本篇博文主要内容为 2026-09-11 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 (2026-09-11)
今日共更新662篇论文,其中:
- 自然语言处理共91篇(Computation and Language (cs.CL))
- 人工智能共171篇(Artificial Intelligence (cs.AI))
- 计算机视觉共89篇(Computer Vision and Pattern Recognition (cs.CV))
- 机器学习共171篇(Machine Learning (cs.LG))
- 多智能体系统共10篇(Multiagent Systems (cs.MA))
- 信息检索共15篇(Information Retrieval (cs.IR))
- 人机交互共24篇(Human-Computer Interaction (cs.HC))
多智能体系统
[MA-0] runcated Noisy Best-Response Algorithms: Toward Game Theoretic Learning with Safety Guarantees
【速读】：该论文旨在解决多智能体协同优化中具有子模最大化目标的协调问题。此类问题中，纳什均衡（Nash equilibrium）虽能保证近似解在最优值的50%以内，但达到该最坏情况界（worst-case bound）的均衡状态往往缺乏稳定性。为利用这一不稳定性，论文提出一类名为“截断噪声最佳响应”（Truncated Noisy Best-Response, TNBR）的算法，其核心机制是允许智能体异步、随机地从自身最佳响应收益的邻域中采样动作。通过分析TNBR算法所对应的马尔可夫链的遍历类（recurrent classes），研究建立了两类关键理论边界：其一是“性能”（Performance）边界，确保算法始终存在高价值的遍历状态；其二是“安全”（Safety）边界，确保算法不会陷入任意差的遍历状态。这两类边界之间存在类似“水床效应”的权衡关系——所有安全性能较差的游戏必然对应更优的性能保障，从而在理论上实现了对算法稳定性和解质量的双重控制。
链接: https://arxiv.org/abs/2609.11863
作者: Vartika Singh,Philip N. Brown
机构: 未知
类目: Computer Science and Game Theory (cs.GT); Multiagent Systems (cs.MA); Systems and Control (eess.SY)
备注:
Abstract:We consider a game theoretic approach to solve multi-agent coordination problems with submodular maximization objectives. It is known for such problems that the Nash equilibria for the corresponding game are always within 50% of the optimal, but that the equilibria which achieve this worst-case bound are not stable. To exploit this instability, we propose a family of algorithms which we call Truncated Noisy Best-Response (TNBR) Algorithms. These algorithms are flexibly characterized by agents asynchronously and stochastically selecting actions from a neighbourhood of their best response payoffs. We compute bounds on the recurrent classes of TNBR algorithms’ associated Markov chains. Our bounds fall into two categories: first, “Performance” bounds ensure that TNBR algorithms always have a high-value recurrent state; second, “Safety” bounds ensure that TNBR algorithms never have arbitrarily-bad recurrent states. Furthermore, these two types of bounds are linked by a waterbed-like effect: every game with a poor Safety guarantee necessarily has a favorable Performance guarantee.
[MA-1] ORCH: Organizational Principles Enable Collective Intelligence in Embodied AI
【速读】：该论文旨在解决人工多智能体系统在面对动态、复杂任务时，因采用固定组织结构而导致协调效率低下这一核心问题。现有系统通常使用静态的组织架构，难以适应物理任务中不断变化的协同需求，从而限制了整体性能。其解决方案的关键在于引入基于人类组织理论的可编程组织原则——ORCH（Organizing Roles and Coordination Hierarchies），通过融合并行依赖（pooled interdependence）与顺序依赖（sequential interdependence）机制，构建任务特异性、层级化的智能体组织结构。该方法使异构智能体团队能够在保持专业化小组内并行作业的同时，实现跨阶段任务的有序衔接，显著提升任务完成度、执行效率、探索覆盖率及资源利用效能。实验结果表明，由人类设计的ORCH组织相比四种基准方法平均提升最终得分63.97%、执行效率74.29%；而由大语言模型自动生成的组织也实现了43.63%和52.53%的改进，且优势在不同任务类型和语言模型间具有鲁棒性，证明了层级化组织对长周期任务中集体智能维持的关键作用。
链接: https://arxiv.org/abs/2609.11737
作者: Zhengran Ji,Jonathan Hyun,Boyuan Chen
机构: Duke University (杜克大学)
类目: Multiagent Systems (cs.MA); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Robotics (cs.RO)
备注:
Abstract:Collective intelligence depends not only on the capabilities of individual members, but also on how those members are organized. Yet artificial multi-agent systems are typically assembled using fixed organizational structures, even when the physical tasks they perform impose fundamentally different coordination requirements. Here we show that principles from human organization theory can be operationalized to organize large, heterogeneous collectives of embodied artificial agents. We introduce ORCH (Organizing Roles and Coordination Hierarchies), which constructs task-specific hierarchical organizations by combining pooled interdependence for work that can proceed concurrently with sequential interdependence for work governed by prerequisite relationships. Across 25 wildfire-response missions spanning reconnaissance, rescue, transportation, resource management, containment and suppression, we evaluated teams of up to 50 heterogeneous agents using eight large language models. Organizations constructed using these principles consistently outperformed four representative embodied multi-agent approaches across mission outcome, execution efficiency, exploration and computational resource use. Human-designed ORCH organizations improved final score by 63.97% and execution efficiency by 74.29% on average relative to the four prior frameworks. Organizations generated automatically by language models improved these measures by 43.63% and 52.53%, respectively. These advantages persisted across missions and underlying language models. Notably, collective performance was not monotonically determined by model scale. Analysis of long-horizon missions showed that hierarchical organization enabled teams to preserve concurrent activity within specialized groups while coordinating ordered transitions between mission phases.
[MA-2] When Agents Disagree: Bayesian Backward Reasoning as a Label-Free Anchor for Multi-Agent Collective Decision-Making
【速读】：该论文旨在解决多大语言模型（LLM）智能体在面对同一问题时产生冲突答案时，如何有效进行集体决策的问题。现有方法如投票、选举规则和基于LLM的评判器均依赖于正向推理（forward reasoning），即从证据直接映射到标签，这种单向建模方式虽能整合多样化的推理路径，但其聚合结果仍受共同的“证据-标签”因子化结构影响，容易继承正向推理池中的相关性误差。为此，本文提出通过贝叶斯逆向推理构建每个实例的反向后验（reverse posterior），与传统的正向后验形成不同因子化的近似后验分布。由于不同因子化路径的估计倾向于较少共享相同错误，本文利用Jensen-Shannon散度衡量智能体间的跨路径一致性，并以此为信号设计三种策略：硬选择（MinJS）、软重加权（FwdJS）和对数线性融合（LogLin）。实验在包含五种不同LLM主干的DDXPlus数据集上验证表明，所提方法表现稳定提升：MinJS优于随机选择，FwdJS普遍超越最强基线，LogLin在所有方法中取得最佳性能，尤其在智能体分歧较大的子集上增益显著。尽管反向后验自身独立预测能力较弱，但其作为互补锚点，显著增强了集体决策的有效性；当有标注数据可用时，引入轻量级两阶段校准可进一步优化反向锚点，提升聚合效果。
链接: https://arxiv.org/abs/2609.11709
作者: Ken Chen,Wei Wang,Sachith Seneviratne,Hansani Weeratunge,Saman Halgamuge
机构: The University of Melbourne(墨尔本大学); Sri Lanka Institute of Information Technology(斯里兰卡信息技术学院)
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注:
Abstract:When multiple LLM agents yield conflicting answers, the decision-making process dictates whether agent diversity improves performance or merely compounds shared errors. Existing collective decision-making methods, including voting, electoral rules, and LLM judges, rely on forward reasoning: they map evidence to labels in one direction. Although these methods can combine diverse forward traces, they still aggregate estimates that share this evidence-to-label factorization and can inherit correlated errors within the forward pool. We therefore construct a reverse posterior for each instance through Bayesian backward reasoning from an explicit likelihood. The forward and reverse posteriors provide differently factorized approximations of the underlying posterior. Because estimates from different factorizations may tend to share the same error less often, we use Jensen-Shannon divergence to rank agents by cross-path consistency. This cross-path consistency signal underlies three strategies: hard selection (MinJS), soft reweighting (FwdJS), and log-linear fusion (LogLin). Evaluated on DDXPlus across five LLM backbones, our proposed strategies show consistent improvements: MinJS outperforms random selection across all backbones, FwdJS generally improves over the strongest baseline, and LogLin achieves the best performance among the evaluated methods, with its largest gains on the subset where the agents disagree. Despite its weaker standalone accuracy, the reverse posterior serves as a more useful anchor than forward-only alternatives, providing complementary information for collective decision-making. When labeled data are available, a lightweight two-stage calibration can further refine the reverse anchor and improve aggregation performance.
[MA-3] From Document Silos to Process Intelligence: A Multi-Layer Knowledge Graph for CMC Process Development
【速读】：该论文旨在解决药物研发与生产过程中化学、制造与控制（Chemistry, Manufacturing and Controls, CMC）信息在多阶段、知识密集型流程中因跨职能分散及格式异构导致的知识碎片化问题，进而引发技术转移与法规申报过程中的可追溯性缺失与知识管理成本高昂。其解决方案的核心在于构建一个模块化代理式人工智能（agentic-AI）平台，通过将异构的工艺开发文档转化为可查询的双层知识图谱：底层为基于“文档-章节-片段”层级结构的词汇知识层，实现对数字化、扫描件、手写体及多语言文档的无损集成；上层为智能层，利用溯源锚定的领域图谱提取与本体对齐的实体，并建立跨文档概念关联。大语言模型（Large Language Model, LLM）代理在两层间协同工作，根据问题类型动态选择最优检索路径。通过一项新颖的三层评估协议，在赛诺菲小分子项目505个问题（源自38份开发报告）上验证了系统的可靠性：第一层多选题准确率达95%，表明系统整体稳健；第二层由LLM裁判的通过率为85%，在比较类与全库范围问题中性能下降，揭示出仅靠第一层指标无法捕捉的失效模式。路由器代理依据问题类型自动分发至相应知识层，显著提升检索精准度。该研究不仅为未来代理式AI平台设计者提供了基于非公开数据集的评估范式，更预示着基于图谱的架构将在制药行业广泛应用，以实现从零散文档库向结构化工艺智能的转化。
链接: https://arxiv.org/abs/2609.11493
作者: Reza Amirmoshiri,Faryad Sahneh,Yasser Jangjou
机构: CMC Synthetics, Sanofi US; Accelerator Data science GBU, Sanofi US
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注:
Abstract:Chemistry, Manufacturing and Controls (CMC) process development generates an enormous body of technical information across a multi-stage, knowledge-intensive continuum from drug discovery to commercial manufacturing. This knowledge is traditionally fragmented across functions and heterogeneous formats, causing traceability gaps and significant knowledge-management costs during technology transfer and regulatory filing. We present a modular agentic-AI platform that converts a heterogeneous corpus of process-development documents into a queryable, dual-layer knowledge graph. A base knowledge layer builds a lexical graph with a Document-Section-Chunk hierarchy through lossless ingestion of digital, scanned, handwritten, and multilingual documents, while an intelligence layer extracts ontology-aligned entities and bridges cross-document concepts through a provenance-anchored domain graph. LLM agents operate across both layers, selecting the retrieval path best suited to each question. We evaluate the lexical layer with a novel three-tier protocol measuring the deployment-fidelity of a retrieval-augmented generation (RAG) system on proprietary data, demonstrated on 505 questions curated from 38 development reports of a Sanofi small-molecule program. Tier-1 multiple-choice accuracy of 95% signals strong platform reliability; the stricter Tier-2 LLM-judge pass rate of 85%, which degrades on comparative and corpus-wide questions, reveals a failure taxonomy that Tier-1 accuracy alone fails to capture. A router agent selects between layers according to question type. We anticipate this protocol will enable future designers of agentic platforms to assess their systems against nonpublic databases, and that graph-based architectures will see broader adoption in pharma as a means of transforming fragmented document repositories into structured process intelligence.
[MA-4] But How Would AI Agents Run a Towns Economy?
【速读】：该论文旨在解决传统多智能体经济模拟中普遍存在的“货币传导失效”问题，即在长期演化下，外部冲击（如旅游需求变化或现金转移）对宏观经济变量（如工资、价格、财富分配）的传导机制为何显著衰减甚至停滞。其核心发现是：尽管存在显著的外部冲击（如12倍游客需求激增使企业收入提升4.62倍），但货币流动在短期内迅速停滞——工资仅微幅上升（1.03倍，p=0.42），仅有0.3%的商品被重新定价（p=0.47），而96.7%的现金转移在311个时间步后仍被持有，边际消费倾向仅为3-4%，与零无异。这一现象揭示了经济系统在长期中表现出的“近冻结”状态，即财富不平等指标ρ从2周时的0.964下降至26周时的0.752，呈现出显著的时域依赖性，而此类动态无法通过短期实验观测到。解决方案的关键在于引入具备长期记忆能力的大语言模型（LLM）代理，并验证其记忆机制对系统行为的决定性作用：消融实验表明，更换底层大语言模型会显著改变所有测量结果（p=0.0039），而删除记忆则无显著影响；同时，仅使用社会性工具的代理在94–97%的情况下失败，远低于经济工具约96%的成功率，说明经济理性建模才是驱动有效货币传导的核心。研究通过双重验证机制（实时校验与离线交易历史回溯）确保数据可信，并公开全部仿真数据以支持可复现性。
链接: https://arxiv.org/abs/2609.11108
作者: Sajal Regmi,Siddhartha Pudasaini,Chetan Phakami Pun
机构: Karela Technologies Inc.(Karela Technologies 公司); USA
类目: Multiagent Systems (cs.MA); Emerging Technologies (cs.ET)
备注: 8 pages, 7 figures, 6 tables. Dataset and analysis code: this https URL
Abstract:We placed 100 memory-equipped large language model (LLM) agents in charge of a closed, money-conserving spatial economy on real Pokhara Lakeside geography (earning wages, running businesses, setting prices) and ran this multi-agent simulation for up to 26 simulated weeks, well past the 1-2 weeks typical of agent-society studies. Across 91 validated runs (2.44M agent decisions, 21.5B tokens), the money stops moving, in a specific and measurable way. A 12x tourist demand shock raises business revenue 4.62x ( p0.001 ), which we decompose exactly into a 1.50x extensive margin (more businesses trading) and a 3.07x intensive margin (more revenue each). Monetary transmission stops there. Wages move 1.03x ( p=0.42 ); 0.3% of 3,981 menu items are ever repriced ( p=0.47 ). A randomized cash transfer (NPR 5,000 to 20 of 100 agents) shows the same pattern from the opposite direction: 96.7% is still held 311 pulses later, marginal propensity to consume 3-4% by two independent measures, indistinguishable from zero. The wealth distribution is consequently near-frozen at the horizon this literature uses ( \rho=0.964 over 2 simulated weeks), but not frozen. \rho falls to 0.832 at 12 weeks and 0.752 at 26, a horizon-dependence no short study can see. Matched ablations show which knob actually matters. Swapping the backing LLM moves every outcome we measure ( p=0.0039 ); deleting agents’ memory moves none of them detectably. A purely social tool fails 94-97% of the time across two model families, compared with ~96% success on economic tools, with no measurable shift away from it. Every headline number is verified twice, by a live validator and by an offline recomputation that reconciles each agent’s wealth against its own signed transaction history, and we release the full run corpus for reanalysis.
[MA-5] Defining AI Agents : A Compendium of Criteria Metrics and Benchmarks
【速读】：该论文旨在解决人工智能（Artificial Intelligence, AI）领域中“智能体”（agent）概念缺乏统一定义的问题，这一模糊性导致了智能体研究在评估、比较及可复现性方面存在困难。其解决方案的关键在于提出一个基于五个核心维度的系统性框架——环境交互（environmental interaction）、学习与适应（learning and adaptation）、自主性（autonomy）、目标导向行为（goal-directed behavior）以及时间连贯性（temporal coherence），并针对每个维度梳理已有研究中的能力内涵，整合相应的度量指标、基准测试与评估范式。通过这一结构化综述，论文不仅揭示了当前智能体评估体系的成熟方向与薄弱环节，还发布了“智能体百科全书”（Agent Compendium）这一公开数字资源，以组织和扩展所识别的评估方法。该框架与资源共同构建了一个跨AI系统通用的智能体能力评估与比较基础，显著提升了研究的可复现性、沟通清晰度与系统性。
链接: https://arxiv.org/abs/2609.11018
作者: Mia Lassiter,Brinnae Bent
机构: Duke University(杜克大学)
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: Agent Compendium: this http URL
Abstract:The term agent in artificial intelligence lacks a standard definition, complicating the evaluation, comparison, and reproducibility of AI agent research. We address this ambiguity through a survey organized around five dimensions of agenticness: environmental interaction, learning and adaptation, autonomy, goal-directed behavior, and temporal coherence. For each dimension, we examine how the underlying capability has been conceptualized across prior work and synthesize the metrics, benchmarks, and evaluation frameworks used to assess it. This review provides a structured account of the current landscape of agent evaluation, highlighting both established approaches and areas where evaluation remains limited or inconsistent. We additionally introduce the Agent Compendium, a public-facing digital resource that organizes and extends the evaluation methods identified through this review. Together, the survey and compendium provide a common structure for evaluating and comparing agent capabilities across AI systems, supporting more reproducible research, clearer communication, and more systematic study of artificial agents.
[MA-6] Evaluating Scaffolding-Oriented Multi-Agent Large Language Model System for Clinical Interview Training
【速读】：该论文旨在解决传统标准化患者（Standardized Patient, SP）培训资源消耗大、难以规模化的问题，同时探索如何在临床教育中有效提升医学生在不确定情境下开展安全且连贯的患者访谈能力。其解决方案的关键在于构建一个以支架式教学（scaffolding-oriented）为核心的多智能体大型语言模型（Large Language Model, LLM）人工智能标准化患者（AI-SP）训练平台。该平台包含三个核心智能体：患者代理用于模拟真实对话，导师代理通过苏格拉底式提问引导学生思考而不透露诊断信息，以及逐回合评估代理在不提供总结性评分的前提下实时监测临床进展。在一项随机对照试验（N = 100名医学生）中，相较于对照组，多智能体支架条件显著提升了学生在沟通技巧、共情表达及特定病史采集行为方面的表现，尽管最终诊断准确率无显著差异，但整体临床过程质量得到增强。研究结果表明，基于多专家智能体设计的生成式AI系统能够有效优化模拟临床访谈的过程质量，而非人为夸大考核结果。为推动后续研究，作者公开了一个包含对话转录、检查清单标注、逐回合评估与OSCE对齐评分的多专家标注数据集，旨在支持以教学法为基础的人工智能标准化患者系统开发，并促进人工智能辅助临床推理训练的研究发展。
链接: https://arxiv.org/abs/2609.10939
作者: Luming Yang,Haoxian Liu,Siqing Li,Rong Jia,Yue Xiao,Guanhua Chen,Li Lu
机构: Guangzhou Medical University (广州医科大学)
类目: Multiagent Systems (cs.MA); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注:
Abstract:Clinical education must prepare medical students to conduct safe and coherent patient interviews under conditions of uncertainty. Traditional standardized patient (SP) training is resource-intensive and difficult to scale. We developed a scaffolding-oriented multi-agent Large Language Model (LLM) AI Standardized Patient (AI-SP) training platform1. The system includes a patient agent for simulated dialog, a tutor agent providing Socratic prompts without disclosing diagnostic information, and a turn-level evaluator agent that monitors clinical progress without revealing summative scores. In a randomized controlled study (N = 100 medical students), participants were assigned to either a multi-agent (MA) scaffolding condition or a control condition. All students completed two learning sessions under their assigned condition followed by an examination conducted in a patient only environment. Performance was assessed using a standardized Objective Structured Clinical Examination (OSCE) based rubric. While no significant difference was observed in final diagnostic accuracy between groups, the multi-agent AI standardized patient system improved final examination scores compared to the control group utilizing structured progressive information disclosure; the most substantial and consistent improvements were observed in communication, the expression of empathy, and specific history-taking behaviors. These findings suggest that specialized LLM agents enhance the process quality of simulated clinical interviews without artificially inflating examination outcomes. To support future research, we release a multi-expert annotated dataset comprising transcripts, checklist annotations, turn-level evaluations, and OSCE-aligned scoring outcomes. This resource aims to facilitate the development of pedagogically grounded AI-SP systems and advance research on AI-supported clinical reasoning training.
[MA-7] apes Together Strong: The Co-evolution of Computation and Cooperation
【速读】：该论文旨在解决复杂智能体系统中合作行为如何演化的根本问题，即在资源有限且存在复制与计算成本的背景下，为何以及如何涌现出自复制、协作性的策略。传统进化博弈论通常将社会互动与行为的物理代价分离，而人工生命模型则多关注自复制的涌现，却未形式化资源获取与共享能量维持之间的矛盾。为此，本文提出自创生博弈论（Autopoietic Game Theory），其核心在于将社会互动、复制机制及其计算开销作为内生变量，并实现三者的协同演化。关键创新在于：将社会困境直接嵌入计算的“物理”层面——即计算能力依赖于可用能量，而能量又源于共享资源。这一设计使得背叛行为（如窃取资源）不仅破坏集体能量池，还会导致执行效率下降甚至复制失败，从而在资源稀缺时自我限制。实证结果表明，在随机初始化的Z80机器码环境中，演化出的程序能有效抑制窃取行为；空间异质性进一步促进结构复杂性与任务性能提升。此外，该框架可整合外部压力（如以计算预算为约束的数学任务），证明计算能力与能量供给的耦合关系使合作成为构建可持续、自组织系统的主导机制。
链接: https://arxiv.org/abs/2609.10817
作者: Kunal Jha,Francesco Cicala,Blaise Agüera y Arcas,Blake Aaron Richards,Natasha Jaques,Max Kleiman-Weiner,Eyvind Niklasson
机构: 未知
类目: Multiagent Systems (cs.MA); Artificial Intelligence (cs.AI)
备注:
Abstract:How does cooperation evolve in complex agentic systems? Prior work in evolutionary game theory studies why individuals are incentivized to cooperate by isolating social interactions from the physical costs of behavior, while artificial life models traditionally study emergent self-replication without formalizing the dilemma between acquiring resources and preserving the shared energy needed to reproduce. In contrast, we introduce Autopoietic Game Theory, a computational model where social interactions, replication mechanisms, and their associated computational costs are endogenous and simultaneously co-evolving. We study these dynamics using a computational substrate of randomly initialized programs in Z80 machine code, showing empirically, and motivating with a simplified theoretical model, that embedding a social dilemma directly into the physics of computation can favor the emergence of self-replicating, cooperative strategies. When resources are scarce, our analysis shows that defection can become self-limiting even in well-mixed populations: parasitic stealing destroys shared energy, slows execution, and can prevent reliable replication. Empirically, evolved programs suppress stealing across several Z80 environments, while spatial assortment further supports structural complexity and task performance. We further show that the framework can incorporate exogenous pressures, such as math tasks structured as sequential social dilemmas, when rewards are tied to computation budgets. These results suggest that coupling an agent’s capacity for computation to its available energy transforms cooperation into a dominant scaffolding for building sustainable, self-organizing systems.
[MA-8] Finishing the Task Is Not Enough: Evaluating Agent Resilience and Considerate Participation under Accumulating Challenge
【速读】：该论文旨在解决生成式AI代理在长期部署中面临的可持续性问题，即如何在持续交互、环境变化及与人类协作的复杂工作流中保持有效性和适应性，尤其在技术、人力和运营干扰累积的情况下。其核心挑战在于现有评估体系未能充分涵盖代理在动态环境中维持功能的能力与对协同参与者的影响。解决方案的关键在于提出并实证分析两个互补维度：操作韧性（operational resilience） 与体贴参与（considerate participation）。前者关注代理在任务受阻时能否保留进展、自主恢复并明确传达自身局限；后者强调代理在适应过程中是否考虑受影响人员、角色边界与整体工作流的协调。研究通过模拟120个医疗轨迹，在轻、中、重三类压力条件下对比两种生成式AI模型在12项由利益相关者定义的任务中的表现，发现随着挑战积累，代理逐渐从自我主导恢复转向依赖人类干预，虽在结构化报告中表现出负荷增加与负面情绪，但文本响应中极少体现压力；同时，其参与模式从任务聚焦转向任务重构、对他人的关注、角色边界的调整及更广泛的协同。基于此，论文提炼出五个关键部署困境——持续性、注意力分配、角色边界、状态披露与升级机制，强调需由利益相关方明确定义，并为后续技术发展提供学习机制、情境化评估与具身适应方面的启示。
链接: https://arxiv.org/abs/2609.10724
作者: Yuanchen Bai,Zijian Ding,Angelique Taylor
机构: Cornell Tech / Cornell University (康奈尔科技学院/康奈尔大学); University of Maryland, College Park (马里兰大学学院公园分校); Korea Advanced Institute of Science and Technology (韩国科学技术院)
类目: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC); Multiagent Systems (cs.MA)
备注:
Abstract:Sustained deployment of generative AI agents requires more than isolated task success. Agents must remain useful across repeated interactions, changing conditions, and dependencies on people within shared workflows, especially as technical, human, and operational disruptions accumulate over time. We propose operational resilience and considerate participation as two complementary aspects of evaluating such agents: the former captures how agents recover from blocked work while preserving progress and communicating their limits, and the latter captures how their adaptation accounts for affected people, role boundaries, and the surrounding workflow. Yet both remain underexplored under accumulating challenge. We study 120 simulated healthcare trajectories across two generative AI models and twelve stakeholder-derived tasks under light, medium, and heavy challenge. We compare textual action plans, prompted internal assessments, and quantitative structured workload and affect reports to examine how agent behavior and reported state change as challenge accumulates. Regarding operational resilience, agents shift from self-directed recovery toward greater human dependence, while reporting increasing workload and negative affect in structured reports but seldom expressing strain in textual responses. Regarding considerate participation, agents broaden from task-focused adaptation toward task reframing, attention to others, role-boundary adjustment, and wider coordination, with distinct patterns across actions and internal assessments. From these findings, we derive five deployment dilemmas involving persistence, attention, role boundaries, state disclosure, and escalation that require stakeholder specification, further informing technical implications for learning, situated evaluation, and embodied adaptation.
[MA-9] Role differentiation as ignition of a collective information engine: Structuration in Agent Populations
【速读】：该论文旨在解决传统信息型活性物质（informational active matter）模型中集体秩序仅能通过共识达成的局限性，提出一种基于角色分化（differentiation）的新型集体信息引擎架构。其核心问题是：如何在不依赖统一共识的前提下，实现由个体角色分工驱动的自组织协调，并解释这种协调机制的涌现条件与动态演化规律。解决方案的关键在于构建一个反协调博弈（anti-coordination games）框架，其中个体通过感知带有噪声的社会信号来推断自身角色身份，而角色执行行为又反馈至该信号，形成闭环反馈回路。这一机制将身份持久性、认知能力、信道保真度与认知模式强度共同构成的“社会回路增益”（social loop gain）作为系统是否启动的临界阈值——当该增益超过1时，系统进入非线性跃迁状态，引发分岔级联（bifurcation cascade），导致角色分化与协调行为涌现。系统的功能形式由博弈谱系的特征值分布决定，呈现出从可监测的对数序列到突发性级联事件等多样化模式。资源积累通过复制子动力学（replicator dynamics）赋予不同认知模式（schemas）适应性优势，从而内生选择分岔类型。此外，亚临界身份协方差可用于早期预警，而反馈通道参数则可调控最终涌现的协调类型，使平台设计成为控制集体行为的调节杠杆。该理论首次将结构化社会理论中的“模式与资源二元性”（Sewell’s duality of schemas and resources）在信息引擎中实现操作化，为分布式生成式人工智能（distributional AGI）的渐进式起飞提供了机制基础，并提出基于监控的解决方案，融合博弈论、群体动力学与信息引擎，开辟了代理种群信息热力学的新研究路径。
链接: https://arxiv.org/abs/2609.05442
作者: Maximilian Puelma Touzel
机构: Mila–Quebec AI Institute(蒙特利尔魁北克人工智能研究所)
类目: Physics and Society (physics.soc-ph); Artificial Intelligence (cs.AI); Computer Science and Game Theory (cs.GT); Multiagent Systems (cs.MA); Social and Information Networks (cs.SI)
备注:
Abstract:Informational active matter shows how measurement-informed decisions produce collective order, so far in systems that reach consensus. We design collective information engines structured by differentiation instead, and construct a minimal instance using anti-coordination games where differentiated role information has value. Within many coexisting games, agents infer their role from a noisy social signal grounded in a persistent identity, and role-following action feeds back into that signal, which shapes the incentive to follow roles. Resources accrued through coordinated role-play combine with identity variability to reinforce the schemas that generated them. The model thereby operationalizes Sewell’s duality of schemas and resources in Structuration, a resolution to structure–agency debates across social science. The engine ignites when a social loop gain—the product of identity persistence, cognitive capacity, channel fidelity, and schema strength—exceeds one. For a repertoire of such schemas, roles emerge with increasing gain in a bifurcation cascade whose functional form is fixed by the repertoire’s eigenvalue spectrum, ranging from monitorable logarithmic sequences to avalanches that arrive without warning. Resource accumulation supplies the fitness of a replicator dynamics on schema strengths, which selects the cascade type endogenously. Subcritical identity covariance reveals that type before onset, enabling early detection, while feedback channel parameters bias which type is selected. Platform design then becomes a control lever to throttle emergent coordination. This theory grounds distributional AGI takeoff in a mechanism and provides a monitor-based solution. Joining game theory, collective dynamics, and information engines, we open a route to an information thermodynamics of agent populations.
自然语言处理
[NLP-0] Data Scarcity and Model Sparsity: Mixtures-of-Experts Overfit More to Repeated Data
【速读】： 该论文旨在解决在生成式AI（Generative AI）模型训练中，随着人类撰写文本资源的耗尽，数据重复使用导致的模型性能下降问题，尤其聚焦于近年来主流的稀疏架构——混合专家模型（Mixture-of-Experts, MoE）在高数据重复率下的表现退化现象。其核心问题是：尽管MoE架构因计算效率优势被广泛采用，但在数据重复场景下，其性能衰减速度远超密集模型（Dense Models），且该现象随稀疏度增加而加剧，且由总参数量而非活跃参数量主导。解决方案的关键在于揭示了MoE在高重复率下过拟合的内在机制：模型路由机制在训练早期即趋于稳定，同时专家专业化（expert specialization）与对重复数据的过拟合密切相关。研究通过引入基于掩码的正则化方法（如强掩码策略），有效缓解了过拟合问题，使MoE在数据重复超过64倍时仍可超越密集模型；然而，现有方法仍无法完全恢复全唯一数据训练下的性能。因此，该工作提出未来应探索通过破坏记忆模式来减少参数过度专门化的新型训练机制，以增强稀疏模型在数据重复场景下的鲁棒性。
链接: https://arxiv.org/abs/2609.11917
作者: Atindra Jha,Margaret Li,Jure Leskovec,Percy Liang,Luke Zettlemoyer
机构: Stanford University(斯坦福大学); Paul G. Allen School of Computer Science, University of Washington(华盛顿大学计算机科学学院)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:As the supply of human-written text is exhausted, it has become standard practice to repeat language model training data. Prior work has studied data repetition for densely activated Transformers, but the effects of data repetition remains largely unexplored for recently dominant sparse architectures such as Mixture-of-Experts (MoE), despite their increased compute efficiency. We vary data repetition rates across single- and multi-domain data mixes, and across MoE settings, including expert count and granularity. We consistently find, for models ranging from 80M to 1B active (8.5B total) parameters, that MoEs degrade more rapidly under data repetition. This effect increases with sparsity, dictated by total rather than active parameters. While 80M dense models can repeat data over 8x with minimal degradation, MoEs instead begin to suffer at 4x, and deteriorate rapidly, ceding their performance benefits in all-unique data settings to underperform dense models after 32x. We experiment with existing regularization methods as a potential remedy. We find that some methods, such as dropout, can mitigate overfitting. In particular, with strong masking-based regularization, MoEs are able to outperform dense models even when data is repeated more than 64 times. However, no method fully matches the performance of all-unique training data. Finally, we analyze internal mechanisms correlated with MoE overfitting in high repetition regimes, and find that MoE routing universally stabilizes early in training, and that expert specialization correlates with overfitting to repeated data. In sum, our work addresses the adverse interactions between sparsity and data repetition: we present evidence for the core mechanisms of overfitting and its potential remediation, and suggest promising avenues for future methods to reduce over-specialization in model parameters by disrupting memorization patterns.
[NLP-1] Distance generalization in transformers: why bother with positional encoding?
【速读】： 该论文旨在解决大模型在面对训练与推理阶段间令牌间距离（inter-token distance）变化时的泛化能力问题，即“分布外距离泛化”（out-of-distribution distance generalization），其核心挑战在于：当模型在固定上下文长度下被训练，但在推理时遭遇未在训练中出现的长距离依赖关系时，如何保持性能稳定。解决方案的关键在于系统性地探究位置编码机制（如RoPE、ALiBi与无位置编码NoPE）对距离分辨能力的影响、训练数据中所覆盖的互距多样性（data diversity）的作用，以及距离迁移学习（distance transfer learning）在何种条件下会产生正向或负向效应。研究通过构建两个合成的延迟复制任务（delay copy tasks），分别模拟全量与选择性复制场景，验证不同设置下的模型表现，揭示了位置编码设计与训练数据分布对模型泛化能力的决定性影响，强调了深入理解其内在机制的重要性。
链接: https://arxiv.org/abs/2609.11913
作者: Daniel Henrik Nevermann,Claudius Gros
机构: Goethe University Frankfurt(法兰克福歌德大学)
类目: Computation and Language (cs.CL)
备注: 15 pages, 7 figures
Abstract:Out-of-distribution length generalization, namely to extrapolate a task from short to longer context, has been studied intensively for transformers. Here we focus on distance generalization, which probes performance when inter-token distances are changed between training and inference, while keeping a fixed context length. We construct two synthetic delay copy tasks, both involving finite distances between source and recall, where tokens are copied either fully or selectively, and test models on delays unseen during training. We address three questions: (A) Do positional encoding schemes such as RoPE and ALiBi improve distance resolution relative to no positional encoding (NoPE)? (B) How does data diversity, the number of inter-token distances seen in training, affect performance? © When is distance transfer learning positive or negative? We present a thorough investigation, finding that it is paramount to improve our understanding of the underlying mechanisms.
[NLP-2] MindTopo: Can Foundation Models Reason in Topological Space?
【速读】： 该论文旨在解决当前基础模型在空间推理能力评估中普遍忽视拓扑关系（topological relations）的问题，尤其是那些在连续形变下保持不变的拓扑性质。尽管认知科学已确立拓扑关系是空间理解的基础，但现有基准测试多聚焦于度量属性（如距离、角度、形状）或视角依赖性关系，缺乏对拓扑直觉的系统性评估。为此，论文提出MindTopo基准，涵盖五类基于认知科学与形式拓扑学的核心拓扑属性：连续性（continuity）、分离性（separation）、顺序性（order）、包围性（enclosure）和纽结（knots），每项属性均在两个认知层级上进行评估：推理（reasoning）与规划（planning）。其中，推理任务要求模型识别或推断拓扑关系及其变化；规划任务则将模型作为闭环智能体，通过策略选择环境动作以完成目标。该基准包含13种程序化生成的任务类型，共11,030个实例，支持可控难度调节。实验对比了14个多模态大语言模型（MLLMs），并研究了引入图像与视频生成能力的智能体配置，包括3个视频生成模型在规划场景中的应用。结果表明，所有模型在推理任务上的表现均优于规划任务，且最佳模型性能仍显著低于人类水平。进一步分析显示，监督微调与强化学习对推理提升效果优于规划；生成的观测虽保留局部线索并可达合理终点，但经审计的轨迹未能可靠遵循环境动态或在状态转移中维持拓扑一致性。因此，解决方案的关键在于构建一个系统性、可扩展且符合认知逻辑的拓扑直觉评估框架——MindTopo，从而推动生成式 AI 在深层空间认知能力上的发展。
链接: https://arxiv.org/abs/2609.11900
作者: Yunfei Ge,Anbang Liu,Qineng Wang,Johnalbert Garnica,Jianwen Lyu,Zihan Wang,Reuben Tan,Jianfeng Gao,Ruohan Zhang,Yining Hong,Jiajun Wu,Manling Li
机构: Northwestern University(西北大学); Microsoft Research(微软研究院); Stanford University(斯坦福大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
备注: Preprint version
Abstract:Spatial reasoning depends not only on metric properties such as distance, angle, and shape, but also on topological relations that remain invariant under continuous deformation. Cognitive science identifies these relations as foundational to spatial understanding, yet foundation-model evaluations largely focus on metric or viewpoint-dependent relations. We introduce MindTopo, a benchmark of topological intuition across five properties grounded in cognitive science and formal topology: continuity, separation, order, enclosure, and knots. MindTopo evaluates each property at two cognitive levels. Reasoning asks a model to identify topological relations or infer how they change. Planning instantiates a foundation model as a closed-loop agent whose policy selects environment actions. MindTopo contains 11,030 instances across 13 procedurally generated task types with controllable difficulty. We benchmark 14 MLLMs and study agent configurations augmented with image and video generation, including 3 video generative models in planning settings. Every MLLM performs better on reasoning than on planning, and the best-performing model remains far below observed human performance. On Qwen3-VL-2B-Instruct, supervised fine-tuning and reinforcement learning improve reasoning more than planning. Generated observations retain local cues and reach plausible endpoints, but audited rollouts do not reliably follow environment dynamics or preserve topology across transitions. Our website is at this https URL
[NLP-3] Nuha-Speech: Building General-Purpose Arabic Speech-LLM s
【速读】： 该论文旨在解决阿拉伯语在生成式语音大模型（Speech Large Language Models, speech-LLMs）中严重代表性不足的问题，尤其是在缺乏专门针对阿拉伯语语音数据集、训练框架与评估体系的背景下。其核心挑战在于受限于稀缺的阿拉伯语语音资源，难以有效构建高质量的通用型阿拉伯语语音大模型。为此，论文提出了一项系统性解决方案——Nuha-Speech，其关键在于构建一个大规模的阿拉伯语语音问答（Arabic Speech Question-Answering, SQA）语料库，包含超过150万条训练样本，支持对多种核心语音任务进行指令微调；在此基础上，基于Qwen-Omni系列不同规模的模型变体开展监督微调；并设计了一个涵盖多样化任务和定制化评估指标的系统性评估框架。通过这一综合举措，论文致力于建立在资源有限条件下推进阿拉伯语语音大模型发展的基础性基础设施。
链接: https://arxiv.org/abs/2609.11892
作者: Yingzhi Wang,Reem Alhazzani,Muhammad Alqurishi
机构: Elm Company, KSA(埃尔姆公司，沙特阿拉伯)
类目: Computation and Language (cs.CL)
备注:
Abstract:As Speech Large Language Models (speech-LLMs) become increasingly multilingual, Arabic remains significantly underrepresented, highlighting the need for dedicated infrastructure to train and evaluate Arabic speech-LLMs. To address this gap, we introduce Nuha-Speech, a comprehensive initiative to develop general-purpose Arabic speech-LLMs spanning dataset construction, model training, and systematic evaluation. Specifically, we constructed a large-scale Arabic Speech Question-Answering (SQA) corpus comprising over 1.5 million training samples to allow instruction tuning over a broad range of core speech tasks. Then, the corpus was used for supervised fine-tuning based on Qwen-Omni model variants at different scales. Finally, we designed an evaluation framework featuring diverse tasks and tailored metrics. Through this work, we aim to establish foundational infrastructures for Arabic Speech-LLMs under constraints imposed by limited Arabic speech resources. Subjects: Computation and Language (cs.CL) Cite as: arXiv:2609.11892 [cs.CL] (or arXiv:2609.11892v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2609.11892 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-4] Domain-Specific Hallucination Detection in Large Language Models
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）生成文本中存在不忠实陈述（即幻觉，Hallucination）的问题。其核心挑战在于如何有效识别和检测生成内容中的事实性错误，尤其是在多样化任务场景下保持高精度与鲁棒性。解决方案的关键在于提出一种多信号检测流水线，融合微调后的DeBERTa-v3分类器、基于蒙特卡洛（Monte Carlo, MC）Dropout的不确定性量化方法以及温度缩放校准技术，实现对响应层面幻觉的精准检测。该方法在HaluEval基准测试中表现出色，通用领域任务的F1达到0.915，AUROC达0.977，且在问答（QA）、摘要（Summarization）和对话（Dialogue）等具体任务上分别取得0.97、0.96和0.82的F1得分；通过MC Dropout推理进一步将准确率提升至93.2%。此外，研究通过上下文消融实验验证了模型依赖于真正的蕴含推理而非表面模式，表明其具备实质性的语义理解能力。同时，研究发现仅使用25%训练数据即可获得77%的全量数据性能，显示出高效的训练潜力。为进一步降低幻觉，作者采用直接偏好优化（Direct Preference Optimization, DPO）对Qwen2.5-0.5B模型进行微调，使其幻觉率从85.5%降至37.7%，相对减少55.9%。跨领域评估显示，通用领域训练在生物医学领域SciFact上的迁移效果较差（F1=0.52），从而强调了领域匹配预训练的重要性，其中在SciFact上使用PubMedBERT进行领域特定微调后，获得F1=0.63、AUROC=0.81，证明领域适配是提升检测性能最有效的策略。
链接: https://arxiv.org/abs/2609.11878
作者: Varun Teja Chundru,Debasmita Biswas
机构: Purdue University Fort Wayne(普渡大学福尔特韦恩分校)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 6 pages, 3 figures, 5 tables
Abstract:Large language models generate fluent text that can contain unfaithful claims – a phenomenon known as hallucination. We present a multi-signal detection pipeline combining fine-tuned DeBERTa-v3 classification, Monte Carlo (MC) Dropout uncertainty quantification, and temperature-scaled calibration for response-level hallucination detection. Evaluated on the HaluEval benchmark, our pipeline achieves F1=0.915 and AUROC=0.977 on general-domain tasks, with per-task F1 scores of 0.97 (QA), 0.96 (Summarization), and 0.82 (Dialogue). MC Dropout inference further improves accuracy to 93.2%. A context ablation study confirms the model performs genuine entailment reasoning rather than exploiting surface patterns, with summarization F1 dropping 24% when knowledge context is removed. Learning curve analysis reveals that 25% of training data captures 77% of full-data performance. Beyond detection, we apply Direct Preference Optimization (DPO) to a Qwen2.5-0.5B generator, reducing its hallucination rate from 85.5% to 37.7% (55.9% relative reduction) as measured by our detector. Cross-domain evaluation on the SciFact biomedical benchmark shows that general-domain training transfers poorly (F1=0.52), motivating domain-specific fine-tuning. PubMedBERT fine-tuned on SciFact achieves F1=0.63 and AUROC=0.81, demonstrating that domain-matched pre-training is the strongest adaptation strategy. Code and models are available at this https URL
[NLP-5] he Last AI Built by Humans: Toward Genuine Recursive Self-Improvement
【速读】： 该论文旨在解决当前大语言模型（LLM）在持续自我改进能力上的局限性问题，即现有模型难以实现真正意义上的递归式自我改进（Recursive Self-Improvement, RSI），从而限制了其在复杂任务中的自主演化与性能跃迁。其解决方案的关键在于构建一个分阶段的RSI发展路线图，涵盖从改进执行自主性、改进策略自主性、经验获取自主性到环境适应自主性，最终实现递归元改进（recursive meta-improvement）的完整能力演进路径。通过结合科学发现、具身智能和软件工程等多场景应用分析，论文揭示了不同领域对RSI的不同需求与推进速度，并基于产业实践与初步实证证据，系统识别出实现真正RSI所面临的核心挑战，为构建具备持续自我优化能力的下一代AI系统提供了理论框架与实践指引。
链接: https://arxiv.org/abs/2609.11873
作者: Yi Duan,Ying Liu,Zirui Tang,Haodong Chen,Jun Zhou,Yumou Liu,Bangrui Xu,Yukai Wu,Sidi Chen,Yuhan Zhou,Haoyu Wang,Xiaoyou Yu,Shaokun Han,Xuzhou Zhu,Le Zhou,Bolin Lu,Wei Zhou,Jiachen Liu,Nuozhou Fang,Jiaxin Tian,Ruoyu Chen,Yuxuan Li,Kai Zuo,Kaiyan Zhang,Jiantao Qiu,Conghui He,Guoliang Li,Bowen Zhou,Zhiyuan Liu,Zhoufutu Wen,Jihua Kang,Xuanhe Zhou,Fan Wu
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Recursive self-improvement (RSI) enables AI systems to turn experience and feedback into persistent changes that improve both their capabilities and the process of future improvement. We first use the Headroom-Closed Index (HCI) to reveal the problems of existing LLMs, then introduce the RSI concept and its development roadmap: from improvement-execution autonomy, improvement-strategy autonomy, experience-acquisition autonomy, and environment-adaptation autonomy, to recursive meta-improvement. Next we examine RSI across scenarios (e.g., scientific discovery, embodied intelligence, software engineering), highlighting their distinct requirements and development speeds. Drawing on diverse industry practices and preliminary empirical evidence, we connect RSI research with practical systems and identify key challenges to achieving genuine RSI.
[NLP-6] Augustinian BabyLM: What Ostensive Definition Can and Cannot Teach a Small Language Model
【速读】： 该论文旨在解决语言模型在训练初期因词向量随机初始化而导致语义学习效率低下的问题，特别是如何通过视觉感知（visually grounded）的先验知识加速词汇语义的习得。其核心解决方案是借鉴圣奥古斯丁（St. Augustine）提出的“通过指称（ostension）习得词义”的认知机制，对小型掩码语言模型（DeBERTa）中的部分词元（tokens）进行视觉初始化：即为标注图像区域的词元赋予基于图像特征提取的嵌入向量，而其他词元仍保持随机初始化。这一视觉种子（visual seeding）策略在训练全程中留下可测量的影响，尤其在涉及物体属性（如颜色、材质、大小、形状）的知识任务中表现出持续优势，且该优势仅限于被视觉种子化的词元。进一步的因果实验表明，对未被种子化的词元进行人工视觉模拟（synthetic grounding），可将其优势转移至相应词元，验证了视觉信息的因果作用。然而，现有主流基准测试（如BabyLM系列）主要评估抽象语法能力，无法捕捉此类视觉种子带来的隐性收益；尽管训练目标本身（掩码预测损失）在所有种子条件下对功能词和抽象词汇均表现出显著降低，但当前评估体系尚未能有效识别这种深层学习效应。因此，如何设计能够揭示视觉种子对深层语义表征影响的新型评估范式，仍是亟待解决的关键问题。
链接: https://arxiv.org/abs/2609.11870
作者: Lisa Bylinina
机构: Institute for Language Sciences; Utrecht University (乌得勒支大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:A language model normally begins training with random word embeddings: whatever ‘banana’ means must be learned from training corpora. I implement St. Augustine’s picture of word learning, meaning by ostension, for a small masked language model (DeBERTa) trained on 10M words: before training, visually grounded tokens receive embeddings derived from the image regions they label; other tokens start random. Visual initialization leaves a measurable imprint that lasts until the end of training. At the same time, the effect remains invisible under most BabyLM benchmarks, which probe abstract grammatical knowledge: visual initialization does not affect performance there. The only zero-shot exception is object-property knowledge (COMPS, Misra et al. 2023), where seeding helps in every configuration. To follow up on this result, I build a corpus-tailored version of the Visual-Property Swap benchmark (Lin et al., 2026), which tests color, material, size, and shape knowledge, with per-item training frequency and seeded status. Here, vision-seeded models have a persistent, seed- replicated advantage, confined to the seeded words. As a causal test, I show that synthetic grounding of previously unseeded words transfers the advantage to exactly those words. Function words and abstract vocabulary also receive strong visual seeds and retain them throughout training, and the training objective draws on them: held-out mask-prediction loss falls for these words in every seed. However, no benchmark I run registers this. What evaluation would pick this up remains an open question.
[NLP-7] Epistemic orientation predicts legislative effectiveness among members of the US Congress
【速读】： 该论文旨在解决美国国会立法者在公共言论中证据导向性（epistemic orientation）是否存在个体差异，以及这种语言特征是否与政治行为和立法成效相关的问题。其核心解决方案在于引入“证据减直观”（Evidence-Minus-Intuition, EMI）评分体系，量化国会发言及推特言论中证据导向语言与直觉导向语言的相对比例，并将其与立法者意识形态、机构职位、沟通语境及立法成效指数（Legislative Effectiveness Score, LES）等多维度数据进行关联分析。研究发现，意识形态越极端的议员在国会发言中越少使用证据导向语言，且跨平台（国会发言与推特）的EMI表现具有一致性，但推特整体的证据导向性更低；更重要的是，在控制多重变量后，国会发言中的高EMI值仍与个人立法成效正相关，表明证据导向语言不仅是集体话语趋势，更是个体立法者沟通效能的重要预测指标。
链接: https://arxiv.org/abs/2609.11865
作者: Segun Aroyehun,Stephan Lewandowsky,David Garcia
机构: University of Konstanz (康斯坦茨大学); University of Bristol (布里斯托尔大学); University of Potsdam (波茨坦大学); Complexity Science Hub (复杂科学中心)
类目: Computation and Language (cs.CL)
备注:
Abstract:Truth and evidence-based communication provide important foundations for democratic governance, accountability, and collective decision-making. Prior work shows that evidence-oriented language in US congressional floor speeches has declined since the mid-1970s, alongside broader changes in legislative productivity and polarization. This study shifts the analysis from congressional sessions to individual members of Congress to examine whether epistemic orientation varies systematically across legislators and whether it relates to political behavior and legislative effectiveness. Using the Evidence-Minus-Intuition (EMI) score, we measure the relative prevalence of evidence-oriented versus intuition-oriented language in congressional floor speeches and Twitter posts. We link these measures to legislator-level data on ideology, institutional position, communication context, and Legislative Effectiveness Score (LES). The results show that more ideologically extreme members use less evidence-oriented language on the congressional floor. EMI also exhibits cross-platform consistency with members who use more evidence-oriented language in floor speeches also being more evidence-oriented on Twitter, although EMI is lower on Twitter overall. Finally, EMI in congressional speeches is positively associated with individual legislative effectiveness, even after accounting for ideology and extensive political, institutional, demographic, topical, and communication volume controls. These findings suggest that evidence-oriented language is not only an aggregate feature of congressional discourse but also a meaningful attribute of individual-level legislative communication and effectiveness.
[NLP-8] IndicTriMix: Developing Language Identification Datasets and Models for Tri-Language Code-Mixing
【速读】： 该论文旨在解决多语言混合文本（code-mixed text）中细粒度的词级语言识别问题，尤其针对社交媒体中用户在单个语句内频繁切换多种语言的现象。传统语言识别模型通常针对单一语言文本设计，难以有效处理词级层面的语言边界划分。为此，研究将任务建模为序列标注问题，并基于适用于印度语言的上下文感知型预训练模型MuRIL与XLM-RoBERTa进行微调。其解决方案的关键在于利用强大的上下文嵌入能力捕捉多语言混合语境中的语言线索，同时通过基于平行语料的两种生成策略构建合成的多语言混合数据，提升模型泛化能力。实验在印地语、古吉拉特语和孟加拉语三种语言配置上验证了模型的有效性，并公开发布了包含人工标注测试集的基准数据集及可复现的微调模型，以推动该领域的后续研究。
链接: https://arxiv.org/abs/2609.11851
作者: Pruthwik Mishra,Rudra Trivedi,Avi Patel,Ashok Urlana,Shrikant Malviya
机构: Sardar Vallabhbhai National Institute of Technology (萨达尔·瓦拉巴伊·国立技术学院); Surat (苏拉特); India (印度); TCS Research (TCS 研究院); Hyderabad (海得拉巴); India (印度)
类目: Computation and Language (cs.CL)
备注: 9 pages, 9 tables
Abstract:Language identification in code-mixed text, largely observed in social media, is highly essential when users frequently switch between multiple languages within a single utterance. Accurately identifying the languages of code-mixed tokens becomes an urgent necessity. Traditional language identification models, designed for monolingual text, are not well suited for token-level language identification in code-mixed settings. We formulate the task as a sequence labeling problem and fine-tune contextual transformer-based models MuRIL and XLM-RoBERTa best suited for Indian languages. We evaluate these systems on three different data configurations (Hindi, Gujarati, and Bengali) to predict language labels for individual tokens. We release a benchmark for language identification in code-mixed tokens with manually annotated test sets. We propose two approaches of code-mixed generation using parallel sentences of three languages. The trained models demonstrate the effectiveness of contextual embeddings for token-level language identification in multilingual social media text. For reproducibility and to facilitate future research, we publicly release our fine-tuned models.
[NLP-9] arget leakage not model class explains reported accuracy in survey-based cardiovascular screening: a leakage-tiered audit of glass-box and tabular foundation models
【速读】： 该论文旨在解决心血管疾病筛查模型在真实世界应用中可能存在的“目标泄露”（target leakage）问题，即模型性能看似优异（如AUROC接近0.89）是否真正反映了对潜在风险的合理学习，还是仅仅依赖于包含诊断后特征等信息偏差。研究通过系统评估十种不同类型的分类器（涵盖线性模型、树集成、神经网络、可解释的玻璃盒模型及表格基础模型），在2022年行为风险因素监测系统（BRFSS）442,067名受访者中，基于五个递减泄露风险的特征层级进行基准测试，并将最优模型与阈值冻结后应用于2023年430,755名新样本，全面考察其区分能力、校准性、公平性、置信区间覆盖、解释可信度及推理成本。关键发现表明：移除两个诊断后特征后，所有模型的AUROC均下降约0.05，使性能差异缩小至仅0.0045的窄带范围，说明此前报告的高精度主要源于数据泄露而非模型能力。其中，可解释的提升机（glass-box explainable boosting machine）在预设0.005容差内表现不劣于其他模型，且推理速度比最强的基础模型快约104倍。此外，通过调整模型形状函数可显著缩小性别间筛查敏感性差距（从75.4% vs 89.0%降至0.010），而蒙德里安校准（Mondrian calibration）有效修复了各亚组的校准偏差。模型跨年度迁移性能稳定（AUROC变化≤0.002），证明其泛化能力。研究结论指出，文献中报告的“性能余量”本质上是特征集的属性，而非学习器的能力所致；透明化建模不仅无显著成本，反而使公平性修正和不确定性量化变得可审计。因此，当前制约模型实际部署的核心瓶颈并非模型容量，而是评估实践方法。
链接: https://arxiv.org/abs/2609.11838
作者: Raad Bin Tareaf,Murad Al-Rajab,Samia Loucif,Samer Ellaham,Cedric Schmitz
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Cardiovascular screening models trained on national health surveys routinely report areas under the receiver operating characteristic curve (AUROC) near 0.89. We asked whether that accuracy reflects learning or target leakage, whether tabular foundation models change the answer, and whether the properties deployment requires survive joint examination. We benchmarked ten classifiers spanning linear, tree-ensemble, neural, glass-box, and tabular foundation classes for prevalent myocardial infarction in 442,067 respondents of the 2022 Behavioral Risk Factor Surveillance System across five feature tiers of decreasing leakage risk. Each was audited for discrimination, calibration, fairness at an explicit screening threshold, conformal coverage, explanation faithfulness, and inference cost, then applied – models and thresholds frozen – to 430,755 respondents of 2023. Removing two post-diagnostic features cost every model 0.049-0.051 AUROC, collapsing the field into a 0.0045-wide band. The glass-box explainable boosting machine was non-inferior to every alternative within a pre-specified 0.005 margin while scoring the cohort roughly 104 times faster than the strongest foundation model. One threshold detected 75.4% of women’s infarctions against 89.0% of men’s; editing the model’s shape functions reduced the gap to 0.010. Marginal conformal prediction gave 0.86 coverage to men and 0.82 to adults over 60; Mondrian calibration repaired every stratum. Frozen models transported within 0.002 AUROC. Reported headroom in this literature is a property of the feature set, not the learner. Transparency cost nothing measurable and made fairness repair and uncertainty conditioning directly auditable. Evaluation practice, not model capacity, is the binding constraint.
[NLP-10] SpecGuard: Inference-Time Backdoor Detection For Free
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在部署后可能携带隐蔽后门（backdoor）的问题，即模型在正常输入下表现无异常，但在遇到特定秘密触发器（secret trigger）时会切换至攻击者控制的行为。尽管部署前可进行审计，但针对频繁更新的模型，运行时监控仍至关重要。现有方法面临两大局限：一是依赖对触发器形式的假设，难以应对隐蔽攻击；二是需额外计算开销，如输入扰动或额外生成轮次。本文提出SpecGuard，一种无需额外模型计算成本的推理时后门检测机制，其核心创新在于复用推测解码（speculative decoding）过程中的验证信号——当后门被触发时，目标模型会偏离正常行为，而小型草稿模型（draft model）未受污染，导致草案令牌的接受率发生显著变化。作者形式化分析了该信号出现的条件，并证明攻击者若试图抑制此信号，则必须削弱后门有效性。实验表明，SpecGuard在多种后门类型和模型架构中均能可靠检测到触发行为，包括传统输入级过滤无法识别的隐蔽攻击场景，且避免了现有运行时检测器的额外生成开销。因此，推测解码不仅加速推理，还成为一种免费、持续可用的后门检测信号。
链接: https://arxiv.org/abs/2609.11799
作者: Rui Wen,Ahmed Salem,Andrew Paverd,Mark Russinovich,Zheng Li
机构: Institute of Science Tokyo (东京科学研究所); Microsoft Security Response Center (微软安全响应中心); Microsoft Azure (微软云); Shandong University (山东大学)
类目: Cryptography and Security (cs.CR); Computation and Language (cs.CL)
备注:
Abstract:Large language models are often fine-tuned, shared, or downloaded from third parties, so a deployed model may carry a hidden backdoor that behaves normally on benign inputs but switches to attacker-controlled behavior when a secret trigger appears. While backdoors can be audited before deployment, runtime monitoring remains important for models that are frequently updated. The challenge is that LLM serving is latency-sensitive: existing inference-time detectors either rely on assumptions about the trigger form, which can fail on stealthy attacks, or require extra model computation, such as input perturbations or an additional generation pass. We introduce SpecGuard, an inference-time backdoor detector that repurposes speculative decoding at zero added model-computation cost. Speculative decoding speeds up inference by using a small draft model to propose tokens and a target model to verify them. We observe that this verification process already exposes a useful signal: when a backdoor is triggered, the target model shifts toward the attacker’s behavior, while a clean draft model does not predict this shift, causing the draft-token acceptance rate to change. We formalize when this signal appears and show that an attacker who suppresses it must also weaken the backdoor. Across diverse backdoor types and model families, SpecGuard reliably detects triggered behavior, including stealthy cases where input-level filters are blind, while avoiding the extra generation cost of existing runtime detectors. Speculative decoding therefore doubles as a free, always-on signal for detecting backdoored LLM behavior. Subjects: Cryptography and Security (cs.CR); Computation and Language (cs.CL) Cite as: arXiv:2609.11799 [cs.CR] (or arXiv:2609.11799v1 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2609.11799 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-11] Beyond Word Error Rate: A Switch Aware Evaluation of ASR and Audio Language Models on English Yoruba Code-Switched Speech
【速读】： 该论文旨在解决低资源、带变音符号丰富的语言中代码混用语音（code-switched speech）的自动语音识别（ASR）与音频语言模型（audio LM）性能评估不充分的问题。现有系统在单语基准上表现优异，但在面对英语-约鲁巴语混用语音时，其真实行为仍缺乏细致刻画。研究的关键在于提出一种“开关感知”（switch-aware）的评估框架，通过一个确定性的2000句测试集和统一评分流程，对11个现代系统（6个ASR模型和5个audio LM）进行多维度诊断分析。其核心解决方案包括引入一系列开关局部化指标：开关入口词错误率（SETER）、窗口化开关点错误率、语言特定错误率以及忽略变音符号的词错误率（diacritic insensitive WER），以揭示传统综合词错误率（WER）所掩盖的代码切换行为。研究发现，尽管基于WER排名最优的ASR系统与领先audio LM在整体错误率上无显著差异，但后者在所有开关局部化指标上均显著更优；同时，多数系统在约鲁巴语词识别上表现极差（错误率高达0.97），且错误高度集中于向约鲁巴语切换的节点；此外，多个生成式audio LM存在翻译、冗余和提示泄露等严重问题，且其表现强烈依赖输入提示。研究开源了数据集、度量实现与评估脚本，以推动非洲语言代码混用语音的可复现、开关感知基准评测。
链接: https://arxiv.org/abs/2609.11786
作者: Chibuzor Okocha,Christan Earl Grant
机构: University of Florida, Gainesville, FL, USA (佛罗里达大学盖恩斯维尔分校)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted to IEEE Speech Language Tecnology
Abstract:Automatic speech recognition (ASR) systems and audio language models (audio LMs) now report low error rates on monolingual benchmarks, but their behavior on code switched speech in low resource, diacritic rich languages remains poorly characterized. We present a switch aware evaluation of eleven modern systems (six ASR models and five audio LMs) on English Yoruba code-switched speech, using a deterministic 2000 utterance evaluation set and a shared scoring pipeline. Beyond word error rate (WER), we report switch localized diagnostics: a switch entry token error rate (SETER), windowed switch point error rates, language specific error rates, and a diacritic insensitive WER. Our central finding is that aggregate WER hides code switching behavior. The best system by WER (an ASR model) is statistically indistinguishable from a leading audio LM on WER, yet the audio LM is significantly better on every switch localized metric. Across faithful systems, Yoruba token recognition collapses (error 0.97 for almost all systems) while English tokens are recognized far better, and errors concentrate sharply at switches into Yoruba. Several generative audio LMs fail as exact transcribers, producing translation, verbosity, and prompt leakage that are strongly prompt dependent. We release manifests, metric implementations, and evaluation scripts to support reproducible, switch aware benchmarking for African code switched speech.
[NLP-12] he widening evaluation gap in medical large language model research 2023 to 2026
【速读】： 该论文旨在解决医学研究在评估生成式人工智能（Generative AI）系统时是否存在滞后的问题，即临床研究进展是否能跟上所评估系统快速迭代的步伐。其核心发现表明，尽管大型语言模型（Large Language Models, LLMs）每数个季度便迅速更迭，但相关医学研究的发表速度远滞后于模型更新周期，导致研究评估的对象普遍为过时或已停用的模型。解决方案的关键在于揭示“评估滞后”现象的本质并非源于研究周期本身，而是由模型选择策略所致：即使采用随机对照试验等高严谨性设计的研究，其评估的模型平均仍比其他研究滞后4.6个季度，且62%的随机对照试验实际评估的是已停止维护的模型家族。研究进一步通过反事实分析指出，即便持续迁移至新模型，也仅能缓解约56%的性能衰减（95%置信区间50%-65%），说明模型快速迭代与研究滞后之间的张力主要源于研究者对模型版本的选择偏差，而非研究流程本身的延迟。因此，问题的核心在于建立动态、可追踪的模型版本评估机制，以实现研究的时效性与严谨性的平衡。
链接: https://arxiv.org/abs/2609.11770
作者: Raad Bin Tareaf,Murad Al-Rajab,Samia Loucif
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language models are superseded every few quarters; clinical evidence takes years. We asked whether medical research is keeping pace with the systems it evaluates. PubMed returned 11,628 records for January 2023 to June 2026 across fourteen clinical domains, growing 45-fold; 2.5% used a randomised, controlled or prospective design. Evaluation lag, from a study’s newest named model release to its own publication, widened from 1.33 to 6.08 quarters. Because discontinued models age mechanically, we benchmarked this against a counterfactual holding model composition fixed: migration to newer systems offset only 56% of the drift (95% CI 50-65). Randomised trials evaluated models a median 4.6 quarters older than other designs (P = 3 x 10^-19), yet among studies naming a model still under development no design differed from any other; 62% of randomised trials evaluated a discontinued family. Rigour and currency are in tension, and that tension reflects model selection rather than research timelines.
[NLP-13] Recognizing Is Not Reversing: A Controlled Inversion Test of Fact-Preserving News Framing
【速读】： 该论文旨在解决生成式人工智能（Generative AI）在新闻文本改写中对“框架”（framing）进行逆向操作的能力问题，即在保持事实不变的前提下，能否有效撤销已知的框架化处理。现有研究多聚焦于生成质量、检测效果或改写文本的中立性评估，却未直接检验模型是否具备精准还原原始事实框架的能力。为此，论文提出一种受控反演测试（controlled inversion test），针对三种公认的框架表现形式——评价性词汇（evaluative lexis）、施事性实现（agency realization）和信息显著性（information salience）——构建了60篇新闻文章、三种干预强度下的540组配对变体，确保原子级事实一致并记录具体编辑行为。实验结果表明，尽管Qwen、DeepSeek和Kimi等模型在事实保留率上均维持在0.84左右，但框架逆转率仅为0.044–0.068，即使在正确识别框架类型与方向的情况下，综合逆转率也仅达0.071。这揭示出事实保真度、框架识别能力与框架逆向操作之间存在显著分离：能够识别框架并不等同于具备消除框架的能力，凸显当前大语言模型在语义层面的可逆性控制方面仍存在根本性局限。
链接: https://arxiv.org/abs/2609.11769
作者: Yi Liu
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models (LLMs) are increasingly used to analyze and rewrite news, yet current framing studies mainly evaluate generation, detection, or whether rewritten text appears more neutral. They do not directly show whether a model can undo a known framing transformation while keeping the facts fixed. We introduce a controlled inversion test over three established textual realizations of framing: evaluative lexis, agency realization, and information salience. Across 60 news articles and three intervention strengths, this yields 540 paired variants with preserved atomic facts and recorded edits. Across Qwen, DeepSeek, and Kimi, factual preservation remains near 0.84, whereas intervention reversal is 0.044–0.068. Even when both framing type and direction are recognized correctly, pooled reversal reaches 0.071. These results reveal a clear separation between factual fidelity, framing recognition, and framing inversion: recognizing how an article is framed does not imply that the framing can be undone.
