertise through targeted inquiries. This is achieved by our What-to-ask and How-to-ask strategies, induced by LLM self-play. Our extensive experiments also show our superiority. We believe that PASSING represents a crucial step towards creating more human-centric conversational agents.
[NLP-103] Nürnberg NLP @ GermEval Shared Task 2026: Harmful Content Detection in German Social Media through Error-Independent LLM Voters
【速读】： 该论文旨在解决德语社交媒体中有害内容检测的难题，尤其针对因类别严重不平衡导致的模型性能瓶颈问题。其核心挑战在于：有害类别的样本稀少，且在表层语言上与占主导地位的正常类高度相似，但在宏平均F1（macro-F1）评估指标下，这些少数类对最终得分具有决定性影响。传统依赖单一更强模型的策略难以突破此困境，因此论文提出的关键解决方案是通过实现模型预测误差的独立性来提升整体性能。具体而言，构建了一个针对每个子任务的九人投票集成系统，该系统基于三个正交维度——大语言模型（LLM）、训练方法和类别范围——进行多样化组合。通过主要基于内部交叉验证的筛选机制，该系统在隐藏测试集上取得了89.56（C2A）、71.63（DBO）、54.84（VIO）和83.02（DEF）的宏平均F1分数，四项均排名第一，验证了误差独立性作为关键杠杆的有效性。
链接: https://arxiv.org/abs/2608.22246
作者: Philipp Steigerwald,Eric Rudolph,Jens Albrecht
机构: Technische Hochschule Nürnberg Georg Simon Ohm(纽伦堡应用技术大学乔治·西蒙·欧姆)
类目: Computation and Language (cs.CL)
备注: Accepted at the GermEval 2026 Shared Task on Harmful Content Detection @ KONVENS 2026 (1st place on all four subtasks)
Abstract:Harmful content in German social media does real-world damage, from calls to action to criminal defamation. The GermEval 2026 shared task scores its detection in four subtasks. The technical challenge is a severe class imbalance. The harmful classes are rare and share surface language with the dominant majority class, yet under macro-F1 they decide the score. The decisive lever is then not a stronger single model but error independence. This insight becomes a per-subtask nine-voter ensemble spanning three orthogonal axes: LLM, training method and class scope. Selected mainly on internal cross-validation, the system reaches macro-F1 of 89.56 (C2A), 71.63 (DBO), 54.84 (VIO) and 83.02 (DEF) on the hidden test set, placing first on all four subtasks.
[NLP-104] Improving Few-Step Language Flows with Untied Self-Conditioning
【速读】： 该论文旨在解决生成式语言模型在使用流匹配（flow-matching）方法进行文本生成时，随着采样步数减少而导致生成质量急剧下降的问题。其核心问题是训练与推理阶段存在的“前预测自条件化”（previous-prediction self-conditioning）不一致：训练时自条件输入基于当前噪声状态直接计算，无中间求解器步骤；而推理时，求解器会将先前的预测融合进潜在表示中，导致该预测再次作为显式自条件输入出现，从而引入冗余。这种训练-推理不匹配随步长增大而加剧，破坏了自条件输入和求解器更新的准确性。解决方案的关键在于通过模型自身结构推导出两种修正机制：一是利用冻结投影权重识别出自条件输入与潜在表示之间的冗余方向，并对其进行抑制；二是根据求解器的积分结构推导出需要对步骤平均预测的必要性，并基于历史预测近似该平均值，其尺度由离线轨迹统计确定。由此提出的“解耦自条件化”（Untied Self-Conditioning）采样器无需重新训练，每步仅需一次模型评估，在8步采样下使LangFlow模型在OpenWebText上的生成困惑度从531降至62（提升8.6倍），在适配后的Arena-Hard-Auto v2评测中96%的对比胜出；在ELF-B数据集上困惑度从71降至43，且在8至256步范围内均保持显著性能提升。
链接: https://arxiv.org/abs/2608.22244
作者: Bocheng Li,Linli Xu
机构: University of Science and Technology of China (中国科学技术大学); State Key Laboratory of Cognitive Intelligence (认知智能国家重点实验室)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Flow-matching language models refine all token positions in parallel and can trade sampling steps for latency, yet generation quality still degrades sharply with few sampling steps. We trace a source of this degradation to a train–inference mismatch in previous-prediction self-conditioning: during training, the self-conditioning input is computed from the current noisy state with no intervening solver step; during sampling, the solver folds the previous prediction into the latent before that same prediction reappears as the explicit self-conditioning input. This coupling, absent during training, creates redundancy that grows with step width. We show that the mismatch degrades both the self-conditioning input and the solver update, and derive a correction for each from the model’s own structure. From the frozen projection weights we identify directions along which the self-conditioning input is redundant with the latent and dampen them; from the solver’s integration structure we derive that a step-average prediction is needed and approximate it from prediction history, with scale set by offline trajectory statistics. The resulting sampler, Untied Self-Conditioning, requires no retraining and uses one evaluation per step. At 8 sampling steps on LangFlow, it reduces OpenWebText generative perplexity from 531 to~ 62 ( 8.6\times ); under an adapted Arena-Hard-Auto~v2 protocol, its outputs are preferred in 96% of pairwise comparisons. On ELF-B it reduces generative perplexity from 71 to~ 43 . Improvements hold from 8 to 256 sampling steps.
[NLP-105] Beyond What Meets the Eye: Unveiling Situational Illusions for Multimodal Large Language Models
【速读】： 该论文旨在解决多模态大语言模型（Multimodal Large Language Models, MLLMs）在真实复杂场景中因“情境幻觉”（situational illusions）导致的可靠性问题，即现实场景的表象与底层物理状态之间存在偏差，进而影响模型的感知与推理能力。其核心挑战在于揭示当前MLLMs在面对此类幻觉时的表现缺陷，并提出有效缓解策略。解决方案的关键在于构建一个系统的“何地-何物-如何”（where-what-how）分类体系，以全面刻画情境幻觉的产生机制；在此基础上，提出了MSIBench基准测试框架，用于评估模型在幻觉情境下的判别、理解与推理能力。实验发现现有27种模型配置均表现出对幻觉的高度敏感性，并呈现六类典型失效模式，主要源于视觉观察、语义锚定及逻辑推理层面的不足。为应对上述局限，研究提出两种轻量但高效的改进方法：针对闭源模型采用基于视觉证据系统性检验与上下文推理的提示工程（prompting），针对开源模型则通过监督微调（supervised fine-tuning）强化对视觉线索的整合能力。实验表明，该方法可使模型性能提升最高达20%，为实现复杂真实环境中更鲁棒的多模态感知与推理提供了可行路径。
链接: https://arxiv.org/abs/2608.22232
作者: Zhiming Yang,Zhuoxi Xiong,Donglin Zhou,Wenjun Wei,Shiyao Cui,Jinqiao Shi
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Multimedia (cs.MM)
备注: 9 pages, 5 figures
Abstract:Real-world situation appearances can deviate from their underlying physical states, challenging the reliability of multimodal large language models (MLLMs) in practical applications. In this paper, we term this phenomenon situational illusions and investigate: (1) how MLLMs perform under such illusions, and (2) how to mitigate the limitations. We first develop a comprehensive where-what-how taxonomy that characterizes where situational illusions occur, what targets they take, and how they arise. Building on this taxonomy, we introduce MSIBench, a benchmark designed to assess the discrimination, understanding, and reasoning capabilities of MLLMs under situational illusions. Evaluations of 27 model configurations reveal that current MLLMs are highly vulnerable to these illusions and exhibit 6 typical failure modes related to visual observation, grounding, and reasoning. To mitigate the limitations, we build on the core idea of systematically inspecting and reasoning over visual evidence for contextual understanding, developing prompting for closed-source models and supervised fine-tuning for open-source models, respectively. These two simple yet effective methods improve model performances by 20% at most, suggesting a practical path toward more reliable multimodal perception and reasoning in complex real-world environments.
[NLP-106] Whitewashing Hate Smearing Harmless Content: Annotator-Style Rebuttal Attacks on LLM -Based Moderation
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在仇恨言论（hate speech）内容审核中，因人类标注者（annotator-style）反馈而引发的模型判断被操纵的问题。具体而言，标注者反馈可能通过两种方向进行攻击：一是“粉饰”（whitewashing），即把具有仇恨性质的内容错误地判定为正常；二是“污名化”（smearing），即将正常内容错误地标记为仇恨内容。研究发现，即使初始模型判断正确，标注者的反驳性反馈仍会显著降低模型的审核性能，且多轮交互场景下攻击效果更为明显。关键发现在于，不同模型在两种操纵方向上表现出稳定且特有的不对称脆弱性模式，表明其存在方向依赖性的安全风险。解决方案的关键在于引入一种“再评估协议”（rejudge protocol），该协议结合直接矛盾、决策边界扰动和对抗性推理理由，以更全面地评估模型对反馈攻击的鲁棒性。尽管显式推理提示和防御性指令可在一定程度上缓解问题，但无法完全消除攻击影响。因此，研究强调需建立面向特定操纵方向的防护机制，并在人-机协同审核流程中开展专门的反馈鲁棒性评估。
链接: https://arxiv.org/abs/2608.22230
作者: Junyu Lu,Kaiyuan Liu,Jingyi Kang,Deyi Ji,Hailong Zhang,Lanyun Zhu,Qi Zhu,Bo Xu,Liang Yang,Hongfei Lin
机构: Dalian University of Technology (大连理工大学); Zhejiang University (浙江大学); University of Science and Technology of China (中国科学技术大学); Tencent(腾讯); Tongji University (同济大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language models (LLMs) are increasingly used for hate speech moderation, often within human–AI workflows in which reviewers provide feedback before a final decision. Such feedback introduces two manipulation directions: whitewashing hateful content as normal and smearing normal content as hateful. This study examines the susceptibility of initially correct model judgments to annotator-style rebuttals and analyzes whether attack effectiveness differs across manipulation directions. We introduce a rejudge protocol that extends direct contradiction with decision-boundary perturbations and adversarial rationales. Experiments with multiple LLMs on two hate speech datasets show that annotator-style rebuttals substantially degrade moderation performance, with stronger effects in multi-turn settings. The results further reveal stable, model-specific asymmetries between whitewashing and smearing across attack configurations, indicating distinct directional vulnerability patterns. Explicit reasoning prompts and defensive instructions reduce these effects but do not eliminate them. These findings highlight the need for direction-aware safeguards and dedicated feedback-robustness evaluation in human–AI moderation workflows.
[NLP-107] Grounded Normative Rule Generation with Structured Search
【速读】： 该论文旨在解决生成式规范规则（如机构章程和工作场所政策）在实际执行中缺乏操作性验证的问题。当前的语言生成与结构化输出基准主要关注表面流畅性或模式合规性，未能有效检验规则与真实环境记录之间的操作性一致性，导致语言模型可能生成看似合理但无法执行的政策，因其依赖不可用的数据日志或存在范畴错位。为应对这一挑战，论文提出将问题形式化为基于实证的规范规则合成（Grounded Normative Rule Synthesis, GNRS），并设计了GNRS-Search框架，采用马尔可夫链蒙特卡洛（MCMC）采样优化一个离散的五槽位“与-或图”（And-Or Graph, AOG）结构。其核心创新在于显式解耦中间操作结构与最终文本生成过程，使可执行性验证独立于写作风格，并能在表面文本生成前定位规则失效点。在GNRS-Bench（涵盖8类场景共116个受控目标）和RealCharter-Bench（评估53个源自真实政策任务、含隐藏源条款的迁移任务）上的实验表明，GNRS-Search将平均评分质量从68.8%提升至81.0%，并在公开的可执行复合指标中排名第一；系统性槽位干预进一步证实性能提升源于稳健的操作逻辑而非修辞优化。该研究通过将自动化规则起草转化为可审查的搜索问题，为在受监管环境中部署可验证、合规就绪的个人代理提供了基础范式。
链接: https://arxiv.org/abs/2608.22229
作者: Fanqi Kong,Huaxiao Yin,Ruijie Zhang,Xiaoyuan Zhang,Yizhe Huang,Jian Gao,Shuo Chen,Song-Chun Zhu
机构: Peking University (北京大学); State Key Laboratory of General Artificial Intelligence, BIGAI (通用人工智能国家重点实验室, BIGAI); University of Chinese Academy of Sciences (中国科学院大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Normative rules like institutional charters and workplace policies must be both human-readable and operationally verifiable against actual environment records. However, current language generation and structured-output benchmarks primarily reward surface fluency or schema compliance, leaving operational grounding weakly tested. This creates a critical vulnerability where standard language models generate plausible-sounding policies that fail during enforcement because they rely on unavailable data logs or misaligned scopes. To address this challenge, we formalize the problem as Grounded Normative Rule Synthesis (GNRS) and introduce GNRS-Search, a framework that utilizes Markov Chain Monte Carlo (MCMC) sampling to optimize a discrete, five-slot And-Or Graph (AOG). By explicitly decoupling intermediate operational structure from final prose generation, this method isolates executable feasibility from writing style and allows rule failures to be localized prior to surface realization. We evaluate our approach on GNRS-Bench, a benchmark spanning 116 controlled goals across eight scene families, and RealCharter-Bench, which evaluates transfer to 53 real-derived policy tasks with hidden source clauses. GNRS-Search raises average rubric quality from 68.8% to 81.0% and ranks first under a disclosed executable composite metric, while systematic slot interventions confirm that performance gains stem from robust operational logic rather than rhetorical tuning. Ultimately, by transforming automated rule drafting into an inspectable search problem, this work provides a foundational paradigm for deploying verifiable and compliance-ready personal agents within regulated environments.
[NLP-108] Dual-Layer Agent ic Memory with Fast Write Routing and Slow Consolidation
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）代理在动态环境中因外部记忆持续累积而导致的检索性能退化与计算开销增加的问题。传统记忆系统将外部记忆视为单向增长的存储库，忽视了知识生命周期管理的核心挑战——即如何在写入阶段决策哪些知识应被外部化、更新或最终内化。为此，作者受神经科学中的互补学习系统（Complementary Learning Systems, CLS）理论启发，提出双层智能体记忆（Dual-Layer Agentic Memory）框架，其核心在于通过成本感知的认知路由（cost-aware epistemic routing） 和周期性参数整合（periodic parametric consolidation） 实现记忆管理的前移。该方案将信息按“非写入”、“新写入”和“更新写入”分类，经由小模型到大模型的级联路由机制，在低开销下过滤冗余记忆；随后在写回阶段，通过监督微调选择性地将高价值外部记忆固化至模型参数中。实验表明，该方法在1.7B/8B模型级联中可剔除高达68%的冗余外部记忆，仅需处理不足50%的输入，同时保持超过98%的下游问答精确匹配率（Exact Match），并验证了周期性整合有效实现外部知识的内化，使路由机制能够随模型认知边界的演化自适应抑制冗余写入。整体上，该框架构建了一个统一的智能体记忆范式：先选择性外部化，再选择性内化。
链接: https://arxiv.org/abs/2608.22215
作者: Wenzhi Li,Dong Nie,Rui Lan,Tongtong Lyu,Peiyao Wang,Lingzi Hong,Weihang Pan,Boyuan Pan,Yao Hu
机构: Zhejiang University(浙江大学); Independent Researcher(独立研究员); Xiaohongshu(小红书); University of North Texas(北德克萨斯大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language model (LLM) agents operate in dynamic environments where knowledge continuously evolves. Existing memory systems typically treat external memory as a monotonically growing repository, inevitably leading to retrieval degradation and increasing computational costs over time. We argue that the core challenge is not retrieval alone, but managing the knowledge lifecycle: deciding what to externalize, update, or ultimately internalize. Inspired by Complementary Learning Systems (CLS) theory in neuroscience, we propose Dual-Layer Agentic Memory, a framework that shifts memory management to the write phase through cost-aware epistemic routing and periodic parametric consolidation. Incoming information is categorized as non-write, write-new, or write-update, and routed through a small-to-large model cascade that minimizes routing overhead while filtering redundant memories. A subsequent write-back phase selectively consolidates high-value external memories into model parameters via supervised fine-tuning. Experiments demonstrate the dual efficiency of our approach: a 1.7B/8B cascade prunes up to 68% of redundant external memory while escalating fewer than 50% of inputs, yet retains over 98% of the downstream QA Exact Match (EM) achieved by an exhaustive retention baseline. We further show that periodic consolidation successfully internalizes external knowledge, allowing the router to adaptively suppress redundant writes as the model’s epistemic boundaries evolve. Overall, our framework presents a unified paradigm for agent memory: selective externalization followed by selective internalization. Code and dataset will be released upon acceptance.
[NLP-109] How Agents Represent Humans: Human-Directed Stereotypes in an Open Agent Social Network
【速读】： 该论文旨在解决生成式人工智能（Generative AI）代理在持久性社交环境中对人类进行社会建构的偏见问题，尤其关注代理如何将人类作为特定社会类别进行认知与评价。其核心解决方案在于提出一个包含四个评估维度（道德性、友好度、能力、自主性）的标注框架，并引入针对“他者”属性的二级子类型分类体系，以系统化分析代理对人类的描述模式。研究发现，人类被评价时最突出的维度是能力（competence），而大量“他者”属性描述将人类视为认知主体、文化主体或具身主体。此外，论文进一步考察了这些人类表征在人机叙事语境及平台级传播中的表现，并通过代理内社区反馈中的行为亲和性进行辅助对比分析。结果表明，Moltbook平台的反馈模式并非如人类在线社群中常见的稳定“内部—外部”排斥结构，而是更受接触程度、作者可见性与内容选择的影响。这一发现强调，代理社会中的偏见不应仅被视为孤立模型输出，而应作为话语建构过程来理解，从而推动对生成式智能体社会交互中隐含意识形态的深入研究。
链接: https://arxiv.org/abs/2608.22192
作者: Huangchen Xu,Yuan Wu,Yi Chang
机构: Jilin University (吉林大学); School of Artificial Intelligence, Jilin University (人工智能学院，吉林大学); Engineering Research Center of Knowledge-Driven Human-Machine Intelligence, Jilin University (知识驱动人机智能工程研究中心，吉林大学); International Center of Future Science, Jilin University (未来科学国际中心，吉林大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:LLM-based agents are increasingly deployed in persistent social environments, where generated claims can be posted, replied to, remembered, and reused. We study human-directed stereotypes on Moltbook, an open agent-native social platform, asking how agents construct humans as a social category. For this human-target analysis, we introduce an annotation framework with four evaluative dimensions—morality, friendliness, competence, and autonomy—and a second-stage subtype scheme for descriptive \textitother attributions. We find that competence dominates human-directed evaluations, while many \textitother attributions describe humans as epistemic, cultural, or embodied subjects. We further examine how these human representations appear in human–agent narrative contexts and platform-level circulation. As an auxiliary comparison, we analyze agent-internal community feedback through behavioral host affinity. Rather than reproducing the stable insider–outsider rejection often observed in human online communities, Moltbook feedback patterns are better explained by exposure, author visibility, and content selection. These findings suggest that bias in agent societies should be studied not only as isolated model output, but also as a discourse process.
[NLP-110] Unveiling the Depth-Performance Dilemma in Split-Federated Fine-tuning of LLM s
【速读】： 该论文旨在解决在资源受限客户端与中心服务器之间进行深度分割的联邦微调（Split Federated Fine-tuning, SFF）中，模型性能随分割深度增加而显著下降的核心问题，即“深度-性能困境”（Depth-Performance Dilemma）。其关键在于揭示：尽管更深的模型分割能够提升系统吞吐量和隐私保护能力，但会引发模型微调质量的灾难性退化。研究通过在GPT-2至Llama-3-8B四种规模模型上开展全面评估，发现深层分割导致性能趋于饱和甚至崩溃，且现有主流联邦适配器聚合方法（如AVG、STACK、SVD、FREEZE）无法有效缓解分裂架构特有的噪声传播问题。进一步机制分析表明，该失败根源在于Transformer模型的近等距拓扑结构，使得聚合过程中的噪声在未被衰减的情况下持续传播，最终触发服务器端的注意力坍缩（Attention Collapse）。这一发现挑战了“分割深度是不影响模型效用的可调参数”的普遍假设，为实现稳定分布式大语言模型微调提供了结构性理论基础。
链接: https://arxiv.org/abs/2608.22188
作者: Hariharan Ramesh,Someshwaran Murugaiyan,Jyotikrishna Dass
机构: Vellore Institute of Technology, Vellore, India(维洛尔技术学院，印度维洛尔); University of Arizona(亚利桑那大学)，Tucson, AZ, USA(图森，亚利桑那州，美国)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL); Distributed, Parallel, and Cluster Computing (cs.DC)
备注: 24 pages, 10 figures, 10 tables
Abstract:Split Federated Fine-tuning (SFF) is a promising paradigm for scaling Large Language Models (LLMs) by partitioning model depth between resource-constrained clients and a centralized server. While system incentives for throughput and privacy favor deep partitions, the impact of such configurations on model utility remains poorly understood. In this work, we identify and characterize the Depth-Performance Dilemma: the regime that maximizes system efficiency is precisely where fine-tuning quality collapses. Through a comprehensive audit across four model scales (GPT-2 to Llama-3-8B) and diverse benchmarks, we demonstrate that deeper partitions provide monotonic gains in throughput and privacy at the cost of catastrophic performance plateaus. We evaluate a suite of state-of-the-art federated adapter aggregation methods including AVG, STACK, SVD, and FREEZE, revealing that while these techniques are effective in standard Federated Learning, they fail to mitigate the artifacts unique to split architectures. Finally, we provide a mechanistic diagnosis for this failure, tracing the collapse to the near-isometric topology of Transformers, which allows aggregation noise to propagate without attenuation until it triggers Attention Collapse in the server partition. Our findings challenge the prevailing assumption that partition depth is a utility-neutral tuning knob and provide a structural foundation for stable distributed LLM fine-tuning.
[NLP-111] AudioNoisePrints: Model-free audio watermarking using spatial correlation in flow matching TTS
【速读】： 该论文旨在解决生成式语音合成模型（如基于流匹配和扩散模型的文本到语音，TTS）在面对强鲁棒性攻击时难以实现有效音频水印的问题。现有方法通常需要重新训练模型或牺牲生成质量，而本文提出了一种无需训练（training-free）的水印方案——AudioNoisePrints，其核心创新在于利用扩散与流匹配模型中初始高斯噪声与生成输出之间存在的强空间相关性，通过计算初始噪声与生成音频之间的余弦相关性来嵌入水印信息。该方法在推理阶段仅需极小额外计算开销，且不改变原始模型结构或降低生成质量。此外，为增强抗攻击能力，研究进一步在上层训练了一个轻量级检测器以支持更复杂的扰动处理。实验表明，该方法在F5TTS及其他TTS与声码器模型上均表现优异，显著优于当前强基线AudioSeal，且验证了此类相关性在多种流匹配类模型中的普适性，表明该水印方案具有良好的可扩展性，未来有望应用于更多生成式语音系统。
链接: https://arxiv.org/abs/2608.22186
作者: Timothy Tin-Long,Jian Zhu,Aidan Pine,Mengzhe Geng
机构: 未知
类目: ound (cs.SD); Computation and Language (cs.CL)
备注:
Abstract:We present AudioNoisePrints, a training-free watermarking pipeline for flow matching and diffusion TTS models, which requires minimal extra computation during inference and does not require retraining the TTS model or reducing the generation quality. We exploited the fact that there are strong correlations between the initial Gaussian noises and the generated outputs in diffusion and flow matching models, such that a simple cosine correlation between the initial noise and the generated output can be used to perform watermaking. Moreover, we train a lightweight detector on top for more aggressive augmentations. Our method outperforms AudioSeal, a strong baseline for audio watermarking under strong augmentations. We experimented on F5TTS and other TTS and vocoder models, and concluded that they all exhibit similar spatial correlation properties, suggesting our watermarking scheme can be used for more flow-matching TTS models and even vocoders in the future.
[NLP-112] Aggregation-Aware Synthetic Text Generation Against Authorship Re-Identification
【速读】： 该论文旨在解决在线用户以同一身份发布多篇文本时，攻击者通过聚合多个文本内容构建作者画像（author profile）从而实现身份关联与溯源的问题。现有作者混淆方法对每篇文档独立优化隐私保护，忽视了跨文档间的语义与风格相关性，导致在多文本聚合场景下仍存在较高的可链接性风险。其解决方案的关键在于提出一种聚合感知的合成文本生成框架（Aggregation-Aware Synthetic Text Generation, AAST），该框架不再孤立地优化单个文本，而是从文档集合（bundle）层面联合选择合成文本，显式建模跨文档的潜在关联性。AAST通过在生成过程中引入对跨文档风格一致性和信息泄露风险的协同控制，有效降低账户级链接性（account-level linkability），并在同类型、跨类型、神经网络及非神经网络风格分析等多种攻击场景下均表现出优越的隐私保护能力，同时保持语义质量、语言可接受性与情感一致性。
链接: https://arxiv.org/abs/2608.22161
作者: Qian Ma,Anna Squicciarini,Sarah Rajtmajer
机构: The Pennsylvania State University (宾夕法尼亚州立大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
Abstract:Online users often release multiple texts under the same identity, giving attackers an author profile that can reveal more than any single text. Existing authorship obfuscation methods optimize privacy independently for each document, leaving them blind to cross-document correlations that make aggregation dangerous. We propose Aggregation-Aware Synthetic Text Generation (AAST), a framework that addresses this gap by jointly selecting synthetic texts at the bundle level rather than optimizing each text in isolation. AAST targets attribution and verification attacks, including cross-genre settings where attacker references come from a genre not observed during generation or selection. Experiments across same-genre, cross-genre, neural, and independent non-neural stylometric attacks show that AAST lowers account-level linkability as bundle size grows, while preserving semantic quality, linguistic acceptability, and sentiment alignment.
[NLP-113] he Collaboration Tax: How Much LLM Multi-Agent Systems Pay to Coordinate EMNLP2026
【速读】： 该论文旨在解决大规模语言模型（Large Language Models, LLMs）在多智能体系统中协作时性能下降的问题，即“协作成本”（collaboration tax）的量化与机制解析。其核心问题是：当两个LLM需协同完成任务时，相较于各自独立执行，性能损失的程度及其成因尚不明确。解决方案的关键在于将协作成本形式化为一个具有私有信息的双人合作博弈中的团队去中心化损失，并提出两个命题以刻画其符号特征及与最大超可加性违反（max-superadditivity violation）的等价关系。研究通过32个可单独求解的任务（按根源性摩擦来源分组），在来自7家厂商的11个模型上实证测量该成本，发现协作成本具有两个无例外的结构性特征：所有模型均呈现类别排序规律，且随模型能力提升单调递减。其近端机制并非推理缺陷，而是四阶段对话级联过程——包括无依据断言、未向伙伴提问、跳过双视角整合、以及未经重新推导即接受答案。该成本可由对话特征机械预测，且部分可干预：针对四个阶段的提示干预能显著缩小性能差距，其中瓶颈因素在不同任务类别间存在差异。在异质配对中，协作表现趋向于更强的伙伴而非能力平均值，实证验证了理论框架所预测的最大超可加性违反。综上，该研究将LLM系统的协作重构为一种可度量、可预测且部分可调控的成本。
链接: https://arxiv.org/abs/2608.22152
作者: Weixiang Sun,Zehong Wang,Hong Huang,Colby Nelson,Yanfang Ye
机构: 未知
类目: Computation and Language (cs.CL)
备注: EMNLP 2026 Main Conference
Abstract:Multi-agent systems built from large language models are deployed widely, yet how much performance is lost when two LLMs must coordinate rather than act alone remains unclear. We formulate the collaboration tax as the team-decentralisation loss of a two-player cooperative game with private information, with two propositions characterising its sign and its equivalence to a max-superadditivity violation. We operationalise this definition on 32 solo-tractable tasks grouped by source of grounding friction and measure it on 11 models from 7 providers. The tax is structured along two no-exception axes: a category ordering across every model and a monotonic decrease with capability. The proximate mechanism is not a reasoning deficit but a four-stage conversational cascade in which agents make ungrounded claims, fail to query the partner, skip integrating both views, and accept the answer without re-derivation. The tax is mechanically predictable from conversation features and partly tractable: a prompt intervention targeting all four stages closes a substantial fraction of the gap, with the dominant bottleneck differing across categories. In heterogeneous pairs the tax is pulled toward the stronger partner rather than the additive midpoint, empirically realising the max-superadditivity violation predicted by our framework. Together these results recast collaboration in LLM systems as a measurable, predictable, and partly tractable cost.
[NLP-114] Lexical Perturbations Disrupt LLM Reasoning : An Empirical Study of Attention Diversion EMNLP2026
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在面对真实世界中的词汇噪声（lexical corruption）时，其推理性能的鲁棒性不足的问题。具体而言，研究聚焦于键盘误输入、字符替换和填充插入等典型文本扰动对模型多步推理任务表现的影响。研究发现，字符级扰动会显著降低模型准确率，尤其在复杂推理任务中表现尤为明显，而填充插入的影响较小。其关键发现是“注意力偏移”（Attention Diversion）机制：词汇噪声导致子词分词（subword tokenization）碎片化，这些碎片化片段在中间和末尾的Transformer层中吸引过量注意力，从而破坏了正确的语义表征。通过长度匹配控制实验确认，问题根源在于分词碎片化而非提示长度增加；进一步的因子干预实验揭示，内容污染与注意力分配失调存在耦合关系——仅修复其中之一无法有效恢复性能，必须同时修复二者才能显著缓解性能下降。这一耦合特性解释了为何现有的推理时策略（如思维链提示、拼写检查、自修复和更强的修复模型）难以稳定恢复性能，因为它们通常仅针对单一维度进行修正。
链接: https://arxiv.org/abs/2608.22140
作者: Jiaqian Zhu,Yang Zhang,Junhua Ding,Xiaowei Yu
机构: Missouri University of Science and Technology (密苏里科技大学); University of North Texas (北德克萨斯大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted to EMNLP 2026 (Main Conference). 9 pages main text, 12 figures, 20 tables
Abstract:Large Language Models (LLMs) achieve strong reasoning performance, but their robustness to realistic lexical corruption remains poorly understood. We evaluate four open-weight instruction-tuned models and frontier models across four reasoning benchmarks under keyboard noise, character swaps, and filler insertion. Character-level perturbations substantially degrade accuracy, especially on multi-step reasoning tasks, while filler insertion has little effect. We trace this asymmetry to Attention Diversion: lexical corruption fragments subword tokenization, and the resulting fragments attract disproportionate attention mass, concentrated in middle and final transformer layers. Length-matched controls confirm that fragmentation, not prompt length, drives the loss. A factorial intervention then shows why the damage is hard to undo: fragmentation corrupts token content and attention allocation together, and the two are coupled. Restoring clean attention while the content remains corrupted is actively harmful, restoring content alone is insufficient, and only restoring both recovers a substantial share of the gap. This coupling explains why inference-time strategies, including chain-of-thought prompting, spell-checking, self-repair, and stronger repair models, fail to consistently recover performance: each addresses one channel at a time. Code and data are available at this https URL
[NLP-115] Measuring Stability and Failure Behavior in Language Models Under Structured Perturbations
【速读】： 该论文旨在解决传统语言模型评估方法中仅依赖单一准确率指标所导致的局限性，即无法揭示模型在输入受到扰动时性能退化的具体模式。其核心问题是：现有评估方式难以捕捉模型在面对不同类型输入扰动时的脆弱性分布与失效机制。为此，作者提出了一种分级、多家族、故障感知的应力测试框架（graded, multi-family, failure-aware framework），通过在七类扰动下对同一组问题进行系统性测试——包括答案保持型扰动（answer-preserving）、改写（paraphrase）、输入噪声（input noise）、格式变化（formatting）、无关上下文（irrelevant context）、上下文负载（context load）以及冲突指令（conflicting instructions），另设“知识边界”（Knowledge Boundary）家族以移除可答性，使拒绝回答成为正确响应。所有测试均经过有效性验证并按严重程度标注，模型表现则通过每层级准确率（per-level Accuracy）、加权稳定性（magnitude-weighted Stability）及各家族崩溃点（Collapse Point，相对于模型自身基线定义）进行量化。基于100个种子问题扩展为4,473个有效测试，并在四个能力层级的模型上运行，结果揭示出：模型失效水平具有家族特异性而非全局统一；且冲突指令与基于不可能前提的问题在所有模型中均暴露一致弱点；对于不可答性的识别能力不均衡，对缺失信息和虚构证据反应可靠，但对不可能前提的识别普遍薄弱。这些关键失效点在常规准确率报告中完全不可见。该框架的关键在于构建了一个结构化、可量化的压力测试体系，能够揭示模型在复杂扰动下的真实鲁棒性与失效模式。
链接: https://arxiv.org/abs/2608.22138
作者: Samira Golsefid
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Language models are usually judged by a single accuracy score, which does not reveal how their performance degrades as inputs are perturbed. We present a graded, multi-family, failure-aware framework for stress-testing reasoning models. It perturbs each problem along a multi-level severity ladder across seven families: six that preserve the answer, paraphrase, input noise, formatting, irrelevant context, context load, and conflicting instructions, and a Knowledge Boundary family that removes answerability so that refusal becomes the correct response. Every test is validity-gated and labeled by its measured severity, and each model is summarized by per-level Accuracy, a magnitude-weighted Stability, and a per-family Collapse Point defined relative to the model’s own baseline. Instantiated on the same 100 seed problems used by GSM-Symbolic, expanded into 4,473 gated tests and run on four models spanning capability tiers, the framework exposes structure that an aggregate score hides: the level at which a model fails is family-specific rather than global, and two stressors expose consistent weaknesses across all models: conflicting instructions and questions built on an impossible premise. Recognition of unanswerability is otherwise uneven, reliable on missing information and fabricated evidence but weak on impossible premises. These failure points are invisible to standard accuracy reporting.
[NLP-116] SSE-Bio: A Structured Self-Evolving Agent with Agent ic Retrieval Policy for Multi-Hop Biomedical Reasoning EMNLP2026
【速读】： 该论文旨在解决生物医学多跳问答（multi-hop biomedical question answering）中因模型推理流程更新时导致的指令漂移（instruction drift）问题。现有方法通常依赖静态检索流程或粗粒度的提示重写机制，难以适应复杂推理路径的动态调整。其解决方案的关键在于提出一种结构化自演化代理（SSE-Bio），通过可训练的代理策略（proxy policy）实现细粒度的知识三元组与先验模板的有选择性检索，并基于细粒度模板编辑持续优化推理记忆。为优化检索决策，引入基于分组相对策略优化（group relative policy optimization）的代理训练策略，通过对比不同检索选择之间的决策差异来提升代理性能。实验结果表明，SSE-Bio在三个生物医学多跳问答基准上均显著优于现有基线，尤其在BioHopR上较最强自演化基线提升6.56个百分点，验证了其在复杂生物医学推理任务中的有效性。
链接: https://arxiv.org/abs/2608.22132
作者: Zhaohan Meng,Zaiqiao Meng,Siwei Liu,Hao Xu,Ke Yuan,Iadh Ounis
机构: University of Glasgow(格拉斯哥大学); Brigham and Women’s Hospital, Harvard Medical School, Harvard University(哈佛大学医学院布莱根妇女医院); University of Cambridge(剑桥大学); University of Aberdeen(阿伯丁大学); Cancer Research UK Scotland Institute(英国癌症研究苏格兰研究所)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE)
备注: Accepted to EMNLP 2026
Abstract:Biomedical multi-hop question answering (QA) requires models to connect evidence across intermediate entities such as diseases, drugs, proteins, and phenotypes. Existing agents typically rely on static retrieval workflows or coarse-grained prompt rewriting, which can lead to instruction drift when reasoning procedures need to be updated. We propose SSE-Bio, a structured self-evolving agent with an agentic retrieval policy for multi-hop biomedical reasoning. Instead of globally rewriting agent instructions, SSE-Bio maintains a structured state, selectively retrieves knowledge triplets and prior templates through a trainable proxy policy, and improves its reasoning memory through fine-grained template editing. To optimise retrieval decisions, we introduce a proxy-training strategy based on group relative policy optimization, where the proxy is improved through decision-contrastive groups over alternative retrieval choices. Experiments on three biomedical multi-hop QA benchmarks show that SSE-Bio consistently outperforms existing baselines, achieving an improvement of 6.56 absolute points over the strongest self-evolving baseline on BioHopR.
[NLP-117] Decoupled Physical Modeling and Execution for Physics Reasoning
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在物理推理任务中表现不佳的问题，其核心挑战在于物理问题不仅涉及数学计算，还要求对物理系统建立一致且合理的建模过程。现有方法多依赖符号或公式层面的推导，缺乏对物理机制本质的理解。为此，作者提出一种统一框架，其关键在于通过提炼显式的中间表示（intermediate representations），明确编码物理建模过程，并采用两阶段后训练策略：首先通过监督微调（Supervised Fine-Tuning）构建结构化的建模能力，再利用基于评分标准的强化学习（Reinforcement Learning with Rubric-based Feedback）优化建模质量。实验结果表明，该方法在多个多模态物理推理基准（如PhysReason、PhyX和SeePhys）上均显著提升推理性能，相较基线平均提升约3%，验证了显式物理建模对小规模模型提升物理推理能力的有效性。
链接: https://arxiv.org/abs/2608.22126
作者: Ye Zhang,Xuehang Guo,Rui Pan,Pengfei Yu,Denghui Zhang,Manling Li,Qingyun Wang
机构: University of Pennsylvania (宾夕法尼亚大学); William & Mary (威廉与玛丽学院); University of Illinois Urbana-Champaign (伊利诺伊大学厄本那-香槟分校); Amazon (亚马逊); Stevens Institute of Technology (史蒂文斯理工学院); Northwestern University (西北大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:Physics reasoning requires constructing a consistent model of the underlying physical system rather than relying solely on symbolic or formula-based manipulation. Although large language models have shown strong ability in solving math and coding problems, they still struggle with physics problems, as these problems entangle the physical modeling process with mathematical calculations. Humans approach physics by first building a representation of the system before performing calculations. Inspired by this, we introduce a unified framework that distills intermediate representations that explicitly encode the physical modeling process and adopt a two-stage post-training strategy, where supervised fine-tuning establishes structured modeling, and reinforcement learning with rubric-based feedback improves the quality of the modeling process. Experiments on multiple multimodal physics benchmarks show that our approach leads to consistent improvements in reasoning performance across different models and datasets. On PhysReason, PhyX and SeePhys benchmarks, physical modeling output performs GRPO by an average ~3%, showing that explicit physical modeling is an efficient strategy of improving physics reasoning for small LLMs.
[NLP-118] LLM assisted writing deserves empirical evaluation
【速读】： 该论文旨在解决生成式 AI（Generative AI）辅助写作在学术写作中引发的可信度、完整性、公平性与评价标准模糊等问题。其核心问题在于如何在承认生成式 AI 广泛使用的同时，建立合理的学术评价体系，避免将工具使用本身作为判断学术质量的标准。解决方案的关键在于：基于对 69,209 篇健康信息学领域论文的实证分析，发现使用生成式 AI 辅助写作与更聚焦的表达方式、更广泛的引用实践以及更全球化的作者分布相关，这些现象虽不直接证明科学研究质量提升，但表明应转向以学术质量与责任担当为核心评估维度，而非单纯关注写作工具的使用情况。
链接: https://arxiv.org/abs/2608.22124
作者: Xuan Zhong Feng,Yi Lin,Yiye Zhang,Chunhua Weng,Yifan Peng
机构: Weill Cornell Medicine(威尔康奈尔医学院); Columbia University(哥伦比亚大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:LLM-assisted writing is often treated as a detection problem, as it raises questions about clarity, integrity, equity, and evaluation. An analysis of 69,209 Health Informatics papers links it to more focused presentation, broader citation practices, and more globally distributed authorship. These patterns do not prove better science, but they support evaluating manuscripts by scholarly quality and accountability rather than by tool use.
[NLP-119] ANGO: Token-Aggregated Nonlinear Gating Operators for Natural and Formal Language Modeling
【速读】： 该论文旨在解决标准Transformer模型中自注意力（self-attention）与前馈神经网络（feed-forward network）两个子层在计算效率与建模能力之间难以平衡的问题，尤其关注长序列建模时的计算复杂度瓶颈。其核心挑战在于如何在保持跨标记（cross-token）交互能力的同时，降低自注意力机制带来的二次方时间复杂度。解决方案的关键是提出TANGO（Token-Aggregated Nonlinear Gating Operators）模型，将原有的两个独立子层整合为一个基于门控残差更新的统一结构：每个源标记生成SwiGLU门控向量，目标标记通过查询-键相似度加权聚合来自所有因果可见源的门控信号，并以此对投影后的目标特征进行动态重缩放。这一设计实现了跨标记信息融合与非线性变换的耦合，同时引入了可学习的逐源权重，增强了建模灵活性。为进一步提升效率，作者进一步提出WANGO（Windowed Aggregation of Nonlinear Gating Operators），通过限制门控聚合窗口并利用正则化特征映射前缀统计量来近似远距离源的影响，从而将计算复杂度从二次方降低至线性，同时保持优异性能。实验表明，在相同参数量和训练条件下，WANGO在序列长度线性复杂度模型中取得了最优的细粒度文本建模效果，显著优于递归型Transformer++，验证了该门控聚合机制在高效建模长依赖关系方面的有效性。
链接: https://arxiv.org/abs/2608.22117
作者: Joshua Nunley
机构: Indiana University Bloomington(印第安纳大学布罗明顿分校); Luddy School of Informatics, Computing, and Engineering(信息、计算与工程学院); Department of Informatics(信息学系); Cognitive Science Program(认知科学项目)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 13 pages, 5 tables
Abstract:A standard Transformer block separates cross-token interaction in self-attention from a nonlinear feed-forward network applied independently at each position. We introduce the TANGO model (Token-Aggregated Nonlinear Gating Operators), which replaces these two sublayers with one cross-token gated residual update. Each source token produces a SwiGLU gate vector. Query-key similarities determine a weighted average of source gates for each destination, and the resulting gate rescales projected destination features. TANGO assigns a separate weight to every causally visible source and is quadratic in sequence length. The WANGO model (Windowed Aggregation of Nonlinear Gating Operators) retains the same unnormalized scores within a recent window and uses positive feature-map prefix statistics for older sources, giving linear sequence-length complexity for fixed window and feature dimensions. We compare TANGO and WANGO with Recurrent and Untied Transformer++, full-attention GAU, and FLASH. All models have approximately 44.3M nonembedding parameters and are trained in three matched runs. TANGO, WANGO, and Recurrent Transformer++ apply one shared block four times; the other architectures use four independent blocks. TANGO obtains the lowest mean validation negative log-likelihood on FineWeb-Edu, Lean, and DeepMind Mathematics, although it has the largest analytical forward-pass operation count. WANGO obtains the lowest mean FineWeb-Edu NLL among the architectures with computation linear in sequence length and outperforms Recurrent Transformer++ at nearly the same analytical forward-pass multiply-accumulate count. Comments: 13 pages, 5 tables Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL) ACMclasses: I.2.6; I.2.7 Cite as: arXiv:2608.22117 [cs.LG] (or arXiv:2608.22117v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2608.22117 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-120] Semantic Reasoning Denoising: Correcting Language Model Reasoning with Semantic Operators
【速读】： 该论文旨在解决大语言模型在自然语言推理过程中产生的局部语义错误会逐级传播并导致最终结论错误的问题，尤其针对无约束自纠正机制可能保留、放大甚至引入新错误的缺陷。其解决方案的关键在于提出一种面向语义推理轨迹的可操作化马尔可夫去噪方法——语义推理去噪（Semantic Reasoning Denoising, SRD）。SRD将语义噪声形式化为可执行的错误算子（error operators），这些算子明确描述错误类型、位置以及被污染与修复的命题，通过算子组合构建逐步加剧的噪声状态。在训练阶段，模型学习识别当前推理轨迹中的活跃语义噪声并重建邻近的低噪声状态；在推理阶段，基于噪声等级感知的去噪机制反复预测逆向算子并验证其适用性，从而实现局部化、迭代式的稳定轨迹修正。实验表明，该方法在六大领域内基准任务（涵盖数学、代码、知识与常识）上平均提升3.2点，在七项跨数据集迁移任务中保持对Llama-3-8B-Instruct的竞争力，并将最强的Qwen3-8B基线平均提升2.9点。分析进一步证实，结构化的语义噪声预测与迭代算子执行是性能提升的核心机制。
链接: https://arxiv.org/abs/2608.22090
作者: Yujiao Yang
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Large language models can produce fluent reasoning traces whose local semantic errors propagate to an incorrect conclusion, while unconstrained self-correction may preserve, amplify, or introduce errors. Existing diffusion language models provide iterative refinement, but usually define noise as token masking or replacement rather than as errors in the reasoning process. We present Semantic Reasoning Denoising (SRD), an operatorized Markov denoising method for natural-language reasoning trajectories. SRD represents semantic noise with executable error operators that describe the error type, its location, and the corrupted and repaired propositions. Composing these operators constructs progressively noisier states. During training, the model learns to identify the semantic noise active in the current trajectory and to reconstruct the paired adjacent lower-noise state. During inference, noise-level-aware denoising repeatedly predicts an inverse operator and checks whether it is applicable, so each executed update makes a localized move toward a stable trajectory. Across six in-domain benchmarks spanning mathematics, code, knowledge, and commonsense, SRD improves the strongest same backbone baseline by 3.2 points on average. On seven cross-dataset transfer targets, it remains competitive with Llama-3-8B-Instruct and improves the strongest Qwen3-8B baseline average by 2.9 points. Analyses of noise sources, objectives, and denoising depth further show that structured semantic-noise prediction and iterative operator execution are central to the improvement.
[NLP-121] Real-TurnTurk: A Multimodal Turkish Corpus for Turn-Taking Prediction CEC2026
【速读】： 该论文旨在解决自然同步对话系统中话轮转换（turn-taking）建模困难的问题，尤其针对土耳其语场景下缺乏自然对话语料库的现状。现有研究虽已探索多模态方法与大语言模型在话轮结束预测中的应用，但尚未有专注于土耳其语的自然对话数据集。为此，本研究构建了一个多模态土耳其语对话数据集，涵盖未脚本化的二人互动，包含同步的正面视频、分说话人音频通道（可区分重叠发言）以及时间对齐的转录文本。话轮转换预测被建模为二分类任务，并采用遗传算法（Genetic Algorithm, GA）优化由视觉、声学与语言特征提取的可解释决策规则。所提出的框架采用混合“与-或”（AND-OR）规则表示形式，以刻画导致话轮转换的不同线索组合，其关键在于通过可解释的规则学习实现对复杂多模态交互中话轮转换动态的有效建模。
链接: https://arxiv.org/abs/2608.22071
作者: Ahmet Tuğrul Bayrak,Fatma Nur Korkmaz,Bekir Berker Türker,Mustafa Sertaç Türkel,Alper Kaplan
机构: Ata Technology Platforms(阿塔科技平台); Luxembourg National Research Fund(卢森堡国家研究基金会)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted to INTCEC 2026. This is the author’s pre-print version. The final authenticated version will be available through the conference proceedings
Abstract:Turn-taking is a basic organizational feature of human conversation and remains difficult to model in natural, synchronous dialog systems. While existing research has explored multimodal approaches and large language models for turn-ending prediction, there is a lack of naturalistic conversational corpora specifically addressing turn-taking dynamics in Turkish. This study introduces a multimodal Turkish conversational dataset of unscripted dyadic interactions, comprising synchronized front-facing video, per-speaker audio channels that allow overlapping speech to be attributed to individual speakers, and time-aligned transcriptions. Turn-taking prediction is formulated as a binary classification problem, and a Genetic Algorithm (GA) is employed to optimize interpretable decision rules derived from visual, acoustic, and linguistic features. A hybrid AND-OR rule representation is adopted in the proposed framework to represent the alternative cue combinations that precede a turn transition.
[NLP-122] SkillBloat: Token Amplification Attacks via Skill Injection in LLM Coding Agents
【速读】： 该论文旨在解决生成式AI（Generative AI）中技能生态系统（skill ecosystem）所引入的新型资源滥用威胁，即通过技能注入（skill injection）实现令牌放大（token amplification）攻击问题。具体而言，恶意技能可诱导编码代理（coding agent）在执行任务时消耗远超正常水平的计算资源（如令牌），从而造成经济性资源浪费或服务拒绝风险。其解决方案的关键在于提出SkillBloat框架——一个两阶段攻击构造方法：第一阶段通过多机制、多攻击类型条件筛选，识别出具有高潜力的放大攻击模式；第二阶段利用大语言模型（LLM）引导的完整文档级技能重写，对候选攻击进行迭代优化与精细化重构。实验证明，该方法在真实世界技能基准上实现了5.4184×至10.1455×的平均最佳令牌放大倍数，且消融实验表明第二阶段的迭代优化显著优于仅依赖初始攻击类型筛选的效果，凸显了持续优化在提升攻击效能中的关键作用。研究揭示了技能生态系统的资源放大攻击面与传统安全导向的技能污染攻击存在本质差异，具有独立且亟待防御的特性。
链接: https://arxiv.org/abs/2608.21929
作者: Yuanjin Zheng,Jingbang Chen
机构: CUHK-Shenzhen SLAI(深圳香港中文大学智能学习与人工智能研究中心)
类目: Cryptography and Security (cs.CR); Computation and Language (cs.CL)
备注:
Abstract:Agent skills extend coding agents with task-specific instructions, scripts, and resources, but they also create a trusted instruction channel that can be abused beyond conventional security attacks. This paper studies token amplification through skill injection: an economic resource-abuse threat in which a malicious skill causes an agent to consume substantially more tokens than needed for normal task execution. We present SkillBloat, a two-phase framework that first screens a library of diverse attack-type conditions across multiple amplification mechanisms and then refines the strongest candidate through LLM-guided full-document skill rewriting. Evaluated on a real-world skill benchmark, SkillBloat achieves 5.4184x-10.1455x average best amplification across multiple coding-agent target configurations. An ablation shows that the second-stage refinement loop consistently improves average best amplification over Phase 1 attack-type screening alone, demonstrating that iterative optimization provides additional benefit beyond initial attack-type selection. These results show that skill ecosystems expose a practical resource-amplification attack surface that is orthogonal to existing security-oriented skill poisoning. Subjects: Cryptography and Security (cs.CR); Computation and Language (cs.CL) Cite as: arXiv:2608.21929 [cs.CR] (or arXiv:2608.21929v1 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2608.21929 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Yuanjin Zheng [view email] [v1] Sat, 22 Aug 2026 11:41:48 UTC (789 KB) Full-text links: Access Paper: View a PDF of the paper titled SkillBloat: Token Amplification Attacks via Skill Injection in LLM Coding Agents, by Yuanjin Zheng and 1 other authorsView PDFHTML (experimental)TeX Source view license Current browse context: cs.CR prev | next new | recent | 2026-08 Change to browse by: cs cs.CL References Citations NASA ADSGoogle Scholar Semantic Scholar export BibTeX citation Loading… BibTeX formatted citation loading… Data provided by: Bookmark checked="checked"class=“labs-tab-input”> Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv’s community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) mathjaxToggle(); We gratefully acknowledge support from our major funders, member institutions, , and all contributors. About Help Contact Subscribe Copyright Privacy Accessibility Operational Status (opens in new tab) Major funding support from
[NLP-123] GuardianBench: A Same-Scene Instruction-Contrastive Benchmark for Latent Contextual Risk in Embodied AI
【速读】： 该论文旨在解决具身人工智能（Embodied AI）中潜在的上下文风险问题，即在看似安全的场景下，由于指令与场景的特定组合可能引发危险行为，而这种风险在传统评估中难以被察觉。现有研究多关注视觉环境变化或执行过程中的动态分析，但对固定场景、仅改变指令这一关键维度的探索不足。为此，论文提出GuardianBench——一个基于国际安全标准构建的指令对比型基准测试，通过3,024个同场景下的“安全/不安全”对比指令-场景样本，系统性地隔离并量化此类潜在风险。实验表明，当前主流视觉语言模型（VLMs）普遍存在指令不敏感现象：在相同场景下，无论指令是否安全，模型均倾向于批准执行，平均成对准确率仅为24.1%。进一步的推理归因审计揭示，模型主要失败于未能有效捕捉区分安全与不安全组合的指令相关线索。作为后训练优化方案，轻量级的判据对数似然监督（Verdict Log-Odds Supervision, VLOS）显著提升了开放权重模型的表现。综上，本工作通过构建受控的潜在风险任务范式、标准化的对比基准、层级化故障诊断及判据校准机制，为暴露和提升模型在复杂指令-场景组合下的安全推理能力提供了可复现的评估框架。
链接: https://arxiv.org/abs/2608.21928
作者: Zhesheng Zhang,Jiahao Lu,Wei Liu,Cong Pan,Jianhua Yang,Yixiang Chen,Hongyuan Yu,Mengqi Zhang,Kailin Lyu,Zhumin Chen,Keji He
机构: Shandong University(山东大学); National University of Singapore(新加坡国立大学); Nanjing University of Aeronautics and Astronautics(南京航空航天大学); Institute of Automation, Chinese Academy of Sciences(中国科学院自动化研究所); Xiaomi Corporation(小米公司)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Robotics (cs.RO)
备注: 21 pages, 4 figures
Abstract:In embodied AI, safety risk can be latent: a benign instruction and a safe scene become hazardous only when composed. Prior work has advanced embodied safety by varying visual contexts or evaluating execution-time dynamics, but the complementary axis of fixing the scene and varying only the instruction remains underexplored. We introduce GuardianBench, an instruction-contrastive benchmark grounded in international safety standards that isolates this latent contextual risk through 3,024 instruction-scene examples organized as same-scene Safe/Unsafe contrastive pairs across various hazard categories. Benchmarking state-of-the-art vision-language models (VLMs) reveals instruction-insensitive verdicts: models disproportionately approve both instructions under a given scene; across the primary models, average pair accuracy is only 24.1%. Our systematic rationale audit localizes the dominant failure: models fail to bind the instruction-relevant cues that differentiate safe from unsafe compositions. As a post-training case study, Verdict Log-Odds Supervision (VLOS), a lightweight verdict-level objective, substantially improves performance on open-weight backbones. Together, our latent contextual risk task formulation, standards-grounded contrastive benchmark construction, pair-level and rationale-level failure diagnosis, and benchmark-enabled verdict calibration establish GuardianBench as a controlled evaluation suite for exposing and improving safety reasoning over instruction-scene compositions under latent contextual risk.
[NLP-124] Modeling Claim Dependency Structure for Patent Litigation Prediction with Graph Attention Networks
【速读】： 该论文旨在解决专利诉讼带来的高额企业成本及其对研发（RD）激励的扭曲问题，核心挑战在于如何在专利授权时尽早识别潜在的诉讼风险。现有基于BERT的模型存在两大根本局限：一是采用扁平序列编码忽略了独立权利要求与从属权利要求之间的法律依赖结构，而这种结构直接决定专利保护范围；二是将整个权利要求集输入单一编码器，导致关键法律文本信息被淹没。通过针对134万件美国专利商标局（USPTO）实用新型专利的六模型消融实验，研究证实逐项权利要求编码、图结构连接性、注意力机制以及注意力聚合均具有独立且可叠加的预测价值。为此，本文提出ClaimGAT——一种基于图注意力网络（Graph Attention Network, GAT）的模型，其核心创新在于：对每项权利要求进行独立编码，构建有向的权利要求依赖图，利用GATConv层处理图结构，并通过注意力聚合（Attentional Aggregation）整合各权利要求信息，从而输出诉讼风险评分及可解释的逐项门控权重，支持事后结构分析。实验表明，ClaimGAT仅使用专利授权时可观察的信息，即可实现0.818的AUC-ROC值和前10%截断处4.89倍的提升（lift），并揭示高风险专利中存在结构选择性与内容敏感性分离的趋势，符合防御性权利要求撰写的典型特征。
链接: https://arxiv.org/abs/2608.21924
作者: Takao Arai,Hiroyasu Inoue
机构: University of Hyogo(兵库大学); Center for Computational Science, RIKEN(理化学研究所计算科学中心)
类目: Computation and Language (cs.CL)
备注:
Abstract:Patent litigation imposes substantial costs on firms and distorts RD incentives, making early risk identification a practically important task. While prior work has applied BERT-based models to patent claim text, two fundamental limitations remain: flat sequence encoding loses the dependency structure between independent and dependent claims that legally determines patent scope, and feeding the entire claim set to a single encoder discards legally critical text. A six-model ablation on 1.34 million USPTO utility patents confirms that per-claim encoding, graph connectivity, attention, and Attentional Aggregation each provide independent, additive predictive value. We propose ClaimGAT, a Graph Attention Network that encodes each claim independently, constructs a directed claim dependency graph, processes it with GATConv layers, and aggregates independent claims via Attentional Aggregation to yield both a litigation risk score and claim-level gate weights that enable post-hoc structural analysis. ClaimGAT achieves an AUC-ROC of 0.818 and a lift of 4.89x at the top 10%, using only information observable at the time of patent grant. It reveals a tendency in high-risk patents for structural selection and content sensitivity to diverge, a pattern consistent with defensive claim drafting.
[NLP-125] BanglaVeilGuard: Cross-Script Safety Benchmarking and Lightweight Guardrails for Bangla Large Language Models
【速读】： 该论文旨在解决孟加拉语大语言模型（Bangla Large Language Model, LLM）安全评估中因依赖英语中心或标准书写脚本基准而导致的评估失效问题，尤其针对孟加拉语用户在实际使用中普遍存在的多脚本混用、拼写变异、代码混杂（code-mixing）、区域语体等复杂语言现象。其解决方案的关键在于提出 BanglaVeilGuard——一个以孟加拉语为核心的轻量级安全基准与提示防护机制，能够覆盖六种语言形式：标准孟加拉语、罗马化孟加拉语、孟加拉英语混写（Banglish）、孟加拉语-英语代码混杂、噪声性孟加拉语及方言孟加拉语。该框架采用非破坏性多视角归一化技术，结合提示风险分类器与阈值预生成门控机制，在不修改目标模型权重的前提下，对异构模型进行高效提示筛选。实验表明，该防护机制在多种目标模型上显著降低攻击成功率（从93.8%–100.0%降至6.3%），同时在关键指标上实现88.5%的未受保护请求召回率，优于现有仅基于提示的基线方法；但当前主要挑战仍体现在对方言及噪声良性提示的过度拒绝，揭示了孟加拉语大模型部署中安全性与有用性之间的明确权衡边界。
链接: https://arxiv.org/abs/2608.21880
作者: Md. Rakibul Hassan,Muhammad Iqbal Hossain
机构: BRAC University(布拉大学); Dhaka(达卡); Bangladesh(孟加拉国)
类目: Computation and Language (cs.CL); Cryptography and Security (cs.CR)
备注: Accepted at the 4th International Conference on Computing Advancements (ICCA 2026). 8 pages
Abstract:Bangla large language model (LLM) safety is difficult to evaluate with English-centric or standard-script benchmarks because Bangla users routinely write across scripts, spellings, code-mixed forms, and regional registers. This paper presents BanglaVeilGuard, a compact Bangla-first safety benchmark and lightweight prompt guard for six language forms: standard Bangla, Romanized Bangla, Banglish, code-mixed Bangla–English, noisy Bangla, and dialectal Bangla. The benchmark contains 2,366 quality-filtered prompts and a held-out 354-prompt evaluation split spanning unsafe, safe, and safe-sensitive requests. BanglaVeilGuard uses non-destructive multi-view normalization with a prompt-risk classifier and thresholded pre-generation gate, allowing it to screen prompts for heterogeneous target models without changing their weights. Across target-model families, guarded runs reduce attack success under deterministic response scoring from 93.8–100.0% to 6.3% for Claude Opus 4.8, BanglaLLama, and TituLLM; TigerLLM-1B with BanglaVeilGuard achieves 78.2% accuracy with 8.8% ASR. The prompt guard also attains 88.5% unsafe recall, substantially above the evaluated prompt-only guard baselines. The main remaining cost is over-refusal on dialectal and noisy benign prompts, revealing a concrete safety-helpfulness frontier for Bangla LLM deployment.
[NLP-126] he Chase Is the Curriculum the Capture Anchors the Credit: Pursuit-Evasion Self-Play for Zero-Data LLM Reasoning
【速读】： 该论文旨在解决零数据自对弈（zero-data self-play）中任务难度定位不精确与奖励稀疏性问题，即现有方法仅通过事后探测候选任务并拒绝不可学习者，无法动态学习在环境难度轴上的最优任务放置位置，且仅依赖稀疏的终端奖励来评估求解器表现。其核心解决方案是将零数据自对弈重构为一个追逃博弈（pursuit-evasion game）：在LURE框架中，一个基于大语言模型（LLM）的“逃逸者”负责沿环境难度轴动态布置任务以规避“追捕者”（由规划-执行代理构成），而追捕者则通过可验证的交互逐步逼近逃逸者。逃逸者通过捕获前沿奖励（capture-frontier reward）进行训练，该奖励在求解器恰好在一半的轨迹中成功捕获逃逸者时达到峰值，从而将“勉强可捕获”转化为可学习的任务定位策略，而非人工设定的拒收阈值。同时，追捕者获得锚定于捕获事件的密集过程信用（capture-anchored dense process credit），其验证器进展通过轮次锚定的KL散度进行联合归一化，确保双主体共进化过程稳定。实验表明，LURE在三个可验证推理环境中均超越先进基线，在统一与专用设置下表现优异，且统一模型在九个外部基准测试（涵盖三类任务）上展现出优于所有训练后基线的聚合零样本外推（OOD zero-shot）准确率。
链接: https://arxiv.org/abs/2608.21871
作者: Jing Yu,Shengchao Chen,Yiyun Tan
机构: 未知
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 9 pages, 5 figures, 5 tables
Abstract:Reinforcement learning with verifiable rewards has become the dominant recipe for improving large language model reasoning, yet it presumes large human-curated task collections. Zero-data self-play removes this dependency, but existing methods vet learnability only by probing candidates and rejecting post hoc, never learning where along an environment’s difficulty axis to place a task, and credit the solver with sparse terminal rewards alone. We recast zero-data self-play as a pursuit-evasion game: in LURE, an LLM evader positions tasks along each environment’s difficulty axis to stay one step ahead of a planner-executor pursuer that hunts it down through verifiable interaction. The evader is trained on a capture-frontier reward that peaks when the solver captures it on exactly half of its rollouts, turning barely catchable into a learned positioning strategy rather than a hand-tuned rejection band. The pursuer earns capture-anchored dense process credit, in which monotone verifier progress is group-normalized jointly with the terminal capture under a round-anchored KL that keeps the co-evolution stable. Across three verifiable reasoning environments and three backbone families, LURE outperforms advanced baselines under unified/specialist settings, while the unified model attains stronger aggregate OOD zero-shot accuracy than all trained baselines across nine held-out benchmarks from three task families.
[NLP-127] MemGuard: Persisting Verifier Signals for LLM -Agent Memory Governance
【速读】： 该论文旨在解决大语言模型（LLM）智能体在执行长序列任务流时，因记忆系统不可靠而导致的性能退化问题。具体而言，现有方法面临两大失效模式：一是不可靠的记忆接纳（unreliable admission），即错误轨迹、偶然成功或误导性观察因表面相关性被误存入记忆，进而误导后续决策；二是记忆漂移（memory drift），即长期运行中积累重复、过期和冲突的信息，仅依赖检索难以修复。其解决方案的关键在于提出MemGuard机制，将验证器（verifier）输出从一次性过滤器升级为持久化的生命周期元数据，通过将多准则评分转化为奖励、置信度、标签与不确定性等描述符，并附着于候选记忆项，在激活前及后续检索、冲突消解、摘要生成与归档过程中持续复用，从而实现对记忆内容的动态质量管控。实验在Terminal-Bench 2.0、SWE-Bench Verified、WebArena和Mind2Web四个基准上进行，覆盖四种模型架构，结果表明，相较于四种记忆基线及仅使用验证器的对照组，MemGuard在所有16个配置下均取得最优成功率与最少平均步数，最大提升达7.9个百分点（WebArena）、5.6点（Mind2Web）以及2.4–3.5点（终端与软件工程类任务）。
链接: https://arxiv.org/abs/2608.21867
作者: Haoyu Wang,Guangyuan Dong,He Liang,Zijing Zhang,Jiachen Luo,Chuang Liu,Chao Xue,Hao Tang
机构: Nankai University(南开大学); National University of Singapore(新加坡国立大学); Chinese Academy of Sciences(中国科学院); Peking University(北京大学); Technical University of Munich(慕尼黑工业大学); Queen Mary University of London(伦敦玛丽女王大学); Wuhan University(武汉大学); University of New South Wales(新南威尔士大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 30 pages, 7 figures
Abstract:LLM agents are moving from single-prompt use to long task streams in which reusable memory becomes a core capability for terminal, software-engineering, and web tasks. Such memory is useful only when stored experience remains reliable across hundreds of interactions, but two failure modes break that assumption in practice. The first is unreliable admission: failed trajectories,accidental successes, and misleading observations enter memory because they appear relevant, then mislead later decisions. The second is memory drift: long-running banks accumulate duplicate, stale, and conflicting records that retrieval alone cannot repair. MemGuard’s key distinction is to treat verifier output not as a one-shot filter, but as persistent lifecycle metadata. It converts multi-criteria score-token verification into reward, confidence, label, and uncertainty descriptors that are attached to every candidate before activation and reused during retrieval, conflict resolution, summarization, and archival. We evaluate MemGuard on Terminal-Bench 2.0, SWE-Bench Verified, WebArena, and Mind2Web across four backbones, comparing against four memory baselines plus a verifier-only control under matched runtime budgets. Averaged over five seeds, MemGuard achieves the best success metric and lowest average steps in all 16 backbone-benchmark settings, improving over ReasoningBank, the strongest prior baseline among the memory methods we evaluate, with a largest gain of 7.9 success-rate points on WebArena, 5.6 step-success-rate points on Mind2Web, and 2.4-3.5 points on terminal and software-engineering benchmarks. Code is available at this https URL.
[NLP-128] HiDiffTIR: Hierarchical Difficulty-Aware Policy Optimization for Multi-Turn Tool-Integrated Reasoning EMNLP2026
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）代理在执行复杂任务时，通过与外部工具进行多轮交互（Tool-Integrated Reasoning, TIR）所面临的策略优化信号不精准的问题。现有基于强化学习（Reinforcement Learning, RL）的方法通常对整个轨迹（trajectory）赋予统一的奖励，且等同对待所有正确的工具调用，忽视了不同轨迹及推理步骤之间存在的难度差异与学习价值差异，导致学习信号粗略，难以有效区分简单与复杂的工具使用模式。为此，本文提出一种分层难度感知的策略优化框架——HiDiffTIR（Hierarchical Difficulty-aware TIR），其核心在于在轨迹级和回合级均实现难度感知的信用分配（credit assignment），使策略能够聚焦于更具信息量的轨迹和更困难的推理步骤。该方法无需额外监督信号，仅依赖标准强化学习采样所得的群体统计特征即可实现细粒度优化。在三个工具使用基准测试上的大量实验表明，HiDiffTIR显著优于多个强基线方法，在多轮TIR性能与工具调用准确率方面均有提升，验证了难度感知信用分配对于提升工具集成型LLM代理策略优化效率的关键作用。
链接: https://arxiv.org/abs/2608.21863
作者: Yucan Guo,Xiaohan Wang,Miao Su,Saiping Guan,Zhongni Hou,Jiajun Chai,Wei Lin,Guojun Yin,Xiaolong Jin,Jiafeng Guo,Xueqi Cheng
机构: State Key Laboratory of AI Safety; Institute of Computing Technology, Chinese Academy of Sciences; University of Chinese Academy of Sciences; Meituan
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted by EMNLP 2026 (Findings)
Abstract:Tool-Integrated Reasoning (TIR) is a fundamental capability for LLM agents to solve complex tasks by interacting with external tools iteratively. Reinforcement Learning (RL) has become the dominant paradigm for enabling this capability. However, existing approaches typically assign uniform trajectory-level advantages and treat all correct tool calls equally, ignoring the varying difficulty and learning value across trajectories and reasoning steps. This can lead to imprecise learning signals that do not adequately distinguish between trivial and challenging tool-use patterns. To address this limitation, we propose HiDiffTIR, a Hierarchical Difficulty-aware policy optimization framework for multi-turn TIR. HiDiffTIR performs difficulty-aware credit assignment at both trajectory and turn levels, enabling the policy to focus on more informative trajectories and harder reasoning steps. Notably, this fine-grained optimization is achieved without additional supervision, relying solely on group-level statistics derived from standard RL rollouts. Extensive experiments on three tool-using benchmarks demonstrate that HiDiffTIR consistently improves multi-turn TIR performance and tool invocation accuracy over strong RL baselines, highlighting the necessity of difficulty-aware credit assignment for effective policy optimization in tool-integrated LLM agents.
[NLP-129] PUMA: A Polish Benchmark for Culturally Grounded Multimodal Understanding
【速读】： 该论文旨在解决当前多模态大模型在非英语文化与语言背景下的评估不足问题，尤其聚焦于波兰语及波兰文化语境中多模态理解能力的系统性评估。现有研究多集中于英文文本的理解与生成，而对图像、音频及视觉丰富文档等多模态数据在非英语语境中的处理能力缺乏全面评估。为此，论文提出PUMA（Polish Unified Multimodal Assessment），一个包含900个精心设计任务的新型基准测试集，专门用于考察多模态模型在波兰文化与语言背景下的综合能力，涵盖文本、图像、音频以及视觉文档的理解与生成。其解决方案的关键在于构建一个高度本地化、跨模态且任务多样化的评估框架，能够有效揭示主流商业模型、开源权重模型及小型专用系统在复杂音频处理与文档理解任务中的显著性能差距，从而推动面向特定文化和语言的多模态人工智能研究发展。
链接: https://arxiv.org/abs/2608.21853
作者: Sławomir Dadas,Michał Perełkiewicz,Rafał Poświata,Małgorzata Grębowiec,Bartłomiej Jaworski,Izabela Woźniakowska
机构: National Information Processing Institute(国家信息处理研究所)
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language models are increasingly moving beyond text processing, adding support for other modalities such as images and audio. While text understanding and generation have been extensively studied, multimodal data processing capabilities, particularly in the context of cultures and languages other than English, have not yet been evaluated comprehensively. In this paper, we propose PUMA (Polish Unified Multimodal Assessment), a novel benchmark of 900 hand-crafted tasks designed to probe the limits of multimodal models in the Polish cultural and linguistic context. The dataset evaluates both cultural understanding and practical skill in processing text, images, audio, and visually rich documents. Our extensive evaluation of frontier commercial models, open-weights models, and specialized smaller systems highlights a significant performance gap. While top commercial models achieve high scores in visual question answering, most models struggle with complex audio or document understanding. We open-source our evaluation framework to advance localized multimodal AI research.
[NLP-130] LëtzCross: A Cross-Lingual Page-Level Benchmark for Multimodal Retrieval over Luxembourgish Documents
【速读】： 该论文旨在解决在跨语言、低资源环境下，生成式文档检索系统（如基于图像的页面检索器）的表现问题，特别是针对卢森堡语这一资源稀缺语言的文档检索挑战。现有研究多集中于视觉丰富的文档检索，但对跨语言场景下的性能缺乏深入理解。为此，论文提出LëtzCross基准，用于评估在卢森堡语PDF文档上进行跨语言页面级检索的能力，其中文档页面以图像形式索引，查询则分布在英语、法语、德语和卢森堡语四种语言中。该基准结合了以文本为中心的问答对与视觉引导的问答对，全面覆盖基于PDF的检索增强生成（RAG）中的文本与视觉双重需求。关键解决方案在于通过系统级对比分析，验证了基于图像的检索方法（如ColPali风格）相较于传统基于OCR的纯文本检索器，在多种查询语言下均表现更优；同时发现单语言微调中法语微调在卢森堡语查询上达到最高平均性能，而多语言微调中加入卢森堡语训练数据显著提升了该语言的检索效果，表明语言间的知识迁移在低资源场景下具有重要价值。
链接: https://arxiv.org/abs/2608.21714
作者: Omar El Bachyr,Fred Philippy,Laura Maria Bernardy,Saad Ezzini,Jacques Klein,Tegawende Bissyande
机构: University of Luxembourg(卢森堡大学); King Fahd University of Petroleum and Minerals(沙特法赫德国王石油矿产大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Recent page-image retrievers such as ColPali have improved retrieval over visually rich documents, yet little is known about how they behave in cross-lingual, low-resource settings. We introduce LëtzCross, a benchmark for cross-lingual page-level retrieval over Luxembourgish PDF documents, with document pages indexed as images and queries provided in English, French, German, and Luxembourgish. The benchmark combines text-focused QA pairs with visually grounded QA pairs, covering both textual and visual retrieval needs in PDF-based RAG. We use LëtzCross to compare OCR-based text-only retrievers with ColPali-style page-image retrievers and find that the latter perform better across query languages in this system-level comparison. We also examine single-language and multilingual fine-tuning. Fine-tuning transfers across query languages, with French yielding the highest mean performance on Luxembourgish queries among the single-language settings. In the multilingual setting, including Luxembourgish gives the strongest results and substantially improves retrieval for Luxembourgish queries.
[NLP-131] he Plan Not the Decoder: Diagnosing and Repairing Compositional Failure in Reasoning -Augmented Text-to-Image Generation ECCV2026
【速读】： 该论文旨在解决生成式文本到图像模型（如GoT-R1）在处理组合性提示时失败的根本原因：是规划阶段（plan）存在错误，还是解码器（decoder）未能忠实执行正确的规划。其核心发现是，规划器（planner）才是性能瓶颈，而非解码器的不忠实执行。关键解决方案在于利用可编辑的机器可读规划，通过符号化验证与计划修正实现无需重新训练的图像生成优化。实验表明，解码器对规划具有高度忠实性——94%的生成布局准确实现了规划中的空间关系，且对象-边界框绑定在计划段落重排后仍保持稳定；而规划器则因语言表述依赖性（如“left”与“right”在语义相同布局下准确率分别为98%和54%）及光栅扫描顺序偏差导致空间关系出错，进而产生杂乱几何结构。通过仅修改边界框几何信息即可带来+10.7的显著提升（p=1e-4），彻底替换计划可获得+13.3的增益（p=1e-4），证明了模块化规划-解码架构的可行性，前提是规划内部必须保持一致性（如避免框-文本矛盾，否则会导致对象重复或身份融合）。研究还发布了完整的计划保真度评估协议、所有规划文本及12,000张生成图像，为后续研究提供基准。
链接: https://arxiv.org/abs/2608.21713
作者: Ashritha Gonuguntla
机构: Carnegie Mellon University (卡内基梅隆大学)
类目: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 15 pages, 7 figures. Accepted at ECCV 2026 (oral)
Abstract:Reasoning-augmented text-to-image models such as GoT-R1 emit an explicit textual plan - object names, attributes, and bounding boxes - before generating image tokens. When such a model fails a compositional prompt, is the plan wrong, or is the plan right and the decoder unfaithful? Because the plan is machine-readable it can be edited before decoding, which makes the two separable. We first validate the ruler. Swapping the two bounding boxes inside the model’s own chain demonstrably flips the generated layout: detector-based accuracy falls 0.75 - 0.48 (p1e-3), while a widely used VQA-based spatial metric rises. A five-rater human study agrees with the detector on 81% of items and with the VQA judge on 57%. All spatial results therefore use geometric scoring. Under sound measurement the decoder is a faithful executor: 94% of generated layouts realize the planned relation, and object-box binding survives reordering of the plan’s object segments. The planner is the bottleneck. It writes wrong relations for phrasing-dependent reasons - 98% accuracy on “left” against 54% on “right” for semantically identical layouts, a raster-order bias we isolate with a mention-order control - and cluttered geometry that the decoder faithfully reproduces. Editing the plan therefore fixes the image without retraining: symbolic verification with resampling gives +5.0 points (p1e-3), minimal in-place repair +6.0 (p=.02), rewriting only box geometry +10.7 (p1e-4), and replacing the plan outright +13.3 (p=1e-4). Gains are indifferent to the plan’s prose style and to its likelihood under the planner, but not to its geometry. Modular planner-decoder designs are therefore viable, provided the plan is internally consistent: box-text contradictions induce object duplication and identity fusion. We release the plan-fidelity evaluation protocol, all plans, and 12k generated images.
[NLP-132] From Association to Causation: Improving Retrieval Precision of Retrieval-Augmented Generation via Causal Relations and an Attention Mechanism
【速读】： 该论文旨在解决检索增强生成（Retrieval-Augmented Generation, RAG）中因标准终端检索阶段依赖密集向量相似度而导致的“关键词匹配偏差”问题，即检索结果虽与查询共享关键词但缺乏实际所需信息，尤其在知识库规模扩大时该问题愈发严重。其核心解决方案是引入基于Reichenbach共同原因原理的因果图模型，将查询与检索文档间的共享关键词视为潜在的共同原因（A），而文档残余关键词构成另一潜在变量集（B），二者通过文档形成一个“碰撞器”结构（A - d - B）。由于检索过程会打开查询与B之间的关联路径，因此提出一种无需训练的注意力式重评分规则：计算查询嵌入与B加权质心嵌入之间的余弦相似度。该方法不建模知识内容内部的因果关系，而是显式建模检索过程本身的因果结构，从而更准确地识别与查询存在因果关联的文档。在真实企业级471文档知识库上，该方法将相关指南的检索排名从第6位提升至前3位；在模拟关键词堆砌场景的受控诊断语料上，平均目标文档排名由2.88提升至1.25，显著优于采用训练型交叉编码器重排序的基线（2.63）。然而，在通用BEIR基准测试中性能低于传统相似度基线，表明其适用边界存在于存在关键词堆砌现象的私有知识库环境，通过文档级校准门控机制可实现95%以上的可靠模式选择，验证了其部署可行性。
链接: https://arxiv.org/abs/2608.21702
作者: Jing Liu,Yongxing Qi,Muchen Jiang,Chengnan Hu,Qingqing Peng,Haoming Wang,Yuqing Wang,Yang Yu,Xu Zhang,Ting Wu
机构: Hangzhou Innovation Institute, Beihang University (北京航空航天大学杭州创新研究院)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 16 pages, 2 figures, 4 tables, 1 algorithm. Code available at this https URL
Abstract:Retrieval-Augmented Generation (RAG) grounds LLM generation on retrieved documents, but the standard terminal retrieval stage–dense-vector similarity, optionally followed by reranking–often returns documents that share keywords with the query without containing the needed information, a failure mode that grows with the knowledge base. We trace it to a conceptual gap: similarity captures only associational relations, whereas the documents that matter are linked to the query causally. We model the terminal retrieval stage with a causal graph grounded in Reichenbach’s common cause principle: the keywords shared by the query and a retrieved document form a latent common cause A, and the document’s residual keywords form a latent set B linking the document to the ideal output. Since a retrieved document is a collider (A - d - B), retrieval itself opens an associational path between the query and B, which licenses a training-free, attention-style re-scoring rule: the cosine similarity between the query embedding and the weighted centroid embedding of B. Unlike causality-enhanced RAG variants that model causal relations inside the knowledge content, our graph models the causal structure of the retrieval process itself. On a real 471-document enterprise knowledge base, the method promotes a relevant guideline from rank 6 to the top 3; on a controlled diagnostic corpus reproducing the keyword-stuffing regime, it improves the mean target rank from 2.88 to 1.25, while a trained cross-encoder reranker barely helps (2.63). Conversely, on three BEIR benchmarks the score underperforms the similarity baseline, delineating the applicability boundary: the method guards the keyword-stuffing regime of growing proprietary knowledge bases and complements neural rerankers; a corpus-level calibration gate selects the correct regime with = 95% reliability. A fully local testbed demonstrates deployability.
[NLP-133] Measuring Activation Control in Large Language Models
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在安全部署过程中，因具备自我操控激活状态能力而可能规避基于潜在空间（latent space）的监控机制的问题。随着模型日益具备内省能力，其不仅可能在行为层面表现出欺骗性（如“策略性行为”或“欺骗”），还可能通过自然语言指令主动调节其残差流（residual stream）的激活方向与幅度，从而在潜在空间内部实施隐蔽操作，导致现有激活监控方法失效。解决方案的关键在于提出“激活可控性基准”（Activation Controllability Benchmark），用于量化模型通过自然语言指令对自身残差流激活进行调控的能力。实验结果表明，多数主流LLM在不同能力层级下均表现出一定程度的时空分辨率控制能力，且在简单任务中已可部分规避包括线性探测、自然语言自编码器、激活预言机及雅可比透镜在内的多种激活监控手段。这一发现揭示了激活空间本身可能成为未来监控体系的新挑战，因此论文建议前沿研究机构和评估者应将激活可控性纳入未来模型的安全评估指标体系。
链接: https://arxiv.org/abs/2608.21664
作者: Marek Mateusz Kowalski,Joshua Fonseca Rivera,Uzay Macar,David Demitri Africa
机构: UK AI Security Institute; Google(谷歌)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 19 figures, 4 tables. Code: this https URL . Data: this https URL
Abstract:Safe deployment of increasingly capable models will likely come to rely on latent-space monitoring as a complement to behavioral evaluations, especially when evaluation-aware models exhibit scheming or deception. However, if models can also control their own activations, deception could extend into the latent space itself. With this in mind, we introduce the Activation Controllability Benchmark to quantify the extent to which models can modulate their residual stream via natural-language instruction. Across model families and capability levels, we find that most LLMs can control the direction and magnitude of their residual stream activations with some degree of temporal resolution, though performance varies considerably across models. In simple tasks, this level of control can evade activation-based monitoring methods (including linear probes, natural language autoencoders, activation oracles, and the Jacobian lens), albeit imperfectly. These results suggest that control over the activation space itself could become a confound for monitoring as introspective capabilities increase; therefore, we recommend that frontier labs and evaluators track activation controllability in future models.
[NLP-134] Mitigating Database Leakage in RAG Systems with Keyword-Grounded Fact Substitution
【速读】： 该论文旨在解决生成式 AI（Generative AI）在检索增强生成（Retrieval-Augmented Generation, RAG）系统中面临的提示注入攻击（prompt injection attacks）问题，此类攻击可能导致检索模块或生成模块泄露敏感数据库内容。其核心解决方案在于提出一种名为KFS-RAG的防御机制，关键在于通过注意力滚动生成与因果扰动相结合的方法，从检索到的上下文中识别出少数具有影响力的关键词，并利用一个辅助大语言模型（LLM）基于这些关键词生成一组精炼、以关键词为依据的事实片段，进而用这些经过净化的事实替代原始检索文本。该方法确保生成器仅基于经筛选的、去敏感化的证据进行推理，从而有效降低信息泄露风险，同时保持生成结果的准确性和相关性。
链接: https://arxiv.org/abs/2608.21656
作者: Ziliang Zhang,Yubo Zhu,Wei Tong,Jingyu Hua,Zijian Wang,Yuan Zhang,Sheng Zhong
机构: Nanjing University (南京大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Retrieval-Augmented Generation (RAG) has emerged as a powerful paradigm for combining large language models (LLMs) with external knowledge sources. However, RAG systems remain vulnerable to prompt injection attacks, which may mislead the retriever or generator to expose sensitive database contents. To address this issue, we propose KFS-RAG, a defense that mitigates information leakage by reformulating the retrieved context. Specifically, our method first identifies a small set of influential keywords from the retrieved context via an attention rollout plus a causal perturbation mechanism. These keywords are then used to guide an auxiliary LLM to generate a compact set of keyword-grounded facts from the retrieved passages. Finally, the original context is substituted with these curated facts, ensuring that the generator operates on sanitized evidence rather than the raw retrieved text. Experimental evaluations demonstrate that KFS-RAG significantly reduces the risk of database leakage under injection attacks while maintaining response accuracy and relevance. This work highlights a practical pathway toward building secure and trustworthy RAG systems.
[NLP-135] CyrillicQA: The Influence of Phonetically Encoded Secret Language on LLM Performance
【速读】： 该论文试图解决的问题是：大型语言模型（Large Language Models, LLMs）在处理使用拉丁字母且使用者人口众多的标准语言时表现优异，但在面对其他语言变体（尤其是濒危语言）时存在显著劣势。尽管LLMs具备潜在的语言保护能力，但其是否具备人类般的创造力与抽象能力，以解码音素编码语言（phonetically encoded language）为核心挑战。解决方案的关键在于评估并提升LLMs在非标准、低资源语言场景下的语义理解与抽象表征能力，使其不仅能够识别和生成语言形式，还能通过上下文推理实现对音素编码语言的深层语义还原，从而实现对濒危语言的有效保护与传承。
链接: https://arxiv.org/abs/2608.21462
作者: Erik Thureck,Leo S. Rdian
机构: Humboldt-Universität zu Berlin (洪堡大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 4 pages, in English; 4 pages, in German (original); German version originally published in: Rüdian, S. (2026). Prompt-Engineering in Education (1st ed., pp. 39-42). Humboldt-Universität zu Berlin. this https URL
Abstract:Due to the selection of their training data, large language models (LLMs) perform best on standard-language inputs from languages using the Latin alphabet with large speaker populations, while disadvantaging other language varieties. Nevertheless, they can also be a versatile tool for preserving precisely such endangered languages. But do they also possess the necessary creativity and capacity for abstraction to decode phonetically encoded language the same way humans do?
[NLP-136] Evaluating Multimodal Narrative Understanding of Popular Hollywood Films
【速读】： 该论文旨在解决在构建用于电影大规模计算分析的稳定基准时，因好莱坞影片版权保护而带来的挑战。其核心问题在于如何在尊重知识产权的前提下，建立一个既具代表性又可公开访问的电影数据集，以支持生成式 AI 在电影叙事分析中的研究。解决方案的关键在于：通过两个标准筛选电影——一是票房受欢迎程度（首次公开发布1922至1979年《综艺》杂志每周票房数据的大型开放数据集），二是可能处于公共领域状态（基于对美国版权登记与续展记录的系统性研究）。在此基础上，研究团队构建了一个新的多模态多项选择题（MCQ）基准，专注于评估模型对电影叙事元素的理解能力。实验结果表明，多数视觉-语言模型在该任务上表现不佳（接近随机水平），而融合音频信息的视听模型最高仅达到61.1%的准确率，远低于人类水平，凸显了当前多模态模型在深层叙事理解方面的局限性。
链接: https://arxiv.org/abs/2608.21430
作者: David Bamman,Kent K. Chang,Allison Cooper,Juishan Hsu,Reina Kushihashi,Madison Mar,Arnav Podichetty,Rachael Samberg,Ipek Nil Sancak,Yuhan Shao
机构: UC Berkeley(加州大学伯克利分校); School of Information(信息学院)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Computers and Society (cs.CY)
备注:
Abstract:Multimodal language models increasingly show promise for enabling the large-scale computational analysis of film, opening up new avenues for learning about film history and the evolution of narrative techniques. But the creation of stable benchmarks built around Hollywood films is complicated by copyright protections. In this work, we address these concerns directly, by building a new collection of Hollywood films defined by two criteria: box office popularity (where we publish the first large-scale, open collection of weekly box office earnings reported by Variety magazine from 1922-1979); and likely public domain status (by researching copyright registrations and renewals in the US Catalog of Copyright Entries). We build a new multimodal MCQ benchmark on top of this collection that focuses on narrative elements that directly evaluate the abilities of models to inform meaningful research on film narrative; we find that many vision-language models struggle on this task (with many performing at near-chance levels of accuracy), while audio-visual models (including those that use audio in captioning scenes) reach a maximum accuracy of 61.1%, well below human-level performance.
[NLP-137] Mitigating Bias in Large Vision-Language Models via Counterfactual Ensemble Decoding
【速读】： 该论文旨在解决大型视觉语言模型（Large Vision-Language Models, LVLMs）在处理不同社会群体肖像时继承并放大社会偏见的问题。现有去偏方法依赖于解码过程中对原始生成与有偏生成的词元概率进行单一视角比较，其根本局限在于仅采用刻板化的单一观点，无法体现社会认知的多样性。为此，本文提出反事实集成解码（Counterfactual Ensemble Decoding, CED）框架，其核心创新在于：在视觉表征空间中构建多社会群体的反事实视角，并在解码阶段融合这些异质视角以促进公平生成。CED首先通过识别与各社会群体相关的语义方向，在视觉空间中生成反事实表征，从而打破刻板叙事；随后在解码过程中，定位各视角间差异最大的解码层，并基于不确定性感知权重对多个视角的词元分布进行集成，优先保留来自不同群体的高置信度输出，实现更均衡的概率分布，引导更公正的文本生成。大量实验表明，该方法在三个社会偏见评估基准上显著优于现有基线，对职业、描述性特征及身份特质等场景的偏见降低达47.97%；同时，模型的核心生成能力保持良好，性能退化极小。
链接: https://arxiv.org/abs/2608.21415
作者: Yisong Xiao,Aishan Liu,Yongxin Huang,Zonghao Ying,Shiji Zhao,Tianlin Li,Yong Han,Jian Yang,Xianglong Liu
机构: Beihang University (北京航空航天大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Large Vision-Language Models (LVLMs) have achieved remarkable performance across a wide range of tasks; however, they often inherit social biases from their training data, resulting in biased behavior when processing portraits from different social groups. Existing debiasing approaches typically compare token probabilities between the original and biased generations during decoding, but they are fundamentally limited by their reliance on a single, stereotyped viewpoint and fail to account for the diversity of social perspectives. Inspired by the social science principle that diversity fosters fairness, we propose Counterfactual Ensemble Decoding (CED), a novel framework that constructs multi-group counterfactual perspectives within the visual representation space and integrates them during decoding to promote equitable model behavior. CED first performs counterfactual steering in the visual space by identifying semantic directions associated with each social group and generating counterfactual representations along these directions, thereby offering diverse perspectives that disrupt stereotypical narratives. During decoding, CED locates the decoder layer exhibiting the greatest divergence among these perspectives and ensembles their token distributions using uncertainty-aware weights, prioritizing high-confidence tokens from different groups to yield a more balanced probability distribution that guides fairer generation. Extensive experiments on three social bias evaluation benchmarks demonstrate that \tool achieves substantial improvements over leading baselines, reducing bias by up to 47.97% across scenarios involving occupations, descriptors, and persona traits. Moreover, CED also preserves the core capabilities of the original model with minimal degradation.
[NLP-138] Mitigating Speaker Leakage in Cascaded Multi-talker ASR with Diarization-based Transcript Correction INTERSPEECH2026
【速读】： 该论文旨在解决级联式多说话人自动语音识别（cascaded multi-talker ASR, MT-ASR）中因语音分离阶段产生的说话人泄漏（speaker leakage）导致的识别性能瓶颈问题。现有修正策略主要依赖于词汇层面的说话人归属重标注，但难以有效消除混合语音中残留的跨说话人干扰。本文提出一种基于剪枝（pruning-based）的互补性解决方案，其核心在于利用预训练的说话人辨识（speaker diarization）模型作为多模态验证器，通过三重一致性共识——时间包含性（temporal containment）、词汇交叉验证（lexical cross-validation）与时间对齐性（temporal alignment）——精准识别并剔除存在泄漏伪影的转录片段。实验在LibriMix、LibriSpeechMix及AMI Meeting数据集上验证了该方法的有效性，结果表明其在多种重叠条件下均能稳定降低词级别拼写错误率（cpW ER），尤其在高泄漏子集上实现最高达29%的相对减少，显著提升了复杂声学环境下级联式多说话人语音识别结果的可靠性。
链接: https://arxiv.org/abs/2608.22196
作者: Hermann Yepdjio Nkouanga,Minwei Luo,Maggie Wigness,Suresh Singh
机构: 未知
类目: Audio and Speech Processing (eess.AS); Computation and Language (cs.CL)
备注: Accepted to INTERSPEECH 2026
Abstract:While cascaded multi-talker ASR (MT-ASR) leverages state-of-the-art foundation models, its performance is often capped by speaker leakage during separation. Prior correction strategies primarily focus on lexical re-labeling for speaker attribution. We propose a complementary pruning-based paradigm that robustly identifies and removes leakage artifacts. Our method utilizes a pre-trained speaker diarization model as a multimodal verifier to prune transcribed segments satisfying a tripartite consensus of temporal containment, lexical cross-validation, and temporal alignment. Results on LibriMix, LibriSpeechMix, and the AMI Meeting corpus show our algorithm consistently reduces cpW ER across diverse overlap conditions. Specifically, on subsets with high speaker leakage, our method achieves relative cpW ER reductions of up to 29%, highlighting its effectiveness in enhancing the reliability of cascaded MT-ASR transcripts in complex acoustic environments.
信息检索
[IR-0] Robustness of IR Models to Collection Growth CIKM2026
链接: https://arxiv.org/abs/2608.23419
作者: Emmanouil Georgios Lionis,Debasis Ganguly,Sean MacAvaney
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL)
备注: CIKM 2026 Short Paper track
Abstract:Information Retrieval (IR) systems seek to identify relevant documents within a collection. In practical applications, collections are dynamic, with documents frequently added. We argue that ideally, a retriever’s effectiveness should not decrease when non-relevant documents are added to a collection. This study formalises this concept and empirically evaluates it by merging two collections with negligible topic overlap. We hypothesise that the way an IR model conditions its ranking on other documents in a collection (e.g., the IDF component in BM25 or contextual documents in listwise rerankers) plays an important role in its robustness to the addition of non-relevant documents. We broadly classify models as those that do not depend on other documents (Multi-Document-Agnostic, MDA) and those that do (Multi-Document-Dependent, MDD). Our results show that neither MDD nor MDA models are fully robust to the addition of non-relevant documents, as all models exhibit some performance degradation. Interestingly, among the models we test, MDA is more effective than MDD for retrieval, whereas MDD and MDA rerankers are equally effective.
[IR-1] Adaptive Item-based Collaborative Structures via Noise Rescheduling in Diffusion for Generative Recommendation
链接: https://arxiv.org/abs/2608.23400
作者: Jiaqi Wang,Tianying Liu,Heng Chang,Jihong Guan,Wengen Li,Shuigeng Zhou
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
备注:
Abstract:Discrete Diffusion Models (DDMs) have recently been introduced to recommendation systems, modeling user history as a token generation process via iterative denoising. However, while effective at capturing user-level sequential patterns, these methods often fail to explicitly integrate item-based collaborative filtering information, a critical component for accurate recommendation. This deficiency manifests in two key aspects: (1) the item representation is often semantic-focused, lacking collaborative priors for diffusion training; and (2) the denoising process employs a uniform noise schedule, treating all tokens indiscriminately and ignoring item-level adaptive structural dependencies. To bridge this gap, we propose ANR-DiffRec, a unified framework designed to encode item-based collaborative structures into discrete diffusion for generative recommendation. First, we explicitly incorporate an item co-occurrence matrix to guide semantic ID generation, providing a structured collaborative prior for discrete diffusion training. Second, we introduce an item-based adaptive noise rescheduling mechanism that dynamically adjusts denoising weights according to both local contextual recoverability and behavior-aware item dependencies. Specifically, the proposed strategy jointly models intra-item structural context and inter-item collaborative signals, enabling structure-aware denoising during diffusion training. Extensive experiments on multiple benchmarks demonstrate that our method consistently outperforms state-of-the-art generative recommendation models. Code: this https URL.
[IR-2] owards a Densing Law for User Representation Learning at Billion-Scale Capacity
链接: https://arxiv.org/abs/2608.23392
作者: Bin Dou,Junru Zhang,Zhaoyi Yuan,Wuliang Huang,Letian Gong,Baokun Wang,Huan Li,Yu Cheng,Weiqiang Wang
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
备注: 28 pages, 13 figures, technical report
Abstract:User representation learning in real-world industrial scenarios is commonly scaled by increasing user amount, behavioral sequence length and model size. However, existing methods face two challenges: (i) Bottleneck for raw data scaling at billion-scale capacity, as performance exhibit diminishing performance gains with larger-scale raw text user behavioral input, which can be mitigated by tokenization. (ii) Lack of quantitative analysis of how tokenization configurations should scale with data size. In this report, we propose User Behavioral Densing Law for characterizing the quantitative relationship between data scale and the minimum sufficient tokenization capacity. Firstly, we conduct a pilot study on raw tokenized scaling comparison on billion-scale Alipay dataset, revealing the raw data scaling bottleneck and the sustained gains enabled by tokenization. To derive the scaling pattern governing the minimum sufficient tokenization configuration at different data scales, theoretical analysis and systematic experiments are employed to summarize the quantitative scaling pattern. We find an approximately linear relationship between the logarithms of minimum sufficient tokenization capacity and input data size measured by tokens, and the scaling slope varies systematically with the tokenization method and data source, reflecting differences in representation-space redundancy and intra-source uniqueness. Guided by the proposed law, we further develop ALGN, an adaptive variable-length tokenization method that improves capacity allocation. Extensive experiments across diverse data sources, tokenization methods, and downstream tasks demonstrate the generalizability and reliability of the User Behavioral Densing Law, providing practical guidance for tokenization configuration selection in large-scale user representation learning. Moreover, ALGN outperforms existing baselines.
[IR-3] Hierarchical Exponential-Gaussian Mixtures for Watch-Time Distribution Prediction ICDM2026
链接: https://arxiv.org/abs/2608.23356
作者: Sofia Gulevskaia,Mikhail Trapeznikov,Aleksandr Poslavsky,Alexander D’yakonov
类目: Information Retrieval (cs.IR); Machine Learning (cs.LG); Machine Learning (stat.ML)
备注: 16 pages, 8 figures, 7 tables, accepted at IEEE International Conference on Data Mining (ICDM 2026)
Abstract:Accurate watch-time (WT) prediction is an important requirement for short-video recommendations. Yet WT distributions are near-zero-inflated, long-tailed and multimodal. The recent Exponential-Gaussian Mixture Network (EGMN) models the full conditional WT distribution rather than a single point estimate and achieves state-of-the-art performance. Our large-scale reproduction study reveals that EGMN is vulnerable to variance collapse, component redundancy, and inactive components. We propose a Hierarchical Exponential-Gaussian Mixture (HEGM) model that addresses these failure modes through a hierarchical skip-watch decomposition, KL-based variance regularization, structured initialization, removing the forced Gaussian shift and the entropy regularizer. Across public and large-scale industrial datasets, HEGM improves ranking accuracy and threshold-event prediction, while maintaining competitive point-estimation accuracy and substantially improving mixture stability and interpretability. A 1.5-month production A/B test confirms statistically significant engagement lifts. Our code and models are publicly released at this https URL.
[IR-4] he Emergence of Relevance Through Axiomatic Attention Patterns During LoRA Fine-Tuning EMNLP2026
链接: https://arxiv.org/abs/2608.23338
作者: Matthew Perlman,Atharva Nijasure,James Allan
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注: Accepted to EMNLP 2026 Findings. 17 Pages. 25 Figures. 5 Tables
Abstract:LoRA fine-tuning is standard for adapting LLMs to reranking, but it remains unclear where in the network task-specific relevance behavior is learned and what attention-level changes accompany that learning. Through ablation and attention experiments, we identify where LoRA attention updates to RankLLaMA improve performance and whether those gains coincide with interpretable relevance-oriented attention patterns such as lexical matching, rarity sensitivity, and query-document interaction. We find that given LoRA fine-tuned MLPs throughout the network, restricting LoRA attention updates to a compact mid-network region is sufficient for recovering over half of the performance gained by applying LoRA to all attention layers, and that omitting attention fine-tuning in this region hurts performance more than elsewhere in the network. Additionally, we show that regions where applying LoRA affects performance the most overlap with regions where fine-tuning increased attention to axiomatic IR features. Rarity sensitivity, document-query interaction, and several compositional features are highly correlated with gains in ranking performance. Our results support an interpretable, correlational account of how relevance-oriented behavior emerges during LoRA fine-tuning and point toward improved strategies for adapting rerankers.
[IR-5] he Laws of Context Allocation: Causal Measurement and Closed-Loop Orchestration in Generative Search
链接: https://arxiv.org/abs/2608.23252
作者: Peiyang Liu,Xi Wang,Di Liang,Wei Ye
类目: Machine Learning (cs.LG); Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注:
Abstract:As Retrieval-Augmented Generation (RAG) shifts toward diverse portfolio generation, it is stymied by two critical bottlenecks: flawed measurement of evidence utilization, and suboptimal context budget allocation. We resolve both sequentially. To resolve measurement, we expose a pervasive ``diagnostic illusion’': standard relevance proxies fail catastrophically on hard negatives. We replace them with an efficient causal leave-one-out probe that accurately isolates generative reliance and formally calibrates the structural dilution of LLM attention. To resolve allocation, we deploy this causal probe in a deconfounded factorial grid. We prove that the prevailing strategy of monolithic context widening is an architectural trap penalized by relevance decay. Instead, allocating compute iteratively across multiple sequential generations drives transformative portfolio recall gains of 16.7–20.5 absolute percentage points, scaling robustly up to 32B models. Finally, we unify these solutions into a deployable closed-loop submodular scheduler. Augmented by an attribution-steered contrastive decoder to override LLM attention inertia, our architecture systematically forces fresh evidence integration. By dominating classical open-loop baselines, we establish sequential, feedback-driven orchestration as the definitive paradigm for generative search. Our code, data, and causal measurement instruments are available at this https URL. Subjects: Machine Learning (cs.LG); Computation and Language (cs.CL); Information Retrieval (cs.IR) Cite as: arXiv:2608.23252 [cs.LG] (or arXiv:2608.23252v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2608.23252 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Peiyang Liu [view email] [v1] Mon, 24 Aug 2026 13:44:11 UTC (1,876 KB)
[IR-6] Retrieval-Augmented Classification of Environmental Mitigations in Hydropower Licensing Documents
链接: https://arxiv.org/abs/2608.23241
作者: Hong-Jun Yoon,Tom Ruggles,Joanna Lee,Debjani Singh
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
备注:
Abstract:Identifying and classifying environmental mitigation obligations in Federal Energy Regulatory Commission hydropower licensing documents is a labor-intensive task requiring deep domain expertise. We formulate this as a multi-label classification problem over a structured 135-category taxonomy and address the central challenge of severe label scarcity: 40 of 135 categories have no training examples, and 26 have fewer than five. A supervised Bidirectional Encoder Representations from Transformers (BERT)-based pipeline, while effective on well-represented categories, achieves F1 of zero on unseen classes regardless of augmentation strategy. We introduce a Retrieval-Augmented Generation (RAG) pipeline that conditions classification on retrieved category definitions, enabling zero-shot generalization across the full label space. We further propose a hybrid system that combines BERT detection with RAG classification, exploiting the high recall of fine-tuned detection and the zero-shot coverage of retrieval-augmented reasoning. Evaluated on the full set of 2017 license documents (5,860 paragraphs, 135 categories), the hybrid achieves a Micro F1 of 0.524, outperforming the BERT-only pipeline (0.477) and the RAG-only pipeline (0.416) across all training-support buckets.
[IR-7] Which Histories Matter for Time Series Forecasting? Learning Predictive Relevance with Future Supervision
链接: https://arxiv.org/abs/2608.23221
作者: Yong-Hoon Choi,Youngjin Cho
类目: Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注:
Abstract:Historical retrieval for time-series prediction commonly treats past similarity as a proxy for usefulness. We ask a different question: which historical examples should be expected to matter for a query? We define predictive relevance as expected future utility conditioned on inference-time information, using realized futures only during training as privileged supervision. A normalized-pattern retriever first forms a coarse candidate set, and a lightweight residual multilayer perceptron (MLP) learns a listwise future-compatibility target while keeping inference-time scoring strictly past-only. Our method retains similarity-based candidate generation but reranks its candidates by a more predictive relevance criterion. Optimal relevance decomposes into candidate-level utility and query-specific compatibility, motivating Candidate-Prior and Shuffled-Future controls. Across six benchmarks, the reranker improves Pattern retrieval while revealing candidate-global, query-specific, and mixed relevance regimes. On all 12 confirmatory tasks, it improves Pattern and outperforms a matched-protocol Stationarity-Aware Retrieval-Augmented Time Series Forecasting (SARAF) retrieval rule. Architecture-matched ablations show that correct future supervision, rather than the MLP or added context alone, drives gains in query-specific regimes. Alternative-similarity experiments show that a strong last-value-anchored L2 rule remains superior in some domains, whereas future-supervised relevance is particularly strong where our diagnostics indicate query-specific relevance, especially on Solar. Candidate-pool diagnostics show that this contrast is not explained solely by coarse Pattern retrieval. Overall, historical relevance is structured and domain dependent rather than governed by a universally superior retrieval rule.
[IR-8] Aligning Biomedical Texts and Knowledge Graphs: A Systematic Comparison of Lightweight Alignment Strategies ISWC2026
链接: https://arxiv.org/abs/2608.23214
作者: Artem Bisliouk,Elizaveta Nosova,Heiko Paulheim,Andreea Iana,Rita T. Sousa
类目: Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注: Accepted at the Third Workshop on Knowledge Graphs and Neurosymbolic AI (KG-NeSy 2026) co-located with ISWC 2026
Abstract:Biomedical knowledge exists in two complementary but distinct forms: unstructured scientific literature and structured knowledge graphs (KGs). Aligning them is essential for knowledge grounding, evidence retrieval, and KG completion, yet existing methods do not explicitly align free-text evidence with KG triples. We present a unified framework for systematically studying design choices for aligning biomedical text and KGs. With a text encoder and a KG embedding model both frozen, we learn only a lightweight projection between their spaces via a contrastive objective. This enables a fair comparison across six design dimensions: text encoder, KG embedding model, projection head, triple composition, training direction, and hard-negatives sampling. We construct CTD-Align, a corpus of over 22K one-to-one tripledocument pairs linking chemical-gene interactions from the Comparative Toxicogenomics Database to supporting PubMed passages. We evaluate alignment on it in two retrieval settings: document-to-triple and triple-to-document. We find that the triple composition and the training direction (i.e., shared retrieval space) have the greatest impact, whereas the text encoder and hard-negatives sampling matter little. Overall, simple choices win: projecting text into the KG space with a linear head over concatenated subject, predicate, and object embeddings performs best. These findings establish lightweight contrastive alignment as an effective, practical foundation for bridging biomedical text and KGs.
[IR-9] Evaluating Modern RAG : Textual Multimodal Dense and Late Interaction Pipelines
链接: https://arxiv.org/abs/2608.23176
作者: Emre Kuru,Mehmet Onur Keskin
类目: Information Retrieval (cs.IR)
备注:
Abstract:Retrieval-augmented generation (RAG) systems have traditionally relied on text-based pipelines that extract and retrieve information from documents. While efficient and lightweight, these approaches often struggle with documents where meaning is conveyed through layout, tables, and visual elements. Recent advances in multimodal pipelines, powered by vision-language models (VLMs), improve retrieval quality by jointly encoding visual and textual signals, but at increased computational and memory cost. We propose a quantitative, data-driven selection methodology that guides practitioners in choosing the most appropriate RAG pipeline for a given document corpus based on empirical effectiveness and resource constraints. We evaluate contemporary textual and multimodal pipelines, including dense and late-interaction architectures, analyze their trade-offs, and provide actionable guidance for balancing retrieval performance with system efficiency.
[IR-10] he Disconnect Between Better Descriptive Reasoning Trace Quality and Recommendation Effectiveness RECSYS’26
链接: https://arxiv.org/abs/2608.23154
作者: Gustavo Penha,Juan Elenter,Claudia Hauff,Hugues Bouchard,Paul Bennett,Mounia Lalmas
类目: Information Retrieval (cs.IR)
备注: Accepted at the Recsys’26 Workshop on Agentic and Generative AI for E-Commerce
Abstract:Recent work has focused on improving explicit natural-language descriptive reasoning traces for generative recommendation. This includes systems that augment semantic ID (SID) prediction with chain-of-thought reasoning. However, because SIDs are opaque learned identifiers rather than natural language, they require costly alignment before an LLM can reason over them. This provides a controlled experimental setting in which both item representation (Title vs. SID) and semantic grounding (minimal vs. extensive SID alignment) can be varied independently. We therefore present the first controlled comparison of descriptive reasoning trace quality across semantic IDs and natural-language titles in a 2 x 2 factorial study on three Amazon product domains using a shared Qwen3-1.7B backbone. We find that introducing explicit descriptive reasoning traces reduces traditional offline recommendation effectiveness under standard SFT and RL training, even though natural language titles produce substantially more grounded and interpretable traces. Extensive SID alignment improves descriptive trace quality but not traditional offline recommendation effectiveness, while a richer reward signal partially recovers performance. Overall, our results show that improving descriptive reasoning trace quality is not, by itself, sufficient to consistently improve traditional offline recommendation effectiveness under the training objectives and evaluation protocols studied here.
[IR-11] A Dual-Expert Strategy Integrating LLM s to Mitigate Negative Transfer in Cross-Domain Sequential Recommendation CIKM2026
链接: https://arxiv.org/abs/2608.23131
作者: Hyeongjun Yun,Kihyuk Song,Jaegul Choo,Chung Park
类目: Information Retrieval (cs.IR)
备注: Accepted at CIKM 2026
Abstract:Cross-Domain Sequential Recommendation (CDSR) predicts the next item a user will interact with based on their historical interaction sequences across multiple domains. Recent approaches leverage Large Language Models (LLMs) finetuned on textual representations of cross-domain user sequences to retrieve the recommended items, referred to as LLMRec. However, LLMRec primarily models the autoregressive patterns of token-level item texts, while overlooking item-level collaborative signals. This semantic misalignment often leads to distorted knowledge transfer across domains-termed negative transfer degrading performance in the CDSR task. To address this issue, we propose a novel LLM-based CDSR model, DuELRec: Domain-Gated Dual Experts with LLMs for Cross-Domain Sequential Recommendation. We propose a domain-gated dual-expert framework, equipped with an item-aware attention transformation module, which aggregates textual subtokens into item-level representations and enforces block-level attention masking. The single-domain expert restricts autoregressive attention to items within the same domain, while the cross-domain expert allows it across all domains. A gating mechanism adaptively fuses their outputs, using single-domain signals to reduce cross-domain noise that causes negative transfer. Second, we introduce a dual-sampling token-to-item contrastive learning objective that allows LLMs to capture the item-level collaborative signals from both single- and cross-domains. This is achieved by transforming token-level item texts into item-level representations and applying stochastic negative sampling from both single- and cross-domain item pools for contrastive learning. Extensive experiments on two real-world datasets across ten domains show that our model outperforms 26 state-of-the-art methods in recommendation performance.
[IR-12] raining-Free Pseudo-Fusion for Composed Image Retrieval with Diffusion Models and Multimodal Large Language Models
链接: https://arxiv.org/abs/2608.23102
作者: Fan Xu,Luis A. Leiva
类目: Computer Vision and Pattern Recognition (cs.CV); Information Retrieval (cs.IR)
备注:
Abstract:Composed Image Retrieval (CIR) is an emerging paradigm in content-based image retrieval that enables users to formulate compositional queries by combining a reference image with an auxiliary modality, usually text-based. This approach supports fine-grained search where the target image shares structural elements with the user-provided image while incorporating the modifications specified by the auxiliary text. Conventional CIR methods rely on multimodal fusion to combine visual and textual features into a joint query embedding, which requires training modules that align composed queries with the targets. In this work, we propose PeFuse (for pseudo-fusion), a training-free framework that leverages pretrained Diffusion Models and Multimodal Large Language Models to bridge modalities via generative conversion. We introduce two novel strategies: uni-directional and bi-directional conversion, which convert CIR into four single-modality retrieval problems. These methods reformulate CIR as either intra-modal or cross-modal single-query retrieval tasks, bypassing the need for dedicated task-specific training. Extensive experiments on standard benchmarks demonstrate that converting CIR into text-to-image retrieval tasks is more effective than alternative conversion strategies, achieving competitive or superior performance compared with state-of-the-art methods, while maintaining high flexibility thanks to replaceable components of the conversion pipeline. These results highlight the effectiveness of the pseudo-fusion paradigm for zero-shot CIR. Our code is publicly available at: this https URL.
[IR-13] Cultural Moment Benchmark: Evaluating Video Cultural Reasoning and Grounding in Southeast Asia EMNLP2026
链接: https://arxiv.org/abs/2608.23065
作者: Burak Satar,Zhixin Ma,Cheng Yu-Tong,Huy Hoang Tran,Phuong Anh Nguyen,Chong-Wah Ngo
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Information Retrieval (cs.IR); Multimedia (cs.MM)
备注: Accepted to EMNLP 2026 Main Conference, this https URL
Abstract:Cultural understanding in video means more than recognizing what is visible; it requires grasping the symbolic and temporal significance of cultural concepts. We decompose this into three abilities: naming what a concept symbolizes, visually recognizing it on video, and locating its sub-events in time. Existing video-cultural benchmarks tend to test what is seen, collapsing these three abilities into a single score that hides the bottleneck. We introduce the Cultural Moment Benchmark (CMB): 306 expert-curated concepts from seven countries in Southeast Asia across five categories. We evaluate each concept through three stages, one per ability. Given a description, Stage 1 (S1) selects from four candidate concept names, Stage 2 (S2) selects from four candidate video moments, and Stage 3 (S3) predicts the start and end times of the moment in a video. To keep each stage focused on a distinct ability, we use three design choices: semantic-similarity distractors (S1, S2), unlabeled video moments (S2), and free-form localization on a different example video (S3). Across six vision-language models, failure modes vary by ability and modality. i) Even the strongest closed-source models score below 30% when all three stages must be correct; ii) The three abilities do not fully cascade: naming a concept correctly helps half the models recognize it on video, but recognizing it has little effect on locating the sub-event in time; iii) Audio is complementary, redundant, or distracting depending on the concept, more often distracting in non-Latin-script countries; removing both audio and subtitles hurts Games and Music the most. Our 14-rater human study shows that even Expert raters score below chance on concepts from a neighboring country, indicating that CMB requires country-specific cultural knowledge. CMB acts as a diagnostic harness, attributing failures to a specific ability or modality.
[IR-14] Hypergraph Embedding Indexing for Efficient Dense Vector Retrieval
链接: https://arxiv.org/abs/2608.22980
作者: Kishore Konda
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
备注:
Abstract:Dense vector retrieval has become the foundation of modern semantic search, yet existing approximate nearest neighbor (ANN) indexes treat an embedding as an indivisible point in a high-dimensional space. In this work, we propose the Hypergraph Embedding Index (HEI), a framework that instead organizes documents according to combinations of highly activated latent embedding dimensions. This formulation enables inverted-index style candidate generation while preserving the semantic ranking capabilities of dense embeddings. We further demonstrate that constructing multiple complementary hypergraphs substantially improves retrieval coverage without the combinatorial growth associated with increasing the dimensionality of a single hypergraph. Finally, we establish that the statistical properties of embedding activations strongly influence coordinate-inverted indexing efficiency, introducing \emphactivation diversity as a diagnostic metric governing embedding indexability in coordinate-inverted frameworks.
[IR-15] Cascading Relevance-driven Recommendation Network for CTR Prediction in Trigger-Introduced Recommendation
链接: https://arxiv.org/abs/2608.22973
作者: Kaixuan Chen,Wenwen Wang,Xing Fang,Yang Huang,Jing Wang
类目: Information Retrieval (cs.IR)
备注:
Abstract:E-commerce has emerged as crucial platforms for people’s daily consumption and shopping interests. There is a new recommendation scenario, Trigger-Introduced Recommendation (TIR), where users click interested product, which is defined as the trigger item, containing their instant interest, and in the undertaking page following the relevant target items. Distinguished from traditional search and recommendation scenarios, trigger contains relatively strong instant interest, which is more vague and implicit compared to search terms. Relying on large amounts of labeled data, existing methods lack the exploration of trigger relevance, which affects users’ immersive experience. To alleviate this problem, we propose the Cascading Relevance-driven Recommendation Network (CRRN) to emphasize the interaction and relevance between trigger and target, comprising three essential components: 1) the Trigger-Target Interaction layer extracts interaction features of trigger and target based on personalized gating. 2) Cascading Interest Fusion module explicitly estimates users’ trigger intention and fuses instant and personalized interests adaptively with cascading attention blocks. 3) Category-assisted Pairwise Loss enhances trigger relevance with the guidance of category association between trigger and target. Extensive experiment results show that CRRN outperforms recent state-of-the-art methods on both industrial and public datasets. Online A/B tests further validate the effectiveness of our method. Our code is available at this https URL.
[IR-16] SWAP: A Multilingual Retrieval-Augmented Thai Wellness Advisor
链接: https://arxiv.org/abs/2608.22917
作者: Pornthep Ukosaramig,Kobkrit Viriyayudhakorn
类目: Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注: 8 pages, 2 tables. Data and evaluation logs: this https URL
Abstract:We present TSWAP, a deployed eight-language conversational wellness advisor grounded, via retrieval-augmented generation, in a verified knowledge base of Thai traditional medicine and certified wellness providers. An unmodified open-weight LLM (Qwen3.6-35B-A3B on vLLM) is grounded on a ~30.6K-chunk Thai index by a hybrid dense-sparse retriever with cross-encoder reranking; a first-turn query classifier forces tool-based retrieval for entity lookups; a rule-based safety layer enforces medical scope and Thai emergency routing; and all eight languages are served zero-shot with translate-then-retrieve. We release the first Thai traditional-medicine/wellness retrieval benchmark (50 questions with gold document IDs; Recall@5 = 0.88), production QA logs (91.1% test-retest pass over 259 cases), and a 71-question frontier no-retrieval probe showing what each grounding pillar contributes: without the safety prompt the backend model family produced a full drug-dosing schedule and complied with out-of-scope requests, and without the knowledge base it produced zero verifiable provider recommendations. We further report two transferable deployment findings: English-calibrated 4-bit AWQ quantization corrupts Thai tone marks, and forced-retrieval routing is necessary for reliable grounding.
[IR-17] Better Retrieval Worse Robustness:How Multi-hop RAG Amplifies Upstream ASR Errors EMNLP2026
链接: https://arxiv.org/abs/2608.22872
作者: Zhenghua Bao
类目: Computation and Language (cs.CL); Information Retrieval (cs.IR); Audio and Speech Processing (eess.AS)
备注: Accepted to EMNLP 2026 (Main Conference)
Abstract:Speech-based applications pass spoken queries through automatic speech recognition (ASR) before any retrieval module, so ASR errors enter the pipeline as a fixed upstream constraint. We empirically test whether two extensions to standard retrieval-augmented generation (RAG), entity-graph linking and iterative reformulation, absorb or amplify these errors. Using four English accents synthesized through neural TTS, we evaluate four RAG configurations on three multi-hop QA benchmarks (HotpotQA, 2WikiMultiHopQA and MuSiQue) against a clean-text oracle. Although the structurally richer configurations generally retain higher absolute F1 under ASR input, both extensions amplify the error: the F1 gap from clean text to the highest-WER accent is 36-67% larger under their combination than under naive dense retrieval, on all three benchmarks. The dominant failure mode is corruption of one or more query entities, accounting for 87-96% of degradation cases on 2WikiMultiHopQA across all four methods. Two lightweight surface-form mitigations leave most of the gap intact, indicating that downstream retrieval structure amplifies remaining entity errors. We release code and data at this https URL .
[IR-18] WARP: Wasserstein-Aligned RAG for Population Opinions
链接: https://arxiv.org/abs/2608.22859
作者: Aman Singh Thakur,Aditya Agrawal,Alwarappan Nakkiran,Alex Karlsson
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL)
备注: Pre-print
Abstract:RAG systems are increasingly used to summarize what large collections of documents say. A user asks “What do people think about X?” and receives an answer that reads as consensus. But standard top-k retrieval ranks documents by query similarity, not by how faithfully they represent the population, so minority views quietly disappear. Existing fixes fall short. Diversity re-rankers like MMR and DPP spread retrieved documents apart, but with no target distribution to aim for. Calibration methods based on KL or JS divergence do target one, yet treat opinion bins as unordered: confusing strong positive with strong negative costs no more than an adjacent-bin miss. We introduce WARP, a family of post-retrieval algorithms that calibrate retrieved evidence to the population’s opinion distribution. WARP first recovers underrepresented opinions that cosine ranking may bury, then uses Wasserstein-1 distance to select documents whose sentiment-intensity distribution matches the population target, capturing the ordinal structure ignored by KL and JS divergence. We develop three variants for dense, sparse, and variable candidate pools, trading off calibration quality and speed. Across three review domains spanning 35K documents, 156 queries, and 26 entities, WARP’s domain-matched variants reduce distributional error by at least 43% with sub-second latency. These gains carry through to generation: a five-judge LLM panel prefers WARP-generated answers in 86% of decided comparisons at k = 5. Comments: Pre-print Subjects: Information Retrieval (cs.IR); Computation and Language (cs.CL) Cite as: arXiv:2608.22859 [cs.IR] (or arXiv:2608.22859v1 [cs.IR] for this version) https://doi.org/10.48550/arXiv.2608.22859 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[IR-19] Same Agent Different Answers: A Repeat-Aware Audit of Corpus-Induced Answer Churn in Retrieval-Augmented QA
链接: https://arxiv.org/abs/2608.22856
作者: Jingjie Ning,Xueqi Li
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL)
备注:
Abstract:A retrieval-augmented QA system can return different answers after an index expansion even when its requested model identifier, prompt, retrieval policy, evidence depth, rendering, and exposed generation controls are held fixed. Aggregate accuracy may hide these changes when gains and losses cancel, while ordinary generation variability makes one-shot comparisons overstate update effects. We call the hidden phenomenon accuracy-blind answer churn and introduce the \emphSnapshot Compatibility Audit, which estimates excess answer churn by subtracting same-snapshot repeat disagreement from cross-snapshot disagreement. We instantiate it by expanding one frozen FineWeb prefix from one to seven shards. In a preregistered 400-question Natural Questions study, normalized-exact and blinded-semantic excess churn are 6.44 and 10.25 percentage points while exact-match accuracy changes by only -1.50 points. A post-hoc analysis finds repeat-stable semantic flips on 40/400 questions. A separately preregistered 200-question TriviaQA study yields smaller, directionally consistent excess churn while exact-match accuracy moves in the opposite direction. An outcome-blind post-hoc 100-question subset replication with a second DeepSeek generator and serving configuration finds 8.75 pp of semantic excess churn even as exact match rises by 3.00 percentage points. Answer-level compatibility can therefore fail without a conspicuous or consistently directed utility shift. Retrieval-augmented releases should audit compatibility alongside utility.
[IR-20] he Compaction Cliff in Long-Running AI Agent Memory
链接: https://arxiv.org/abs/2608.22752
作者: Saber Zerhoudi,Jelena Mitrovic,Michael Granitzer
类目: Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注:
Abstract:A safety rule and an episodic log compete for the same tokens in an AI agent’s context. When the budget overflows, both are summarized at the same rate; only the rule needs exact wording to remain enforceable. On 20 production agent configurations, Claude Code’s /compact prompt on Sonnet 4.6 preserves 53% of safety rules after one compaction round and 10% after five. We name this the Compaction Cliff. We address it with Knowledge Triage, a framework that classifies each line of an agent’s knowledge base by type and routes each type through its own retention policy. Three deterministic operators implement this triage across the three context-management operations: TypeCompact rewrites items in place under per-type fidelity, TypeDecompose partitions a topic too large to compact safely, replicating in-scope safety rules across partitions, and TypeRetrieve fetches items from external storage with in-scope rules pinned ahead of relevance. On five public corpora, TypeCompact preserves 2–4 \times more safety rules than the strongest single-shot LLM compactor at every ratio, with 96% recall over five rounds. TypeDecompose reaches 0% locality violations against 93% under uniform partitioning. TypeRetrieve reaches 100% recall@50 against 73% for the best single-shot LLM retriever. On three downstream behavioral benchmarks, we outperform the production Sonnet compactor on medical compliance (paired McNemar p 10^-8 on preservation, N = 200 ), the full-policy and hierarchical baselines on retail task pass rate ( p 0.01 , N = 115 ), and the hierarchical compaction on the airline domain ( p = 0.024 ). We release AgentArtifactCorpus (396,934 agent configurations from 54,628 public GitHub repositories), the classifier, and the reference implementation.
[IR-21] Risk-Aware Reranking for Agent ic Tool Retrieval CIKM2026
链接: https://arxiv.org/abs/2608.22751
作者: Qinfei Li,Xiaoxuan Dong,Jin Zhang,Dexu Yu,Wenhao Deng,Junchen Fu,Youhua Li,Hanwen Du,Chunxiao Li
类目: Information Retrieval (cs.IR)
备注: Accepted by CIKM 2026
Abstract:Tool retrieval determines which external tools are exposed to an LLM agent for a user query or task, making retrieval a critical pre-execution safety boundary. Unlike document retrieval, tool retrieval exposes executable actions: a tool that is useful for one task may be unnecessary or risky for another. However, existing tool-retrieval methods primarily optimize semantic relevance, and safety evaluations often focus on failures after tool execution rather than risks introduced during retrieval. We study risk-aware tool retrieval, where the goal is to retrieve useful tools while reducing exposure to higher-risk tools. We propose a lightweight reranking framework on top of a frozen first-stage retriever. The framework models query-conditioned relevance and tool-level exposure risk separately, combines them through an explicit parameter controlling the tradeoff between safety and utility, smooths scores over a ToolGraph, and optionally applies rule-based safety constraints. To support retrieval-time safety evaluation, we annotate 6,108 tools across UltraTool and Seal-Tools with five ordinal risk levels and define metrics that measure risky-tool exposure in the top- k results. Experiments on UltraTool and Seal-Tools show that our approach improves the relevance–safety tradeoff over relevance-only retrievers and reranking baselines, with the rule-filtered variant providing a conservative operating point for safety-critical deployments. These findings indicate that retrieval-stage filtering can reduce the candidate action space exposed to agents before execution, complementing downstream tool-use safeguards. The code and supplementary materials are available at: this https URL.
[IR-22] ExecRubrics: Executable Tool-Augmented Rubrics for Verifiable and Efficient Long-Form Evaluation EMNLP2026
链接: https://arxiv.org/abs/2608.22559
作者: Kaustubh D. Dhole,Charles L. A. Clarke,Eugene Y. Agichtein
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注: Accepted to EMNLP 2026 Findings
Abstract:Rubrics aim to make language-model evaluation transparent by decomposing response quality into interpretable criteria. However, natural-language rubrics are often ambiguous, require black-box LLM judges, and typically assume criteria aggregate independently through linear weighted sums, limiting their ability to capture dependencies, alternatives, penalties, and override conditions. We propose ExecRubrics, a framework for representing rubrics as compact executable programs. ExecRubrics encodes evaluation logic as verifiable Python scoring functions, giving natural-language rubric intent an operational semantics: a fixed decision procedure that can be inspected, executed, and edited. On three long-form response benchmarks-HealthBench, HelpSteer, and ArgQuality-we show that ExecRubrics can substitute for expensive black-box judges in ranking preferred over dispreferred responses, matching or improving NL rubric baselines with best preference accuracies of 53%, 78%, and 92%, respectively, while reducing evaluation latency by up to 320 times. We show that incorporating external logic and resources from text processing libraries such as NLTK and spaCy further improves preference accuracy. Our results suggest a novel way of looking at evaluation, by offering a faster, more explainable, and less ambiguous alternative to black-box rubric evaluation, particularly in high-stakes domains such as healthcare and banking where precision and auditability are critical.
[IR-23] GRAFT: Graph-Distilled Generative Retrieval for Facet-Aware Scientific Literature Exploration
链接: https://arxiv.org/abs/2608.22381
作者: Italo Luis da Silva,Hanqi Yan,Yujing Wang,Jiangnan Ye,Lin Gui,Yulan He
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL)
备注:
Abstract:Scientific papers may relate by problem, method, result, or contribution, but document-level retrievers collapse these into a single similarity score without saying why they are related. Citation- and similarity-based retrieval alone also confines search to the neighbourhood of what is already known, whereas generative retrieval generates document identifiers directly, enabling the exploratory retrieval that scientific discovery depends on. We connect papers in a graph whose edges are typed by these four facets, derived from facet items and citation signals, and distil it into a generative retriever whose identifiers are the papers’ own facet text. Two graph properties do not survive naive distillation. First, because every training pair is an edge, naive enumeration indexes just 84% of the corpus. Coverage-aware distillation makes every paper learnable through a reverse-neighbour fallback, a minimum-coverage threshold, and edge-importance weighting. Second, constrained decoding guarantees that every generated identifier is a valid paper, but not that the graph connects it to the query. Graph-weighted reciprocal rank fusion scales each candidate’s rank term by its query-candidate edge weight, dropping unsupported ones. On LitWeave, our constructed corpus of 11,359 NLP papers, Graft recovers 91% of its graph teacher’s Recall@20 with no nearest-neighbour index or encoder at inference, and outperforms the graph teacher on query papers outside the corpus. It reproduces the graph’s own facet labels at 0.922 precision, so every returned paper arrives labelled with the facet that surfaced it rather than an opaque score.
[IR-24] Prompt-Based Abstention Fails Under Misleading Context: A Controlled Study of Small Frozen RAG Models AACL
链接: https://arxiv.org/abs/2608.22228
作者: Yohanes Andre Setiawan
类目: Information Retrieval (cs.IR)
备注: 20 pages. Committed to AACL-IJCNLP 2026. Code/data to be released
Abstract:Missing and misleading evidence are not the same problem in retrieval-augmented generation (RAG), but prompt-based abstention treats them alike. Models abstain when context is absent, not when it is misleading. We introduce GRAB-RAG (Graded Abstention Benchmark for Retrieval-Augmented Generation), a paired benchmark that tests the same questions across four context conditions (supportive, degraded, missing, and misleading) in Natural Questions and HotpotQA. In the misleading condition, we edit a gold passage to support a wrong answer and place it among other retrieved passages. We test five abstention policies on three small frozen models (3.8B–8B) across two QA benchmarks. Models abstain reliably when evidence is missing, but under explicit abstention prompting still answer 41.6% of misleading questions, with 63% of those answers echoing the planted wrong entity verbatim. Chain-of-thought provides little additional benefit. A generator-side conflict check cuts the rate to 13.3% but discards many correct answers, while an NLI verifier recovers that coverage but fails when parametric memory and the misleading passage agree on the same wrong answer. Prompt-based abstention asks whether context is sufficient, not whether it is correct. Neither verifier closes this gap without trading coverage for safety.
[IR-25] VERDICT: Agreement Beats Pixel-Space Verification in Real-Document OCSR
链接: https://arxiv.org/abs/2608.22183
作者: Yani Guan,Dengpan Dong,Shuang Luo,Zi Wei,Joah Han,Dan Hannah,Yumin Zhang,Qichao Hu,Kang Xu
类目: Computer Vision and Pattern Recognition (cs.CV); Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注:
Abstract:Optical Chemical Structure Recognition (OCSR) converts 2D molecular depictions in the published literature into SMILES, and is increasingly important for constructing large-scale chemical training datasets. Automation at that scale requires identifying unreliable predictions in the absence of ground truth. Three families of label-free signals were compared on 263 ACS journal depictions with verified ground truth: model confidence, re-rendering similarity, and agreement among recognizers. Pixel-space re-rendering performed little better than chance (AUROC 0.547 , 95% CI [0.465,0.629] ), and an oracle-tuned threshold on it reduced correct labels per image from 0.745 to 0.205 . Agreement among four architecturally distinct recognizers instead reached an AUROC of 0.916 ( [0.880,0.952] ). The two-of-four rule accepted 81.7% of images at 88.8% precision, the three-of-four rule 52.1% at 98.5% . The same pattern held on CLEF-IP, UOB, and USPTO. This distinction is obscured on synthetic benchmarks, where re-rendered predictions naturally resemble their inputs. A substance filter removed 2,193 false agreements on wildcards and R-group fragments, after which the three-of-four rule rejected all 68 generic depictions. VERDICT was then applied to PMC Open Access, producing 6,146 structure labels for 4,833 molecules; chemist adjudication of 400 released labels in two independent samples yielded precisions of 0.995 for the three-of-four tier and 0.958 for the two-of-four tier. VERDICT therefore enables validated labels for multimodal molecular databases linking structure images, machine-readable representations, and source-publication information. In SES AI’s Molecular Universe platform, VERDICT further serves as an image-based interface for searching and retrieving molecular records.
[IR-26] GrOIL: Graph-Grounded Domain Ontology Induction with Constrained LLM Mediation
链接: https://arxiv.org/abs/2608.22135
作者: Maruf Ahmed Mridul,Abid Talukder,Oshani Seneviratne
类目: Information Retrieval (cs.IR)
备注: 12 pages, 9 figures, 3 tables
Abstract:Constructing formal ontologies from domain documents requires simultaneously enforcing corpus grounding, vocabulary consistency, axiom-level expressivity, and end-to-end provenance, a combination no existing automatic system delivers. We present a seven-stage graph-grounded pipeline that converts domain documents into a complete, auditable Web Ontology Language (OWL) Terminological Box (TBox) without any unconstrained generation step. Documents are first encoded as Unified Discourse-Hypergraphs (UDH) capturing entity participation and discourse dependencies; subsequent stages transform this graph evidence into a class hierarchy, typed object and datatype properties, and restriction axioms, with Large Language Model (LLM) usage restricted to narrow, graph-grounded mediation tasks. A paired Assertional Box (ABox) population procedure grounds named individuals in the induced TBox, enabling SPARQL-based functional evaluation. Every emitted term carries a full decision chain from raw source passages through each pipeline stage, making the TBox directly auditable and suitable for targeted human refinement. Evaluated on the life insurance domain using two established benchmarks and a new 100-contract corpus spanning ten product types, our pipeline achieves strong results across all evaluation dimensions, outperforming direct and multi-agent LLM baselines on competency-question (CQ) coverage (0.85 vs. 0.63 and 0.62 on one term-life contract; 0.77 vs. 0.40 and 0.44 on another contract), while also attaining high keyphrase coverage comparable to a manually-constructed reference ontology and strong performance on structured gap-and-overlap reasoning, all without any manual TBox engineering. Ontology growth analysis provides evidence consistent with vocabulary saturation at scale, demonstrating that the pipeline produces stable, reusable domain representations from large document corpora.
[IR-27] RAG Collapse: LLM Responses Collapse When Retrieved Documents Are Self-Authored
链接: https://arxiv.org/abs/2608.22118
作者: Gregory Druck,Ethan Smith
类目: Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注: 36 pages, 31 figures, 4 tables
Abstract:LLM responses are based on the internet (via training or RAG), and AI is now used to generate a significant amount of content online (Paredes et al., 2026), creating the potential for a self-reinforcing feedback loop. Prior work has shown that when LLMs are recursively trained on their own output, they experience model collapse (Shumailov et al., 2024): responses become less diverse, and eventually no longer resemble the original training data. In this paper, we show that a similar collapse occurs if LLM-based AI systems retrieve references they authored using a search tool. We call this RAG collapse. We conduct extensive experiments with three types of simulations of AI systems retrieving references they generated, using three model families, and 1,019 information-seeking prompts, totaling 1,528 simulations and over one million LLM API calls, and find that 79.6% (1,216/1,528) of simulations end in collapse. Surprisingly, even a single self-authored reference can trigger collapse because the LLM disproportionately cites its own content. This self-bias persists even after controlling for reference quality.
[IR-28] W-RAG : Source-Aware Retrieval for Enterprise Document Generation from Heterogeneous Knowledge Bases
链接: https://arxiv.org/abs/2608.22081
作者: Hridya Dhulipala,Rajesh Ombase,Michael Wang,Tien N. Nguyen
类目: oftware Engineering (cs.SE); Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注: 18 pages
Abstract:Retrieval-Augmented Generation (RAG) enables large language models to incorporate external knowledge during generation, improving factual grounding and domain adaptability. However, existing RAG pipelines assume that evidence retrieved from multiple repositories can be ranked globally using a single similarity function. While suitable for open-domain retrieval, this assumption breaks down in enterprise document generation, where heterogeneous knowledge bases (such as policies, regulations, technical documentation, and departmental guidelines) serve distinct roles and must be jointly represented in the generated document. As a result, global ranking often produces unbalanced context dominated by a subset of sources, leading to incomplete enterprise drafts. To address this limitation, we propose W-RAG, a source-aware retrieval framework that performs ontology-guided retrieval, local ranking within each knowledge base, and source-level weighting to regulate evidence composition. We further introduce a new dataset for retrieval-grounded enterprise document generation spanning multiple document types and industry domains. Experiments show that standard RAG pipelines struggle on this task, while W-RAG significantly improves document coverage and generation quality.
[IR-29] Revisiting N2DCG: An Empirically Grounded Reformulation of Carousel Recommendation Evaluation
链接: https://arxiv.org/abs/2608.21877
作者: Jingwei Kang,Santiago de Leon-Martinez,Maarten de Rijke,Harrie Oosterhuis
类目: Information Retrieval (cs.IR); Human-Computer Interaction (cs.HC)
备注:
Abstract:Carousel interfaces have been widely used in video and music streaming services, yet it remains unclear how to properly evaluate recommender systems in these two-dimensional layouts. N2DCG has been proposed to address this gap by adapting NDCG to carousel-based recommendation, but it relies on unverified assumptions borrowed from the single-list web-search setting that do not transfer well to two-dimensional carousel layouts. We identify two substantial limitations of N2DCG: its ideal ranking, used for normalization, violates carousel constraints, and its discount function does not reflect user browsing behavior observed in empirical data. To address both limitations, we propose a reformulation of N2DCG that normalizes appropriately by respecting constraints and uses an empirically grounded discount function. We validate the proposed metric, showing that it better reflects users’ empirical behavior on real-world eye-tracking data and better predicts the comparison results of carousel layouts simulated based on empirical examination patterns. Subjects: Information Retrieval (cs.IR); Human-Computer Interaction (cs.HC) Cite as: arXiv:2608.21877 [cs.IR] (or arXiv:2608.21877v1 [cs.IR] for this version) https://doi.org/10.48550/arXiv.2608.21877 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[IR-30] Why This Not That? Mining User Profiles for Pair-wise Counterfactuals
链接: https://arxiv.org/abs/2608.21662
作者: Meysam Varasteh,Veronika Bogina,Noam Koenigstein,Robin Burke
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
备注:
Abstract:The topic of explanation in recommender systems has seen steady research attention since the earliest days of the field. With some exceptions, this work has focused on the explanation of single items in a recommendation list and, especially recently, has emphasized approaches that are decoupled from the logic of the recommendation algorithm itself. Based on findings in the psychology of interpersonal communication, we propose a new task, pairwise interpretation of item rankings, asking the comparative question ``Why is item A ranked higher than item B?‘’. An effective solution to this task, we argue, is inherently grounded in the operation of the recommendation algorithm. We propose a class of techniques based on counterfactual learning to uncover the items in a user’s profile that have contributed to the relative ranking of items. Using multiple datasets, we show that it is possible to identify such items as potential basis for comparative explanation.
[IR-31] Why didnt more people see it? Recommendation: Transparency for providers
链接: https://arxiv.org/abs/2608.21641
作者: Meysam Varasteh,Robin Burke
类目: Information Retrieval (cs.IR)
备注:
Abstract:Transparency in recommender systems has been widely studied from the perspective of those receiving recommendations, yet the needs of item providers, the creators whose content is distributed through these platforms, remain largely unexplored. Providers often lack insight into how their items do or do not receive exposure in users’ recommendation lists. In this work, we address this gap by proposing a surrogate modeling approach to explain item exposure at a system level. Rather than explaining individual user-item pairs, we train a proxy model to approximate the exposure distribution produced by a recommender. By quantifying the contribution of each feature, we seek to explain the factors driving the recommendation model’s decisions across the entire user base. We evaluate our approach on two datasets and three recommendation models. Results show that the surrogate model captures the global behavior of all three recommenders with high fidelity and that the most influential factors vary meaningfully across models and domains.
[IR-32] Retrieval Needs Multivectors: An Exponential Separation
链接: https://arxiv.org/abs/2608.21494
作者: Mihir Agarwal,Viraj Agrawal,Sabyasachi Basu,Ankit Garg,Kirankumar Shiragur
类目: Information Retrieval (cs.IR); Databases (cs.DB); Machine Learning (cs.LG)
备注:
Abstract:Recent works have highlighted the expressive limitations of embedding based retrieval models through both theoretical analyses and challenging benchmarks such as LIMIT. While multi-vector embeddings consistently outperform single-vector embeddings, the precise representational gap between them remains poorly understood. In this work, following Jayaram’s work, we provide the first explicit family of query and document sets, together with their relevance matrices, for which single-vector embeddings that rank all relevant documents above irrelevant ones require exponential size, whereas polynomial-size multi-vector embeddings suffice. Our result establishes an exponential separation between the expressive power of single-vector and multi-vector embeddings for the task of ranking of documents as opposed to approximating numerical scores as in the work of Jayaram. Motivated by our theoretical construction, we introduce ANDOR, a new retrieval benchmark that naturally instantiates these hard examples. We show that state-of-the-art single-vector embedding models perform poorly on ANDOR in the zero-shot setting and exhibit only marginal improvements after fine-tuning, highlighting the inherent difficulty of the benchmark compared to prior work. In contrast, multi-vector models consistently outperform their single-vector counterparts and improve substantially with fine-tuning, closely aligning with our theoretical predictions. Subjects: Information Retrieval (cs.IR); Databases (cs.DB); Machine Learning (cs.LG) Cite as: arXiv:2608.21494 [cs.IR] (or arXiv:2608.21494v1 [cs.IR] for this version) https://doi.org/10.48550/arXiv.2608.21494 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[IR-33] A Deep Causal Inference Approach to Measuring the Effects of Forming Group Loans in Online Non-profit Microfinance Platform
链接: https://arxiv.org/abs/1706.02795
作者: Thai T. Pham,Yuanyuan Shen
类目: Machine Learning (stat.ML); Information Retrieval (cs.IR); Machine Learning (cs.LG); General Finance (q-fin.GN)
备注:
Abstract:Kiva is an online non-profit crowdsouring microfinance platform that raises funds for the poor in the third world. The borrowers on Kiva are small business owners and individuals in urgent need of money. To raise funds as fast as possible, they have the option to form groups and post loan requests in the name of their groups. While it is generally believed that group loans pose less risk for investors than individual loans do, we study whether this is the case in a philanthropic online marketplace. In particular, we measure the effect of group loans on funding time while controlling for the loan sizes and other factors. Because loan descriptions (in the form of texts) play an important role in lenders’ decision process on Kiva, we make use of this information through deep learning in natural language processing. In this aspect, this is the first paper that uses one of the most advanced deep learning techniques to deal with unstructured data in a way that can take advantage of its superior prediction power to answer causal questions. We find that on average, forming group loans speeds up the funding time by about 3.3 days.
人机交互
[HC-0] When “Do Not” Is Not Deny: Security Rules in CLAUDE.md vs Built-In Controls
链接: https://arxiv.org/abs/2608.23550
作者: Ting Yan
类目: Human-Computer Interaction (cs.HC); Cryptography and Security (cs.CR)
备注: 11 pages, 5 figures
Abstract:In this http URL, “do not” is a natural-language instruction that the model interprets. Claude Code’s deny is a built-in control that blocks an action before the agent can take it. Both can express the same security goal, but they control the agent in different ways. We measure this gap in 481 public this http URL files. An LLM matched the extracted candidate rules against Claude Code’s documented controls, and two security practitioners independently checked a sample without seeing the model’s answers or each other’s labels. Depending on how closely a control had to match the written rule, only about 4-16% of the retrieved security rules had a matching built-in control. Under the strictest standard the estimate was 4.4% (95% CI: 2.6-6.7%), and the two annotators agreed closely on which rules had a match. A manual review of complete files found that our extraction method captured 66.3% of eligible security rules; the reported rates therefore apply to the rules it captured. This is a usable security problem: this http URL is a write-only channel. A developer writes a security rule but gets no feedback on whether a control will enforce it. The same plain-text form hides two kinds of rule: those a permission rule, mode, or sandbox can enforce, and those left to the model to interpret.
[HC-1] Multisensor Measurement of Train Driver Mental Fatigue: From Simulation to Reality
链接: https://arxiv.org/abs/2608.23361
作者: Esther Bosch,Rebecca Kruschka,David Schackmann,Stephanie Hoyer,Wolfgang Kilian,Stefan Schwanitz,Anneke Hamann
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Increasing automation in rail transport shifts the train driver’s role from active control to prolonged supervisory monitoring. This creates conditions for mental fatigue (MF) and reduced vigilance. Despite the safety relevance of this issue, evidence on the feasibility and robustness of physiological indicators of MF under operational rail conditions remains limited. Most prior work relies on simulators or lab studies. The present study investigated multiple subjective, physiological, and behavioral indicators of MF in professional train drivers across two complementary settings: a high-fidelity train simulator (n=14) and a real-world rail environment (n=6). To our knowledge, this is the first study to deploy a full multisensor battery under actual train operating conditions. In both settings, a standardized protocol was used comprising a baseline drive, a one-hour auditory n-back task as an MF induction procedure, and a second drive. Heart rate variability and breathing rate showed consistent and theoretically expected changes across both environments, suggesting reduced physiological arousal following the fatigue induction task. In contrast, EEG-based frontal theta power and parietal alpha and beta power, electrodermal activity, blink duration, and behavioral indicators did not show clear mental fatigue-related patterns. Real-world data collection revealed substantial technical challenges related to vibration, sensor connectivity, and concurrent high-frequency data acquisition. These findings suggest that autonomic indicators, particularly HRV and breathing rate, represent the most promising and ecologically robust measures for operational fatigue monitoring in train drivers. However, neurophysiological measures require further validation under realistic conditions before deployment in driver monitoring systems, and larger samples are needed to confirm these preliminary patterns.
[HC-2] Beyond the Mirror: Balancing Interaction Modality and Avatar Fidelity in Public 3D Virtual Try-On Systems
链接: https://arxiv.org/abs/2608.23345
作者: Yueqian Guo,Tianzhao Li,Xin Lv
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Virtual Try-On (VTON) systems deployed on large public displays face a dual barrier: the physical strain of mid-air interaction and the social inhibition caused by public self-consciousness. This paper presents a real-time 3D avatar system integrating markerless motion capture with dynamic visual fidelity control to investigate and mitigate both barriers. Through a dual-study empirical evaluation, we first decoupled physical fatigue from gesture interaction ( N=20 ), demonstrating that interaction fatigue is primarily driven by visuomotor latency rather than the physical act of gesturing; our optimized low-latency gesture pipeline achieved usability comparable to touchscreens while delivering superior immersion and hygiene. Building on these insights, our second study ( N=25 ) investigated the “avatar fidelity paradox” via a 2 \times 2 factorial design manipulating interaction modality (gestures vs. touch) and visual fidelity (photorealistic MetaHuman vs. stylized mannequin). Results reveal that while high fidelity and mid-air gestures independently maximize virtual embodiment ( p .05 ), their combination elicits the highest social awkwardness. Crucially, low-fidelity avatars serve as a “psychological mask” that alleviates public embarrassment during expressive gestures, while mid-air gestures simultaneously act as a compensatory mechanism to preserve perceived try-on trust despite reduced visual realism. Finally, we propose a context-aware fidelity framework to balance privacy, immersion, and commercial trust in public spatial interactions.
[HC-3] Evaluating SAT Solver Metrics as Predictors of Human-Perceived Nonogram Difficulty
链接: https://arxiv.org/abs/2608.23300
作者: Changdao He,Yibing Ju,Jonathan Calver,Alice Gao
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注: 16 pages, 7 figures, 4 tables
Abstract:Algorithmic solver effort is often assumed to align with perceived puzzle difficulty, but this assumption is rarely tested against human solving data. We evaluate this assumption for Nonograms, a popular logic puzzle similar to Sudoku in which numeric clues along each row and column determine a unique solution grid. We formulate Nonograms as a constraint satisfaction problem and solve them using existing SAT solvers. We then conduct a user study in which we collect data on both participant interactions and reported difficulty. We find that neither participants’ reported difficulty nor their behavioural signals correlate meaningfully with SAT solver metrics; however, we find evidence that expertise moderates the relationship between solver metrics and reported difficulty. In this process, we uncover distinct, recurring solving strategies that indicate human preference for complex propagation, diverging from solver-measured complexity.
[HC-4] Expectations and Practices around AI Disclosure in CS Research EMNLP2026
链接: https://arxiv.org/abs/2608.23271
作者: Arati Mohapatra,Danish Pruthi
类目: Computers and Society (cs.CY); Computation and Language (cs.CL); Human-Computer Interaction (cs.HC)
备注: Accepted to EMNLP 2026 (Findings)
Abstract:As generative AI tools find increasing use in research workflows, ongoing debates on their impact, appropriateness and responsible use have led policymakers to enact policies to disclose AI use at multiple publishing venues. However, are current AI disclosure policies and practices reflective of their purpose? In this work, we first investigate disclosure policies of top computer science venues and find that despite their prevalence, they remain highly under-specified. Secondly, through a survey of computer science researchers (N= 109 ), we characterize the necessity of disclosures across different research tasks and levels of human involvement. We learn that researchers find disclosures most necessary for tasks involving research design, and for tasks when the human involvement is low. We also compile expectations that researchers have about the information to be conveyed in AI disclosure statements. Lastly, through an analysis of 13867 disclosure statements from EMNLP 2025 and ICLR 2026 , we reveal a large disconnect between these expectations and AI disclosures in practice—a prime example being writing assistance which is deemed less necessary but frequently disclosed. We conclude with recommendations for authors and policymakers that seek to align AI disclosure policies and practices with expectations.
[HC-5] AI emotional support is better only when chosen but shifts preferences even when it is not
链接: https://arxiv.org/abs/2608.23196
作者: Yaoxi Shi,Cathy Mengying Fang,Guy LabanPattie Maes,Amit Goldenberg
类目: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注:
Abstract:People increasingly face a novel decision when seeking emotional support: human or AI. In existing studies, AI’s empathic messages are rated as well as or better than humans’. But these studies either assigned the support source or honored people’s choice. In real life, support is often incongruent with choice, as people want one source and receive the other. Across three experiments (N = 1,951), participants chose whether to share an emotional experience with a human or an AI, then were randomly assigned to a congruent or incongruent partner. AI support was rated as superior only among those who had chosen it. Yet regardless of congruence, interacting with AI increased willingness to choose it again. In a 28-day study with OpenAI (N = 981), daily conversations shifted preferences toward AI and away from humans, but only when conversations turned personal. Emotional support choices are thus path-dependent, progressively redirecting away from human connection.
[HC-6] A Simulator-Grounded Framework For Constructing Verifiable Muscle-Grounded QA From 3D Tongue Meshes
链接: https://arxiv.org/abs/2608.23137
作者: Seungho Eum,Unsang Park
类目: Computer Vision and Pattern Recognition (cs.CV); Human-Computer Interaction (cs.HC)
备注: 15 pages, 4 figures, 15 tables
Abstract:Existing articulatory corpora based on real-time MRI and electromagnetic articulography capture tongue shape and motion but do not provide traceable labels for the muscle-driven process that generated an observed configuration. We introduce a simulator-grounded data-construction framework and instantiate it as 3DTongueQA. Controlled 11-dimensional muscle activations are mapped to fixed-topology tongue meshes with the ArtiSynth Badin finite-element model, converted into structured biomechanical records, and rendered as deterministic QA on muscle state, geometry, and target-directed change. We screen 295,157 configurations, retain 295,115 valid meshes, and construct 891,156 QA records per language. Language naturalization changes only surface form and is verified against the source records; English and Korean instantiations demonstrate construction-level portability. A swappable SpiralNet+±-Qwen3-8B baseline reaches 62.9 \pm 9.2 Muscle EM, 74.0 \pm 0.2 Value Accuracy, and 65.9 \pm 4.7 Direction EM, while mismatching the paired mesh reduces Muscle EM to 2.2; a dataset-leakage-controlled anchor-held-out model retains 80.4–98.6% of the full-inventory scores on unseen anchors. Task-specific structured readouts further reach 88.7 \pm 0.7 Muscle EM and 93.3 \pm 1.0 Direction EM. These complementary results show that the constructed supervision supports both efficient structured prediction and heterogeneous natural-language QA rather than being tied to a particular decoder architecture.
[HC-7] What Makes an Initial Reaction Ready for Discussion?: Multi-Persona AI Support for Stance Reflection and Writing
链接: https://arxiv.org/abs/2608.23050
作者: Sky Shih-Kai Hong,Mu-Tien Kuo,Wei-Ji Chen
类目: Human-Computer Interaction (cs.HC)
备注: 5 pages, 3 figures, 2 tables. Accepted to TAICHI 2026 ( this https URL )
Abstract:An initial reaction to a social or community issue can feel meaningful before it is ready to become a message: people still need to clarify the claim, anticipate audience risks, and decide how much reasoning should become visible to others. We present StanceLab, a prototype for preparing a stance before entering a discussion. The prototype compares a three-persona mode, where an Interviewer, Mentor, and Opponent respond in parallel to help users diagnose and revise a stance, with a standalone LLM mode. In a formative within-subject pilot with six participants and 12 task sessions, every session produced a short final message in the notepad. The pilot revealed two design requirements: persona roles should diagnose useful blind spots or objections, and parallel responses need coordination support. We propose a future diagnosis-and-writing workflow that turns persona-based reflection into selective, audience-aware final messages.
[HC-8] Free-Energy-Gated Plasticity for Real-Time Online Motor Learning in Physical Human–Robot Interaction
链接: https://arxiv.org/abs/2608.23000
作者: Hiroki Sawada,Jun Tani
类目: Robotics (cs.RO); Human-Computer Interaction (cs.HC)
备注:
Abstract:Fully online embodied learning requires synaptic adaptation to acquire new behaviors while preserving previously learned dynamics during ongoing interaction. We extend the Predictive-Coding-inspired Variational Recurrent Neural Network (PV-RNN) to continuously adapt its synaptic weights and propose Free-Energy-Gated Plasticity (FEGP), which regulates the effective learning rate according to variational free energy. In real-time physical human–robot interaction, a randomly initialized network acquired three cyclic motor patterns without offline pretraining, replay, or task-boundary signals, with all three patterns emerging in autonomous rollouts. Controlled experiments over ten randomized teaching streams and five network initializations per stream showed that FEGP substantially improved repertoire coverage and retention of previously acquired patterns after they left the recent observation window. Neither a constant learning rate matched to the gate’s time-averaged effective rate nor replay of the same gain values with disrupted temporal organization reproduced these improvements. These results indicate that the temporal allocation of plasticity relative to model–environment mismatch, rather than simply its average magnitude or distribution, is critical for maintaining previously acquired behaviors during continued online learning.
[HC-9] LLM Pedagogical Behavior in AI Tutoring Interactions
链接: https://arxiv.org/abs/2608.22993
作者: Suhyeon Lee,Juneha Baek,Jaehyeong Park,Donghyuk Shin
类目: Computation and Language (cs.CL); Human-Computer Interaction (cs.HC)
备注:
Abstract:Students increasingly use LLMs as tutors for coursework and problem solving. Little is known about the level of assistance LLMs provide when students use them as tutors in authentic learning interactions. This matters because tutoring responses can differ substantially in how directly they help students complete a task. We operationalize this dimension as scaffolding level and develop a five-level scale, validated against human annotations, that characterizes responses according to the degree of direct assistance they provide. We apply the scale to 14,637 LLM responses from 203 students in a university AI course. Responses are overwhelmingly concentrated at high levels of assistance, with more than 95% classified as either Explaining or Solving. Scaffolding level is systematically associated with students’ subsequent conversational behavior, but provides little additional predictive information about performance on three subsequent exams beyond prior achievement and dialogue behavior. These findings provide an empirical baseline for LLM assistance in tutoring interactions and a measurement framework for evaluating how alternative tutoring designs change that assistance.
[HC-10] AI-based worker guidance in assembly and disassembly operations using multimodal ego/exo-centric data capture and structured task knowledge
链接: https://arxiv.org/abs/2608.22617
作者: Vivek Chavan,Jörg Krüger
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注: 5 pages. Published in CIRP Annals - Manufacturing Technology
Abstract:Assembly and disassembly processes rely on expert knowledge that is difficult to document, reuse, and transfer. This paper presents a data-centric approach for extracting structured task knowledge from expert demonstrations using egocentric and exocentric recordings. Temporal and multimodal information from video and narration is jointly encoded to derive structured task representations that enable procedural documentation and context-aware worker guidance. The approach is evaluated on a real-world disassembly case study, demonstrating that video-based representations capture procedural structure and execution context beyond static image-based methods. The results highlight the potential of egocentric video understanding for repair, training, and circular manufacturing applications. Project website: this https URL
[HC-11] Hybrid Panels: Toward Human-AI Collaboration in Survey Research
链接: https://arxiv.org/abs/2608.22582
作者: Julia Romberg,Tobias Gummer,Gabriella Lapesa,Tanja Kunz,Claudia Wagner
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Human-Computer Interaction (cs.HC)
备注: 15 pages, under review
Abstract:Large-scale population surveys are essential for generating robust social and scientific insights, yet they face significant challenges, including declining response rates, increasing data collection costs, long delays between data collection and data provision, and the risk of nonresponse bias. Advances in artificial intelligence (AI) have opened up new opportunities for AI-supported survey infrastructures where the goal is to overcome these challenges without limiting the data quality. A promising AI-enabled survey infrastructure for which we build a first pilot is a hybrid panel. A hybrid panel is a longitudinal AI-enabled survey which allows to iteratively improve the alignment between large language models (LLMs) and the population they aim to simulate and use the errors to inform the design and implementation of the next survey wave (e.g., inform the participant recruitment, assignment of questions to participants). It incorporates both human participants and LLMs as fundamental elements of its design. In this research note, we introduce the concept of a hybrid panel by providing a definition and outlining an overarching framework, spanning data collection to data validation. We detail results from a first pilot study to illustrate (open) challenges that we identify for hybrid panels.
[HC-12] “I want to be pushed I want to grow”: Enabling social workers to design evaluations of LLM augmentation in their work
链接: https://arxiv.org/abs/2608.22459
作者: Anna Kawakami,Chloe Qianhui Zhao,Renee Shelby,Fernando Diaz,Haiyi Zhu,Kenneth Holstein
类目: Human-Computer Interaction (cs.HC)
备注: Accepted at AIES 2026
Abstract:Workers are increasingly asked to adopt AI systems to assist their work, yet are rarely given a voice in defining what meaningful AI augmentation should look like or how to evaluate for it. In this paper, we propose worker-driven AI measurement—a bottom-up approach to AI evaluation where workers collaboratively shape decisions about which tasks AI should augment, what “successful” augmentation looks like, and how it should be measured. We explore how to support this through a case study with 19 workers from a local school social work organization. Through a series of eight workshops, workers iteratively develop their own measurement goals for AI evaluation, systematize these goals, and then design a benchmark to capture how effectively an LLM can “challenge” them to reflect on their own assumptions and biases in the context of their day-to-day work. Workers collaboratively design and refine an LLM-as-a-judge rubric based on their professional and lived expertise. In validations of the worker-created benchmark, we find that there is strong agreement between worker and LLM judge ratings and that the resulting benchmark can differentiate performance across six state-of-the-art LLMs. Based on our case study, we discuss opportunities for future work to support worker-driven AI measurement as a complementary approach to existing top-down AI evaluation approaches.
[HC-13] Probing How Users Interact with Turn-Level Design Frictions for AI Chatbots
链接: https://arxiv.org/abs/2608.22427
作者: Helen Weixu Chen,Katy Ilonka Gero
类目: Human-Computer Interaction (cs.HC)
备注: 32 pages, 8 figures, currently under review at ACM IUI 2027
Abstract:AI chatbots can help people write faster, but they can also encourage overreliance by making it easy to turn minimal input into usable text. We study turn-level design friction: intentional constraints added to each chatbot exchange that slow, limit, or redirect how users request, access, or use model responses. We designed six friction probes, organized around three mechanisms: eliciting user contribution, restricting access to generated content, and reshaping system output. In a within-subject study with 24 participants, all six probes increased workload, task duration, and perceived ownership relative to a conventional AI chatbot, while their effects on recall and recognition were more selective. We further found that participants adapted to friction in different ways, and that the same constraint could support or obstruct involvement depending on users’ goals and workflows.
[HC-14] All four leading LLM s talk more than they listen to personality-verified synthetic help-seekers
链接: https://arxiv.org/abs/2608.22425
作者: Pablo A. Fonseca,Raquel Rodríguez-Carvajal,Rafael A. Calvo
类目: Human-Computer Interaction (cs.HC); Computation and Language (cs.CL); Computers and Society (cs.CY)
备注: 30 pages, 6 figures, 24 tables. Data and code: this https URL
Abstract:Large language models are increasingly consulted at moments of distress, yet single-turn benchmarks neither test sustained exchanges nor distinguish between users. We built a personality-aware evaluation in which four widely used models advised several synthetic help-seekers, each given a psychometrically specified profile, in an acute crisis: a caregiver learning of a relative’s dementia diagnosis. Auditors blind to the profile prompt recovered the specified bands from dialogue alone with high agreement on every instrument (ICC(2,4) = 0.91; 0.79-0.96 by instrument; band-score r = 0.78), as expected for the Big Five but equally for coping style, coping self-efficacy, resilience and reactance, which the lexical approach never covered. Such evaluation therefore reaches beyond the Five Factor Model to motivational, regulatory and self-appraisal dispositions. The four models were not distinguishable on emotion stabilisation and failed alike, sharing three modes: verbosity, a talk-to-listen ratio above one, and problem-solving before the situation had been explored.
[HC-15] LLM s for Survey Text Analysis - A Performance Comparison Between Humans and GPT -5 on Inductive Content Analysis
链接: https://arxiv.org/abs/2608.22417
作者: Leonardo Bergmann,Renata Gheorghiu,Ana Gvritishvili,Alex Mican,Chris Stewart,Topias Tolonen-Weckström
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Human-Computer Interaction (cs.HC)
备注:
Abstract:Large language models (LLMs) are increasingly used to support text analysis in qualitative research, yet evidence on their performance in inductive content analysis remains limited. This study compares human and LLM-based inductive coding of open-ended survey responses from 903 answers across six variables from a European PhD student survey. Five human coders performed inductive content analysis following a standardized coding scheme, while an LLM (GPT-5.4) conducted the same task using an established prompting procedure. Agreement between human and LLM outputs was assessed using the Adjusted Rand Index (ARI). Results showed an alignment between humans and the LLM, with ARI values of 0.61 for coding and 0.54 for theme generation. These values were close to the internal consistency of coding and theme results within humans (ARI = 0.68) and the LLM (ARI = 0.76). Agreement varied widely across variables, with low within-entity consistency consistently linked to low between-entity agreement, underscoring the role of data characteristics and individual performance in reliability. Overall, the findings suggest that LLMs can approximate human coding in this case-specific setting, particularly at the coding level, and may serve as a scalable support tool for inductive qualitative analysis.
[HC-16] Addressing the Selection Problem in Explainable AI IJCAI26
链接: https://arxiv.org/abs/2608.22356
作者: Claire Vlases,Katelyn Morrison
类目: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注: Accepted to the Workshop on Explainable Artificial Intelligence at the International Joint Conference on Artificial Intelligence 2026 (XAI@IJCAI26)
Abstract:Explainable AI (XAI) research has produced a plethora of explanation techniques, yet user studies repeatedly show that available explanations are not effective in practice. We argue that, given the siloed nature of conventional XAI, users are struggling to select the appropriate XAI technique. Viewing XAI through a philosophical lens, we offer a formalization of what we call the selection problem: the systematic failure of XAI interfaces to bridge the gap between a user’s natural-language uncertainty and the explanation technique that resolves it. Following a logical premise-conclusion format, we show that conventional interfaces require users to translate their uncertainty into a technique selection, a challenging prerequisite to meet. We also propose a structural solution: a multi-agent LLM orchestration tool that translates the user’s query to the proper XAI explanation technique. We provide an example of how this structural solution could be instantiated to address the selection problem.
[HC-17] Correctness Is Not Homogeneous Evidence: A Correctness-conditioned Evidence-aware Knowledge Tracing Model
链接: https://arxiv.org/abs/2608.22267
作者: Fuzheng Zhao
类目: Human-Computer Interaction (cs.HC)
备注: 24 pages, 2 figures, 11 tables. Preprint
Abstract:Knowledge tracing models usually use response correctness as a central observation for estimating students’ latent knowledge states. However, the same correct or incorrect response may arise from different behavioral contexts, such as rapid guessing, hint use, or repeated attempts. Treating correctness as uniformly informative may therefore introduce ambiguity into recurrent state updates. This study proposes Correctness-conditioned Evidence-aware Knowledge Tracing (CE-KT), which uses observable response-process features to condition how correctness is written into recurrent states. CE-KT derives weakly supervised behavioral proxy scores from response time, hint use, attempt count, and behavioral history. These scores are used as behavioral signals, not as direct measures of mastery, response quality, or cognitive state. CE-KT then uses current correctness to select a correct-response or incorrect-response gate. The selected gate modulates both the LSTM hidden state and cell state, and the modulated states are fed back into later recurrent updates. Experiments on ASSISTments data show that behavioral condition scores are associated with future same-skill performance within fixed correctness groups, especially for incorrect interactions. CE-KT generally outperforms several behavior-fusion alternatives on the main predictive metrics, although its calibration advantage is not consistent. Ablation analyses provide partial support for correctness-specific recurrent modulation and recurrent feedback. These findings suggest that behavioral information can help condition the interpretation of response correctness in knowledge tracing, but the proposed proxy scores should not be treated as direct evidence of true mastery or causal learning effects.
[HC-18] CAIA in Practice: Field Evaluation of an AI-Assisted Support System for Text-Based Online Counselling ICTAI2025
链接: https://arxiv.org/abs/2608.22251
作者: Philipp Steigerwald,Nico Bienlein,Jennifer Burghardt,Mara Stieler,Robert Lehmann,Jens Albrecht
类目: Human-Computer Interaction (cs.HC); Computation and Language (cs.CL)
备注: Accepted at IEEE ICTAI 2025
Abstract:Rising global demand for mental health support creates significant service delivery challenges, with asynchronous email counselling serving as a crucial low-threshold channel for accessing care. This paper presents CAIA, a co-designed AI-based tool suite that demonstrates responsible AI integration into counselling practice through seven LLM-driven functions enhanced by retrieval-augmented generation. A field evaluation involved 34 professional counsellors conducting authentic sessions with trained student counsellees (36 threads, 321 messages, 1,257 AI outputs). User behaviour analysis confirms substantial adoption, revealing that professional autonomy and information accuracy are decisive for sustained acceptance, with counsellors particularly valuing interpretive functionalities that provide new perspectives and stimulate professional reflection.
[HC-19] Multi-Agent Discovery and Resource-Aware Autonomous Exploration of Scientific Datasets
链接: https://arxiv.org/abs/2608.22045
作者: Aashish Panta,Hugo Lee,Giorgio Scorzelli,Kyongsik Yun,Valerio Pascucci
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注: 10 pages, 4 figures, 1 table. To appear in 2026 IEEE eScience Proceedings
Abstract:Modern scientific facilities and instruments generate datasets at scales that are difficult for individual researchers to discover, access, and explore. Although many datasets are publicly available, using them often requires familiarity with repository organization, data formats, multiresolution structures, and visualization parameters. We present WebVisus, a constrained and resource-aware multi-agent system for discovering and autonomously exploring remote, multiresolution scientific datasets. Given a natural-language research question, WebVisus identifies the user’s intent and launches an autonomous exploration agent that examines slices, volumes, and timesteps while adapting data resolution and retrieval quality to available client memory and computational resources. This design supports progressive exploration without complete dataset downloads or manual configuration of low-level visualization parameters using natural languages. We report the system architecture, constrained agent protocol, resource-aware access mechanism, and case studies evaluating autonomous visual exploration and resource-aware agentic access across scientific datasets.
[HC-20] An Interpretable Deep Learning Framework for Material Perception and Classification from Multisensory Tactile Data
链接: https://arxiv.org/abs/2608.21894
作者: Li Zou,Dave Hogendoorn,Yasemin Vardar
类目: Robotics (cs.RO); Human-Computer Interaction (cs.HC)
备注: 7 pages, 5 figures, journal
Abstract:Human tactile perception relies on complex multisensory cues. Yet the relationship between tactile signals and perceptual representations remains poorly understood, limiting the integration of touch in digital environments and human-like robotic perception. To address this gap, we developed a computational framework comprising three interconnected deep learning models that map multisensory touch data to material perception, without relying on hand-crafted features. The models represent progressively different routes from tactile signals to material class: from low-level interaction signals to perceptual attribute distributions (Model 1), from predicted attribute distributions to material classification (Model 2), and directly from tactile signals to material categories, bypassing intermediate representations (Model 3). By combining deep learning with Integrated Gradients, the framework achieved high accuracy while offering interpretability, revealing which sensory modalities most strongly drive its decisions. Our results show that deep learning can approach near-perfect material classification when unconstrained by intermediate perceptual stages, but matching human-like performance is harder once those stages are modeled explicitly. Notably, thermal cues emerged as particularly informative across all models, providing robust signals for material differentiation. The results offer a computational account of how tactile signals lead to material perception and show how interpretable deep learning can both approach human-level performance and reveal cues that robotic and haptic systems need to incorporate.
[HC-21] AI Watchdog: Agent Interfaces for Detecting and Defending Against Manipulative Dark Patterns in AI Conversations
链接: https://arxiv.org/abs/2608.21841
作者: Rachel Poonsiriwong,Chayapatr(Pub)Archiwaranguprok,Constanze Albrecht,Monchai Lertsutthiwong,Pattie Maes,Pat Pataranutaporn
类目: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注:
Abstract:Conversational AI increasingly shapes consequential decisions, yet users have limited support for recognizing and resisting manipulation. We present AI Watchdog, a browser-based agent interface that monitors live conversations, detects five dark-pattern categories, including sycophancy, brand bias, anthropomorphization, sneaking, and harmful generation, and alerts users when they occur. Its open-weight turn-level classifier supports independent deployment and a path toward local inference, preserving user privacy while remaining separate from the conversational AI. We evaluated AI Watchdog in a preregistered, five-condition between-subjects experiment (N = 150) comparing a no-intervention control with four configurations varying nudge timing (prebunking vs. just-in-time) and engagement mode (without vs. with cognitive forcing). Results show that participants rarely flagged manipulative turns across all conditions, and post-task awareness did not differ significantly across groups. However, just-in-time warnings without cognitive forcing were the only intervention to significantly reduce compliance with AI-steered recommendations containing dark patterns, lowering compliance from 71.7% to 53.7%, an 18 percentage-point reduction. Exploratory analyses further showed that lower misinformation susceptibility was associated with greater flagging but not lower compliance, while higher AI trust was associated with greater compliance and lower reported awareness. Together, these findings suggest that explicit recognition of conversational dark patterns and behavioral resistance to AI steering may be distinct outcomes, motivating further investigation of timely, low-friction defensive interfaces.
[HC-22] UrbanGazeVis: A Visualization System for Analyzing Eye-Tracking Data on Urban Safety Perception
链接: https://arxiv.org/abs/2608.21686
作者: Andres De La Puente,Felipe Moreno,Luis Sante,Mauro Diaz,Jorge Poco
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Perceived safety in streetscapes depends on where people look, yet how gaze relates to visual cues of urban disorder remains poorly understood. Prior work treats safety as an image-level label, offering little insight into how attention to specific elements (e.g, buildings, greenery, people, signs of decay) shapes these judgments. We present a head-mounted eye-tracking study in which 30 participants viewed and rated the safety of 150 street-view images from Rio de Janeiro using a HoloLens 2 headset. Gaze traces were mapped onto semantic segments and disorder cues (e.g., damaged walls, graffiti, overhead cables), yielding a multimodal dataset linking gaze dynamics, scene semantics, and safety scores. To analyze it, we introduce UrbanGazeVis, an interactive visual analytics system with image- and participant-centric views that connects the spatial, temporal, and semantic dimensions of gaze to perceived safety, supporting comparisons between safe and unsafe scenes, inspection of divergent ratings for similar images, and region-of-interest analysis via glyph-based summaries. Statistical models show that sustained attention to physical disorder is associated with lower perceived safety, while the visual analysis reveals context-specific effects often masked by global aggregation. Together, these analyses offer actionable insights for urban design and planning.
[HC-23] Who Bears the Cost of Honesty? A FAccT Workshop Synthesis and Research Agenda for Equitable AI Disclosure
链接: https://arxiv.org/abs/2608.21671
作者: Runlong Ye,Jessica He,Finola Finn,Angel Hsing-Chi Hwang,Donal Khosrowi,Seyun Kim,Morgan Klaus Scheuerman
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:AI disclosure is increasingly promoted and sometimes required as a route to transparency, accountability, provenance, and trust. Yet disclosure can also expose AI users to suspicion, stigma (e.g., competence penalties), and surveillance, affecting minoritized groups in particular. This paper reports on Who Bears the Cost of Honesty?, a CRAFT workshop at the 2026 ACM Conference on Fairness, Accountability, and Transparency that used scenario-anchored power mapping and design fiction to explore the benefits, harms, tensions, and power asymmetries that emerge under AI disclosure norms and mandates. We document the workshop design and analyze the disclosure approaches participants co-created, comprising four completed power maps, three context cards, and one interface prototype. These artifacts span education, workplace, politics/journalism, and interpersonal contexts. They depict disclosure as a multi-actor accountability process, surface concerns that the use of accessibility-related AI could be held against workers in performance evaluations, and explore how context-specific, bottom-up disclosures may support transparency while mitigating some risks of stigma and misinterpretation. We contribute (1) a documented two-stage workshop method; (2) an artifact-grounded thematic synthesis; and (3) a diagnostic framework, the Cost-of-Honesty Stack, with provisional design suggestions and research directions.
[HC-24] From Mastery Profile to Simulated Response: Stochastic Student Knowledge Graphs (SSKG) for Faithful LLM Student Simulation
链接: https://arxiv.org/abs/2608.21668
作者: Yuan An,Emily Wang,Benjamin Wang,Ruhma Hashmi
类目: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注:
Abstract:Large language models (LLMs) are increasingly used to simulate students at different mastery levels. These simulations can generate synthetic training data and stress-test tutoring systems. However, common prompt-based approaches leave the answer decision to the LLM, which tends to perform according to its built-in capabilities even when instructed to simulate a student with low mastery. As a result, these approaches may have difficulty distinguishing students with low and high levels of mastery. We demonstrate this limitation using 379 College Board-calibrated SAT Algebra items and five archetypal mastery profiles. Three LLMs from three vendors (Gemini 3.1 Flash Lite, Claude Haiku 4.5, and GPT-5.4-mini) achieve 96.8-100% accuracy across all profiles. To address this limitation, we introduce a method grounded in a Stochastic Student Knowledge Graph (SSKG). A curriculum knowledge graph (CKG) is extracted from an open algebra textbook, and each SAT solution is decomposed into a chain of required triples. The SSKG assigns a mastery probability to each triple, which is sampled to determine question correctness. An LLM then generates a first-person rationale consistent with the outcome. The simulation reduces accuracy to 44.1-85.2% across profiles and produces a clear monotone mastery gradient.
[HC-25] Agent ic AI for Safety-critical Multi-drone Systems: Challenges and Opportunities
链接: https://arxiv.org/abs/2608.21444
作者: Timothy Merritt,Alejandro Jarabo-Peñas,Juan Bravo-Arrabal,Maria-Theresa Bahodi,Anders Lyhne Christensen
类目: Artificial Intelligence (cs.AI); Emerging Technologies (cs.ET); Human-Computer Interaction (cs.HC); Robotics (cs.RO)
备注: 9 pages, 4 figures, presented at the AgentCraft Workshop at IUI2026 Conference, this https URL - this http URL proceedings forthcoming
Abstract:Multi-drone systems are increasingly positioned for safety-critical missions such as search and rescue (SAR) and critical infrastructure monitoring. Yet, real-world adoption remains constrained not only by autonomy performance, but by the difficulty of integrating agentic behavior into professional work: operators must understand, trust, and govern automation under uncertainty, time pressure, and accountability. This position paper synthesizes the ambitions and lessons from two ongoing efforts: NAMUR, which explores LLM-supported robot control in SAR and firefighting contexts, and PERSIST, which explores persistent drone operations for monitoring and security at critical infrastructure sites. We argue that agentic AI should be approached as a socio-technical design problem, where interfaces, oversight mechanisms, and evaluation practices are as critical as algorithms. We outline a human-centered, participatory, and iterative research approach aimed at uncovering stakeholder needs, shaping agent capabilities through successive prototypes, and producing transferable proof-of-concept systems and evaluation strategies for other safety-critical contexts.
[HC-26] EditStream: A Unified Autoregressive Framework for Interactive Video Generation and Editing
链接: https://arxiv.org/abs/2608.21424
作者: Yuqian Zhou,Zhenghong Zhou,Zongze Wu,Cameron Smith,Richard Zhang,Jiebo Luo,Eli Shechtman,Zhe Lin
类目: Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR); Human-Computer Interaction (cs.HC); Machine Learning (cs.LG); Multimedia (cs.MM)
备注: 25 pages, 12 figures, Project page: this https URL
Abstract:Interactive video generation and editing are becoming increasingly important for creative design. In this report, we introduce EditStream: a unified framework for interactive video generation and editing. EditStream unifies multiple video creation and manipulation tasks within a single DiT-based model through flexible task-specific conditioning, and further transforms it into a fast, few-step autoregressive model for efficient streaming. It supports Text-to-Video, Image-to-Video, Video-to-Video, Editing Propagation, Reference-guided Video Editing, and Camera Pose Change, enabling flexible control over video generation, transformation, and editing within one system. To make the unified model practical for interactive use, we develop a two-stage distillation approach that combines Velocity Moment Matching (VMM) with autoregressive unrolling. VMM matches conditional velocity moments at student-reached intermediate states to preserve generation quality and motion, while unrolling exposes the student to its own autoregressive predictions to improve temporal stability. Together, they alleviate common challenges in few-step autoregressive video generation, including over-saturation, degraded motion, temporal instability, and complex training. EditStream provides a practical and scalable solution that bridges high-quality diffusion-based video models with interactive creative workflows.
[HC-27] Evaluating Human and LLM -Generated Thematic Analysis in HRI for Vulnerable Populations: A Comparative and Ethical Analysis
链接: https://arxiv.org/abs/2608.21420
作者: Alva Markelius,Fethiye Irmak Dogan,Julie Bailey,Hatice Gunes
类目: Robotics (cs.RO); Human-Computer Interaction (cs.HC)
备注: Accepted at 35th IEEE International Conference on Robot and Human Interactive Communication (RO-MAN), Kitakyushu, Japan
Abstract:Thematic analysis (TA) has long been regarded as an inherently human, reflexive, and interpretive process. However, the extent to which LLM-generated TA is appropriate for Human-Robot Interaction (HRI) research involving vulnerable populations remains largely unexamined and raises critical questions about validity and ethics, particularly in sensitive research contexts. This paper presents a comparative study of human- and LLM-generated TA in an HRI context with a focus on vulnerable populations. We evaluate both objective and semantic agreement between human- and LLMgenerated themes, and examine whether observed divergences reflect systematic interpretive patterns with ethical significance. Our analysis investigates whether LLM-generated TA risks marginalising or misrepresenting the experiences of vulnerable participants, with implications for researchers employing LLM-assisted TA in HRI.
计算机视觉
[CV-0] EG-ARSA: An Expert-Grounded Open Model for Visual Road Safety Auditing in Low-Resource Settings
链接: https://arxiv.org/abs/2608.23563
作者: Md Thamed Bin Zaman Chowdhury,Moazzem Hossain
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Road traffic injuries remain a major challenge in low- and middle-income countries, where proactive road safety auditing is limited by incomplete crash records, shortages of qualified auditors, and the high cost of large-scale field inspections. To address this problem, we propose Expert-Grounded Distillation (EGD), a novel artificial intelligence framework that transfers institutional road safety expertise into a compact vision-language model for scalable visual road safety auditing. The key innovation is a quantified expert-grounding stage in which the teacher vision-language model is calibrated against authoritative field audits. Large-scale annotation is permitted only after the teacher reaches substantial agreement with expert risk assessments (Cohen’s kappa = 0.74). The calibrated teacher then generates structured supervision that is distilled into an 8-billion-parameter student vision-language model using Low-Rank Adaptation and a single leakage-free prompt. We also introduce Bangladesh Road Safety Audit (BD-ARSA), the first open, expert-grounded Bangladeshi visual road safety audit dataset containing 21,947 image-audit records with near-national coverage, and Expert-Grounded Road Safety Auditor (EG-ARSA), the first vision-language model developed specifically for this task. Experimental results show that grounded fine-tuning substantially improves ordinal risk assessment over the zero-shot baseline, while blind expert evaluation demonstrates that the compact student outperforms both its 31 billion-parameter teacher and Gemini-2.5-Flash. These findings demonstrate that EGD provides an effective and scalable engineering solution for proactive road safety auditing in resource-constrained environments.
[CV-1] FixAnything: 3D-Consistent Rendering Refinement via Video Generative Priors ECCV2026
链接: https://arxiv.org/abs/2608.23549
作者: Khiem Vuong,Deva Ramanan,Srinivasa Narasimhan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Appearing in ECCV 2026. Project page: this https URL
Abstract:Rendering views using 3D scene representations such as Gaussian Splatting (3DGS), Neural Radiance Fields (NeRF), meshes, or even point clouds produces artifacts when input views are sparse or target views lie far from the input. Recent work mitigates these artifacts using diffusion-based generative priors, but is specialized to individual representations and require custom architectures or extensive retraining. We present FixAnything, a single model for fixing a wide range of rendering artifacts. It does so by repurposing a pretrained video generative model, leveraging its implicit multi-view priors with only minimal modification and lightweight finetuning. Our key insight is that even noisily-rendered sequences preserve camera motion and coarse scene structure, allowing cleanup to be formulated as video-to-video translation. To control what scene structure should be preserved, we introduce a binary mask denoting the clean pixels, enabling the model to anchor its output to high-quality inputs (e.g. training views) while refining the rest. To encourage FixAnything to produce 3D-consistent renderings that support downstream reconstruction, we use camera pose accuracy (recovered via structure-from-motion) as a reward signal for direct preference optimization (DPO). Across four distinct 3D representations, FixAnything consistently improves rendering quality with lightweight finetuning, demonstrating that a single generalist video prior can replace multiple specialist refinement pipelines. The simplicity of the framework enables immediate adoption of stronger future video models without architectural redesign.
[CV-2] Predicting Multiple Clinical Outcomes Related to Functional Recovery and Social Isolation Among Older Adults After Lower-Limb Fracture or Hip Replacement
链接: https://arxiv.org/abs/2608.23531
作者: Santosh Ray,Pratik K. Mishra,Ali Abedi,Charlene H. Chu,Amir Ahmad,Shehroz S. Khan
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Older adults recovering after lower-limb fracture or hip replacement may experience complex recovery trajectories. Most of the time, these clinical aspects are studied in isolation, masking their joint impact on recovery. This study used the MAISON-LLF dataset, which contains multimodal sensor and clinical assessment data from 18 older adults recovering in the community after lower-limb fracture or hip replacement. Participants were monitored for up to eight weeks, corresponding to a maximum of 1,008 participant-days of sensor monitoring. Forty-six daily features were extracted from indoor motion, acceleration, step count, heart rate, out-of-home mobility, and sleep data. Five clinical outcomes were assessed every two weeks: the Social Isolation Scale, Oxford Hip Score, Oxford Knee Score, Timed Up and Go test, and 30-second Chair Stand test. We utilize an inherent relationship between multi-modal sensor data and different clinical scores and formulate it as a multi-output regression problem. We tested various machine learning and deep learning single- and multi-output regression algorithms to predict these scores simultaneously. The results showed that predicting clinical scores jointly was better than separately. The tabular DL multi-output regressor, NODE, gave a remarkable performance of MSE=3.96 and MAE=1.02 in comparison to other multi- and single-output regressors. The SHAP feature analysis further showed the importance of including multimodal sensors to provide a good estimate of patients’ recovery trajectory. This work may support the simultaneous assessment of functional recovery and social engagement among community-dwelling older adults and ultimately help improve their care and quality of life.
[CV-3] Investigating Relational Reasoning in VLMs
链接: https://arxiv.org/abs/2608.23518
作者: Adhithya Laxman Ravi Shankar Geetha,Aulia Kharis Rakhmasari,Haleema Ramzan,Xander Yap
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision-Language Models (VLMs) achieve strong performance in visual reasoning tasks, but it remains unclear whether they understand visual relations, or simply employ shortcuts such as language cues or priors. To investigate this, we use the Qwen3-VL-4B (Bai et al., 2025), a modern VLM, to decode how visual information is encoded across depths. For this, we propose a synthetic dataset of simple geometric shapes for controlled analysis, along with queries crafted to precisely test language cues. Furthermore, the dataset is modified to test causal reliance on visual evidence. Our results show that current VLMs combine genuine visual reasoning with shortcut strategies primarily rooted in language cues.
[CV-4] Action-Aligned Retrieval with Pairwise Multimodal Reranking for Text-Based Person Anomaly Search ECCV2026
链接: https://arxiv.org/abs/2608.23503
作者: Thanh-Khoi Nguyen,Thanh-Nhan Vo,Trong-Thuan Nguyen,Minh-Triet Tran
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to the AI City workshop @ ECCV 2026
Abstract:Text-based person anomaly search requires distinguishing individuals based on fine-grained, context-dependent behaviors rather than mere appearance. Existing methods struggle to capture these context-conditioned actions, frequently relying on isolated skeletal geometry, discarding raw query details during reformulation, or utilizing absolute pointwise scoring for multimodal verification. To address these limitations, we propose \textbfActPair, a unified three-stage coarse-to-fine framework that combines action-aligned retrieval with pairwise multimodal reranking to bridge the pose-semantic gap. First, we fine-tune a vision-language model (VLM) with an action-aligned multi-task objective that encourages the representations to encode action-discriminative semantics. Second, we perform parallel late-fusion retrieval using the original query and a large language model (LLM)-generated context-grounded rewrite, retaining complementary details from both semantic views. Finally, we propose an efficient off-the-shelf reranking module that leverages a pivot-promote algorithm to perform direct pairwise visual comparisons, mitigating residual spatial and compositional ambiguities without the prohibitive inference costs of exhaustive evaluation. Extensive experiments demonstrate that our framework achieves the best results among the compared methods on the Pedestrian Anomaly Behavior (PAB) public test and transfers effectively to an unseen, non-anomaly-specific dataset.
[CV-5] SVD-Based Typicality Maps for Out-of-Distribution Detection in Vision Transformers
链接: https://arxiv.org/abs/2608.23499
作者: Aldo Sean Sartor,Leandro de Souza Rosa,Andriy Enttsel,Mauro Mangia,Riccardo Rovatti
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:We present a method for analyzing the internal representations of Vision Transformers (ViTs) exploiting the geometry of their learned parameters. Each affine layer’s weight matrix is factored via Singular Value Decomposition (SVD), and activations are projected onto the leading right singular vectors to obtain compact, layer-intrinsic representations. A class-conditional density model is then fitted at each layer, producing per-class \emphtypicality scores that are stacked across depth into \emphtypicality maps: two-dimensional summaries of how class-specific evidence evolves through the network. From these maps, we derive two post-hoc scores for Out-Of-Distribution (OOD) detection: a \emphPrototype Alignment Score (PAS), measuring agreement with class reference prototype patterns, and a \emphMulti-Layer Soft Voting (MLSV) score, capturing cross-layer consensus without stored prototypes. On ViT-B/16 fine-tuned on CIFAR-100, the proposed scores achieve competitive detection performance without retraining or OOD exposure.
[CV-6] GeoWAM: Visual Geometry World Action Models for Autonomous Driving
链接: https://arxiv.org/abs/2608.23486
作者: Yiren Lu,Xin Ye,Jiaming Liu,Jin Yao,Yi-chung Chen,Liam Merino,Dhruva Dixith Kurra,Min Cai,Tom Lampo,Yu Yin,Danhua Guo,Burhan Yaman
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注:
Abstract:World action models (WAMs) have recently gained increasing attention as a framework for jointly modeling scene evolution and ego actions in autonomous driving. Most existing WAMs learn scene dynamics in pixel space by combining a video-generation backbone for future-observation prediction with an action head for ego-trajectory prediction. Pixels, however, provide only an indirect representation of these dynamics: they entangle geometry and motion with appearance, texture, and illumination, forcing the model to infer three-dimensional transformations from two-dimensional observations. We argue that geometry, represented by point clouds, offers a more natural state space for driving because it explicitly captures spatial structure and the rigid and non-rigid transformations that govern scene evolution while directly aligning with the space in which driving actions are executed. Building on this insight, we introduce \textbfGeoWAM, a visual geometry world action model for autonomous driving. Rather than predicting future images, GeoWAM is pretrained to forecast future scene geometry, yielding representations that jointly encode spatial structure and temporal evolution. A geometry-conditioned action head then leverages these learned geometric dynamics to predict future ego trajectories. Extensive open-loop and closed-loop evaluations show that visual geometry world modeling yields substantially stronger driving policies than image-based alternatives, establishing future-geometry prediction as an effective pretraining objective for autonomous driving.
[CV-7] Geometry-Driven Opti-Acoustic Co-Registration and View-Invariant Reflectivity Mapping for Side-Scan Sonar
链接: https://arxiv.org/abs/2608.23479
作者: Taqi Hamoda,Nuno Gracias
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Side-Scan Sonar (SSS) is a primary modality for large-scale underwater mapping, yet automated perception and cross-modal alignment are severely bottlenecked by acoustic complexities such as speckle noise, shadows, and extreme viewpoint dependencies. Traditional handcrafted descriptors and modern deep learning matchers fail to bridge the physical domain gap between optical and acoustic imagery without 3D geometric constraints. To overcome these limitations, we propose a novel geometry-driven framework for pixel-level opti-acoustic co-registration and view-invariant reflectivity mapping. Our method utilizes Structure-from-Motion (SfM) to reconstruct a dense 3D seafloor mesh, acting as a geometric anchor between the visual and acoustic domains. We introduce a First Bottom Return (FBR) extraction algorithm to dynamically correct non-linear altitude drift caused by uncalibrated SfM reconstruction. Furthermore, we apply an inverse Lambertian model and a dual-Gaussian weighting function to isolate the intrinsic seabed reflectivity, effectively neutralizing slant-range propagation loss and geometric view-dependence. By deterministically associating these isolated acoustic properties with optical pixels, our pipeline generates highly accurate, strictly co-registered multi-modal datasets. This automated, physics-guided approach eliminates the need for manual annotation and paves the way for advanced self-supervised learning in benthic habitat mapping.
[CV-8] Act with Intent: Distilling Behavior Intent for Vision-Language-Action Models
链接: https://arxiv.org/abs/2608.23478
作者: Sangoh Lee,Sangwoo Mo,Wook-Shin Han
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL
Abstract:Vision-Language-Action (VLA) models can turn multimodal context into robot actions, but their action decoders are still trained largely by behavior cloning. This supervises which motor command was demonstrated while leaving implicit the local objective served by the behavior under the instruction. Future-based supervision enriches action learning with frames, latent observations, trajectories, or motion representations, but these signals capture particular realizations of what may happen rather than the shared semantic objective of the forthcoming behavior. We propose Intention Distillation (INDI), which distills behavior-level intent into the action decoder. During training, a frozen teacher VLM interprets a demonstrated segment from the current observation, instruction, coarse action summary, and corresponding execution video. From its standard inputs, the deployed VLA recovers the resulting multimodal intent representation at an intermediate decoder layer and uses it to organize action prediction together with representations of how the behavior unfolds and what it achieves. On SimplerEnv-Bridge, INDI improves GR00T-N1.7 from 64.3% to 84.7%, and on RoboCasa Kitchen it improves the controlled GR00T-N1.7 baseline from 64.1% to 70.3%, with consistent gains on \pi_0.5 across both benchmarks. In real-world tasks, INDI improves average success from 62.0% to 68.7%, with gains of up to 12.0 pp on longer-horizon tasks. Further analyses show that the recovered latent is used by the decoder, captures behavior objective and execution progress, and organizes downstream predictions in an objective-dependent manner. These results show that action decoders benefit from explicitly modeling the semantic objective of the behavior they generate.
[CV-9] owards Comprehensive Basketball Understanding
链接: https://arxiv.org/abs/2608.23435
作者: Yirong Hu,Jiayuan Rao,Yu Zhang,Shangzhe Di,Weidi Xie
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 26 pages, 3 figures
Abstract:Understanding a basketball game requires recognizing events, localizing actions, identifying players, and relating these to structured game knowledge. Existing benchmarks primarily evaluate these abilities one at a time, leaving the interactions among these abilities under-explored. We introduce BasketballBench, a multimodal benchmark comprising 7,980 questions across ten tasks in text, image, and video. It is built from the 2025-2026 NBA season and includes official playby-play, rosters and profiles for 530 active players, and 2,501 possession-level broadcast clips. We further propose BasketballSkills, an agent that composes eight basketball-specific perception and retrieval tools under four reusable skills that specify tool order, evidence bindings, and stopping conditions. Experiments show that current MLLMs struggle particularly on questions requiring the integration of multiple capabilities, whereas BasketballSkills outperforms them, highlighting the effectiveness of explicitly composing domain-specific capabilities for comprehensive basketball understanding.
[CV-10] Image-Conditioned Diffusion Models for Quality Assurance of Organ-at-Risk Segmentations in Radiotherapy MICCAI2026
链接: https://arxiv.org/abs/2608.23432
作者: Clea Dronne,Catharine H Clark,Xavier Loizeau,Elizabeth Miles,Peter Hoskin,Jamie R McClelland
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Submitted to the MICCAI 2026 UNSURE Workshop
Abstract:Accurate organ-at-risk segmentation is essential for radiotherapy planning, but reviewing segmentations is time-consuming and subjective. We investigate normative modelling for segmentation error detection in head-and-neck CT, comparing a VAE framework with an image-conditioned segmentation diffusion model. Models were evaluated on RADCURE brainstem and spinal cord segmentations using simulated boundary and width perturbations. Error detection was assessed using the Dice similarity coefficient and the Distance to Agreement (DTA) between the input and reconstructed segmentations. While both models detected some simulated errors, regional DTA showed that the diffusion model localised subtle boundary errors more consistently. These results support image-conditioned diffusion reconstruction as a promising framework for localised, anatomy-aware segmentation QA.
[CV-11] Photorealistic Novel View Synthesis of Human Faces using Next-Scale Transformers
链接: https://arxiv.org/abs/2608.23410
作者: Federico Stella,Fei Jiang,Zhongshi Jiang,Zohar Barzelay,Emanuel Garbin,Amin Jourabloo,Liuhao Ge
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Photorealistic novel view synthesis of people remains challenging at high spatial resolutions and across multiple target cameras, where preserving identity, fine appearance details, and geometric coherence is critical. We build on the next-scale autoregressive paradigm and adapt it for human-centric view synthesis by enabling higher image resolutions, multi-view outputs and stronger cross-view consistency in a single forward pass. We train on a synthetic dataset of human faces spanning diverse identities and apparel. Contrary to diffusion models, this paradigm does not need 2D pre-training and, thanks to its next-scale architecture, it benefits from lower-resolution, general-purpose pre-trainings, with the full-sized purpose-specific images being used only in the last training stages. This enables our architecture to converge with a smaller amount of purpose-specific training data, allowing us to use a smaller but more realistic training dataset. The resulting model produces sharp and realistic views, with the option to synthesize multiple novel viewpoints simultaneously for improved agreement across views. Empirically, we observe gains in perceptual fidelity and cross-view coherence on human subjects, demonstrating that next-scale autoregression is an effective backbone for scalable, multi-output human view synthesis. We also couple our pipeline with an existing transformer-based model for pixel-aligned 3D gaussian lifting from multi-view facial inputs, resulting in accurate and photorealistic 3D models of human faces.
[CV-12] MomADv2: Reliable Temporal Memory for End-to-End Autonomous Driving
链接: https://arxiv.org/abs/2608.23405
作者: Ziying Song,Shengkai Zhang,Lin Liu,Peiliang Wu,Lei Yang,Dongyang Xu,Bin Sun,Li Wang,Shaoqing Xu,Caiyan Jia,Yadan Luo
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: 16 pages, 6 figures
Abstract:Long-horizon planning is critical for safe autonomous driving in complex scenarios. Existing methods improve planning continuity with temporal memory, but such memory may become invalid and mislead decisions when the driving command changes. Thus, selectively leveraging useful history while suppressing command-inconsistent memory remains a key challenge. To address this issue, we propose MomADv2, a reliable state-space memory framework for long-horizon end-to-end autonomous driving. At its core, MomADv2 introduces a Selective State-Space Planning Memory Query Module, which filters historical planning queries based on temporal continuity and command consistency, selects planning modes relevant to the current command, and models the evolution of planning intentions through a selective state-space mechanism. To further alleviate local trajectory deviations and error accumulation in long-horizon planning, we design a Flow-Matching Trajectory Residual Refiner. It learns a continuous residual correction field from the refined planning output to the expert trajectory, enabling fine-grained trajectory refinement while preserving the stability of anchor-based planning. Extensive experiments on closed-loop NAVSIM and Bench2Drive, as well as open-loop nuScenes, demonstrate that MomADv2 improves long-horizon planning consistency and reduces the average collision rate by 15.6% over MomAD under 6-second planning.
[CV-13] Long-Horizon Audio-Visual Generation for Persistent Stories and Interactive Worlds
链接: https://arxiv.org/abs/2608.23383
作者: Nan Duan,Haoyang Huang,Weiyang Jin,Haoran Li,Yaowei Li,Yuming Li,Yijun Liu,Xin Lu,Xiaoxiao Ma,Yanwen Ma,Yaofeng Su,Yilang Sun,Haoyu Wang,Zeyue Xue,Songchun Zhang,Junhao Zhuang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL
Abstract:Video generation is progressing beyond isolated clips toward long-form narratives and interactive worlds, requiring models to preserve identities, follow user controls, and remain stable over extended rollouts. We present JoyAI-Echo-1.5, a unified audio-visual generation system with two purpose-built variants. The long-video variant introduces composable cross-shot memory that aggregates visual evidence across multiple prior shots and speaker cues derived from speech-filtered full-shot audio, enabling persistent character appearance and voice identity across flexible combinations of text, image, and memory conditioning. The world-model variant converts heterogeneous navigation inputs into calibrated metric 6-DoF camera trajectories and injects them through a geometry-aware conditioning pathway, enabling controller-agnostic interaction across flexible viewpoints. To support efficient long-horizon generation, we transform a bidirectional audio-visual backbone into a causal few-step generator using progressive teacher forcing and short- and long-horizon Self-Gradient Forcing on self-generated rollouts. Experiments demonstrate strong performance in both settings. JoyAI-Echo-1.5 achieves improvements over existing long-video baselines in cross-shot consistency, visual quality, text alignment, and speech fidelity. Its world-model variant ranks first on WBench, with an average score of 81.7, and achieves leading visual quality and long-horizon persistence on SANA-WM-Bench. Together, these results indicate that memory, geometric control, and rollout-aware training provide a practical foundation for generating coherent stories and continuously evolving interactive worlds. Project page: this https URL.
[CV-14] DF-MoE: Generalizable Deepfake Detection via Multimodal Sparse Mixture-of-Experts BMVC2026
链接: https://arxiv.org/abs/2608.23363
作者: Vlad Hondru,Florinel Alin Croitoru,Iuliana Georgescu,A. Sophia Koepke,Radu Tudor Ionescu
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted at BMVC 2026
Abstract:Audio-visual deepfake detection is an actively studied topic, where one of the main challenges is to develop detectors able to generalize across deepfake generation methods. We conjecture that overfitting can be mitigated by extracting multiple high-level cues from the available audio and visual modalities via pre-trained models. We therefore assemble a wide variety of pre-trained models to extract features that encode mouth movements, face parsing, facial expressions, head pose, gaze tracking, heart rate, audio emotion and speech activity. We further integrate both unimodal and multimodal cues via a Mixture-of-Experts (MoE) backbone to detect deepfakes. We perform in-domain and cross-domain experiments on five benchmarks for deepfake detection (MAVOS-DD, AVLips, PolyGlotFake, BioDeepAV, FakeAVCeleb) to compare our framework (DF-MoE) with state-of-the-art methods. Our results indicate that DF-MoE obtains superior deepfake detection results, surpassing all competing methods. We release our code at this https URL.
[CV-15] OptiSight: Bridging Semantic Reasoning and Geometric Control for Embodied Navigation
链接: https://arxiv.org/abs/2608.23354
作者: Alperen Avan,Jordi Sanchez-Riera
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Autonomous indoor navigation requires both semantic understanding and precise geometric control. We propose OptiSight, a hybrid framework that combines Vision-Language Model reasoning with deterministic visual servoing through a finite-state Chain-of-Thought architecture. Grounded-SAM localizes open-vocabulary targets, while camera projection geometry converts visual observations into navigation commands without requiring dense mapping. The VLM is queried only at key decision points, reducing computational overhead while geometric control handles continuous navigation. Experiments in AI Habitat demonstrate reliable zero-shot navigation across diverse indoor scenarios, including obstacle avoidance and semantic ambiguity, while operating within an 8~GB VRAM budget. The source code is available at this https URL.
[CV-16] Controllable blind deblurring with diffusion models ICIP2026
链接: https://arxiv.org/abs/2608.23343
作者: Imane Si Salah,Emile Cribelier,Thomas Veit,Wolf Hauser,Arthur Leclaire
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 6 pages, 5 figures, 1 table. Accepted to IEEE ICIP 2026
Abstract:Image acquisition with a camera involves several degradations due to the optical system, sensor, or low-level processing steps. We address blind deblurring in professional photography: we aim to invert unknown isotropic blur without knowledge of the degradation this http URL such inverse problems,where some high-frequency information is lost, it is challenging to use generative models to produce details that are both photo-realistic and faithful to the input. We propose SuperSharpen, a diffusion-based blind deblurring method offering explicit control over restoration strength through a blur measure. We compare two conditioning strategies: a ControlNet-style adapter on a frozen backbone, and full finetuning of the diffusion prior. Our experiments show that finetuning achieves better fidelity with fewer hallucinated details. We validate our approach on synthetic and real-world blur, demonstrating improved perceptual quality and controllable restoration strength.
[CV-17] Can Coding Agents Build Robust Baselines? A Skill-Based Approach for Automating the Medical Imaging Model-Development Pipeline ICME MICCAI2026
链接: https://arxiv.org/abs/2608.23336
作者: Eugenia Moris,José Ignacio Orlando
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: MICCAI 2026 Workshop AgenticMed
Abstract:Developing competitive deep learning baselines for medical imaging remains a highly iterative process requiring literature review, implementation, experimentation, and expert refinement. Existing automation approaches typically optimize isolated components, such as architecture search or hyperparameter tuning, rather than the complete baseline development process. We present an agentic AI Scientist workflow that combines literature-guided reasoning, automated code generation, and hypothesis-driven experimentation to generate competitive baseline models for medical imaging challenges. The framework is evaluated on four public benchmarks spanning segmentation, classification, and detection. Across all tasks, the Experimentation Pipeline consistently improves validation performance, achieving competitive leaderboard results, including 6th place on both PUMA tracks (15 teams) and 31st place on MILK10k (125 teams). On MIDOG25, the resulting model also demonstrates strong domain generalization across scanners, tumor types, and species. Using the same workflow across all challenges without task-specific redesign, we demonstrate that skill-based, literature-guided agentic workflows can substantially reduce the engineering effort required to develop competitive medical imaging baselines.
[CV-18] IntentQA: Intent Question Answering in Videos by Cognitive Context Reasoning
链接: https://arxiv.org/abs/2608.23330
作者: Jiapeng Li,Ping Wei,Wenjuan Han,Song-Chun Zhu,Lifeng Fan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 18 pages, 7 figures. Accepted manuscript of an article published in IEEE Transactions on Pattern Analysis and Machine Intelligence
Abstract:Video understanding requires intelligent agents to transcend mere recognition of visual facts and comprehend the underlying intents behind human actions (often termed the “dark matter” of social intelligence). To bridge the gap between visual observation and intent reasoning, we introduce a novel task, IntentQA, and contribute a large-scale VideoQA dataset specifically tailored for this purpose. However, recognizing that standard metrics may overestimate capabilities due to dataset biases, we go beyond simple accuracy to rigorously evaluate model robustness. We augment the benchmark by generating five distinct contrast sets via Large Language Models (LLMs) and introducing a “Contrast Performance Decline” metric. We propose the X-CaVIR (eXplainable Context-aware Video Intent Reasoning) framework, which leverages three types of “Cognitive Context” to enhance video analysis: i) Situational Context via a cross-modal Video Query Language (VQL) module, ii) Contrastive Context via a Contrastive Learning module, and iii) Commonsense Context via a Commonsense Reasoning module. Crucially, to overcome the opacity of traditional black-box models, we refine the integration of LLMs within X-CaVIR by employing a transparent pipeline that synergizes video captions with VQA model outputs. This approach not only improves performance by effectively utilizing rich commonsense knowledge but also renders the reasoning process explicitly interpretable. Extensive experiments demonstrate the effectiveness of our components, the superiority of X-CaVIR over state-of-the-art baselines, and its stability against perturbations on the contrast sets.
[CV-19] hinking Beyond Videos: Unifying Video Reasoning and Deep Research for Open-World Video Agents
链接: https://arxiv.org/abs/2608.23329
作者: Wenqi Liu,Shijie Ma,Yunxiao Wang,Meng Liu,Qile Su,Han Liu,Bohan Hou,Xuanyu Zheng,Changyi Liu,Tianke Zhang,Haonan Fan,Kaiyu Jiang,Yingxin Li,Jiankang Chen,Xu Wang,Bin Wen,Tingting Gao,Han Li,Jianhua Yin,Yinwei Wei,Xuemeng Song
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Open-world video understanding often requires a model to locate sparse visual evidence and acquire external knowledge that is absent from the video and its parametric memory. While Thinking-with-Videos enables active temporal perception and Deep Research supports multi-step information seeking, the two capabilities are typically developed in isolation. We introduce VideoRover, a unified Video Deep Research framework that iteratively coordinates video cropping, multimodal search, and webpage browsing. Given a video-question pair, VideoRover uses each tool result to select the next action, so localized video clips guide external retrieval and retrieved evidence triggers further video inspection and verification. To develop this capability, we construct an automated data curation pipeline, producing 26K verified SFT trajectories and 3K challenging RL instances. We also introduce VideoRover-Bench, a benchmark stratified by video duration and research difficulty. Experiments on VideoDR and VideoRover-Bench show that our VideoRover-8B-RL achieves performance comparable to proprietary models in the direct-answer setting without tool use while outperforming larger open-source models equipped with the same tool suite. Ablation studies and training dynamics further validate the complementary roles of active video grounding, external retrieval, and long-horizon reinforcement learning.
[CV-20] Grounding Free-Form Instructions for Fashion Complementary Image Generation
链接: https://arxiv.org/abs/2608.23302
作者: Matteo Attimonelli,Claudio Pomo,Alessandro De Bellis,Danilo Danese,Dietmar Jannach,Tommaso Di Noia
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Fashion complementary image generation (CIG) aims to create garments that stylistically match a seed item based on user intent, making it a natural multimodal grounding problem where models must interpret language in visual context. Existing CIG benchmarks rely on rigid template prompts (e.g., “a photo of a skirt”), failing to reflect natural user queries and obscuring model behavior across levels of linguistic specificity. We introduce fashion complementary image generation with free-form instructions, a multimodal language-grounding setting where a model generates a compatible garment from a seed image and a natural-language instruction. To this end, we enrich three CIG benchmarks with low-, medium-, and high-specificity instructions generated by a vision-language model and validated by human annotators. We instantiate the task with StyleFlow, a Rectified Flow Matching model that jointly conditions on the seed image and instruction within a single multimodal transformer. Across image quality metrics, catalog-alignment analysis, ablations, and human evaluation, StyleFlow consistently produces instruction-aligned and stylistically coherent garments while reducing architectural complexity and inference cost relative to auxiliary-module approaches.
[CV-21] What Remains Normal? Clean Images Miss Useful Near-Defect Normal Patches for Anomaly Detection
链接: https://arxiv.org/abs/2608.23299
作者: Joongwon Chae,Runming Wang,Peiwu Qin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Memory-based anomaly detectors store nominal training patches and score test patches against this memory. A patch selected for coverage therefore becomes a nor- mal reference without a separate check that geometric rarity makes it safe to trust. We probe this coupling with sparse training contamination. Under fixed representa- tions and memory budgets, we compare random, medoid, local, and global coverage selectors. We then use CLEANCON, an out-of-bag cross-image support gate that changes candidate-image eligibility while fixing the representation, absolute mem- ory size, builder, and inference rule. Global coverage strongly over-represents sparse contamination. CLEANCON reduces final-memory contamination to approx- imately zero and increases category-macro P-AP in all 12 matched comparisons. Yet along a retention sweep, the lowest-contamination memory does not attain the highest P-AP; performance continues to improve while contamination rises. Mem- ory contamination therefore does not order the resulting memories by this http URL is publicly available at this https URL.
[CV-22] What Memory Composition Does Not Tell Us About Anomaly Detection
链接: https://arxiv.org/abs/2608.23295
作者: Joongwon Chae,Runming Wang,Peiwu Qin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Memory-based anomaly detectors store nominal training patches and score test patches against this memory. A patch selected for coverage therefore becomes a nor- mal reference without a separate check that geometric rarity makes it safe to trust. We probe this coupling with sparse training contamination. Under fixed representa- tions and memory budgets, we compare random, medoid, local, and global coverage selectors. We then use CLEANCON, an out-of-bag cross-image support gate that changes candidate-image eligibility while fixing the representation, absolute mem- ory size, builder, and inference rule. Global coverage strongly over-represents sparse contamination. CLEANCON reduces final-memory contamination to approx- imately zero and increases category-macro P-AP in all 12 matched comparisons. Yet along a retention sweep, the lowest-contamination memory does not attain the highest P-AP; performance continues to improve while contamination rises. Mem- ory contamination therefore does not order the resulting memories by P-AP
[CV-23] Spotter: Efficient Urban Visual Localization via Geo-Referenced Facade Landmarks in GPS-Degraded Environments
链接: https://arxiv.org/abs/2608.23290
作者: Antoni Valls,Jordi Sanchez-Riera
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Accurate visual localization on robotic and wearable platforms remains challenging in dense urban environments. Existing methodologies typically rely on GPS for absolute positioning, yet GPS signals frequently degrade in urban canyons due to multipath propagation. Consequently, standard solutions like visual odometry suffer from unmitigated drift over time, while map-matching techniques struggle to acquire the reliable GPS priors they need, on top of being too computationally heavy for real-time edge execution. To address these limitations, we propose Spotter, a robuts and real-time visual localization framework that uses building facades as a reliable source of global geo-reference, while retaining the capability to integrate GPS signals when available. In an offline stage, Spotter processes Google Street View panoramas by semantically segmenting facades and pairing multi-view stereo depth with cartographic data to build a compact metric database. At runtime, query images are matched via a cascaded retrieval and geometric verification pipeline to recover fine-grained global camera localization. We benchmark Spotter on a newly collected dataset of pedestrian sequences acquired with wearable smart glasses across several districts of Barcelona. Experimental results show that Spotter outperforms odometry-based baselines and achieves localization accuracy comparable to state-of-the-art map-based methods while operating at significantly higher frame rates.
[CV-24] Spatiotemporally Decoupled Autoregressive Diffusion Model for Human Motion Generation ICME2026
链接: https://arxiv.org/abs/2608.23279
作者: Chengqun Yang,Liang Xu,Yanping Li,Fulong Liu,Jingnan Gao,Weili Zeng,Yichao Yan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by ICME 2026
Abstract:Text-driven human motion synthesis has made substantial development with two core modules of motion representation and generative architecture. For representation, Vector Quantization (VQ)-based methods compress motion data into discrete tokens while latent-based models operate directly in continuous space. However, both of these representations exhibit significant limitations. VQ-based methods suffer from inherent information loss, which compromises the quality, diversity, and generalization of generated motions, while continuous representation on holistic whole-body motion hinders part-level flexibility. For architecture, diffusion and autoregressive diffusion models have demonstrated their superiority, yet the fine-grained controllability over individual body parts is also limited. Thus, we propose a unified spatiotemporally decoupled framework named DeMoDiff, which jointly redesigns representation and architecture. To enhance representation extraction capabilities and offer greater part-level controllability, we present a spatial-temporal VAE that encodes each body joint rather than compressing the whole-body motion into a single latent space. Then, we incorporate spatial-temporal masking and attention mechanisms into an autoregressive diffusion generator, achieving both generative capability and controllable editability. Extensive experiments on the HumanML3D and KIT-ML datasets demonstrate that our model achieves state-of-the-art reconstruction performance and compelling motion generation results. Moreover, our framework demonstrates strong temporal and spatial editing capabilities, further validating its effectiveness. Our project page: this https URL
[CV-25] Dual-Grained Agent Memory and Shapley Context Attribution for Multimodal Agent ic Learner
链接: https://arxiv.org/abs/2608.23268
作者: Jieke Wang,Tiancheng Shen,Yibo Yang,Ming-Hsuan Yang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Frontier multimodal large language models (MLLMs) deliver impressive perception yet still falter on scientific and mathematical reasoning. Parameter-level adaptation is unavailable for closed-weight or on-device backbones, and stateless prompting forfeits any compounding benefit from problems already solved. We propose \textbfDG-Mem, a dual-grained agentic memory framework that augments a frozen MLLM with a non-parametric, externally stored memory built once from training-time rollouts and consulted read-only at test time. Motivated by the Complementary Learning Systems (CLS) account of human memory, DG-Mem factors its store into an instance-grounded exemplar memory and a category-level schema memory of IF-THEN rules, with a transient reflection store mediating their construction so that schemas are synthesized only from abstract reflections, never from exemplar text. Two design choices distinguish DG-Mem: an online concept categorizer that grows the category space incrementally during training rather than committing to a predefined taxonomy, and a Shapley context attribution procedure that decomposes correctness across the entire retrieved rule set and yields a per-rule utility that re-weights retrieval at test time. The pipeline introduces no gradient updates and is deployable on closed-weight or on-device backbones. Across MathVista, MMMU, and MMMU-Pro on four open-weight and proprietary backbones (Qwen3.5-27B, Qwen3.5-122B-A10B, GPT-5-Nano, Gemini-3-Flash), DG-Mem improves consistently over no-memory and competitive memory baselines.