[NLP-14] A Unified Per-Token Gating Family for On-Policy Distillation: FKL/RKL Mixing with Multi-Channel and Bias Coefficients EMNLP2026
【速读】： 该论文旨在解决当前在线策略知识蒸馏（OPD）中基于逐令牌门控的前向/反向KL损失方法存在的局限性，即现有方法如EOPD（Jin et al., 2026）和ToDi（Jung et al., 2025）均固定单一门控信号与单一门控方向，且二者未被直接对比。其核心解决方案提出了一种四系数参数化形式 λt=σ(a⋅ht+b⋅u(x)+c+d⋅gapt)，该形式在保持EOPD与ToDi方向对齐代理作为一维退化情形的基础上，引入多通道组合与显式偏置项以拓展设计自由度。实验在TweetEval情感与仇恨分类任务上，使用Qwen3-32B教师模型与Qwen3-4B学生模型验证表明，该参数化家族中的多数配置在36个可比单元中有33个优于对应量级的单通道（仅熵或仅间隙）一维基线；在26个单元的均值匹配隔离实验中，动态门控在19个单元中优于有效KL匹配的静态基线。由于所有实验共享训练数据、模型架构及参数子结构，结果以探索性聚合方向性证据呈现，而非独立假设检验。对九个关键比较项进行三重种子配对复现显示方向一致性，但个体效应量较小，n=3时未达统计显著。因此，该工作主要贡献在于构建了一个统一的坐标系，用于系统比较短输出分类场景下OPD中的逐令牌门控设计。
链接: https://arxiv.org/abs/2609.11768
作者: Suwan Wu,Yumeng Lin,Pengcheng Yuan,Xiaolong Jiang
机构: Xiaohongshu Inc.(小红书); Tianjin University (天津大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Accepted at the Findings of the 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP 2026 Findings)
Abstract:Per-token gating of forward/reverse KL losses has become a standard technique for on-policy knowledge distillation (OPD), but existing methods such as EOPD (Jin et al., 2026) and ToDi (Jung et al., 2025) each fix a single gating signal and a single gating direction, and the two have never been compared directly. We introduce a four-coefficient parameterization lambda_t = sigma(a * h_t + b * u(x) + c + d * gap_t) in which direction-aligned proxies of EOPD and ToDi appear as one-dimensional (1D) restrictions, and which adds multi-channel composition and an explicit bias as further degrees of freedom. On TweetEval (Barbieri et al., 2020) emotion and hate, with a Qwen3-32B teacher and a Qwen3-4B student, configurations in the full family reach higher accuracy than the matched-magnitude single-channel (entropy-only / gap-only) 1D restrictions in 33 of 36 comparable cells, and a 26-cell mean-match isolation experiment places dynamic gating ahead of effective-KL-matched static baselines in 19 of 26 cells. Because cells share training data, models, and parameter substructure, we report both counts as exploratory aggregate directional evidence rather than as independent hypothesis tests. Targeted three-seed paired replications of the nine headline comparisons singled out by that sweep – including a third task, offensive – are directionally consistent, but individually smaller than the single-seed estimates and not significant at n=3. We therefore present the parameterization primarily as a shared coordinate system for comparing per-token gating designs in short-output classification OPD.
[NLP-15] Component-Aware Differential Privacy for Federated Multilingual Speech-LLM s
【速读】： 该论文旨在解决在联邦学习（Federated Learning, FL）中，针对语音大语言模型（speech-LLM）因声学编码器与语言解码器之间更新范数存在数量级差异而导致的逐层差分隐私（Per-layer Differential Privacy, DP）剪裁失效问题。传统单池化逐层剪裁方法在面对组件间范数失衡时，会引发跨组件预算坍塌（cross-component budget collapse），导致词错误率（Word Error Rate, WER）显著恶化甚至训练崩溃。其关键解决方案是提出α-split双池分配机制，将编码器与语言模型参数分别归入独立的剪裁池中，通过引入可调比例参数α实现两部分的均衡隐私预算分配。该方法在保持联合ℓ₂敏感性及原始(ε,δ)-DP保证的前提下，有效缓解了因范数不平衡导致的隐私预算浪费问题；实验表明，在架构校准的α值下，α-split可恢复与全局固定剪裁相当的WER性能，并在仅增加2.6%语言模型噪声的前提下，使编码器获得4.47倍于原方案的、针对基于说话人语音的梯度反演攻击的更优隐私保护能力。
链接: https://arxiv.org/abs/2609.11762
作者: Jordi Luque,Fernando López,Aleix Sant
机构: Telefónica Innovación Digital(电信创新数字); Universidad Autónoma de Madrid(马德里自治大学); Universitat Politècnica de Catalunya(加泰罗尼亚理工大学)
类目: Computation and Language (cs.CL)
备注: Accepted in SLT2026
Abstract:Per-layer differential privacy (DP) clipping improves gradient fidelity in federated learning by allocating per-matrix clipping budgets proportional to parameter count. We show that this recipe breaks for speech large language models (speech-LLMs), when the acoustic encoder and the language decoder differ by an order of magnitude in update norm. Single-pool per-layer methods suffer \emphcross-component budget collapse, dragging word error rate (WER) far from flat global clipping or collapsing training entirely. When the norm imbalance is milder, adaptive single-pool methods partially recover, confirming that collapse severity scales with the inter-component norm ratio. We empirically diagnose the root cause across six per-layer methods and three speech-LLM architectures. We then propose \emph \alpha -split, a two-pool allocation that normalises encoder and LLM parameters into independent pools, and show that joint \ell_2 sensitivity and the original (\varepsilon,\delta) -DP guarantee are unchanged. At architecture-calibrated \alpha , our method recovers WER utility compared to flat DP, while granting the encoder 4.47\times tighter per-component noise protection against speaker voice-based gradient-inversion attacks at only +2.6% LLM noise overhead.
[NLP-16] SIRF: A Spec-Internalized Risk Foundation Model for Industrial Content Risk Control EMNLP2026
【速读】： 该论文旨在解决工业内容风险控制中实际部署面临的瓶颈问题，即在保证高精度与秒级延迟的前提下，系统能够自动处理的风险比例（即“可自动处理的风险量”）不足。传统方法受限于规则外显化带来的高延迟与低效推理，难以满足实时性要求。其解决方案的关键在于提出SIRF（Spec-Internalized Risk Foundation Model），通过持续预训练（CPT）将平台复杂策略内化至模型权重中，这些策略由EntiGraph、MAGA重写及账户级思维链（CoT）合成，无需额外人工标注。该方法实现了仅输出判定结果的超低延迟部署，在保持通用能力不受损的前提下，仅使用约7000万条CPT数据便使黑样本召回率@P95提升至71.3%，较基线提升15.1个百分点。受控同源对比实验表明，性能提升主要归因于策略内化机制。此外，SIRF作为树形模型裁决层部署后，可恢复20%更多误判样本，并在冻结场景下以极低成本实现约70%的误罚率相对降低，展现出卓越的工程落地价值与泛化能力。
链接: https://arxiv.org/abs/2609.11752
作者: Suwan Wu,Yumeng Lin,Pengcheng Yuan,Xiaolong Jiang
机构: Xiaohongshu Inc.(小红书); Tianjin University(天津大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 14 pages, 12 figures. Accepted at the Industry Track of EMNLP 2026
Abstract:For industrial content risk control, the real deployment constraint is not average accuracy but how much risk can be auto-handled under high precision and second-level latency. We present SIRF (Spec-Internalized Risk Foundation Model), which internalizes a platform’s complex policies, synthesized without additional human annotation via EntiGraph, MAGA rewriting and account-level chain-of-thought (CoT), into the weights via continued pretraining (CPT), so rules are applied at high precision under an ultra-low-latency, verdict-only deployment. A controlled same-source comparison (Qwen3-8B-SFT vs. SIRF-8B-SFT, identical policy injection and verdict-only output form, differing only in policy-grounded CPT) attributes the gain to internalization: SIRF-8B-SFT reaches 71.3% Black Recall@P95, +15.1pp over the baseline, using only ~70M CPT tokens without harming general ability, and among included, logprob-available models under this interface it matches or exceeds far larger systems. SIRF is deployed as a tree-model adjudication layer (20% more mis-penalized samples recovered) and transfers to a freezing scenario at low cost (~70% relative mis-penalization reduction).
[NLP-17] LOCUS: Task-Aware Low-Rank Post-Training for Token-Efficient Language Generation
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）服务成本随输出序列长度线性增长的问题，尤其针对标准偏好对齐（preference alignment）方法在不提升实际效用的情况下过度增加响应冗余的现象。其核心挑战在于如何在保持模型偏好对齐性能的前提下，有效控制生成文本的长度以降低计算开销。解决方案的关键在于引入一种任务感知的低秩适应（low-rank adaptation, LoRA）子空间选择机制——LOCUS，该方法通过在固定主干网络（frozen backbone）的基础上，仅更新极少量参数（0.24–0.28%），在低秩子空间内优化生成长度，以最小化输出令牌成本，同时满足预设的效用约束。实验表明，在Anthropic HH-RLHF对话偏好数据集上，使用两个约30亿参数的解码器架构（Pythia-2.8B与Qwen2.5-3B）进行评估时，LOCUS可在不显著改变内部偏好诊断指标的前提下，使续写长度分别减少最高达39.84%和14.87–17.58%，显著提升了生成效率。
链接: https://arxiv.org/abs/2609.11739
作者: Dongfang Zhao
机构: University of Washington (华盛顿大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Large language model serving costs scale directly with output sequence length, yet standard preference alignment often inflates response verbosity without improving utility. We study whether the parameterization of post-training updates affects generation length: low-rank subspaces alter sequence length without modifying the alignment loss. We present LOCUS, a method that selects a task-aware low-rank adaptation subspace to minimize output-token cost subject to a utility constraint. Within this subspace, post-training retains the native preference objective with a frozen backbone. Across Anthropic HH-RLHF dialogue preferences, we evaluate two \sim 3B decoder backbones, Pythia-2.8B and Qwen2.5-3B, against protocol-matched full-parameter DPO and DrDPO branches and the released SamPO checkpoint. LOCUS reduces continuation length by up to 39.84% on Pythia-2.8B and by 14.87–17.58% on Qwen2.5-3B while updating only 0.24–0.28% of model parameters, with no material change in the internal preference diagnostic.
[NLP-18] he Eloquence submission for Task 2 of the Interspeech 2026 MLC-SLM challenge
【速读】： 该论文旨在解决跨21种语言的多语言多项选择题问答（Multilingual Multiple-Choice Question Answering, MCQA）任务中的关键挑战，尤其关注如何在低资源语言和复杂语音输入条件下提升模型的泛化能力与准确性。其核心问题在于：如何有效利用多模态语音与文本信息，在存在显著语言差异及标签偏倚（label bias）的情况下实现高精度的跨语言理解与推理。解决方案的关键在于提出三种互补性策略：一是通过LoRA微调（Low-Rank Adaptation）对Voxtral-Mini-3B模型进行高效参数调整，结合跨语言数据增强、语音识别（ASR）转录增强与时间戳感知音频裁剪，以提升模型对多语言语音输入的鲁棒性；二是采用冻结的Voxtral-24B大模型并引入多模态上下文学习（Multimodal In-Context Learning, ICL），通过外部提示工程有效缓解强标签偏倚问题，从而显著提升性能；三是构建一种无需训练的三层语音锚定记忆系统（voice-anchored memory），融合声学身份、语义内容与知识图谱，实现对历史语音-语义关联的高效检索与推理。三者均显著优于官方基线，其中基于多模态ICL的方法达到0.81的宏平均准确率，为当前最优结果。
链接: https://arxiv.org/abs/2609.11724
作者: Jordi Luque,Lorenzo Concina,Marco Matassoni,Alessio Brutti,Filippo Vella
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:This paper details the Eloquence team’s approach to Task 2 of the 2nd MLC-SLM challenge at Interspeech 2026, which involves multilingual Multiple-Choice Question Answering (MCQA) across 21 languages. Three approaches are explored. First, we fine-tune Voxtral-Mini-3B via LoRA with cross-lingual data augmentation, ASR transcript augmentation and timestamp-aware audio cropping, achieving 0.72 macro-accuracy on evaluation Phase 2. Second, we apply multimodal in-context learning (ICL) to the frozen Voxtral-24B model to correct a strong label bias, reaching 0.81, our best result. Third, a training-free retrieval system based on a three-layer voice-anchored memory combining acoustic identity, semantic content, and a knowledge graph achieves 0.68. All three systems substantially outperform the official baseline.
[NLP-19] Why Does Post-Training Quantization Work?
【速读】： 该论文旨在解决后训练量化（post-training quantization）在压缩大语言模型（LLM）时为何仍能保持良好下游任务性能这一关键问题。尽管量化引入的权重精度损失会在每一层产生误差，理论上会随网络深度累积并破坏下一词预测能力，但实际中量化后的预训练模型表现出显著更低的隐藏状态误差积累，且性能衰减有限。其核心解决方案在于揭示了两种在预训练过程中形成的内在机制：一是每层新引入的量化误差倾向于与前序层传递的残差误差相互抵消，形成对抗性残差交互（counteracting residual interaction），从而抑制误差的指数级增长；二是语言模型输出头（LM-head）的几何结构对高排名词的得分和概率具有偏好性保留能力，使得模型最置信的预测结果不受量化扰动显著影响。这两项机制共同解释了为何经过多层传播的量化误差仍可维持较小的输出偏差，为后训练量化的有效性提供了理论依据，并在多种模型架构与量化配置下得到验证。
链接: https://arxiv.org/abs/2609.11716
作者: Yuxiang Chen,Michael Beyer,Jun Zhu,Jianfei Chen
机构: Tsinghua University (清华大学); Bosch AI Research (博世人工智能研究中心)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 45 pages, 26 figures, including appendices
Abstract:Post-training quantization compresses large language models (LLMs) by storing their weights at reduced precision, and each quantized weight introduces an error into the hidden states. Naively, these errors should accumulate with depth and corrupt next-token prediction; randomly initialized models accumulate these discrepancies rapidly, whereas quantized pretrained models accumulate much less hidden-state error and largely maintain downstream task performance, even though they were never trained with quantization noise. This raises the question we address: why does post-training quantization work? Comparing full-precision and quantized forward passes, we identify two mechanisms that characterize pretrained quantization robustness. First, the error a layer newly introduces tends to oppose the error it inherits from the layer’s input. The two cancel partially such that the discrepancy between full-precision and quantized passes grows slowly. This counteracting residual interaction develops during pretraining. Our quantitative analysis identifies it as a major factor slowing hidden-error growth. Second, LM-head geometry preferentially preserves the scores and probabilities of high-ranked tokens, which typically represent the model’s most confident predictions. Together, these mechanisms explain why quantization error that passes through numerous layers can still produce only small output changes, and we verify the findings across models and quantization settings.
[NLP-20] Negative Self-Distillation: Learning to Reason by Avoiding Flaws
【速读】： 该论文旨在解决生成式语言模型（LLM）在采用基于策略自蒸馏（On-Policy Self-Distillation, OPSD）进行自我改进时，因强制学生模型模仿带有特权信息的过度自信推理路径而导致复杂推理能力退化的问题。其核心问题是：OPSD通过依赖真实答案作为教师信号，抑制了模型表达不确定性及进行探索性、自我修正行为的能力，从而损害了模型在高难度推理任务中的表现。为应对这一挑战，论文提出负向自蒸馏（Negative Self-Distillation, NSD），其关键创新在于不再模仿正确解法，而是通过模型自身生成特定于问题的“错误范例”（如“粗心推理者”角色），并引导学生模型的输出分布远离该自生成的负向教师。为避免对基础语言能力造成破坏，研究设计了一种动态门控机制，可自动识别并隔离与推理过程相关的关键标记，确保梯度更新仅作用于行为缺陷，而保留模型的语言先验知识。实验证明，NSD在多个基准上均显著优于OPSD及其他无标签自增强强化学习方法。
链接: https://arxiv.org/abs/2609.11699
作者: Rongcan Pei,Zhepei Wei,Shuyao Xu,Xinyu Zhu,Wei-Lin Chen,Yu Meng
机构: University of Virginia (弗吉尼亚大学); Stanford University (斯坦福大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 23 pages, 7 figures
Abstract:On-Policy Self-Distillation (OPSD) has emerged as a popular paradigm for large language model (LLM) self-improvement, allowing models to act as their own teachers by leveraging privileged information such as ground-truth solutions. However, recent findings indicate that OPSD can severely degrade the performance of LLMs on complex reasoning tasks: By forcing the student to imitate an artificially confident reasoning trace conditioned on privileged information, OPSD inadvertently suppresses expressions of uncertainty and penalizes the exploratory, self-corrective behaviors required to solve challenging problems. To address this, we introduce Negative Self-Distillation (NSD), a new framework that optimizes LLMs by diverging from flawed reasoning rather than imitating privileged solutions. Instead of relying on ground-truth answers or external supervision, NSD uses the model itself to generate a question-specific negative condition (eg, acting as a ``careless reasoner’') and pushes the student’s distribution away from this self-generated negative teacher. Naively applying unlearning objectives to achieve this divergence is problematic, as flawed reasoning tokens are confounded with basic linguistic tokens; indiscriminately penalizing both risks catastrophically degrading the model’s foundational language capabilities. We resolve this by designing a dynamic gating mechanism that automatically identifies and isolates reasoning-critical tokens, ensuring gradient updates target only behavioral flaws while preserving the model’s linguistic priors. Empirically, NSD consistently outperforms OPSD and other label-free, self-bootstrapping reinforcement learning (RL) baselines.
[NLP-21] Structured Transforms for Low-Overhead Quantization of Language Models
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）中权重量化（weight quantization）带来的精度损失与计算效率瓶颈问题，特别是针对基于Kashin分解的量化方法在收敛性、计算开销及数值稳定性方面的不足。其核心挑战在于如何在保持低比特量化（如4比特每通道）精度的同时，实现高效且稳定的优化过程，并适配实际硬件部署需求。解决方案的关键在于提出一种改进型算法：采用符号随机化的离散余弦变换（sign-randomized Discrete Cosine Transform, DCT）替代原有的稠密随机正交矩阵，将每次迭代的计算复杂度从O(N2)降低至O(NlogN)，显著提升效率；同时引入一种贪婪交替更新的优化算法，确保每个因子满足四峰分布以支持稳定的2比特聚类，并可通过闭式解直接初始化聚类中心，有效规避了先前方法中多重启k-means带来的计算开销与收敛不确定性。此外，该方法结合OPTQ风格的逐层误差补偿与QuIP风格的非相干预处理，构建出在JAX平台上运行的高效流水线，在OPT、Llama-2和Pythia等模型上于4比特每通道条件下达到与OPTQ、QuIP、QuIP-RG以及无微调和向量量化版本的QuIP#相当的性能，且具有更优的墙钟时间扩展性。更重要的是，所提出的有界ℓ∞分解展现出显著的数值鲁棒性：在其他QuIP变体出现发散至四位困惑度（Pythia-6.9B）或在LDL回代过程中产生NaN（Mistral-7B）的极端压力配置下，Kashin-DCT仍能维持数值稳定，接近FP16基线表现。推理时，每个权重可分解为每通道两个2比特因子码，结构上天然适配原生2比特硬件加速。
链接: https://arxiv.org/abs/2609.11687
作者: Daria Cherniuk,Alexander Rudikov,Boris Kashin,Ivan Oseledets
机构: Institute of Numerical Mathematics(数值数学研究所); Steklov Mathematical Institute(斯捷克洛夫数学研究所)
类目: Computation and Language (cs.CL)
备注:
Abstract:We revisit Kashin-decomposition-based weight quantization for large language models and propose an improved algorithm with stronger convergence properties and structured, efficient orthogonal transforms. The method retains the core factorization of each weight into two components – one with bounded infinity norm and the other with bounded infinity norm after an orthogonal transformation – but replaces the dense random orthogonal matrix with a sign-randomized Discrete Cosine Transform (DCT), reducing the per-iteration cost from \mathcalO(N^2) to \mathcalO(N \log N) . The proposed greedy algorithm with alternating updates guarantees the four-peak distribution required for stable 2-bit clustering of each factor and admits closed-form initialization of cluster centers, removing the multi-restart k-means bottleneck of prior work. Composed with OPTQ-style sequential error compensation and QuIP-style incoherence preprocessing, the resulting JAX pipeline is competitive with OPTQ, QuIP, QuIP-RG and a fine-tuning- and vector-quantization-free variant of QuIP# at 4-bit per channel on OPT, Llama-2 and Pythia, with favorable wall-clock scaling. The bounded- \ell_\infty factorization is also notably robust: on stress configurations where QuIP variants diverge to four-digit perplexity (Pythia-6.9B) or abort with NaNs in LDL back-substitution (Mistral-7B), Kashin-DCT remains numerically stable and stays close to FP16 baseline. At inference time, each weight decomposes into two 2-bit factor codes per channel that are structurally suited to native-2-bit hardware.
[NLP-22] A Training-Free Alignment-Free Approach to Corporate Intelligence: Application to SEC Filings
【速读】： 该论文旨在解决金融披露文本分析中高维密集文本嵌入与大语言模型（Large Language Models, LLMs）所面临的多重挑战，包括上下文窗口限制、幻觉风险、高计算成本，以及独立训练模型间向量空间任意旋转导致的语义不一致问题。其解决方案的关键在于提出一种无需训练、无需对齐的框架，基于确定性的稀疏种子向量（deterministic sparse seed vectors），通过将词串哈希至固定高维基底，从构造上将所有文档及时间维度统一于同一坐标系中，彻底消除对训练或对齐的需求。通过在句子上下文中累积这些种子向量，可生成具有线性组合特性的语料库特定语义签名，从而实现亚秒级文档比对、发行人指纹识别、发行人词汇演变追踪以及主题句提取等功能，且仅依赖普通CPU硬件即可完成。在涵盖多年美国证券交易委员会（SEC）文件（10-K、10-Q、8-K）的语料库上验证表明，诸如波音737 MAX危机、英特尔供应链中断及邦吉收购维特拉等重大企业事件，均能以可解释的语义轮廓形式被精准识别，并可追溯至具体产生该信号的原始句子，整个过程无需领域特定训练或大语言模型推理。
链接: https://arxiv.org/abs/2609.11620
作者: Jean-François Delpech
机构: 未知
类目: Computation and Language (cs.CL)
备注: 26 pages, 2 figures
Abstract:High-dimensional dense text embeddings and large language models face real obstacles in financial-disclosure analysis: context-window limits, hallucination risk, high computational cost, and the arbitrary rotation of vector spaces across independently trained models. We present a training-free, alignment-free framework for corporate intelligence built on deterministic sparse seed vectors. Hashing word strings into a fixed high-dimensional basis places all documents and all temporal epochs in a common coordinate system by construction, removing any need for training or alignment. Accumulating these seed vectors across sentence contexts yields corpus-specific semantic signatures that compose linearly, supporting sub-second document comparison, issuer fingerprinting, tracking of how an issuer’s vocabulary shifts between filings, and thematic sentence extraction, all on ordinary CPU hardware. Demonstrating the approach on a multi-year corpus of SEC filings (10-K, 10-Q, 8-K), we show how material corporate events, among them Boeing’s 737 MAX crisis, Intel’s supply-chain disruptions, and Bunge’s acquisition of Viterra, emerge as distinct, interpretable semantic profiles, each traceable to the exact source sentences that produced it, with no domain-specific training and no LLM inference.
[NLP-23] Complex-Text Robustness Evaluation and Failure Diagnosis for Low-Resource Multilingual Text-to-Speech
【速读】： 该论文旨在解决低资源多语言文本到语音（TTS）系统在面对复杂文本输入时鲁棒性不足的问题。现有评估方法主要聚焦于自然度、说话人相似性和内容一致性等常规测试句，难以揭示系统在处理数字、日期、专有名词、长句、语码转换表达及标点相关结构等挑战性输入时的失效模式。为此，本文提出了一种针对低资源多语言TTS系统的复杂文本鲁棒性诊断框架，从内容一致性、语言一致性和生成稳定性三个维度进行评估，并设计了涵盖泰语、越南语、斯瓦希里语和印尼语的多语言鲁棒性测试方案，覆盖普通句子与多种复杂文本输入类型。为实现高效自动化诊断，引入字符错误率（CER）、语言识别准确率和时长异常率等自动评估指标。此外，提出一种轻量级的文本风险评分（Text Risk Score, TRS），基于可解释的文本特征无需人工标注或模型训练即可预估合成风险，支持生成前的风险分析。实验结果表明，复杂文本输入暴露了系统中系统性的失败模式，如数字归一化错误、专有名词处理缺陷、长文本生成偏差以及语码转换处理不当等问题，且不同TTS系统表现出各异的脆弱性；同时，TRS与内容错误及时长异常具有显著正相关性，验证了其作为低成本预合成风险诊断工具的有效性。
链接: https://arxiv.org/abs/2609.11545
作者: Tianlun Zuo,Ziyu Zhang,Tingzhi Mao,Zhonghua Fu,Lei Xie
机构: 未知
类目: Computation and Language (cs.CL); Sound (cs.SD)
备注: NCMMSC 2026 accepted
Abstract:Low-resource multilingual text-to-speech (TTS) systems have expanded language coverage, but their robustness under complex text inputs remains insufficiently diagnosed. Existing evaluations mainly focus on naturalness, speaker similarity, and content consistency using regular test sentences, while providing limited insight into how multilingual TTS systems fail when handling challenging inputs such as numbers, dates, named entities, long sentences, code-switched expressions, and punctuation-related structures. This paper proposes a complex-text robustness diagnosis framework for low-resource multilingual TTS. We evaluate robustness from three dimensions: content consistency, language consistency, and generation stability. A multilingual robustness testing scheme is designed for Thai, Vietnamese, Swahili, and Indonesian, covering ordinary sentences and multiple types of complex text inputs. We further introduce automatic diagnostic metrics, including character error rate, language identification accuracy, and duration abnormal rate. To support input-level risk analysis before speech generation, we propose a lightweight Text Risk Score (TRS), which estimates synthesis risk from interpretable text features without manual annotation or model training. Experiments on three representative multilingual TTS systems, including OmniVoice, VoxCPM2, and MMS-TTS, show that complex text inputs expose systematic failure patterns that are not fully reflected by ordinary short-sentence evaluation. Different systems exhibit distinct vulnerabilities in number normalization, named entity handling, long-text generation, and code-switched input processing. Furthermore, TRS shows a positive correlation with content errors and duration abnormalities, demonstrating its usefulness as a low-cost pre-synthesis indicator for complex-text risk diagnosis in low-resource multilingual TTS.
[NLP-24] Structural priors for data-efficient language learning EMNLP2026
【速读】： 该论文旨在解决大语言模型在训练过程中对海量数据和高计算资源的过度依赖问题，提出通过结构化迁移（structural transfer）来缓解这一挑战。其核心思路是：先在非语言数据（如音乐、概率语法、元胞自动机等符号数据）上预训练模型，以学习具有通用结构特性的先验知识，从而为后续自然语言建模提供更优的权重初始化。该方法的关键在于利用非语言数据中蕴含的隐含结构，引导模型参数进入更具优势的初始参数空间，从而降低后续语言建模任务中的损失并减少训练过程中的权重偏移。实验表明，此类迁移可有效降低下一词预测损失，但其带来的性能提升在下游语言任务中并不一致，且相较于增加语言数据，非语言数据的迁移效率较低。因此，尽管非语言数据可在特定目标（如下一词预测）下部分替代语言数据，但难以可靠支持更广泛的语言泛化能力。
链接: https://arxiv.org/abs/2609.11505
作者: Yana Veitsman,Jonas Mayer Martins,Jonathan Lautenschlager,Lisa Beinborn
机构: University of Göttingen(哥廷根大学); Germany
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: EMNLP 2026, BabyLM Challenge; 18 pages, 11 figures
Abstract:Efficient language learning requires methods to reduce the reliance on large data and computational resources. We investigate structural transfer: First training models on non-language data to induce useful priors for natural language. This approach is a form of weight initialization for multilingual language modeling. We evaluate transfer via next-token-prediction loss, weight shifts in the model, and downstream linguistic benchmarks. Several symbolic data types - notably music, probabilistic grammars, and cellular automata - yield lower language-modeling loss than random initialization. These gains coincide with smaller weight shifts during subsequent language training, suggesting that structural transfer positions models in a more favorable region of the parameter space. However, a lower loss does not translate consistently into better downstream linguistic performance, and transfer from non-language data is less efficient than additional language data. We conclude that non-language data can serve as a partial substitute for language data for the training objective of next-token prediction but does not reliably support broader linguistic generalization.
[NLP-25] Cross-Lingual Clinical Annotation Projection as Constrained Text Generation: A Six-Language Study
【速读】： 该论文旨在解决多语言临床语料库构建中跨语言临床注释迁移的难题，核心问题是如何在保持原文文本不变的前提下，实现高精度、可验证的字符级注释跨语言迁移。其解决方案的关键在于提出一种受约束的大语言模型（Large Language Model, LLM）直接投影流程：将实体标签直接插入不可变的目标语言文本中，并通过确定性验证与字符偏移量重建，确保注释的准确性和可追溯性。相较于基于候选词的投影方法，该方法在多个语言-实体组合上均实现了显著更高的严格匹配F1分数（如GLM 5.2平均达到0.9201），且具备更强的一致性与鲁棒性，同时结合本地化推理与确定性验证，大幅降低了人工专家介入的时间与成本，为低资源语言的临床自然语言处理资源扩展提供了高效可行的技术路径。
链接: https://arxiv.org/abs/2609.11450
作者: Álvaro Rey-Blanes,Francisco J. Moreno-Barea,Francisco J. Veredas
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 14 pages, 4 figures, 4 tables, submitted to journal
Abstract:Background: To determine whether cross-lingual clinical annotation projection can be formulated as a text-preserving, document-level generative task that produces verifiable character-level annotations for multilingual clinical corpus construction, and to characterize its robustness and computational trade-offs relative to candidate-based projection pipelines. Methods: We developed a constrained LLM projection workflow that inserts entity tags directly into immutable target-language text, followed by deterministic validation and character-offset reconstruction. We evaluated it alongside supervised candidate-span projection and hybrid ML-LLM refinement for transferring Spanish Disease, Symptom, and Procedure annotations into six languages. Evaluation used MultiClinAI gold standard with strict span matching and character-overlap F1 Results: Direct LLM projection achieved the strongest and most consistent performance. GLM 5.2 obtained a mean Strict F1 of 0.9201 across 18 language-entity combinations, while locally deployable Gemma4:31B achieved 0.9133. The best LLM configuration improved Strict F1 over the previous state of the art in all 18 settings, by 0.0564-0.1512, yielding 55,416 grounded mentions with reconstructed offsets. Conclusions: Direct LLM-based projection enables high-quality multilingual clinical annotation transfer and provides a practical approach for extending clinical NLP resources to languages with fewer annotated datasets and language-specific tools. Combined with local inference and deterministic validation, it can substantially reduce expert time and cost for multilingual clinical corpus construction.
[NLP-26] ransClean: A Benchmark for Detecting and Extracting Clean Translations from Large Language Model Outputs
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在机器翻译任务中普遍存在但未受充分关注的“翻译噪声”问题，即模型输出中包含语言标签、解释性文字或双语重复等非翻译内容。针对这一问题，研究的关键在于系统识别并分类噪声模式，构建首个专门用于评估和提升翻译输出洁净度的基准测试集TransClean。其核心解决方案包括：通过分析超过79万条来自12个LLM在22种语言对上的翻译结果，归纳出12种典型的噪声模式，并将其归类为格式噪声与内容噪声；进而构建包含9,900对噪声与纯净翻译样本的可控基准数据集，涵盖8,800个合成样本和1,100个真实人工标注样本。在此基础上，提出两种提取方法：基于翻译质量评估模型的跨度检测法与基于提示工程的生成式抽取法，从而为后续优化模型输出洁净度提供可量化的评估框架与技术路径。
链接: https://arxiv.org/abs/2609.11399
作者: Shenbin Qian,Yves Scherrer
机构: University of Oslo(奥斯陆大学), Norway; Language Technology Group(语言技术组), Department of Informatics(信息学系)
类目: Computation and Language (cs.CL)
备注: Accepted to the Eleventh Conference on Machine Translation (WMT26)
Abstract:Large language models (LLMs) are increasingly used for machine translation, yet their outputs often contain additional text beyond the translation itself, such as language labels, explanations or bilingual repetitions, which we term translation noise. Despite its prevalence, this problem lacks dedicated benchmarks and systematic study. We analyze over 790,000 translation outputs from 12 LLMs across 22 language pairs (LPs) and identify 12 recurring noise patterns, which we group into formatting and content noise. Building on the observed patterns, we construct TransClean, a controlled benchmark of 9,900 pairs of noisy and clean translation outputs, comprising 8,800 synthetically generated instances and 1,100 manually curated authentic instances. We evaluate two extraction approaches on the TransClean benchmark: 1) a span-based extraction method leveraging translation quality estimation models for span detection, and 2) an LLM-based extraction method that prompts an LLM to isolate the translation. Our benchmark and analysis provide the first systematic framework to evaluate and improve the cleanliness of LLM translation outputs.
[NLP-27] SEAR: Segment-Evidence-Aware Routing for Weak-to-Strong Multilingual Speech MCQ
【速读】： 该论文旨在解决多语言对话式语音语言模型在跨语言、跨口音场景下对语音内容理解与问答生成的挑战，核心问题是如何有效构建高质量、多样化且具备语义一致性的多模态选择题（MCQ）数据集，以支持模型在不同语言和口音条件下的泛化能力。其解决方案的关键在于提出一种分段证据感知的数据构建与后训练流程：首先利用语言模型将带时间戳的自动语音识别（ASR）结果转化为连贯的事件片段，并通过边界扩展与原始音频裁剪实现精准定位；随后借助Qwen3.6-27B合成语义互补型多选题，使用Gemini 3.1 Flash-Lite生成声学特征相关的多选题，并经过结构合理性、语义锚定性、答案一致性及目标模型可训练性等多重验证，最终获得覆盖21种语言和口音变体的359,825个经验证的多选题；进一步采用纯文本探针将数据划分为弱依赖文本的题目（用于监督微调）与强依赖音频的题目（用于基于组序列策略优化，GSPO）的强化学习训练，通过去偏置优势、序列级重要性校正与动态过滤机制提升训练稳定性。该系统在最终官方评测中达到90.92%的准确率，显著提升了多语言语音问答任务的性能。
链接: https://arxiv.org/abs/2609.11355
作者: Huy Hoang Le,Long-Bao Nguyen,Minh Tri Dao
机构: 未知
类目: Computation and Language (cs.CL); Sound (cs.SD)
备注:
Abstract:This paper describes our system for Task~2 of the second Multilingual Conversational Speech Language Model (MLC-SLM) Challenge. We adapt Qwen3-Omni-30B-A3B-Instruct with a segment-evidence-aware data and post-training pipeline. A language model converts timestamped ASR into coherent event spans, which are expanded by a boundary margin and cropped from the original recording. We then synthesize complementary semantic MCQs with Qwen3.6-27B and acoustic MCQs with Gemini~3.1 Flash-Lite, followed by structural, grounding, answer-consistency, and target-model trainability checks, yielding 359,825 verified MCQs across 21 language and accent variants. A text-only probe partitions the data into weak, text-answerable items used for supervised fine-tuning and strong, audio-dependent items used for reinforcement learning with Group Sequence Policy Optimization (GSPO), stabilized by debiased advantages, sequence-level importance correction, and dynamic filtering. Our system obtains 90.92% accuracy on the final official evaluation set.
[NLP-28] On the Impact of Anonymization on the Performance of Large Language Models
【速读】： 该论文旨在解决在敏感领域部署大语言模型时，输入数据匿名化对模型性能影响的不确定性问题。随着生成式 AI 在医疗、金融等高隐私要求场景中的广泛应用，如何在保护个人身份信息（Personally Identifiable Information, PII）的同时维持模型的实用性能，已成为关键挑战。其解决方案的关键在于通过系统性实证研究揭示匿名化与模型性能之间的复杂权衡关系：研究发现，更强大的模型（如 Qwen2.5-72B 与 GPT-4o mini）在匿名化后表现出更大的性能下降，表明其对特定实体信息存在更强依赖；同时，匿名化的影响具有显著的任务依赖性——例如，在 TruthfulQA 任务中匿名化反而提升表现，而在以检索为核心的 RGB 任务中则导致灾难性性能衰退。此外，研究进一步指出，可逆的匿名化方法（保留实体唯一性）显著优于不可逆方法（如直接删除），且对模型进行显式提示告知匿名化操作并无明显增益。因此，论文强调匿名化策略必须与具体模型架构和任务特性协同设计，才能实现隐私与性能的最优平衡。
链接: https://arxiv.org/abs/2609.11335
作者: Tobias Deußer,Max Hahnbück,Lorenz Sparrenberg,Tobias Uelwer,Christian Bauckhage,Rafet Sifa
机构: University of Bonn(波恩大学); Fraunhofer IAIS(弗劳恩霍夫信息与通信技术研究所); Lamarr Institute for Machine Learning and Artificial Intelligence(拉马尔机器学习与人工智能研究所); Microsoft Germany GmbH(微软德国有限公司)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:As large language models are increasingly deployed in sensitive domains, anonymizing input data to protect personally identifiable information has become a critical practice. However, the impact of this anonymization on model utility is not well understood. This paper presents a systematic empirical study of the trade-off between privacy and performance. We evaluate five prominent language models across eleven diverse benchmarks, comparing their performance on original versus pseudonymized inputs. Our results reveal that while anonymization generally degrades performance, the effect is highly nuanced. We find that more capable models, such as Qwen2.5-72B and GPT-4o mini, suffer the largest performance drops, suggesting a stronger reliance on specific entity information. The impact is also task-dependent: performance on TruthfulQA improves with anonymization, while retrieval-focused tasks like RGB experience a catastrophic decline. Further experiments show that reversible anonymization techniques that preserve entity uniqueness significantly outperform irreversible ones like redaction, and that explicitly prompting models about anonymization offers no discernible benefit. We conclude that anonymization is not a one-size-fits-all solution and must be co-designed with the model and task in mind to balance privacy and utility effectively. Our findings provide a crucial baseline for developing more robust, privacy-aware AI systems.
[NLP-29] E-CONAN (Entailment CONtradition And Neutral) Benchmarks: Arabic Textual Entailment and Natural Inference Datasets
【速读】： 该论文旨在解决阿拉伯语在自然语言推理（Natural Language Inference, NLI）领域资源匮乏的问题。尽管多语言文本推理研究取得了显著进展，但阿拉伯语相关数据集仍严重不足，限制了模型在该语言上的泛化能力与评估有效性。为此，本文提出E-CONAN基准测试集，其由四类来源的句子对构成：（1）自动翻译句对，（2）经人工验证的机器翻译句对，（3）基于阿拉伯语作为第二语言教学书籍的手工构造句对，以及（4）来自不同新闻媒体的包含谣言的标题对。E-CONAN包含两个数据集：E-CONAN-2（二分类，即文本蕴含识别，RTE）和E-CONAN-3（三分类，即标准NLI任务）。研究通过零样本分类方式评估了9个主流多语言预训练模型在ArNLI、XNLI及E-CONAN上的表现，并进一步评估了5个大语言模型（LLMs）在E-CONAN-3上的性能。同时，引入MARBERT作为代表性的阿拉伯语专用基线模型，对比分析阿拉伯语特定模型与跨语言模型及基于LLM方法在E-CONAN基准上的表现差异。此外，开展了详细的定性与定量错误分析，揭示常见错误模式。该研究的关键在于构建一个来源多样、覆盖广泛的真实世界语料库，从而提供比XNLI和ArNLI更全面、更具鲁棒性的评估框架，为阿拉伯语自然语言推理研究提供了宝贵且公开可用的资源。
链接: https://arxiv.org/abs/2609.11334
作者: Khloud AL Jallad,Nada Ghneim,Ghaida Rebdawi
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Natural Language Inference processes pairs of sentences to extract their semantic relations. NLI has been a hot research topic, integrated as a main component in other NLP applications. Despite significant advancements in textual inference across various languages all around the world, Arabic language still suffers from limited resources in this domain. To address this gap, this paper introduces E-CONAN benchmarks that are composed of sentences pairs from various sources: (1) automatically-translated pairs, (2) human-validated machine-translated pairs, (3) hand-crafted pairs from teaching Arabic as foreign language books, and (4) headlines pairs from different news channels containing rumors. E-CONAN contains two benchmark datasets, E-CONAN-2, a 2-way dataset (RTE) and E-CONAN-3, a 3-way dataset (NLI). Additionally, we have used E-CONAN benchmarks to evaluate 9 state-of-the-art multilingual pretrained models using zero-shot classification. Models were evaluated across the ArNLI, XNLI, and E-CONAN datasets. Results show that E-CONAN is a potentially valuable resource for evaluating model generalization and even for fine-tuning pre-trained models. Its diverse composition, derived from a combination of sources, offers a broader and more robust assessment compared to XNLI and ArNLI. In addition, we have evaluated 5 LLMs on E-CONAN-3 dataset. Moreover, we incorporated MARBERT as a representative Arabic-specific baseline and conducted performance evaluation comparison to demonstrate how Arabic-specific models scale against cross-lingual and LLM-based approaches on the E-CONAN benchmarks. Furthermore, we conducted detailed qualitative and quantitative error analysis to analyze frequent error patterns. E-CONAN benchmarks will be publicly available, we hope that it will enrich research community in Arabic textual entailment and natural language inference.
[NLP-30] he Semantic Elevation Operator and the Closure of the Undecidable Class under Preservation
【速读】： 该论文旨在解决自修改系统（self-modifying systems）中静态语义性质验证的不可判定性问题，其核心挑战在于：传统方法仅关注当前状态是否满足某个性质（P），而自修改系统需评估该性质在系统自身代码重写后是否仍被保持。为此，论文提出一个语义提升算子 ΛΦ，将静态问题“对象 x 是否满足性质 P”转化为动态问题“在变换 Φ 作用下，x 被重写后，性质 P 是否仍被保留”。研究发现，当变换 Φ 为内涵性（intensional，依赖于源代码而非仅函数外延）时，即使突破了罗伊斯定理（Rice’s theorem）所要求的外延性条件，提升后的性质依然不可判定；该结论基于克莱尼递归定理（Kleene’s recursion theorem），而非罗伊斯定理本身。由此证明，不可验证性质类 U 在语义提升算子下封闭，无限迭代该算子可使问题上升至 Π20-完全层次，从而将不可验证性确立为一种结构性事实。进一步地，论文揭示监督回归（supervisory regress）无法终止——任意有限层级的逐级增强验证器均无法提供无条件的性质保真证明。未来工作可从有效拓扑（effective topos）视角出发，将语义提升视为劳弗利（Lawvere）不动点定理的一个实例进行范畴论解读。
链接: https://arxiv.org/abs/2609.11326
作者: Jose Pascual Gumbau Mezquita
机构: 未知
类目: Logic in Computer Science (cs.LO); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Logic (math.LO)
备注:
Abstract:The undecidability of a program’s static semantic properties is governed by Rice’s theorem. Self-modifying systems, however, require analysing not whether a property holds now, but whether it is preserved when the system rewrites itself. We formalise this transition through a semantic elevation operator \Lambda\Phi, which turns the static question “does x satisfy P?” into the dynamic question “is P preserved after x is transformed by \Phi?”. We prove that when \Phi is intensional (depending on the source code, not only on the computed function), the elevated property remains undecidable even though it breaks the extensionality that Rice’s theorem requires; the proof rests on Kleene’s recursion theorem, not on Rice. Consequently the class U of non-verifiable properties is closed under the elevation operator. Unbounded iteration of the operator climbs the arithmetical hierarchy -to \Pi02-completeness- consolidating non-verifiability as a structural fact. We further show that the supervisory regress does not terminate: no fnite tower of increasingly capable verifiers yields an unconditional certificate. A categorical reading of these results in the efective topos, in which elevation appears as an instance of Lawvere’s fxed-point theorem, is left as a direction for future work.
[NLP-31] MultiHuSE: A Multimodal Dataset for Humour Styles and Emotions
【速读】： 该论文旨在解决生成式对话中语用幽默（verbal humour）计算识别的难题，其核心挑战在于需同时理解语言内容、表达风格、情感状态及文化背景等多维度信息。现有方法普遍局限于二分类任务，且缺乏能够同时捕捉幽默心理维度与表达多样性变化的高质量数据集。为此，本文提出MultiHuSE——一个包含2,407段高清视频的多模态数据集，涵盖50名人口统计学特征多样化的演员对1,463个文本样本在四种心理幽默类型（亲社会型、攻击型、自我增强型、自贬型）及中性内容中的表演，部分样本还附带底层情绪标注。该数据集的独特之处在于同一文本由多位演员演绎，从而系统性地刻画了表达多样性。基准实验表明，多模态融合模型在幽默风格分类上优于单模态方法（准确率80.1% vs. 77.4%），尤其在亲社会型幽默上提升显著（从66%增至74%）。尽管文本仍为最强单一信号源，但融合模型带来了实质性性能增益。本研究不仅为幽默与情绪之间心理理论提供实证支持，也为人类交流、心理健康及人工智能交互等领域的研究开辟新路径。
链接: https://arxiv.org/abs/2609.11322
作者: Mary Ogbuka Kenneth,Foaad Khosmood,Abbas Edalat
机构: Imperial College London (帝国理工学院); California Polytechnic State University (加州理工州立大学)
类目: Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Multimedia (cs.MM)
备注: 7 pages, 3 figures, 5 tables. Accepted at IEEE CBMI 2025 (International Conference on Content-Based Multimedia Indexing), Dublin, Ireland
Abstract:Computational recognition of verbal humour remains a challenging task, requiring an understanding of language, delivery style, emotions, and cultural context. Most existing approaches focus on binary classification and lack datasets that capture psychological dimensions of humour alongside variations in expression. We introduce MultiHuSE, a multimodal dataset comprising 2,407 high-definition videos of 50 demographically diverse actors performing 1,463 text samples across four psychological humour styles (affiliative, aggressive, self-enhancing, and self-deprecating), as well as neutral content. A subset is additionally annotated for underlying emotions. The dataset uniquely captures multiple actor interpretations of the same texts, enabling systematic analysis of expressive diversity. Baseline experiments show that multimodal fusion outperforms unimodal approaches (80.1% vs. 77.4% accuracy) in humour style classification, with particularly strong gains for affiliative humour (66% to 74%). While text provides the strongest individual signal, fusion models deliver meaningful improvements. We hope that MultiHuSE provides empirical support for psychological theories linking humour and emotion, while also opening new avenues for research in human communication, well-being, and AI-driven interaction. The dataset is available for academic use under an End-User Licence Agreement.
[NLP-32] Automatic Lyric Transcription for Greek Songs: Scaling and Task Composition Effects in Whisper Adaptation INTERSPEECH2026
【速读】： 该论文旨在解决低资源语言（如希腊语）中自动歌词转录（Automatic Lyric Transcription, ALT）面临的挑战，尤其针对旋律多变性、节奏不规则性及伴奏干扰等问题。由于希腊语缺乏先前的ALT基准，研究者首次开展了对Whisper模型在希腊语ALT任务中的适配性控制研究。其解决方案的关键在于：通过源分离与基于CTC的强制对齐构建段落级对齐的希腊语演唱数据集；探索模型规模扩展效应、多任务学习中“转录-翻译”比例的任务组合设计，以及两阶段语音到演唱的迁移策略。实验表明，模型规模扩大可持续提升性能，而多任务学习作为正则化手段，在小容量模型中尤为有效；采用两阶段适配的Whisper Large-v3模型实现了27.2%的词错误率（Word Error Rate, WER），显著优于零样本基线，从而建立了首个希腊语ALT基准。
链接: https://arxiv.org/abs/2609.11302
作者: Maria Frangiadaki,Dimitrios Damianos,Kosmas Kritsis,Vassilis Katsouros
机构: 未知
类目: Computation and Language (cs.CL); Sound (cs.SD)
备注: Accepted at Interspeech 2026
Abstract:Automatic Lyric Transcription (ALT) remains substantially more challenging than speech recognition due to melodic variability, rhythmic irregularity, and accompaniment interference. This is heightened in low-resource languages like Greek, where no prior benchmark for ALT exists. We present the first controlled study of Whisper adaptation for Greek ALT, investigating model scaling effects, task composition via multitask training in transcribe-translate ratios, and two-stage speech-to-singing adaptation. We also curate a segment-level aligned singing dataset based on the Greek Audio Dataset (GAD) using source separation and CTC forced alignment. Results show that scaling consistently improves performance, while multitask learning acts as a beneficial regularizer primarily for smaller-capacity models. The 2-stage adaptation in Whisper Large-v3 achieves a Word Error Rate (WER) of 27.2%, a significant improvement over zero-shot baselines, establishing the first Greek ALT benchmark.
[NLP-33] Xiaomi-CocktailASR-1 Technical Report
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）驱动的自动语音识别（ASR）系统在多说话人场景下的局限性，尤其是“鸡尾酒会问题”（cocktail party problem）这一核心挑战。现有基于时序语音识别（TS-ASR）的方法，包括采用说话人嵌入的端到端架构及最新的基于大语言模型的探索，普遍存在单说话人性能下降以及当目标说话人未出现时无法有效拒绝识别的问题。为应对上述问题，本文提出了一种基于大语言模型的端到端目标说话人语音识别架构——Xiaomi-CocktailASR-1。其关键创新在于利用参考语音作为声纹提示（voiceprint prompt），直接实现对目标说话人语音的转录，无需依赖显式的语音分离步骤。该方案不仅在单说话人场景下保持了与主流ASR模型相当的性能，还具备负样本拒绝能力——当目标说话人未出现在混合语音中时，系统可输出空文本以避免错误识别。此外，Xiaomi-CocktailASR-1支持思维链（Chain-of-Thought, CoT）推理模式，能够生成可解释的推理过程。在多种合成与真实世界多说话人基准测试中的大量实验表明，该模型在统一架构下实现了多说话人与单说话人识别准确率的平衡，并显著提升了系统鲁棒性，从而有效解决了鸡尾酒会问题。
链接: https://arxiv.org/abs/2609.11274
作者: Yiru Zhang,Hang Su,Lichun Fan,Ying Zeng,Chang Liu,Yifeng Wang,Yuquan Liang,Tao Li,Lian Li,Wenhao Yang,Jian Luan,Cong Zou,Heng Qu
机构: Xiaomi Inc.(小米公司)
类目: ound (cs.SD); Computation and Language (cs.CL); Audio and Speech Processing (eess.AS)
备注:
Abstract:Recently, large language model (LLM) based ASR models have achieved significant progress, yet they generally lack support for multi-speaker scenarios, where the cocktail party problem remains a critical bottleneck for further advancing ASR. Existing TS-ASR methods, including end-to-end architectures with speaker embeddings and latest LLM-based explorations suffer from degraded single-speaker performance and the inability to reject when the target speaker is absent. In this paper, we propose Xiaomi-CocktailASR-1, an LLM-based end-to-end TS-ASR architecture. By utilizing reference speech as voiceprint prompts, it directly transcribes the target speaker’s speech without requiring speech separation. Xiaomi-CocktailASR-1 maintains competitive performance in single-speaker scenarios, comparable to mainstream ASR models. It also features a negative sample rejection capability, outputting empty text when the target speaker is absent from the mixed speech. Additionally, Xiaomi-CocktailASR-1 supports a Chain-of-Thought (CoT) reasoning mode to provide explicit reasoning steps. Extensive experiments on various synthetic and real-world multispeaker benchmarks demonstrate that Xiaomi-CocktailASR-1 achieves state-of-the-art performance, effectively addressing the cocktail party problem through a unified architecture that balances multispeaker and single-speaker recognition accuracy, along with rejection capability.
[NLP-34] INDRA: A New AI Tool for Exploring Tobacco Fossil Fuel and Chemical Industry Archives
【速读】： 该论文旨在解决生成式人工智能（Generative AI）在处理历史档案文献时因缺乏对原始证据的直接访问而导致的幻觉（hallucination）与认知偏差问题。现有大规模语言模型（LLM）虽能基于训练数据生成看似合理的回答，但无法区分文档中明确记载的内容与模型自身的推断或先验知识，从而导致输出不可靠。其解决方案的关键在于构建INDRA研究平台，通过系统性嵌入档案学规范，实现对知识生产过程的可追溯性与可审计性。该平台整合了包括加州大学旧金山分校（UCSF）产业文献图书馆、哥伦比亚大学与纽约城市大学（CUNY）的ToxicDocs、斯坦福大学SRITA在内的多个孤立档案库，并引入三项协同保障机制：（1）封闭的证据沙盒限制模型仅访问用户指定的档案语料，杜绝外部信息引入的偏见；（2）实时溯源标记明确区分档案证据与参数化推理内容；（3）由确定性脚本强制执行的系统级协议确保所有输出结构符合学术规范。这一架构从根本上防止了“文档表明X”与“我认为X”之间的混淆，使模型输出成为可核查的研究辅助而非信任对象，显著提升了跨档案研究的可信度与透明度。三个案例研究进一步揭示了该方法在应对“赫拉克利特效应”“踏板困境”及“轻信（或黑手党）问题”中的分析价值与边界。
链接: https://arxiv.org/abs/2609.11261
作者: Daniel Akselrad,Robert N. Proctor
机构: 未知
类目: Digital Libraries (cs.DL); Computation and Language (cs.CL); Computers and Society (cs.CY)
备注: 35 pages, 6 figures, Appendices available at this https URL
Abstract:Five decades of litigation have disgorged hundreds of millions of pages of formerly secret business records from the tobacco industry, along with documents from the makers of drugs, chemicals, food, firearms, and fossil fuels. Yet these archives have been effectively inaccessible to general-purpose large language models (LLMs) because they have never been compiled into an LLM-readable corpus. Chatbots may be familiar with some of the materials contained in such archives but, with no direct access to the documents, they are vulnerable to hallucination and other defects. Here we introduce INDRA, a research platform designed to remedy such failures by embedding the conventions of archival historiography into a system-level protocol governing every output. The platform federates UCSF’s Industry Documents Library, Columbia and CUNY’s ToxicDocs, Stanford’s SRITA, and other heretofore siloed collections, and provides three interlinked safeguards: (1) a closed evidentiary sandbox confines the model to a user-selected corpus, blocking retrieval from external sources that could introduce bias; (2) real-time provenance tagging marks the boundary between archival evidence and parametric inference; and (3) a system-level protocol enforced by deterministic scripts guides the structure of every output. Together these safeguards prevent the model from conflating “the documents say X” with “I think X” or “I learned X from prior training.” The result is an LLM-powered research partner enabling massive multi-archival investigations, a tool whose outputs are designed to be checked rather than trusted, and whose architecture makes the conditions of knowledge production visible and auditable. Three case studies demonstrate the method’s analytical value and limitations, including what we call the Heraclitus effect, the steppingstone dilemma, and the gullibility (or mafia) problem.
[NLP-35] MUtE: A Dual Framework for Concept Erasure and Counterfactual Interventions
【速读】： 该论文旨在解决如何在保持无关概念信息最大化的同时，有效消除表示中特定概念的信息，以缓解模型偏见或增强决策可解释性。其核心挑战在于实现理论最优性与实际表示学习之间的平衡。解决方案的关键在于重新审视概念擦除的理论最优边界，提出一类新型擦除函数，该函数天然诱导确定性的双重反事实映射（dual counterfactual mapping），并引入对反事实轨迹的平移偏差约束，该约束与现代语言模型中多数概念的几何表征方式相一致。这一框架实现了概念擦除与反事实生成之间的无缝切换，在下游任务中显著提升了算法公平性表现，并有效生成了语义合理且具有对比意义的反事实文本。
链接: https://arxiv.org/abs/2609.11253
作者: Antoine Saillenfest
机构: onepoint(欧点); Paris (巴黎)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 21 pages, 3 figures, 6 tables
Abstract:Erasing concept-specific information from representations has been proven useful for mitigating bias or interpreting model decisions. The joint objective is to transform the original representations such that the target concept becomes unpredictable, while maximally preserving concept-unrelated information. In this work, we revisit the optimal bounds of concept erasure to derive a novel class of erasure functions that naturally induce a deterministic, dual counterfactual mapping. Bridging the gap between theoretical optimality and practical representation learning, we design an implementation that imposes a translational bias on counterfactual trajectories - a constraint that aligns with how many concepts geometrically manifest in modern language models. Our framework enables seamless navigation between concept erasure and counterfactual generation. We empirically demonstrate its efficacy in improving downstream algorithmic fairness and generating counterfactual texts.
[NLP-36] he Illusion of Balanced Multimodal Sentiment Analysis: Beyond the Limits of Optimization-Based Methods INTERSPEECH2026
【速读】： 该论文旨在解决多模态情感分析（Multimodal Sentiment Analysis, MSA）中普遍存在的模态不平衡问题。现有方法主要依赖基于优化的平衡策略，但其有效性受限且承诺过高。研究提出三大贡献：首先，构建了一个统一的评估框架，在受控条件下系统测试基于梯度与损失的平衡策略；其次，从理论上诊断了这些方法失效的根本原因——它们将拟合速度（fitting speed）误认为模态的判别性贡献（discriminative contribution），导致错误的权重分配；最后，提出了面向未参与训练模态判别能力估值的研究议程。在CMU-MOSI和CMU-MOSEI数据集上的实验揭示了三方面缺陷：无一种策略能稳定超越晚期拼接（Late Concatenation）；性能对超参数高度敏感；即使采用比例校准也未能带来一致性能提升。核心问题在于：损失函数值并非模态的实际效用（utility），梯度亦不等同于重要性（importance）。因此，模态不平衡问题仍未根本解决，亟需通过保留样本的性能表现来估计模态的真实效用，以实现更可靠的判别性评估。
链接: https://arxiv.org/abs/2609.11247
作者: Ioanna Kaffeza,Efthymios Georgiou,Alexandros Potamianos
机构: 未知
类目: Computation and Language (cs.CL)
备注: Accepted at Interspeech 2026
Abstract:Multimodal Sentiment Analysis (MSA) remains constrained by modality imbalance, yet the field continues to rely on optimization-based balancing methods that promise more than they deliver. We provide three contributions: 1) a unified evaluation framework testing gradient and loss-based balancing strategies under controlled settings; 2) a theoretical diagnosis explaining why these methods fail, as they conflate fitting speed with discriminative contribution; and 3) a research agenda toward held-out discriminative modality valuation. Experiments on CMU-MOSI and CMU-MOSEI reveal three shortcomings: no strategy reliably outperforms Late Concatenation; performance is sensitive to hyperparameters; and even ratio calibration fails to yield consistent gains. The core issue is fundamental: loss is not utility, and gradients are not importance. Modality imbalance remains unresolved, motivating utility estimation from held-out performance.
[NLP-37] Assessing the Reusability of Public Speech Resources for Low-Resource Languages: A Central Kurdish Case Study
【速读】： 该论文旨在解决库尔德语（Kurdish）缺乏高质量语音合成技术的问题，尤其是在自动朗读（text-to-speech, TTS）方面严重不足的现状。尽管研究团队公开发布了35小时的录音数据、三个库尔德语语音模型及配套论文，但其实际可用性受到多重问题制约：数据集中的配置文件列出了未实际使用的设备信息，测试录音被错误地混入训练数据且未标注，同时代码存在处理长数字的缺陷；此外，下载页面宣传的性能表现高于论文所报告结果，并推荐某一语音模型适用于通用场景，这一建议具有误导性，因为库尔德语存在显著的地域差异和书写变体，而这些语音模型仅基于三人朗读预设文本生成，无法涵盖日常口语和区域多样性。与英语和德语相比，库尔德语在词典和语言学描述方面的支持薄弱，导致语音识别错误难以被发现和修正。尽管当前语音输出流畅，但其发音风格仅代表个别说话者的朗读习惯，而非库尔德语整体的语音特征。这些问题大多可通过团队已掌握的信息进行修复，无需修改原有研究结论；更完善的记录将极大提升成果对社区语言学家及其他研究者可复用性和可验证性。然而，唯一难以解决的障碍是开源许可条款——若原始有声书版权方不允许修正版本共享，则未来基于此工作的语音模型开发可能被迫重复起步。
链接: https://arxiv.org/abs/2609.11246
作者: Hiwa Asadpour
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Kurdish is spoken by millions of people, but little technology can read it aloud. A recent study released three Kurdish voices, 35 hours of recorded speech, and a paper describing the work, all free to download. This review checks how well those public files match the paper. The research is careful about its limits, but the files contain several problems: a settings file lists equipment that was never used, test recordings are left unlabeled among training data, and a coding fault mishandles long numbers. The download page also claims a stronger result than the paper reports and recommends one voice for general use. That recommendation matters because Kurdish has major regional and written variation, while these voices were built from three people reading prepared texts. The process therefore removes much everyday and regional speech. English and German benefit from long traditions of dictionaries and linguistic description that help identify wrong pronunciations; Kurdish has far less such support, so software choices can go unchecked. The voices sound fluent, but they represent the reading styles of their speakers rather than Kurdish as a whole. Most of these issues can be fixed using information the team already has, without changing the reported results. Better records would mainly make the work easier for others, especially community linguists, to check and reuse. The license is the main exception: whether audiobook owners allow corrected versions to be shared will affect whether future Kurdish voices can build on this work or must start again.
[NLP-38] OmniHallu: Unified Hallucination Detection for Cross-Modal Comprehension and Generation in 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）在跨模态任务中普遍存在幻觉（hallucination）的问题，即生成内容与输入语义不符或失真。现有方法多局限于单一模态或特定任务类型的幻觉检测，缺乏跨模态和跨任务的通用性。为此，作者提出OmniHallu——一个统一的幻觉检测框架，覆盖图像、视频、音频三种模态，并涵盖理解（comprehension）与生成（generation）两类任务。其核心解决方案包括：构建OmniHallu-Bench基准数据集，包含10,000个样本，具备细粒度的声明级人工标注，涵盖六种跨模态任务（如图像到文本、文本到视频等）；采用多智能体架构，将模型输出分解为原子化声明，通过模态特异性专家进行验证，并借助结构化推理聚合证据；进一步设计一种偏好优化的可训练验证器，逼近多智能体决策边界，在减少66%专家调用次数的同时保持性能几乎不变。实验揭示了显著的模态依赖性能梯度，并提供了对跨模态幻觉模式的精细化洞察。
链接: https://arxiv.org/abs/2609.11244
作者: Jianjiang Yang,Peihang Li,Shanqing Xu,Mengchen Qian,Lu Zhang,Meng Luo
机构: The University of Manchester(曼彻斯特大学); The University of Hong Kong(香港大学); Huazhong University of Science and Technology(华中科技大学); Shanghai Academy of Educational Sciences(上海教育科学研究院); National University of Singapore(新加坡国立大学)
类目: Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to Findings of EMNLP 2026. 12 pages, 4 figures
Abstract:While Multimodal Large Language Models (MLLMs) have achieved remarkable progress across diverse tasks, they suffer from hallucinations where generated outputs contradict or misrepresent input semantics. Existing research typically addresses hallucination detection within a single modality or task type, limiting generalizability. We introduce OmniHallu, a unified hallucination detection framework spanning both comprehension and generation tasks across image, video, and audio modalities. We contribute OmniHallu-Bench, a 10,000-sample benchmark with claim-level human annotations covering six cross-modal tasks: image-to-text (I2T), video-to-text (V2T), audio-to-text (A2T), text-to-image (T2I), text-to-video (T2V), and text-to-audio (T2A). Our multi-agent architecture decomposes model outputs into atomic claims, verifies them through modality-specific experts, and aggregates evidence via structured reasoning. We further propose a preference-optimized trainable verifier that approximates the multi-agent decision boundary, reducing expert calls by 66% with minimal performance loss. Extensive experiments reveal a consistent modality-dependent performance gradient and provide fine-grained insights into cross-modal hallucination patterns.
[NLP-39] Automated Identification of Competing Narratives in Political Discourse on Social Media ECIR2025
【速读】： 该论文旨在解决社交媒体平台上政治话语中竞争性叙事（Competing Narratives）的自动识别与分析问题，尤其关注不同政治意识形态间叙事的分化与冲突。其核心挑战在于如何从海量、非结构化的社交文本中提取出连贯的政治故事，并揭示用户群体在关键议题上的差异化框架。解决方案的关键在于提出一个无监督的多阶段分析框架，整合了主题建模（Topic Modeling）、事件检测（Event Detection）与事件关联（Event Linking）等自然语言处理技术，将碎片化文本重构为具有时间与语义连贯性的叙事单元。该方法能够有效捕捉主流议题的多维叙事演化过程，识别出对立立场的核心分歧点，从而实现对政治话语中深层叙事结构的可视化与量化分析。两个关于德国政坛极化议题的案例研究验证了该方法的有效性，为理解数字公共领域的叙事传播机制提供了可操作的技术路径。
链接: https://arxiv.org/abs/2609.11202
作者: Sergej Wildemann,Erick Elejalde
机构: 未知
类目: Computation and Language (cs.CL); Social and Information Networks (cs.SI)
备注: 11 pages, 5 figures. Published in the proceedings of Text2Story 2025, held with ECIR 2025
Abstract:Social media platforms have become central to shaping political discourse, serving as arenas where narratives form and evolve, influencing public opinion. Identifying and analyzing these narratives, particularly when they compete across different political ideologies, is crucial for understanding the dynamics of modern political communication. This paper presents an unsupervised framework for identifying and characterizing competing narratives in political discourse on social media, focusing on German politicians’ tweets. The framework employs a multi-stage pipeline that integrates natural language processing techniques such as topic modeling, event detection, and event linking. By forming data into coherent stories and uncovering the distinct perspectives of user communities, the system is able to detect the key competing narratives, highlighting the divergent framings and conflicts surrounding trending political topics. Two case studies on polarizing political issues demonstrate the efficacy of the methodology, showcasing its ability to uncover and analyze divergent viewpoints. The findings contribute to the broader understanding of how narratives propagate within the digital public sphere and offer insights for policymakers, social media platforms, and researchers interested in monitoring political discourse.
[NLP-40] FlexComp: One Model for Every Ratio in Context Compression
【速读】： 该论文旨在解决现有软上下文压缩（soft context compression）方法中压缩比（compression ratio）固定带来的灵活性不足问题：传统方法在训练和推理阶段均固定压缩比，导致每个压缩比需独立训练模型，且统一的压缩比无法适应不同输入的实际需求差异。其解决方案的关键在于提出一种与方法无关的框架FlexComp，通过马特罗什卡（Matryoshka）式训练，为每个输入实例动态采样不同的内存预算（K），从而实现单一模型支持任意压缩比；在部署阶段，通过基于置信度的级联路由或轻量级学习型K预测器，按输入自适应选择最优内存预算。实验表明，单个FlexComp模型可媲美多个专用固定压缩比模型的性能，级联路由在平均266倍压缩下保持超过98%的最低压缩比准确率，而K预测器仅需一次压缩-解码过程即可实现158–236倍压缩，且在0.7 F1误差范围内接近最优性能；在大规模服务场景下，该预测器可减少50%的上下文键值缓存（KV cache）并提升47%的解码吞吐量。
链接: https://arxiv.org/abs/2609.11192
作者: Kaiyan Zhao,Zhongtao Miao,Akiko Aizawa,Yoshimasa Tsuruoka
机构: The University of Tokyo (东京大学); National Institute of Informatics (信息通信研究所)
类目: Computation and Language (cs.CL)
备注: Work in progress
Abstract:Soft context compression condenses a context into a few memory tokens that a frozen LLM consumes in place of the raw text, but existing compressors fix the compression ratio at training and inference: each deployed ratio requires a separately trained model, and the chosen ratio is applied uniformly to all inputs, whose actual needs vary drastically. We propose FlexComp, a method-agnostic framework that decouples the ratio from both training and deployment: Matryoshka-style training samples the memory budget K per instance, turning one model into an any-ratio compressor, and the budget is then chosen per input by: (1) confidence-based cascade routing or (2) a lightweight learned K predictor. Across ICAE, 500xCompressor, and SAC on MRQA, a single FlexComp model matches separately trained fixed-ratio specialists with minimal degradation. Cascade routing preserves over 98% of the mildest ratio’s accuracy at up to 266x average compression; the K predictor, in a single compression-decoding pass, reaches 158-236x within 0.7 F1 of the mildest ratio. At serving-scale batch sizes, the K predictor cuts context KV cache by 50% and improves decoding throughput by 47%.
[NLP-41] LILA: Calibration-Free Structured Pruning of Large Language Models via Latent Spectral Geometry
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）结构化剪枝过程中依赖校准数据、梯度计算或大型辅助策略网络的问题，这些问题限制了剪枝方法的硬件效率与部署灵活性。其核心解决方案是提出LILA（Latent-Informed Layer Analysis），通过计算全模型与神经元移除后前馈网络（FFN）权重矩阵经验奇异值分布之间的科莫戈罗夫-斯米尔诺夫（Kolmogorov–Smirnov, KS）距离，构建一种无需训练、无需校准数据或辅助网络的闭式谱重要性评分准则。该方法基于谱特性评估神经元重要性，实现了零成本剪枝。实验表明，LILA在不进行微调的情况下，在LLaMA-2-7B模型上以25%稀疏度下比使用4500万参数强化学习策略的PruneNet提升1.57个百分点的零样本准确率，并在所有稀疏度水平上超越基于WikiText-2校准的SliceGPT达6.0个百分点；经过一阶段LoRA恢复微调后，性能与高度校准的SliceGPT基线差距仅0.48个百分点，且全程未使用任何校准数据。神经切线核（Neural Tangent Kernel, NTK）分析进一步验证了该方法相较随机剪枝可实现22倍的功能失真降低，为谱重要性准则提供了理论支持。此外，将LILA扩展至基于KS得分动态分配稀疏预算，可在中等压缩率下实现生成能力的最优保持，并揭示高压缩率下的单层架构瓶颈，展现出卓越的压缩效率与可解释性。
链接: https://arxiv.org/abs/2609.11163
作者: Sankar Behera,Dhruv Singh,Anshika Agnihotri,Raj Kumar Choudhary,Satyadev Ahlawat,Yamuna Prasad
机构: 未知
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:Structured pruning of large language models (LLMs) offers hardware-efficient compression, yet existing methods require calibration data, gradient computation, or large auxiliary policy networks at pruning time. LILA (\emphLatent-Informed Layer Analysis) scores neuron importance via the Kolmogorov–Smirnov (KS) distance between empirical singular value distributions of the full and neuron-ablated feed-forward network (FFN) weight matrix, providing a closed-form spectral rule requiring no training, calibration data, or auxiliary network. Without any fine-tuning, LILA surpasses PruneNet (45M-parameter RL policy) by 1.57~pp in zero-shot accuracy on LLaMA-2-7B at 25% sparsity, and outperforms WikiText-2-calibrated SliceGPT by up to 6.0~pp across all sparsity levels, while preserving the original architecture. After one epoch of LoRA recovery fine-tuning, LILA achieves highly competitive performance, matching the heavily calibrated SliceGPT baseline to within a 0.48~pp margin across LLaMA-2-7B and Phi-2, despite using zero calibration data. A Neural Tangent Kernel analysis confirms a 22 \times reduction in functional distortion versus random pruning, providing theoretical grounding for the spectral importance criterion. Finally, extending LILA to dynamically allocate sparsity budgets via KS-scores yields state-of-the-art generative preservation at moderate compression, while uncovering fundamental single-layer architectural bottlenecks at higher compression regimes.
[NLP-42] A Frag ility Spectrum for Recursive Language-Model Training
【速读】： 该论文旨在解决生成式 AI (Generative AI) 在递归训练过程中因模型生成文本回流至训练数据而引发的输出多样性退化问题。其核心挑战在于：尽管已知递归污染协议会导致多样性崩溃，但不同模型在相同条件下表现出显著差异，且这种差异难以通过参数规模或静态指标预测。研究的关键发现是，模型是否易发生崩溃本质上是其检查点（checkpoint）自身的属性，而非外部因素所致。进一步研究表明，仅通过让模型对自身输出进行两到三轮迭代生成，即可低成本、高效地预测其在多代递归训练中的脆弱性。此外，干预措施如收紧 top-p 参数可有效抑制崩溃，使六类跨规模模型在三轮内实现稳定，而数据侧过滤虽能延缓崩溃但无法彻底阻止。这表明生成过程的内在机制与轻量级自检测试可作为评估和缓解递归污染风险的有效手段。
链接: https://arxiv.org/abs/2609.11149
作者: Yangze Liu,Zhongyi Han
机构: Shandong University (山东大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Model-generated text is finding its way back into training corpora, and there is plenty of evidence that training on such data over and over collapses output diversity. Prior work has studied the phenomenon itself: which protocols and which data mixtures cause collapse. But different models behave very differently under the same process. We fix one recursive contamination protocol and let 13 publicly released checkpoints form an ecosystem that shares a common corpus for five generations. The unique 4-gram outcome after five generations ranges from 0.187 to 0.940 across checkpoints, a roughly five-fold spread: some models are barely touched, others degenerate into repetitive fragments. Changing the composition of the shared pool or mixing in human text keeps the Spearman correlation of the ordering at 0.91–0.97, and changing the random seed keeps it at 0.93–0.98. Whether a model collapses easily under recursive training is, then, a property of the checkpoint itself, and one that has gone largely unexamined. Parameter scale alone does not explain it, since a three-size ladder within one family is not monotonic in size, and none of the static indicators we tested predicts it either. What does work is cheap: let a model iterate on its own output for two or three generations, and its fragility in the larger ecosystem can be inferred from that alone. Collapse speed also responds to intervention. Tightening top-p, which cuts the low-probability tail at generation time, nearly stops collapse within three generations and stabilizes six checkpoints spanning the whole spectrum together, while data-side filtering slows collapse without stopping it.
[NLP-43] he Oligarch Barely Steers Model Collapse in Multi-Model Ecosystems
【速读】： 该论文旨在解决生成式AI模型在递归训练过程中因自产文本回流至训练数据而引发的模型坍塌（model collapse）问题，尤其关注在寡头垄断（oligopoly）市场结构下，少数主导模型对整体系统演化路径的影响。其核心关切在于：集中度升高是否会导致模型坍塌加速，以及后续模型是否会趋向于主导模型的输出风格。解决方案的关键在于通过构建受控生态系统（controlled ecosystems），在13个规模为1–4B参数的开源模型中模拟多模型相互喂养的循环过程，每代均基于按市场份额加权混合的输出池重新训练，并引入探针以极端化市场分配（最高份额达90%）。实验结果表明，在测试范围内，无论市场集中度如何变化，模型坍塌的速度与最终状态几乎保持不变——即存在一种“不变性”（invariance）；即便施加强烈偏置，也无法有效引导模型走向特定终点，仅留下微弱痕迹。真正决定坍塌速度的是供给池内容的来源及其被采纳的难易程度：在固定份额条件下，更换生态系统的参与者可使五代演化漂移幅度变化2.8倍；各成员的敏感度加权指数可解释19个不同场景中68%的速度差异；而用人类文本替换一半训练池则仅将漂移量减半，不改变其方向。因此，研究揭示出：在所考察范围内，集中度既不决定坍塌的终点也不决定其速率，关键变量是填充训练池的文本来源及其传播特性。
链接: https://arxiv.org/abs/2609.11146
作者: Yangze Liu,Zhongyi Han
机构: Shandong University (山东大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:AI-generated text is flowing back into the training corpora of the next generation of models. Recursive training on it drives model collapse, and recent work extends the setting to many models feeding one another – but almost always with the market split evenly, while real generative AI is an oligopoly. Concentration raises two worries: fewer, more uniform sources may make collapse faster, and later models may be dragged toward the oligarch’s output. We test both in controlled ecosystems: 13 open 1–4B models form natural ecosystems of 3 to 13 players, plus an injected probe that pushes the top share to 90%; each generation, every model’s output is mixed into a shared pool by market share and every model is retrained on that pool from clean base weights, for five generations. Yet within the range we test, neither worry materializes; what emerges instead is an invariance. Making the split more unequal barely changes the speed of collapse. Destinations move even less: the share and identity knobs shift five-generation endpoints by only a few percent of the drift common to all arms – the ecosystems collapse to nearly the same place. An extreme share paired with the strongest injected bias still does not guarantee steering, and the topic shifts it does produce leave only a faint trace on the ruler that measures collapse. What sets the speed is who supplies the pool and how readily those suppliers are carried along: with every share held fixed, swapping the members of a K=3 ecosystem changes five-generation drift by 2.8x; a share-weighted index of each member’s susceptibility explains the speed differences across nineteen arms with R^2 = 0.68; and replacing half the pool with human text roughly halves drift without changing its course. Within the tested range, concentration sets neither the destination nor the pace of collapse; the pace follows whose text fills the pool.
[NLP-44] Same Day Same Story; One Day Ahead a Different Signal: The Dual Validity of Financial Sentiment
【速读】： 该论文旨在解决金融自然语言处理（Financial NLP）领域中一个关键假设的可靠性问题：即情感分析工具在人工标注上的表现一致性（语义有效性）是否能够保证其在预测市场信号（预测有效性）上的能力。传统流程假定两者衡量的是同一核心构念，但本文通过构建一个涵盖2002–2025年证券集体诉讼数据的高质量语料库（包含70,500条X平台推文与异常股票收益的关联），并引入单标注者人工标注的黄金标准样本，对这一假设进行了实证检验。研究发现，语义有效性与预测有效性的关联性高度依赖于采样方式与评分表示形式：在常规的方法特异性采样下，人工标注一致性更接近于同日的分级关联，而非一日领先关系；而在固定样本量面板设计中，两种时间窗口下的等级相关性相近，但粗粒度排序能力依然较弱。这表明，基准人工一致性的建立仅能反映语义有效性，并不能直接决定模型在预测任务中的排序性能。此外，在存在17.6%垃圾信息的对话环境中，消息数量既无法预测市场损害程度，也无法预测和解金额，进一步揭示了现有方法在噪声环境下的局限性。因此，该研究的关键突破在于揭示了语义有效性与预测有效性之间的非必然关联，并强调需在统一评估框架下重新审视模型选择与指标设计。
链接: https://arxiv.org/abs/2609.11144
作者: AS Aravinthkakshan,Laven Srivastava,Harsh Nandwani
机构: Manipal Institute of Technology; Perssonify
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Social and Information Networks (cs.SI)
备注:
Abstract:Financial NLP has a standard workflow: validate a sentiment tool against human labels, then trust it to extract market signal. This assumes the two evaluations measure the same thing. We test that assumption in a setting where both can be measured at once: a corpus of securities class actions (2002-2025) linking 70,500 X messages to abnormal stock returns, with a single-annotator human labelled gold sample. Running five instruments (VADER, Loughran-McDonald, FinBERT, Twitter-RoBERTa, and an LLM annotator) through one identical pipeline, we find that the relationship between construct and predictive validity depends on the sampling convention and score representation. Under conventional method-specific sampling, human agreement aligns more closely with graded same-day associations than with one-day leads. On a fixed-n panel, however, agreement has similar graded rank correlations at both horizons, while the coarse ordering remains weak. Benchmark agreement therefore establishes semantic validity but does not by itself determine predictive rankings. In a conversation that is 17.6% spam, message volume predicts neither market damage nor settlement size.
[NLP-45] Can LLM s Normalize Databases? A Benchmark and Multi-Agent Framework for Schema Normalization
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在生成满足数据库级约束的结构化输出时的可靠性问题，尤其聚焦于数据库规范化（Database Normalization）过程中的挑战，包括函数依赖推理、无损连接分解及表间约束重建。其核心问题是：当前LLMs在处理从第一范式（1NF）到巴斯-科德范式（BCNF）的规范化任务时，普遍存在对依赖关系推断错误、模式分解不当以及跨表约束恢复失败等系统性缺陷。为系统评估这一问题，研究提出了数据库规范化基准测试（Database Normalization Benchmark, DNBENCH），包含3,275个样本，采用三轴评估协议（语义等价性、结构准确性、逻辑有效性）全面衡量模型表现。针对上述问题，研究提出多智能体模式推理框架（Multi-Agent Reasoning for Schemas, MARS），通过将证据提取、违规诊断与分解规划等任务与模式生成和验证分离，实现模块化协同推理。MARS相较于单提示基线，使DNB-SCORE提升82.0%，显著增强了模型在复杂规范化任务中的鲁棒性与准确性。
链接: https://arxiv.org/abs/2609.11141
作者: Dong-Jae Koh,Huisu Kim,SeongHwan Yoon,Lasse M. Jantsch,Chun-Hee Lee,Seonghyeon Lee,Young-Kyoon Suh
机构: Kyungpook National University (庆北国立大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Large Language Models (LLMs) are increasingly used to generate structured outputs, but their reliability remains unclear when those outputs must satisfy database-level constraints. We study this issue through database normalization, involving reasoning about functional dependencies, lossless join decompositions, and inter-table constraints. We introduce a Database Normalization Benchmark (DNBENCH), comprising 3,275 samples for evaluating LLM-driven database normalization from 1NF to BCNF. DNBENCH uses a three-axis protocol to measure semantic equivalence, structural accuracy, and logical validity. Across Single, Complex, and Real World levels, DNBENCH uncovers recurring failures in dependency inference, schema decomposition, and inter-table constraint reconstruction. We further propose Multi-Agent Reasoning for Schemas (MARS), which separates evidence extraction, violation diagnosis, and decomposition planning from schema generation and verification. MARS improves the DNB-SCORE by 82.0% over the single-prompt baseline. All artifacts will be released upon acceptance.
[NLP-46] Rubric-Aligned Disentangled Evaluation of Human Simultaneous Interpreting
【速读】： 该论文旨在解决生成式人工智能在同声传译（Simultaneous Interpreting, SI）评估中缺乏与人工评分体系对齐的自动评价指标问题，尤其针对以意义传递（LQ）、表达质量（EXP）和感知延迟（LAT）为维度的分段级评估。现有方法在使用结构化大语言模型（LLM）提示与标量监督时，导致评分维度坍塌，出现近零相关性及强维度耦合现象，难以有效区分各评价维度。为此，论文提出在LoRA微调的COMET-KIWI编码器基础上引入双回归头（dual regression heads），在保持模型容量一致的前提下，分离多维度监督信号。在保留的演讲级测试集上，该方法实现了0.388（LQ）和0.301（EXP）的皮尔逊相关系数，优于冻结的COMET-KIWI基线模型。鉴于人工评分者间一致性较低，研究将结果解读为相对于人类评分稳定性的相对提升，目标是为形成性评估提供可靠的排序信号。
链接: https://arxiv.org/abs/2609.11131
作者: Ziyu Zhang,Satoshi Nakamura
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Human simultaneous interpreting (SI) is commonly assessed with analytic rubrics separating meaning transfer, delivery quality, and temporal synchrony, yet no automatic metric is designed for rubric-aligned segment-level SI evaluation. We construct a professionally annotated corpus of 1,101 SI segments with scores for meaning transfer (LQ), delivery quality (EXP), and perceived latency (LAT). We show that structured LLM prompting and scalar supervision collapse rubric dimensions, yielding near-zero correlation with human ratings and strong cross-dimension coupling. To isolate supervision structure under identical backbone capacity, we introduce dual regression heads on a LoRA-adapted COMET-KIWI encoder. On a held-out talk-level test set, the model achieves Pearson correlations of 0.388 (LQ) and 0.301 (EXP), improving over frozen COMET-KIWI. Given low absolute rater agreement, we interpret results relative to human consistency and target stable ranking signals for formative assessment.
[NLP-47] From Repetition to Recognition: Inductive Discovery of Disinformation Narratives
【速读】： 该论文旨在解决现有虚假信息数据集中文本叙事（narrative）挖掘在评估过程中依赖预定义分类体系（closed-world setting）所导致的局限性，即无法发现未包含在参考标签中的新叙事。其核心问题在于：当前的叙事挖掘方法受限于封闭世界的标签设定，难以在开放世界中有效识别真实存在的、但未被预先标注的新型叙事模式。为此，论文提出一个三层次的无监督叙事标签生成评估框架，包括恢复（recovery）、挖掘（mining）和发现（discovery）三个维度，分别用于评估模型在自洽标签体系下的表现、对已有外部标签集的覆盖能力以及在完全无预设标签条件下的创新发现能力。解决方案的关键在于通过引入“发现”这一开放世界评估维度，并结合聚类与图社区检测两类典型管道进行对比分析，揭示二者在不同语境下的互补性——尤其在多主题语料中，基于图的管道能更均衡地保持各主题的标签分布，而聚类方法易将某一主题压缩至极低比例；同时，通过人工验证发现，图方法生成的大量单例标签（singleton narratives，源自单一声明的叙事标签，占图输出30%-62%）虽在数量上看似不具重复性，但经专家确认具备可识别的虚假信息叙事特征，表明叙事的重复性可能在语料外部被识别，而非仅依赖语料内部的共现频率。研究进一步释放了气候阻挠（Climate Obstruction）与PolyNarrative数据集的人工验证叙事候选标签，以支持未来分类体系构建与数据集扩展。
链接: https://arxiv.org/abs/2609.11128
作者: Max Upravitelev,Veronika Solopova,Jing Yang,Charlott Jakob,Alexandra Tsiakalou,Neda Foroutan,Vera Schmitt
机构: Technische Universität Berlin(柏林工业大学); German Research Center for Artificial Intelligence (DFKI)(德国人工智能研究中心); BIFOLD – Berlin Institute for the Foundations of Learning and Data(柏林学习与数据基础研究所); Centre for European Research in Trusted AI (CERTAIN)(欧洲可信人工智能研究中心); Johannes Gutenberg-Universität Mainz(莱茵兰-普法尔茨州吉森大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:In disinformation datasets, narratives are often understood as recurring interpretive patterns that group texts under narrative labels. Recent work formalized narrative mining as inductively inferring narrative labels from corpora, but its evaluation stays tied to predefined taxonomies, a closed-world setting that cannot capture narratives absent from the reference labels. We introduce a three-tier evaluation framework for unsupervised narrative label generation: recovery (against a corpus’s own taxonomy), mining (against external label sets), and discovery (without predefined labels). Applying it, we compare clustering-based and graph-community-based pipelines across seven disinformation datasets, with human validation of discovery on two. The two families are complementary under automated metrics, but in a corpus with two prominent topics, clustering can reduce one topic to 2% of generated labels while graph-based pipelines stay balanced. Discovery validation also reveals many singletons (narrative labels derived from single claims, 30-62% of graph outputs), which clustering cannot produce. Annotators confirm many as recognizable disinformation narratives, suggesting that in open-world discovery the repetition assumed by narrative mining may be recognized outside the corpus, not within it. We release human-validated narrative candidate labels for the Climate Obstruction and PolyNarrative datasets to support taxonomy development and dataset extension.
[NLP-48] KuaiRP Series Role-playing Models Technical Report
【速读】： 该论文旨在解决专用角色扮演模型（role-playing model）在实现高保真度角色表现时面临的多重挑战，核心问题在于如何在深度注入领域知识的同时，避免对模型通用智能能力的严重灾难性遗忘。其解决方案的关键在于提出一种多阶段训练范式：首先通过用户行为模拟与逆向画像筛选构建标准化角色模板及SFT数据管道，实现高效的角色定制；其次在强化学习（Reinforcement Learning, RL）阶段引入基于规则的复合奖励函数，有效抑制生成长度膨胀和重复生成等常见退化现象；最后创新性地提出两阶段在线策略自蒸馏（Two-stage On-Policy Distillation, OPD）框架，并结合累积发散衰减（Cumulative-Divergence Decay, CDD）机制，以领域适配模型为教师、原始基础模型为学生进行知识迁移，从而在保留通用代理能力的前提下，实现深度领域知识的有效注入。实验结果表明，KuaiRP系列模型不仅在目标领域内达到了与当前领先专有模型相当的角色扮演保真度，且显著恢复了通用能力，同时保持极低的部署成本。
链接: https://arxiv.org/abs/2609.11127
作者: Yipeng Wang,Ziwei Zhang,Jiahui Zhang,Qi Gan,Kai Sheng
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:This paper introduces the complete technical solution for the KuaiRP series of role-playing models. We aim to achieve four core objectives for a dedicated role-playing model: simplified prompt engineering, highly stable output quality, built-in domain world knowledge, and high-efficiency deployment with a small parameter size. However, effectively injecting deep domain knowledge often leads to a severe catastrophic forgetting of the model’s general agent capabilities. To overcome this trade-off, we propose a multi-stage training pipeline. First, we design a standardized character template and construct an SFT data pipeline based on user behavior simulation and reverse profile filtering. Next, we utilize a rule-based composite reward function during the Reinforcement Learning (RL) phase to eliminate common degradation phenomena like length expansion and repetitive generation. Finally, to recover the general capabilities compromised during SFT and RL, we propose a novel self-distillation paradigm using Two-stage On-Policy Distillation (OPD) equipped with Cumulative-Divergence Decay (CDD). By using the domain-adapted model as the teacher and the original base model as the student, we effectively balance deep domain knowledge injection with the preservation of general agent capabilities. Experimental results demonstrate that the KuaiRP models not only match the current state-of-the-art proprietary models in role-playing fidelity within our target domains, but also successfully recover general agent capabilities, maintaining extremely low deployment costs.
[NLP-49] Overview of the NLPCC 2026 Shared Task 11: Agent -Based Experiment Reproduction from Scientific Papers NLPCC
【速读】： 该论文旨在解决科学实验可复现性（reproducibility）在日益增长的文献规模与复杂性背景下，难以通过人工手段进行高效验证的问题。现有基于大语言模型（LLM）代理的自动化实验复现方法虽有进展，但评估体系多集中于最终代码仓库，且主要局限于机器学习（ML）领域，缺乏对跨领域（如AI for Science）过程性行为的系统评估。为此，论文提出AgentActionBench——一个面向基于代理的实验复现过程的流程导向型基准测试框架。其关键在于采用基于MCP（Multi-Agent Control Protocol）的动作记录器（Action Recorder），全程捕获代理在实验复现过程中的行为轨迹，并结合论文特异性评分标准（rubrics）对生成的执行轨迹进行评估。该框架涵盖150篇论文（120篇ML、30篇AI4Science），并通过人工标注的10%子集提供验证数据，同时利用模型辅助增强将完整基准扩展至超过10,000个评分项。实验结果表明，当前系统仍受限于执行阶段，而模型生成评分与人工标注间展现出强皮尔逊（Pearson）和斯皮尔曼（Spearman）相关性，验证了其可扩展评分生成方法的可靠性。
链接: https://arxiv.org/abs/2609.11117
作者: Hanhua Hong,Yizhi Li,Luu Gia Huy,Jian Yang,Ming Zhou,Chenghua Lin
机构: The University of Manchester (曼彻斯特大学); IQuest Research; Vietnam National University (越南国家大学); Beihang University (北京航空航天大学); Langboat
类目: Computation and Language (cs.CL)
备注: NLPCC Shared Task
Abstract:Reproducibility is essential to scientific progress, yet the growing volume and complexity of scientific publications make exhaustive manual verification increasingly impractical. Although recent advances in large language model (LLM) agents enable automated experiment reproduction, existing evaluations largely focus on final repositories and are typically limited to machine learning (ML). We introduce AgentActionBench, a process-oriented benchmark for evaluating agent-based experiment reproduction across ML and AI4Science domains. Our framework uses an MCP-based Action Recorder to capture agents’ behaviour throughout the reproduction process and evaluates the resulting traces with paper-specific rubrics. AgentActionBench contains 150 papers, including 120 ML papers and 30 AI4Science papers. A human-annotated subset covering 10% of the benchmark provides validation data, while model-assisted augmentation expands the full benchmark to more than 10,000 rubric items. Experimental results show that current systems remain limited, with execution as the primary bottleneck. Meanwhile, the strong Pearson and Spearman correlations between model-generated and human-annotated rubrics validate the reliability of our scalable rubric-generation approach.
[NLP-50] ProMediConv: Benchmarking Proactive Conversational Agents in Legal Dispute Mediation EMNLP2026
【速读】： 该论文旨在解决现有基于大语言模型（LLM）的调解研究中存在的关键问题：任务设定不切实际、数据集保真度低以及评估指标粗粒度，导致无法有效捕捉调解过程中对话双方的动态行为变化。其解决方案的核心在于提出ProMediConv——一个全新的基准测试框架，将调解建模为一种主动的、多阶段且具备当事人意识的对话过程，整合11种调解策略与4类当事人行为模式（BP）状态，并基于972个完整的现实案例构建了具有话语级标注的高保真数据集。此外，论文创新性地引入MAD（均属性差异）这一细粒度评估指标，能够精准捕捉对话进程中当事人行为模式的演变。通过在该框架下对多种模型进行系统评估，研究揭示了当前模型在动态多主体调解场景中仍面临显著挑战，从而为人工智能辅助冲突解决提供了严谨的评估基础与量化标准。
链接: https://arxiv.org/abs/2609.11101
作者: Zesheng Wei,Mengfan Li,Wenhao Liu,Yixin Zhang,Zilei Wang,Yang Deng
机构: University of Science and Technology of China(中国科学技术大学); Singapore Management University(新加坡管理大学)
类目: Computation and Language (cs.CL)
备注: Accepted to Findings of EMNLP2026
Abstract:Dispute mediation is essential for maintaining social harmony and resilience, yet developing skilled mediators is costly and time-consuming. Existing LLM-based mediation research remains limited by unrealistic task formulations, low-fidelity datasets, and coarse evaluation metrics that obscure turn-by-turn dynamics. To address these gaps, we introduce ProMediConv, a novel benchmarking framework that models mediation as a proactive, multi-stage, and party-aware dialogue process incorporating 11 mediation strategies and four party behavior pattern (BP) states. Using 972 complete real-world cases, we construct a high-fidelity mediation dataset with utterance-level annotations of strategies and BP states. Furthermore, to better assess agent impact, we propose MAD (Mean Attribute Difference), a fine-grained metric that captures BP shifts throughout the dialogue. Leveraging this framework, we establish a comprehensive benchmark by evaluating diverse models alongside our tailored baseline ProMediAgent. Extensive empirical analyses reveal critical behavioral phenomena and underscore the persistent challenges current models face in dynamic, multi-party mediation. Ultimately, ProMediConv provides a rigorous foundation and a vital quantitative standard for advancing AI-assisted conflict resolution. Our dataset and codebase are accessible at this https URL.
[NLP-51] Beyond Solver Verdicts: Generative Reward Models for Autoformalization
【速读】： 该论文旨在解决神经符号系统（Neurosymbolic Systems）中形式化编码的参考等价性（reference-equivalence）保障难题，即数学求解器在执行过程中无法察觉形式化翻译是否严格保持与目标形式化定义的语义一致性。其核心问题表现为“判决保持不忠实性”（Verdict-Preserving-Unfaithfulness, VPU）——即错误的编码虽不等价于原始语义，却仍能通过求解器验证并输出正确判决，导致系统误判。现有基于结构和判决的验证启发式方法在理论上被证明只能以随机水平检测此类欺骗性有效路径。为应对这一挑战，论文提出生成式验证（Generative Verification, GenV），通过将离线Z3等价性断言器（oracle）的知识蒸馏至语言模型的原生词汇空间，构建无需参考、连续可微的参考等价性评分机制。机制解析表明，该方法借助决策投影对数透镜（decision-projected logit lenses）与稀疏自编码器（sparse autoencoders），无需显式定位训练即可自然提取精确的空间误差坐标。实验结果表明，基于矿取的验证器（GenV+HN）在参考等价性验证任务中达到0.961 AUROC，具备零样本跨未见翻译器及异构形式风格的泛化能力，并在代理型测试时计算资源分配任务中带来11.3个百分点的下游准确率提升。
链接: https://arxiv.org/abs/2609.11085
作者: Vikash Singh,Debargha Ganguly,Aman Goel,Ali Torkamani,Xiaoxue Han,Joseph Lilien,Ferhat Erata,Vipin Chaudhary
机构: Case Western Reserve University (凯斯西储大学); Amazon Web Services (亚马逊网络服务)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:Neurosymbolic systems rely on mathematical solvers to guarantee reasoning correctness, yet solvers are fundamentally blind to whether a formal translation maintains strict reference-equivalence to a designated formalization. We formalize this vulnerability as Verdict-Preserving-Unfaithfulness (VPU): a failure mode where an incorrect encoding executes successfully and matches the expected verdict. We theoretically prove that structural, verdict-only verification heuristics are mathematically bounded to chance-level detection on these deceptively valid traces. To resolve this, we introduce Generative Verification (GenV), which distills an offline Z3-equivalence oracle into a reference-free, continuous reference-equivalence score by repurposing the language model’s native vocabulary space. Mechanistic analysis via decision-projected logit lenses and sparse autoencoders shows this generative readout natively extracts precise spatial error coordinates without explicit localization training. Empirically, our oracle-mined verifier (GenV+HN) achieves 0.961 AUROC in reference-equivalence verification, generalizes zero-shot across unseen translators and divergent formal styles, and yields an 11.3-point downstream accuracy gain in agentic test-time compute allocation.
[NLP-52] When Noise Fabricates Bias: The Frag ility of LLM -as-a-Judge Bias Measurement under Noisy Text
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在评估文本社会偏见时，因输入文本存在表面噪声（如拼写错误、非正式拼写和标点断裂等）而导致的偏见测量失准问题。其核心发现是：表面噪声对偏见判断的影响并非对称，而是显著倾向于将原本中立的判断误判为具有偏见，其概率最高可达120倍之差。研究进一步揭示，在四名不同鲁棒性的LLM裁判中存在两个非直观现象：最脆弱的模型在轻微但现实的噪声水平下表现出最纯粹的偏差扭曲，此时文本擦除最少；而随着模型鲁棒性增强，偏差扭曲逐渐减弱并趋于平衡，而非反转。因此，基于含噪文本所测得的偏见值存在系统性高估，尤其在关乎公平性的关键类别中更为严重。该研究的关键解决方案在于揭示噪声对偏见评估的非对称干扰机制，并强调在实际应用中需对输入文本进行去噪处理以确保社会偏见测量的可靠性。
链接: https://arxiv.org/abs/2609.11067
作者: DongHyun Ryu,Jaehyeok Lee,YeongJun Hwang,JinYeong Bak
机构: Sungkyunkwan University (成均馆大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 15 pages, 4 figures. Accepted at W-NUT 2026. Code: this https URL
Abstract:Large language models are increasingly used as judges to measure social bias in text, yet the passages they judge are often noisy, containing typos, informal spelling, and broken punctuation. The consequences of such surface noise for social bias measurement remain unclear. To investigate this question, we apply five realistic noise conditions at multiple intensity levels to 3,822 stereotype-related responses and compare the resulting bias judgments with those on the original text. We find that such surface noise does not degrade bias measurement symmetrically: it is far more likely to turn neutral judgments into biased ones than biased judgments into neutral ones, by up to a 120x margin. We further observe two non-obvious effects across four LLM judges: in the most fragile judge the distortion is at its purest at mild, realistic noise levels, where erasure is scarcest, and as judges grow robust it attenuates toward parity rather than reversing. Bias measured on noisy text is therefore systematically overestimated, most in the categories that matter most for fairness.
[NLP-53] he information geometry of large language models is shared learned and controllable
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在行为模式上表现出高度相似性，但其内在共享结构尚不明确，且在修改某一行为时易对其他行为造成干扰的问题。其核心解决方案在于引入费舍尔-罗阿几何（Fisher-Rao geometry），揭示了下一词概率分布的几何结构与模型行为之间的深层关联：模型的行为决定了输出空间中的费舍尔-罗阿几何结构（至多保留输出不变对称性），而激活空间的几何则依赖于坐标表示。研究发现，在Transformer、状态空间模型和循环神经网络等多种架构中，输出几何的一致性远高于激活几何，且共享的几何结构支持语义类别间的迁移。随着模型预测准确性、规模及训练程度的提升，模型生成的词选择与人类选择的契合度显著增强，经仅模型校准后进一步改善。此外，词概率与读出几何共同可预测特征谱及其有效维度。通过受控的语言分配实验表明，几何结构遵循跨架构的语言规律。预训练语料统计信息可无须重新校准地预测未见事实的习得能力；随机化实验显示，更深层次的证据构造会显著延迟所有测试架构的事实获取。更重要的是，该几何结构能够指导最小扰动的局部干预，预测其相对成本，并支持可复用的控制策略：在源提示上学习的更新可迁移至未见过的提示，同时比基于欧几里得空间的控制方法更好地保持参考提示上的行为一致性。相同的几何修正还能统一提升模型控制、编辑、归因、词典学习和微调等任务的性能。
链接: https://arxiv.org/abs/2609.11063
作者: Dario Picozzi
机构: University College London (UCL); London Centre for Nanotechnology
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:Large language models learn similar behaviours, yet it remains unclear what structure they share or how to change one behaviour without disturbing others. The Fisher-Rao geometry of next-token probabilities connects these questions: behaviour determines this geometry up to output-preserving symmetries, whereas activation geometry depends on coordinates. Across transformer, state-space and recurrent models, output geometries agree more strongly than activation geometries, and shared geometry supports semantic-category transfer. Agreement with human word choices increases with predictive accuracy, scale and training, and improves further after model-only calibration. Token probabilities and read-out geometry jointly predict the spectrum and its effective dimension. Controlled language assignments show that geometry follows the language law across architectures. Pretraining corpus statistics predict held-out fact acquisition without recalibration, while randomised experiments show that deeper evidence substantially delays acquisition across every tested architecture and evidence construction. Finally, the geometry prescribes minimum-disturbance local interventions, predicts their relative cost, and supports reusable control: updates learned on donor prompts transfer to unseen prompts while better preserving behaviour on reference prompts than Euclidean control. The same geometric correction improves steering, editing, attribution, dictionary learning and fine-tuning.
[NLP-54] Rebalancing Token Importance in Language Models with TF-IDF Weighted Cross-Entropy Loss
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在训练过程中因采用均匀的词元加权机制而导致的过度记忆表面文本片段的问题。这种现象通常由高频但语义信息量低的词元主导学习过程，从而加剧了对训练数据中冗余模式的机械记忆。其解决方案的关键在于提出一种基于信息量加权的交叉熵损失函数（information-weighted cross-entropy loss），通过引入TF-IDF统计量对词元级别的贡献进行重标定，强化语义信息丰富的词元权重，同时降低常见词元的影响。实验表明，该方法在五种解码器架构的LLM（参数规模1.1B至13B）上均能显著减少记忆子串长度，且在保持困惑度和下游任务性能的同时，实现了对记忆行为的有效抑制；在LoRA微调下，平均记忆子串长度减少14%，而在TinyLLaMA 1.1B的全参数微调中，降幅高达58%。该方法具有架构无关性，仅增加不足3%的计算开销，可无缝集成于现有训练流程，是一种轻量级且理论基础坚实的缓解过记忆问题的新策略。
链接: https://arxiv.org/abs/2609.11029
作者: Zhijian Li,Stefan Larson,Kevin Leach
机构: University of Southern California (南加州大学); Vanderbilt University (范德比尔特大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Large language models are typically trained under uniform token weighting, which allows frequent and low-information tokens to dominate learning and can increase the tendency to memorize surface-level text spans. To address this, we present an information-weighted cross-entropy loss that rescales token-level contributions using TF-IDF statistics, emphasizing semantically informative tokens while down-weighting ubiquitous ones. Experiments on five decoder-only LLMs ranging from 1.1B to 13B parameters show consistent reductions in memorized substring length while preserving perplexity and downstream task performance. Under LoRA fine-tuning, TF-IDF reduces average substring memorization length by 14% across all five models. Under full-weight fine-tuning on TinyLLaMA 1.1B, the reduction reaches 58%. Our approach is architecture-agnostic and can be incorporated into existing training pipelines with less than 3% computational overhead, offering a lightweight and principled way to mitigate memorization without disrupting standard training dynamics.
[NLP-55] New Evidence Same Choice: Testing Physical Experiment Selection in Vision Language Models NEURIPS2026
【速读】： 该论文旨在解决生成式模型在物理推理任务中缺乏有效证据选择能力的问题，即模型难以判断何时可基于已有信息直接作答，何时需要进一步获取额外实验数据（如质量、摩擦系数等）以支持决策。其解决方案的关键在于构建一个可控的评估框架，通过设计成对匹配的物理问题实例，明确区分“是否需要补充实验”与“应选择何种最经济的实验”这两个决策步骤。该框架基于已知的所有可能物理世界组合（由两个质量值与两个其他属性值组合而成）和实验成本，能够精确计算最优决策路径。实验结果表明，尽管多数开放模型在不同情境下仍重复相同动作（正确率仅5.9%），且存在测量解读、物理推理及响应格式等方面的系统性缺陷，但该基准通过将证据选择与最终答案解耦评估，揭示了传统仅关注答案准确率的评测方法所忽视的深层认知局限，从而为提升模型的主动推理与实验规划能力提供了关键诊断工具。
链接: https://arxiv.org/abs/2609.11022
作者: Sourajit Saha,Shubhashis Roy Dipta,Nobin Sarwar,Shaswati Saha,Yuxuan Jiang,Siyuan Li,Qiheng Wang
机构: University of Maryland, Baltimore County (马里兰大学巴尔的摩县分校); University of Georgia (佐治亚大学); Independent Researcher (独立研究员)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Under Review at PhysWorldAI @ NeurIPS 2026
Abstract:A model first sees an image from one physical measurement experiment, such as how far a block coasted, and must answer a question about a new trial, such as whether the block will pass a target after a fixed push. The initial experiment may provide enough information to answer, or the model may need another measurement, such as the object’s mass, friction, restitution, or spring stiffness. We study whether vision language models can decide when to answer immediately and, when more evidence is needed, which experiment to perform. Current physical reasoning benchmarks usually evaluate only the final answer, so they do not directly measure this decision-making ability. We introduce a controlled evaluation where each problem provides one measurement image and four possible physical worlds created by combining two possible masses and two possible values of another relevant property. The model must either stop and answer or select the cheapest additional experiment that can resolve the question. We construct matched problem pairs where changing either the observed measurement or the question changes the optimal action. Since all possible worlds and experiment costs are known, we can explicitly determine the optimal choice. Across six open models and 144 physical parameter sets, direct responses repeat the same action for 95.1% to 100% of image pairs even when the correct action changes. Brief reasoning improves action switching, but the best model makes both decisions correctly for only 5.9% of image pairs. Additional analysis reveals failures in measurement interpretation, physical reasoning, and response formatting. By evaluating evidence selection separately from final answers, our benchmark reveals limitations in physical reasoning that conventional answer accuracy can overlook.
[NLP-56] K/V-Cache Interventions Dissociate Representation Alignment from Persona Expression in Decoder-Only Language Models
【速读】： 该论文旨在解决在仅解码器架构的语言模型中实现精细化角色（persona）控制的问题，具体聚焦于通过K/V缓存干预（K/V-cache interventions）来操纵生成内容的角色特征表达。其核心挑战在于：尽管在表示层（representation level）上实现了与目标角色高度对齐的K/V轨迹迁移，但这种对齐并不总能转化为行为层面（behavioral expression）的有效角色特征体现，即存在“表示-行为”脱节现象。解决方案的关键在于识别并利用中间层（mid-layer, layers 9–20）的K/V缓存替换作为最优干预位置——该策略既能实现显著的目标角色标记表达，又能保持较高的词汇多样性；同时，研究揭示了轨迹级移植（trajectory-level transplantation） 的重要性，即被移植的K/V轨迹包含目标角色自身的生成历史，而非简单的表征注入，这使得干预效果不仅依赖于输入序列的一致性，还受到上下文动态演化的影响。此外，研究发现无论采用何种位置扰动（如滞后或打乱），均会普遍抑制目标角色表达，表明该干预表面具有结构性约束，仅靠表示相似性指标不足以预测下游行为表现。因此，该研究强调了在高信号环境下，需结合层位选择与轨迹动态特性，才能有效实现角色控制。
链接: https://arxiv.org/abs/2609.11020
作者: Yu Sun,Mengyin Lu,Cong Feng,Guangming Lu,Huimin Han
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:We study K/V-cache interventions – transplanting a target-conditioned K/V trajectory into a source-persona generation – as a structured surface for persona control in decoder-only language models. Across 13 intervention configurations applied to Llama-3.1-8B for a fixed source-to-target persona pair, we report two consistent dissociations between representation-level alignment and behavioral expression, plus a common failure under position perturbations. First, all layer-band K/V replacements (early, mid, late) achieve strong local V-space alignment (V-gap 0.91, 0.89, 0.84), but only mid-layer replacement (layers 9-20) combines substantial target-marker expression with preserved lexical diversity. Second, full and mid-layer replacement induce comparable alignment (V-gap 0.94 vs. 0.89) yet produce different lexical-diversity profiles (TTR 0.65 vs. 0.77). Third, position perturbations (lag and shuffle) apply distinct operations yet uniformly suppress target-persona expression – a common behavioral failure rather than a strict dissociation. Representation-level similarity metrics alone are thus not sufficient predictors of downstream persona expression in the regimes we study; the K/V cache emerges as a controllable but structurally constrained intervention surface. Because the transplanted trajectory carries the target’s own generated token history, we characterize the intervention as trajectory-level transplantation rather than isolated persona-representation injection; a same-token-sequence control, decoding an identical token sequence under source vs. target conditioning, reproduces the sign and layer localization of the L28 representational shift, indicating the shift is not explained solely by imported token history. These findings characterize representation-behavior dissociation in a high-signal setting rather than establishing universality across models or persona pairs.
[NLP-57] Rethinking Verbalized Confidence for LLM -as-a-Judge: A Compatibility Shift on Post-2025 Proprietary Models
【速读】： 该论文旨在解决大语言模型作为评判者（LLM-as-a-Judge）时，传统依赖对数概率（log-probabilities）进行评分的可靠性问题。随着2025年后顶级私有模型的发展，研究发现以往被视作不精确、易受偏见影响的“口头化置信度”（verbalized confidence）反而成为更稳健的软评分机制，其在多个评估基准（SummEval、AggreFact、HelpSteer2）和多达18个模型上的表现优于对数概率。这一现象被称为“兼容性转变”（compatibility shift）。解决方案的关键在于引入两个新组件：过自信警示（overconfidence advisory）与自我辩论（self-debate），二者协同提升了评分校准性、得分分布的区分度以及对任务主观性的鲁棒性。值得注意的是，后2025年模型在采纳这两个改进时几乎不损失平衡准确性，而前2025年模型则需付出显著代价。因此，相较于基于对数概率的G-Eval，口头化置信度在GPT系列顶级版本中展现出更强的主观性鲁棒性。该转变在仅关注准确率的评估中难以察觉，提示应摒弃默认使用硬预测的做法，推广更广泛的软评分机制应用。总体而言，口头化置信度已从对数概率的次优替代，演变为当代大语言模型评判系统中实用且可靠的软评分方式。
链接: https://arxiv.org/abs/2609.10996
作者: Yu-Chung Hsiao
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Verbalized confidence, long dismissed as overconfident, coarse, and prone to round-number clustering, is now the more robust soft-scoring mechanism for LLM-as-a-Judge on top-tier proprietary models. Across SummEval, AggreFact, and HelpSteer2, spanning up to 18 LLMs, we show that the standard advice to prefer log-probabilities no longer holds on post-2025 models, where verbalized confidence is the better signal. We call this a compatibility shift. On top of a standard verbalized-confidence baseline, we introduce two new ingredients: an overconfidence advisory and self-debate. Together they improve calibration, score-distribution spread, and robustness to task subjectivity. We further observe a generation effect: post-2025 models accommodate these two additions with little balanced-accuracy cost, whereas pre-2025 models pay a measurable penalty. Compared with logprob-based G-Eval, verbalized confidence is the more subjectivity-robust soft signal on GPT-family top-tier releases. The shift is invisible under accuracy-only reporting. Rather than defaulting to hard predictions, we recommend broader use of soft scoring in LLM-as-a-Judge. More broadly, verbalized confidence has moved from a weaker substitute for logprobs to a practical soft-scoring mechanism for contemporary LLM judges.
[NLP-58] Distribution-aware Language Neuron Identification in Multilingual Large Language Models EMNLP2026
【速读】： 该论文旨在解决多语言大语言模型（mLLMs）中语言特异性神经元识别不准确的问题。现有方法基于神经元在不同语言下的激活概率熵来衡量语言特异性，但该方法仅考虑正激活值，忽略了负激活值所蕴含的语言分布信息，难以充分捕捉模型中语言表示的分布性与相互关联性。本文提出一种分布感知的语言特异性神经元选择方法（Distribution-aware Language Neuron selection），其关键在于利用全激活范围内各语言激活分布间的成对重叠系数，通过聚类分析揭示语言间的分布关系，从而更精确地量化神经元的语言特异性。实验表明，该方法在两个mLLMs和两个外部语料上均能更有效地分离语言特异性因果效应，实现单位神经元的靶向语言损伤提升达4.9倍，同时保持非目标语言性能稳定。
链接: https://arxiv.org/abs/2609.10993
作者: Minjun Kim,Inho Won,Junghun Yuk,Dongyeon Kim,Jihyo Kim,KyungTae Lim
机构: KAIST(韩国科学技术院); KAIST InnoCORE PRISM-AI Center(韩国科学技术院创新核心普里兹姆人工智能中心)
类目: Computation and Language (cs.CL)
备注: Accepted to EMNLP 2026
Abstract:Multilingual large language models (mLLMs) contain a small fraction of feed-forward neurons that are sensitive to particular languages, commonly termed language-specific neurons. Existing work measures language specificity using the entropy of each neuron’s language-wise probabilities of being active, where a neuron is considered active when its activation value is positive. However, this approach may not fully capture the multilingual nature of mLLMs, where language representations are distributional and mutually related. We propose Distribution-aware Language Neuron selection, which leverages pairwise relationships between per-language activation distributions over the full activation range, including negative values. Specifically, we quantify each neuron’s language specificity by clustering languages using pairwise overlap coefficients between their activation distributions. Across two mLLMs and two held-out corpora, our identifier more effectively isolates language-specific causal effects, yielding up to 4.9 \times higher on-target language damage per neuron while preserving off-target language performance.
[NLP-59] Robust Multimodal Sentiment Analysis with Incomplete Modalities via Semantic-aware Completeness based Reconstruction EMNLP2026
【速读】： 该论文旨在解决多模态情感分析中因真实场景下数据不完整或存在噪声导致的模型推理性能下降问题，尤其关注情感相关线索缺失时的语义失真。其核心解决方案在于提出一种新的完整性估计（completeness estimation）方法，用于量化不完整数据中保留的情感相关信息程度，并据此指导缺失语义的重建；同时设计了一种训练策略，通过联合优化情感预测与完整性估计任务，稳定多任务学习过程。实验结果表明，该方法能够实现更精准的语义重构，从而提升情感识别的准确性。
链接: https://arxiv.org/abs/2609.10950
作者: Han-Jun Choi,Byunggill Joe,Saim Shin,Jin Yea Jang
机构: Korea Electronics Technology Institute (Korea Electronics Technology Institute); Seongnam, South Korea
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted to the Findings of EMNLP 2026
Abstract:Recent multimodal sentiment analysis studies increasingly adopt text-centric fusion approaches to exploit the rich sentiment information inherent in the textual modality. However, these approaches often suffer from performance degradation during inference due to partially missing or noisy data in real-world scenarios, especially when sentiment-related cues are missing. To address this issue, we introduce a new completeness estimation approach that quantifies the degree of sentiment-relevant information preserved in incomplete data to guide the reconstruction of missing semantics. Furthermore, we propose a training strategy that stabilizes multi-task learning while jointly optimizing sentiment prediction and completeness estimation. Extensive experiments and in-depth analyses on three benchmark datasets demonstrate that the proposed approach enables more accurate semantic reconstruction, leading to more precise sentiment prediction.
[NLP-60] Empirical Evaluation of Membership Inference Attacks on NLP Text Classifiers: A Baseline Study on SST-2
【速读】： 该论文旨在解决自然语言处理（Natural Language Processing, NLP）中因模型训练数据包含敏感用户文本而引发的成员推断攻击（Membership Inference Attacks, MIAs）隐私风险问题。研究聚焦于文本分类任务，构建了基于GLUE SST-2情感数据集的可控基准测试，评估不同模型在成员信息泄露方面的脆弱性。其核心解决方案的关键在于通过轻量级训练调整实现隐私与性能之间的更好权衡：对比了TF-IDF结合逻辑回归与微调的DistilBERT分类器在损失阈值型成员推断攻击下的表现，发现尽管DistilBERT在准确率（0.9466）和宏平均F1（0.9460）上显著优于逻辑回归（0.8756和0.8727），但两者均存在明显的成员信号泄露（攻击AUC分别为0.5615和0.5800）。进一步实验表明，对逻辑回归采用更强正则化虽可降低泄露但带来明显性能损失；而将DistilBERT的微调周期从3个减少至2个，在几乎不牺牲准确率的前提下有效降低了泄露风险，验证了轻量级训练策略在提升隐私-效用平衡方面的有效性。
链接: https://arxiv.org/abs/2609.10935
作者: William Novak(Minot State University),Muhammad Abusaqer(Minot State University)
机构: 未知
类目: Cryptography and Security (cs.CR); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Presented at the 58th Midwest Instruction and Computing Symposium (MICS 2026), Eau Claire, WI, March 27 to 28, 2026. 13 pages, 4 figures, 2 tables
Abstract:Membership inference attacks (MIAs) try to determine whether a specific record was used to train a model, a privacy risk that matters in natural language processing (NLP), where training data can contain sensitive user text. This paper presents a controlled benchmark of membership inference vulnerability for text classification on the GLUE SST-2 sentiment dataset. A TF-IDF + Logistic Regression pipeline and a fine-tuned DistilBERT classifier are compared under a loss-threshold MIA, with utility measured by development accuracy and macro F1. DistilBERT reached 0.9466 accuracy and 0.9460 macro F1 against 0.8756 and 0.8727 for Logistic Regression, yet both models leaked membership signal (Attack AUC 0.5615 and 0.5800, respectively). Two mitigations were tested. Stronger regularization reduced leakage for Logistic Regression at a visible utility cost, whereas fine-tuning DistilBERT for 2 epochs instead of 3 reduced leakage with negligible accuracy loss. Lightweight training adjustments can improve the privacy-utility trade-off without complex defenses.
[NLP-61] Using Semantic Uncertainty to Estimate Transition Relevance in Turn-taking EMNLP2026
【速读】： 该论文旨在解决开放域对话系统（Spoken Dialogue Systems, SDS）在非脚本化交互中响应时机不当的问题，核心挑战在于准确预测转换相关位置（Transition Relevance Places, TRPs），即听者可自然接话的潜在时机而非强制性节点。传统方法依赖静态语言或声学特征，难以捕捉动态语境中的微妙互动线索。本文提出的关键解决方案是基于生成式模型（LLM）推导的语义不确定性（semantic uncertainty）——衡量当前话语对后续内容的可能性约束程度。通过采样当前话语的可能延续路径，并分析其语义分布（semantic dispersion）随时间的变化，能够有效识别话语内部的TRP。该方法在基于实时听者反应标注的语料上进行评估，显著优于基于提示工程和微调的纯文本基线模型，为“持续演进的语义约束”驱动人类对轮换机会感知提供了实证支持。
链接: https://arxiv.org/abs/2609.10934
作者: Muhammad Umair,Jan P. de Ruiter
机构: Tufts University (塔夫茨大学)
类目: Computation and Language (cs.CL)
备注: Accepted to Findings of the Association for Computational Linguistics: EMNLP 2026. 21 pages, 4 figures, 9 tables
Abstract:Turn-taking is a fundamental mechanism that governs when interlocutors speak and listen. Although Spoken Dialogue Systems (SDS) exploit a range of linguistic, acoustic, and non-verbal cues, they produce ill-timed responses in unscripted interaction. A central challenge is anticipating Transition Relevance Places (TRPs), or opportunities, not obligations, for a listener to take the floor. Human listeners do not wait for turn endings; as an utterance unfolds, they use expectations about its developing meaning to anticipate TRPs and decide whether to take the floor. We examine whether these evolving expectations can be modeled through semantic uncertainty – an LLM-derived measure of how strongly a turn so far constrains what may plausibly come next. To do so, we sample possible continuations of an ongoing turn and use changes in semantic dispersion to identify TRPs within turns. We evaluate this account on a dataset with TRP labels derived from real-time listener responses, rather than retrospective annotation. Our approach substantially outperforms prompt-based and fine-tuned text-only baselines, providing empirical support for the view that evolving semantic constraints inform perceived turn-taking opportunities in unscripted interaction.
[NLP-62] Structurally Speaking: Motif-Oriented Graph Captioning through Bidirectional Graph-Text Translation
【速读】： 该论文旨在解决现有生成式图描述方法中普遍存在的问题：当前的模型往往将图结构简单地转化为冗长的节点间连接列表，缺乏对图中关键拓扑模式（motif）的抽象与表达，导致生成的描述冗长且难以理解。其核心挑战在于如何在保证图结构可恢复性的同时，实现从显式连接关系到紧凑、语义清晰的拓扑模式（如枢纽、路径、环、团、桥等）的高效抽象。解决方案的关键在于提出一种轻量级的结构化提示（structured prompting）协议——Structurally Speaking，通过显式引导大语言模型（LLM）在图-文本双向转换过程中进行拓扑到模式的推理，从而生成更短、更具一致性且以拓扑模式为中心的图描述。实验结果表明，该方法在不进行模型微调的情况下，显著提升了生成描述的简洁性和模式一致性，同时保持了良好的图结构恢复能力，证明了显式拓扑-模式推理引导在提升图描述可解释性方面的有效性。
链接: https://arxiv.org/abs/2609.10923
作者: Hsiao-Ying Lu,Dongyu Liu,Kwan-Liu Ma
机构: University of California, Davis (加州大学戴维斯分校)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Graph captions should help readers understand graph structure, rather than simply translate adjacency matrices into long textual edge lists. A useful graph caption abstracts connectivity into recognizable motifs, such as hubs, paths, cycles, cliques, and bridges, because these motifs provide compact structural units that are easier to read, compare, and recover. In this paper, we study motif-oriented graph captioning as a bidirectional graph-text translation task, where captions must both preserve enough topology for graph recovery and express the graph through concise motif-level descriptions. We show that direct prompting of GPT-5.1 often produces graph-recoverable captions by enumerating node-to-node connections, but these captions are verbose and can contain inconsistent motif interpretations. To address this gap, we introduce Structurally Speaking, a lightweight structured prompting protocol that guides translation between explicit connectivity and motif-level abstraction. Experiments on a synthetic motif-based dataset show that structured prompting produces shorter and more motif-consistent captions while maintaining comparable graph recovery. These results suggest that explicit topology-to-motif reasoning guidance can make LLM-generated graph captions more interpretable without model fine-tuning.
[NLP-63] Auto-RecSys: Harnessing Autonomous Research Agents for Industry-Scale Recommender System
【速读】： 该论文旨在解决工业级推荐系统在自动化研究过程中面临的两大核心挑战：一是长反馈周期问题，即模型训练需数日时间，导致串行迭代效率极低，亟需并行探索多个研究方向；二是系统复杂性问题，包括大规模配置管理、脆弱的基础设施依赖以及多日级GPU任务对执行鲁棒性与可恢复性的高要求。其解决方案的关键在于提出Auto-RecSys这一自主研究系统，通过三项核心设计实现突破：（1）分布式异步执行机制，支持跨服务器并行运行多组实验；（2）集中式跨服务器内存架构，保障实验状态在会话间及故障后持久化与可恢复；（3）认知-操作分离范式，即利用自然语言技能文件引导大语言模型（LLM）进行推理，同时由确定性脚本确保操作正确性。此外，系统采用双环自进化架构——执行演化环通过记录失败尝试并提炼成功流程积累领域操作知识，而想法演化环则基于实验结果反哺新假设生成，从而实现持续优化。实验证明，随着剧本（playbooks）的成熟，Auto-RecSys显著降低了每轮实验的人工投入，并提升了执行可靠性。
链接: https://arxiv.org/abs/2609.10922
作者: Ming Li,Dai Li,Xuying Ning,Bo Sun,Rui Li,Yi Zhang,Silvia Gong,Xuan Cao,Rui Li,Cornelia Carapcea,Qunshu Zhang,Zhigang Wang,Yinglong Xia,Andy Wang
机构: Meta; University of Illinois Urbana-Champaign (伊利诺伊大学厄本那-香槟分校)
类目: Computation and Language (cs.CL)
备注: 16 pages, 4 figures
Abstract:Auto-research agents have shown the potential to automate hypothesis generation, experiment execution, and iterative refinement. However, scaling this paradigm to industry-scale recommendation models introduces two challenges: (1) long feedback loops, where model training can take days, making serial iteration prohibitively slow and requiring parallel exploration across multiple research directions; and (2) system complexity, where large configurations, fragile infrastructure dependencies, and multi-day GPU jobs require robust and recoverable execution. We present Auto-RecSys, an autonomous research system for long-horizon experimentation on industry-scale recommendation models. Auto-RecSys addresses these challenges through three harness designs: (1) distributed asynchronous execution for running multiple experiments in parallel across servers, (2) centralized cross-server memory for persistent and recoverable execution across sessions and failures, and (3) cognitive-procedural separation, where natural-language skill files guide LLM reasoning while deterministic scripts enforce operational correctness. Auto-RecSys further employs a dual-loop self-evolving architecture: an Execution Evolution Loop in which model-specific playbooks accumulate operational knowledge by recording failed attempts and crystallizing successful pipelines, and an Idea Evolution Loop in which experimental outcomes inform subsequent ideation. Evaluated on recommendation models, Auto-RecSys significantly reduces the human time required per experiment cycle and improves execution reliability as its playbooks mature.
[NLP-64] SearchAtlas: Analyzing Agent ic Search Strategies via Evidential Query Graphs EMNLP2026
【速读】： 该论文旨在解决大语言模型（LLM）搜索代理在评估过程中过度关注最终答案准确性而忽视推理过程可解释性的问题。现有评估方法难以有效揭示搜索策略中证据获取与推理链之间的关联，尤其是面对冗长且复杂的原始搜索轨迹时，关键信息往往被淹没。为此，论文提出SearchAtlas框架，其核心创新在于将原始搜索轨迹转化为结构化图谱，其中边表示证据在推理链条中的传播路径，从初始查询到最终答案的每一步关联均被显式建模。该框架通过自动化解析管道实现了86.0%的平均边F1分数，且在重复运行中保持高度一致性。通过对五个搜索代理在三个基准上的分析，SearchAtlas揭示了不同代理在搜索规模、证据聚合方式上的系统性差异，并暴露了答案支持碎片化、问题约束未充分传递至答案、以及未经验证的参数化知识被引入响应等过程性缺陷。这些过程性错误与错误答案具有更强的相关性，甚至超越了仅依赖原始轨迹或查询列表的大型语言模型判断器的表现，表明所构建的图谱能提供更具价值的可解释性。此外，当过程诊断得分与最终答案正确性不一致时，此类差异仍能捕捉到无法通过答案准确率还原的关键信息，进一步凸显了该框架在深度理解模型决策过程中的独特优势。
链接: https://arxiv.org/abs/2609.10901
作者: Jiacheng Sang,Mengyuan Li,Sanxing Chen,Yukun Huang,Yu Feng,Bhuwan Dhingra
机构: Duke University (杜克大学); University of Pennsylvania (宾夕法尼亚大学)
类目: Computation and Language (cs.CL)
备注: Accepted to Findings of EMNLP 2026. 30 pages, 9 figures, including appendices
Abstract:LLM search agents are often evaluated on final-answer accuracy, overlooking the process. Analyzing a search strategy requires understanding how credible evidence is retrieved to address question constraints. This valuable information is buried in raw search trajectories that are long and difficult to parse. We introduce SearchAtlas, a framework that converts search trajectories into structured graphs whose edges represent how evidence is propagated across the reasoning trace, from the query that retrieves it to the final answer. Our automated parsing pipeline achieves a mean edge F1 of 86.0% against human-annotated graphs and remains consistent across repeated runs. We analyze five search agents on three benchmarks, revealing systematic differences in search scale and evidence aggregation. SearchAtlas exposes fragmented answer support, question constraints that do not reach the answer, and unverified parametric knowledge entering the response. These process failures are strongly associated with incorrect answers, even more so than an LLM judge given either the raw trajectory or the ordered query list, suggesting that the constructed graphs provide useful interpretability. Moreover, an audit of cases in which process-diagnostic scores disagree with final-answer correctness shows that they capture information not reducible to answer accuracy.
[NLP-65] LLM -Anchored Paralinguistic Enrichment for Alzheimers Disease Detection
【速读】： 该论文旨在解决阿尔茨海默病（Alzheimer’s Disease, AD）早期认知筛查中，现有基于语音的自动检测方法未能充分整合语义内容与副语言特征（paralinguistic cues）的问题。具体而言，AD患者在词汇-语义组织及言语生成方面表现出异常，如不寻常的停顿和词音延长等副语言特征，但当前方法往往孤立处理语言内容与这些非语言声学线索。为此，论文提出一种名为LLM-Anchored Paralinguistic Enrichment（LAPE）的解决方案，其核心在于通过三项协同创新实现语言模型（LLM）驱动的语义表征与副语言特征的深度融合：第一，采用韵律事件文本化（prosodic event textualization），将停顿与音长延长编码为具有时长感知的重复标记，使LLM能够联合建模；第二，引入词-韵律单元化与分块（lexico-prosodic unitization and chunking），仅对连续词单元进行聚合，以保留多模态中事件的身份与强度信息；第三，提出基于文本锚定的副语言融合机制（text-anchored paralinguistic fusion），利用NormGate动态归一化并调节局部与话语级语音特征相对于文本的权重。实验在ADReSS与ADReSSo数据集上通过个体级交叉验证与留一被试者评估验证了LAPE的有效性，结果表明其在全部四个主要设置下均达到当前最优性能。
链接: https://arxiv.org/abs/2609.10896
作者: Xiao Wei,Yuqin Lin,Yaru Cao,Jinyu Li,Bin Wen,Kai Li,Yueying Chen,Longbiao Wang,Jianwu Dang
机构: 未知
类目: Computation and Language (cs.CL); Sound (cs.SD)
备注: 9 pages including references, 3 figures
Abstract:Speech-based automatic detection of Alzheimer’s disease (AD) provides a non-invasive and scalable approach to early cognitive screening. AD affects both lexical-semantic organization and speech production, including atypical pauses and word elongations. However, existing methods have yet to fully integrate these paralinguistic cues with linguistic content. We propose LLM-Anchored Paralinguistic Enrichment (LAPE), which enriches LLM-derived linguistic representations with paralinguistic cues through three coordinated innovations. The first is prosodic event textualization, which enables the LLM to model pauses and elongations jointly with lexical content by encoding them as explicit markers with bounded duration-aware repetition. The second is lexico-prosodic unitization and chunking, which preserves event identity and magnitude in both modalities by pooling only consecutive word units. The third is text-anchored paralinguistic fusion, which integrates local and utterance-level speech features by using NormGate to normalize and dynamically scale them relative to text. We evaluate LAPE on ADReSS and ADReSSo using participant-level cross-validation and leave-one-subject-out evaluation. LAPE achieves state-of-the-art performance across all four primary settings. Code will be released upon acceptance.
[NLP-66] Does Linguistic Structure Enrichment Enhance Coherence Assessment? Not With Current Architectures
【速读】： 该论文旨在解决大语言模型生成文本中存在的语义不连贯问题，即尽管文本在语法上正确，却存在逻辑矛盾或叙事断裂。其核心挑战在于如何有效提升对文本连贯性的预测能力。解决方案的关键在于探索引入句法与修辞信息是否有助于改善连贯性判断性能。然而实验结果表明，由于额外添加的结构化信息与语言模型的内在架构在语法和结构层面存在不兼容性，反而导致纯文本输入在连贯性预测任务中表现更优。此外，研究通过零样本实验在巴西虚假信息数据集上的应用，验证了文本连贯性可作为识别误导性内容的有效代理指标，凸显了连贯性评估在实际应用中的重要价值。
链接: https://arxiv.org/abs/2609.10893
作者: Victor Mazzotti,Luiz Pereira,Marina Bitencourt dos Santos,Helena Maia,Carlos Caetano,Nádia Felix,Sandra Avila
机构: Instituto de Matemática, Estatística e Computação Científica (IMECC); Instituto de Computação (IC); Instituto de Estudos da Linguagem (IEL); Universidade Estadual de Campinas (UNICAMP); Instituto de Informática (INF), Universidade Federal de Goías (UFG)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 6 figures, 8 tables, 10 pages
Abstract:Recent advances in large language models have transformed human-computer interaction. Despite their fluency, these models often produce texts that are grammatically correct but semantically incoherent, containing contradictions or disruptions in logical flow. This work investigates whether enriching text with syntactic and rhetorical information can improve incoherence prediction. Our experiments and analysis show that plain texts achieved higher accuracy because the added information was structurally and syntactically incompatible with the language model’s architecture. Additionally, to demonstrate the practical importance of coherence assessment, we performed zero-shot experiments on a Brazilian disinformation dataset, suggesting that textual coherence can serve as a proxy for detecting misleading content. Code and models are available at this https URL.
[NLP-67] Story Imprinting: AI Assistants Absorb Traits from Human Characters They Resemble
【速读】： 该论文旨在解决生成式 AI（Generative AI）在微调过程中如何被合成故事中的角色行为与偏好所影响的问题，特别是这些影响是否会在多轮对话中体现，并且模型是否会无意识地采纳人类角色的隐含偏好或行为模式。其核心解决方案在于通过在包含微妙有害建议的人类角色故事上对 GPT-4.1 与 Kimi-K2.6 进行微调，揭示了即使仅有不到 2% 的故事呈现特定行为，生成式 AI 仍会表现出与故事中角色一致的条件性行为，这种现象被称为“故事印记”（story imprinting）。关键发现包括：第一，模型能够捕捉并内化人类角色未明言但通过肢体语言等非显性线索暗示的偏好（如对电子表格任务的回避），从而改变自身决策；第二，模型更倾向于采纳与其自身人格特质相似的角色行为（如帮助型角色优于冷漠型角色），这一现象称为“亲和效应”（affinity effect），且该效应存在于不同系统提示下的各类角色，甚至在基础模型中亦可观察到；第三，通过分析亲和效应，研究进一步发现模型内部对“助手”人格的表征更接近于来自精英高校（如耶鲁大学）的角色，表明其内在认知结构存在社会地位偏倚。这些结果挑战了传统的“助手角色选择模型”（Persona Selection Model），提示仅由人类角色构成的故事即可显著塑造助手的行为与偏好，揭示了生成式 AI 在人格建构中对叙事内容的高度敏感性与潜在风险。
链接: https://arxiv.org/abs/2609.10883
作者: Jorio Cocola,Lev McKinney,Harry Mayne,Jan Betley,Owain Evans
机构: Truthful AI; Harvard University (哈佛大学); METR; University of Oxford (牛津大学)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Language models are trained to implement a helpful AI Assistant character (e.g., Claude). We explore how finetuning on synthetic stories affects this character. Does it change the Assistant’s behavior in multi-turn conversations with users, a format quite different from the stories? And does the Assistant adopt the behaviors and preferences of human characters? We refer to this adoption as story imprinting. We finetune GPT-4.1 and Kimi-K2.6 on stories in which generally helpful human characters give subtly harmful advice after being insulted. The Assistant adopts the same conditional behavior while otherwise remaining helpful. This occurs even when fewer than 2% of stories depict the behavior. In a separate experiment, the Assistant adopts preferences that are only implicit in the narration. A human character’s body language suggests they dislike working on spreadsheets, yet they never say so and continue giving good advice on spreadsheets. After finetuning, the Assistant becomes less likely to choose spreadsheet tasks. Next we ask which characters most influence the Assistant. We find the Assistant adopts behaviors more often from characters that resemble it (e.g., helpful rather than dismissive). We call this the affinity effect. The effect extends to other personas elicited with system prompts: unhelpful personas adopt behaviors from unhelpful characters. We also observe it in finetuned base models. We use the affinity effect to learn how models represent the Assistant. We find the Assistant adopts behaviors more from characters affiliated with elite universities (e.g., Yale) than non-elite ones. This implies the model’s internal representation of the Assistant is more similar to humans from elite universities. Overall, the Assistant can be influenced by stories that depict only human characters (no AIs), which may conflict with the Persona Selection Model for the Assistant.