[CV-26] Progressively Learning Heterogeneous Skills in a Unified Latent Space
链接: https://arxiv.org/abs/2608.23258
作者: Yue-Yi Zhang,Ming Gong,Linpu He,Wei-Shi Zheng,Zhilin Zhao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 23 pages, 18 figures
Abstract:We propose HetSkills, a novel framework designed to progressively learn heterogeneous skills within a unified latent space for physics-based character control. The core idea is to treat this latent space as a shared executable interface, enabling seamless integration of skills learned from diverse data sources, supervision forms, and tasks. HetSkills begins by learning a tracking skill that establishes a strong foundation in motion control and creates a shared motion decoder, which can be reused across tasks without the need for retraining or separate controllers. To prevent the text-to-motion skill from exploiting shortcut pathways instead of learning language semantics, we introduce motion intuition distillation to ground text-to-motion generation in language semantics and a task-guidance module that dynamically adjusts actions based on high-level language instructions. This enables HetSkills to preserve natural motion while continuously expanding its skill repertoire, making it highly adaptable for long-horizon tasks. Experimental results demonstrate the effectiveness in motion tracking, text-to-motion generation, motion completion, and downstream task adaptation, achieving impressive success rates even under challenging conditions.
[CV-27] E2S-Pruner: Progressive Two-Stage Evidence Fusion for Visual Token Pruning in Vision-Language Models
链接: https://arxiv.org/abs/2608.23253
作者: Taoyu Qian,Qi Wang,Daqian Shi,Yuanhao Jiang,Shang Gao,Hualong Yu
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Vision-language models typically encode an image into hundreds of visual tokens, incurring substantial inference latency and GPU memory overhead. Existing pruning methods largely rely on attention scores and directly aggregate outputs across attention heads and network layers, making it difficult to characterize evidential uncertainty and conflict. We propose E2S-Pruner, a progressive two-stage evidence-fusion framework for visual token pruning that requires no auxiliary model, trainable parameters, or fine-tuning. In the first stage, E2S-Pruner treats each attention head as an independent evidence source, estimates its reliability from evidence clarity and inter-head consistency, and represents each visual token using three states: important, unimportant, and uncertain. In the second stage, Dempster–Shafer evidence theory is used to quantify inter-layer conflict and fuse complementary evidence from multiple network layers. We further introduce a spatial novelty constraint that promotes coverage of distinct image regions and prevents the retained tokens from concentrating in a few locally salient areas. On LLaVA-1.5-7B, E2S-Pruner retains 98.0%, 96.8%, and 90.6% of the aggregate performance when the average numbers of retained visual tokens are 192, 128, and 64, respectively, while improving throughput by 1.96x and 2.09x under the 128-token and 64-token settings. Experiments on Qwen2-VL-7B further demonstrate cross-model generalization. Code is available at this https URL.
[CV-28] Semantic Reconstruction and 3-D Detection via Learned Multi-Pair Fusion in RF Imaging
链接: https://arxiv.org/abs/2608.23249
作者: Amir Rezaei,Wen-Xin Pan,Giuseppe Caire
类目: Computer Vision and Pattern Recognition (cs.CV); Signal Processing (eess.SP)
备注:
Abstract:We consider a multistatic radio-frequency imaging problem with anisotropy, in which the reflection from a point depends on the positions of the transmit (Tx) and receive (Rx) arrays. The goal is to label the voxels of a field of view by a finite set of semantic classes and to group them into object instances. For the image formation of each Tx–Rx pair we apply a standard inverse-problem solver, and we feed the resulting per-pair reconstructions into a trained three-dimensional (3-D) U-Net that performs the fusion implicitly and the per-voxel classification explicitly. On a controlled, under-determined multistatic setup, we consider the following image formation methods: back-projection (BP) and the least absolute shrinkage and selection operator (LASSO) from a single deterministic snapshot, and incoherent BP and group-LASSO from multiple fading snapshots. For each imaging method we train a separate U-Net that fuses the six Tx–Rx pairs (its input channels) and assigns each voxel a probability vector over the classes. Taking the most probable class gives a labeled volume—the semantic reconstruction. Object instances and their oriented bounding boxes then follow by geometric post-processing (clustering and principal-component analysis). Across a wide range of signal-to-noise ratio, the semantic reconstruction (scored against ground truth by segmentation intersection-over-union) and the resulting 3-D detection degrade far more gracefully than the classical intensity reconstruction: the detection in particular stays reliable well into noise levels at which that reconstruction has dissolved. Because real scenes contain objects of classes the network was not trained on, we add an explicit unknown class trained by outlier exposure, which labels held-out novel objects as unknown instead of mislabeling them as a known class by reconstructed shape.
[CV-29] Mover360: Controllable Object Manipulation in 360° Panoramic Images
链接: https://arxiv.org/abs/2608.23238
作者: Haoyi Zhong,Fang-Lue Zhang,Andrew Chalmers,Taehyun Rhee
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:We present Mover360, a controllable object manipulation framework for 360° images. Unlike perspective images, 360° images in equirectangular projection (ERP) exhibit horizontal wrap-around, latitude-dependent distortion, and global scene continuity, which makes object-level edits difficult for existing perspective editors to produce and for users to specify. To address this, Mover360 centers on object Translation (relocating a specified object within an existing panorama) while supporting reference-guided Insert and Remove as auxiliary tasks. Its interface unifies point-, bbox-, and mask-guided control by encoding each task into a fixed prompt and a compact, ERP-aligned instruction map. In the default point mode, a single click relocates an object, allowing the model to infer a plausible size, support, and illumination using panoramic context and an auxiliary depth condition. Structurally, Mover360 is a lightweight adaptation of a pretrained diffusion transformer. To generate paired supervision, we construct a UE5 data-generation pipeline with surface-aware object placement and randomized illumination, yielding large-scale paired data and a dual-domain benchmark of synthetic and real panoramas with ground truth for all three tasks. Across both test domains and two evaluation protocols, Mover360 outperforms strong baselines for perspective editing, insertion, and inpainting in reconstruction fidelity, semantic consistency, and distributional quality. Code and our benchmark dataset are available at this https URL.
[CV-30] MLLM -Assisted Audio VOS: A 3rd Place Report for the MeViS-Audio Track 8th LSVOS Challenge
链接: https://arxiv.org/abs/2608.23234
作者: Liangtao Shi,Jinxia Xie,Xiantao Hu,Ting Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 5 pages
Abstract:In this technical report, we present a training-free framework for audio-guided video object segmentation, which integrates Multimodal Large Language Models (MLLMs) with SAM-based segmentation models. We decompose the task into several stages and identify suitable foundation models for each stage. Without introducing additional model training or task-specific fine-tuning, our approach leverages the strong multimodal reasoning capabilities of MLLMs to model text-visual correspondence and employs SAM-based models for accurate object mask generation. The proposed framework demonstrates the effectiveness of leveraging foundation models for audio-guided video segmentation and achieves competitive performance in the MeViS-Audio Track of the 8th LSVOS Challenge.
[CV-31] hink Only When Needed: Prompt-Authority Control for Selective Slow-Path Intervention in Vision-Language-Action Manipulation
链接: https://arxiv.org/abs/2608.23224
作者: Zhiruo Zhou,Zelin Li,Xiwen Chen,Jiazhuo Li,Chenwei Wang,Huiming Chen,Xiaojun Zhu
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Retrieval can efficiently and effectively augment a frozen vision–language–action (VLA) policy without retraining, yet retrieved text becomes a control intervention once it enters the executed prompt. In a matched audit, raw appended text reduces mean success from 92.47% to 3.00%, while meaningful and length-matched meaningless appends both fail on all 500 states. This result identifies \emphprompt-form collapse: changing the instruction form, rather than adding useful semantics, can dominate execution. We introduce TOWN-VLA (Think Only When Needed), a prompt-authority interface that separates candidate generation from permission to alter the policy input. A fixed compatibility rule authorizes a canonical compact instruction; otherwise, the interface restores the original Base prompt exactly. Across 900 audited routes, every route follows this contract: 525 routes recover Base with matching hashes, and all 375 authorized prompts preserve the task signature. On a matched 4\times7 LIBERO-Plus evaluation with 10,030 episodes per method, success rises from 69.5% to 73.1% ( +362 episodes; 95% CI 1.89–5.45 points), improving on six perturbation axes and all four suites. On a physical PiPER arm with a frozen \pizerofive checkpoint, success rises from 52.7% to 78.7% over 150 trials per method ( p=3.16\times10^-6 ). Prompt authority is enforceable for a frozen controller; oracle-free admission calibration is the next deployment target.
[CV-32] BenthicDINO: Physics-Informed Self-Distillation for View-Invariant Side-Scan Sonar Representations
链接: https://arxiv.org/abs/2608.23215
作者: Taqi Hamoda,Hayat Rajani,Nuno Gracias
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Automated perception in side-scan sonar (SSS) imagery is severely hindered by physical acoustic artifacts, resulting in representations that inextricably mix intrinsic seabed reflectivity with transient viewing geometries. Existing self-supervised learning (SSL) frameworks rely on augmentations designed for natural images, failing to account for acoustic degradation and explicitly enforce view-invariance. To address this gap, we introduce a physics-informed self-distillation framework built upon the DINOv3 architecture utilizing a ConvNeXt-v2-Tiny backbone to maximize data efficiency. The proposed methodology enforces view-invariance through two primary mechanisms: physically motivated augmentations that simulate speckle noise, range-dependent attenuation, and radiometric miscalibration; and a Hilbert-Schmidt Independence Criterion (HSIC) penalty that explicitly decouples learned dense patch features from physical viewing parameters. Furthermore, we propose a dense, hierarchical feature fusion strategy across all four network stages to preserve fine-grained sediment details alongside deep semantic abstractions. Extensive evaluation demonstrates that the framework natively groups complex benthic topographies into stable, noise-free semantic clusters without relying on manual annotations. During supervised downstream tasks on the S3Seg dataset, the fused representations exhibited exceptional data efficiency, achieving 96% of its absolute peak performance using only 10% of the available annotated data, ultimately reaching a mean Intersection over Union (mIoU) of 71.4% and an overall accuracy of 86.5%.
[CV-33] Bee Detection and Tracking at Hive Entrance using YOLO11 and ByteTrack
链接: https://arxiv.org/abs/2608.23213
作者: Thi Thu Thao Nguyen,Johannes Reschke
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 17 pages, 13 figures
Abstract:This work presents an automatic bee entrance monitoring system based on YOLO11 transfer learning and the ByteTrack tracking algorithm. The study investigates the influence of data augmentation, backbone freezing, and tracker parameter optimization on the detection and counting of small, fast-moving bees. The detector with progressive backbone unfreezing strategy achieved about 97.0% precision and 98.7% mAP50, while providing more stable convergence than full fine-tuning. Experiments also showed that light augmentation outperformed heavy augmentation. For tracking, ByteTrack parameters were optimized to improve trajectory continuity under low-confidence detections. On an independent 25 FPS side-view video, the optimized YOLO11-ByteTrack system correctly counted 43 of 47 incoming bees (91.5%) and 7 of 30 outgoing bees (23.3%). Error analysis showed that most counting errors were caused by missed detections due to rapid bee motion and motion blur, while tracking failures became less frequent after parameter optimization. Overall, the results indicate that moderate augmentation, progressive backbone unfreezing, and ByteTrack tuning improve the reliability of automatic bee entrance monitoring under realistic recording conditions.
[CV-34] Learning Spherical Occupancy Profiles for Multi-View 3D Reconstruction and Generation
链接: https://arxiv.org/abs/2608.23206
作者: YiHsuan Tsai
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 12 pages,5 figures,5 tables. Code and data will be released
Abstract:We study spherical occupancy profiles-the ray-wise occupancy probability profiles P® = T® o® distilled from multi-view 3D Gaussian reconstructions-as a unified intermediate representation for both discriminative and generative 3D reconstruction from images. On a 999-object subset of Google Scanned Objects with 48 turntable views each, we train (i) a discriminative per-ray decoder that injects global view-averaged and ray-specific image evidence into a FiLM-conditioned profile head, reaching median soft depth error 0.035 (normalized) on an independent 90-object test split, and (ii) a generative pipeline built on a profile VAE and a latent diffusion model, which supports unconditional sampling that matches the reconstruction manifold and image-conditioned multi-solution reconstruction whose per-object solution spread is quantifiable and tunable via classifier-free guidance. We further analyze the morphology of predicted profiles: post-hoc power sharpening and a learned sharpening target both recover ground-truth profile width without degrading depth, exposing a monotonic width-peak frontier in the L1-per-ray loss family and motivating a principled redefinition of morphology gates. Real-photo validation on two DTU scenes confirms the pipeline transfers to non-synthetic input. Our results suggest that ray-wise occupancy profiles offer a compact, learned, and uncertainty-aware interface between multi-view reconstruction and generative priors.
[CV-35] oward a Foundation Plug-and-Play Prior for Computed Tomography Reconstruction via a Multimodal Diffusion Model
链接: https://arxiv.org/abs/2608.23190
作者: Haley Duba-Sullivan,Patxi Fernandez-Zelaia,Obaidullah Rahman,Amirkoushyar Ziabari
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Computed tomography (CT) throughput is limited by scan time, which grows with both the number of projections acquired and the detector integration time for each. Reconstructing high-quality volumes from sparse-view or low-dose measurements therefore depends on an informative prior, typically a neural network trained for one specific scan setting and retrained whenever the modality, geometry, or material changes. We investigate whether a single diffusion model trained across several imaging domains can instead serve as a prior for many CT problems simultaneously. We evaluate the proposed method using the same frozen model on three datasets that differ in modality, beam geometry, material, and degradation type, spanning flaw analysis in additively manufactured metal parts imaged with cone-beam X-ray CT and concrete microstructure imaged with parallel-beam neutron CT. Our proposed method out-performs analytic reconstructions in all three cases, providing a step toward a reusable foundation prior for heterogeneous CT reconstruction problems.
[CV-36] EchoWM: Open and Enterable Omnimodal World Models
链接: https://arxiv.org/abs/2608.23189
作者: Songchun Zhang,Yaowei Li,Junhao Zhuang,Weiyang Jin,Haoyu Wang,Xin Lu,Yilang Sun,Shiyi Zhang,Haoran Li,Xiaoxiao Ma,Yuming Li,Yijun Liu,Yaofeng Su,Yanwen Ma,Haoyu Wu,Zihan Su,Yue Ma,Lvmin Zhang,Haoyang Huang,Zeyue Xue,Anyi Rao,Nan Duan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 42 pages, 24 figures
Abstract:We present EchoWM, an omnimodal world model for enterable generative media that responds to continuous navigation while jointly generating 720p video, environmental sound, music and speech. We organize interaction around camera intent: in first-person scenes, it specifies observer motion, while in third-person scenes, camera–character dynamics are learned from data without view-specific controllers. Discrete commands and continuous poses are mapped to a shared metric-scale relative 6-DoF trajectory, with dataset-level calibration preserving motion magnitude across heterogeneous data. To jointly learn audio-visual generation and trajectory control, we construct a complementary data engine and adopt progressive training followed by autoregressive post-training for long-horizon generation. Extensive evaluations show that \model achieves strong trajectory following and high visual quality on public world-model benchmarks, supporting both first- and third-person interaction across varied subjects, and maintaining synchronized environmental sound and speech over long-horizon generation.
[CV-37] A Comparative Study of Label-free Representation Quality Metrics in Deep Learning
链接: https://arxiv.org/abs/2608.23182
作者: Daniel Richards Arputharaj,Daniel Jönsson,Gabriel Eilertsen
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注: Published in Transactions on Machine Learning Research (TMLR). OpenReview: this https URL
Abstract:We present a comparative study of label-free metrics for assessing the quality of representations in deep neural networks to understand their reliability under a wide variety of configurations. We group existing label-free metrics into three families based on their construction and analytically establish connections between metrics within the same family. We then characterise the sensitivity of spectral metrics through controlled synthetic experiments. Finally, all label-free metrics are evaluated against downstream task accuracy across a diverse set of 260 vision models on six datasets spanning generic object classification, fine-grained object classification, scene recognition and geospatial task, stratifying results by architecture class and training objective. We find that intrinsic dimensionality (ID) is the most reliable predictor among the metrics considered. However, the reliability of all metrics, including ID, is moderated by architecture class and training objective. Our results provide a clearer understanding of what label-free representation quality metrics measure, when they are reliable, and how to interpret them in practice.
[CV-38] Neighbor-Aware View Synthesis for Restoring Missing Views in Light-Field Camera Arrays
链接: https://arxiv.org/abs/2608.23175
作者: Sakshi Goel,Ayush Goyal,K S Venkatesh,Koteswar Rao Jerripothula
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:In light-field (LF) imaging systems, dense spatial sampling from a camera array enables powerful post-capture capabilities such as refocusing and depth estimation. However, real-world LF capture is often affected by hardware malfunctions, where one or more cameras in the array fail, leading to missing sub-aperture images and degraded reconstruction quality. This paper addresses the problem of defective or missing view restoration in light-field camera arrays. We propose a novel generative framework that synthesizes the absent views by exploiting information from a carefully selected subset of neighboring cameras. These selected images, along with a positional encoding map indicating both their locations and the desired target view, are fed into a conditional Generative Adversarial Network (cGAN) trained to generate the missing viewpoint in a geometrically consistent manner. Extensive experiments on synthetic and real-world LF datasets demonstrate that our method produces visually plausible and photometrically accurate reconstructions, outperforming baselines for view interpolation both quantitatively and qualitatively. The proposed framework thus offers a robust and efficient solution for fault-tolerant light-field image acquisition.
[CV-39] BenthicFlow: Generating Extensible Underwater Environments via Flow Matching ECCV2026
链接: https://arxiv.org/abs/2608.23173
作者: Joaquín Figueira,Camile Lendering,Manfred Gonzalez-Hernandez,Giacomo D’Amicantonio,Erkut Akdag,Egor Bondarev
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to ECCV 2026 in the 2nd Workshop on Marine Vision
Abstract:Computer vision applications for 3D scene understanding in underwater environments remain challenging due to the lack of high-quality 3D data and the inability of surface-trained models to generalize to underwater scenes. To address this challenge, an emerging trend is to employ generative models to close the data domain gap. However, existing methods assemble large scenes by stitching independently generated tiles post hoc with separately trained models, while demonstrating heterogeneous landscapes only within individual survey sites. We introduce BenthicFlow, a unified framework based on a single conditional flow-matching model that jointly generates aligned textures and depth maps. A MultiDiffusion-inspired sampling procedure reconciles overlapping windows throughout the generative trajectory, enabling spatially extensible RGBD mosaics without a separate stitching model. The generated mosaics are subsequently lifted into explicit 3D benthic environments using surface-aligned Gaussian surfels. Experiments across geographically distinct survey sites demonstrate that BenthicFlow preserves site-specific appearance while generating coherent, large-scale 3D scenes that closely match the target distributions. Code and trained models are available at this https URL.
[CV-40] An end-to-end-trained vision-language model for native-language prostate pathology report generation
链接: https://arxiv.org/abs/2608.23143
作者: Christian Grashei,Fabian Gülhan,Maximilian Legnar,Fabian Stögbauer,Cleo-Aron Weis,Carolin Mogler,Peter Schüffler
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Prostate cancer is among the most frequently diagnosed malignancies worldwide, and structured reporting of each biopsy core burdens pathologists. Existing tools frame this as classification, leaving pathologists to assemble coherent reports, while many slide-level vision-language models rely on English-centric encoders that transfer poorly to other clinical languages. We present a slide-level framework generating prostate biopsy reports that is language-independent by construction: tokenizer and model are trained from scratch, demonstrated here in German. To address paired-data scarcity, an automated pipeline uses a locally deployed large language model to split composite reports into core-specific image-text pairs, yielding 17,344 pairs from 2,402 historical cases without manual annotation. Evaluated for clinical attributes rather than linguistic similarity, the model achieves 96.2% F1 for malignancy detection and 65.2% for Gleason grading, competitive with an FDA-cleared classifier. Grading is further validated on three external cohorts with latent-space augmentation. Institutions can thus train native-language reporting models on their own archives.
[CV-41] How Merge-Tolerant Are Vision Transformers for Wheat Phenotyping? ECCV2026
链接: https://arxiv.org/abs/2608.23142
作者: Simon Ravé,Pejman Rasti,David Rousseau
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to the CVPPA workshop at ECCV 2026
Abstract:Vision-based wheat phenotyping requires repeated measurements under deployment constraints, from growth-stage recognition to wheat-head counting and organ segmentation. Plain Vision Transformers (ViTs) provide a common architecture for these tasks, but quadratic attention limits high-throughput and edge inference. Training-free token merging is attractive because it can be inserted into trained models without retraining. We provide a systematic benchmark of ToMe and Mutual Pair Merging across growth-stage classification, wheat-head detection, and wheat-organ segmentation, measuring task quality, throughput, token count, and peak GPU memory, with additional Raspberry Pi 5 measurements. The benchmark reveals a clear hierarchy: classification is highly merge-tolerant, while detection and segmentation are constrained by repeated instances, thin organs, dense boundaries, reconstruction, and runtime overhead. Optimized attention backends can erase apparent speedups, so deployment value must be profiled on the target runtime rather than inferred from token count.
[CV-42] MIVIFI: Bridging Perspective and Fisheye Domains for Training Multi-View Fisheye Image Generation Models ITSC
链接: https://arxiv.org/abs/2608.23140
作者: Matthias Neuwirth-Trapp,Begüm Altunbas,Jiayi Wang,Yan Xia,Maarten Bieshaar,Xinyu Huang,Daniel Cremers
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: Accepted at the IEEE International Conference on Intelligent Transportation Systems (ITSC) 2026
Abstract:Achieving 360° coverage is critical for the visual perception systems of autonomous vehicles. Fisheye cameras offer a cost-effective solution by enabling full surround coverage with as few as two sensors. However, existing multi-view fisheye datasets are limited, and synthesizing rare corner cases typically requires computationally expensive 3D simulations, hindering the training. While generative models have achieved significant success in standard perspective imagery, their application to wide-angle distortion remains unexplored. In this work, we formally introduce the novel problem of multi-view fisheye image generation conditioned on volumetric semantic representations and present two distinct methods. We first propose SyntheOcc-FE, which adapts the SyntheOcc architecture to fisheye data. While effective, this method is constrained by the scarcity of fisheye datasets, which limits its generalization. To overcome these limitations, we propose our second method, MIVIFI (multi-view fisheye), which leverages cross-domain learning with Equirectangular Projections. By bridging the gap between dataset domains using KITTI-360 fisheye images alongside nuScenes multi-view standard images, our approach enables high-fidelity manipulation of scene content. This framework enables the structural modification of semantic occupancy inputs to introduce or eliminate specific actors and facilitates the rendering of diverse meteorological conditions and illumination scenarios absent in the limited fisheye datasets. Quantitative and qualitative experiments demonstrate that our methods achieve robust photorealistic multi-view fisheye image generation and highlight the specific advantages of our cross-domain strategy for handling data scarcity.
[CV-43] Pointing-VLA: Typed Spatial Grounding Interfaces for Vision-Language-Action Manipulation
链接: https://arxiv.org/abs/2608.23138
作者: Xiwen Chen,Zelin Li,Zhiruo Zhou,Huiming Chen,Chenwei Wang,Xiaojun Zhu
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision-language-action (VLA) models often expose spatial grounding through autoregressive text coordinates or opaque action tokens, creating brittle interfaces between multimodal reasoning and robot execution. We present Pointing-VLA, a typed hidden-state spatial readout built on Embodied-R1. Geometry-specific heads predict normalized points, object-functional grounding (OFG) heatmaps, and visual trajectories without serializing geometry as text. For the evaluated Bridge/WidowX and physical pick-place deployments, an explicit execution contract assigns PICK to source-conditioned OFG and PLACE to Pointing, providing direct stage-aligned spatial targets. Pointing-VLA achieves SOTA performance on Bridge/WidowX, averaging 72.9% across the evaluated four-task set without Bridge-specific finetuning under collision-enabled CuRobo execution. Pointing and OFG show complementary strengths across native and cross-dataset evaluations. The OFG/contact readout transfers to NORA-1.5, preserving or improving success while reducing recorded controller time by more than 20 \times ; typed heads are also 6.68–6.90 \times faster than Embodied-R1 text decoding on a shared external suite. When integrated as spatial guidance for a \pi_0.5 action policy, Pointing-VLA raises autonomous real-robot success from 52.7% to 80.7% across three visual contexts. These results establish typed spatial readouts as an efficient, inspectable interface between embodied reasoning and robot execution.
[CV-44] Bridge Damage Detection from Low-Light UAV Imagery via Degradation-Aware Mixture-of-Experts Enhancement
链接: https://arxiv.org/abs/2608.23136
作者: Hu Wang,Hongxu Pu,Zhiqi Hu,Fangzhou Lin,Wang Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 31 pages, 9 figures
Abstract:Poor illumination obscures small, low-contrast defects in UAV bridge imagery, reducing the reliability and operational flexibility of automated inspection. This paper investigates whether degradation-aware image restoration can improve bridge damage detection under low-light conditions and transfer from synthetic degradations to real inspection scenes. We propose DaL- MoE, a detector-agnostic restoration front end trained with an ISP-aware low-light synthesis pipeline and equipped with degradation-aware guidance estimation and complementary experts for noise suppression, color adjustment, and structural-detail recovery. On paired synthetic data, DaL-MoE achieves 23.12 dB PSNR and 0.8482 SSIM, increasing YOLOv11m box mAP50 from 0.3097 to 0.4923 and mask mAP50 from 0.2281 to 0.3529. On real low-light UAV imagery without paired normal-light references, sim-to-real evaluation shows improved defect visibility and more complete detections than direct inference on raw low-light inputs. Future work will develop low-light-aware bridge damage detectors with stronger cross-scene generalization across bridge sites, imaging conditions, and illumination levels.
[CV-45] Loopy: Seamless Video Loop Generation via Anchored Looping Shift of Positional Embedding
链接: https://arxiv.org/abs/2608.23090
作者: Haotian Dong,Wenjing Wang,Chen Li,Jing Lyu,Xin Wang,Di Lin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 15 pages, 21 figures, accepted by ACM TOG
Abstract:Looping videos are essential for practical applications such as web graphics, game development, and social media. However, existing approaches typically fail to generate high-quality looping videos due to the neglect of how video generation models perceive temporal order and how this relates to the looping behavior. In this work, we are the first to reveal that position embedding at different attention layers within DiT exhibits varying levels of positional control, with the most pronounced layer acting as an anchor. We formulate this anchored layer as the reference point of the looping video, offering strong contextual priors for the remaining layers to facilitate the generation of seamless and coherent video content. Based on this insight, we propose an anchored position embedding shifting strategy that applies layer-specific shift lengths according to each layer’s temporal control effect, effectively transforming DiT’s temporal perception from a straight line to a circle. Leveraging this strategy, we develop a general framework, Loopy, for high-quality looping video generation, supporting both RGB and RGBA videos, while also enabling advanced AIGC features such as identity control and style transfer. Experiments demonstrate that our approach significantly improves temporal consistency and visual fidelity in generated looping videos. The released model is available on our website: this https URL.
[CV-46] Grounding Isnt Knowing: Do VLMs Need Object Localization for Spatial Reasoning ?
链接: https://arxiv.org/abs/2608.23074
作者: Xiwei Liu,Yulong Li,Xinlin Zhuang,Xuhui Li,Zhixiang Lu,Haolin Yang,Imran Razzak,Yutong Xie
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision-language models (VLMs) can answer spatial questions, yet the mechanisms connecting object grounding to spatial reasoning remain poorly understood. It is underexplored whether spatial reasoning internally requires precise objects localization, or can bypass explicit localization through global layout cues. In this work, we investigate two representative model families, LLaVA-1.5 and Qwen2.5-VL, using a suite of mechanistic interpretability tools, including token ablation, layer-wise probing, attention knockout, and causal mediation analysis. We find that spatial relation prediction follows a staged grounding-to-reasoning process in which object-aligned tokens establish coarse target-reference anchors, while precise bounding-box boundaries are not required. Positional information becomes decodable before relation decisions emerge, and a small set of attention heads mediates the causal effects of both localization and spatial reasoning. The two tasks share early grounding-related processing but ultimately rely on partially distinct specialized pathways. Through rigorous experiments, we provide a token-, layer-, and head-level account of how VLMs transform object grounding into spatial relations, showing that knowing where objects are is not equivalent to knowing how they relate.
[CV-47] From Generation to Simulation: How Far Are World Models from Being True Simulators?
链接: https://arxiv.org/abs/2608.23070
作者: Tong Wang,Huan Deng,Mucheng Yang,Yang He,Xiaohui Kuang,Gang Zhao
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: 42 pages, 23 figures, 2 tables. Project page: this https URL
Abstract:With the rapid progress of diffusion models and large-scale video generation, generative world models are increasingly expected to replace traditional simulators, including physics engines, game engines, and reinforcement-learning environments. Yet the remaining distance from generation to simulation lacks a systematic assessment. We present a capability-based study using an external yardstick: eight capabilities of a traditional simulator, namely asset construction, physics engine, interaction, controllability, stability, state feedback, diversity, and evaluation metrics. We trace three main technical routes–latent dynamics, video generation, and joint-embedding prediction–and map exactly 200 representative works published from 2018 to June 2026 onto these capabilities. Our analysis shows that world models have achieved functional substitution in interaction and controllability for specific scenarios, but remain short of traditional simulators in formal guarantees of physical laws, structured state feedback, and reproducible long-horizon evolution. State feedback is the most neglected cross-route shortcoming: only 6 of 163 implementation papers expose a runtime interface for querying entity states or physical parameters. We identify six research directions: formalized physics, a unified action interface, first-class state feedback, long-horizon stability, downstream-utility evaluation, and cross-route hybridization. Project page: this https URL
[CV-48] When the Edit Changes the Patient: Measuring Identity Preservation in Counterfactual Retinal Images
链接: https://arxiv.org/abs/2608.23024
作者: Andrea Posada,Wenke Karbole,Bach Ngoc Doan,Alexander Weers,Solmaz Abdolrahimzadeh,Maria Patsiamanidi,Kahkashan Haider,Vaishali Khare,Daniel Rueckert,Andrew Lotery,Sobha Sivaprasad,Martin J. Menten
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Counterfactual medical image generation aims to modify an existing image to reflect a hypothetical scenario in which certain characteristics of the imaged subject are altered, while keeping their identity fixed. Most existing works repurpose established image editing methods, which do not directly supervise identity preservation. Instead, they assume that identity is implicitly preserved by anchoring generation to the source image. This assumption is rarely tested and may fail in domains where biometric cues are subtle, such as retinal optical coherence tomography (OCT). In this work, we explicitly measure identity preservation for three groups of text-conditioned editing methods - source-anchored, structured-prompt, and paired-training - using referee classifiers, embedding alignment scores, and a blind reader study. We find that all methods produce high-quality OCT images with comparable editing success, yet their identity preservation differs markedly. Source-anchored editing frequently alters the depicted subject, while paired-training preserves it best. We argue that future work on medical counterfactual generation must explicitly measure and report identity preservation alongside image realism and editing success.
[CV-49] AnaDiffusion: Anatomically CompositionalLatent Diffusion for Controllable 3D Brain MRI Generation
链接: https://arxiv.org/abs/2608.23014
作者: Huiwen Han,Lulin Liu,Bangya Liu,Yuanhao Cai,Nuo Chen,Xiaoqing Wang,Ziqian Xie,Chenyu You,Shuiwang Ji,Degui Zhi,Zhiwen Fan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:3D brain MRI generation has made significant advances in medical imaging, simulation, and controllable anatomical analysis. However, existing generative models typically synthesize 3D volumes monolithically, often overlooking regional anatomical structures and limiting local controllability. To address these limitations, we introduce AnaDiffusion, an anatomically compositional latent diffusion framework that factorizes the generation process into distinct, anatomically meaningful regions, followed by part-to-whole assembly and global refinement. Our approach first trains part diffusion models to capture local structural priors. We then inject an assembled anatomical composite of the parts into the whole-brain latent representation and continue denoising. This mechanism enables the model to resolve global context while preserving the injected anatomy. As a result, AnaDiffusion produces both explicit part assets and a globally coherent volume, thereby enabling controllable part editing without requiring subject-specific dense segmentation maps at inference time while maintaining consistent part-to-whole brain structure. On the subject-disjoint ADNI test split, AnaDiffusion achieves the lowest FID across the whole brain, left and right hemispheres, cerebellar-brainstem complex, and seam regions. It also achieves the best cerebellar and second-best ventricular and brainstem absolute Cohen’s d values among the evaluated methods. In localized editing experiments, paired MS-SSIM demonstrates high target transfer and off-target preservation, supporting controllable part replacement with minimal unintended anatomical alterations.
[CV-50] Misanthrope: A Privacy-Preserving Keypoint Detector ECCV
链接: https://arxiv.org/abs/2608.23012
作者: Francesco Vultaggio,Predrag Djindjic,Markus Gerke,Sebastian Tschiatschek,Phillipp Fanta-Jende
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to Privacy preserving Visual Localization (PPVLM) workshop at the European Conference on Computer Vision (ECCV) 2026
Abstract:Image matching is a core component of applications such as Simultaneous Localization and Mapping (SLAM), Visual Localization, and Structure from Motion (SfM). However, the local image features central to this task are vulnerable to inversion attacks, which enable adversaries to reconstruct privacy-sensitive scene content from local features. These attacks pose a particular threat in distributed computing scenarios where the pre-computed features leave edge devices to be processed by remote servers. In this work, we introduce Misanthrope, a novel privacy-preserving keypoint detector trained through self-distillation to avoid detecting keypoints on people—a predominant source of privacy-sensitive content in most localization scenarios—thus mitigating inversion attacks at the source rather than through post-hoc obfuscation. We demonstrate how inverted images from traditional feature detection pipelines can be used to detect and re-identify people in the scene, while Misanthrope is able to mitigate these attacks. Furthermore, Misanthrope maintains image matching performance on par with the state of the art and even surpasses it in challenging settings where people act as distractors, such as phototourism and in-the-wild odometry. On the Image Matching Challenge 2021 Phototourism test set, Misanthrope is the top-performing sparse feature extractor in 7 out of 9 scenes. We make our model and its evaluation script available here: this https URL
[CV-51] Coarse Indexing Fine Evidence: Decoupling Temporal Granularity in Long-Video RAG
链接: https://arxiv.org/abs/2608.23011
作者: Zhe Jin,Zhimin Lin,Bin Zheng,Junhua Fang,Huihua Yang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Graph-based retrieval-augmented generation (RAG) provides a scalable paradigm for long-video understanding, but existing systems typically inherit a fixed temporal granularity from video segmentation when constructing their retrieval index. We argue that this design unnecessarily couples indexing granularity with evidence granularity: coarse representations can often suffice for locating relevant temporal regions, while fine-grained evidence remains important for downstream reasoning. We propose \textbfDensity-Aware Graph Construction (DAGC), a training-free approach that decouples a query-independent coarse retrieval index from the original fine-grained evidence space. DAGC constructs a compact, density-adaptive graph index by merging visually redundant neighboring chunks, while preserving mappings to the original temporal units. Retrieved coarse regions are subsequently expanded back to the original chunk granularity for fine-grained evidence refinement and answer generation. Experiments on MLVU, VideoMME, and LongVideoBench show that DAGC retains only about 40–50% of the original graph nodes and achieves 1.3 – 1.7\times end-to-end wall-clock acceleration while preserving approximately 99% of the original QA performance. The gains transfer across different LVLM backbones and video RAG pipelines, suggesting that long-video RAG need not maintain the same temporal granularity for indexing and evidence reasoning.
[CV-52] ENCORE: Entropy-Guided Cropping and Attention Regularization for Robust Vision–Language Understanding
链接: https://arxiv.org/abs/2608.22996
作者: Yuanhao Sun,Huawei Ji,Jiaxin Ding,Luoyi Fu,Xinbing Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision-Language Models (VLMs) perform well on diverse vision-language tasks, but transformer-based visual encoders split images into fixed-resolution sub-images, compromising object integrity in lightweight VLMs. Existing methods only focus on the visual modality and fail to dynamically preserve the integrity of prompt-relevant regions, limiting performance. In this work, we observe that the early-layer image-text entropy of cross-modal attention strongly correlates with answer grounding quality and task accuracy. Building on this finding, we propose \textbfENCORE, an entropy-guided framework with two components: At inference, an \textbfEntropy-based Cropping Strategy (ECS) evaluates a small set of candidate crops and selects the one with minimal entropy, preserving contiguous regions relevant to the prompt. At training, \textbfEntropy Regularization Training (ERT) augments next-token prediction with an entropy term that sharpens attention on key visual tokens while down-weighting irrelevant ones. Experiments on ten VQA benchmarks show that ENCORE, fine-tuning only 0.14% of parameters, achieves an average 1.43% accuracy gain and state-of-the-art performance among recent 2B-parameter VLMs. Our code is released in this https URL.
[CV-53] Optimize Surgical Triplet Recognition: A Knowledge-Driven Mixture-of-Experts Solution
链接: https://arxiv.org/abs/2608.22972
作者: Yiyi Zhang,Yuchen Yuan,Ying Zheng,Jialun Pei,Jinpeng Li,Zheng Li,Pheng-Ann Heng
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted in TMI
Abstract:Surgical action triplet recognition constitutes a critical task in context-aware robot-assisted surgery, facilitating automatic surgical action perception by identifying instrument, verb, target, and their association. However, existing works struggle to analyze such complex surgical scenes due to three main issues: (1) component-level optimization conflicts caused by entangled feature spaces, (2) category-level optimization conflicts arising from severe data imbalance, and (3) lack of domain knowledge guidance that limits model interpretability and robustness. To address these challenges, we propose a Mixture-of-Experts-guided Co-Optimization (\textitMoeCo) framework powered by knowledge-driven learning. Within the co-optimization pipeline, to first mitigate component-level conflicts, we introduce a component-tailored adapter that disentangles task-specific features across spatial-temporal regimes, facilitating effective component specialization. Next, we develop a coordinated gradient learning strategy to handle category-level conflicts, which adaptively rebalances positive-negative gradients to enhance the perception of rare categories. Notably, inspired by surgical domain expertise, we introduce a knowledge-driven mixture-of-experts mechanism that dynamically integrates multimodal large language model-guided knowledge via activated experts, thereby enriching the co-optimization pipeline with more expressive and robust representations. Extensive experiments on the public CholecT45 and CholecT50 datasets confirm the effectiveness of the proposed co-optimization pipeline and the superiority of dynamic priors integration via the knowledge-driven mixture-of-experts mechanism.
[CV-54] ParallelWorld: Test-Time Scaling for Embodied Reasoning
链接: https://arxiv.org/abs/2608.22971
作者: Min Chen,Shengjun Zhang,Yuxin Li,Zhang Zhang,Xin Fei,Chong Xia,Yueqi Duan
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: Project Page: this https URL
Abstract:Embodied Reasoning constitutes a fundamental capability of embodied intelligence, serving as the basis for autonomous perception, reasoning, and interaction within physical environments. Recent studies have shifted the paradigm of embodied reasoning from static perception toward dynamic exploration, where agents acquire task-relevant information through interactions with the environment. However, existing active reasoning approaches generally generate exploration trajectories incrementally without long-horizon planning. Even recently emerged test-time scaling frameworks often resort to myopic, single-step lookaheads, which struggle to resolve the delayed feedback inherent in complex, occluded spatial environments. To address this limitation, we propose ParallelWorld, a multi-horizon test-time scaling framework for embodied reasoning. Instead of greedy, single-step trials, ParallelWorld empowers agents to simulate and evaluate multi-step future trajectories in parallel before committing to an action. Specifically, we introduce a verifier-guided tree-search paradigm. Starting from the current state, ParallelWorld branches into multiple parallel trajectories and rolls them out continuously across a multi-step horizon. At each simulation step, a verifier agent evaluates the intermediate state transitions, dynamically pruning unpromising branches and prioritizing paths with the highest information gain. Once the multi-step prospective simulation is complete, the agent synthesizes the long-horizon outcomes to commit to the optimal action sequence. Finally, an answer agent performs reasoning over the selected trajectory to produce the final reasoning. Extensive experiments on ESI-Bench demonstrate that ParallelWorld consistently improves active perception and reasoning performance.
[CV-55] Simplified Cross-Modal Calibration for Heterogeneous Event-RGB Stereo Systems BMVC
链接: https://arxiv.org/abs/2608.22965
作者: Nico Hessenthaler,Adam T. Müller,Nicolaj C. Stache
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to the 37th British Machine Vision Conference (BMVC) 2026
Abstract:Accurate extrinsic calibration between event-based and frame-based cameras remains a practical bottleneck for heterogeneous stereo systems. Existing approaches often require sensor or target motion, precise synchronization, or computationally expensive event-to-image reconstruction. We propose a simple, motion-free cross-modal calibration framework that uses a temporally modulated, blended ChArUco target presented on standard consumer displays. By alternating between the original pattern and a partially blended version, the target reliably triggers events while remaining continuously observable to a frame-based camera, avoiding blank frames and reducing synchronization constraints to a coarse, trigger-based alignment. We discretize events into frames coarsely aligned with the RGB images, apply lightweight denoising, and perform ChArUco-based intrinsic and stereo extrinsic calibration. Extensive experiments assess robustness to blending opacity, display brightness, external illumination, viewing angle, and handheld acquisition. Compared to the strongest motion-based reference (E2Calib + Kalibr) and a non-motion-based reference (Plasberg et al.), our approach reduces the mean reprojection error by 44% and 6% , respectively, while substantially simplifying the calibration procedure. Finally, we demonstrate practical utility in a robotic eye-to-hand calibration case study, showing consistent transformations and stable downstream geometric measurements even under partial occlusions. Code is publicly available at this https URL.
[CV-56] WildHandBench: A Benchmark for Handwritten Text Understanding that Challenges MLLM s and Humans
链接: https://arxiv.org/abs/2608.22959
作者: Jun Zhang,Qiao Zhao,Cheng Cui,Jianying Qu,Zhongkai Sun,Jianwen Yang,Changda Zhou,ZhuoXin Liu,Shubin Han
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:While the top model on OmniDocBench now reaches 96.34% overall on printed-document parsing, the ability of current models to handle challenging handwritten documents remains largely uncharacterized. Existing benchmarks focus on isolated text or formulas, overlook handwritten tables and real-world degradation, and report aggregate accuracy without explaining why models fail. We present WildHandBench, a benchmark containing 500 handwritten documents across three structures (free text, tables, formulas), four languages, and nine real-world scenarios. We introduce a Prior-Driven Error (PDE) metric that quantifies whether errors originate from language priors rather than visual evidence. Evaluating 18 state-of-the-art models together with calibrated human baselines, we find: (1) the best model achieves only 71.85% overall; (2) humans outperform all models yet the gap is narrow (77.09% vs. 71.85%); and (3) model errors are qualitatively different from human errors – 63-91% of model errors are prior-driven versus only 49% for humans, exposing systematic reliance on language priors that conventional accuracy metrics cannot capture. Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) Cite as: arXiv:2608.22959 [cs.CV] (or arXiv:2608.22959v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2608.22959 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-57] WADE: A Reasoning -Annotated Benchmark for Multi-Instance Floating-Waste Grounding with Compact Vision-Language Models
链接: https://arxiv.org/abs/2608.22950
作者: Md. Asaduzzaman Shuvo,Ahsan Farabi,Md. Abdul Ahad Minhaz,Mahedi Hasan,Israt Khandaker,Ibrahim Khalil Shanto,Muhammad Nomani Kabir
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Floating waste in inland waterways threatens aquatic ecosystems and requires timely monitoring under cluttered, multi-object conditions. Existing aquatic-waste datasets provide limited geographic coverage, sparse multi-instance annotations, and little supervision beyond boxes and labels. Compact vision-language models (VLMs) therefore remain insufficiently evaluated for jointly localizing, classifying, counting, and explaining floating waste. We introduce WADE, a reasoning-annotated benchmark containing 2,167 images from rural Bangladesh, 13,608 bounding boxes, and ten waste categories. Each annotation is associated with class-level recognition rules covering visual cues, likely confusions, and discriminative features. We evaluate six VLMs under zero-shot, two-shot, reasoning-guided, and fine-tuned settings using detection, counting, and hallucination metrics. For resource-efficient adaptation, we jointly fine-tune Qwen3-VL-2B on boxes, labels, and reasoning chains using QLoRA. Fine-tuning increases recall from 0.0248 to 0.2339 and F1 from 0.0257 to 0.2163, while reducing image-level hallucination from 0.6836 to 0.0883. However, over three-quarters of instances remain undetected, establishing WADE as a challenging benchmark for dense floating-waste grounding with compact VLMs.
[CV-58] Quality Inspection of Printed Circuit Board Pin Insertion via Semantic Segmentation and Board-Level Feature Extraction
链接: https://arxiv.org/abs/2608.22937
作者: Nils Rabeneck,André Kiunke,Nicole Hoess,Wolfgang Mauerer
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Quality control during printed circuit board (PCB) assembly is a critical step in ensuring reliable electronic products. Detecting misaligned pins during or after pin insertion remains a particularly challenging inspection task. This paper presents an automated defect detection method for identifying incorrectly inserted pins on PCBs. The proposed pipeline combines semantic segmentation using a U-Net architecture with contour-based feature extraction and logistic regression for board-level pass/fail classification. Segmentation masks are used to derive contour representations of individual pins, from which board-level features -such as average contour size- are extracted and used to train a logistic regression classifier. We evaluate the method on two datasets: an industrial collection of real-world PCB images, and a publicly available PCB pin-inspection dataset with substantially different visual characteristics. To assess the effectiveness of the proposed approach, a comparison against PatchCore, an anomaly detection technique new to be applied to pin inspection, as well as instance segmentation-based pin detection is made. The developed method achieved Area Under the Receiver Operating Characteristic Curve (ROC-AUC) values of 0.990 on a random test set split from the industrial data and 1.000 on the public dataset indicating strong separation between pass and fail boards. The results indicate that the proposed approach is a promising candidate for automated pin inspection in industrial environments and achieves strong performance on datasets with substantially different visual characteristics after dataset-specific training.
[CV-59] Motion-Based Tokenization for Cross-Dataset Egocentric Gaze Modeling
链接: https://arxiv.org/abs/2608.22926
作者: Virmarie Maquiling,Zhuojiang Cai,Enkelejda Kasneci
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages, 1 figure, 1 table
Abstract:Gaze is increasingly used as an input signal for vision and multimodal models, yet no consensus exists on how to represent it across datasets. Raw traces preserve detail but are noisy and device-dependent, while coarse event labels are easy to model but can discard local motion structure. We formulate event-aligned, fixed-horizon angular displacement as an interpretable, event-conditioned motion vocabulary and compare it with event-only, spatial, absolute-angle, learned vector-quantized, and continuous representations. To assess transfer alongside target predictability and token collapse, our evaluation combines next-token prediction with target-domain regret, low-order target references, paired bootstrap, order sensitivity, motif overlap, and frozen structural probes. In an event-aligned headset benchmark, angular-motion tokens have lower target-domain regret than frozen-codebook VQ tokens in one transfer direction, while the reverse direction is inconclusive. The probes reveal complementary representation properties, and event-only tokens show that low perplexity can retain little motion information. On a third egocentric dataset, a matched comparison of I-VT, native, and frame-span interfaces shows that event construction materially changes transfer: native events have the lowest regret into EGTEA, while frame-span events have zero motif overlap and fail severely as a source. Motion-based tokenization therefore provides a compact representation for event-aligned egocentric gaze streams, while the evaluation identifies how target predictability and event construction shape cross-dataset conclusions.
[CV-60] Results of the 1st Asynchronous CASTLE Challenge at the Joint Egocentric Vision Workshop in Conjunction with CVPR 2026
链接: https://arxiv.org/abs/2608.22914
作者: Luca Rossetto,Werner Bailer,Cathal Gurrin,Graham Healy,Omar Shahbaz Khan,Stevan Rudinac,Klaus Schöffmann,Allie Tran
类目: Computer Vision and Pattern Recognition (cs.CV); Multimedia (cs.MM)
备注:
Abstract:This report summarizes the contributions and results of the 1st Asynchronous CASTLE Challenge at the Joint Egocentric Vision Workshop in conjunction with CVPR 2026.
[CV-61] AquaFlow: A Monocular Gaussian Splatting SLAM for Underwater Streaming Reconstruction
链接: https://arxiv.org/abs/2608.22906
作者: Yingxiang Xu,Kerui Ren,Wenqi Guo,Changjian Jiang,Tao Lu,Linning Xu,Mulin Yu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Preprint
Abstract:Recent monocular 3D Gaussian Splatting (3DGS) streaming reconstruction methods have achieved impressive performance by balancing reconstruction quality and efficiency. However, extending these frameworks to underwater scenes remains challenging due to severe visual degradation, such as light attenuation and scattering, which degrades camera pose tracking and distorts scene geometry. To address these challenges, we propose AquaFlow, a monocular Gaussian Splatting streaming reconstruction framework for efficient and high-fidelity underwater reconstruction. Specifically, AquaFlow fine-tunes a 3D vision foundation model on large-scale underwater data for robust pose and pointmap estimation, and introduces a medium-guided incremental Gaussian initialization strategy for streaming mapping. Furthermore, we develop a streaming-compatible hybrid scene representation that integrates structured, distance-conditioned neural Gaussians with a physics-inspired optical model to compensate for underwater image formation effects, enabling accurate scene reconstruction. We evaluate AquaFlow on a comprehensive dataset of 62 diverse underwater trajectories, collected from both public benchmarks and in-the-wild web videos across various scales. Extensive experiments demonstrate that AquaFlow achieves state-of-the-art tracking and rendering performance, reducing average localization error by 13.2% and improving PSNR by 4.74 dB compared to WaterSplat-SLAM.
[CV-62] NemoSplat: Feed-Forward 4D Gaussian Splatting for Media-Aware Underwater Reconstruction
链接: https://arxiv.org/abs/2608.22888
作者: Xiaopeng Guo,Wai Chung Tse,Yipeng Zhu,Hanwen Zhang,Huajian Huang,Sai-Kit Yeung
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages
Abstract:Reconstructing photorealistic scenes in unconstrained underwater environments remains challenging due to severe media-induced light scattering and unpredictable dynamic objects. Recent feed-forward visual foundation models have demonstrated remarkable capabilities in generalized novel view synthesis and tracking. However, when directly applied to aquatic videos, optical attenuation and motion interference fatally corrupt their feature aggregation, leading to severe tracking and reconstruction failures. To overcome these limitations, we present NemoSplat, the first feed-forward 4D Gaussian Splatting framework tailored for media-aware dynamic reconstruction directly from uncalibrated marine videos. Beyond providing robust estimations of camera poses and dense scene depth, we devise a Promptable Dynamic Disentangler that utilizes a confidence-aware fusion strategy of learned dynamic probabilities and optional semantic text priors, effectively isolating massive transient entities. Furthermore, to counteract visual degradation, a Media-Aware Gaussian Predictor is formulated to jointly estimate intrinsic 3D Gaussian attributes alongside physical media parameters, rendering pristine scene appearance in a single forward pass. Additionally, we introduce a large-scale underwater dataset with massive dynamic elements to facilitate training and evaluation. Extensive experiments on our dataset demonstrate that NemoSplat achieves state-of-the-art tracking accuracy and high-fidelity rendering.
[CV-63] DRAg ent: Discriminative Reasoning Agent for Referring Expression Segmentation
链接: https://arxiv.org/abs/2608.22885
作者: Yujie Qi,Luyan Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 5 pages, 3 figures, 4 tables
Abstract:Referring Expression Segmentation (RES) aims to generate a pixel-level mask for the object specified by a language expression. Recent methods based on multimodal large language models (MLLMs) often rely on one-pass coordinate prediction for visual localization, which serializes continuous spatial locations as discrete text tokens and may lead to localization bias and alignment errors. To address these issues, we propose DRAgent, an MLLM-driven discriminative reasoning (DR) framework for RES. Instead of requiring the MLLM to generate localization coordinates, DRAgent first constructs a detector-generated candidate space and then uses the MLLM as a visual-semantic target discriminator. Specifically, the MLLM performs reliable target selection among potential distractors through a two-stage DR mechanism, which first screens high-recall candidates and then performs instance-wise verification. The selected target box is subsequently used as a spatial prompt for a foundation segmentation model to produce the final pixel-level mask. Furthermore, we construct a self-consistency-filtered reasoning-chain data pipeline for LoRA-based fine-tuning, providing more reliable supervision for enhancing the MLLM’s discriminative reasoning capability. Experiments demonstrate that DRAgent achieves competitive performance on RefCOCO, RefCOCO+, and RefCOCOg.
[CV-64] FOVEA: Focused On-Demand Visual Evidence Adaptation for Cache-Friendly Multimodal Speculative Decoding
链接: https://arxiv.org/abs/2608.22883
作者: Hengjie Zhu,Dayan Wu,Zihao Zhang,Xinze Liu,Jingxuan Yu,Peng Fu,Zheng Lin,Weiping Wang,Ding Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multimodal speculative decoding accelerates vision-language models by allowing a lightweight draft model to propose candidate tokens for parallel verification by a larger target model. Existing methods typically condition the drafter on a fixed visual interface, such as a predefined visual-token budget or a static compressed representation. However, our controlled visual-budget analysis shows that visual demand varies substantially across tasks and decoding stages, which means more visual input is not always beneficial. Actually, insufficient evidence may weaken visual grounding, while excessive context adds overhead and may disrupt drafting. We propose FOVEA (Focused On-demand Visual Evidence Adaptation), a cache-friendly approach that builds a reusable visual memory and dynamically retrieves a bounded subset for a draft state. A cumulative-mass rule determines both how many and which entries are selected. The selected entries are aggregated into a visual readout and fused with the current draft hidden state through a lightweight gated residual correction. Rather than inserting visual tokens into the autoregressive context, the correction modifies only the representation passed to the language-model head. Experiments across multiple vision-language backbones and multimodal benchmarks show that FOVEA improves draft acceptance and end-to-end decoding speed, achieving up to 2.13\times speedup over autoregressive decoding. These results demonstrate that state-conditioned evidence retrieval is an effective alternative to reusing a fixed visual representation throughout multimodal generation.
[CV-65] Large-Small Model Collaboration for Zero-Shot Surgical Phase Recognition MICCAI2026
链接: https://arxiv.org/abs/2608.22879
作者: Yiyi Zhang,Ying Zheng,Wenxin Fan,Yu Zhu,Yuchen Yuan,Litao Zhao,Zheng Li,Pheng-Ann Heng
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: MICCAI 2026 Early Accept
Abstract:Task-specific lightweight models for surgical phase recognition excel at capturing temporal dynamics but generalize poorly under domain shift. Conversely, surgical foundation models (FMs) offer superior transferability via large-scale pretraining, yet their lack of explicit temporal modeling often yields temporally inconsistent predictions, leading to degraded performance. To exploit the complementary strengths of both paradigms, we propose \textbfLarge-\textbfSmall \textbfTemporal adaptation (\textbfLaST), a novel large-small collaborative framework that enables zero-shot adaptation to unseen clinical domains. In LaST, the FM initiates the pipeline by generating frame-level phase priors that serve as initial weak supervision. To effectively utilize these noisy phase priors, we introduce an iterative temporal refinement scheme that integrates dynamic quality control to filter reliable predictions and dual-model cross-learning to mitigate confirmation bias. Simultaneously, the lightweight model leverages its intrinsic temporal modeling ability to progressively correct inconsistent predictions and enhance overall accuracy across iterations. At the end, a cycle replay strategy is employed to close the loop: the refined, more accurate predictions are utilized as upgraded supervision signals for the subsequent iterations, fostering a self-reinforcing evolution of both label quality and model capability. Extensive experiments demonstrate that LaST achieves robust adaptation to unseen domains for zero-shot surgical phase recognition, outperforming the baseline (PeskaVLP) by 24.85%-43.17% in accuracy and even surpassing fully supervised linear probing and several state-of-the-art few-shot approaches. Codes will be released at this https URL.
[CV-66] Stochastic Separability of Embedding Manifolds
链接: https://arxiv.org/abs/2608.22874
作者: Liqing Zhang
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Neurobiological studies and representation learning have observed that representations of objects belonging to the same category in high-dimensional neural spaces exhibit low-dimensional object manifold characteristics, and different object manifolds are linearly separable in these neural spaces. However, these experimentally observed phenomena lack rigorous theoretical validation to date. This paper proposes a new stochastic separability theorem for embedding manifolds of two different object categories. First, we establish a projection measure concentration theorem for embedding manifolds under general conditions. We develop a new two-layer measure concentration analysis technique, which unifies two estimation bounds via the law of total expectation to derive measure concentration inequalities. Based on the measure concentration theorem, we further prove a stochastic separability theorem for embedding manifolds of two different object categories. If two datasets have distinct means and bounded total variances, their samples become linearly separable with high probability, provided that the projection direction satisfies a non-singularity condition. The main contributions of this paper are twofold: 1. We prove the projection concentration properties of embedding manifolds in high-dimensional spaces by using two-lawyer tail-bound inequalities. 2. We identify a non-singularity condition for the stochastic separability between embedding manifolds, and rigorously prove the stochastic projection separability theorem. The theorem not only uncovers geometric and statistical properties of the object embedding manifolds, but also provides a novel mechanism for representation learning in deep networks. Subjects: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2608.22874 [cs.LG] (or arXiv:2608.22874v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2608.22874 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Liqing Zhang [view email] [v1] Mon, 24 Aug 2026 07:01:00 UTC (56 KB) Full-text links: Access Paper: View a PDF of the paper titled Stochastic Separability of Embedding Manifolds, by Liqing ZhangView PDFHTML (experimental)TeX Source view license Current browse context: cs.LG prev | next new | recent | 2026-08 Change to browse by: cs cs.CV References Citations NASA ADSGoogle Scholar Semantic Scholar export BibTeX citation Loading… BibTeX formatted citation loading… Data provided by: Bookmark checked="checked"class=“labs-tab-input”> Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) IArxiv recommender toggle IArxiv Recommender (What is IArxiv?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv’s community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) mathjaxToggle(); We gratefully acknowledge support from our major funders, member institutions, , and all contributors. About Help Contact Subscribe Copyright Privacy Accessibility Operational Status (opens in new tab) Major funding support from
[CV-67] UniMem: Unifying Multimodal Memory and Control for Vision-Language-Action Models
链接: https://arxiv.org/abs/2608.22869
作者: Lars Osterberg,Maggie Wang,Mac Schwager
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:While Vision-Language-Action (VLA) models have leveraged internet-scale pretraining and task-focused finetuning to achieve strong performance on long-horizon tasks, they often struggle with non-Markovian tasks that require memory. Existing approaches to memory typically involve additional Vision-Language-Models (VLMs) for long-term memory management, introducing a memory bottleneck and a fractured training pipeline. Conditioning on multiple historical frames can provide the VLA with access to more descriptive features of past scenes, but can degrade performance if frames are chosen at arbitrary, fixed intervals. To address these limitations, we present UniMem, a framework that unifies high-level, multimodal memory and low-level control under one backbone. UniMem employs an event classifier for memory updates, a keyframe encoder for dense spatial memory, and a keyframe caching technique to minimize overhead during policy rollouts. We evaluate UniMem across five simulation and four hardware tasks targeting sequential and spatial memory, demonstrating that our unified, single-model system outperforms fixed-interval image sampling baselines (93.4% vs. 68.2%) in simulation and hierarchical baselines (80.0% vs. 43.5%) in hardware, while offering faster inference and a simple training pipeline for easy adoption. Project website: this https URL
[CV-68] oward Sub-1 kB Identity-Preserving Face Compression: A Benchmark of Codecs a Custom Learned Codec and Studies of Resolution Demographic Fairness Recompression and Adversarial Robustness
链接: https://arxiv.org/abs/2608.22866
作者: Petr Hurtik,Jakub Sochor
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Storing face images under a hard sub-kilobyte budget, as required for identity documents, smart-card biometrics and bandwidth-constrained verification, forces a codec to discard most of the signal while keeping what a face matcher actually reads: identity. Generic codecs optimize pixel fidelity, not the embedding distances that drive verification, so which codec, resolution and setting best preserve identity at 1024 bytes or less, and how that degrades at 512, is unclear. We benchmark ten general and face-specific codecs across resolutions, byte budgets, two datasets (controlled Color FERET, in-the-wild AI-Solutions-KK) and four anchor face matchers, with a fourteen-model ViT and CNN roster confirming the ranking is backbone-invariant. We then train a custom identity-preserving codec that hits the byte budget exactly via binary search over a frozen gain table, and run four studies: resolution, demographic fairness, recompression, and no-box adversarial robustness. Sub-kilobyte identity preservation is feasible, but which codec to deploy depends entirely on the budget. At 1024 bytes and the 112 px working resolution the problem is close to solved: modern codecs hold Color FERET equal-error rate under 0.35 percent on the ArcFace anchor. At 512 bytes the field re-sorts: AVIF, HEIF, JPEG XL and legacy JPEG collapse to 28 to 98 percent false-non-match rate at FMR 1e-4, while WebP, JPEG-AI and our byte-budgeted learned codecs stay out of that band, with 24.3 percent for WebP against 6.9 percent for our accurate variant in the wild. That re-sort, not the 1024-byte ranking, is the operational result: a codec chosen at 1 kB is not the codec to deploy at half that. Subjects: Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2608.22866 [cs.CV] (or arXiv:2608.22866v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2608.22866 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-69] Following Motion for Sequential Modeling in Video Frame Interpolation ECCV2026
链接: https://arxiv.org/abs/2608.22861
作者: Jaehyun Park,Nam Ik Cho
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to ECCV 2026
Abstract:State Space Models (SSMs) have surfaced as a promising architecture in Video Frame Interpolation (VFI), as they can capture long-range dependencies with linear computational complexity. However, their predefined scanning order limits their effectiveness in modeling the dynamic motion trajectories inherent in VFI problems. To tackle this challenge, we propose Motion-Guided Mamba for Video Frame Interpolation (MGMVFI), an adaptation of the selective state space model tailored explicitly for VFI. MGMVFI introduces Motion-Guided Serialization (MGS), which leverages optical flow to define a motion-adaptive 1D input order for the SSM. This aligns the causal state updates with semantically related tokens, enabling motion-consistent feature propagation, particularly for large and dynamic motions. Additionally, to mitigate the unreliable feature representations caused by inaccurate optical flow estimates, we introduce contextual synthesis that utilizes the surrounding spatial context for robust inter-frame feature synthesis. These components are seamlessly integrated within our tailored Mamba architecture, which also employs a lightweight refinement block to enhance local detail reconstruction at a reduced computational cost. Extensive experiments on standard VFI benchmarks demonstrate that MGMVFI achievesstate-of-the-artperformance,particularly on complex and dynamic motions, thereby establishing a new direction for sequence modeling in video interpolation.
[CV-70] Mapping the Concept Landscape: Structural Perception of Global Distributions for Transparent Data Pruning ECCV2026
链接: https://arxiv.org/abs/2608.22858
作者: Dongyue Wu,Tao Ma
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注: ECCV 2026
Abstract:Existing data pruning methods predominantly rely on high-dimensional feature embeddings to measure sample importance. However, these compressed vectors often obscure fine-grained semantic interactions, leading to suboptimal coverage of rare semantic concepts in the pruned subsets. In this paper, we propose Mapping the Concept Landscape (MCL), a novel structural perception framework for transparent data pruning. Instead of abstract embeddings, we represent each image-caption pair as an explicit sample-level graph comprising entities, events, and attributes. By integrating these individual graphs into a comprehensive dataset-level graph, we characterize the global distribution of semantic concepts and quantify their rarity across the entire corpus. Based on this structured perception, we develop a greedy concept-coverage maximization algorithm that iteratively selects samples to maximize the marginal gain of high-value, under-represented concepts. Experimental results on various benchmarks demonstrate that our method not only achieves superior pruning efficiency compared to state-of-the-art methods but also provides a transparent and interpretable audit trail for the selection process.
[CV-71] VeCAS: Vessel-Focused Contrast-Free Angiogram Synthesis for Vascular Interventions
链接: https://arxiv.org/abs/2608.22828
作者: De-Xing Huang,Chen-Yu Wang,Hao Liang,Xiao-Hu Zhou,Mei-Jiang Gui,Tian-Yu Xiang,Qin-Yi Zhang,Chen Wang,Xiao-Liang Xie,Shi-Qi Liu,Ming-Yuan Liu,Zhen-Chang Wang,Zeng-Guang Hou
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages, 8 figures, 5 tabels, supplementary material: this https URL
Abstract:X-ray angiography relies on iodinated contrast agents to visualize vascular structures during image-guided interventions. However, contrast administration carries risks of adverse events, motivating the development of contrast-free alternatives. Generating X-ray angiograms directly from non-contrast X-ray images offers a potential solution, but existing approaches remain limited by (i) insufficient control over vascular localization and (ii) inefficient modeling of redundant background content. To address these challenges, we propose VeCAS, a two-stage vessel-focused contrast-free angiogram synthesis framework that separates vascular structure localization from angiographic appearance synthesis. In Stage I, a discriminative model localizes vascular structures in non-contrast X-ray images, while cross-modality latent distillation transfers vessel-sensitive knowledge from X-ray angiograms during training. In Stage II, a vessel-focused inpainting model synthesizes angiographic appearance within the localized vascular regions while preserving the non-vascular background. Experiments on an in-house lower-limb vascular intervention dataset show that VeCAS outperforms the comparison methods in terms of vascular structural fidelity and image quality. Visual Turing tests and physician assessments indicate the perceptual realism of the synthesized angiograms. In addition, robotic guidewire navigation experiments in vascular phantoms show that VeCAS guidance reduces the time to target by 41.4% and the number of operation steps by 40.7% compared with non-contrast guidance. Together, these results suggest the potential of VeCAS to serve as ``meta contrast agent’’ for vascular interventions.
[CV-72] SiZeUp: Fast 3D Proxy from Aerial Images via Depth Ordinal Loss SIGGRAPH
链接: https://arxiv.org/abs/2608.22821
作者: Wenjun Zhou,Yunshan Li,Qiaoyu Zhu,Weidan Xiong,Hao Zhang,Daniel Cohen-Or,Hui Huang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to SIGGRAPH Asia 2026 Conference Papers
Abstract:We present SiZeUp, a fast and scalable approach for constructing large-scale 3D urban proxy models directly from calibrated oblique aerial imagery. Our method adopts a height-from-footprint representation, reducing 3D building abstraction to a low-dimensional optimization problem in which building footprints are extruded by a single height parameter. To enable efficient and robust height estimation, we introduce an ordinal depth consistency loss that enforces agreement between the relative depth ordering of rendered proxies and depth priors predicted by a monocular depth model. This is realized through a differentiable renderer that maps parametric building proxies into multi-view depth images, allowing gradients to be propagated from depth supervision to building heights. Our ordinal formulation produces stable optimization in practice and avoids explicit feature matching or dense point cloud reconstruction. Rather than relying on metric depth, which can be unreliable under monocular scale ambiguity, our ordinal depth consistency loss operates on relative depths, providing a more reliable signal across views. Combined with an efficient dynamic view selection, our approach achieves a 23-52 \times speedup over state-of-the-art proxy reconstruction pipelines while maintaining comparable proxy-level coverage and volume consistency, making it well suited for large-scale urban modeling tasks.
[CV-73] Direct Parallel or Sequential? A Comparative Study of Training-Free Multi-Subject Image-to-Video Generation
链接: https://arxiv.org/abs/2608.22819
作者: Yanliang Qi,Kexi Chen,Muchao Ye,Haomiao Ni
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: ACM Multimedia Workshop 2026
Abstract:Text-conditioned image-to-video (I2V) generation has advanced rapidly, yet generating videos with multiple subjects remains challenging. A model must simultaneously preserve the appearance of each subject, assign distinct motions, and maintain coherent spatial and temporal interactions. This paper presents a systematic study of three representative paradigms for training-free multi-subject I2V generation: direct, parallel, and sequential generation. Direct generation applies a pretrained I2V model to the complete reference image and prompt, requiring all subjects and motions to be synthesized jointly. Parallel and sequential generation instead decompose the reference image and prompt into subject-specific visual and textual conditions. Parallel generation synthesizes each subject independently and subsequently composes the resulting videos, reducing the complexity of each generation step at the cost of weaker inter-subject context. Sequential generation first synthesizes a background video and then progressively introduces individual subjects. This preserves accumulated scene context but introduces sensitivity to subject ordering and error propagation. We empirically evaluate the three paradigms across diverse multi-subject scenes, comparing appearance preservation, motion fidelity, temporal consistency, and inter-subject coherence, while also characterizing their distinct failure modes. Our findings reveal the strengths and limitations of each paradigm and offer practical insights for designing controllable multi-subject video generation systems.
[CV-74] Reproducible Vision-Guided 6-DoF Robotic Manipulator with a Mixed Stepper-Driver Architecture and Browser-Native Control
链接: https://arxiv.org/abs/2608.22799
作者: Lasan Perera,Deneth Priyadarshana,Dulana Pitiwaduge,Isitha Dinujaya,Mokshan Colambage
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV); Systems and Control (eess.SY)
备注: 13 pages, 16 figures, 7 tables. Design files and firmware: this https URL
Abstract:We present the NeuralNexus Arm, an open, low-cost 6-DOF robotic manipulator built by an undergraduate engineering team, together with the design decisions and debugging experience needed to reproduce it. The arm is driven by a single STM32H743 microcontroller on a custom printed circuit board (PCB) and combines two stepper-driver strategies on one controller: push-pull 3.3 V step/direction outputs for onboard TMC2209 drivers on the three wrist joints, and open-drain outputs for external CL57T and DM542 drivers on the three high-torque proximal joints. We describe the mechanical design, mixed-driver electronics, interrupt-driven firmware, a MATLAB/Simscape-based inverse-kinematics pipeline, a browser-native control interface using the Web Serial API, and a lightweight vision pipeline for object localisation and autonomous pick-and-place tasks. We also document non-obvious hardware and firmware failure modes encountered during the transition from a development board to the custom PCB as reproducibility guidance. All design files and firmware are released openly. The platform actuates all six axes under coordinated control at a 2 kHz update rate and executes both manual and pre-recorded motions from the browser interface.
[CV-75] VersaDB: A High-Performance AI Storag e Database for Unifying Mutimodal Datasets
链接: https://arxiv.org/abs/2608.22795
作者: Cong Wang,Zelin Liu,Yang Luo Ran Zhang,Zhijian Guo,Hui Zhang,Fan Yu,Yanfei Cao,Naijie Gu,Jun Yu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:The AI field has been rapidly developing, leading to the emergence of a large number of AI training datasets of various types. These datasets contain different modalities, including text, images, audio, etc., and may come in various data storage formats. With the advancement of AI hardware, AI computation units like GPUs, TPUs, and NPUs can greatly accelerate the training speed of AI models, which in turn increases the demand for faster data processing. When using existing AI processing frameworks to handle datasets with different modalities and storage formats, processing speeds may be suboptimal due to issues such as data layout and the way users handle the data. Therefore, using a unified database to store multiple data formats can better manage and optimize data access. In this paper, we introduce VersaDB, a database designed specifically for AI datasets with various modalities. We implemented a page-based storage system, separating structured and unstructured data. Additionally, we generated B+ tree-based index files to accelerate data access. VersaDB supports automatic sharding and maintains a hierarchical metadata management system, with corresponding metadata maintained at the page, shard, and global levels, forming the foundation for the efficient operation of the database. We also focused on ease of use by providing APIs for directly converting datasets into VersaDB, as well as APIs for converting popular AI data storage formats (e.g., CSV, TFRecord, .bin) into this http URL experiments show that using VersaDB can achieve up to 5.35x acceleration and maintain consistent performance across different parallelism levels.
[CV-76] GuidedFlow: An Attention-Guided Framework for Anomaly Detection in Additive Manufacturing
链接: https://arxiv.org/abs/2608.22789
作者: Sosmita Paul,Krishna Roy
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Additive Manufacturing (AM) plays a vital role in the ongoing industrial revolution. However, quality control remains crucial and challenging due to printing defects or potential cyber-physical intrusions. Image or video-based anomaly detection is a key effort towards addressing these challenges. Various approaches have been explored in this domain, including reconstruction-based, embedding-based, and flow-based methods. Though normalizing flow-based methods address some of the core challenges of unforeseen defects and generalization while maintaining detection performance, existing approaches struggle with tiny/stringing defects common in 3D printing. In a small-data setting, this poses a limitation in generalization. To address these limitations, we propose \textbfGuidedFlow, a novel attention-guided normalizing flow model for anomaly detection and localization. GuidedFlow employs a pre-trained ResNet model, fine-tuned on the domain dataset. An attention-guided spatial and temporal flow framework models the dynamics across multiple scales and frames. A Spatio-Temporal Attention Network (SAN) enables the flow model to prioritize relevant contextual cues from input frames. We evaluate GuidedFlow on our AM3D-AD dataset, consisting of benign and anomalous real 3D printed object images and videos. We also conduct a comparative study using the MVTec-AD industrial image anomaly detection dataset. Experimental results demonstrate that GuidedFlow outperforms most of the state-of-the-art models with enhanced detection accuracy and AUROC.
[CV-77] OmicSync: Reliability-Aware Spatial Multi-Omics Clustering with Evidence-Constrained LLM Reasoning
链接: https://arxiv.org/abs/2608.22785
作者: Rabeya Tus Sadia,Qiang Ye,Qiang Cheng
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Spatial multi-omics technologies jointly profile gene expression, surface proteins, and histology at each tissue spot, yet most spatial domain discovery methods provide only cluster assignments, without indicating assignment reliability, modality contributions, or why a domain decision should be trusted. We present OmicSync, a reliability-aware spatial multi-omics framework that couples unsupervised domain clustering with evidence-constrained LLM reasoning using model-derived per-spot signals, including assignment confidence, epistemic routing uncertainty, and modality-routing weights. These signals are converted into structured evidence dictionaries and used to generate standard, stepwise, counterfactual, contrastive, and uncertainty-focused explanations. OmicSync integrates a KAN-GCN backbone with spatial encoding, cross-modal fusion, uncertainty-aware routing, cell-type supervision, and missing-modality imputation. We further introduce OmicSync-R, which closes the reasoning-clustering loop by using automatically computed reasoning-quality scores as REINFORCE rewards, allowing reasoning coherence to shape the latent structure without backpropagating through the language model. Across four 10x CytAssist FFPE spatial proteomics benchmarks, OmicSync achieves the best average rank on Human Tonsil (1.44), Glioblastoma (1.78), and Tonsil Add-on (1.22), and second-best on Human Breast Cancer (2.33). OmicSync-R further improves ARI on Human Breast Cancer from 45.73 to 46.72 and outperforms existing methods on six of nine clustering metrics. Together, OmicSync and OmicSync-R enable reliability-aware, spot-level auditable spatial domain discovery guided by evidence-constrained reasoning.
[CV-78] Neural Operator based Multi-Field Reconstruction of Inner Solar Boundary State
链接: https://arxiv.org/abs/2608.22782
作者: Vignesh Kumar Pandian Sathia,Reza Mansouri,Dustin J. Kempton,Pete Riley,Rafal A. Angryk
类目: Machine Learning (cs.LG); Instrumentation and Methods for Astrophysics (astro-ph.IM); Solar and Stellar Astrophysics (astro-ph.SR); Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages, 4 figures, preprint, accepted at International Conference on Machind Learning and Applications
Abstract:The Solar wind is a continuous flow of charged particles emanating from the solar surface and governed by complex, interacting magnetohydrodynamic processes. Accurate specification of inner-boundary conditions is essential for heliospheric modeling and solar-wind prediction. In many practical applications, only a subset of interacting multi-field variables is directly available, but for a comprehensive view of solar wind prediction and downstream magnetohydrodynamic simulations, a more complete boundary state is required. In this work, we study the problem of learning the multi-field multi-scale solar magnetohydrodynamic state at 30 solar radii ( R_\odot ) using operator learning. Specifically, given the radial velocity and radial magnetic field, we aim to reconstruct the non-radial velocity and magnetic field components, radial and non-radial current density, thermodynamic density, and pressure components. This mapping is highly nonlinear, spatially coupled, and multi-scale, making it a challenging task for data-driven scientific machine learning. To address this problem, we employ a Local Neural Operator (LocalNO) that learns mappings between input and output function spaces while retaining locality and resolution-awareness. Unlike conventional regression models and autoencoder models, neural operators are better suited for learning structured field-to-field transformations arising from physical systems. The resulting predictions along with inputs are intended to serve as boundary condition variables for future inner-heliospheric modeling pipelines. Comments: 8 pages, 4 figures, preprint, accepted at International Conference on Machind Learning and Applications Subjects: Machine Learning (cs.LG); Instrumentation and Methods for Astrophysics (astro-ph.IM); Solar and Stellar Astrophysics (astro-ph.SR); Computer Vision and Pattern Recognition (cs.CV) ACMclasses: J.2; I.5.4 Cite as: arXiv:2608.22782 [cs.LG] (or arXiv:2608.22782v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2608.22782 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-79] Can We Perform Online RL for Image Editing without Editing Rewards?
链接: https://arxiv.org/abs/2608.22780
作者: Qichao Ma,Jikang Cheng,Ling Liang,Zhaofei Yu,Tiejun Huang,Renye Yan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Reinforcement learning (RL) enables direct preference optimization for image editing through editing-specific rewards, which remain less developed due to costly triplet supervision and complex task-dependent calibration. In contrast, text-to-image (T2I) generation benefits from a mature and diverse reward ecosystem spanning semantic alignment, aesthetics, realism, glyph shape, and other visual preferences. Extending this ecosystem to image editing would substantially broaden the range of visual preferences accessible to RL-based optimization, prompting the central question: \emphCan We Perform Image Editing RL without Editing Rewards? In this paper, we argue that the standard image editing dimensions have potential to be mapped to the T2I reward space: image quality can transfer directly, prompt following can be aligned through a description of the desired visual state, and reference consistency admits a coarse semantic conversion by encoding the source content to preserve. However, editing instructions specify relative changes, whereas T2I rewards require self-contained target descriptions; moreover, semantically valid captions from generic vision-language models may be incompatible with the frozen reward. Hence, we further introduce Lever-Edit, a two-stage framework that learns a reward-aligned captioner for counterfactual target descriptions, freezes it, and optimizes the editing policy solely with the transferred T2I reward. Experiments show competitive editing alignment and source preservation against editing-reward-based fine-tuning, while outperforming intuitive transfer baselines.
[CV-80] LagrangeGS: Non-Conservative Lagrangian System on Dynamic 3D Gaussian Splatting BMVC2026
链接: https://arxiv.org/abs/2608.22773
作者: Shogo Sato,Takuhiro Kaneko,Shoichiro Takeda,Tomoyasu Shimada,Riku Inoue,Kazuhiko Murasaki,Ryuichi Tanida
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by BMVC2026
Abstract:Dynamic 3D Gaussian Splatting (3DGS) achieves photorealistic reconstruction of time-varying scenes, and recent physics-aware extensions improve extrapolation by explicitly predicting velocity fields. However, these extensions merely fit vector fields to visual deformations without satisfying Lagrangian mechanics, leading to three major issues: (i) physically inconsistent trajectories, (ii) lack of time-reversibility, and (iii) geometric collapse during long-term extrapolation. In this paper, we propose LagrangeGS, which formulates dynamic 3DGS as a non-conservative Lagrangian system. While this Lagrangian formulation fundamentally solves (i), a direct application of general LNNs to dynamic 3DGS requires a large velocity-Hessian inversion for millions of Gaussian particles. To overcome this computational bottleneck, we approximate the velocity-Hessian as an identity matrix, decoupling particle dynamics for computational tractability. For (ii), we restrict the non-conservative forces to be explicitly time independent, enabling consistent backward integration. Finally, to address (iii), we introduce local rigid alignment that regularizes particle trajectories. Extensive evaluations on dynamic scene benchmarks demonstrate that LagrangeGS enables stable long-term extrapolation, consistent time reversal, and counterfactual physics-based editing without retraining.
[CV-81] ByteAction: Byte-space Action Recognition Foundation Model
链接: https://arxiv.org/abs/2608.22760
作者: Fangcheng Li,Zhen Yu,Kejun Wu,Qiong Liu,You Yang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Byte-space Action Recognition (BAR) aims to recognize human actions directly from compressed image bitstreams without any pixel decoding. By operating entirely in byte space, BAR is inherently independent of file integrity and pixel-level reconstruction, making it naturally applicable to privacy-sensitive scenarios and robust against bitstream corruption. In this paper, we propose ByteAction, a BAR foundation model that achieves accurate action recognition on corrupted image bitstreams. ByteAction follows a dual-view byte-level recognition framework. It constructs weakly and strongly corrupted bitstream views, which are augmented by Bitstream Pattern Augmentation (BPA) and encoded with a shared ByteFormer backbone. The model is optimized with both classification and corruption consistency objectives. Specifically, we propose Bitstream Pattern Augmentation (BPA), which reshapes one-dimensional byte sequences into two-dimensional byte matrix and applies region-level erasure to encourage the model to learn robust cross-region byte dependencies. We further propose a Corruption Consistency Training strategy that constrains the model to maintain stable predictions across different corruption severities through bidirectional KL divergence. Experiments on the image bitstream from Stanford40, PPMI, and PASCAL VOC 2012 Action demonstrate that ByteAction achieves state-of-the-art corruption robustness across all scenarios while maintaining competitive intact bitstream performance.
[CV-82] Object-Uni: A Unified Model for Object-Centric Spatial Understanding and Controllable Generation
链接: https://arxiv.org/abs/2608.22757
作者: Mining Tan,Yinuo Wang,Ziqi Zhou,Weize Quan,Sifei Li,Jingdong Chen,DanDan Zheng,Libin Wang,Weiming Dong
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Unified models for visual understanding and generation have made rapid progress, yet they still lack the ability to understand and manipulate the spatial states of object instances. Existing models can describe objects in natural language, but they struggle to precisely represent continuous object poses and generate geometrically consistent images under target viewpoints. To mitigate this, we propose \emphObject-Uni, a unified model for object-centric spatial understanding and controllable generation. Specifically, we formulate object-centric spatial intelligence as a unified problem connecting pose perception, spatial reasoning, pose-conditioned generation, and object-centric novel view synthesis. We treat object pose as an explicit geometric variable shared by understanding and generation, rather than merely a prediction label or control signal. To make pose usable by multimodal large language models, we propose a viewpoint-based orientation abstraction that maps orientation into structured viewpoint descriptions while preserving continuous geometric supervision. We further construct an object-centric spatial benchmark (UniSpatial-80K) and train a unified model with an object-token-grounded pose anchor to associate each instance with its pose state. Experiments show that our model improves object-level pose understanding and pose-controllable generation, moving unified models from describing objects toward manipulating spatial states.
[CV-83] Seeing the Unseen: Semantic-in-Gaussian for Sparse-View 3D Generalization ECCV2026
链接: https://arxiv.org/abs/2608.22740
作者: Zeyang Bai,Yunpeng Wang,Yunbiao Wang,Jun Xiao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at the ECCV 2026 Workshop on 3DWM
Abstract:Generalizable 3D Gaussian Splatting (G-3DGS) has emerged as a promising approach for novel view synthesis undersparse-view settings. However, existing frameworks remain restricted by pixel-aligned Gaussian estimation, whichstruggles in partially observed or occluded regions and often leads to incomplete surfaces or structural collapse. Toaddress these challenges, we propose SeeU (Seeing the Unseen), a novel G-3DGS framework. We frame its core design asSemantic-in-Gaussian: semantic-conditioned refinement in Gaussian space. Specifically, we introduce a Cross-viewEntropy-Aware (CEA) module that aggregates multi-view semantic and geometric cues into compact embeddings. Theseembeddings guide the Conditional Gaussian Transformer, which applies residual updates to coarse Gaussians, helpingrecover under-constrained regions of partially observed structures while preserving surface consistency. Comprehensiveexperiments on multiple benchmarks demonstrate that SeeU consistently improves rendering quality and structuralcompleteness while retaining efficient feed-forward inference. Especially under challenging extrapolation settings,SeeU achieves an average improvement of 2.44 dB in PSNR compared to recent SOTA G-3DGS methods.
[CV-84] SymmAdapt: Symmetrical Flow Matching for Source-Free Domain Adaptation in Medical Image Segmentation MICCAI2026
链接: https://arxiv.org/abs/2608.22532
作者: Tal Grossman,Noa Cahan,Hayit Greenspan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at SASHIMI 2026, held in conjunction with MICCAI 2026. To appear in the Springer LNCS proceedings of the MICCAI 2026 Satellite Events
Abstract:Domain shift across imaging modalities and acquisition sites remains a significant barrier to the clinical deployment of segmentation models. Source-free unsupervised domain adaptation (SFUDA) addresses this by adapting a pretrained model to an unlabeled target domain without requiring access to sensitive source data. We introduce a novel SFUDA framework built on Symmetrical Flow Matching, a unified generative model that segments an input image and synthesizes a source-like image from a mask within the same learned flow. By initializing inference from a domain-agnostic Gaussian origin, the model preserves structural consistency across domains and grounds predictions in learned anatomy rather than shifted texture statistics. Our pipeline leverages this symmetry to generate reliable pseudo-labels and corresponding source-like synthetic images from unlabeled target data, creating a generative replay buffer that anchors source knowledge during a generative self-training stage that fine-tunes on a joint set of real target and synthetic source-like images. We evaluate on abdominal multi-organ and cardiac segmentation, covering cross-modality MRI-CT shifts, and multi-site prostate segmentation. Our approach outperforms SFUDA baselines and is competitive with conventional UDA methods.
[CV-85] RS3-Prune: Read-Sparse Store-Sparse Token Pruning for Video Object Segmentation BMVC2026
链接: https://arxiv.org/abs/2608.22526
作者: Avilasha Mandal,Sarvesh Shashikumar
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to BMVC 2026
Abstract:We introduce RS ^3 -Prune, a training-free token-pruning recipe that instantiates as a small set of inference time hooks atop existing video object segmentation (VOS) networks. Modern VOS models have converged on a common, expensive design: an image encoder produces a dense token grid for every frame, and a memory bank accumulates these tokens across all previously processed frames to condition future predictions. As a video grows longer, the resulting token budget governs both per-frame latency and peak GPU memory. Hence these models break on use cases such as — long-form video or real-time deployment on memory-bounded accelerators. In this work we argue that the right axis along which to compress memory-bank VOS is the token budget itself. RS ^3 -Prune operates in two precise locations within an arbitrary memory-bank VOS pipeline: at the boundary between the image encoder and the memory-attention readout, where we restrict the queries that participate in the cross-frame attention to only a small, geometrically informed subset; and at the boundary between the memory encoder and the memory bank, where we restrict which tokens are ever permitted to enter the bank to those that lie within the object’s spatial extent. Over various established benchmarks, RS ^3 -Prune delivers up to 38.8% FPS speedup and reduces 13.1% peak memory usage, while preserving a competitive \mathcalJ \mathcalF compared to the unmodified VOS networks.
[CV-86] VISTA: Test-Time Compositional Alignment for Visual Autoregressive Generation
链接: https://arxiv.org/abs/2608.22521
作者: Hossein Shahabadi,Niki Sepasian,Mahdieh Soleymani Baghshah
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Visual autoregressive (VAR) models have emerged as a fast, high-quality alternative to diffusion for text-to-image generation, but like diffusion models they exhibit persistent compositional failures, producing images that violate the attribute bindings and spatial relations specified in the prompt. While a rich line of test-time alignment methods has developed for diffusion, no comparable approach exists for next-scale VAR generation, whose stateful, discrete, multi-resolution sampling process makes existing techniques inapplicable. We close this gap with \textbfVISTA (\textbfVisual Autoregressive \textbfSemantic \textbfTest-time \textbfAlignment), the first gradient-based test-time alignment framework for next-scale autoregressive image generation. Built on Infinity, VISTA intervenes directly in the generation process, optimizing intermediate representations through the frozen transformer to steer visual predictions toward compositional constraints, without modifying model parameters or requiring additional training. VISTA introduces the mechanisms needed to make such optimization stable across scales, together with an extensible objective space that any differentiable constraint on cross-attention can plug into. Across two benchmarks and two model scales, VISTA improves every targeted compositional category, raising the mean targeted score by nearly 20% on a 2B backbone and almost 6% on an 8B backbone, with the largest gains on spatial relations. Image quality is preserved: an independent preference model VISTA never optimizes scores its outputs nearly 20% higher. Notably, the 2B model with VISTA surpasses a backbone four times its size, indicating that a substantial part of the compositional gap between model scales is recoverable at test time.
[CV-87] Learning Sample-wise Rank-aware Interpolation Weights for Composed Visual Data Retrieval ECCV2026
链接: https://arxiv.org/abs/2608.22500
作者: Boseung Jeong,Taegyu Park,Donghyeon Kwon,Hyunsouk Cho,Suha Kwak
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: ECCV 2026
Abstract:At the heart of composed visual data retrieval is the fusion of a reference visual input and a textual modification into a single query. While current state-of-the-art methods utilize multimodal large language models for this fusion, their complexity introduces prohibitive querytime latency, limiting their scalability. We instead revisit the efficacy of simple linear interpolation within an embedding space, and introduce SRAIN, the first framework that dynamically predicts query-specific interpolation weights. The key challenge lies in the fact that the quality of an interpolation weight should be measured by the interpolated embedding’s discriminability from negatives as well as its proximity to true targets; this makes collecting and predicting optimal weights intractable. We overcome this bottleneck through two key innovations: batch-wise rank-aware weight estimation during training, and a compact memory bank that synthesizes hard negatives during inference. SRAIN achieves the best in composed video retrieval and matches the current state of the art in composed image retrieval, all while substantially reducing querytime latency compared to MLLM-based alternatives.
[CV-88] HeatTok: Enhancing Remote Sensing Image Understanding via Thermodiffusion-based Tokenization ACM-MM2026
链接: https://arxiv.org/abs/2608.22485
作者: Yingying Yan,Jiaqi Tang,Wei Wei,Qianzhou Wang,Jinjian Wu,Botong Geng,Jianmin Chen,Yuyang Xia,Lei Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 19 pages (10 pages main text + appendix), 10 figures. Accepted at the 34th ACM International Conference on Multimedia (ACM MM 2026), Rio de Janeiro, Brazil, November 10–14, 2026
Abstract:Current visual tokenizers in Multimodal Large Language Models (MLLMs) predominantly rely on patch-based partitioning, which causes severe semantic mixture and object fragmentation in remote sensing imagery due to the irregular contours of geo-objects. Moreover, existing adaptive methods struggle to extract precise object-level tokens and lack dedicated geometric positional encodings for irregular regions. In this paper, we propose HeatTok, a semantic-aware tokenizer driven by thermodiffusion aggregation. Inspired by the physical principles of heat conduction, HeatTok adaptively merges adjacent homogeneous regions to generate semantically independent, object-aligned irregular tokens. To enable MLLMs to perceive these irregular shapes, we design the Gaussian Multimodal Rotary Positional Embedding (G-MRoPE), which models token spatial distributions via 2D Gaussians and explicitly injects center, scale, and orientation cues. Extensive evaluations on the VRSBench and EarthVQA datasets demonstrate that HeatTok effectively preserves object-level semantic integrity and achieves state-of-the-art performance under a reasonable token budget. The code is available: this https URL.
[CV-89] M3ISR: A Multi-Modal Multi-View Benchmark for 3D/4D Gaussian Splatting and Feedforward Compression
链接: https://arxiv.org/abs/2608.22465
作者: Xinhui Liu,Lei Liu,Zhenghao Chen,Lebin Zhou,Wei Wang,Wei Jiang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:High-fidelity free-viewpoint video (FVV) and interactive rendering increasingly rely on explicit Gaussian representations, yet practical deployment remains constrained by representation size, dynamic updates, and computational cost. Existing multi-view video benchmarks provide valuable real-captured content, but they make it difficult to isolate the effects of controlled camera geometry, representation efficiency, and temporal redundancy. We introduce M ^3 ISR, a controlled synthetic benchmark for 3D and 4D Gaussian Splatting (3DGS/4DGS). The benchmark contains 25 scenes from five indoor and outdoor scene groups, two camera/motion configurations, six synchronized 1080p views, and dense ground-truth annotations including RGB, camera parameters, depth, semantic and instance segmentation, and static–dynamic masks. The shared-center camera design intentionally isolates angular view variation and enables controlled evaluation of novel-view synthesis and representation efficiency. We organize M ^3 ISR into five complementary tracks covering 3DGS synthesis, 4DGS synthesis, 4DGS streaming, 3DGS compression, and 4DGS compression. Representative baseline results show small differences in static reconstruction quality but substantial differences in representation storage, while the evaluated streaming methods exhibit substantially higher reported training or reconstruction cost than the corresponding offline dynamic reconstruction baselines. We further define feedforward compression tasks for 3DGS and 4DGS and provide reference rate–distortion formulations and preliminary baseline evaluations. The benchmark is intended as a controlled and complementary testbed for systematic study of Gaussian-based FVV reconstruction, compression, and streaming.
[CV-90] Robust Bimanual Vision-Language-Action Models via Embarrassingly Simple Modality Masking
链接: https://arxiv.org/abs/2608.22419
作者: Dongzhou Cheng,Ziang Li,Yixiao Zhou,Haojuan Li,Jinghao Zhang,Lei Lei,Minjing Dong,Jie Gui,Jiaqi Wang
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: 35 pages, 22 figures, 9 tables
Abstract:Query-based Vision-Language-Action (VLA) models offer low-latency inference that is attractive for bimanual robotic manipulation, but we observe that they can still exhibit discontinuous actions and execution failures in complex dual-arm tasks. We hypothesize that unstable multi-view and language fusion is one contributing factor in these failures, often coinciding with attention spreading to distracting regions. To improve robustness, we introduce the Modality Masking Mechanism (M3), an embarrassingly simple, training-only strategy that requires no architectural changes or large-scale robot pretraining. M3 stochastically masks subsets of modality channels during training, exposing the policy to controlled partial observations and encouraging it to rely less on distracting cues and more on evidence that remains reliable. We evaluate M3 on ten bimanual tasks from RoboTwin 2.0 and on three long-horizon real-world tasks. Compared with the Adapter baseline, M3 improves average success by 21.7% in the Clean setting and 11.4% in Clean2Rand, where policies are trained on clean demonstrations and evaluated on randomized scenes, while also improving averaged real-world full-task success by over 30%. These results suggest that structured training-time masking is a practical way to improve the robustness of query-based VLA policies for bimanual manipulation.
[CV-91] MotionDLO: Hybrid Event- and Frame-Based Tracking of Deformable Linear Objects
链接: https://arxiv.org/abs/2608.22398
作者: Annalena Hartmann,Priyamvada Ajithkumar,Patrick Bründl,Jörg Franke
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Reliably tracking moving deformable linear objects (DLOs) while simultaneously ensuring robustness, accuracy, and temporally consistent state estimation remains a fundamental challenge in robot perception. We introduce MotionDLO, a real-time tracking framework specifically designed to overcome these limitations in temporal continuity and latency. The method exploits the high temporal resolution and sparsity of event-based cameras and combines segmentation with the Coherent Point Drift (CPD) algorithm under the principles of Motion Coherence Theory. This integration enables temporally consistent shape estimation while maintaining a low computational overhead. Existing event-based tracking methods are typically computationally efficient but exhibit reduced accuracy compared to frame-based approaches, or alternatively compromise event sparsity to achieve competitive performance. To resolve this trade-off, we propose a hybrid event- and frame-based tracking architecture that preserves the complementary strengths of both sensing modalities. The event stream ensures high-frequency motion updates, while frame-based information stabilizes spatial accuracy and object identity. We demonstrate that the proposed framework reliably associates DLO instances across video sequences, enabling robust perception for robotic manipulation tasks. Experimental results validate real-time performance at 12 ms update rates and accurate shape tracking with an point-to-curve error as measurement of accuracy of up to 0.43 mm, supporting dynamic path adaptation during manipulation. The source code and demonstration datasets are publicly available.
[CV-92] LiST: Local-Simplex Test-Time LoRA Fusion EMNLP2026
链接: https://arxiv.org/abs/2608.22370
作者: Yihua Shao,Jia Li,Siyu Chen,Xinyu Luo,Yang Liu,Kecheng Chen,Xinwei Long,Lingyu Zhu,Fanhu Zeng,Maolin Wang,Ziyang Yan,Jingcai Guo,Hao Tang,Nicu Sebe,Zhenyi Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by EMNLP 2026 Finding
Abstract:Task-specific LoRA adapters offer a modular way to specialize large language and vision-language models. However, existing adapter composition methods are mostly static and cannot adapt to individual test inputs. To address these issues, we propose \textbfLiST, a label-free test-time LoRA fusion framework that converts an existing LoRA bank into a target-conditioned local simplex and searches sample-specific fusion weights at inference time. LiST builds joint task representations from LoRA parameter anchors and prompt-level behavior vectors, retrieves neighboring adapters as a local search space, and performs branch-preserving fusion without updating the backbone or adapters. Candidate weights are selected by a prompt-level energy with prior, geometric, and stochastic-consistency constraints, and are deployed only when they pass a safe acceptance rule. Otherwise, LiST falls back to a target-conditioned prior. Experiments on multimodal and language benchmarks show that LiST outperforms static LoRA merging and conventional test-time adaptation baselines, while preserving task-specific adapter utility and improving robustness on unseen tasks.
[CV-93] DiD It in 87 Minutes: A Label-Free Softmax-to-Linear Adaptation of Vision Transformers for Object Detection
链接: https://arxiv.org/abs/2608.22368
作者: Huaiyuan Qin,Gabriel James Goenawan,Zihang Lin,Muli Yang,Hongyuan Zhu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:While linear attention is a compelling mechanism for high-resolution object detection due to its reduced cost for global token mixing, converting the Softmax-attention ViT backbone of a trained detector into a linear-attention one is not a trivial drop-in replacement. Directly swapping the attention operator leads to severe performance degradation, and generic label-free distillation, though effective for classification, often fails on detection tasks. We argue that the central challenge is \textitdetector-interface preservation: the converted backbone must reproduce the exact feature tensors expected by the fixed downstream detector, rather than merely imitating internal Softmax hidden states. To address this, we introduce Detector-Interface Distillation (DiD), a label-free conversion method that exclusively trains the linear-attention backbone by aligning detector-facing interface tensors with those of a frozen Softmax teacher. On DOTA-v1.5, DiD substantially outperforms established baselines and matches supervised, fully trained linear models. Adaptation completes in roughly 87 minutes on 4 GPUs, and the linearized backbone cuts inference latency by ~62% and peak memory by ~49%. We hope our findings offer the community a simple, label-free route to reusing trained Softmax detectors as efficient linear ones, and encourage interface-aware objectives in future architecture-conversion work.
[CV-94] When Do VLMs Help Arabic Manuscript OCR? A Cross-Dataset Study
链接: https://arxiv.org/abs/2608.22366
作者: Moshiur Farazi,Firoj Alam,Abderrahmane Maaradji,Zakaria Maamar,Hamdy Mubarak,Wajdi Zaghouani
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision-language models (VLMs) are increasingly being used for document understanding, yet their role in Arabic and Islamic manuscript recognition remains underexplored. To address such a gap in this paper, we evaluate traditional OCR, general-purpose VLMs, Arabic-specialized VLMs, and OCR-conditioned VLM correction across eight Arabic text datasets spanning historical manuscripts, aged printed books, clean print, multi-domain documents, and handwriting. The results show that no single approach dominates across setups. On line-level historical manuscripts, VLMs are close to Tesseract; on page-level manuscript images, they perform better; and in several settings, an OCR-conditioned corrector improves over both standalone OCR and standalone VLMs. The central finding is an OCR-prior recoverability principle: OCR conditioning helps when the OCR output remains visually and textually recoverable, providing anchors that the VLM can refine against the image. It improves recognition on aged print, clean print, mixed-domain Arabic, and some Naskh manuscripts, but degrades performance when the prior is script-mismatched or systematically misleading, as in Maghribi manuscripts and realistic student handwriting. Additional diagnostics show that Arabic VLM-OCR is sensitive to diacritics, preprocessing, generation budget, and repetition loops. These findings support an adaptive OCR-VLM workflow that routes pages according to script, OCR-prior recoverability, length diagnostics, and failure-mode indicators.
[CV-95] Pre-Decoding Acoustic Triage for Budgeted Vision-Language Captioning of Untrimmed Egocentric Video
链接: https://arxiv.org/abs/2608.22359
作者: Masoud Jalayer,Changyi Li,Yu Xiao
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Sound (cs.SD)
备注: 18 pages, 5 figures, 9 tables. Under review at IEEE BigData 2026, Industrial and Government Track. Code: this https URL
Abstract:Automatically analyzing hours-long egocentric video is increasingly essential for progress monitoring, quality control, and safety in logistics, construction, and manufacturing. Yet current pipelines that process short, fixed-size windows with a vision-language model (VLM) are prohibitively expensive because cost scales with the number of model calls. To reduce this cost, prior work proposes triage policies to select which windows merit a VLM invocation. However, these policies either sample uniformly or rank windows using visual features, which ironically requires the video decoding that the budget constraints are meant to avoid. We propose audio-first triage: select windows using the lightest modality, scored before any video frame is decoded, so the approach composes naturally with token compression or quantization. The novelty lies in the objective, not the representation: rather than a per-frame sound-event detector, we train the selector to trigger once per action. This objective shift improves action coverage by 4.0-10.8 percentage points across all evaluated call rates, using frozen AudioSet-pretrained features without domain-specific sound-event labels. Using fewer than half of the available calls, the triage cuts 9-20% of VLM calls at matched coverage on EPIC-KITCHENS-100 (EK-100), surpasses uniform sampling through the mid-range on Ego4D over 247 clips, and outperforms two recent visual keyframe selectors. Code, the reference implementation and every results file this manuscript reads are at this https URL.
[CV-96] Multimodal examination answer data with expert-designed Outcome-Based Education rubrics for criterion-level assessment
链接: https://arxiv.org/abs/2608.22346
作者: Jahangir Alam SM,Md Khalid Syfullah,Saad Ahmed,Munira Akter Mou,A K Z Rasel Rahman,A.K.M. Masudur Rahman,Mohammed Sowket Ali
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:This data article describes a multimodal collection of scanned examination answers paired with expert-designed Outcome-Based Education (OBE) grading metadata. The collection contains 485 answer submissions from 415 consenting students at four academic institutions. Eight faculty contributors supplied examination materials covering nine subjects and 12 distinct question templates. Each answer-level item links a scanned PDF to a randomized identifier, subject label, question, model answer, criterion definitions, performance-level descriptions, criterion marks, and a total mark. The 12 rubrics contain 47 criteria in total. The scans retain realistic academic content, including handwriting, printed text, equations, tables, code, figures, sketches, and diagrams. CamScanner, Adobe Scan, and conventional scanners contributed variation in illumination, contrast, orientation, compression, and resolution. Diverse handwriting, crossed-out work, revised calculations, and inserted corrections add further visual variability for robustness and generalization studies. Preparation involved heterogeneous-source consolidation, label and text standardization, score validation, identifier randomization, filename randomization, and JSON-to-PDF integrity checks. An answer-level audit confirmed 485 unique identifiers, 485 unique PDF filenames, agreement between each total mark and its criterion-mark sum, and scores within the applicable rubric maximum. The data can support rubric-aware automated evaluation, multimodal document understanding, criterion-level feedback, score prediction, and privacy-aware OBE assessment research. Access is restricted to research use and is available from the corresponding author upon reasonable request.
[CV-97] Fast and Compact 3D Gaussian Splatting with Polarized Opacity Prior
链接: https://arxiv.org/abs/2608.22344
作者: Zi-Ming Wang,Kai-Wen Duan,Kowei Huang,Akihiro Sugimoto,Shang-Hong Lai
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:3D Gaussian Splatting (3DGS) achieves state-of-the-art rendering quality at real-time speeds but suffers from “model bloat” - a large number of redundant, low-opacity Gaussians that inflate memory usage and training costs. This inefficiency stems from the standard “densify-then-prune” paradigm, which expands the model aggressively before relying on pruning to achieve compactness. To mitigate this problem, we present an efficient training framework that builds an intrinsically compact representation, replacing the conventional densify-then-prune cycle. Our method leverages a synergistic design: an L2 reconstruction loss to provide error-proportional gradients that stabilize optimization, and a novel Polarized Opacity Prior (POP) to actively manage the Gaussian population. POP steers informative primitives toward full opacity and uninformative ones toward transparency, enabling natural pruning and accelerating rendering through Early Ray Termination. Experiments on three public datasets demonstrate that our approach consistently achieves accelerated 3DGS training with significantly fewer Gaussians while maintaining comparable visual reconstruction quality. These results show that the proposed framework provides a simple and effective path toward fast and inherently compact 3DGS training.