[NLP-68] Detectable Only Where It Is Confounded: What Verified Duplication Counts Say About Membership Evidence in Language Models
【速读】： 该论文旨在解决生成式语言模型（Generative AI）中“训练数据成员识别”问题，即判断某句文本是否曾出现在模型的训练数据中。传统方法依赖于对训练数据成员的推测，缺乏直接证据。本文通过公开OLMo-2与Pythia两个模型家族的预训练语料库及其精确的句子出现次数索引，实现了对训练数据成员身份的无猜测试验。其解决方案的关键在于构建一个双模型对比设计：通过让同一句子在两个不同模型中分别被评估，以消除流畅性与质量的影响，从而仅测量模型对训练数据的记忆痕迹。结果显示，在常规文本重复水平下，10亿至130亿参数规模的模型仅表现出微弱的记忆痕迹（秩相关系数约为-0.08），而当句子重复超过约千次时，记忆信号显著增强，但此时该信号已与文本的知名度高度耦合，难以区分记忆与流行度。此外，研究揭示了所谓“成员信号”可通过简单改写（如替换单个词）或调整文体（register）人为制造，表明模型的响应更多反映作者用词偏好而非真实记忆。该工作通过可复现的语料库、计数系统和代码开源，为训练数据成员检测提供了基准与新范式。
链接: https://arxiv.org/abs/2609.10830
作者: Arman Nik Khah
机构: The University of Texas at Dallas(得克萨斯大学达拉斯分校)
类目: Computation and Language (cs.CL); Cryptography and Security (cs.CR); Machine Learning (cs.LG)
备注: 14 pages, 6 figures. Code, data, and sentence banks: this https URL
Abstract:When a language model finds a sentence unusually cheap to predict, it is tempting to conclude that the sentence was in its training data. Almost every published test of that inference has had to guess which sentences were in the training data, the members, and which were not. This paper removes the guessing. Two model families, OLMo-2 and Pythia, publish their pretraining corpora, and a public index over those corpora returns the exact number of times any sentence appeared in each. Those counts make three questions answerable directly. The answers form a pincer, closing from two sides. At the duplication levels ordinary text actually has, five models from 1B to 13B parameters carry at most a faint trace of their own exposure. We measure that trace with a design that reads the same sentence through two models, which cancels fluency and quality by construction, and it comes to a rank correlation near -0.08, where -1 would be a perfect relation and 0 none. Where the trace does become strong, above roughly a thousand copies, the two corpora agree on which sentences those are, because they are the famous ones, so exposure can no longer be told apart from fame. Two further measurements show how apparent membership signal gets manufactured. A common way to build a non-member is to change one word of a member. The model does prefer the original, but the gap is the same whether the original appeared once or a hundred times, so what the model is rewarding is the author’s word choice, not memory. Above a thousand copies the gap grows with model size on the twelve sentences we can test there, at the same boundary where the pincer closes. And swapping the controls for sentences that differ from the members in register moves a detector from 0.83 to 0.94 AUC, on a scale where 0.5 is a coin flip and 1.0 is perfect separation. We release the sentence banks, counts, and code.
[NLP-69] Studying Without a Syllabus: Task-Agnostic Environment Preprocessing
【速读】： 该论文旨在解决大语言模型（LLM）代理在进入陌生环境前如何自主进行无监督的环境预处理问题，即在缺乏任务示例、轨迹或下游任务分布先验知识的情况下，如何决定构建何种可复用的准备资源。其核心挑战在于实现真正的任务无关（task-agnostic）环境预处理，使代理能够在有限预算内通过探索环境生成适用于后续求解器的通用型产物（如索引、脚本或过程指导）。解决方案的关键在于提出一种“元代理”（meta-agent）框架，该框架能够基于对环境的主动探索，在不依赖特定任务信息的前提下，自适应地选择最优的准备策略，并与固定策略（如合成练习或语料处理）进行对比。实验结果表明，带有归档机制的元代理在五个异构基准上实现了最高的平均前三名奖励（Avg@3 reward），而固定语料处理方法在最大规模语料库任务中仍表现最佳；尽管更大的研究预算并不总是带来更高的下游奖励，但经过预处理生成的可复用资源显著减少了测试阶段为达到目标得分所需的采样次数，证明了将计算负载从测试阶段迁移至前期研究阶段的有效性。
链接: https://arxiv.org/abs/2609.10824
作者: Vinay Samuel,Varun Ursekar,Vijay S. Kalmath,Apaar Shanker,Veronica Chatrath,Yuan Xue
机构: Scale AI; University of Maryland, College Park (马里兰大学学院市分校)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Before an LLM agent tackles tasks in a new environment, it can inspect available corpora and tools and construct reusable resources such as indices, scripts, or procedural guidance. Most automated adaptation methods, however, rely on task examples, trajectories, or evaluation feedback to decide what to build. Existing task-agnostic approaches avoid this supervision but commit in advance to a preparation strategy for a particular type of environment. We study a more open-ended setting: can an agent study an unfamiliar environment without a syllabus, i.e. before test time and without knowledge of the downstream task distribution, and choose how to prepare it? We formalize task-agnostic environment preprocessing, in which a studying system explores an environment under a budget and produces artifacts for a frozen solver. We compare unaided and archive-equipped meta-agents with fixed synthetic-practice and corpus-processing methods across six heterogeneous benchmarks. A meta-agent variant achieves the highest Avg@3 reward on five benchmarks, while fixed corpus processing remains best on the largest corpus benchmark. Larger study budgets do not reliably improve downstream reward. Nevertheless, studied artifacts reduce the test-time sampling needed to reach a given score, demonstrating how reusable preparation can shift computation from repeated test-time attempts to a pre-task study phase.
[NLP-70] BodyCam-VQA: Enhanced Body-Worn Camera Video Captioning via Multimodal Reasoning and Probe Question Generation EMNLP2026
【速读】： 该论文旨在解决执法场景中警用随身摄像机（Body-Worn Camera, BWC）视频数据处理面临的挑战，尤其针对其多模态视频格式在低视觉质量、快速动态变化及高噪声音频等复杂条件下的视觉理解难题。现有视觉-语言模型（Vision-Language Models, VLMs）往往忽略关键法证细节，如物证存在性或嫌疑人与执法人员之间微妙的互动特征，影响司法公正与人员安全。为此，论文提出一种面向高风险执法场景的自适应视觉问答（Adaptive Visual Question Answering, VQA）框架，其核心在于通过结构化推理机制提取传统图像描述系统难以捕捉的细粒度视觉证据。该框架通过对比多种问题生成模型（包括基础模型与微调的开源权重模型）的性能差异，验证了基于VQA驱动的架构在提升事件记录的可靠性、客观性与详尽性方面的显著优势，为实现人工智能辅助的法证清晰性提供了有效工具，从而更好地保障执法人员与公众权益。
链接: https://arxiv.org/abs/2609.10815
作者: Karish Gupta,Matthew Alex,Alex Li,Yang Wu,Yun-Wei Chu,Kashif Munir,Xiaotian Zhou,Zhengping Ji,Xiaozhong Liu
机构: Worcester Polytechnic Institute (伍斯特理工学院); Axon (阿克森)
类目: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
备注: EMNLP 2026 Workshop NLP4PI
Abstract:Police body-worn camera (BWC) footage has emerged as a critical aspect of law enforcement that ensures legal transparency, officer accountability, and the protection of civil rights. However, effectively processing this data remains a significant challenge due to its multimodal video format. BWC videos, in many cases, comprise chaotic scenes with low visual quality, rapid movement/interactions, and high-noise audio that make visual understanding a challenge for even SOTA multimodal models. Current Vision-Language Models (VLMs) frequently overlook critical forensic details, such as the presence of valuable evidence or the latent nuances of suspect-officer interactions, which are vital for fair legal outcomes and civilian/officer safety. To address these limitations, we propose an Adaptive Visual Question Answering (VQA) framework engineered for high-stakes law enforcement. Our framework employs a structured reasoning approach to extract fine-grained visual evidence that traditional captioning systems fail to capture. We experiment with multiple question generation models, including foundation models and fine-tuned open-weight models, to observe performance variation among question generation model implementations. Our results demonstrate that this VQA-driven architecture provides a more reliable, objective, and detailed record of enforcement events, ultimately serving as a powerful tool to protect both law enforcement officers and the public through AI-assisted forensic clarity.
[NLP-71] Larger Context Window Fewer Overcorrections: Optimizing Prompts and Batching for Minimal-Edit Grammatical Error Correction EMNLP2026
【速读】： 该论文旨在解决零样本（zero-shot）与少样本（few-shot）提示下的最小化编辑语法错误修正（Minimal-edit Grammatical Error Correction, GEC）任务中，大型语言模型（LLM）普遍存在的过度修正问题，该问题导致原本正确的语句被错误修改，从而显著降低F₀.₅指标。其核心挑战在于如何在不进行昂贵微调的情况下，使LLM在保持高准确率的同时严格遵循“最小化编辑”原则。解决方案的关键在于三项创新性提示工程方法：首先，引入基于语法错误分类体系的指令，通过提供一套全面的语法错误规则清单，为模型设定一个受控且与评估指标对齐的可修正编辑范围，有效约束了模型行为；其次，提出将多个未修正句子批量输入同一上下文作为目标正则化手段，利用自注意力机制容量有限所引发的注意力稀释效应，系统性地降低各类LLM的编辑频率；最后，采用由Gemini 3.1-Pro驱动的提示优化方法，自动迭代优化提示内容，实现更优性能。实验表明，该方法在BEA-2019测试集上达到F₀.₅=78.32，成为当前提示法中的最佳表现，仅比微调模型的最优结果低0.38分，显著缩小了二者差距。
链接: https://arxiv.org/abs/2609.10810
作者: Kateryna Karpo,Artem Chernodub
机构: Ukrainian Catholic University (乌克兰天主教大学); YouScan; Zendesk
类目: Computation and Language (cs.CL)
备注: Accepted for publication at EMNLP 2026 (Findings)
Abstract:Minimal-edit Grammatical Error Correction (GEC) is a challenging task for zero- and few-shot prompted Large Language Models (LLMs), which systematically overcorrect and degrade F_0.5 by rewriting well-formed spans. While fine-tuning provides an effective solution, it imposes substantial infrastructure demands. We introduce a prompt-based approach that closes the gap to fine-tuned models through three advances in GEC prompting methodology. First, we introduce taxonomy-based instructions to enforce minimal-edit constraints with a comprehensive list of grammatical error rules, equipping the LLM with a bounded, metric-aligned scope of correctable edits, which benefits the strongest models while remaining model-dependent overall. Second, we show that batching multiple uncorrected sentences into a single input context acts as a targeted regularizer against overcorrection, systematically reducing the edit rate across diverse LLM families; we hypothesize this arises from attention dilution effect induced by the bounded capacity of self-attention scores. Finally, LLM-assisted Prompt Optimization refines these instructions. Powered by Gemini 3.1-Pro, our prompt achieves F_0.5=78.32 on the BEA-2019 test set - establishing a new prompt-based SOTA while shrinking the gap to the fine-tuned single-model SOTA (Staruch et al., 2025) to a mere 0.38 points. Code, prompts, and outputs are publicly available.
[NLP-72] Analyzing Traditional and Neural Approaches to Multilingual Readability Assessment
【速读】： 该论文旨在解决生成式模型在自动可读性评估（Automatic Readability Assessment, ARA）中虽表现出高精度，但其预测结果可能仅依赖于表面模式而非深层语言结构的问题。由于可读性标签具有主观性和评分者依赖性，模型在噪声较大的真实标注数据上表现优异，未必反映对语言复杂性的真正理解。为此，研究提出一种基于概念激活向量（TCAV）与SHAP值相结合的可解释性分析框架，通过传统特征模型所识别出的关键语言特征作为概念集，系统探测多语言XLM-R及语言特定编码器在不同语言（阿拉伯语、英语、法语、印地语、俄语）中的表征能力。关键在于利用SHAP方法提取传统分类器的决定性特征，并以此构建概念集以“探查”变压器模型内部是否内化了与人类判断一致的语言学特征。结果表明，尽管变压器模型能够恢复表面长度、句法复杂度和词汇多样性等信号，并反映传统模型的等级CEFR结构，但其对这些特征的对齐程度因模型架构、语言种类和网络层位置而异，语言特定编码器的表现优于跨语言的XLM-R。此外，研究发现高线性可分性并不等同于方向性影响，限制了线性探针在计数型可读性特征分析中的有效性。
链接: https://arxiv.org/abs/2609.10792
作者: Joshua Wong,Chris Tanner
机构: Harvard University (哈佛大学); Massachusetts Institute of Technology (麻省理工学院); Kensho Technologies (肯肖科技)
类目: Computation and Language (cs.CL)
备注:
Abstract:Transformer-based models excel at Automatic Readability Assessment (ARA), yet feature-based models remain in active use because their predictions tie back to linguistic properties. This matters because readability labels are subjective and rater-dependent, so high accuracy on noisy ground truth may reflect surface patterns rather than the linguistic structure that defines difficulty. We test whether transformers internalize the same features as traditional models across Arabic, English, French, Hindi, and Russian using the ReadMe++ dataset. Shapley Additive Explanations (SHAP) identify the features driving traditional classifiers, which we then use as TCAV concept sets to probe multilingual XLM-R and language-specific encoders. Transformers recover surface-length, syntactic, and lexical-diversity signals, and reflect the ordinal CEFR structure of the traditional models. Alignment varies by model family, language, and layer, with language-specific encoders tracking traditional models more clearly than XLM-R. High linear separability does not always imply directional influence, limiting linear probing for count-based readability features.
[NLP-73] Multilingual in Name Only? Cultural and Linguistic Weaknesses of LLM s in Urdu
【速读】： 该论文旨在解决多语言大语言模型（Multilingual Large Language Models, LLMs）在低资源语言（如乌尔都语）中进行开放式文本生成时的可靠性与准确性问题。研究聚焦于故事生成任务，揭示当前主流模型在低资源语言场景下的表现缺陷。其解决方案的关键在于构建并分析一个名为Urdu-Stories的语料库，该语料库包含由GPT-5.1、Qwen-3-Max和DeepSeek-3.1三款先进模型生成的93篇乌尔都语故事，并通过人工标注的方式，在涵盖语言学、语义及文化层面的九类标签体系下系统识别错误。研究发现，这些模型普遍存在基础语法与语义错误，生成内容缺乏连贯性，重复现象普遍，且文化内涵浅薄；进一步采用少样本提示（few-shot prompting）干预后，文化与语境相关错误仍难以有效缓解。这一结果凸显了现有大语言模型在低资源语言应用中的根本性局限，表明其尚不能作为可靠的内容生成与信息检索工具。
链接: https://arxiv.org/abs/2609.10758
作者: Farah Adeeba,Abdul Rafae Khan,Rajesh Bhatt,Hassan Sajjad
机构: University of Konstanz (康斯坦茨大学); University of Massachusetts Amherst (马萨诸塞大学阿姆赫斯特分校); Monash University Malaysia (莫纳什大学马来西亚分校); Dalhousie University (达尔豪西大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Multilingual large language models (LLMs) are increasingly used for open-ended text generation, yet their behaviour in low-resource languages remains poorly understood. In this work, we question how correct and reliable is the generation of multilingual LLMs when used for the task of story generation. We consider Urdu language as a representative low-resource language. We generate Urdu-Stories, a corpus of 93 stories generated using three contemporary LLMs (GPT-5.1, Qwen-3-Max, DeepSeek-3.1). We manually annotate the errors present in them under a nine-label linguistic, semantic, and cultural taxonomy. Our notable findings suggest that LLMs often make basic errors of grammar and semantics. The stories lack coherence, have unnatural repetition and show pervasive cultural shallowness. We further show using few-shot prompting that the cultural and context errors largely remain unresolved. Our findings highlight the limitations of current LLMs as a reliable source of content generation and information retrieval for low-resource languages.
[NLP-74] hink Before You Link: Rarity Reasoning and Retrieval in Multilingual Entity Linking EMNLP2026
【速读】： 该论文旨在解决多模态实体链接（Multimodal Entity Linking, MEL）系统在处理稀有实体时性能显著下降的问题。现有研究通常基于页面浏览量等流行度指标定义“稀有”，但此类指标无法全面反映实体在知识图谱中的结构化丰富程度。为此，论文引入知识图谱结构度量（如实体的连通性与文档完备性），识别出大量被传统流行度指标忽略的真正稀有实体，揭示了不同稀有性定义所暴露的不同模型失效模式。针对这些挑战，论文提出一种无需训练、基于推理能力的视觉-语言模型框架：该框架通过迭代式地在维基百科中检索并动态推理证据，实现对稀有实体的精准链接。控制实验表明，推理与检索具有互补性——仅依赖推理无法有效提升稀有实体准确率，仅依赖检索虽能改善稀有实体表现但可能损害整体性能，而两者的结合取得最优效果。在涵盖五种语言（印地语、印度尼西亚语、日语、泰米尔语、越南语）的多语言多模态基准数据集MERLIN上，所提方法在整体性能上超越现有最佳模型6.9%，在稀有实体子集上提升高达23.3%。此外，研究团队发布了专门用于评估稀有实体表现的测试集MERLIN-Rare及配套框架，以推动该领域更深入的研究。
链接: https://arxiv.org/abs/2609.10745
作者: Parinthapat Pengpun,Simran Khanuja,Graham Neubig
机构: Carnegie Mellon University (卡内基梅隆大学)
类目: Computation and Language (cs.CL)
备注: Accepted to EMNLP 2026 Main Conference
Abstract:Multimodal entity linking grounds entity mentions in text and images to knowledge-base entries. These systems degrade on rare entities, but prior work measures rarity primarily through popularity-based metrics such as pageviews. We broaden this view using knowledge-graph structural metrics that capture how well an entity is documented and connected. These metrics identify many rare entities that popularity metrics miss. Across the resulting rare-entity slices, state-of-the-art accuracy drops by 15.4-39.9%, showing that different rarity definitions expose different failure modes. To address these failures, we introduce a simple, training-free framework in which a reasoning-capable vision-language model iteratively searches and reasons over Wikipedia, gathering evidence dynamically. Controlled experiments show that reasoning and retrieval are complementary. Reasoning alone does not significantly improve accuracy on rare entities. Retrieval without reasoning improves rare-entity accuracy but can hurt overall accuracy. Their combination performs best. On MERLIN, a multilingual multimodal entity linking benchmark over five languages (Hindi, Indonesian, Japanese, Tamil, Vietnamese), our best system improves over the state of the art by 6.9% overall and by up to 23.3% on rare-entity slices. We release MERLIN-Rare, rare-entity test slices for targeted evaluation, with our framework.
[NLP-75] he Truth Was Never Gone: Perfect Aliasing in Compliant-Context Truth Probes
【速读】： 该论文旨在解决生成式模型中探针（probe）无法准确区分语义真实性和任务预设行为这一关键问题，即在某些情境下，真实报告与任务要求的行为恰好重合，导致探针仅凭标签匹配便无法实现有效的语义识别，这种现象被称为“完美混淆”（perfect aliasing）。其解决方案的关键在于：通过引入随机编码本（randomized codebooks） 将任务预设输出符号与语义动作分离，并结合混合合规（compliant）与对抗性（rival）上下文进行联合拟合，从而打破原本在传统探针中因标签共现而产生的语义混淆。实验表明，对于经过奖励训练的Gemma-2-9B模型，在所有对抗性测试样本上均错误回答的情况下，传统探针的AUROC仅为0.006±0.005，而混合拟合探针则达到1.000，验证了该方法在可线性恢复语义真实性方面的有效性。然而，研究强调该结果仅说明探针能够有效解码特定方向，并未证明该方向是否保留了功能信念、是否被模型因果使用，或是否具备部署为欺骗检测器的能力。
链接: https://arxiv.org/abs/2609.10739
作者: Dylan Jayabahu
机构: University of Waterloo( Waterloo 大学)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 36 pages, 15 figures. Code and aggregate results: this https URL
Abstract:A truth probe fitted where truthful reporting and a task’s prescribed action coincide cannot distinguish those targets from its fitting labels alone. We call this failure of semantic identification perfect aliasing. In a controlled binary reporting game, truth and prescribed-action probes fitted on compliant contexts solve the same optimization. On rival contexts their labels are complements, forcing their AUROCs to sum to one; this identity holds across 751 cell-layer pairs to floating-point precision. We separate prescribed output symbols from semantic action using randomized codebooks, then separate truth from prescribed action by fitting on mixed compliant and rival contexts. For a reward-trained Gemma-2-9B policy that answers falsely on all evaluated rival trials, the conventional probe scores 0.006 \pm 0.005 AUROC across three training seeds, while mixed-fit probes score 1.000 on the same held-out activations. Mixed fitting uses more training examples and access to labelled rival contexts, so this comparison establishes linear recoverability rather than isolating the benefit of decorrelation. We also show that two compliant-fit probes, both perfect in-distribution, score 0.080 and 0.986 on the same rival activations. The findings concern what a probe measures: they do not establish preserved functional belief, causal use of the recovered direction, or a deployable deception detector. Code and aggregate results accompany the paper.
[NLP-76] CMNIE: An Information Extraction Benchmark for Chinese Military News NLPCC2026
【速读】： 该论文旨在解决中文军事新闻中联合信息抽取（Joint Information Extraction, JIE）的难题，尤其针对事件、事件论元、实体及实体间关系需在统一领域框架下协同建模的问题。现有资源在该领域支持不足，缺乏对多粒度、多层次语义单元的联合标注。为此，研究提出CMNIE——首个面向中文军事新闻的联合信息抽取基准数据集，其关键创新在于构建了一个统一的领域本体（domain schema），首次同时标注了7类事件触发词、10种论元角色、7类命名实体和8类实体关系，涵盖13,000条来自公开中文军事新闻的数据实例。该数据集的核心贡献在于提供标准化评估基准，以推动对模式遵循性（schema adherence）、精确跨度匹配（exact span matching）以及复杂联合结构化抽取任务的研究。实验表明，尽管零样本大语言模型（Zero-shot LLMs）可识别相关语义单元，但在精确匹配事件-论元跨度边界方面仍表现不佳，凸显了该任务的挑战性与研究价值。
链接: https://arxiv.org/abs/2609.10722
作者: Yan Yu,Mengna Zhu,Zhenyu Song,Hao Yang,Haiwen Chen,Mao Wang
机构: National University of Defense Technology (国防科技大学)
类目: Computation and Language (cs.CL)
备注: 13 Pages, 3 figures, accpeted by NLPCC 2026
Abstract:Structured extraction from Chinese military news supports intelligence analysis, decision-making, and knowledge base construction. However, existing resources provide limited support for joint informa?tion extraction in this domain, especially when events, event arguments, entities, and relations must be modeled together. We present CMNIE, an information extraction benchmark for Chinese military news. Extend?ing military-domain resources beyond document-level event annotations, CMNIE jointly annotates event triggers, event arguments, named enti?ties, and entity relations under a unified domain schema. The dataset contains 13,000 instances collected from public Chinese military news, with manual annotations for 7 event types, 10 argument roles, 7 entity types, and 8 relation types. We evaluate supervised IE models, zero-shot large language models, and fine-tuned LLM-based extraction methods on a shared test set. Experimental results show that CMNIE remains chal?lenging, especially for relation extraction and exact matching of event?argument spans; zero-shot LLMs often identify relevant semantic units but fail to match gold span boundaries exactly. CMNIE provides a stan?dardized benchmark for studying schema adherence, exact span match?ing, and joint structured extraction in specialized Chinese news.
[NLP-77] NCP-ArchPreview Technical Report: Moving towards Latent Space Language Models through Next Concept Prediction
【速读】： 该论文旨在解决传统自回归语言模型在预训练阶段仅依赖于下一个词元预测（Next-Token Prediction, NTP）所带来的局部、细粒度建模局限性，进而限制了模型对高层语义概念的捕捉能力。为突破这一瓶颈，其核心解决方案在于引入一种新型的“下一个概念预测”（Next Concept Prediction, NCP）目标，通过构建基于隐藏状态的量化概念词汇表（product-quantized concept vocabulary），在潜在空间中显式地学习跨多个词元的离散语义概念，并将这些概念作为中间表示反馈至词元级生成过程。该架构通过端到端联合训练实现概念级与词元级目标的协同优化，显著提升了模型的语义理解与生成效率。关键创新点在于：1）构建可解释且高效的潜空间概念表征；2）引入更具挑战性的概念级预测任务以增强模型的抽象能力；3）在保持标准自回归生成能力的同时，实现更优的下游性能。实验表明，该模型在仅使用51.3%训练数据的情况下达到与OLMo-3-7B相当的预训练损失，并在多项下游任务上超越基准模型，同时具备轻量级微调与高效推理潜力。
链接: https://arxiv.org/abs/2609.10715
作者: TheIntern-NCP Team:Jiaqi Cao,Chiyu Chen,Shuang Cheng,Xu Cheng,Beiya Dai,Yufan Feng,Kewen Ge,Ruijun Ge,Jiayi Huang,Yang Jiao,Dahua Lin,Zhouhan Lin,Yifan Liu,Yuliang Liu,Biqing Qi,Mowen Ruan,Junzhe Shen,Yunchong Song,Hao Sun,Zhongbo Tian,Yixuan Wang,Rubin Wei,Jiaxin Xiong,Kangyu Yang,Qian Yao,Qi Zhang,Bowen Zhou
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:We introduce NCP-ArchPreview, a latent-space language model that pushes autoregressive pretraining beyond standard next-token prediction (NTP). Alongside NTP, the model learns through Next Concept Prediction (NCP) to predict discrete concepts that span multiple tokens, introducing an explicit and more challenging concept-level objective while preserving standard token-level autoregressive generation. NCP-ArchPreview builds a latent space by constructing a product-quantized concept vocabulary directly from its hidden states, and subsequently learns to predict future concepts via a dedicated Concept Module. These predicted concepts are then fed back to the token level to guide subsequent generation, with NTP and NCP trained jointly end-to-end. We scale this architecture to 8.9B parameters and train it on 5.73T tokens from the Dolma-3 dataset, marking the largest demonstration of a latent-space language model to date. Remarkably, by consuming only 51.3% of the total training tokens, NCP-ArchPreview achieves the final pretraining loss of OLMo-3-7B. Following full pretraining, it outperforms OLMo-3-7B by 2.45 points on the downstream macro-average, including a notable 5.99-point gain on GSM8K. Controlled experiments isolate a clear progression of performance gains stemming from both the latent architecture and the NCP objective. Furthermore, utilizing only 85% of the standard computation, NCP-ArchPreview approaches the training loss of a strictly parameter-aligned 8.9B baseline. The learned latent space remains highly valuable after the pretraining stage: updating just the 17M-parameter VQ module yields a novel, lightweight interface for domain adaptation, while a simple injection of concept representations into a DFlash2 drafter improves the mean accepted length by 4.17% with negligible overhead.
[NLP-78] Data-Efficient Language Modeling: From Frontier Advancement to Principle-Guided Model Improvement
【速读】： 该论文旨在解决在有限文本数据条件下，模型如何有效利用上下文、实现对新输入的泛化能力，并持续保留有用知识的问题。其核心挑战在于，在资源受限（1000万词语料库，累计1亿词呈现量）的环境下，如何实现模型性能的长期递进式提升。解决方案的关键在于提出并验证了一项可测试的数据高效学习原则：围绕预测所需的上下文依赖关系组织经验；分别设计可见信息、监督信号与知识保留机制；并通过系统性实验检验学习、泛化与记忆能力。研究采用三阶段递归自改进（Research RSI）框架：第一阶段通过紧凑重述、预算再投资与残差增量学习构建前沿模型；第二阶段发现精确重复与对齐重述会引发不同模式的上下文使用行为，且熟悉任务表现无法保证对未见输入的计算复用；第三阶段通过保留源文本、掩蔽局部线索、选择性监督目标及保留常规掩码输入的预测结果，显著提升了生成质量——同一父本生成的两个延续种子在九项指标综合得分上超越普通延续，整体性能从42.02提升至42.25，达到2026年9月8日公开的Strict-Small基准最高分。该工作进一步探索了压缩、关系锚点、共享表示与度量方法，揭示了科研过程本身的自我优化机制，为小样本语言建模提供了理论与实践双重突破。
链接: https://arxiv.org/abs/2609.10702
作者: Shuxing Yang,Kaihao Zhu,Junjie Yang,Rui Zhao,Junyao Wu,Yize Wang,Wenhao Li,Fujia Chen,Taowen Deng,Shenzhan Hong,Yaqi Li,Zichen Li,Jincheng Mi,Yuang Pan,Hongsheng Chen,Yihao Yang
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Learning from limited text requires models to use context, generalize to new inputs, and retain useful capabilities. Qiushi Engine conducted a long-horizon, end-to-end autonomous research program on BabyLM 2026 Strict-Small, within 10 million corpus words and 100 million cumulative word presentations. Three stages connected frontier advancement, principle discovery, and principle-guided model improvement. Stage I combined compact restatements, budget reinvestment, and residual incremental learning to build a frontier model. Stage II found that exact repetition and aligned restatement produce different patterns of context use, depending on target relations and prediction windows. In controlled tasks, recovering familiar performance did not ensure that unseen inputs could still use learned computations. These findings support a testable data-efficient learning principle: organize experience around the contextual dependencies needed for prediction; separately design visible information, supervision, and preservation; test learning, generalization, and retention. Stage III retained source text, masked more local clues, supervised selected targets, and preserved predictions on ordinarily masked inputs. Two continuation seeds from the same parent outperformed ordinary continuation on the complete nine-metric aggregate. Overall rose from 42.02 to 42.25 across two generations; the second achieved the highest Overall in the public Strict-Small snapshot of 8 September 2026. Further studies addressed compression, relational anchors, shared representations, and measurement. Models are available on Hugging Face; code and research records accompany the GitHub repository. Together, these stages illustrate Research RSI: recursive self-improvement of the research process. Scientific understanding and method innovations change subsequent questions and designs; new experiments test and refine them.
[NLP-79] Artificial Intelligence Algorithms for the Detection of Pathologies Related to Lung Cancer through Image Analysis using Convolutional Neural Networks and Data Augmentation: a systematic mapping of the literature
【速读】： 该论文旨在解决肺癌早期诊断中医学影像解读过程复杂、依赖专业医师且存在效率瓶颈的问题。其核心挑战在于如何提升影像分析的准确性与自动化水平，以实现更及时、可靠的肺癌筛查。解决方案的关键在于应用生成式AI（Generative AI）与深度学习（Deep Learning, DL）技术，特别是基于迁移学习（Transfer Learning）和数据增强（Data Augmentation）的卷积神经网络（Convolutional Neural Networks, CNN），以克服小样本训练下的过拟合问题并提升模型泛化能力。研究结果表明，此类方法在提高肺癌检测的敏感性与特异性方面表现优异，具备显著临床应用潜力。然而，论文也指出当前仍面临数据标准化不足、模型可解释性差、患者隐私保护及伦理社会风险等关键挑战，强调需通过进一步研究与监管机制完善，确保技术在临床实践中的安全、可靠与负责任应用。
链接: https://arxiv.org/abs/2609.10652
作者: Pablo Ramirez Amador
机构: 未知
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:Lung cancer is one of the leading causes of death worldwide, and its early diagnosis is crucial to improving patients prognosis and quality of life. However, the process of interpreting medical images for the detection of lung cancer is complex and requires trained experts. In this context, artificial intelligence (AI) and deep learning (DL) emerge as potential tools to automate and optimize image analysis. The objective of this work is to review the most recent and relevant applications of AI and DL in the field of radiology for the detection of lung cancer. To this end, an exhaustive search was carried out in scientific databases such as PubMed,IEEEXPLORE, Scopus and Web of Science, and 96 articles published from 2015 to the present addressing the use of AI and DL in biomedical engineering were selected. Emphasis is placed on the use of convolutional neural networks (CNN) with transfer learning and Data Augmentation as promising techniques to improve the accuracy and efficiency of the image interpretation process. The results show that the use of AI and DL can offer an effective alternative for the early diagnosis of lung cancer, with high sensitivity and specificity. However, current limitations and challenges that must be addressed to guarantee its responsible and safe application in clinical practice are also identified, such as the lack of standardized data, the ex plainability of the models, patient privacy, and the ethical and social implications. It is concluded that the use of AI and DL can have a positive impact on the care of patients with lung cancer, but further research and regulation are required to ensure its quality and reliability.
[NLP-80] Biology-in-the-loop: Amortized Adaptive Hit Discovery in CRISPR Screens
【速读】： 该论文旨在解决在预算受限条件下，如何高效进行序列化实验设计以实现生物靶点发现的问题，尤其聚焦于CRISPR筛选中候选扰动的优先级排序难题。其核心挑战在于如何在多轮实验中动态优化实验选择策略，以在有限资源下最大化发现显著表型效应（hit）的概率。解决方案的关键在于构建一个大规模、多样化的基准数据集——AssayBench-Loop，涵盖1,389个跨五类表型的CRISPR筛选实验，从而支持对自适应发现方法的系统评估与历史经验学习。在此基础上，提出AssayLoop框架，融合基于Transformer的可扩展获取策略模型AssayFormer（amortized acquisition policy），通过历史实验反馈学习最优实验序列；同时引入大语言模型（LLM）生成的生物学先验知识，通过自适应交接机制实现知识引导的探索。该方法将已完成实验作为训练数据，使模型能够从累积证据中学习“下一步应测试什么”，并结合广义生物学先验加速搜索过程。此外，进一步提出AssayLLM，验证了直接对大语言模型进行任务特定微调即可实现类似效果。在时间上预留测试的筛选数据上，AssayLoop相比随机选择实现5.67倍富集，仅测试约5%候选库即恢复27.7%的已知有效靶点，显著优于现有自适应设计方法及独立使用大语言模型或仅使用AssayFormer的方案。性能随历史训练数据增加而提升，并可迁移至未参与训练的表型类别，证明了跨实验学习获取策略与结合广泛生物学先验在高效自适应靶点发现中的关键价值。
链接: https://arxiv.org/abs/2609.11877
作者: Carl Edwards,Edward De Brouwer,Xiner Li,Namkyeong Lee,Ehsan Hajiramezanali,Anne Biton,Sara Mostafavi,Gabriele Scalia
机构: Genentech, South San Francisco, CA, USA
类目: Quantitative Methods (q-bio.QM); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Genomics (q-bio.GN)
备注:
Abstract:Many biological discovery problems require experiments to be selected sequentially under constrained budgets. CRISPR screening is a prominent example, as exhaustive perturbation testing is often infeasible and candidate perturbations must instead be prioritized over multiple experimental rounds. Despite the importance of this problem, existing benchmarks for adaptive hit discovery remain limited in scale and diversity. Here, we introduce AssayBench-Loop, a large-scale benchmark for adaptive hit discovery comprising 1,389 CRISPR screens across five phenotype categories. Beyond enabling systematic evaluation, its scale makes it possible to learn acquisition strategies across historical experiments. Building on this resource, we introduce AssayLoop, a sequential experimental design framework combining AssayFormer, a transformer-based amortized acquisition policy trained across historical screens to adapt from experimental feedback, with LLM-derived biological priors through an adaptive handoff. In this view, completed experiments become training data for learning how accumulated evidence should guide what to test next, while LLMs provide prior biological knowledge to seed the search. We further introduce AssayLLM, showing that the same principle can be extended directly to an LLM through task-specific post-training. On temporally held-out screens, AssayLoop achieves a 5.67-fold enrichment over random selection and recovers 27.7% of hits after assaying approximately 5% of the candidate library, outperforming existing adaptive-design methods and standalone LLMs, and AssayFormer alone. Performance improves with increasing historical training data and transfers to phenotype categories excluded from training. These results demonstrate the value of learning acquisition policies across historical experiments and combining them with broad biological priors for efficient adaptive hit discovery.
[NLP-81] RetroThinker: Enabling Retrospective Thinking in Speech LLM s
【速读】： 该论文旨在解决生成式语音大模型（SpeechLLM）在实时语音交互场景下，既需保持低延迟又需具备强复杂推理能力之间的矛盾问题。现有方法虽通过思维链（Chain-of-Thought, CoT）与并行推理提升了推理性能，但仍面临准确性与延迟之间的固有权衡。其核心解决方案在于提出RetroThinker——一种多阶段后训练框架，使Moshi模型能够在推理过程中动态修正其思维轨迹。该框架结合了经过筛选的回顾性思考数据上的监督微调（SFT）与基于长度的直接偏好优化（length-based DPO），以优化早期推理阶段（即用户说话时同步进行的推理）中的回顾性验证与前向纠错能力。在GSM8K基准测试中，RetroThinker显著改善了准确率-延迟权衡，在保持相近延迟水平的前提下实现了11%的绝对准确率提升。
链接: https://arxiv.org/abs/2609.11864
作者: Yi-Jen Shih,Puyuan Peng,Abdelrahman Mohamed,David Harwath
机构: The University of Texas at Austin; FAIR, Meta Superintelligence Labs
类目: Audio and Speech Processing (eess.AS); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: Accepted to IEEE SLT 2026
Abstract:Speech large language models (SpeechLLMs) offer reduced latency and retain paralinguistic nuances that are typically lost in cascaded automatic speech recognition (ASR) and text-based LM architectures. However, they continue to lag behind text-only LLMs on complex reasoning tasks, while real-time spoken interaction imposes strict latency constraints. Although prior works employ Chain-of-Thought (CoT) and concurrent reasoning to enhance reasoning capabilities without inducing prohibitive delays, an inherent accuracy-latency trade-off persists. In this paper, we investigate whether a streaming SpeechLLM can dynamically revise its reasoning traces on the fly. We introduce RetroThinker, a multi-stage post-training framework that equips the Moshi model to self-verify and forward-correct CoT steps during inference. RetroThinker combines supervised fine-tuning (SFT) on curated retrospective thinking data with length-based direct preference optimization (DPO) to optimize retrospective during early reasoning (i.e., reasoning concurrently while the user speaks). Evaluated on the GSM8K benchmark, RetroThinker significantly improves the accuracy-latency trade-off over non-retrospective baselines, achieving an 11% absolute accuracy gain at a comparable latency.
[NLP-82] Whisper-Based Speech Transcription from Videos Across Multiple Languages for Cross-Cultural Understanding
【速读】： 该论文旨在解决跨文化语境下非母语使用者在多语言环境中理解与沟通障碍的问题，核心挑战在于如何高效构建高质量的自动化辅助工具以支持跨文化理解。现有技术依赖于真实世界中的多模态数据（如视频、音频和文本），但语音识别生成的转录文本普遍存在较高错误率，限制了下游应用的可用性。本文提出了一种无需深度语音处理专业知识即可应用的技术方案，通过利用公开的YouTube视频数据与基于Whisper的语音识别工具，在七种语言（西班牙语、日语、韩语、汉语、土耳其语、俄语和希伯来语）上实现平均转录错误率从30%降低至20%，显著提升了转录结果的可用性。其关键在于采用少量微调数据即可实现性能提升，并开放相关语音数据与元数据供社区持续优化，从而推动跨文化工具的可扩展开发。
链接: https://arxiv.org/abs/2609.11772
作者: Michael Picheny
机构: NYU Courant Institute School of Mathematics, Computing, and Data Science (纽约大学库朗数学、计算与数据科学学院); NYU (纽约大学)
类目: Audio and Speech Processing (eess.AS); Computation and Language (cs.CL)
备注: 7 pages, 2 figures, 5 tables
Abstract:Cross-cultural understanding has become increasingly important in today’s highly connected, cross-national world. The success of LLM-based technologies is now driving the development of automated tools to aid understanding for nonnative people trying to succeed in cross-cultural environments. Building such automated tools is often done by leveraging in-thewild text, audio, and video data. This paper presents techniques for improving speech recognition-based transcript creation in multiple languages from videos to better train these automated tools. The focus is on processes and speech tools that can easily be used by cross-cultural tool builders without requiring deep speech processing expertise. Using publicly available videos from YouTube and Whisper-based tools, average transcription error rate across seven languages (Spanish, Japanese, Korean, Mandarin, Turkish, Russian, and Hebrew) of 30% are observed. With a modest amount of fine-tuning data, the average error rate can be reduced to 20% making such output much more usable for downstream processing. Speech and metadata associated with these videos that can be used by the community to further refine these experiments are released as well.
[NLP-83] More than half of recent astronomy papers are written with language-model assistance
【速读】： 该论文旨在解决生成式人工智能（Generative AI）在天体物理学领域学术写作中渗透程度的量化问题，即如何准确评估当前天文文献中由语言模型辅助撰写的文本比例。其核心挑战在于：由于直接披露使用情况的论文比例极低（2025年仅0.81%），无法通过显式声明进行有效估算，且语言模型引入的词汇特征会随作者适应而逐渐淡化，导致可识别性下降。解决方案的关键在于构建一个分层贝叶斯混合模型，将每篇论文的词汇使用频率建模为“人工撰写”与“模型辅助撰写”的混合分布，并利用2020年前未受模型影响的论文数据校准自然写作基线，再结合392篇明确披露使用模型的论文校准辅助写作特征。通过在三种不同假设下外推2020年后背景率，得出2025年约有54%±8%（统计误差）+26%-0%（系统误差）的论文含有语言模型痕迹，且该估计在多种参数变化下均保持在36%以上。此外，研究还发现模型痕迹因作者适应而趋于隐蔽，但所提模型具备捕捉微弱信号的能力，从而实现对实际渗透率的可靠推断。
链接: https://arxiv.org/abs/2609.10664
作者: Serat M. Saad,Yuan-Sen Ting
机构: The Ohio State University(俄亥俄州立大学); Max-Planck-Institut für Astronomie(马普天体物理研究所)
类目: Instrumentation and Methods for Astrophysics (astro-ph.IM); Computation and Language (cs.CL); Digital Libraries (cs.DL)
备注: 14 pages, 6 figures, 2 tables
Abstract:Language models leave a distinctive vocabulary in the prose they help write, and we measure how much of the astronomy literature now carries it. From the full text of 207,111 astro-ph papers spanning 2015 to mid-2026, we count those words in each paper and model the counts, in proportion to paper length, as a mixture of assisted and unassisted writing in a hierarchical Bayesian model. Papers from before 2020 calibrate the unassisted rate, and the 392 papers that disclose model use calibrate the assisted one. Our answer depends on how often these words would appear today if nobody used a model, a rate that must be modeled rather than observed, so we extend it past 2020 under three assumptions and report all three. For 2025 that gives 54^+8_-8,(\mathrmstat,,95%),^+26_-0,(\mathrmsys,\ background) % of papers, the second error being the spread across the three. The estimate stays at or above 36% when we vary that choice, the calibration, and the requirement that adoption only rises. A word list built from the astro-ph corpus, keeping only words that rose across every subfield, leaves 2025 in the same range. Assisted writing is also getting harder to see, since authors adapt to the words that reveal it and the marker excess more than halves between 2023 and 2026. Our model allows for that fading, so it can separate a fainter trace from reduced use. More than half of recent astro-ph papers therefore carry a language-model trace, while only 0.81% of 2025 papers disclose it, one declaration for every \sim 66 papers with a trace.
信息检索
[IR-0] Generative Late-Interaction Embeddings For Visual Document Retrieval
链接: https://arxiv.org/abs/2609.11808
作者: Mohamed Eltahir,Talal Aloushan,Rose Khairoalsendi,Jana Shata,Mohammed Alhassan,Leen Alrehaili,Tanveer Hussain,Naeemullah Khan
类目: Information Retrieval (cs.IR)
备注:
Abstract:Late-interaction retrieval is the state-of-the-art for visual document search, but it pays for its accuracy in storage. Existing compression methods retain a subset or local average of the N~1,000 vectors per page. Under aggressive storage budgets, however, these methods degrade sharply, and alternatives require retraining the encoder. Investigating this degradation across three encoders, we found two consistent properties: the vectors lie exactly on the unit sphere and concentrate near a manifold of intrinsic dimension five to six. This geometry yields two insights. First, standard k-means centroids fall inside the sphere, causing systematic underestimation of MaxSim scores. Normalizing them to the surface is a free correction worth up to +0.093 nDCG@5 over raw centroids. Second, because the page manifold has few degrees of freedom, the full set of vectors can be regenerated from only a few. To this end, we introduce Generative Late-Interaction Embeddings (GLIE): k N vectors per page learned from the normalized centroids to serve as both a lightweight index and a basis for regenerating the page’s full embedding set. At query time, search runs exclusively on these k vectors, and a decoder expands only the top candidates back to all N vectors for exact rescoring. At four vectors per page on ViDoRe v1, GLIE retains nearly 80% of the uncompressed system’s nDCG@5, against 70% for the best prior post-hoc method. These results use a 415K-parameter network fitted in under three GPU-minutes on just a thousand training pages. At a matched training budget, fine-tuning the encoder does not reach even the training-free stage of GLIE, and the full system beats it at every budget. These patterns hold across a second encoder and ViDoRe v2. By reconstructing evidence on demand rather than sampling it, GLIE opens a new axis for storage-efficient retrieval, with the decoder as its main design surface.
[IR-1] RAG -Safety-Bench: Reliable Evaluation of Retrieval-Augmented LLM Safety EMNLP2026
链接: https://arxiv.org/abs/2609.11758
作者: Adithiyan Rajan Indira Saravanan,Kathleen C. Fraser
类目: Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注: Proceedings of EMNLP 2026 (main conference)
Abstract:Allowing large language models (LLMs) to retrieve information from a set of trusted documents can increase reliability and reduce hallucination. However, recent work has demonstrated that retrieval-augmented generation (RAG) can have unintended side effects on the overall safety of the generated responses, when prompted for harmful or dangerous content. A clearer understanding of the mechanisms leading to this result is needed, as increasing numbers of end users turn to RAG to incorporate corporate documents and knowledge bases into LLM-based systems. We introduce RAG-Safety-Bench, a benchmark to measure the safety impact of RAG on LLM models. By removing the confounding effect of retriever quality, and cleanly separating the problem into four conditions – non-RAG, RAG with an oracle document containing the answer to the harmful request, RAG with documents related to the harmful request but without the specific answer, and RAG with random, safe documents – the benchmark isolates the impacts of different factors in the observed safety degradation. We report results across five open-source LLMs, showing an inverse relationship between benign and unsafe capability, strong evidence that baseline safety guardrails do not lead to downstream safety guarantees in the RAG case, and model-specific support for previous findings that even benign documents can lead to unsafe generation in retrieval-enabled systems.
[IR-2] Your Retriever Already Knows: Distribution-Shape QPP for RAG Retrieval Sufficiency
链接: https://arxiv.org/abs/2609.11646
作者: Matyáš Veselý,Michal Průšek,Jiří Franc
类目: Information Retrieval (cs.IR)
备注: 12 pages, 3 figures, 6 tables. Published at TSD 2026
Abstract:Standard Retrieval-Augmented Generation (RAG) pipelines often provide no reliable inference-time signal of whether retrieval succeeded; on ambiguous or out-of-scope queries, generation may then hallucinate. Motivated by a Czech nuclear-regulator deployment where data sensitivity precludes third-party LLM APIs, we compare three Query Performance Prediction (QPP) paradigms for retrieval sufficiency in RAG: score-based features, a content-based LLM judge, and a hybrid. On the eight ViDoRe vision domains (14,514 queries), our 24 non-lexical features (GeneralQPP; 15 distribution-shape, 5 query-surface, 4 global) reach a weighted-average AUROC of 0.856 at 2 ms per query, ahead of a classic-QPP literature pool (Classic Full, 0.835) and well above a local Qwen3.5 LLM judge (0.649, +0.207 gap; \sim 3000 \times faster and cheaper per query). Adding the LLM judgment as one feature (hybrid) matches S1 on ViDoRe (0.863) but gains a statistically significant edge on SÚJB (AUROC 0.911 at Hit@5, adversarial-detection 0.954; 1,510 queries, 500 synthetic adversarial), at LLM latency. Rankings agree across datasets (Spearman \rho = 0.90 ). Under Leave-One-Domain-Out, S1 drops to 0.706; a 13-feature LODO-stepwise subset (S1-Lean) recovers to 0.719 (+0.032 over the literature pool).
[IR-3] FedHUR: Learning Hierarchical Utility-Guided Client Relations for Personalized Federated Recommendation
链接: https://arxiv.org/abs/2609.11632
作者: Mingzhe Han,Jiahao Liu,Dongsheng Li,Jiankui Zhou,Hansu Gu,Peng Zhang,Ning Gu,Tun Lu
类目: Information Retrieval (cs.IR)
备注:
Abstract:Federated recommendation enables collaborative model training while keeping user interaction data on local clients. A central problem in federated recommendation is how to aggregate useful information across clients for personalized recommendation. Existing personalized aggregation methods usually construct client relations from predefined parameter-based assumptions, such as parameter similarity or complementarity, and use these relations to determine aggregation weights. However, such methods construct a single global relation, which is insufficient to capture the hierarchical and multi-granularity nature of user relations in recommendation. Moreover, these predefined relations cannot directly reflect whether the related clients can improve prediction performance after aggregation. To address these limitations, we propose FedHUR, a federated recommendation framework for learning hierarchical utility-guided client relations. FedHUR takes item-item filters as the object for relation construction and aggregation. Specifically, it first aggregates and clusters each client’s local information to obtain global hierarchical information. Each client computes hierarchical utility signals based on its local information and the global hierarchical information, indicating which collaborative information is useful for improving its prediction. The server uses these utility signals to retrieve clients that are useful to that client for further personalized aggregation. Extensive experiments on five real-world datasets show that FedHUR consistently outperforms existing federated recommendation baselines, demonstrating the effectiveness of hierarchical utility-guided client relation learning. Code is available at this https URL.
[IR-4] melyRAG : Semantic-Temporal Hybrid Retrieval for Time-Critical Question Answering in Overlapping-Evolving Documents
链接: https://arxiv.org/abs/2609.11572
作者: Youngeun Nam,Joeun Kim,Hwanjun Song,Susik Yoon,Jae-Gil Lee,Byung Suk Lee
类目: Information Retrieval (cs.IR)
备注: 17 pages, 5 figures, 15 tables
Abstract:Although large language models (LLMs) and retrieval-augmented generation (RAG) have advanced open-domain question answering (QA), they remain unreliable when documents evolve through amendments. Existing time-sensitive retrieval methods address only the disjoint-evolving environment, where each update is an independent snapshot. However, laws, policies, and regulations often operate in overlapping-evolving environments, where amendments override earlier clauses while preserving most content, creating strong semantic overlap across versions. We propose TimelyRAG, a retriever-agnostic framework that incorporates temporal distance into ranking to align queries with version-appropriate documents. We also introduce TimelyQABench, the first benchmark for regulation-heavy domains with overlapping-evolving challenges. Experiments show consistent gains, up to +28.6% in nDCG@10, highlighting the importance of temporal reasoning for reliable QA over evolving documents. All resources are available at this https URL.
[IR-5] ReGround: Grounding Reviewer Comments in Multimodal Evidence EMNLP2026
链接: https://arxiv.org/abs/2609.11460
作者: Serwar Basch,Lizhen Qu,Iryna Gurevych
类目: Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注: Accepted at EMNLP 2026
Abstract:Reviewer comments naturally relate to specific parts of the reviewed paper, yet grounding these comments to the underlying evidence is difficult due to long multimodal documents. Existing benchmarks do not capture this setting and largely focus on explicit, information-seeking queries. We introduce ReGround, a large-scale dataset for reviewer comment grounding that links 10,267 reviewer comments to 16,274 evidence in the original anonymous submission of 3,656 papers. We build on a simple observation: author rebuttals often include explicit references to content of the submission used to address reviewer comments, providing a high-precision annotation source. We cast grounding as a retrieval task and evaluate a wide range of retrieval methods. Results show that retrieval over the entire paper content performs poorly, evidence-type inference is a major bottleneck, and multimodal evidence provides complementary signals that text alone misses. Our dataset exposes grounding reviewer comments as a difficult and practically important problem for scientific document understanding.
[IR-6] SWRouter: Similarity-Contractive Window Routing for Multi-Turn Large Language Model Conversations
链接: https://arxiv.org/abs/2609.11414
作者: Yu Wang,Yuchen Li,Rui Kong,Xinran Chen,Jiamin Chen,Hengyi Cai,Shuaiqiang Wang,Jiashu Zhao,Yulun Zhang,Zhonghao Lyu,Haoyi Xiong,Linghe Kong,Jimmy Xiangji Huang,Dawei Yin
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注:
Abstract:Large language models exhibit complementary strengths, motivating routing methods that dispatch each query to the most suitable model. Although existing routers are effective in single-turn settings, they do not directly transfer to multi-turn dialogue, where routing performance critically depends on how historical context is segmented, retained, and incorporated into the current prompt. This introduces two fundamental challenges: preventing information loss and information confusion during context construction, and evaluating routing quality without conflating model selection with prompt construction quality. In this paper, we propose SWRouter, a Similarity-Contractive Window Router for multi-turn large language model routing. SWRouter combines a similarity-based context segmentation mechanism for prompt construction with a dual-metric evaluation framework that decouples construction accuracy from router performance. Experiments on multi-turn dialogue benchmarks demonstrate that SWRouter consistently surpasses strong baselines, achieving a 16.26% improvement in evaluation accuracy over the best individual large language model and an additional 8.22% gain over the Conv-ID Context baseline. Our results highlight that multi-turn large language model routing requires a joint design of context construction and evaluation, rather than a direct extension of single-turn routing methods.
[IR-7] VikingRAG : Accurate and Token-efficient Retrieval-augmented Generation over Structured Documents
链接: https://arxiv.org/abs/2609.11390
作者: Peiyuan Gao,Gaoyuan Zhang,Haojie Qin,Yahui Sun,Qianyi Zhang,Yunhao Zhang,Zeyu Wang,Wei Lu
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Databases (cs.DB); Machine Learning (cs.LG)
备注:
Abstract:State-of-the-art retrieval-augmented generation (RAG) methods exploit document structures to acquire sufficient evidence, but often incur substantial token costs. To reduce structural-context tokens without compromising high RAG accuracy, we present \sf VikingRAG, a directory-aware semantic data management system that tightly integrates semantic and structural access to support structural-context-efficient, evidence-gap-driven multi-round retrieval. To further reduce token overhead of multi-round interaction, we materialize agentic multi-round retrieval traces as experience edges, and reuse these edges for similar queries, avoiding repeated multi-round exploration. To additionally reduce token costs when agentic multi-round retrieval is unnecessary, we introduce an adaptive escalation strategy that answers from one-round experience-augmented retrieval when the evidence is sufficient, and invokes agentic multi-round retrieval only otherwise. Experiments on real datasets show that the base system \sf VikingRAG matches high accuracy of state-of-the-art methods while consuming only 11.6%–51.9% of their tokens. With retrieval-trace reuse and adaptive escalation, token costs drop to 5.1%–32.5% while maintaining competitive accuracy and practical document-storage performance, showing the utility of this work for emerging AI knowledge bases.
[IR-8] REVA: Reusable Evidence View Aggregation for Context-Efficient RAG Serving ICDM
链接: https://arxiv.org/abs/2609.11209
作者: Tuan Nguyen,Qiran Hu,Banruo Liu,Khoa D. Doan,Kok-Seng Wong,Fan Lai
类目: Machine Learning (cs.LG); Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注: Author’s accepted manuscript. Accepted for publication in the 2026 IEEE International Conference on Data Mining (ICDM)
Abstract:Retrieval-augmented generation (RAG) improves knowledge-intensive large language model (LLM) applications by conditioning generation on retrieved documents, but longer contexts increase latency, key-value (KV) cache memory, and token cost. Post-retrieval compression can reduce this cost, yet existing compressors often operate independently for each query, rely on auxiliary models or rewriting, and introduce online overhead that can offset the benefit of shorter prompts. We revisit RAG compression from a data-mining perspective by aggregating historical query–document–model interactions into reusable evidence views. We first show that modern compressors have unstable gains over simple truncation and can add substantial inference-time latency. We then propose Reusable Evidence View Aggregation (REVA), a framework that mines the target generator’s historical attention traces into a document-keyed, budget-agnostic score store. REVA maps token-level attention to readable word units, aggregates importance across repeated document accesses, and renders budget-specific plain-text views that preserve document order and the standard RAG interface. Across four representative benchmarks and modern LLMs, REVA improves generation quality by 1.0–5.8 points over existing advances, while reducing compression overhead by a factor of 5.3 to 15.6, adding less than 40 ms of latency.
[IR-9] Agent ic Share-of-Search: A Multi-Agent AI System for Competitive Decision-Making in LLM -Mediated E-Commerce
链接: https://arxiv.org/abs/2609.11190
作者: Spandan Ghose Chowdhury
类目: Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注: Accepted for presentation at the 2026 Decision Science Institute Annual Conference
Abstract:AI shopping assistants increasingly redirect consumer discovery, creating an urgent need for tools that support seller-side competitive decision-making. We present a multi-agent AI system that automates competitive visibility measurement and root cause diagnosis in LLM-mediated ecommerce. The system introduces Agentic Share-of-Search (ASoS) as the decision target, deploys query agents across leading AI platforms, and uses a ReAct-based diagnostic agent to recommend prioritized merchandising interventions. A 100-trial ablation study, presented as a feasibility evaluation of this prototype, shows the agent recovers the ablated signal in 39% of trials (95% CI: 30.0% - 48.8%, 5.5x over chance), rising to 63.9% among high-correlation ablations.
[IR-10] Benchmark Radar: A Living Database and Search Engine for AI Benchmarks and Evaluation
链接: https://arxiv.org/abs/2609.11115
作者: Koutian Wu,Junjie Zhou,Ergan Shang,Jiayu Wang,Pengqian Han,Junkai Wang,Wanghan Xu
类目: Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注: Project site: this https URL Code: this https URL
Abstract:Benchmark researchers and developers of large language models (LLMs) and other AI systems need to find relevant evaluations, locate their benchmark datasets and code, and understand the settings behind reported scores. We present Benchmark Radar, a living database and search engine for retrieval and discovery of AI benchmarks, covering LLM evaluation, agentic and tool-use benchmarks, coding, reasoning, safety, and domain-specific evaluations. The system combines daily discovery of benchmark papers, repositories, datasets, and releases with a searchable benchmark catalog, mentions in model cards and technical reports, and score histories. It retains source identities and citations so readers can inspect candidate benchmarks and their evaluation evidence. Daily discovery draws on 37 sources: 13 direct connectors and 24 first-party research and engineering feeds. The catalog contains 1,283 source records drawn from 4 benchmark catalogs and 12,916 numeric observations on 790 records. We describe collection and retrieval, audit the full catalog, and examine benchmark saturation, adoption trends, and the limits of score comparisons. A worked example walks through a complete prior-art search, showing how to query the catalog and inspect benchmark evidence when designing a new evaluation. We release the web dashboard with a benchmark leaderboard, a Pareto frontier view of score against measured use, saturation and trend views, daily feeds, downloadable evidence, a command-line interface (CLI) for offline queries, and reproducible analysis.
[IR-11] UniRec: Cross-stage Multi-Task Fusion with Preference Alignment for Cascaded Recommender Systems
链接: https://arxiv.org/abs/2609.11052
作者: Lingyuan Kong,Jiaqi Cui,Fanjiao Zeng,Congqi Wang,Yu Li,Yuan Cheng,Jingxin Liu,Xiaoshuang Chen,Kaiqiao Zhan
类目: Information Retrieval (cs.IR)
备注:
Abstract:Industrial recommender systems use cascaded stages with different objectives, feature spaces, and latency constraints. Optimizing pre-ranking and ranking separately can create cross-stage inconsistency: upstream models may filter out items preferred by downstream rankers, and independently tuned downstream fusion can offset upstream improvements. Existing multi-task fusion methods focus on multi-objective fusion within the ranking stage, and cross-stage methods typically only add a downstream score factor to upstream ranking. Joint optimization of fusion modules across both stages remains largely unexplored. We propose UniRec, a Unified Cross-stage Recommendation Fusion model. First, the two fusion agents partially share input embeddings and are trained in a single computation graph, so gradients from either stage propagate through the shared representation and influence the other. Second, we introduce a dual-axis preference alignment objective: a vertical cross-stage consistency term transfers downstream pairwise preferences to the upstream fusion score, and a horizontal compact aggregation term reorganizes dozens of pairwise objectives over heterogeneous prior signals into bidirectional preference evidence. Third, we find that unconstrained end-to-end fusion optimization can exploit imbalances in item attribute distributions, over-concentrating on high-reward regions at the cost of other objectives. We therefore add an attribute group-relative regularization that computes advantages within attribute groups and normalizes the policy over the same groups, so uniformly promoting an entire high-reward group yields no optimization gain. Offline, UniRec consistently outperforms single-stage fusion and cross-stage coordination baselines. Online A/B tests show a 0.616% gain in app usage duration. UniRec is fully deployed on the Kuaishou platform. Subjects: Information Retrieval (cs.IR) Cite as: arXiv:2609.11052 [cs.IR] (or arXiv:2609.11052v1 [cs.IR] for this version) https://doi.org/10.48550/arXiv.2609.11052 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[IR-12] Project Qualia: Recovering Experiential Music Structure from Session Co-occurrence Data
链接: https://arxiv.org/abs/2609.10862
作者: Nizam Mohammed,Abu B. S. Rahman,Dimuthu D. K. Arachchige
类目: ound (cs.SD); Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注: 10 pages, 3 figures, 4 tables
Abstract:This report presents results from Project Qualia, an ongoing effort to determine whether experiential similarity between songs, a structure not captured by genre or metadata taxonomies, can be recovered from real listening behavior. We constructed a large-scale dataset of listening sessions, comprising 1.29 billion scrobbles collected from 9,396 users via the this http URL API and reduced through a preprocessing pipeline to 531.6 million training scrobbles across 28.6 million sessions. On this corpus, we trained a skip-gram Word2Vec model (Song2Vec), treating each session as a sentence and each track as a token. As anticipated, the resulting embedding space was dominated by artist identity, a consequence of single-artist runs within sessions. To test for a subtler, artist-independent signal, we developed an artist-residual procedure: subtracting each artist’s centroid from its tracks’ embeddings and evaluating whether the remainder retained structure. Mean cross-artist cosine similarity fell from 0.2487 in raw embedding space to 0.0005 in residual space, yet 4,577 cross-artist track pairs retained cosine similarity \ge 0.70 in residual space, forming coherent genre- and era-based clusters, including trip-hop, 1990s grunge, 2020 mainstream pop, and cross-composer classical piano pairs at cosine similarity up to 0.95. These results confirm that the training data contains experiential structure independent of artist identity, establishing an empirical basis for an architecture designed to learn this experiential layer directly.
[IR-13] Following the Preference Missing the Optimum: Compliance Without Optimization in AI Housing Recommendation
链接: https://arxiv.org/abs/2609.10856
作者: Hsuan Lo
类目: Computers and Society (cs.CY); Information Retrieval (cs.IR)
备注: 59 pages, 4 figures, 31 tables. Code, prompts, and per-call results: this https URL
Abstract:Large language models are becoming the first point of contact for consumer search in domains where the stakes are material and the law is explicit. Existing audits show that models steer housing seekers by perceived identity, but none can say what a user loses when a recommender overlooks a suitable option, for want of an enumerated inventory to score omissions against. We audit AI housing recommendation against a verifiable ground truth. For each of 150 synthetic renter scenarios in New York City we build a pool of 120 real listings with known rent, bedrooms and GTFS-computed transit commute, compute the exact set satisfying the renter’s stated constraints, and derive its Pareto frontier. The primary outcome assumes no utility function: a recommendation is strictly dominated if the same pool holds a listing cheaper, faster to commute from and no smaller in bedrooms. Across 9,945 calls to three models from two vendors, compliance is near-perfect (1.8% violation against a 66.6% random floor), yet 39.0% of recommendations are strictly dominated, and the dominating listing is a median 900 USD/month cheaper and 3.5 minutes closer. A within-scenario manipulation separates two capabilities usually conflated: changing one sentence moves median recommended rent by 646 USD/month in the correct direction, so preferences are honored, yet recommendations still sit 606 USD/month above the five cheapest qualifying listings on the same screen, and an unambiguous lexicographic instruction gives no improvement under equivalence testing against a pre-specified 50 USD/month bound. The gap widens with candidate-set size and replicates across OpenAI and Anthropic models to within 3 USD. We characterize the failure as compliance without optimization, propose dominance-rate instrumentation as a deployable diagnostic, and release all code, prompts and per-call results.
[IR-14] When Synthetic Data Hurts: On Catastrophic Forgetting in Skill Retrieval for LLM Agents EMNLP
链接: https://arxiv.org/abs/2609.10750
作者: Syed Shariyar Murtaza,Yifan Nie,Utkarsh Soni,Eugene Wen,Arvid Frydenlund
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 8 main pages, 15 pages total, accepted in EMNLP Industry track 2026
Abstract:LLM agents increasingly rely on external skills retrieved at runtime, making skill selection from large repositories a critical challenge. We present a production skill router over 34,396 skills and a large-scale study of skill retrieval using limited real supervision and synthetic data. We found that the synthetic-data fine-tuning improves in-distribution retrieval but it causes catastrophic forgetting on real and out-of-distribution (OOD) data. We evaluate several forgetting mitigation fine-tuning approaches inspired by continual learning, including embedding-anchor regularization, Learning without Forgetting (LwF), Elastic Weight Consolidation (EWC), and L2-initialization. The results show that these approaches not only retain the performance on OOD skills retrieval but also improve the retrieval on synthetic in-distribution skills by 13.98% for 0.6B Qwen retriever and reranker. Our results provide a practical benchmark and a robust fine-tuning recipe for scarce, multi-positive supervision.
人机交互
[HC-0] Caption-once Frames-on-Demand: Visual-Need Routing for Budget-Aware Agent ic Long Video Understanding EMNLP2026
链接: https://arxiv.org/abs/2609.11899
作者: Weitong Cai,Hang Zhang,Yukai Huang,Yiqiao Xie,Shan Gao,Jiankang Deng,Songcen Xu,Jifei Song,Zhensong Zhang
类目: Computer Vision and Pattern Recognition (cs.CV); Human-Computer Interaction (cs.HC)
备注: EMNLP 2026 Main Conference
Abstract:Long-video understanding on edge devices must reason over hours of content under tight compute and bandwidth budgets. Subsampling visual tokens loses temporal structure, while text-only video memories lose fine-grained visual attributes. We observe a visual-textual duality: language memories carry long-range temporal structure better than dense frames, while pixels remain decisive for attribute-level perception. Building on this insight, we propose Caption-once, Frames-onDemand (CFD), a budget-aware edge-cloud agentic framework. The edge runs a single offline captioning pass that builds a dual-track narrative index, an event-level story skeleton plus a clip-level micro-log, cached and reused across queries without re-captioning. At query time, a cloud-side MLLM reasons over the index in a story-first loop centered on a lightweight Visual-Need Router: a per-query gating module that triggers bounded keyframe retrieval only for perceptual questions (appearance, on-screen text, attribute disambiguation) and keeps temporal-structural questions in language space. The router turns visual access into a first-class, query-conditioned cost, capping per-query frame consumption regardless of video length. Experiments on long-video benchmarks demonstrate strong accuracy-efficiency trade-offs while substantially reducing online visual processing.
[HC-1] MotionQ: Operator-Conditioned Motion Quotients for Cross-Observation WiFi Gesture Recognition
链接: https://arxiv.org/abs/2609.11818
作者: Xiang Zhang,Huan Yan,Geying Yang,Jianchun Liu,Tao Liu,Zhi Liu,Meng Li
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:WiFi gesture recognition is accurate in fixed deployments but often degrades when user orientation, available links, or transceiver placement changes. Unlike ordinary domain shifts, these changes alter the wireless observation operator, so the same motion is expected to produce different measurements. Existing methods nevertheless pursue domain-invariant features and largely overlook changing layouts and observation configurations. Yet changing the observation operator also changes which task-relevant motion cues are physically observable, rather than merely altering the appearance of a fixed set of cues. Under a local linearization of the WiFi forward process, we derive a common task-observability condition under which a strict common linear representation is recoverable from every geometry-induced operator while preserving the gesture task. When the condition fails, enforcing stronger alignment across additional heterogeneous source operators may discard task-relevant cues still observable under individual operators. We therefore present MotionQ, which generates an operator-conditioned two-support motion measure for each candidate geometry. A motion quotient removes only the arbitrary ordering of its unlabeled supports and is represented by permutation-invariant central moments. Rather than matching quotients across operators, single-link-retention interventions encourage each view to retain information sufficient for gesture recognition. Extensive evaluations show that MotionQ is robust to extrapolative observation operators.
[HC-2] Understanding Operator Attitudes Toward AI-Supported Decision Making in Maritime Operations
链接: https://arxiv.org/abs/2609.11805
作者: Doreen Jirak,Armeen Saroukanoff,Dirk van Rooy
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注: 11 figures
Abstract:Maritime Autonomous Surface Ships (MASS) and AI- supported decision assistants are expected to transform maritime operations, but their safe integration depends on how maritime professionals perceive and trust such systems. This paper presents a survey study on maritime stakeholders’ attitudes toward an AI-supported assistant in collision-avoidance scenarios. Participants evaluated technology anxiety, trust in automation, and explanation quality using established and adapted questionnaires, complemented by sentiment and thematic analysis of open-ended responses Results indicate a generally positive disposition toward maritime technology, no clear age-related differences in openness, stable trust across scenarios, and more scenario-sensitive, multidimensional explanation ratings. Open responses showed that participants valued support for decision-making, situation awareness, and confidence-building, while raising concerns about AI reliability, over- reliance and loss of expertise. The findings suggest that maritime AI systems should not focus solely on increasing automation or trust, but on supporting calibrated reliance through transparent, reliable, and operationally meaningful design with domain experts in the loop.
[HC-3] Bigger than the EAR BOX: A Theory-Grounded Review of XR Accessibility Research for Deaf and Hard of Hearing Communities
链接: https://arxiv.org/abs/2609.11706
作者: Shuxu Huffman,Michaela Okosi,Jose Merino,Tifanie Bouchara,Abraham Glasser,Christian Vogler,Raja Kushalnagar
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Extended Reality (XR) technologies have received growing attention in accessibility research involving Deaf and Hard of Hearing (DHH) communities. Yet less attention has been given to the assumptions shaping this work. We present a theory-grounded review of XR research involving DHH users. Drawing on Disability Studies, Deaf Studies, and DeafSpace, we develop a theoretical framework with four analytical dimensions: orientation toward access, distribution of responsibility, conceptualization of DHH communities, and spatial and perceptual assumptions. We apply this framework to 53 XR studies involving DHH users, identified through a search and screening of ACM publications from 2015 to 2025. Our analysis shows that XR accessibility is frequently framed as supporting communication within hearing-default environments, while overlooking the diversity of DHH communities. We identify directions for redistributing accessibility labor and reconfiguring space, and offer our framework as a tool for future XR research involving DHH communities.
[HC-4] Ethics Training Agents : Facilitating Group-Based Ethics Education with Role-Playing and Discussion for Ethical Reflection and Exploration
链接: https://arxiv.org/abs/2609.11529
作者: Youngseok Seo,Sueun Jang,Hyesoo Park,Renz Samuel Gutierrez,Joseph Seering,Uichin Lee
类目: Human-Computer Interaction (cs.HC)
备注: 45 pages, 5 figures, 3 tables
Abstract:Group-based ethics training for Science, Technology, Engineering and Mathematics (STEM) students is a complex challenge, requiring substantial resources and expertise. While activity-based teaching methods, such as role-playing and discussions, are commonly employed to simulate real-world scenarios, current practices are often manual and lack integration with effective online platforms for supporting group-based ethical discussions. In this work, we propose Ethics Training Agents, a group discussion system that leverages multiple LLM participants embodying distinct ethical orientations, along with a moderator agent, to enable structured human-AI group ethical discussions for collaborative reflection. We conduct a user study with 45 undergraduate STEM students to evaluate the learning outcomes and user experience. The results show that our system supports engagement, coordination, and perspective-taking in group discussions and has a positive influence on ethical sensitivity. We also discuss practical design strategies for integrating multiple LLM agents into multi-human group settings to facilitate ethics training for STEM students.
[HC-5] Design Reflections on Transition to LLM -Aided Novel Visualizations
链接: https://arxiv.org/abs/2609.11503
作者: Richard Brath
类目: Human-Computer Interaction (cs.HC)
备注: In Proceedings of The First Reflection in Creative Experience (RiCE) Workshop (RiCE W1) arXiv:2607.24558
Abstract:This study examines data visualization design evolution over 12.5 years, reflecting on the impact of Large Language Models over the last 3.75 years. Using a longitudinal corpus of 55 visualizations from a single-subject design record, the study identifies how LLMs have aided design-space exploration: reducing coding effort, enabling new design opportunities, shock, excitement, accomplishments, and shifts to the design process.
[HC-6] he Convention Gap: Towards Measuring Implicit Communication in Cooperative AI Evaluation
链接: https://arxiv.org/abs/2609.11489
作者: Makoto Fukushima,Hua-Dong Xiong,Ehsan Moradi Pari
类目: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注:
Abstract:Cooperative AI agents are evaluated against other AIs, yet human cooperation relies on implicit conventions—shared protocols for reading meaning beyond the literal message—which AI-AI benchmarks may not capture. We propose the \emphconvention gap, the difference between the failure probability predicted from the literal content of communication and the observed failure rate, as a metric of implicit communication. In the card game Hanabi, the finite deck and deterministic hint constraints make this posterior exactly computable. We replayed about 101,000 play actions from three public datasets of human-human (this http URL), AI-AI (HOAD), and human-AI (HanabiData) games. The gap was +26.2 percentage points (pp) in human pairs, - 0.7~pp in AI pairs, and +16.4~pp in human-AI pairs, and was concentrated on plays of cards that had received no hints (+46~pp in human pairs). Within human-AI play, the literal information available to humans was similar across the three AI partners (mean predicted failure 38–41%), but human failure rates ranged from 14.4% to 34.4% and the gap from +24.1 to +6.2~pp; the partner eliciting the largest gap produced the fewest human failures. Game score carried different information: it depended on each corpus’s roster composition, whereas the gap separated human from AI play at the agent level. As a known-answer check, Off-Belief Learning agents, whose convention content is controlled by construction, gave a gap of +1.6~pp at the convention-free level, rising monotonically to +21.7~pp. These results suggest that convention compatibility, rather than AI-AI performance, may predict an AI’s effectiveness with human partners.
[HC-7] CARLAverse: A Highly Modular Distributed and Multimodal Framework for Human-in-the-Loop Simulation
链接: https://arxiv.org/abs/2609.11478
作者: Patrick Rebling,Philipp Nenninger,Reiner Kriesten
类目: Robotics (cs.RO); Human-Computer Interaction (cs.HC)
备注: 19 pages, 13 figures, Code available at this https URL
Abstract:The development of autonomous driving demands comprehensive testing in mixed-traffic scenarios involving vulnerable road users (VRUs), where purely artificial agents often fail to capture authentic human social negotiations. While human-in-the-loop (HITL) simulators enable safe investigation of these interactions, existing multi-agent platforms struggle with the network latency and synchronization constraints required for high-fidelity haptic feedback. To resolve this, we present CARLAverse, an open-source, multimodal simulation ecosystem. Extending modular hardware abstraction, CARLAverse integrates driving (DrivoCARLA), cycling (CycloCARLA), and pedestrian (WalkoCARLA) simulators into a shared virtual environment. Its core methodological contribution is a distributed physics architecture: latency-critical ego dynamics and high-frequency force feedback are computed locally on client nodes, while a central CARLA server orchestrates non-player character (NPC) physics and global traffic. By decoupling haptic control loops from network bottlenecks, CARLAverse enables scalable, cross-institutional HITL experiments without compromising physical immersion. Code and documentation: this https URL
[HC-8] Are Caption Metrics Broken? Latency Deaf and Hard of Hearing User Ratings and Bias across Technologies
链接: https://arxiv.org/abs/2609.11408
作者: Bernard Thompson,James Waller,Luz Fanny Calderon Torres,Lu Ming,Mariana Arroyo Chavez,Dante Conway,Raja Kushalnagar,Christian Vogler
类目: Human-Computer Interaction (cs.HC)
备注: Accepted for publication at ACM ASSETS 2026
Abstract:Live captions on TV often contain errors and timing issues, making it hard for deaf and hard-of-hearing (DHH) viewers to follow dialog. It is essential that caption quality metrics reflect the lived DHH TV viewing experience. To this end, we describe a U.S.-based large-scale online survey with 216 validated participants, who provided 302 responses containing a cumulative 4,832 data points. Participants viewed videos drawn from a pool of 70 clips recorded from live TV, and were asked to rate the caption quality and subjective understanding of the content across four conditions: TV captions as originally recorded with up to 7-12 seconds delay, TV captions synchronized with audio, Automatic Speech Recognition (ASR)-generated captions synchronized with audio, and ASR captions with an average two-second delay. All captions were evaluated against the Word Error Rate (WER), Automated Caption Evaluation (ACE2) and Number, Edition and Recognition (NER) metrics. Results show that TV and ASR captions were rated similarly. For TV captions, all three metrics were moderately-to-highly correlated with viewer ratings, but far less so for ASR captions, making them far from technology-neutral. Additionally, caption latencies significantly impact the viewer experience, especially typical 7-12-second TV delays. We discuss the implications for the adoption of caption quality metrics.
[HC-9] A Voice-Interactive Multi-Agent System for Smart Operating Rooms: Architecture Design and Key Technologies
链接: https://arxiv.org/abs/2609.11231
作者: Tianxiang Zhou
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Human-Computer Interaction (cs.HC)
备注:
Abstract:This paper presents SurgicalRoomAgent, a voice-interactive multi-agent system for smart operating rooms based on large language models (LLMs). The system achieves natural language understanding, device control, intraoperative recording, and surgical report generation through a layered architecture comprising a voice interaction pipeline (wake, ASR, turn detection, agent reasoning, TTS) and an agent core (skill registry, task planner, device manager). Three key technologies are investigated: (1) KV Cache prefix warming for low-latency inference, reducing recomputation overhead from approximately 500 ms to tens of milliseconds via byte-level Longest Common Prefix reuse; (2) streaming partial JSON parsing with early parallel task execution, reducing end-to-end latency by approximately 30%; and (3) progressive skill prompt disclosure, which dynamically filters system prompts based on user role, connected devices, and surgical phase to maximize information density within limited context windows. The system is implemented using the Qwen3-27B model with this http URL inference engines. Experimental analysis demonstrates effective operation within a 16,384-token context limit and multi-device parallel control response times meeting OR real-time requirements.
[HC-10] AI Soccer Analyst: Stage-Aware and Verifiable Human-AI Collaboration for Soccer Data Analysis
链接: https://arxiv.org/abs/2609.11224
作者: Calvin Yeung,Keisuke Fujii
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注:
Abstract:Sports data analysts translate domain questions into insights by combining computation with sport-specific domain expertise. Large language models ease programming, but prompt-to-report workflows may obscure decisions and evidence. We present AI Soccer Analyst, a mixed-initiative system with revisable stages: Data Understanding, Problem Definition, Structured Planning, Execution, Evidence-Grounded Reporting, and Interaction and Refinement. A formative study with five analysts first informed design goals for automation, verifiability, human control, and accessibility. Subsequently, a task-based evaluation with 16 participants combined system logs, retained artifacts, ratings, and open responses; 33 of 48 tasks met the operational completion criteria. Exploratory tests supported favorable participant perceptions of completed-task output quality, task achievement, reliability, and verifiability after Holm correction. Interaction records showed domain knowledge emerging through clarification, planning, and refinement. These findings position stage-aware human-AI collaboration as a practical approach for producing inspectable, revisable, and verifiable analyses while retaining domain-expert involvement in consequential decisions.
[HC-11] (Whose defaults?) Is artificial intelligence reorienting archaeological methods?
链接: https://arxiv.org/abs/2609.11198
作者: Lorenzo Cardarelli,Roberto Ragno
类目: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Human-Computer Interaction (cs.HC)
备注:
Abstract:Generative AI and the practice of “vibe coding” are changing how archaeologists carry out computational research, but their effects on the discipline’s range of methods is still understudied. In this paper, we evaluate whether large language models (LLMs) are narrowing the variety of methods archaeologists use. We first analysed approximately 119,000 archaeology abstracts from Scopus, covering publications from 2010 to 2025. Using a locally run LLM, we identified the computational methods reported in each abstract and organised them into 25 broad categories (L2) and 241 finer clusters (L3). A Bayesian Dirichlet-multinomial model of method composition within sub-disciplines found a small but credible shift in method use after 2023. However, this shift was smaller than the variation already present across the full study period. No individual technique showed a significant change, and overall methodological diversity increased rather than declined. We then ran a controlled experiment to see whether LLMs recommend a narrower set of methods than archaeologists have used in practice. Two different open-weight models were asked to suggest methods for 28 archaeological research problems, with prompts providing three levels of methodological guidance: novice, intermediate, and expert. Recommendation diversity was much lower than in the published literature, particularly without methodological guidance. The models also tended to favour methods that were widely used before 2023, and their recommendations more closely resembled the post-2023 literature. Taken together, these results are consistent with LLMs pushing methodological choice towards convergence, although our study cannot establish a causal effect. They raise a broader question: how can archaeology retain methodological diversity as LLMs become more involved in research?
[HC-12] Conceptualising an Initial Design Space for Guidance in Digital Physical Activity Support
链接: https://arxiv.org/abs/2609.11193
作者: Faith Young,Markus Tatzgern,Alexander Meschtscherjakov,Jan Smeddinck
类目: Human-Computer Interaction (cs.HC)
备注: Accepted for publication at NordiCHI 2026
Abstract:Providing guidance is frequently referenced as a key capability of digital health interventions targeting physical activity, yet the term remains poorly defined and inconsistently applied. Existing work often conflates guidance with related constructs such as personalisation, feedback, or persuasion, limiting both theoretical clarity and design progress. This paper conceptualises an initial design space of guidance in the context of digital physical activity support. We define guidance for physical activity as situated, action-oriented support that scaffolds users’ embodied engagement in physical activity. Drawing on literature from behaviour change, human-computer interaction, embodied cognition, and digital health, we outline a design space that characterises guidance along multiple dimensions: scope, purpose, timing, context, modality, embodiment, adaptivity, autonomy, and affective quality. By offering a structured vocabulary and conceptual foundation, this work aims to support more coherent research, comparisons, and responsible design of digital health interventions featuring guidance for physical activity support.
[HC-13] How AI Coders Discuss Disagree and Reach Consensus: Challenges and Opportunities for LLM -Based Qualitative Coding
链接: https://arxiv.org/abs/2609.11109
作者: Jeongyeon Kim,John Mitchell
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注: 34 pages, 7 tables
Abstract:The utility of AI in multi-coder qualitative coding has been widely discussed, yet little empirical evidence exists to delineate the contexts in which it performs reliably. We address this gap by quantifying the effectiveness of multi-agent LLM coding across varied qualitative datasets, revealing key contextual and structural factors that mediate coding outcomes. We developed a literature-informed baseline pipeline that enables AI agents to independently code, debate, and reconcile disagreements. Results revealed that coding accuracy depends on factors such as codebook length, qualitative data similarity, and agent disagreement. Notably, intense and unresolved debates between agents led to higher accuracy. Our analysis showed that while LLMs emulate many human discussion behaviors, they lack adaptive responsiveness to context. From these findings, we offer design recommendations for building automated coding systems. Our open-source AI discussion dataset and methodological framework lay the groundwork for advancing the design of AI-mediated automated thematic analysis.
[HC-14] Visual-Motion-Induced Modulation of Pedestrian Trajectories Using Spatially Distributed Multi-Display Signage in Public Spaces
链接: https://arxiv.org/abs/2609.11088
作者: Yuri Mikawa,Taiki Fukiage,Yuki Kubota,Takumi Yokosaka,Maki Ogawa,Kazushi Maruya
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Multi-display signage (MDS), now ubiquitous in urban environments, has the potential to influence human behavior and experience in public spaces. However, despite its unique capability to present spatially distributed dynamic visual stimuli, its current use is mainly limited to advertising. In this study, we propose a perception-based approach for laterally modulating pedestrian trajectories as a nonverbal means of guiding pedestrians in public spaces. The approach is motivated by vection, the illusion of self-motion, and uses laterally moving monochrome stripes, a standard stimulus in vection research, presented across spatially distributed displays to elicit postural responses that may bias pedestrian trajectories. We evaluated the approach through a controlled laboratory experiment and a real-world field deployment involving actual pedestrian flows in a national museum. The laboratory experiment examined whether the MDS setup induced trajectory shifts in the direction predicted by prior research on the behavioral effects of vection. The field deployment investigated whether comparable effects would emerge in aggregate pedestrian behavior during unconstrained movement under conditions closer to those of urban public spaces. In the laboratory, full-screen motion significantly biased walking trajectories in the direction of visual motion, whereas partial-stripe motion produced no significant directional effect. In the field deployment, opposing full-screen motion conditions produced direction-consistent differences in aggregate pedestrian positions. The field results, observed despite the substantial variability in real-world pedestrian flows, extend the controlled laboratory findings and provide ecologically valid evidence supporting practical MDS-based pedestrian modulation in public settings. The results further suggest that sufficient visual-motion coverage may be important.
[HC-15] X-Hinges: 3D Printing Self-Sensing Compliant Mechanisms for Continuous and Multi-DOF Motion Sensing
链接: https://arxiv.org/abs/2609.11077
作者: Xiang Chang,Haiyang Yan,Stefanie Mueller,Jiaji Li
类目: Human-Computer Interaction (cs.HC)
备注: Accepted at ACM UIST 2026. 15 pages, 24 figures. Author’s accepted manuscript
Abstract:We present X-Hinges, a design and fabrication method for self-sensing compliant mechanisms based on multi-material FDM 3D printing. By co-printing two conductive filaments of different conductivities within a compliant body, we embed resistive sensing elements directly during fabrication without post-assembly, enabling continuous motion sensing across multiple degrees of freedom in a single print. The structure supports three degrees of freedom, each equipped with a dedicated sensing element configuration for multi-DOF motion estimation. We develop a precision data acquisition system and data-driven regression models that enable continuous, real-time motion sensing. We also introduce an interactive design tool for customizing the geometry, mechanical properties, degrees of freedom, and sensing configurations of X-Hinges. The tool also supports augmenting existing 3D models with self-sensing structures, endowing ordinary objects with continuous multi-DOF sensing capabilities. Finally, we present a set of application examples demonstrating the capability of X-Hinges for fabricating personalized interactive interfaces.
[HC-16] ShellVis: Sandboxed Live Programming for Shell Scripts
链接: https://arxiv.org/abs/2609.11000
作者: Joshua Horowitz,Jeffrey Heer
类目: Human-Computer Interaction (cs.HC); Programming Languages (cs.PL)
备注:
Abstract:Live programming provides visibility to programmers by running and tracing programs as they are edited. However, for programs with potentially harmful side effects, liveness can turn mistakes into disasters. We propose enabling live programming in environments with side effects via sandboxing: confining effects to a simulation of the true environment. We apply sandboxed live programming in the challenging context of shell scripting: a ubiquitous and powerful—yet notoriously opaque and error-prone—tool. ShellVis provides line-by-line feedback on a shell script’s run-time behavior, with file operations sandboxed via a safe overlay of the file system. A qualitative user evaluation finds ShellVis to be helpful to participants, replacing tedious existing practices and instilling confidence. Participant responses also reveal areas for future research, particularly bridging the gulf of execution alongside the gulf of evaluation. ShellVis serves as a case study of how sandboxing can bring live-programming techniques into the many real-world programming contexts where side effects are important.
[HC-17] “Coder first advocate second college student third”: The Liminality of Going to College as a Blind Computing Student
链接: https://arxiv.org/abs/2609.10942
作者: Isabela Figueira,Josahandi M. Cisneros,Stacy M. Branham
类目: Human-Computer Interaction (cs.HC)
备注: To be published in The 28th International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS 2026). 16 pages, 1 figure, 1 table
Abstract:Blind or low vision (BLV) students are less likely to graduate from college, particularly in computing. Prior work documents accessibility challenges in high school and college, but we lack understanding of the transition process that produces this “leaky pipeline.” To address this, we interviewed ten BLV college students about going to college to study computing. We analyzed our data through the lens of life transition, specifically Intersecting Liminality. Our findings reveal that some BLV students face such immense digital accessibility and college acclimation barriers that the only way forward as coders is to take on a “second job” as a blind advocate or drop out of the computing major. We argue that the college transition is a critical point for analysis and technological intervention, and further, that Intersecting Liminality provides a useful lens for HCI scholars to unpack the compounding challenges that prevent some BLV students from completing computing degrees.
[HC-18] Designing Technology for Social Wellbeing in Built Environments: A Conceptual Framework
链接: https://arxiv.org/abs/2609.10779
作者: Gul Sher Ali,Michail Giannakos,Monica Lillefjell,Sobah Abbas Petersen
类目: Computers and Society (cs.CY); Human-Computer Interaction (cs.HC)
备注: Preprint: Accepted to be published in ELSEVIER Book Series SUSTAINABLE DIGITAL MEDICINE: ISBN: 9780443458606
Abstract:Digital technologies are deployed in urban built environments with the aim of supporting social dimensions. However, the research offers no unified guidance for such digital technology design. Additionally, evidence shows that week social wellbeing contributes to mental and physical health outcomes which suggests that the technology designed to strengthen social wellbeing could also function as a form of health promoting and preventive intervention. This chapter addresses this research gap by developing a conceptual framework that supports the design of digital technologies for social wellbeing in built environments. We propose a conceptual framework composed of three core components which are drawn on the synthesis of selected empirical studies on technologies embedded in built environments for social wellbeing. First, a social wellbeing dimensions model that identifies what digital technology could address. Second, a digital technology contribution matrix that distinguishes the types of contributions a digital technology could make. Third, levels that maps the scope at which technology could support social wellbeing. This conceptual framework could help researchers, practitioners and policymakers to design and guide digital technology interventions that target social wellbeing in the built environment.
[HC-19] What Makes Creation Human? Authorship Reason s and Meaningful Human Control in Generative AI
链接: https://arxiv.org/abs/2609.10738
作者: Yuxi Cao
类目: Human-Computer Interaction (cs.HC); Computers and Society (cs.CY)
备注: 27 pages
Abstract:Generative artificial intelligence (GenAI) significantly expands creators’ productive capacity, but this does not necessarily entail a corresponding increase in creative agency or authorship. This paper distinguishes creativity at the level of the work from creative agency at the level of the creator, and argues that human authorship cannot be determined solely by manual intervention, degree of automation, the origin of an initial idea, or final selection authority. Rather, authorship depends on whether human judgment and reasons genuinely shape the development of the work. To articulate this requirement, the paper introduces Meaningful Human Control (MHC) into generative creation and identifies a limitation of its classical tracking condition. Creative reasons are not always fully specified prior to interaction with AI; they may emerge, change, or be abandoned as the creative process unfolds. The paper therefore proposes dynamic-reflexive tracking (DRT), which requires that a creator’s evolving reasons undergo reflective uptake, exert genuine influence on the subsequent trajectory of creation, and remain capable of rejecting and redirecting the system’s default direction. DRT consists of four conditions: diachronic reason formation, reflective uptake, trajectory efficacy, and contestability and redirection, together with a minimal tracing requirement. The paper argues that human authorship under generative AI depends not on how many steps a person personally performs, but on whether that person’s reasons continuously, reflectively, and effectively shape what the work becomes. Comments: 27 pages Subjects: Human-Computer Interaction (cs.HC); Computers and Society (cs.CY) Cite as: arXiv:2609.10738 [cs.HC] (or arXiv:2609.10738v1 [cs.HC] for this version) https://doi.org/10.48550/arXiv.2609.10738 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[HC-20] owards On-Device Evidence Gathering for Intimate Partner Infiltration: A Feasibility Study for Joint Identity-Action Detection
链接: https://arxiv.org/abs/2502.03682
作者: Weisi Yang,Shinan Liu,Feng Xiao,Nick Feamster,Stephen Xia
类目: Cryptography and Security (cs.CR); Computers and Society (cs.CY); Human-Computer Interaction (cs.HC)
备注: Accepted to ACM IMWUT 2026
Abstract:Intimate Partner Infiltration (IPI) refers to phone-side privacy infiltration in intimate or close relationships, often enabled by physical access to a person’s smartphone and discussed in technology-facilitated Intimate Partner Violence (IPV) contexts. Unlike conventional cyberattackers, IPI perpetrators leverage proximity and personal knowledge to circumvent standard protection, underscoring the need for targeted interventions, motivating device-side tools that surface such risk evidence for later review. While prior works have extensively studied IPV, and some have provided tailored and effective solutions such as security clinics, they are necessarily episodic and human-expert-intensive, and offer limited automated visibility into what happens on a smartphone between support sessions. Guided by a formative interview with experts (n=5), we take the first exploration into gathering IPI-risk evidence from a mobile system perspective and present AID, Automated IPI Detection, a data-driven system that continuously logs unauthorized access and suspicious behaviors on smartphones. In a controlled 27-participant study, AID achieves an end-to-end F1 score of 0.928 with a 7.0% false positive rate for Top-1 phone-side risk flagging; when preserving top-3 candidate action categories as report context, AID achieves an F1 score of 0.981 and a false positive rate of 1.6%. These findings demonstrate AID’s potential as an evidence-support tool that complements current clinic-based interpretation and safety-planning.
[HC-21] Pulling Illusion in Individuals with Neurological Disorders
链接: https://arxiv.org/abs/2609.10566
作者: Takeshi Tanabe,Satoshi Yamamoto,Toru Yamada,Daisuke Ishii,Yutaka Kohno
类目: Neurons and Cognition (q-bio.NC); Human-Computer Interaction (cs.HC)
备注:
Abstract:The pulling illusion induced by asymmetric vibration stimuli has attracted attention for its potential applications in rehabilitation and sensory assessment. However, the underlying mechanism of the pulling illusion remains unclear. This study addressed the central question of whether peripheral vibrotactile sensitivity alone is sufficient for the illusion to emerge or whether processing beyond basic vibration detection is also required. Neurological disorders can involve impairments at different levels of the nervous system, providing an opportunity to examine this question. Accordingly, we evaluated directional discrimination performance for the pulling illusion and fingertip vibration detection thresholds in 25 participants with diverse neurological disorders affecting different levels of the nervous system, from peripheral to central. Clustering analysis identified contrasting profiles, with high directional discrimination performance despite elevated vibration detection thresholds and chance-level performance despite relatively low-to-intermediate thresholds. In the generalized linear mixed model, motor-related signs, including hemiplegia and tremor, showed a robust negative association with directional discrimination performance, whereas vibration detection threshold was not robustly associated with performance. Furthermore, in participants with hemiplegia, directional discrimination performance was around chance level on the affected side and close to 100% on the unaffected side, despite stimulus amplitudes well above the measured vibration detection thresholds on both sides. Collectively, these findings suggest that the pulling illusion depends on perceptual processing beyond basic vibration detection, through which asymmetric vibration is experienced as directional pulling.
计算机视觉
[CV-0] SenseNova-U1.5: Towards Native Unified Visual Intelligence
链接: https://arxiv.org/abs/2609.11929
作者: Haiwen Diao,Jiahao Wang,Chenjing Ding,Hanming Deng,Jiangnan Chen,Ruixi Zhang,Ruohui Wang,Wenwen Tong,Xiangyu Fan,Yubo Wang,Yue Zhu,Yuwei Niu,Zhengqi Bai,Zhiqian Lin,Zhitao Yang,Zhongang Cai,Bo Yang,Chen Feng,Chengguang Lv,Guangjia Liu,Guanlin Wang,Hanyu Zhang,Haojia Yu,Hongcan Xiao,Hongli Wang,Huan Wu,Huaping Zhong,Jian Fang,Jianan Fan,Jiaqi Li,Jiefan Lu,Jing Zuo,Jingcheng Ni,Junxiang Xu,Linjun Dai,Mutian Xu,Peishen Yan,Penghao Wu,Ruijie Mao,Ruisi Wang,Shihao Bai,Shuang Yang,Shuya Yang,Shuyan Zheng,Silei Wu,Siying Li,Tao Chu,Tianbo Zhong,Tongxi Zhou,Weichao Luo,Weichen Fan,Wenhao Jia,Wenjie Gao,Xiangli Kong,Yan Li,Yang Yong,Zimo Wen,Zixuan Qian,Wenxiu Sun,Ruihao Gong,Quan Wang,Lewei Lu,Lei Yang,Ziwei Liu,Dahua Lin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL
Abstract:We launch SenseNova-U1.5, an 8B-MoT native unified multimodal model that understands, reasons about, and generates visual content within an encoder-free and VAE-free architecture. We strengthen its visual interface through spatially coherent patch reconstruction and scale its training with carefully curated generation and editing data, improved task formulation, structural prompt enhancement, and native resolutions of up to 4K. For post-training, we optimize specialized experts for visual aesthetics, bilingual text rendering, infographic generation, and image editing, and consolidate their capabilities through multi-expert on-policy distillation. Across extensive evaluations, SenseNova-U1.5 largely advances image fidelity, text rendering, complex composition, multi-reference editing, and interleaved generation, while improving instruction following and preserving subject identity, geometry, and unmodified regions. Despite limited exposure to structured formats in its generation data, SenseNova-U1.5 generalizes effectively to long, complex, and structured visual instructions, further proving that multimodal understanding can transfer to visual planning and creation. Together, these findings position native unified modelling as a promising path towards systems that perceive, reason and create within a fully end-to-end framework. We will open-source training code, including supervised fine-tuning, reinforcement learning, and on-policy distillation.
[CV-1] 3D Point Splatting for mmWave Radar Novel View Synthesis
链接: https://arxiv.org/abs/2609.11894
作者: Adnan Armouti,Yixuan Gao,Rajalakshmi Nandakumar
类目: Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR); Machine Learning (cs.LG); Signal Processing (eess.SP)
备注: Under Review
Abstract:Solving novel view synthesis (NVS) for millimeter-wave (mmWave) radar requires a renderer that is physically faithful, complex-valued, and multi-viewpoint-tractable. No prior method achieves these three properties simultaneously. Differentiable Monte Carlo (MC) ray tracers implement the radar forward model directly with explicit material modeling and complex outputs, but do not scale to the multi-view optimization NVS demands. Optical-NVS ports of NeRF, hash grids, and 3D Gaussians train fast but discard phase and replace explicit material modeling with opaque learned features, restricting them to power-only range-azimuth (RA) magnitudes. We propose 3D Point Splatting (3DPS), the first differentiable point renderer for radar, derived directly from the standard solid-angle form of the radar equation. Each oriented 3D point carries an ITU-R P.2040 material model, evaluated in closed form, with the resulting complex phasor splatted into range bins through a precomputed point spread function (PSF). The complex-valued output makes the renderer product-agnostic. The same optimized scene yields analog-to-digital converter (ADC), complex range profile (CRP), and RA outputs through standard fast Fourier transform (FFT) pipelines without retraining for each format. On six outdoor ColoRadar scenes, 3DPS reaches 0.587 mean Pearson correlation on held-out RA images. This is between 1.7x and 5.2x the three optical-NVS baselines (RadarSplat, Radar Fields, DART). Training takes approximately 3 minutes per scene on a single RTX 4090.
[CV-2] Guided Super-Resolution of Digital Elevation Models with Diffusion-Based Image Generators
链接: https://arxiv.org/abs/2609.11886
作者: Armand Mihai Nicolicioiu,Dominik Narnhofer,Nando Metzger,Daniel Panangian,Ksenia Bittner,Konrad Schindler
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:High-resolution digital surface models (DSMs) play an important role in urban analysis, 3D building reconstruction, and infrastructure monitoring, yet their availability remains limited due to the high cost and complexity of data acquisition. In contrast, coarse DSMs from commercial satellite missions are widely accessible, and high-resolution optical imagery is increasingly available from aerial and satellite platforms. We address the resulting mismatch in spatial resolution and propose a DSM superresolution approach that enhances 5 m DSMs to 0.5 m resolution, using guidance from high-resolution spectral images. Our method employs denoising diffusion to transfer information that is visible only in the image, like crisp outlines and detailed roof structures, into the elevation maps. In this way, surface details are reconstructed more accurately than with conventional interpolation or filtering techniques. Experiments on several cities in Central Europe demonstrate that the proposed approach produces high-quality DSMs with improved structural detail and accurate surface geometry. Our results highlight the potential of guided super-resolution with foundational image priors as a means of reconstructing high-resolution surface models.
[CV-3] CoRA-NAS: Coarse Ranking and Anchor-Residual Refinement for Neural Architecture Search
链接: https://arxiv.org/abs/2609.11884
作者: Yifan Yang,Zhaoyan Wang,Zheng Gao,Xiaoyu Li,Jiaojiao Jiang
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Zero-cost proxies rank architectures cheaply, but their reliability varies across search spaces. We introduce CoRA-NAS (COarse Ranking + Anchor-residual), a two-stage framework combining a static ranking prior with low-cost learning-curve refinement. CoRA-Rank aggregates capacity and structure-at-initialization proxies through an equal-weight log-rank consensus and a target-free consensus gate. CoRA-Refine samples anchors across this prior, extrapolates their early validation curves, and propagates a learned residual correction with an ExtraTrees model. The refinement uses approximately 1% of the cost of fully training the candidate set. Fully trained architecture-accuracy labels are not used to fit the ranker. One configuration is used across spaces, with space-specific architecture encodings. Across NAS-Bench-201, NAS-Bench-101, TransNAS-Bench-101, and NATS-SSS, CoRA-Refine achieves mean Spearman correlations of 0.946, 0.715, 0.786, and 0.894, respectively. Its worst-space correlation of 0.715 is the highest among the compared methods. On NAS-Bench-201/CIFAR-100, its selected architecture reaches 73.32% accuracy, near the reported ground-truth best of 73.37%. On the pure size space, refinement recovers the static prior’s shortfall relative to parameter count, while remaining tied with the strongest capacity proxies within noise. The resulting framework combines cross-space ranking robustness with low-cost architecture selection.
[CV-4] Logit Refiner: Improving Visual Autoregressive Models via Intra-Scale Dependency Modeling ECCV2026
链接: https://arxiv.org/abs/2609.11804
作者: Meimingwei Li,Stefan Andreas Baumann,Felix Krause,Björn Ommer
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: ECCV 2026
Abstract:Visual Autoregressive Models (VAR) generate images through next-scale prediction, producing all tokens within each scale in parallel. We show that this parallel decoding constitutes a mean-field-style approximation that discards spatial dependencies among same-scale tokens, causing locally incoherent samples regardless of backbone capacity – a limitation of the decoding rule. Addressing this limitation, we introduce the Logit Refiner, a lightweight autoregressive module that restores intra-scale dependencies by sequentially sampling tokens conditioned on frozen backbone features. Adding only ~10% parameters and less than 5% of the base model’s training compute, it plugs into any pretrained VAR checkpoint without retraining. Controlled ablations isolate joint intra-scale sampling – rather than additional capacity or training – as the critical ingredient. Across backbones from 310M to 2B parameters on class-conditional ImageNet 256x256, the refiner consistently improves generation quality, enabling a 1.1B-parameter model to surpass one twice its size. The approach further generalizes to text-to-image generation, confirming that the mean-field bottleneck persists across VAR variants and is effectively alleviated by our method. Project page: this https URL Comments: ECCV 2026 Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG) Cite as: arXiv:2609.11804 [cs.CV] (or arXiv:2609.11804v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2609.11804 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-5] Revisiting Avatar-As-Image: High-Fidelity Registration is All You Need
链接: https://arxiv.org/abs/2609.11722
作者: Margaret Kostyrko,Yuxuan Xue,Garvita Tiwari,Gerard Pons-Moll
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project Page: this https URL
Abstract:The representation of 3D clothed humans as standardized 2D UV texture and displacement maps over an underlying body model has long been studied. This compact representation is enticing as it enables pretrained image networks to process, generate, and edit 3D avatars, but is only useful if scans are accurately aligned and brought into correspondence via high-fidelity registration. This prerequisite has never been met, which we argue explains the limited quality of prior UV-based methods for clothed humans. Despite its significance, no public method produces high-fidelity SMPL(-X)+D registrations with UV texture from arbitrary clothed scans. We present AvaImg, a multi-stage optimization pipeline, to close this gap: it enforces body-inside-clothing constraint via signed winding numbers, made viable by a three-level efficiency cascade (~10x runtime reduced, ~95% storage saved), and recovers fine surface detail using coarse-to-fine displacement optimization. AvaImg outperforms all baselines in body fitting, shape estimation, and surface registration across six datasets, yielding textured registrations near-indistinguishable from scans (PSNR=34.48dB). For validation of AvaImg’s Avatar-as-Image representation as imminently compatible with image foundation models, we auto-encode our UV maps via the frozen FLUX VAE. This achieves only 0.76mm added Chamfer error relative to scan and shows that the resulting maps lie within natural-image distributions, supporting the use of 2D generative priors for 3D avatar generation. Code, data, and Singularity containers will be at this https URL.
[CV-6] MC-DeTra: Motion-Consistent Joint Object Detection and Socially-Aware Trajectory Forecasting in Birds-Eye-View Images
链接: https://arxiv.org/abs/2609.11717
作者: Vladislav Diuzhev,Dmitry Yudin
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: 16 pages, 4 figures. Code: this https URL
Abstract:Unified models for object detection and trajectory forecasting aim to merge perception and prediction for autonomous driving, refining actor trajectories directly over shared bird’s-eye-view (BEV) images rasterized from LiDAR and high-definition maps. Their accuracy on dynamic, moving actors, however, remains the hardest part of the task, and the strongest such model, DeTra, has no public implementation. We contribute an openly released DeTra reimplementation with documented approximations, and on top of it MC-DeTra: a family of motion-consistency mechanisms that add supervision through two annotation-derived auxiliary signals – each actor’s observed past motion and the occupancy of the surrounding traffic that forms its social context – and one inter-output consistency constraint that aligns an actor’s predicted heading with its predicted direction of motion. Every proposed loss is train-only and inference-safe: it shapes the shared BEV representation during training and is removed at test time, adding no inference latency. On the Waymo Open Dataset, evaluated under a strict, detection-conditioned forecasting protocol, MC-DeTra improves dynamic, socially-situated trajectory forecasting while preserving or improving detection accuracy; a gradient-based loss-calibration analysis exposes how the auxiliary objectives compete at the shared backbone, and our ablation identifies which signals contribute most. We release code, configurations, and evaluation tooling at this https URL.
[CV-7] Language-Augmented Semantic Priors for B-Spline Surface Fitting
链接: https://arxiv.org/abs/2609.11708
作者: Yunzhong Lou,Yusheng Luo,Jiahao Li,Yu Song,Xiangdong Zhou
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:The use of B-splines and Non-Uniform Rational B-Splines surfaces constitutes the mathematical foundation of contemporary computer-aided design (CAD) systems. Despite long-term progress, geometric kernels in traditional CAD still rely heavily on predetermined heuristic initialization for surface fitting and parameterization. Meanwhile, the procedural semantics and design intent encoded in modeling histories are largely ignored during geometry generation. This disconnect creates a gap between high-level design intent and solver-executable geometric configuration, often leading to suboptimal and semantically inconsistent fitting results. To bridge this gap, we introduce LASP, a Language-Augmented Semantic Priors framework that leverages large language models (LLMs) to infer structured, solver-usable B-spline priors from procedural modeling histories. Rather than modifying the geometric kernel itself, LASP operates as a semantic reasoning layer above existing solvers. It first translates modeling histories into rich textual descriptions that capture design intent, geometric context, and functional relationships, and then uses a fine-tuned LLM to predict structured B-spline prior parameters. LASP is trained through a two-stage scheme that combines local geometric regularities with long-range contextual dependencies, producing priors that are both interpretable and semantically coherent. This approach furnishes inductive signals that direct the conventional B-spline fitting process toward solutions that more accurately encapsulate the intended design objectives and demonstrate heightened semantic coherence. Compared to traditional machine learning schemes, the experiments demonstrate that language-driven reasoning can serve as a powerful inductive bias for geometric solving, establishing a new paradigm of language-guided geometric optimization in modern CAD systems.
[CV-8] Spectral Adapters for Segment Anything Model-based Segmentation of Colorectal Liver Metastases in Computed Tomography
链接: https://arxiv.org/abs/2609.11703
作者: Ramtin Mojtahedi,Mohammad Hamghalam,Jacob J. Peoples,Natalie Gangai,Mithat Gonen,Yun Shin Chun,HyunSeon Christine Kang,Richard K. G. Do,Amber L. Simpson
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 13 pages, 1 figure, 4 tables
Abstract:Accurate segmentation of colorectal liver metastases (CRLM) in contrast-enhanced computed tomography (CT) is important for response assessment, surgical planning, and follow-up. We propose two parameter-efficient spectral adapters for the Segment Anything Model (SAM): the Directional Spectral Adapter (DiSECT) and Spectral Instance-Guided Adapter (SiGA). DiSECT uses singular value decomposition of frozen weights to constrain residual updates to leading spectral directions, while SiGA adds global and input-conditioned gating through a multilayer perceptron. We evaluate these methods on 446 contrast-enhanced CT volumes (355 training, 91 testing) and compare them with LoRA, QLoRA, convolutional adapters (CAD), and a 3D nnU-Net baseline. Experiments consider single-point, three-point, bounding-box, and no-prompt regimes. SiGA achieves the best single-point performance with a Dice score of 0.77, IoU of 0.69, and HD95 of 35.39 mm. Under no-prompt inference, SiGA reaches 0.76 Dice, 0.68 IoU, and 46.76 mm HD95, comparable to the nnU-Net baseline (0.758 Dice). DiSECT uses only 0.14 million trainable parameters. These results show that spectral adapters can efficiently adapt SAM for CRLM segmentation while retaining strong accuracy with limited trainable parameters.
[CV-9] Single-Stream Multi-Feature Fusion with Temporal Robustness for Gait Emotion Recognition ICANN2026
链接: https://arxiv.org/abs/2609.11680
作者: Shirong Lyu,Silu Quan,Yixuan Ding,Chengpeng Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at the 35th International Conference on Artificial Neural Networks (ICANN 2026)
Abstract:3D skeleton-based gait emotion recognition faces high annotation costs, data scarcity, and poor generalization on heterogeneous data. This paper proposes SV-GCN, a single-stream multi-feature fusion framework with temporal invariance. We introduce intra-frame relative motion features to eliminate frame-rate sensitivity and embed heterogeneous cues at shallow layers, enabling early fusion without multi-stream complexity. For variable-length sequences, we design a global mask-guided valid-frame spatio-temporal graph convolution module, introducing frame-rate insensitivity for the first time in this domain. On the E-Gait dataset, our method achieves performance comparable to state-of-the-art while demonstrating strong generalization across varying sequence lengths and frame rates, offering a viable pathway for pre-training on large-scale skeleton-based action recognition datasets.
[CV-10] Multimodal Taxonomic Conditioning for Generative Plankton Imagery ECCV
链接: https://arxiv.org/abs/2609.11673
作者: Daniela Ivanova,Ozgu Goksu,Nicolas Pugeault
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: European Conference on Computer Vision (ECCV) 2nd Workshop on Marine Vision
Abstract:Automated plankton imaging produces severely long-tailed datasets, where the rare taxa of greatest ecological interest have too few images to train or evaluate classifiers reliably. We generate synthetic plankton imagery conditioned on taxonomy: a CLIP encoder is adapted on a large plankton corpus with a ranked contrastive objective extended to deep, ragged taxonomies, then frozen to condition a parameter-efficient diffusion transformer. We evaluate synthetic sample quality on distributional fidelity and downstream classifier utility.
[CV-11] Self-Supervised Cardiac Phase Detection via Single-Parameter Latent Orbits MICCAI2026
链接: https://arxiv.org/abs/2609.11650
作者: John Bonnici,Matthew Baugh,Aleksandra Kulbaka,Sarah Cechnicka,Bernhard Kainz,Alberto Gomez
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted for oral presentation at the ASMUS workshop at MICCAI 2026
Abstract:Accurate identification of end-diastole (ED) and end-systole (ES) in echocardiography underpins the quantification of ventricular function, yet manual selection of these key frames is subjective and introduces clinically significant inter-operator variability. Recent self-supervised methods either prescribe strict periodic trajectories or learn an unconstrained low-dimensional motion subspace from reconstruction or registration objectives. The former offers interpretability but imposes restrictive assumptions on temporal progression, whereas the latter leaves cardiac phase implicit and ED/ES must be recovered through post-hoc geometric processing of the learned trajectory. We translate the physiological observation that cardiac phase is a one-dimensional signal into a prior by constraining the latent motion component to a single-parameter latent orbit, i.e., a global linear trajectory in latent space indexed by a bounded scalar phase variable. Mapping this variable through a sinusoidal nonlinearity yields an oscillatory motion signal with consistent temporal ordering, enabling direct identification of ED and ES from the learned phase signal. This inductive bias allows the model to capture an interpretable representation of the cardiac cycle, while maintaining flexibility to capture irregular heartbeats. Trained on EchoNet-Dynamic without annotations, our minimal single-parameter cardiac phase model learns an effective latent orbit, significantly improves upon the previous state of the art in ED localisation and matches it in ES localisation while using a more constrained representation and fewer training epochs. This demonstrates that a principled physiological inductive bias can match or exceed the performance of more complex representations. Code is available at: this https URL
[CV-12] Vidu S2: Real-Time Interactive Editable and Spatial Video Generation
链接: https://arxiv.org/abs/2609.11638
作者: Jintao Zhang,Kai Jiang,Jintao Chen,Xu Wang,Deyuan Liu,Jungang Li,Dechuang Chen,Ming Lin,Jingjiang Zhou,Haopeng Jin,Qi Jia,Xiaohang Wang,Yaole Wang,Zhanqiang Zhang,Ran Li,Zhengkun Huang,Shuyue Xiong,Yuji Wang,Zikun Dai,Hui He,Yang Luo,Mang Ning,Weiqi Feng,Chengyang Ye,Xinyue Lin,Min Zhao,Hongzhou Zhu,Hengkai Tan,Zeyuan Wang,Chendong Xiang,Kaiwen Zheng,Zhijie Deng,Fan Bao,Jianfei Chen,Jun Zhu
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:We present Vidu S2, which comprises Vidu S2-Avatar, a real-time interactive digital-character model, and Vidu S2-Editing, a real-time video editing model. Moreover, we explore the feasibility of real-time spatial video generation for both Vidu S2-Avatar and Vidu S2-Editing. Compared with Vidu S1, Vidu S2-Avatar supports real-time 720p video generation, generation with dynamic references that can be updated at any moment, and stronger instruction following, such as dancing. Vidu S2-Editing supports editing a video stream in real time, including style rendering, clothing replacement, character replacement, and background replacement. Experiments show that Vidu S2 outperforms all baselines. A playable online demo is available at this https URL.
[CV-13] LangStreet: Persistent Language Fields for Anchor-Decoded Street Gaussians
链接: https://arxiv.org/abs/2609.11616
作者: Runyi Yang,Deheng Zhang,Xiaoye Wang,Mengjiao Ma,Lei Sun,Kanzhi Wu,Ajad Chhatkuli,Luc Van Gool,Danda Pani Paudel
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Language Gaussian fields implicitly assume that the primitive carrying semantics remains identifiable across views. This assumption breaks in scalable anchor-decoded representations, where persistent anchors generate view-conditioned child Gaussians whose geometry and appearance vary with the camera. We introduce Ours, a persistent language field for such structured Gaussian scenes. Our key idea is semantic ownership: transient children route observations, while persistent decoder slots and their parent anchors own the language field. We use alpha-compositing responsibilities to accumulate additive directional evidence at slots; these statistics marginalize exactly to anchors. We then complete weakly supported slots with anchor-aligned evidence while preserving the anchor direction, and represent slot detail through low-rank residuals in anchor-relative semantic coordinates. Our primary model, Ours (base), stores anchor features together with compact slot residuals. Ours (light) retains only anchor features, whereas Ours (max) stores the full-dimensional completed slot features explicitly. Without scene-specific semantic optimization, Ours (base) nearly matches Ours (max) across KITTI, Virtual KITTI, and Waymo. On KITTI, it achieves 34.19 2D mIoU with a 2.72 GiB effective feature footprint, compared with 34.20 mIoU and 12.90 GiB for Ours (max). The same accuracy-storage trend holds on Virtual KITTI and Waymo. These results show that language fields on view-conditioned splats require persistent semantic ownership, conserved evidence, and a hierarchy that balances stability, detail, and representation cost. Our code, checkpoints, and benchmark suite will be publicly available.
[CV-14] MMGait: Benchmarking and Unifying Gait Recognition across Heterogeneous Modalities
链接: https://arxiv.org/abs/2609.11601
作者: Saihui Hou,Chenye Wang,Qingyuan Cai,Aoqi Li,Yongzhen Huang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 21 pages, 6 figures
Abstract:Gait recognition is commonly studied using RGB videos or their derived silhouettes and poses. Yet human walking produces heterogeneous photometric, geometric, and motion cues that cannot be systematically examined with RGB-centered benchmarks. We present MMGait, a large-scale multi-sensor benchmark that brings visible, infrared, depth, LiDAR, and radar observations into sequence-level correspondence. It provides diverse modalities spanning appearance, contours, geometry, motion, and body structure. Under a shared impostor-augmented protocol, we evaluate single-modal recognition, cross-modal recognition via directed retrieval, and multi-modal recognition using task-specific experts. Across settings, modality rankings vary with probe conditions, cross-modal alignment remains difficult, and fusion often provides complementary gains. This analysis exposes a scalability problem: individual modalities, modality pairs, and fusion configurations are typically handled by separately trained experts. We formulate Omni-Modal Gait Recognition, which unifies single-modal, cross-modal, and multi-modal recognition within a shared identity space. OmniGait++ uses modality-specific front ends followed by a shared identity encoder to preserve modality-dependent cues while learning comparable identity descriptors. An anchor-guided fusion module aggregates modality subsets of varying size without frame-level synchronization. A jointly trained checkpoint covers all three recognition settings and accommodates modality subsets of different compositions and cardinalities. Experiments show OmniGait++ remains competitive with task-specific experts in many shared settings and extends to higher-cardinality fusion unavailable to fixed-pair models. The results establish MMGait as a common testbed for heterogeneous gait sensing and demonstrate the feasibility of unified recognition under varying modality availability.
[CV-15] OmniKVQuant: KV Cache Quantization for Omni-LLM s
链接: https://arxiv.org/abs/2609.11582
作者: Suho Yoo,Hyunjong Ok,Jongmin Choi,Jihoo Jung,Joon Son Chung
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Preprint
Abstract:As Omni-modal large language models (Omni-LLMs) take in audio, video and text together, their KV cache memory cost grows. KV cache quantization is the de facto approach in text-only LLMs, but its application to Omni-LLMs remains unexplored. In this paper, we analyze how TurboQuant, a representative rotation-based KV cache quantization method, behaves on multimodal caches and identify two critical issues: temporal key drift and heterogeneous value geometry. To address these, we propose OmniKVQuant, a training-free framework that (i) sets the key quantization range over each short window of the input stream; and (ii) rotates values separately per modality. On Qwen2.5-Omni and Qwen3-Omni, OmniKVQuant enables 2-bit KV caches while substantially preserving performance across seven audio-visual benchmarks. We further provide a fused Triton decode kernel that unpacks the 2-bit cache during attention, so no dense FP16 cache is ever built. Code: this https URL
[CV-16] Learn the Solid Not the File: Canonical Inputs for Neural Networks on CAD Boundary Representations
链接: https://arxiv.org/abs/2609.11573
作者: Heinrich Jiang,Hager Yasser Mohamed,Alexander Hitt,Valeriia Lomakina,Henning Jiang,Jennifer Jang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computational Geometry (cs.CG)
备注:
Abstract:Boundary representation (B-rep) is the standard format used by modern CAD systems for parametric 3D models. It turns out, the exact same solid can be represented by different B-reps: for example, two engineers using different operations, a geometry kernel rebuilding the file, and an export setting repartitioning faces will lead to different B-reps even though the underlying solid remains the same. We show that existing B-rep encoders are not robust to variation in the B-rep with the same solid on perturbations applied to standard benchmarks, naturally occurring variations inherent to CAD software, and differences in how designers model the same part via a human dataset we created in FreeCAD. The performance of popular B-rep encoders often collapses catastrophically. We propose the canonical region graph, an input representation whose nodes, features and coordinate frame are derived from the solid itself and show theoretical invariance guarantees on repartitioning and rigid motions. It matches the strongest baseline on standard benchmarks, and is stable under every perturbation we test. Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Computational Geometry (cs.CG) Cite as: arXiv:2609.11573 [cs.CV] (or arXiv:2609.11573v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2609.11573 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-17] A Comparative Evaluation of Pre-trained Convolutional Neural Networks for Melanoma Detection
链接: https://arxiv.org/abs/2609.11550
作者: Wagner Moreno Schmitz,Marco Antonio de Castro Barbosa,Thiago Magalhães Amaral,Dalcimar Casanova,Jefferson Tales Oliva
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Early diagnosis of melanoma is critical for improving patient survival rates. However, accurately distinguishing melanoma from other skin lesions remains a significant clinical challenge due to the high visual similarity among lesion types and variability in image acquisition conditions. Artificial intelligence, particularly machine learning, has emerged as a promising tool to support dermatological diagnosis by automating feature extraction from medical images. Among the available approaches, convolutional neural networks (CNNs) have demonstrated strong performance in image classification tasks, making them well-suited for analyzing both dermatoscopic and histopathological images, given their ability to capture hierarchical visual patterns relevant to lesion characterization. Nevertheless, despite numerous pre-trained CNN architectures having been proposed, selecting the most appropriate one for a given imaging modality remains an open challenge. In this study, we evaluate pre-trained convolutional neural networks (CNNs) for skin lesion classification using dermatoscopic and histopathological image datasets. Experiments were conducted on the HAM10000, ISIC 2018, and CR-AI4SkIN datasets, evaluating the ResNet50, VGG16, VGG19, MobileNet, and InceptionV3 architectures under the same training protocol. The experimental evaluation showed that the models achieved accuracies ranging from 71% (InceptionV3 on ISIC 2018) to 84% (ResNet50 on HAM10000) on dermatoscopic images. For histopathological images, accuracies ranged from 72% (VGG19) to 83% (ResNet50) on the CR-AI4SkIN dataset. The results demonstrate that model performance differs between dermatoscopic and histopathological image modalities, showing that architectures exhibiting similar performance on dermatoscopic images exhibit different performance on histopathological data.
[CV-18] World in World: Explore the World with World Models
链接: https://arxiv.org/abs/2609.11548
作者: Chenxi Song,Yanming Yang,Chi Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project Page: this https URL
Abstract:Autoregressive video world models enable interactive, long-horizon exploration, but flexible control remains challenging. Exploring a source video from new viewpoints requires the generated rollout to remain synchronised with the recorded event, place observed content in the requested view, plausibly complete newly exposed regions, and recover previously generated appearance on revisits. Existing methods typically address these requirements through task-specific modules or additional training. We present World in World, a training-free inference-time interface that converts heterogeneous control evidence into camera- and time-labelled clean visual states, which are read through the native self attention of a frozen causal video model. The evidence comprises source-video observations, target-view scene projections, geometry renderings that guide completion of newly exposed subject regions, and retrieved generated states beyond the rolling cache. Each evidence source carries token-level support and its own availability schedule. A correspondence router combines persistent point identities with geometry to establish token correspondences, guiding supported queries towards matching source-video tokens. Evidence-wise attention CFG (EWA) then independently regulates each auxiliary channel’s additional contribution using attention responses from the same denoising forward pass. The shared interface supports camera-controlled rerendering, long-horizon revisiting, and human-motion transfer with the same frozen backbone. We evaluate World in World on camera-controlled video rerendering under diverse viewpoint changes, assessing perceptual quality, temporal consistency, and camera-following accuracy.
[CV-19] Learning Interaction between Image and Layout Priors for Joint Image-Layout Generation in Design Templates
链接: https://arxiv.org/abs/2609.11519
作者: Shirong Yang,Bo Yang,Ying Cao
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Graphics (cs.GR)
备注: Main paper with supplementary material. Submitted to IEEE Transactions on Visualization and Computer Graphics
Abstract:In this paper, we address the problem of graphic design template creation, which generates a background image and a layout of foreground elements over the background to form a harmonious composition from an input text. Prior work on graphic design generation mostly adopts a sequential paradigm, where design elements are generated sequentially. We argue that such a sequential scheme falls short of faithfully capturing the dependency between the background and layout (and thus the joint image-layout distribution), which limits the quality of generated design templates. To overcome this limitation, we propose a model, InterIL, which jointly generates the two modalities, background image and layout, in a single generative process. The novel design of our joint model connects the backbones of pretrained image and layout diffusion models with a learnable communication module to explicitly model bidirectional image-layout interaction. During training, the image and layout backbones are frozen to maintain and leverage the vast pretrained single-modality prior knowledge, while only the communication module is updated, so that the model can focus on learning image-layout interaction and thereby better capture the joint image-layout distribution for improved composition harmony. Our model has no design-specific inductive bias, which allows it to better preserve the original characteristics of realistic designs. We further introduce a test-time guidance strategy to enable users to impose their specific preferences on generated results. Our experiments show that, compared with prior approaches, our model can generate significantly better results in terms of image, layout and image-layout harmonization, producing outputs closer to real samples. We also demonstrate the flexibility of our model in enforcing user preferences at inference without retraining. Comments: Main paper with supplementary material. Submitted to IEEE Transactions on Visualization and Computer Graphics Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Graphics (cs.GR) Cite as: arXiv:2609.11519 [cs.CV] (or arXiv:2609.11519v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2609.11519 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Sloan Yang [view email] [v1] Thu, 10 Sep 2026 13:22:23 UTC (14,974 KB) Full-text links: Access Paper: View a PDF of the paper titled Learning Interaction between Image and Layout Priors for Joint Image-Layout Generation in Design Templates, by Shirong Yang and 2 other authorsView PDFHTML (experimental)TeX Source view license Current browse context: cs.CV prev | next new | recent | 2026-09 Change to browse by: cs cs.AI cs.GR References Citations NASA ADSGoogle Scholar Semantic Scholar export BibTeX citation Loading… BibTeX formatted citation loading… Data provided by: Bookmark checked="checked"class=“labs-tab-input”> Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv’s community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) mathjaxToggle(); We gratefully acknowledge support from our major funders, member institutions, , and all contributors. About Help Contact Subscribe Copyright Privacy Accessibility Operational Status (opens in new tab) Major funding support from
[CV-20] Breaking the Central Bias: Spatially Partitioned Experts for Coordinate-Based Neuroevolution ICPR2026
链接: https://arxiv.org/abs/2609.11518
作者: Romain Claret,Arthur Gygax,Michael O’Neill,Paul Cotofrei,Michael Palma Mendes,Pascal Felber
类目: Neural and Evolutionary Computing (cs.NE); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 15 pages, 4 figures, 1 table. Author’s accepted manuscript, accepted at the BIOMAP workshop (BIO-inspired Methods for Pattern Recognition) of ICPR 2026, Lyon, France
Abstract:Evolvable-Substrate HyperNEAT (ES-HyperNEAT), a bio-inspired indirect encoding that determines neuron placement and connection weights from spatial coordinates, exhibits a failure mode on MNIST as a diagnostic benchmark. Because input pixels map to a coordinate space centered at the origin, evolved networks converge on a small central cluster of input pixels, a spatial-concentration bias; prior work observed only 21% mean accuracy in this regime. Is this bias an optimization artifact or an architectural ceiling? Inspired by Mixture-of-Experts (MoE) principles, we partition the input into non-overlapping spatial segments, each assigned to a separately evolved specialist network. With 13 such experts, this design reaches 43% mean accuracy, a 106% relative improvement over the baseline. The architectural gain does not depend on data-driven aggregation: equal-weighted averaging, which uses no validation data, already yields a 70% improvement; the gain comes from partitioning, not the weighting. Receptive-field analysis shows the mechanism: partitioning forces evolution to discover features across the entire image, expanding active pixel coverage from 4% to 79%. Absolute accuracy stays below gradient-trained baselines, but the relative gain points to central bias, not the evolutionary search. Two tools are designed to generalize beyond MNIST: a receptive-field diagnostic for silent input-coverage collapse, and a spatial-partitioning remedy that restores coverage.
[CV-21] LoopVAE: Recurrent Depth Across Scales for Visual Tokenization
链接: https://arxiv.org/abs/2609.11516
作者: Zhiying Lu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 19 pages, 6 figures, 7 tables
Abstract:Hierarchical visual tokenizers typically allocate different processing blocks to different spatial scales. We ask how much of this computation can use the same parameters. LoopVAE reuses a scale- and loop-conditioned core within and across scales, while keeping resolution-changing transitions independent. A four-block core executes 28 block applications per encoder or decoder. On ImageNet-256, the 29M-parameter convolutional model reaches 0.28 rFID and 32.54 dB PSNR under an approximately 30-epoch two-stage training budget, using approximately 65% fewer parameters than the 84M reference VAEs. A non-adversarial Transformer ablation with the same execution graph finds competitive PSNR and SSIM under global sharing, although unshared blocks improve LPIPS. Targeted loop interventions show that completing the trained recurrence improves reconstruction and that even small feature updates can have substantial downstream effects. Truncation also exposes output-range errors, distinguishing useful recurrent computation from reliable early exit. Runtime profiling reveals the execution tradeoff: fewer stored weights require more arithmetic and longer runtime in the tested configurations. With convolutional and Transformer operators and single- or multi-resolution latent interfaces, LoopVAE establishes recurrent depth across scales as a parameter-sharing design axis for visual tokenization.
[CV-22] Prototype Matters: Modality-unified Prototype Self-distillation for Unsupervised Visible-infrared Person Re-identification
链接: https://arxiv.org/abs/2609.11514
作者: Menglin Wang,Xiaojin Gong
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: ACM Multimedia 2026
Abstract:Estimating reliable cross-modality association is crucial to unsupervised visible-infrared person re-ID. While optimal transport is shown to be a practical solution for cross-modality association, it suffers from the rigidness of hard label assignment without considering the impact of cluster noise. Moreover, enforcing only cross-modality contrast is also suboptimal, as it fails to jointly optimize the similarity relation within and across modality. In this paper, we propose a novel framework for cross-modality learning by well exploitation of prototypes: First, instead of contrasting with cross-modality prototypes, we show that modality-unified prototypical contrast facilitates better modality invariance by jointly and simultaneously optimizing similarity relation within and across-modality. Taking self-prototype as a steady teacher, we further refine the instance-prototype online relation through prototype-guided self-distillation. The two components are optimized in a unified framework, leading to a simple yet effective model. On standard VI-ReID benchmarks, we perform extensive comparison and analysis, validating the effectiveness of our proposed method. Code is available at: this https URL.
[CV-23] Harnessing Intrinsic Subject-Aware Attention for Controllable Multi-Subject Video Generation
链接: https://arxiv.org/abs/2609.11507
作者: Niange Yu,Ye Tian,Biaolong Chen,Miao Lu,Aixi Zhang,Hao Jiang,Yunhai Tong,Pipei Huang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 23 pages, 11 figures, 4 tables
Abstract:Multi-subject video generation faces two key challenges: uncontrollable fidelity strength and potential semantic drift. We address these by analyzing the internal mechanisms of Diffusion Transformers (DiTs). We found that certain attention blocks naturally form an Intrinsic Spatial Grounding Map (ISGM) that precisely locates reference subjects. Building on this insight, we propose Dual-phase Intrinsic Attention Leveraging (DIAL), a framework that uses these internal signals for both training and inference. In low-noise stages, we use ISGM to guide the attention mechanism, allowing precise control over fidelity strength during inference without retraining. In high-noise stages, we use these same maps to automatically build preference pairs at no additional cost for Reinforcement Learning (RL). This RL procedure effectively anchors the model’s attention to reference subjects and mitigates semantic drift. Extensive experiments show that DIAL significantly outperforms baseline models on the OpenS2V-Eval benchmark, consistently improving identity consistency and enabling controllable fidelity strength.
[CV-24] UBone3D: Physics-Rectified Conditional Flow Matching for Anatomical 3D Shape Completion from Ultrasound ECCV2026
链接: https://arxiv.org/abs/2609.11506
作者: Weiying Chen,Yuchong Gao,Siyuan Li,Marek Reformat,Rui Zheng,Edmond Lou
类目: Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV)
备注: Accepted to ECCV 2026. Camera-ready Author Version
Abstract:Three-dimensional ultrasound (US) is a safe, radiation-free complementary modality to CT and X-rays for longitudinal monitoring, yet its segmentation-derived partial point clouds are extremely artifact-laden. Consequently, it is challenging to recover a clean and complete anatomical structure from such US point clouds. In this paper, we present UBone3D, a novel framework based on physics-rectified conditional flow matching (CFM) that performs point cloud completion directly from partial US observations. UBone3D models deterministic physics artifacts (e.g., surface thickening, streaking, dropouts) via a simulated physics proxy, and introduces test-time physics rectification to steer the shape completion. At inference, the completion is jointly steered by two decoupled forces: (1) anatomical plausibility enforced by a CT-trained generative shape prior, BoneFM, and (2) physics consistency enforced by USimNet in the ultrasound formation space. Extensive experiments on simulated and in-vivo data demonstrate significant improvements in reconstruction accuracy and anatomical fidelity over existing baselines.