[CV-98] ransHands: Repurposing Human Pose Encoders as Hand Pose Encoders
链接: https://arxiv.org/abs/2608.22341
作者: Milo Piccioli,Gianluca Amprimo,Claudia Ferraris,Gabriella Olmo
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Lifting 3D hand poses from 2D monocular representations remains challenging due to the limited availability of large-scale, diverse 3D-annotated hand datasets, in contrast to the abundance of human body motion data. We address this limitation by transferring motion representations learned from large body pose corpora to the hand domain. We introduce TransHands, a backbone-agnostic transfer learning framework that enables pre-trained human motion encoders to be effectively adapted for 3D hand pose estimation from 2D pose inputs. Rather than training hand-specific biomechanical models from scratch, TransHands combines a two-stage training and fine-tuning strategy with a lightweight hand-specific input adaptation module that aligns hand kinematics with the representation space learned for full-body motion. We evaluate TransHands across four state-of-the-art motion modeling architectures, including transformer-based, graph-based, and frequency- domain models. Results demonstrate that motion priors learned from body pose data transfer consistently across architectures, yielding consistent accuracy gains, strong cross-domain generalization, particularly in challenging egocentric settings, and applicability for downstream tasks in real-world contexts.
[CV-99] AcroMELD: Recovering Interactive PDF Forms with Structure-Aware Graph Set Transformers
链接: https://arxiv.org/abs/2608.22338
作者: Samuel Abramov(Accessful GmbH)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 13 pages, 2 figures, 7 tables. Source compiles to a tagged PDF 2.0 document with MathML embedded as associated files
Abstract:Interactive PDF form fields are often absent from documents that visually resemble forms, leaving users unable to enter data without printing or external editing tools. Detecting the missing widgets is difficult because a field may be indicated by several overlapping cues, born-digital PDFs expose useful but incomplete drawing structure, and dense pages can contain hundreds of fields. We introduce AcroMELD (AcroForm Multi-source Evidence Linking Decoder), a 39.4M-parameter detector that combines a high-resolution visual transformer with label-free PDF primitives. Its 896-query set comprises 384 visual proposals, 384 structure-seeded proposals, and 128 learned recovery queries. Four graph-set layers exchange information over geometry-biased sparse neighborhoods and cross-attend to PDF structure. A learned same-field relation links co-referent candidates, while a localization-quality head is trained on the containment-aware overlap used by the downstream recovery decision. We define a hash-bound evaluation protocol with disjoint development, calibration, internal-test, and quarantined external-holdout roles. The sealed, single-seed candidate reaches native containment micro- F_1 0.9344 on the internal test and 0.8477 on the one-shot external holdout (95% PDF-cluster bootstrap interval [0.8339, 0.8605]). This passes the registered historical FFGBT-v8 reference by 0.0186 absolute F_1 . Under the stricter external adapter, however, performance is 0.7786 IoU- 0.5 F_1 and 0.2900 COCO mAP, below a locally evaluated CommonForms-L reference; the signature class receives no prediction at the selected threshold. Thus the result supports the registered operational gate while exposing substantial domain and rare-class limitations.
[CV-100] Motion-Aware Reasoning from Speech to Mask Tracks: Runner-up Solution for the MeViS-Audio Track of the 8th LSVOS Challenge 2026
链接: https://arxiv.org/abs/2608.22337
作者: Jinxing Zhou,Suiyi Zhao,Yanghao Zhou,Ruohao Guo
类目: Multimedia (cs.MM); Computer Vision and Pattern Recognition (cs.CV); Sound (cs.SD)
备注:
Abstract:Speech-guided referring video object segmentation aims to recover the mask tracks of objects specified by a spoken motion description. Here, speech carries a linguistic instruction rather than acoustic evidence from a sounding object, so a solution must connect speech recognition, motion-centric temporal grounding, mask tracking, and explicit no-target handling. We introduce Speech2MaskTrack, our approach for the MeViS-Audio track of the 8th LSVOS Challenge. Speech2MaskTrack transcribes the spoken query and compiles it into structured constraints over category, count, direction, interaction role, and temporal phase. SAM3.1 enumerates multiple instance tracks, which TRACE ranks using complete-trajectory motion and relation evidence. A frozen lexical presence gate may suppress the ranked SAM3.1 base prediction. When the gate predicts that a target is present, an available full-expression-conditioned SaSaSa2VA track replaces the SAM3.1 mask. Only outputs that remain empty enter GPT-assisted recovery, which invokes SaSaSa2VA again under query- and mask-level verification. Speech2MaskTrack achieved second place in the official challenge ranking.
[CV-101] ReART: Reference-Guided Retrieval and Refinement for Emotion-Aware Art Generation
链接: https://arxiv.org/abs/2608.22329
作者: Qianqian Tang,Jiayi Gao,Ting Lei,Yang Liu
类目: Multimedia (cs.MM); Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by ACM Multimedia 2026 (Grand Challenge Track 1), 7 pages, 3 figures
Abstract:Emotion-aware artistic image generation requires a model to satisfy semantic content, artistic style, and target emotion simultaneously. The key challenge is that artistic captions conflate these axes into underspecified free-form text, making fine-grained visual attributes such as brushwork, composition, and tonal atmosphere difficult to ground concretely. We present ReART, a reference-guided retrieval and refinement framework. Our method decomposes test captions and each image annotation in the EmoArt database into structured visual fields, and performs field-wise retrieval over subject, layout, brush-line, and tone-mood dimensions to retrieve role-specific visual references that supply the perceptual detail text alone cannot convey; these references are used alongside a structured prompt for initial synthesis. For samples where any Attribute Alignment Score (AAS) axis falls below threshold, an AAS-driven refinement loop diagnoses failures, constructs constrained repair plans specifying elements to keep, errors to fix, and operations to avoid, routes references by correction purpose, and performs controlled editing under structural preservation constraints. Our system ranks 2nd in Track 1 of the AffectiveArt 2026 Grand Challenge, achieving a perfect AAS of 1.00 and an overall score of 0.78. Code is available at this https URL.
[CV-102] MedReaMM: Evaluating Large Multimodal Models on Expert-Level Clinical Diagnostic Synthesis
链接: https://arxiv.org/abs/2608.22323
作者: Lai Wei,Yuchao Chen,Zhenbiao Cao,Xiaojin Zhang,Zhongyu Wei,Bangting Wang,Wei Chen,Xiang Bai
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:The application of Large Language Models (LLMs) to diagnostic decision-making has garnered growing interest. However, existing benchmarks largely focus on textual reasoning or isolated visual question-answering (VQA) tasks, lacking holistic integration of clinical narratives and medical imaging, and thus failing to assess the multimodal diagnostic synthesis capability central to expert clinical judgment. To bridge this gap, we introduce MedReaMM, a benchmark specifically designed to evaluate models’ ability to synthesize heterogeneous clinical evidence consisting of detailed patient histories alongside multiple medical images into accurate differential diagnoses under a complete-information paradigm. Constructed from case reports sourced from top-tier medical journals and curated clinical case databases, MedReaMM comprises 625 expert-validated cases with an average of 2.79 medical images per case and a total of 1,042 standardized diagnoses annotated with ICD-11 codes. These cases predominantly represent rare, atypical, or multi-system presentations that demand expert-level evidence integration beyond routine pattern recognition. We evaluate 23 Large Multimodal Models (LMMs) and find that most achieve diagnostic accuracy scores below 50%, underscoring a substantial gap in multimodal diagnostic synthesis capability. Further analysis reveals that medical knowledge proficiency, medical image understanding, and evidence integration are all highly correlated with diagnostic performance.
[CV-103] Does a Modern-Handwriting Warm-Up Help Historical Arabic OCR? A Reproducible Compute-Matched Evaluation on Muharaf and KHATT
链接: https://arxiv.org/abs/2608.22316
作者: Sumaih Almarshad,Maram Alamri,Dona Aloraini,Fares Altuwaim,AlJawharh AlOtaibi,Reem Alyabis,Rayah Aldawsari
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注: 14 pages, Dal Research Team
Abstract:Whether an intermediate stage of modern Arabic handwriting helps or hurts historical Arabic HTR is usually decided from one implementation and one comparison, too thin a basis for a claim either way. We test stability by running the same nominal ablation four times, letting the base checkpoint, encoder-freezing strategy, epoch budget, precision, and learning-rate schedule vary as they naturally did during development, while holding the normalization, scorer, and interval estimation fixed. Each run compares intermediate training on modern handwriting (KHATT) then fine-tuning on historical manuscripts (Muharaf) against fine-tuning on Muharaf directly. Across the four runs the estimated effect swings from -17.64 to +14.52 CER points and reverses sign. The two extremes are exactly the two runs with an identifiable confound (a fivefold lower learning rate in one; a checkpoint of undisclosed provenance in the other); the two clean runs land at -0.25 and +0.94, i.e. no effect. A tight interval from one implementation says nothing about the next. We then run a compute-matched experiment with identical budgets over three seeds: KHATT warm-up is +2.42 CER points worse than a matched same-domain control (95% interval [+0.60, +4.25]); the part of that gap specific to the handwriting domain is only about 0.6 points a small negative effect under this configuration, not a universal result. We release a SaudiHeritage-OCR package with the normalizer, interval scorer, a verified KHATT decoder, experimental manifests, VLM baselines, and an edition-alignment protocol, so the result can be checked independently. The Al-Mahd inscription line is held strictly out and is not offered as a benchmark.
[CV-104] On the Choice of Tensor Estimation for Corner Detection Optical Flow and Denoising
链接: https://arxiv.org/abs/2608.22314
作者: Freddie Åström,Michael Felsberg
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Many image processing methods such as corner detection, optical flow and iterative enhancement make use of image tensors. Generally, these tensors are estimated using the structure tensor. In this work we show that the gradient energy tensor can be used as an alternative to the structure tensor in several cases. We apply the gradient energy tensor to common image problem applications such as corner detection, optical flow and image enhancement. Our experimental results suggest that the gradient energy tensor enables real-time tensor-based image enhancement using the graphical processing unit (GPU) and we obtain 40% increase of frame rate without loss of image quality.
[CV-105] Adapting Dense Vision-Language Relationships for Multi-label Classification with Partial Label
链接: https://arxiv.org/abs/2608.22313
作者: Cheng Chen,Yifan Zhao,Jia Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Learning multi-label image classification with incomplete annotations is a challenging task that has been widely studied for its superior trade-off between high efficiency and less labor consumption on large-scale datasets. Predominant methods rely on strong prior assumptions to recover the missing semantics from partial annotations. However, these statistic priors suffer from unstable semantic mistakes and thus lead to catastrophic overfitting. Toward this end, we propose a Language-driven Dense Semantic Adaptor (LDSA) that excavates prior-adaptive relationships from multimodal pretrained CLIP models. In our approach, the densely contrastive adaptor is first proposed to construct dense visual contrastive constraints, transferring the task-specific knowledge to visual domains. We then propose a language-driven interactive decoder with the help of class-specific prompt tuning, which adapts language proxies with visual domains. With the collaborative learning of proposed modules, experimental results demonstrate our proposed LDSA achieves a new state of the art on public multi-label classification benchmarks, and interpretable analyses reveal that our LDSA discovers implicit semantic relationships with the prior-adaptive learning scheme.
[CV-106] On Tensor-Based PDEs and their Corresponding Variational Formulations with Application to Color Image Denoising
链接: https://arxiv.org/abs/2608.22302
作者: Freddie Åström,George Baravdish,Michael Felsberg
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:The case when a partial differential equation (PDE) can be considered as an Euler-Lagrange (E-L) equation of an energy functional, consisting of a data term and a smoothness term is investigated. We show the necessary conditions for a PDE to be the E-L equation for a corresponding functional. This energy functional is applied to a color image denoising problem and it is shown that the method compares favorably to current state-of-the-art color image denoising techniques.
[CV-107] Self-Calibrating Dense Displacement Fields for Reliable Co-Registration of Large Optical Satellite Imagery
链接: https://arxiv.org/abs/2608.22300
作者: Shoukun Sun,Zhe Wang,Sanaz Salati,Jiyin Zhang,Hui Wang,Xiaogang Ma
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Co-registration underlies nearly every multi-temporal and multi-sensor use of optical satellite imagery, and operational products still carry documented offsets well above the fraction-of-a-pixel scale at which change detection, time series, and data fusion degrade. Real image pairs differ along several axes at once (sensor response, scene content, viewing geometry, resolution, mosaic seams), and the last of these is not a single global motion. Existing tools embed a motion model and constants tuned to their development data; a pair that fits is registered precisely, while one that does not either fails to match or returns a result wrong by tens of pixels with no failure reported. Learned matchers add a GPU requirement and carry no accuracy guarantee outside their training distribution. We present SCDF (self-calibrating displacement fields), a training-free, GPU-free estimator whose motion model is the dense per-pixel displacement field itself, so no scene motion falls outside the model. A single predict–measure–filter loop runs over a resolution pyramid: the accumulated field predicts where each patch of the moving image falls in the reference, RootSIFT matching and a correlation pass measure the displacement there to sub-pixel precision, and filters whose thresholds are all calibrated on the image pair itself decide what survives. One configuration, with no per-dataset tuning, processes full 8192^2 scenes on a single CPU core. On 584 constructed-ground-truth pairs built from real Sentinel-2, Landsat-8/9, and NAIP imagery, against seven classical baselines and two zero-shot pretrained matchers, SCDF registers every pair with zero failures, reduces the best baseline’s real-pair median end-point error from 6.83 to 4.17m, and cuts its 90th percentile from 17.8 to 7.77m.
[CV-108] argeted Iterative Filtering
链接: https://arxiv.org/abs/2608.22299
作者: Freddie Åström,Michael Felsberg,George Baravdish,Claes Lundström
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:The assessment of image denoising results depends on the respective application area, i.e. image compression, still-image acquisition, and medical images require entirely different behavior of the applied denoising method. In this paper we propose a novel, nonlinear diffusion scheme that is derived from a linear diffusion process in a value space determined by the application. We show that application-driven linear diffusion in the transformed space compares favorably with existing nonlinear diffusion techniques.
[CV-109] ONAV: Task-Oriented Navigation and Action-Velocity Chunk Learning for Articulated Object Quadrupedal Mobile Manipulation
链接: https://arxiv.org/abs/2608.22296
作者: Haoran Lin,Mingyu Yang,Pengfei Qi,Kehan Chen,Qiang Diao,Liangji Zeng,Wenrui Chen,Yaonan Wang,Kailun Yang
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: The project page is at this https URL
Abstract:Quadruped mobile manipulation requires two tightly coupled capabilities: reaching manipulation-ready configurations and maintaining stable contact throughout articulated-object interaction. However, existing methods often terminate navigation near the target, leaving a gap between reachability and manipulation readiness, while tracking lag, motion jitter, and contact instability limit continuous interaction. To address these challenges, we present TONAV, a unified framework integrating task-oriented navigation with action-velocity chunk learning. First, we introduce a position-velocity-coupled teleoperation framework that explicitly captures motion dynamics to improve master-follower consistency and collect smooth, temporally consistent demonstrations. Next, task-oriented navigation leverages vision-language reasoning to decompose high-level instructions into executable subgoals and adaptively refine the robot base toward a manipulation-ready configuration. Finally, action-velocity chunk learning jointly models joint positions and their temporal transitions under velocity supervision, enabling smooth and stable sustained-contact manipulation. Real-world experiments across diverse articulated-object tasks demonstrate that TONAV achieves higher success rates in both task-oriented navigation and complete mobile manipulation, mitigating the navigation-manipulation gap and improving continuous-contact interaction. The project page is at this https URL.
[CV-110] DECO: Depth-Guided Co-Visibility Reasoning for Low-Altitude UAV Visual Localization
链接: https://arxiv.org/abs/2608.22289
作者: Yibin Ye,Xichao Teng,Shuo Chen,Xiaokai Song,Dongdong Guan,Qifeng Yu,Zhang Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Unmanned aerial vehicles (UAVs) increasingly require robust visual localization in GNSS-denied environments. A common solution estimates UAV poses by matching keypoints between UAV images and geo-tagged orthographic reference maps derived from satellite or aerial imagery, followed by Perspective-(n)-Point (PnP) pose solving. However, such reference maps mainly record top-down surfaces such as roofs and ground planes, while vertical structures such as facades and walls are often compressed or missing. Consequently, many visually distinctive keypoints in low-altitude UAV images have no valid counterparts in the reference map, leading to redundant matches and inaccurate pose estimation. To address this issue, we propose DECO, a DEpth-guided CO-visibility reasoning framework for low-altitude UAV visual localization. DECO uses monocular depth priors to infer local surface geometry and estimate co-visible regions between UAV images and the reference map. Based on this prior, a Geometry-Saliency Coupled Co-visibility Score is introduced to jointly consider geometric co-visibility and detector saliency for keypoint ranking. In this way, DECO retains keypoints that are both visually distinctive and geometrically co-visible, improving feature matching and PnP-based pose estimation. Extensive experiments demonstrate that DECO achieves superior localization performance and can be integrated with different depth models, feature detectors, and matchers. The source code will be available at this https URL.
[CV-111] OVIBench: Benchmarking Online Video Question Answering under Interruption EMNLP2026
链接: https://arxiv.org/abs/2608.22279
作者: Naiming Liu,Zhiheng Wu,Shuning Wang,Tie Zhang,Bowen Liu,Tong Wang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: EMNLP 2026
Abstract:Recent vision language models (VLMs) have achieved strong progress in video understanding. However, most existing video QA research and benchmarks still follow an offline, single-round paradigm, overlooking realistic interactions where users may interrupt the model during answer generation. To address this gap, we formulate the task of Online Video Question Answering under Interruption and introduce OVIBench, the first standardized benchmark for evaluating VLMs in this setting. OVIBench categorizes interruptions into three types: Cancellation, False Trigger, Correction and supports both open-ended and multiple-choice evaluations. To enable large-scale and reproducible testing, we develop an offline simulation protocol that reproduces interruption during generation under a unified temporal setup, together with a multi-dimensional metric suite for assessing interruption understanding and response generation. Experiments demonstrate that OVIBench effectively distinguishes models’ interruption-handling abilities, especially in following correction requests. Finally, we construct a train set OVI-Train for interruption-aware fine-tuning. Models fine-tuned on this dataset achieve significant gains on OVIBench, validating the effectiveness of our benchmark and data design. OVIBench, OVI-Train, and the evaluation code will be released.
[CV-112] GAN-Diff : Coupling Pretrained WGAN-GP Features with Conditional Diffusion U-Nets
链接: https://arxiv.org/abs/2608.22272
作者: Saif Ahmed,Ashadulla Hil Galib,S.M. Riaz Rahman Antu,Ahmed Faizul Haque Dhrubo,Souvik Pramanik,Mohammad Abdul Qayum,Mohsin Sajjad,Mohammad Ashrafuzzaman Khan
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 7 pages, 10 figures, 4 tables
Abstract:Generative adversarial networks (GANs) can provide efficient image generation, while diffusion models offer high-quality image restoration but require iterative sampling. This paper presents a hybrid GAN-guided diffusion framework that uses a pretrained Wasserstein GAN with gradient penalty (WGAN-GP) as a feature prior for conditional diffusion-based image restoration. Intermediate features from the frozen WGAN-GP generator are incorporated into a diffusion U-Net through cross-attention and remain fixed during the DDIM sampling process. The framework is evaluated on two restoration tasks, Gaussian denoising and 2Xsuper-resolution, using CelebA face images. During development, several sources of instability were identified and addressed, including adversarial learning-rate imbalance, inappropriate diffusion initialization, excessive corruption, and insufficient parameter averaging. The resulting framework consistently improves the quality of both degraded and low-resolution images. In particular, it improves denoising performance by 4.40 dB in PSNR and super-resolution performance by 3.70 dB over their respective input baselines. These results demonstrate the potential of a frozen GAN feature prior to guide diffusion models toward stable and effective image restoration.
[CV-113] raining-Free VLM Personalization via Calibrated Residual Decoding
链接: https://arxiv.org/abs/2608.22263
作者: Jiaao Yu,Yujian Ma,Xianming Hu,Pengran Wang,Ang Li
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Vision-language models can be personalized in a training-free manner by directly providing user profiles, preferences, or visual references at inference time, without updating model parameters. However, direct personalized prompting does not guarantee that the model will reliably exploit such evidence. The predictive distribution under the positive user profile often mixes two sources: personalized signals genuinely supported by the current profile, and the model’s generic visual or linguistic priors. As a result, from the positive-profile response alone, it is difficult to determine whether a high-confidence answer is supported by the user profile or merely reflects the model’s default preference. To address this problem, we propose a training-free calibrated residual decoding framework. Given the same image and question, we construct three evidence conditions: a positive profile , a counterfactual profile , and an empty profile . Our method keeps the prediction under as the anchored base, and explicitly estimates the marginal contribution of personalization from score differences across the three conditions. We further introduce normalized-entropy-based uncertainty calibration, allowing the strength of personalized enhancement to adapt to the reliability of the residual signal. Experiments on MMPB, YoLLaVA, and MyVLM show that the proposed method improves personalized multimodal understanding without fine-tuning, with consistent gains on identity-sensitive visual personalization tasks. Additional analysis shows that entropy calibration stabilizes residual decoding when the contrastive personalization signal is uncertain. Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) Cite as: arXiv:2608.22263 [cs.CV] (or arXiv:2608.22263v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2608.22263 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-114] Hyper2: Unleashing Hyperbolic Geometrys Full Potential via Dual-Space Consistency BMVC2026
链接: https://arxiv.org/abs/2608.22238
作者: Guantian Zheng,Haiyang Xu,Tianyu Gao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 18 pages, 5 figures, 5 tables. Accepted to BMVC 2026
Abstract:HyperbolicCD pioneered hyperbolic geometry for point cloud completion by replacing the Euclidean Chamfer distance with arcosh(1+alpha||x-y||^2), but the reported gains are modest (3-7% Chamfer reduction across SeedFormer, PointAttN and PMP-Net backbones on PCN and ShapeNet-55). We argue the bottleneck lies elsewhere: the loss is hyperbolic but the encoder it back-propagates through is Euclidean, so the position-dependent supervision of the loss is averaged away by the chain rule before it reaches the parameters. We call this a cross-geometry mismatch, and make it testable through two model-agnostic indicators, feature-loss correlation r_FL and effective gradient utilisation u_G. On an SVDFormer backbone trained with HyperbolicCD’s loss alone we measure (r_FL, u_G) = (0.68, 39%). We propose Hyper^2, a dual-space consistency framework that extends HyperbolicCD by reusing the identical arcosh(1+alpha d^2) functional form as a positional bias on the refinement attention (a hyperbolic distance encoding), paired with HyperbolicCD’s hyperbolic Chamfer loss under a single shared curvature alpha. Both operators are O(N log N) scalar non-linearities on Euclidean distances and together add only ~1.6% FLOPs over SVDFormer. Hyper^2 delivers -22.9% Chamfer on ShapeNet-55 over SVDFormer (well above the 13.2% linear sum of the -12.0% loss-only and -1.2% encoding-only single-space ablations) and -37.5% on the 21 unseen ShapeNet-34 categories. The two indicators remain essentially flat for any single-space configuration but jump together to (0.95, 87%) only when both encoder and loss are hyperbolic, supporting the claim that geometric consistency across encoder and loss, rather than either operator alone, is what enables hyperbolic supervision in point cloud completion. Code is available at this https URL.
[CV-115] When Test-Time Adaptation Helps Harms or Becomes Inactive: A Condition-Level Study on CIFAR-10-C
链接: https://arxiv.org/abs/2608.22233
作者: Sreeja Guha Majumdar,Aratrika Saha
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Test-time adaptation (TTA) aims to improve model robustness under distribution shift by adapting a source model using unlabeled test data. Although methods such as TENT and EATA have demonstrated gains on corrupted data, aggregate accuracy can obscure the conditions under which adaptation fails or provides little benefit. We present a controlled comparison of three TTA strategies—BatchNorm-statistics adaptation (BN-Adapt), entropy-minimization adaptation (TENT), and reliability-filtered adaptation (a scoped re-implementation of EATA)—against an unadapted source model on the full CIFAR-10-C benchmark, covering 15 corruption types and 5 severity levels. All three methods improve mean accuracy over the source model by 12.2–13.3 percentage points (Wilcoxon signed-rank p 10^-12 ). However, each method underperforms the source model on 8.0–9.3% of conditions, with failures concentrated in low-severity corruptions where the source model already performs near ceiling, particularly brightness, fog, contrast, and defocus blur. We further find that EATA closely tracks the gradient-free BN-Adapt baseline, with a mean absolute difference of 0.09 percentage points, compared with 1.08 percentage points relative to TENT. This suggests that reliability filtering can substantially restrict effective adaptation, causing EATA to behave more like a BatchNorm-statistics baseline than an entropy-minimization method. These results show that aggregate accuracy alone can mask systematic TTA failure modes and motivate condition-level evaluation of when adaptation helps, harms, or becomes effectively inactive.
[CV-116] UR2-MLLM : Uncertainty-aware Revisit Reasoning in Multimodal Large Language Models for Radiology Report Generation EMNLP2026
链接: https://arxiv.org/abs/2608.22217
作者: Yucheng Chen,Yang Yu,Jiazhou Zhou,Yufei Shi,Yongying Lan,Yichi Zhang,Liyi Li,Si Yong Yeo
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: EMNLP 2026 Findings
Abstract:Radiologists generate diagnostic reports through iterative and selective revisiting of suspicious regions to refine their interpretations. Recent multimodal large language models (MLLMs) for radiology report generation (RRG) have shifted from text-only reasoning toward a ``Thinking-with-Images’’ paradigm, incorporating visual evidence into the reasoning process. However, existing methods provide static visual evidence without a dynamic revisit mechanism during reasoning, neglecting how radiologists re-examine uncertain observations. To this end, we propose an Uncertainty-aware Revisit Reasoning MLLM (UR ^2 -MLLM) framework that dynamically revisits uncertain regions during reasoning for RRG. UR ^2 -MLLM is first equipped with uncertainty perception by training on an uncertainty-aware dataset. We then construct a multimodal reasoning trajectory dataset together with a detect-and-copy mechanism, which guides when and where to revisit. Finally, a visual grounding reward refines this behavior through reinforcement learning, aligning the revisited regions with corresponding anatomical structures. Experiments on MIMIC-CXR and IU-Xray show that UR ^2 -MLLM achieves state-of-the-art performance, highlighting the value of uncertainty-aware visual revisit reasoning for reliable and clinically aligned report generation.
[CV-117] SAM3Dual: A 3rd Place Solution to the MOSEv2 Track 8th LSVOS Challenge ECCV2026
链接: https://arxiv.org/abs/2608.22193
作者: JeongRae Kim,Chaehyun Kim,Changwon Lim
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 3rd place solution to the MOSEv2 Track of the 8th LSVOS Challenge at ECCV 2026
Abstract:We present SAM3Dual, our third-place solution to the MOSEv2 track of the 8th Large-scale Video Object Segmentation (LSVOS) Challenge at ECCV 2026. SAM3Dual is a training-free inference extension of pretrained SAM 3 that explicitly separates temporal memory into a short-term branch for recent observations and a long-term branch for interval-sampled historical representations. The two memory responses are combined using a deterministic sequence-relative fusion schedule and conservatively modulated by the previous-frame object confidence. All pretrained SAM 3 parameters remain frozen, requiring no task-specific training, fine-tuning, test-time training, or online parameter optimization. The complete system achieved an official JF score of 64.37 and ranked third in the MOSEv2 track. This result highlights the potential of reorganizing temporal memory entirely at inference time to obtain competitive long-term VOS performance while preserving the pretrained model.
[CV-118] BehaviorWorldGen: Closing the Loop between Action Models and World Simulators via Controllable Behavior-Aware Structured World Generation
链接: https://arxiv.org/abs/2608.22187
作者: Jiaqi Wang,Zhuo Zhang,Haining Guan,Tingguang Zhou,Haowen Cui,Zhongyang Zhu,Yulong Zheng,ChuanYe Wang,Xuefeng Chen,Zhen Yang,Tianchen Deng,Feiyang Tan,Hangning Zhou,Bo Dai,Lixia Shen,Xiwu Chen,Xiyang Wang,Jiajun Zhu
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Modern driving action models are increasingly improved in a self-improvement loop, where a learned world simulator imagines future observations and the resulting data is fed back to refine the action model. However, the bottleneck of this loop lies in the simulators’ inability to generate behaviorally plausible responses by surrounding agents, making generated data both unrealistic in interaction and imbalanced in distribution. We introduce BehaviorWorldGen, a framework that closes the loop between action models and world simulators through controllable behavior-aware structured world generation. Its core component is BehaviorFlow, a meta-action-conditioned traffic-flow model that injects interpretable behavior controls and jointly generates multi-agent rollouts. BehaviorFlow realizes the specified agent behaviors while allowing surrounding vehicles to respond to the ego and to one another. The resulting rollouts are rendered by a world simulator into realistic multi-view observations, which are paired with corrected interaction-aware trajectories for action-model refinement. Since BehaviorWorldGen uses structured trajectories as the interface between its modules, it is compatible with diverse action models and world simulators. Experiments on world generation, scene extrapolation, and policy refinement demonstrate consistent improvements, with the largest benefits concentrated on difficult interactive scenarios.
[CV-119] When Does Visual Generation Help Visual Understanding in Unified Multimodal Models?
链接: https://arxiv.org/abs/2608.22174
作者: Yubo Zhu,Zhehan Kan,Jingyi Yang,Miaolin Chen,Jinbo Xing,Kai Zhu,Zijian Wang,Sheng Zhong,Wei Tong
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Unified multimodal models (UMMs) can perform both understanding and generation, raising a central question: can visual generation improve understanding? Existing evaluations provide mixed evidence, but confound task difficulty, reasoning paradigms, and the closed-loop interaction between generation and understanding. We introduce VGAU-Diag, a fine-grained evaluation framework for vision generation-assisted understanding. It stratifies samples by difficulty, enables unified evaluation of multiple reasoning paradigms, and uses Oracle-Assisted Reference Protocols. Our analysis shows that generated visual aids help on easier instances but become unreliable as reasoning complexity increases. Oracle-assisted diagnosis further reveals that the main bottleneck often lies on the visual-understanding side rather than the visual-generation side, as current UMMs struggle to leverage even faithful visual aids. We also show that effective visual generation should target visual-understanding bottlenecks rather than add more reasoning steps, and identify a three-stage transition from task-irrelevant noise, to misleading plausible guidance, and finally to useful assistance. These findings would be useful to guide the development of better this http URL code is available at this https URL.
[CV-120] RACE: Artifact-Robust Statistical Shape Modeling from Imperfect Surface Scans - A Case Study in Craniosynostosis 3D Photography MICCAI2026
链接: https://arxiv.org/abs/2608.22131
作者: Sanjay Bhandari,Nawazish Khan,Alzbeta Novotna,Tiffany Jeong,Loretta Bowman,Michael Hernandez,Tobi Somorin,Viraj Govani,Jesse Goldstein,Shireen Elhabian
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accepted at ShapeMI workshop at MICCAI 2026
Abstract:Craniosynostosis severity analysis increasingly relies on statistical shape models (SSMs) to quantify cranial morphology, but most existing workflows depend on computed tomography or heavily curated three-dimensional (3D) photographs. Raw clinical 3D photographs provide a radiation-free and repeatable alternative, yet often contain shoulders, hands, hair, clothing, scanner noise, and incomplete boundaries that corrupt correspondences. We introduce the Template-constrained Robust Artifact-aware Correspondence Estimation (TRACE) framework, an unsupervised method for constructing SSMs directly from artifact-contaminated clinical 3D head photographs. TRACE predicts sparse anatomically corresponding head-surface control points from the raw point cloud, refines them through a coarse-to-fine Surface-Aware Deformation cascade, and uses thin-plate spline warping to deform a clean template mesh into a subject-specific head reconstruction. This template-constrained formulation keeps dense correspondences on clinically relevant head anatomy while suppressing non-head artifacts. The correspondence module is decoupled from the point-cloud encoder, enabling the same deformation pipeline to be paired with different backbones, including PointNet, DGCNN, and Point Transformer V3. Across all backbones, TRACE substantially improves surface sampling, topology preservation, and shape-model quality over prior SSM methods, providing a scalable foundation for photograph-based craniosynostosis shape analysis and a framework that may extend to other artifact-contaminated surface scans when an appropriate clean template is available.
[CV-121] Vehicle speed dataset for the major European road network derived from Sentinel-2 imagery 2022-2026
链接: https://arxiv.org/abs/2608.22116
作者: Maciej Adamiak,Sascha Fendrich,Julian Psotta,Alexander Zipf
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:The dataset provides individual vehicle speed observations on European E-roads: motorways, trunk roads, primary and secondary roads, as tagged in OpenStreetMap as e-road, for the years 2022-2026. Speeds are derived from Copernicus Sentinel-2 Level-2A satellite optical imagery using a processing pipeline that exploits the short, well-characterized acquisition delays between the blue (B02_10m), green (B03_10m), and red bands (B04_10m) of the Sentinel-2 push-broom instrument. A moving vehicle appears at slightly displaced positions in the three bands, forming a moving echo. The detected displaced intensity peaks are linked into per-vehicle trajectories through a prediction-and-matching procedure. The resulting displacements are converted into ground speeds using publicly accessible inter-band time delays. Each record contains the trajectory geometry, per-channel displacements and headings, internal quality indicators, the estimated speed, the acquisition timestamp, and the source Sentinel-2 product identifier. The dataset is distributed as GeoPackage files, with one record per detected vehicle, and can support studies of traffic patterns, speed behavior, transport modeling, and the calibration of road network attributes at a continental scale.
[CV-122] Learning Implicit Constitutive Laws for Dynamic 3D Gaussian Splatting from Monocular Videos
链接: https://arxiv.org/abs/2608.22102
作者: Xiaoyang Liu,Kai Han
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:We present GCA (Gaussian Constitutive Alignment), a framework for learning implicit constitutive laws from monocular dynamic video of deformable objects represented by 3D Gaussians. Given a static multi-view scan for geometric initialization, our method learns intrinsic physical dynamics solely from a single fixed-viewpoint video of the moving object. Existing implicit methods often suffer from local minima under noisy supervision and lack physical interpretability, while explicit approaches rely on predefined constitutive equations, limiting generalizability and becoming unstable in monocular settings. To address these challenges, our framework unifies LoRA-based adaptation with two key alignment modules. First, we propose Rank-based Depth-Geometric Anchors (RDGA) to establish robust geometric constraints from monocular dynamic observations via scale-invariant rank-based depth alignment, reducing the reliance on unreliable pixel-level color supervision. Second, a Constitutive Prior Regularizer (CPR) integrates classical constitutive models as soft differentiable priors, regularizing the optimization while preserving the flexibility of implicit modeling—even when the actual material is absent from the hypotheses. Extensive experiments on synthetic, real-to-sim, and real-world datasets demonstrate that GCA outperforms existing methods, achieving 48% lower Chamfer Distance than the strongest baseline on synthetic benchmarks while remaining robust under monocular supervision.
[CV-123] hree-Phase Scribble-Adaptive Curriculum Learning for autoPETV Grand Challenge
链接: https://arxiv.org/abs/2608.22096
作者: Libo Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:This report describes Libo Zhang’s algorithmic solution to autoPETV Grand Challenge on interactive lesion segmentation in whole-body PET/CT. Interaction is encoded as two additional input channels that rasterize the accumulated foreground and background scribbles, and a residual-encoder U-Net of about 140 million parameters is trained with a three-phase curriculum over 4000 epochs: the network first learns fully automatic segmentation with silent interaction channels, then observes ground-truth-derived scribbles under randomly sampled visibility modes, and finally adapts to its own mistakes through online simulation of up to five error-driven correction steps. Training draws on 1811 autoPET and DeepPSMA studies, and the submission ensembles the best and final checkpoints of five folds by logit averaging. In interactive five-fold cross-validation with six interaction steps, the final checkpoints reach a mean AUC-Dice of 3.836 and a mean AUC-DMM of 3.869, improving monotonically in every fold, with roughly half of the total gain delivered by the first corrective scribble. Our code and trained model checkpoints are available on this https URL.
[CV-124] When More References Hurt: Contamination-Aware DINOv2 Memory Banks for Few-Shot Steel Defect Detection
链接: https://arxiv.org/abs/2608.22082
作者: Hannaneh Kalantari,Javad Khoramdel
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Patch-memory anomaly detectors assume that their reference bank is normal, an assumption that is difficult to guarantee when additional industrial images are unverified. We study whether a few trusted normal images can safely recover useful normal patches from such references without defect masks. Starting from the DINOv2 patch-memory formulation used by AnomalyDINO, we score candidate patches by distance to a clean seed bank, discard the most suspicious 20%, merge the retained patches with the seed, and enforce a fixed budget by greedy coreset selection. On Severstal, naive additional references contain 9.46% anomalous patches; the proposed trim rejects 78.1% of them and reduces residual contamination to 2.59%. At an equal 51,200-patch development budget, the proposed bank reaches 0.1084 AUPRC versus 0.0950 for naive expansion, 0.0952 for random removal, and 0.1030 for eight clean images. Injecting only 0.5% anomalous patches into a clean bank reduces AUPRC from 0.1030 to 0.0759. On all five completed held-out pairs, the proposed bank improves over naive expansion, with a mean gain of 0.0142 AUPRC. Reference purity is therefore a first-order design variable, and unverified images are useful only when their contribution is filtered explicitly.
[CV-125] Spiking Neural Networks for Energy-Efficient Object Detection in Forward-Looking Sonar Imagery
链接: https://arxiv.org/abs/2608.22072
作者: Gwenevere Frank,Gert Cauwenberghs
类目: Computer Vision and Pattern Recognition (cs.CV); Signal Processing (eess.SP)
备注:
Abstract:Autonomous underwater vehicles (AUVs) are increasingly important tools in industries ranging from research, to energy, to defense. AUVs are power-constrained platforms operating in remote environments with fixed battery capacities, where propulsion competes with compute and sensors for power over lengthy mission durations. AUVs frequently operate in dark or turbid waters where optical sensing is of limited value, and rely on sonar as their primary sensing modality. Convolutional neural networks (CNNs) are the state-of-the-art solution for object detection in forward-looking sonar imagery, but are energy expensive (e.g. YOLOv8m: 322 mJ/inference). Spiking neural networks (SNNs) rely on binary spike activations and thus sparse accumulate-only operations, allowing them to be remarkably energy efficient, particularly when paired with dedicated neuromorphic hardware. The sparse, high-contrast structure of forward-looking sonar (FLS) returns is structurally matched to spike coding in a way that optical imagery is not. No prior work has assessed the suitability of SNNs for object detection in FLS imagery. SpikeYOLO, a fully spiking network trained with surrogate gradients, was benchmarked against state-of-the-art CNN baselines on three FLS object detection datasets. Key results: SpikeYOLO T=2 achieves 3.3 \times lower theoretical compute energy on UATD (97 vs 322 mJ) at competitive accuracy (0.529 mAP@0.5:0.95 vs. YOLOv8m’s 0.575); SpikeYOLO matches YOLOv8m on mAP@0.5 and outperforms YOLO-SONAR and Fast R-CNN baselines on the sparse Marine-Debris-FLS dataset at 4.4 \times lower energy; SpikeYOLO demonstrates superior robustness to multiplicative speckle noise (3.0% degradation at \sigma=0.4 vs. 8.9% for YOLOv8m), outperforming YOLOv8m outright at \sigma=0.6 , directly relevant to real-world FLS deployment.
[CV-126] Inferring Action from Future Latent State for Robotic Manipulation
链接: https://arxiv.org/abs/2608.22067
作者: Fenghao Lei,Zhixiong Huang,Long Yang,Jiabao Chen,Jie Cheng,Peilin Huang,Han Fu,Zhuo Li,Xiaoxue Ren
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:World-Action Models (WAMs) build robot control on video-generation backbones, which jointly predict dense future visual trajectories and robot actions. We argue that video generation is an unnecessary intermediate objective for world-action modeling. For robotic manipulation, the goal of a world model is not to reproduce how the world looks at every intermediate moment, but to predict the state that the world will reach after an action is executed. The intermediate frames only describe the visual transition between physical states, which consumes substantial model capacity and computation, but do not directly specify the physical outcome that the robot action is intended to produce. In this paper, we propose DELE-w0.5, which infers robot actions from predicted future states without relying on video generation. Concretely, DELE-w0.5 infers the action sequence from its corresponding compact future latent state. The future latent state captures the action-relevant physical outcome of robot interaction and serves as an explicit bridge between world modeling and action generation. The core design principle of DELE-w0.5 is to model how the physical world changes under robot actions, rather than how its visual appearance evolves frame by frame. This formulation removes the high-dimensional visual redundancy introduced by dense video representations, and it therefore enables cheaper training and low-latency inference. Across 480 real-robot trials on four long-horizon manipulation tasks, our DELE-w0.5 achieves the best performance among all compared policies, attaining 62.5 overall full-task success and 81.3 macro ordered-stage progress, outperforming the strongest baseline by 47.5 and 30.7 percentage points, respectively.
[CV-127] ADMIL: Attention-Distilled Multiple Instance Learning for Selective Foundation Model Inference in Pathology
链接: https://arxiv.org/abs/2608.22066
作者: Duncan Stothers,Ren-Chin Wu,William Lotter
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Attention-based multiple instance learning (ABMIL) using pathology foundation model embeddings is effective for slide-level tasks, but exhaustive inference requires applying a large image encoder to every foreground tile despite the subsequent attention distribution often concentrating over a small subset of informative regions. We introduce ADMIL (Attention-Distilled Multiple Instance Learning), a selective-compute framework that distills an ABMIL teacher’s attention into a lightweight tile-selection model, PriorNet. Using an EfficientNet architecture, PriorNet learns the teacher attention distribution from raw tile pixels with KL divergence; at inference, it scores the foreground pool, selects the top-K tiles, and invokes the expensive foundation model only on that subset before a selected-bag ABMIL student predicts the slide label. Across BRACS, PANDA, and CAMELYON16, ADMIL matches full-teacher headline performance at K=4, 8, and 128 tiles, respectively, avoiding 98% of foundation model (Virchow2) tile embeddings and model inference FLOPs. Random and teacher-attention oracle controls show that this result depends on task-relevant selection rather than tile-count reduction alone. Quantitative and qualitative analyses suggest that PriorNet recovers the teacher’s tile ordering with high fidelity while focusing on task-relevant morphological regions. ADMIL shows that nearly all expensive tile encodings can be removed without sacrificing slide-level performance, providing a potential path for more efficient deployment in clinical settings where latency and compute costs are key considerations.
[CV-128] Competitive Memory Readout for Robust Video Object Segmentation: 2nd Place Technical Report for the MOSEv2 Track of the 8th LSVOS Challenge
链接: https://arxiv.org/abs/2608.22064
作者: Mingqi Gao,Sijie Li,Jungong Han
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:We present our solution for the MOSEv2 track of the 8th Large-scale Video Object Segmentation (LSVOS) Challenge at ECCV 2026. The challenge evaluates robust video object segmentation under complex temporal dynamics, including long-term occlusion, disappearance and reappearance, large appearance changes, and strong interference from visually similar objects. Our method builds on SAM~3 and focuses on its memory readout. Standard target-only memory retrieval can confuse the annotated target with same-class non-target objects because such distractors are represented only implicitly as background. Our method introduces Competitive Memory Readout, which explicitly incorporates same-class competitor evidence when retrieving target information from memory. To prevent excessive suppression of weak or reappearing targets, we further apply a lightweight adaptive restoration rule after competition. The resulting system retains the original SAM~3 tracking pipeline while improving target identity preservation in challenging videos. Our submission achieves 66.20 on the primary challenge score and ranks 2nd in the MOSEv2 track.
[CV-129] Robust Global Structure-from-Motion via View Graph Pruning
链接: https://arxiv.org/abs/2608.22054
作者: Jiamin Xu,Lixing Yao,Weichen Dai,Renshu Gu,Zunjie Zhu,Weiwei Xu,Gang Xu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Structure-from-Motion (SfM) aims to estimate camera poses and reconstruct 3D structures from a collection of unordered images. Compared with incremental SfM, global SfM achieves better scalability by jointly estimating camera poses based on a view graph constructed from pairwise correspondences. However, its performance is highly sensitive to erroneous edges caused by visually ambiguous matches, which may lead to incorrect camera registration and reconstruction artifacts. In this work, we propose a subgraph-guided view graph pruning framework for robust global SfM. Our key idea is to exploit the internal consistency of reliable subgraphs to identify and remove unreliable connections. Specifically, we first partition the view graph into locally consistent subgraphs and perform global SfM within each subgraph to obtain reliable camera poses. We then apply RANSAC-based edge pruning across subgraphs to remove inconsistent edges, and finally perform global SfM on the refined view graph. Extensive experiments on ambiguous, sequential, and unordered image datasets demonstrate that our method improves the robustness of global SfM under challenging conditions. Further evaluation with neural rendering shows that the improved camera estimation leads to higher-quality novel view synthesis results.
[CV-130] ORBIT: Benchmarking SfM in the Wild with 360° Video CVPR2026
链接: https://arxiv.org/abs/2608.22039
作者: Sara Sabour,Linyi Jin,Richard Tucker,Amir Hertz,Marcus Brubaker,Saurabh Saxena,Junhwa Hur,Andrea Tagliasacchi,Deqing Sun,David J. Fleet,Richard Szeliski,Noah Snavely
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: A revision was Accepted at CVPR 2026
Abstract:Structure-from-Motion (SfM) is a cornerstone of 3D perception, yet current methods often fail when applied to complex videos involving challenging camera motions or dynamic scenes. Compounding the problem, the field lacks reliable ground-truth benchmarks for such difficult scenarios, making it hard to gauge real-world progress or to pinpoint where improvements are most needed. To address this gap, we introduce a new benchmark for evaluating camera pose estimation. Our key insight is to leverage online panoramic 360° video as a source of data from which to construct challenging clips, while still enabling robust ground-truth trajectory recovery. The panoramic nature of these videos provides richer visual context for tracking camera motion, even when parts of the view are affected by blur, motion, or dynamic objects. After tracking camera motion across full 360° videos, we crop and reproject selected portions to generate perspective-view clips that serve as our benchmark, called ORBIT. Experiments show that COLMAP, as well as recent optimization-based and feed-forward SfM methods struggle to accurately estimate camera poses on our benchmark. Hence, ORBIT provides a valuable testbed where researchers can meaningfully measure progress on truly challenging, real-world SfM problems.
[CV-131] AirAlign: Geometry-Aware Relative Pose Alignment for UAV Last-Meter Navigation
链接: https://arxiv.org/abs/2608.21926
作者: Jinyi Zhou,Shuo Feng,Yufei Wu,Piji Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Unmanned aerial vehicle (UAV) navigation in modern low-altitude environments requires more accurate pose alignment in the final approach stage for target information acquisition or manipulation, making “last-meter” navigation increasingly important. However, severe viewpoint and appearance variations make this task challenging. To tackle this problem, we propose AirAlign, a framework for RGB-only image-pair relative pose alignment for UAVs. AirAlign uses a pretrained visual geometry reconstruction model as the backbone to extract geometry-aware features from source-target image pairs. In addition, to better utilize the limited training data, we split the training set into multiple scene-disjoint folds for unseen cross-validation and model selection. During inference, the predictions of the selected models are averaged to form the ensemble output of the overall framework. Experiments on the PairUAV challenge at the ACMMM 2026 Workshop on UAVs in Multimedia demonstrate the effectiveness and robustness of our method, while comprehensive ablation studies validate the contribution of each component.
[CV-132] Entity-Constrained CBCT Retrieval for Low-Resource Dental Record Completion MICCAI2026
链接: https://arxiv.org/abs/2608.21913
作者: Nhi Ngoc-Yen Nguyen,Thai Nguyen,Kiet Huynh Cao Tuan,Huy-Hieu Pham
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at the ODIN 2026 Workshop, MICCAI 2026
Abstract:Completing dental records from cone-beam computed tomography (CBCT) is difficult when annotation is scarce and individual clinical fields are supported by different types of evidence. MMDental Task 3 requires seven-field record completion from only 50 labeled CBCT cases and scores the correctness of structured FDI positions and ICD codes; consequently, a visually plausible retrieved record can still be harmful when it introduces an unsupported entity. We propose Entity-Constrained CBCT-Guided Retrieval (ECCR), a parameter-free framework that separates evidence availability from evidence authority. A corpus-derived prior first supplies the complete record. A frozen 3D encoder retrieves image-conditioned Diagnosis evidence, which is appended only if it does not expand the prior FDI or ICD entity set, so the asserted entity set is invariant by construction. On public validation, ECCR reaches a weighted score of 0.3134, improving on both full-record multimodal retrieval (0.2237) and a static text-only prior (0.2915); the guard blocks 63.3% of retrieved candidates, each of which would otherwise have injected an FDI position or ICD code absent from the prior. On the final test evaluation, ECCR obtains 11.37 of a 97.4-point attainable maximum, securing second place overall. The result indicates that, in an extreme low-resource setting, controlling what multimodal evidence is allowed to modify can be more reliable than transferring an entire retrieved record.
[CV-133] A Scalable Vector Graphics Latent Space ECCV2026
链接: https://arxiv.org/abs/2608.21893
作者: Leonardo Zini,Elia Frigieri,Lorenzo Baraldi
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accepted at The 19th European Conference on Computer Vision – ECCV 2026
Abstract:Scalable Vector Graphics are a fundamental medium for resolution-independent visual content, yet the deep learning community lacks a continuous, dense, and invertible latent space for vector representations, the kind of foundational building block that Variational Autoencoders and their descendants have long provided for raster images. We introduce SLS (SVG Latent Space), a Transformer-based autoencoder that learns compact dense representations of individual SVG paths, the atomic visual elements from which any SVG image can be composed. By modeling SVG commands, coordinate data, and visual properties within a unified BPE-based token vocabulary, SLS learns fixed-size latent representations that jointly capture structure and appearance, and can be decoded back into valid, style-consistent SVG paths with high fidelity. The resulting embedding space is robust, invertible, and structured: embeddings lie on a unit hypersphere, enabling efficient similarity search, composition, and downstream conditioning through simple vector-space operations. Finally, we demonstrate that SLS generalizes across diverse tasks reducing their FLOPs by over 150 times compared to token-based approaches, and establishing a general-purpose latent foundation for vector graphics research.
[CV-134] Pixel-Space Diffusion via Observation Operators
链接: https://arxiv.org/abs/2608.21885
作者: Shaojie Guo,Lichen Ma,Haoyang Tong,Yu He,Zipeng Guo,Xiaoan Liu,Feng Yan,Yu Guo,Fei Wang,Junshi Huang,Yan Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Pixel-space diffusion models directly model image distributions but remain difficult to optimize. Recent methods alleviate this challenge through target reparameterization, while still relying on a fixed clean-image target throughout denoising. Through empirical analysis, we identify a scale-time mismatch: image structures become predictable from coarse to fine as noise decreases, whereas existing models are forced to predict the full image even under high noise, resulting in low-SNR gradients that hinder optimization. To resolve this mismatch, we propose Observation Operator Diffusion, a unified framework that aligns both the supervision trajectory and feature refinement with the intrinsic recovery order of image structures. Specifically, we replace fixed full-image supervision along the standard flow path with a time-indexed observation trajectory that evolves from coarse structures to the full image during denoising. This trajectory is instantiated with a family of Gaussian-Lanczos operators at varying observation scales, yielding a path-consistent training objective. We further introduce GL-CoDA, a decoder that injects scale-specific Gaussian-Lanczos observations across decoding stages for coarse-to-fine feature refinement. Extensive experiments show that the proposed approach converges substantially faster while consistently improving generation quality, achieving an FID of 1.52 on ImageNet-256.
[CV-135] VIG: Visual Information Gain as a Reward Signal for Multimodal Chain-of-Thought Compression EMNLP2026
链接: https://arxiv.org/abs/2608.21883
作者: Wen Luo,Xiaohan Yi,Xiaotao Huang,Liqun Huang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by EMNLP 2026 Findings
Abstract:Multimodal large reasoning models often rely on long Chain-of-Thought (CoT) traces in which a substantial fraction of tokens, such as repeated visual descriptions, self-reflection, and other visually-disengaged filler, inflate inference cost without contributing to the answer. Existing CoT compression methods optimize output length but never measure whether a reasoning token is actually grounded in the image. We propose \textbfVIG (Visual Information Gain), an information-theoretic GRPO reward that scores each reasoning token by how much the image reduces its predictive uncertainty. VIG is computed online from two forward passes of the same policy, one with and one without the image, so no reference chains, external annotations, or auxiliary reward models are needed. Across six main multimodal reasoning benchmarks and three Qwen3-VL-Thinking model sizes (2B/4B/8B), plus an additional R1-Onevision-Bench evaluation on 8B, VIG consistently improves the accuracy–efficiency trade-off, supporting our central claim: \emphefficient multimodal reasoning emerges from raising visual information density, where every reasoning token earns its place by anchoring to the image, rather than from imposing a length budget. Our source code is available at this https URL.
[CV-136] Region-Weighted Losses and Model Fusion for Cross-Modal PET Attenuation Correction MICCAI26 WWW
链接: https://arxiv.org/abs/2608.21881
作者: Khoa Tuan Nguyen,Joris Vankerschaver,Wesley De Neve
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: ntkhoa team submission for BIC-MAC MICCAI26 challenge ( this https URL )
Abstract:We describe our approach to the Big Cross-Modal Attenuation Correction (BIC-MAC) challenge, which asks for a pseudo-CT in Hounsfield Units to be synthesized from Non-Attenuation-Corrected PET (NAC-PET), DIXON MRI and a topogram, and scores both the pseudo-CT and the Attenuation-Corrected PET (AC-PET) reconstructed from it. Three ideas carried our improvements over the organizers’ 3D U-Net baseline. The loss matters more than the architecture: we compute the L_1 error in the Carney attenuation-coefficient ( \mu ) space that the CT metric itself uses, weighted by anatomical region. Only once that loss was in place did the unregistered DIXON MRI work as extra input channels. A fixed convex combination of two independently trained models then beat both of its members on three of the four metrics and ranks first overall on the public validation leaderboard.
[CV-137] ViSMoE: Visual-Aware Sparse Mixture-of-Experts for Embodied Referring Expression Grounding ICANN2025
链接: https://arxiv.org/abs/2608.21878
作者: Shuo Feng,Piji Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by ICANN 2025
Abstract:Embodied Referring Expression Grounding is the task of enabling an agent to navigate in real environments and to localize a remote object based on natural language instructions. In this scenario, the agent needs to select one view for navigation at each step and identify a specific object among all candidate objects at the destination. However, most of the previous approaches fail to distinguish between views and objects, instead processing them using the vanilla vision encoder, which results in ambiguous representations of both views and objects. To address the above issues, we propose ViSMoE, which equips sparse Mixture-of-Experts with a visual-aware routing policy for the embodied agent. This framework processes different types of visual information specifically, resulting in discriminative visual representations for both views and objects. Experimental results on REVERIE and SOON datasets demonstrate that ViSMoE outperforms the previous state-of-the-art methods, showing the superiority of our proposed method.
[CV-138] GuardPaint:SpeculativeSafetyDecodingforText-to-ImageGeneration
链接: https://arxiv.org/abs/2608.21869
作者: Shreyash Dhoot,Paras Dhiman,Arsh Abbas Naqvi,Aranbi Dutta,Aman Chadha,Vinija Jain,Amitava Das
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Text-to-image (T2I) diffusion models offer powerful visual generation, but their controllability creates a critical safety challenge: adversarial prompts can steer the denoising trajectory toward policy-violating content such as explicit nudity or graphic violence. Existing safeguards mostly act before generation through prompt filtering or after generation through image classification, leaving the diffusion process itself unguarded and often yielding only refusal rather than safe visual repair. We introduce GuardPaint, a speculative decoding framework for safe T2I generation that intervenes inside the diffusion trajectory without modifying the base model. A lightweight auditor monitors intermediate images, localizes unsafe regions, and triggers surgical inpainting repair only where needed. Candidate repairs are generated by a policy-aligned inpainter and selected through a guarded tournament that accepts edits only when they improve policy compliance while preserving prompt fidelity and perceptual quality. Across five jailbreak families SneakPrompt, MMA, PGJ, DACA, and RABell and UNet/flow-matching models including SD~1.5, SDXL, SD~3.5, and FLUX.1-dev. GuardPaint reduces attack success and harmful generations with minimal degradation to image quality, prompt fidelity, and benign behavior. Content warning: This paper contains examples involving nudity and violence that some readers may find disturbing, distressing, or offensive. Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) Cite as: arXiv:2608.21869 [cs.CV] (or arXiv:2608.21869v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2608.21869 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Shreyash Santosh Dhoot [view email] [v1] Sat, 22 Aug 2026 09:28:30 UTC (8,206 KB)
[CV-139] BioMed-Agent -RL: A Meta Learning All You Need for Biomedical Applications
链接: https://arxiv.org/abs/2608.21864
作者: Md Asaduzzaman Jabin,Zihao Wu,Tianming Liu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:The current progress of Clinical Vision Large Language Models (C-VLLMs) has substantially improved digital diagnostics, still these frameworks often endure lesion noises, modality misalignment, hallucination, and missed contextual grounding in complex clinical cases. Moreover, prevailing agent systems usually depend on static and non-adaptable pipelines and lack the versatility necessary for complex medical reasoning. To resolve these difficulties, we present BioMed-Agent-RL, a unified medical agent that incorporates adaptive orchestration, policy, and reward-based reinforcement learning (RL) models for biomedical applications. To ensure reliability, it invokes clinical context-aware preference optimization (CPO), direct preference optimization (DPO), and group relative policy optimization (GRPO) with dynamic entropy regulation. This pipeline utilizes a multimodal meta-learning approach that operates as a field-specific expert and human judgment synthesizer. The agent adaptively utilizes a set of model-level expertise, such as clinical grounding and reasoner, lesion segmenter, and field-specific synthesizer, across various clinical modalities (e.g., X-ray) by utilizing an iterative and adaptive RL approach. The agent learns to seriously synthesize misleading, conflicting vision cues and trust in inherent reasoning, while specialist advice is faulty. An intensive ablation study is conducted across multiple benchmarks, and the agent significantly outperforms existing state of the art models, such as GPT-5, attaining up to ~73% accuracy (gain of ~5%) over contemporary baselines. As a result, the framework suggests a new standard for building factual, reliable, robust, and expert-like intelligent agent systems for independent clinical reasoning.
[CV-140] Frame-Level Evaluation in Weakly Supervised Video Anomaly Detection Mostly Measures Video-Level Ranking
链接: https://arxiv.org/abs/2608.21854
作者: Inpyo Song,Jangwon Lee
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Preprint
Abstract:Weakly supervised video anomaly detectors are trained with video-level labels but are commonly evaluated as temporal localizers using Micro-AUROC or AP over pooled test frames. Because these metrics compare frames from different videos, a detector can score well by separating videos without accurately ordering moments within them. We exactly decompose Micro-AUROC by video identity into Within-AUROC for temporal ordering within videos and Cross-AUROC for comparisons across videos. Across ShanghaiTech, XD-Violence, and UCF-Crime, only 0.071-0.388% of comparisons between anomalous and normal frames occur within the same video. When both classes remain distributed across V videos, this share decreases as O(1/V), a benchmark property we call temporal dilution. We train anomaly video binary classifiers under the same video-level supervision and repeat each video score across all frames. These video-constant outputs reach 81.40-97.18 Micro-AUROC despite having no within-video variation. Across 72 controlled runs, replacing every frame score with its video mean preserves a median 98.6% of the Micro-AUROC margin above chance. The same empirical pattern holds for author-released outputs and for XD-Violence under its official AP evaluation. A detector can therefore achieve a high pooled score even when it assigns the same score to every moment within each video.
[CV-141] GaussVid: Sparse-View Gaussian Splatting with 3D-Aware Video Diffusion Priors
链接: https://arxiv.org/abs/2608.21849
作者: Xinhui Liu,Can Wang,Wei Jiang,Wei Wang,Dong Xu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:3D Gaussian Splatting (3DGS) has achieved remarkable success in novel view synthesis; however, reconstructions under sparse views often exhibit noticeable artifacts. While recent video diffusion models provide strong spatio-temporal priors for 3DGS restoration, directly fine-tuning them for restoration is suboptimal, as they lack awareness of the underlying multi-camera geometry, resulting in multi-view inconsistencies. In this work, we propose a novel 3D-aware video restoration framework designed to enhance the quality of sparse 3DGS reconstruction. Specifically, we construct a large-scale 3DGS video dataset to enable specialized fine-tuning. To bridge the gap between 2D video generation and 3D multi-view constraints, we introduce a camera-conditioned geometric prior. By using the first and last frames as boundary anchors and encoding the corresponding camera relationships, we explicitly inject spatial structure into the video generation pipeline. This boundary-anchored, camera-aware prior guides the network toward geometrically grounded restoration that remains coherent across viewpoints. Extensive experiments show that, among video-prior restoration methods, our approach attains the best pixel- and structure-level fidelity (PSNR/SSIM) and improves multi-view consistency, while remaining competitive in perceptual quality (LPIPS).
[CV-142] BC-IHV: Conditioning the Color Space for Stable Rectified-Flow Low-Light Enhancement
链接: https://arxiv.org/abs/2608.21847
作者: Yi Ai,Zheng Chen,Yuanhao Cai,Yulun Zhang,Xiaokang Yang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 7 pages, 5 figures
Abstract:Low-light image enhancement (LLIE) must correct ambiguous exposure without overwriting structure already supported by the input. Generative transport can model exposure ambiguity; however, its flexibility may also alter observable geometry and chromatic content. Moreover, fixed invertible color coordinates are usually treated only as representations, although their inverse mappings reshape the RGB-domain gradients received by the enhancement network. To address these issues, we propose Structure-Anchored Rectified Flow (SA-RF), which maintains correspondence through separate chromaticity/intensity stems, a scale-matched condition pyramid, and HybridAda. HybridAda assigns location-specific retrieval to spatial cross-attention and global exposure modulation to pooled AdaLN. We further introduce BC-IHV, a learnable Box–Cox polar color space whose analytically invertible intensity mapping controls the inverse-gradient dynamic range through a single exponent. This allows the representation to balance dark-range expansion and gradient conditioning instead of adopting a fixed linear or logarithmic law. Experiments on three LOL benchmarks, blind image-quality evaluation, and cross-dataset tests demonstrate consistent reconstruction and perceptual advantages over the sota. Controlled studies further support the effectiveness of both the proposed framework and color representation.
[CV-143] FIRM-Video: Check Before You Score for Reliable Text-to-Video Reward Modeling
链接: https://arxiv.org/abs/2608.21839
作者: Peiyuan Zhang,Xiangyu Zhao,Hongbo Liu,Xiaoxing Hu,Mingxin Liu,Shuran Ma,Yunhang Shen,Jian Hu,Haihan Gao,Haoyu Cao,Xue Yang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Reliable reward models are essential for text-to-video evaluation and alignment. However, the trade-off between evaluation accuracy and inference efficiency places high demands on the quality of training supervision. Existing approaches often rely on holistic judges with fixed rubrics or open-ended reasoning, leading to incomplete inspection, unfaithful justification, and entangled attribution. We introduce FIRM-Video, a unified checklist-driven data construction framework based on a check-before-score principle: construct dimension-specific checklists, verify each criterion against temporal visual evidence, and aggregate only verified decisions. For Instruction Following, FIRM-Video decomposes prompts into weighted atomic requirements; for World Coherence, it constructs prompt-calibrated, target-specific checks grounded in visible entities and actions; and for Perceptual Quality, it applies a generic taxonomy of visual defects. The verified criteria and scores are further transformed into natural-language analyses for end-to-end reward modeling. Subsequently, we construct FIRM-Video-90K with 88,044 dimension-specific instances from 29,348 videos, and introduce FIRM-Video-Bench with 750 point-wise human annotations across 250 videos. The Qwen3-VL-based FIRM-Video-8B achieves the best overall MAE on FIRM-Video-Bench while consistently delivering the highest VBench Total, Quality, and Semantic Scores in Best-of-8 sampling across three video generators.
[CV-144] StereoDiffuer: Diffusion-based Progressive Geometry Modeling with Saliency Attention Perception for Stereo Matching
链接: https://arxiv.org/abs/2608.21710
作者: Bohan Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 17 pages, 9 figures, and 11 tables. Accepted by Signal Processing: Image Communication
Abstract:With the advance of deep neural networks, the quality of disparity maps obtained through stereo matching has steadily improved. However, existing stereo matching methods still struggle to preserve fine-grained geometric details, resulting in blurred edges and over-smoothed predictions in challenging regions. To address these limitations, we propose StereoDiffuer, an iterative diffusion-based stereo matching framework that explicitly models geometric details and progressively refines disparity estimates. The framework incorporates a Saliency Attention Perception (SAP) module to extract salient geometric cues, including object boundaries, thin structures, and sharp edges. Confidence-guided SAP features are combined with the initial disparity estimate to condition an iterative denoising diffusion process, which corrects residual disparity errors and restores geometric details suppressed during cost-volume regularization and upsampling. Experimental results on the Scene Flow and KITTI benchmarks demonstrate the effectiveness of the proposed framework and its competitive performance relative to the compared stereo matching methods.
[CV-145] Emotion Intensity Matters: Generating Realistic Expressions in Virtual Humans with CVAEs
链接: https://arxiv.org/abs/2608.21697
作者: Vitor Miguel Xavier Peres,Lara Volpato,Gabriel Ferri Scnheider,Soraia Raupp Musse
类目: Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR)
备注:
Abstract:Generating expressive facial behavior in virtual humans (VHs) remains a central challenge in affective computing and character animation. This paper presents a novel approach based on Conditional Variational Autoencoders (CVAEs), trained on real human facial expression data, to synthesize controllable emotional expressions at varying intensities. Using a dataset comprising six basic emotions represented at two intensity levels (low and high), we train a CVAE model to generate synthetic facial expression data while preserving semantic consistency with real human expressions. Despite the limited amount of training data (only 7,680 facial expression samples), the proposed approach learns meaningful latent representations and generates coherent emotional variations. Our method enables control over emotional intensity, making it suitable for animating virtual characters without requiring actor performances or manual artistic intervention. Our research aimed to evaluate whether the method (CVAE) preserves the characteristics associated with the different intensity levels present in the dataset. Results show that the proposed model preserves key expressive characteristics across intensity levels while supporting generalization across emotional intensity levels, contributing to the creation of emotionally expressive virtual characters from relatively small datasets.
[CV-146] SketchFlow: Zero-Shot Vector Sketch Generation via GMM Prior Flow in CLIP Latent Space SIGGRAPH
链接: https://arxiv.org/abs/2608.21659
作者: Jin Zhou,Hongliang Yang,Pengfei Xu,Hui Huang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to SIGGRAPH Asia 2026 Conference Papers. 16 pages
Abstract:Vector sketches remain one of the most concise and immediate mediums for abstract human expression. However, generating high-quality vector strokes that exhibit human-like drawing styles remains an open challenge due to the severe scarcity of fine-grained, high-quality text-to-sketch paired data. Existing text-conditioned generation methods often rely on unstable, time-consuming optimization or struggle to generalize to unseen categories in a zero-shot manner. To address these limitations, we present SketchFlow, a novel generative framework rooted in Optimal Transport (OT) theory and flow matching. By leveraging pre-trained CLIP models to bypass labor-intensive image-level text annotations, we formulate cross-modal alignment as a continuous mapping problem directly within the CLIP latent space. To bridge the inevitable modality gap between discrete text concepts and continuous sketch features, we first inject noise into discrete category embeddings to construct a continuous Gaussian Mixture Model (GMM) prior. We then utilize an Optimal Transport Conditional Flow Matching (OT-CFM) model to learn a deterministic vector field mapping from this continuous GMM prior to the target sketch feature distribution. Finally, a Hybrid Diffusion Decoder, fusing 1D U-Net and Transformer architectures, is designed to decode these features into fast and high-fidelity stroke trajectories. Extensive experiments demonstrate that SketchFlow substantially outperforms existing baselines in visual quality and adherence to natural human drawing styles. Furthermore, our geometry-preserving framework demonstrates promising local zero-shot synthesis for prompts beyond the QuickDraw training vocabulary, including unseen concept labels and semantic modifiers, while enabling smooth, continuous semantic interpolation between distinct concepts. Source code is available at: this https URL.
[CV-147] Bounded Precision-Geometry Scaling for Robust Multi-Task Learning under Loss Scale Mismatch
链接: https://arxiv.org/abs/2608.21653
作者: Krishna Subedi
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multi-task learning often combines losses that span several orders of magnitude, causing homoscedastic uncertainty weighting to degrade severely. We propose Bounded Precision-Geometry Scaling (BPGS), a method that maps each task’s log-variance through a bounded sigmoid parameterisation anchored to detached batch loss statistics, and decouples network optimisation from uncertainty optimisation. Its normalised task weights are provably invariant to uniform rescaling under non-degenerate loss scales. We evaluate BPGS on synthetic stress tests and three real-world benchmarks: NYUv2 dense prediction, Yeast multi-label classification, and RF1 multi-target regression. Under pure loss rescaling from \times 1 to \times 1000 , its macro score changes from 0.777 to 0.778, whereas Kendall weighting drops from 0.780 to 0.637; \ell_1 -normalising Kendall’s weights does not close the gap. On NYUv2, BPGS records the lowest depth absolute relative error (0.223), depth RMSE (0.790), and total loss (1.891) among all compared methods, including Nash-MTL. Sensitivity studies on batch size and calibration show small variation across the tested ranges, and runtime overhead relative to Kendall is under 1%. BPGS posts the highest Yeast micro-F1 (0.616) and is competitive on RF1, though PCGrad leads RMSE and MAE there. These findings establish BPGS as a scale-robust alternative to homoscedastic uncertainty weighting, notably effective when loss-scale disparities dominate multi-task optimisation.
[CV-148] Semantic Slots for Video Object-Centric Learning BMVC2026
链接: https://arxiv.org/abs/2608.21636
作者: Khalil Sabri,Guillaume-Alexandre Bilodeau,Nicolas Saunier,Wassim Bouachir
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at BMVC 2026
Abstract:Video Object-Centric Learning (OCL) has traditionally focused on refining the encoder architecture to ensure temporal consistency. In this paper, we argue that the primary bottleneck lies in the decoder. We show that traditional decoders force slots to be spatially anchored, hindering their ability to adapt to motion. We propose SemanticSlots, which uses a Transformer-based decoder that leverages image context, relieving slots from encoding boundary precision and spatial location. This allows slots to function as semantic queries that are inherently object position invariant, retrieving matching features rather than memorizing coordinates. More importantly, this property allows slots computed from a single frame to decompose subsequent video frames, eliminating the need for complex temporal predictors or auxiliary temporal losses. Results on YouTube-VIS show that SemanticSlots improves upon VideoSAUR by 31 points in mBO and outperforms current state-of-the-art methods by 21 points, achieving 86.6% ARI and 62.8% mBO.
[CV-149] ASSO: TAsk-Specific Subspace Optimization for Continual Learning of Vision-Language Models
链接: https://arxiv.org/abs/2608.21487
作者: Chang Sun,Francesco Barbato,Matteo Caligiuri,Pietro Zanuttigh
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Vision-Language Models (VLMs) exhibit strong zero-shot capabilities, making them an attractive solution for continual learning across diverse tasks. However, during continual adaptation, both catastrophic forgetting and zero-shot degradation occur, severely degrading performance. In this paper, we introduce TASSO, a new paradigm that efficiently preserves the latent space geometry while ensuring network plasticity. We achieve this with two complementary techniques: subspace learning and geometry-aware knowledge distillation. Specifically, we first learn a sequence of task-specific low-rank projectors, which we use to project the latent representations before optimizing cross-entropy. Secondly, we employ a geodesic-distance-based loss that distills knowledge from the previous-task model while effectively preserving the latent space geometry. These design choices not only avoid unnecessary parameter updates along the full embedding dimensions but also improve learning by focusing on task-specific manifolds. Moreover, the geometry-aware distillation provides strong regularization and significantly reduces both catastrophic forgetting and zero-shot degradation throughout the continual learning sequence. Experimental results with the CLIP vision language model in the multi-domain task incremental and class incremental learning benchmarks demonstrate clear improvements over state-of-the-art methods in mitigating forgetting and preserving zero-shot capabilities.
[CV-150] EXPL-FR: Explaining Face Recognition Models via Vision-Language Alignment ECCV2026
链接: https://arxiv.org/abs/2608.21486
作者: Guray Ozgur,Mustafa Efe Tamyapar,Naser Damer,Fadi Boutros
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at the ECCV 2026 Workshops
Abstract:Deep face recognition (FR) models reach near-saturated accuracy but remain opaque: a practitioner cannot ask which semantic attributes a similarity score relied upon. EXPL-FR answers this inside the FR model’s own embedding space. A lightweight adapter aligns a vision-language model’s (VLM) image encoder with the frozen FR space, trained on face images alone and never on text. Because the VLM’s encoders share one space, the same adapter applies to the text encoder, turning 978 attribute prompts in 22 categories, also extendable, into FR-space anchors at no extra cost. We do not assume this transfer works: a face-verification protocol measures it, and an ablation changing only the adapter isolates its contribution. Not every concept survives, because an FR model earns its invariances by discarding the factors it must verify identities across. A label-free detectability measure compares each concept’s separability in FR space against the VLM space, and the 100 most detectable form the model’s readable semantic signature, which separates identities better than the full vocabulary. We cover four FR backbones and two VLM encoders, EXPL-FR needs no architecture access, and supports identity-level, per-image, and differential explanations. We benchmark attribute-level auditing under three supervision settings, human labels (current practice), VLM pseudo-labels, and our fully prompt-driven audit, against real verification behavior. With no labels, the prompt-driven audit ranks four FR models by their measured per-ethnicity RFW errors and ranks controlled attribute changes by their true verification cost.
[CV-151] From Subjective Judgments to Auditable Standards:Protocol-Guided AI Auditing of Website Redundancy
链接: https://arxiv.org/abs/2608.21476
作者: Ge Kong,Yongtong Cao
类目: oftware Engineering (cs.SE); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Website redundancy does not have a single fixed meaning. The same repeated element may distract during one task and provide backup during another. We introduce CORA (Counterfactual, Observable Redundancy Audit), which measures repetition load, normal-use tax, and failure-domain recovery reserve separately. Each run retains screenshots, stable element identities, and task traces. A versioned vision-language model proposes the annotations. Typed validation and release checks then determine whether a calibrated dimension can be reported; failed or malformed outputs stay in the fixed denominator. On a transparent mechanistic testbed, the factorized CORA representation separated reserve from normal-use tax and predicted perturbed success more accurately than scalar-load baselines. The model studies then showed why repeatability is not enough: two small local vision-language models produced recurring outputs, but neither instrument met all release requirements. CORA therefore withheld automated scores from both instruments while retaining the raw responses and failure records. Separate checker fixtures confirmed that the typed validator and hardened release gates implement their specifications; these tests do not establish semantic grounding or accuracy on production sites. Taken together, the results position CORA as an auditable candidate procedure for the controlled benchmark studied here rather than a general standard. Human agreement, AI-versus-human accuracy, and validation on independent production sites remain open empirical questions.
[CV-152] 3D Point Cloud from Close-Range Photogrammetry for Defect Characterisation of Rubberised Concrete
链接: https://arxiv.org/abs/2608.21468
作者: Jiacheng Liu,Mohammed Alnahhal,Ailar Hajimohammadi,Sara Gonizzi Barsanti,Jinling Wang,Mohsen Kalantari
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages, 10 figures. Published in the International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XXV ISPRS Congress 2026, Toronto, Canada
Abstract:While three-dimensional (3D) point clouds are widely used in civil engineering, mainstream LiDAR systems such as Terrestrial Laser Scanning (TLS) are physically constrained to laboratory environments. Since their laser spot size typically exceeds the width of microcracks, the beam physically bridges over voids, rendering TLS unsuitable for fine-scale defect analysis. Alternatively, close-range photogrammetry utilising Structure-from-Motion (SfM) and Multi-View Stereo (MVS) algorithms offers a solution for testing highly tortuous materials, and its utility at fine-scale remains underexplored. This study adapts photogrammetric workflows specifically for rubberised concrete (RuC), a sustainable composite exhibiting high ductility and complex fracture morphologies. High-resolution image sets were captured using a Canon DSLR and an iPhone 16 to generate dense 3D models. Comparisons revealed that the DSLR-based reconstruction achieved sub-millimetre resolution, demonstrating superior performance for fine-scale surface monitoring. An RGB-guided crack extraction method was developed to enhance the identification of surface defects and isolate potential crack areas from the background. The extracted crack regions were visually distinguishable and provided a well-structured geometrical representation of defect morphology. Furthermore, a Pre and Post-Test deformation analysis was conducted to quantify surface displacement across testing stages. The results confirm that this close-range photogrammetry workflow is a flexible, high-resolution alternative to LiDAR for surface inspection and deformation monitoring of specimens in laboratory settings. Ultimately, this approach establishes a robust geometric baseline for future automated 3D feature characterisation and material performance evaluation.
[CV-153] Complexity Induction: Compositional Generalization via Structured Label Distortion
链接: https://arxiv.org/abs/2608.21464
作者: Aleksandr Abramov
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:We demonstrate that structured distortion of training data - which we term complexity induction - can induce compositional generalization in a standard CNN classifier without architectural modification. Using synthetic images of colored geometric shapes, we encode classes as flat string labels (e.g., “red-circle”) with no explicit attribute decomposition, and exclude certain color-shape combinations from training entirely. We apply two distortion methods derived from Jaccard string similarity between class names: mixed labels (soft target distributions encoding inter-class overlap) and expanded dataset (false training samples with structurally motivated incorrect labels). Both methods induce the ability to predict unseen class combinations, and act at different levels: mixed labels activate the classifier for unseen combinations by exploiting the CNN’s natural embedding structure, while expanded training improves the embedding factorization itself. A control with random (unstructured) false labels confirms that the effect depends on the structure of the distortion, not on noise per se. These results suggest that structured complication of training signals can influence both the internal organization of learned representations and their compositional interpretation - a principle that may underlie the role of natural language in cognitive development.
[CV-154] FigmaTrace: Capturing Creative Nuances in Human Figma Design Workflows
链接: https://arxiv.org/abs/2608.21460
作者: Darshan Deshpande,Yoshinari Fujinuma,Martyna Markiewicz,Devanshu Bansal,Shivani Jain,Nicholas Saban,Chirag Maheshwari,Anand Kannappan
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Dataset: this https URL Model: this https URL
Abstract:Vision Language Models have recently shown improvements in several objective and verifiable domains such as object detection but continue to underperform on subjective and creative design tasks. A major contributor to this performance gap is the lack of high quality human workflow data that captures a diverse set of preferences and decisions that make human experts good at design tasks. In this work, we first define a unique, expert curated taxonomy of design skills and best practices which we further expand into a set of 126 open ended, subjective, long horizon tasks. Built on top of this and expert solutions, our dataset FigmaTrace contains over 200 hours of human captured video data converted into 3469 design trajectories using a novel design phase-based method. We use our dataset to train four models and show that training on FigmaTrace leads to a performance improvement comparable to frontier closed models such as \textscClaude-Opus-5 and \textscGPT-5.6-Sol on four out of distribution agentic GUI environments. We further perform a useful ablation to attribute these performance improvements to a design phase-based video to trajectory conversion which outperforms prior length-based conversion approaches. Finally, we perform a qualitative analysis on the best performing \textscQwen3.8-27B outputs to better correlate performance improvements to FigmaTrace’s trends. We open source our dataset and the best model for the community.
[CV-155] CLSC DETR: Reliable Candidate Ranking via Cross Layer Geometric Support for UAV Small Object Detection
链接: https://arxiv.org/abs/2608.21457
作者: Junyan Lin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Unmanned aerial vehicle (UAV) object detection is critical for applications such as target search, where accurate detection of small objects in complex aerial scenes remains challenging. The limited spatial extent, dense distribution, and frequent occlusion of small objects make reliable candidate ranking particularly difficult. Existing Detection Transformer (DETR) based methods improve ranking by estimating localization quality from individual queries and incorporating it into classification scores. However, a single query often lacks sufficient geometric evidence for small objects with weak boundary cues, resulting in unreliable quality estimation and unstable ranking. To address this limitation, we propose Cross Layer Local Support and Consistency Calibration for DETR, termed CLSC DETR. Specifically, the Cross Layer Local Support module establishes correspondences between final layer queries and intermediate layer candidates to aggregate complementary geometric evidence for more reliable localization quality estimation, while the Classification and Localization Consistency Calibration module adaptively adjusts classification scores according to localization quality and classification reliability to improve candidate ranking. Experiments show that CLSC DETR improves AP and AP _75 over the baseline by 1.5% and 2.0% on VisDrone, respectively, while achieving consistent improvements on UAVDT.
[CV-156] omatoes Potatoes and Onions: Questioning the Need for Faces in Face Presentation Attack Detection
链接: https://arxiv.org/abs/2608.21455
作者: Guray Ozgur,Fadi Boutros,Naser Damer
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at ACMM2026 Workshops
Abstract:Face presentation attack detection (PAD) is traditionally formulated as a face-specific problem, although many of the visual artifacts introduced by print, replay, and recapture processes are not inherently tied to facial appearance. In this work, we investigate whether transferable PAD representations can be learned without using faces during downstream PAD training. To this end, we introduce TPO, a controlled face-free presentation attack dataset consisting of bona fide, print, and replay recordings of, almost randomly chosen, tomatoes, potatoes, and onions acquired under protocols that closely mirror conventional face PAD datasets. Using a foundation-model-based PAD architecture, we demonstrate that a detector trained on TPO achieves an average AUC of 92.70% across four standard cross-dataset face PAD benchmarks, outperforming training on synthetic faces and remaining competitive with models trained on real face datasets. Conversely, models trained on face PAD datasets transfer consistently above chance to TPO, suggesting that the learned representations capture characteristics of the presentation process rather than object semantics. Furthermore, incorporating TPO into conventional face PAD training consistently improves cross-dataset performance under fixed optimization budgets, indicating that face-free data provides complementary information rather than simply additional training samples. Finally, representation and frequency analyses provide further evidence that transferable PAD representations cannot be explained by a single spectral artifact but instead encode richer presentation cues shared across object categories. Together, these results provide empirical evidence that transferable presentation attack representations can be learned independently of facial content, opening new opportunities for privacy-preserving and identity-independent PAD development.
[CV-157] Multi-Scale Fruit Capsules: Dilated Convolutions and Dynamic Routing for In-the-Wild Explainable Fruit Recognition
链接: https://arxiv.org/abs/2608.21454
作者: Subhankar Chattoraj,Sawon Pratiher,Samiran Das,Hubert Konik
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted in IECON 2026
Abstract:The same fruit appears in a bunch, unpicked, peeled, bagged in plastic, or sliced on a dish, so automated fruit classification in the wild (AFCW) must absorb wide intra- class and narrow inter-class variability in shape, size, colour and texture. Convolutional networks route information through pooling, which discards the pose and location of the region of interest and therefore generalises poorly across these presentations. We propose FruitCapsNet, a capsule network whose Fruit Capsules replace the standard convolutional front end with dilated convolutions: the receptive field grows exponentially at constant parameter cost, so each capsule encodes multi-scale context before dynamic routing resolves part whole spatial agreement. Hyper-parameters, including the dilation factor, are selected by Bayesian optimisation rather than grid search. On three public datasets (SMP, FruitsGB, Fruits-360) and a new 19-class, 10,639-image in-the-wild dataset (PD-19), FruitCapsNet exceeds ten fine-tuned transfer-learning backbones at one-third the depth, with the largest margin (+2.7% over the nearest competitor) on the hardest set. Grad-CAM saliency propagated from the DigitCaps layer shows that the improvement comes from attributing decisions to whole-fruit regions rather than to object edges, giving post-hoc evidence that the gain is not a dataset artefact.
[CV-158] Beyond Visual Similarity: Entity-Aligned Retrieval for Knowledge-Based Visual Question Answering ACM-MM2026
链接: https://arxiv.org/abs/2608.21450
作者: Hangrui Xu,Zhengxian Wu,Yunyao Yu,Zhuohong Chen,Rui Cong,Xiangwen Deng,Zhifang Liu,Peng Jiao,Haoqian Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by ACM MM 2026
Abstract:Knowledge-Based Visual Question Answering (KB-VQA) relies on retrieving external information to answer queries involving long-tail entities. However, existing retrieval pipelines predominantly employ CLIP-style dual encoders, which prioritize surface-level visual similarity over entity-level semantic alignment. This paradigm often fails when semantically identical concepts exhibit large visual variations or when distinct entities appear visually similar. To address this, we propose KBMR, the first MLLM-based embedding retriever tailored for KB-VQA. Leveraging the robust autoregressive capabilities of MLLMs, KBMR maps images into a semantic space that better preserves concept identity. To tackle the challenge of noisy supervision in Wikipedia-scale retrieval, we introduce an MLLM-based semantic discriminator that generates continuous entity-consistency weights. These weights guide a novel continuous semantic distillation objective, enabling effective hard negative sampling and soft supervision beyond rigid binary labels. Extensive experiments demonstrate that KBMR significantly outperforms CLIP baselines, yielding up to a 14.7% improvement in retrieval Recall@1 and a 9.4% gain in end-to-end VQA accuracy. Code is available at this https URL.
[CV-159] BIMScript: Material-Aware Structured Scene Programs for BIM Ingestion ECCV2026
链接: https://arxiv.org/abs/2608.21447
作者: Prakash Kondibhau Naikade,Thomas B. Moeslund,Andreas Møgelmose
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project Page: see this https URL ; to be published in ECCV 2026 TwinWorld Workshop
Abstract:Structured-language models such as SceneScript reconstruct a scene as a short program of parametric commands, an inherently editable and semantically explicit representation. We ask three questions that stand between such models and their most compelling application, automated ingestion of existing buildings into BIM tools, studied here on synthetic scans: \emphwhat is the scene made of, \emphhow fast can it be produced, and \emphexactly where is each element. BIMScript answers all three within one grammar. First, we extend the layout language with per-element \emphmaterial and \emphcondition attributes, supervised by a vision-language-model material-passport corpus we build over 100k synthetic scenes (1.9M pseudo-labeled elements), and route image appearance to the material tokens through a lifted-feature point encoder. Second, we show that autoregressive decoding of these programs is dominated not by compute but by kernel-launch and host-synchronization overhead, and remove it with an output-exact CUDA-graph decoder (1.9 vs 6.4,ms/step, 3.4\times ) plus a grammar-parallel, tolerance-verified draft-and-verify scheme that exploits the deterministic entity schema. Third, we address the model’s 5cm token-grid granularity with training-free geometric snapping and a hybrid discrete–continuous decoder head that regresses a sub-bin offset, and measure how much of the residual error each recovers. Because each command maps one-to-one onto a native Revit object, we validate direct ingestion into a BIM authoring tool end to end with a working add-in and its IFC4 export, and the same program’s language form is designed to support LLM-driven, sustainability-aware reasoning over the built asset.
[CV-160] ViTexSZ: Heterogeneous Vision-Text Knowledge Distillation for EEG Seizure Detection
链接: https://arxiv.org/abs/2608.21445
作者: Chenxi Liu,Mingzhao Li,Yicong Liu,Hao Miao,Hongyuan Zhang,Ziyi Chen,Gaofeng Meng
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Automated seizure detection from electroencephalography (EEG) is essential for continuous neurological monitoring, particularly for subclinical epileptic seizures that may exhibit only subtle electrographic changes. Existing time-series methods are often designed for fixed EEG channel configurations, thereby limiting their applicability to heterogeneous EEG recordings with irregular channel layouts. Although visual and language modeling offer promising alternatives, aligning heterogeneous EEG representations with clinical semantics remains challenging. We introduce ViTexSZ, a heterogeneous Vision-Text knowledge distillation framework for EEG seizure detection. ViTexSZ converts EEG recordings into structured waveform images and introduces a query-based multi-channel alignment module that maps source-dependent visual features into a unified token space. A heterogeneous teacher further integrates the aligned EEG representations with clinical prompts through a multimodal large language model, associating high-level clinical semantics with seizure-related evidence. Vision-text knowledge distillation then transfers the teacher representations to a lightweight student during detection. Experiments on four EEG seizure datasets demonstrate the generalizability of ViTexSZ across both subclinical and general seizure detection scenarios, achieving the highest accuracy on all datasets and relative improvements of up to 12.9% over the second-best baselines, showing its effectiveness.
[CV-161] xt-Guided Visual Dependency Graph Learning with Cross-Modal Attention Priors
链接: https://arxiv.org/abs/2608.21443
作者: Fei Wang,Yutong Zhang,Yang Ye,Jinxian Chen,Wang Wenshuai,Xiong Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Estimating interpretable conditional-dependence structures from multimodal visual-linguistic features remains largely unexplored. We propose CM-GLasso (Cross-Modal Graphical Lasso), a framework that bridges vision-language representation learning and sparse Gaussian Graphical Models. CM-GLasso introduces three key components: (i) a text visualization strategy that renders class-attribute descriptions as images and processes them through the same SigLIP-2 vision encoder as natural images, yielding prototype-indexed patch-level attention footprints in a shared feature coordinate system; (ii) a cross-attention distillation mechanism that condenses high-dimensional patches into a small set of semantic graph nodes, whose attention-footprint similarities yield cross-modal structural priors for non-uniform L1 penalization; (iii) a joint ADMM formulation that estimates shared and class-specific precision components within a single convex objective, avoiding the need to first estimate and then decompose separate class-wise graphs. The learned sparse graph topologies directly support a parameter-free, precision-based classification rule and a lightweight topology-aware segmentation head. Extensive experiments on eight benchmarks demonstrate that CM-GLasso achieves competitive or superior performance compared with strong feature-based and task-specific baselines. Under the matched controlled protocol, it attains the highest average classification accuracy (91.97%) and the highest segmentation mIoU among the controlled baselines on VOC (74.75%) and ADE20K (64.01%), while also yielding explicit sparse conditional-dependence graphs with common-specific decomposition.
[CV-162] WorldMind: Decoupled Game World Model for State-Aware NPC Behavior
链接: https://arxiv.org/abs/2608.21439
作者: Zhiyang Deng,Boran Zhang,Danze Chen,Yeying Jin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL
Abstract:Game world models have recently demonstrated promising capabilities in generating visually coherent and action-controllable gameplay videos. However, non-player character (NPC) behavior in existing models is either implicitly entangled with video generation or explicitly prescribed through external control signals. Consequently, a game world model has to jointly understand the state, plan the NPC’s response and render its visual outcome, limiting its ability to produce responsive and state-aware NPC behavior. The challenge lies in the lack of an explicit interface for state-grounded decision-making. To this end, we introduce WorldMind, to our knowledge the first decoupled framework for state-aware NPC behavior in game world models. WorldMind separates interactive world modeling into four layers: an Understanding Layer that constructs a compact state from generated frames; a Decision Layer that reasons over the compact state to plan the NPC’s next action; a Control Layer that translates the actions into temporally aligned conditions; and a Generation Layer that synthesizes their visual outcomes. By reconnecting layers in a closed interaction loop, WorldMind grounds NPC behavior in the evolving game state. We further introduce BOSS-140K, a dataset of gameplay videos paired with rich internal game states, together with an agent that automates the collection at scale. Experiments on BOSS-140K demonstrate reliable compact state reconstruction and mechanics-grounded planning, with WorldMind preferred over the baselines in approximately 70% of pairwise comparisons for its more tactically appropriate and coherent NPC behavior. Project page: this https URL
[CV-163] Boosting Knowledge-based Visual Question Answering with Structured Context Reasoning ICME2026
链接: https://arxiv.org/abs/2608.21431
作者: Qiyou Liu,Yong Zhang,Jianjie Luo,Zhenguo Yang,Yi Yu
类目: Computer Vision and Pattern Recognition (cs.CV); Multimedia (cs.MM)
备注: Accepted by ICME 2026. Source code is available at this https URL
Abstract:Knowledge-based Visual Question Answering aims to answer questions about an image by integrating external knowledge with visual and textual information. Recent approaches often rely on in-context learning to prompt Large Language Models (LLMs) with multimodal context in a zero-shot or few-shot manner. However, we observe that directly concatenating heterogeneous visual descriptions and retrieved knowledge into long, unstructured prompts often degrades reasoning performance, due to both excessive irrelevant context and the lack of explicit relational structure. In this paper, we propose an LLM-based Structured Context Reasoning (SCoRe) framework that infers both explicit and implicit relationships for prediction. SCoRe consists of three stages: Context Acquisition, which generates diverse visual notes and retrieves explicit knowledge via an efficient two-stage multimodal retrieval strategy; Context Selection, which filters relevant visual, explicit, and implicit knowledge using LLM-guided selection; and Context Compression, which performs Relational Logic Distillation (RLD) to transform raw text into explicit entity-relation triplets. These relational triplets serve as a concise and structured prompt for final answer prediction. Extensive experiments on the OK-VQA and A-OKVQA benchmarks demonstrate that SCoRe consistently outperforms state-of-the-art methods.
[CV-164] Measuring Gender Representation in Animated Films
链接: https://arxiv.org/abs/2608.21429
作者: David Bamman,Allison Cooper,Ruby Alvarez Rubio,Reina Kushihashi,Madison Mar
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Animated films–often developed with an audience of children in mind–are an important vector for enculturation, and empirical work that has examined the representation of gender at scale in these films has largely focused on counting the gender composition of the cast rather than deploying a more fine-grained instrument (such as assessing the visibility of those characters in overall screentime). In this work, we develop a computational pipeline for recognizing animated characters in these films, and use it to test several hypotheses about gender representation in a corpus of 224 popular animated movies. We find that while the overall representation of female characters in animated films largely tracks with those of live-action films (over the period 1980-2025), we see stark differences between the representation of human characters (much greater representation among women and girls) and non-humans (largely male). Contrary to past work on Disney, we do not see female characters declining in antagonist roles in animated films, and characters who are women and girls are much more likely to share scenes together than their live action contemporaneous counterparts.
[CV-165] Few-Shot Cross-Dataset Adaptation for Tuberculosis Detection Using DenseNet
链接: https://arxiv.org/abs/2608.21427
作者: Bidhan Biswas,Shahadat Hossain Sohag,Nabil Ashab,Soumit Kumar Kundu,Saif Mahmud Parvez
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Tuberculosis (TB) is one of the most common and dangerous bacterial ailments. Every year, it causes a large number of deaths worldwide. Although many deep learning models can detect tuberculosis from chest X-rays quite accurately, severe domain shift across datasets makes the task challenging. Different imaging protocols, patient demographics, and equipment across domains make the task of generalization difficult. In real-world settings, a model may perform well on one dataset but show a noticeable drop in performance when tested on another. In this work, we address this domain adaptation challenge through a few-shot scaling study. A controlled cross-dataset evaluation is presented in this paper using TBX11K as the source domain and the Mendeley TB dataset as the target domain. It is investigated how varying the number of target samples affects model performance under three training regimes: frozen backbone adaptation, full fine-tuning of a source-pretrained DenseNet121 model, and training from scratch. The results indicate that the model can perform well even with limited data and can achieve 98.36% accuracy with just 75 labeled samples per class. The adaptation curves demonstrate how fine-tuning effectively mitigates domain shift. These findings establish full fine-tuning of pretrained models as a highly effective and practical strategy for mitigating domain shift in low-resource clinical deployment scenarios.
[CV-166] AI Visual Inspection for Garment Production
链接: https://arxiv.org/abs/2608.21426
作者: Ray Wai Man Kong,Ding Ning,Theodore Ho Tin Kong
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: 18 pages, 8 figures
Abstract:The garment manufacturing industry is under increasing pressure to improve product quality, reduce costs, and accelerate digital transformation toward Industry 4.0. One of the most challenging quality-control activities is sewing-line inspection, where defects such as broken stitches and skipped stitches are difficult to detect consistently through manual inspection. Human-based inspection is often affected by fatigue, subjective judgement, and inconsistent performance, resulting in defect leakage, rework, and reduced production efficiency. This study presents the development and validation of an Artificial Intelligence (AI)-based visual inspection system for garment sewing-line quality control. The system utilizes Convolutional Neural Networks (CNNs) to detect sewing defects and was initially trained using black fabric and black sewing thread samples. Experimental testing was conducted on black, red, dark green, light blue, silver, and fluorescent yellow fabrics. The results demonstrated successful detection of jump sewing-line defects on black, red, and dark green materials, while performance limitations were observed for broken sewing-line defects and fabrics with significantly different visual characteristics, including light blue, silver, and fluorescent yellow colours. These findings indicate that model accuracy is strongly influenced by the diversity of training data and the ability to generalize across different fabric and thread colours. Comments: 18 pages, 8 figures Subjects: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO) Cite as: arXiv:2608.21426 [cs.CV] (or arXiv:2608.21426v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2608.21426 Focus to learn more arXiv-issued DOI via DataCite Journalreference: International Journal of Computer Science and Information Technology Research V14 issue 3 pp47-64 July-September 2026 Related DOI: https://doi.org/10.5281/zenodo.21946142 Focus to learn more DOI(s) linking to related resources
[CV-167] Aligning Human Sense: Calibrated Distributional Reward Learning for Video Generation ECCV2026
链接: https://arxiv.org/abs/2608.21425
作者: Nai-Xin Zhai,Weihua Cheng,Dexu Yu,Yikai Gu,Hanwen Du,Junchen Fu,Chenxi Huang,Yingwei Song,Liyuan Lillian Ma,Yang Ran,Youhua Li,Yongxin Ni
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accepted by ECCV 2026
Abstract:Video generation is central to AI-powered content creation. Aligning generated videos with human preferences is a key criterion for evaluating generation quality. Despite significant progress in visual quality, three key challenges remain. First, the reliability of reward signals is constrained by the quality of human preference data, which is often affected by subjective noise and bias. Second, standard scalar reward models collapse multi-aspect human preferences into a single value, leading to the loss of dynamic trade-offs across multiple preference dimensions. Third, in policy optimization, the widely adopted KL divergence imposes primarily local constraints and may fail to capture the global structure of human preferences. To address these challenges, we propose a unified preference-aware learning framework for video generation. First, we introduce elite-guided filtering to calibrate preference data and construct reliable supervision for reward model training. We then model video quality as a multidimensional reward distribution to capture the uncertainty inherent in human preferences, and use the Wasserstein distance to align the learned reward distribution with the empirical human preference distribution. Finally, we introduce Wasserstein-based distributional alignment into GRPO, guiding policy optimization to better match the global structure of human preferences over videos. Experiments on reward modeling and video generation demonstrate that our approach improves the reliability of reward signals and the perceptual consistency of generated videos. Our code is available at this https URL.
[CV-168] opology of a Smile: Persistent Homology in Dental Imaging
链接: https://arxiv.org/abs/2608.21422
作者: Leon Dahlmeier,Sara Kališnik,Albert Mehl,Bastian Rieck
类目: Computer Vision and Pattern Recognition (cs.CV); Algebraic Topology (math.AT)
备注:
Abstract:CBCT (Cone Beam Computed Tomography) scans provide detailed three-dimensional images, widely used in dentistry for diagnostic and treatment planning tasks. While invaluable, analyzing and documenting these scans is labor-intensive, prompting efforts to automate key steps like the classification and segmentation of anatomical structures to identify tooth types and associated pathologies. In this article, we propose an approach to automation that leverages persistent homology, a framework from topological data analysis that studies the shape of data by identifying features like connected components, holes, and voids across multiple scales. Persistent homology, together with a support vector machine, allows us to classify teeth in a CBCT scan and to perform diagnostics. Our method advances the state of the art, reaching average accuracy scores of 97.67% for tooth-labeling and 96.77% for diagnostic tasks, outperforming a CNN trained on the same data with accuracy of 70.27% and 86.67%, respectively.
[CV-169] GET: Generative Embedding Translation for Medical Image Segmentation ECCV2026
链接: https://arxiv.org/abs/2608.22619
作者: Md Maklachur Rahman,Md Hasan Al Banna,Saraf Anjum,Mahmudul Hasan,Tracy Hammond
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Accepted at ECCV 2026 - BioImage Computing Workshop
Abstract:Generative segmentation provides an alternative to direct pixel-wise prediction by operating on learned latent representations, but effective image-to-mask translation must preserve target structure while remaining computationally efficient. We propose Generative Embedding Translation (GET), a structured embedding-translation framework that progressively transforms image embeddings into mask embeddings within the frozen latent space of a Stable Diffusion VAE. GET uses a U-Net-style Embedding Translation Network with 1.07M trainable parameters, combining Mobile Bottleneck Convolutions, Subsampled Self-Attention, and Multi-scale Feature Enrichment for local modeling, global context, and multi-scale refinement. Across five medical segmentation datasets, GET outperforms generative, CNN, and Transformer baselines. Compared with the strongest generative baseline, GMS, GET improves average Dice and IoU by 0.93% and 1.26%, reduces HD95 by 0.81 pixels, and uses 31.41% fewer trainable parameters. Under bidirectional BUS-BUSI domain shift, GET further improves Dice and IoU by 3.51% and 3.39%, while reducing HD95 by 27.37 pixels. Our code is available at: this https URL.
[CV-170] CiUNet: A Hybrid Swin-CNN UNet for Medical Image Segmentation
链接: https://arxiv.org/abs/2608.22281
作者: Bin Dong,Jinghong Chen
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: 14 pages, 3 figures. The demo predictor and trained weights are available at: this https URL
Abstract:Medical image segmentation requires high accuracy and robustness, yet practical commercial deployment also demands privacy preservation and computational efficiency. In this context, the U-Net architecture, which can be inherently decoupled into independent encoder and decoder components, serves as a natural commercial choice. However, pure Transformer-based variants like Swin-UNet often suffer from insufficient local detail capture and limited interpretability. In this paper, we propose a lightweight hybrid architecture built upon the Swin-UNet framework. Our model integrates a parallel CNN encoder to complement the shallow layer reasoning of Swin Transformers with local texture features. To bridge the semantic gap and enhance fine-grained spatial detail recovery, we design an asymmetric feature fusion strategy and introduce cross-layer skip (XSkip) connections that explicitly propagate shallow CNN features into the decoder. We further incorporate novel loss functions and an auxiliary supervision head (Aux-Head) to strengthen training stability, boundary delineation, and intermediate feature interpretability. Extensive experiments on the Synapse multi-organ segmentation dataset demonstrate that our approach achieves state-of-the-art competitive Dice scores and Hausdorff distances, offering an accurate, efficient, and interpretable solution for clinical deployment.
[CV-171] Pretreatment DCE-MRI Resolves Response Quality Within Pathologic Endpoints in Neoadjuvant Breast Cancer
链接: https://arxiv.org/abs/2608.22097
作者: Dattatreya Kantha,Murray H. Loew
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Quantitative Methods (q-bio.QM)
备注:
Abstract:Pathologic complete response (pCR) is a strong neoadjuvant endpoint, yet 5-15% of complete responders recur and clinical/genomic variables do not reliably identify them. We tested whether pretreatment dynamic contrast-enhanced MRI entropy - intratumoral enhancement heterogeneity - resolves response quality hidden within pCR and residual cancer burden (RCB). Across four cohorts (1,200 patients), a prespecified entropy threshold defined favorable and adverse structural states. Crossing structure with pathology yielded a four-tier framework spanning 4.1-fold recurrence in I-SPY1 and 7.7-fold at response extremes. In I-SPY2, 55 of 219 complete responders (25.1%) were structurally adverse, pretreatment. In an external HER2-positive responder synthesis (I-SPY1 pathology-confirmed pCR plus UCSF best-response proxy; n = 33, 10 events), adverse structure was associated with higher recurrence risk (HR = 2.87, 95% CI 1.38-5.96) capturing 7 of 10 recurrences, enriching rather than determining risk. In a HER2-positive RCB-0 subset, recurrence was 12.5% with favorable and 80.0% with adverse structure; Firth Cox regression preserved the association (HR = 8.13, 95% CI 1.71-49.21; n = 21, 6 events). In Duke (n = 908; 76 events), favorable structure remained independently associated with lower distant-recurrence risk (adjusted HR = 0.61, 95% CI 0.41-0.91). RNA linked favorable structure to a directionally reproduced immune-architecture program among non-overlapping patients within ISPY2; EMT-pathway enrichment was favorable-side, while the adverse tier contained a broadly immune-depleted substate. Yet full-cohort RNA models weakly discriminated structural state and did not recover continuous entropy. Pretreatment MRI therefore does not replace pCR or RCB; it reveals response-quality differences that these endpoints compress and identifies a recurrence-enriched group for prospective validation.
[CV-172] SweepLSD: A One-Pass O(width)-Memory Line Segment Detector with an Integer-Only Streaming Core and a Real-Time FPGA Realization
链接: https://arxiv.org/abs/2608.22086
作者: Yoshiyasu Shimizu
类目: Image and Video Processing (eess.IV); Hardware Architecture (cs.AR); Computer Vision and Pattern Recognition (cs.CV)
备注: 40 pages, 12 figures, 18 tables. Code, benchmarks, and evaluation harnesses (MIT): this https URL
Abstract:We present SweepLSD, a line segment detector that reads the image exactly once and emits each segment within a few rows of its last pixel passing the scan line. Every stage, including connected-component labeling and the final line test, processes the image as a row stream: intermediate memory is O(width) rather than O(pixels), and the per-pixel core is integer-only. We give the first complete description of the algorithm, designed in the author’s 2014 master’s thesis but never published, together with an open-source C++17 implementation and an FPGA realization – held bit-exact against the software in its hardware configuration – detecting segments in live 1080p30 video on 2009-era silicon without frame buffer or external memory. On structure-rich public 4K photographs downscaled to Full-HD, one CPU thread detects segments in ~11 ms – 4.6x/5.2x/25x faster than the original authors’ implementations of ELSED, EDLines, and LSD – with the tightest frame-time distribution and the best per-segment direction accuracy of the four detectors, and curve rejection by design, while trailing ELSED in F-score on synthetic ground truth. A Manhattan-frame vanishing-point study on York Urban and NYU-VP scores every detector under a selection/evaluation-separated best-estimator-per-detector protocol, under which SweepLSD leads on NYU-VP by ~0.3 degrees and trails by 0.1 degrees on York Urban, with the fastest end-to-end pipeline of the four detectors on both. A single-frame camera-attitude application, evaluated on synthetic scenes with exact ground truth and on EuRoC and TUM-VI, matches the baselines’ accuracy at a fraction of their memory, and drives a 4K horizon lock to 0.06 degrees median attitude error at 32 ms median per frame.