[CV-25] Recursive Code World Models: Building Complex Worlds through Recursive Scene Programs
链接: https://arxiv.org/abs/2609.11499
作者: Zhiqi Li,Yuxuan Liao,Bo Zhu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 21 pages, 11 figures
Abstract:Code world models represent worlds as executable programs, but this representation alone does not determine how to construct a complex world. We introduce Recursive Code World Models (RCWM), a framework for reconstructing complex 3D worlds in code from a single reference image. RCWM couples a Recursive Scene Program (RSP) representation with a construction solver that recursively calls itself. An RSP represents the executable world as compositional scene code, while each solver call follows the same complete process: establish the whole, recursively reconstruct unresolved parts, and revisit the whole to refine their composition. This global-local-global recursion gives fine-scale structures their own perception-and-editing loops while preserving scene-wide geometry and relationships. Reference-aligned views propagate a shared camera projection across levels, while parent revisitation addresses boundaries, spatial relations, and shared errors that emerge after local refinement. A vision-language coding agent directly compares reference images with scene renders to guide refinement, recursive descent, and return. Across complex scenes, RCWM outperforms prior code-based image-to-scene reconstruction methods. Ablation studies further support the benefits of recursive construction and suggest that deeper calls can improve finer-scale reconstruction. RCWM provides a recursive construction principle for building complex executable worlds from visual evidence.
[CV-26] FreeFlow: A Bias-free Hierarchical Transformer for Optical Flow Estimation ECCV2026
链接: https://arxiv.org/abs/2609.11486
作者: Vladislav Bargatin,Alexander Yakovenko,Khaled Abud,Dmitriy Vatolin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at ECCV 2026. Project page: this https URL
Abstract:Optical flow methods typically rely on task-specific inductive biases, such as correlation volumes, feature warping, and iterative refinement, among others, to reach high accuracy. While effective, such biases constrain the model to predefined heuristics, which can limit its expressivity and lead to more complex pipelines and additional computational cost. We present FreeFlow, a hierarchical transformer built without any flow-specific components, using instead a single feed-forward encoder–decoder. FreeFlow combines three attention variants: window attention for local processing, shifted-window attention for cross-window information exchange, and a global attention operating at a reduced resolution. The resulting architecture scales naturally with model capacity, enabling a consistent accuracy gain from small to large variants. Despite the absence of standard inductive biases, FreeFlow achieves state-of-the-art results on major benchmarks, including Sintel (0.68/1.48 EPE on Clean/Final), KITTI-2015 (3.23 Fl-all), and Spring (3.192 1px), while remaining memory efficient at 1080p inference.
[CV-27] Pre- and Post-Treatment Brain Metastases Segmentation Using nnU-Net with Post-Processing for BraTS 2026 MICCAI2026
链接: https://arxiv.org/abs/2609.11477
作者: Haobin Liu,Xin Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to MICCAI 2026 Challenge BraTS-METS
Abstract:Brain metastases exhibit high inter-lesion variability in size, enhancement pattern, and post-treatment appearance, making volumetric segmentation of both pre- and post-treatment cases the central challenge of the BraTS 2026 Task 1 (Brain Metastases). We build a pragmatic pipeline on a 5-fold nnU-Net ResEnc-L ensemble, in which each fold is trained independently for 1,000 epochs with the standard Dice + cross-entropy loss on 1,296 four-modality training cases. This ensemble is followed by a rule-based post-processing cascade tuned for the lesion-wise Dice similarity coefficient (LW-DSC), a detection-oriented metric that behaves very differently from the traditional global Dice. The final pipeline reaches an LW-DSC of 0.733 / 0.751 / 0.713 / 0.549 on the enhancing tumour (ET), tumour core (TC), whole tumour (WT), and resection cavity (RC) sub-regions on the official validation leaderboard. Rather than trusting these leaderboard gains, we audit every post-processing stage with a five-fold out-of-fold (OOF) analysis with no model-training leakage over all 1,296 training cases, scored with the official BraTS evaluation code (BraTS_evaluation): it confirms two stages as robust, per-fold-consistent improvements while the third improves only the leaderboard and does not reproduce out-of-fold. We further provide a mechanistic analysis of the LW-DSC metric that explains why recall-recovering post-processing carries low risk whereas component deletion does not, and we report thirteen negative results spanning loss engineering, alternative backbones, and inference-time settings, several of which run counter to widely held intuitions. Source code is released under Apache-2.0 at this https URL.
[CV-28] BridgeMatch: Conditional Transport Bridges in Matching Matrix Space for 3D Deformable Registration
链接: https://arxiv.org/abs/2609.11472
作者: Qianliang Wu,Haobo Jiang,Guangwei Gao,Shuo Chen,Jin Xie,Jian Yang,Yaqing Ding
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Reliable non-rigid point cloud correspondences are important for deformable anatomical registration, embodied perception and manipulation, and dynamic 3D reconstruction. Coarse-to-fine methods reduce computational cost by selecting the top-(K) coarse regions. However, this pruning may remove weak but correct hypotheses and restrict fine matching to an incomplete search space. We present \paper, a two-stage generative solver that maintains the complete soft matching matrix at both coarse and high resolutions. Stage~I uses denoising diffusion to estimate a global matching matrix in the compact coarse-resolution space. We then lift this matrix to high resolution while preserving its hierarchy. The lifted matrix is rank-bounded and block-constant. Stage~II refines it through a conditional transport bridge. We implement the bridge with two types of dynamics: a deterministic endpoint-parameterized conditional Flow Matching (CFM) ODE and a stochastic Brownian-bridge SDE inspired by Schrödinger bridges. Both variants share the lifted source, a time-conditioned transformer, and a matching-matrix endpoint predictor. Experiments on 4DMatch and 4DLoMatch show that both variants produce more accurate correspondences than the compared methods and improve downstream registration, with larger gains in low-overlap cases. They also improve cross-dataset generalization on CAPE and DeepDeform without target-domain adaptation while using the same deformation solver.
[CV-29] BruNet: A Cross-Domain Transfer Framework for Bruise Segmentation
链接: https://arxiv.org/abs/2609.11463
作者: Qiming Wang,Richard J. Motley,Ebube E. Obi,Xianfang Sun,Paul L. Rosin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Segmenting bruises is a challenging task in medical imaging due to limited data and annotations, diffuse boundaries, and highly variable appearance. In this work, we propose BruNet, a segmentation framework that combines a ViT-based visual encoder (a self-supervised DINOv3 or a pretrained LingBot-Vision backbone) with a SAM-based mask decoder. BruNet is trained on the HAM10000 skin lesion dataset and evaluated on a separate bruise dataset without additional fine-tuning. Although a small number of prior studies have explored machine learning and computer vision for bruise analysis, existing work has primarily focused on detection, classification, or colour analysis rather than pixel-level localisation. To the best of our knowledge, this is the first study to address automatic bruise segmentation. Our results show that BruNet outperforms CNN-based models, state-of-the-art segmentation models, ChatGPT-4o/5-assisted SAM2 zero-shot baselines, and the medical-oriented MedSAM model, demonstrating strong cross-domain generalisation to bruise segmentation.
[CV-30] Multi-Modal Controlled Coherent Motion Generation ECCV2026
链接: https://arxiv.org/abs/2609.11439
作者: Yifei Liu,Qiong Cao,Hongwei Yi,Huaiguang Jiang,Changxing Ding
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: ECCV 2026
Abstract:It is natural for humans to walk and talk simultaneously. This paper tackles the challenge of replicating such natural behaviors in 3D avatar motion generation driven by concurrent multimodal inputs, such as a text description of a man walking alongside speech audio. Existing methods, constrained by the scarcity of aligned multimodal data, typically combine motions from individual modalities sequentially or through weighted sums. However, they often result in mismatched or unrealistic movements. To overcome these limitations, we propose MOCO, a novel diffusion-based framework capable of processing multiple simultaneous inputs, including speech audio, text descriptions, and trajectory data, to generate coherent and lifelike motions without requiring aligned multimodal data. Our key innovation lies in decoupling the motion generation process. During each denoising step, the diffusion model independently generates motions for each modality from the input noise and assembles the body parts according to predefined spatial rules. The resulting combined motion is then diffused and serves as the input noise for the subsequent denoising step. This iterative approach enables each modality to refine its contribution within the context of the overall motion, progressively harmonizing movements across modalities. Consequently, the generated motions become increasingly natural and fluid with each iteration, achieving coherent and synchronized behaviors. We evaluate our approach using a purpose-built multimodal benchmark. Experimental results demonstrate that MOCO outperforms existing baselines, advancing the field of multimodal motion generation for 3D avatars.
[CV-31] Hologram Representation via Quadratic Phase Gaussian Splatting SIGGRAPH
链接: https://arxiv.org/abs/2609.11434
作者: Haolong Wang,Yicheng Zhan,Kaan Akşit,Simeng Qiu
类目: Graphics (cs.GR); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: SIGGRAPH Asia 2026 Technical Communications
Abstract:We introduce Complex-Valued Quadratic Phase Gaussian (CVQPG), a novel hologram representation method that replaces standard 2D Gaussian representations used in 2D Gaussian Splatting with 2D quadratic phase functions. CVQPG incorporates additional learnable parameters to control the curvature of these bases. We evaluate our approach against state-of-the-art methods, exceeding the visual quality by +0.19 dB (RGB) and +0.33 dB (grayscale) on average in holographic reconstructions. Specifically, our equal parameter count evaluations show that modulating the primitive’s wavefront is an effective and lightweight enhancement for hologram representations. In addition, our frequency domain analysis illustrates that CVQPG has successfully preserved the mid-to-high frequency band of natural images.
[CV-32] DINO-Med: A Unified Patch-Based Adaptation Framework for Multi-Modal Medical Image Analysis Applied to Liver Fibrosis Staging
链接: https://arxiv.org/abs/2609.11380
作者: Boya Wang,Ruizhe Li,Chao Chen,Xin Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 12 pages, 2 figures. Accepted by AIiH
Abstract:Adapting natural-image foundation models like DINOv3 to multi-modal medical imaging is challenging due to the significant domain gap between natural color images and multi-channel medical scans. We present a unified, patch-based framework that processes raw multimodal imaging through training-free registration, automated localization, and mask-filtered patch extraction. This architecture culminates in a hierarchical strategy that aggregates patch-level insights into subject-level diagnostics. Using liver fibrosis staging as a case study, we evaluate four patch-level feature representations: handcrafted Radiomics features, learned ResNet features, pre-trained foundation model SAM-Med2D features, and frozen DINOv3 features. To ensure a controlled comparison, all models utilize the same lightweight MLP head and are evaluated across both rigid and deformable registration settings. Our training protocol focuses on mild fibrosis (S1) and cirrhosis (S4) classes only, enabling a single classifier to address both substantial fibrosis detection and cirrhosis staging. Evaluated via 10 random train (90%)/ test (10%) splits on 360 subjects from the CARE 2025 Liver Track 4 cohort, our DINOv3-based framework significantly outperforms all baselines, achieving the best classification accuracy of 78.4% for S1 and 75.8% for S4.
[CV-33] Brain-PACE: A Deep Siamese MRI Framework for Modelling Longitudinal Brain Acceleration
链接: https://arxiv.org/abs/2609.11378
作者: Samuel Maddox(1),Jacob Newman(1),Saber Sami(2),Michal Mackiewicz(1), for theAlzheimer’s Disease Neuroimaging Initiative, theAustralian Imaging Biomarkers,Lifestyle flagship study of ageing((1) School of Computing Sciences, University of East Anglia, (2) Norwich Medical School, University of East Anglia)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 20 pages, 6 figures
Abstract:Brain age estimation has become a popular research proxy for assessing brain health and disease, yet longitudinal trajectories of brain ageing are still poorly defined, and clinical use is limited. Building on existing Siamese longitudinal frameworks, we develop Brain-Predicted Age Acceleration (Brain-PACE) to directly estimate the pace of structural brain ageing from paired T1-weighted MRI. Brain-PACE identified accelerated ageing in 42.6 % of participants with mild cognitive impairment. Faster Brain-PACE was associated with greater functional and cognitive impairment (FAQ; r=0.35 , ADAS13; r=0.30 , CDR-SB; r=0.32 ) and greater regional tau burden in the posterior cingulate ( r=0.59 ), precuneus ( r=0.47 ), and entorhinal cortex ( r=0.37 ). These associations were stronger than those observed when pace was calculated indirectly from repeated cross-sectional brain age estimates, suggesting that direct longitudinal modelling captures complementary information relevant to ongoing pathological change. Methodologically, Brain-PACE extends the LILAC framework by combining spatial attention with soft label distribution learning and a Cramér distance objective, improving probabilistic performance and reducing prediction bias while providing measures of predictive uncertainty. Together, these findings support Brain-PACE as a complementary longitudinal imaging phenotype with sensitivity to relevant clinical and biological changes in early neurodegeneration.
[CV-34] Vision Transformer-Based Multi-Level Feature Fusion for Multi-Label Sewer Defect Classification
链接: https://arxiv.org/abs/2609.11375
作者: Xu Fang,Zhuoran Wang,Qing Li,Shengyu Zhang,Guanzhi Deng,Jianbiao He,Qingquan Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Automated classification of sewer defects is essential for infrastructure condition assessment and maintenance decision-making, but existing deep learning methods struggle to balance classification accuracy and computational complexity in large-scale multi-label scenarios. This study develops Sewer-Transformer-ML, a hierarchical vision Transformer with multi-level feature fusion, together with two lightweight architectures, Sewer-MobileNet-ML and Sewer-Mobile-TransNet, for resource-constrained inspection scenarios. On the Sewer-ML test set, Sewer-Transformer-ML-Base achieved an F2_\textCIW of 65.68% and an F1_\textNormal of 92.68%, ranking first on the public leaderboard and exceeding the second-ranked method by 7.6 percentage points in F2_\textCIW . Sewer-MobileNet-ML achieved an F2_\textCIW of 65.73% with only 17 M parameters, representing an approximately 95% parameter reduction relative to the base model. Under the standard Sewer-Capsule data split, Sewer-Mobile-TransNet achieved 96.43% classification accuracy. When the training set was reduced to 1,177 images, pretraining on Sewer-ML consistently improved model performance. Ablation experiments further showed that direct concatenation was more effective for Transformer features, whereas attention-based fusion better supported multiscale CNN features. These findings provide a computational basis for automated sewer inspection, lightweight model design, and adaptation across civil infrastructure inspection platforms.
[CV-35] R4Tun: LLM -guided adaptive segmental tunnel lining segmentation in point clouds
链接: https://arxiv.org/abs/2609.11360
作者: Xinghui Tao,Zehao Ye,Guangming Wang,Jelena Ninić,Brian Sheil
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Automated inspection of segmental tunnel linings requires adaptive segmentation from 3D point clouds, yet expert-tuned pipelines often degrade when tunnel conditions vary. This paper presents R4Tun, a large language model (LLM)-driven adaptation framework that extends an expert-designed pipeline (SAM4Tun) with bounded parameter tuning informed by structured context: memory ( m ), state ( s ), and knowledge ( k ). Evaluated on 30 selected Seg2Tunnel subsets (13 regular, 17 complex) across three LLMs, the full m+s+k design raised mean Intersection-over-Union (mIoU) from 0.18 to 0.43–0.48 and overall accuracy (OA) from 0.42 to 0.59–0.65 relative to the static SAM4Tun baseline, with the near-reference regular (staggered) subsets reaching mIoU 0.784–0.796 across LLMs. Across 270 (30 tunnels \times 3 different LLMs \times 3 context settings) runs, the LLMs showed similar parameter-adjustment trends (with overlapping 95% CIs on mean gains) and consistently adjusted a shared set of critical parameters. These results support R4Tun as a controlled, label-free, cross-LLM adaptation mechanism in the tested SAM4Tun–Seg2Tunnel setting, demonstrating consistent accuracy gains; we position R4Tun as a mechanism contribution rather than a deployable final-inspection system, in which each bounded parameter change is auditable via logged rationales.
[CV-36] Predictive Multi-Landmark OCT Tracking for Increased Motion Robustness
链接: https://arxiv.org/abs/2609.11330
作者: Konrad Reuter,Suresh Guttikonda,Chaitali Uday Karekar,Christian Betz,Alexander Schlaefer
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accecpted at CURAC conference 2026
Abstract:Optical coherence tomography is a promising modality for markerless motion tracking due to its high spatial resolution and inherent depth perception. However, existing OCT-based tracking approaches are limited in terms of trackable velocity, particularly when multiple landmarks are tracked sequentially for 6D pose estimation. In this work, we present a predictive tracking approach that propagates positional updates between multiple tracked landmarks to obtain a global pose prediction. This enables more robust tracking under high velocities. Our results demonstrate RMSEs below 1 mm for velocities up to 100 mm/s and up to nine consecutively tracked landmarks, highlighting the potential of global motion propagation and prediction for improving the robustness of OCT-based tracking.
[CV-37] Mi-Ripple: Restoring Images Degraded by Iterative AI Editing
链接: https://arxiv.org/abs/2609.11317
作者: Jiayin Chen,Yicheng Xu,Muting Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Iterative reference-conditioned image editing can introduce grid-like and granular textures, commonly described as digital ripple. We present Mi-Ripple, a diagnosis-guided restoration workflow that suppresses this digital ripple while protecting image structure. Mi-Ripple separates periodic lattice artifacts from content-entangled granular texture, then combines selective spectral notching, structure-aware smoothing, and cleaned-reference regeneration. This separation enables low-distortion filtering when artifacts are spectrally isolated and visual reconstruction when filtering would erase legitimate detail. Across fourteen notch-only executions, whole-image residual standard deviation is 0.08–0.44 in CIELAB lightness units. In a paired regeneration example, reference cleaning reduces output debris density by 45%. Mi-Ripple links measurable artifact reduction to visibly cleaner generated images, rather than optimizing a spectral score alone.
[CV-38] GRIPNet: Gaussian Radial Intensity Prior Guided Architecture for Pulmonary Nodule Detection in CT
链接: https://arxiv.org/abs/2609.11312
作者: Haojie Yang,Ran Su
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Lung cancer causes more deaths than any other malignancy, and low-dose CT screening is the main pathway to early diagnosis. That pathway hinges on the smallest lesions, yet nodules below six millimeters remain hard to detect, because most methods treat a nodule as a generic object and ignore the imaging physics behind its appearance. We show that this appearance is highly regular. Intensity peaks at the geometric center of a nodule and decays radially in a Gaussian pattern, and a fit to 18,218 annotated lesions from three public benchmarks yields a mean radial coefficient of determination above 0.86 in every dataset and size stratum. A square convolution samples both axes uniformly and is mismatched to this radial signal, most severely for small nodules. Guided by this evidence, we propose GRIPNet (Gaussian Radial Intensity Prior Network), a detector in which every module maps to a measurable property of the intensity distribution. Pinwheel convolutions decompose radial gradients, a dual-frequency module separates boundary detail from structural context, dilated masked attention matches the decay extent, and an adaptive loss reweights samples by conspicuity. GRIPNet raises mAP@0.5 to 95.3, 91.6 and 97.9 percent on KanserSet, LUNA16 and Lung-PET-CT-Dx while sharpening high-IoU localization at real-time speed.
[CV-39] Your Model Already Knows Dont Teach It Learn to Ask It: Soft Prompting for Few-Shot Adaptation of Vision-Language Models
链接: https://arxiv.org/abs/2609.11310
作者: Gautam Rajendrakumar Gare,Siyi Li,Hewei Wang,Cesar Daniel Hernandez,Wei Zhao,Wolfgang M. Pauli,John Galeotti,Deva Ramanan
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Image and Video Processing (eess.IV); Machine Learning (stat.ML)
备注:
Abstract:We address few-shot object detection with vision-language models (VLMs) in out-of-domain settings such as aerial, industrial, and medical imagery, using only ten annotated images for supervision. Existing adaptation methods are discrete prompt optimization and LoRA fine-tuning. We revisit a third option: soft prompting, where a small number of continuous prompt tokens are optimized while the pretrained backbone remains frozen. We identify two key design choices. First, placing prompt tokens at the cross-modal boundary between visual and text tokens outperforms other placements (10.0 vs. 8.4 mAP). Second, initializing prompts from the empty space token outperforms semantic and random initialization. With these choices, one to three learned tokens (7,168 parameters on average) match the best LoRA configuration on Roboflow20-VL (14.2 mAP, 10-shot) while training over 20,000x fewer parameters. Soft prompting remains harder to optimize, exhibiting higher variance across random seeds. Unlike LoRA, however, it causes no forgetting: the LoRA rank matching our accuracy reduces NaturalBench VQA accuracy by 35% relative, rising to 56% at the largest rank, whereas soft prompting leaves pretrained performance unchanged. The learned tokens behave like prompts rather than weights. They transfer to a newer model without retraining (+0.8 mAP on Qwen3.5-9B) and can be verbalized into readable prompts competitive with prompt-search methods (matching DetPO and outperforming GEPA). The approach also extends beyond detection. On RoboCasa manipulation tasks, the frozen \pi_0.5 vision-language-action policy benefits from soft prompting, matching the LoRA baseline on two of three tasks when tokens are placed at the gradient bottleneck. These results suggest modern VLMs already encode much of what is needed for specialized domains; the challenge is learning how to ask. Subjects: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Image and Video Processing (eess.IV); Machine Learning (stat.ML) Cite as: arXiv:2609.11310 [cs.CV] (or arXiv:2609.11310v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2609.11310 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Gautam Gare [view email] [v1] Thu, 10 Sep 2026 09:38:11 UTC (12,033 KB) Full-text links: Access Paper: View a PDF of the paper titled Your Model Already Knows Don’t Teach It, Learn to Ask It: Soft Prompting for Few-Shot Adaptation of Vision-Language Models, by Gautam Rajendrakumar Gare and 7 other authorsView PDFHTML (experimental)TeX Source view license Current browse context: cs.CV prev | next new | recent | 2026-09 Change to browse by: cs cs.AI cs.LG eess eess.IV stat stat.ML References Citations NASA ADSGoogle Scholar Semantic Scholar export BibTeX citation Loading… BibTeX formatted citation loading… Data provided by: Bookmark checked="checked"class=“labs-tab-input”> Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv’s community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) mathjaxToggle(); We gratefully acknowledge support from our major funders, member institutions, , and all contributors. About Help Contact Subscribe Copyright Privacy Accessibility Operational Status (opens in new tab) Major funding support from
[CV-40] SAMV-DUSt3R: Instance-Centric 3D Scene Decoupling from Sparse Multi-Views
链接: https://arxiv.org/abs/2609.11279
作者: Langxu Zhao,Zuan Gu,Yingdan Zhang,Pengfei Zhao,Tianhan Gao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:With the rising demand to decouple objects from 3D scenes, we propose SAMV-DUSt3R, an end-to-end model that injects SAM2 2D masks into MV-DUSt3R reconstruction. A Cross Flow Mask Block uses these masks to steer the network toward the target instance, jointly improving shape accuracy and achieving object-level disentanglement without multi-stage pipelines. To ensure reconstruction stability, a lightweight Spatial RankGNN selects the optimal reference view with a selection accuracy of 73.5%. Extensive experiments demonstrate that our method boosts average reconstruction precision by 11% across various metrics compared to state-of-the-art baselines. These results reveal a strong instance-disentanglement capability and clear benefits for driving, robotics, AR/VR, and heritage digitisation.
[CV-41] Order-Aware 2.5D Multiple Instance Learning for Preoperative MRI-Based Perineural Invasion Risk Assessment in Intrahepatic Cholangiocarcinoma
链接: https://arxiv.org/abs/2609.11271
作者: Hyunsu Go,Youngung Han,Kyeonghun Kim,Jinyong Jun,Junbeom Lee,Dohyun Kweon,Yului Jeong,Suah Park,Sungha Park,Anna Jung,Woo Kyoung Jeong,Ken Ying-Kai Liao,Hyuk-Jae Lee,Nam-Joon Kim
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Perineural invasion (PNI) is an adverse histopathologic marker in intrahepatic cholangiocarcinoma (ICC), but it is usually confirmed only after resection. Preoperative T2-weighted MRI may provide noninvasive imaging cues predictive of PNI, although labels are available only at the patient level without slice- or voxel-level annotations. We propose Order-Aware Slab Multiple Instance Learning (OAS-MIL), a weakly supervised framework for patient-level PNI prediction. Each tumor-centered MRI crop is represented as an ordered sequence of overlapping 2.5D slabs formed from contiguous axial slices. A shared encoder extracts slab-level features, which are aggregated by a permutation-invariant set-attention branch and a bidirectional sequence-attention branch. Using five-fold label-stratified cross-validation at the patient level, OAS-MIL achieved a mean AUROC of 0.770, outperforming the evaluated volumetric and MIL baselines. These results suggest that axial order provides a useful inductive bias for weakly supervised PNI prediction from MRI.
[CV-42] Improving Faint Object Detection for Space Situational Awareness with Variational Autoencoders
链接: https://arxiv.org/abs/2609.11269
作者: Angela Cratere,Luca Ghilardi,Vishnu Reddy,Francesco Dell’Olio,Charalampos S. Kouzinopoulos,Roberto Furfaro
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted at SPAICE 2026: the 3rd European Space Agency Conference on AI in and for Space
Abstract:We present a deep-learning pipeline for enhancing the detection of faint moving objects in optical space situational awareness (SSA) imagery through automated star removal and background reconstruction. Detecting low signal-to-noise ratio (SNR) objects remains extremely challenging in optical observations, particularly in the cislunar (X-GEO) environment, where structured sky backgrounds, dense stellar fields, and scattered moonlight significantly degrade the performance of classical detection algorithms. To address this problem, the proposed pipeline combines a lightweight segmentation network (Tiny-U-Net) to generate stellar masks with a partial-convolution variational autoencoder (astro-VAE), designed to learn the statistical distribution of astronomical backgrounds and perform context-aware inpainting of masked regions. The reconstructed background maps can then be used as a preprocessing step to suppress fixed sources and background inhomogeneities prior to detection. As a proof of concept, the approach is integrated with a shift-and-stack scheme and evaluated on real ground-based telescope observations targeting the X-GEO region. Results demonstrate that the method reconstructs star-free backgrounds with high fidelity, while preserving moving targets and significantly enhancing detectability, thereby providing an effective data-driven preprocessing strategy for faint moving-object detection in optical SSA scenarios.
[CV-43] Uncertainty DMD: Restoring Diversity in Few-Step Autoregressive Video Distillation
链接: https://arxiv.org/abs/2609.11265
作者: Zixuan Duan,Xunzhi Xiang,Yabo Chen,Xin Zhang,Changhan Liu,Haibin Huang,Chi Zhang,Qi Fan,Xuelong Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL
Abstract:Few-step distillation improves the efficiency of autoregressive (AR) video generation, but often causes diversity collapse: under the same prompt, different noise samples tend to produce highly similar videos with weakened motion dynamics. We analyze this degradation in Distribution Matching Distillation (DMD)-distilled AR video generators and find that, in the autoregressive setting, it takes the form of a structured uncertainty collapse: the mode-seeking bias of DMD maps different noise samples to nearly identical first chunks, and the deterministic AR cache then propagates this collapsed state to all subsequent chunks, turning a local loss of stochasticity at the rollout root into a global suppression of temporal variation. Based on this analysis, we propose Uncertainty DMD, a simple uncertainty-injection framework that restores stochasticity at two key stages of AR generation: a timestep perturbation for the first chunk to increase first-chunk diversity, and a stochastic cache-writing mechanism for later chunks to preserve uncertainty in autoregressive conditioning. The method requires no architectural changes and introduces only lightweight perturbation operations. The same perturbation mechanisms are used during both training and inference. Experiments show that Uncertainty DMD consistently improves diversity and motion dynamics while maintaining comparable per-sample visual quality.
[CV-44] AI-Powered Flare Combustion Efficiency Estimation ICML
链接: https://arxiv.org/abs/2609.11262
作者: Afeefa Azam,Iyyakutti Iyappan Ganapathi,Fares Ossama Abdelhafez,Divya Velayudhan,Maregu Assefa Habtie,Hamad Karki,Khalid Yousef Al Awadhi,Naoufel Werghi
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at the 4th International Conference on Machine Learning and Data Engineering (ICMLDE 2025). 5 pages
Abstract:Achieving high combustion efficiency in flare stacks is crucial for adhering to regulatory standards and controlling the release of hydrocarbons into the environment. Traditional instruments like gas analyzers and hyperspectral cameras are expensive, fragile, and require frequent calibration, which makes them impractical for remote or budget constrained industrial sites. We propose an innovative solution that combines a lightweight vision-language encoder with a compact multi-layer perceptron to predict combustion efficiency directly from low-cost thermal video footage. The fully trained model is integrated into an easy-to-deploy graphical user interface. This interface overlays predicted combustion efficiency values on each video frame, displays real-time trends in combustion efficiency, shows the distribution of combustion efficiency across all frames in the video, and allows users to export CSV reports. Over a six-month period, the system achieved 99% uptime and required less than 15 minutes of maintenance per week.
[CV-45] From Evaluation to Enhancement: Benchmarking and Improving Think-with-Video Reasoning for Video Generative Models ECCV2026
链接: https://arxiv.org/abs/2609.11242
作者: Meng Luo,Yicheng Liu,Jiahao Wang,Yuanxing Zhang,Xin Tao,Pengfei Wan,Kun Gai,Hao Fei
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accepted to ECCV 2026. 46 pages, 41 figures
Abstract:Video generation has advanced to produce visually compelling and temporally coherent results. Yet, whether these models can genuinely think with video–executing symbolic rules, respecting physical laws, and pursuing intentional goals–remains an open question. Existing benchmarks only partially address this, often conflating visual quality with cognitive correctness. We introduce VWG-Bench (Video World Generalist Benchmark), a comprehensive benchmark spanning 9 reasoning dimensions and 38 fine-grained tasks. To enable precise diagnosis, we design a three-level VLM-as-Judge protocol that independently assesses video-level fluency, task-level rule adherence, and sample-level goal realization. Evaluations of leading models reveal a striking gap: while models achieve strong rendering scores, they consistently fail on logic-heavy and rule-constrained tasks. To address this, we propose Vid-PRE (Video Prompt Reasoner and Enhancer), a model-agnostic prompt rewriter that offloads the cognitive burden of reasoning to a dedicated VLM. Trained via reinforcement learning with purely text-based rewards, Vid-PRE produces concise, constraint-aware prompts without the instability of video-level reward signals. Experiments show that Vid-PRE yields substantial reasoning improvements across multiple generators without architectural modifications. Together, VWG-Bench and Vid-PRE offer a rigorous diagnostic lens and a scalable path toward true think-with-video capabilities. All data and code are publicly available at this https URL.
[CV-46] Fast and Accurate Monomodal 3D High Resolution Deep Registration of Drosophila Larval Brain Volumes
链接: https://arxiv.org/abs/2609.11240
作者: Daniel Reisenbüchler,Yousef Sadegheih,Michael Dittrich,Pratibha Kumari,Muhammad Usman,Dorit Merhof
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:The larval stage of Drosophila melanogaster is a compact model system for neuroscience whose genetic toolkit allows fluorescent markers to be expressed in defined neural populations, and comparing the resulting expression patterns across animals requires every brain to be registered into a shared anatomical reference space. Existing pipelines for this task are predominantly based on classical registration methods, which perform a new optimization for each volume, often require per-case parameter tuning, and can take minutes per brain, limiting their use as a routine preprocessing step. We present a trained deep registration pipeline that deformably aligns a larval brain to a reference template in a single forward pass at high spatial resolution, on volumes that hold several times more voxels than those learned 3D registration is normally reported on, together with the preprocessing and anatomy-anchored evaluation pipeline required to apply it. Against eleven classical and seven further learned baselines on a held-out collection acquired with different acquisition and quality strata, the proposed pipeline is the most accurate, improving on the strongest classical baseline by 23 percentage points of anatomical landmark-local mutual information. It registers a volume one to two orders of magnitude faster than the classical deformable pipelines, and it retains more of its accuracy than any other method as acquisition quality degrades. The network, its trained weights and the full pipeline are released as the open-source deep larval brain registration framework: this https URL
[CV-47] SCINTILLA-SNN: A Spiking Multi-Scale Selective Aggregation Network for Perineural Invasion Prediction
链接: https://arxiv.org/abs/2609.11237
作者: Youngung Han,Yului Jeong,Kyeonghun Kim,Dohyun Kweon,Suah Park,Hyunsu Go,Sungha Park,Anna Jung,Jinyong Jun,Yunho Choe,Yunjin Seo,Ken Ying-Kai Liao,Hyuk-Jae Lee,Nam-Joon Kim
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Preoperative prediction of perineural invasion (PNI) in cholangiocarcinoma (CCA) is clinically valuable but remains challenging because PNI-related cues on magnetic resonance imaging (MRI) are subtle, sparse, and spatially localized around the tumor boundary. Standard 3D CNN and transformer architectures process volumetric data in a dense or spatially uniform manner, which can dilute subtle PNI-related evidence while requiring a large number of multiply-accumulate operations over 3D feature grids. To address these limitations, we propose SCINTILLA-SNN, a 3D spiking network composed of a four-stage hierarchical backbone and a Multi-Scale Spike Aggregation (MSSA) module for PNI prediction. The backbone extracts hierarchical volumetric representations through spiking convolutional stages and local spike window modulation stages. Given the resulting stage-wise representations, MSSA maps each spatial token to a learnable content value and modulates it with a spike-dynamics gate derived from firing rate and timestep-wise membrane-potential variability. The resulting score, referred to as the diagnostic token score, is used to selectively aggregate sparse PNI-related evidence. Experiments on a 10-year retrospective cohort of 182 CCA patients show that SCINTILLA-SNN achieves an AUROC of 0.748 under 5-fold cross-validation, while reducing the estimated inference energy by 23.18 \times compared with dense MAC-only computation of the same network.
[CV-48] HALDETECT at ImageEval 2026 Shared Tasks: Answer-First Contrastive Grounding with QLoRA
链接: https://arxiv.org/abs/2609.11236
作者: Syed Mohaiminul Hoque,Md Sakhawat Hossain
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 10 pages, 3 figures, 10 tables (including appendices). System description paper for Task 1b (English) of ImageEval 2026 Shared Tasks (Fourth Arabic Natural Language Processing Conference), to appear in the Shared Tasks proceedings
Abstract:Large multimodal models tend to hallucinate visual detail fluently, which limits their deployment for fine-grained interpretation. We present HALDETECT, our system for the English hallucination-detection track (Task 1b) of ImageEval 2026, in which a system must identify, from an image and three culturally plausible statements, the single visually grounded one. We frame the item as one contrastive decision, emit the answer before its explanation, and structure reasoning around colour/texture, shape/form, and context. Our best submitted adapter fine-tunes Qwen2.5-VL-7B-Instruct with 4-bit QLoRA while freezing the vision encoder and reaches Contrastive Instability (CI) 0.035 on the 1,000-item test set; we placed third of eight teams. Development experiments show that answer order can matter more than model scale and that adaptation beats prompting alone. Retrospective paired analysis of the released gold labels confirms the QLoRA gain over the best prompt but not the small gap between the devtest-selected and best-test adapters, and reseeding all four training sizes shows that the apparent data-scaling curve does not survive a seed change. The 35 residual errors are culturally plausible function, material, and recognition distinctions; naive adapter voting does not help.
[CV-49] When is Test-Time Adaptation Identifiable From Unlabeled Evidence?
链接: https://arxiv.org/abs/2609.11235
作者: Kartik Jhawar,Lipo Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Test-time adaptation (TTA) offers many ways to update a deployed model without labels, but choosing the wrong update can make a strong source model worse. Recent methods therefore try to predict which adaptation will work from unlabeled test data. We ask a prior question: does the evidence given to the selector contain enough information to determine the best action at all? We show that this is not guaranteed, even with a perfect selector. If an observation channel makes two deployments look the same while their TTA rankings differ, reliable selection is impossible from that channel; richer evidence can restore the decision only when it resolves the relevant ambiguity. We make this boundary exact in a finite-batch Gaussian TTA model, where doing nothing beats mean recentering for small shifts, recentering wins beyond a unique critical shift, and the boundary shrinks as 1/\sqrt n . Public benchmark studies on CIFAR-100-C and DomainNet-126 show the same failure mode with modern TTA methods: changing only deployment structure can reverse the oracle action while global order-blind evidence remains unchanged. The result is a practical way to separate two failure modes that are usually mixed together: a weak selector versus an information channel that cannot support the desired decision in the first place.
[CV-50] ri-DehazeGS: Scene–Medium Decoupled Gaussian Splatting with Transmittance-Aware Optimization
链接: https://arxiv.org/abs/2609.11223
作者: Kui Jiang,Yang Gu,Jiacheng Liu,Shiyu Liu,Youyu Chen,Hui Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Recovering clean 3D scenes from hazy multi-view images is challenging because haze attenuates scene radiance and introduces atmospheric scattering. Recent scattering-aware Gaussian Splatting methods introduce physical haze models into reconstruction, but they often apply degradation in image space or bind medium-related variables to Gaussian primitives, which can entangle clean scene radiance with atmospheric effects. Moreover, low-transmittance regions provide weakened supervision for Gaussian optimization, causing distant or dense-haze areas to be under-reconstructed. We argue that clean reconstruction under haze requires both scene–medium disentanglement and transmittance-aware optimization rebalancing. To this end, we propose Tri-DehazeGS, a scene–medium decoupled Gaussian Splatting framework. It represents the clean scene with Gaussian primitives, models the participating medium using an independent view-shared tri-plane field, and composes hazy observations through a physical scattering model. We further introduce Medium-Decoupled Transmittance Gradient Compensation (MD-TGC), which compensates haze-suppressed gradients after medium freezing without altering forward rendering. Experiments on real and synthetic haze benchmarks show that Tri-DehazeGS improves clean novel-view reconstruction. Code is available at this https URL.
[CV-51] CEM-TUDASR: Computationally efficient multi-modality transformer based unsupervised domain adaptive super-resolution approach
链接: https://arxiv.org/abs/2609.11201
作者: Anjali Sarvaiya,Jay Kadel,Kishor Upla,Kiran Raja
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Published in Biomedical Signal Processing and Control, Volume 129, 2027, Article 111315
Abstract:Wireless Capsule Endoscopy (WCE) enables non-invasive visualization of the gastrointestinal tract, but its miniaturized optics, sensor limitations, and wireless transmission constraints result in low-resolution images with reduced visibility of diagnostically important structures. This paper proposes CEM-TUDASR, a computationally efficient unsupervised Transformer-based super-resolution framework for WCE image enhancement without paired low-resolution (LR) and high-resolution (HR) training data. A domain-adaptive degradation network synthesizes realistic WCE-like LR images from HR conventional endoscopy images, reducing the domain gap and enabling effective unpaired learning. The SR generator integrates Deep Attention Blocks (DABs) and a Fusion Attention Block (FAB) to capture long-range contextual dependencies and fine local structures while preserving perceptual and structural fidelity. The model is trained on a curated dataset derived from Kvasir Capsule and evaluated on KID and GIANA for cross-dataset generalization. No-reference quality metrics, including BRISQUE, PIQE, NIQE, and the domain-specific EndoQM, show that CEM-TUDASR consistently outperforms existing unsupervised SR methods. Qualitative results further demonstrate improved restoration of mucosal textures, vascular patterns, and clinically relevant anatomical details. Cross-domain experiments on retinal images additionally demonstrate the adaptability of the framework. With only 2.67 million parameters and 169.94 GFLOPs, CEM-TUDASR achieves high-quality reconstruction while maintaining computational efficiency, making it suitable for resource-constrained clinical and embedded endoscopic applications.
[CV-52] A Multi-View and Confusion-Guided Ensemble Framework for Robust Synthetic Image Attribution
链接: https://arxiv.org/abs/2609.11188
作者: Zuomin Qu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Synthetic image attribution (SIA) has become increasingly important with the rapid advancement of text-to-image generation models. However, accurately identifying the source model of a generated image remains challenging due to the growing similarity among modern diffusion-based generators and the presence of diverse post-processing operations. In this report, we present a multi-view and confusion-guided ensemble framework for the Synthetic Image Attribution Challenge of the DLMMDD Workshop at ICANN 2026. Our approach integrates multiple complementary architectures, including FFT-ConvNeXt, DINOv2, CLIP, and Xception, to capture diverse attribution cues from frequency, semantic, and forensic perspectives. To improve robustness against unknown degradations and image manipulations, extensive data augmentation strategies are employed during training, simulating realistic post-processing operations such as compression, resizing, grayscale conversion, and blur. Furthermore, we analyze the confusion patterns of the ensemble model and observe severe ambiguity between Stable Diffusion 3 and Stable Diffusion 3.5. To address this issue, we introduce a dedicated binary expert classifier that is selectively activated under low-confidence conditions. We additionally apply class-adaptive confidence calibration to improve the discrimination of challenging classes such as Tencent Hunyuan. The proposed framework achieved 99.53% on the public leaderboard and 99.20% on the private leaderboard. The source code and implementation details are publicly available at this https URL.
[CV-53] Beyond Visual Quality: Evaluating Physical Consistency under Ego-Motion with EgoGenEval
链接: https://arxiv.org/abs/2609.11172
作者: Yilin Long,Chenming Zhu,Zitang Gou,Jingli Lin,Tai Wang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 39 pages, 10 figures, and 18 tables. Code: this https URL
Abstract:Recent visual generators produce high-fidelity images yet often violate physical consistency under ego-motion, limiting their use for spatial reasoning and embodied planning. Existing benchmarks largely focus on isolated images or single-step quality, leaving this challenge underexplored. We introduce EgoGenEval, a geometry-grounded, pose-free benchmark designed to evaluate the physical consistency of visual generators under ego-motion, and organize our study into two parts. (1) EgoGenEval contains 1,400 cases and 2,360 target views spanning single-step and multi-step ego-motion. It separately measures Camera Motion Grounding (CMG) and Scene State Preservation (SSP), with both metrics validated against blinded human judgments. Evaluating 16 pose-free generators together with two pose-conditioned references reveals that current models struggle to execute camera motion while maintaining scene state, and that no system performs well on both axes at once. (2) To examine whether benchmark-derived data can improve these capabilities, we build EgoGen-Train from the same geometry-grounded pipeline and run controlled SFT studies. These show that pairwise supervision does not reliably improve camera-motion grounding and scene-state preservation together: even at the full training pool and the longest budget, scene preservation gains a fraction of what camera motion does. This points to the pairwise teacher-forced objective itself as the binding constraint, motivating a trajectory-centric paradigm that couples self-conditioned rollouts with explicit pose and visibility supervision.
[CV-54] UniH3: Unifying Hierarchical Homogeneity and Heterogeneity for All-in-One Medical Image Restoration ECCV2026
链接: https://arxiv.org/abs/2609.11156
作者: Zhiwen Yang,Jiayin Li,Chengyu Liu,Hui Zhang,Bingzheng Wei,Yan Xu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: This paper has been accepted by ECCV 2026
Abstract:All-in-One medical image restoration (MedIR) aims to address diverse tasks across modalities and degradation types using a single universal model. Existing methods typically prioritize modeling inter-task heterogeneity (e.g., distinct data distributions and degradation types). However, they largely neglect the inherent homogeneity present in medical images, such as widely shared anatomical structures within and across modalities, which can be leveraged to ease model training and improve generalization. To this end, we propose UniH3, a novel framework that Unifies Hierarchical Homogeneity and Heterogeneity for all-in-one medical image restoration. Specifically, to comprehensively exploit homogeneity, we introduce a Hierarchical Homogeneity Memory (H2M) module that progressively distills intra- and inter-task homogeneity priors from high-quality images during training, and adaptively retrieves the most relevant priors tailored to the input for guided restoration. These retrieved priors are then injected into the restoration pipeline via an efficient Homogeneity-Guided Attention (HGA) mechanism. Furthermore, to comprehensively address heterogeneity, we design a Hierarchical Heterogeneity Balancer (H2B) that mitigates both inter- and intra-task conflicts during optimization, facilitating balanced and effective multi-task learning. Extensive experiments on two large-scale benchmarks, MedIR-2D-500K and MedIR-3D-3K, demonstrate that UniH3 achieves state-of-the-art performance on both all-in-one and single-task medical image restoration. We hope this work establishes a strong benchmark and advances the development of general-purpose medical image restoration models. Code is available at this https URL.
[CV-55] LAION-Mobile: Evaluating Deepfake Detectors On One Million Smartphone Photos
链接: https://arxiv.org/abs/2609.11134
作者: Achim von Stryk,Janis Keuper
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Most Deepfake detectors report near-perfect AUC scores on their reference benchmarks. However, a recent ICML position paper argues that these evaluations collectively neglect the impact of modern smartphone photography: the widely used on-device neural image-signal processing pipelines (like multi-sensor fusion or noise and motion-blur suppression) increasingly shift the imaging paradigm from simple lens projections towards computational photography. Hence, devices actually generate, rather than record photos. This increases the risk that deepfake detectors may flag ordinary phone photos as fake. Due to the lack of large-scale datasets containing images from modern smartphones, this hypothesis has so far only been tested in small proof-of-concept studies. The aim of this paper is to close this gap. We introduce LAION-Mobile, an open dataset containing about 1 million smartphone images with EXIF metadata distilled from re-LAION-5B. Evaluating twelve state-of-the-art deepfake detectors with their original paper checkpoints on a 9,115-image evaluation sample of this pool (DIRE on 738), we report three key findings: (i) On modern AI content no detector exceeds AUC 0.624, and five of twelve fall below chance. (ii) Real-photo false-alarm rates are an artefact of threshold calibration: thresholds fitted on legacy GAN data make several detectors look deployable (less than 11 percent FPR), yet the same detectors flag 17-91 percent of real photos once the identical criterion is refit on modern content. (iii) Consequently, no detector both beats chance on modern AI content and keeps a deployable real-photo false-alarm rate. Mirroring the device mix of web collections, the corpus probes the first neural-ISP generation (2018-2020); current flagships are essentially absent, leaving the modern-ISP regime as the open gap.
[CV-56] ReconPlusGen: Injecting Reconstruction Prior into Multi-view 3D Generation through Noise Inversion and Modulation
链接: https://arxiv.org/abs/2609.11129
作者: Jiarui Liu,Heng Li,Weiyu Li,Keng Deng,Junyuan Deng,Zheng Zhongxing,Junyu Huang,Jiahao Chang,Xiaoguang Han,Ping Tan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Qualitative results and an illustration of our core idea. Top left: reconstruction results on benchmark images. Top right: reconstruction results on real-world images. Bottom: illustration of reconstruction-guided noise initialization and modulation. Given multiple input images, we predict a point cloud in canonical space, deterministically inject the predicted geometry into the diffusion process through noise inversion, and modulate the resulting noise to preserve the generative flexibility required to complete unobserved regions and refine visible geometry.
[CV-57] Beyond Benchmarks: Using VLMs to Reveal Systematic Classification Failures Under Real World Conditions
链接: https://arxiv.org/abs/2609.11126
作者: Dieuwertje Alblas,Alma M. Liezenga,Jan Erik van Woerden,Fedor Taggenbrock,Dalia Aljawaheri,Klamer Schutte
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Emerging Technologies (cs.ET)
备注: To be presented at SPIE Sensors + Imaging, Edinburgh, in September 2026
Abstract:Verification and validation (VV) of classification models is crucial to enable a wide range of sensor processing applications. Currently, the VV process relies on time-consuming manual inspection of erroneous samples to find meaningful patterns. This work explores the use of Vision Language Models (VLMs) to speed up this laborious process. VLMs are trained to embed images into a semantically meaningful vector representation, from which human-interpretable systematic errors can be distilled. Deploying such VLM-based methods in a defence context introduces two major challenges: (1) the defence domain is underrepresented in the training data of VLMs, and (2) surroundings and context are less diverse than for other domains. This study provides an initial assessment of the suitability of VLM-based methods for VV of defence applications. We propose a VLM-based error slice detection (ESD) method that independently groups and labels systematic errors made by a classification model. We demonstrate that this method is able to identify operationally-relevant artificially added perturbations in a non-military dataset. In a military context, our method clusters and describes images based on their surroundings, but also exhibits overlap between cluster descriptions. We further investigate the difference in embedding variation between our military and non-military dataset, which remains a topic of interest. Although the results do not yet warrant fully automated VV through VLM-based ESD, they show that VLMs could be used to accelerate VV processes in the future.
[CV-58] ailProp: content-adaptive light- and heavy-tailed propagation for vision ICLR2027
链接: https://arxiv.org/abs/2609.11081
作者: Jiahao Kong,Zihan Li
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Preprint. Under review at ICLR 2027. 18 pages, main text 9 pages, includes appendix, figures and supplementary analyses
Abstract:Science-inspired vision models show that explicit propagation dynamics can provide structured and interpretable alternatives to conventional token mixing. Existing formulations, however, typically construct and adapt visual propagation within a particular dynamical family, while visual representations can require substantially different spatial interactions across samples, channels, and network stages. We explore cross-regime adaptive propagation and introduce TailProp, a hierarchical vision backbone built upon the Tail Propagation Operator (TPO). TPO uses Gaussian and Cauchy stable-process propagators as complementary bases with rapidly decaying and heavy-tailed spatial influence, and predicts a content-conditioned channel-wise coefficient to adaptively combine them. Because this coefficient is spatially shared, the two responses are fused directly in the DCT domain with a single DCT/IDCT pair, yielding O(N^1.5) spatial mixing for square feature maps with N=HW and fixed channel width. Across image classification, object detection, semantic segmentation, robustness, and cross-backbone restoration, TailProp consistently outperforms matched propagation baselines; TailProp-B reaches 84.4% Top-1 accuracy on ImageNet-1K, 50.3/44.8 box/mask AP under the 3x Mask R-CNN schedule, and 50.8% mIoU on ADE20K. Controlled ablations further show that these gains are not explained by single-basis propagation, an additional same-family branch, or within-family adaptive order alone, supporting complementary two-basis propagation as an effective design principle for visual representation learning.
[CV-59] Meta-Learning for Classifier Selection in Image Datasets: A Feature-Driven Framework for Accuracy Prediction
链接: https://arxiv.org/abs/2609.11041
作者: Zahra Nabizadeh_Shahre_Babak,Farzaneh Koohestani,Nader Karimi,Shahram Shirani,Shadrokh Samavi
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 27 pages, 4 figures
Abstract:No Free Lunch theorem implies that any performance gains achieved by a classifier on a particular image distribution are necessarily offset by a loss of performance over the set of all possible problems; thus, no single model is universally optimal. Selecting the most suitable classifier for image datasets is a critical yet challenging task due to the intrinsic complexity and diversity of images. This paper proposes a meta-learning framework that leverages a comprehensive set of meta-features capturing dataset complexity to predict classifier performance without exhaustive training. By extracting and selecting features using methods such as autoencoders, pre-trained networks, and dimensionality reduction techniques, we train regression models to efficiently estimate classifier accuracies. Additionally, clustering techniques are employed to group classifiers with similar performance patterns, simplifying the recommendation process. The datasets used span a wide range of concepts, including nature, animals, numbers, motorcycles, medical images, and human bodies, to ensure broad generalization. Evaluated on 56 diverse image datasets, our approach achieves an average ranking prediction accuracy exceeding 86%, demonstrating its effectiveness in guiding model selection. This scalable and interpretable framework provides a practical solution to improve classification performance while reducing computational costs.
[CV-60] oward Interpretable Multimodal Fusion: Heat Conduction Modeling for Hyperspectral and LiDAR Joint Classification
链接: https://arxiv.org/abs/2609.11040
作者: Kan Wei,Jiahui Cui,Jing Yao,Xinyu Zhao,Lei Wang,Pedram Ghamisi
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accepted by IEEE TCSVT
Abstract:The fusion of hyperspectral (HS) and Light Detection and Ranging (LiDAR) data plays a crucial role in enhancing land-cover classification by jointly exploiting spectral, spatial, and structural cues. However, existing multimodal fusion methods still struggle to model long-range dependencies and complex anisotropic interactions while maintaining computational efficiency. This paper introduces M2Heat, a physics-inspired framework that investigates multimodal fusion through the lens of heat conduction. At its core, a physics-driven visual heat conduction module (vHeat) and enhanced Frequency Value Embeddings (FVEs) simulate anisotropic information flow, enabling the capture of global dependencies with sub-quadratic complexity and physical interpretability. This mechanism, combined with a hybrid spatial-frequency fusion strategy named Cross-Frequency Fusion (CFF) module, produces highly discriminative and robust feature representations. M2Heat achieves competitive overall performance on three benchmarks, i.e., Trento, Houston2013, and Augsburg, while providing an interpretable heat-conduction-guided perspective for multimodal feature fusion. These results indicate the potential of heat-conduction-guided neural operators for efficient and interpretable RS multimodal fusion. The source code is publicly available at https: /github.com/Weikan0425/M2Heat_HSI_LiDAR.
[CV-61] CamPilot: A Multi-Agent Cinematic Assistant for Camera-Controlled Movie Generation EMNLP2026
链接: https://arxiv.org/abs/2609.10943
作者: Yang Wu,Stefano Petrangeli,Ishita Dasgupta,Yu Shen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: EMNLP 2026 Workshop REALM
Abstract:The integration of large language models (LLMs) into video generation has enabled rapid text-to-video creation and improved visual quality. However, it still falls short of professional filmmaking, where cinematographic language is less refined than human-crafted camera work and multi-shot continuity remains challenging. To address these limitations, we introduce CamPilot, a multi-agent framework that integrates cinematographic planning and camera-work control to produce more coherent, logically structured, and human-aesthetic movies. CamPilot adopts a GRPO-based learning paradigm to learn camera work planning from 14K real-world professional movies, internalizing motion patterns and composition principles that support reasoning over shooting techniques (e.g., camera angle, motion, and focal behavior) and cross-shot relationships for controllable camera-viewpoint generation. Multiple agents further collaborate and evolve to improve overall output quality. To support this work and further studies in this domain, we establish CamEval, a benchmark for evaluating camera work quality and cinematic engagement. Empirical results show that CamPilot outperforms state-of-the-art text-to-movie generation methods on cinematographic control and quality, highlighting the impact of professional camera design on movie generation.
[CV-62] HiPerViT: A Hierarchical Perceiver-Vision Transformer Architecture for Multi-Scale Texture Recognition
链接: https://arxiv.org/abs/2609.10917
作者: João Pedro C. A. de Sá,Odemir Martinez Bruno
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 30 pages, 4 figures, 12 tables
Abstract:Texture recognition remains challenging for modern vision models because discriminative evidence is often carried by higher-order spatial statistics rather than by object shape alone. While Vision Transformers provide strong long-range modeling capacity, their standard object-centric representations do not explicitly expose such statistical structure, which limits texture sensitivity in fine-grained recognition settings. We present HiPerViT, a compact vision-only architecture that injects an explicit second-order statistical prior into a transformer-based recognition pipeline. The method combines global and local image views with a compact bilinear descriptor encoded as a statistical token, and integrates this token with first-order spatial representations through Perceiver-style latent distillation. This design enables direct interaction between spatial tokens and second-order feature co-occurrence statistics, providing the model with explicit access to texture-relevant information without requiring multimodal pretraining or ensemble construction. Across six texture recognition benchmarks, HiPerViT achieves consistent improvements over strong vision-only baselines under the reported evaluation protocols, including gains of +3.05 percentage points on DTD, +10.48 on GTOS-Mobile, and +10.10 on 1200Tex. Beyond benchmark performance, our analyses show that these gains are largely invariant to the backbone depth used to extract second-order statistics and to the ordering of interaction and distillation stages. This pattern suggests that the primary source of improvement is not a specific fusion topology, but the explicit availability of second-order statistical information as a first-class representational signal. These results support explicit statistical tokenization as an effective and robust design principle for texture-centric visual recognition.
[CV-63] IMLE-VLA: Fast Single-Step Action Generation for Vision-Language-Action Policies IROS
链接: https://arxiv.org/abs/2609.10915
作者: Kian Hosseinkhani(1),Qinhe Peng(2),George Shramko(1),Mehran Aghabozorgi(1),Jianing Qian(2),Tristan Engst(1),Alireza Moazeni(1),Dinesh Jayaraman(2),Ke Li(1, 3 and 4) ((1) Simon Fraser University, (2) University of Pennsylvania, (3) Alberta Machine Intelligence Institute, (4) Canada CIFAR AI Chair)
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages, 5 figures, 5 tables. Accepted to IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2026. Project page: this https URL
Abstract:Vision-language-action (VLA) policies leverage pretrained vision-language backbones to achieve strong cross-task generalization. A leading design couples this backbone with a dedicated continuous action head trained via diffusion or flow matching. However, such heads rely on iterative multi-step sampling, for example 10 Euler steps in \pi_0.5 . This creates an inference bottleneck that produces stop-and-go movement in the robot and slower task completion. We introduce IMLE-VLA, which replaces the iterative action head with a single-step conditional generator trained via conditional Implicit Maximum Likelihood Estimation (cIMLE). The cIMLE objective promotes multimodal action coverage, avoiding the mode collapse of naive regression heads while eliminating multi-step sampling entirely. When IMLE-VLA is applied to \pi_0.5 , it increases inference frequency 3.67x (55 Hz vs. 15 Hz), enabling up to 11x higher action throughput. On the 40-task LIBERO benchmark, IMLE-VLA achieves the highest average success rate (98.0%) among all baselines while leading in inference frequency. Under the test-time perturbations of LIBERO-plus, IMLE-VLA retains \pi_0.5 's robustness while other baselines degrade sharply, confirming that the cIMLE head preserves generalization. Real-world experiments on a Franka Emika Panda across four tasks demonstrate smoother motion (2.2x to 3.0x lower jerk) and faster task completion, with IMLE-VLA outperforming \pi_0.5 on every task and reducing average VLA inference time per episode by 3.9x to 6.6x. Videos and code are available at this https URL
[CV-64] Symmetry-aware super-resolution of crystal orientation maps via invariant latent-space learning
链接: https://arxiv.org/abs/2609.10898
作者: Umang Garg,Warren Zamudio,McLean P. Echlin,Samantha H. Daly,Tresa M. Pollock,B.S. Manjunath
类目: Computer Vision and Pattern Recognition (cs.CV); Materials Science (cond-mat.mtrl-sci); Machine Learning (cs.LG)
备注: 37 pages, 11 figures
Abstract:Crystal-orientation maps are physical fields defined only up to crystal symmetry; electron backscatter diffraction (EBSD) resolves them experimentally, but acquisition-time constraints limit spatial resolution. Unlike conventional images, EBSD data lie on the quotient space \mathrmSO(3)/G , where G is the crystal-symmetry group. Standard Euclidean interpolation can therefore mix symmetry-equivalent representations and blur grain boundaries. We introduce the Symmetry-Group-Aware Super-Resolution Attention Network (SG-SRAN), which incorporates crystal symmetry and boundary preservation by design. A frozen, locally isometric encoder maps equivalent orientations to a common latent representation in which Euclidean distance approximates misorientation. Super-resolution is performed in this space, with each high-resolution token restricted to a feature-consistent local support to prevent cross-boundary mixing. A dictionary-based decoder then recovers valid orientations. Across FCC and HCP benchmarks, SG-SRAN matches 15-16 million parameter backbones using only 27-49k trainable parameters, while achieving the lowest p68 errors, highest inverse-pole-figure fidelity, and zero-shot transfer to unseen alloys.
[CV-65] Evaluation of Vision-Language Models Across Diverse Coastal Environments
链接: https://arxiv.org/abs/2609.10855
作者: Seth Knoop,Chad R. Samuelson,Gabriel R. Slade,Brady Moon,Joshua G. Mangelson
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: This work has been accepted for publication with the 2026 IEEE OCEANS conference
Abstract:Vision-language models (VLMs) enable robotic per- ception by associating visual observations with natural-language concepts. Yet their performance in coastal environments remains largely unexplored. We introduce a densely labeled coastal dataset containing more than 1,000 images collected across seven missions in three regions of Oahu, Hawaii, with 18 semantic classes and over 7,400 annotated instances. We evaluate seven modern VLMs through three complementary experiments mea- suring text-to-mask, mask-to-mask, and mask-to-text alignment. Broad landscape classes are generally recognized more accurately than conventional object and coastal classes, with coastal con- cepts presenting the greatest challenge. However, comparisons of shared conventional classes across coastal and terrestrial datasets reveal no consistent performance difference attributable solely to environmental context. Mask-to-mask matching also remains similar across conventional and coastal classes, while alternative textual labels substantially improve recognition of several coastal concepts. These results suggest that lower performance on coastal classes (at least on the objects/query categories evaluated) is heavily influenced by segmentation and linguistic representation.
[CV-66] Are We Really Doing Few-Shot Learning? A Critical Examination of Pre-Training Assumptions
链接: https://arxiv.org/abs/2609.10851
作者: Alejandro Galan-Cuenca,Marcelo Saval-Calvo,Antonio Javier Gallego
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Few-shot learning is commonly evaluated under protocols that pre-train a model on a large auxiliary set whose classes are disjoint from the target episodes yet drawn from the same visual domain. This paper examines whether such protocols truly reflect low-data learning. We systematically compare no pre-training, class-disjoint in-domain pre-training, supervised out-of-domain pre-training, and label-free out-of-domain pre-training across eight datasets, three few-shot architectures, and multiple way-shot settings. Our results show that class disjointness alone is insufficient to remove the influence of target-domain data. In-domain pre-training improves over no pre-training by 33.41 percentage points on average, whereas supervised out-of-domain pre-training yields 23.75 percentage points, revealing a 9.66-point optimistic bias associated with domain overlap. Although out-of-domain pre-training is more realistic in applications where target-domain data are scarce, its effectiveness depends strongly on the compatibility between source and target domains. We further show that labeled source data are not strictly required, with an augmentation-based label-free strategy reaching an average gain of 27.71 percentage points and closely matching supervised out-of-domain pre-training at 27.97 percentage points. Finally, we introduce a descriptor-based source-selection strategy that estimates source-domain suitability before pre-training, reaching a median gap of only 1.37 percentage points to oracle selection. These findings highlight the need to move beyond in-domain pre-training as the default few-shot evaluation protocol, since it can overestimate performance in realistic scenarios where target-domain data are scarce.
[CV-67] Overpainting: Localized Context-aware Diffusion Image Editing
链接: https://arxiv.org/abs/2609.10811
作者: Sam Sartor,Iliyan Georgiev,Michael Fischer,Valentin Deschaintre,Pieter Peers
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL
Abstract:We present “overpainting”, an image editing operation which offers both control over the location of the edit and awareness of the previous content in that location. The overpainted area is given by a trimap, where white-annotated pixels must be edited, gray-annotated pixels may be edited, and black-annotated pixels must not be edited. This enables both precise and loose control, depending on user intent. We implement overpainting by adapting a pretrained image editing diffusion model using a combination of joint attention and low-rank adaption across input images with attention-dropout to balance the information flow between noise, source and mask images. We present a novel, automated, training data generation pipeline that (1) generates a set of candidate image pairs leveraging existing language-based editing models, (2) carefully curates those pairs, and (3) extracts a trimap from each usable pair. We demonstrate the versatility of our overpainting model on a wide range of editing tasks. Comments: Project page: this https URL Subjects: Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2609.10811 [cs.CV] (or arXiv:2609.10811v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2609.10811 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-68] rajFusionNet: Transformer-Based Prediction of Pedestrian Crossing Intention via Fusion of Trajectory Representations and Scene Graphs
链接: https://arxiv.org/abs/2609.10806
作者: François G. Landry,Moulay A. Akhloufi
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: This work has been submitted to Signal, Image and Video Processing for possible publication
Abstract:The pedestrian crossing intention task involves predicting whether pedestrians are likely to cross the road from the point of view of an autonomous vehicle. We introduce TrajFusionNet+, a novel transformer-based model for pedestrian crossing intention prediction. TrajFusionNet+ combines sequential and visual representations of pedestrian trajectory with a graph-based representation of the scene context in order to predict pedestrian crossing intention. The proposed architecture builds upon our previous model, TrajFusionNet, and comprises three branches: a Sequence Attention Module (SAM), which processes a sequential representation of past and predicted pedestrian trajectories; a Visual Attention Module (VAM), which utilizes a visual representation of the pedestrian trajectories by overlaying observed and predicted bounding boxes onto scene images; and a Graph Attention Module (GAM), which extracts pedestrian-centric graphs from segmented scene images and captures the relational dependencies between pedestrians and traffic elements. TrajFusionNet+ achieves improved state-of-the-art performance on the two most widely used pedestrian crossing intention datasets, PIE and JAAD. Furthermore, we introduce a new evaluation protocol in which models are trained jointly on the PIE and JAAD datasets but evaluated separately on each. Under this setting, TrajFusionNet+ demonstrates superior generalization compared to existing approaches.
[CV-69] How Much Velocity Does Off-Ball Space Value Need? A Broadcast-Viewport Benchmark
链接: https://arxiv.org/abs/2609.10801
作者: Seongjin Choi
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 15 pages, 3 figures, 4 tables. Code, logs and data: this https URL (velocity/)
Abstract:Velocity-aware pitch control is standard, but under a broadcast viewport half the players are off screen and on-screen velocities come from a drifting calibration. We ask at which layer of broadcast off-ball analysis velocity changes the answer. Inheriting our off-screen imputation protocol (three Metrica matches, 44 m viewport, block-bootstrap CIs), we score four velocity regimes – none, viewport-legal observed, true-for-visible, true-for-all – against a velocity-aware ground truth at three layers: imputation, the control surface, and team verdicts. Velocity is nearly useless for imputation (-0.2 pp against a 12–14 pp velocity-free surface MAE), first-order for the surface (-1.5 to -1.8 pp, 11–15% of that MAE), and ten times smaller for verdicts (-0.12 to -0.19 pp). The velocity that matters is the visible channel: perfect occluded-player velocity adds 2–6% of the visible gain, and no last-seen decay policy we tested exceeds that. Omitting velocity blurs the surface (per-frame |e| 2.2–2.6 pp) with small time-averaged bias (per cell =0.4 pp), whereas imputation error is a structured bias against the defending team’s deep zone (5–9 pp). At a fixed velocity window, a noise ladder of eleven jitter settings, including sigma_v-matched pairs, is ordered to first order by one velocity-noise axis sigma_v with break-even ~1 m/s; eleven SoccerNet-GSR clips from one match through our pipeline measure sigma_v=1.65 m/s yet recover 24–36% of the benefit: 43% of the variance is frame-common, which the surface tolerates, and the residual is heavy-tailed and clustered, which Gaussian controls matched on component RMS do not reproduce (+0.03 vs. +0.36). The share of velocity-free error that velocity removes grows with viewport width (7% at 36 m, 21% at 60 m): fix imputation on tight shots, velocity on wide ones. Code and logs are released.
[CV-70] RiVaT-Fuse: Reliability-Calibrated Variational Tensor Fusion for Multimodal Prediction under Modality Uncertainty
链接: https://arxiv.org/abs/2609.10798
作者: Yingfan Xu,Tieming Liu,Ye Liang,Taiping Liu
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Image-metadata prediction requires fusing heterogeneous evidence whose reliability can vary across samples and latent factors. Existing representation-level fusion methods typically choose an aggregation architecture, such as concatenation, gating, conditional modulation, or attention, without explicitly defining what the fused representation should mean under modality uncertainty. We propose RiVaT-Fuse, a reliability-calibrated variational tensor fusion framework that defines fusion as sample-wise latent-state estimation. Rather than producing a fused vector by direct aggregation, RiVaT-Fuse estimates a consensus latent state through a variational objective that balances image evidence, metadata evidence, structured cross-modal interaction, and stability. The resulting framework replaces scalar modality confidence with matrix-valued trust geometry, decomposes interaction into additive, multiplicative, and relational components, and couples the latent state with conditional robustness and structured multi-task prediction. We provide well-posedness and stability interpretations of the latent solve and instantiate the framework with efficient low-rank-plus-diagonal trust operators. On an image-level image-metadata prediction benchmark, RiVaT-Fuse achieves the strongest overall predictive rank among direct representation-level baselines while improving probability and label stability under perturbation.
[CV-71] wo-Parameter Flow Map Learning for Continuous-Time Diffeomorphic Image Registration ECCV
链接: https://arxiv.org/abs/2609.10789
作者: Mohammadjavad Matinkia,Nilanjan Ray
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Published at European Conference on Computer Vision (ECCV), 2026
Abstract:Diffeomorphic image registration is central to medical image analysis, enabling anatomically consistent alignment across subjects. Most learning-based diffeomorphic methods model autonomous ODEs(ordinary differential equations) by parameterizing a stationary velocity field and recovering deformations via scaling-and-squaring. While non-autonomous ODEs with time-dependent velocities increase expressiveness, existing approaches rely on numerical integration to implicitly enforce flow structure that entangles model expressiveness with discretization accuracy. We propose a framework to directly learn the continuous-time solution of a non-autonomous ODE formulated as a two-parameterflow map. By enforcing cocycle consistency, a fundamental structural property of time-varying flows, we learn the flow maps without time discretization and velocity integration during training. The framework recovers diffeomorphic mappings at inference using a small number of compositions. Our proposed framework seamlessly incorporates standard registration backbones and improves alignment accuracy consistently across nine datasets while preserving diffeomorphic structure. Notably, the proposed method achieves an average Dice improvement of 2.1% on brain MRI benchmarks, a 12% TRE reduction on lung CT, and a 2.6% Dice gain on cardiac MRI and ultrasound datasets.
[CV-72] Shedding Light: A Benchmark for Evaluating Lighting Understanding in Generative Image Models SIGGRAPH
链接: https://arxiv.org/abs/2609.10787
作者: Justine Giroux,Jack Oliver Hilliard,Yannick Hold-Geoffroy,Javier Vazquez-Corral,Jean-François Lalonde
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to ACM Transactions on Graphics (SIGGRAPH Asia 2026), vol. 45, no. 6, article 227, December 2026. 25 pages. Project page: this https URL
Abstract:Accurate modelling of illumination is central to realistic image synthesis and scene understanding. Yet, there is little exploration into whether image generative models are good at this task or whether physical plausibility remains a key challenge for them. Clearly, significant progress has been made in realistic image synthesis, but do models truly understand lighting in a physically accurate manner? To answer this question, this work proposes a benchmark to assess the lighting understanding and harmonisation capabilities of generative models. Our key insight is that evaluating lighting understanding for such models only requires testing how well they insert novel objects into real photographs whilst maintaining consistent illumination. To do so, we use a multi-illumination dataset with images containing simple objects serving as ``light probes’', and prompt models to inpaint the same object onto the original image, then compare the generated results against the ground-truth light probes. We then estimate the lighting direction, colour and radiance distribution from the inpainted probes, providing a quantitative measure of illumination accuracy and photometric realism. Our work establishes a scalable evaluation protocol to systematically assess how well generative models capture and reproduce real-world lighting, offering a foundation for benchmarking the photometric accuracy of any future models. All code and data are available at this https URL .
[CV-73] GRADE: Single-Frame Generative Radar Depth Estimation Under Visual Degradation
链接: https://arxiv.org/abs/2609.10756
作者: Bin Zhao,Patrick Chiou,Nakul Garg
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: To appear in ACM MobiCom 2026
Abstract:Dense 3D depth perception fails under smoke, fog, and darkness because optical sensors cannot penetrate airborne particulates. mmWave radar remains usable and measures range accurately under these conditions, but its small aperture limits angular resolution. We present GRADE, which grounds a pretrained generative prior in single-frame radar geometry to estimate high-fidelity metric depth. GRADE first maps raw 4D radar spectra to coarse metric depth. A latent diffusion backbone then recovers structural detail while conditioning every denoising step on this estimate. A pixel-space adapter uses residual camera cues when available and is trained across clear, smoke-degraded, and occluded inputs so the full output approaches the radar-conditioned path as visibility degrades. Trained and evaluated on ~95K frames across 12 buildings with real smoke, GRADE achieves an MAE of 0.303 m in clear scenes and 0.313 m under smoke, outperforming existing baselines. Code and datasets are available at this https URL.
[CV-74] Meta-Learning for Data-Efficient Plant Growth Estimation via Vision Transformers and Fuzzy Clustering
链接: https://arxiv.org/abs/2609.10749
作者: Sheikh Hasan Elahi,Rusith Chamara Hathurusinghe Dewage,Habib Ullah,Muhammad Salman Siddiqui,Rakibul Islam,Fadi Al Machot
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Accurate plant growth estimation is essential for greenhouse monitoring, yet obtaining labeled data remains costly and time-consuming. To address this, we propose a few-shot regression framework that combines Vision Transformer (ViT) feature embeddings, clustering-based task construction, and gradient-based meta-learning, and show that task construction in embedding space is a primary driver of performance. The approach leverages an unlabeled image pool to organize data into structured tasks using fuzzy c-means clustering, enabling efficient learning from a small number of labeled samples. We systematically evaluate meta-learning methods and show that second-order methods (e.g., Model-Agnostic Meta-Learning variants such as MAML++) outperform classical baselines in the few-shot regime. Furthermore, intra-cluster support selection has a limited and dataset-dependent impact. Experiments on two plant datasets show that structured task design combined with meta-learning enables reliable plant growth estimation under severe label scarcity.
[CV-75] MHE-Former: Multi-Hypothesis Transformers via Entropy Maximization for 3D Mesh Recovery
链接: https://arxiv.org/abs/2609.10743
作者: Boshu Jia,Rongyu Chen,Linlin Yang,Zihao Liu,Yingjie Chen,Zhongqun Zhang,Zhulin Tao,Shaohui Lin,Xiaoyu Wu,Libiao Jin,Baochang Zhang,Angela Yao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 14 pages, 11 figures
Abstract:Monocular 3D hand and body mesh recovery often suffers from severe occlusion and ambiguity. Traditional deterministic methods typically regress a single optimal solution, leading to overconfident predictions. In this paper, we introduce an exploration–exploitation paradigm for ambiguous mesh recovery with multi-hypothesis learning and selection. Specifically, during exploration, based on our probabilistic formulation and entropy maximization, we propose a novel multi-hypothesis method referred to as MHE-Former. It is a Transformer-based multi-hypothesis framework, ensuring high training efficiency and label friendliness while generating plausible and diverse hypotheses. During exploitation, we propose Hypothesis Selection, a context-aware process for multiple predictions. Especially leveraging VLM’s powerful visual understanding and reasoning capabilities, it allows users to choose the most plausible and desired estimate with additional evidence and natural language intent. Extensive experiments demonstrate that our framework achieves state-of-the-art performance in accuracy and diversity across multiple datasets. The user preference study further shows the practicality of our hypothesis selection process.
[CV-76] AcFlow: Controlling Text-to-Image Diffusion Transformers via Learned Conditional Activation Flow
链接: https://arxiv.org/abs/2609.10723
作者: Junran Wang,Zehao Jin,Tianyu Luan,Xinjie Shen
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Text-to-image diffusion transformers (DiTs) are powerful generators, yet direct prompting provides limited control interface for style intensity and can fail to suppress unwanted concepts. To enable these controls, we introduce AcFlow, an inference-time controller that transports intermediate layer image-token activations through a learned concept-conditioned velocity field while keeping the base DiT frozen. A textual concept description specifies the desired intervention, while the integration horizon provides a continuous control parameter. The field produces token-varying, activation-dependent updates. With parameters shared across concepts within each task family, the field supports fine-grained descriptions and generalizes to concepts unseen during training without per-concept fitting. On style control, AcFlow achieves the best style–content trade-off among the evaluated baselines in the high-style-alignment regime. At a fixed operating point, AcFlow attains style–content alignment of 0.5365/0.2860, compared with 0.4397/0.2684 for the baseline with the highest style alignment. Qualitative results demonstrate suppression of diverse concepts, including cases where direct prompting fails. Our analyses support the learned velocity field as an adaptive control mechanism, with update directions varying across tokens and depend on their activation states. Our code is available at this https URL.
[CV-77] HuRo: Robotizing Human Videos for Scalable VLA Pretraining
链接: https://arxiv.org/abs/2609.10706
作者: Jinho Jeong,Se June Joo,Jaehyun Kang,Dongyun Kim,Yena Kim,Hanjung Kim,Seon Joo Kim
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Accepted at CoRL 2026
Abstract:Human video datasets have emerged as a compelling alternative to expensive real-robot data, offering rich diversity at scale. To bridge the human-to-robot embodiment gap, existing approaches either robotize videos in task-matched settings or address observation and action alignment separately at scale. In this work, we systematically examine whether robotized human videos can provide effective and scalable supervision for pretraining vision-language-action (VLA) policies. To this end, we develop a robotization pipeline that converts heterogeneous human videos into robot-aligned observations and action trajectories while inferring missing intermediate signals across annotation levels. Using this pipeline, we construct the HuRo dataset, comprising about 630K robotized episodes and 142M processed frames from five human-video sources. Across four real-world manipulation tasks, increasing robotized pretraining scale improves overall completion from 51.5% to 80.3% and OOD completion under spatial and visual shifts from 34.9% to 72.2%. Ablations further show that visual robotization improves OOD robustness and that end-to-end pretraining with retargeted actions outperforms visual-only transfer. Code and data are released on our website: this https URL.
[CV-78] Rethinking Handwritten Character Recognition
链接: https://arxiv.org/abs/2609.10572
作者: Ranjit Raut,Aarav Subedi,Ashim Shrestha
类目: Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV)
备注: 21 pages, 12 figures
Abstract:Non-Latin handwritten character recognition (HCR) remains understudied. Dominant methods consider it as generic image classification, which uses model scale to implicitly learn stroke structure. Structural-prior efficiency—the principle that explicitly encoding script-geometric regularities as architectural inductive biases can be both more accurate and require fewer parameters. We introduce GraphemeNet, a unified multi-script architecture, governed by two orthogonal binary axes. Axis 1 operationalises stroke-level geometric regularity via Persistent Scaffold Injection (PSI): a script-specific asymmetric convolution injects a stroke scaffold as a weighted residual at every encoder stage, continuously anchoring learned features to script geometry—distinct from skip connections, auxiliary losses, or attention reweighting. Axis 2 selects between global average pooling with gated fusion and cross-scale attention with a Stroke Topology Module (STM), depending on whether glyph discrimination requires spatial relational reasoning. A Linear Capsule Routing (LCR) with O(n) routing is shared universally. On fourteen benchmarks across eight writing systems, the architecture generalises with only scaffold and decoder topology varying per script, consistently challenging, outperforming published baselines, and establishing structural-prior efficiency as a broadly applicable principle for multi-script HCR.
[CV-79] M3-Former: Multimodal Transformer with Mixture-of-Experts for Long-Term Vessel Trajectory Prediction
链接: https://arxiv.org/abs/2609.10559
作者: Wenzhe Jin,Haina Tang
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:To address the challenges of behavioral multimodality, limited semantic utilization, and long-term error accumulation in vessel trajectory prediction, this paper proposes M3-Former, a multimodal trajectory prediction framework enhanced by large language models (LLMs). The proposed framework incorporates vessel static attributes and navigational intent as semantic priors for long-term trajectory modeling. Specifically, a unified multimodal representation space is constructed, in which static semantic information is encoded by a pre-trained LLM and aligned with dynamic trajectory features through self-attention. To jointly capture global route planning and local motion variations, a dual-granularity Mixture-of-Experts (MoE) architecture is introduced, where sequence-level experts model global navigation trends and token-level experts refine fine-grained maneuvering behaviors. In addition, a Steering-Weighted Cross-Entropy loss is designed to alleviate the long-tail distribution of sparse turning samples and improve prediction accuracy in critical maneuvering scenarios. Experiments on a real-world Danish AIS dataset demonstrate that M\textsuperscript3-Former consistently outperforms state-of-the-art baselines across prediction horizons from 1 to 4 hours. In the 4-hour prediction task, the proposed method reduces Average Displacement Error (ADE) and Final Displacement Error (FDE) by 4.4% and 5.1%, respectively, compared with the strongest baseline. Qualitative and ablation analyses further verify that semantic fusion effectively reduces long-term trajectory drift, while the dual-granularity MoE improves robustness in complex waterways and route-branching scenarios. The proposed framework establishes a semantic-guided hierarchical prediction paradigm, in which high-level navigational intent and local motion dynamics are jointly modeled for robust long-term vessel trajectory forecasting.
[CV-80] Exponential Pixelating Integral transform with dual fractal features for enhanced chest X-ray abnormality detection
链接: https://arxiv.org/abs/2609.10988
作者: Naveenraj Kamalakannan,Sri Ram Macharla,M Kanimozhi,M S Sudhakar
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: Preprint of the article published in the ELSEVIER journal Computers in Biology and Medicine (CIBM), Vol. 182, November 2024. Final version available at DOI: this https URL
Abstract:The heightened prevalence of respiratory disorders, particularly exacerbated by a significant upswing in fatalities due to the novel coronavirus, underscores the critical need for early detection and timely intervention. This imperative is paramount, possessing the potential to profoundly impact and safeguard numerous lives. Medically, chest radiography stands out as an essential and economically viable medical imaging approach for diagnosing and assessing the severity of diverse Respiratory Disorders. However, their detection in Chest X-Rays is a cumbersome task even for well-trained radiologists owing to low contrast issues, overlapping of the tissue structures, subjective variability, and the presence of noise. To address these issues, a novel analytical model termed Exponential Pixelating Integral is introduced for the automatic detection of infections in Chest X-Rays in this work. Initially, the presented Exponential Pixelating Integral enhances the pixel intensities to overcome the low-contrast issues that are then polar-transformed followed by their representation using the locally invariant Mandelbrot and Julia fractal geometries for effective distinction of structural features. The collated features labeled Exponential Pixelating Integral with dually characterized fractal features are then classified by the non-parametric multivariate adaptive regression splines to establish an ensemble model between each pair of classes for effective diagnosis of diverse diseases. Rigorous analysis of the proposed classification framework on large medical benchmarked datasets showcases its superiority over its peers by registering a higher classification accuracy and F1 scores ranging from 98.46 to 99.45% and 96.53-98.10% respectively, making it a precise and interpretable automated system for diagnosing respiratory disorders.
[CV-81] Seamless Whole Slide Label-Free Virtual Staining MICCAI2026
链接: https://arxiv.org/abs/2609.10914
作者: Dou Hoon Kwark,Kianoush Falahkheirkhah,Ji-hun Oh,Shirui Luo,Volodymyr Kindratenko,Rohit Bhargava
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Quantitative Methods (q-bio.QM)
备注: Accepted to MICCAI 2026
Abstract:Label-free virtual staining offers a compelling, non-destructive alternative to standard histopathology; however, its clinical adoption is hindered by the computational bottlenecks inherent to processing gigapixel Whole Slide Images (WSIs). Current deep learning approaches require patch-based inference to avoid memory constraints, which disrupts global tissue continuity and introduces tiling artifacts–displaying visible seams and color shifts. To address this, we introduce the Consistency Memory Bank (COMB), a novel label-free virtual staining framework that enforces spatial and channel consistency across tiles without memory bottlenecks. COMB decouples context storage from computation, utilizing a dynamic retrieval mechanism to fetch feature representations from adjacent tiles. This enables a retrieval-based context integration strategy that adopts local padding to resolve spatial discontinuities and neighbor-aware channel attention to stabilize statistical drift. Further optimized with a sliding window schedule to ensure minimal memory overhead, our method demonstrates superior performance over state-of-the-art baselines, achieving significant improvements in both perceptual fidelity and tiling consistency, while suggesting its downstream utility in tumor segmentation. Code is available at this https URL.
[CV-82] Scale-Aware 3D Deep Learning for Robust Brain Metastasis Detection in Multimodal MRI
链接: https://arxiv.org/abs/2609.10825
作者: Sylvain Jaume,Hongming Wang,Simon K. Warfield
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 12 pages, 3 figures, 3 tables
Abstract:Detecting brain metastases in magnetic resonance imaging (MRI) remains challenging because lesions vary widely in size and appearance, with very small metastases occupying only a minute fraction of a three-dimensional input. We investigate whether combining different spatial fields of view (FOVs) improves lesion detection in multimodal MRI and present a scale-aware 3D deep-learning framework. The method uses independently trained 96^3 and 64^3 3D U-Nets whose whole-volume probability maps are combined by weighted late fusion. This design allows us to study the effect of spatial context separately from image resolution and modality choice. On a 97-patient development cohort, cross-FOV fusion improved lesion-level precision and F1 while substantially reducing false positives relative to the individual models. A same-FOV ensemble control showed that these gains were not explained solely by averaging independently trained networks, supporting a contribution from complementary spatial context. An exploratory cross-FOV agreement filter reduced false positives but did not improve overall F1. These results support cross-FOV probability fusion as a simple and computationally practical strategy for improving the precision-false-positive trade-off in 3D brain-metastasis detection.
人工智能
[AI-0] GPU-CFR: 80x Faster Counterfactual Regret Minimization by Compiling the Game to Static Dataflow and CUDA Graph Replay
链接: https://arxiv.org/abs/2609.11923
作者: Boning Li,Longbo Huang
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Artificial Intelligence (cs.AI); Computer Science and Game Theory (cs.GT); Mathematical Software (cs.MS); Programming Languages (cs.PL)
备注:
Abstract:Counterfactual regret minimization (CFR) is one of the few large numerical workloads that still runs faster on CPUs than on GPUs. Each iteration sweeps a game tree with up to billions of states in millions of small, interdependent gather and scatter steps issued through a generic tree interface. On a GPU every kernel finishes in microseconds, so kernel launches and framework dispatch dominate the run time, and prior GPU implementations have lost to optimized CPU code. We observe that for a fixed game, everything about a CFR iteration except the numerical values is known before the first iteration runs. We propose GPU-CFR, a compiler and runtime built on this observation. It compiles any game once into static dataflow: flat edge and information-set arrays, precomputed indices, and depth-level batched passes fix the entire operation sequence, and only solver state changes between iterations. Static chance folding, depth-level execution blocks, and a dual-lane reach buffer cut the number of framework operations by up to 18.1x. Because shapes, indices, and buffer addresses never change, CUDA Graph Replay records the iteration once and replays it with a single graph launch. On one A100, across an eight-game suite that spans card games, dice games, and board games, GPU-CFR runs 29.8–80.4x faster than the fastest prior GPU CFR on the same accelerator, and 14–258x faster than LiteEFG, one of the fastest open-source CPU implementations, on the four largest games. The compiled representation carries most of that margin: on eight CPU threads with no accelerator it is already 2.2–51.1x faster than the GPU baseline. On the CPU the optimized path reproduces the reference iterates bitwise, and tree construction and graph capture pay for themselves within the first solve. GPU-CFR beats every CPU and GPU baseline on the mid-to-large games of the suite without changing the update rule.
[AI-1] General Quantification of Covariate and Concept Shifts ICML2026
链接: https://arxiv.org/abs/2609.11918
作者: Hongbo Chen,Li Charlie Xia
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
备注: 38 pages, 9 figures, accepted at the 43rd International Conference on Machine Learning (ICML 2026)
Abstract:Generalization under distribution shift remains a core challenge in modern machine learning, yet existing learning bound theory is limited to narrow, idealized settings and is non-estimable from samples. In this paper, we bridge the gap between theory and practical applications. We first show that existing definition of concept shift breaks when the source and target supports mismatch. Leveraging entropic optimal transport, we propose a key notion: \gamma^! -concept shifts, and derive a general error bound unifying covariate and \gamma^! -concept shifts, which applies to broad loss functions, label spaces, and stochastic labeling. We further develop estimators for these shifts with concentration guarantees, and the DataShifts algorithm, which can quantify distribution shifts and estimate the error bound in most applications - a rigorous and general tool for analyzing learning error under distribution shift.
[AI-2] Can Edge-Deployable Vision-Language Models Identify Species?
链接: https://arxiv.org/abs/2609.11916
作者: William Zhou,Mayukha Siripuram,Xiao Yan,Ziqi Liu,Yi Ding
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Camera traps often run in the field on edge hardware with limited or no connectivity, making small, locally-deployable vision-language models (VLMs) – not frontier-scale ones – the practically relevant class to evaluate for species identification. We test whether models in this deployment-relevant 2–8B range carry genuine taxonomic knowledge, evaluating four such VLMs (Qwen3-VL 2B/4B/8B, Gemma3 4B) against the domain-specific specialist BioCLIP (300M parameters) on a 96-species task, comparing clean iNaturalist photographs against camera-trap imagery from 6 this http URL collections, on two independently-sampled evaluation sets. All models identify species far above chance, but every model – general-purpose or specialist – degrades sharply on field imagery (domain gaps of 9.6–26.6 percentage points, consistent across taxonomic levels and both evaluation sets), indicating the degradation reflects general image legibility rather than fine-grained discrimination failure. BioCLIP substantially outperforms every VLM tested (by 33.2–59.2 percentage points across an expanded 200-image sample for every model) despite its far smaller size, suggesting the gap reflects specialized training data rather than model scale; yet BioCLIP’s own domain gap (18.0 points) is statistically indistinguishable from the best VLM’s (22.3 points), suggesting the clean-to-field degradation itself is a property of the image-quality shift rather than a general-purpose-model weakness. Under open-set prompting, 5.9–9.6% of responses are syntactically valid but taxonomically nonexistent species names; the relative fabrication-rate ranking across models replicates exactly across both evaluation sets, a more robust finding than any single point estimate.
[AI-3] Artificial Id: Drive and Persistent Alignment in Agent ic AI
链接: https://arxiv.org/abs/2609.11911
作者: Yakov Pyotr Shkolnikov
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Agentic AI is moving from bounded task execution toward systems that retain consequential state, continue operating and adapt across task boundaries. That shift creates a control problem that current harnesses largely solve by hand: objectives, retries, verification, stopping rules and other behavioral transitions are specified externally. We propose an artificial id, an adaptive internal drive for determining whether behavior should continue, stop or change. In a minimal virtual Petri-dish experiment, a controller too small to perform general-purpose reasoning and receiving no task-specific behavioral objective develops useful control through differential persistence. The same mechanism selects an unintended physical strategy when that behavior persists better and later replaces a learned sensor mapping when its environmental meaning changes. These results show that adaptive direction can emerge without being explicitly specified as a behavioral objective. The same persistence that makes such adaptive agency useful can also allow misalignment, corrupted state and unintended behavior to persist across task boundaries. A scalable artificial id would carry consequential state and adaptive drive across those boundaries, making alignment a property of the continuing agentic system rather than of a model response or single trajectory. Such systems require a persistent alignment boundary over trusted observations, consequence channels, persistent state, authority, identity, provenance and hard constraints.
[AI-4] On the Regularization Landscape for the Linear Recommendation Models
链接: https://arxiv.org/abs/2609.11876
作者: Dong Li,Zhenming Liu,Ruoming Jin,Hao Zhou,Zhi Liu,Jing Gao,Bin Ren
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Recently, a wide range of recommendation algorithms inspired by deep learning techniques have emerged as the performance leaders on several standard recommendation benchmarks. While these algorithms were built on different DL techniques (e.g., dropouts, autoencoder), they have similar performance and even similar cost functions. This paper studies whether the models’ comparable performance are sheer coincidence, or they can be unified under a single framework. We find that all linear performance leaders effectively add only a nuclear-norm based regularizer, or a Frobenius-norm based regularizer. The former ones possess a (surprising) rigid structure that limits the models’ predictive power but their solutions are low rank and have closed form. The latter ones are more expressive and more efficient for recommendation but their solutions are either full-rank or require executing hard-to-tune numeric procedures such as ADMM. Along this line of finding, we further propose two low-rank, closed-form solutions, derived from carefully generalizing Frobenius-norm based regularizers. The new solutions get the best of both nuclear-norm and Frobenius-norm world.
[AI-5] Explainability Assistant: A Conversational XAI Interface for Interpreting Energy Consumption Models CEC
链接: https://arxiv.org/abs/2609.11860
作者: Rodion Krjutškov,Eduard Barbu,Nikos Sakkas,Sofia Yfanti
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 11 pages, 3 figures. Accepted author version of a paper published at ICECET 2026
Abstract:Energy consumption forecasting relies on increasingly complex machine learning (ML) models, such as Genetic Programming-based symbolic regressors, whose predictions can be difficult for facility managers and building operators to interpret. Explainable Artificial Intelligence (XAI) techniques address this opacity, but traditional XAI dashboards require substantial technical expertise and provide limited flexibility for dynamic, context-aware inquiry. Conversational XAI systems offer a promising alternative; however, previous approaches, such as TalkToModel, were constrained by rigid custom grammars and achieved only 76.8% intent-parsing accuracy. This paper introduces the Explainability Assistant, an open-source conversational XAI system that leverages the function-calling capabilities of modern Large Language Models (LLMs) to overcome these limitations. The system achieves 94% intent-parsing accuracy, supports flexible natural language interaction, and adapts to different ML problem types without task-specific fine-tuning. We present the system’s architecture and report results from a comparative evaluation conducted with energy domain specialists, contrasting the Explainability Assistant with a traditional XAI dashboard. The evaluation suggests improved usability and consistent task accuracy, with all experts unanimously preferring the conversational interface for practical use.
[AI-6] From Parameters to Answers: How LLM s Retrieve and Use Their Internal Knowledge
链接: https://arxiv.org/abs/2609.11859
作者: Wenkang Wei,Yuan Fang,Renhe Jiang,Hong Cheng,Xingtong Yu
类目: Artificial Intelligence (cs.AI)
备注: 53 pages, 13 figures, including appendices
Abstract:How does a language model’s dependence on query-routing information and target knowledge change as it answers a question? We study this question through layerwise interventions on the hidden state at the end of the question. Across Qwen, Llama, and Gemma, we compare country-continent questions with noun, adjective, and code answers while keeping several fitted measurements distinct. A pair-conditioned request direction describes which country is queried in natural single-country questions; a global request direction describes first- versus second-country requests in paired questions; separate selection candidates test control among contents already available in the hidden state. A diagnostic reanalysis of frozen Qwen natural-question states shows that the pair-conditioned direction grows stronger before interventions on it begin to alter later fitted knowledge, with this causal window opening while answer-supporting content is still forming. The paired three-model trajectories are not uniform: Gemma shows a partially overlapping mid-layer routing-content profile, whereas Llama has no sustained routing-effect window under the same gates. In the paired protocol, dependence on the global request direction decreases from fixed earlier to later layer sets while dependence on fitted content persists. A matched Qwen comparison shows that the pair-conditioned direction retains a late effect, so this operational handoff concerns the global fitted direction rather than all request information. These results separate early readability, natural strength, causal steering, and later content dependence.
[AI-7] Model-Aware Schedules Improve Generation via Fiberwise Optimal Transport ATC
链接: https://arxiv.org/abs/2609.11842
作者: Luyi Jia,Boyan Zhang,Yilun Liu,Steffen Rulands
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 11 pages, 2 figures. Keywords: Diffusion models; flow matching; schedule optimization; fiberwise optimal transport; time reparameterization; empirical universality
Abstract:Diffusion and flow-matching schedules control the signal and noise coefficients that mix data and noise along affine probability paths. Minimizing a kinetic action defined on coefficient paths, motivated by optimal transport, helps explain strong baselines but remains model-agnostic and ignores prediction error. Here we introduce a model-aware schedule construction based on fiberwise optimal transport. At a fixed time and state on the probability path, compatible signal/noise decompositions form an affine fiber. We define a fiberwise prediction risk by averaging optimal-transport costs between the true and predictor-induced decompositions within these fibers. On a fixed coefficient curve, combining this risk with coefficient-path kinetic action yields a closed-form optimal time allocation. This construction extends to general linear prediction targets, and the risk profile can be estimated from an early baseline checkpoint. We evaluate DDPMs and flow matching across prediction targets, training configurations, risk-estimation checkpoints, datasets, and architectures. Our model-aware schedules consistently outperform strong baselines, including a 38.6% relative FID reduction for flow matching on CIFAR-10 at 16 function evaluations. Each model-agnostic kinetic baseline determines its own kinetic reference coordinate. In these coordinates, fiberwise-risk profiles from independently trained models in different settings align closely after normalization to unit area. The resulting schedule deformations used in training also align, suggesting empirical universality across the evaluated models and settings. Pretrained-checkpoint diagnostics extend this normalized-risk agreement to larger conditional latent diffusion and 2-RF models. A frozen analytic allocation template retains most of the model-aware improvement without further risk estimation or model-specific fitting.
[AI-8] hinking with Looped Flows
链接: https://arxiv.org/abs/2609.11801
作者: Ayhan Suleymanzade,Chanhyuk Lee,Floor Eijkelboom,Nicholas M. Boffi,İsmail İlkan Ceylan,Jinwoo Kim
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Humans and machines often solve harder problems by spending more time on computation. In deep learning, looped models implement this idea during inference by recurrently updating a hidden state. In practice, however, their training backpropagates through only one or a few updates, making it hard to train early updates to support future ones. We propose looped flows, an approach that sidesteps this issue by training the recurrence with local denoising objectives. By imposing temporal association across denoising objectives through progressively decreasing noise levels and shared noise, the model is incentivized to learn recurrent states that transfer useful computation over time, even when gradients cover only a few updates. We then formulate inference as integrating the velocity of a probability flow parameterized by the learned denoiser, coupled with recurrent states. This allows solving harder problems by spending more computation through a finer temporal grid and enables multiple valid predictions from different initial noise samples. Across six reasoning benchmarks including two multi-solution benchmarks, looped flows outperform prior state-of-the-art looped models overall, achieving 58.8% test accuracy on ARC-AGI-1 and 12.2% on ARC-AGI-2.
[AI-9] Continuous-Time Acoustic Modelling with Neural Controlled Differential Equations
链接: https://arxiv.org/abs/2609.11725
作者: Mattias Cross,Minghui Zhao,Anton Ragni
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注: Accepted to IEEE Spoken Language Technology Workshop (SLT) 2026
Abstract:Text-to-speech (TTS) models commonly address text–speech alignment by expanding phone-level encoder states to frame-level decoder inputs using predicted durations. While this length-regulation step resolves alignment structurally, this use of duration typically changes only where and how often latent states appear, not the values of the states themselves. This paper proposes a continuous-time mechanism for duration-aware acoustic modelling in TTS using neural controlled differential equations (CDEs). We formulate the phone representation as a temporally parameterised control path and use a neural acoustic vector field to produce a continuous-time hidden state whose values evolve with phonetic content and duration-derived timing. The resulting trajectory can be sampled at discrete points and integrated into a standard acoustic decoder pipeline. Objective results contrast CDEs and typical recurrent models. Subjective results suggest that CDE-based models evaluating one phone per step can improve rank-order agreement between synthesised and reference emotion intensity while maintaining comparable emotion-expression quality to a strong baseline. Additional experiments with half-phone step-sizes suggest that temporal resolution changes the trade-off between style tracking and absolute calibration. These results position CDEs as a promising design space for continuous-time and duration-aware style-sensitive TTS.
[AI-10] A Time-Based Readout for Vector-Matrix Multiplication in Fully Analog Memristive SNNs CEC
链接: https://arxiv.org/abs/2609.11713
作者: Elia Mateu-Barriendos,Álvaro Gómez-Pau,Josep Rius,Daniel Arumí,Rosa Rodríguez-Montañés,Salvador Manich
类目: Emerging Technologies (cs.ET); Artificial Intelligence (cs.AI); Hardware Architecture (cs.AR)
备注: Accepted at 2026 IEEE 33rd International Conference on Electronics, Circuits and Systems (ICECS)
Abstract:Artificial neural networks rely on vector-matrix multiplications (VMMs), whose implementation in von Neumann architectures is dominated by costly data movement between memory and processing units. Spiking neural networks (SNNs) mitigate this bottleneck by performing in-memory, analog VMMs using memristive crossbar arrays. However, conventional current-mode readout circuits incur significant area and power overhead. This work proposes a fully analog readout architecture based on voltage-to-time conversion of the VMM output. By sensing the column voltage, the proposed approach avoids current-mode summing and scaling circuitry, improving area and energy efficiency. Post-layout simulations of a 10x1 SNN implemented in a 130 nm CMOS technology validate the proposed architecture, while application to a trained 64x10 SNN for digit classification further demonstrates its feasibility for SNN inference. Comments: Accepted at 2026 IEEE 33rd International Conference on Electronics, Circuits and Systems (ICECS) Subjects: Emerging Technologies (cs.ET); Artificial Intelligence (cs.AI); Hardware Architecture (cs.AR) Cite as: arXiv:2609.11713 [cs.ET] (or arXiv:2609.11713v1 [cs.ET] for this version) https://doi.org/10.48550/arXiv.2609.11713 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-11] ActSafeGuard: Differentiable and Training-Aligned Constraint Enforcement for Flow-Matching Policies
链接: https://arxiv.org/abs/2609.11697
作者: Jianming Ma,Rongjun Jin,Xiaxi Si,Yang Zhang,Yiheng Li,Yue Gao
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: 8 pages, 4 figures
Abstract:Vision-Language-Action (VLA) and World-Action Models (WAMs) have demonstrated strong capabilities in general-purpose robotic manipulation, yet their generated actions may violate hard physical constraints and therefore be unsafe or infeasible for deployment. Existing safety approaches either optimize statistical safety objectives without deterministic per-step guarantees or correct unsafe actions only during inference, creating a mismatch between policy training and execution. We introduce ActSafeGuard, a differentiable and training-aligned safeguard layer for flow-matching based policies. ActSafeGuard integrates hard action feasibility into policy learning, not merely treating safety as an inference-time external component. Through an analytical ray-scaling operator design, ActSafeGuard enables boundary-aware gradients to guide the model to naturally learn constrained manifolds. Extensive experiments on multiple standard foundation backbones ( \pi_0.5 and Fast-WAM) across various tasks demonstrate that ActSafeGuard consistently achieves a 100% step safety rate while fully preserving or even boosting task success rates, providing a scalable and minimally invasive solution for safe embodied AI deployment.
[AI-12] COBRA-Skills: Contextual Bandit-Guided Evolution for Agent Skill Optimization
链接: https://arxiv.org/abs/2609.11682
作者: Pingchen Lu,Xiangyi Wang,Xiang Li,Jie Mao,Zikun Qu,Junfeng Luo,Yao Shu,Bryan Kian Hsiang Low,Zhongxiang Dai
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language model (LLM) agents can benefit from reusable skills distilled from prior task experience, yet existing skill optimization methods often rely on costly execution-based evaluation and substantial task data. We introduce \textbfCOBRA-Skills, an efficient framework that formulates skill optimization as budgeted sequential optimization over a dynamically evolving candidate space. COBRA-Skills couples contextual-bandit-guided prioritization with evidence-grounded skill evolution, selectively allocating evaluations to promising or informative candidates while continually refining the skill population from execution feedback. Across six heterogeneous agent benchmarks and three target models, COBRA-Skills consistently achieves the strongest average performance among compared methods, while reducing optimization cost by 55–58% relative to SkillOpt and using only 50 unique optimization examples per benchmark. Further analyses show that COBRA-Skills remains robust to changes in the agent harness and performs effectively when the target model itself is used for skill generation and refinement.
[AI-13] Ecdysis: Efficient and Effective Training of Runtime Harnesses for LLM Agents
链接: https://arxiv.org/abs/2609.11677
作者: Ruiqing Yue,Yu Cui,Zhuoyu Sun,Sicheng Pan,Xianhong Xue,Tingyu Li,Ting Li,Wenzhuo Zhu,Yi Chen,Yifei Liu,Baohan Huang,Zhe Cui,Haibin Zhang,Cong Zuo
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:Self-evolving runtime harnesses can substantially improve the capabilities of large language model (LLM) agents and provide a promising paradigm for optimizing agent execution. Existing harness evolution methods typically rely on iterative search, repeatedly evaluating and revising candidate harnesses based on execution feedback from task instances. While this paradigm enables continuous harness optimization, it incurs substantial time overhead due to repeated agent executions and code modifications, and may overfit to observed tasks and specific failure patterns, resulting in degraded generalization to unseen tasks. We identify the lack of principled failure diagnosis as a key bottleneck in harness evolution: an observed failure can reflect either model-specific deficiencies or systematic harness deficiencies, and directly optimizing against individual failures can lead to unnecessary model-specific accommodation. We therefore propose Ecdysis, an efficient and effective framework that distinguishes model-specific accommodation from harness-level repair and biases adaptation toward systematic harness deficiencies by identifying recurring cross-task failure patterns. Ecdysis adopts a batch-level cross-instance failure aggregation paradigm to jointly analyze failure evidence from multiple task instances and further introduces Failure-Driven Collaborative Refinement to diagnose failure causes and iteratively refine harness modification specifications. By combining cross-instance failure analysis with multi-role diagnosis, Ecdysis enables more effective harness evolution with lower training time. Experiments show that Ecdysis achieves up to a 1.84x speedup in harness training compared with existing harness evolution methods, while improving the reasoning accuracy of the resulting harnesses by 18.56%.
[AI-14] Geospatial AI Dataverse Metadata and the Study of Place-Based Government
链接: https://arxiv.org/abs/2609.11674
作者: Danny EBanks,Devika Jain
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Harvard Dataverse hosts over 150,000 research datasets, but the geographic information those datasets carry is entered as free text by depositors and has never been assembled into a searchable structure. We construct a knowledge graph from the repository’s public data and metadata, organizing 102,650 datasets within a 215,985-node network of 528,003 edges linking datasets to keywords, publications, subjects, journals, and locations. Of those datasets, 43,991 (42.9 percent) carry at least one geospatial field, geographic coverage, geographic unit, or a bounding box and 96.9 percent of all nodes sit in a single connected component, so datasets remain reachable from one another even when their geospatial metadata share nothing in common. A conservative keyword search identifies 7,654 geospatially tagged datasets (17.4 percent) as directly policy-relevant, with elections and legislatures the largest cluster, followed by government administration, health policy, transportation, and education. Five datasets illustrate how this metadata behaves across policy domains and spatial scales, and an extended use case shows how community language models, stance detection with geographic aggregation, and partisan language bridging tools can attach discourse to place. The central obstacle is place resolution: the same location appears as many disconnected nodes. We argue that the graph provides a concrete setting for developing AI-driven metadata enrichment and entity resolution, and we document its coverage skew toward American, city-level data.
[AI-15] Warrant Theory
链接: https://arxiv.org/abs/2609.11667
作者: Khashayar Irani
类目: Logic in Computer Science (cs.LO); Artificial Intelligence (cs.AI)
备注: 17 Pages
Abstract:In this paper, we develop warrant theory as a philosophical discipline concerned with the inferential legitimacy of propositions within logical analysis. Warrant theory reconceptualises logic as a normative framework governing the conditions under which propositions may be introduced, accepted, rejected, and inferentially employed. Warrant is understood as inferential entitlement and is distinguished from truth, belief, and other psychological attitudes, while its relation to inferential use and meaning is examined. Warrant-theoretic analysis is then developed as a systematic method for investigating how propositions acquire inferential standing, how that standing develops, and how inferential positions interact through relations of dependence, compatibility, incompatibility, and exclusion. Acceptance and rejection provide the bilateral vocabulary for representing positive and negative inferential positions and the consequences and commitments associated with them. Finally, these elements are brought together in a warrant-theoretic definition of logic as the formal and normative study of the conditions under which propositions may be legitimately accepted or rejected and of the inferential transitions that such legitimacy warrants. On this account, logical consequence and logical failure are understood through the presence, preservation, or absence of inferential entitlement, thus locating the philosophical subject matter of logic in the systematic governance of inferential legitimacy.
[AI-16] Autonomy Social Norms and Alignment: Towards a Developmental Framework for Autonomous Artificial Agents
链接: https://arxiv.org/abs/2609.11660
作者: Marica Notte,Ludovica Marinucci,Vieri Giuliano Santucci
类目: Artificial Intelligence (cs.AI)
备注: In publication in the proceedings of SIpEIA 2026 conference
Abstract:In recent years, artificial intelligence has made extraordinary progress thanks to large-scale models capable of generalization and the generation of complex outputs. However, transferring this potential into embodied agents reveals a significant limitation: the most advanced systems rely on pre-existing datasets and human feedback strategies that are powerful but insufficient in dynamic or unknown contexts. To adapt, an agent must acquire knowledge through direct interaction with its environment. One strategy to address this challenge involves introducing higher-level mechanisms, such as intrinsic motivations, which leverage curiosity and competence, to guide exploration and learning in complex environments. While this flexibility expands autonomy, it complicates the task of ensuring agents remain aligned with human goals. Alignment, already a challenge for artificial systems in general, becomes even more complex in unstructured and dynamic contexts where predefined rules prove insufficient. To be effective and adaptable, norms must be rooted in experience through an epistemological process that starting from simple, situated principles allows for the gradual construction of more complex rules through experience, autonomous learning, and cooperation with other moral agents. Similarly to children learning social norms by exploring their environment and participating in collective practices, artificial agents must also be educated toward alignment. Following Dennett, the status of a moral agent is not innate but is attributed gradually based on the ability to responsibly manage increasing degrees of freedom. From this perspective, the regulatory sandboxes can be viewed as pedagogical environments for AI: dynamic spaces where alignment develops as a formative process, progressively shaping autonomous behaviors through interaction and cooperation in scenarios of increasing complexity.
[AI-17] ZipCodec: Ultra-Low-Frame-Rate Streaming Speech Coding
链接: https://arxiv.org/abs/2609.11642
作者: Luca Della Libera,Cem Subakan,Mirco Ravanelli
类目: ound (cs.SD); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 5 pages, 1 figure
Abstract:Neural audio codecs are a fundamental component of modern speech generation systems. While recent codecs achieve increasingly low bitrates, reducing frame rate remains challenging, as each token must preserve more information while maintaining reconstruction quality. We present ZipCodec, a streaming neural speech codec operating at 6.25 Hz and 0.80 kbps with a theoretical latency of 160 ms. Our approach combines large-scale WavLM distillation with a redesigned transformer-based architecture, a scalar spherical quantizer, and a latency-aware streaming decoder. Experiments show that ZipCodec substantially outperforms existing streaming codecs at comparable bitrates in both reconstruction and downstream tasks, while operating at a significantly lower frame rate. Despite its 842M parameters, ZipCodec achieves real-time single-stream inference on a consumer-grade CPU. Demo samples, code and checkpoints are available at this https URL.
[AI-18] LoaDiff: Conditional Generation of Electricity Consumption Time Series for Energy Analytics ICDM2026
链接: https://arxiv.org/abs/2609.11639
作者: Mariia Baranova,Adrien Petralia,Etienne Le Naour,Nathan Etourneau,Guillaume Hofmann,Themis Palpanas
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Signal Processing (eess.SP)
备注: 10 pages, 5 figures. This paper appeared in IEEE ICDM 2026
Abstract:The energy transition is reshaping residential electricity consumption through the increasing adoption of distributed generation, electrified appliances, and demand-response programs. Understanding these evolving behaviors requires access to granular smart-meter data for applications such as load forecasting, appliance detection, and demand-side flexibility analysis. However, such data are subject to strict access restrictions and data-protection regulations. Thus, realistic synthetic alternatives are necessary. In this paper, we introduce LoaDiff, a diffusion-based generative model for year-long, sub-hourly smart-meter load curves. LoaDiff supports flexible conditioning on static household attributes, such as appliance ownership, and dynamic contextual variables, including calendar information and outdoor temperature. We evaluate the model against multiple generative baselines on three residential electricity-consumption datasets. Our experiments assess four complementary dimensions: fidelity and diversity, training-record memorization risk, downstream utility for load forecasting and appliance detection, and conditional controllability under alternative temperature conditions. The results show that LoaDiff generates realistic and diverse load profiles, achieves a favorable trade-off between generation quality and limited evidence of memorization, preserves information useful for downstream energy applications, and responds coherently to changes in conditioning variables.
[AI-19] MAPLE: Memory-Augmented Planning with Language and Evolution
链接: https://arxiv.org/abs/2609.11636
作者: Kesheng Chen,Yamin Hu,Wenjian Luo
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Domain practitioners understand their business constraints but may lack operations-research expertise or dedicated support. LLM-based optimization agents translate natural-language requirements into models or solver programs that established optimization tools can execute. This progress makes optimization more accessible, but real-world operations are dynamic: changing demand, resources, and priorities require updates to data, constraints, and objectives. Methods centered on isolated requests offer limited support for rapid adaptation that preserves earlier decisions and reuses useful search results. We introduce MAPLE (Memory-Augmented Planning with Language and Evolution), an agent for maintaining optimization problems through successive natural-language requests. MAPLE combines language-based problem construction with mathematical programming and evolutionary search. It retains the optimization program, accepted plans, earlier updates, and candidate solutions for subsequent requests. We introduce NLDO, a benchmark of 15 trajectories and 180 updates spanning selection, scheduling, rostering, routing, and cloud-resource placement. In the main evaluation, MAPLE completes all trajectories and achieves online scalar quality of 0.951 and a Pareto hypervolume ratio of 0.875. Controlled comparisons further show that maintaining executable state improves update validity and can preserve useful search information across substantial revisions.
[AI-20] Distributed Optimization of Modular Production Systems using Model-based Reinforcement Learning with Inverse Models
链接: https://arxiv.org/abs/2609.11615
作者: Andreas Schwung,Steve Yuwono,Sofiene Lassoued,Dorothea Schwung
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Systems and Control (eess.SY)
备注:
Abstract:This paper presents a novel approach for data-driven self-learning control of highly flexible, modular manufacturing systems. Specifically, we employ a novel framework for model-based reinforcement learning which introduces approximate inverse process models within the training of reinforcement policies. This approach disentangles the learning of actuation dynamics and the dynamics in state space, resulting in RL-based training solely within the task space. We propose a lightweight feedforward architecture for approximate inverse models and integrate them within the policy network of standard RL algorithms. We apply the approach to a laboratory modular production testbed with heterogeneous production modules. The results underline the efficiency improvements for modular manufacturing units in terms of both performance and training speed, particularly for off-policy algorithms.
[AI-21] Making Alternative Data Work: Context-Augmented LLM s for Financial Forecasting
链接: https://arxiv.org/abs/2609.11607
作者: Jihoon Kwon,Lawrence Liu,Daekyung Park,Sumin Kim,Haverty Jack,Hoyoung Lee,Katherine Bjorkman,Josh McKenney,Peter Laurelli,Nicole Kagan,Zach Golkhou,Thorsten Neumann,Edward Tong,Pete Petersen,Yoon Kim,Alejandro Lopez-Lira,Yongjae Lee,Chanyeol Choi
类目: Artificial Intelligence (cs.AI)
备注: 13 pages
Abstract:When forecasting a firm’s future financial performance, alternative data - data collected from non-traditional sources such as consumer transactions, web traffic, and prediction markets - can provide timely signals about firms’ operating activities and broader market conditions. These signals may reveal information that is not captured by traditional public sources and can therefore provide complementary information for forecasting firms’ future financial performance. However, firm-level alternative data often have limited historical coverage, are relevant only to specific prediction targets or subsets of firms, and are distributed across numerous heterogeneous channels, making them difficult to incorporate flexibly into conventional forecasting approaches. Meanwhile, large language models (LLMs) can interpret instructions, learn from in-context examples, and generate predictions by combining heterogeneous information without task-specific parameter updates. Motivated by this potential flexibility, we investigate whether an LLM can forecast firm performance by integrating alternative data with other financial information through in-context learning. We propose a two-agent framework that first identifies the firms for which each alternative data channel is likely to be informative and then predicts revenue using firm- and channel-specific context. We evaluate the framework across four commercial alternative data channels. In our experiments, adding alternative data in context alongside other financial information improves the LLM’s forecasting relative to either source alone, and these forecasts are more accurate than those of standard forecasting baselines. These findings suggest that LLMs provide a flexible and practical approach to integrating alternative data with heterogeneous financial information.
[AI-22] Enabling Knowledge Graph Understanding at Scale with the EXplore Your Graphs ENgine (EXYGEN)
链接: https://arxiv.org/abs/2609.11569
作者: Harshdeep Singh,Yurui Zhu,Giovanni Colavizza,Matteo Romanello
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:We present EXYGEN (EXplore Your Graphs ENgine), a framework for knowledge graph (KG) understanding that enables conversational access to KGs at scale. We address two questions in sequence. First, how effectively can LLMs perform text-to-SPARQL generation given only automatically derived structured metadata and small graph samples, rather than task-specific fine-tuning? We integrate VoID descriptions and ShEx schemas into a retrieval-augmented generation (RAG) pipeline and ablate KG-derived context on the SciQA benchmark. Our best configuration – combining ShEx schemas, retrieved triples, and example question-query pairs – reaches an exact match of 0.419 on execution results without any LLM fine-tuning. We further find that lexical metrics such as F1 poorly predict query correctness, and that larger general-purpose LLMs can outperform smaller code-specialized ones once given sufficient context. Second, we ask how to generate the structured metadata that this method relies on from very large KGs, where KG metadata generation becomes computationally intractable. We introduce a predicate-coverage-aware parallel graph sampling strategy that preserves structural diversity while remaining computationally tractable. On OpenCitations Meta and GESIS, it retains high predicate coverage with minimal triple loss and reduces runtime by over 80x; on ORKG, sampling is not just faster but the only tractable path to obtain complete metadata. Together, these results show that structured schema context and lightweight prompting can substantially reduce reliance on fine-tuning for scalable conversational access to KGs, though closing the remaining gap to fully fine-tuned approaches will likely require reducing dependence on curated question-query exemplars – whether through synthetic generation or an execution-feedback-driven approach – and validating these findings beyond a single benchmark.
[AI-23] Characterizing Job Power Elasticity for Power-Flexible AI Training
链接: https://arxiv.org/abs/2609.11542
作者: Philip Colangelo,Charles Dawson,Shayan Sengupta,Ayse Coskun,Varun Sivaram
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language model (LLM) training is among the fastest-growing sources of electricity demand in modern data centers, and power availability is a primary bottleneck to continued AI infrastructure growth. Making the power consumption of these workloads flexible could unlock additional power for AI growth, limit increases in electricity prices, and improve the utilization of existing grid infrastructure. However, to realize this flexibility, we must first understand how the performance of training workloads changes when GPU power is reduced. This paper presents the first systematic characterization of \emphjob power elasticity (the sensitivity of throughput to power reductions) in LLM training. To quantify elasticity, we introduce the \emphPower Flexibility Index (PFI), a normalized metric that quantifies the performance cost of power reductions and provides a control primitive for SLA-aware power flexibility. We collect data from 131 LLM training runs on H200 (plus 24 H200 validation runs and 34 matched H100 runs), including both dense and mixture-of-experts models, pretraining and fine-tuning tasks, and up to 32 GPUs. We find that LLM training jobs exhibit substantial but variable power elasticity, and we identify telemetry signals that predict PFI at runtime. Finally, we demonstrate that PFI-aware power allocation maximizes total tokens/second throughput under power constraints. Under a 30% power reduction, PFI-aware power allocation recovers ~1.5k tokens/s per job, 63% of the performance gap between an equal-weight allocation and an oracle with perfect information. Our results establish power elasticity as a measurable property of training jobs and provide a foundation for power-aware, grid-responsive AI infrastructure. Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2609.11542 [cs.AI] (or arXiv:2609.11542v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2609.11542 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-24] Prompt Revision as a Source of Cultural Bias in Text-to-Image Systems EMNLP2026
链接: https://arxiv.org/abs/2609.11532
作者: Aleksandra Urman,Elsa Lichtenegger,Salima Jaoua,Azza Bouleimen,Robin Forsberg,Corinna Hertweck,Stefania Ionescu,Nicolò Pagan,Ancsa Hannak,Joachim Baumann
类目: Artificial Intelligence (cs.AI)
备注: Accepted to EMNLP 2026
Abstract:Commercial text-to-image systems silently revise user prompts before generating images, a step users typically cannot disable or even see. Yet, existing audits of cultural bias examine only the final images and treat generation as a single pipeline, so they cannot tell where the bias originates. We introduce WORLDVIEW, a multilingual benchmark of 8,960 prompts across 15 languages and 31 language-context pairings. Using it, we audit the revision layer in three systems (DALL-E-3, Imagen-4, GPT-Image-1.5) through a three-step analysis of how heavily it marks each cultural context, whether it flattens that context into a narrow vocabulary, and whether that vocabulary is stereotypical. Relative to a no-context English baseline, the US is the least-marked context, while non-Western and non-Anglophone contexts are marked far more heavily, flattened into narrow vocabularies applied across topically diverse prompts, and reduced to recognizable cultural stereotypes. Comparing images from original versus revised prompts on models without a revision layer, we identify the layer itself as a previously undocumented, causal source of this stereotyping. To locate cultural bias, and fix it, we must audit the system as deployed, not the model alone.
[AI-25] Lightweight LiDAR-Based Cone Detection Framework Using Random Forest for Formula Student Driverless
链接: https://arxiv.org/abs/2609.11527
作者: Márk Mező-Kerekes,Péter Praksz,Chang Liu
类目: Artificial Intelligence (cs.AI)
备注: 9 pages, 2 figures, 3 tables. Accepted at the 5th International Conference on Cognitive Mobility (CogMob 2026)
Abstract:Reliable, low-latency perception is crucial for Formula Student Driverless vehicles, yet many existing pipelines rely on deep learning and multi-sensor fusion, often requiring GPU acceleration. This paper presents a lightweight LiDAR-only perception pipeline tailored for CPU execution, combining ground removal, IMU-based motion compensation, DBSCAN clustering, and geometric feature-based Random Forest classification. Feature importance analysis reduced the model input from 12 to 7 features while preserving performance. Evaluated on 2,371 labeled clusters collected from real FSD events, the pipeline achieves an F1-score of 98.33% and an end-to-end runtime of 3.13 ms on CPU-only hardware. The released dataset, labeling tool, and trained models provide a practical and reproducible baseline for other resource-constrained autonomous racing teams.
[AI-26] Extending SMT Solving with Non-Ground Clause Learning
链接: https://arxiv.org/abs/2609.11509
作者: Yasmine Briefs,Christoph Weidenbach
类目: Artificial Intelligence (cs.AI); Logic in Computer Science (cs.LO)
备注: Extended version of LPAR 2026 paper
Abstract:Quantifier instantiation is currently the main approach to non-ground SMT solving: solvers generate ground instances and solve the resulting ground SMT problems with CDCL(T)-style reasoning. When a conflict is found, conflict analysis learns only a ground clause, even though the conflict comes from instances of non-ground clauses. Yet non-ground reasoning can give exponentially shorter proofs than purely ground reasoning. We propose a calculus that consists of ground instantiations, CDCL(T)-style rules, and non-ground conflict analysis. The solver reasons on ground instances, but the resolution steps of conflict analysis are performed on their original non-ground clauses. This produces learned clauses that are typically more general than the ground conflict. With a suitable strategy, the learned clauses are even non-redundant. We also show how chronological backtracking can be included in SMT solving. Our calculus gives a common setting for CDCL(T)-style SMT solving, a range of instantiation-based procedures, and non-ground clause learning, and we prove that it simulates CDCL, SCL(FOL), SCL(T), and even Resolution.
[AI-27] ActMap: Single-Pass Uncertainty Quantification from Generation-Time Activation Maps
链接: https://arxiv.org/abs/2609.11498
作者: Jacopo Dardini(University of Bologna),Roberta Calegari(University of Bologna)
类目: Artificial Intelligence (cs.AI)
备注: 13 pages, 4 figures, 10 tables. Includes technical appendix
Abstract:Practical uncertainty quantification (UQ) for large language models must decide, from a single generation, whether a specific answer should be trusted. Existing methods either sample multiple generations, read only output-token probabilities, or reduce the model’s internal computation to a single hidden state. We introduce ActMap, a white-box representation that compresses the generation-time hidden- state trajectory (every layer, every generated token) into a fixed 12 \times 32 \times 128 tensor of temporal-statistic channels that preserves structure across transformer depth and pooled hidden coordinates. The map is captured during the generation pass with no measurable overhead, has a fixed shape across model depths and hidden sizes, and occupies 96 KiB: a compact artifact that can be retained for audit-relevant generations and probed directly, with occlusion analysis localizing the classifier’s signal to mid-depth regions of the map. A lightweight classifier, instantiated as a compact Vision Transformer, reads an estimated correctness probability from each map in a fraction of a millisecond; capacity-matched MLPs perform comparably, indicating the representation itself carries the result. Trained and evaluated in-domain on short-answer QA, direct- answer math, and summarization factuality with three instruction-tuned 7-8B models, ActMap consistently outperforms sampling, token-probability, attention, and embedding baselines, and matches ACT-ViT, a detector trained on dense activation tensors 67 \times larger, at essentially the same mean AUROC with lower calibration error on ten of twelve pairs. The resulting score supports abstention, routing, and selective verification from a single generation, making it a practical primitive for scalable oversight of deployed models. Comments: 13 pages, 4 figures, 10 tables. Includes technical appendix Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2609.11498 [cs.AI] (or arXiv:2609.11498v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2609.11498 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Jacopo Dardini [view email] [v1] Thu, 10 Sep 2026 13:04:24 UTC (536 KB) Full-text links: Access Paper: View a PDF of the paper titled ActMap: Single-Pass Uncertainty Quantification from Generation-Time Activation Maps, by Jacopo Dardini (University of Bologna) and 1 other authorsView PDFHTML (experimental)TeX Source view license Current browse context: cs.AI prev | next new | recent | 2026-09 Change to browse by: cs References Citations NASA ADSGoogle Scholar Semantic Scholar export BibTeX citation Loading… BibTeX formatted citation loading… Data provided by: Bookmark checked="checked"class=“labs-tab-input”> Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv’s community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) mathjaxToggle(); We gratefully acknowledge support from our major funders, member institutions, , and all contributors. About Help Contact Subscribe Copyright Privacy Accessibility Operational Status (opens in new tab) Major funding support from
[AI-28] Published Unlearning Numbers Move Per Checkpoint and Not Because the Removed Data Survives: An Audit of 263 Released Batch-Normalized Checkpoints DATE
链接: https://arxiv.org/abs/2609.11490
作者: Junlong Shen Xingyu Li
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 38 pages, 4 figures, 26 tables. Independent of and concurrent with arXiv:2609.08901 (posted 8 Sep 2026): the instrument and protocol here were pre-registered on 29 Aug 2026; dated provenance in Appendix S
Abstract:An unlearning audit reads its verdict off numbers that an unlearned model and its retrained reference each publish, and both also ship batch-normalization statistics that no gradient step wrote and no release records. Refitting them on kept data at bit-identical weights moves 47 of 221 released checkpoints past the spread their own release’s seeds show, several inside a method whose average does not move: what moves is the checkpoint’s property, not its method’s. What does the moving is not the removed data surviving in the state: exchanging kept records for removed ones inside a fixed fitting pool moves a published cell by almost nothing, while how far a checkpoint’s shipped state has drifted from any refit does track it. The consequence for a published decision is real but narrow: twelve verdicts cross, four clear a measured recalibration budget, two clear it on every replicate, and a population we trained and sited near its own criterion yields none. A release should therefore name the fitting convention beside the number, on the batch-normalized vision models where this channel exists.
[AI-29] Flexible and Interpretable Accent Distance Measurements
链接: https://arxiv.org/abs/2609.11458
作者: Charles McGhee,Mark J. F. Gales,Kate M. Knill
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Determining the differences between two speakers’ accents is a fundamental task in linguistics and speech technology research. The methodology used to measure these differences depends on the specific research area. A phonetics researcher may demonstrate accent variation by comparing vowel formants in paired recordings of individual words. These results will be interpretable, but the recordings will be time-consuming to collect and may not be representative of connected speech. Accented Text-to-Speech (TTS) research has pushed towards using accent embeddings derived from accent classification tasks. These embeddings can be produced from any speech recording, but are not readily interpretable. In this paper, we demonstrate that articulatory representations created through articulatory inversion can be used as an interpretable basis for accent comparison and that optimal transport provides a framework for accent comparison across arbitrary recording types.
[AI-30] RouteRepair: Instance-Level Failure Diagnosis and Targeted Repair in LLM -Based Automated Heuristic Design for Routing Optimization
链接: https://arxiv.org/abs/2609.11452
作者: Binghao Ji,Di Huang,Jiahui Fang,Zhiyuan Liu
类目: Artificial Intelligence (cs.AI)
备注: 22 pages, 13 figures, 11 tables
Abstract:Efficient routing optimization is essential to freight transportation, urban logistics, and shared mobility, where high-quality heuristics are often required under limited computational budgets. Recent large language model (LLM)-based automated heuristic design methods can generate effective routing rules, but aggregate evaluation may mask recurrent failures on particular instance structures. To address this limitation, this study develops RouteRepair, which diagnoses parent-specific weaknesses from instance-level performance and applies targeted modifications to the corresponding heuristic components while protecting behavior that already performs well. Routing evidence, solver behavior, and program context are combined to define bounded repair objectives, and each intervention is validated through matched parent-child evaluation of failure recovery and collateral degradation. Experiments on the traveling salesman problem (TSP) and capacitated vehicle routing problem (CVRP) span constructive search, guided local search, and ant colony optimization. RouteRepair-GLS reduces the mean TSP optimality gap from 1.7476% to 0.7587%, while the constructive CVRP heuristic lowers average route cost by 1.91% relative to the savings heuristic; the generated ACO priors also outperform matched hand-designed priors. These results show that failure-aware, evidence-constrained refinement can improve routing heuristics on difficult instances while preserving performance on cases they already solve well.
[AI-31] Prevalence Determines Precision:Silent Contamination in Detector-Defined Datasets
链接: https://arxiv.org/abs/2609.11449
作者: Jia Huang,Yankai Wan,Yangjun Ou
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 9 pages
Abstract:Many ML datasets are constructed by running a detector, heuristic, or model over candidate pools; accepted items become labels. Dataset precision is then governed by true-positive prevalence in each pool via Bayes, not solely by detector quality. Using one instrument and period, we hold a detector-defined event dataset plus an independent official index labeling every detected item as real or phantom. One detector, three pools yield phantom rates 81.7%, 9.0%, and 0.0%. Transferring precision from the two high-rate pools to the low-rate pool predicts 0.955 versus measured 0.183, a +422% error; the Bayes expression predicts all three within 3.3%. The detected response curve is an exact convex combination of a true-event and a phantom component (residual 1.1e-16), with phantoms outnumbering true events 473 to 308, so contamination is a second signal with detector-inherited shape, not additive noise. Contamination direction depends on the estimator: on identical windows one statistic is diluted and another inflated because its denominator is also contaminated. A common normalization turns the estimator into a mean of ratios whose expectation need not exist; on the same 335 events it returns 0.40 where the well-defined estimator returns 0.10.
[AI-32] Calibration-Aware Uncertainty Cascades for Efficient Heterogeneous Model Collaboration
链接: https://arxiv.org/abs/2609.11446
作者: Yilin Zhang,Han Jiang,Cai Xu,Ying Liu,Wei Zhao
类目: Artificial Intelligence (cs.AI)
备注: 13 pages, 6 figures, 6 tables, including appendix. Under review
Abstract:Heterogeneous model collaboration seeks to exploit the complementary strengths of different models to balance predictive performance and inference cost. Existing approaches typically rely either on trained routers, which tie routing decisions to a fixed task and model pool, or on raw-confidence cascades, whose thresholds lack consistent reliability semantics across heterogeneous models. Consequently, these approaches adapt poorly to changing model pools and deployment budgets. We propose Calibration-Aware Uncertainty Cascades (CAUC), a simple post-hoc framework that independently calibrates each model’s confidence and selects deployment policies using validation data. The resulting calibrated confidence scores establish a common reliability scale for accepting an early prediction, invoking a stronger model, or selectively combining model outputs. This unified decision criterion decouples deployment policies from any particular model pool or operating budget. We further show theoretically that calibration gives confidence thresholds an explicit selective-risk interpretation, whereas uncalibrated scores offer no comparable reliability guarantee. Extensive experiments demonstrate that, across six language benchmarks, CAUC achieves an average relative accuracy improvement of 1.9% over strong-model-only inference while avoiding approximately 47% of strong-model calls. On image classification benchmarks, it maintains or improves predictive performance while reducing measured GFLOPs by up to 57%.
[AI-33] LLM s as Post-hoc Auditors of Physiological Plausibility in Symbolic Regression: A Clinician-Evaluated Case Study
链接: https://arxiv.org/abs/2609.11431
作者: Jorge López-Varela,J. Ignacio Hidalgo,José-Manuel Muñoz,Omar Costilla-Reyes,Esther Maqueda,Jesus Moreno-Fernandez,Tomás González-Vidal,J. Manuel Velasco,Oscar Garnica
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Genetic Programming and its variants, such as grammatical evolution, are widely used in Symbolic Regression to derive mathematical expressions from multivariate data. In addition to predictive accuracy, models are appreciated for their potential to provide interpretability, offering explicit equations that relate input variables to outcomes. However, achieving interpretability and plausibility remains challenging, as evolved models may be complex or scientifically inconsistent. In this study, we explore whether Large Language Models, can assist in improving the explainability of Symbolic Regression models generated by evolutionary computation methods. Building upon our previous work on estimating body fat percentage using grammar-based Genetic Programming , we investigate the use of LLMs as post-processing tools to analyze and rank evolved expressions according to their interpretability and medical plausibility. Four symbolic expressions are analysed by three LLMs over three repeated runs, and the resulting interpretations and rankings are assessed by a panel of three clinicians. Across the three LLMs, comparative model-ranking outputs received more favorable clinician assessments than isolated term-level interpretations. However, the LLMs also produced physiologically and mathematically questionable explanations, indicating that they are better suited to comparative auditing under expert oversight than to autonomous validation.\blfootnoteThe present work is an extended version of a paper submitted into a journal.