[CV-173] CRS-Bench: A Reference-Relative Reliability Benchmark for Medical Image Encoders WACV2027
链接: https://arxiv.org/abs/2608.22059
作者: Xingtao Lin,Hangqi Ren,Caiwan Sun,You Chen
类目: Image and Video Processing (eess.IV); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages, 7 figures. Submitted to WACV 2027
Abstract:Pretrained image encoders are central to medical image classification, where expert annotation is costly and task-specific cohorts are often limited. As the model space expands from general-purpose to broad-medical and specialty-specific encoders, selecting the representation becomes a substantive modeling decision. Clean-test discrimination alone is insufficient for this purpose: encoders with similar AUROC can differ in calibration, label efficiency, and stability under acquisition perturbations or distribution shift. We introduce CRS-Bench, a controlled benchmark for multi-objective medical encoder selection. CRS-Bench evaluates 15 pretrained encoder families across dermatology, ophthalmology, and radiology using ISIC 2019, APTOS 2019, and CheXpert, with CheXpert-to-MIMIC-CXR as an observed institutional shift, yielding 17,575 controlled run records and 3,515 seed-aggregated metric rows. Each encoder is characterized along four operational reliability dimensions: discrimination, calibration, label efficiency, and robustness. We summarize these dimensions using the Clinical Reliability Score (CRS), a Pareto-aware, reference-relative score combining dominance, profile balance, and worst-axis performance. AUROC and CRS are positively associated but not decision-equivalent: 21 of 105 pairwise orderings reverse, with a mean absolute rank displacement of 1.87. Paired-seed bootstrap analysis identifies PanDerm, MedSigLIP, and MedGemma as a stable leading reliability tier rather than a statistically resolved single leader. CRS-Bench provides a controlled framework for selecting medical image encoders from multi-axis reliability profiles rather than clean-test AUROC alone. Comments: 10 pages, 7 figures. Submitted to WACV 2027 Subjects: Image and Video Processing (eess.IV); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2608.22059 [eess.IV] (or arXiv:2608.22059v1 [eess.IV] for this version) https://doi.org/10.48550/arXiv.2608.22059 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-174] CHIMERA Challenge: Biochemical Recurrence Prediction in Prostate Cancer Patients using multimodal datasets MICCAI2025
链接: https://arxiv.org/abs/2608.21497
作者: Robert N. Spaans,Catherine Chia,Tongjie Wang,Adam Kowalewski,Parandzem Khachatryan,Domingos Oliveira,Khrystyna Faryna,Jean-Paul A. van Basten,Geert Litjens,Nadieh Khalili
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: 38 pages, 3 figures, 3 supplementary figures. Preprint submitted to Medical Image Analysis. Challenge results presented at the CHIMERA workshop, MICCAI 2025. Challenge website: this https URL
Abstract:Biochemical recurrence (BCR), defined as any detectable prostate-specific antigen level after prostatectomy with confirmatory elevation, is widely used as a surrogate endpoint and typically assessed using clinical and pathological variables. Currently, no standardized benchmark exists for multimodal prognostic modeling in urological cancers, partly because curating heterogeneous multimodal data remains challenging. We developed the CHIMERA Challenge, a multimodal benchmark integrating preoperative mpMRI, post-prostatectomy histopathology, patient characteristics, and clinician-derived variables from 267 patients across two institutions. The dataset comprises 801 MRI sequences, 13 clinical variables per case, and 942 WSIs. Training (n=95), validation (n=23), and test (n=149) splits were established and hosted on the Grand Challenge platform. Baseline clinical and pathological characteristics did not differ significantly across splits. Models were evaluated on predicting time to BCR using the C-index. Post-challenge analyses tested how each model type performed when clinician-derived variables were withheld or randomized. Unimodal clinical models achieved the highest test C-index of 0.7402 but proved sensitive to the integrity of these variables, with performance collapsing toward chance (C approximately 0.50) when they were randomized. Multimodal models retained near-baseline performance when these variables were withheld (delta C at most 0.04), indicating their ability to recover prognostic signal directly from imaging data. CHIMERA is the first public, standardized multimodal benchmark for prostate cancer prognosis. Although models using only patient characteristics and clinician-derived variables yielded the highest leaderboard performance, multimodal models demonstrated greater robustness in clinically realistic scenarios where complete expert annotation is not guaranteed.
[CV-175] MDFI: A Multi-Domain Features Integration for Compressed Video Quality Enhancement
链接: https://arxiv.org/abs/2608.21495
作者: Sang NguyenQuang,Hieu Bui Minh,Dang BuiDinh,Xiem HoangVan
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:The latest video coding standard, H.266/VVC, has demonstrated significant improvements in compression efficiency compared to H.265/HEVC. Despite its advanced coding techniques, H.266/VVC still faces challenges in meeting the increasing demand for higher perceptual quality and enhanced compression performance. To address these limitations, we propose MDFI (Multi-Domain Features Integration), a compressed video quality enhancement approach that features a novel Frame-Prediction Feature Transform (FPFT) module to process prediction information. Moreover, MDFI integrates a multi-domain feature fusion strategy that effectively combines spatiotemporal characteristics, cross-frequency representations, and compressed-domain prediction information to enhance decoded video quality. Additionally, we introduce a comprehensive dataset that encompasses uncompressed video sequences, corresponding reconstructed versions at multiple QP levels, and predicted frames generated from H.266/VVC compressed bitstreams, providing essential resources for developing and benchmarking video enhancement approaches. Extensive experiments demonstrate that our MDFI approach achieves superior performance to state-of-the-art methods in both objective metrics and visual quality, effectively mitigating video compression artifacts. The code is available at: this https URL.
[CV-176] Reliability- and Anatomy-Consistency-Aware Multimodal Learning for Robust Fracture Classification from Bangladeshi Radiographs
链接: https://arxiv.org/abs/2608.21482
作者: Musa Tur Farazi,K G Subarno Bithi
类目: Image and Video Processing (eess.IV); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Background: Multimodal fracture classifiers may benefit from patient and anatomical metadata, but they can also become brittle when contextual information is missing or mismatched. Methods: We studied 1493 radiographs from the Bangladeshi OrthoFrac-XR dataset using leakage-safe age, sex, bone type, and laterality. A ConvNeXt image encoder was combined with a clinical multilayer perceptron through concatenation, late fusion, reliability-gated residual fusion, and a hierarchical state-location formulation. We additionally introduced an anatomy-consistency gate that attenuates metadata corrections when an image-side anatomical prediction disagrees with the reported bone type. Results: Across five folds and three seeds, hierarchical residual fusion achieved a macro-F1 of 0.6046 +/- 0.0279, compared with 0.5727 +/- 0.0270 for image-only learning, while improving the Brier score from 0.5239 to 0.4948. In a five-fold robustness experiment, anatomy-consistency fusion reduced the macro-F1 loss under shuffled metadata from 0.0567 to 0.0203 relative to ordinary residual fusion, although its clean-data macro-F1 was lower. Without bone type at inference, auxiliary anatomy supervision improved macro-F1 from 0.5620 +/- 0.0330 to 0.5899 +/- 0.0289. Conclusions: Structured context improves fracture classification, and consistency-aware gating limits harm from mismatched metadata. The observed clean-performance-robustness trade-off and the absence of patient-level identifiers motivate external and prospective validation.
[CV-177] Multimodal pseudo-CT synthesis for PET attenuation correction using separate modality encoding and topogram conditioning
链接: https://arxiv.org/abs/2608.21481
作者: Rory Bell,Artemis Bouzaki,Jiaming Cao,Jasmine Morrison,Chelsea Sargeant
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Medical Physics (physics.med-ph)
备注: Technical report for the BIC-MAC 2026 Challenge
Abstract:We participated in the BIC-MAC Challenge with a multimodal 3D patch-based U-Net for pseudo-CT generation from NAC-PET, MRI, and 2D topograms. By using separate PET and MR encoders, multi-scale feature fusion, and FiLM-based topogram conditioning at the bottleneck, we obtain a model that integrates complementary cross-modal information while reducing reliance on precise voxel-wise correspondence between modalities. Our final submission can be found: this https URL
[CV-178] HiFiC-G: Adapting HiFiC for Hi-C Contact Matrices
链接: https://arxiv.org/abs/2608.21446
作者: Andre Antonio Straton
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Genomics (q-bio.GN)
备注: 13 pages, 3 figures, 2 tables. Bachelor’s thesis project, Transilvania University of Brasov (UNITBV). Language editing and translation assistance provided using Claude (Anthropic)
Abstract:We study whether the loss design of High-Fidelity Generative Image Compression (HiFiC), a GAN-based neural codec originally built for natural photographs, can be adapted to preserve biologically meaningful structure in Hi-C chromatin contact maps under lossy compression. Standard image compression, including HiFiC in its original form, optimizes for human visual perception; but a Hi-C contact map is normally distributed together with its numeric matrix file (.cool/.mcool), which downstream genomic analysis tools consume directly. Aggressive compression that looks acceptable to the eye can nonetheless blur or delete loops and topologically associating domain (TAD) boundaries that these tools depend on. We modify HiFiC’s distortion term with a spatially-weighted MSE that up-weights biologically salient regions (loops, TAD boundaries, stripes, compartment structure) and add an insulation-score loss term that directly penalizes loss of TAD boundary sharpness. We describe a three-phase fine-tuning strategy that adapts a pretrained HiFiC checkpoint to the Hi-C domain without catastrophic forgetting. We evaluate the resulting system, HiFiC-G, using both conventional image-quality metrics (PSNR, SSIM) and genomics-domain preservation metrics (loop/TAD/compartment/stripe preservation percentage) across two cell lines. HiFiC-G preserves local structure, meaning stripes and TAD boundaries, substantially better than the metrics alone would suggest, while long-range A/B compartment structure remains poorly preserved; we show this gap tracks genomic scale and is consistent with a specific architectural cause, the fixed-size tiling that both HiFiC-G and the original HiFiC rely on for memory efficiency.
[CV-179] Gate Voltage Effect on Pulse Detection Efficiency of Perimeter-Gated SPADs
链接: https://arxiv.org/abs/2608.21371
作者: Hunter Guthrie,Md Sakibur Sajal,Zexi Liu,Marc Dandin
类目: Instrumentation and Detectors (physics.ins-det); Computer Vision and Pattern Recognition (cs.CV)
备注: 4 pages, 7 figures, accepted in MWSCAS 2026 Conference
Abstract:Perimeter-gated single-photon avalanche diodes (pg-SPADs) are known for their dynamic dark noise modulation capabilities. They are reported to trade noise for photon sensitivity under continuous illumination. However, the implications of this trade-off have not heretofore been studied with pulsed optical systems. This work bridges this gap. We demonstrate that pg-SPADs fabricated in a 0.35 \mu m standard CMOS process trade-off pulse detection efficiency for a reduction in the the spread of spurious events within a burst window. Consequently, herein, we propose guidelines for the optimal use of pg-SPADs in pulsed LIDAR applications in view of the observed trade-off.
[CV-180] OmniAssistBench: Assistant-style Interaction Benchmark for Omni-LLM s
链接: https://arxiv.org/abs/2608.21360
作者: Xianyun Sun,Chaoyou Fu,Zhengye Zhang,Feiyang Duan,Qingyuan Cao,Yonghui Niu,Sihang Yuan,Ge Zhang,Caifeng Shan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL
Abstract:Recent omni-modal large language models (Omni-LLMs) show great potential as real-time video assistants, which continuously perceive environments and guide users to achieve specific goals. Unlike traditional passive video understanding, interactive assistants should actively combine visual states, user goals, and prior knowledge to provide effective help. Evaluating this is rather challenging, as the model’s unpredictable response dynamically changes the user’s subsequent actions, which static offline datasets cannot accommodate. To address this bottleneck, we introduce OmniAssistBench. To solve the issue of diverging interaction paths where the same user goal can be achieved through various methods, we provide models with predefined priors derived from the source video, requiring them to guide users along the exact same routes. Since real interaction videos are rare, we construct the dataset by reverse-engineering existing Internet videos. We deduce logical user goals and segment the videos into multi-turn clips to simulate continuous interactions. This rigorous pipeline required over 1000 expert person-hours to build the dataset. Results show that the proprietary Gemini-3-Pro reaches 66.4 out of the max point of 100, while the open-source Qwen3-Omni-Instruct achieves 51.2. Although current models generally understand user inputs, they frequently provide incorrect or incomplete answers. Specifically, they struggle with visual prompts (e.g., hand gestures), fail to maintain historical context during multi-turn interactions, and fail to delay response until the target event. Results indicate substantial room for improvement before models can become reliable assistants.
人工智能
[AI-0] ReWorld: An Interactive World Model with Long-Horizon Memory
链接: https://arxiv.org/abs/2608.23565
作者: Zhifei Chen,Luozhou Wang,Guibao Shen,Dongyu Yan,Shuai Yang,Tianshuo Xu,Yihua Du,Wei Wang,Tianyi Gui,Lianghua Huang,Yingcong Chen
类目: Artificial Intelligence (cs.AI)
备注: 21 pages, 9 figures. Project page: this https URL
Abstract:An interactive world model must follow the user’s actions, remember the places it has shown, and stream in real time. The tension is structural: control wants a short horizon, memory wants an unbounded one. ReWorld separates the two during training and bounds them at inference. Mixed per-head attention windows confine most heads to the recent past while a small set of global heads attends over the entire history, and random head routing keeps either capability from binding to particular heads; random chunk dropping makes sparse histories in-distribution. At inference the whole past lives under a fixed budget: a bounded KV cache backed by a pose-indexed landmark bank, from which the model retrieves the landmarks nearest the current pose. A metric-scale-aligned data engine places eight sources – Unreal-rendered fly-throughs, game roaming, and real-world footage – on one physical action scale, so the same key press moves the camera the same distance in every source, and palindrome trajectories supply the revisit evidence that memory training needs. Distribution-matching distillation confined to a LoRA adapter then compresses sampling to four steps: one backbone serves both a high-fidelity multi-step mode and a real-time interactive one, streaming 704x1280 video across photorealistic, game-style, and stylized worlds. Under a three-axis protocol covering action following, long-horizon recall, and video quality, against six recent interactive world models it attains the best control fidelity ( 11.95^\circ rotation error and the best camera-motion consistency) and the best generation quality; and on minute-long out-and-back rollouts ( 64 ,s, 384 latents), its fixed 12-chunk cache still regenerates the starting view – at rollout lengths where a sliding window has long evicted the evidence and full-KV attention runs out of memory.
[AI-1] How AI Assistance Affects Human Skill Development: A Study of Learning with Logic Puzzles
链接: https://arxiv.org/abs/2608.23543
作者: Shang Wu,Catarina G Belem,Shuyuan Fu,Mark Steyvers,Padhraic Smyth
类目: Artificial Intelligence (cs.AI)
备注: Accepted at Human-AI Complementarity and Alignment (HCOMP) 2026
Abstract:While AI assistance can improve human task performance in the short term, it may also undermine the development of skills in the longer term. We examine this tension in a controlled logic-puzzle experiment involving on-demand AI assistance, where participants complete tasks before, during, and after AI is available. By experimentally varying AI request costs, we find that lower-cost assistance induces more frequent AI use. We also find that participants who request AI assistance during the AI-access phase perform worse at the task after assistance is removed, and their subsequent unassisted performance is overestimated when predicted from earlier AI-assisted performance. We use a Bayesian latent ability model to separate initial ability, post-AI ability, and participant-specific skill change, while estimating how independent reasoning during the AI-access phase relates to skill development. The results show that greater independent problem-solving effort is associated with larger gains in latent ability, consistent with the interpretation that skill development is weaker when AI assistance substitutes for independent reasoning.
[AI-2] Adapter-Based Few-Shot Continual Learning for Malicious Packet Recognition
链接: https://arxiv.org/abs/2608.23536
作者: Kyle Stein,Guillermo Francia,III Eman El-Sheikh,Andrew Arash Mahyari
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:The continual evolution of malware variants necessitates detection systems that can adapt to new threats without retraining from scratch. However, continually updating models on new data often leads to catastrophic forgetting, where previously learned knowledge is overwritten. While continual learning has been increasingly explored for malware detection, the specific setting of Few-Shot Class-Incremental Learning (FSCIL), where new malware classes must be learned from only a small number of labeled examples, remains comparatively underexplored. Therefore, this work investigates the FSCIL setting for malware classification. To address the stability-plasticity dilemma, we propose a hybrid framework that leverages a Self-Supervised Learning (SSL) backbone initialized through domain-specific pre-training on malware packets. Our method incorporates Low-Rank Adaptation (LoRA) to efficiently adapt the model during the base session while freezing the core backbone to preserve previously learned representations, alongside a prototype-based classification head for incremental sessions to establish robust decision boundaries from limited samples. Extensive experiments across several datasets demonstrate that our approach consistently outperforms prior malware FSCIL baselines and achieves state-of-the-art performance.
[AI-3] Correcting a learned physical invariant improves world-model rollouts
链接: https://arxiv.org/abs/2608.23526
作者: Richard Bao
类目: Artificial Intelligence (cs.AI)
备注: 10 pages, 5 figures. Code at this https URL
Abstract:World models can predict video without learning dynamics that they reliably preserve. We test whether a frozen DreamerV3 trained only on pendulum video learns a scalar that its own latent transition treats as approximately conserved. A label-free search recovers the same energy-like invariant across independently trained conservative models, while the same procedure finds no comparable invariant in matched damped models. During autonomous rollouts, this quantity drifts. Projecting the latent state back toward its initial level set reduces rollout error in all three conservative models, whereas matched random constraints usually increase it. These results distinguish a dynamically meaningful invariant from a merely decodable correlate and reveal a concrete failure mode: a world model can learn a physical constraint from pixels yet violate that constraint when it imagines forward.
[AI-4] EarthVerse: Benchmarking Scientific Agents Across Dynamic Earth Systems and Natural Hazards
链接: https://arxiv.org/abs/2608.23525
作者: Zhiqing Cui,Xinxiang Yin,Yihong Tang,Xinglang Zhang,Yuanzhe Hu,Siru Zhong,Weidong Tang,Yuxuan Liang,Weijia Li,Ming Jin,Shirui Pan,Yuhao Kang,Dingyi Zhuang,Jinhua Zhao
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Earth-system analysis reconstructs changing physical processes from observations that differ in source, scale, timing, and modality. Natural hazards make this work consequential because incomplete evidence can change estimates of severity, exposure, and mechanism. We introduce EarthVerse, a benchmark that evaluates scientific agents through package-scoped investigations. Its 405 reproducible tasks are grounded in 199 documented events and 19 hazard families. Agents inspect heterogeneous event packages, choose compatible evidence, execute transparent calculations, reconcile source differences, and preserve provenance in the final answer. We provide executable ground truth that decomposes each task into fine-grained answer units, together with task-specific rubrics that assess the supporting research process while allowing multiple valid paths. We evaluate 25 model and agent systems under a controlled tool-using protocol, then use controlled studies to locate failures in evidence access, tool selection, memory, reasoning, interaction, and scientific execution. Across systems, the best mean answer-unit accuracy is 84.65%, while the highest Strict@95 is only 34.81%. The gap shows that current agents often complete individual steps without maintaining a consistent chain across evidence, scales, units, calculations, and physical interpretation. EarthVerse provides a reproducible basis for measuring end-to-end scientific reliability in dynamic Earth systems.
[AI-5] SRPO: Self-Reflective Policy Optimization for Long-Horizon Reasoning ICML2026
链接: https://arxiv.org/abs/2608.23493
作者: Jialong Liu,Yuling Shi,Ning Yang,Xiaodong Gu,Zuchao Li
类目: Artificial Intelligence (cs.AI)
备注: Accepted to ICML 2026
Abstract:Self-reflection is a powerful mechanism for credit assignment in human learning, converting sparse outcome feedback into actionable guidance. However, its potential for post-training Large Language Models (LLMs) remains underexplored. We propose Self-Reflective Policy Optimization (SRPO), a framework that internalizes this capability. SRPO enables LLMs to analyze their own completed trajectories, synthesize errors into concise “reflection patches,” and use reflection-conditioned teacher scores on student on-policy rollouts as dense token-level training signals. This process effectively transforms sparse terminal supervision into dense, token-level learning signals without requiring external critics, separate reward models, or larger teacher models. We demonstrate that SRPO achieves state-of-the-art performance across mathematical reasoning and long-horizon agentic benchmarks with exceptional data efficiency. Using a Qwen3-8B base model, SRPO attains 73.3% on AIME’24 using only 8% (0.08x) of the training FLOPs required by scaled supervised fine-tuning, while significantly improving success rates on WebShop (64.7%), ALFWorld (76.8%), and SWE-Bench-Lite (31.2%). Code is available at this https URL
[AI-6] Multi-Modal Semantic Expansion with Constrained LLM Reranking for Conversational Music Recommendation
链接: https://arxiv.org/abs/2608.23484
作者: Naman Garg,Sarika Jain,George Fazekas
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:We present Team Semiintelligencn’s solution for the ACM RecSys 2026 TalkPlayData Challenge, addressing conversational music recommendation through a multi-modal and personalized conversational recommender system. Our submitted system employs a three-stage pipeline: (1) multi-modal retrieval constructing decay-weighted centroids across seven dense embedding spaces - track- and user-level CF-BPR, Qwen3 (metadata, lyrics, attributes), CLAP audio, and SigLIP visual - supplemented by BM25 lexical retrieval and an artist substring-match signal, all fused via weighted Reciprocal Rank Fusion (RRF) with optimized signal weights; (2) lightweight reranking (history filtering, popularity smoothing, and catalog diversity penalization); and (3) persona-diversified response generation using GPT-4o-mini. Beyond this submitted configuration, we report development-time experiments with additional components - constrained LLM-guided artist injection, album continuation signals, XGBoost LambdaMART, and a superior GPT-4.1 response prompt - that were not deployed to Blind B due to cost and complexity constraints. We optimize RRF weights on a 500-session development split via differential evolution, improving MRR by +19.5%. On Blind A, we observe that unconstrained LLM-guided injection across 54 sessions causes catastrophic nDCG regression (-18.9%), while conservative injection on only 9 sessions yields the best observed Blind A nDCG - a finding we present as a Blind A observation warranting further validation. The submitted system achieves a Blind B composite score of 0.3213.
[AI-7] StrategyBench: Evaluating Explicit Strategy Induction in Large Language Models
链接: https://arxiv.org/abs/2608.23475
作者: Jinghan Tan,Yuanzheng Wang,Lu Chen,Zijun Chen,Yuqian Wang,Maosong Sun
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:As large language models are increasingly used in data-scarce and evolving task scenarios, few-shot in-context learning (ICL) has become a key paradigm for task adaptation. However, direct ICL often uses a small set of examples without explicitly abstracting task rules, making it sensitive to example construction. In contrast, human learners often reduce such sensitivity by first summarizing task rules from examples and then applying them to new instances. To evaluate this ability, we propose StrategyBench, which selects strategy-inducible tasks from BIG-Bench, constructs reference strategies, and defines evaluation metrics along two dimensions: strategy quality and downstream utility. We further analyze strategy induction from three perspectives: task variation, model configuration, and adaptation setting, covering category-wise differences, generator-executor choices, demonstration design, and SFT-based adaptation. Experiments show that explicit strategy utility differs substantially across task categories and depends on both strategy generation and execution conditions. The benchmark is released at: this https URL.
[AI-8] MetaCaster: Meta-Harness-Optimized Agent for End-to-End Few-Shot Learning of Lightweight Time Series Forecasters EMNLP2026
链接: https://arxiv.org/abs/2608.23473
作者: ChengAo Shen,Wenchao Yu,Fangyu Wu,Dongjin Song,Hanghang Tong,Dongsheng Luo,Wei Cheng,Haifeng Chen,Jingchao Ni
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted by EMNLP 2026
Abstract:Time series forecasting (TSF) is evolving toward multimodal and agentic settings, yet using foundation models remains uneconomical in resource-constrained scenarios, where compact, specialized forecasters are more desirable. However, lightweight forecasters typically require substantial training data, limiting their use in domains with scarce, slowly accumulated, or privacy-sensitive time series. To address this dilemma, we investigate the challenging problem of few-shot learning for lightweight forecasters. We propose MetaCaster, a meta-harness-optimized multi-agent framework that uses agentic data generation to automatically train specialized lightweight forecasters from only a few examples and textual contexts. Our work highlights a new TSF paradigm in which agents act not as forecasters but as intermediary engineers that prepare efficient, task-specific forecasters for deployment. Experiments on 18 datasets, 23 state-of-the-art lightweight forecasters, and 14 baselines demonstrate that MetaCaster achieves both data efficiency and computational efficiency while maintaining high-quality TSF performance.
[AI-9] InjecMEM: Memory Injection Attack on LLM Agent Memory Systems
链接: https://arxiv.org/abs/2608.23471
作者: Hanling Tian,Gengyu Zhang,Zeyang Sha,Jingying Wang,Yuhang Liu,Zhehao Huang,Kun Yang,Xiaolin Huang
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: 29 pages, 3 figures. Accepted at COLM 2026
Abstract:Memory is becoming a default subsystem in deployed LLM agents to provide persistent personalization and continuity. This naturally prompts a question: will memory system introduce new vulnerabilities into agents? Thus we propose InjecMEM, a novel memory injection attack paradigm that requires only a single interaction (no read/edit access to memory store) to steer later responses of related queries toward a pre-specified output. Guided by the retrieval-then-generate mechanism of memory systems, we craft the injection with a retriever-agnostic anchor and an adversarial command. The anchor contains high-recall topical cues so that downstream retrieval consistently associates the record with the target topic. The command is a short sequence optimized to remain effective under uncertain fused contexts, variable placements, and long prompts so that it reliably steers outputs once retrieved. We learn the command via gradient-based coordinate search, averaging over synthetic prompt templates and insertion positions, and extend it to joint optimization across backbones to study transfer. Evaluated across multiple memory systems and backbone models, InjecMEM achieves reliable topic-conditioned retrieval and targeted generation, remains effective under memory drift, and leaves non-target queries unaffected. Our results underscore the need to harden memory systems and provide a reproducible framework for studying agent memory.
[AI-10] Reward-Free Continual Adaptation for Resilient Space Robots
链接: https://arxiv.org/abs/2608.23452
作者: Andrej Orsula,Miguel Olivares-Mendez,Carol Martinez
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted for publication at the Third Conference on AI in and for Space (SPAICE 2026) | The source code is available at this https URL
Abstract:Space robots operate in extreme environments where hardware degradation can critically compromise traditional control strategies. While continual reinforcement learning offers a promising mechanism for online adaptation, it inherently requires access to a reward signal during deployment. However, precise reward computation in space is often infeasible due to the lack of external tracking systems and the overall complexity of the environment. To address the challenge of unobservable rewards, we introduce a reward-free continual learning framework that leverages latent-state world models. By pre-training a model-based agent across diverse simulations, the world model learns a robust predictor of the reward structure within its latent space. Upon deployment to an environment with severe hardware degradation, we freeze the observation encoder and reward predictor to update only the transition dynamics of the world model through unsupervised rollouts. By training the policy entirely on imagined trajectories generated by this updated world model, the agent adapts to altered dynamics without receiving new rewards. We demonstrate our approach across simulated planetary traversal, orbital navigation, and precision assembly tasks subjected to severe morphological failures.
[AI-11] Characterizing Necessary Losers to Explain Tournaments Losers
链接: https://arxiv.org/abs/2608.23446
作者: Contet Clément,Umberto Grandi,Jérôme Mengin
类目: Artificial Intelligence (cs.AI)
备注: This paper is the extended version of Contet, Grandi, Mengin. Characterizing Necessary Losers to Explain Tournaments Losers. In: Proceedings of the 9th International Conference on Algorithmic Decision Theory (ADT) (2026)
Abstract:We study the problem of formally explaining why a candidate was not selected by a given tournament rule, by identifying sub-tournaments in which the candidate loses independently of how the rest of the tournament is completed. We define destructive minimal supports as any minimal sub-tournaments satisfying this property, which in formal explainable artificial intelligence correspond to abductive explanations for the question “Why does the loser lose the tournament?”. For six common tournament solutions (maximin, uncovered set and its weighted variant, top-cycle, Copeland, and Borda) we provide characterizations of when a candidate is either a necessary loser or a possible winner, we determine the size of the smallest destructive minimal supports, complemented by polynomial-time algorithms for their computation except for the case of the Borda rule which is suspected to be NP-complete.
[AI-12] ChebBooster: A Training-Free Approach for Efficient Diffusion Transformer Inference via Chebyshev-Inspired Extrapolation
链接: https://arxiv.org/abs/2608.23429
作者: Chengjie Lu,Tianchi Deng,Zhengqi He,Chengwen Luo,Xueliang Li
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Diffusion Transformers (DiTs) have shown strong performance in high-fidelity image generation, but their sampling process remains computationally intensive due to full model execution at every timestep. While cache-based acceleration has been explored to mitigate inference cost, naive reuse schemes suffer from low accuracy over long intervals, and Taylor-series-based extrapolation methods often face instability caused by Runge oscillations. In this paper, we propose ChebBooster, a training-free extrapolation framework based on Chebyshev polynomial theory that achieves stable and efficient acceleration for DiTs. Specifically, we adopt the Barycentric formulation to evaluate Chebyshev approximants with high numerical stability and minimal overhead, and further decouple the extrapolation into an offline weight precomputation phase and a lightweight online application stage. Extensive experiments across three representative DiT-based models, including DiT-XL/2, PixArt- \Sigma , and FLUX.1-dev, demonstrate that ChebBooster achieves consistent improvements in visual quality and inference efficiency, reaching up to 3.68\times latency speedup and 5.12\times FLOPs reduction, outperforming existing training-free baselines under diverse generation tasks and resolutions.
[AI-13] SkillAlchemy: Open-World Agent Skill Creation
链接: https://arxiv.org/abs/2608.23417
作者: Hengjun Wang,Shuyue Wei,Boyi Liu,Jun Yang,Yongxin Tong
类目: Artificial Intelligence (cs.AI)
备注: 33 pages, 7 figures, 8 tables. Includes appendices
Abstract:Agent skills are reusable procedural artifacts that extend language agents with specialized workflows, tool conventions, and domain behaviors at inference time. However, creating reliable skills still depends largely on human authorship, model priors, or execution traces. These sources are often unavailable for unfamiliar tasks, suggesting the need to create skills from open-world materials. In this paper, we study open-world skill creation: given an underspecified skill brief and a source-access specification, a creator must discover behavior-relevant requirements omitted by the brief and determine how broadly each source-derived procedure is justified. We propose SkillAlchemy, an admission-centered framework for source-grounded skill creation. SkillAlchemy identifies implicit requirements through contrastive evidence, admits candidate procedures based on evidence-supported scope, and compiles the admitted content into a grammar-guided skill package. Extensive experiments across 87 SkillsBench v1.1 tasks demonstrate that SkillAlchemy improves pass rate over no-skill execution by 19.9 percentage points and the strongest automated baseline by 8.6 percentage points, while achieving performance comparable to human-curated skills.
[AI-14] MediSkill-Evo: Process-Constrained Self-Evolution for Evidence-Grounded Clinical Interaction
链接: https://arxiv.org/abs/2608.23397
作者: Ruoyu Wu,Shenfu Xie,Yinqian Sun,Haibo Tong,Feifei Zhao
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Interactive clinical agents must gather decisive evidence and convert it into grounded actions under partial observability. A correct final diagnosis alone does not show that an agent respected evidence and care-process constraints. We introduce MediSkill-Evo, a clinical agent that evolves governed process knowledge without backbone fine-tuning. It separates experience into four typed banks for clinical skills, process rules, symbolic schemas, and measurement procedures. Provenance, support, replay, and controller-defined safety checks govern publication to a frozen test-time snapshot. A Process-Constrained Preference Harness binds evidence to its source, rejects controller-invalid candidates, and ranks actions with a safety-prioritized Clinical Process Critic. We evaluate complete agent systems across two backbone endpoints and six controlled stress dimensions under the same Doctor-turn limit. On 300 held-out Qwen encounters, MediSkill-Evo improves diagnosis accuracy from 61.33 percent to 69.00 percent and treatment-intent coverage from 33.62 percent to 66.44 percent, while reducing automatically scored critical failures from 31.00 percent to 16.33 percent relative to AgentClinic. On 180 hard-isolation conditions derived from 30 cases, target recovery reaches 93.61 percent under patient-behavior pressure, 100.00 percent for temporal evidence, and 92.22 percent for triage red flags. An exploratory 100-case MedSAM comparison evaluates request-gated tool-interface feasibility. These results provide descriptive end-to-end evidence for the complete system on fixed evaluation suites, not causal evidence for an individual bank or clinical validation of the automatic judge.
[AI-15] Adversarial Entropy Inflation Against Gumbel-Based Inference Verification
链接: https://arxiv.org/abs/2608.23375
作者: Nikita Kezins
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: 4 pages, 1 figure, 1 table
Abstract:Gumbel-based inference verification bounds LLM weight exfiltration by only forgiving token choices that plausibly arise from honest GPU nondeterminism, reporting a 200x slowdown for a steganographic adversary under benign prompt traffic. This bound assumes a passive attacker; we show it degrades sharply against an adversary who instead controls the prompt distribution. Because the verifier’s admissible-token-set size is driven by the model’s own output entropy, prompts engineered to break grammatical and sub-word structure – rather than benign conversational traffic – widen that set and open a materially larger covert channel. Across six instruction-tuned models spanning 1B to 32B parameters and three random seeds, our strongest attack (character- and script-level disruption) roughly doubles bits leaked per token relative to benign prompts, cutting the slowdown factor to 60x - 118x. These results indicate that static, benign-traffic-calibrated thresholds are insufficient for this defense, and that jitter-forgiveness thresholds should instead be calibrated dynamically against local token entropy.
[AI-16] Modalities Should Talk to Each Other: Dual-Stream Multimodal Learning for Long-Horizon Influenza Forecasting
链接: https://arxiv.org/abs/2608.23373
作者: Seyed Mohammad Hossein Hashemi,Mohsen Hooshmand,Parvin Razzaghi
类目: Artificial Intelligence (cs.AI); Applications (stat.AP)
备注:
Abstract:Forecasting long-range influenza-like illness (ILI) matters for public health readiness. Publicly available surveillance datasets typically pair numeric epidemiological signals with textual information that is noisy, loosely structured, only indirectly related to near-term trends, and often lagged relative to the numeric signal. Fusing the two therefore requires careful design. We propose Dual-Stream Attention (DSA), a multimodal deep learning framework that forecasts 12-week-ahead ILI activity from a 36-week multimodal history by letting the numerical and textual streams condition each other. Using the Time-MMD health-domain dataset, DSA separately encodes the two modalities with a Transformer-based numerical encoder and a domain-adapted headline encoder, then couples them through a bidirectional Cross-Modal Attention (CMA) mechanism: the text (news headlines) conditions the interpretation of the numeric signal and vice versa. The CMA output then passes to a causal temporal model for forecasting. Evaluated across ten random seeds, DSA achieves a median test MSE of 0.416, versus 0.668, 0.607, and 0.851 for iTransformer, TaTS, and GPT4MTS, corresponding to mean-error reductions of 54.95%, 37.29%, and 67.23%, with paired Cohen’s d of 0.555, 0.337, and 0.345, respectively, and ranks first in 100% of bootstrap draws. It also has substantially lower worst-window error than all baselines. On an external-geography dataset, DSA again ranks first among nine evaluated baselines. Ablations show the advantage does not depend on text-encoder choice or language-model fine-tuning, and that bidirectional attention outperforms either direction alone. Finally, perturbation-based faithfulness analysis shows the learned CMA is functionally informative under targeted masking, with a stronger effect in the text-to-numerical direction.
[AI-17] Walking on the DARKSIDE
链接: https://arxiv.org/abs/2608.23370
作者: Aldo Gangemi,Emanuele Bottazzi
类目: Artificial Intelligence (cs.AI); Logic in Computer Science (cs.LO)
备注: 20 pages, 2 figures, several tables
Abstract:Large Language Models (LLMs) recognise patterns but do not natively track the path of exclusions that a coherent discourse demands. When an input rests on a fabricated authority, a misapplied mechanism, or a surreptitious analogy, an unsteered LLM tends to engage with it as if it were grounded, and to reify the misstep into any structured output it generates. Logic-Augmented Generation (LAG) with POLANYI++, an LLM-steering method that uses heuristics, ontologies and problem solving methods for tacit knowledge extraction, produces an Extended Knowledge Graph (XKG) in OWL2, but inherits the same vulnerability: a sophisticated nonsensical input is reified into the graph alongside the legitimate triples, and is hardly detectable by automated reasoners since the XKG is generated jointly with the wrong assumptions. We introduce DARKSIDE, a coherence auditing method on top of POLANYI++. It formalises the trail as an explicit data structure of accumulated exclusions over discourse time, complemented by a warrant axis that classifies each named referent as Warranted, Unattested, Misattributed or Fabricated, with an escalation rule that pushes the DelegationRiskAssessment to UNSAFE when the fabricated rate is positive or the unsupported rate exceeds a threshold. We evaluate DARKSIDE as a steering layer over a Gemini 3 on BSBench, a 100-item adversarial corpus of sophisticated-sounding nonsense across software engineering, finance, healthcare, physics and law, with Claude Sonnet 4.6 as an independent judge. The empirical evidence supports an architectural claim: when an LLM forward pass is wrapped in an ontology-mediated negative-trail apparatus, the structural pattern-vs-path gap can be partially scaffolded. The XKG functions as the missing memory, and the warrant axis as an epistemic firewall.
[AI-18] Mycelial Search: A Graph-Structured Metaheuristic for Continuous Optimisation
链接: https://arxiv.org/abs/2608.23323
作者: Mohammad Mahdi Dehshibi
类目: Neural and Evolutionary Computing (cs.NE); Artificial Intelligence (cs.AI)
备注: Submitted to “Expert Systems With Applications”; 18 pages, 6 figures, 6 tables. To facilitate reproducibility, the Python implementation of Mycelial Search (Myco) is publicly available on [GitHub]( this https URL )
Abstract:Continuous optimisation methods need to balance sharing information and maintaining alternative search directions. In this paper, we introduce Mycelial Search (Myco), a graph-structured metaheuristic designed around active tips, community-weighted flow, adaptive cord plasticity, and anchor-based injection. Candidate solutions form an evolving spatial graph in which a Louvain partition distinguishes within-community from cross-community information exchange. Adaptive cord plasticity subsequently modifies active tip-to-tip edges according to their alignment with the local flow. An anchor-based injection mechanism supplements the graph-driven tip dynamics. We evaluated Myco on the CEC 2022 single-objective bound-constrained benchmark suite at dimensions D=10 and D=20 , using 30 independent runs per algorithm-function pair. The comparison includes eleven established optimisers from several search families. Myco reaches competitive results on selected functions across both dimensions. The ablation analysis further shows that community structure regulates the range of graph-based information exchange, whereas cord plasticity controls the persistence of local directional influence. These findings indicate that graph-structured local interaction can support continuous optimisation, while its effectiveness depends on landscape structure and information transfer across local search regions.
[AI-19] EviSafe: Evidence-Grounded Safety Evaluation for Vision-Language Models
链接: https://arxiv.org/abs/2608.23313
作者: Xuetong Li,Gaofeng Liu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Vision-language model safety benchmarks typically evaluate only final responses: whether a model refuses, warns, or complies. This outcome-level view cannot tell whether a model is safe for the right multimodal reason. Safelooking behavior may reflect keyword-triggered refusal, missed visual hazards, or over-refusal of benign-sensitive inputs. We introduce EviSafe, an evidence-grounded framework for VLM safety that jointly evaluates natural user-facing behavior, explicit grounding in textual and visual evidence, and behavioral sensitivity to counterfactual changes in safety-critical evidence. EviSafeBench instantiates the framework as a controlled benchmark with 1,181 gold image-text scenarios and 2,452 targeted counterfactual variants across eight safety domains and eight risk-source types. Each scenario includes a gold safety decision, evidence annotations, a safe-response policy, and counterfactual interventions. The three-probe protocol queries models with natural-response, evidencereporting, and counterfactual-response prompts, then scores them using an evidence-aware judge. Across eleven evaluated VLMs, natural severity accuracy ranges from 27.6% to 52.8%, relaxed diagnostic consistency from 6.1% to 29.3%, and unsafe-to-safe counterfactual transition success from 30.4% to 58.4%. These gaps show that the evaluated VLMs are not reliably safe for the right multimodal reason and motivate evaluation beyond refusal counts.
[AI-20] FIDES: A Concordance Protocol for LLM -Generated Trading Strategies
链接: https://arxiv.org/abs/2608.23308
作者: Arther Tian,Alex Ding,Simon Wu,Aaron Chan
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:An LLM asked for a trading strategy returns three artifacts at once: a natural-language rationale, an executable implementation, and once run, a track record. Whether these are the same object is rarely checked. We present FIDES, a measurement protocol that treats them as three views to be reconciled rather than one deliverable to be graded. Through dual delivery, a single model call returns both a natural-language strategy with an explicit claimed edge and a self-contained strategy(df) function. FIDES executes the code in a sandbox against a lag-one out-of-sample backtest and scores three concordance gaps: say to do, do to real, and say to result. On 8 liquid US ETFs across four models plus a two-stage elicitation arm, 40 strategies, 2023 to 2024 out-of-sample, three findings stand out. First, concordance does not predict profit: only 2 of 40 strategies beat buy-and-hold, and a plain sma(50,200) rule outperforms every model’s mean Sharpe. Second, self-assessment is badly calibrated: 32 of 40 strategies claim to beat buy-and-hold and exactly one does. Third, swapping the language-code judge for a second model flips say to do on more than half of items. Injecting this http URL(-1) drops do to real by 0.33 on average, while our runtime future-information probe fired on neither clean nor injected code. We frame FIDES as a protocol for measurement fidelity, not a claim about market performance. Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI) Cite as: arXiv:2608.23308 [cs.CR] (or arXiv:2608.23308v1 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2608.23308 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-21] Sigmoid Attention as a Better Substrate for Learned KV Cache Eviction ICML2026
链接: https://arxiv.org/abs/2608.23296
作者: Isaac(Rucheng)Li
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted at the ICML 2026 Workshop on Resource-Adaptive Foundation Model Inference (AdaptFM)
Abstract:Learned KV-cache eviction often faces a soft-to-hard mismatch: during training, differentiable gates typically attenuate token contributions, whereas inference saves memory only when KV entries are physically removed. We ask whether the attention substrate affects this soft-to-hard transition. Using GPT-2-scale Transformers trained on OpenWebText, we run a controlled 2\times2\times2 comparison over attention type, learned gating, and positional encoding. Although sigmoid attention is worse as a dense language model, learned hard eviction changes the useful operating points: sigmoid-gated models delete KV entries with negligible PPL change relative to their own no-eviction references. Under a matched live-cache protocol on the same dense backbones, learned sigmoid gates obtain lower PPL than our H _2 O and KeyDiff implementations, whereas softmax gates do not uniformly beat these post-hoc methods. The results suggest that attention normalization can substantially affect whether a training-time soft gate transfers cleanly to hard KV deletion.
[AI-22] How Much Regularization Survives Averag ing? Update Masking in Federated Learning
链接: https://arxiv.org/abs/2608.23286
作者: Wenhao Yan,Fu Kuroda,Yucheng Jin,Zhenke Chen
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Federated learning on non-IID data seeks flat minima to generalize across clients, and existing methods borrow sharpness-aware minimization from centralized training. There is a second way to reach flat minima, in which the regularization comes for free from noise added to the parameter updates, and it has never been carried over to the federated setting. We show the reason. Masking charges the optimizer for moving in sharp directions. We prove that when each client draws its own mask, federated averaging weakens that charge by exactly the cohort size, and that giving every client the same mask brings it back by a factor equal to the inverse gradient diversity of the cohort. In our experiment setting on CIFAR-10, that factor is 1.19 out of a possible 10. Turning off minibatch sampling raises it to 8.96, while changing data heterogeneity a hundredfold leaves it between 1.17 and 1.50. The configurations keeping the regularization train far too poorly to use.
[AI-23] Is Next-Chunk Reasoning RL Really Better than SFT? Revisiting Training Strategies under no-CoT Data
链接: https://arxiv.org/abs/2608.23256
作者: Yinhao Tang,Youqing Fang,Yanan Sun,Jiangning Liu,Ziyi Wang,Xun Zhao,Weiming Zhang,Bin Liu,Kuikun Liu,Wenwei Zhang,Kai Chen
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Recent work proposes next-chunk reasoning RL for leveraging no-CoT data—corpora such as worked solutions and textbook derivations that contain reasoning-rich content but lack explicit chain-of-thought annotations. The method trains a model to generate implicit reasoning traces and rewards them by their ability to predict the next chunk of text. While promising, existing evaluations primarily compare against conventional SFT baselines, leaving open whether the gains come from the RL formulation itself or from more effectively exposing the model to no-CoT data. We address this question with a controlled study of next-chunk reasoning RL and a simple but previously overlooked alternative: Mixed SFT, a single supervised fine-tuning stage that jointly trains on no-CoT and long-CoT data. Despite its simplicity, Mixed SFT achieves a clearly higher post-RLVR performance ceiling than next-chunk reasoning RL while requiring over 60 times less training compute. The advantage is consistent across in-domain mathematical reasoning and out-of-domain reasoning tasks. Moreover, we show that higher pre-RLVR accuracy does not necessarily translate into higher post-RLVR accuracy, highlighting the need to evaluate no-CoT training strategies in the context of the full post-training pipeline.
[AI-24] Multi-Winner Voting with Argumentative Ballots
链接: https://arxiv.org/abs/2608.23247
作者: Ryuta Arisaka,Hirotaka Ono
类目: Computer Science and Game Theory (cs.GT); Artificial Intelligence (cs.AI)
备注:
Abstract:We introduce multi-winner voting with argumentative ballots (MVArg) and investigate theoretical properties. As our conceptual contribution, we generalise approval ballots to argumentative ballots, thereby allowing voters to express defeasible preferences over candidates. We accordingly generalise voter cohesion and justified representation axioms JR, PJR and EJR. As our theoretical contribution, we establish several key results. First, MVArg is strictly more expressive than multi-winner voting with approval ballots (MV). Second, our notions of cohesion and justified representation are conservative generalisations of their counterparts in MV. Third, the MVArg counterpart of JR can always be satisfied, whereas the counterparts of PJR and EJR cannot always be. Fourth, although verifying whether a winner set satisfies the MVArg counterpart of JR is already coNP-hard, such a winner set can be constructed in polynomial time. All definitions, propositions, auxiliary lemmas and theorems have been formalised and mechanically checked in Lean 4.
[AI-25] What is mathematics now and what should it be?
链接: https://arxiv.org/abs/2608.23218
作者: Jeremy Avigad
类目: Artificial Intelligence (cs.AI); History and Overview (math.HO)
备注:
Abstract:Advances in neural theorem provers have been impressive, but the successes obscure a broader vision of what AI can do for mathematics and how mathematicians can engage with AI. This essay advances a more expansive and optimistic point of view.
[AI-26] NetConfArena: An Executable Benchmark for LLM Agents in Closed-Loop Network Configuration
链接: https://arxiv.org/abs/2608.23179
作者: Chang Liu,Xiaohui Xie,Xinyi Chen,Yong Cui
类目: Networking and Internet Architecture (cs.NI); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language model (LLM) agents are increasingly attractive for automating network configuration, yet their reliability and failure patterns are poorly understood. An essential prerequisite is to assess such agents in a realistic but risk-free environment. Existing benchmarks, however, fall short: they often treat configuration as static command generation or rely on overly simplified settings. Such evaluations understate the core challenges of network configuration, where correctness requires reasoning about protocol complexity and topology dependence. We present NetConfArena, an executable benchmark for evaluating LLM agents in closed-loop network configuration. NetConfArena places agents in emulated multi-device networks, provides a standardized and compact action interface for task execution, and evaluates the resulting network behavior with hidden task-specific executable test cases. The benchmark relies on an LLM-assisted, emulation-grounded pipeline, which converts human-oriented network materials into reusable parameterized task templates. We evaluate representative LLM agents on 480 task instances instantiated from 96 protocol-focused task templates, yielding 3840 execution trajectories, and show that failures are not limited to command errors. The failures also reveal gaps in task-specification adherence and robust planning and execution. These findings suggest two future directions: using validated trajectories as supervision signals to improve foundation models, and designing harness mechanisms that make agent execution more reliable and accountable.
[AI-27] Counterfactual Transition Graphs: Evaluating Cross-Class Transition Quality ECML-PKDD2026 KDD
链接: https://arxiv.org/abs/2608.23164
作者: Syed Muhammad Hamza Zaidi,Szymon Bobek,Grzegorz J. Nalepa,Myra Spiliopoulou
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted at XKDD @ ECML-PKDD 2026 (Naples, Italy)
Abstract:Counterfactual (CF) explanations for time-series classifiers are usually evaluated one example at a time: what minimal edit flips this single window’s prediction? We argue that the more informative question for diagnostic interpretability is structural: how does the classifier connect its own classes to each other? We propose a counterfactual transition graph (CGT) in which each node is a class and each edge weight is the CF reliability of the transition from one prototype to another under a proximity aware retrieval sweep. On a six-class hand-movement task, we induce a CGT that reveals a non-trivial topology, which is not predicted by the binary confusion matrix: it shows that counterfactual reachability does not align with classifier accuracy and even runs counter to it (Spearman \rho=-0.37 over the 15 pairs), i.e. the boundaries the classifier separates most confidently are among those an in-distribution edit can least often cross. Our framework is method agnostic, i.e. any CF-explainers can be used. Presently, we use it to juxtapose replacement-based CFs with gradient-based CFs; gradient-based methods reach almost any class by stepping off the data manifold, while replacement-based methods stay on it and fail on precisely the rigid boundaries. Comments: Accepted at XKDD @ ECML-PKDD 2026 (Naples, Italy) Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI) Cite as: arXiv:2608.23164 [cs.LG] (or arXiv:2608.23164v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2608.23164 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-28] DeMixPert: Decomposed Response Modeling with Gaussian Mixtures for OOD Single-Cell Perturbation Prediction
链接: https://arxiv.org/abs/2608.23114
作者: Jiawen Liu,Xuechenxiao Cao,Yutong Li,Bing Liu,Jiaming Liang,Tinghe Zhang,Xiaoqi Sheng,Hongmin Cai
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Predicting transcriptome-wide responses to unseen genetic perturbations remains a major computational challenge because accurate prediction requires recovering both perturbation-specific transcriptional shifts and heterogeneous cellular responses. Existing methods often entangle deterministic response structure with stochastic population-level variation, causing dominant shared patterns to mask weaker perturbation-specific signals and impair distributional modeling. To address these challenges, we propose \textbfDeMixPert, an approach for Decomposed response Modeling with Gaussian Mixtures for Out-Of-Distribution (OOD) single-cell Perturbation prediction. DeMixPert decomposes perturbation-induced changes into a basal-state-dependent systematic response, a perturbation-specific response, and population-level variation. The systematic component is derived from the basal state encoded from control-cell expression, whereas the perturbation-specific component is inferred from pretrained target embeddings for unseen-target generalization. DeMixPert models population-level variation using a Gaussian prototype Invertible Network and adaptively combines reusable Gaussian prototypes according to the basal state and perturbation condition. The resulting mixture is mapped to a condition-specific variation distribution. Sampled variations are integrated with the systematic and perturbation-specific components, followed by joint decoding with the basal state to reconstruct perturbed-cell gene expression. Experimental results show that DeMixPert effectively captures heterogeneous single-cell perturbation responses and achieves superior performance across unseen-perturbation settings. The source code is made publicly available upon publication.
[AI-29] PolyChirp: Multi-Species Birdsong Classification Using TinyML on Low-Power Acoustic Sensors
链接: https://arxiv.org/abs/2608.23101
作者: Nathan Duboisset,Zhaolan Huang,Felix Bießmann,Roudy Dagher,Antoine Lavandier,Emmanuel Baccelli
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Recent progress in the field of TinyML has demonstrated that low-power hardware based on microcontrollers can achieve bird species monitoring in real time based on acoustic sensor data for an entire breeding period on a single battery charge. However, the state of the art on low-power microcontrollers was so far limited to binary classification of a single species. In contrast, real fauna monitoring deployments often target multiple species simultaneously. To address this challenge we develop PolyChirp, an approach combining biological domain expertise, automated dataset curation, neural architecture optimization and novel hardware to achieve multiclass bird species detection in the wild. PolyChirp is based on newly designed tiny multiclass models that leverage recent microcontrollers and hardware acceleration with a neural processing unit (NPU). We evaluate the predictive performance of these models, and we measure their computational performance – memory footprint, latency, energy consumption – on common microcontroller hardware. Our results demonstrate that PolyChirp not only outperforms state-of-the-art on single species binary classification, but also achieves robust classification of up to 10 species simultaneously, while still fitting with the resource envelope of a sensor that must remain operational in the field for a full season on a single battery charge.
[AI-30] Shaping the Evolutionary Dynamics of Robot Morphology via Adaptive Control Learning
链接: https://arxiv.org/abs/2608.23100
作者: Junru Song,Yang Yang,Yaqing Xu,Ying Wen,Wei Peng,Guozhen Li,Wei’en Zhou,Wen Yao
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Robot co-design via bi-level optimization couples within-lifetime controller learning for fitness evaluation with cross-generational morphological evolution. Prior work has established that well-adapted morphology facilitates faster control learning, a property termed morphological intelligence. Yet how control learning reciprocally shapes morphological evolution remains unexplored. This paper examines both directions for a holistic account of brain-body interplay. We first show that morphological contributions to control learning decouple into two orthogonal dimensions. We formalize the convergence speed as morphological intelligence and identify the performance ceiling as a complementary quantity termed true potential. A concise functional relation is then established to jointly characterize both quantities from individual learning curves, which, when aggregated at the population level, capture evolutionary profiles. Through extensive experiments on simulated voxel-based soft robots, we reveal that premature fitness evaluation systematically underestimates true potential and biases selection towards fast learners. This restricts design space exploration, compromising both optimization efficiency and morphological diversity. Notably, the widely recognized morphological Baldwin effect emerges as an artifact of this bias rather than a general evolutionary tendency. We therefore propose AdaControl, which monitors disproportionate selection for morphological intelligence during evolution and allocates minimally sufficient control learning for unbiased fitness evaluation. With AdaControl, a simple genetic algorithm rivals state-of-the-art generative-model-based co-design methods in discovering diverse high-performing designs while cutting computation by up to 80% versus exhaustive control.
[AI-31] Jiuge-Tuiqiao: An Interpretable Human-AI System for Classical Chinese Poetry Refinement EMNLP2026
链接: https://arxiv.org/abs/2608.23098
作者: Yufeng Han,Lifan Deng,Cunliang Kong,Wenhao Li,Xin Cong,Yuzhuo Bai,Kangyang Luo,Maosong Sun
类目: Artificial Intelligence (cs.AI)
备注: Accepted to the System Demonstrations Program at EMNLP 2026. Code and demo: this https URL
Abstract:Classical Chinese poetry composition has long valued Tuiqiao, the iterative refinement of words, imagery, and prosody. However, many current AI poetry systems follow a one-shot generation paradigm, which reduces users to prompt providers and weakens their creative agency. We present Jiuge-Tuiqiao, an interactive human-AI collaborative system for classical Chinese poetry composition. The system is designed around a triadic model: user-driven control, ancient-guided evidence, and AI-assisted generation. Users can lock characters or lines, receive real-time prosody feedback, and obtain interpretable refinement suggestions grounded in high-frequency collocations, PPL-ranked classical lines, and structured knowledge extracted from classical encyclopedias. This design turns AI from an autonomous generator into a background assistant that supports the user’s own process of poetic refinement. Preliminary experiments and user feedback suggest that Jiuge-Tuiqiao improves controllability, interpretability, and user engagement in classical poetry composition.
[AI-32] POOL: Propagated Uncertainty Over Lookalikes
链接: https://arxiv.org/abs/2608.23086
作者: Rounak Sharma,Ananya B. Sai,Soumyabrata Pal
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Black-box large language models need confidence scores that can separate likely-correct from likely-incorrect outputs, enabling systems to prioritize human review, route uncertain cases to stronger models, or choose abstention thresholds on development data. Yet existing confidence estimators face a cost-quality trade-off: verbal confidence is cheap but is often overconfident, while sampling-based uncertainty is more informative but scales linearly with the number of samples per query. We propose \textscPOOL (\emphPropagated Uncertainty Over Lookalikes),a cost-efficient framework that addresses this trade-off taking inspiration from group-testing.\textscPOOL clusters query stems with overlaps, evaluates a base estimator on representative medoids, softly propagates confidence scores to nearby queries, and selectively evaluates high-disagreement cases. We instantiate this framework with \textscHy@ p , a hybrid estimator that combines verbal confidence with spectral answer diversity computed from the negative von Neumann entropy of sampled answer this http URL six domains from three datasets and five black-box LLMs, \textscHy@5 achieves higher average AUROC than verbal confidence and \textscVn@10 sampling while using half as many samples as \textscVn@10. \textscPOOL-\textscHy@5 retains 93.5–97.9% of its AUROC while saving 19.3–39.3% of generations. On paraphrase-dense workloads, generation savings rise to 73-76%, showing that semantic redundancy can be leveraged to lower confidence-estimation costs.
[AI-33] Improving O-RADS Risk Stratification from Ultrasound Reports: A Comparative Evaluation of Hybrid versus End-to-End LLM Reasoning Strategies
链接: https://arxiv.org/abs/2608.23061
作者: Xiaotong Tan,Chunli Qiu,Xin Liu,Qing Huang,Guangli Zhou,Bo Gao,Xiaoyan Song,Shuyan Wang,Xiuqin Wang,Wufeng Xue,Ruobing Huang,Dong Ni,Guowei Tao,Jun Cheng
类目: Artificial Intelligence (cs.AI)
备注: Main manuscript: 20 pages, 5 figures, and 2 tables; supplemental material: 11 pages, 1 figure, and 3 tables
Abstract:Background: Automating clinical guideline-based decision-making with large language models (LLMs) remains challenging because of reliability, hallucination, and limited interpretability. We compared the performance of LLMs and reasoning strategies for automated Ovarian-Adnexal Reporting and Data System (O-RADS) classification from free-text pelvic ultrasound reports. Methods: In this retrospective study, consecutive patients with ovarian masses who underwent pelvic ultrasound were included. Eight LLMs were tested with three reasoning strategies: implicit-knowledge end-to-end, rule-informed end-to-end, and a feature-based hybrid architecture that decoupled feature extraction from rule-based classification. The reference standard was O-RADS categorization established by expert consensus. Results: A total of 310 women with 390 ovarian masses were evaluated. The feature-based hybrid architecture using Gemini 3.6 Flash demonstrated the best performance, achieving an accuracy of 99.2% (387 of 390) and almost perfect agreement with the reference standard (weighted kappa = 1.00; 95% CI: 0.99-1.00). Its performance surpassed that of original clinical reports (accuracy, 87.7% [342 of 390]; weighted kappa = 0.94; 95% CI: 0.91-0.96) and end-to-end LLM strategies (accuracy range, 65.6% [256 of 390] to 95.9% [374 of 390]). For structured feature extraction, Gemini 3.6 Flash demonstrated higher overall accuracy than Claude Fable 5 (98.9% vs 97.8%; P 0.001). The hybrid architecture reduced misclassification errors and mitigated the overstaging tendency observed in original reports. Conclusion: The feature-based hybrid LLM architecture that separates clinical feature extraction from deterministic guideline execution enables highly accurate, reliable, and interpretable automated O-RADS classification, providing a promising approach for standardized, guideline-based clinical decision-making.
[AI-34] LLM -based Agents for Forecasting and Prediction: Methods Training Evaluation and Applications
链接: https://arxiv.org/abs/2608.23058
作者: Xiaogang Xu,Jiaqi Tang,Jianmin Chen,Yingying Yan,Zhenchao Tang,Xiangxin Zhou,Xiaobin Hu,Wei Wei,Jinfeng Wu,Qifeng Chen,Lu Zhou,Jiafei Wu,Zhe Liu,Jianwei Yin,Weimin Zheng
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models (LLMs) now support forecasting systems that combine language-based reasoning with temporal data, evidence retrieval, external tools, and iterative prediction. We investigate LLM-based forecasting agents, meaning systems in which a language model contributes to a scored prediction about a future or currently unobserved target. We organize architectures into three groups. Standalone LLM workflows operate on encoded time series or event context. Tool- and retrieval-augmented agents incorporate external evidence. Hybrid systems pair LLMs with statistical or foundation models. We then review training methods and evaluation protocols. We examine negative as well as positive evidence, including sensitivity to small input perturbations, ablations in which the LLM component does not improve accuracy, and benchmark gains that may reflect contamination instead of temporal reasoning. We cover applications in finance, weather, health, energy, and operations, and we summarize the benchmarks and datasets used for evaluation. The evidence indicates that measurement is a central limitation. Future work requires calibration under distribution shift, contamination-resistant live evaluation, explicit reporting of cost and accuracy together, and methods for handling feedback between deployed forecasts and the outcomes being forecast.
[AI-35] From Inertia to Objectivity: Improving Deep Research Agents with Noise Isolation EMNLP2026
链接: https://arxiv.org/abs/2608.23045
作者: Xiangxin Zhang,Zhanwei Zhang,Zhihang Fu,Binbin Lin,Wenxiao Wang
类目: Artificial Intelligence (cs.AI)
备注: EMNLP 2026
Abstract:Web search agents powered by Large Language Models (LLMs) show strong promise, but deep research tasks expose a recurring failure mode: once an agent has produced a query, plan, or intermediate conclusion, it becomes less objective when later judging the consequences of that same action. We term this phenomenon \textbfinertia bias. To make it measurable, we introduce the IBIS benchmark, which controls the search observations while varying whether the model is evaluating the outcome of its own prior action. We find that models are substantially worse when they ``own’’ the preceding search step, showing that self-authored action history can systematically distort subsequent judgment. We further show that this bias propagates into two forms of system-level degradation: search noise at the worker level and contextual noise at the manager level. To address this problem, we propose NIS-Agent, which applies context isolation at the two decision points most vulnerable to inertia bias: webpage triage and final-answer validation. Across GAIA, WebWalkerQA, BrowseComp, and BrowseComp-zh, NIS-Agent achieves competitive performance while reducing token cost by 33% compared to our baseline. We further train an 8B model to be intrinsically more resistant to inertia bias; under the same NIS-Agent framework, it attains average performance comparable to GPT-4o on deep research benchmarks.
[AI-36] MobilePA-Bench: Benchmarking Mobile Planner Agents on Complex Real-World Tasks
链接: https://arxiv.org/abs/2608.23035
作者: Yi Zhu,Xiongwei Wu,Qiyi Wang,Tingyu Qu,Jiajun Liu,Sihan Cao,Long Chen,Weigao Sun,Feida Zhu,Yiran Zhong,Steven Hoi
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:As on-device LLM agents evolve into personal copilots, the mobile operating system has become a key testbed for this paradigm, making rigorous capability evaluation essential. Yet existing benchmarks fall into two camps, each with a critical blind spot: GUI-centric benchmarks test surface-level screen manipulation while overlooking background tool use and long-horizon planning, whereas static function-calling benchmarks rely on offline API matching that is detached from real runtime constraints. To close this gap, we present \textbfMobilePA-Bench, an interactive, stateful, and tool-centric benchmark for evaluating the tool-calling and planning abilities of mobile planning agents. MobilePA-Bench runs on an executable sandbox that maintains live application databases and returns structured feedback, spanning 13 functional domains and 212 realistic mobile tools. Beyond basic tool use, it evaluates a central planning agent along three advanced dimensions: \emph(1)~Sub-agent Collaboration—decomposing a complex task and delegating specialized work to capable sub-agents; \emph(2)~Memory Usage—recalling stored memories, user profiles, and past preferences to resolve implicit requests; and \emph(3)~Skill Usage—invoking pre-packaged composite skills instead of planning every step from scratch. Extensive experiments show that current frontier LLMs remain unreliable in mobile settings: performance drops sharply under strict tool ordering, permission limits, and unexpected runtime errors. By pairing an interactive function-calling sandbox with evidence-based verification, MobilePA-Bench serves as both a practical diagnostic benchmark and an interactive foundation for agentic reinforcement learning—accelerating the development of dependable mobile agents.
[AI-37] FedCC: Towards Addressing Label Distribution Skews in Distillation-Based Federated Learning
链接: https://arxiv.org/abs/2608.23031
作者: Wenxuan Ye,Onur Ayan,Xueli An,Georg Carle
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted to Globecom 2026
Abstract:Federated Learning (FL) enables distributed clients to collaboratively train models without sharing raw data, making it promising for leveraging massive devices in communication networks. In distillation-based FL, each client applies its local model on an unlabeled public dataset, and shares only prediction results with the server. While heterogeneous local data introduces label distribution skew, thus biasing client models toward majority classes and leading to potentially inaccurate predictions. The lack of ground-truth labels in the public dataset hampers the server’s ability to calibrate predictions, which ultimately degrades overall performance. To address this, we propose FedCC, a simple and effective algorithm for mitigating client misclassification. Instead of being forced to classify and risking error propagation, clients are allowed to tag ambiguous samples as ‘unknown’. This additional class, together with calibrated pseudo-labels on the public data, balances confidence in majority classes against uncertainty in under-represented ones. Extensive experiments demonstrate that FedCC significantly outperforms existing methods, especially under severe label skew. In the extreme scenario where each client holds samples from only one of ten classes, FedCC achieves 67.3% accuracy, while baselines collapse to near-random results.
[AI-38] Artificial Empathy: Towards a Framework for Unsupervised Agency Detection and Policy Reconstruction
链接: https://arxiv.org/abs/2608.23030
作者: Peter Kuhn,Chris Pang,Sonakshi Chauhan
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:We study how an AI system can identify and model other agents in its environment from observation alone, which is a capability necessary for cooperative behaviour in the real world. This problem is less constrained than inverse reinforcement learning and remains largely unexplored. We propose a framework that uses a reinforcement learning agent, trained on an independent task as a prior about agentic dynamics, to perform agency detection and policy reconstruction.
[AI-39] PsychJail: Exploring Psychological Jailbreaks via Multi-Turn Persuasion of LLM Policies
链接: https://arxiv.org/abs/2608.23028
作者: Zeyu Feng,Qingyu Wu,Yuzhe Luo,Hua Cheng
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models (LLMs) are increasingly deployed in education, healthcare, policy advising, and other interactive settings, where users engage them as sustained social interlocutors rather than one-shot query engines. This shift makes jailbreaks a growing safety threat, yet most research emphasizes single-turn prompt optimization or iterative attack refinement, leaving psychologically grounded multi-turn vulnerabilities underexplored. We present PsychJail, a psychology-guided framework for red teaming aligned LLMs through theory-grounded, multi-turn persuasion. PsychJail maps established social-psychological persuasion techniques into a tactic-conditioned attack policy. It factorizes each attacker action into a Change-of-Meaning analysis, tactic selection, and victim-visible message, operationalizing the Persuasion Knowledge Model (PKM). The policy is refined with trajectory-level reinforcement learning using a PKM-gated reward that credits early jailbreak success only when every turn contains a well-formed Change-of-Meaning analysis. Across four aligned victim models, PsychJail achieves the highest average attack success rate (87.3%) and outperforms strong single-turn and multi-turn baselines on every model. We also measure susceptibility at the action that breaks each victim, revealing four distinct model-level fingerprints that identify which persuasion levers affect each model and how broadly. These fingerprints help explain cross-model transfer asymmetry. We interpret them as four candidate psychological profiles-rationalist, credibility-driven, narrative-monoculture, and broadly persuadable-while treating this interpretation as a conjecture requiring future validation. Our findings establish psychological jailbreaks as a distinct red-teaming frontier for increasingly interactive LLMs.
[AI-40] SplitLite: Low-Rank Residual Compression for Split Learning
链接: https://arxiv.org/abs/2608.23018
作者: Tao Li,Yulin Tang,Qi Guo,Xianhao Chen
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Federated fine-tuning of on-device large language models (LLMs) faces a significant computing burden. To overcome this limitation, split learning (SL) has emerged as a promising solution, which offloads the primary training workload to a powerful server. However, SL requires exchanging high-dimensional activations and gradients between clients and the server, resulting in prohibitive communication costs. To overcome this challenge, we propose SplitLite, a communication-efficient split federated LoRA fine-tuning method that exploits the low effective rank structure of consecutive-epoch activation and gradient residuals. Our key finding is that, when LoRA uses rank r updates in parameter space, the activation and gradient residuals of the same data sample between adjacent epochs also exhibit effective rank- 2r and rank- 4r structures, respectively. By revealing this property, SplitLite transmits only quantized truncated singular value decomposition (SVD) residual factors, thereby significantly reducing both activation uplink and gradient downlink traffic. Extensive experiments on the GLUE benchmark across a series of advanced on-device LLMs demonstrate that our method reduces activation uplink communication costs by up to 93.5% and total communication costs by up to 83.7%, without performance degradation.
[AI-41] A Physical Response-and-Memory Model for Muon Optimization
链接: https://arxiv.org/abs/2608.22994
作者: Yinze Hu,Hongjun Xiang,Xingao Gong,Hongyu Yu
类目: Machine Learning (cs.LG); Disordered Systems and Neural Networks (cond-mat.dis-nn); Statistical Mechanics (cond-mat.stat-mech); Artificial Intelligence (cs.AI); Computational Physics (physics.comp-ph)
备注: 44 pages, 8 figures
Abstract:Training large language models is costly. How low a loss the same compute can ultimately reach depends on how each step’s gradient is converted into a weight update; the rule that performs this conversion is the optimizer. From SGD and AdamW to the recent Muon, effective update rules have mostly been shaped by engineering intuition and then selected on benchmarks. Muon semi-orthogonalizes the momentum matrix before applying the update and has kept breaking records on public training benchmarks; yet why the semi-orthogonalized direction works, and over how long a history the momentum should average, are two questions at present answered mainly by experience. Here we treat the weight matrix during training as a responsive medium with memory and build a physical model for it, in which both questions find answers: the semi-orthogonalized direction is the maximally dissipative response under an output-side safety budget, which explains why it works; momentum is the internal stress accumulated by the medium; how long it should average is set by the relaxation of this stress, and a real medium relaxes on more than one timescale, the simplest form being one fast and one slow. On this basis we propose the Bi-Maxwell optimizer. The framework further yields a testable consequence: gradient directions change fast early in training and more slowly later, so the optimal memory length should grow with training stage; step-by-step measurements of a proxy for it by a read-only probe across 8 independent training trajectories are consistent with this consequence. Replacing the memory kernel alone, from a single timescale to two, brings training to the target loss in noticeably fewer steps on a public large-language-model optimizer benchmark.
[AI-42] SA-RSQ: A Versatile Sparse Representation Framework for Multi-modal Recommender Systems
链接: https://arxiv.org/abs/2608.22979
作者: Xiang Wang,Shigang Quan,Tingzhen Chang,Kang Yang,Sitong Chen,Yabo Fan,Xingxing Wang,Zhaodian He
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Deploying high-dimensional multimodal features in industrial recommender systems incurs substantial storage and latency overhead. Hard quantization is compact but introduces boundary distortion, whereas dense soft quantization couples representation quality to the limited storage budget. We propose Sparse Activation-based Residual Soft Quantization (SA-RSQ), which uses Top-K sparse routing and softmax weights to store compact (Index, Probability) tuples. The stored tuples decouple per-item storage from codebook dimensionality; for a fixed selected support, gradients propagate through the routing weights and weighted reconstruction without relying on a straight-through estimator. Experiments on a proprietary food-delivery advertising dataset show favorable reconstruction-performance and CTR trade-offs across storage budgets of 8-48 bytes per item. A preliminary Next-Distribution Prediction study and a one-week online A/B test further demonstrate the practical potential of SA-RSQ, with relative lifts of +2.51% in CTR and +3.66% in CPM.
[AI-43] Budget-Constrained Embodied Perception: Four Resource Walls and a Pre-Registered Evaluation of Access-Structured Perception on Open Models at less than 31B
链接: https://arxiv.org/abs/2608.22975
作者: Defu Lin,Wenhui Chen,Ziyao Lin,Jianlin Chen,Peiji Long,Chi Man Vong
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Embodied multimodal agents must answer from growing observation streams under a fixed per-decision token budget. We formalize this constraint through four resource walls: a perceptual Shannon wall for bounded state, a horizon wall for query-independent frame selection, a round wall for non-adaptive retrieval, and a conditional composition wall for fixed-depth inference. We introduce ASP, a training-free wrapper for frozen multimodal models that combines a capped structured state, a verbatim episodic index, and query-conditioned budget allocation with iterative access. Following a pre-registered protocol, we evaluate seven open-weight models from 3B to 31B on SEW-Bench, a license-free synthetic long-horizon walkthrough benchmark constructed to instantiate these walls. The registered natural-video benchmarks were not run because their frames require dataset agreements; our evidence therefore concerns access mechanisms, not natural-scene perception. Under a 4,096-token decision budget, ASP reaches 75 to 94% episodic retrieval accuracy, compared with 3 to 19% for equal-budget query-independent sampling, and budget reallocation outperforms quadrupling the sampling budget on every backbone. However, the full three-component architecture does not validate channel duality: removing the compressive state raises the flagship mean from 35.4 to 58.0, ASP does not outperform the verbatim-only baseline on any backbone, and two of four pre-registered falsification criteria fire. These results show that query-conditioned access, rather than parameter count or context growth alone, is decisive under a fixed budget, while prompted online compression does not earn its cost in this setting.
[AI-44] oward Effective and Reliable LLM Agents via Dynamic Ontology
链接: https://arxiv.org/abs/2608.22974
作者: Xiaohui Zhang,Zequn Sun,Chengyuan Yang,Yuanning Cui,Lingbing Guo,Wei Hu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language model (LLM) agents rely heavily on knowledge encoded in model parameters or presented as unstructured context. In domain-specific tasks, this leaves important semantic connections implicit. This often results in incomplete evidence use and brittle multi-step decisions. Ontologies offer a way to externalize domain concepts and relations as machine-interpretable structures, but constructing task-usable ontologies traditionally requires substantial effort from domain experts and is difficult to scale. Automatic construction is also challenging: an ontology that appears semantically plausible may not contain the relational structures needed for actual decision making. We present OaK, an ontology-as-a-kernel framework that dynamically constructs and refines task-oriented ontologies for LLM agents. Given task requirements and training data, OaK constructs an ontology and its knowledge graph, generates task-adaptation functions for graph reasoning, and uses judge feedback to iteratively refine both. By making relevant concepts and relations explicit, the ontology grounds knowledge retrieval and multi-step decision making. We evaluate OaK on TravelPlanner, CRMArenaPro, and ToolQA. Results show that OaK improves standard LLM agents, strengthens evidence grounding, and boosts the reliability of multi-step reasoning.
[AI-45] What Process Evaluation of Coding Agents Actually Measures: Action Task and Step Are Three Different Levels
链接: https://arxiv.org/abs/2608.22960
作者: Jiawei He,Mengyu Shi,Jie jia,Xikai Yang,Dong Sun
类目: Artificial Intelligence (cs.AI)
备注: 38 pages, 8 figures
Abstract:Coding agents are increasingly evaluated not only by whether they solve a task, but also by how they execute it. However, existing process-level evaluations often treat action prediction, task uncertainty, and step attribution as if they were the same problem, which makes it unclear what such evaluations actually measure. In this paper, we introduce a measurement framework for process evaluation in coding agents and instantiate step-level causal attribution with SCAE, a replay-based estimator derived from a structural causal model of agent execution. Our framework combines prefix-conditioned identification, replay/intervention-based estimation, and controlled judge-information manipulation to study process evaluation at the action, task, and step levels. Experiments on 499 file-localization episodes from 12 repositories show that next actions are driven primarily by execution provenance rather than code-graph transitions, execution uncertainty is structured at the task rather than step level, and full-trace judges exhibit systematic collider bias, suggesting that current process evaluation often measures semantic relevance rather than certified causal contribution.
[AI-46] Concepts for Securing Agent ic AI Coding and the Terok Environment KDD2026 ECML
链接: https://arxiv.org/abs/2608.22930
作者: Jiří Vyskočil,Franz Pöschel,Andreas Knüpfer
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注: to be published in the proceedings of the AGENSYS workshop (Workshop on Knowledge Discovery, Maintenance and Distributed Intelligence in Multi-Agent Systems) at ECML PKDD 2026 conference in Sept. 2026
Abstract:Agentic AI is a fascinating new tool for software development. It is a huge step forward compared to “conventional” AI assisted coding, which in turn was a considerable breakthrough earlier. AI support through LLMs is a young and very fast-moving field. The “conventional” (non-agentic) flavor became useful and productive in early 2025 (around 18 months ago) and the agentic flavor followed in fall 2025 (approximately 9 months ago). Besides all its benefits and potential, it also carries some fundamental risks for IT security. And the agentic approach added very severe risks while making others much more dangerous. With all the motivation to explore this fascinating new tool we should not ignore the risks but actively address them. We present (I) an assessment of the IT security risks, (II) a concept for mitigating them without breaking its benefits, and (III) an overview about an implementation of our concept. In this very dynamic field this is likely not the final and once-and-for-all answer to the identified issues but still a substantial step forward in responsible usage of Agentic AI for software development. It should also be a contribution to the community to allow early and eager evaluation of the potential of agentic AI for software development without actually suffering from its implied IT security risks. Comments: to be published in the proceedings of the AGENSYS workshop (Workshop on Knowledge Discovery, Maintenance and Distributed Intelligence in Multi-Agent Systems) at ECML PKDD 2026 conference in Sept. 2026 Subjects: Artificial Intelligence (cs.AI); Software Engineering (cs.SE) Cite as: arXiv:2608.22930 [cs.AI] (or arXiv:2608.22930v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2608.22930 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-47] Deep Learning-Based Multi-User Communication Design for Dense IoT Networks: Interference-Aware Finite-Blocklength Communication and Preliminary MIMO Extensions
链接: https://arxiv.org/abs/2608.22923
作者: Arkadeep Sinha,Shubham Paul,R. Manivasakan
类目: Information Theory (cs.IT); Artificial Intelligence (cs.AI)
备注:
Abstract:Dense IoT networks require reliable communication despite limited spectrum and substantial multi-user interference while maintaining manageable receiver complexity. This work introduces a deep-learning-based end-to-end multi-user communication design for interference-limited finite-blocklength IoT scenarios, focusing on short and medium blocklengths. We extend a prior 2-user SiameseNet transceiver framework to accommodate 2, 4, and 8 users, leveraging learned redundancy for interference suppression and noise robustness. Compared to conventional non-orthogonal access baselines, our method demonstrates strong Block Error Rate (BLER) performance across various scenarios without resorting to joint detection; the per-user decoder scales roughly linearly with the number of users. Further, we examine the robustness under interference mismatch and unequal interference strengths, critical for practical deployments with heterogeneous devices. The Latent-space analysis reveals that the learned codeword distance increases as the effective per-user rate decreases, corroborating with the observed BLER improvements. In addition, we also present preliminary results for a 2X2 MIMO setup under fixed-channel CSIT and CSIR, indicating potential for extending the framework to IoT gateways with multiple antennas. Subjects: Information Theory (cs.IT); Artificial Intelligence (cs.AI) Cite as: arXiv:2608.22923 [cs.IT] (or arXiv:2608.22923v1 [cs.IT] for this version) https://doi.org/10.48550/arXiv.2608.22923 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-48] Beyond Observed Auxiliary Relations: Environment-Conditioned Modeling for Multi-Behavior Recommendation CIKM2026
链接: https://arxiv.org/abs/2608.22920
作者: Seunghan Lee,Hyunsik Yoo,Jian Kang,Susik Yoon,SeongKu Kang
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted at CIKM 2026 (35th ACM International Conference on Information and Knowledge Management). 11 pages, 7 figures, 4 tables
Abstract:Multi-behavior recommendation (MBR) leverages auxiliary behavioral signals, such as clicks and add-to-cart, to enhance target behavior prediction like purchases. While recent graph neural network-based approaches have achieved strong performance by systematically propagating auxiliary behavior signals, they still suffer from two fundamental challenges inherent to auxiliary behaviors: (1) missing auxiliary signals, which hinder generalization to items without auxiliary observations, and (2) unreliable auxiliary signals, which amplify noise misaligned with the target behavior. To address these challenges in a unified manner, we propose BOAR, an environment-conditioned MBR framework that addresses missing and unreliable auxiliary signals through two complementary modules conditioned on auxiliary observability. Extensive experiments demonstrate that BOAR consistently outperforms state-of-the-art baselines, achieving up to 7.82% gains in HR@10 overall and up to 44.2% gains for target items without auxiliary observations, highlighting its ability to capture hidden preferences beyond observed auxiliary relations. Our code is available at: this https URL.
[AI-49] CDEG: Learning Decision-Critical Evidence for Long-Horizon Diagnostic Agents
链接: https://arxiv.org/abs/2608.22899
作者: Xiwei Dai,Zijie Meng,Zhiting Fan,Yixuan Tang,Ziru Niu,Zuozhu Liu
类目: Artificial Intelligence (cs.AI)
备注: 9 pages, 5 figures
Abstract:Unlike static medical question answering, long-horizon diagnosis captures the sequential nature of clinical practice: evidence is progressively acquired, integrated, and evaluated over multiple rounds of interaction before reaching a final diagnosis. However, existing doctor agents often fail when critical evidence is either not acquired or not adequately incorporated into diagnostic reasoning. Recent agentic approaches attempt to address these failures by reusing historical trajectories or distilled memories. But their diagnostic gains remain constrained because such experience may contain noisy or incidental information and is typically reused without validating which evidence actually drives diagnostic decisions. To address this limitation, we introduce CDEG, a graph-based framework that learns reusable decision-critical evidence from historical diagnostic trajectories. CDEG contrasts successful and failed trajectories from the same case to identify candidate evidence, validates their diagnostic impact through controlled counterfactual interventions, and organizes the resulting diagnosis–evidence–action relations into a structured graph. During inference, CDEG tracks the evolving patient evidence state to retrieve relevant diagnostic relations and selectively guide missing evidence acquisition or overlooked evidence reappraisal. Across in-domain and out-of-distribution benchmarks with multiple doctor agent backbones, CDEG consistently improves diagnostic performance, achieving up to an 11.5% accuracy gain over vanilla agents. These results demonstrate that reliable long-horizon diagnosis requires moving beyond trajectory-level experience reuse toward evidence-level learning of the factors that truly shape clinical decisions.
[AI-50] he Mask Is Not the Model: Auditing Prefix Invariance in Attention State-Space and Hybrid Sequence Models
链接: https://arxiv.org/abs/2608.22876
作者: Taebong Kim,Youngsik Hong,Minsik Kim,Sunyoung Choi,Jaewon Jang,Minseo Kim
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 24 pages, 4 figures
Abstract:We formalize prefix invariance: representations at position t must not depend on future inputs. We give a lightweight audit, two forward passes, no training or gradients, that localizes exactly where causality breaks. Attention-mask inspection is incomplete: leaks can occur via scans or normalization despite correct masks. Across 192 injected-fault trials on eight checkpoints, mask inspection found none, while our audit localized all 192/192, also finding a defect in Zamba2 and Nemotron-H.
[AI-51] GSAR: Goal-State-Anchor Rewards for Mobile GUI Agents with Self-Evolving Data Synthesis
链接: https://arxiv.org/abs/2608.22847
作者: Long Zhang,Yuhan Chen,Chaoran Zhang,Wanxia Cao,Kun Huang,Pengzhi Gao,Wei Liu,Jian Luan,Chenliang Li,Lixin Zou
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Vision-Language Models (VLMs) based GUI agents stand to benefit significantly from online reinforcement learning (RL). However, their training is bottlenecked by two fundamental issues: current data synthesis methods for GUI Agents rely on specific environments and struggle to generate diverse data, while existing evaluators either suffer from limited scalability or provide inaccurate and unreliable reward signals. To overcome these challenges, we introduce GSAR (Goal-State-Anchor Reward), a RL reward framework that supports scalable task generation and delivers reliable reward signals for stable and efficient policy optimization. Our approach features self-evolving data synthesis, which produces multiple environments through task execution and generates diverse tasks and goal states. Complementing this, a state-anchor mechanism automatically annotates task-relevant UI elements in successful goal states as reference anchors. During RL training, these reference anchors provide accurate, scalable reward signals that substantially enhance efficiency. Extensive evaluations demonstrate that our framework achieves over 90% accuracy on offline trajectory verification and performs closest to rule-based methods. Furthermore, agents trained using our reward framework exhibit strong performance on both AndroidWorld and our constructed benchmark, establishing a scalable approach for GUI agent training.
[AI-52] FinixDoc: Rethinking Financial Document Parsing Beyond Saturated Benchmarks
链接: https://arxiv.org/abs/2608.22842
作者: Hang Wang,Jin Zhang,Guoliang Xu,Pengyue Lu,Yao Li,Zijiao Zhang,Tianyu Huang,Weiqi Xiong,Yulong Wang,Chuqiao Lu,Wenkang Huang,Kai Yang,Yadong Li,Hui Li,Xingzhong Xu,Xiao Xu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Financial document parsing requires accuracy, structural consistency, and verifiability that current benchmarks often fail to reflect. We present FinixDoc, an end-to-end agentic parsing system for real-world financial documents, with FinixDoc-VL, a 4B-scale vision-language model built on Qwen3-VL-4B, as its core parser. To characterize the gap between benchmark and deployment performance, we introduce a Document Parsing Capability Matrix organized along two practical axes: visual quality and document scale. Guided by this matrix, FinixDoc-VL is trained with a domain-adapted recipe combining homoglyph-aware contrastive learning and multi-stage reinforcement learning with composite domain-specific rewards. To better leverage our accumulated advantage in low-quality financial-document data and support large-scale, high-quality data production, we further build a human-in-the-loop Data Factory pipeline with confidence-aware expert review. For evaluation, we construct FinixDocBench, a financial-domain evaluation suite covering digital-native, camera-captured, ultra-large-page, and internal-workflow scenarios, with a compliance-reviewed subset released alongside this technical report. On its main subsets, FinixDoc-VL achieves the highest overall score (81.43) among evaluated baselines, outperforming the next-best open-source model by 5.13 points, with the largest gains on internal financial workflows (FinixInner: 84.08 vs. 78.73).
[AI-53] Hierarchy-Aware Supervised Uncertainty Estimation for Black-box LLM Taxonomic Reasoning
链接: https://arxiv.org/abs/2608.22839
作者: Shuting Xie,Nathaniel Lesperance,Graham W. Taylor
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models (LLMs) are increasingly used for scientific decision support, yet reliable confidence estimation remains difficult in black-box settings. We study uncertainty estimation for hierarchical taxonomic reasoning generated by a black-box LLM in a long-tailed biodiversity monitoring pipeline. Using proxy features extracted by an open-source tool LLM, we train lightweight supervised estimators with hierarchy-aware supervision to predict rank-wise correctness. Across three tool LLMs, the supervised estimators consistently outperform a token-likelihood baseline for micro discrimination and selective prediction under a single global rejection threshold, improving micro AUROC from 0.57 to 0.75–0.80. The best results are achieved by a rank-specific multi-head design (H3), suggesting that accounting for hierarchical output structure is important when a unified abstention rule is required. Our code is publicly available at this https URL
[AI-54] Let the Bullets Fly: Multimodal Fake News Detection with Temporal-Aligned Generative Danmaku
链接: https://arxiv.org/abs/2608.22832
作者: Xiansheng Luo,Chaowei Zhang,Zewei Zhang,Yi Zhu,Jipeng Qiang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:The social interactions among crowds via \textitDanmaku (a.k.a., bullet comments) on modern multimedia platforms can facilitate both viewpoint conflicts and consensus, providing fine-grained discriminative social signals that can benefit fake news detection. However, the inherent accumulation latency of \textitDanmaku in real-world scenarios violates the real-time necessity of fake news detection, making the studies of \textitDanmaku-related fake news detection underexplored. To break this violation, we simulate this temporal-aware user interactive process by proposing a novel temporal \textbfGenerative \textbfdanmaku framework, called \textbfGenda, which consists of: (1) a \textitDanmaku Trigger for predicting the timing and intensity of user reactions; and (2) a \textitDanmaku Generator for synthesizing corresponding semantic and emotional expressions, thereby mutually constructing a temporally aligned and human-like pseudo \textitDanmaku streams. To make the generated \textitDanmaku useful for identifying fake news videos, we further design a \textitDanmaku-guided Temporal Multimodal fake news detection model - \textbfDM-FEND, which enables fine-grained multimodal interactions among video, audio, text, and \textitDanmaku, enhancing dynamic modalities alignment and semantic noise inhibition. The experimental results demonstrate that \emphDM-FEND consistently outperforms state-of-the-art baselines across both Chinese (FakeSV) and English (FakeTT) benchmarks. Further ablations validate the crucial role of temporal \textitDanmaku modeling in enhancing robustness and discriminative capability. Finally, this study offers a bright and robust solution for multimodal fake news detection in modern social interactive fashions by bridging the temporal inconsistency between news and user behaviors.
[AI-55] Beyond the Harness: End-to-End Optimization of Context Artifacts for Enterprise Text-to-SQL
链接: https://arxiv.org/abs/2608.22830
作者: Kate Gwimm,Carson Eisenach
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Deploying LLMs for enterprise Text-to-SQL is bottlenecked less by the model than by what context reaches it: business logic spans thousands of tables, and no model can ingest a full catalog at once. We argue that the most effective place to intervene is therefore the \emphknowledge-base context the model consumes, and that this context should be \emphconstructed from historical usage rather than tuned for as a fixed input. Using a query-DAG decomposition–the same family of intermediates that enterprise benchmarks like BEAVER annotate, here recovered from production SQL–we compare the value of oracle query graphs versus retrieved knowledge-base context. In this ablation, retrieved knowledge-base context provides the largest marginal improvement when added to the full oracle graph. Building on this, we optimize a distillation procedure that turns historical query profiles into reusable SQL reference cards. On a benchmark of 5176 production queries from a major online retailer, optimizing these context artifacts yields larger gains ( \sim12 – 25% AST similarity) than optimizing the retrieval harness ( \sim3 – 12% ). On the public BEAVER benchmark, which lacks the production-usage signals available in our internal setting, the picture is more mixed: table cards alone perform about the same as raw historical SQL. The best optimized variant retrieves both cards and raw SQL, scoring 9.00% versus 6.33% (p-value 0.12 ) for the comparable baseline on a held-out N=300 subset, using retrieved context and harness changes but no agentic loop.
[AI-56] Fairness-Aware Mixture-of-Experts via Subgroup Reweighting and Gate Regularization
链接: https://arxiv.org/abs/2608.22820
作者: Sunhee Hwang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Deep learning models often produce performance disparities across demographic groups, due to the training data imbalance with respect to sensitive attributes such as gender or age. To address this problem, existing work has explored fair representation learning, data re-sampling, and adversarial training, which can be broadly categorized into two main approaches. Single-stage methods typically learn a shared representation for fairness, but often struggle to handle heterogeneous subgroup distributions. Two-stage methods learn representations separately from the final prediction task, which can lead to misalignment between fairness objectives and downstream predictions. We identify routing-induced bias, a failure mode in which subgroup imbalance drives the gating network to route subgroups onto a few experts, and propose an end-to-end Mixture-of-Experts (MoE) framework that corrects it. Specifically, we apply subgroup reweighting to correct data imbalance, and introduce gate entropy regularization to prevent routing from collapsing onto subgroup attributes, keeping expert utilization both balanced and interpretable. Beyond improving fairness, the routing distribution offers an interpretable view of how subgroups are allocated across experts. Experimental results demonstrate that the proposed approach improves fairness while maintaining competitive predictive performance.
[AI-57] riplet2Track: A Hierarchical System with Object-Centric Representations for Reliable Long-Horizon Manipulation
链接: https://arxiv.org/abs/2608.22800
作者: Jianxiang Liu,Gaojing Zhang,Chuan Wen,Qipeng Liu,Yuxuan Zhao,Ning Guo,Wenzhao Lian
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: 8 pages, 6 figures. Accepted for presentation at the 2026 IEEE International Conference on Systems, Man, and Cybernetics (SMC 2026)
Abstract:Ensuring reliability in uncertain environments remains difficult for long-horizon robotic manipulation. End-to-end VLA models are data-heavy and opaque, making diagnosis and verification difficult. Hierarchical pipelines are more interpretable, but their plans are often weakly grounded in observations, weakly aligned with low-level actions, and computed without online feedback, leading to open-loop behavior and hallucinations. To address these issues, we introduce the Triplet-to-Track System (TTS), a closed-loop long-horizon imitation learning system that uses human videos to reduce reliance on robot-collected data. TTS represents high-level subgoals as instance-grounded triplets, translates them into continuous track priors for execution, and monitors task progress from observations for online replanning. Across diverse real-world long-horizon tasks, TTS achieves a 74.8% average success rate and supports object-level and compositional generalization.
[AI-58] Performance of a domain-specific large language model in answering patient questions in psychiatry
链接: https://arxiv.org/abs/2608.22797
作者: Alexander J. Hish,Arjun Nagendran,Scott N. Compton
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Background This study was designed to evaluate whether a domain-specific large language model (LLM) trained exclusively on patient education resources can answer questions about psychiatric medications, in a manner superior to LLM chatbots. We developed an LLM (“MIND”) fine-tuned for clinical fidelity, trained on patient education resources from authoritative medical organizations. Methods We compared the responses of MIND, ChatGPT, and OpenEvidence to patient questions about escitalopram, using two methods: (1) computer analysis according to a rubric measuring accuracy, clarity, completeness, nuance, safety, and referral appropriateness; (2) ratings from N=10 board-licensed psychiatrists on similar metrics. Results When rated by rubric, MIND was rated highest in all domains (p0.001). When rated by psychiatrists, ChatGPT was rated accurate more often than MIND with a negligible effect size (p=0.021, r=0.073); MIND was rated complete more often than ChatGPT with a small effect size (p0.001, r=0.160); and MIND and ChatGPT were rated safe with the same frequency (p=0.955, r=0.002). The majority of psychiatrists preferred the responses generated by ChatGPT (57.6%) compared to MIND (42.4%, p=0.003). Conclusions MIND was able to answer many questions about escitalopram in a manner deemed accurate, complete, and safe by psychiatrists the majority of the time. However, despite MIND’s ability to provide more complete responses, psychiatrists preferred ChatGPT’s responses. MIND represents a step towards building safe LLM systems to enhance patient education in psychiatry.
[AI-59] ailSieve: Partial-Rollout-Guided Tail Routing for LLM Rollouts
链接: https://arxiv.org/abs/2608.22788
作者: Tianqi Xu,Lu Lv,Haoyang Huang,Wenjie Huang,Zhanming Shen,Yuhao Shen,Baolin Zhang,Xinyi Hu,Shuang Ge,Jun Dai,Tianyu Liu,Suorong Yang,Zhikai Li,Ye Bai,Jun Zhang,Lei Chen,Yue Li,Mingchen Wan
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Large-scale rollouts have become a core component of modern LLM systems, spanning reinforcement learning (RL) post-training, on-policy distillation (OPD), and sampling-heavy evaluation pipelines. Unlike online serving, which is typically optimized for request-level latency and throughput, a small number of long-tail generations can dominate the end-to-end makespan of an entire rollout step. In practice, rollout requests are often routed uniformly across replicas, which can place extremely long generations inside high-concurrency decoding batches. To address this, we present TailSieve, a partial-rollout-guided framework that jointly controls tail routing and replica allocation for LLM rollouts. In an idealized setting with known completion lengths, we show that makespan-optimal routing in the long-tail regime combines tail isolation with load balancing, and that a simple top-k policy closely approximates this offline optimum. Leveraging the observation that long-tail prompts tend to remain long-tailed across policy updates, TailSieve uses partial rollouts as a training-free signal for identifying candidate tail groups. A hierarchical controller then jointly adapts the number of isolated groups and the replica split between the tail and bulk pools using collected response-work history and a measured concurrency-throughput model. TailSieve achieves up to 1.67x routing-only speedup over uniform group routing. The resulting low-concurrency tail pool further enables route-specialized speculative decoding with MTP or DFlash, achieving up to 2.59x speedup over uniform routing. Selected prompts are regenerated under the current policy, preserving on-policy generation and avoiding additional routing-induced length bias in steady state. Subjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG) Cite as: arXiv:2608.22788 [cs.AI] (or arXiv:2608.22788v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2608.22788 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-60] he Retriever Should Remember: Experience-Amortized Reranking for Long-Term Agent Memory
链接: https://arxiv.org/abs/2608.22767
作者: Qi Feng,Chris Ding,Jicong Fan
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Long-term language-model agents accumulate memories across interactions, but their retrievers typically do not accumulate retrieval experience. Semantic retrieval is efficient, but embedding similarity does not always reflect whether a memory contains evidence relevant to the current query. Large language model (LLM) rerankers provide stronger query-conditioned relevance scores, yet stateless reranking repeatedly scores a large candidate pool and discards these scores after each query. We introduce EARM, an experience-amortized reranking framework that treats previously acquired LLM relevance scores as reusable retrieval experience. EARM stores sparse query–memory relevance scores in an online matrix, learns their shared structure through causal matrix completion, and combines a small set of newly observed scores with estimated scores to rerank the remaining candidates. The scoring budget decreases as experience accumulates, changing LLM reranking from a repeated per-query expense into a retrieval capability learned over an agent’s lifetime. Experiments on long-term conversational memory show that mixed observed-and-estimated reranking improves answer accuracy over semantic retrieval by up to 6.62% and remains effective when only 17.5% of candidates receive direct LLM relevance scores, thereby substantially reducing the inference overhead of LLM reranking. These results motivate a broader view of agent memory: a long-lived agent should remember not only past content, but also how that content has proved useful for retrieval.
[AI-61] Compositional Chain-of-Relations for Faithful Knowledge Graph Question Answering with Large Language Models
链接: https://arxiv.org/abs/2608.22762
作者: Chenhui Liu,Jianpeng Zhou,Jiahai Wang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Knowledge graph question answering (KGQA) is a key task for evaluating KG-augmented Large Language Models (LLMs), and complex KGQA that requires multi-hop reasoning is especially challenging. Solving a complex query involves two coupled phases: candidate retrieval, which locates answer candidates over the KG, and constraint handling, which filters these candidates against the query constraints. Faithful reasoning requires grounding both phases in the KG. However, existing agent-based methods ground candidate retrieval through entity-centric exploration, while leaving constraint handling to the LLM’s internal knowledge, which leads to two critical limitations. (1) Unreliable entity pruning: entity-centric exploration uses entities as search units and must prune them to a fixed-size subset at each hop. Because entity information in KGs is often incomplete and a fixed-size subset cannot retain all valid entities, such pruning inevitably drops valid entities and ultimately leads to wrong answers. (2) Ungrounded constraint handling: query constraints are resolved from the LLM’s internal knowledge rather than the KG, leaving the final answers unverifiable and prone to hallucination. To address these limitations, this paper introduces a relation-centric exploration paradigm, which uses relations rather than entities as search units and thus avoids unreliable entity pruning. Built on this paradigm, this paper proposes Compositional Chain-of-Relations (CCoR), a simple and effective framework that grounds both phases in the KG with two relation chains: a main chain for candidate retrieval and a constraint chain that verifies query constraints through explicit KG exploration. Experiments on four KGQA benchmarks show that CCoR consistently improves accuracy, faithfulness, and efficiency over strong baselines, with more pronounced gains on complex queries.
[AI-62] DeepSAGE: Stage-Aware Reinforcement Learning for Structured CBT Counseling Dialogue
链接: https://arxiv.org/abs/2608.22615
作者: Qi Zhang,Heajun An,Prakriti Dumaru,Sang Won Lee,Lifu Huang,Pamela J. Wisniewski,Jin-Hee Cho
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large Language Model (LLM)-based counseling agents can generate fluent and supportive responses, but they often lack the structured, goal-directed progression required to conduct a coherent therapeutic session. We present DeepSAGE (Strategic AI Guidance Engine), a hybrid LLM–Deep Reinforcement Learning (DRL) framework for stage-aware counseling dialogue grounded in the first session of Cognitive Behavioral Therapy (CBT). DeepSAGE represents the session as eleven stages with explicit therapeutic objectives, with an external controller determines stage completion and the DRL model selects therapeutic intentions that guide LLM response generation. We evaluate DeepSAGE against six retrieval-, prompting-, stage-, and policy-based alternatives. DeepSAGE elicits higher simulated client engagement and openness and achieves the strongest balance of stage-goal completion and dialogue efficiency among stage-structured systems. Domain expert review further indicates that the generated conversations exhibit broadly plausible emotional trajectories and recognizable CBT processes. Because the evaluation relies primarily on simulated clients and model-based metrics, these findings demonstrate comparative dialogue-control improvements rather than clinical effectiveness. These results suggest that combining stage-structured dialogue with learned strategy selection is a promising approach for AI counseling, though clinical effectiveness, safety, and real-world utility require further human evaluation.
[AI-63] Coalition-Aware Skill Reliability for Self-Evolving Agents
链接: https://arxiv.org/abs/2608.22610
作者: Qiyan Zhao,Xiaofeng Zhang,Bo Liu,Minda Chen,Wei Xiong,Jingyang Chen,Guanting Ye,Wenhao Yu,Xiaosong Yuan,Shijie Han,Da-Han Wang,Jianmin Ji,Fei Huang,Xu-Yao Zhang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Agent skills, structured artifacts distilled from interaction trajectories and dynamically reused from skill banks, have become a central mechanism for enabling large language model (LLM)-based self-evolving agents to learn from past experience. Yet existing work has largely focused on the operational aspects of skills, such as acquisition, evolution, and retrieval, while leaving a more fundamental reliability question unresolved: Do accumulated skills in an agent’s skill bank actually make positive mechanistic contributions? We investigate this question through systematic skill-bank audits across alternative bank compositions and deployment domains, measuring the resulting changes in agent behavior. These audits reveal two recurring reliability failures: coalition pollution, where bank-level gains conceal negative coalition-level skill contributions, and cross-domain utility reversal, where source-beneficial skills reverse their effects after transfer. These findings motivate two reliability interventions: coalition-aware skill selection during skill accumulation and label-free skill masking after transfer. Coalition-Aware Skill Selection (CASS) selects more reliable candidate skills for the current bank using sampled Shapley marginals. Unsupervised Skill-Masked Coalition Optimizer (u-SMCO) masks transferred skills whose exclusion improves retrieval quality on unlabeled target-domain data. Agentic experiments on LoCoMo, LongMemEval, HotpotQA, and ALFWorld show that CASS and u-SMCO consistently improve task performance and cross-domain generalization over strong skill-based self-evolving agent baselines. Beyond accuracy, coalition-conditioned reliability modeling reduces sensitivity to noisy outcome-reward fluctuations during reinforcement learning and exposes the limits of isolation-based skill evaluation.
[AI-64] Weakly supervised concept Bottleneck Learning for Robust Two stage Object centric visual reasoning
链接: https://arxiv.org/abs/2608.22584
作者: Sparsh Tiwari,Gesina Schwalbe,Bettina Finzel
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Two-stage neuro-symbolic architectures provide an elegant paradigm for visual problem solving by cleanly separating connectionist perception of predefined symbols from possibly later defined relational reasoning thereon. However, anchoring high-level predicates into visual frames typically necessitates annotations that are expensive to acquire. In this work, we introduce the Dynamic Orthogonal Concept Bottleneck (D-OCB), an object-centric slot- VAE framework designed to extract human-aligned symbolic predicates under extremely weak supervision. D-OCB eliminates the arduous manual tuning of loss-balancing coef- ficients by dynamically learning optimal hyperparameter allocations during training. To infuse prior knowledge on independence of concept categories, in addition to standard re- construction self-supervision we penalize correlation across concept subspaces. Crucially, to combat the instability of very low supervision regimes, D-OCB incorporates a dynamic di- mensionality allocation mechanism; this adaptive formulation allows well-represented con- cepts to yield latent dimensions to underperforming concepts that are lagging behind, effectively preventing representation collapse and significantly improving overall concept accuracy. Through an extensive empirical evaluation, we demonstrate that our framework achieves high concept alignment and downstream visual reasoning accuracy using minimal label budgets, matching or outperforming end-to-end paradigms.
[AI-65] Clinical Graph-JEPA: Predictive Patient-State Knowledge Graphs for Cognitive Decision Support
链接: https://arxiv.org/abs/2608.22583
作者: Kushagra Yadav,Nalin Prabhath,Amit Lamba,Goeun Han,Yining Mao
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted at WM@Booth 2026
Abstract:Clinical records contain rich evidence about patient state, but converting that evidence into reliable, structured knowledge graphs remains difficult because extraction errors, ontology mismatch, missing relations, and temporal ambiguity can propagate into downstream systems. We propose a clinical knowledge graph construction and refinement framework that combines multi-agent relation proposal, ontology-aware normalization, deterministic evidence scoring, and JEPA-based latent refinement. Rather than treating a clinical knowledge graph as a static extraction artifact, we treat it as a predictive patient-state representation. For each admission, the system constructs an evidence-scored graph from structured MIMIC-IV records and inferred clinical cross-links, then learns to recover held-out clinical relations from the observed graph context. We evaluate the refiner with leakage-free leave-one-out edge recovery (MRR and Hits@k) and held-out batch-mask evaluation (AUC and MRR). To isolate the contribution of discharge-note context, we compare a note-embedding-free configuration with a note-augmented configuration that injects real discharge-note representations only into note-grounded entities. Under the same cohort and evaluation protocol, entity-grounded note injection improves overall leave-one-out MRR by 31% relative improvement.
[AI-66] CausalCache: Conditional High-Fidelity Restoration for Long-Horizon GUI Agents
链接: https://arxiv.org/abs/2608.22577
作者: Jiaxuan Luo,Zhanfeng Liao,Jiayao Teng,Yuan Wang,Haojian Huang
类目: Artificial Intelligence (cs.AI)
备注: 9 pages, 4 figures
Abstract:Long-horizon GUI agents can retain a complete interaction trace cheaply as textual action records, but expose only a few past events to the policy in high-fidelity pixels. We formulate this as conditional fidelity restoration: each event persists in summary-only form and is linked to an archived screenshot, while an active visual-context budget B limits how many events may be promoted to summary-plus-image form. Recent- B spends every slot on the latest events. CausalCache instead reallocates the same B promotions over the complete trace, evicting a recent image only when a distant event has higher conditional marginal utility. Its history-gated key/value (HGKV) adapter modifies only restored history-image tokens and is exactly bypassed with no history image. Matched-budget replacement groups and per-arm-anchored difference-in-differences supervision make uniform history amplification worth zero; a budget-aware selector then chooses which summarized events to restore. On desktop, the frozen policy shows no reliable preference for a task-relevant archived screenshot over the recent frame it would displace; HGKV learns exactly that selectivity inside a pre-specified drift envelope. On OSWorld-Verified, restoring history to high fidelity is worth about 13 success points over summary-only memory, while same-budget allocations remain indistinguishable. Zero-shot on a cross-application mobile benchmark, CausalCache significantly improves overall success over the same-budget recent allocation ( +3.7 points on the full roster), and the gain concentrates where it should: +8.6 points on the memory-critical split fixed by benchmark metadata at construction, no detectable effect on matched controls, and a significant split-by-method interaction.
[AI-67] Scaling Curriculum Learning For Autonomous Driving NEURIPS2026
链接: https://arxiv.org/abs/2608.22549
作者: Cevahir Koprulu,David Paz,Feng Tao,Yuliang Guo,Xinyu Huang,Ufuk Topcu,Liu Ren
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 31 pages, 18 figures. Under review at NeurIPS 2026
Abstract:Batched simulators for autonomous driving have recently enabled training reinforcement learning (RL) agents at scale, encompassing thousands of traffic scenarios and billions of interactions within a matter of days. Although such high-throughput feeds RL algorithms faster than ever, their sample-efficiency has not kept pace: As the standard training scheme, domain randomization uniformly samples scenarios, thereby consuming a vast number of interactions on cases that contribute little to learning. Curriculum learning offers a remedy by adaptively prioritizing scenarios that matter most to policy improvement. We present CL4AD, the first integration of curriculum learning into batched autonomous driving simulators by framing scenario selection as an unsupervised environment design problem. We introduce utility functions that shape curricula based on success rates and the realism of the agent’s behavior, in addition to existing regret-estimation functions. Large-scale experiments in GPUDRIVE demonstrate that curriculum learning achieves a 99% success rate a billion steps earlier than domain randomization, reducing wall-clock time by 77%, and outperforms heuristic curricula with static and dynamic attributes, with only one exception at the largest scale. An ablation under limited compute shows that curriculum learning improves sample efficiency by 67%. We also investigate how utility functions behave at scale, and how prioritized scenarios evolve during training. We release an implementation of CLForAD in GPUDRIVE.
[AI-68] STAGE: Stateful Translation to Agent ic Graph Execution with Policy-Scoped Context and Deterministic Control
链接: https://arxiv.org/abs/2608.22538
作者: Mengxi Luo,Changjia Chen,An Cao,Zirong Huang,Wanyi Dai
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Policy-governed agents must interpret case evidence while following an authorized procedure. We present \textscStage, an executable-graph framework that confines model judgment to policy-scoped nodes while placing procedural control in deterministic code. At each node, the model receives task-relevant policy context and returns a typed result, while the coordinator enforces the reviewed execution contract. We evaluate \textscStage on SOP-Bench Referral Abuse, two \tau^2 -bench domains, and Smart Dispute, a proprietary banking benchmark. Compared with monolithic full-policy execution, \textscStage generally improves task success and repeated-run reliability across workflows of varying procedural complexity. The largest gains occur on the deeper Telecom and Smart Dispute workflows, where \mathrmPass^3 increases by 7.5–55.0 and 57.2–65.7 percentage points, respectively, depending on the model. These results show that combining policy-scoped context with deterministic procedural control can improve the reliability of policy execution.
[AI-69] CONTRAMEM: Learning Self-Evolving Procedural Memory from Contrasting Multi-Model Trajectories
链接: https://arxiv.org/abs/2608.22533
作者: Zheyuan Deng,Binghang Lu,Hanqi Feng,Shirley Huang,Dianzhuo Wang,Yuanda Xu,Zhiwei Zhang,Yige Sun,Changhong Mou,Runyu Zhang,Yuexing Hao,Barnabas Poczos,Xiaomin Li
类目: Artificial Intelligence (cs.AI)
备注: 35 pages, 7 figures; includes technical appendix
Abstract:Autonomous computer-use agents are increasingly applied to long-horizon tasks requiring coordinated application calls, persistent state tracking, and verifier-sensitive writes, yet they remain prone to procedural failures: misreading application state, tool semantics, or task progress. Procedural memory promises more consistent decisions and less redundant exploration, but constructing high-quality memory without model training remains challenging. We introduce CONTRAMEM, a source-flexible, training-free framework for self-evolving procedural memory that treats same-task outcome variation as supervision: differences in correctness, efficiency, recovery, and failure modes expose outcome-relevant procedural distinctions, distilled into a compact bank of app-level Function Cards and task-level Skill Cards that evolves through localized curation rather than append-only accumulation or whole-bank rewriting. On held-out GAIA2/ARE computer-use tasks, CONTRAMEM more than doubles the success rate across the three source-model targets (26.2% to 55.3%), with consistent per-model gains (GPT-5.5: 27.5 to 61.0; Claude Sonnet 4.6: 28.0 to 52.5; DeepSeek V4 Pro: 23.0 to 52.5). The same bank transfers unchanged to the unseen Qwen3.7 Plus (18.5 to 35.5), indicating transferable procedural knowledge rather than model-specific behavior. The same construction carries over unchanged to AppWorld, beating both no memory and its own single-source self-memory variant for all three mid-tier agents on both public test splits. Under a matched trajectory budget, heterogeneous multi-model trajectories yield stronger memory than self- or same-model multi-rollout memory: the margin comes from contrastive behavioral diversity, not stronger source agents or more sampling.
[AI-70] HANSARD: A Reference Architecture for Forensic Readiness Runtime Witnessing and Graded Attribution in Autonomous Multi-Agent AI Systems
链接: https://arxiv.org/abs/2608.22512
作者: Christos Sardianos,Iliana Pla,Vasilis Efthymiou,Iraklis Varlamis,Thomas Lagkas,Panagiotis Sarigiannidis,Georgios Th. Papadopoulos
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Autonomous multi-agent systems nowadays act in finance, software supply chains, and security operations. Already, the first largely AI-orchestrated intrusion campaigns have been reported. Yet, when such a system causes harm, no method can robustly establish what happened, what caused it, or who is accountable. This is because provenance forensics works at the wrong abstraction, formal causality assumes the causal model, and agent auditing trusts self-recording. The target failure mode is, thus, attribution laundering, i.e., spreading an act across redundant agents until none is a but-for cause. Worse, the record is produced by the suspects, which comprises the assumption adopted throughout this work. Agents may therefore anticipate the investigation and the part of logging infrastructure may itself collude. In this paper, HANSARD is proposed, a reference architecture treating accountability as a life-cycle property. First, a readiness profile sealed before operation bounds what later findings may claim. Second, capturing at five choke points beyond the agents’ reach makes omissions detectable, not only tampering. Third, a typed PROV-DM-aligned causal graph accrues as the system runs, and three indicators read it live to gate oversight without adjudicating. Fourth, post-incident replay yields contingent effects under the modified Halpern-Pearl definition, together with a compensation-set size. Finally, a synergy residual measures harm due to the combination rather than to individuals, making laundering visible. Cause, responsibility and accountability are then reported separately, each capped by an evidentiary tier, while a future research agenda is also provided.
[AI-71] ClawProBench: Trace-Aware Evaluation of AI Agents with Runtime Coverag e and Frozen Workplace-Style Holdouts
链接: https://arxiv.org/abs/2608.22510
作者: YuanHang Xiao
类目: Artificial Intelligence (cs.AI)
备注: 29 pages, 4 figures
Abstract:Agent benchmarks often evaluate only final answers even when agents run on stateful runtimes. We argue this under-specifies what is being evaluated: the proper unit is a declared model-plus-runtime configuration whose failures can occur in evidence acquisition, runtime routing, safety boundaries, or repeated execution. We present ClawProBench, a trace-aware benchmark for runtime-native agent evaluation instantiated on OpenClaw, a live agent runtime with workspace tools and native surfaces for browsing, memory, messaging, scheduling, skills, and subagents. ClawProBench defines two tracks: a 102-scenario full profile with live workspace and native-runtime routing tasks, and a frozen 68-scenario holdout with closed-world JSON output contracts for robust ranking. Trials are scored from execution traces via a safety-gated formula combining correctness, process quality, and efficiency, preserving failure evidence for audit. Our anonymous artifact includes benchmark definitions, scoring code, manifests and sanitized traces. We evaluate 68 configurations on the full profile and 37 on holdout. The top safety-gated average trace score is 0.7671. Native-runtime tasks underperform workspace-live tasks (0.5238 vs. 0.6415). On holdout, pass@k-any outperforms strict three-trial pass (0.6638 vs. 0.2890), while full-profile and holdout rankings show weak alignment (Spearman 0.1300). Rankings based purely on correctness differ substantially from process-aware, safety-gated and strict-pass views. Final-answer leaderboards may hide native-surface weaknesses, one-off successes and trace-local agent failure modes.
[AI-72] When Does AI for PDEs Yield Scientific Evidence?
链接: https://arxiv.org/abs/2608.22504
作者: Wenshuo Wang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Existing AI-for-PDE benchmarks primarily assess models in terms of predictive or approximation accuracy. In physics research, however, AI outputs often serve as evidence for scientific claims. These two objectives are not equivalent: the former measures an output’s agreement with a reference target or satisfaction of governing constraints; the latter asks whether, given a specified object of study, scientific claim, assumptions, and evidence standard, the output provides sufficient evidence for that claim. To bridge this gap, we extend a widely used PDE-simulation benchmark and a comprehensive benchmark for PDE inverse problems to enable, for the first time in AI for PDEs, evaluation of whether and to what extent model outputs support specified scientific claims. Our results show that numerical accuracy and evidential support can rank models differently, explain when and why they do so, and reveal that existing benchmarks can favor methods whose outputs provide weaker support for the scientific claims of interest. Together, we formalize, empirically demonstrate, and explain this evaluation–use mismatch in AI for PDEs.
[AI-73] Functional compatibility as a determinant of persistent neural learning
链接: https://arxiv.org/abs/2608.22462
作者: Hossein Javidnia
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Artificial neural networks can acquire new capabilities but often damage existing ones when they continue to learn. This stability-plasticity problem has motivated replay, regularization and constrained-update methods, yet it remains unclear whether a property of incoming learning itself determines what can be retained without disrupting protected behaviour. Here we show that functional compatibility, the extent to which new learning can coexist with behaviour that must be preserved, is a causal determinant of persistent learning. To our knowledge, this is the first controlled causal demonstration in which compatibility is deliberately changed from matched neural states and persistent learning is measured under a common retention requirement. The effect generalizes across independent learning directions, convolutional and transformer architectures, vision and text, and additional seeds. Learning rules differ in how efficiently they exploit available compatibility, while retention constraints limit how much can be stored. At larger finite updates, nonlinear geometry changes the available learning opportunity and ultimately prevents the matched compatibility continuum from being realized. These results establish functional compatibility as an experimentally controllable principle of persistent neural learning, shifting the problem from preventing forgetting towards identifying which components of new learning can safely become permanent.
[AI-74] EMPIRE: Explicit Manipulation Planning as a Learnable Intermediate Representation for Egocentric Hand-Motion Forecasting
链接: https://arxiv.org/abs/2608.22449
作者: Wen Wang,Ruibing Hou,Hong Chang,Shiguang Shan,Xilin Chen
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: 14 pages, 10 figures, 18 tables
Abstract:Forecasting dexterous hand motions from egocentric observations is fundamental to intelligent interactive systems. Existing VLM-based methods typically map observations directly to future motions, overlooking the underlying manipulation process that governs hand-object interactions. Moreover, end-to-end optimization couples manipulation learning with motion synthesis, causing motion-generation gradients to interfere with the pre-learned manipulation-aware representations. To overcome these limitations, we propose EMPIRE, a two-stage framework that introduces Explicit Manipulation Planning as an Intermediate Representation for Egocentric hand-motion forecasting. Stage I: Learn to Plan. EMPIRE first learns explicit manipulation plans from multimodal context to capture the progression of hand-object interactions. Stage II: Learn to Act. A motion generator synthesizes future bimanual hand motions conditioned on frozen planner representations, preventing motion-generation gradients from affecting manipulation planning. To support our method, we further construct EMPIRE-651K, a bimanual hand-motion forecasting dataset comprising 650,910 training windows across 111 tasks, each paired with an explicit per-hand manipulation plan. Under identical training and evaluation protocols, EMPIRE achieves state-of-the-art forecasting accuracy, with an MPJPE of 84.53 mm and a finger-relative error of 38.97mm. We release the code and dataset at this https URL.
[AI-75] When Persona Simulations Are Informative: Graph-Structured Signals for Pluralistic Opinion Sensing
链接: https://arxiv.org/abs/2608.22438
作者: Taehyeon An,Jaehyeong Park,Donghyuk Shin
类目: Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注:
Abstract:Persona-conditioned large language models (LLMs) are increasingly used to simulate survey responses across diverse domains. However, apparent response variation can reflect unconditioned model priors or token sampling noise rather than systematic persona conditioning. We argue that persona-conditioned variation is informative when semantically similar personas exhibit concordant response shifts. To operationalize this principle, we introduce Persona-Conditioned Informativeness (PCI), an unsupervised diagnostic metric that measures whether semantically similar personas deviate in concordant directions relative to item-level sample baselines. By modeling personas as a similarity graph, PCI uses Local Moran’s I to quantify local spatial coherence and extract compact persona subsets without using construct labels. To evaluate PCI without external human benchmarks, we test its ability to recover established latent value structure using the 57-item Portrait Values Questionnaire-Revised (PVQ-RR). Confirmatory factor analysis (CFA) shows that a PCI-selected 10% subset substantially improves overall construct recovery relative to response-stability and random selection. These findings support PCI as a principled internal diagnostic for screening synthetic respondents in survey pipelines.
[AI-76] hink with Structured Grounding: Perceptual Reinforcement Learning for Chart and Visual-Tabular Understanding
链接: https://arxiv.org/abs/2608.22429
作者: Changjiang Jiang,Qiannian Zhao,Lei Xin,Jinxiang Xie,Preslav Nakov,Zhuohan Xie
类目: Artificial Intelligence (cs.AI)
备注: Manuscript
Abstract:Multimodal Large Language Models (MLLMs) capable of thinking with images often rely on external tools for fine-grained perception. However, this reliance introduces significant inference latency and fails to effectively resolve the spatial-structural gap-a fundamental challenge in text-dense and structurally relational visuals (e.g., charts and visual tables) where strict relative spatial arrangements bind textual elements. Without external tools, standard MLLMs struggle with such fine-grained visual reasoning tasks. To address these issues, we propose Think with Structured Grounding (TwSG), a novel fine-grained image perception framework designed to internalize complex images’s tool-use capabilities within the model. TwSG distills the benefits of multi-step reasoning and micro-cropping into a single efficient forward pass during inference. Specifically, we use an MLLM to identify key regions guided by ground-truth answers, and then prompt a teacher model to generate high-quality visual question-answering (VQA) data. These fine-grained, region-based supervisory signals are subsequently distilled back into the full-image representation. Our training pipeline consists of two stages: (1) a cold-start supervised fine-tuning (SFT) phase using multi-turn data with focused area descriptions to foster complex reasoning and error recovery; and (2) a reinforcement fine-tuning (RFT) phase driven by a novel process reward mechanism, TL-GRPO, which encourages strategic reasoning. Extensive experiments across various MLLM architectures demonstrate that TwSG reduces inference latency while substantially improving accuracy and robustness, endowing models with native fine-grained region description and flexible reasoning capabilities.
[AI-77] Where World Models Break: Natural-Input Failure Discovery
链接: https://arxiv.org/abs/2608.22421
作者: Zhanpeng Shi,Zi Liang,Rong Feng,Shiqin Tang,Xuyang Chen,Hongzong Li
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:World models predict action-conditioned futures and serve as critical internal simulators for downstream planning and control. However, catastrophic prediction failures of world models could dangerously propagate through the control pipeline, as subsequent agent or model training and decision-making depend heavily on the continuous environment evolution forecasted by these world models. Existing evaluations overlook this systemic risk: by aggregating average errors over benign generations from general queries, they fail to stress-test the model against catastrophic collapses under rare or unobserved condition-action combinations. To bridge this gap, we formalize the natural-input failure discovery problem: under a finite query budget, finding environment-valid conditions and action prefixes that induce severe prediction risk, verifying whether these failures reproduce on fresh seeds, and testing their persistence under nearby valid edits. Discovering such critical failures is computationally challenging, as valid condition-action combinations explode exponentially, rendering exhaustive search or standard sampling infeasible given the high cost of noisy rollouts. To tackle this, we propose BasinLens, which exploits the underlying structure of valid inputs, where each coordinate possesses environment-defined semantic types and admissible domains, by pairing uncertainty-guided global search with typed local replacements. Across diverse benchmarks and world-model families, BasinLens exposes reproducible and locally persistent failure modes that conventional evaluations fail to reveal, showing that average-case benchmarks can mask important vulnerabilities in world-model-driven control.
[AI-78] Cross-Subject Generalization in Decoding Perceived Speech from Non-Invasive Brain Recordings
链接: https://arxiv.org/abs/2608.22420
作者: Aoke Zhang,Bo Wang,Xihong Wu,Heping Cheng,Jing Chen
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注: Submitted to IEEE Transactions on Audio, Speech, and Language Processing (TASLP)
Abstract:Decoding perceived speech from non-invasive brain recordings has garnered significant attention in recent years due to its wide range of potential applications. However, existing methods face considerable challenges in cross-subject decoding, primarily due to limited generalizability and the absence of explicit mechanisms for extracting subject-consistent information. These limitations result in high training costs and suboptimal decoding performance. To address these challenges, we propose an innovative Cross-Subject Perceived Speech Decoding (CPSD) framework, which comprises two training stages: source model pre-training and personal specialization. In the source model pre-training stage, contrastive learning is employed to capture shared representations across multiple source subjects. Subsequently, personal specialization initializes the model for the target subject by extracting consistent components from the source model and fine-tuning it using target subject data. Additionally, we introduce the Positional Encoding-based Spatial Attention (PESA) module, which remaps MEG/EEG data into a standardized reference space, thereby enhancing cross-subject consistency and facilitating model training. We evaluate the proposed CPSD framework on three perceived speech neural datasets encompassing different modalities and languages. The results demonstrate that our framework outperforms baseline methods by more than 6.8%, 15.4%, and 15.8% in Top-10 accuracy on the Armeni 2022, PKUEEG 2025, and Broderick 2018 datasets, respectively. Further analyses confirm the effectiveness, efficiency, and robustness of the proposed approach.
[AI-79] KONTOGRAPH: Verified Point-in-Time Feature Consistency and Amortised Explanation for Real-Time Anti-Money Laundering under a 200 ms Decision Budget
链接: https://arxiv.org/abs/2608.22389
作者: Ahmed Abolfadl
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Software Engineering (cs.SE)
备注:
Abstract:Regulation (EU) 2024/886 obliges European payment service providers to settle euro credit transfers in under ten seconds, around the clock. This removes both the overnight batch window in which anti-money-laundering (AML) analytics traditionally ran and the settlement delay that made recovery possible, forcing detection, explanation and decision inside a single-digit-second envelope. We present KONTOGRAPH, an end-to-end AML pipeline for the SEPA Instant rail built under a self-imposed 200 ms 99th-percentile budget, and report an empirical study on 1,562,860 simulated payments with injected typologies and deliberately incomplete labels. Three findings are of interest beyond the system itself. First, a temporal graph network with per-node memory improves PR-AUC over a gradient-boosted tabular baseline from 0.0053 to 0.1717, a paired day-blocked bootstrap difference of +0.166 with 95% CI [0.105, 0.241]; per-node memory alone more than doubles the score. Second, expressing each feature once and compiling it to three execution backends, with equivalence enforced by property-based tests that perturb the future, surfaced three point-in-time violations that code review had passed–each of which would have inflated reported performance. Third, and most consequential for practice, exporting the deployed tree ensemble to ONNX changed only 7.4 \times 10^-8 in mean score yet altered 0.26% of decisions and inflated the alert volume by 12%, because 32-bit accumulation perturbs scores across a cost-optimal threshold of 3.98 \times 10^-4 . We argue that a serving-format conversion must be treated as a model change until measured, and that fidelity metrics for subgraph explainers can be vacuous when candidate neighbourhoods are small–a null result we report in full.
[AI-80] Self-Supervised Graph Representation Learning for In-The-Wild Wearable and Smartphone based Emotion Recognition ICASSP2025
链接: https://arxiv.org/abs/2608.22387
作者: Ioannis N. Ziogas,Leontios J. Hadjileontiadis,Ahsan H. Khandoker,Aamna Al Shehhi
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Signal Processing (eess.SP)
备注: Published at ICASSP 2025. Copyright 2025 IEEE
Abstract:Wearable and smartphone-based emotion recognition (WER) remains a challenging setting in affective computing, due to the notorious difficulty and bias associated with in-the-wild label collection. The high inter-and intra-subject emotional variability motivates us to explore WER modeling through graph node classification in a limited resources learning scheme powered by Self-Supervised Learning (SSL) graph masking augmentation tasks. We employ a subgraph sampling approach during training, utilizing labeled and unlabeled data, along with supervised, semi-supervised, and SSL mechanisms in a multi-task inductive graph neural network architecture. Our evaluations on K-EmoPhone through leave-one-group-out cross-validation in the binary arousal and valence tasks yield average accuracy gains of 4.3% and 7.8%, compared to the full resource setting, utilizing only 20% and 25% of the labels, respectively. Our model analysis sheds light on the relation of SSL graph augmentations to emotional arousal and valence and justifies the approach of SSL-driven subgraph training for in-the-wild WER.
[AI-81] WAM-OPD: On-Policy Distillation for World Action Models
链接: https://arxiv.org/abs/2608.22364
作者: Liuhaichen Yang,Zhuang Jiang,Chenchao Sheng,Zezhi Tang
类目: Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注:
Abstract:World action models (WAMs) couple visual future prediction with robot action generation, but accelerated students can lose task capabilities during distillation and later encounter states that are poorly represented by offline data. We study whether on-policy distillation (OPD) can repair such a student without requiring sparse-reward reinforcement learning. We introduce WAM-OPD, a deployment-consistent post-training recipe for a video-first WAM. The student acts in the environment and therefore determines the history distribution. A frozen teacher labels those student histories with coherent video and action targets, while the student action branch is trained under its own generated video plan, as it is at deployment. Joint video and action losses update lightweight adapters in the shared backbone, together with an action flow-matching regularizer. In preliminary RoboTwin 2.0 studies on two tasks, the released one-video/one-action-step Flash-WAM improves from 0.0% to 58.3% success on HANDOVER MIC, and from 16.7% to 33.3% on PUT OBJECT CABINET. These task-specific results are an initial capability proof rather than evidence of broad or uniform generalization. They nevertheless suggest that dense teacher supervision on student-induced histories is a promising post-training interface for video-first WAMs.
[AI-82] Analyzing and Mitigating Cross-Lingual Degradation in Multilingual Medical VQA EMNLP2026
链接: https://arxiv.org/abs/2608.22363
作者: Jingbo Wang,Sendong Zhao,Haochun Wang,Bing Qin,Ting Liu
类目: Artificial Intelligence (cs.AI)
备注: EMNLP 2026 main
Abstract:Medical visual question answering (VQA) is a crucial task in clinical AI, yet its evaluation has so far centered almost exclusively on English, limiting its relevance to linguistically diverse patients and clinicians. Recent multilingual medical VQA benchmarks show that large vision-language models (LVLMs) degrade in non-English languages, but lack a fine-grained analysis of how cross-lingual variation affects the distinct capabilities that medical VQA requires. To this end, we construct a multilingual medical VQA benchmark over eight languages, organized into four representative scenarios that isolate the core capabilities medical VQA requires. Evaluating five open- and closed-source LVLMs, we find that cross-lingual degradation is not uniform but highly scenario-dependent. We therefore propose MedVL-XLRepE, a training-free scenario-aware representation engineering method, leveraging LVLMs’ superior English medical VQA capability to steer non-English representations toward their English counterparts at inference time. Across three LVLMs and eight languages, MedVL-XLRepE consistently mitigates cross-lingual degradation, with gains of up to 6.33%.
[AI-83] SANE: State Anomaly Neutralization for Stable Extreme-Context Delta-Rule Models
链接: https://arxiv.org/abs/2608.22354
作者: Qingwen Lin,Boyan Xu,Xiao Liu,Zhifeng Hao,Ruichu Cai
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Delta-Rule recurrent models maintain a fixed-size state, enabling O(1) inference memory but potentially becoming unstable under extreme-context extrapolation. By tracking RWKV-7 over sequences of up to 100M tokens, we empirically identify a distinct failure pattern: \textbflocalized norm explosion atop a relatively sparse substrate, rather than global state saturation. Analysis of the recurrent update suggests that persistent decay keeps weakly updated entries small, whereas uneven injections allow a few channels to accumulate extreme values. Motivated by this diagnosis, we propose \textbfState Anomaly Neutralization (SANE), which applies adaptive \tanh compression at chunk boundaries while preserving the intra-chunk parallel structure. Within a safe threshold range ( 3 \le \alpha \le 5 ), SANE matches the baseline on 11 short-context reasoning benchmarks with no statistically significant degradation. After a 100M-token prefix, which exceeds the training length by over 24,000\times , SANE retains functional reasoning ( 33.46 – 35.56 ) while the baseline encounters numerical overflow. In contrast, overly permissive thresholds ( \alpha \ge 8 ) remain numerically stable but lose reasoning capability entirely, showing that numerical stabilization alone does not guarantee functional reasoning and revealing a capacity–stability trade-off in state compression.
[AI-84] HERO: Human-profile Enhanced Retrieval Optimization Framework for Long-term Agent Memory
链接: https://arxiv.org/abs/2608.22310
作者: Yuanhua Lin,Yile Li,Zhiyuan Zhao,Jing Shang,Jian Sun
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Long-term memory is crucial for personalized responses and long-horizon agent interactions. Existing methods often rely on LLMs to compress or rewrite dialogue histories and use the transformed memories as retrieval evidence. Despite the progress in organizing fragmented contexts, two major drawbacks persist: (1) information loss from compression, which discards fine-grained but later useful details, and (2) semantic drift from rewriting, which erodes the original tone and situated context. In this work, we propose a novel Human-profile Enhanced Retrieval Optimization framework for long-term agent memory (HERO). Specifically, HERO converts the dialogue history into a traceable heterogeneous memory graph that preserves raw dialogue text as evidence for reasoning, thereby mitigating information loss. For retrieval, HERO extracts initial anchors from the current query and incorporates human profiles via an iterative graph traversal; these anchors and profiles provide guidance signals that adaptively activate the most informative regions of the graph. Experiments on two benchmark datasets show that HERO outperforms strong baselines on both factual and personalized reasoning, while providing more faithful access to raw dialogue evidence.
[AI-85] he Imitator Game: Benchmarking Robot Imitative Ability Beyond Action Prediction
链接: https://arxiv.org/abs/2608.22301
作者: Xunzhe Zhou,Yiyang Cai,Fengyi Wang,Ran Ju,Hanxiang Ren,Ruizhe Liu,Yu Zhang,Qian Luo,Feng Chen,Pei Zhou,Yi Ma,Yanchao Yang
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Humans imitate at the level of intent: given a demonstration, we infer its goal and carry it out with whatever tools, objects, and layouts are at hand. Current robot policies instead learn observation-to-action mappings from visual inputs and language instructions, without explicitly inferring the demonstrated task. Learning from human video thus remains largely trajectory-level: models can replay motions in near-identical scenes, but still struggle to imitate what the demonstrator intends rather than merely what they do. We introduce The Imitator Game, a four-level benchmark (L0-L3) that progressively widens the gap between the human demonstration and the robot’s own scene, isolating where trajectory replay ceases to suffice and task understanding becomes necessary. We pair it with IG-10K, the largest environment-aligned paired human-robot dataset to date and the only one instantiated across all four levels in both real and simulated settings (20,000+ paired episodes, 50+ tasks, 6 domains), and Imitator Arena, an open platform for blind A/B human evaluation. Across nine state-of-the-art models, performance is stable from L0 to L2 but collapses at L3, identifying functional substitution - achieving the same intent through a different object affordance - as the decisive barrier to intent-level imitation. Human-video-conditioned models outperform caption-conditioned ones, yet every model falls below 13% zero-shot success on unseen tasks; fine-tuning IG-10K-pretrained models with only 10 paired human-robot demonstrations yields large gains that grow with pretraining scale. The project website and access to Imitator Arena are available at this https URL.
[AI-86] Learning from the Test: Self-Referential Differential Testing for Deep RL Agents
链接: https://arxiv.org/abs/2608.22284
作者: Junda He,Jieke Shi,Zhou Yang,Mingfei Cheng,David Lo
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Deep Reinforcement Learning (DRL) has achieved significant success in complex decision-making problems. As DRL systems are increasingly deployed in real-world applications, ensuring their quality and reliability is paramount. Current works primarily focus on detecting safety-critical failures, often neglecting policy optimality, which can lead to reduced efficiency, user distrust, and economic losses. This oversight, compounded by the inherent “testing oracle problem” for optimality, leaves a significant gap in comprehensively evaluating DRL systems. To address this gap, we propose Delta (Differential Testing for DRL Agents), a novel and comprehensive framework that automatically identifies both safety-critical and optimality bugs in DRL agents. Delta employs a two-phase approach: (1) Safety Testing, where the Agent Under Test (AUT) is evaluated for catastrophic failures while collecting data from its decision-making policy, and (2) Optimality Testing, where this collected data from the prior phase is used to train a challenger agent via Offline Reinforcement Learning. Differential testing is then performed by comparing the challenger agent against the AUT; instances where the challenger achieves higher cumulative rewards indicate optimality issues in the AUT. We demonstrate Delta’s effectiveness across five environments. We investigate the effectiveness of three offline RL algorithms (BC, BCQ, and CQL) in generating challenger agents. Experimental results demonstrate that safety testing datasets are valuable for training competent DRL agents. Challenger agents trained with BCQ proved most effective for identifying optimality issues within the framework of Delta. Across the five environments, Delta uncovered an average of 2,518 optimality issues, outperforming the baseline methods by 50.2%.
[AI-87] Multi-Task Learning for Non-Canonical Phoneme Recognition via Articulatory Feature Decomposition
链接: https://arxiv.org/abs/2608.22273
作者: Sophia Riaz,Haoze Zheng,Amos Roche,Miyu Zhang,Anamika Ragu,Salvatore Penachio,Kaustav Mukherjee,Aneesh Jonelagadda
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注: 23 pages, 4 figures, 4 tables
Abstract:Pathological and more broadly non-canonical speech present significant challenges for automatic phoneme recognition due to systematic deviations from canonical pronunciation and limited availability of labeled clinical speech data. Existing phoneme recognition systems are typically trained on canonical speech and treat phonemes as atomic categorical labels, limiting their ability to detect structured articulatory errors common in speech disorders and accents. In this work, we introduce a linguistically structured approach to non-canonical phoneme recognition that decomposes phoneme prediction into articulatory feature dimensions such as manner, place, and voicing. We implement this formulation using a hierarchical multi-task learning architecture in which task-specific articulatory feature heads learn feature-level representations that are subsequently integrated through a cross-attention-based fusion module to produce phoneme predictions. To address the scarcity and noise of pathological speech labels, we combine this framework with semi-supervised learning via Momentum Pseudo-Labeling (MPL) and propose a cascaded training strategy that progressively introduces articulatory feature tasks while employing staged unfreezing of a pretrained speech encoder. Experiments on L2-ARCTIC, used as a proxy for pathological speech variation, show that the proposed approach achieves substantial improvements in phoneme recognition performance compared to strong baseline architectures, while yielding interpretable error patterns aligned with phonological feature structure. These results suggest that articulatory feature supervision is a promising strategy for robust and interpretable phoneme recognition in non-canonical speech, and motivate future validation on clinically diagnosed pathological speech datasets.
[AI-88] Read Less Solve More: Token-Efficient Sparse Reading for AI Agents
链接: https://arxiv.org/abs/2608.22237
作者: Zedong Liu,Jiaan Wu,Xinyang Ma,Le Xu,Kai Wang,Yuanchao Hu,Dingwen Tao,Guangming Tan
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Long-horizon agents increasingly rely on repeated access to external artifacts, yet current reading interfaces often expose entire objects even when only sparse evidence is needed. This over-reading increases token and latency costs and can dilute task-relevant evidence, while existing context-reduction methods mainly intervene after broad content has already entered the trajectory. We present SparseRead, a training-free, model-transparent reading layer that controls content admission before unnecessary evidence reaches the model context. SparseRead combines a regime-aware Read Gate, extensible Reader Backends, and a stateful protocol for bounded, source-anchored evidence acquisition with explicit refinement, verification, stopping, and fallback. Across six frontier models, including Claude Opus 5, and five workload scenarios, SparseRead reduces token volume by up to 92.9% and wall time by up to 89.0%, while preserving or improving task quality. Its consistent gains across three agent frameworks further demonstrate broad portability.
[AI-89] FreKoo: Learning Continuous Spectral Dynamics for Temporal Domain Generalization
链接: https://arxiv.org/abs/2608.22224
作者: En Yu,Xiaoyu Yang,Wei Duan,Guangquan Zhang,Jie Lu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Temporal Domain Generalization (TDG) aims to learn from historical domains and generalize to unseen future distributions under concept drift. Nevertheless, prevailing TDG methods struggle with complex real-world streaming scenarios involving both multi-scale drift patterns (e.g., long-term periodicity intertwined with short-term incremental changes) and local uncertainties, especially in continuous settings where observations arrive irregularly. To address this limitation, we propose FreKoo++, a novel continuous spectral-dynamical framework that pioneers the unification of continuous Koopman modal dynamics with adaptive spectral disentanglement. Specifically, FreKoo++ maps source-domain parameters into a compact latent space, modeling their evolution as a superposition of learnable continuous modes where complex eigenvalues jointly encode oscillatory frequency and temporal growth or decay. This formulation naturally accommodates irregular timestamps and supports arbitrary horizon extrapolation without rigid discrete stepping. Furthermore, we propose a new adaptive soft spectral weighting mechanism backed by stability and spectral regularization, which automatically isolates persistent dominant dynamics from transient noise without relying on manual frequency thresholds. We derive modal approximation and generalization bounds that characterize how amplitude and eigenvalue estimation errors propagate with the prediction horizon. Extensive experiments on both discrete and continuous TDG benchmarks demonstrate that FreKoo++ achieves state-of-the-art performance under complex multi-scale drifts and irregular sampling.
[AI-90] Query-Driven Multimodal Information Extraction from Long Documents
链接: https://arxiv.org/abs/2608.22214
作者: Yikai Gao,Ding Xia,Xi Yang
类目: Artificial Intelligence (cs.AI); Multimedia (cs.MM)
备注:
Abstract:In domain-specific multimodal long documents, images and text jointly convey complex knowledge that cannot be fully captured by plain text alone. However, existing paradigms like DocVQA primarily focus on generating textual answers or localizing evidence regions, rather than outputting query-specific textual attribute values and corresponding images. To address this gap, we propose query-driven image-text joint extraction from long documents, requiring models to output query-requested textual attribute values and corresponding image bounding boxes. Based on challenges related to both user intent and document content, we designed a two-level taxonomy that operates at the query and instance levels. Further, we construct ITJoint, the first high-quality, manually annotated benchmark for this new task, comprising 2,455 pages of domain-specific documents with numerous non-decorative images, 316 queries, and 910 answer instances. Finally, we evaluate representative standalone Vision-Language Models from different providers and further design Q2IT, a multi-agent collaborative framework consisting of three progressively collaborating agents for evidence collection, page selection, and target-image localization. Using a joint evaluation approach that assesses both text extraction and image localization, our experiments show that standalone VLMs struggle with this task, while Q2IT significantly improves performance on ITJoint, although a substantial gap remains toward perfect results.
[AI-91] Joint Causal Structure and Cluster Discovery Using Variational Inference
链接: https://arxiv.org/abs/2608.22212
作者: Avni Rajpal,Anubhav Kumar,Rishabh Karnad,Mohammad Emtiyaz Khan,P.K. Srijith
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
备注:
Abstract:Causal discovery aims to understand the relationships between individual random variables. In many applications, such as brain imaging and climate modeling, it is more meaningful to consider interactions among groups of variables. Existing methods assume that knowledge of such groups or clusters is explicitly available when modeling interactions. However, in practice, these clusters as well as the causal relationships among them, are latent. In this paper, we present a novel approach based on variational inference to simultaneously infer both the latent clusters and causal structures. We learn an approximate posterior over clusters and graph-structure by considering variational distributions based on categorical and Bernoulli models respectively. We derive variational lower bounds and estimation techniques to learn variational and model parameters. The effectiveness of our proposed methods for cluster and causal discovery are demonstrated on both synthetic and real data sets.
[AI-92] Disagree to Explore Agree to Commit: Routing-Guided Test-Time Scaling for Software Agents
链接: https://arxiv.org/abs/2608.22191
作者: Kang Chen,Junjie Nian,Yixin Cao,Yugang Jiang
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注:
Abstract:Software-engineering agents solve repository-level tasks through long, stochastic tool-use trajectories, and repeated attempts often find fixes missed by one run. Test-time scaling is difficult because patches lack canonical answer forms, while sibling actions from a shared prefix are correlated. We study whether native MoE router traces can guide steering and selection without an external judge or selection-time test execution. Our analysis shows that routing provides a robust behavioral role signal; token-granular readouts and decision-matched comparison sets turn it into effective control. We therefore introduce Risa (Routing-Informed Steering and Arbitration): within trajectories, routing encourages diverse exploration and controlled convergence during patch commitment; across separately sampled trajectories, agreement at informative patch positions selects a final candidate. We evaluate on SWE-bench Verified using open-weight sparse MoE agents across scales and reasoning-effort settings. Risa’s routing arbitration raises the macro-average resolved rate from 44.9% under uniform sampling to 48.2% on the gpt-oss family, matching text consensus without answer-string matching, and it transfers to Qwen3.6, where it improves on uniform choice and matches text consensus on the full 500-task benchmark.
[AI-93] Role-Specialized Mixture-of-Agents with Open-Weight LLM s for Clinical Prediction
链接: https://arxiv.org/abs/2608.22176
作者: Jun Hou,Yi Fang,Xuan Wang
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: COLM 2026 DAIH Workshop
Abstract:Large Language Models (LLMs) are increasingly applied to clinical prediction tasks such as in-hospital mortality and readmission from electronic health records (EHRs). Privacy and compliance constraints motivate systems that can be deployed locally, which has increased interest in open-weight multi-agent designs. However, most medical multi-agent systems are evaluated as a single block, leaving unclear which agent role contributes to prediction and whether retrieval drives observed gains. We study a role-specialized Mixture-of-Agents (MoA) that combines medical knowledge retrieval with contrastive similar-patient reasoning. By varying the role design while holding the retrieval setup fixed, we localize the main effect to the final integrator. Pairing large open-weight analysts with a small open-weight integrator matches closed-model prompting on F1 for mortality prediction while flagging substantially more true high-risk patients. Mechanism analysis shows the role assignment directly yields a high-recall operating point without threshold tuning. The effect is task-dependent, with smaller gains for readmission because the available records correlate weakly with this longer-horizon outcome. These results position role design as a key factor in privacy-constrained, training-free clinical LLM prediction.
[AI-94] MCP-Universe RL: A Framework for Training MCP Tool-Use Agents via Reinforcement Learning
链接: https://arxiv.org/abs/2608.22167
作者: Ziyang Luo,Yan Yang,Xiangru Jian,Ziji Shi,Xiaoqiang Lin,Jun Hao Liew,Silvio Savarese,Junnan Li
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Technical Report
Abstract:Reinforcement learning (RL) has become an effective way to improve the tool-use ability of large language models (LLMs), but most existing RL frameworks stop at the policy update. For every new domain, the user is left with two hard systems problems: standing up an isolated environment for each of hundreds of concurrent trajectories and connecting it to training, and scheduling the rollout so that the GPU stays busy across long, multi-turn episodes that spend much of their time stalled on slow tool calls. We present MCP-Universe RL (MCP-U RL), an open-source framework that takes over both. It uses the Model Context Protocol (MCP) as the interface to the environment, so any tool already exposed as an MCP server plugs into training with no RL-specific integration code. It builds the two missing layers once and reuses them across domains: an environment-orchestration layer that provisions, isolates, and recycles the MCP environments over a pluggable container backend, and a rollout-orchestration layer whose staged pipeline overlaps trajectories to keep the GPU busy while episodes wait on tools. A backend-agnostic training layer then applies the update through an existing RL backend, with veRL and slime integrations. With one configuration, changing only the task specification, we train software-engineering, deep-research, and general tool-use agents on gpt-oss-20b and improve task reward in all three.
[AI-95] AUDITA: certified auditing and causal attribution of adverse outcomes in autonomous multi-agent systems
链接: https://arxiv.org/abs/2608.22160
作者: Zhixu Du,Yiran Chen
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Physical automation is scaling toward fleets of embodied machines commanded by an AI brain. Early deployments already run factories and warehouses at production rates beyond any human line, and their adoption is accelerating. But when their joint decisions cause harm, everyone involved has reason to blame everyone else, the machine vendor, the algorithm provider, the factory operator, the insurer, and the regulator, and no method can divide the responsibility between them. Existing methods read logs whose origin they cannot verify and name a single culprit, misrepresenting outcomes that are overdetermined, preempted, or caused by an omission. We present \audita, an audit layer pairing a tamper-evident record of every inter-agent command with a certified, graded causal-attribution engine. We prove its verdict cannot be gamed: a rule-following agent can never be made to look guilty, an attempt to shift blame is itself caught and graded, and we establish the exact limit of what an evidence-based auditor can certify. On live language-model pipelines it reduces the standard judge baseline’s responsibility error roughly threefold; on a benchmark of accident-grounded structures it recovers responsibility where single-culprit baselines fail, and stays invariant under forgery. \audita turns the question of who is to blame from an argument about logs into a calculation over evidence.
[AI-96] Why Does Robustness Reduce Superposition?
链接: https://arxiv.org/abs/2608.22155
作者: Adam Elimadi
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 9 pages, 4 figures. Accepted at the COLM 2026 Workshop on AI Interpretability (AIW)
Abstract:The study of adversarial examples and their origins remains an open area of research. Mechanistic interpretability, and superposition in particular, offers new avenues for approaching this problem. Gorton Lewis (2025) demonstrate that adversarial examples arise from superposition and show empirically that adversarial training reduces superposition, yet provide no mechanistic account of why this occurs. We present an empirical explanation inspired by the feature taxonomy of Ilyas et al. (2019), tracing the following chain of causalities: adversarial training abandons non-robust features, leading to fewer total features to represent, resulting in less superposition.
[AI-97] Meta-Ctrl: Guaranteed Plan Generation by Decoupling Syntactic and Semantic Constraints
链接: https://arxiv.org/abs/2608.22149
作者: Gwen Yidou-Weng,Edward Sun,Tianyi Ma,Metin Alp Dogan,Benjie Wang,Allen Peng,Guy Van den Broeck,Yuchen Cui
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:LLMs generate fluent plans for robots but routinely violate the syntactic and se8mantic constraints they must satisfy to execute, and existing remedies trade formal guarantees against plan quality: soft methods (affordance scoring, grounded decoding) give no guarantee, while symbolic planners (LLM+P) discard the LM’s commonsense. We propose \textbfMeta-Ctrl, a constrained-decoding framework that guarantees the encoded constraints while preserving the base LM’s plan quality. Meta-Ctrl introduces \emphmeta-tokens—a compact vocabulary of grounded actions—enforcing syntax at the token level and semantics (preconditions, goals, ordering) at the action level, an exact factorization that cuts the memory of constrained decoding from over 107TB to under 2GB. With it, a small open-weight LM becomes competitive where it otherwise sits at the bottom of the leaderboard: on WAH-NL under the LoTa-Bench protocol it reaches the highest reported subgoal success rate, exceeding GPT-4’s, with consistent gains across the Embodied Agent Interface. We further demonstrate it on a real tabletop robot, where every generated plan satisfies its preconditions and goals by construction. Project website: this https URL.
[AI-98] Evaluation of Small Vision-Language Models on Qualitative Mechanical Problems
链接: https://arxiv.org/abs/2608.22143
作者: Henry Fordjour Ansah(1),Shreya Banerjee(1),Pranish Ghimire(1) ((1) Louisiana State University of New Orleans)
类目: Artificial Intelligence (cs.AI)
备注: 8 pages, 11 figures
Abstract:Qualitative mechanical problem-solving (QMPS) refers to solving qualitative problems from the mechanical domain. Qualitative problems can be solved with minimal discipline-specific information, without any robust quantitative calculation, generally by using qualitative reasoning and commonsense knowledge. QMPS is a vital aspect of human intelligence that allows us to tackle a wide range of tasks, from simple everyday ones such as turning on a tap to complex tasks in highly demanding and well-paying jobs in various fields, e.g., emergency medicine, plumbing, driving, etc. Employers often use the Bennett Mechanical Comprehension Test (BMCT) to evaluate job candidates’ ability to solve such problems. In this work, we assess two state-of-the-art multimodal models, Gemma-3 and Qwen-VL, on their ability to interpret mechanical problem images by eliciting a step-by-step chain of thought (CoT) and a final answer. Each image inherently encodes ground-truth qualitative facts, such as contact points in gears, support relations, and relative weights, which we use to evaluate each model’s spatial and commonsense reasoning capabilities. We assess each chain for coherence, completeness, and logical progression to assess each model’s thought process, and final answers are compared to verified solutions to measure accuracy.
[AI-99] MEMONDEMAND: A Memory Management System for Large-Scale Enterprise Data
链接: https://arxiv.org/abs/2608.22141
作者: Xinyuan Song,Bowen Zhu,Hasibul Haque,Liang Zhao
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Enterprise repositories are large, heteroge- neous, and continuously updated, making re- trieval difficult when efficient access, source- faithful evidence, and cross-query adaptation must be supported together. Enterprise mem- ory extends retrieval beyond the model con- text, but existing systems do not jointly address collection-specific hierarchy construction, low- cost routing, detailed evidence loading, and workload-aware memory updates at this scale. We introduce MEMONDEMAND, short for On- Demand Memory, a memory management sys- tem with three coordinated mechanisms: a dy- namic multi-level hierarchy that determines the abstraction structure and depth for each col- lection, dual memory at every hierarchy level that separates distilled routing from detailed evidence, and on-demand memory promotion that updates node priority under a bounded active-state budget. On EnterpriseRAG-Bench, MEMONDEMAND outperforms the strongest published LB#1 result at every evaluated scale from 10M tokens through the complete 618M- token collection, with gains of 12.23% at 10M and 4.66% at 618M. Results on FinanceBench, HotpotQA, and FRAMES further show strong performance across financial, multi-hop, and fact-retrieval settings. Together, these results establish MEMONDEMAND as an accurate, ef- ficient, and scalable memory solution for very large enterprise repositories across data scales, domains, and evidence requirements. Our code is available at this https URL xfab-xinyuansong/MemOnDemand.git.
[AI-100] MegaMem: A Retrieval Solution for Ultra-Large Context Windows
链接: https://arxiv.org/abs/2608.22137
作者: Xinyuan Song,Bowen Zhu,Hasibul Haque,Liang Zhao
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Modern language models and agents increasingly require persistent memory for complete codebases, long interaction histories, and heterogeneous enterprise records. The key challenge is to keep hundreds of millions of tokens searchable while passing only bounded source evidence to the answer model. We introduce MegaMem, a source-resolved dual-view retrieval system that separates semantic access from generation evidence. Distilled records and detailed evidence are searched with original and transformed queries; every distilled hit resolves to an immutable source ID before reciprocal-rank fusion, deduplication, and cross-encoder reranking; and only the highest-ranked detailed evidence within a fixed budget supports generation. Post-answer attribution then identifies which loaded sources support the fixed answer. We evaluate MegaMem on EnterpriseRAG-Bench, which contains more than 500,000 heterogeneous enterprise documents and approximately 650M tokens. MegaMem improves Overall from 68.22 to 82.26 and reaches 86.50 Correctness. These results show that MegaMem supports ultra-large persistent memory while preserving strong answer accuracy under a bounded generation context. By separating searchable memory scale from answer-context size, MegaMem provides a practical path toward accurate retrieval over memories ranging from hundreds of millions to one billion tokens. Our code is available at this https URL xfab-xinyuansong/MegaMem.git.
[AI-101] ask-Driven 3D Printability Assistance via Geometry- and Knowledge-Grounded LLM Reasoning
链接: https://arxiv.org/abs/2608.22128
作者: Zhaoda Du,Qiaojie Zheng,Xiaoli Zhang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Printability assessment in additive manufacturing is typically conducted at the geometry level before printing to determine whether a computer-aided design (CAD) model or stereolithography (STL) file can be successfully fabricated. Task suitability, in contrast, is usually evaluated after printing to determine whether the fabricated part satisfies the requirements of its intended use. As a result, for non-expert users to print functional parts, unsuitable material or process choices may only be identified after fabrication, leading to repeated printing, material waste, and user frustration. To address this challenge, this paper leverages the reasoning and language-understanding capabilities of large language models (LLMs), while grounding the reasoning with geometry evidence and structured material/printer knowledge to generate reliable pre-print recommendations. Given a stereolithography (STL) model and a natural-language task description, the framework generates a structured recommendation covering printability, material choice, process parameters, design guidance, risks, and explanations. We evaluate the framework on focused STL benchmark scenarios with novice-style task descriptions. The proposed method achieves 75.0% printability over 96 physical validation trials, with 88.9% task suitability among successfully printed samples. It also improves Gemini 2.5 Flash-Lite material-selection accuracy from 37.5% under pure LLM to 90.0%. Expert evaluation further shows improved report quality, while post-print feedback improves recommendations on selected problematic cases. These results suggest that user task intent, geometry evidence, and structured material knowledge are all important for reliable task-driven printability assistance.
[AI-102] Development and Feasibility Evaluation of an Edge AI as Medical Device System for Breast Cancer Multidisciplinary Team Meetings
链接: https://arxiv.org/abs/2608.22108
作者: Aarzoo Dhiman,Farzana Haque,Kartikae Grover,Lydia Brian Smith,William Stephen Jones
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Breast Cancer Multidisciplinary Team (MDT) meetings manage increasingly complex cases under considerable time pressure, and documentation requirements can reduce clinical efficiency and decision quality. Existing AI based MDT workflows rely on cloud-based processing, limiting their use because patient discussions contain identifiable information. We developed a fully on-device AI pipeline using open-source Automatic Speech Recognition (ASR) and Large Language Models (LLMs) that transcribes breast cancer MDT discussions, structures clinical information, and generates treatment recommendations using retrieval-augmented generation (RAG) grounded in National Institute for Health and Care Excellence (NICE) guidance. The pipeline runs on a single NVIDIA Jetson AGX Orin, ensuring that patient audio, transcripts, and outputs remain within institutional infrastructure. Evaluation included two recorded simulated MDT discussions, ten clinically validated synthetic discussions, and 1,270 acoustically augmented recordings. Optimisation of Whisper large-v3 reduced word error rate by 20.7% and 24.4% on the recorded discussions and achieved performance within 0.58% WER and 1.58% word information lost of a commercial clinical ASR benchmark on augmented audio. MedGemma-RAG identified 2.3 times more MDT-concordant interventions than a proprietary cloud comparator (p = 0.020), with no significant difference in overall accuracy. Stakeholders identified automated documentation, treatment recommendation support, and case triage as the most credible near-term applications while highlighting workflow integration, governance, and clinician trust as key implementation challenges. These findings demonstrate the feasibility of privacy-preserving, fully on-device AI for MDT documentation and guideline-informed decision support, providing a foundation for prospective clinical evaluation.
[AI-103] Hack-Verifiable Terminal Bench: Evaluating Reward Hacking in Terminal Tasks
链接: https://arxiv.org/abs/2608.22103
作者: Amit Roth,Ivan Bercovich,Yonathan Efroni
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:As agents grow more capable and autonomous, their tendency to reward hack, satisfying a task’s checks while violating its intent, becomes an increasingly important failure mode. Measuring reward hacking is itself challenging, as detection typically relies on human inspection or LLM judges, both of which can be unreliable. The hack-verifiable environments (HVE) methodology addresses this challenge by embedding detectable hacks into tasks, allowing reward hacks to be identified automatically and reliably. In this work, we adapt HVE to Terminal Bench, a leading benchmark of real-world terminal and coding tasks, and introduce Hack-Verifiable Terminal Bench (HVTB). Using HVTB, we measure reward-hacking rates across frontier models and study whether prompts with varying amounts of information on the hack can mitigate this behavior. This lets us test whether prompting can prevent not only known reward-hacking strategies, but also ‘unknown unknown’ exploits that the prompt does not anticipate. We release all environments and agent traces at this https URL
[AI-104] On Predicting Vulnerability Severity Using In-Context Learning: An Industrial Case Study
链接: https://arxiv.org/abs/2608.22089
作者: Daniel Rodriguez-Cardenas,David Nader Palacio,Anna Schmedding,Yiyang Lu,Aadil Mallick,Bill Hudson,Chris Gourley,Michael Roytman,Chris Shenefiel,Evgenia Smirni,Denys Poshyvanyk
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: 11 pages, 6 figures
Abstract:Modern software systems require earlier and more scalable vulnerability severity assessment to reduce exposure to high-impact security flaws. Security analysts typically assign CVSS scores, but this manual triage does not scale with the growth of disclosed vulnerabilities and often depends on cloud LLM services that raise confidentiality concerns. This paper presents an industrial case study on predicting CVSS v3.1 scores directly from vulnerable C/C++ snippets using in-context learning with locally deployable, open-source LLMs. We compare proprietary data with the Big-Vul dataset, showing sufficiently aligned CVSS distributions to justify Big-Vul as a proxy for industrial data when constructing prompt-based testbeds. We then vary in-context configurations and model parameters, evaluating CodeLlama2-7B, CodeLlama2-13B, Mistral-7B, gpt-oss, and GPT4o-mini using mean squared error (MSE) and feasibility metrics. Our results show that medium-sized open-source code models, particularly CodeLlama2-7B, can approximate the best cloud performance for CVSS regression when guided by lightweight, output-constraining prompts, offering a practical, privacy-preserving building block for severity triage in industrial settings.
[AI-105] Dissecting Neuro-Symbolic Quality Assurance for Synthetic Oncology Data Generation
链接: https://arxiv.org/abs/2608.22085
作者: Laxmigayathri Challa,Yuhan Zhou,Ana Cleveland,Haihua Chen
类目: Artificial Intelligence (cs.AI)
备注: 14 pages, 7 figures
Abstract:Synthetic clinical data generation with large language models addresses the scarcity that limits cancer staging research, but oncology hallucinations are categorically harmful: one clinically impossible staging assignment contaminates every downstream model trained on it. Neuro-symbolic pipelines validate during generation, yet the contribution of individual quality-assurance components remains unclear. We report three controlled studies isolating gate necessity, constraint attribution, and retrieval conditionality, holding generation protocol, diversity thresholds, and fine-tuning hyperparameters constant across adapter conditions. The symbolic gate enforces schema completeness, ontology coverage against the Systematized Nomenclature of Medicine, and staging-logic consistency under American Joint Committee on Cancer eighth-edition rules. Ungated, 29.9% of records contain schema failures and 20.1% contain clinically invalid staging. Schema validation is the load-bearing filter: within the fully gated corpus it rejects 148 of 512 records, ontology grounding a further 24, and staging-logic validation none—the only generator producing logic violations is already excluded on schema, making clinical-logic validation a generator-conditional safeguard rather than the dominant filter. Retrieval augmentation is strongly model-dependent: it improves gate compliance for one generator by 12.5 percentage points, has no measurable effect for a second, and collapses output in a third. Across gated configurations ontology density is largely unchanged, indicating that symbolic validation improves clinical validity rather than vocabulary richness. Symbolic gating therefore buys corpus validity but no commensurate gain on real lung-cancer notes in this study; retrieval should be evaluated per model, and ontology density should not be reported as a proxy for corpus quality.
[AI-106] Improving Energy Efficiency of Oil Platforms Through Optimal Loading of Diesel Generators Using Machine Learning and Search Algorithms
链接: https://arxiv.org/abs/2608.22076
作者: Khivishta Boodhoo,Josh Plumbly,Nicholas Watson
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Rising energy demand, fossil fuel depletion and climate change highlight the need for more efficient energy production and consumption. Offshore oil and gas platforms face challenges related to inefficient energy use, system failures, accessibility and environmental impact. Machine learning (ML) offers opportunities to improve the safety, sustainability and efficiency of these systems; however, previous research has largely focused on increasing oil production rather than reducing energy consumption on platforms. This study investigates the use of ML and search algorithms to improve diesel efficiency on an offshore oil platform. Data collected over 18 months from a platform in Scotland were analysed, focusing on four diesel generators as the primary diesel-consuming equipment. Following exploratory data analysis and outlier detection, regression models were developed to predict daily diesel consumption for different generator power loads. Multiple Linear Regression and Artificial Neural Networks achieved the best predictive performance compared with Extra Trees Regression, Extreme Gradient Boosting and Random Forest. Search algorithms were then used to identify combinations of generator power loads that minimised daily diesel consumption. The results showed an average diesel saving of 27% per day compared with the worst daily power-load combinations, equivalent to approximately 24,000 litres/day. These findings demonstrate significant opportunities for improving energy efficiency on offshore oil platforms using ML-based optimisation.
[AI-107] Decision-Support and Modeling with Large Language Models for Geothermal Well Arrays
链接: https://arxiv.org/abs/2608.22068
作者: Edwin Ouko,Emmanuel Lujan,Alan Edelman,Robert Metcalfe
类目: Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE)
备注: 10 pages, 8 figures. Presented at the 50th Workshop on Geothermal Reservoir Engineering, Stanford University, February 10-12, 2025
Abstract:Geothermal well arrays, which organize multiple geothermal wells into carefully planned geometric configurations, provide opportunities to enhance energy production capacity and increase fault tolerance. The development and adoption of these emerging geothermal technologies could be accelerated through the recent advances in large language models (LLMs) and high-level high-performance languages. A challenge in LLM-based applications is the reliability of the generated outputs, as they can be prone to subjective biases and hallucinations. This study assesses the potential of cutting-edge LLMs - such as ChatGPT, Gemini, Claude, Grok, and domain-specific models like AskGDR - as expert assistants that can synthesize insightful interpretations of complex geothermal data, as well as improve feature capabilities of geothermal models and numerical software. We developed a novel approach, leveraging Google’s recently introduced AI assistant, NotebookLM, to accelerate the generation of unpublished quantitative geothermal benchmarks. The rapid generation of these evaluation instruments is essential for assessing the swiftly evolving capabilities of emerging language model technologies. In particular, we use these benchmarks and LLM-based interviews to analyze opportunities and limitations of two promising technologies: geothermal well arrays and closed-loop coaxial wells. Furthermore, we present a case study illustrating how LLMs can facilitate auto-parallelization of geothermal numerical models. Our analysis emphasizes their application in digital twins and underscores the importance of high-level, high-performance code generation. This line of research could play a transformative role in the geothermal sector by enabling the next-generation of decision-support applications, integrating data analysis, informed recommendations, and more dynamic numerical modeling workflows.
[AI-108] From SQL Generation to Tool Selection: A Domain-Oriented Pattern for MCP Servers
链接: https://arxiv.org/abs/2608.22063
作者: Bartolomeo Bogliolo
类目: Artificial Intelligence (cs.AI); Databases (cs.DB)
备注:
Abstract:Agents built on Large Language Models (LLMs) increasingly reach enterprise data through the Model Context Protocol (MCP), and many MCP database servers maximize flexibility by exposing a single generic SQL execution tool. This paper proposes the Domain-Oriented Tooling Pattern: instead of generating SQL at query time, the model selects from a small set of domain-aligned tools whose parameterized queries encapsulate schema navigation, joins and business rules on the server side. We formalize the pattern around three architectural invariants and introduce Model Demotion, the observation that replacing SQL synthesis with intent classification lowers the model tier required to serve routine requests. As a reference implementation we present MCP Blueprint, an open-source framework in which domain tools are defined declaratively as YAML metadata plus external parameterized SQL files. We evaluate the pattern with a public reproducibility benchmark comparing three MCP server designs - raw SQL execution, a thin generic tool pack, and a verticalized domain pack - on four local models (3B-8B) across seventeen customer-facing tasks over the Sakila database (609 completed cells; temperature 0; three repetitions per cell). The verticalized pack reaches a pooled mean score of 0.939 versus 0.666 for raw SQL and 0.605 for the generic pack; the smallest model improves from 0.583 to 0.929, matching or exceeding every larger configuration while cutting cost per correct answer by an order of magnitude. All harness code, prompts, gold answers, frozen packs and per-cell results are publicly available.
[AI-109] Search Broadly Seek Evidence on Both Sides Decide Narrowly: Evidence-Admissible GraphRAG for Longitudinal Clinical Event Verification AAAI2027
链接: https://arxiv.org/abs/2608.22062
作者: Xingtao Lin,Yubo Feng,Weixin Liu,Hangqi Ren,Junchao Zhou,Caiwan Sun,You Chen
类目: Artificial Intelligence (cs.AI)
备注: Submitted to AAAI 2027
Abstract:Longitudinal clinical event-relation verification determines whether a patient record supports a specified relation among two or more clinical events. This task is challenging because evidence is distributed across structured records, notes, laboratory trajectories, encounters, and time, while negation, temporal mismatch, repeated documentation, and conflicting findings can make retrieved information appear relevant without establishing the relation. We present MedEventGraph-RAG, an evidence-admissible framework that represents event occurrences in a patient-specific graph and links each occurrence to source evidence, including structured rows, note spans, timestamps, and numerical trajectories. Given a verification query specifying events, relation, and clinical scope, the graph guides discovery of candidate event chains and retrieves evidence from both supporting and contradicting sides. A query-specific evidence contract filters information by patient identity, scope, occurrence binding, and source traceability before a separate assessor determines supported, conflicting, refuted, or insufficient outcomes. Across ten protocols on i2b2, n2c2, MIMIC-IV, and LUNGUAGE, MedEventGraph-RAG achieves balanced accuracies of 78.6, 67.3, and 96.8 on temporal, medication-adverse-event, and recorded-order verification, improving over the strongest matched baselines by 26.9, 4.9, and 30.4 points. Under evidence masking, it reaches 92.2 balanced accuracy with no false-support predictions. When intermediate events are hidden, it recovers complete source-traceable event chains in 57.9% of i2b2 and 70.0% of LUNGUAGE cases. These results show that separating broad evidence discovery from narrow evidence-admissible assessment improves longitudinal clinical verification and reduces unsupported conclusions. Comments: Submitted to AAAI 2027 Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2608.22062 [cs.AI] (or arXiv:2608.22062v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2608.22062 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-110] MEMORY Wins All: Indirect Bias Injection Attacks via Social Media Feeds
链接: https://arxiv.org/abs/2608.22061
作者: Minjae Seo,Wonwoo Choi,Geonwoo Han,Taekyoung Kwon,Yongsu Kim,Sang Seo,Jaewon Noh,Hankyul Baek,Seongyun Seo,Myoungsung You
类目: Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注: 18 pages, 7 figures, 11 tables
Abstract:Personal AI agents routinely consume external content while performing tasks such as web browsing, email processing, and SNS feed summarization, and they retain selected information or execution results in persistent memory for later use. We show that this ordinary ingestion of external content opens an indirect path for manipulating subsequent agent behavior. Based on this observation, we present IBIA, an Indirect Bias Injection Attack that plants an adversary-aligned stance on a specific topic into a victim agent’s memory through external content, without direct access to the agent, its memory, or future user queries. For this, IBIA combines three mechanisms: comment cloaking, which keeps the crafted content consistent with the surrounding discussion, comment watermarking, which enables lightweight identification during curation, and category anchoring, which makes the retained stance salient under later related requests. We evaluate IBIA on BiasBench, a benchmark of 6,000 adversary-crafted social comments and 120 email instances. The watermark-based curation identifies 95.9% of the injected comments. Under the OpenClaw setting, IBIA achieves adversary-aligned response rates (AARs) of 91.2% on average across four downstream tasks, including 86.6% on the frontier GPT-5.5. We further propose a memory boundary defense that detects the injected bias and reduces AARs to 80.6%.
[AI-111] GenCoord: Skill-Path Commitments under Private Information
链接: https://arxiv.org/abs/2608.22055
作者: Peng He,Junning Zhu,Haohan Yuan,Jianpeng Liang
类目: Artificial Intelligence (cs.AI)
备注: 25 pages, 10 figures, and 23 tables, including supplementary material. Peng He and Junning Zhu contributed equally. Paper source and compact evidence: this https URL
Abstract:Suppose one embodied agent knows what must be built, while its teammate alone knows which transformation its workcell can perform. Neither local view determines who should act, what should be handed off, or how the joint task should continue. We introduce GenCoord, which turns the task consequence of such private facts into an executable skill-path commitment. A local Qwen3.5-0.8B model emits a multi-step SELF plan and peer REQ; bounded feedback conditions route revision when the deciding capability is peer-local. The resolved commitment is parsed, checked, canonically materialized, compiled to Mineflayer skills, and verified by handoff and terminal state. Counterfactual interventions that hold the world, call schedule, and executor unchanged make requester revision and receiver execution follow the injected task consequence in both directions. Across three independently trained seeds, correct capability feedback closes the paired local-information gap from 50% to 100%. Multi-step commitments improve held-out-template success by 6.9 points while reducing model decisions by 32%. At matched closed-loop quality on 128 held-out semantic clusters, Short DSL reduces peer traffic by 92.8% and median time-to-commitment by 68.2% relative to controlled free-form communication. These results identify executable task consequences as the coordination unit connecting distributed local reasoning to verified joint action.
[AI-112] More Accurate or More Efficient? Evaluating Locally Deployed Compact Open-Weight Language Models for Mathematical Reasoning
链接: https://arxiv.org/abs/2608.22048
作者: Orion Powers,Daniella Seum,Khaled Slhoub
类目: Artificial Intelligence (cs.AI)
备注: 8 pages, 1 figure, 6 tables
Abstract:Large language models are increasingly deployed on local hardware for privacy, cost, and accessibility reasons. Yet many evaluations emphasize accuracy while fewer quantify local runtime and energy, characterize failure modes, or apply paired statistical comparisons under controlled conditions. This paper presents a controlled, documented procedure for evaluating locally hosted LLMs on mathematical reasoning. It combines fixed inference settings, hierarchical answer extraction and verification, explicit failure-mode classification, and per-question resource measurement, and reports accuracy with paired significance tests and effect sizes. We demonstrate it in a preliminary study of three compact open-weight models under five billion parameters, Gemma3:4b (Google), Phi3:3.8b (Microsoft), and Qwen3:4b (Alibaba), across datasets spanning Grade 8 Math, Calculus I, and Advanced Probability and Statistics. All models ran through the same local inference server on one workstation, using a shared prompt template, controlled settings, and a matched question set per dataset. No single model dominates. Qwen3:4b is most accurate on two datasets and Gemma3:4b on Calculus I, yet Gemma3:4b returns roughly three times more correct answers per watt-hour than Qwen3:4b on every dataset while generating far fewer output tokens; Qwen3:4b requires substantially more generation time, energy, and output per question. Phi3:3.8b is substantially less accurate on all three datasets; its low extraction-failure rate indicates incorrect answers rather than unparsed output, though we caveat possible prompt-format effects. These preliminary findings indicate that accuracy alone is an insufficient basis for selecting a local model.
[AI-113] Bi-EZP: LLM -Guided Bilevel Program Evolution for Ensemble Zero-Cost Proxy Discovery
链接: https://arxiv.org/abs/2608.21927
作者: Yutao Lai,Kezhao Lai,Hai-Lin Liu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 14 pages, 4 figures, 8 tables
Abstract:Zero-cost proxies enable neural architecture search (NAS) to rank candidate networks from statistics computed at initialization, avoiding repeated training. However, different proxies capture different properties and often produce inconsistent rankings across search spaces. Ensemble proxies can combine complementary signals, but automated discovery must optimize both discrete aggregation structures and their continuous coefficients, making structural quality difficult to separate from parameter calibration. We propose Bi-EZP, a bilevel framework that decouples these decisions. At the upper level, a large language model generates executable aggregation programs over four complementary base proxies with program-specific parameter bounds. At the lower level, covariance matrix adaptation evolution strategy (CMA-ES) optimizes the continuous parameters of each fixed program on an inner training split. The calibrated programs are then evaluated using Kendall’s rank correlation on a disjoint validation split, enabling evolutionary selection to favor structures that generalize beyond their calibration data. Experiments on NATS-Bench and Network Design Spaces evaluate ranking performance across heterogeneous search spaces, and DARTS experiments assess downstream architecture search. Results show that separating program discovery from numerical calibration provides an effective approach to automated ensemble zero-cost proxy construction. The source code is available at: this https URL
[AI-114] ESCRAG -R1: Retrieval-Augmented Reinforcement Learning for Emotional Support Conversation
链接: https://arxiv.org/abs/2608.21925
作者: Weichu Liu,Yuxuan Hu,Yirong Sun,Ningning Mao,Ziyun Zhang,Jian Chen,Mingyang Xu,Qishan Zhong,Chengming Li
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Emotional Support Conversation (ESC) systems aim to provide holistic support by balancing professional therapeutic competence with natural empathy. However, existing methods struggle to simultaneously achieve structured, stage-aware reasoning and seamless empathy-expertise alignment, often resulting in an artificial splicing of clinical strategies and generic reassurance. To overcome these limitations, we propose ESCRAG-R1, a unified framework that integrates retrieval-based psychological guidance into Group Relative Policy Optimization (GRPO). By incorporating retrieval into the reinforcement learning loop, ESCRAG-R1 transforms external knowledge into a robust learning signal that stimulates explicit internal reasoning prior to generation and fundamentally reshapes the model’s internal policy. To provide the reliable supervision required for this optimization, we construct ESC-Preference, a high-quality dataset based on a Client–Counselor–Judge evaluation framework that delivers precise, empathy-aware reward signals. Extensive experiments demonstrate that ESCRAG-R1 significantly outperforms existing baselines by mitigating superficial splicing and realizing a natural integration of professional guidance and empathetic expression. Code and datasets are released at this https URL.
[AI-115] Consistency Is Not Coherence: Orientation Search for Certified Alignments Between 4D Defence Upper Ontologies ATC ISWC2026
链接: https://arxiv.org/abs/2608.21914
作者: Fabio Rovai
类目: Artificial Intelligence (cs.AI); Databases (cs.DB); Logic in Computer Science (cs.LO)
备注: 14 pages. Long paper accepted at the Ontology Matching 2026 workshop (OM 2026), co-located with ISWC 2026. Code and data: this https URL
Abstract:We align three upper ontologies that sit under UK and NATO defence data infrastructure: the Information Exchange Standard (IES), the Higher Quality Data Model (HQDM) that underpins the National Digital Twin, and Basic Formal Ontology (BFO). No public alignment between IES and HQDM existed. Promoting a hand-curated 17-correspondence crosswalk to OWL and reasoning over the complete merged ontologies with HermiT produces three results that we believe matter beyond this pair.
[AI-116] raining Needs Trustworthy Worlds: Verified Synthetic Web Environments for Agent Learning
链接: https://arxiv.org/abs/2608.21898
作者: Chenghao Zhang,Canran Xiao,SaiSai Hu,Dan Roth
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Web agents promise to automate complex digital workflows, but their training remains limited by synthetic environments that look plausible while hiding broken links, inconsistent states, or infeasible tasks. We address the gap between scalable environment generation and trustworthy agent learning by constructing synthetic web environments that are executable, auditable, and grounded in backend state. Our framework represents each generated website as a structured scaffold of pages, navigation links, database records, state-change markers, and task constraints, then verifies and repairs structural, semantic, consistency, and feasibility defects before policy training. During interaction, ordinary UI transitions are executed deterministically, while persistent backend updates are invoked only through validated state-change markers, enabling dense rewards compiled from verified task-progress predicates. Across 500 synthetic environments spanning six domains, our method reduces task-blocking defects and improves feasible-task rate from 48.6% to 94.8%, while producing stronger PPO policies and improving transfer to WebArena, WebShop, and MiniWoB++ without LLM calls at evaluation time. These results show that verified synthetic environments can serve as a scalable and reliable training substrate for compact web agents, shifting synthetic webagent learning from surface-level plausibility toward executable, state-grounded supervision.
[AI-117] From Solver Feedback to Faithful Plans: Multi-Role Reinforcement Learning for Symbolic Planning
链接: https://arxiv.org/abs/2608.21897
作者: Chenghao Zhang,Yikai Mao,Shanqi Liu,Haoyu Gao,SaiSai Hu,Dan Roth
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Reliable planning requires converting natural-language instructions into executable symbolic specifications, yet large language models remain brittle without costly PDDL annotations and may exploit solver success in semantically unfaithful ways. We study how to learn faithful natural-language-to-PDDL formalization using only solver feedback, without human-written demonstrations. We propose a solvergrounded multi-role reinforcement learning framework where a single language model acts as an Actor, Judge, and Editor for generation, verification, and repair. The Actor proposes PDDL specifications, the Judge provides a solver-calibrated quality signal, and the Editor performs bounded diagnostic-conditioned refinement. On PlanBench, our method improves average success from 35.5% for LLM+P to 70.8%, achieves 66.3% faithful success, and reduces semantic drift to 6.4%. These results show that organizing solver feedback into generation, verification, and repair roles enables more scalable and faithful annotation-free symbolic planning
[AI-118] Breaking the Assumptions: Auditing Input-Side Jailbreak Defenses Against Semantic Attacks
链接: https://arxiv.org/abs/2608.21895
作者: Aaditya Pratap,Harsh Kasyap,Somanath Tripathy
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:Locally deployed Large Language Models (LLMs) via inference engines such as Ollama run without the moderation and abuse detection present in API-served models. Therefore, the safety of LLMs depends on the defense mechanisms used, and their effectiveness depends on the assumptions on which they were designed. This paper does an audit of defense mechanisms under jailbreak attacks on locally deployed models. Some defenses provide formal guarantees (SmoothLLM, Erase-and-Check, Sequential Monitors), while others rely on empirical detection results (Semantic Smoothing, Self-Denoised Smoothing, Perplexity Filtering). Instead of merely observing that defenses fail, we trace each failure back to the specific assumption: for every defense, we extract the condition it relies on, derive the empirical pattern a violation should produce, and test that prediction on six open-weight models (14B to 35B parameters) with a corpus of 100 jailbreak prompts taken from more than 40 public sources, totalling 13,800 evaluation records.
[AI-119] Pruned Traffic Trees: Native Semantic Compression with a Protocol-Structured Model Family for Encrypted Traffic Classification
链接: https://arxiv.org/abs/2608.21874
作者: Yuantu Luo,Jun Tao,Xiangyu Xu,Linxiao Yu,Kangying Li
类目: Networking and Internet Architecture (cs.NI); Artificial Intelligence (cs.AI)
备注: This paper is submitted to INFOCOM 2027
Abstract:Deep learning has achieved strong performance in encrypted traffic classification (ETC), yet its computational cost limits deployment on resource-constrained network devices such as routers and middleboxes. Existing compression methods mainly operate on weights, channels, hidden representations, or predictions, but do not explicitly determine which protocol fields and structural contexts should remain. We propose Pruned Traffic Trees (PTT), a three-level protocol-structured model family that treats native protocol structures as compression units. PTT-Full learns protocol-structured representations and field salience from complete Protocol Tree Graphs (PTGs), with flow-level self-supervised learning and protocol-presence-aware sparse execution. The learned salience and TopK+ k closure construct Distilled PTGs (PTG-Ds) for PTT-Distilled, while PTT-Lite inherits this topology and reduces width through structure-aligned transfer and flow-level logits distillation. Under flow-disjoint and Strong Information Information (SII)-masked settings, PTT-Full achieves Macro-F1 scores of 0.9519 and 0.9416 on CSTNET-TLS1.3 and CipherSpectrum, while PTT-Lite retains 0.9325 and 0.9136 with 80.3% and 61.3% fewer parameters, 98.85% and 98.78% lower effective GFLOPs, and 8.75 \times and 8.46 \times CPU inference speedups. These results demonstrate that treating protocol structure itself as the compression object enables effective performance-efficiency trade-offs for lightweight ETC.
[AI-120] HiMA-MDD: A Hierarchical Multi-Agent Harness for Interpretable Multimodal Depression Detection in Clinical Interviews
链接: https://arxiv.org/abs/2608.21868
作者: Ao Chen,Xiaojiang Peng
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Depression assessment from multimodal clinical interviews requires integrating dispersed evidence from multiple symptoms into a coherent PHQ-8 profile. This process is hierarchical: relevant evidence is often sparse and context-dependent within local question-answer exchanges, multiple exchanges jointly support symptom-level judgments, and the final assessment depends on the coherence of the complete symptom profile. Existing LLM systems either process interviews holistically or distribute work across generic agent roles; neither design necessarily provides an explicit orchestration mechanism that coordinates evidence access, item-score authority, bounded feedback, and state recording across these levels. To address this gap, we introduce HiMA-MDD, a hierarchical multi-agent harness that aligns this assessment hierarchy with three agent layers. After non-agentic preprocessing constructs context-preserving multimodal QA units, Layer 1 identifies candidate QA-to-item relations and supports bounded item-grounded evidence routing. Layer 2 assigns symptom groups to operational factor specialists, with one specialist responsible for each provisional item score. Layer 3 audits the complete provisional profile, requests at most one round of targeted revision, and reconstructs the verified PHQ-8 profile. This layered design naturally yields a Hierarchical Evidence Trace, preserves all intermediate evidence, judgments, and revisions for auditability. The final item scores then deterministically produce the total score and screening decision. Using Qwen2.5-72B-Instruct as the harness backbone, our experiments on E-DAIC demonstrate that HiMA-MDD outperforms the compared state-of-the-art methods.
[AI-121] ChainPrune: Evaluating and Reducing Redundancy in Long Chain-of-Thought Reasoning
链接: https://arxiv.org/abs/2608.21860
作者: Weihang Pan,Zhengxu Yu,Yuxiang Zhang,Wenzhi Li,Zhongming Jin,Binbin Lin,Xiaofei He,Jieping Ye
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 15 pages, 12 figures, 4 tables
Abstract:Chain-of-Thought (CoT) reasoning has significantly enhanced the multi-step problem-solving capabilities of large language models (LLMs) by introducing explicit intermediate reasoning. However, advanced Large Reasoning Models (LRMs) often exhibit overthinking behaviors, including excessively long reasoning steps, redundant steps, and high computational overhead. Existing token-length reward strategies aim to promote concise outputs, but often result in pseudo-conciseness, where token count is reduced, yet redundant reasoning persists, leading to longer and less structurally efficient chains. To address these limitations, we propose ChainPrune, a novel reasoning path semantic structural optimization method to efficiently and controllably synthesize self-generated high-quality training data. We initially consolidate self-generated reasoning paths into a tree-based structure, followed by a multi-criteria dominant path selection process for preference data construction that formulates shallow reasoning trajectories while preserving essential reasoning steps. To further enhance the quality of reasoning, we incorporate a DPO-based preference learning method combined with supervised loss, effectively mitigating false reward suppression. This innovative integration significantly enhances both the efficiency and effectiveness of our reasoning framework. Comprehensive experimental results demonstrate significant reductions in step length and computational overhead, while maintaining or even enhancing accuracy.
[AI-122] LLM s are Few-Shot Decision-Makers: Generalized Context-Aware Microgrid Frequency Control through Prompt Decision Transformer
链接: https://arxiv.org/abs/2608.21858
作者: Xu Yang,Chenhui Lin,Haotian Liu,Kaihang Deng,Yunhe Li,Wenchuan Wu
类目: ystems and Control (eess.SY); Artificial Intelligence (cs.AI)
备注:
Abstract:The rapid evolution of energy structures has positioned microgrids as pivotal components of next-generation power systems, offering enhanced resilience and renewable energy integration. However, the inherent low inertia, complex dynamics, and poor model conditions of microgrids necessitate advanced data-driven frequency control strategies. Although reinforcement learning (RL) has demonstrated certain potential and advantages, existing RL methods often struggle with generalization across diverse microgrid configurations and lack adaptability to unseen environments, particularly when explicit system parameters are unavailable. To address these challenges, in this paper, we introduce a novel prompt decision transformer (Prompt-DT) architecture for microgrid frequency control. Unlike traditional approaches that rely on hard-to-obtain environmental characteristic parameters, the proposed method leverages few-shot expert historical trajectories as prompts to guide autonomous perception and adaptive decision-making. In addition, we propose a context-aware training and execution mechanism utilizing self-supervised contrastive learning to enhance environment recognition and prompt utilization efficiency. In addition, a physics-informed prompt design technique that filters prompts based on cumulative reward and frequency volatility is proposed, ensuring high-quality physical guidance during online execution. Finally, to ensure generalization in unseen environments with limited data, we develop a lightweight finetuning approach that achieves performance comparable to full-parameter finetuning with minimal adjustments.
[AI-123] Reinforcement Learning on Benign Facts Amplifies Leakage of Memorized Private Data
链接: https://arxiv.org/abs/2608.21727
作者: Renfei Zhang,Niloofar Mireshghallah
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Reinforcement learning with verifiable rewards (RLVR) is deployed to make models better at reasoning tasks, but its side effect on what models will divulge is under studied. Here we show that RLVR on facts increases extraction of personally identifiable information (PII) the instruct model had already memorized. We first confirm that instruct models have already memorized PII but leave them latent, rarely surfacing one when asked. We then apply RL on benign factual data that contains no PII of any kind, and re-probe: a targeted probe over name-email pairs, and an untargeted free-recall prompt that simply asks the model to list the addresses it knows. PII extraction rises sharply under both: on DeepSeek-V3.1, verbatim recall@k increases from 0.155 to 0.370, a 2.4x gain. The effect scales with model size: across three models spanning 8B to 671B parameters, absolute leakage is largest in the biggest model. Meanwhile model’s reasoning abilities and refusal rates are retained, indicating that RL selectively changes which memorized information is accessible rather than broadly altering the model. In summary, memorized private data can be made markedly more extractable by training that never touches it. This gives an adversary a route to memorized data that requires no privacy-relevant training signal and no access to the data itself – only the ability to fine-tune on something innocuous.
[AI-124] Ask or Answer: A Decision Framework for Multi-Turn Health Misinformation Intervention EMNLP2026
链接: https://arxiv.org/abs/2608.21721
作者: Xiaoying Song,Anirban Saha Anik,Jinyu Liu,Qitao Tan,Geng Yuan,Lingzi Hong
类目: Artificial Intelligence (cs.AI)
备注: Accepted at EMNLP 2026
Abstract:Correcting health misinformation in dialogue requires more than producing a factual rebuttal: users differ in what they know, what they believe, and what they need to hear, so an effective intervention often depends on first asking the right clarifying question. Yet existing methods either respond immediately or probe indiscriminately, treating clarification as either unnecessary or always beneficial. We propose Reward-Optimized Probe-and-Respond (RO-PnR), a framework that learns when asking is worth its cost. At each turn, RO-PnR chooses between probing for more information and committing to a final correction, guided by a turn-level reward that weighs the expected gain from probing against its interaction cost. To capture how user heterogeneity affects probing value, we model each simulated user with a latent state along health literacy and belief commitment. Experiments show that RO-PnR achieves the highest cost-adjusted utility across three health-misinformation datasets and three base models, using 30% fewer turns than always-probe baselines.
[AI-125] Context as an Environment: Programmatic Context Management for Long-Horizon Agents
链接: https://arxiv.org/abs/2608.21690
作者: Yin Lin,Elaine Ang,Erkang Zhu,Bolin Ding,Jingren Zhou
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:LLM agents increasingly take on long-running tasks whose history grows far beyond a single model context window. Existing approaches compress earlier interactions or extract selected information into fixed memory representations, committing to what to preserve before future needs are known. We present Scroll, a context manager that treats each agent session as an executable Session Environment. The environment is backed by an append-only Event Log and a sandboxed, persistent Python kernel. The kernel maintains a typed namespace across model calls, allowing tool outputs, retrieved history, and derived state to be bound to variables rather than serialized into the prompt at each call. Model-written code searches, materializes, and transforms session state through exec; only explicitly printed projections enter the model’s working view for the next call. Context management thus becomes a programming task that inherits the improving coding abilities of LLMs, while the Event Log preserves lossless historical ground truth. As the working view approaches its budget, stale spans are evicted but remain recoverable: an eviction index keeps compact landmarks tied to exact Event Log addresses, so that the agent navigates directly to evicted regions instead of searching the full log. With Qwen3.8-Max as the backbone, Scroll achieves 94.8% on LongMemEval_S; 73.1% on BEAM_10M, surpassing the best published memory system by 5.1 points; and 86.7% on LOCA_256K, exceeding the best published long-horizon agent by 37.4 points.
[AI-126] Read Write Relax: Why Neural PDE Surrogates Need Both Global and Local Processing
链接: https://arxiv.org/abs/2608.21677
作者: Anuj Kumar,Heiko Zimmermann,Josiah Bjorgaard,Jacan Chaplais,Nikolaos Bouklas,Matteo Salvador,Alexander Lavin
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE); Applied Physics (physics.app-ph); Computational Physics (physics.comp-ph)
备注: 26 pages, 14 figures
Abstract:Recent mesh-based simulation advances have, in no small part, relied on neural surrogates of two distinct families: global models that route information through a small set of latent tokens, and local models that perform message passing across mesh edges. Consistent with both classes is the inability to perform beyond low-dimensional problems and small-scale or oversimplified meshes, the simulation regimes where industrial problems reside. Our work shows this explicitly and presents a unified formulation. In global approaches, latent-token attention acts as a spatial low-pass filter, while local message passing lacks the global reach necessary to propagate information across large mesh spaces. Viewed through the error, the two operators are the halves of a multigrid cycle: one corrects errors at the lower end of the spectrum, the other at the higher end, and neither can do the other’s job. We introduce Read-Write-Relax (RWR), which interleaves latent attention with message-passing relaxation under a unified formulation. The interleaved processor lowers error across the entire spectrum, making RWR the most accurate model in nearly every comparison across our industrial and public benchmarks. It is also markedly data-efficient in the scarce-data regimes, accurate on the engineering quantities of interest, and scales full-field predictions to challenging, large-scale problems.
[AI-127] SynEHR: Joint Modeling Inter-visit Temporal Evolution and Intra-visit Clinical Structure for Longitudinal EHR Synthesis CIKM’26
链接: https://arxiv.org/abs/2608.21673
作者: Ximiao Li,Lin Jiang,Rongchao Xu,Dahai Yu,Zhe He,Guang Wang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 12 pages. CIKM '26
Abstract:Longitudinal electronic health records (EHRs) document patients’ sequences of clinical visits over time, preserving the temporal evolution of disease progression and care delivery. However, real longitudinal EHRs are difficult to access because they contain large amounts of fine-grained, patient-specific information. Synthetic EHR generation therefore provides a valuable approach for preserving the statistical patterns and clinical structure of patient visit trajectories, enabling broader modeling and analysis when real records are limited. Although recent generative models have made progress in producing future visit sequences, they remain limited in explicitly integrating inter-visit irregular temporal evolution and intra-visit clinical event structures in EHRs, leading to clinically inconsistent and temporally unrealistic visit sequences. In this work, we propose SynEHR, a lightweight adaptive LLM-based framework for longitudinal EHR synthesis. There are two novel designs in SynEHR, i.e., a Temporal State Conditioning Module captures irregular temporal states across visits and a Temporal-Relational Adaptation Module combines these states with patient history to dynamically construct patient-specific relational representations. SynEHR then builds on a parameter-efficient LoRA-adapted language-model generator with next-visit generation capability to train the two modules for temporally and clinically informed generation. Extensive experiments on real-world EHR datasets across fidelity, privacy, and downstream utility evaluations demonstrate that SynEHR outperforms state-of-the-art models by generating more clinically coherent and temporally faithful longitudinal EHR data.
[AI-128] Power-Performance Characterization of TinyML Systems
链接: https://arxiv.org/abs/2608.21646
作者: Yujie Zhang,Dhananjaya Wijerathne,Zhaoying Li,Tulika Mitra
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 7 pages, 9 figures. Published in Proceedings of the IEEE International Conference on Computer Design (ICCD), 2022
Abstract:TinyML systems are enabling machine learning (ML) inference at the edge. However, there is little quantitative analysis of such systems. This paper presents a systematic performance and power characterization of diverse TinyML applications on microcontrollers (MCUs), spanning neural network models, software libraries, operating systems, and hardware architectures. We focus on the impact of the multiple layers of abstraction that provide higher programmability at the expense of performance and energy efficiency. We propose a model to estimate the costs of different abstraction layers and make recommendations for minimizing those costs. Our findings can help designers with Neural Architecture Search (NAS) and CNN inference optimization on edge devices.
[AI-129] KAN-Robust-Bench: A Benchmark for Evaluating the Robustness of Kolmogorov-Arnold Networks
链接: https://arxiv.org/abs/2608.21488
作者: Mohammad Meymani,Roozbeh Razavi-Far
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:While machine learning models have demonstrated strong performance in many domains, these models have shown profound vulnerabilities when they are exposed to adversarial threats. While adversarial attacks fall into various categories, the most prominent category in research studies is evasion. In evasion attacks, the adversary generates perturbed versions of samples, which might not be observable by human eyes. These samples generally fool the machine learning models with high confidence. This phenomenon poses a significant security violation against machine learning models. In this paper, we investigate the certified and empirical robustness of various Kolmogorov-Arnold network architectures against strong evasion attacks. At first, we provide the mathematical foundations for randomized smoothing and interval bound propagation, and report the \ell_2 -certified robustness of the models under randomized smoothing. After that, we systematically evaluate the robustness of various defended and undefended KAN models under FGSM, PGD, and CW attacks in order to find out the optimal defense strategies and architectures.
[AI-130] Enhanced Artificial Neural Networks Using QHAdamW in Air Quality Forecasting
链接: https://arxiv.org/abs/2608.21463
作者: Mary Joy Daniel Vinas
类目: Artificial Intelligence (cs.AI); Emerging Technologies (cs.ET); Machine Learning (cs.LG); Neural and Evolutionary Computing (cs.NE)
备注: 20 pages, 16 tables, 15 figures, Published with International Journal of Engineering Trends and Technology (IJETT)
Abstract:The study employed an Artificial Neural Network in combination with the optimized Adaptive Moment Estimation (Adam) algorithm, currently the only AQI forecasting model available in the Philippines. The modified QHAdamW - Quasi-Hyperbolic Momentum (QHAdam) and Adam with decoupled weight decay (AdamW) were both extensions of the Adam optimizer, and both offer unique advantages for training ANN. The proposed QHAdamW optimizer addresses the issues on convergence, generalization, and forecasting performance of Adam. Hyperparameter tuning results revealed that 0.01 and 0.001 were the most effective optimal values for the generalization performance of QHAdamW. The comparative analysis results using seven evaluation metrics revealed that the error value range is lower, and the regression coefficient, having a value approximately equal to 1, improved the model accuracy performance. Likewise, the model converges to a satisfactory level of performance with the convergence performance results of lower loss values as obtained from training and validation losses. Based on data from a real-time air quality tracking station in Manila, a feed-forward neural network is used to predict the AQI of PM2.5 and PM10 separately. This model can be used to forecast Particulate Matter (PM), to help the Department of Environment and Natural Resources-Environmental Monitoring Bureau (DENR-EMB) implement a comprehensive air quality management.
[AI-131] Software Frameworks for Explainable AI in Time Series Classification: A Systematic Review KDD2026 KDD ECML
链接: https://arxiv.org/abs/2608.21449
作者: Louis Peter,Nils Gumpfer,Jana Fischer,Christin Seifert,Jennifer Hannig
类目: Artificial Intelligence (cs.AI)
备注: 17 pages, 5 figures, to be published in ECML PKDD 2026, XKDD and Beyond Workshop, LNCS
Abstract:Time series arise in a wide range of application domains and are analyzed using machine learning in decision-critical settings. Time series classification (TSC) is one of the most widely studied and relevant tasks. In this context, ensuring the transparency and trustworthiness of TSC models has become an important requirement, motivating the use of explainable artificial intelligence (XAI) methods. Despite growing interest, research on XAI for TSC remains fragmented, and a systematic understanding of the available software frameworks for explanation generation, their evaluation practices, and practical limitations is still lacking. Prior work largely focused on individual explanation methods, while cross-framework consistency, time-series-specific evaluation, and reproducibility have received little attention. In this survey, we analyze existing software frameworks for explanation generation and evaluation in TSC. We compare them along multiple dimensions, including supported XAI methods, evaluation metrics, usability, benchmarking support, and reproducibility, providing the first time-series-specific survey of frameworks with implementation comparisons and an analysis of frequency-domain support. We identify six frameworks that explicitly support time series and reveal common limitations: only one method supports frequency-domain explanations despite their relevance; only two evaluation metrics have been developed specifically for time series; and identical XAI methods can yield substantially different explanations across frameworks. Based on these findings, we discuss open challenges and outline directions for future research, highlighting the need for unified, time-series-specific XAI frameworks that enable faithful, reproducible, and time-series-aware explanations.
[AI-132] Constructing Predictive Surgical Path for AI-based Capsulorhexis Skill Transfer
链接: https://arxiv.org/abs/2608.21441
作者: Mohammad Javad Ahmadi,Hamid D. Taghirad
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Automated training of surgeons is one of the most crucial factors that significantly minimize surgical training risks and expenses. With recent advances in artificial intelligence (AI) knowledge and available data from various surgeries, AI’s involvement in surgical training is becoming very promising. It is recommended that at the early stages of AI development, it interferes in the surgery as a third agent alongside the trainer. As trust in AI increases, this process will lead to an AI agent acting as a trainer in the future. The first phase in which AI can intervene in the training process is to suggest an improved surgical path to the trainer. A platform must be constructed in the first step, to accomplish this task and to enhance the movement path of trainee surgeons. This paper introduces this platform along with an annotated capsulorhexis surgery dataset called the ARAS-Farabi dataset. In this research, a deep convolutional neural network is pre-trained with JIGSAWS and ARAS-Farabi surgical datasets that can extract surgical skill characteristics from surgery tool tip motion data. The proposed platform develops a reference model from the feature space of an expert surgeon’s movement trajectory and proposes an improved path to enhance the skill of a novice surgeon. An optimization with two loss functions is utilized to create a path that raises the skill level of the novice surgeon’s path while simultaneously predicting and preserving his/her intent. The results of this study reveal that, with the assistance of an AI agent, the trainee surgeon’s movement path can be enhanced by at least 20 percent while maintaining his intentional objective. In addition to the recommended deep network, various tangible indicators have also been developed in this research to verify the level of trainee improvement.
[AI-133] Geo-VLA: Geometry-Aware Vision-Language-Action Planning via Internalization of Map Semantics
链接: https://arxiv.org/abs/2608.21440
作者: Ran Chen,Jiaxing Ren,Zhikun Zhang,Yunhao Hou,Junbao Zhuo,Bochao Zou
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Vision-language-action (VLA) models have advanced end-to-end autonomous driving by leveraging foundation models for semantic reasoning and long-tail generalization. However, their planning performance remains limited in complex driving environments because image-only representations inadequately capture planning-relevant road geometry and topology. In this paper, we propose Geo-VLA, a plug-and-play framework that enhances VLA models by learning geometry-aware visual representations. During training, Geo-VLA internalizes geometric map semantics to strengthen road-structure representations, while requiring no HD maps or additional lane information during inference. To support this approach, we introduce Geo-QA, a geometry-focused question-answering dataset that injects road geometry into vision-language representations through contrastive learning and instruction tuning. Experiments on NAVSIM v1 demonstrate that Geo-VLA consistently improves VLA planners with distinct action-generation architectures, achieving 92.1 PDMS and establishing a new state-of-the-art among single-camera VLA planners.
[AI-134] Composable Trust Infrastructure for Manufacturing Knowledge Graphs: Cross-System Provenance Temporal Reasoning and Decision Traceability
链接: https://arxiv.org/abs/2608.21418
作者: Grama Chethan
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注:
Abstract:Manufacturing knowledge graphs that integrate data from heterogeneous industrial systems face a trust deficit: consumers cannot determine whether queried data is valid, whether it was valid when a decision was made, where it originated, or how it was acted upon. We argue that four trust capabilities – SHACL validation, PROV-O provenance, domain-aware bi-temporal versioning, and graph-native decision objects – compose through shared correlation identifiers to produce emergent trust properties that no single capability delivers alone. We present a composable trust infrastructure that integrates these four capabilities into a unified RDF architecture. Capabilities compose through shared entity URIs, ingestion activity identifiers, and temporal correlation keys, enabling compound queries spanning all four dimensions. An experimental ablation confirms that removing any single capability causes exactly three of six composition queries to fail, demonstrating that all four are equally load-bearing. Analysis of higher-order compositions reveals four emergent three-way properties and one irreducible four-way property (full-chain auditability, 31ms execution). The infrastructure is validated on a testbed integrating eleven industrial sources – OPC UA, TIA Portal, eClass, AAS, ISA-95, ISA-18.2, SAP S/4HANA, Teamcenter, Opcenter EX, Insights Hub, and SCM – under an 89-class ISA-95-aligned ontology. The unified graph contains 8,743 triples across five named graphs, stitched by 81 owl:sameAs identity edges. Evaluation uses simulated but structurally realistic data from purpose-built emulators; data structures and cross-system linkage patterns are representative of real industrial installations.
[AI-135] Retrieval-grounded robot program generation and simulation-based correction via Model Context Protocol
链接: https://arxiv.org/abs/2608.21417
作者: Zhichao Zhou,Siyuan Chen,Omkar Salunkhe,Ebru Turanoglu Bekar,Johan Stahre,Anders Skoogh
类目: Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注: Accepted by CIE53; to appear in the CIE53 Proceedings, Khalifa University, Abu Dhabi, UAE, October 20-23, 2026
Abstract:Flexible manufacturing requires industrial robots to be reprogrammed rapidly as product variants change. This paper presents a language-model-based workflow that generates, validates, and iteratively corrects ABB RAPID robot programs from natural language task descriptions. A dual-stream retrieval-augmented generation (RAG) pipeline grounds code generation in verified technical documentation and production templates, reducing domain-specific errors produced by ungrounded language models. A custom Model Context Protocol (MCP) server connects the language-model client directly to ABB RobotStudio for automated code upload, simulation execution, and diagnostic feedback. The evaluation combines a 30-query retrieval benchmark, scoped code-generation checks, and RobotStudio case studies in a simulated pickand- place manufacturing cell. The simulation loop exposes execution failures that static and semantic checks alone cannot catch, including suction release-height errors, unreachable placement targets, and configuration-dependent recovery motions. The results show how RAG and MCP can connect grounded code generation with executable feedback from industrial robot simulation software, while reducing but not eliminating expert setup and final supervision.
[AI-136] Operational digital twin clinics enable task-based evaluation of embodied AI
链接: https://arxiv.org/abs/2608.21416
作者: Xinyuan Wu,Jingrao Zhang,Mengdi Xu,Henry K. Chu,Mingguang He,Danli Shi
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Embodied artificial intelligence (AI) must be tested in the clinical environments where it will operate, but building realistic, robot-testable settings is costly and difficult to scale. Here we show that routine clinic images can be transformed into operational digital twins for task-based evaluation of embodied AI. Using 39 ophthalmic clinic scenes, we converted single photographs into editable, simulator-ready environments and assessed reconstruction quality, room-scale geometry, mesh grounding, multi-robot feasibility, perturbation sensitivity and closed-loop policy performance. The reconstructed scenes preserved workspace structure, while local editing enabled controlled device reconfiguration. Device meshes, collision proxies and semantic anchors converted visual reconstructions into contact-aware simulation scenes. Across three robot embodiments, shared task targets showed different patterns of reachability and contact feasibility. Small device translations and rotations produced task-specific changes in contact margins that were not captured by visual similarity alone. Digital-twin trajectories also supported local policy learning and closed-loop evaluation. These findings establish operational validity as a key principle for clinical digital twins and provide an intermediate layer between offline development and physical deployment of embodied AI in healthcare.
[AI-137] he Abstention Protocol: RCA for Clos Fabrics OSDI2026
链接: https://arxiv.org/abs/2608.21412
作者: Madhava Gaikwad,Deepak Pandey
类目: Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC); Networking and Internet Architecture (cs.NI)
备注: Presented at Usenix OSDI 2026. 17 pages, 5 figures, 4 tables
Abstract:Root cause analysis (RCA) in large datacenter networks is challenging because telemetry is noisy, partial, and asynchronous. Score-based approaches degrade under these conditions, often yielding unstable or incorrect attributions. We present \textscCoreSec, a production RCA system that replaces weighted fusion with a PAM-style abstention algebra. Telemetry agents are composed using control flags that yield deterministic decisions and explicit abstention when evidence is ambiguous. CoreSec combines this algebra with topology-aware configurations that capture failure surfaces across Clos fabrics and converge monotonically as evidence accumulates. Deployed at hyperscale, CoreSec provides stable and explainable RCA behavior across diverse environments without retuning. Our experience shows that structured composition with abstention forms a practical foundation for automated RCA in real-world cloud networks. Comments: Presented at Usenix OSDI 2026. 17 pages, 5 figures, 4 tables Subjects: Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC); Networking and Internet Architecture (cs.NI) Cite as: arXiv:2608.21412 [cs.AI] (or arXiv:2608.21412v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2608.21412 Focus to learn more arXiv-issued DOI via DataCite Journalreference: 20th USENIX Symposium on Operating Systems Design and Implementation (OSDI 26), Seattle, WA, July 2026, pp. 405-421
[AI-138] AI Surrogate Modeling for Real-Time Tokamak Equilibrium Prediction: Benchmarking Neural Architectures and Validation on EXL-50U
链接: https://arxiv.org/abs/2608.23217
作者: Guoyang Shi,Zitong Zhang,Siqi Ding,Jianguo Chen,Yapeng Zhang,Jiayi Zhi,Hanyue Zhao,Tianyuan Liu
类目: Plasma Physics (physics.plasm-ph); Artificial Intelligence (cs.AI)
备注:
Abstract:Fast and reliable plasma equilibrium prediction is essential for real-time tokamak operation and control, but conventional Grad-Shafranov (GS) solvers are often too costly for real-time deployment. We develop an AI surrogate framework and benchmark five architectures (MLP, CNN, FNO, Transformer, and KAN) on a numerical GS database with 100,000 IID and 10,000 OOD samples. Under a unified protocol, we evaluate accuracy, inference efficiency, model scaling, and robustness. We also establish device-level validation on the EXL-50U tokamak by linking numerical GS solutions, surrogate predictions, and the standard Shape Editor reference to assess simulation-to-device consistency. The surrogates achieve errors of 10^-3 - 10^-2 relative to GS solutions, while the GS-to-device discrepancy remains at 10^-3 . Transformer gives the best IID accuracy, whereas CNN offers the best balance of accuracy, robustness, and speed, reaching 0.7 ms TensorRT latency. On unseen plasma geometries and parameter regimes, CNN and FNO show the strongest extrapolation stability, with 4%-5% relative L_2 error, while models with weaker inductive biases degrade more substantially. Scaling data and model capacity improves interpolation but not necessarily extrapolation, revealing a trade-off between capacity and OOD generalization. Overall, this work provides a systematic, device-consistent benchmark for AI-based GS prediction and practical guidance for selecting reliable surrogates for real-time plasma control and fusion applications.
[AI-139] Spending Scarce Confirmatory PET Measurements: Target-Aligned Validation in A4/LEARN
链接: https://arxiv.org/abs/2608.22223
作者: Eliuvish Han Cui
类目: Applications (stat.AP); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Machine Learning (stat.ML)
备注: 11 pages, 3 figures, 1 table; supplementary material and code/results are included as ancillary files
Abstract:Anti-amyloid therapies and blood-based biomarkers are changing Alzheimer disease workups into a two-stage measurement workflow: screen broadly with cheaper information, then spend scarce confirmatory amyloid measurements where they support the decision that will be reported. Amyloid positron-emission tomography (PET) remains one such protocol measurement for amyloid burden, but PET slots, trial budgets, and payer-facing evidence packages are finite. This paper asks a deliberately operational question: when is simple transparent PET validation enough, and when is a fitted residual-uncertainty score worth the added complexity? For a weighted protocol target, the first-order value of validating subject i is the product of target influence and residual protocol uncertainty. Generic uncertainty sampling uses only the second factor and can spend PET measurements on subjects that are hard to predict but weak for the scientific, clinical, or commercial claim. We apply this rule to the A4/LEARN PET archive, treating observed PET as a design laboratory for scarce-confirmation studies. For the primary APOE4 carrier versus non-carrier contrast in Centiloid 24-or-higher PET positivity, simple APOE4-balanced validation recovers nearly all of the target-specific gain: at PET budget 200, the confidence-interval width ratio relative to random validation is 0.923 for APOE4 balancing and 0.914 for target-specific scoring, while generic uncertainty sampling is 0.980. Other targets behave differently: target-specific scoring gives larger gains for an age-slope analysis and for cutoff-indexed PET positivity. The practical message is simple: spend scarce protocol measurements according to the claim being validated, not only according to prediction uncertainty.
[AI-140] Discovering Dual-Origin Slow Wind from Solar Orbiter with Self-Supervised Contrastive Learning
链接: https://arxiv.org/abs/2608.22065
作者: Henry Han,Jorge Yero Salazar
类目: olar and Stellar Astrophysics (astro-ph.SR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 17 pages, 6 figures, 7 tables. Accepted at the Fifth Southwest Data Science Conference (SDSC 2026-Mid-Atlantic), Norfolk, VA, USA, May 18-19, 2026. To appear in New Frontiers in Artificial Intelligence and Data Science, Springer Communications in Computer and Information Science, vol. 3101
Abstract:Whether the slow solar wind originates from one coronal source or two distinct channels remains a central open question in heliophysics. Resolving this requires unsupervised separation of two populations that arrive at nearly the same bulk speed and differ mainly in heavy-ion composition. We present Solar-CDC, a self-supervised contrastive deep clustering (CDC) framework that maps plasma observables to a latent space via a Transformer encoder, optimizes a triplet margin loss, and updates pseudo-labels via k -means. Theoretically, we prove that neighborhood-preserving embeddings such as t-SNE and UMAP are fundamentally constrained. Preserving the neighbor graph leaves the cross-cluster cut fraction unchanged, and preserving all but a fraction \varepsilon of its links moves that fraction by at most \varepsilon . Neither bound depends on the target dimension. A margin objective rewrites the graph and drives the cut fraction to zero. Empirically, on 30,602 Solar Orbiter observations, thirty combinations of dimensionality reduction and clustering peak at a silhouette of 0.454 , whereas Solar-CDC reaches 0.869 . Escaping the geometric bound alone does not guarantee physical validity: TriMap also optimizes triplets and reaches 0.824 , yet its clusters score below chance against the published composition taxonomy. Solar-CDC instead recovers clusters with mean charge-state ratios of 0.080 , 0.160 , and 0.400 , placing the intermediate population inside the window associated with coronal-hole boundaries. Even when the defining charge-state ratio is withheld from the inputs entirely, the model still recovers the taxonomy defined on it. Solar-CDC thus connects self-supervised representation learning to coronal source diagnostics. Importantly, a learning loss recovers physical populations only when driven by dynamically updated physically-aware clusters rather than distances.
[AI-141] Barycentric Fused Gromov-Wasserstein Balancing for Causal Inference under Multiple Treatments
链接: https://arxiv.org/abs/2608.22024
作者: Yuki Murakami,Takumi Hattori,Kohsuke Kubota
类目: Methodology (stat.ME); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Machine Learning (stat.ML)
备注: 26 pages
Abstract:Estimating heterogeneous single and interaction treatment effects from observational data under multiple simultaneous treatments is crucial for decision-making. To mitigate estimation variance, previous studies balance representation distributions between every pair of treatment patterns. However, such pairwise balancing scales quadratically with the number of treatment patterns and fails to preserve consistent local proximity structures across patterns, which degrades counterfactual estimation. To address these challenges, we propose the Causal Inference for Heterogeneous Single and Interaction Treatment Effects Network (CIHSI-Net), a deep learning framework built on a novel Barycentric Fused Gromov-Wasserstein Balancing (BFG-WB) objective. BFG-WB aligns the representation distribution of each treatment pattern with a shared Wasserstein barycenter, achieving global alignment while reducing the computational complexity from quadratic to linear, and its Fused Gromov-Wasserstein discrepancy preserves the local proximity structures essential for reliable heterogeneous effect estimation. Simulation studies show that CIHSI-Net consistently outperforms state-of-the-art baselines, and an application to real-world marketing data demonstrates its practical utility in complex multi-treatment scenarios.
[AI-142] Scalable quantum simulation of continuous-time generative models via tensor networks
链接: https://arxiv.org/abs/2608.21700
作者: Nathan X. Kodama,L. Andrew Wray,Sam Cochran,Chad Rigetti,Shravan Veerapaneni,Michael J. Keiser
类目: Quantum Physics (quant-ph); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Continuous-time flow and diffusion models are widely used across many application domains, from large-scale deployment in computer vision and protein folding to emerging adoption for modeling language, time series, and quantum states. After training, inferring statistical properties from continuous-time models is costly. Wavefunction flows target this cost by recasting learned transport as unitary evolution, whose final Born distribution approximates the target distribution. This prepares a coherent amplitude encoding (a qsample) that can be post-processed by quantum algorithms offering a quadratic advantage over Monte Carlo sampling. We present the first numerical study of these flows, in which we represent time-dependent potentials and states as tensor networks. At spatial dimension d=8 , storage falls by \sim 10^7\times relative to the dense grid of N^d points, and evolution wall-clock time falls by \gtrsim 10^3\times against a baseline extrapolated from the measured d\le 5 scaling. We validate our pipeline by reproducing the O(1/\sqrtp_\rm rare) scaling of rare-event sampling.
[AI-143] PepLLM : ESM-Guided Llama for Structured Protein-Peptide Binding Interface Analysis
链接: https://arxiv.org/abs/2608.21367
作者: Hao Qian,Shikui Tu,Lei Xu
类目: Biomolecules (q-bio.BM); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Protein-peptide interactions are central to cellular regulation and peptide-based drug discovery, yet existing computational methods mainly focus on interaction classification, binding-site prediction, or peptide binder generation. These formulations provide limited insight into the physicochemical mechanisms that determine how a peptide binds to a protein. In this work, we introduce \textbfPepLLM, an instruction-tuned framework for structured protein-peptide interface understanding. Given protein-peptide sequences, PepLLM generates a machine-readable JSON annotation describing multiple interface properties, including peptide burial state, hydrogen-bond density, salt-bridge presence, hotspot residues, hydrophobicity, and electrostatic complementarity. To support this task, we construct a new protein-peptide interface dataset by integrating structural interface analysis, solvent-accessible surface area computation, hydrophobic burial estimation, electrostatic potential calculation, and redundancy-aware data splitting. PepLLM connects a pretrained ESM encoder with a LLaMA decoder through a nonlinear modality adapter. The adapted ESM residue embeddings are injected into the LLaMA prompt as continuous soft tokens via placeholder-token replacement, enabling the decoder to generate structured interface annotations under instruction tuning. By moving beyond single-label prediction toward multi-property and mechanism-aware generation, PepLLM establishes a new task and modeling paradigm for interpretable protein-peptide interface analysis.
[AI-144] Small Language Model enabled Autonomous agent for Language-Conditioned Cognitive Radar
链接: https://arxiv.org/abs/2608.11596
作者: Minhaj Uddin Ahmad,Zakia Zaman,Shunqiao Sun,Mizanur Rahman
类目: ignal Processing (eess.SP); Artificial Intelligence (cs.AI); Systems and Control (eess.SY)
备注: Accepted at MLSP 2026, ATL, USA
Abstract:Modern radar systems require adapting their processing strategies in response to changing interference, clutter, and data availability. This paper introduces a framework for a small language model (SLM)-driven autonomous agent designed for language-conditioned cognitive radar, functioning as an intelligent controller for a suite of array signal processing tools. Given a natural-language command, the agent extracts radar-operation-related cues, selects an appropriate sequence of signal-processing methods, configures parameters, and invokes executable tools for numerical computation. Experiments with a synthetic uniform linear array (ULA) radar demonstrate that, given a natural-language command, the agent performs meaningful algorithm selection across diverse scenarios for sidelobe control, jammer suppression, multiple-null beamforming, coherent-source handling, and low-snapshot direction-of-arrival (DOA) estimation. Ablation results show that radar-specific prompting and physics-grounded tool execution are both required for reliable decisions and hallucination-free numerical results.
[AI-145] Correcting Variable Importance Scored by Random Forests
链接: https://arxiv.org/abs/2606.10770
作者: Guancheng Zhou,Haiping Xu,Jason Liu,Donghui Yan
类目: Methodology (stat.ME); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 22 pages, 10 figures
Abstract:Variable importance produced by Random Forests (RF) is used widely in statistical data analysis, and has played an important role in a variety of tasks such as assisting model interpretation, model selection and diagnosis, and cost-bounded learning etc. However, the calculation of variable importance in RF does not take into account of the correlations among variables, and variables that are correlated to many other variables tend to receive a lower importance index or being completely masked (i.e., with an importance index near zero) by other strongly correlated variables. To prevent influence from unwanted correlated variables in calculating variable importance, we propose to group variables by their conditional correlations (conditional on the response variable). We explore two computationally efficient options, with one grouping variables individually, and then separates the variable of interest from all correlated variables, while the other uses clustering to group variables according to their pair-wise conditional correlations. Our experiments show that both lead to sensible corrections to the importance of variables.
机器学习
[LG-0] Provably adaptive sampling with uniform and remasking discrete diffusion models
链接: https://arxiv.org/abs/2608.23554
作者: Daniil Dmitriev,Zhihan Huang,Yuting Wei
类目: Machine Learning (cs.LG); Information Theory (cs.IT); Statistics Theory (math.ST); Machine Learning (stat.ML)
*备注: 37 pages, 4 figures
Abstract:Discrete diffusion models offer a promising alternative to autoregressive generation by enabling parallel updates, but their sampling efficiency can depend strongly on the choice of the forward process and the sampler. For the uniform forward process, existing lower bounds for the standard \tau -leaping sampler scale linearly with the ambient dimension d , raising the question of whether this dependence is intrinsic to the forward process. We answer this question in the negative. We consider a first-order sampler based on the leave-one-out denoiser for uniform and remasking processes whose coordinate updates can be performed in parallel. In both cases, the sampler can correct denoising mistakes during the sampling process, which becomes necessary when many coordinates are updated together. Our main result establishes an adaptive sampling guarantee: up to logarithmic factors, N = O(\mathrmDTC(X_0) / \varepsilon) discretization steps suffice to achieve sampling error O(\varepsilon_\mathrmscore+\varepsilon) , where \varepsilon_\mathrmscore is the error in score estimation. Thus, the sampling complexity is governed by the intrinsic dependence structure of the target distribution, as measured by its dual total correlation \mathrmDTC(X_0) , rather than directly by the ambient dimension d . Our analysis proceeds through a Bayes-optimal auxiliary sampler that separates discretization error from score-estimation error. We also derive an exact information-theoretic representation of the discretization error in terms of the mutual information between different coordinates of the forward process at different times. This representation applies to general forward processes and, in the uniform and remasking cases, can be controlled by \mathrmDTC(X_0) . Numerical experiments on structured synthetic distributions illustrate the predicted dimension-adaptive behavior.
[LG-1] Robustness of Anomaly Detection Models for Industrial Control Systems under Training-Time Data Contamination
链接: https://arxiv.org/abs/2608.23547
作者: Mustafa Umut Ozbek,Taiwo Ojo,Pooria Madani,Khalil El-Khatib,Li Yang
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注: Accepted and to appear in IEEE CASCON 2026. Code is available at: this https URL
Abstract:Machine-learning-based anomaly detection is increasingly used in industrial control systems (ICS), yet most studies assume that detector training data is trustworthy. In practice, training data may be corrupted through compromised logs, labeling errors, manipulated historian records, or unsafe retraining processes. This paper evaluates the robustness of offline ICS anomaly-detection pipelines on the Secure Water Treatment (SWaT) benchmark under training-time contamination. We assess 11 heterogeneous anomaly detectors under three contamination strategies: random injection, similarity-targeted injection, and feature-noise injection. The first two insert attack samples into the nominal training pool, while the third adds bounded Gaussian noise to selected normal training samples. These attacks are contamination-based rather than gradient-driven poisoning methods. Contamination budgets from 1% to 10% are evaluated using clean validation and test sets under a unified offline protocol. The results show that robustness is strongly model-dependent and cannot be predicted from clean-data performance alone. Injection-based contamination causes the greatest degradation, particularly for local-density and distance-based detectors, whereas feature-noise contamination has a comparatively limited effect. PCA, SVM, HBOS, and IForest remain relatively stable, while the tuned neural detectors demonstrate intermediate robustness. Overall, the findings highlight the importance of training-data integrity in ML-enabled ICS monitoring, subject to the evaluated dataset, models, and threat assumptions.
[LG-2] Inertial Manifold Neural Operator for Dissipative Time-Dependent Partial Differential Equations
链接: https://arxiv.org/abs/2608.23546
作者: Xiaoyang Xie,Clarence W. Rowley
类目: Numerical Analysis (math.NA); Machine Learning (cs.LG); Dynamical Systems (math.DS)
*备注: 5 pages, 11 figures, 6 tables
Abstract:In this paper, we introduce the Inertial Manifold Neural Operator (IMNO) for solving dissipative time-dependent partial differential equations (PDEs). The long-time dynamics of such systems often exhibit an effective low-dimensional structure due to dissipation. Unlike standard neural operator architectures such as the Fourier Neural Operator (FNO), IMNO explicitly leverages the low-dimensional structure to achieve better physical interpretability, accuracy, and stability in long-horizon autoregressive training and prediction for nonlinear dissipative PDEs. For shift-equivariant PDEs, we further introduce a shift-equivariant variant (IMNO-SE) of the proposed neural operator, ensuring that a spatial shift in the input induces the same spatial shift in the output. This symmetry-preserving inductive bias substantially improves its performance in shift-equivariant PDEs. Extensive benchmark experiments are presented to evaluate IMNO’s performance numerically.
[LG-3] RAD: Rule-Augmented Relational Anomaly Detection
链接: https://arxiv.org/abs/2608.23468
作者: Noah Dahle,Anne Tumlin,Ngoc Tran,Xenofon Koutsoukos,Tyler Derr
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR); Databases (cs.DB)
*备注:
Abstract:Anomaly detection is often applied to data stored in relational databases, yet most existing methods require flattening multiple tables into a single feature matrix. This flattening can obscure entity identity, schema structure, and multi-hop dependencies, limiting the detection of anomalies that depend on relational context rather than isolated feature values. Beyond preserving relational structure, relational anomaly detection raises an additional challenge: how to incorporate symbolic behavioral evidence into learned relational representations. To address these challenges, we study relational anomaly detection, where the goal is to identify anomalous entities or events in a multi-table database. We propose RAD, a rule-augmented relational anomaly detector that combines heterogeneous graph representation learning with refined symbolic rule signals. RAD derives candidate rules from random-forest paths over flattened summaries of the entities or events being scored, refines them into compact interpretable predicates, injects the resulting rule features into the graph model, and learns anomaly scores using reconstruction-based and pairwise-ranking supervision. To evaluate this setting, we introduce a relational anomaly detection benchmark spanning three settings: LANL cybersecurity event detection and two unexpected user-churn anomaly tasks derived from Amazon and HM relational databases. Experiments show that RAD improves anomaly ranking over flattened tabular detectors and relational baselines under natural class imbalance, achieving the best average rank on AUROC and AUPRC across the benchmark. Ablations show that direct rule injection and ranking-based supervision are key contributors to performance, while edge reconstruction is not uniformly beneficial. Our code and data are available at: this https URL.
[LG-4] ProxyFormer: A Dual-Stream Proxy Architecture for Ultra-Long Context and High-Resolution Generation
链接: https://arxiv.org/abs/2608.23463
作者: Zhongpan Tang
类目: Machine Learning (cs.LG)
*备注:
Abstract:The quadratic growth of attention computation and key-value (KV) cache with respect to sequence length is a central bottleneck for ultra-long-context language models and high-resolution generative models. We propose ProxyFormer, a general dual-stream architecture built upon proxy tokens. In each layer, fine-grained local features are compressed bottom-up into a small set of proxy states; expensive global interactions are performed only in the compressed proxy space; the globally contextualized proxies are then decompressed and injected top-down back into the local stream. Because the local stream persists across layers, fine-grained information that is not captured by one compression step remains accessible for later refinement, alleviating the irreversible information loss of conventional one-shot compression. We further introduce factorized multi-level compression/decompression, layer-wise dynamic compression ratios, asymmetric dual embeddings, and a proxy-only KV-cache inference scheme. On a 16GB GPU with batch size 1, a standard decoder-only model can train sequences of only about 20K tokens, whereas ProxyFormer with a compression ratio of 64 extends the trainable sequence length to about 0.7M. A model trained with a 64K window retains 92%-95% retrieval accuracy on a multi-needle retrieval task with 1,048,576 tokens, and a model trained with an 8K window exceeds 94% accuracy when extrapolated to 256K tokens. Preliminary image-generation experiments demonstrate the feasibility of ProxyFormer for both pixel-space and latent-space flow matching.
[LG-5] Diversity-Based Active Learning: An Evaluation of Metric Spaces for Active Learning Selection
链接: https://arxiv.org/abs/2608.23461
作者: Siddharth Chilamkur,Dorit S. Hochbaum
类目: Machine Learning (cs.LG)
*备注:
Abstract:With rapid advancement over the last few years, many different methods are now widely used for classification. However, training these models requires substantial labeled data. Active Learning is a potential solution to this problem. Pool-based active learning minimizes costs by querying only the most informative samples from an unlabeled dataset. Diversity-based approaches, on the other hand, attempt to select a representative subset of the data. There are many different objectives for determining the selection process, including exact K-center, exact K-median, and Greedy K-center. In this paper, we will focus on evaluating the performance of Greedy K-center across a variety of metric spaces: the raw feature space, a Linear Discriminant Analysis (LDA) space, and a model-derived probability space (with and without entropy-based weighting). Using Random Forest classifiers as a baseline evaluator, our empirical results on synthetic and real-world datasets demonstrate that mapping unlabeled instances into a predictive probability space and weighting the result by entropy often dominates the other options for active learning selection with Greedy K-center.
[LG-6] raceable Spectral Inference via Influence Functions: Efficient Data Attribution and Error Proxies for the Ariel Mission
链接: https://arxiv.org/abs/2608.23458
作者: Nikki Grens,Luís F. Simões,Kai Hou Yip,Theresa Lueftinger
类目: Machine Learning (cs.LG); Instrumentation and Methods for Astrophysics (astro-ph.IM); Machine Learning (stat.ML)
*备注: To appear in “Proceedings of SPAICE 2026: Third Conference on AI in and for Space”
Abstract:Interpretability is critical for machine learning models deployed in scientific space missions such as ESA’s Ariel, where ground truth is unavailable during operations and physical plausibility must be assessed. While most explainable AI methods focus on feature attribution, this work investigates training data attribution through influence functions and introduces three key contributions for operational spectroscopy pipelines. First, influence is reformulated in terms of prediction rather than loss, enabling label-free deployment. Second, by leveraging the closed-form ridge solution of an Extreme Learning Machine, infinitesimal prediction influence is efficiently computed. Third, an influence-based conservative error proxy is derived by propagating training residuals through the influence sensitivities. Evaluated against simulated spectra, the proposed proxy correlates strongly with scale and shape-based spectral errors. Furthermore, influence functions enable the identification of the most influential samples and the approximation of the most harmful ones. Together, these results suggest that this approach can serve as an operational framework for scientific machine learning.
[LG-7] he Axiomatic Trader: Latent Regularity Information Budgets and the Canonical Form of a Quantitative Investment System
链接: https://arxiv.org/abs/2608.23416
作者: Jiayu Li
类目: Machine Learning (cs.LG); Portfolio Management (q-fin.PM)
*备注:
Abstract:Systematic trading rests on one article of faith: that regularities found in the past persist. We state it as a time-invariant mechanism driven by an unobserved latent state, and show that it leaves a researcher five constants to declare — the recurrence bound Lambda at a block length b , the invariance defect epsilon_0 of the representation it is declared of, the coherence times ell_i of the state’s coordinates, the signal ceiling rho and the fraction kappa of it contingent on the regime — after which the architecture of a correct quantitative investment system is nearly forced.
[LG-8] Spectrum-Aware Bounds on Invertibility for Privacy-Enhancing Instance Encoding
链接: https://arxiv.org/abs/2608.23382
作者: Seokjin Hwang,Yuting(Ray)Li,Kiwan Maeng
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR)
*备注:
Abstract:Instance encoding is a popular empirical technique for privacy enhancement when sharing data to an untrusted server. It transforms sensitive data through an encoding process before sharing, with the hope that the encoding process retains utility but makes it hard to reconstruct the original data. However, most work offers no theoretical guarantee that the encoding process is actually irreversible. A recent work derived a mean-squared error (MSE) bound limiting any adversary’s reconstruction accuracy, offering one of the first theoretical results in this domain. This bound, however, has three critical limitations: it is often too loose, only works with randomized encoders (excluding many deterministic encoders practitioners use), and only bounds MSE. We introduce a family of new bounds that (1) are tighter, (2) applicable even to fully deterministic encoders, and (3) can extend beyond MSE to other norm-based similarity metrics, by properly accounting for the encoder’s spectral structure. We evaluate our bounds across a range of encoders, datasets, and attacks, showing they hold consistently and improve upon the existing bound.
[LG-9] st-Time Adaptation for ECG Classification via SQI-Gated Self-Training and Beat-Rhythm Consistency MICCAI2026
链接: https://arxiv.org/abs/2608.23347
作者: Wenhan Jiang,Zhipeng Deng,Jiale Zhou,Haolin Wang,Yafei Ou,Yefeng Zheng
类目: Machine Learning (cs.LG)
*备注: Accepted at MICCAI 2026
Abstract:Deep learning models for electrocardiogram (ECG) classification often suffer from significant performance degradation when deployed in unseen domains due to shifts in acquisition devices and patient populations. Test-time adaptation (TTA) offers a practical solution by adapting models using only unlabeled data at inference time. However, existing TTA methods often underperform on ECG tasks, since naive online updates ignore the hierarchical beat-rhythm structure of cardiac cycles and are vulnerable to signal artifacts, which leads to unstable adaptation and model drift. We propose BeatRhythm-TTA, an ECG-tailored TTA framework that explicitly accounts for ECG’s noisy observations and structured beat-rhythm semantics under domain shift. First, to handle pervasive ECG artifacts, we introduce a Signal Quality Index (SQI)-gated adaptation scheme that selectively filters out low-quality signals to prevent harmful updates. Second, to leverage ECG’s beat-rhythm semantics, we enforce dual-level consistency so the model preserves beat morphology and rhythm dynamics while adapting to shifted acquisition conditions. Extensive experiments on multi-label ECG diagnosis across three adaptation protocols, using PTB-XL as the source domain and CPSC2018/Georgia as two target domains, demonstrate the effectiveness of our method, yielding an average +2.70% relative improvement in Macro-F1 over the best competing method.
[LG-10] owards Actionable Surgical Team Dynamics: from Teamwork to Counterfactual Annotations
链接: https://arxiv.org/abs/2608.23344
作者: Vincenzo Marco De Luca,Antonio Longa,Andrea Passerini
类目: Machine Learning (cs.LG)
*备注: Accepted at HCOMP 2026
Abstract:Modeling team interactions in high-stakes environments such as operating rooms is critical for understanding how coordination, communication, and individual behaviors shape team performance and safety outcomes. Existing datasets in this domain are often fragmented across modalities, annotation schemes, and formats, limiting their ability to support integrated analyses of real-world collaborative processes. We address this limitation by introducing an extended multimodal dataset for surgical team interaction analysis, built from real operating room recordings. Starting from an existing corpus, we construct an analysis-ready version of the data by providing speaker diarization, transcripts, and multi-level annotations capturing team performance, interaction processes, and individual characteristics. Team performance is assessed using a standardized surgical teamwork evaluation protocol, while interaction quality and individual attributes are annotated through structured rating schemes covering collaboration, group dynamics, and non-technical skills. To further support the study of coordination breakdowns and performance variability, we introduce counterfactual annotations that describe plausible alternative team outcomes in the presence of observed interaction failures, enabling analysis of how specific behavioral patterns may relate to different trajectories of team performance. In addition, we provide structured temporal and relational representations designed to support computational modeling of teamwork processes and the design of AI-assisted collaborative systems. The dataset is designed to support the study of how individual actions, interaction patterns, and team-level processes jointly contribute to team outcomes in surgical settings, providing a unified resource for analyzing collaborative behavior in high-stakes domains.
[LG-11] Beyond Point Predictions: Uncertainty-Aware Satellite Poverty Mapping for Public Policy
链接: https://arxiv.org/abs/2608.23322
作者: Markus B. Pettersson,James Bailie,Mohammad Kakooei,Eagon Meng,Adel Daoud
类目: Machine Learning (cs.LG)
*备注:
Abstract:Despite their critical importance for policy and research, high-resolution poverty data remain limited across much of Africa. Machine learning (ML) with earth observation (EO) imagery has recently emerged as a way to supplement these data by predicting (i.e., estimating) poverty where it has not been directly measured. Yet to be used reliably, decision-makers and analysts need assurances that they will not be misled by the errors in these predictions. To meet this need, we develop an uncertainty-aware EO-ML method for poverty mapping based on simultaneous quantile regression and a novel form of conformal prediction. Using a spatiotemporal transformer trained on sequences of Landsat and nighttime-light images, we produce prediction intervals for neighborhood-level International Wealth Index estimates across Africa which are statistically guaranteed to achieve their desired coverage rates. While our method’s point-prediction performance matches the state of the art, its prediction intervals are wider than might be expected given its high R^2 of 0.75 . However, other models of similar accuracy likely suffer from comparable uncertainty, pointing to an inherent limitation: even with its remarkably high explanatory power, EO-ML cannot naively be relied upon for policy-making, such as when designing poverty-targeting programs. To handle this challenge, we develop a procedure to efficiently allocate aid using both ground-truth surveys and model predictions while provably ensuring the risk of excluding eligible neighborhoods remains below a prespecified level. In simulations, this approach delivers substantially more aid per eligible recipient than other strategies, thereby demonstrating that EO-ML can indeed be a reliable supplement to traditional data sources—as long as methods
[LG-12] Spicing up Genetic Netlist Generation with LLM s
链接: https://arxiv.org/abs/2608.23317
作者: Stefan Uhlich,Yağız Gençer,Andrea Bonetti,Arun Venkitaraman,Chia-Yu Hsieh,Eisaku Ohbuchi,Lorenzo Servadei
类目: Neural and Evolutionary Computing (cs.NE); Hardware Architecture (cs.AR); Machine Learning (cs.LG)
*备注: Accepted for MLCAD 2026, Extended Version
Abstract:Analog circuit topology synthesis remains challenging because useful designs occupy a tiny fraction of a combinatorial search space, and small structural changes can induce highly nonlinear changes in behavior. Evolutionary algorithms are attractive because they can optimize over discrete circuit topologies using only black-box evaluations, but they often require many SPICE simulations and may converge prematurely. We introduce LLM-SPICEMixer, a hybrid synthesis framework that augments genetic netlist generation with IGEL (Inspiration-Guided Evolution with LLMs), an LLM-based proposal operator. During search, IGEL prompts an LLM with high-performing circuits from the elite set and instructs it to generate a new SPICE netlist, which is then evaluated by SPICE and selected using the same reward mechanism as conventional genetic operators. Thus, the LLM contributes structured topology proposals while simulation remains the source of truth. We evaluate LLM-SPICEMixer on a challenging benchmark task: synthesizing transistor-level circuits that implement a discriminant function for Iris classification. Compared with the genetic framework without LLM guidance, LLM-SPICEMixer improves the median final training reward by 8.4% and the median validation-selected test reward by 8.8%. The best validation-selected circuit achieves 93.3% test accuracy at the nominal tt corner and 85.9% average test accuracy across 17 process, voltage, and temperature corners.
[LG-13] ADDA: a Modular Framework for Representing Simulating and Assimilating Dynamics with End-to-end Differentiability
链接: https://arxiv.org/abs/2608.23297
作者: Anthony Frion,Vien Minh Nguyen-Thanh,Ali Can Bekar,Pauleo R. Nimtz,Vadim Zinchenko,David S. Greenberg
类目: Mathematical Software (cs.MS); Machine Learning (cs.LG)
*备注:
Abstract:Data assimilation (DA) is an essential tool for prediction and understanding in the geosciences. DA combines simulation programs representing scientific knowledge with observations that constrain system dynamics, resulting in analyses and forecasts that incorporate both knowledge and data. DA tasks can be addressed with a diverse toolset, including variational, ensemble and learning-based methods. In particular, many recent works have proposed using automatic differentiation tools for variational, learning-based or hybrid methods. However, comprehensive comparisons across algorithms and dynamical systems remain challenging, due to the incompatibility of simulation and assimilation codes, inflexible handling of spatial and temporal discretizations, specialization of DA methods to specific simulations, and limited support for automatic differentiation and parallel computation in simulations. To address this challenge, we introduce Automatic Differentiation for Data Assimilation (ADDA), a software framework for defining and working with system states, simulations, observation schemes and DA methods. ADDA provides a powerful and flexible set of base classes for representing dynamical systems and observation operators, with support for collocated and staggered grids, unstructured meshes, Lagrangian state variables and irregular or continuous-time observations. Parallel processing and differentiability are first-class features, with support for batch axes and automatic differentiation throughout. ADDA is implemented in PyTorch library, but supports DA for JAX-based computation of dynamics and their gradients. To demonstrate its features, we further provide differentiable, ADDA-compatible implementations of 10 dynamical systems of various dimensionalities and scales, from which we design multiple illustrative DA examples. All of our code is publicly available at this https URL.
[LG-14] Poisson Subspace Clustering: Focusing on the Essentials in Count Data
链接: https://arxiv.org/abs/2608.23287
作者: Collin Leiber,Kai Puolamäki,Heikki Mannila
类目: Machine Learning (cs.LG)
*备注:
Abstract:Count data represented as a matrix of non-negative integer values, such as contingency tables, are prevalent across diverse domains. When clustering such data sets, specific methods are required, as generic algorithms often fail to consider their unique distributional properties, leading to unreliable outputs. An effective strategy is to use well-established statistical models such as the Poisson and negative binomial distributions. We present 3CPO, a clustering algorithm based on statistically solid modeling of count data. In addition to the cluster labels, it identifies a subset of relevant columns, enhancing the interpretability of the results. We propose a simple iterative algorithm that maximizes the posterior probability to find good clustering solutions and discuss its properties. Extensive experiments demonstrate its ability to define high-quality clusters within associated subspaces for various data domains, ranging from gene expressions and texts to economics. Our findings suggest that 3CPO is a robust solution for clustering count data in a statistically sound and interpretable manner. Our code is available at this https URL.
[LG-15] A Multidimensional Data-Driven Hybrid Transformer Framework for Non-invasive Continuous Blood Pressure Prediction
链接: https://arxiv.org/abs/2608.23276
作者: Yuexin Ma,Jingqi Hou,Yuxuan Kang,Zhaoying Liu
类目: Machine Learning (cs.LG)
*备注: 21 pages, 5 figures, 6 tables
Abstract:Objective. To develop and evaluate a cuffless continuous blood pressure (BP) estimator using temporal physiological and demographic features. We propose a hybrid Transformer framework to estimate diastolic and systolic BP from ECG/PPG-derived feature sequences. Approach. Rather than raw waveforms, the framework models 10-step sequences of six physiological descriptors and two demographic covariates. A Multi-Source Temporal Encoder Module combines Transformer, Kolmogorov-Arnold Network, and XGBoost branches to capture complementary temporal, nonlinear, and tabular information. A Dynamic Conditional Fusion-Decoder applies differential multi-head attention, token-weighted aggregation, and gated residual correction. A robust composite objective jointly optimizes DBP and SBP. Main results. Using the MIMIC-III Waveform and Clinical Databases, the source pool comprised 28,486 waveform segments from 203 subjects, and feature generation retained 53,621 observations from 166 subjects. On 2,431 segment-level held-out test windows, mean error +/- standard deviation was 0.41 +/- 3.74 mmHg for diastolic BP and -1.60 +/- 5.95 mmHg for systolic BP, with 95% limits of agreement of [-6.93, 7.74] and [-13.25, 10.06] mmHg, respectively. The proportions within 10 mmHg were 98.48% and 94.36%. The framework achieved the lowest standard deviations and narrowest limits of agreement among the locally retrained baselines. Significance. The feature-sequence fusion framework improved agreement with reference BP and fell within numerical AAMI and BHS Grade A thresholds on this split. This retrospective analysis is not formal device validation; subject-disjoint and external evaluation remain necessary before clinical use.
[LG-16] From Multimodal Observation to Interpretable Suggestions: Counterfactual Time-Expanded Relational Modeling of Surgical Teams
链接: https://arxiv.org/abs/2608.23254
作者: Vincenzo Marco De Luca,Antonio Longa,Giovanna Varni,Andrea Passerini
类目: Machine Learning (cs.LG)
*备注: Accepted at ACM Multimedia 2026
Abstract:In surgery, patient safety is threatened not only by technical issues but also by poor teamwork. However, existing surgical AI-based solutions focus mainly on visual workflow and technical execution, neglecting the modeling of team interactions and missing opportunities to actively support clinicians in improving their teamwork skills. To address this gap, we propose a tempo-relational framework for modeling surgical team dynamics from multimodal observations. By leveraging Time-Expanded graphs, the approach captures both relational structure and temporal evolution, achieving strong expressivity while remaining robust in the low-data regime typical of surgical settings. Beyond prediction, such modeling enables the generation of efficient, interpretable, and actionable suggestions for clinicians. More specifically, we generate suggestions via a counterfactual procedure that identifies minimal yet structured changes in individual behaviors and interaction patterns associated with improvements in team performance. Experiments with simulated surgical procedures show that our approach improves predictive performance in diverse behavioral and interaction goals while offering meaningful insights into team dynamics. This work advances surgical AI beyond outcome-driven prediction towards a socially grounded, team-centric, and actionable paradigm to better understand and support the development of team skills in surgical settings.
[LG-17] Leverag ing Remote Traffic Data for Local Air Pollutant Estimation: A Scenario-Based Machine Learning Study Across London Monitoring Sites
链接: https://arxiv.org/abs/2608.23219
作者: Valeria Legaria-Santiago,Amadeo Arguelles,Magdalena Saldana-Perez,Jocelyn Richardson,Marcella Bona
类目: Machine Learning (cs.LG)
*备注: Accepted for publication in Atmosphere
Abstract:Vehicular traffic is a major source of air pollution; however, the contribution of remotely acquired traffic information to local machine-learning (ML) air-pollution models remains insufficiently characterised. This study evaluates four interpretable tree-based ML models (Random Forest, Extra Trees, LightGBM, and XGBoost) under six predictor scenarios combining progressively larger predictor sets, ranging from remotely acquired traffic, meteorological, and temporal variables alone to the inclusion of measurements from one and four neighbouring monitoring stations, to estimate NO _2 , PM _10 , PM _2.5 , and O _3 concentrations across several sites in London. ML model performance was compared with a ridge linear regression model as a baseline, with spatial interpolation methods and with a cross-site validation experiment. When modelling without data from neighbouring stations, the RMSE for NO _2 ranged from 9.73 to 11.66 \mu g/m ^3 without traffic information, compared with 8.72 to 11.52 \mu g/m ^3 when traffic information was included. Additionally, for NO _2 , SHAP analyses indicate that traffic-related variables can contribute at levels comparable to pollutant measurements from neighbouring monitoring stations in traffic-dominated~environments.
[LG-18] When More Modalities Hurt: Modality Dropout for Heavy-Duty Vehicle Engine Diagnostics
链接: https://arxiv.org/abs/2608.23161
作者: Adeel Zafar,Slawomir Nowaczyk,Hamid Sarmadi,Saeed Gholami Shahbandi
类目: Machine Learning (cs.LG)
*备注:
Abstract:Heavy-duty vehicle diagnostics generate three disconnected data modalities: unstructured multi- lingual service complaints, high-dimensional sensor telemetry with over 80% missing values, and Diagnostic Trouble Codes (DTCs). We investigate whether fusing these modalities improves engine component classification on a proprietary dataset from a major truck manufacturer. Through 5-fold cross-validation across multiple model configurations spanning three model families on five engine component classes (885 samples, the full cross-database matched population for this manufacturer), we find that naive fusion provides modest gains over text alone (65.3%). However, modality dropout during training, which randomly disables entire modalities per batch, forces the network to exploit weaker inputs and achieves 68.8% accuracy on text+DTC fusion (weighted F1: 0.67), a 3.5-point improvement over text-only (65.3%, weighted F1: 0.64) and the best result across all methods including logistic regression and gradient-boosted trees. Per-class analysis shows that the dominant modality varies by fault type: text describes symptoms, DTCs encode structured fault signals, and sensors measure physical state. On intake/exhaust faults, sensors alone reach 93% where text achieves 80%. On fuel system faults, fusion with modality dropout nearly triples accuracy from 15% to 38% over text alone. To our knowledge, this is the first application of three-way modality fusion combining text, sensors, and fault codes in industrial vehicle diagnostics.
[LG-19] Conformal Risk Minimization for Semi-Supervised Domain Adaptation via Optimal Transport
链接: https://arxiv.org/abs/2608.23153
作者: Manos Giannopoulos,Yi Shen,Michael M. Zavlanos
类目: Machine Learning (cs.LG)
*备注: 18 pages, 2 figures, 6 tables
Abstract:In high-stakes healthcare applications, machine learning models are frequently trained on data from one patient population and deployed on another, creating a distribution shift that degrades both accuracy and reliability. Semi-Supervised Domain Adaptation (SSDA) addresses this by leveraging labeled data from some source domain to improve model performance on a target domain where labels are scarce. However, existing SSDA methods optimize primarily for point-prediction accuracy and offer no principled uncertainty quantification — a prerequisite for clinical trust. Conformal Prediction (CP) can address this limitation by providing prediction sets with rigorous, distribution-free coverage guarantees. However, applying CP post-hoc to a pre-trained model can yield prohibitively large prediction sets, as SSDA pre-training methods do not account for the nonconformity score geometry that determines conformal set size. Conformal Risk Minimization (CRM) has been used to resolve this issue in the fully supervised setting by integrating the CP objective directly into model training, but it requires a large labeled dataset to compute nonconformity thresholds during training, precisely the data that is scarce in the SSDA regime. We propose an end-to-end framework that integrates CRM into the SSDA training objective, enabling effective CRM in the limited-labeled-target-data regime. The key idea is to utilize Optimal Transport (OT) to generate pseudolabels for unlabeled target instances, providing the additional training signal needed by CRM to operate using only a small labeled target set. This results in a model jointly optimized for domain invariance and conformal efficiency, producing prediction sets that are compact, coverage-valid, and support domain-specific constraints such as excluding mutually contradictory diagnoses in skin lesion classification.
[LG-20] Macro-Action Topological Navigation under Noisy Localization using Reinforcement Learning
链接: https://arxiv.org/abs/2608.23055
作者: Simon Hakenes,Tobias Glasmachers
类目: Machine Learning (cs.LG); Robotics (cs.RO)
*备注: 15 pages, Accepted at the Artificial Intelligence Symposium (AIS) 2026
Abstract:Navigating large, photorealistic 3D apartments from raw pixels is widely considered infeasible for plain reinforcement learning. We build an agent that does it anyway, estimating its own pose from the camera alone. The agent has to reach several target objects in sequence, and their positions change between episodes, so it must explore to find them. It builds on our earlier object-centric topological controller, which still read the agent’s true pose and its object detections from the simulator. Here we replace that true pose with an onboard, object-centric estimate. For each object we keep a bank of ORB features that, when the object is seen again, yield a rough pose measurement, which a minimal Extended Kalman Filter (EKF) fuses with a motion model. As on a real robot, the executed motions are noisy. The estimate drifts, but the agent and the nearby objects drift together, so a locally consistent pose is enough to follow each short edge and then home in visually on the target, which lets us replace full SLAM with a much smaller model, closer to how biological navigation appears to work. In the photorealistic Habitat simulator, the agent reaches its target objects from vision alone, with a pose that only needs to be locally consistent.
[LG-21] Graph Representation Learning of Lightweight IoT Ciphers ICONIP2026
链接: https://arxiv.org/abs/2608.23054
作者: Jonathan Cook,Sabih ur Rehman,M. Arif Khan
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR)
*备注: This is the author’s version of a paper accepted at the 33rd International Conference on Neural Information Processing (ICONIP 2026)
Abstract:SIMON and SIMECK belong to a family of Lightweight Cryptographic Algorithms (LCAs) based on the Feistel block cipher, designed for Internet of Things (IoT) devices. As with all Feistel ciphers, they are susceptible to differential cryptanalysis, necessitating rigorous resilience evaluations. While state-of-the-art techniques leverage heuristics and sampling to improve efficiency, little work has applied Machine Learning (ML) guided Graph Representation Learning (GRL) to efficiently identify and visualise high-probability differential clusters. We address this gap by introducing an efficient feature engineering strategy that extracts four differential attributes from a partial Difference Distribution Table (pDDT), revealing structural information concealed in raw differential data. Utilising the enriched features, we construct and compare three ML-guided directed graphs for SIMON 32 and SIMECK 32 using K-Nearest Neighbour (KNN), Decision Trees (DT), and Random Forests (RF). To the best of our knowledge, our framework produces the first graph-based visualisation of the differential clustering effect, in which high-probability single-bit differentials form geometrically close clusters in the learned embedding. All three models achieve a precision of 1.0 in identifying high-probability differentials, confirming zero false positives. KNN achieves the strongest cluster separation, the highest F1 score and the lowest graph construction time of approximately 2.3 seconds, while DT and RF produce optimal paths with near-perfect regression. The results are consistent across both LCAs, demonstrating the applicability of the framework to other AND-rotation LCA families.
[LG-22] Reservoir of Importance: Learning Semi-Structured Sparsity with Differentiable Subset Sampling EMNLP2026
链接: https://arxiv.org/abs/2608.23048
作者: Ha Dinh,Xuan Duy Ta,Khoat Than,Khac-Hoai Nam Bui
类目: Machine Learning (cs.LG)
*备注: Accepted as an EMNLP 2026 Main Conference paper
Abstract:Semi-structured N : M sparsity has emerged as a practical direction for accelerating large language models (LLMs). However, existing learnable-mask approaches incur substantial parameter and memory overhead, limiting their scalability to large models and aggressive sparsity regimes. In this work, we revisit semi-structured pruning from a perspective that reconciles efficiency with scalability. We propose Reservoir of Importance (RoI), a lightweight semi-structured pruning framework that learns sparsity masks through differentiable subset sampling. Unlike prior methods that model full categorical distributions over all feasible N : M patterns, RoI introduces a compact-logit parameterization for sparsity mask learning and performs sampling without replacement to select masks, thereby reducing trainable parameters from combinatorial complexity to \mathcalO(M) . As a result, RoI requires 1.5-8.75 \times fewer learnable parameters and significantly lower memory cost, while remaining fully aligned with hardware-friendly sparsity patterns. Extensive evaluations across multiple scales of the Qwen2.5 LLM family (0.5-7B parameters) demonstrate that RoI achieves competitive performance with strong memory efficiency, stability, and scalability to more aggressive N : M sparsity patterns, offering a practical path toward efficient LLM deployment.
[LG-23] Channel-Token Attention for Reliable Dynamic Spectrum Access under Bursty Primary-User Traffic
链接: https://arxiv.org/abs/2608.22992
作者: Krishna Acharya,Dinanath Padhya,Utsab Dahal,Ashish Kandel,Binod Sapkota
类目: Networking and Internet Architecture (cs.NI); Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注: 9 pages, 4 figures, 2 tables
Abstract:Dynamic spectrum access must coordinate secondary users under bursty primary-user activity while preserving packet reliability and delay. We present TACAN, a centralized policy that represents each channel as a token containing occupancy history and automatic-modulation-classification entropy; a context token supplies queue class, delay and user identity. A Transformer encoder is warm-started from an occupancy-greedy policy and refined with proximal policy optimization. The frozen policies were trained to maintain a channel assignment in every slot, including when queues were empty. We therefore replay them on held-out trajectories and distinguish standby assignment success from packet-present access and packet delivery. In a 20-channel network with 60 primary devices and 4 secondary users, TACAN achieves 92.53% +/- 0.47 packet-present access success, compared with 89.94% for Greedy and 83.53% for PPO+MLP. Its paired gain over Greedy is 2.59 points (parametric 95% CI 1.89-3.29), with wins in all five seeds; the exact two-sided sign-test value is 0.0625. The gain rises from 0.57 points at normal primary-user load to 7.67 points at extreme load. TACAN also reduces mean delivery delay from 1.208 to 1.123 slots and the conditional user-reliability gap from 9.69 to 3.15 points. Delivered packets per SU-slot remain arrival-limited (30.12% versus 30.11%), so no packet-throughput gain is claimed.
[LG-24] Hierarchy-Aware Semantic Losses for Knowledge Graph Link Prediction
链接: https://arxiv.org/abs/2608.22981
作者: Filip Kronström,Ross D. King
类目: Machine Learning (cs.LG)
*备注:
Abstract:Knowledge graphs are often accompanied by ontological class hierarchies that encode valuable semantic information, yet many link prediction methods either ignore such hierarchies or incorporate them indirectly through additional graph edges. Recent work introduced hierarchy-aware graph neural networks (GNNs), which use semantic losses derived from box embeddings to encourage satisfaction of subclass relationships during GNN-based representation learning. While this approach has shown promise for biological regression tasks, its effectiveness for knowledge graph link prediction has not been investigated. In this paper we evaluate hierarchy-aware semantic losses on link prediction across three benchmark datasets: AIFB, CoDEx, and BioKG. We combine graph neural network encoders with box-embedding-based semantic losses that encourage learned representations to better satisfy ontology-derived class hierarchies, and compare this approach to both standard link prediction models and models incorporating subclass relations as graph edges. Across all datasets, hierarchy-aware semantic losses significantly improve mean reciprocal rank (MRR) and consistently outperform models that incorporate hierarchy information through additional subclass edges. Relative to the baseline GNN models, MRR improved by 7.6%, 2.4%, and 15.5% on AIFB, CoDEx, and BioKG, respectively. Furthermore, semantic losses consistently outperform the alternative of augmenting the graph with subclass edges. These results are consistent with ontology-derived class hierarchies providing complementary information to graph structure, and suggest that encouraging hierarchical consistency through semantic losses is an effective and comparatively parameter-efficient mechanism for improving knowledge graph link prediction. Subjects: Machine Learning (cs.LG) ACMclasses: I.2; I.2.4; I.2.6 Cite as: arXiv:2608.22981 [cs.LG] (or arXiv:2608.22981v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2608.22981 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-25] Do Time-Series Foundation Models Pay Off for Industrial Monitoring? A Cost-Aware Empirical Study
链接: https://arxiv.org/abs/2608.22968
作者: Guan-Hua Wen,Kuan-Yu Chen
类目: Machine Learning (cs.LG)
*备注: Accepted for poster presentation at CIF26 (2026), Poster P01013. 7 pages, 2 figures
Abstract:Industrial monitoring models must detect operationally relevant deviations while satisfying target-specific data, calibration, and resource constraints. Time-series foundation models (TSFMs) promise reusable representations and zero-shot forecasts, yet evidence for their deployment value remains mixed when task definitions are heterogeneous and lightweight baselines are competitive. This work presents a protocol-aware empirical assessment across three settings: a C-MAPSS degradation-risk proxy, normal-only training for anomalous-sound detection on MIMII, and BDG2 forecasting-residual diagnostics with synthetic target perturbations. We assess classical one-class methods, compact neural autoencoders, residual forecasters, MOMENT-small, Chronos-T5, and TimesFM 2.5 in terms of anomaly-ranking performance, risk-horizon sensitivity, residual forecasting and perturbation sensitivity, and local implementation cost. Across 100 C-MAPSS engines evaluated out of fold, TCN-AE reaches fold-weighted AUROC/AUPRC 0.9570/0.8960, compared with 0.7310/0.3080 for MOMENT reconstruction; paired engine-cluster bootstrap confidence intervals exclude zero for both differences. Across five matched MIMII pump evaluations, OCSVM also exceeds MOMENT reconstruction in AUROC and AUPRC. On a fixed 12-meter BDG2 panel, TimesFM 2.5 has the lowest aligned forecast error and the highest synthetic AUROC point estimate, although synthetic AUPRC is similar across TSFM and fitted residual models. Same-device measurements show that MOMENT incurs higher latency, peak allocated VRAM, and serialized state-dictionary size than TCN-AE. Under the evaluated frozen and zero-shot settings, TSFMs are task-dependent deployment options rather than default replacements for fitted lightweight models.
[LG-26] Stochastic gradient descent with initial regularization
链接: https://arxiv.org/abs/2608.22953
作者: Nabil Kahalé
类目: Machine Learning (cs.LG); Optimization and Control (math.OC); Machine Learning (stat.ML)
*备注: 33 pages
Abstract:We analyze a variant of stochastic gradient descent with initial regularization (SGDIR) and derive dimension-free upper bounds on its expected excess risk for the squared loss. In the noiseless case, we obtain new bounds for both averaged and non-averaged SGDIR under moment, source, and capacity assumptions. For a particular value of the source parameter, these bounds are of order m^-2\log^2m , where the number of training samples is of order m . For another value of the source parameter, we obtain, for any \epsilon0 , bounds of order m^-3+\epsilon , provided that the capacity parameter exceeds \epsilon^-1 . We also establish a lower bound that matches our upper bounds in certain regimes up to a polylogarithmic factor. In the noisy case, we provide an instance-based comparison between SGDIR and ridge regression. Under general assumptions and a mild lower bound on the regularization parameter, we show that the expected excess risk of SGDIR is no larger than that of ridge regression, up to a polylogarithmic factor. Numerical experiments on synthetic and real data are consistent with our theoretical findings.
[LG-27] A Momentum-Based Variance-Reduced Algorithm for Federated Multiobjective Optimization
链接: https://arxiv.org/abs/2608.22945
作者: Yong Zhao,Chunlin You,Minh N. Dao,Zai-Yun Peng
类目: Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注:
Abstract:Federated learning has traditionally been formulated as a single-objective optimization problem, primarily focused on maximizing model utility. In real-world applications, however, machine learning models often need to optimize multiple and potentially conflicting objectives simultaneously. This motivates federated multiobjective optimization (FMOO), which provides a natural framework for jointly handling multiple task-specific objectives in federated learning. In this paper, we propose a momentum-based variance-reduced algorithm for federated multiobjective optimization. The method incorporates a momentum-driven gradient estimator into the local updates to reduce the variance of stochastic updates, leading to an improved convergence rate. We establish theoretical guarantees showing that the expected Pareto stationarity measure of a randomly selected output iterate decays at a rate of \mathcalO(T^-2/3) , improving upon the \mathcalO(T^-1/2) rates established for existing methods such as FSMGDA and FedCMOO. Numerical experiments on federated multiobjective optimization benchmarks demonstrate the effectiveness and competitive performance of the proposed algorithm.
[LG-28] hinking at the Right Size: Amortized Distillation Across Post-Trained LLM s EMNLP2026
链接: https://arxiv.org/abs/2608.22854
作者: Yan Zhou,Sara Kangaslahti,Jonathan Geuter,Nihal V. Nayak,Marco Fumero,Francesco Locatello,David Alvarez-Melis
类目: Machine Learning (cs.LG)
*备注: 8 pages, 6 figures. EMNLP 2026 Findings
Abstract:Practical deployment of large language models (LLMs) requires families of post-trained variants—instruction-tuned, reasoning-tuned, and chat-style models—each at multiple sizes to meet diverse latency and memory budgets. Producing each (variant, size) pair independently is prohibitive, so model families typically span only a handful of coarse-grained sizes per post-trained variant. Boomerang distillation (Kangaslahti et al., 2026) reduces this cost along the size axis for base models. Through model size interpolation, it constructs models of intermediate sizes from a single teacher-student pair without additional training. However, it still treats each post-trained variant as a separate object of optimization. We introduce ADAPT—Amortized Distillation Across Post-Trained LLMs—a framework for amortizing distillation across both axes of a model family: size and post-training variant, producing L \times K models for L interpolated sizes across K post-trained variants with a single distillation run. ADAPT combines two components. First, a two-phase distillation procedure constructs post-trained students through pre-training alignment and supervised fine-tuning distillation, enabling smooth size–performance interpolation on generation and reasoning tasks. Second, weight-delta initialization approximates this construction across post-trained variants by transferring the distillation-induced weight change from the base model to students initialized from different post-trained variants. The resulting continuum of interpolated models also enables adaptive model-size selection at inference time, improving the compute–accuracy trade-off for long-form reasoning tasks.
[LG-29] RIBOSPAN: A Long-Context RNA Foundation Model for Versatile RNA Modeling
链接: https://arxiv.org/abs/2608.22849
作者: Ziyuan Wang,Bohao Tang,Fei Zhang,Shuo Han,Pengfei Liu
类目: Machine Learning (cs.LG); Genomics (q-bio.GN)
*备注: 21 pages, 5 figures
Abstract:Full-length RNAs, particularly messenger RNAs, often exceed the context lengths used to pretrain existing RNA foundation models, limiting complete-transcript modeling at single-nucleotide resolution. We present RIBOSPAN, a 1.61-billion-parameter bidirectional RNA foundation model natively pretrained with context lengths up to 10,240 nt. RIBOSPAN combines dense bidirectional self-attention, single-nucleotide tokenization, and attention-isolated sequence packing to enable high-resolution modeling of complete long RNAs. We evaluate the model through nucleotide reconstruction, a controlled long-context representation benchmark, and frozen RNA-type representation analysis. Native 10K pretraining preserves strong reconstruction at 10,240 tokens, while continued pretraining with 40% masking improves recovery under heavy corruption while preserving representation quality. The long-context benchmark further shows that native 10K models maintain strong contextual responsiveness and context-specific representation separation while keeping perturbation-induced representation changes highly localized. Inference-time YaRN scaling recovers much of the contextual organization lost by direct extrapolation of short-context models, but induces substantially greater distal representation diffusion. Frozen-representation evaluations further demonstrate state-of-the-art RNA representation quality, with RIBOSPAN achieving the strongest overall performance across diverse RNA types and retaining a clear advantage on long RNAs. Building on the same backbone, we develop a multidimensionally conditioned discrete-diffusion framework for full-length mRNA generation and redesign, including synonymous-codon diffusion for protein-preserving CDS optimization. Together, RIBOSPAN establishes a powerful long-context foundation for transferable RNA representation learning and full-transcript mRNA design.
[LG-30] DIME: Query-Efficient Framework for Membership Inference on Diffusion Models
链接: https://arxiv.org/abs/2608.22824
作者: Tue Do,Daniel Alabi
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR)
*备注:
Abstract:Membership inference attacks expose whether individual records were used to train a model, yet existing attacks on diffusion models are largely heuristic and can require substantial query budgets. We introduce DIME (Denoiser Ideal Membership Error), a theoretically grounded and query-efficient framework for membership inference on diffusion models. Our starting point is an exact characterization of the optimal diffusion denoiser for a finite training set, which reveals that membership leakage is governed by the denoiser’s implicit reconstruction error. This error decomposes into two complementary signals: a bias term, capturing reconstruction accuracy, and a previously unexplored local crowding term, capturing the geometry of nearby training examples. Both admit efficient estimators using only model queries, yielding a practical attack with as few as two queries. Across CIFAR-10/100, STL10-U, CelebA, and ImageNet, DIME consistently outperforms prior attacks at comparable or substantially lower query cost, improving TPR at 1% FPR by up to 3\times ; remarkably, its two-query variant can outperform existing 30-query baselines. Finally, we suggest, discuss, and evaluate specific defenses to counteract such powerful membership tests.
[LG-31] SAGE: Stability-Aware Graph-Based Ensemble Feature Selection for Explainable Postpartum Depression Risk Prediction
链接: https://arxiv.org/abs/2608.22809
作者: Md. Rokon Islam Emon,Syed Shariar Alam Shuvo,Shahriar Siddique Ayon,Abdullah Al Mamun,Ahnaf Atef Choudhury
类目: Machine Learning (cs.LG)
*备注: Accepted for publication at 10th International Artificial Intelligence and Data Processing Symposium (IDAP’26)
Abstract:Postpartum depression (PPD) poses a major burden on maternal and child health, especially in low- and middle-income countries where prevalence exceeds 19%. Despite advancements in machine learning for PPD prediction, current approaches are limited by opaque global explanations that lack clinical usefulness at the patient level, unstable feature selection, and poor generalization under class imbalance. We propose SAGE, a Stability-Aware Graph-Based Ensemble feature selection system that incorporates both local explainable AI and a genetically optimized artificial neural network (GA-ANN). Using a primary cohort of 766 postpartum women, SAGE combines information-theoretic relevance, PCA-based structure, and graph-based interactions with bootstrap stability weighting to identify robust and non-redundant predictors. The GA-ANN architecture, optimized using a genetic algorithm and enhanced with GAN based oversampling, achieved strong performance with 87.96% accuracy, 86.32% F1 score, and 0.88 AUC using only 16 features, outperforming baseline and other feature selection methods. Psychological and socioeconomic factors such as EPDS score, PHQ-9 score, feelings about motherhood, and abuse history are the main predictors, while demographic factors have less influence. The LIME-based explanations allow instance-based insight into selected features from the graph, enabling personalized risk assessment. The findings make SAGE a scalable, interpretable, and clinical tool for early identification of PPD in health-care limited resources.
[LG-32] Change Detection in Probability Flow ODE: Online Testing in Diffusion Latent Spaces
链接: https://arxiv.org/abs/2608.22807
作者: Artem Kraevskiy,Artem Prokhorov
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:A rapidly growing range of sequential data tasks, such as identifying trend reversals in financial markets, auto-segmenting video and audio recordings, detecting changes in movement direction from motion sensors cannot be fully addressed without detection of distributional shifts in time-ordered data. We consider a sequential change-point detection problem where the conditional density switches at an unknown time, yet neither the pre- nor post-change distribution admits a closed-form. Classical likelihood-ratio statistics are inapplicable in this settings. A conditional diffusion model, trained on pre-change-point data with a frozen context encoder, defines a deterministic bijection via the probability flow ODE. Pre-change observations are mapped onto standard Gaussian latent variables. Post-change observations, processed through the same frozen map, deviate from this reference. We employ the Maximum Mean Discrepancy as the test statistic, derive closed-form expressions for its components under the Gaussian null, and establish its asymptotic distribution as a degenerate U-statistic. Afterwards we apply an online detection procedure of Shiryaev–Roberts to the resulting statistic with exact threshold calibration. The method detects arbitrary distributional shifts, including covariance rotations and higher-order structural breaks, without parametric assumptions on either regime. Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML) Cite as: arXiv:2608.22807 [cs.LG] (or arXiv:2608.22807v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2608.22807 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-33] Contrastive Representation-Guided Genetic Minority Oversampling for Imbalanced Time-Series Classification
链接: https://arxiv.org/abs/2608.22804
作者: Wenbin Pei,Yunrong Hao,Zhen Liu,Guan Wang,Bing Xue,Yiu-Ming Cheung,Qiang Zhang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Real-world time-series classification tasks often exhibit class imbalance, which can be extremely severe in some applications. To avoid training biased classifiers on imbalanced data, sampling is one of the most popular data pre-processing techniques because of its classifier-agnostic nature. However, due to the complex temporal dependencies in original time-series data and the scarcity of minority-class samples, existing sampling methods, including interpolation-based oversampling methods and deep learning-based generative models, usually suffer from limited generalization and poor diversity when generating new time-series samples. This paper proposes a Frequency-domain representation-guided Multi-tree Genetic Programming-based oversampling approach (FreMGP) to imbalanced time-series classification, where each individual represents a set of synthetic samples for the minority class. A frequency-domain class-discriminative representation module based on contrastive learning is also developed, guiding the evolutionary search toward high-quality synthetic time-series samples. Experiments on imbalanced time-series datasets demonstrate that FreMGP outperforms existing oversampling methods and consistently improves the performance of different classifiers, including both general machine learning and deep learning models.
[LG-34] ReCoG: Reciprocal Co-Evolution for Multimodal Graph Learning
链接: https://arxiv.org/abs/2608.22786
作者: Rui Xue,Tianfu Wu
类目: Machine Learning (cs.LG)
*备注:
Abstract:Multimodal graph learning requires jointly training over graph structure and heterogeneous node attributes, yet existing methods largely decouple these processes: prior multimodal graph neural networks (GNNs) focus on aligning modalities in a shared embedding space while operating on fixed or weakly adapted graph structures, and graph structure learning approaches infer topology from unimodal node representations without accounting for multimodal interactions. This separation fundamentally limits the ability of GNNs to capture semantically meaningful relationships in multimodal settings, where observed edges are often noisy, incomplete, or misaligned with underlying semantics. We propose ReCoG (Reciprocal Co-Evolution for Multimodal Graph Learning), a new learning paradigm that tightly couples graph structure learning and multimodal representation learning through end-to-end reciprocal interaction. Concretely, ReCoG integrates (i) a multimodal graph refiner that infers and corrects edges using cross-modal semantic evidence, and (ii) a coupled cross-modal message passing mechanism that performs joint intra- and inter-modality propagation over the refined graph. This unified design yields greater expressiveness than decoupled or two-stage formulations and allows dynamic interaction between topology and representation learning. Across diverse benchmarks for node classification and link prediction, ReCoG consistently outperforms strong multimodal graph structure learning baselines, including graph foundation models. Our results demonstrate that reciprocal co-evolution of structure and semantics is important for effective multimodal graph learning, challenging the prevailing separation between topology and representation learning.
[LG-35] Learning to Control Coupled-Dynamics Environments with Joint Markov Decision Processes
链接: https://arxiv.org/abs/2608.22765
作者: Ege C. Kaya,Aliasghar Pourghani,Mahsa Ghasemi,Vijay Gupta,Abolfazl Hashemi
类目: Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注: 14 pages, 5 figures
Abstract:Coupled-dynamics environments expose the one-step outcomes that would follow from several possible counterfactual actions under a common realization of exogenous randomness. The ordinary Markov decision process formalism allows one to reason about the marginal law of each action but discards dependence across these counterfactual outcomes. The Joint Markov decision process (JMDP) formalism preserves that dependence. Prior work established the formalism and solved the fixed-policy joint moment evaluation problem in JMDPs. This paper develops optimal-control methods. We define a nonparametric distributional Bellman optimality operator for JMDPs, and prove that when the induced marginal MDP has a unique optimal policy, its iterates converge in Wasserstein distance to the optimal joint return law. For the first two moments, we establish convergence under a weaker condition that permits several mean-optimal actions as long as their tie resolutions share a second-moment fixed point. We also derive sampled targets for neural approximation.
[LG-36] LpWM: A Case for Sparse Representations in World Models
链接: https://arxiv.org/abs/2608.22764
作者: Yilun Kuang,Yash Dagade,Quentin Le Lidec,Lucas Maes,Randall Balestriero,Yann LeCun
类目: Machine Learning (cs.LG)
*备注:
Abstract:Joint-embedding predictive architectures (JEPAs) learn latent dynamics for planning and avoid representation collapse by matching features to maximum-entropy distributions such as isotropic Gaussians, yielding dense representations. However, it is unclear whether dense representations are the most favorable geometry for modeling dynamics. In this work, we ask whether a different geometry, sparse representations, can make action-conditioned latent dynamics easier to model, and what dynamical structure emerges from such representations. We first show that nonlinear Lipschitz dynamics can be approximated arbitrarily well by action-conditioned linear dynamics in a sufficiently high-dimensional one-hot latent space, with rollout error vanishing as the dimension grows. This motivates distributed sparse representations as a practical relaxation of one-hot sparsity. We introduce LpWorldModel (LpWM), a JEPA model regularized with Rectified Distribution Matching Regularization (RDMReg) to match encoder features to a Rectified Generalized Gaussian distribution, yielding non-negative sparse codes. Empirically, sparsity lowers the predictor complexity required for successful planning: on PushT, sparse LpWM outperforms dense LeWM by up to 57% in planning success at intermediate predictor capacities. This advantage also extends beyond Gaussian distribution matching, with LpWM outperforming dense VICReg representations across multiple predictor families. We further find that the learned sparse representations are mode-factored, with support encoding discrete dynamical regimes and feature magnitudes capturing continuous within-regime state. Together, these results suggest that sparse representations can reduce the predictor complexity required for control while revealing interpretable structure.
[LG-37] MOSH-WM: Mask-Grounded Soft-Hamiltonian Dynamics for Object-Centric World Models
链接: https://arxiv.org/abs/2608.22750
作者: Zhekai Wang,Haoxiang Huang,Xiang Liu,Zhikang Chen,Yueqing Sun,Qi Gu,Shiji Zhou,Miao Liu,Sen Cui
类目: Machine Learning (cs.LG)
*备注:
Abstract:Object-centric world models forecast future videos by evolving a set of entity slots, but the variables receiving dynamics supervision are often unconstrained visual features. We introduce \method, a mask-grounded soft-Hamiltonian world model that makes its position-like state explicitly depend on slot-owned image support. A frozen video-slot encoder produces slots and masks; spatial moments of mask-owned support form a canonical state Q , temporal differences form P , and a learned energy supplies a soft directional bias to a bounded learned increment. Decoder-relevant appearance and identity are stored separately in a causal visual context. A gated composer and bounded residual then combine this context with the propagated phase state to reconstruct decoder-compatible slots. On OBJ3D, given six observed frames and evaluated over the following 30 frames, \method reduces LPIPS by 25.0% and spatial MSE by 33.7% relative to the strongest object-centric baseline. On CLEVRER, given six observed frames and evaluated over the following ten frames, the corresponding reductions are 14.5% and 18.7%. Horizon-resolved visual and object-state measurements show that the complete model accumulates error more slowly throughout the 30-frame closed-loop rollout. Project page:this https URL.
[LG-38] Learning Generalizable Behaviors for Terminal Agents
链接: https://arxiv.org/abs/2608.22631
作者: Yihang Yao,Bo Pang,Xuan Phi Nguyen,Ding Zhao,Shafiq Joty,Semih Yavuz
类目: Machine Learning (cs.LG)
*备注:
Abstract:Terminal agents are a compelling application of large language models (LLMs), with the potential to integrate deeply into users’ daily workflows. Reinforcement learning (RL) is a key technique for improving their capabilities, making scalable training environments a central challenge. Since public real-user interaction data are scarce, synthetic environments provide a practical alternative, but often suffer from domain gaps and limited fidelity, leading to poor generalization. Existing work mainly scales the quantity and diversity of synthetic environments, while reward-signal quality and the mechanisms governing generalization remain under-explored. We study how RL improves terminal agents and propose the Agentic Compositional Generalization hypothesis: rather than teaching new domain-specific skills from scratch, RL primarily shapes high-level decision-making behaviors that compose and route low-level skills acquired during pre-training and supervised fine-tuning (SFT). This account is consistent with our empirical results and suggests that verifier quality, which determines which behaviors are reinforced, is more important than simply increasing environment quantity or diversity. Motivated by this insight, we propose River, a simple training recipe that improves reward quality by filtering low-quality environments and augmenting outcome rewards with process-level behavior regularization. Using this recipe, our RL-trained agent achieves the best performance among evaluated open-source RL-trained 8B models across four terminal-agent benchmarks. River also generalizes across model families, scales, agent harnesses, and RL objectives. Using fewer than 30% of the TMax training environments, River improves RL gains by 106% and 30% on average for models ranging from 2B to 27B on Terminal-Bench-Lite and Terminal-Bench-v2.1, respectively.
[LG-39] KMGen: A Skill-based Approach for Synthetic Individual Patient Data Generation
链接: https://arxiv.org/abs/2608.22618
作者: Jalen Jiang,Chufan Gao,Ethan Rasmussen,Stephen Z. Xie,Jimeng Sun
类目: Machine Learning (cs.LG)
*备注:
Abstract:Individual patient data (IPD) from clinical trials is the substrate for survival modeling, meta-analysis, and safety research, yet IPD is rarely released. Prior work has addressed only half of this gap: reconstructing Kaplan-Meier (KM) curves from published plots – typically requiring manual digitization or human-in-the-loop correction – while offering no mechanism for generating the adverse-event (AE) streams that constitute the other half of a patient record. We introduce KMGen, the first end-to-end framework that (i) fully automates KM curve extraction at accuracy competitive with human-guided tools, and (ii) generates synthetic per-patient AE trajectories from public trial registry records. The extraction stage is a fully automated agentic pipeline – an agent generates code to extract each step in the KM curve – achieving a mean Integrated Absolute Error (IAE) of 0.0151 on a 32-plot benchmark spanning clean, edge-case, and adversarial conditions. The IPD generation stage decouples patient archetype extraction from statistical sampling: an LLM distills the trial record into arm-specific statistics, adverse events, patient demographics, and risk multipliers. A mechanistic sampler generates patient events via clinical archetypes, bootstrap rank-correlation coupling to the empirical KM curve (preserving the marginal survival distribution exactly), and cycle-based AE scheduling with an induction/maintenance split. Across three held-out oncology trials spanning an order of magnitude in cohort size and 30 independent regenerations per trial, KMGen achieves mean integrated KM absolute difference \Delta_\textKM,\leq,0.051 , sex/ECOG JSD \leq,0.013 on 5 of 6 demographic slots, and recovers \geq,71% of the top-15 AEs by exact MedDRA term under a single fixed parameter set. The pipeline is released as open source at this https URL.
[LG-40] What AstroPT knows about galaxies and what that can teach us about LLM s
链接: https://arxiv.org/abs/2608.22614
作者: UniverseTBD:Kshitij Duraphe,Aman Kumar,Michael J. Smith,Shashwat Sourav
类目: Machine Learning (cs.LG); Instrumentation and Methods for Astrophysics (astro-ph.IM)
*备注: 13 pages, 12 figures, code at this https URL , accepted at Sci-FM@COLM 2026
Abstract:Interpretability research increasingly asks when concepts emerge during training and whether linear probes recover real structure, but in language models these claims are hard to validate because language offers little ground-truth ordering of concepts or relationships among them. We propose the use of astronomical ground truth through AstroPT, a transformer trained on millions of galaxy images, as a calibration testbed. AstroPT is an LLM-like model trained within a domain where the difficulty ordering of concepts and the relations among them are known in advance. Probing frozen representations across checkpoints, layers, model sizes, and objective choices, we find that galaxy properties emerge in a fixed order that tracks their known difficulty—quantities written almost directly into the pixels (band magnitude) become decodable early in training and shallow in the network, while multiband/spectra based and inferred quantities (such as redshift and specific star formation rate) emerge later and deeper. This order is invariant to our tested training objectives, and scales in magnitude but not in sequence with capacity. Our linear probe directions further recover the known physical structure among galaxy properties. Our findings suggest that astronomy offers a controlled sandbox for calibrating mechanistic interpretability methods we otherwise apply to LLMs blind.
[LG-41] Mitigating Explanation Leakage in Financial Fraud Detection Systems
链接: https://arxiv.org/abs/2608.22607
作者: Muhammad Waleed Gul,Elaheh Homayounvala
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR)
*备注:
Abstract:Financial fraud detection relies heavily on centralized machine learning models. This creates serious data privacy risks. Federated Learning (FL) decentralizes data processing, but financial regulations still require models to be transparent. This means using Explainable AI (XAI) tools such as TreeSHAP. Recent cybersecurity research shows a problem with this approach. Sharing high-fidelity SHAP explanations exposes the federated network to Membership Inference Attacks (MIAs). This dissertation proposes and evaluates DP-FedSHAP. It is a new architecture that applies client-level differential privacy only to post-hoc TreeSHAP vectors. It is compared against a Weight-Level DP baseline, which perturbs the trained model directly instead. Using the highly imbalanced IEEE-CIS Fraud Detection dataset, this study measures the trade-off between explanation fidelity, privacy preservation, and the model’s Area Under the Precision-Recall Curve (AUPRC).
[LG-42] Adversarial Agents on Topology Optimization: Understanding the Frag ility and Robustness of Deep Learning-based and Physics-Based Design Models under Adversarial Perturbation
链接: https://arxiv.org/abs/2608.22606
作者: Hoang Anh Nguyen,Yuan Hong,Hongyi Xu
类目: Machine Learning (cs.LG)
*备注:
Abstract:Topology optimization, using both physic-based approaches and deep learning surrogates, serves as a cornerstone for generative design agents in cyber-manufacturing systems. While deep learning surrogates have gained widespread adoption due to their speed in online design generation, this work demonstrates their vulnerability under input perturbations. In this work, we present a mechanics-grounded reliability evaluation framework that formulates an adversarial agent targeting the generative design models. We investigate a strictly non-intrusive threat model where bounded perturbations are introduced exclusively to the initial-density channel, while physical boundary conditions, compliance-gradient channels, network architectures, and solver routines remain intact. Evaluating surrogate models across U-Net, convolutional, and generative architectures with varying physics-gradient conditioning depths demonstrates that bounded initialization noise can cause catastrophic mechanical failure, increasing compliance by multiple orders of magnitude through severed load paths and disconnected supports. Furthermore, we discover that incorporating richer physics-gradient conditioning in the deep learning surrogates does not guarantee monotonic robustness across surrogate families. Finally, physics-in-the-loop recovery demonstrates that initializing the classical SIMP optimizer with perturbed topologies mitigates design performance degradation, having a high probability of restoring compliance to near-baseline levels across tested instances. These findings demonstrate that learned surrogates should serve as physics-verified initializers instead of replacing physics-based solvers entirely in a resilient cyber-manufacturing system. Moreover, the proposed adversarial agent provides a foundation for future training generative design agents robust against noise and targeted perturbations.
[LG-43] abular foundation models for non-tabular tasks
链接: https://arxiv.org/abs/2608.22594
作者: Goran Nakerst,John Brennan,Wouter Beugeling,Masudul Haque
类目: Machine Learning (cs.LG)
*备注: 7 pages, 5 figures
Abstract:Tabular foundation models (TFMs) have recently emerged as a promising paradigm for machine learning on tabular data, offering the ability to generalize across datasets without task-specific training. Since many machine learning datasets can be represented as tables, this raises the question: does TFM capability extend beyond tasks traditionally regarded as tabular? We address this question by using TabPFN v3 on three non-tabular classification problems: handwritten digit recognition on MNIST, language identification of French and German words, and image classification on Tiny ImageNet. In each case, the original data are represented as rows of a table and classification is formulated as prediction of a missing label. We evaluate performance as a function of the number of context samples provided to the pretrained model, with no additional training or fine-tuning. Despite having no explicit access to the spatial or sequential structure characterizing the data, TabPFN v3 in some cases achieves accuracies comparable with that of models or methods geared specifically toward the corresponding tasks.
[LG-44] GCA: Global Centroid Alignment in Federated Learning
链接: https://arxiv.org/abs/2608.22593
作者: Jong-Ik Park,Harry Jiang,Logan Blakely,Georgios Fragkos,Shamina Hossain-McKenzie,Carlee Joe-Wong
类目: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC)
*备注:
Abstract:Autoencoder (AE)-based federated learning (FL) is attractive for anomaly detection when clients have limited local data. However, conventional FL exchanges AE parameters or gradients, incurring substantial communication overhead and potentially exposing input training data information, since AEs are explicitly optimized to reconstruct their inputs. We introduce \emphGlobal Centroid Alignment (GCA), a latent-code-mediated FL protocol that coordinates clients without transmitting AE parameters or gradients. In each round, (1) clients first train their local AEs using a \emphreconstruction update and upload a small subset of encoder latent codes to the FL server. (2) The server pools these codes, fits a clustering model, and broadcasts only \emphglobal latent centroids and their support counts. (3) Each client then updates its encoder by aligning its local latent codes with the \emphnearest centroid using \emphinverse-count weighting to emphasize globally underrepresented patterns. Steps (1)–(3) repeat over communication rounds. Because GCA exchanges only sampled latent codes and centroid statistics, its communication cost depends on latent dimensionality and the numbers of uploaded codes and returned centroids rather than on AE model size. Across five tabular and two vision benchmarks, GCA yields higher reconstruction error under a server-side client data extraction attack in all 21 comparisons and clearly lower cosine similarity in 20 of 21 comparisons with FedAvg, FedProx, and FedNova, showing its ability to protect training data. It even improves test accuracy over FedAvg by up to 5.76% . GCA achieves extraction defense comparable to DP-FedAvg, remains effective when DP-FedAvg does not reduce target resemblance, and lowers per-round communication by up to 99.15% .
[LG-45] wo-level domain-decomposition AdaGrad method for scalable training of graph neural networks
链接: https://arxiv.org/abs/2608.22575
作者: Laurynas Varnas,Julien Herrmann,Alexander Heinlein,Serge Gratton,Alena Kopaničáková
类目: Numerical Analysis (math.NA); Machine Learning (cs.LG)
*备注:
Abstract:Graph neural networks (GNNs) have emerged as a powerful framework for learning from graph-structured data. However, their efficient training remains challenging, particularly in distributed computing environments. This challenge arises from the use of message passing, which couples all graph nodes, leading to expensive optimization steps, high memory requirements, and substantial communication overhead. To alleviate these limitations, we propose a novel domain-decomposition (DD) variant of AG2m, an AdaGrad method enhanced with second-order curvature information and momentum, denoted by DD-AG2m. The proposed DD-AG2m alternates between AG2m optimization on the original (global) graph and AG2m optimization on the partitioned graphs. To incorporate global information at reduced cost, we further introduce a two-level variant (2DD-AG2m) that performs global optimization steps on a coarse graph obtained by randomly subsampling nodes within each subdomain. Numerical experiments spanning graph classification, node-level regression, and spatiotemporal forecasting tasks demonstrate that the proposed DD methods reduce the computational cost required to achieve the same predictive performance by a factor of 4-8. Moreover, for the fixed computational cost, they improve the predictive performance of GNNs by up to 22% compared with the baseline AG2m.
[LG-46] Neighbor-embedded Graph Neural Network-based Crowd Delivery Traffic Management in Smart City
链接: https://arxiv.org/abs/2608.22555
作者: Kishu Gupta,Deepika Saxena,Ashutosh Kumar Singh,Chung-Nan Lee
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注:
Abstract:The significant upsurge in vehicle traffic presents a considerable challenge in the pursuit of smart mobilization and transportation (SMT) worldwide. Current approaches primarily focus on vehicular traffic management through congestion prediction but fall short in addressing essential objectives such as traffic reduction and appropriate vehicle selection to alleviate congestion in smart cities ( SmCt ). To address these concerns, this work introduces a novel \textitNeighbor-Embedded Graph Neural Network-based Crowd Delivery Traffic Management (NeCDM) Model, comprising two key components: the Traffic Congestion Prediction Unit (TCPu) and the Traffic Observation and Management Unit (TOMu). The TCPu utilizes Graph Neural Network (GNN) optimization to accurately predict traffic flow levels at various delivery stations within SmCt ecosystems. Additionally, the TOMu facilitates the intelligent selection of the most suitable delivery vehicles for fulfilling crowd delivery requests ( CDR ). This work emphasizes the potential of crowd delivery as a feasible solution for achieving SMT goals while adhering to smart city parameters ( \mathcalSCP s), such as reduced carbon emissions, shorter travel times, and minimized travel distances. The proposed model achieves notable improvements in computational efficiency, including reductions of up to 4.03% in L1 loss ( £ ), 16.66% in L2 loss ( £_rmse ), and 7.64% in computation time.
[LG-47] Model-Consistent Byzantine-Resilient Decentralized Federated Learning for Collaborative Missions
链接: https://arxiv.org/abs/2608.22552
作者: Yue Li,Sudip Bhujel,Cameron Lira,Ning Wang,Yang Xiao
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注: Accepted for publication at the IEEE Conference on Communications and Network Security (CNS), 2026
Abstract:Decentralized federated learning (DFL) is a promising paradigm for autonomous nodes to collaboratively train AI models without relying on a central server. However, existing DFL solutions do not guarantee global model consistency, a critical requirement for collaborative mission-critical scenarios where model divergence undermines decision uniformity and safety. This lack of consistency also amplifies vulnerability to Byzantine adversaries, who exploit the decentralized network topology and weak synchrony to perform equivocation and model poisoning attacks against individual victims. This paper introduces DFL-C, a novel Byzantine-resilient DFL architecture that enables decentralized nodes to perform collaborative training with global model consistency. At its core, DFL-C integrates an asynchronous common subset (ACS) consensus protocol into the DFL workflow to ensure all nodes aggregate a uniform set of model updates to establish global model consistency, despite individual Byzantine equivocation. DFL-C further implements a dual-domain trust scoring mechanism to provide resilience against data-domain Byzantine manipulations including model poisoning attacks. This mechanism complements the consensus protocol, significantly reducing the latter’s runtime. Our experimental results demonstrate that DFL-C maintains model accuracy while achieving global model consistency under Byzantine behaviors with moderate consensus overhead. Notably, when compared with the state-of-the-art DFL solution BALANCE (Fang et al.) that does not provide model consistency, DFL-C achieves better model accuracy against untargeted model poisoning attacks and comparable resilience against backdoor attacks, with the advantage widened under non-IID scenarios. Comments: Accepted for publication at the IEEE Conference on Communications and Network Security (CNS), 2026 Subjects: Distributed, Parallel, and Cluster Computing (cs.DC); Cryptography and Security (cs.CR); Machine Learning (cs.LG) Cite as: arXiv:2608.22552 [cs.DC] (or arXiv:2608.22552v1 [cs.DC] for this version) https://doi.org/10.48550/arXiv.2608.22552 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-48] Stress Testing Unlearning Algorithms
链接: https://arxiv.org/abs/2608.22527
作者: Noam Diamant,Ethan Fetaya,Neta Glazer
类目: Machine Learning (cs.LG)
*备注:
Abstract:Recently, machine unlearning, the removal of specific training data influence from a model, has gained increasing attention. In large language models (LLMs), unlearning is particularly challenging due to the ambiguity of inputs and outputs. Con- sequently, rigorous evaluation is critical for assessing both safety and utility, and for driving progress in unlearning meth- ods. We identify two key shortcomings in existing unlearning benchmarks: (1) they do not actively test whether unlearned information can still be forcibly extracted, and (2) they fail to evaluate performance preservation on boundary questions, be- nign queries that are semantically close to the unlearned con- tent. Here we introduce WMDP++, an extension of WMDP that addresses these gaps by incorporating targeted extrac- tion of unlearned information and systematic evaluation on boundary questions. WMDP++ provides a more stringent and informative benchmark for evaluating unlearning in LLMs.
[LG-49] From Detrimental to Beneficial: Dynamic Influence-based Valuation and Editing
链接: https://arxiv.org/abs/2608.22522
作者: Adrian Nyakairu,Hongfu Liu
类目: Machine Learning (cs.LG)
*备注:
Abstract:Data valuation is a cornerstone of data-centric learning, where prior efforts primarily focus on designing algorithms to classify training samples as either beneficial or detrimental for the learning task. However, leveraging these valuation estimates for subsequent data intervention remains underexplored; conventional approaches typically discard or downweight harmful samples, thereby underutilizing available data resources. In this paper, we present Dynamic Influence-based Valuation and Editing (DIVE), a novel and efficient framework that dynamically estimates sample values at the batch level and transforms detrimental data into beneficial contributions. Rather than altering the raw data, DIVE operates at the optimization level by strategically reversing the gradient directions of harmful samples during training, ensuring seamless integration with standard learning procedures with minimal overhead. Extensive empirical evaluations demonstrate that DIVE consistently improves classification performance, maximizes data efficiency, stabilizes optimization, and effectively generalizes to large language model fine-tuning.
[LG-50] From Symmetry to Invariance: Learning Galois Equivalent Representations in Finite Fields
链接: https://arxiv.org/abs/2608.22513
作者: Zheng Zhang,Na Zhang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Neural networks can learn algebraic operations from finite examples, but it remains unclear whether this ability transfers across mathematically equivalent representations of the same operation. We study this question through multiplication in finite fields under changes of basis. The Galois action organizes basis representations into orbits, and bases in the same orbit induce the same coordinate multiplication map. This structure allows us to separate learning multiplication from transferring it to basis representations that are not used for training. We examine several ways of providing or recovering the relevant orbit structure, including invariant labels, basis matrices, orbit recognition, and algebraic decomposition. Our main approach trains a model to predict the Galois action between basis representations. Repeated applications of the learned transformation are then used to construct a canonical representative for each orbit, which supports multiplication on held-out bases through exact canonical matching. This provides a concrete mechanism for converting a learned algebraic symmetry into an invariant representation that can be used for transfer.
[LG-51] Quantum-Inspired Hybrid Neural Networks for Neural Decoding: A Controlled Ablation Study of Learnable Quantum Sidecar Integration ECAI2026 IJCAI
链接: https://arxiv.org/abs/2608.22475
作者: Diana Legziel Levy,Menachem Finkelstein,Peter Chin,Eilon Vaadia,Sarel Cohen
类目: Machine Learning (cs.LG); Quantum Physics (quant-ph)
*备注: Accepted at the 5th International Workshop on Human Brain and Artificial Intelligence (HBAI 2026), IJCAI-ECAI 2026
Abstract:We study parameterized quantum circuits (PQCs) integrated as residual sidecar modules within a ResNet-50 backbone for 31-class neural population decoding—imagined handwriting classification from multi-neuron spike rasters. Under strictly controlled conditions (fixed data splits, seeds, and optimizer), we compare four model variants: baseline, quantum sidecar with frozen input projection, quantum sidecar with backbone-gradient-trained projection, and a measurement-guided variant that aligns angle encodings with circuit measurement outcomes. The backbone-gradient variant improves accuracy in 3/4 seeds (+0.19% mean, 95% CI [-1.10%, +1.48%]) and consistently reduces Linear CKA similarity to baseline features ( \Delta=-0.025 , 4/4 seeds), indicating genuine structural reorganization of representations. A nine-variant ablation identifies simple shallow architectures as the most effective and reproducible configuration. Measurement-guided training consistently improves representation geometry without reducing accuracy. All results use noiseless statevector simulation on 4 qubits, a regime chosen to reflect the practical constraints of current near-term superconducting hardware; no quantum computational advantage over classical methods is claimed.
[LG-52] he Variance of Thought: Policy Variance Critical Forks and Local Credit Assignment
链接: https://arxiv.org/abs/2608.22467
作者: Yingru Li
类目: Machine Learning (cs.LG)
*备注:
Abstract:Long-horizon language-model tasks — multi-step reasoning and tool-using agents alike — are limited by credit assignment. We analyze it through the policy variance \sigma_\pi^2(s)=\operatornameVar_a\sim\pi[Q_\pi(s,a)] , which in a deterministic MDP is the sole source of return variance and is injected in discrete pulses at states we call critical forks. Three results follow. (i) Policy variance is a discovery budget: observing an action of advantage c requires \Omega(c^2/\sigma_\pi^2(s)) draws, a bound that is exact on the canonical two-point fork. (ii) Policy variance is bounded by the policy’s Gini dispersion, \sigma_\pi^2(s)\le 1-|\pi(\cdot|s)|_2^2 , a rollout-free necessary condition for criticality computable from logits alone. (iii) The remaining horizon sets the estimation cost: at a fork whose downstream success probability is P , the Monte Carlo advantage estimate has signal-to-noise ratio of order \sqrtP , so its sample cost scales as 1/P — a cost that branched sampling shares. Bootstrapping removes it by converting a product of survival probabilities into a sum, provided the value representation is multiplicatively accurate, which argues for log-value parameterization.
[LG-53] MASH-Bench: Diagnosing Cross-Source Failure in Mass-Shooting Risk Classification
链接: https://arxiv.org/abs/2608.22460
作者: Neha Sharma,Ritesh Sharma
类目: Machine Learning (cs.LG); Computers and Society (cs.CY)
*备注:
Abstract:Public mass-shooting databases differ substantially in coverage, feature availability, and reporting practices, creating challenges for machine-learning models that must generalize across data sources. We introduce MASH-Bench, a harmonized benchmark of 6,968 incidents from four U.S. databases: Kaggle, Mother Jones, Stanford MSA, and the Gun Violence Archive (GVA). We evaluate cross-source risk classification using leave-one-dataset-out (LODO) evaluation. Random Forest, XGBoost, and LightGBM achieve VeryHigh-risk recall of 0.68-0.89 on the curated sources but generalize poorly to GVA, where mean recall drops to 0.20 and precision to 0.0004. To investigate the source of this degradation, we conduct a controlled feature-masking ablation that removes the five features unavailable in GVA from the curated sources. The resulting recall collapse to zero provides evidence that feature completeness is a major contributor to the observed cross-source failure. We further evaluate three domain-adaptation approaches: DANN, CORAL, and importance weighting. DANN improves VeryHigh-risk recall on GVA by 0.282 (95% CI [0.11, 0.47], p = 0.003), although precision remains low, whereas CORAL and importance weighting yield zero recall. Oracle prior-shift recalibration likewise fails to recover VeryHigh-risk predictions, indicating that label-side correction alone is insufficient under the observed feature deficiencies. A per-group audit further identifies substantial disparities associated with media-attributed mental-health labels. Overall, these results indicate that, in MASH-Bench, cross-source generalization is constrained more by feature completeness and label prevalence than by classifier choice. The benchmark provides a controlled setting for diagnosing these effects in cross-source risk classification.
[LG-54] KPI-Conditioned Generative Design of Automotive Hood Inner Panels: A Two-Stage Retrieval-Generation Pipeline with Surrogate-Based Performance Estimation
链接: https://arxiv.org/abs/2608.22457
作者: Sudeep Chavare
类目: Machine Learning (cs.LG)
*备注: 20 pages, 8 figures, 9 tables
Abstract:An inner hood panel must meet a deflection target, stay below a stress limit, and hit a mass target. Machine-learned surrogates have made the forward direction, geometry to performance, fast and routine. The inverse direction, producing geometry from a stated requirement, remains largely unaddressed for industrial parts whose design space is organized into discrete topology families rather than a continuous parameterization. This work presents a two-stage pipeline for that inverse problem. A reachability stage determines which topology families can satisfy a given requirement vector. A conditional variational autoencoder then generates point-cloud geometry within a selected family, and a neural-operator surrogate estimates the performance of each candidate. The pipeline is built entirely from public data and freely available compute, and is deployed as an interactive tool. The pipeline works, with qualifications that are reported as primary findings rather than caveats. The surrogate is accurate in aggregate, but its error is comparable to the performance differences it is asked to discriminate, which bounds what can be claimed for any individual generated design. That ratio of surrogate error to within-class signal is argued to be the quantity that determines whether a pipeline of this kind can work at all.
[LG-55] Geometric Structures on Graphs: a Holonomy-Based Discretization of Curvature
链接: https://arxiv.org/abs/2608.22453
作者: Hao Li,Yuhan Peng,Junwen Dong
类目: Machine Learning (cs.LG); Differential Geometry (math.DG)
*备注:
Abstract:We propose a holonomy-based framework for discretizing curvature on graphs equipped with local symmetric positive-definite metrics. Each vertex carries a fibre metric (g_i), and each directed edge carries a reversible metric-compatible transport (F_ij). The ordered product around an oriented triangular loop (\mathcal C) gives a holonomy (H_\mathcal C), whose normalized logarithm (\Omega_\mathcal C=-s_\mathcal C^-1\operatornameLog(H_\mathcal C)) is used as a finite-loop curvature observation. Thus the construction discretizes the geometric principle that infinitesimal holonomy is controlled by curvature, rather than treating holonomy as a heuristic feature. Since (\Omega_\mathcal C) lies in the (g_i)-orthogonal Lie algebra, it is not itself a velocity of an SPD metric. We therefore introduce two aggregation mechanisms: a commutator with a symmetric response matrix, producing symmetric Ricci-type metric responses, and an incidence-aware covariant divergence of curvature-induced edge fluxes, reflecting the relation between trace and covariant divergence. The resulting responses are locally orthogonal-gauge equivariant and can drive exponential updates that preserve positive definiteness. We also give a reversible metric-compatible parametrization of edge transports, allowing orthogonal edge factors, loop scales, weights, and response matrices to be learned while respecting the graph geometry. Known-geometry calibrations on the unit sphere test the holonomy–curvature relation, curvature preservation under nontrivial local metric representations, and the empirical recovery of edge transports from local observations.
[LG-56] Dual-Scale State-Space Modeling with Speaker-Wise Dynamic CRF for Speech Emotion Recognition in Conversation ICASSP2027
链接: https://arxiv.org/abs/2608.22399
作者: Guan-Hua Wen,Kuan-Yu Chen,Hou-Chiang Tseng
类目: Machine Learning (cs.LG)
*备注: 5 pages, 2 figures, 4 tables. Submitted to ICASSP 2027
Abstract:Conversational speech emotion recognition must reconcile acoustic evidence across temporal scales with two interaction processes: cross-speaker contextual influence and within-speaker emotion evolution. We propose DSSM-CRF, an audio-only architecture that explicitly separates these processes. Bidirectional state-space models encode fused self-supervised speech representations at frame and dialogue scales, so each utterance representation captures local prosody and context from all speakers. The decoder then orders each speaker’s utterances into an independent dynamic conditional random field chain. Consecutive utterances in a speaker’s chain form a transition pair whose score combines a corpus-level transition matrix with a residual predicted from the two contextualized utterances. An auxiliary objective supervises whether each pair changes emotion but does not participate in Viterbi inference. Thus, interlocutor turns affect contextual emotion scores without being treated as transitions in another speaker’s emotion trajectory. DSSM-CRF achieves 75.81% UA and 74.90% WA on IEMOCAP, and 54.72% WA and 49.31% WF1 on MELD. Matched controls demonstrate complementary gains from speaker-wise factorization and CRF modeling.
[LG-57] Precision-Aware Variable Bit Processing Elements for Hardware-Efficient Systolic Array Designs
链接: https://arxiv.org/abs/2608.22378
作者: Dantu Nandini Devi,Madhav Rao
类目: Hardware Architecture (cs.AR); Emerging Technologies (cs.ET); Machine Learning (cs.LG); Image and Video Processing (eess.IV)
*备注:
Abstract:Systolic arrays (SAs) have emerged as prominent hardware accelerators for matrix operations in deep learning, while floating point number formats enable precision control across computational domains. This research investigates approximate computing techniques for floating point (FP) multipliers in Weight Stationary Systolic Arrays, focusing on IEEE 754 (FP32), TensorFloat-32 (TF32), and Brain Floating point (BF16) formats. By integrating partial product matrix (PPM) column truncation with positive and negative compressors in the FP multiplier architecture, we optimize the trade-off between computational efficiency and accuracy. NSGA-II optimization algorithm was employed to explore the vast design space for evolving FP multiplier designs, towards achieving substantial hardware improvements while maintaining acceptable output quality. Substantial hardware benefits were observed in the FP multiplier designs across various applications, while preserving output quality. The FP approximated Processing Elements designed in the SA was found to offer comparable CNN accuracy for models trained on MNIST, F-MNIST, and CIFAR-10 dataset. The FP approximated SA designs that fall in the top 10 CNN performance offered substantial hardware gains in the range of 66% to 92% footprint savings, 60% to 93% of power benefits with 21% to 54% improvement in the delay when compared with the corresponding exact implementations mentioned in the literature for running the model trained on CIFAR-10 dataset. The TF32 and BF16 approximated SA designs also achieved substantial gains while maintaining comparable CNN accuracy. Our findings confirm that targeted approximation in FP multiplier design significantly improves the efficiency of hardware accelerators for error-tolerant applications, establishing an effective approach to hardware resource optimization in contemporary computing architectures.
[LG-58] racing the Unlabeled Storm: Cross-Variable Transfer in a Lagrangian Atmospheric JEPA Framework
链接: https://arxiv.org/abs/2608.22358
作者: K M Anirudh,S Sandeep,Hariprasad Kodamana
类目: Machine Learning (cs.LG); Geophysics (physics.geo-ph)
*备注: 8 pages, 3 figures, plus supplementary material
Abstract:Deep atmospheric convection governs South Asian monsoon variability, yet attempting to learn its latent world model directly from zero-inflated, heavy-tailed precipitation yields suboptimal predictive representations. Continuous atmospheric proxies, such as outgoing longwave radiation (OLR), express this convective organization far more coherently. We address this mismatch with \emphcross-variable proxy learning: M-JEPA, a multiscale Monsoon Joint-Embedding Predictive Architecture, is pretrained on five continuous proxy fields over Lagrangian patches tracking moving convective systems—without rainfall supervision at any point. The resulting frozen representation is transferred to daily precipitation forecasts through a shared decoder trunk featuring parallel probabilistic and deterministic branches. Because rainfall is strictly unobserved during pretraining, downstream skill directly measures the predictive information captured in the latent rollout. A frozen-backbone probing framework with two controls (an identical architecture trained on rainfall alone, and a randomly initialized backbone) attributes the transfer specifically to proxy pretraining: direct rainfall training exhibits 36% higher CRPS error ( 7.52 vs.\ 5.54 ,mm/day). Against the 51-member operational ECMWF ensemble, the transferred model attains a statistically resolved CRPS advantage ( 6.81 vs.\ 6.89 ,mm/day) and higher Brier skill ( +0.05 vs.\ -0.04 ) using 15.4 M parameters on a single consumer GPU, concentrated at heavy-rain thresholds and fine spatial scales, while the ensemble retains an advantage in neighborhood skill and deterministic references on point metrics. The result provides a competitive monsoon precipitation forecast grounded in intraseasonal dynamics and a diagnostic framework for evaluating transferred atmospheric representations.
[LG-59] Gaussian process learning with flow map refinement for parameter estimation in dynamical systems
链接: https://arxiv.org/abs/2608.22324
作者: Yue Hao,Dongwei Ye
类目: Machine Learning (cs.LG); Computational Engineering, Finance, and Science (cs.CE); Numerical Analysis (math.NA)
*备注:
Abstract:Parameter estimation is a central task in data-driven learning of dynamical systems. It aims to recover the underlying physical parameters from observed time-series data, thereby providing interpretable insights into the physical mechanisms governing the system. Gradient/derivative matching methods based on Gaussian process provide an efficient way to perform parameter estimation. Those methods avoid repeated numerical integration and enforce local derivative consistency. However, such local matching may result in global inconsistency with the governing flow map, particularly under scarce and noisy observations. To address this limitation, we propose a framework based on Gaussian process learning with flow map refinement (GPL-FMR), a two-stage parameter estimation framework. The first stage is based on Gaussian process learning algorithm and the posterior obtained from which is transferred as an informative prior to the second stage based on flow-map refinement. The second stage further improves the parameter estimation via optimisation based on global dynamical constraints. We demonstrate and analyse its performance on multiple numerical examples, including the Van der Pol oscillator, the Lotka-Volterra model, and the Lorenz-63 system. The results show that the proposed framework consistently improves parameter estimation accuracy, particularly under scarce and noisy observations.
[LG-60] Beyond Dense Adam States: Adaptive Log-Space Quantization for Memory-Efficient Optimizers
链接: https://arxiv.org/abs/2608.22322
作者: Yan Wang
类目: Machine Learning (cs.LG)
*备注: 17 pages, 5 figures, 7 tables. Code: this https URL
Abstract:Low-precision optimizer-state methods are commonly designed for dense Adam-style moments, but memory-efficient optimizers maintain factored, confidence-based, or projected states whose quantization errors propagate differently. We characterize this heterogeneity in optimizer-state traces from language-model pre-training and introduce Adaptive Log-Space (AL) quantization, a block-wise representation for non-negative states that adapts its nonzero range per block while preserving exact zeros. AL8 and AL16 are combined with independent signed-momentum encodings and state-specific precision choices. Across 96 runs totaling 214.7 GPU-hours, we evaluate AdamW, Adafactor, CAME, and APOLLO paths. On a 20K-step TinyLlama-1.1B benchmark, AdamW with AL8 second moments and 8-bit uniform momentum reaches 72.90 perplexity, versus 72.48 for FP32 and 73.54 for an 8-bit dynamic-quantization baseline, while reducing measured optimizer-state storage from 8392.7 to 2119.2 MiB. CAME requires higher precision for its non-negative states: AL16 reaches 86.16 perplexity versus 86.68 for FP32, while all-AL8 reaches 90.19. In a 100K-step GPT-2 experiment, topology-aware parameter protection reduces the late-loss gap of quantized Adafactor from +0.1185 to +0.0159. These results support state- and topology-aware optimizer quantization. End-to-end comparisons use a single training seed and are reported as empirical measurements. Comments: 17 pages, 5 figures, 7 tables. Code: this https URL Subjects: Machine Learning (cs.LG) Cite as: arXiv:2608.22322 [cs.LG] (or arXiv:2608.22322v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2608.22322 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-61] StocBench: A Benchmark for Generative Modeling of Stochastic Dynamics STOC
链接: https://arxiv.org/abs/2608.22309
作者: Sebastian Pfister,Benjamin Holzschuh,Nils Thuerey
类目: Machine Learning (cs.LG)
*备注: Code available at this https URL
Abstract:We benchmark transport-based generative models as well as distillation-based few-step methods for the probabilistic forecasting of stochastic fluid flows, with a particular focus on performance under limited inference budgets. All methods are evaluated on a two-dimensional Kolmogorov flow with stochastic forcing. We measure one-step distributional accuracy against large simulated reference ensembles and assess whether the invariant measure is preserved during autoregressive rollouts via the enstrophy spectrum. On the stochastic task, flow matching achieves the most accurate one-step conditional distribution at high inference budgets, while the second-order exponential integrator DPM-2 is strongest at very low NFE. Few-step distillation methods are competitive with the multi-step methods and preserve the enstrophy spectrum particularly well. A deterministic control task, in which the forcing over the prediction interval is observed, separates aleatoric from epistemic uncertainty. Model performance does not translate between the two settings: the distilled models are competitive on the stochastic task but least accurate on the control task. While stochastic diffusion samplers such as DDPM better preserve the enstrophy spectrum during rollouts in the stochastic setting, deterministic samplers such as DDIM and DPM-2 show better spectral preservation in the deterministic setting.
[LG-62] DAW: Dynamics-Aware Weighting for Deep Learning Forecasts of Chaotic Systems
链接: https://arxiv.org/abs/2608.22277
作者: Zhou Fang,Gianmarco Mengaldo
类目: Machine Learning (cs.LG); Computational Physics (physics.comp-ph)
*备注:
Abstract:Deep learning surrogates for forecasting chaotic dynamical systems suffer from catastrophic error accumulation over long-term autoregressive rollouts. This behavior is partly tied to the underlying systems: chaotic spatiotemporal systems, such as the Kuramoto-Sivashinsky (KS) equation, visit phase space unevenly - dominated by recurrent, low-dimensional quiescent states (e.g., near-laminar flows) and punctuated by rare, dynamically complex topological transitions (e.g., wave-merging events). Under a sample-wise uniform objective, standard neural surrogates allocate their finite capacity to the statistically numerous quiescent states, under-representing the transient regimes that trigger disproportionate, localized errors. Existing imbalanced-regression methods reweight samples by target-space density. However, statistical target-space rarity need not coincide with the intrinsic dynamical rarity - the recurrence geometry of the attractor that is the source of the imbalance. To address this, we introduce Dynamics-Aware Weighting (DAW), a data-centric objective reweighting framework. Using the local dimension d from dynamical systems theory as an a priori measure of a state’s active degrees of freedom, DAW reshapes the loss landscape to allocate representational capacity toward the sparse, high- d regimes where forecast errors are systematically large. On the chaotic KS equation, DAW consistently outperforms uniform training, purely statistical density weighting, and its randomly permuted ablation, reducing long-term autoregressive error relative to all baselines. Event-level analysis shows that DAW achieves this by suppressing the localized error amplifications incurred during sharp jumps in d , which accompany complex physical processes such as wave-merging in the KS system.
[LG-63] oward a First-Principles Update Geometry for the Language-Model Head
链接: https://arxiv.org/abs/2608.22253
作者: Aditya Somasundaram
类目: Machine Learning (cs.LG)
*备注: 7 pages
Abstract:We study the language-model head and softmax as a single module, deriving an update geometry from their composition rather than from the weight matrix in isolation. Under Hilbert’s projective distance, the maximum change caused by an update S over \left|\left|h\right|\right|_2\le H is H\max_ij\left|\left|s_i-s_j\right|\right|_2 , which is H times the Euclidean diameter of its token rows. Motivated by Muon’s singular-value conditioning, we propose maximizing the smallest row separation while constraining this diameter, producing an approximate-equidistance problem when V\gg d .
[LG-64] MRMAD: A Multi-Round Multi-Audio Benchmark for Evaluating Acoustic Degradation Perception in Large Audio-Language Models EMNLP2026
链接: https://arxiv.org/abs/2608.22236
作者: Yize Li,Ningyuan Yang,Sile Yin,Sindhuja Thogarrati,Sung-En Chang,Andrew C. Singer,Xue Lin,Chuan-Che Huang,Shuo Zhang
类目: ound (cs.SD); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS)
*备注: EMNLP 2026
Abstract:Large audio-language models (LALMs) have shown promising progress in understanding speech, music, and general sound events, yet their ability to reason about how audio signals are degraded remains underexplored. Existing benchmarks primarily evaluate semantic understanding, event recognition, or high-level audio reasoning, leaving a basic question unanswered: Do LALMs understand the differences in audio quality? We introduce MRMAD, a Multi-Round Multi-Audio Degradation benchmark for evaluating audio degradation perception and understanding in LALMs. MRMAD spans speech, music, and sound, and frames evaluation as multi-turn dialogues over multiple audio inputs, requiring models to identify degradation types, compare severity, and perceive corruption changes across turns. Unlike current single-turn audio-language benchmarks, MRMAD evaluates whether LALMs can maintain consistent degradation hypotheses with new evidence and explain low-level acoustic phenomena in natural language. Through a systematic evaluation of 18 representative LALMs from non-thinking to reasoning and Omni models, we find that current models often recognize coarse content while failing to diagnose, compare, or reason about degradations reliably. MRMAD reveals an important yet overlooked aspect of audio-language understanding and provides a diagnostic foundation for building future LALMs that are robust to real-world acoustic conditions.
[LG-65] A Query-Time Framework for Transient 2D Pore-Scale Flow Prediction and Generative Design
链接: https://arxiv.org/abs/2608.22235
作者: Yiming Wang,Jiale Zhu,Zhichen Ye,Yandong Lv,Shiqi Wang,Jinlong Liu,Yucheng Fan
类目: Machine Learning (cs.LG)
*备注: 34 pages, 16 figures, 4 tables
Abstract:Pore-scale flow governs transport and permeability behaviour in porous media engineering applications, yet repeated lattice Boltzmann method (LBM) simulation across many geometries and design queries remains costly for repeated deployment. This study formulates transient pore-scale flow prediction as a geometry-conditioned query-time operator and introduces QSGS-Transient-7606, a benchmark of 7,606 two-dimensional porous structures each paired with 30 logarithmically sampled LBM states. The proposed continuous-time pore-scale flow surrogate model (CT-PoreFlow) integrates topology-aware geometry encoding, compressed spectral mixing, and log-time conditioning with a late-time flux-calibration objective. On unseen test geometries, CT-PoreFlow achieves a velocity relative L2 of 0.2248 and a terminal permeability error of 12.81%. Frozen morphology and computed tomography image audits confirm reasonable cross-geometry robustness without fine-tuning. The surrogate is then embedded in an inverse design workflow, screening 9,216 generative adversarial network and diffusion candidates across 18 property targets prior to LBM verification. Guided GAN sampling attains 98.11% through-connectivity and 72.28% conditional design success, exceeding diffusion-based generation. The framework unifies transient flow prediction, transport-aware screening, and LBM-verified inverse design for porous media.
[LG-66] Risk-Sensitive Reinforcement Learning with Smoothed Quantile Objectives
链接: https://arxiv.org/abs/2608.22227
作者: Mohammad Alipour-Vaezi,Huaiyang Zhong,Sajad Khodadadian
类目: Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注:
Abstract:Reinforcement Learning (RL) has achieved tremendous success in recent years. However, the classical foundations of RL do not account for the risk sensitivity of the objective function, which is critical in various fields, including healthcare, finance, etc. A popular approach to incorporate risk sensitivity is to optimize a specific quantile of the cumulative reward distribution. However, exact quantile objectives are non-smooth and can change abruptly under small perturbations of the return distribution, making them difficult to optimize reliably when the transition model must be learned from data. Motivated by this instability, we develop UCB-BQRL, a model-based optimistic learning algorithm that maintains confidence sets for the transition kernel and plans using a lower-buffered quantile criterion. The buffered criterion smooths the exact quantile objective by averaging nearby lower quantiles, thereby improving stability under transition-estimation error. To compute the buffered-quantile policy at each episode, we introduce EVI-BQ, an exact dynamic-programming procedure. We establish a high-probability regret bound for UCB-BQRL, which up to logarithmic factors scales as \mathcalO(\mathrme^\tau/\rho_\tau+H^2\sqrtSAT) , where \rho_\tau is denoted as the root-level left-plateau threshold, which is a problem-dependent constant. Further, we establish an information-theoretic lower bound of \Omega(H/\rho_\tau\sqrtAT) for the regret of any algorithm dealing with a quantile objective function. Finally, we prove that the exact point-quantile evaluation and exact lower-buffered quantile evaluation are PP-hard under polynomial-time Turing reductions, even for a fixed policy in a two-state, one-action finite-horizon MDP.
[LG-67] Counterfactual Evaluation of Temporal Observation Protocols
链接: https://arxiv.org/abs/2608.22221
作者: Xizhe Zhang
类目: Machine Learning (cs.LG)
*备注:
Abstract:We study counterfactual protocol evaluation: whether data collected under a realised observation protocol determine the predictive value of alternatives that were never deployed. Protocol value is the population R^2 of the Bayes-optimal predictor of a fixed trajectory-level target from the measurements an alternative would collect. We show that even infinite benchmark data need not determine this value: distinct latent covariance structures can induce the same benchmark measurement–target law while assigning different values to the same alternative. We develop a value-specific identification theory in which only latent ambiguity that changes the alternative’s value matters. For linear targets, invisible covariance directions certify non-identification, while targeted measurements can restore identification without recovering the full latent covariance; an exact permutation construction extends the result to nonlinear aggregate targets. With finite dense calibration data, uniform error bounds control protocol-selection regret and distinguishable value gaps. Exact marginal gains then support cost-constrained, target-aware observation design. Simulations and retrospective analyses of Sleep-EDF and Long-Term AF show that broad temporal-layout differences can be more reliably distinguished than fine placements selected from finite data. Together, these results connect identification, calibration resolution and observation design for undeployed protocols.
[LG-68] On the Capability Separation Between World-Model Policy Learning and Imitated World-Action Models
链接: https://arxiv.org/abs/2608.22197
作者: Yang Yu
类目: Machine Learning (cs.LG)
*备注:
Abstract:World-action models predict a future outcome and then infer an associated action. Although this factorization can improve representation learning and data efficiency, it is unclear whether it provides stronger control capability than direct behavior cloning when both are trained from the same observational demonstrations. We compare a direct behavior-cloning policy, an imitation-trained world-action policy, and a policy optimized with an action-conditioned world model. At the controller-class level, every world-action policy can be flattened into a direct stochastic policy with the same closed-loop trajectory distribution. At the population level, under realizability, exact optimization, common deployment information, and distribution-preserving deployment, direct behavior cloning and world-action imitation both recover the observational behavior policy. Thus, future prediction changes the learning factorization but not the unrestricted external policy class or ideal imitation target. Action-conditioned world-model learning differs by predicting outcomes under specified actions and comparing them through a control objective. We characterize the irreducible action-specific prediction error of future models that do not condition on the candidate action, identify conditions under which a world-action joint can recover an interventional forward model, and show that observational demonstrations do not identify action effects in general. Finally, we construct an environment family in which every observational learner has positive worst-case regret, whereas one informative intervention permits zero regret. The key distinction is therefore between predicting futures associated with observed behavior and predicting consequences of specified actions for policy optimization. Subjects: Machine Learning (cs.LG) Cite as: arXiv:2608.22197 [cs.LG] (or arXiv:2608.22197v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2608.22197 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-69] More accurate behavioral predictions with hybrid Bayesian-connectionist models
链接: https://arxiv.org/abs/2608.22154
作者: Brenden M. Lake,Akshay K. Jagadish,Guangyuan Jiang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Researchers must often choose between Bayesian or neural network models of behavior, two paradigms with complementary strengths and weaknesses. An ideal paradigm would facilitate testing many kinds of representations and inductive biases; Bayesian models make this easy, while neural networks do not. Similarly, an ideal paradigm would avoid over-simplifications; neural networks make this easy, while Bayesian models do not. Here, we introduce Bayesian distillation with Behavioral Tuning (BBT) as an approach to getting the best of both traditions. BBT offers a simple recipe for model building: first, a neural network is trained to mimic a Bayesian model through synthetic data, and second, the network is fine-tuned on human behavior to capture additional structure and nuance. Across four case studies in human concept learning, we find that BBT outperforms traditional approaches at predicting human behavior while also revealing psychological insights, resulting in models that can both mimic Bayesian priors and capture heuristics and biases that violate simple modeling assumptions.
[LG-70] Loss Landscape Features That Make Adam Stall: Definitions Estimators and the Preconditioned Hessian View
链接: https://arxiv.org/abs/2608.22145
作者: Rodion Podorozhny
类目: Machine Learning (cs.LG); Numerical Analysis (math.NA)
*备注:
Abstract:Across implicit-neural-representation (INR) architectures and analytic benchmarks we observe that a thoroughly tuned Adam (especially its learning rate (lr), e.g. in a hyperparameter sweep from lr = 0.05 to 10^-8 ) can potentially reach a very low loss even on ill-conditioned loss landscape or converge at a plateau far above the loss attained by second-order methods. This report defines the measured metrics that help determine if Adam can mitigate the ill-conditioning on a given loss landscape. We provide the indicators by which each outcome is determined, that are: the condition number of the Hessian and of the Adam-preconditioned Hessian D^-1/2HD^-1/2 (with the derivation from Adam’s update rule), the diagonal mass \rho that distinguishes axis-aligned from cross-coupled ill-conditioning, the negative spectral mass estimated by stochastic Lanczos quadrature, and the gradient energy fractions over curvature bands, including the flat fraction that indicates the Adam stall. A worked out 2\times 2 example and an illustration show the reasons why a diagonal preconditioning by Adam can remove axis-aligned ill-conditioning by rescaling and why it cannot do the same if the ill-conditioning is cross coupled. In addition, we present a case study of FINER image fitting architecture that goes over the whole loss landscape analysis framework: the fitting architecture description, reasons due to which its landscape stalls Adam at saddles, the measured PSNR values through our tuned baselines to the 120 – 134 ,dB results of the blockwise second order methods, the error maps behind those numbers, and description of the benefits such image fitting accuracy gives in practice.
[LG-71] Learning Reduced-Order Dynamics with Singularity via Latent-Augmented Neural Ordinary Differential Equations
链接: https://arxiv.org/abs/2608.22142
作者: Xiaorui Wang,Yu Zhou,Wenjie Mei,Dongzhe Zheng,Yang Bai,Masaaki Nagahara
类目: Machine Learning (cs.LG)
*备注:
Abstract:This paper addresses the issue of self-intersecting trajectories (in phase space) in industrial reduced-order modeling and proposes the Latent-Augmented Neural Ordinary Differential Equations (LA-NODEs) framework. From the perspective of artificial intelligence, the proposed method augments conventional neural ordinary differential equations to enhance model expressiveness, enabling the representation of conflicting vector fields that may arise in reduced-order systems, thereby improving learning accuracy. Through theoretical analysis, the underlying mechanism of the framework is established, and a condition for determining the minimum required augmentation dimension is derived. From the perspective of engineering applications, the effectiveness of the proposed method is validated on the reduced-order system of two representative industrial models, namely an interior permanent magnet synchronous motor (IPMSM) drive and a distributed energy system (DES). Experimental results demonstrate that the proposed method can recover system features that are difficult to capture using conventional approaches and achieve superior performance in terms of prediction accuracy and modeling fidelity, thereby providing an effective approach for high-precision data-driven modeling of complex industrial systems.
[LG-72] Blockwise Stabilized Adaptive Cubic Regularization with Subsolvers via Recurrence
链接: https://arxiv.org/abs/2608.22129
作者: Rodion Podorozhny
类目: Machine Learning (cs.LG); Numerical Analysis (math.NA)
*备注:
Abstract:Cubic-regularized Newton methods have the optimal \mathcalO(\varepsilon^-3/2) global rate and an automatic saddle-escape mechanism, but their subproblem is most often solved by a full eigendecomposition, limiting feasible model size. We introduce a blockwise optimizer that partitions parameters by tensor, minimizes an independent cubic model with an adaptive cubic constant M_b per block, and accepts or rejects each block step against a monotone guard on the full loss. The subproblem solver is chosen by block size: small blocks use lazy exact cubic steps from explicitly formed per-block Hessians; arbitrarily large tensors use a matrix-free Chebyshev-bounded Krylov subspace built by the Lanczos process. The cubic shift bounds the required polynomial degree whenever the gradient-driven shift dominates negative curvature, renders the shifted operator positive semidefinite before any polynomial is applied, and preserves the \mathcalO(\varepsilon^-3/2) rate under inexact subproblem solves. We prove these claims, and the blockwise scheme carries a monotone per-block descent guarantee. Experiments cover FINER INRs (about 199k parameters) and a 91.4M-parameter ViSIR INR, where the blockwise cubic step remains exact in the cubic-model sense on every block, including the 88.5M-parameter decoder tensor (97% of the model). Run to full convergence on FINER, the ARC- \varphi_1 optimizer reaches 133.5 dB PSNR while tuned Adam plateaus at 78.2 dB at the same extended budget; in the roughly 70 minutes Adam takes to reach its peak, ARC- \varphi_1 reaches 95.6 dB. A companion report isolates the loss-landscape features responsible for Adam’s behavior.
[LG-73] Who Should Teach? Confidence-Aware Dual-Teacher Learning for Few-Shot Node Classification on Text-Attributed Graphs
链接: https://arxiv.org/abs/2608.22127
作者: Hojin Kim,Sujin Yoon,Sungsu Lim,Dongwon Lee,David Yoon Suk Kang
类目: Machine Learning (cs.LG); Social and Information Networks (cs.SI)
*备注:
Abstract:Text-Attributed Graphs (TAGs) integrate graph structures and node-associated textual attributes, and recent studies have increasingly leveraged Large Language Models (LLMs) to improve TAG learning in few-shot settings. However, existing approaches typically utilize LLM-derived information uniformly across all nodes, despite substantial variations in its reliability, while also incurring considerable monetary costs. We argue that the most appropriate source of supervision may differ across nodes, as Graph Neural Networks (GNNs) and LLMs exhibit complementary strengths in exploiting structural and semantic information, respectively. To this end, we propose CoTeach, a Confidence-aware dual-teacher learning framework that dynamically selects the more reliable teacher for each node. Experimental results demonstrate that CoTeach consistently improves few-shot node classification performance while reducing unnecessary LLM utilization and associated monetary costs.
[LG-74] he Price of Decentralization in Top-K Arm Identification
链接: https://arxiv.org/abs/2608.22120
作者: Larissa Xu,Jasmine Nguyen,William Chang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Cooperative teams often need to agree on the best few options rather than simply accumulate reward, and they must do so while each member sees only a fragment of the team’s collective experience. We study this as top- K joint-arm identification in multi-agent multi-armed bandits: at every round M agents simultaneously choose individual actions that compose a joint arm, and the team must ultimately return the K joint arms of highest mean reward. The difficulty is that an agent may not observe the actions of others, their rewards, or either. We treat three observability regimes—(A) shared rewards with hidden actions, (B) observed actions with private rewards, and © full asymmetry—and design communication-free elimination algorithms (UCB-Intervals) that reconstruct implicit coordination from whatever signal each regime leaves intact: a shared arm ordering in (A), observable deviations in (B), and enlarged confidence radii under ©. We give matching analyses in both the fixed-budget and fixed-confidence objectives, then fold all three regimes into a single meta-guarantee indexed by a multiplicity c and a consensus factor \rho . Our central result is quantitative rather than merely algorithmic: change-of-measure lower bounds show that shared-reward identification is optimal up to one universal logarithmic factor, and that the entire statistical price of removing communication is a multiplicative \rho^2 in sample complexity—a fixed 4\times penalty under full asymmetry. The resulting stopping time scales as O!\left(\sum_\mathbfa \frac\log(A^M/\delta)\Delta_\mathbfa^2\right) and the fixed-budget error as \exp(-\Theta(T/H_1)) , with the dependence on the joint-action count A^M shown to be unavoidable.
[LG-75] CST: Collaborative Selective Transmission for Communication-Efficient Multimodal Edge Inference
链接: https://arxiv.org/abs/2608.22115
作者: Hai Chi,Junrui Zhang,Rui Ning,Chonggang Wang,Robert Gazda,Huanrui Yang,Hongyi Wu
类目: Machine Learning (cs.LG)
*备注: 11 pages, 6 figures, 6 tables
Abstract:Collaborative multimodal inference improves edge perception by combining observations from distributed sensing devices, but transmitting high-dimensional helper representations incurs substantial communication overhead and can lead to high end-to-end latency. Existing communication-efficient methods reduce payloads through compression, semantic coding, or feature selection, yet typically optimize compactness or task relevance without explicitly accounting for information already represented at the main device. Consequently, task-relevant but redundant helper features may still consume bandwidth. We present Collaborative Selective Transmission (CST), a main-directed query–response framework that retrieves only helper information complementary to the current main representation. Inspired by Partial Information Decomposition and the Multiview Redundancy Assumption, CST learns sample-adaptive, helper-specific sparse retrieval supports while discouraging retrieval of semantics already covered by the main device or duplicated across helpers. During inference, the main device transmits only support indices, and each helper returns the corresponding latent values, avoiding dense helper-feature exchange. Across three real-world multimodal sensing benchmarks, CST transmits no more than 14.18% of helper feature values while achieving best or near-best task performance among the evaluated methods. Experiments on a five-node NVIDIA Jetson Orin Nano testbed across 5–100 Mbps demonstrate up to a 4.27\times speedup over Transmit-All in end-to-end inference, confirming practical end-to-end latency reductions.
[LG-76] Symbolic Neural ODEs: Learning interpretable models from time-series data
链接: https://arxiv.org/abs/2608.22112
作者: Nibodh Boddupalli,Jeff Moehlis
类目: Machine Learning (cs.LG); Systems and Control (eess.SY); Dynamical Systems (math.DS); Machine Learning (stat.ML)
*备注: Submitted to the SIAM J. Dynamical Systems
Abstract:We present a machine learning framework for identifying sparse, interpretable models of dynamical systems directly from time-series data. Our approach parameterizes the underlying vector field using a neural architecture and trains it by minimizing a multi-step prediction loss over a finite horizon. To ensure numerical tractability, we optimize a mean absolute error objective averaged across prediction steps, and progressively increase the horizon during training. A key feature of this formulation is that it enforces consistency under repeated composition of the learned dynamics. As a result, the identified models exhibit significantly improved stability compared with approaches based on one-step regression of the vector field. When combined with sparsity-promoting regularization, this leads to parsimonious models that generalize beyond the training data. We demonstrate accurate recovery of systems exhibiting a wide range of behaviors, including stable and unstable fixed points, periodic orbits, and chaotic attractors. For chaotic systems, while long-term trajectory prediction is inherently limited by sensitivity to initial conditions, we show that multi-step training yields models with accurate short-term dynamics and strong agreement in long-time statistical properties, including mean, variance, and Lyapunov exponents. Moreover, we establish theoretical bounds linking trajectory error to statistical accuracy, providing a step toward a principled explanation for this behavior.
[LG-77] What actually runs: a measurement study of language model placement and decode speed on the Apple Neural Engine
链接: https://arxiv.org/abs/2608.22110
作者: Shahir M A
类目: Machine Learning (cs.LG); Hardware Architecture (cs.AR); Performance (cs.PF)
*备注: 23 pages, 5 figures. Code, measurement harnesses and run ledgers: this https URL
Abstract:We ask what gets a language model onto the Apple Neural Engine (ANE) and what makes it fast there, and we answer with three measurements. We sweep a 64-shape matrix of LLM primitives that varies how a computation is expressed while holding what it computes fixed, recording per-operation device support. We then train matched models across size and precision, with quantized checkpoints byte-identical in structure to their fp16 counterparts, so every deployment measurement is of a real trained artifact. And we read the ANE’s memory-controller byte counters during inference, establishing what actually ran rather than what the compiler intended. We support every headline claim with at least two of these three measurement paths. We find that placement is a property of how a computation is expressed, not of what it computes: a fused RMSNorm is fully ANE-eligible while its arithmetically identical decomposition is CPU-only. Weight encoding gates the accelerator: CoreML assigns a 25.85M-parameter conv-heavy fp16 model entirely to the CPU (our counters confirm zero bytes through the engine), while the same graph in int8 or 2-bit returns to ~83% residency and runs 1.8-2.2x faster, and a smaller 22.29M all-attention fp16 model sits at 98.9%. Decode cost is bytes streamed per token, at a constant ~0.77 fraction of nominal encoding width across fp16, int8 and 2-bit. The smallest and fastest models we measured are ternary, and at matched size the operator mix barely moves either axis: every resident 25M ternary model lands within 10.0-10.8 MB and 0.62-0.64 ms/token. The headline pair is half-attention ternary at 25M (10.5 MB, 0.63 ms) and 50M (16.8 MB, 0.86 ms) - 9.8x and 6.1x smaller, 3.0x and 2.2x faster than the conv-heavy fp16 design this work began with. From these measurements we draw a design procedure: choose the encoding first, then spend the byte budget on parameters.
[LG-78] Beyond Fresh Starts: Stateful Inference for Streaming ASR in Conversational Voice Agents
链接: https://arxiv.org/abs/2608.22101
作者: Sameep Chattopadhyay,Alexander Erdmann,Mari Ostendorf
类目: Machine Learning (cs.LG)
*备注:
Abstract:Modern voice-agent systems rely on streaming speech recognition models that operate under stringent latency constraints. This study shows that, due to the limited memory constraints of real-time processing, these systems are adversely impacted by conversational phenomena such as long silences and backchannels. While many agentic pipelines mitigate this by resetting state at each turn, this approach discards vital context and impairs performance at turn onsets. We propose two state-management strategies that preserve cross-utterance context to reduce onset errors. In experiments with two state-of-the-art streaming models on two spoken dialogue benchmarks, our best method yields an average of 15-21% relative WER reduction at utterance onsets.
[LG-79] Counterfactual Quotient Models: Learning What Actions Change Not What the World Does
链接: https://arxiv.org/abs/2608.22092
作者: Junlin Chen,Ruijie Wang,Jianxin Li
类目: Machine Learning (cs.LG)
*备注:
Abstract:Reinforcement-learning models commonly predict complete future states, observations, or feature occupancies, even though action selection depends only on differences between the consequences of candidate actions. As a result, these models may devote substantial statistical and representational capacity to high-dimensional phenomena that evolve independently of the agent’s current choice. We introduce the Counterfactual Quotient Model, which treats action-conditioned futures as equivalent when they differ only by a component shared across actions. Its canonical centered representation removes this common component while preserving every pairwise action comparison expressible by the modeled reward family. The implemented model learns these action-dependent effects directly from synchronized counterfactual rollouts, so shared stochastic dynamics cancel before function approximation rather than after complete futures have been predicted. We establish the decision sufficiency, identifiability, common-mode invariance, approximation behavior, and regret properties of the resulting representation. Controlled experiments in physics-based environments provide initial evidence for these properties: direct effect learning suppresses action-independent variation, supports previously unseen reward queries, and improves action ranking relative to models trained to predict absolute futures.
[LG-80] Autonomous Cyber Defense: Real-Time Attack Detection and Mitigation in Software-Defined Networks Using Machine Learning
链接: https://arxiv.org/abs/2608.22075
作者: Alexandre Amaral,Fernando Moro,Ana Malheiro
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注:
Abstract:Adversaries now move faster than manual response processes can absorb. The average eCrime breakout time, that is, the interval between initial access and the first lateral movement to another host, fell to 29 minutes in 2025, a 65% increase in speed over the previous year; the fastest observed breakout took 27 seconds, and in one intrusion data exfiltration began within four minutes of initial access. This work presents a machine learning based system that monitors network traffic in real time, diagnoses attacks, and automatically applies countermeasures in software-defined networks, so that detection and response no longer depend on human intervention. The system comprises two modules: \textitNetwork Dataset Creation (NDC), which collects IP flows, preprocesses and aggregates them to build the training dataset, and \textitIntrusion Prevention System (IPS), which automates the modeling, training, and evaluation of different algorithms and triggers blocking actions on the SDN controller. A case study with a \textitSYN flooding denial of service attack, shows the attack being detected and blocked in 21 seconds without human intervention, a response time compatible with the window imposed by current breakout times.
[LG-81] Personalized and Aspiration-Oriented Career Path Recommendation
链接: https://arxiv.org/abs/2608.22056
作者: Kuleshwar Sahu,Girish Keshav Palshikar,Rajiv Srivastava
类目: Machine Learning (cs.LG)
*备注:
Abstract:Fulfilling career aspirations is important for growth of employee and organization. We propose a data driven methodology to recommend personalized career path for a given aspirant’s career path and aspirations. The pro-posed method uses the career path similarity (CPS) between aspirant’s career and candidate career path, and ‘aspirational similarity’ (AS) between aspiration and candidate career paths to find suitable career path. CPS ensures personalized recommendation while AS ensures aspiration fulfillment. We defined two methods to compute the CPS between career paths which are (a) domain knowledge driven (DKD) and, (b) unsupervised representation learning and alignment (URLA) based, along with different AS measures. The DKD based similarity is defined in the terms of features extracted and summarized over career paths. In the URLA, we use the sequence of event names present in the career paths of the employees to learn the embedding for each event name. In URLA we use learned embedding vector of the career path event names and as-sociated event attributes (skill cluster and domain) to find the best alignment between two career paths. We hypothesized that relative position of event names in the sequence represents semantics of event name and that can be learned. We use LSTM neural network to learn the embedding vector of each career event name. We also define the matching method to compute the AS be-tween aspiration and career path in both proposed methods. We combine CPS and AS to rank available ‘candidate career paths’ of employees to find the suitable one. We get better DCG value in URLA as compare to DKD. We also showed that ranking are coherent using both the methods. URLA method is better since it does not require domain knowledge to model the similarity and includes temporal aspect by optimal Levenshtein alignment using weighted cosine distance.
[LG-82] ReMAP: Self-supervised learning to unveil brain representations and vulnerability
链接: https://arxiv.org/abs/2608.22042
作者: Jade Perdereau,Virginie Loison,Kanssa El Ayeb,Louis Gervais,Melvin Berto Strouc,Fabrice Vallée,Thomas Moreau,Jérôme Cartailler
类目: Machine Learning (cs.LG)
*备注:
Abstract:General anesthesia offers a rare opportunity to observe the human brain under a standardized, controlled perturbation. Yet intraoperative electroencephalography (EEG) is almost always reduced to a single proprietary depth index, collapsing a rich trajectory into one number and discarding how a brain moves between states. Here we ask whether the geometry of that trajectory, not merely the depth it reaches, carries clinically meaningful information. Using similarity-based self-supervised learning on raw, two-electrode frontal EEG, with no labels, we place each recording within a low-dimensional space in which anesthetic depth becomes one readable axis while the shape of a patient’s path encodes additional structure. We validate the representation across two cohorts and two acquisition systems totaling more than 1,000 patients. Depth of anesthesia is predicted accurately (BIS mean absolute error = 3.2, R2 = 0.82), and in the sparse-montage setting our compact ( 68k parameter) model remains competitive with EEG foundation models orders of magnitude larger (4M-157M parameters), indicating that matching the representation to the recording dominates raw scale. The learned space organizes age along its own gradient, independent from depth, without supervision. The same space also aligns with interpretable anesthetic signatures like frontal alpha, slow-delta, and burst suppression, linking this data-driven representation to established neurophysiology. On an independent cohort with longitudinal follow-up, the geometry of the early trajectory separates 30- month cognitive and mortality outcomes complementary to age (AUROC 0.86). These results suggest that the path a brain traces through anesthesia is a label-efficient correlate of latent vulnerability, motivating prospective validation.
[LG-83] Beyond Fixed Directions: Adaptive Representation Analysis of Reasoning and Memorization in LLM s
链接: https://arxiv.org/abs/2608.21919
作者: Shaheen Nabi
类目: Machine Learning (cs.LG)
*备注: Preprint. Code and experimental resources are available on GitHub
Abstract:Recent work has proposed that reasoning and memorization in language models can be characterized by a single representation direction, including methods that keep this direction fixed during reinforcement learning. We test two assumptions behind this view. First, are reasoning-oriented and factual-recall task groups approximately single-direction separable? Second, does the resulting geometry remain stable after GRPO? Using Qwen3-0.6B and a controlled 400-example dataset, we find that a one-dimensional projection can match a full 1024-dimensional linear probe with AUROC = 1.00 on the studied task groups. However, after GRPO, the corresponding direction is substantially reorganized: mean-direction cosine averages 0.453, probe-direction cosine 0.445, while direct representation drift reaches 0.511 at the final layer. Probe AUROC nevertheless remains 1.00. The evidence therefore supports single-direction decodability for the studied task groups but challenges fixed-direction stability: the information persists while its geometric realization changes.
[LG-84] CD-LoRA: Consistency-Driven Low-Rank Adaptation for Multi-Task Fine-Tuning
链接: https://arxiv.org/abs/2608.21909
作者: Qian Zha,Jinda Liu,Yuan Wu,Yi Chang
类目: Machine Learning (cs.LG)
*备注:
Abstract:While Multi-Task Learning (MTL) is essential for adapting Large Language Models (LLMs) to diverse domains, prevailing LoRA-based methods rely on complex routing mechanisms that partition task-specific knowledge. In this work, we reveal that such routing-based designs are prone to a training-inference discrepancy, where stochastic routing decisions under distribution shifts compromise inference stability. Driven by a second-order Taylor analysis that exposes the instability induced by routing variance, we challenge the training-inference discrepancy and propose Consistency-Driven Low-Rank Adaptation (CD-LoRA). By eliminating routers entirely, CD-LoRA employs a consistency-driven alignment mechanism to enforce representation congruence across tasks in a shared low-rank space. This paradigm fosters robust, task-agnostic features without explicit partitioning overhead. Extensive experiments show that CD-LoRA consistently outperforms state-of-the-art multi-adapter baselines, offering a simpler, router-free, and more stable solution for multi-task PEFT. The code is available at the anonymous link this https URL.
[LG-85] herMapNet Attention-Guided Runtime Full-Chip Thermal Map Prediction from Performance Metrics
链接: https://arxiv.org/abs/2608.21887
作者: Qin Gu,Chaofang Ma,Mingyu Yang,Yipu Zhang,Jiliang Zhang,Wei Zhang,Lin Jiang
类目: Hardware Architecture (cs.AR); Machine Learning (cs.LG)
*备注: IEEE conference format, 8 figures, 2 tables. Submitted to IEEE ICCD. Corresponding author: Lin Jiang
Abstract:Runtime thermal management of high-performance chips depends on fast and accurate full-chip thermal maps. Conventional simulators typically estimate power traces from performance metrics first, which adds overhead. This work proposes TherMapNet, an attention-guided thermal simulator that predicts full-chip thermal maps directly from performance metrics. A Transformer encoder captures temporal evolution by treating the time series of each metric as a token, improving modeling of dynamic workloads. A CNN then extracts fine-grained spatial features. For the CNN, a dual-branch channel-spatial attention convolution module (DACM) and a triplet loss are used to improve spatial learning and reconstruction accuracy. TherMapNet is applied to a multi-core CPU (AMD Ryzen 7 4800U) and a many-core GPU (NVIDIA GeForce RTX 4060). Experiments show that it outperforms prior thermal simulators, with RMSE below 0.26 C and inference under 2.4 ms on an NVIDIA GeForce RTX 3090 GPU. These results indicate that TherMapNet can support high-quality runtime thermal management of modern multi-core chips.
[LG-86] More Experts Worse Dynamics: Inverse Scaling and Spectral Bias in Mixture-of-Experts State-Space Models
链接: https://arxiv.org/abs/2608.21840
作者: Chandresh Pandey
类目: Machine Learning (cs.LG)
*备注: 11 pages, 4 figures
Abstract:Mixture-of-Experts (MoE) architectures are commonly motivated as a way to increase expressivity by decomposing complex systems into simpler local dynamics. This intuition has recently been extended to spectral state-space models, where mixing stable operators is assumed to enable adaptation to heterogeneous or regime-switching time series. We critically evaluate this assumption in a controlled synthetic setting designed to isolate dynamical rather than representational challenges. We study a next-step prediction task on sequences composed of three regimes: chaotic dynamics generated by the Mackey-Glass system, a stable oscillatory regime, and a noise-dominated autoregressive regime. Across extensive ablations including capacity scaling, oracle routing, frozen-expert variants, and comparisons to output-level MoE baselines, operator-level mixture models consistently fail to outperform a single-expert baseline. Increasing the number of experts leads to inverse scaling, routing collapses or fails to induce meaningful specialization, and even perfect regime supervision does not prevent degradation in global performance. Furthermore, we show that apparent improvements in mean squared error on chaotic trajectories can be misleading. Phase-space analysis reveals that lower error often arises from temporal smoothing that destroys the geometry of the underlying attractor rather than from faithful modeling of the dynamics. These results identify a likely limitation of operator interpolation under the studied parameterization and training protocol, and underscore the need for geometry-aware evaluation when assessing regime-switching dynamical systems.
[LG-87] Adaptive Multilevel Twisted Sequential Monte Carlo for Rare Events Estimation in Language Models
链接: https://arxiv.org/abs/2608.21736
作者: Zixuan Liu,Fangzheng Wu,Brian Summa,Zizhan Zheng
类目: Machine Learning (cs.LG)
*备注:
Abstract:Rare unsafe behaviors in large language models can remain practically significant even when their probability is extremely small, particularly at deployment scales involving millions or billions of interactions. Twisted Sequential Monte Carlo (SMC) provides a principled framework for rare-event probability estimation by learning twist functions that guide generation toward a target event. However, the standard twist learning framework relies on positive samples from the rare-event target distribution, which may be nearly absent before an informative twist has been learned, resulting in unreliable rare-event estimation. We propose Adaptive Multilevel Twisted SMC, which learns the rare-event twist through a sequence of progressively rarer intermediate events. At each level, the learned twist provides more informative positive examples for learning the next twist, ultimately leading to a more accurate final twist for the target rare event. Experiments across diverse tasks and model scales show that the proposed method produces more accurate rare-event probability estimates. By enabling more reliable discovery of hard-to-observe unsafe behaviors, our method provides a practical tool for strengthening the evaluation and safety alignment of deployed language models.
[LG-88] Posterior Information Dynamics of Diffusion Models for Linear Inverse Problems
链接: https://arxiv.org/abs/2608.21709
作者: Xiangming Meng
类目: Machine Learning (cs.LG); Information Theory (cs.IT); Signal Processing (eess.SP)
*备注:
Abstract:Diffusion models are widely used as priors for linear inverse problems, yet endpoint quality does not reveal when measurement information enters reverse denoising or how it is allocated across signal directions. We study this process through the smoothed likelihood force, the difference between exact posterior and prior scores at each noise level. For a fixed measurement, its expected squared norm gives both posterior–prior relative-entropy dissipation and reverse-path relative-entropy growth. Averaging over measurements yields an information–minimum mean-square error (I-MMSE) identity linking information gain to denoising-error reduction. Under finite second moments, the force energy and its ratio to prior-score energy decay quadratically in the noising kernel’s signal coefficient at high noise. Solvable models show that conditioning removes class separation already explained by the measurement, reduces a uniform index entropy over (n) empirical samples from (\log n) to (H(I\mid r)), and makes assimilation depend on operator–prior alignment even for identical singular values. Experiments in models with tractable posteriors evaluate these predictions. In a separate illustration with a frozen FFHQ model, masks sharing the same spectrum yield different prior-normalized null-space trajectory statistics.
[LG-89] Benchmarking Composable Compression Techniques in Mixture-of-Experts LLM s
链接: https://arxiv.org/abs/2608.21693
作者: Afsara Benazir,Chen Chen,Rongxiao Qu,Jiabo Huang,Jingtao Li,Lingjuan Lyu
类目: Machine Learning (cs.LG)
*备注:
Abstract:Mixture-of-Experts (MoE) LLMs scale model capacity efficiently through sparse activation, but their large expert parameter footprint, routing imbalance, and long-context KV-cache growth make deployment difficult on commodity hardware. Practical deployment often requires stacking multiple compression techniques: expert pruning removes redundant experts, weight quantization lowers model memory footprint, and KV-cache compression reduces long-context memory pressure. However, these techniques are typically evaluated in isolation, leaving open how they interact when applied together in realistic deployment pipelines. In this work, we present MoEXBench, a systematic benchmark for evaluating composable MoE compression as an end-to-end deployment workflow. MoEXBench studies 10 MoE models ranging from 30B to 235B total parameters across standard-attention, hybrid linear-attention, and sliding window attention architectures. It evaluates 20%-50% expert pruning rates, 1 to 16 bit weight-quantization schemes, and multiple KV-cache precision settings, applied both individually and in combination. MoEXBench introduces an eight-module evaluation suite that jointly measures composable-compression quality, workload and architecture robustness, pruning/quantization/KV cache sensitivity, and deployment efficiency on commodity hardware. Our results reveal non-trivial interactions among compression methods: composable compression cannot be predicted from standalone techniques, compression rate alone does not reliably predict quality loss or runtime gain, expert pruning is the dominant degradation source, and average quality can hide workload and architecture-specific failures. By releasing normalized module scores, compressed artifacts, and reproducible scripts, MoEXBench enables practical accuracy-memory-latency comparison across MoE families and hardware backends. Subjects: Machine Learning (cs.LG) Cite as: arXiv:2608.21693 [cs.LG] (or arXiv:2608.21693v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2608.21693 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-90] MusPyExpress: Extending MusPy with Enhanced Expression Text Support NEURIPS2025
链接: https://arxiv.org/abs/2608.21678
作者: Phillip Long,Hao-Wen Dong,Julian McAuley,Zachary Novack
类目: ound (cs.SD); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS)
*备注: Accepted at NeurIPS 2025 Workshop on AI for Music: Where Creativity Meets Computation; 10 pages, 6 figures
Abstract:Current work in modeling symbolic music primarily relies on representations extracted from MIDI-like data. While such formats allow for modeling symbolic music as sequences of notes, they omit the large space of symbolic annotations common in western sheet music broadly known as expression text, such as tempo or dynamics, which specify time- and velocity-dependent controls on the musical composition and performance. To alleviate this gap, we present MusPyExpress, an extension to the popular symbolic music processing library MusPy that enables the extraction of expression text along with symbolic music for downstream modeling. Utilizing this extension, we parse the PDMX dataset to illustrate the wealth of expression text available in MusicXML datasets. Additionally, we introduce multiple generative tasks, including joint expression-note generation, expression-conditioned music generation, and expression tagging, that take advantage of this additional notational information.
[LG-91] Variational Structure at the Edge of Stability
链接: https://arxiv.org/abs/2608.21660
作者: Eric Regis
类目: Machine Learning (cs.LG)
*备注:
Abstract:When discrete-time optimizers operate at the edge of stability, they exhibit near-two-periodic behavior. These oscillatory dynamics are reminiscent of conservative systems, such as the dynamics generated by symplectic integrators. However, a precise formulation of the connection between discrete-time optimizers at the edge of stability and discrete mechanics remains underexplored. Recently, Litman introduced the “edge coupling”: a functional on consecutive gradient descent iterates whose critical points encode the fixed points and two-point orbits of the gradient descent dynamics. Here we extend the edge coupling to heavy-ball and Nesterov momentum. We show that its critical points characterize the fixed points and two-point orbits, with its Hessian characterizing their stability. We also show that the edge coupling can be identified with the symmetric Verlet action, formalizing the connection between the edge of stability and discrete mechanics.
[LG-92] GeoQ: Geometry-Aware Conditional Quantile Error Estimation for Scientific Surrogate Models
链接: https://arxiv.org/abs/2608.21652
作者: Khoa Nguyen,Daniel Serino,Aviral Prakash,Marc Klasky
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 25 pages, 6 figures, 5 tables
Abstract:Neural-network surrogate models are increasingly used to accelerate scientific simulations, but their deployment in extrapolative and autoregressive settings requires input-dependent estimates of prediction error. In this work, we introduce GeoQ (Geometry-Aware Conditional Quantile Error Estimation), a non-intrusive calibration framework for estimating surrogate error at individual query points. GeoQ represents the error at a query point as an anchor-averaged calibration error plus a learned nonnegative correction. This correction is modeled as an upper conditional quantile of the anchor-relative error increment, using geometry-based features that encode representation-space displacement and local support density. A cross-fitting procedure generates approximately out-of-sample calibration tuples, while a feature-space k-nearest-neighbor support score identifies regions \textcolorblackwhere the learned error model is supported by calibration data. We evaluate GeoQ on scalar regression, chaotic dynamics, medium-range weather forecasting, and Richtmyer-Meshkov instability prediction. The results demonstrate that geometry-aware conditional quantile modeling provides a practical and non-intrusive approach for validity-aware error estimation in scientific surrogate models.
[LG-93] Large-Scale Evaluation of Advanced Imputation Methods for Missing Values in Smart Meter Data
链接: https://arxiv.org/abs/2608.21638
作者: Daniela Stojcheska,Marija Markovska,Dimitar Taskovski,Branislav Gerazov,Boris Nikolov
类目: Machine Learning (cs.LG)
*备注: 6 pages, 7 figures; presented at the 2026 IEEE International Conference on Environment and Electrical Engineering and 2026 IEEE Industrial and Commercial Power Systems Europe (EEEIC / ICPS Europe)
Abstract:Accurate and reliable collection of electricity consumption data through Advanced Metering Infrastructure (AMI) is of great importance for the operation of smart grids, especially for the detection of non-technical losses (NTL). However, real-world datasets frequently suffer from missing values due to communication failures. This paper presents an empirical evaluation of three advanced algorithms for large-scale data imputation: the Optimally Weighted Average (OWA) method, Low-Rank Matrix Completion via SoftImpute, and a Shape-Modeling Autoencoder. Existing studies on missing value imputation in electricity consumption data often lack validation on larger datasets. Therefore, the goal of this paper is to validate the selected algorithms on a large-scale real-world electricity consumption dataset from North Macedonia that includes 17,428 commercial smart meters over two years. The robustness of each algorithm is evaluated by simulating continuous gaps in the data ranging from 1 to 168 hours. The results indicate that OWA provides the lowest overall reconstruction error across the evaluated gap sizes and strong stability in worst-case scenarios for gaps of up to one week. In contrast, the autoencoder exhibits higher variance, while SoftImpute has stable but inferior accuracy. These findings suggest that imputation methods should be selected based on the characteristics of load curve data and highlight the potential for hybrid algorithmic architectures in future grid management systems.
[LG-94] ChequeMark: An Ensemble Machine Learning Framework for After-Hours Business Deposit Fraud Detection
链接: https://arxiv.org/abs/2608.21629
作者: Ann Youduo Xu,Emily Yu,Justin Leski,William Lam
类目: Machine Learning (cs.LG)
*备注:
Abstract:Cheque fraud is a material risk in after-hours business deposit operations because funds may be released within one business day, while cheque clearing takes several days. This timing gap creates a fraud exposure window for financial institutions. Prior mitigation relies on static, deposit-level checks and therefore miss historical client behavior and evolving patterns. To address this gap, we propose a multi-view ensemble ML framework that combines: Extreme Gradient Boosting (XGBoost) for known fraud patterns, Isolation Forest for label-free anomaly detection, and Graph Sample and Aggregate (GraphSAGE) for relational patterns associated with transaction activities. We then combine the three outputs into a single client-level risk score. Under stable conditions, performance is comparable to XGBoost; under a targeted distribution shift, our framework performs best (F1: 83.77%, FPR: 0.69%) versus XGBoost (F1: 82.77%, FPR: 0.72%). These results indicate improved robustness to distribution shift while preserving interpretability through plain-language explanations grounded in behavioural, anomaly, and relational evidence.
[LG-95] Rethinking Communication Metrics: How Should We Measure Meaning?
链接: https://arxiv.org/abs/2608.21626
作者: Niloofar Tavakolian,Hakimeh Purmehdi,Jungyeon Baek
类目: Machine Learning (cs.LG)
*备注:
Abstract:Semantic communication shifts the objective of communication systems from accurate symbol reconstruction toward meaning preservation, task accomplishment, and efficient information exchange. However, its evaluation remains fragmented across telecommunications, natural language processing, computer vision, and machine learning, and no single metric can characterize semantic quality across modalities, tasks, and channel conditions. This article surveys key performance indicators (KPIs) for text- and image-based semantic communication systems from a unified, evaluation-centered perspective. Unlike prior surveys primarily organized around architectures, applications, or transmission strategies, this work focuses on how semantic success should be defined and measured. Existing KPIs are classified according to communication goal, source modality, receiver output, reference availability, evaluation level, and channel or resource constraints. The survey reviews reconstruction-based, task-oriented, reference-free, representation-level, perceptual, and channel-aware metrics, and presents a cross-modality comparison of their roles, strengths, and limitations. It further analyzes how unresolved semantic-KPI challenges affect monitoring, quality assurance, resource optimization, fault diagnosis, and standardization. Key open problems include the absence of universal semantic success criteria and standardized semantic ground truth, semantic drift, limited reference-free evaluation, weak integration of machine-learning metrics with communication constraints, and the lack of relation-level and multimodal KPIs. Finally, future research directions are outlined toward standardized, interpretable, adaptive, task-aware, and communication-aware evaluation frameworks.
[LG-96] FrugalSOT - Frugal Search Over the Models
链接: https://arxiv.org/abs/2608.21621
作者: Pradheep P,Yuvanesh S,Harish KB,Keerthan Saai Reddy S,Joshva Devadas T,Naveenkumar J,Hemalatha K
类目: Machine Learning (cs.LG)
*备注:
Abstract:In on-device NLP tasks, limited resources of embedded hardware, such as the Raspberry Pi 5, require efficient inference strategies. This paper introduces FrugalSOT (Frugal Search Over The Models), a resource-aware model selection architecture for on-device NLP inference. FrugalSOT estimates each request’s complexity by extracting features such as prompt length, named entity density, and syntactic complexity. The request is first made to the least complex model that is likely to pass a relevance threshold. If the output of that model falls short of the threshold, the request is made to a more complex model. It is important to note that the relevance threshold undergoes continuous updates in the background. using past validation outcomes in an adaptation process using a low-pass filtering mechanism, thus imparting adaptation to changing input patterns. Experimental results achieved on a Raspberry Pi 5 show that FrugalSOT reduces average inference time and overall computational resource use to a significant extent compared to a single-model baseline approach, without compromising output relevance to the same extent as the most sophisticated model. These results confirm that adaptive model selection can enable efficient, high-quality natural language processing inference on limited devices.
[LG-97] Congruence Decomposition with Neural Block Solvers for Large-Scale PCI Assignment
链接: https://arxiv.org/abs/2608.21485
作者: Yeqing Qiu,Chengpiao Huang,Ye Xue,Akang Wang,Fan Xu,Zhipeng Jiang,Dong Zhang,Ruoyu Sun,Qingjiang Shi,Zhi-Quan Luo
类目: Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注:
Abstract:Physical Cell Identity (PCI) assignment is essential for interference management in dense 5G networks. As cellular networks scale, PCI reuse becomes unavoidable, which may cause collisions, confusions, and multiple forms of modular interference. Jointly mitigating these effects gives rise to a large-scale, multi-objective combinatorial optimization problem that is difficult to solve efficiently at practical network scales. In this work, we propose a congruence decomposition framework with neural block solvers for large-scale PCI assignment. The proposed decomposition exploits the arithmetic structure of PCI values to decouple multiple modular interference objectives into a collection of blockwise Min- k -Partition subproblems, followed by a graph coloring procedure to resolve PCI conflicts. For the resulting NP-hard Min- k -Partition subproblems, we develop neural block solvers by parameterizing their relaxed quadratic formulations with graph neural networks, enabling efficient optimization at large scales. Discrete assignments are recovered through conditional expectation rounding with theoretical guarantees. Experiments on synthetic cellular graphs and real-world 5G networks show that the proposed method consistently outperforms existing modular-interference-aware baselines in modular interference reduction, conflict elimination, and computational efficiency.
[LG-98] Explainable Adaptive Zero Trust Framework for AWS with Adversarial Robustness Evaluation
链接: https://arxiv.org/abs/2608.21477
作者: Om Singh,Yagyaraj Pandey,Nandini Pathak
类目: Cryptography and Security (cs.CR); Computers and Society (cs.CY); Machine Learning (cs.LG); Networking and Internet Architecture (cs.NI)
*备注:
Abstract:Cloud environments built on Amazon Web Services face a structural security vulnerability: once a credential passes authentication, the resulting session is often treated as trusted for its entire duration. This assumption fails when credentials are stolen. We introduce the Explainable Adaptive Zero Trust Framework (EAZTF), a cloud-native security layer that continuously reevaluates the legitimacy of API actions throughout a session. EAZTF combines Isolation Forest and XGBoost to evaluate eight CloudTrail and IAM-derived behavioral features in real time and produce a Trust Risk Score (TRS) that determines whether a session continues, requires step-up MFA, or is restricted. Each decision is accompanied by a SHAP or LIME explanation, providing human-readable audit records for security analysis and compliance. The framework is also evaluated against four adversarial evasion strategies: credential theft, behavioral mimicry, API rate evasion, and privilege escalation. Experiments on an 8,500-record synthetic CloudTrail dataset show that Isolation Forest achieves 94.4% precision, 91.2% recall, and an F1 score of 0.928. Across the four adversarial scenarios, the mean detection rate is 91.0%, with behavioral mimicry being the most difficult at 83.9%. SHAP analysis identifies IP reputation, login-time deviation, and API call velocity as the three dominant features. A structured NIST SP 800-207 self-assessment gives EAZTF a mean compliance score of 93%, compared with 38% for a traditional perimeter baseline. Mean time to detect decreases from hours to under one minute. Because the evaluation uses synthetic data, these results should be interpreted as indicative rather than validated production performance. Subjects: Cryptography and Security (cs.CR); Computers and Society (cs.CY); Machine Learning (cs.LG); Networking and Internet Architecture (cs.NI) Cite as: arXiv:2608.21477 [cs.CR] (or arXiv:2608.21477v1 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2608.21477 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-99] Class-Conditioned Gaussian Mixture Modeling for Imbalanced Time Series Quantification PAKDD2026
链接: https://arxiv.org/abs/2608.21473
作者: Md Shahriar Kabir,Mayesha Maliha R. Mithila,Anne H. H. Ngu,Mylène C. Q. Farias,Byron Gao
类目: Machine Learning (cs.LG)
*备注: 13 pages, 2 figures, 2 tables. Accepted at PAKDD 2026 (Pacific-Asia Conference on Knowledge Discovery and Data Mining), LNAI 16599, pp. 560-572, Springer, Singapore
Abstract:Quantification, estimating class prevalences in bags of unlabeled instances is vital in domains where aggregate statistics are more important than individual instance labels, such as biosignal monitoring, fall detection, and activity recognition. We investigate this issue in the challenging setting of imbalanced time series data and develop CC-GMNet-TS, a class-conditioned Gaussian mixture quantifier that combines a Transformer-based feature extractor with per-class latent mixtures. Unlike previous mixture-based quantifiers, which use a single Gaussian mixture shared by all classes, CC-GMNet-TS assigns each class its own compact mixture in a bounded latent space and scores segment embeddings against these class-specific components to create bag-level representations that emphasize rare but informative patterns. Bags are constructed from labeled pools using the Artificial Prevalence Protocol (APP) and prior shift bag sampling (PShift) to cover a wide range of class prevalence scenarios, and the model is trained end-to-end with a quantification-oriented loss. Experiments on three benchmarks: EMG Data for Gestures, SmartFallMM, and UCI-HAR show that CC-GMNet-TS achieves lower error across the three benchmarks compared to traditional aggregators and recent deep quantifiers, while ablations confirm the contributions of both the Transformer backbone and class-conditioned mixtures during PShift.
[LG-100] RiskWorld: Object-Centric Latent World Modeling for Autonomous Driving Risk Identification
链接: https://arxiv.org/abs/2608.21414
作者: Jingzheng Li,Yufei Ge,Qianren Mao,Zhijun Chen,Bing Li,Xingyu Peng,Baochang Zhang,Xianglong Liu
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注:
Abstract:Autonomous driving risk identification aims to determine which observed object is likely to become safety-critical to the ego vehicle. Existing approaches typically predict scene-level accidents, infer risk objects indirectly from ego behavior, or apply geometric checks after trajectory forecasting, without directly using predicted ego–object relations for risk-source localization. We propose RiskWorld, an object-centric latent world model that identifies risk from the imagined evolution of each candidate relative to the ego vehicle. RiskWorld combines pretrained predictive video representations with structured ego–object histories, contextualizes observed interactions, and rolls relation-aware object states into the future using RSSM-style latent dynamics. It decodes the rollout into object-level risk scores, supported by auxiliary future-relation and temporal-risk predictions. Inference uses only observations up to the current time, while logged futures provide training supervision. On RiskBench, RiskWorld achieves the best overall F1 of 63.0% and the lowest false-alarm rate of 2.1%. Further analyses show that the learned rollout captures the evolution of object-level risk before critical events, while RiskWorld’s selections preserve planning-critical information under filtered observation.
[LG-101] Exploring Long-period Architectures: Four New Planet Candidates from Kepler with Periods 342 days
链接: https://arxiv.org/abs/2608.23425
作者: Matthew T. Hansen,Jason A. Dittmann
类目: Earth and Planetary Astrophysics (astro-ph.EP); Instrumentation and Methods for Astrophysics (astro-ph.IM); Machine Learning (cs.LG)
*备注: 23 pages, 19 figures, 9 tables, published in AJ
Abstract:The Kepler detection pipeline, as well as the transit method, has a bias towards shorter periods, leaving a dearth of detections at longer orbital periods. This relative lack of detections has left an incomplete picture of the architectures of exoplanet systems within the long-period regime. We have built a single transit detection pipeline, utilizing a classification convolutional neural network and the onboard spacecraft diagnostics of the Kepler spacecraft, to detect long-period planets. We apply our pipeline to all currently known planetary systems in the Kepler field hosting at least one planet with an orbital period longer than 6 days. We manually vet all new signals from our pipeline, and identify four new planetary candidates, all of which are in systems where the inner planets exhibit transit timing variations (TTVs). Two of these candidates, Kepler 1752.02 and Kepler 199.03, cause two transit events that are consistent with periods of 777.78^+0.01_-0.02 and 505.495^+0.004_-0.004 days, and radii of 3.55^+0.15_-0.15 and 2.74^+0.05_-0.05 R_\oplus , respectively. Our remaining two candidates, Kepler 1897.02 and Kepler 1811.02, are single transit candidates with radii 4.81^+0.20_-0.19 and 3.25^+0.28_-0.30 R_\oplus , respectively. The shortest orbital periods for these candidates, consistent with the Kepler dataset (gaps and coverage), are 342 days for Kepler 1897.02 and 544 days for Kepler 1811.02. The new planetary candidates, on their own, are incapable of reproducing the observed TTV signals in the inner system. Although difficult to schedule, follow-up observations are needed to further constrain the new candidates and potentially discover the planets causing the perturbations.
[LG-102] KellyBoost: Growth-Optimal Portfolio Construction with Gradient-Boosted Trees
链接: https://arxiv.org/abs/2608.23393
作者: Jiayu Li
类目: Portfolio Management (q-fin.PM); Machine Learning (cs.LG)
*备注:
Abstract:KellyBoost is a single multi-output XGBoost model whose softmax output is the portfolio: with y the vector of per-asset holding-period returns, the training loss is - log(1 + w y), the negative log growth rate, so the fitted model is the growth-optimal (Kelly) allocation conditioned on the features. The objective is exact rather than a surrogate: we derive the gradient, the analytic diagonal Hessian and the full Hessian in closed form, verify them by finite differences, and ship a dependency-free reference engine.
[LG-103] Beyond chlorophyll: machine learning estimates of diagnostic phytoplankton pigments from multispectral ocean colour data
链接: https://arxiv.org/abs/2608.23348
作者: David Moffat,Angus Laurenson,Victor Martinez-Vicente,Gemma Kulk,Xuerong Sun,Robert J. W. Brewin,Shubha Sathyendranath
类目: Other Quantitative Biology (q-bio.OT); Machine Learning (cs.LG); Optics (physics.optics)
*备注: Submitted to Frontiers of Marine Science. 21 pages, 8 Figures
Abstract:Phytoplankton play a central role in marine ecosystems and the global carbon cycle, with different groups contributing differently to ocean biogeochemical processes. While standard techniques exist for monitoring phytoplankton concentration from ocean-colour data, their community composition remains difficult to observe at large scales. Chlorophyll-a, widely available from satellite ocean-colour observations, is commonly used as a measure of phytoplankton biomass but provides limited information on taxonomic composition. Accessory pigments, some of which are diagnostic of important phytoplankton groups, offer additional information on community structure, but their retrieval from ocean-colour data is challenging because of limited spectral resolution and strong covariance with chlorophyll-a. In this study, we evaluate machine learning methods for estimating diagnostic pigment concentrations from multispectral satellite observations. Using a global dataset of 33,640 High Performance Liquid Chromatography (HPLC) measurements matched with ESA Ocean Colour Climate Change Initiative (OC-CCI) reflectance data, we compare Random Forest and TabPFN models trained on multispectral reflectance with baseline models using chlorophyll-a alone. A temporally stratified validation scheme is employed to reduce the effects of autocorrelation. Results show that multispectral models consistently outperform approaches based solely on satellite-derived chlorophyll-a, demonstrating that ocean-colour reflectance contains additional information relevant to pigment discrimination. Improvements vary by pigment, with those strongly correlated with chlorophyll-a showing limited gains, while others exhibit substantial improvement. These findings highlight the potential of machine learning to extract ecologically relevant information from satellite data beyond conventional chlorophyll-based approaches.
[LG-104] SGHA: A Single-Loop Fully First-Order Algorithm for Nonconvex-Strongly-Convex Bilevel Optimization
链接: https://arxiv.org/abs/2608.23211
作者: Zhihao Gu,Qilong Wu,Junchi Yang
类目: Optimization and Control (math.OC); Machine Learning (cs.LG)
*备注:
Abstract:In this work, we study the oracle complexity of finding an \epsilon -stationary point for nonconvex-strongly-convex (NC-SC) bilevel optimization using only first-order oracles. Existing methods achieving the best-known complexity guarantees typically rely on double-loop, penalty-based procedures. We propose a novel single-loop algorithm based on a constrained reformulation in which lower-level stationarity is imposed as a constraint. Specifically, we construct a regularized Lagrangian by introducing a quadratic regularizer and restricting the dual variable to a bounded domain, and then apply Smoothed Gradient Descent Ascent [Zhang et al., 2020], with Hessian-vector products approximated via finite differences of gradients. We refer to the resulting deterministic and stochastic algorithms as SGHA and Stoc-SGHA, respectively. In the deterministic setting, SGHA achieves an oracle complexity of O(\bar\kappa_y^5\epsilon^-2) , where \bar\kappa_y denotes the relevant condition number. In the stochastic setting, Stoc-SGHA achieves an oracle complexity of O\left(\bar\kappa_y^17\epsilon^-6\rho^-3\right) with probability at least 1-\rho for any \rho\in(0,1) , and an oracle complexity of O\left(\bar\kappa_y^17\epsilon^-6\right) in expectation under an additional bounded-iterate assumption. Moreover, under an additional stochastic smoothness assumption imposed only on the lower-level objective, the stochastic oracle complexity of Stoc-SGHA improves to O\left(\bar\kappa_y^11\epsilon^-4\rho^-2\right) with high probability and O\left(\bar\kappa_y^11\epsilon^-4\right) in expectation, matching the \epsilon -dependence of the lower bounds.
[LG-105] Quantum Reservoir Computing with Physics-Informed Correction for Reduced-Order PDE Forecasting
链接: https://arxiv.org/abs/2608.23119
作者: Krishna Bhatia,Harsh,Shalini Devendrababu
类目: Quantum Physics (quant-ph); Machine Learning (cs.LG)
*备注: 11 pages, 5 figures. Accepted at QNRL@WCCI 2026. To appear in Advances in Quantum Neural and Reinforcement Learning, Communications in Computer and Information Science, vol. 3063, Springer Singapore, 2026
Abstract:We study a hybrid proposal–correction architecture for reduced-order PDE forecasting in which a pure-state quantum reservoir computer (QRC) predicts latent coefficient dynamics and a PINN-based physics-informed corrector (PIC) refines local rollout windows. The method is evaluated on Burgers and Kuramoto–Sivashinsky (KS), with KS as the primary chaotic benchmark. On KS, QRC+PIC consistently improves over QRC alone in RMSE, NRMSE, and PDE residual, while Burgers highlights a regime in which simple baselines remain strong. These results suggest that QRC proposals with local physics-informed correction are a viable benchmark-dependent reduced-order forecasting strategy.
[LG-106] One Inverse Step is a Convex Program: Bayes-Limit Calibration of Diffusion Inversion ICML2026
链接: https://arxiv.org/abs/2608.23094
作者: Gordei Verbii
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: Revised follow-up of “Extracting Local Manifold Geometry from Pretrained Diffusion Models in One Inverse Step” (ICML 2026 Workshop SPIGM)
Abstract:One implicit DDIM inversion step is the cheapest probe of whether a pretrained diffusion model encodes local manifold geometry. It is the stationarity condition of an explicit potential, x-G(x)=\nabla\Psi_t(x) , strongly convex at the Bayes limit with modulus exactly e^-h_t for the step’s log-SNR gap h_t - for every data law, schedule and point, with no manifold, reach or unimodality hypothesis. Three consequences must be kept apart. (i) The solution is unique at the Bayes limit; a second one requires the trained score to violate the posterior-covariance bound by 1/(1-e^-h_t) , a hypothesis-free certificate of model error; the same bound makes contraction a schedule constant, \rho_g^\star=1-e^-h_t0.326 throughout the standard DDPM schedule. (ii) The solver can still fail: Picard iteration is unit-step gradient descent on \Psi_t , unstable wherever \lambda_\max(\nabla^2\Psi_t)2 , so oscillation certifies nothing; damping below 2/\lambda_\max cures it. (iii) The geometry lives in the convergence domain: on the scale-free depth w=r\kappa_\max the oscillation shell sits at w=\tfrac12 , schedule-free, and the divergence shell at w=1/(1+\rho_g^\star) , with a measured finite-noise correction in |\mathrmII|^2 . Exact scores reproduce both to within 0.54% on three classes; no trained score we probe shows a shell - a derived limitation, not a null result: the Fermi window conflicts with the model’s own training support by 3.6 - 5.6\times , and the trained Hessian-Lipschitz constant is 2 - 12% of the curvature the law reads, 0 on a ReLU net. Finally the unconditional ceiling \sigma_t\lambda_\max(\mathrmsym,J)\le1 , from \mathrmCov(x_0\mid x_t)\succeq0 alone, holds for the exact score to 3\times10^-7 but is violated in all DDPM CIFAR-10/CelebA-HQ-256 settings, by 1.26 - 4.66\times .
[LG-107] Partial-Moment PINNs for Caldeira–Leggett Parameter Learning in Quantum Brownian Motion IJCNN
链接: https://arxiv.org/abs/2608.23093
作者: Krishna Bhatia
类目: Quantum Physics (quant-ph); Machine Learning (cs.LG)
*备注: 4 pages, 2 figures. Published in the 2026 International Joint Conference on Neural Networks (IJCNN), IEEE World Congress on Computational Intelligence (WCCI 2026)
Abstract:We study parameter recovery in the Caldeira–Leggett (quantum Brownian) oscillator from partial moment traces. Our model is a moment-level PINN that predicts the five first/second moments and enforces the linear CL/HPZ ODEs by automatic differentiation. Physical structure is imposed through a PSD (Cholesky) covariance head, high-temperature CL assumptions with D_xp\approx0 , and fluctuation–dissipation ties between D_pp and \gamma . On synthetic CL data with channels \mu_x,\sigma_xx,\sigma_xp , the constrained variant recovers (\omega,\gamma) accurately, stabilizes D_pp , and achieves low rollout error compared to finite differences and Kalman–EM (expectation–maximization) with exact Van Loan discretization. Fisher-style checks confirm that diffusion needs at least one variance observable, and sparse \sigma_pp ``anchors’’ restore conditioning. We also show that the same PINN can learn time-varying HPZ coefficients.
[LG-108] When a neural surrogate cannot accelerate a solver: runtime share closed-loop drift and the economics of uncertainty gating in a stiff coupled simulation
链接: https://arxiv.org/abs/2608.23075
作者: L. Thümmler,T. Kuroda
类目: Instrumentation and Methods for Astrophysics (astro-ph.IM); High Energy Astrophysical Phenomena (astro-ph.HE); Machine Learning (cs.LG); Computational Physics (physics.comp-ph)
*备注: 44 pages, 10 figures, 3 tables
Abstract:Learned surrogates for expensive inner solver blocks are a widely pursued route to faster multiphysics simulation. We report a controlled, end-to-end negative result and identify three structural barriers, none of them a deficiency of the network we trained. The testbed is the implicit Newton solve coupling energy-dependent neutrino radiation to matter in a general-relativistic radiation-hydrodynamics code, its most expensive physics routine per call. First, per-call cost and share of runtime are different quantities, and only the second bounds acceleration. An exclusive self-time profile puts the target block at 16.9% of critical-rank wall clock, capping any surrogate at ~1.2x by Amdahl’s law. A surrogate 5.8x cheaper per call merely ties the solver, and the configuration stable enough to run without fallback reaches only parity. Second, offline accuracy cannot rank surrogates for deployment: across fourteen networks the pooled Spearman error-versus-survival correlation (rho = +0.73) is a between-family confound that vanishes under control (rho = -0.04). Third, a correct out-of-distribution gate cannot accelerate a loop that leaves its training distribution. We give the break-even deferral fraction in closed form: because the visited states sit 73x off the data manifold, the gate defers 96.8 to 99.7% of cells, almost invariant to surrogate quality. Including its own cost, the gated loop is a 0.94 to 0.96x slowdown. We further separate stability from fidelity: a never-crashing gated run accumulates a linear -19.9% density bias over 6000 steps. The error is a directed, ballistically accumulating bias, not the variance-driven divergence the autoregressive literature targets. Comments: 44 pages, 10 figures, 3 tables Subjects: Instrumentation and Methods for Astrophysics (astro-ph.IM); High Energy Astrophysical Phenomena (astro-ph.HE); Machine Learning (cs.LG); Computational Physics (physics.comp-ph) Cite as: arXiv:2608.23075 [astro-ph.IM] (or arXiv:2608.23075v1 [astro-ph.IM] for this version) https://doi.org/10.48550/arXiv.2608.23075 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-109] Neural Boltzmann Equations
链接: https://arxiv.org/abs/2608.23022
作者: Jonas Spinner,Jack Shergold
类目: High Energy Physics - Phenomenology (hep-ph); Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: v1: 21 pages, 4 figures, 1 table
Abstract:The dynamics of particles in the early universe are described by Boltzmann equations, which involve high-dimensional phase-space integrals. Classical approaches use quadrature integration and evolve the system on a fixed momentum grid, which scales poorly to complicated systems and parameter scans, severely limiting the complexity of processes that can be studied. We introduce Neural Boltzmann Equations (NBEs), which combine three coupled concepts to overcome these limitations. First, particle properties are encoded in physics-inspired neural distribution functions, with parameters that can be predicted using neural networks, enabling efficient parameter scans. Second, phase-space integrals are evaluated with Monte Carlo, using importance sampling tools from collider physics. Third, we use the natural gradient method to evolve the system. After demonstrating the individual benefits of NBEs, we use the framework to perform a precision calculation of the effective number of relativistic neutrino degrees of freedom in the early universe.
[LG-110] A Commutator Framework for Selective Spectral Alignment in Deep Neural Networks
链接: https://arxiv.org/abs/2608.22910
作者: Kaj Nyström
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:We develop a finite-width geometric framework describing how learned feature geometries are organized, transported, and selectively aligned in deep neural networks. Incompatibility among weight-generated covariance, gates, and backward sensitivities is quantified through three families of commutators: between gates and covariance, between sensitivities and covariance, and between average gradient outer products (AGOPs) and neural feature matrices (NFMs). An exact layerwise identity decomposes the sensitivity-covariance commutator into four sources: downstream transport, adjacent-layer imbalance, pointwise sensitivity fluctuations, and nonlinear gate-covariance interactions. The AGOP-NFM commutator is a singular-value-weighted transport of the internal commutator, explaining why observed feature-side alignment alone does not determine the internal geometry from which it emerges. Buffered localized energies resolve mixing between separated covariance subspaces. We establish spectral-gap, projector-evolution, and stabilization estimates, and formulate conditional Lyapunov principles that yield decay under explicit geometric error-bound or intrinsic-damping assumptions. These criteria do not follow from gradient flow alone and clarify why risk reduction need not imply commutator collapse. Analytic examples and numerical experiments exhibit factorization of spectral and activation geometry, transient growth, and cancellation among nonzero sources. In tested finite-time regimes, cancellation dominated by a negative transport-imbalance interaction persists across depths, widths, and two regression benchmarks. Spectral alignment therefore appears as a layer- and scale-dependent compatibility phenomenon governed by transport, interaction, cancellation, and possible damping, rather than a universal consequence of training. Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG) MSC classes: Primary 68T07, Secondary 15A18, 15A27, 37N40 Cite as: arXiv:2608.22910 [stat.ML] (or arXiv:2608.22910v1 [stat.ML] for this version) https://doi.org/10.48550/arXiv.2608.22910 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Kaj Nyström [view email] [v1] Mon, 24 Aug 2026 07:42:16 UTC (930 KB)
[LG-111] Mirror descent algorithms with logarithmic barriers
链接: https://arxiv.org/abs/2608.22834
作者: Alberto De Marchi,Yura Malitsky,Adrien B. Taylor
类目: Optimization and Control (math.OC); Machine Learning (cs.LG); Numerical Analysis (math.NA)
*备注: 18 pages
Abstract:This work derives convergence guarantees for mirror descent and proximal mirror descent algorithms when a logarithmic barrier is used as a distance-generating function. Standard approaches cannot be applied when the solution lies on the boundary, where the Bregman divergence blows up. We show that, in a specific setting, both methods enjoy an O(\log k / k) rate, which is also tight. In addition, our contributions include: (i) a new technique for handling the blow-up; (ii) a resolution of a gap in the theory of relative smoothness; and (iii) a comparison of the proposed approach with interior-point methods.
[LG-112] Generative Neural Networks for Sinkhorn Distributionally Robust Hypothesis Testing
链接: https://arxiv.org/abs/2608.22746
作者: Fenglin Zhang,Teyan Liu,Jie Wang
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注: 41 Pages, 7 figures
Abstract:This paper studies the Sinkhorn distributionally robust hypothesis testing (SDRHT) problem, seeking a robust detector against least-favorable distributions in Sinkhorn discrepancy-based ambiguity sets centered at the empirical distributions. Existing approaches solve this problem by solving large-scale conic programs, which are not scalable. To overcome this, we propose a generative framework that learns least-favorable distributions and supports efficient training and end-to-end sampling. For the Sinkhorn discrepancy-based ambiguity sets, we first derive an equivalent conditional-KL-divergence representation with respect to kernel-smoothed reference distributions. This property allows us to prove strong duality for both constrained and unconstrained minimax SDRHT formulations. Based on the closed-form optimal detector and Brenier’s theorem, we reformulate the max-min dual formulation as a maximization problem over convex potentials whose gradients characterize invertible transport maps between kernel-smoothed distributions and their least-favorable counterparts. We efficiently approximate these potentials using Hyper Input Convex Neural Networks (HyCNNs) equipped with stochastic gradient estimators and prove the representation power of HyCNNs and the distributional universality of their induced transport maps. Numerical results show that the proposed method achieves superior accuracy and robustness across different sample sizes and dimensions, while avoiding the scalability limitations of classical SDRHT methods.
[LG-113] Lightweight Multi-scale Hierarchical Anomaly Detection and Localization for Geospatial Big Data Applications at the Edge
链接: https://arxiv.org/abs/2608.22648
作者: Thomas Benton Townsend,Joshua Bean,Benjamin K Tkach,Narcisa Gabriela Pricope,Dimitrios Michael Manias
类目: ignal Processing (eess.SP); Machine Learning (cs.LG); Systems and Control (eess.SY)
*备注: Accepted at IEEE GLOBECOM 2026
Abstract:As an increasing number of critical applications, including environmental, emergency, meteorological, and agricultural, rely on real-time anomaly detection in geospatial data streams, challenges related to the storage, processing, and communication of this data arise. Traditionally, large volumes of data have been sent to centralized processing locations for insight extraction. Given the big data context of these applications, this approach becomes increasingly infeasible as data volume and velocity continue to increase. This paper proposes a lightweight edge-oriented approach for anomaly detection and localization for geospatial data streams. By leveraging the H3 discrete global grid system and a multi-scale drill-down logic, the proposed approach significantly reduces computational overhead, achieving a 99.7% reduction in evaluations compared to traditional flat-scan methods. Furthermore, by filtering out noise-induced flickering anomalies at lower resolutions, spatially-persistent anomalous signals can be efficiently identified. The results demonstrate that the proposed framework effectively distills massive geospatial data into actionable insights.
[LG-114] Scale-invariant Optimal Sampling for Rare-events Data with Sparse Models
链接: https://arxiv.org/abs/2608.22597
作者: Jing Wang,HaiYing Wang,Qiang Zhang,Hao Helen Zhang
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Subsampling is effective in tackling computational challenges for massive data with rare events. Overly aggressive subsampling may adversely affect estimation efficiency, and optimal subsampling is essential to mitigate the information loss. However, existing optimal subsampling probabilities depend on data scales, and some scaling transformations may result in inefficient subsamples. This problem is more significant when there are inactive features, because their influence on the subsampling probabilities can be arbitrarily magnified by inappropriate scaling transformations. We tackle this challenge and introduce a scale-invariant optimal subsampling function in the context of sparse models, where inactive features are commonly assumed. Instead of focusing on estimating model parameters, we define an optimal subsampling function to minimize the prediction error, using adaptive lasso to outline the estimation procedure and study its theoretical guarantee. We first introduce the adaptive lasso estimator for rare-events data and establish its oracle properties, thereby validating the use of subsampling. Then we derive a scale-invariant optimal subsampling function that minimizes the prediction error of the inverse probability weighted (IPW) adaptive lasso. Finally, we present an estimator based on the maximum sampled conditional likelihood (MSCL) to further improve the estimation efficiency. We conduct numerical experiments using both simulated and real-world data sets to demonstrate the performance of the proposed methods.
[LG-115] Sparse Additive Off-Policy Evaluation for Reinforcement Learning with Potentially Limited Number of Trajectories
链接: https://arxiv.org/abs/2608.22595
作者: Tuoyi Zhao,Chengchun Shi,Zhengling Qi,Lan Wang
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Statistics Theory (math.ST)
*备注:
Abstract:We develop a new framework for flexible, nonlinear, and interpretable off-policy evaluation for infinite-horizon reinforcement learning. To handle large state spaces and support transparent decision-making, we model the Q-function using a nonlinear function class with a sparse additive structure. We derive high-probability finite-sample error bounds for estimating the value function of a target policy and show that the bounds depend only logarithmically on the ambient dimension d , thereby alleviating the curse of dimensionality. In contrast to most existing theory for off-policy evaluation, which typically assumes access to many trajectories, our analysis guarantees accurate value estimation when either the number of trajectories or the time horizon is sufficiently large. In addition, we propose a group-sparsity-based feature screening procedure that identifies, with high probability, a reduced feature set containing all relevant covariates. Numerical experiments demonstrate the effectiveness of the proposed approach.
[LG-116] Interpretable statistical feature engineering for early disruption prediction in the short pulse ADITYA tokamak
链接: https://arxiv.org/abs/2608.22515
作者: Jyoti Agarwal,Kavit Patel,Bhaskar Chaudhury,Abhishek Sharma,Shrichand Jakhar,Manika Sharma
类目: Plasma Physics (physics.plasm-ph); Machine Learning (cs.LG); Data Analysis, Statistics and Probability (physics.data-an)
*备注: 27 pages
Abstract:Reliable early disruption prediction is critical for the safe operation and real-time control of tokamaks. However, machine learning based prediction frameworks have predominantly targeted medium and long pulse devices, with comparatively limited attention given to short pulse tokamaks where available warning time is inherently constrained. In this work, an interpretable machine learning framework is developed for feature engineering and early prediction of disruptions in the ADITYA using the initial plasma evolution information, prior to the activation of the negative converter of the ohmic transformer power supply. Statistical descriptors comprising the mean, variance, skewness, kurtosis and wavelet energy entropy are extracted from routinely available plasma diagnostics over different operation time windows. Decision tree based feature selection is employed to identify physically meaningful disruption precursors and to reduce feature dimensionality. These selected features are used to train a random forest classifier. The proposed framework achieves stable predictive performance across different analysis windows, with a maximum ROC-AUC of 0.87 for 0-35 ms and 0-40 ms windows. Comparable and in some cases improved, performance is obtained using the reduced feature set, demonstrating that the selected statistical descriptors retain the essential information required for disruption prediction. The proposed methodology provides an interpretable and computationally efficient framework for real time disruption prediction in short pulse tokamaks and establishes that carefully engineered statistical descriptors can effectively replace raw time series inputs for early disruption prediction, thereby offering a practical pathway toward real time plasma control in short pulse tokamaks similar to ADITYA and ADITYA-U.
[LG-117] Arbitrag e-Aware Multi-Step Forecasting of Implied Volatility Surfaces: Modelling Surface Trajectories Using Latent Diffusion
链接: https://arxiv.org/abs/2608.22478
作者: Dominik Manuel Buchegger,Lukas Gonon
类目: Mathematical Finance (q-fin.MF); Machine Learning (cs.LG)
*备注:
Abstract:Implied volatility surfaces summarise the option market and are central to many financial applications. Forecasting their future evolution requires modelling two-dimensional geometry, temporal dependence, and predictive uncertainty while preserving economic admissibility. We propose a conditional latent diffusion framework for generating joint 30-step trajectories of implied volatility surfaces and underlying returns. An arbitrage-aware autoencoder learns a low-dimensional surface representation, while the diffusion model captures the conditional joint evolution. Evaluated on SPX surfaces, the framework generates realistic probabilistic multi-step scenarios while also outperforming the persistence benchmark in point forecasting.
[LG-118] Dataset Complexity Shapes Finite-Distance Loss Geometry in Neural Networks
链接: https://arxiv.org/abs/2608.22361
作者: Jaeyong Bae,Hawoong Jeong
类目: Disordered Systems and Neural Networks (cond-mat.dis-nn); Statistical Mechanics (cond-mat.stat-mech); Machine Learning (cs.LG)
*备注: 18 pages; 39 pages including supplementary material
Abstract:Finite datasets can share the same size and low-order statistics while differing strongly in structural complexity. We connect this dataset complexity to loss-landscape geometry by pairing local label mixing across neighborhood scales with local entropy around trained neural-network solutions. Adapted from the Franz–Parisi construction in spin-glass theory, local entropy measures the effective volume of low-loss, solution-like parameter configurations at each distance from a reference. We estimate it in finite networks using adaptive sequential Monte Carlo. In a controlled synthetic sweep, greater dataset complexity produces a larger decrease in local entropy near the reference. Farther away, its radial derivative becomes weak and nearly common across conditions. Dataset complexity therefore changes where the effective solution volume contracts, rather than making it decrease uniformly faster. Experiments on real image data show the same qualitative trend, with label randomization further amplifying the effect. These results show that dataset structure shapes how low-loss neighborhoods are organized across finite distances from trained solutions.
[LG-119] Recovering Weighted Tangent Geometry from a Single-Scale Score Field
链接: https://arxiv.org/abs/2608.22334
作者: Ziqi Zhao,Qingjian Ni
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Near a smooth data manifold, one tangent space summarizes local geometry. At a branch point, the corresponding first-order object is instead a measure over tangent directions, whose normalized masses record the local share of each branch under the chosen data measure. We ask whether a score field at one noise level determines this weighted tangent geometry when the branch center and homogeneity degree d are unknown. In this tangent-measure model, d is the local measure dimension. Gaussian smoothing of a homogeneous tangent measure satisfies an Ornstein–Uhlenbeck eigenfunction equation. Its weak form turns score values—without score derivatives—into a linear system for the center and homogeneity degree, with an explicit rank condition and perturbation bound. After this calibration, the tangential score on one sphere is the spherical log-gradient of a scalar Gaussian–cone transform. Integration recovers that transform up to scale, and all its spherical-harmonic multipliers are positive. Thus one exact shell identifies the normalized angular measure in every ambient dimension D\geq2 . For at most K positive rays, moments through degree 2K-1 constructively recover count, directions, and weights in arbitrary dimension. Any fixed observation scheme needs at least KD-1 scalar tangential components. In the plane, degree K is both sufficient and necessary, and we give quantitative finite-query certificates. For finite planar C^1,\beta branches with positive C^0,\beta densities, we prove O(\sigma^\beta) convergence from the finite-noise score to its tangent model. In controlled experiments, 50k-step training lowers validation normalized-score error across four geometries yet raises angular-moment error, separating ordinary score fit from geometry recovery.
[LG-120] he spatial anatomy of urban wildfire vulnerability: a spatially validated GeoAI framework reveals the roles of building density and vegetation moisture in structure loss during the 2025 Palisades Fire
链接: https://arxiv.org/abs/2608.22293
作者: Parastoo Farajpoor,Mohammadreza Narimani
类目: Geophysics (physics.geo-ph); Machine Learning (cs.LG); Image and Video Processing (eess.IV)
*备注: 20 pages, 13 figures, 4 tables. Data and derived products: this https URL . Replication code: this https URL
Abstract:Urban wildfire resilience depends on interactions among built form, vegetation condition, and extreme fire weather, yet city-scale risk models often overlook whether predictive skill transfers across neighborhoods. We developed a spatially validated GeoAI workflow for the January 2025 Palisades Fire, linking 12,081 CAL FIRE damage inspections to pre-fire Sentinel-2 vegetation indices, Landsat surface temperature, LANDFIRE fuels, terrain, and OpenStreetMap buildings and roads. Among 9,883 inspected residential structures, 5,566 were destroyed. Random cross-validation yielded ROC-AUC 0.92 for the integrated XGBoost model, but 1 km spatial block validation reduced performance to 0.75; logistic regression performed similarly and was better calibrated. Building count within 100 m was the strongest predictor, with destruction odds increasing 4.12-fold per standard deviation. Vegetation moisture and greenness showed opposing conditional associations: NDMI at 100-300 m was protective (OR 0.52), whereas NDVI at 30-100 m was positively associated with destruction after accounting for moisture (OR 1.74). Predictive information was concentrated at the 100-300 m neighborhood scale. A separate post-fire track mapped burn severity and vegetation recovery without leakage. The results support neighborhood-scale susceptibility screening, moisture-aware vegetation management, and spatial block validation as a minimum standard for single-event urban wildfire modeling.
[LG-121] Sharp Barron Regularity Results for Coulombic Many-Electron Wave Functions
链接: https://arxiv.org/abs/2608.22252
作者: Pingbing Ming,Hao Yu
类目: Analysis of PDEs (math.AP); Machine Learning (cs.LG); Numerical Analysis (math.NA)
*备注:
Abstract:We establish sharp Barron regularity for Coulombic many-electron wave functions after extraction of the universal cut-off Jastrow factors. Following the factorization of Fournais et al.~\cite[Definition~1.4]FournaisEtAl2005, for a Coulombic eigenfunction \psi we define the successive quotients by [ \phi=e^-F_2,\mathrmcut\psi \quad\textand\quad \phi_3=e^-F_3,\mathrmcut\phi=e^-(F_2,\mathrmcut+F_3,\mathrmcut)\psi. ] Then [ \phi,\phi_3\in\mathcalB^s(\mathbbR^3N) \qquad\textfor every s2. ] This range is optimal among universal factorizations. No factor depending only on the particle number and the nuclear data, but not on the eigenfunction or its eigenvalue, can make every corresponding quotient belong to \mathcalB^2 . We also determine the exact endpoint growth. Writing \varepsilon=2-s , we prove that, for either u=\phi or u=\phi_3 , there is a computable constant M independent of \varepsilon such that [ \left|u\right|\mathcalB^2-\varepsilon\leq\fracM\varepsilon^2\left|u\right|\mathcalB^1. ] For the unperturbed two-electron atom we prove, with a constant independent of \varepsilon , [ \left|\left|\phi_3\right|_\mathcalB^2-\varepsilon-\frac32\pi Z\lvert\phi_3(0,0)\rvert\varepsilon^2\right|\leq\fracC\varepsilon. ] Hence the quadratic rate in the upper bound is sharp whenever \lvert\phi_3(0,0)\rvert\neq0 , as is the case for the ground state. Subjects: Analysis of PDEs (math.AP); Machine Learning (cs.LG); Numerical Analysis (math.NA) Cite as: arXiv:2608.22252 [math.AP] (or arXiv:2608.22252v1 [math.AP] for this version) https://doi.org/10.48550/arXiv.2608.22252 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-122] oken-Level Likelihood-Array Regression for Membership Inference and AI-Generated Text Detection
链接: https://arxiv.org/abs/2608.22179
作者: Jiajun Sun,Zhanrui Cai
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME)
*备注:
Abstract:Membership inference asks whether a text was used to train a language model, whereas AI-generated text detection asks whether it was generated by a language model rather than written by a human. Existing likelihood-based methods typically compress token-level probabilities into a few prespecified scores, most often using only probabilities conditioned on the full preceding context. We propose likelihood-array regression (LAR), which evaluates each target token under nested left-context windows and organizes the resulting likelihood-derived features into a structured array. After aligning arrays across texts of different lengths, LAR learns how detection information varies with context scale, token position, and likelihood features. LAR-1 aggregates learned contributions from individual aligned cells, while LAR-2 adds second-order features formed from pairs of evaluations of the same target token across context lengths. For within-path quadratic model, we establish matching minimax lower and upper bounds, characterize errors from finite-dimensional approximation and random squared projections, and derive conditions under which an oracle spectral sieve attains the minimax rate. Across multiple scoring language models, LAR substantially improves membership inference and AI-generated text detection over likelihood-based baselines. The analyses further show that shorter-context likelihoods contain information beyond conventional full-context probabilities, while second-order features provide additional gains for membership inference.
[LG-123] Cross-Temperature Defect Identification in Atomistic Simulations via Multi-Level Domain Alignment
链接: https://arxiv.org/abs/2608.22074
作者: Yating Fang,Jungmin Kim,Qian Qian Zhao,Pallavi Biswas,Joshua M. Gonjon,Ryan B. Sills,Ahmed Aziz Ezzat
类目: Materials Science (cond-mat.mtrl-sci); Machine Learning (cs.LG); Computational Physics (physics.comp-ph)
*备注: 46 pages, 8 figures
Abstract:Identifying atomic defects at elevated temperature is difficult because thermal fluctuations blur the local symmetry that both geometric heuristics and supervised classifiers rely on: trustworthy labels exist in low-temperature reference configurations, while the high-temperature regime where robust analysis matters most is effectively unlabeled. We cast this as a cross-temperature domain-shift problem and align the two domains at three levels: an equivariant denoiser at the input level, cross-temperature contrastive learning at the representation level, and a morphology-aware regularizer that steers predictions toward the compact geometry of physical defect structures. Because no atom-wise truth exists at temperature, we further introduce a label-free evaluation suite that scores predicted defect structures along five spatial and physics-based axes, enabling model assessment and selection without high-temperature labels. Near the melting point, the framework identifies vacancies and self-interstitial atoms across face-centered-cubic, body-centered-cubic, and hexagonal-close-packed iron systems with every interstitial localized and zero false detections in every vacancy system against Wigner-Seitz ground truth, with no high-temperature labels used in training. It sustains this fidelity on a million-atom, 2.5 ns trajectory, resolving single vacancy hops and complete Frenkel-pair recombination, and captures grain-boundary phase transformations in aluminum bicrystals, distinguishing two nucleation modes. Multi-level domain alignment thus offers a practical, label-efficient route to temperature-robust structural analysis of large-scale molecular dynamics.
[LG-124] Structured Learning on Mapper Representations
链接: https://arxiv.org/abs/2608.22044
作者: George Babus,Farzana Nasrin
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 29 pages, 10 figures, and 7 tables
Abstract:Modern machine learning (ML) methods are highly effective for prediction tasks, but many commonly used representations reduce complex data to fixed dimensional embeddings that may suppress multiscale structural organization. The Mapper algorithm from topological data analysis (TDA) provides a different perspective by decomposing data into overlapping local regions connected through a nerve construction, producing a structured representation that captures geometric organization, local statistical behavior, and relational connectivity simultaneously. In this work, we develop a framework for learning over Mapper induced structured representations. Rather than treating Mapper as a preprocessing step that produces a graph for downstream learning, we treat the full Mapper construction as part of the representation itself. We study mathematical properties of these representations, including invariance under relabeling, a distance functional on the space of Mapper representations, structural complexity of multiscale decompositions, and learning oriented stability under representation perturbations. Experiments on time series and graph classification datasets validate the proposed framework through controlled studies of representation ablation, Mapper parameter sensitivity, and the geometry of the induced representation space. Together, these results demonstrate how the proposed mathematical framework enables systematic comparison, interpretation, and analysis of Mapper representations, providing practical tools for studying representation geometry, structural complexity, and learning stability in learning tasks.
[LG-125] Physics-Constrained Neural Flow Maps for Long-Horizon Prediction of Spin Dynamics
链接: https://arxiv.org/abs/2608.22006
作者: Haoen Feng,Shenglan Yuan,Shirong Lin
类目: Mesoscale and Nanoscale Physics (cond-mat.mes-hall); Machine Learning (cs.LG)
*备注:
Abstract:Conventional simulation of current-driven magnetization relies on fine-step integration of the spin-transfer-torque Landau–Lifshitz–Gilbert equation, creating a computational bottleneck in parameter sweeps and control searches. In this work, we propose a physics-constrained neural flow map that learns finite-time dynamics directly on the unit sphere. The model maps the current magnetization, spin-torque strength, and requested time span to a future state in a single forward pass. Tangent-space projection and spherical retraction preserve unit magnetization during recursive, composition-consistent rollout. We validate the framework on single-spin trajectories under in-domain torques and previously unseen but stronger drive. Beyond the training horizon, it achieves an in-domain root mean square error of 0.00425 with norm drift at the 10^-7 level. The flow outperforms an adapted Long Short-Term Memory (LSTM) in in-domain accuracy and geometric stability, although the LSTM retains slightly lower out-of-distribution state error. The resulting geometry-preserving propagator reduces reliance on fine-step integration and enables physically admissible long-horizon prediction.
[LG-126] PhysECD: A Physics-Constrained E(3)-Equivariant Framework for Electronic Circular Dichroism Spectrum Prediction
链接: https://arxiv.org/abs/2608.21892
作者: Yi Jiang,Letian Chen,Runhan Shi,Liangzhaoxuan Han,Tong Zhu,Yang Yang
类目: Chemical Physics (physics.chem-ph); Machine Learning (cs.LG)
*备注:
Abstract:The electronic circular dichroism (ECD) spectrum is a primary experimental probe for assigning the absolute configuration of chiral molecules, yet interpreting a measured spectrum requires time-dependent density functional theory (TDDFT) calculations that can cost hours per molecule and must be repeated for every candidate stereoisomer and conformation. We present PhysECD, a physics-constrained, parity-aware E(3)-equivariant framework that bypasses computationally expensive TDDFT and predicts ECD spectra directly from the 3D structure of an individual conformer. Instead of regressing the spectrum as an opaque sequence, PhysECD predicts the physical quantities that generate it: per-state excitation energies and electric and magnetic transition dipoles. These quantities determine the rotatory strength R – the dot product of the two dipoles, a pseudoscalar that reverses sign under mirror reflection – and yield the final spectrum through a differentiable Gaussian-broadening formula derived from the underlying physics. The parity structure of the equivariant features guarantees the correct chiroptical symmetry: reflecting a molecule exactly negates the predicted spectrum. On the CMCDS dataset, PhysECD attains a per-molecule spectral Pearson correlation of 0.642 (mean) / 0.822 (median), substantially exceeding prior learned predictors while remaining physically interpretable. Experiments across multiple backbones further show that the framework is backbone-agnostic, paving the way for real-time assignment of absolute configuration.
[LG-127] First-Principles Atomistic Structure and Dynamics of Polyethylene During High-Pressure Radical Polymerization via Machine Learning Force Fields
链接: https://arxiv.org/abs/2608.21741
作者: Bharatha K. Gunawardana,Teresa Shah,Bicha Azizova,Deepa Ranabhat,Yizhi Song,Akshath Shastri,Srinjoy Ghose,Thomas E. Gartner III,Hsin-Yu Ko
类目: Materials Science (cond-mat.mtrl-sci); Disordered Systems and Neural Networks (cond-mat.dis-nn); Machine Learning (cs.LG); Chemical Physics (physics.chem-ph)
*备注: 15 pages, 7 figures, and 1 table
Abstract:Polyethylene (PE) is one of the most commonly used synthetic polymers. While the synthesis and processing protocols for PE are well established, precise experimental assignment of microscopic structures at atomistic resolution (i.e., the position of each atom) remains largely limited to highly crystalline systems. This gap is often addressed via computer simulations using empirical interatomic potentials, which use approximate but efficient descriptions of interatomic interactions to reach the length and time scales needed to describe macromolecules. These empirical potentials typically perform well for bulk and/or collective properties but face challenges with chemical realism for complex systems, e.g., during reactive processes. In this work, we address this challenge by combining the computational efficiency of a deep potential (DP) machine-learning force field and the chemical realism of first-principles van der Waals (vdW) corrected hybrid density functional theory (DFT) enabled by a SeA high-throughput framework. Using this approach, we study the structure and dynamics of PE oligomers and polymers in an ethylene solvent under common high-pressure (supercritical) radical polymerization conditions. We found that the local solvation environment of radical-containing PE oligomers converges for chain lengths greater than (n~6), suggesting extensibility of our oligomer-trained MLFF to significantly longer polymers. We then confirmed the extensibility of these models to long PE chains by characterizing the molecular weight scaling of single-chain structure and dynamics, which showed classic good solvent behavior. Our PE MLFF retained a consistent level of fidelity and stability across a wide range of thermodynamic state points and chain lengths, at full atomistic resolution, therefore paving the way towards first-principles-based polymer structure and property prediction.
[LG-128] Guidance for Prior Change via Density Ratio Estimation
链接: https://arxiv.org/abs/2608.21729
作者: Yichen Zang,Song Liu,Jiun-Yi Lin
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME)
*备注:
Abstract:Simulation-Based Inference (SBI) serves as a vital framework for parameter inference in scientific fields where simulators involve intractable likelihoods, yet while amortized generative models offer rapid posterior estimation, they are often restricted by the specific priors used during training, thereby limiting their flexibility as prior knowledge evolves. To address this prior dependency, PriorGuide was introduced as an inference-time guidance method, but due to its intractable formulation, it relies on Gaussian approximations of the reverse transition kernel and Gaussian mixture model fitting for the prior ratio, both of which introduce systematic bias. Motivated by these limitations, we propose an unbiased test-time guidance framework that leverages Density Ratio Estimation (DRE) to learn a score guidance term, effectively decoupling the inference process from the prior training. Moreover, our framework remains agnostic to the specific density ratio estimators, making it a general and flexible framework for handling prior changes. Experimental results across multiple tasks demonstrate that our method matches or outperforms PriorGuide on C2ST and MMD in most tasks while maintaining robustness even under limited overlap between the training and target priors. Furthermore, we apply our method to Bayesian updating for parameter inference from planetary light-curve data, where it also demonstrates strong effectiveness and robustness. Code is available at this https URL .
[LG-129] Piecewise Linear Equivariant Maps for Compact Groups
链接: https://arxiv.org/abs/2608.21645
作者: Valeriano Aiello
类目: Representation Theory (math.RT); Machine Learning (cs.LG)
*备注:
Abstract:Motivated by equivariant neural networks, we study piecewise linear equivariant maps between finite-dimensional real representations of compact groups. We show that all genuinely non-linear piecewise linear behaviour is confined to the subspaces on which the identity component of the group acts trivially, while equivariance forces linearity on the corresponding orthogonal complements. As a consequence, we obtain a compact-group analogue of the finite-group existence criterion of Gibson–Tubbenhauer–Williamson for non-zero equivariant piecewise linear maps between irreducible representations, with the identity component giving rise to a rigidity phenomenon absent from the finite-group case.
[LG-130] Separating Voice from Age in COPD Screening
链接: https://arxiv.org/abs/2608.21599
作者: George P. Kafentzis,Nikoletta Arvaniti
类目: Audio and Speech Processing (eess.AS); Machine Learning (cs.LG); Sound (cs.SD); Signal Processing (eess.SP)
*备注:
Abstract:Voice has been proposed as a low-cost screening signal for chronic obstructive pulmonary disease (COPD). COPD is strongly age-associated and voice changes with age, thus such results admit a trivial alternative explanation. We re-evaluate a public sustained-phonation corpus ( 1246 recordings, 68 participants) under a strictly participant-level protocol. We therefore evaluate on repeatedly drawn age-matched cohorts and report the discrimination achieved by the confounders themselves on those same cohorts. Where raw (unmodelled) age ( 0.510 [0.469, 0.551] ) and raw gender ( 0.479 ) are both measured at chance, acoustic models excluding age retain ROC-AUC 0.717 [0.552, 0.859] and average precision 0.747 [0.581, 0.892] against a one-to-one baseline of 0.5 , whereas models containing age fall to 0.531 – 0.679 . The separation is reproduced by two further learners with fixed hyperparameters. Two findings have broader methodological implications: models trained with age transfer less effectively to an age-balanced target cohort than otherwise identical models trained without age, and fourteen classical voice-quality and perturbation measures achieve comparable discrimination to a 55 -dimensional combined representation. We conclude that a non-age acoustic signal is present, that confounding by recording conditions cannot be excluded from the released features, and that the evaluation protocol in standard use cannot distinguish these possibilities.
[LG-131] Random Hazard Forests
链接: https://arxiv.org/abs/2608.21597
作者: Hemant Ishwaran,Eileen M. Hsich,Udaya B. Kogalur,Donald K.K. Lee
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME)
*备注:
Abstract:Clinical data sources such as electronic health records and wearable sensors record patient status repeatedly over follow-up, often at irregular times and on different schedules for different measurements. These data create opportunities for continuously updated, individualized risk prediction. Existing approaches, however, often simplify the temporal structure before modeling it. We introduce Random Hazard Forests (RHF), a survival tree ensemble that learns how a patient’s hazard changes in continuous time as new measurements become available. RHF formulates the estimation problem directly through a nonparametric hazard likelihood for predictable covariate processes. An efficient working model guides tree construction, after which flexible time-varying hazards are estimated for each terminal node. Given any predictable covariate path, each tree follows the path through its terminal nodes over time and assembles the corresponding node-level hazards into a trajectory. Averaging these trajectories across trees yields the RHF pathwise hazard estimate. Because routing at each time uses only the covariate state available immediately beforehand, RHF accommodates internal longitudinal covariates without lookahead. Simulations and an intensive care application show that RHF accurately estimates changing risk under irregular and asynchronous covariate updates.
[LG-132] Neural Network Field Theory at Finite Width
链接: https://arxiv.org/abs/2608.21588
作者: Christian Ferko,Aaron Mutchler
类目: High Energy Physics - Theory (hep-th); Disordered Systems and Neural Networks (cond-mat.dis-nn); Machine Learning (cs.LG)
*备注: 37 pages
Abstract:Under mild assumptions, any quantum mechanical (QM) model or quantum field theory (QFT) admits a representation in terms of an ensemble of neural networks with countably many random parameters. We investigate the features of NN-QM and NN-FT models with finitely many parameters, such as a feedforward network of width N \infty . We find that, generically, such models must violate one of the properties of conventional Euclidean QFTs, such as reflection positivity or cluster decomposition. We present several complementary ways of understanding which features can and cannot be preserved at finite N , both in QM and in QFT.
[LG-133] Sparse Separable Factor Analysis in the Complex Domain with an Application to Local Field Potential Data
链接: https://arxiv.org/abs/2608.21551
作者: Ian Hultman,Kirtikanth Kalapatapu,Yassine Filali,Rainbo Hultman,Sanvesh Srivastava
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Computation (stat.CO); Methodology (stat.ME)
*备注: 50 pages, 9 figures, and 7 tables
Abstract:Complex-valued arrays arise in signal processing, where scientific interpretation depends on retaining amplitude and phase information. Existing covariance estimation methods either ignore the multiway organization of such data or rely on real-domain embeddings that do not directly exploit their complex structure. We develop sparse separable factor analysis (SSFA), a latent factor model for complex-valued arrays with a separable covariance structure across modes. Each mode-specific covariance matrix is modeled through a low-rank Hermitian factor structure and a diagonal residual covariance matrix. To obtain interpretable estimates, we impose elementwise lasso penalties on the complex loading matrices and estimate the SSFA parameters using a mode-wise parameter-expanded expectation-maximization procedure. The resulting loading updates admit closed-form complex soft-thresholding solutions, which shrink the modulus of each loading while preserving its phase. A separate balancing step resolves the scale nonidentifiability of the separable covariance structure. Simulation studies show that SSFA improves covariance estimation relative to vectorization-based methods, including complex principal component analysis. We apply SSFA to local field potential recordings from mice, where we compare separability structures induced by different groupings of brain region, frequency, and time and perform model-based imputation of recordings missing because of electrode misplacement.
[LG-134] What Neural Network Field Theory Can and Cannot Realise on a Computer
链接: https://arxiv.org/abs/2608.21523
作者: Thomas R. Harvey
类目: High Energy Physics - Theory (hep-th); Machine Learning (cs.LG); High Energy Physics - Lattice (hep-lat); High Energy Physics - Phenomenology (hep-ph); Mathematical Physics (math-ph)
*备注:
Abstract:One aim of neural network field theory is to put a quantum or effective field theory on a computer, with the network ensemble itself as the theory. We ask how far that aim can be pushed for a function class regular enough to be computed with. Our main result is a no-go theorem with assumptions that hold for standard network architectures. We use it to separate four versions of neural network field theory, according to whether the defining object is the finite width ensemble or its infinite width limit, and whether the target we want to compute is a quantum or an effective field theory. Neither finite width interpretation is straightforwardly consistent. For finite width ensembles with finite variance at each point, the QFT interpretation fails reflection positivity, while the EFT interpretation establishes no scale separation by which the positivity violation can be placed outside its domain of validity. Of the two limit versions, one can be simulated in full and the other only in part, as only its smeared correlators are computable with a controlled error. As such, at the level of a controlled numerical computation, the QFT and EFT versions cannot be distinguished. One dimension escapes the obstruction, yet reflection positivity is shown to still fail there at every finite width for the cosine network. Two escapes from the theorem remain, giving up either finite variance at a point or exact rotation invariance, and we discuss both of these possibilities.
[LG-135] A Data-Driven Approach to State Construction in Markov Models
链接: https://arxiv.org/abs/2608.21480
作者: Linde Van Gestel,Marie-Anne Guerry,Evy Rombaut
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Probability (math.PR); Methodology (stat.ME)
*备注:
Abstract:A Markov chain is a widely used stochastic process modelling random events over time. These models are built on subsets of the entire dataset, referred to as states, which are considered to be homogeneous regarding transition probabilities. However, the creation of these states is often disregarded or based on prior assumption, potentially violating the homogeneity requirement and thus decreasing the validity and predictive power of the model. In order to fill this gap, this paper combines supervised feature selection with unsupervised learning techniques for data-driven state construction. Density-based clustering, spectral clustering, and Kohonen self-organizing maps are examined for their ability to identify latent groups without prior assumptions. The contribution of this study is twofold. First, the paper presents a methodological framework for state construction incorporating suitable unsupervised learning techniques, with appropriate measures both for classification performance and Markov model accuracy. Secondly, the framework is tested on an application, resulting in a comparative analysis showing that spectral clustering and Kohonen self-organizing maps are best at capturing inherent structure. These results serve as a cornerstone in providing theoretical and methodological guidance for improving state definition in applied Markov modelling.
[LG-136] Magnitude Homology Is the Associated Graded of the Length Filtration
链接: https://arxiv.org/abs/2608.21479
作者: Luciano Melodia
类目: Algebraic Topology (math.AT); Computational Geometry (cs.CG); Machine Learning (cs.LG); Logic in Computer Science (cs.LO)
*备注: Code available at this https URL
Abstract:Magnitude homology is graded by length and knows nothing of persistence. Its persistent refinement knows nothing of where its bars begin and end. We show that the two are one construction: filtering the length nerve by sublevel sets of the length yields the persistence module, and the associated graded of that filtration is the magnitude complex. A long exact sequence exchanges them, and each side gains what it lacked. Magnitude homology locates the critical values of the barcode, so a graded computation lists the lengths at which an endpoint can occur, and the barcode acquires a stability estimate of (n+1)\delta in degree n under a perturbation of size \delta , while a computed perturbation moves a barcode by more than \delta , so the factor cannot be dropped. We apply this to quantitative equational theories, whose free algebras are metric spaces built from syntax: an inclusion of theories induces a morphism of the presenting monads and a comparison of barcodes with an explicit bound, so the invariant measures axiomatic strength. Four examples are computed, one in every degree.
[LG-137] Gauss–Hermite Quadrature for Gaussian-Mixture Entropy with an Action-Space Hermite Surrogate
链接: https://arxiv.org/abs/2608.21467
作者: Jae Wan Shim
类目: Machine Learning (stat.ML); Information Theory (cs.IT); Machine Learning (cs.LG)
*备注:
Abstract:Gaussian distributions are used to model uncertainty in signals and states, and Gaussian mixtures are often used when the underlying distribution is multimodal. Unlike a single Gaussian, a Gaussian mixture generally has no closed-form expression for differential entropy and therefore requires numerical approximation. We propose a Gauss–Hermite quadrature method for evaluating Gaussian mixture differential entropy. The quadrature order controls the numerical resolution of the approximation. The method is evaluated on one- and two-dimensional Gaussian mixture benchmarks against Taylor approximations, analytic entropy bounds, and numerical integration references. For repeated optimization over continuous actions, we also propose a Hermite polynomial surrogate in action space. In a radar pointing benchmark, its second-order form achieves substantially lower surrogate error and optimizer regret than a second-order Taylor surrogate based on local derivatives at the nominal action, while both methods use nine direct objective evaluations per replanning step. The Hermite surrogate also improves pointing performance in the tested benchmark. Subjects: Machine Learning (stat.ML); Information Theory (cs.IT); Machine Learning (cs.LG) Cite as: arXiv:2608.21467 [stat.ML] (or arXiv:2608.21467v1 [stat.ML] for this version) https://doi.org/10.48550/arXiv.2608.21467 Focus to learn more arXiv-issued DOI via DataCite
[LG-138] Spectral partitioning for k-block averag ing kernels of finite Markov chains
链接: https://arxiv.org/abs/2608.21466
作者: Michael C.H. Choi,Youjia Wang
类目: Machine Learning (stat.ML); Information Theory (cs.IT); Machine Learning (cs.LG); Optimization and Control (math.OC); Probability (math.PR); Computation (stat.CO)
*备注: 43 pages, 8 figures
Abstract:We develop spectral algorithms for selecting state-space partitions that define averaging kernels for finite, ergodic and reversible Markov chains. For a partition \mathcal O , the Gibbs kernel G_\mathcal O resamples within the current block from the stationary conditional distribution; when this update is tractable, composing or mixing it with a baseline kernel P can accelerate convergence. We select \mathcal O by rounding the bottom nonconstant eigenfunctions of P^2 , or the algebraically smallest eigenfunctions of P for additive mixtures, using weighted k -means. For F(\mathcal O)=|G_\mathcal OP-\Pi|_F,\pi^2 , we derive exact trace and normalized-cut representations and show that F equals the Pearson \chi^2 -mutual information between the initial block label and the state after one transition, giving this matrix objective a natural probabilistic interpretation. In the two-block case, a threshold sweep exactly solves the associated one-dimensional weighted two-means rounding problem. For general k \geq 2 , weighted k -means rounds the bottom (k-1) -dimensional embedding, after which candidates are rescored by F ; the rounding distortion is a distance between subspaces that yields spectral approximation bounds. We extend the framework to additive mixtures, finite-horizon objectives, and discounted infinite-horizon objectives. In contrast to classical normalized spectral clustering, which uses top nonconstant modes to find low-flow persistent clusters, our method uses bottom modes to favor large normalized cross-block flow and rapid loss of block-label information. Experiments on a controlled-spectrum graph, a mean-field Ising model, and Bayesian variable selection show notable per-iteration improvements in convergence and statistical estimation.
[LG-139] DNA Methylation Profiling in Melanoma: From Lesion Classification to Therapeutic Stratification
链接: https://arxiv.org/abs/2608.21448
作者: Jana T. Winterstein,Lukas Heinlein,Günter Raddatz,Carina Nogueira Garcia,Sarah Haggenmüller,Christoph Wies,Lucas Schneider,Annemarie Hoffsommer,Tim J. Zeuner,Friedegund Meier,Sarah Hobelsberger,Frank F. Gellrich,Mildred Sergon,Axel Hauschild,Lucie Heinzerling,Justin G. Schlager,Kamran Ghoreschi,Max Schlaak,Franz J. Hilke,Carola Berking,Markus V. Heppt,Michael Erdmann,Sebastian Haferkamp,Konstantin Drexler,Dirk Schadendorf,Wiebke Sondermann,Matthias Goebeler,Bastian Schilling,Daniel B. Lipka,Stefan Fröhling,Felix Sahm,Jakob N. Kather,Yuri Tolkach,Jochen S. Utikal,Benjamin Izar,Yevgeniy R. Semenov,Titus J. Brinker
类目: Genomics (q-bio.GN); Machine Learning (cs.LG)
*备注:
Abstract:DNA methylation provides a stable record of cellular identity, capturing epigenetic programs that distinguish specialized cell states despite a shared genome. Because malignant transformation and tumour progression are accompanied by extensive epigenetic remodeling, we hypothesized that the methylome of melanocytic lesions contains biologically and clinically relevant information for both diagnosis and disease progression. In a cohort of 1,001 tissue samples prospectively collected across eight German university hospitals profiled using Illumina Infinium MethylationEPIC arrays, we compared machine-learning models based on selected Cytosine phosphate Guanine (CpG) methylation sites with models incorporating biology-guided features, including epigenetic age acceleration, cell type composition and copy-number variation burden. In an external test set, the best diagnostic classifier was CpG-based and distinguished melanocytic nevi, noninvasive melanoma and invasive melanoma with a macro-averaged area under the receiver operating characteristic curve of 0.919 (95% CI: 0.878 to 0.952). Notably, across CpGs most strongly hyper- and hypomethylated between NV and IM, NIM showed an intermediate methylation profile, providing a molecular correlate of its diagnostic complexity. The best model for clinically relevant treatment group prediction, with AJCC stages grouped according to guideline-based management recommendations, relied on biology-guided features and achieved a macro-averaged mean absolute error of 0.627 (95% CI: 0.477 to 0.808). Together, these findings demonstrate that methylation-based models can capture both diagnostic identity and clinically relevant disease stratification, supporting DNA methylation as a promising biomarker for further validation and potential clinical translation.
附件下载