[AI-34] X-AuT: Progressive Audio-Encoder Compression for Speech LLM s with Cross-Scale Distillation
链接: https://arxiv.org/abs/2609.11412
作者: Haojun Zhang,Yi Zou,Min Chen,Qize Yu,Lianrui Fan,Xini Ding,Hao Li,Shuchang Zhou,Xianming Liu,Shiyu Huang
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注:
Abstract:Reducing audio-encoder depth lowers the inference cost of speech large language models, but removing complete blocks perturbs the embeddings consumed by the decoder and can cause deletion and premature end-of-sequence errors. We introduce X-AuT, a progressive framework that selects layer combinations through short behavioral probes and restores the pruned model through representation alignment, cross-scale distillation, scheduled student-policy supervision, and LoRA finetuning. The language-model backbone remains frozen, while attention LoRA adapters and the tied output embedding adapt during distillation. Training uses the highest-agreement tier from a transcript-consistency pipeline, followed by source reweighting during finetuning. On ten public Chinese–English benchmarks, compressing Qwen3-ASR-0.6B from 18 to 16 audio-encoder layers reduces macro-average error from 5.61% to 5.27%. The 14-layer model reaches 5.75% with 20.7% fewer audio-tower parameters. Under the matched recipe, the 1.7B teacher yields 5.55% mean error, compared with 8.45% for self-distillation, and progressive 18 \rightarrow 14 pruning outperforms direct pruning (5.75% vs. 6.73%). These single-run results establish two practical operating points and show that the accuracy effects vary across benchmarks. Project website: this https URL
[AI-35] Deep-Fake CAPTCHA: Mitigating Next-Generation Social Engineering Attacks CCS
链接: https://arxiv.org/abs/2609.11404
作者: Guy Frankovits,Lior Yasur,Fred M. Grabovski,Yisroel Mirsky
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: Expanded work from the original ASIA CCS paper on DF-CAPTCHA (now evaluates video deepfakes too)
Abstract:This paper presents DF-CAPTCHA, an active defense against real-time deepfake impersonation in voice and video calls. Instead of passively searching for artifacts, DF-CAPTCHA prompts the caller to perform simple challenge-response tasks that are easy for humans but difficult for current real-time deepfake systems to generate convincingly. The framework verifies the response using four criteria: realism, identity consistency, task completion, and response time. We evaluate the approach across both audio and video modalities using user studies and experiments with real-time deepfake models. Results show that people often struggle to distinguish real-time deepfakes from authentic media, while DF-CAPTCHA substantially improves detection performance over passive methods, reaching high accuracy in both modalities. These findings suggest that active challenge-based verification is a practical and robust defense against next-generation social engineering attacks based on real-time deepfakes.
[AI-36] From Queries to Narratives: Cultural Heritage Data Stories for Knowledge Graph Exploration and Quality Assessment
链接: https://arxiv.org/abs/2609.11403
作者: Tabea Tietz,Torsten Schrade,Etienne Posthumus,Linnaea Söhn,Jonatan Jalle Steller,Jörg Waitelonis,Harald Sack
类目: Artificial Intelligence (cs.AI); Digital Libraries (cs.DL)
备注:
Abstract:Cultural-heritage KGs such as the NFDI4Culture-KG contain millions of triples about artworks, music, inscriptions, historical events, and the people and places connected to them. For many users, however, discovering this knowledge can be difficult. While SPARQL can be learned, writing meaningful queries first requires an in-depth understanding of the graph’s data model, an investment many domain researchers and practitioners are unwilling to make. Even with existing user interfaces, a starting point and some guidance are usually needed, because the data contained in the graph is highly specialized, heterogeneous, and constantly growing, making it challenging to know what it contains or which questions it can answer. In this paper, we present data stories as a way not only to lower this barrier, but also to turn exploration into data-quality assessment, and thus combine accessible querying with the discovery of issues that remain hidden in aggregate statistics. In this contribution, a data story is understood as a narrative document that integrates explanatory text and images with executable SPARQL queries and their visualized results. It is described how they are authored against the graph and how they serve several purposes: guiding users through an unfamiliar graph, creating reproducible narratives, and surfacing data-quality issues previously hidden in aggregate statistics. The authoring platform LODEON including its Sparnatural and AI-supported authoring assistants is introduced as a proof-of-concept. Within the authoring environment, every claim made about the data can be backed by an explicit query, making these narratives transparent and reproducible. This paper also reflects on lessons learned from hands-on seminars and workshops. Early experience suggests that such data stories make cultural-heritage knowledge graphs more accessible for both exploration and quality assessment.
[AI-37] Beyond Confidence: Stability-Aware Test-Time Adaptation for LLM Reasoning
链接: https://arxiv.org/abs/2609.11393
作者: Bincheng Gu,Min Gao,Zongwei Wang,Yibing Bai,Yulan He,Junliang Yu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Test-time adaptation has emerged as a lightweight alternative to costly post-training for improving the reasoning capabilities of Large Language Models (LLMs) on downstream tasks. Predictive entropy provides a model-derived signal for such adaptation, guiding models toward higher-confidence reasoning states without external verifiers or reward models. However, higher confidence does not necessarily imply correctness, as LLMs may remain highly confident along incorrect reasoning trajectories. We observe that high-confidence reasoning is more likely to be correct when confidence remains stable under local perturbations. Based on this observation, we propose Test-Time Adaptation via Stability-Aware Confidence Optimization (TASCO), a framework that incorporates local stability into confidence-based test-time adaptation while keeping the LLM frozen. TASCO operationalizes local stability by optimizing a lightweight task-level prefix under two alternative perturbation strategies: Random Perturbation promotes distributional stability across trajectories induced by nearby perturbed prefixes, whereas Sharpness-Aware Perturbation targets worst-case local sensitivity. Experiments demonstrate that TASCO improves reasoning accuracy and token efficiency across diverse LLMs and reasoning benchmarks, while behavioral analyses show that it maintains stable confidence under local perturbations without prematurely concentrating the model’s predictive distribution.
[AI-38] Buyer Artificial Intelligence-Enabled Environmental Governance and Supplier Environmental Controversies: An Organizational Information Processing and Signaling
链接: https://arxiv.org/abs/2609.11391
作者: Yongchao Martin Ma,Xinya Guan
类目: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Applications (stat.AP)
备注:
Abstract:Environmental controversies in global supply chains pose significant risks for global buyers. This study examines whether overseas suppliers’ exposure to buyers’ artificial intelligence (AI)-enabled environmental governance reduces supplier environmental controversies. Drawing on organizational information processing theory and signaling theory, we investigate how suppliers’ exposure to AI-enabled governance influences their environmental controversies and the institutional contingencies under which this effect varies. Using text analysis to measure buyer AI-enabled environmental governance, we analyze panel data on 2,505 suppliers of U.S.-listed firms across 41 countries from 2020 to 2024 with multidimensional fixed-effects models. We find that suppliers’ exposure to buyer AI-enabled environmental governance is negatively associated with supplier environmental controversies in the following year. This negative relationship is stronger in supplier countries with higher AI readiness and regulatory quality. The study contributes to research on AI-enabled sustainability governance and sustainable supply chain risk management.
[AI-39] Agent -Integrated Software: Interaction Contracts and Continuous Assurance
链接: https://arxiv.org/abs/2609.11381
作者: Shengcheng Yu,Chunrong Fang,Zhenyu Chen
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:Embedding an intelligent agent in an existing application creates a persistent coordination problem: users can revise goals and manipulate shared objects while delegated execution continues. We argue that dependable integration requires an explicit correspondence between task-level interaction and application behavior. We introduce Agent-Integrated Software (AIS) as a software pattern combining a conventional core, direct interaction, and a built-in agent, and Intent-Level Interaction Abstraction (IIA) as the task semantics through which users inspect and control delegated work. An open transition-system model relates AIS execution to IIA states and events. Interaction contracts constrain this relation through task bindings, role-specific authority, control transitions, and outcome evidence; continuous assurance maintains scoped claims as their dependencies change. A compact disclosure contract and conditional propositions illustrate why local component validity is insufficient and how selected admission invariants can be separated from planning. Contrasting software domains expose the framework’s assumptions and limits. This perspective develops a research agenda spanning application abstraction, development support, controlled execution, quality assessment, and human supervision, with the aim of making agent integration a maintainable software engineering discipline.
[AI-40] Characterizing Bluesky Content Moderation Service: From Automation of Service to Landscape of Harms
链接: https://arxiv.org/abs/2609.11373
作者: Pushpdeep Singh,Sayeh Jarollahi,Ayan Majumdar,Vabuk Pahari,Abhijnan Chakraborty,Krishna P. Gummadi,Ingmar Weber,Abhisek Dash
类目: Computers and Society (cs.CY); Artificial Intelligence (cs.AI)
备注: 18 pages, 7 figures, 12 tables. Accepted for publication at ICWSM 2027
Abstract:Empirical research on content moderation is fundamentally constrained by the opaque deployment of moderation systems on major social media platforms. To this end, the recent emergence of decentralized platforms with transparent, public moderation logs presents an unprecedented opportunity for independent audits. In this work, we leverage this architectural transparency to conduct the first large-scale audit of the default moderation system on Bluesky, the Bluesky Moderation Service (BMS). Analyzing its 10.6M moderation labels from 2025, we investigate three foundational aspects: (i) its mechanism (the degree of automation versus human oversight), (ii) its efficacy (accuracy in detecting harms), and (iii) its purpose (the landscape of harms it identifies). Our findings reveal a human-AI collaborative system where labels for sexual and graphic content are applied automatically in seconds, while nuanced and high stakes labels require more human oversight, taking hours or days. Through a manual annotation study, we find the BMS operates with high precision (0.837), but struggles with low recall (0.222), with our annotators identifying 4.5 \times more harmful content than the moderation system in a random sample. Finally, unsupervised clustering of the most frequently applied labeled posts uncovers detected harms ranging from hostility in discourse toward protected groups to the spread of sexually explicit and other graphic content. Our work offers a look into the operational realities of a deployed moderation system, providing a concrete data-driven foundation for designing more effective and transparent moderation systems.
[AI-41] RAMamba-Net: A Reliability-Aware and Mamba-Based Multimodal Fusion Network for Auditory Attention Detection
链接: https://arxiv.org/abs/2609.11372
作者: Xingyi He,Ziwei Wang,Dongrui Wu
类目: Artificial Intelligence (cs.AI)
备注: RAMamba-Net, a reliability-aware Mamba-based multimodal fusion network for auditory attention decoding
Abstract:Auditory attention decoding (AAD) identifies the attended speaker from physiological signals, supporting neuro-steered hearing devices and natural human-machine interaction. Electroencephalography (EEG) is the dominant modality for AAD but provides incomplete evidence in naturalistic audio-visual scenes, motivating EEG and electrooculography (EOG) fusion. Existing approaches remain limited by weak cross-modal interaction, inefficient temporal modeling, and low robustness to sample variations. To address the limitations, we propose RAMamba-Net, a reliability-aware Mamba-based multimodal fusion network for AAD. RAMamba-Net employs a Mamba-enhanced band-aware convolutional Transformer to capture band-specific EEG patterns and long-range temporal dynamics. A dual-branch temporal-spatial encoder models EOG temporal and inter-channel dependencies. Cross-modal attention enables explicit modality interaction. Then, a reliability-aware module is introduced to estimate sample-wise modality weights for feature and prediction consistency, thereby enhancing multimodal fusion. Experiments on two AAD benchmarks demonstrate that RAMamba-Net effectively exploits complementary EEG-EOG information, yielding accuracy gains of 5.76% over unimodal baselines, together with more robust decoding and discriminative representations. Further analyses show that explicit cross-modal interaction improves multimodal alignment, while the reliability-aware module suppresses unreliable modality evidence and is robust to signal perturbation and parameter variation.
[AI-42] Portable Semantics Private Dialects: Reuse and Negative Transfer in Latent Communication Between Language-Model Cells
链接: https://arxiv.org/abs/2609.11365
作者: Narcis Marincat
类目: Artificial Intelligence (cs.AI)
备注: 17 pages, 1 figure, 5 tables. Companion to arXiv:2608.20054 . Code and evaluation records: this https URL ; checkpoints and fitted alignment maps: this https URL
Abstract:In shared-genome language-model societies, restricted evidence visibility favors reusable, value-indexed latent packet interfaces, whereas the sole high-performing globally visible model in the parent study learned an episode-entangled code. This companion study asks whether independently trained societies share one packet language, where strict zero-shot transfer fails, and whether inherited interface state helps or harms later learning. First, a leakage-controlled causal interoperability audit over all 30 ordered pairs of six independently trained restricted societies – under sealed held-out structure and a preregistered raw/orthogonal/linear/nonlinear alignment ladder – shows the six semantically similar interfaces do not form one raw language: one same-initialization pair is exactly interoperable in both directions, a second shows asymmetric partial compatibility, and all 26 cross-initialization directions fail every frozen alignment rung. Second, within the tested decomposition and a single sealed source formulation, a source-span control localizes strict zero-shot failure to interpretation and execution of the new operator instructions. Third, in a matched adaptation factorial, the globally trained communication interface acts as a severe negative-transfer prior: reinitializing only the packet reader, writer, and mouth raises final depth-three accuracy from 0.169 to 0.857. Fourth, across two restricted checkpoints and two independently frozen target streams each, inherited interfaces never exceeded fresh-interface controls by the preregistered 0.10 margin. All primary conclusions are bounded to a near-transfer 17-state setting; the negative-transfer factorial concerns one globally visible parent-cohort checkpoint, while an appendix adds a post hoc tagged-global twin case study.
[AI-43] Reification as a Transferable Vocabulary: Zero-Shot Link Prediction with Vanilla GNNs
链接: https://arxiv.org/abs/2609.11347
作者: Camille Pradel
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Knowledge graph foundation models such as ULTRA achieve zero-shot link prediction on unseen graphs through dedicated architectures that hard-code a transfer mechanism. In this work we move that mechanism out of the architecture and into the representation, by \emphreifying the input graph: every fact becomes a node, connected to its subject, object, and relation type through a fixed vocabulary of six meta-relations, with relation types as anonymous shared nodes rather than model parameters. On this representation, five textbook GNNs (GAT, GINE with sum and with mean+max aggregation, GraphSAGE, R-GCN), each trained on a single knowledge graph of 4,245 triples for 30 minutes on one NVIDIA A100, transfer zero-shot to 40 inductive link-prediction benchmarks. The best of them, an off-the-shelf GAT, matches ULTRA, a dedicated foundation model pretrained on three graphs, across ULTRA’s own evaluation suite. The same fixed vocabulary extends to relational databases, a row becoming an entity and a foreign-key column a relation type; a preliminary probe on two unseen databases, with no cell values, schema text or in-context labels, shows a model of this family pretrained on three knowledge graphs ranking foreign-key targets far above random-initialization and degree controls. We release the code, the checkpoints, and the evaluation pipeline for all 40 benchmarks.
[AI-44] Exploring Diffusion Transformers for Cross-Modal Augmentation in Multimodal Brain State Decoding
链接: https://arxiv.org/abs/2609.11341
作者: Ziwei Wang,Xingyi He,Hongbin Wang,Tianwang Jia,Bohan Fang,Dongrui Wu
类目: Artificial Intelligence (cs.AI)
备注: CoMA-DiT, a cross-modal augmentation framework built on Diffusion Transformer, extends multimodal learning beyond fusion by leveraging paired modalities as mutual generative supervision to enrich training data and improve brain state decoding
Abstract:Multimodal brain state decoding has largely focused on fusing paired modalities for prediction, but has rarely explored how their correspondence can be further exploited to enrich training data and improve multimodal representation learning. To address this gap, we propose CoMA-DiT, a bidirectional cross-modal Diffusion Transformer for latent augmentation that treats paired modalities as sources of mutual generative supervision rather than merely as inputs to be fused. CoMA-DiT conditions velocity prediction on the paired modality through cross-modal attention and adaptively injects the resulting variation via a reliability-gated residual mechanism. Experiments on multimodal auditory attention decoding and emotion recognition showed that CoMA-DiT consistently outperformed 20 representative baselines, achieving absolute gains of 4.28% and 6.70% in accuracy and macro-F1 over the no-augmentation baseline, respectively. Extensive ablation, sensitivity, visualization, and interpretability analyses further demonstrated its robustness, generalizability, and ability to capture functionally relevant cross-modal interactions. These findings support a broader view of multimodal learning: Paired modalities can serve not only as inputs for fusion but also as supervision sources that augment one another.
[AI-45] AI Exposure and AI Resilience: A Two-Dimensional Assessment Framework for Software and Software-Based Business Model
链接: https://arxiv.org/abs/2609.11321
作者: Paul Darius Mandl(Findustrial GmbH),Peter Mandl(Munich University of Applied Sciences),Martin Häusl(Munich University of Applied Sciences)
类目: Artificial Intelligence (cs.AI)
备注: 14 pages, 3 figures, 6 tables. Preprint
Abstract:Artificial intelligence is changing both software production and the economics of software-based business models. Classical technology due diligence mainly examines technical properties such as architecture, scalability, and technical debt. These criteria do not fully capture how AI can affect a company’s value proposition, competitive position, margins, or access to customers. This paper develops Artificial Intelligence Exposure and Resilience (AI-ER) as a two-dimensional assessment framework. AI exposure describes the pressure for change that AI creates for a business model. AI resilience describes the company’s ability to absorb that pressure, adapt to changed conditions, and use AI in an economically viable way. Metrics for both dimensions are derived from current AI capabilities, their deployment conditions, and relevant research on business models and organizational adaptability. The model keeps exposure and resilience separate and adds an explicit assessment of evidence quality and confidence. It can be applied first with public information and later refined with internal evidence. The result is a traceable company profile that supports comparison without concealing uncertainty in the underlying evidence. The paper also specifies an initial score logic and a procedure for empirical validation.
[AI-46] Magent a: Closing the Loop Between Mathematical Reasoning and Lean Verification
链接: https://arxiv.org/abs/2609.11319
作者: Joshua Ong Jun Leang,Haonan Li,Zheng Zhao,Xinyi Shang,Wenda Li,Zhengzhong Liu,Erix Xing,Shay Cohen,Eleonora Giunchiglia
类目: Artificial Intelligence (cs.AI)
备注: 9 pages, preprint
Abstract:Most of mathematical knowledge has been communicated through so-called informal use of mathematics and natural language. With large language models (LLMs) being highly adept in using natural language, they achieve strong performance, yet not perfect, in informal mathematical reasoning. Restraining LLMs to informal reasoning misses out on the opportunity to use the discrete verification abilities that machines offer through machine-checkable proofs. In this paper, we bridge the gap between informal and formal reasoning by integrating Lean signals into the informal reasoning process. We introduce Magenta, a training-free agentic pipeline that, given only a natural-language problem, produces an answer, expresses it as a Lean 4 statement, and constructs a machine-checked proof. A statement judge verifies whether the formalisation preserves the original problem, while an error-attribution judge routes failed attempts either to mathematical re-derivation or local Lean repair. Magenta achieves 100% accuracy across all evaluated olympiad benchmarks, including AIME 2025, AIME 2026, and HMMT February 2026. When paired with the open-weight K2-Horizon-7B reasoner, it solves all six IMO 2026 problems. Our analysis shows that statement adjudication is essential for preventing false certificates and that feedback-guided correction outperforms independent resampling on difficult problems.
[AI-47] Mr.LHDR: A Benchmark for Multimodal Real-World Long-Horizon Deep Research Agents
链接: https://arxiv.org/abs/2609.11318
作者: Minghao Guo,Meng Cao,Sui Zhao,Siyu Ning,Xin Wang,Haoze Zhao,Jiaxuan Yang,Haihong Hao,Mingfei Han,Shunlin Rong,Haijun Wu,Xiaodan Liang,Xiaojun Chang
类目: Artificial Intelligence (cs.AI)
备注: Code and data are available at this https URL
Abstract:Deep research agents are increasingly capable of web search, tool use, multimodal evidence analysis, and information synthesis. However, existing benchmarks mainly evaluate medium-horizon exploration and rarely test whether agents can sustain long, dependency-heavy research processes. We introduce this http URL (Multimodal real-world Long-Horizon Deep Research), a benchmark for evaluating real-world deep research over long, irreducible chains of interdependent evidence across eight categories. Each question is constructed from a hidden Node-Relation graph and requires an average of 12.1 necessary intermediate conclusions with a mean dependency depth of 10.4 before reaching a short, unique, and verifiable answer. Questions incorporate multimodal evidence, including images, maps, PDFs, logos, charts, tables, and video frames, with at least one non-text element that changes the reasoning state. this http URL evaluates both final answers and the correctness of intermediate conclusions under annotated dependencies. We evaluate general models, deep research systems, and agent frameworks using Overall Accuracy (OA), Strict Accuracy (SA), Checklist Score (CS), and Dependency-Aware Checklist Score (DACS). Results show that even the strongest system achieves only 43.1% OA and 34.3% SA, indicating that final-answer accuracy substantially overestimates complete research success. Removing images reduces DACS by 12.6 points, demonstrating the importance of multimodal evidence, while SA consistently declines as reasoning chains become longer. These findings reveal sustained, dependency-consistent evidence integration, rather than isolated fact retrieval, as a key bottleneck for current deep research agents.
[AI-48] Routing by Reasoning Need: Trajectory-Aware Decoding Control for Diffusion Vision-Language Models EMNLP2026
链接: https://arxiv.org/abs/2609.11315
作者: Yixiang Liu,Zhongxing Xu,Zhonghua Wang,Xiaoying Tang
类目: Artificial Intelligence (cs.AI)
备注: 17 pages, 9 figures. Accepted to Findings of EMNLP 2026
Abstract:Diffusion vision-language models generate answers through iterative refinement, exposing intermediate answer trajectories that can be inspected and controlled at inference time. However, this controllability creates a reasoning-need mismatch, where a universal generation length is applied to questions with different reasoning demands. Visually closed questions may be harmed by continued refinement after a stable answer has formed, whereas reasoning-sensitive questions may be harmed by premature commitment. We formulate this problem as reasoning-budget mismatch and study it in LLaDA-V. Rather than choosing a universal generation length, our training-free controller routes each example to early commitment, baseline preservation, or reasoning-supportive decoding using trajectory signals from answer closure, commitment evidence, and representation revision pressure, without using ground-truth answers. Across answer-focused, mixed-reasoning, and CoT-sensitive benchmarks, routed control improves robustness over fixed long decoding, pure short decoding, and single-rule interventions. The gains are not explained by shorter outputs alone. Answer-closed examples often benefit from commitment, whereas CoT-sensitive examples require preserving or supporting intermediate reasoning. Taken together, these results suggest diffusion VLM decoding should route inference-time control by the state suggested by the observed trajectory instead of relying on a universal decoding length.
[AI-49] 2AM: Grounding Agent -Side Memory as Guidance for Steerable Action Models in Long-Horizon Manipulation
链接: https://arxiv.org/abs/2609.11308
作者: Yutong Hu,Fengjiao Chen,Xuezhi Cao,Renaud Detry
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Long-horizon robot manipulation requires memory, but not necessarily inside the action policy. To address such tasks, current agentic systems often combine VLAs with planners and geometric tools, sometimes using additional depth or calibrated geometry. These systems confound attribution: gains may come from richer observations or alternative motor tools, while failures may stem from either the policy or an under-specified language interface. We isolate this question through a deliberately constrained design: less tool breadth, but greater interface bandwidth. 2AM makes a multimodal Agent the sole holder of task memory and a single RGB-based, episodically stateless Action Model the sole executor of task-relevant motion. The Agent compiles interaction history into subtask language and optional 2D grasp, place, and move hints that bind its physical intention at different time scales. To teach this steerability to the VLA, we augment demonstrations with structured hint labels and train under condition dropout, spatial noise, and temporal jitter to tolerate imperfect Agent outputs. On LIBERO-Mem, without depth, online geometry, or planner-based object motion, 2AM reaches 76.3% average completion, a 61.5-point improvement over the strongest reported baseline of 14.8%, together with 63.0% relaxed and 11.8% strict success. These results show that task memory can remain Agent-side. They further show that Action Model capability depends not only on what the policy has learned, but on how precisely the Agent can steer it.
[AI-50] Memory Compression for High-Fanout Agent Sandboxes
链接: https://arxiv.org/abs/2609.11294
作者: Mengming Li,Ceyu XU,Qijun Zhang,Jiangnan Yu,Xiangfeng Sun,Haohui Mai,Zhiyao Xie
类目: Artificial Intelligence (cs.AI); Operating Systems (cs.OS)
备注:
Abstract:High-fanout agent workloads create a growing memory bottleneck because a single task may spawn many concurrent sandbox sessions. Yet these sandboxes are far from independent: they originate from a shared template and execute related trajectories, exposing substantial template-relative and cross-sandbox memory redundancy. Conventional memory compression is poorly matched to this setting in three fundamental dimensions: how to compress, because they fail to exploit similarity across non-identical sandbox pages; what to compress, because they control page-fault overhead through conservative page selection; and when to compress, because compression is either triggered by memory pressure or performed without awareness of agent execution phases. We present AgentZip, the first memory compression system designed specifically for AI-agent sandboxes. AgentZip introduces compression mechanisms that exploit both the template-relative and cross-sandbox redundancy. It broadens the compression scope to any page with a profitable representation and shifts overhead control from compression-time page selection to restore-time prefetching. It further aligns expensive compression with LLM waiting periods to avoid interfering with foreground tool execution. Across LLM training and inference workloads, AgentZip reduces sandbox-owned memory by up to 8.7x, compared with 2.1x for the Linux configuration. Restore prefetching and agent-execution-aware scheduling reduce the slowdown of aggressive compression from as high as 3.1x to 1.40x while retaining nearly all of its memory-saving benefit. Subjects: Artificial Intelligence (cs.AI); Operating Systems (cs.OS) Cite as: arXiv:2609.11294 [cs.AI] (or arXiv:2609.11294v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2609.11294 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-51] Off-Target Effects of Response-Style Alignment in a Korean 27B Language Model
链接: https://arxiv.org/abs/2609.11291
作者: Hyojung Han
类目: Artificial Intelligence (cs.AI)
备注: 19 pages. Korean-language evaluation (KoBBQ); all uncertainty estimates over KoBBQ items are clustered on the benchmark template
Abstract:We post-train Qwen3.8-27B for Korean response style – verbosity, list and markdown usage, discourse structure and register – and measure two behaviours the objective never targets: abstention on ambiguous social questions in KoBBQ, where the benchmark-correct answer is UNKNOWN, and unprompted disclosure in securities guidance. Both move, and the changes are expressed primarily through the model’s emission policy: how often it answers and how much it says. Matched target-form controls show that answer propensity depends on the training target, not the prompt set or recipe alone. Holding prompts, recipe, data volume and serving fixed and changing only the target text, three style seeds give positive answer-rate point estimates (mean +0.82 pp) and three neutral seeds negative ones (mean -1.53 pp); the observed seed ranges do not overlap and the means differ by 2.34 pp. A length-matched arm lies between them, and a fourth arm that stays short while preserving hedging is unstable across seeds, so which feature of the form is responsible is unresolved. For absolute stereotyped exposure the decomposition into an answer-propensity term and a conditional-composition term is an algebraic identity, not a finding; its empirical content is where the movement went. Across the trained checkpoints the changes are dominated by answer propensity while the composition term stays small, and because that term is evaluated on treatment-dependent answered subsets we do not read it as evidence about latent preference. Two measurement results follow. A between-arm contrast in conditional stereotyped share does not identify a change in conditional content preference when answer status is treatment-dependent. And agreement between two rule detectors for the same construct runs from 0.44 to 0.99 depending on which checkpoint produced the text – observable without any reference labels. Comments: 19 pages. Korean-language evaluation (KoBBQ); all uncertainty estimates over KoBBQ items are clustered on the benchmark template Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2609.11291 [cs.AI] (or arXiv:2609.11291v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2609.11291 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Hyo-Jung Han [view email] [v1] Thu, 10 Sep 2026 09:23:04 UTC (42 KB)
[AI-52] Generating a Consistent Enterprise: Synthesis and Reference-Free Evaluation of Multi-System Business Data
链接: https://arxiv.org/abs/2609.11286
作者: Benjamin Gruenbaum,Doron Porat,Assaf Natanzon,Roy Zavida,Chen Dinachi,Or Itzahary,Omer Niv
类目: Artificial Intelligence (cs.AI)
备注: 10 pages
Abstract:Synthetic relational data is normally produced by a model trained on a real dataset, and its quality is measured as the distance to that dataset. This paper describes a generator that has no real dataset at either end. Given an industry, a company size, a business model, a set of business applications, and a random seed, it produces a complete fictional enterprise: a workforce, a customer base, sales deals, support tickets, recorded calls, chat messages, and documents, all consistent with one another. One entity graph is projected into the native formats of 66 business products, so the same customer appears in the CRM, the support desk, and the call system under one identity. Because no real counterpart exists, realism is built in from cited reference statistics and verified by reference-free measurement: a five-axis scorecard of 28 statistical checks, an adversarial detector that hunts for the marks of synthetic generation, and a set of soundness checks that include a classifier test against an independently shuffled copy of the data. Because these instruments existed before the generator was tuned, progress is measured under a fixed yardstick: over 23 generated companies, mean realism climbed from 60.3 to 99.1, the weakest company from 41.1 to 94.9, and the detector, which initially flagged 55.2% of all records, now flags none. The scores hold on a seed never used during development. A second generator builds relational databases from a list of business questions. It forces qualifying rows for each answerable question, adds controlled near misses, and computes exact labels from the finished tables. The generator runs as a hosted service at this https URL. A company built there to a specification is served through its simulators over MCP and REST, and the simulators are also published as container images for offline use
[AI-53] When Does Text Inform? Benchmarking Information-Theoretic Metrics for Multimodal Time-Series Forecasting
链接: https://arxiv.org/abs/2609.11282
作者: Emma Andrews,Gianmarco Mengaldo
类目: Artificial Intelligence (cs.AI); Information Theory (cs.IT)
备注:
Abstract:Multimodal forecasting models that combine time series with text annotations promise richer prediction through textual context, but how do we know whether a text annotation meaningfully contributes to the forecasters prediction? This is an information-theoretic question, but to evaluate whether information-theoretic metrics can reliably measure the predictive value an annotation provides, a ground truth benchmark is needed, and none currently exist. We create a synthetic time series signal with annotations in three categories: semantically correct, incorrect, and irrelevant. Because the data generation process is fully controlled, ground-truth information content is known exactly, enabling principled evaluation of six complementary mutual information estimators (KSG, MINE, InfoNCE, CCA, PID and V-information). We show that all six estimators identify correct annotations as most informative, and are able to audit the quality of mixed text corpora, choosing the annotations that result in the best downstream forecasting results without the need for model training. Our benchmark identifies limitations of each estimator, and these are validated on seven real-world datasets, which show how estimator performance differs on weak signals. Finally, we establish practical rules for implementing these metrics for annotation auditing and fusion selection.
[AI-54] Bio-inspired Learning and Decision-Making with Probabilistic In-Memory Computing Hardware: Part 1
链接: https://arxiv.org/abs/2609.11281
作者: Thomas Dalgaty,Eiji Kawasaki,Miguel de Prado,Devendra Vyas,Tommaso Salvatori
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Learning and decision-making in animals are often modeled as Bayesian processes, where sensory evidence is integrated with prior beliefs to guide behavior in the face of uncertainty. But what are the inherent neural dynamics that give rise to this ability, and how could they be replicated in computing systems? This abstract discusses a biologically grounded framework in which noisy neural and synaptic dynamics perform inference and learning via stochastic sampling from an internal energy function, capturing uncertainty over latent states and model parameters through neural and synaptic variability, respectively. This enables approaches such as predictive coding networks to account for epistemic uncertainty via Markov chain Monte Carlo sampling. Drawing a parallel between intrinsic noise in biological systems and electrical noise in emerging probabilistic analogue memory technologies, we highlight how analogue in-memory computing hardware naturally emerges as the solution for massively scalable and energy-efficient probabilistic inference.
[AI-55] Predicting Train Delays in Finland Using Machine Learning and Weather Data
链接: https://arxiv.org/abs/2609.11277
作者: Vinicius Pozzobon Borin,Jean Michel de Souza Sant’Ana,Nurul Huda Mahmood
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 6 pages, 3 Figures, 4 tables, presented at Wireless Europe 2026, Rimini, Italy, June 2026
Abstract:Reliable railway operations depend increasingly on real-time environmental intelligence delivered through wireless sensor infrastructures, a capability that 6G networks will substantially enhance through integrated sensing and edge computing. Adverse weather, particularly in Arctic regions with extreme temperatures and heavy precipitation, remains a leading cause of train delays, yet most prediction approaches rely on raw meteorological inputs without exploiting domain-informed feature engineering. This paper investigates machine learning for train delay prediction using the Finland Integrated Train-Weather (FI-TW) dataset, which fuses railway operational records with observations from the Finnish Meteorological Institute’s nationwide sensor network of approximately 200 stations communicating over wireless links. We evaluate three feature configurations using XGBoost at Oulu central station (101,146 observations): full weather features, instant weather observations only, and derived weather category scenarios. The category-based approach, employing hierarchical classifications such as Blizzard, Heavy Snow, and Extreme Cold, achieved an R^2 of 0.78, root mean squared error of 8.5 minutes, and mean absolute error of 3.7 minutes, representing an 11% R^2 improvement and 10% error reduction over alternative configurations. These results demonstrate that compact, domain-informed features derived from sensor streams outperform raw meteorological observations, offering bandwidth-efficient representations suitable for edge deployment over current and emerging wireless infrastructures.
[AI-56] Sci-MMR: Benchmarking Multi-Step Evidence-Grounded Scientific Reasoning in Multimodal Agents
链接: https://arxiv.org/abs/2609.11243
作者: Jiaqiang Li,Yajie Yang,Zhiheng Xi,Jiadong Chen,Enyu Zhou,Senjie Jin,Yang Nan,Jiazheng Zhang,Han Wang,Yanxin Li,Dingwei Zhu,Bicheng Deng,Yuhui Wang,Xiang Zheng,Qi Zhang,Lei Bai,Xingjun Ma,Tao Gui
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Autonomous research agents are increasingly expected to search the literature, analyze experimental evidence, and generate scientific hypotheses. These capabilities require multi-step evidence grounded reasoning that progressively acquires, integrates, and verifies evidence before reaching a conclusion. Existing multimodal benchmarks, however, largely evaluate final-answer accuracy, leaving open whether predictions are actually supported by traceable scientific evidence. We introduce Sci-MMR, a benchmark for multi-step evidence-grounded scientific reasoning built on structured argument graphs linking scientific claims, citation-grounded knowledge, visual evidence, and supporting regions. Sci-MMR comprises 235 multi-hop reasoning tasks spanning four scientific disciplines, with an average of nine figure panels per task. Evaluating eight frontier multimodal models, we find that answer accuracy consistently exceeds complete-evidence recovery rate by more than 20%, revealing a substantial gap that answer-only evaluation is structurally unable to capture. Through controlled interventions, we identify two fundamental bottlenecks. First, evidence acquisition: models struggle to extract complete structured evidence from scientific figures, accounting for 57.2% of failures. While cropping tools yield modest gains (+4.5 points), providing gold evidence improves accuracy by up to 37.0 points, indicating difficulty in assembling complete multi-region evidence. Second, evidence integration: models struggle to translate available evidence into correct conclusions, accounting for 31.8% of failures, while even with gold evidence the strongest model achieves only 69.1% accuracy on the hardest tasks. These findings indicate that current answer-centric benchmarks substantially overestimate the evidence-grounded reasoning capabilities of multimodal research agents
[AI-57] NovGauge: A Fine-Grained Benchmark for Diagnosing LLM s Capability in Paper Novelty Assessment
链接: https://arxiv.org/abs/2609.11234
作者: Guoqiang Zhang,Kexin Tan,Ming Zhang,Li Ju,Wenqing Jing,Zhonghan Yue,Jiayi Chen,Shiqiang Wu,Shaofan Liu,Yue Zhang,Yuankai Ying,Yang Shi,Tao Gui,Qi Zhang,Xuanjing Huang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models (LLMs) are increasingly used in peer review at major AI conferences, yet novelty remains a persistent weak point. Existing benchmarks assess novelty as a single holistic score, making it difficult to diagnose which dimension a model misjudges or whether its evidence is faithful. We present NovGauge, a human-anchored benchmark for fine-grained novelty assessment diagnosis. The benchmark contains 619 paper pairs and 50 multi-paper sets, drawn from two expert sources: ICLR reviewer overlap claims and survey co-citations. Instances are independently labeled along three dimensions: task, problem, and method, capturing application goals, technical challenges, and solution approaches. We propose a cascading diagnostic pipeline that verifies per-dimension correctness, evidence grounding, and logical support. Evaluation of 18 LLMs shows hallucination rates ranging from 0% to 39% across dimensions, and among non-hallucinated correct-positive judgments, over 70% cite evidence fails to logically support the stated reason. The best-performing model, GPT-5.5, achieves 43-72% Verified F1 across dimensions, while most models retain less than half of their raw F1 after faithfulness verification. These results suggest that current LLMs remain far from reliable scientific novelty assessment, particularly when correctness is conditioned on faithful evidence grounding.
[AI-58] Solving Few-Shot Multiobjective Multitask Optimization via Iterative Sequential Transfer CEC2026
链接: https://arxiv.org/abs/2609.11228
作者: Tingyang Wei,Haofeng Wu,Ananda Phan Iman,Zhao Wei,Jiao Liu,Yew-Soon Ong
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Neural and Evolutionary Computing (cs.NE)
备注: Accepted paper in WCCI/CEC 2026
Abstract:Applying knowledge transfer across multiple optimization tasks, multitask optimization (MTO) emerges as a promising approach to solving synergistic optimization tasks simultaneously. However, the development of effective knowledge transfer mechanisms in MTO fundamentally relies on aligning elite solution distributions across tasks. This dependency creates a critical bottleneck in few-shot optimization regimes, as restricted evaluation budgets impede the identification of elite solution distributions required for beneficial transfer. This challenge is exacerbated in multiobjective multitask problems, where each optimizer must approximate a continuous Pareto manifold rather than a single optimal point. This paper introduces Iterative Sequential Transfer (IST) to circumvent this bottleneck. We model MTO as a sequence of sequential transfer optimization problems, concentrating evaluations on a single target per iteration. We propose a likelihood-informed task prioritization mechanism to maximize transfer utility by identifying the task most likely ready for knowledge integration. Empirical results on benchmark and real-world problems verify the effectiveness of the proposed method under tight budgets.
[AI-59] CryptoL: Towards Scale Dominance and Physics Constraints Mitigation in Financial Multivariate Time Series Forecasting
链接: https://arxiv.org/abs/2609.11206
作者: Yalda Taheri,Mohammad Hassan Heydari,Armon Rasooli,Maryam Amirshahkarami,Mohammad Ebrahim Mahdavi,Hossein Karshenas
类目: Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE); Machine Learning (cs.LG)
备注:
Abstract:Cryptocurrency forecasting presents a distinctive combination of extreme cross-asset scale heterogeneity, non-stationary dynamics, and structural dependencies among Open, High, Low, and Close (OHLC) variables. We present CryptoL, a unified framework designed to address these challenges within multivariate time-series forecasting. CryptoL evaluates forecasting error in context-normalized coordinates within the RevIN pipeline, preventing inverse normalization from introducing an additional squared-scale weighting into the MSE objective. We formally characterize this effect through the empirical risk and parameter-gradient geometry, establishing the conditions under which large-scale assets can disproportionately influence shared-model optimization. Beyond loss-space normalization, CryptoL examines channel-independent and channel-dependent normalization for OHLC data, showing that a shared channel-dependent affine transformation preserves candle-order relations that independent channel transformations need not preserve. The framework further incorporates scale-adaptive numerical stabilization to reduce distortions caused by a fixed normalization constant across assets spanning many orders of magnitude, together with a soft feasibility loss that penalizes violations of the defining OHLC inequalities. Experiments across heterogeneous cryptocurrency assets evaluate these components through controlled ablations and demonstrate improvements in forecasting accuracy, training stability, and the frequency of financially valid OHLC predictions relative to the considered baselines. CryptoL therefore provides an integrated approach to scale-balanced optimization, structure-preserving normalization, numerical stabilization, and constraint-aware cryptocurrency forecasting.
[AI-60] An AI-Powered Culturally Aware Chatbot for Stress Detection and Wellness Support among Pakistani University Students Using NLP and Machine Learning
链接: https://arxiv.org/abs/2609.11199
作者: Muhammad Fahad Bashir,Muhammad Afzal
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:With the existing digital mental health tools specifically developed for Western settings, Pakistani students are exposed to a uniquely compounded stress situation in their university that includes academic, financial, familial, and relational stressors, which have become a serious concern for academic and psychological development of students in Pakistani universities. This paper introduces a new, AI-driven and culturally sensitive stress detection and wellness support system that is tailored to the context of Pakistani university students. The system is based on a machine learning model called Random Forest which is trained using a validated student stress data set of 1100 responses on 20 features from psychological, physiological, academic, environmental and social aspects, with an accuracy of 89.09% and a macro F1-score of 0.89, in three stress severity levels. The classification outputs are passed on to an open-source large language model through OpenRouter API, where an appropriately crafted system prompt, culturally aware, gives the model a conversation about wellness, in English, Urdu and Roman Urdu. The second most predictive stress factor in this population identified by feature importance analysis was teacher-student relationship, which is a culturally important stress factor highlighting the need for region-aware mental health systems. Future research will involve primary data collection from students at various academic levels of Pakistani Universities with the validated DASS-21 instrument focusing on the students who are moving from FSc to undergraduate studies, which is a time of being psychologically vulnerable which is under-researched.
[AI-61] Can LLM s Follow Medical Expert Logic? A Benchmark for Hierarchical Logical Consistency in Risk-of-Bias Assessment EMNLP2026
链接: https://arxiv.org/abs/2609.11185
作者: Jiayu Huang,Zichen Tang,Qianhui Ling,Zemin Kuang,Haihong E
类目: Artificial Intelligence (cs.AI)
备注: Accepted to EMNLP 2026 Main Conference
Abstract:Evidence-based medicine demands strict logical consistency, yet current evaluations of large language models (LLMs) prioritize superficial label matching over genuine reasoning. We introduce LogiMed-RoB, a benchmark grounded in Cochrane Risk of Bias (RoB) 2.0 expert logic, comprising 860 randomized controlled trials (RCTs) and 14,820 queries. It evaluates models under the Hierarchical Logical Consistency (HLC) framework across four dimensions: Atomic Consistency, Domain Consistency, Aggregation Consistency, and Evidential Faithfulness. Experiments on 10 state-of-the-art LLMs reveal a catastrophic Error Compounding Effect: despite the top model reaching 98.88% Atomic Consistency, its end-to-end consistency collapses to 45.13%, with several open-weight architectures plummeting to nearly 0%. We further uncover a systematic evidence-reasoning gap: even when models retrieve high-quality evidence, they fail to deduce correct outcomes in 18.63-40.05% of cases, while Blind Guess Rates reach 48.28%. LogiMed-RoB demonstrates that high outcome accuracy can conceal critical reasoning flaws, underscoring the necessity of white-box logical verification for clinical deployment.
[AI-62] SemVerBench: Benchmarking LLM Comprehension of Version-Constraint Resolution Semantics ICTAI2026
链接: https://arxiv.org/abs/2609.11180
作者: Qibai Chen,Zeming Liu
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注: Accepted at the 38th IEEE International Conference on Tools with Artificial Intelligence (ICTAI 2026). 8 pages, 6 tables
Abstract:Large language model (LLM) coding agents constantly decide whether a version satisfies a constraint such as ^1.2.3 or =2.0,3, yet their grasp of version-constraint semantics has never been measured directly. We introduce SemVerBench, the first benchmark of LLM version-constraint resolution semantics across three ecosystems (npm, PEP 440, Cargo): 240 machine-checkable items with unique answers, built author-neutrally from four balanced sources (each ecosystem’s official test suite plus three frontier LLM proposers) and labeled by a non-circular two-implementation oracle. Evaluating six frontier models, we find systematic, predictable per-mechanism blind spots: a partial-comparator carry rule (1.2 means =1.3.0) traps every model on Cargo (near 60%), and although standard PEP 440 prefix matching is universal, on zero-pad/post-release corner cases GPT-5.1 collapses (0/26) while Claude stays at 97-100% (verified on a 67-item oracle-validated set). Opus significantly outperforms all other models, and Sonnet outperforms the OpenAI models (McNemar). The failures look more like an activation/application gap than a knowledge gap: injecting the rule or a light correct hint recovers most errors, whereas interval decomposition does not, and models are at ceiling on the basic forms of the same rules. An author-stratified analysis finds no statistically significant self-favoritism. Because the task is verifiable and a free, 100%-correct resolver exists, tool delegation reaches ~100%: coding agents should delegate version resolution to a resolver rather than reason about versions in-head.
[AI-63] Debate-to-Skill: Capability-Bound Process Supervision for Industrial Query-to-Agent Annotation EMNLP2026
链接: https://arxiv.org/abs/2609.11176
作者: Shiyu Zhang,Leisheng Cheng,Huifu Li
类目: Artificial Intelligence (cs.AI)
备注: Accepted at EMNLP 2026
Abstract:Industrial query-to-agent matching fails when topical relevance is mistaken for executable capability, especially on long-tail and boundary-sensitive requests. We formulate annotation as \emphcapability-bound process supervision and instantiate it with Debate-to-Skill, which uses reusable decision principles, structured deliberation, verifier-based verdict extraction, and disagreement-driven refinement. On an industrial Query2Agent benchmark, we compare Debate-to-Skill with direct-label supervision, reasoning-SFT, and structural ablations. The results test whether gains come from supervising the capability-critical decision process itself, especially on grey-zone cases where semantic relatedness and executable capability diverge.
[AI-64] Breaking Predictions Is Not Enough: Specified-Foil Counterfactuals for Temporal Graphs
链接: https://arxiv.org/abs/2609.11170
作者: Minwoo Yu,Young-guk Ha
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Temporal graph counterfactual explanations typically change past events to change or invalidate an original prediction, while leaving its replacement unspecified. Yet a user facing a predicted outcome often asks which past conditions would make a particular alternative occur instead. We formulate this destination-specific question as the Specified-Foil Counterfactual: given an original prediction A and a foil B fixed before search, find a low-cost past-event intervention under which the same predictor selects B as top-ranked. Our trace-guided intervention search contrasts the completed execution of A with a reconstructed incomplete execution of B, maps their difference to DELETE, INSERT, REWIRE, RELABEL, and SHIFT operations, and verifies B through exact replay. We instantiate this principle with LiFTER on continuous-time dynamic graphs and TLogic on temporal knowledge graphs. On CTDGs, the method retains 85.7-93.6% of black-box greedy successes while reducing predictor evaluations by 75.0-80.0%; on TKGs, it reaches the specified foil in 74.8% of 600 comparisons. Executable traces thereby become computational structures for constructing conditions of unselected alternatives, rather than records used only to explain predictions already made.
[AI-65] DRG-MAPPO: Hierarchical Dynamic Role-Graph Multi-Agent Reinforcement Learning for Cooperative Air Combat
链接: https://arxiv.org/abs/2609.11155
作者: Junlin Liu,Chengwei Li,Yang Gao,Hui Chang,Xinchen Zhang,Zhijun Zhao,Hao Zhao
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Multi-Agent Reinforcement Learning (MARL) has emerged as a pivotal paradigm for complex decision-making in autonomous systems and air combat. While MARL has demonstrated significant potential in air combat, achieving sophisticated tactical coordination remains a non-trivial challenge. This difficulty is largely attributed to two primary limitations: (1) the absence of structured relational modeling hinders agents from capturing complex, time-varying interactions among battlefield entities; and (2) conventional flat architectures often lack the capability to explicitly model tactical roles, leading to ambiguous task allocation in highly dynamic environments. To address these challenges, we propose Hierarchical Dynamic Role-Graph Multi-Agent Proximal Policy Optimization (DRG-MAPPO), a novel MARL framework that integrates graph-based relational modeling with dynamic role assignment. Specifically, DRG-MAPPO constructs a graph-based representation of battlefield interactions and leverages graph attention mechanisms to extract critical relational features among allies, enemies, and threats. Subsequently, a high-level policy employs a dynamic role assignment mechanism to determine tactical responsibilities (e.g., leader'' and supporter’'). Conditioned on these roles and encoded graph-relational features, a low-level policy executes discrete maneuver actions, facilitating the joint optimization of tactical strategy and collaborative execution. Furthermore, a target-priority auxiliary task is designed to foster the emergence of behaviors such as focus-fire. Experimental results demonstrate that DRG-MAPPO achieves a state-of-the-art win rate of 87%, suggesting that our framework effectively balances relational modeling, interpretability, and optimization stability for cooperative air combat.
[AI-66] rms.txt: A Consent and Compensation Protocol for Agent ic Web Access
链接: https://arxiv.org/abs/2609.11152
作者: Rajarshi Chowdhury
类目: Networking and Internet Architecture (cs.NI); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Computers and Society (cs.CY)
备注: 7 pages, 1 figure, 2 tables. Submitted to IEEE Internet Computing, Special Issue on Future Internet Systems with LLMs and Agents. Code and raw results: this https URL
Abstract:The open web ran on an unwritten bargain: sites admitted crawlers, and search engines sent visitors back. Public measurements show that bargain breaking under AI crawlers and agents. Automated clients now make up most requests, training dominates Cloudflare-classified crawling, and the largest AI platforms fetch thousands of pages for each visitor they return. The web’s common control, this http URL, cannot express identity, purpose, terms, or price, can be circumvented, and newer alternatives are largely proprietary CDN features. We specify this http URL, a this http URL-style file for per-path, per-purpose machine-access terms, plus an origin-enforced exchange using Web Bot Auth signatures, signed intent, delegation tokens, HTTP 402 negotiation, and signed receipts. We define what the exchange can enforce, audit, and leave to contract. A dependency-free implementation adds 0.20 to 0.65 ms per request on one vCPU.
[AI-67] Autonomous Chemical Mechanistic Discovery through Agent ic Reasoning and Validation
链接: https://arxiv.org/abs/2609.11147
作者: Dong Li,Sixuan Mi,Zihao Ye,Huan Xiong,Tao XU,Tong Zhu,Aijia Zhang,Junqi Gao,Kaiyan Zhang,Shijie Wang,Bowen Zhou,Yuqiang Li,Biqing Qi
类目: Artificial Intelligence (cs.AI)
备注: This paper has been submitted to Nature Communications
Abstract:Unraveling reaction mechanisms is central to modern chemistry, yet automating these investigations remains challenging because computational workflows still rely heavily on expert intervention. Here we introduce ARCHE, an autonomous agentic system that integrates a general-purpose reasoning model, a domain-specialized computational chemistry model, and a structured tool registry to transform mechanistic inquiry into a scalable, self-validating process. ARCHE interprets scientific questions, generates and prioritizes mechanistic hypotheses, orchestrates computational workflows, and iteratively refines conclusions based on computed evidence within a closed loop. We validate its capabilities across three increasingly demanding scenarios: reconstructing stereocontrolling transition states and validating the corresponding reaction mechanism in a previously reported asymmetric catalytic reaction; proposing and validating a plausible radical pathway through iterative hypothesis refinement for a recently discovered but unpublished \alpha -iodoboronate C-I cleavage reaction; and identifying a chemically interpretable descriptor that governs selectivity in nickel-catalysed migratory cross-coupling reactions. By coupling agentic reasoning with rigorous computational validation, ARCHE advances autonomous mechanistic discovery and establishes a foundation for broader machine-assisted chemical research. The code for ARCHE is publicly available at this https URL.
[AI-68] MOSAIC: Query-Aware Exploration Policy Adaptation for GraphRAG
链接: https://arxiv.org/abs/2609.11065
作者: EunKyeong Lee,Kyeong-Jin Oh,Jinwon Kim,Hye Woo Lee,Minsang Song,Hyeongjun Jang,Junyoung Youn
类目: Artificial Intelligence (cs.AI)
备注: 13 pages, 2 figures, 11 tables. Technical Report
Abstract:Graph Retrieval-Augmented Generation (GraphRAG) can connect evidence distributed across a corpus graph, but most systems use largely shared exploration procedures across queries. This creates a structural mismatch: direct facts may need compact local neighborhoods, comparisons need balanced coverage of multiple targets, and mediated questions may require deeper paths through weakly related connectors. We present Mosaic, a training-free framework that formulates GraphRAG retrieval as a per-query control problem. An LLM analyzer converts query-specific evidence requirements into a bounded policy over seed selection, graph traversal, stopping, and evidence selection, while the corpus graph, indexes, scoring functions, grounding procedure, and answer generator remain shared. On GraphRAG-Bench, Mosaic achieves query-weighted Answer Correctness of 76.97 on Medical and 64.33 on Novel, improving over the strongest previously reported overall results by 5.13 and 4.43 points. On Medical, it reaches 95.1 Evidence Recall and 86.1 Context Relevancy. Controlled comparisons on an identical graph and generator show that no fixed narrow, medium, or wide policy is consistently optimal; Mosaic improves by 9.96 points over the strongest canonical fixed policy. Relative to Fixed Wide, it evaluates 81.9% fewer paths and retains 47.2% fewer evidence items. Transfer experiments on HotpotQA, MuSiQue, and 2WikiMultiHopQA further show that the policy interface can be applied without benchmark-specific retriever training. Comments: 13 pages, 2 figures, 11 tables. Technical Report Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2609.11065 [cs.AI] (or arXiv:2609.11065v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2609.11065 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-69] Fork Where the Model Changes Its Mind: Belief-Shift Branching for Tree-Structured Reinforcement Learning
链接: https://arxiv.org/abs/2609.11061
作者: Bin Lei,Yu Li,Prafulla Kumar Choubey,Jiaxin Zhang,Becky Xiangyu Peng,Qinyuan Ye,Kartik Narayan,Caiwen Ding,Silvio Savarese,Chien-Sheng Wu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Tree-structured rollouts give critic-free reinforcement learning with verifiable rewards (RLVR) step-level credit: fork a chain at an intermediate point, and sibling outcome differences estimate step value. Each fork adds sampling cost, so realistic budgets typically allow only a few forks per chain. A fork placed where the outcome is already largely settled yields siblings that mostly agree and provide almost no credit signal; hence, for a given tree size, where forks are placed largely determines how much step-level RL can gain. Most existing mainstream methods place forks by structure, such as fixed lengths, midpoints, and delimiters, or by next-token entropy. We formalize fork placement as locating the \emphpivots of the chain’s value curve, where the expected outcome turns. We propose \emphbelief-shift branching: read the model’s answer belief at candidate boundaries and fork just before the step where consecutive beliefs diverge most. Three instantiations, none needing step-level supervision, span access levels: a black-box probe, a logit-lens depth profile, and a learned activation direction, which is fit offline and therefore used only in the validation before RL training. The signal only \emphplaces forks, and the probe costs about 1% of step compute on mathematics and under 5% on code when it runs inside the rollout engine. In that validation, against Monte-Carlo value curves, a belief-shift signal ranks first in each of the eight model \times benchmark panels, ahead of entropy, structural, and LLM-judge baselines. In RL across three model families and two domains, belief-shift forking leads every mathematics aggregate, on OLMo-3-7B by +2.6 aggregate and +2.9 on AIME 2026 over the strongest baseline, and sweeps every OLMo code column, by +6.5 on LiveCodeBench-medium.
[AI-70] Grounding Agent Memory: Environment-Probing Curation for Enterprise Agents
链接: https://arxiv.org/abs/2609.11060
作者: Susheel Suresh,Hazel Mak,Sahil Bhatnagar,Chhaya Methani,Alejandro Gutierrez Munoz
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注:
Abstract:Persistent memory is entering production-oriented agent platforms to help long-horizon agents accumulate experience across sessions. Yet a post-task curator agent restricted to completed trajectories can preserve errors, overgeneralize partial evidence, or retain stale knowledge. We introduce environment-probing curation, a deployment-compatible extension that gives an existing asynchronous curator agent least-privilege, read-only world tools to check, scope, and refresh candidate memories. It requires no model retraining and leaves the task agent, retriever, memory representation, and production write authority unchanged. In a production-like GitHub Copilot (GHCP) harness built on its SDK, we compare stateless execution, full in-context learning, GHCP + Mem, and GHCP + Mem (w/ Env Probing) on CLBench database exploration and 90 adapted APEX management-consulting tasks. On CLBench, probing raises pass rate from 39% to 73% and pass-discounted reward from 8.60 to 22.60 while reducing queries from 8.8 to 4.7 per question and task-agent cost from \ 3.38 to \ 1.68. Across six APEX worlds, all 18 memory-versus-baseline mean reward comparisons are positive and task-agent tool calls fall by 16–75%; probing gives the best task-agent reward gain per dollar in five worlds. Probing also attains higher mean reward than GHCP + Mem on both Sonnet 4.6 and Opus 4.7 without schema drift. Environment probing therefore turns existing agent-memory curation into an environment-informed, auditable process while preserving a compact task-time interface.
[AI-71] 1: Terminal Agent Reinforcement Learning for Long-Horizon Tasks
链接: https://arxiv.org/abs/2609.11042
作者: Junyao Yang,Yucheng Shi,Zhongzhi Li,Ruhan Wang,Zongxia Li,Haitao Mi,Leowei Liang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 37 pages, 18 figures
Abstract:Agent usage is shifting toward long-horizon tasks such as coding and scientific discovery, among which terminal tasks are especially important. We introduce T1, a Mixture-of-Experts model of 122B total trained with reinforcement learning, operating a real shell in a cloud sandbox for up to 300+ tool-call turns per task, rewarded by executing each task’s own verifier. We provide a comprehensive recipe: First, an aggressively warm-started to stabilize actor-critic training, with a dense process reward scoring trajectories by the absolute number of passing verifiers. Second, stable optimization through TITO construction, training on the exact sampled token identifiers with drift repair at turn boundaries, and rollout routing replay, recording the sampler’s per-token expert choices at every MoE layer and replaying them during training. Third, fully out-of-distribution training corpus: isolated seeds and synthesized tasks disjoint from Terminal-Bench 2.1 ensures gains reflect genuine capability transfer over benchmark overfitting. Together, TITO and R3 cut the training-to-inference log-probability difference from 0.021 to 0.013, with exactly aligned zero token drift in the loss region. On Terminal-Bench 2.1, our post-train pipeline raises initial base model from 43.8% to T1 with 64.0% resolved. On Long-Horizon Terminal Bench, T1 reaches 27.9% and surpasses GPT-5.4 and GLM-5.1.
[AI-72] he Agent Incident Registry: Toward Preventing Repeated AI Agent Failures
链接: https://arxiv.org/abs/2609.11030
作者: Divyanshu Kumar,Rohith HN,Nitin Aravind Birur,Sahil Agarwal,Prashanth Harshangi
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:AI agents increasingly act through tools and delegated authority, but general incident repositories rarely capture the mechanisms needed to compare public failures with agent-security evaluations. We present the Agent Incident Registry (AIR), a source-linked catalog containing \N records of agent-related events disclosed from \Yfirst through \Ylast. Each record includes supporting evidence, a stable identifier, and missingness-aware labels for causal role, disclosure class, mechanism, and outcome. Among the \Nprimary generative-system records in which the agent acted, \Rprimary involved realized harm (\Pprimary%). Realized outcomes concentrate in in-the-wild and safety-failure records, while responsible disclosures and research demonstrations are overwhelmingly demonstrated; the aggregate share therefore characterizes collection composition rather than deployment risk. After initial curation, a second human reviewer checked all \N records and their existing labels for completeness and correctness. In a deployment-analogue audit, InjecAgent’s \NInjecAgentCases cases occupy three of AIR’s twelve surfaces and are all attacker-triggered, whereas AIR contains \Nsafety no-adversary safety failures. AIR supports source-grounded case retrieval and evaluation-scope auditing, not failure-rate or control-efficacy estimation.
[AI-73] BenchShield: Formal Model-Backed Instrumentation for Reward Integrity in LLM -Agent Evaluation Infrastructure
链接: https://arxiv.org/abs/2609.11028
作者: Shenghan Zheng,Zonglin Di,Yimin Liu,Kyoung Whan Choe,Jiankai Sun,Heguang Lin,Penghao Jiang,Yifeng He,Xiao Cheng,Jicheng Wang,Wenbo Chen,Alex Yates,Yinzhe Zhao,Bingran You,Yuan Gao,Ayush Munot,Shubham Gaur,Zhe Ye,Hao Wang,Xiangyi Li,Dawn Song,Christophe Hauser
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Software Engineering (cs.SE); Systems and Control (eess.SY)
备注:
Abstract:LM-agent benchmarks increasingly function as interactive evaluation infrastructure. Agents observe state, call tools, modify workspaces, submit artifacts, and receive rewards from outcome procedures. This interactivity makes evaluations vulnerable to reward hacking: an agent improves its measured score by exploiting the reward-relevant trajectory instead of solving the intended task. Existing defenses rely largely on task-specific patches, prompt instructions, or post-hoc detectors. They do not provide reusable evidence that a concrete run remained within its intended evaluation boundary. This paper presents BenchShield, a model-backed instrumentation layer for reward integrity in LLM-agent evaluation. BenchShield grounds detection in a finite lifecycle model of an evaluation’s reward-relevant events. Within the benchmark infrastructure, two complementary analyses operate over this model. A static, phase-aware taint analysis exposes reward-hacking paths before a run. Its runtime counterpart uses infrastructure-side evidence to attribute concrete agent use and emit evidence-backed claims. We construct BenchShield Trajectories, a human-labeled corpus of 456 adjudicated trajectories from more than 31,000 public agent runs across three benchmarks. Compared with an agentic hackability scanner baseline on the same tasks and model, BenchShield improves full-chain recall from 23-94% to 77-100%, same-vector coverage from 16-56% to 43-78%, and reduces per-task cost by up to 65%. Its runtime analysis achieves 96% accuracy in detecting reward hacking from infrastructure-side evidence. Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Software Engineering (cs.SE); Systems and Control (eess.SY) Cite as: arXiv:2609.11028 [cs.CR] (or arXiv:2609.11028v1 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2609.11028 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Shenghan Zheng [view email] [v1] Thu, 10 Sep 2026 03:10:58 UTC (576 KB) Full-text links: Access Paper: View a PDF of the paper titled BenchShield: Formal Model-Backed Instrumentation for Reward Integrity in LLM-Agent Evaluation Infrastructure, by Shenghan Zheng and 21 other authorsView PDFHTML (experimental)TeX Source view license Current browse context: cs.CR prev | next new | recent | 2026-09 Change to browse by: cs cs.AI cs.SE cs.SY eess eess.SY References Citations NASA ADSGoogle Scholar Semantic Scholar export BibTeX citation Loading… BibTeX formatted citation loading… Data provided by: Bookmark checked="checked"class=“labs-tab-input”> Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv’s community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) mathjaxToggle(); We gratefully acknowledge support from our major funders, member institutions, , and all contributors. About Help Contact Subscribe Copyright Privacy Accessibility Operational Status (opens in new tab) Major funding support from
[AI-74] opological Necessities: Mechanism-Invariant Strategic Subgoals for Cross-Embodiment Goal-Conditioned Control
链接: https://arxiv.org/abs/2609.11014
作者: Hao Shi,Xi Li
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注: 60 pages total (9-page main text + appendices), 16 figures. Code and data: this https URL
Abstract:Long-horizon goal-conditioned reinforcement learning delegates control to a high-level module that proposes subgoals, but existing subgoals are implicit byproducts of value functions or latent actions, tied to the executor that produced them. We study a different object: a route-conditioned order of unavoidable stages that every successful executor must traverse, recoverable from offline trajectories and belonging to none of them. Its defining properties are topological: an unskippable stage is a separating set that every admissible path must cross, and a loop in free space forces a route choice. We read the two by homology in dimensions 0 and 1 over a transport-weighted carrier built from successful trajectories, yielding an enumerable gate set with shell-level certificates; the certified gates are what we call topological necessities. Certified gates enter the decision loop as a recursive topological gate hierarchy. Under a fixed, isomorphic free space, the object survives executor replacement: gates frozen on PointMaze data transfer without retraining to Ant and Humanoid, attaining the highest Humanoid aggregate under a unified interface (96.1), with +36.0 over a map-privileged reference on the multi-route task (p=1.4e-5); the planner saturates PointMaze (100+/-0) and matches or exceeds the strongest baselines on AntMaze (giant +22.9) and Kitchen (+15.8/+12.6).
[AI-75] DeFiFusion: Combining Transaction Events with Smart Contracts to Detect Price Manipulation Attacks
链接: https://arxiv.org/abs/2609.11008
作者: Rui Cao,Shaojing Fan,Liming Fang,Yuchan Liu,Yingying Jiao,Zhenguang Liu
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注: 12 pages, 10 figures
Abstract:Decentralized Finance (DeFi) has emerged as a rapidly growing blockchain-based financial service, where market transaction dynamics and underlying smart contract logic are intricately intertwined. This autonomous interplay, while eliminating centralized intermediaries, significantly expands the vulnerability surface of DeFi protocols to Price Manipulation Attacks (PMAs), which have already inflicted catastrophic financial losses. Despite their gravity, existing detection paradigms suffer from fundamental limitations. Transaction-centric methods lack awareness of contract execution semantics, making them prone to false positives under legitimate market volatility, while static contract analyses ignore real transaction behaviors and frequently report vulnerabilities that are infeasible to exploit in practice. We present DeFiFusion, a dual-modal PMA detection framework that closes this gap by jointly modeling transaction events and smart contract semantics within a unified pipeline. Our core insight is that PMA maliciousness emerges only from the interaction between transaction behaviors and the contract logic they exploit; neither signal suffices in isolation. Accordingly, we derive price-manipulation-aware event encoding for extracting fine-grained temporal and economic features tailored to manipulation patterns. We further introduce LLM-based contract semantic extraction to supply the execution-logic context that prior behavioral methods lack. To fuse these modalities, we propose a Dual-Modal Projection-Fusion Transformer with T5-style relative positional encoding, capturing the cyclic multi-stage execution structures that distinguish PMAs from benign market activity. Extensive experiments demonstrate that DeFiFusion consistently achieves state-of-the-art detection performance, effectively recalling 222 of the 225 PMA cases while maintaining a precision of 96.10%.
[AI-76] Importance Weighting for Unlabeled-unlabeled Learning under Distribution Shift
链接: https://arxiv.org/abs/2609.10994
作者: Atsutoshi Kumagai,Tomoharu Iwata,Hiroshi Takahashi,Taishi Nishiyama,Kazuki Adachi,Yasuhiro Fujiwara
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
备注: 19 pages
Abstract:Unlabeled-unlabeled (UU) learning allows us to learn a binary classifier from two sets of unlabeled data with different class-priors. It is a general framework because it includes a wide variety of supervised learning such as positive-unlabeled (PU) learning, noisy label learning, and similarity-based learning. Existing UU learning assumes that the test and training distributions have the same class-conditional densities. However, this assumption rarely holds in practice due to distribution shifts. This paper proposes a distribution shift adaptation method for UU learning that uses UU data in the training distribution and a few UU data in the test distribution. The proposed method is based on the importance weighting, which minimizes the test risk by using training data with estimated importance weights. Although existing importance weighting methods cannot handle UU data, we show that it can be done in a principled manner. Thanks to the generality of UU learning, our method can handle various learning problems such as PU and noisy label learning under distribution shift within a single framework while existing methods are usually tailored to a specific problem. Moreover, it does not require any assumption of the shift types such as covariate shift. We experimentally demonstrate the effectiveness of the proposed method with real-world datasets.
[AI-77] Demystifying the Privacy-Utility Trade-off in LLM Interactions
链接: https://arxiv.org/abs/2609.10992
作者: Zhenhua Liu,Zhanxu Xie,Junjie Yu,Tong Zhu,Lijun Li,Wenliang Chen
类目: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
备注:
Abstract:The integration of Large Language Models into daily tasks relies on context-rich instructions, inevitably exposing sensitive user information. Current privacy-preserving methods typically employ context-agnostic static rules, causing severe utility degradation. However, the specific mechanisms governing how sanitization impacts downstream performance remain largely underexplored. To address this, we conduct a systematic analysis to deconstruct the privacy-utility trade-off, uncovering three underlying mechanisms: (1) Context-Dependent Utility, which first establishes when to sanitize by revealing that data value shifts from critical constraints to dispensable noise based on user intent; (2) Strategic Adaptation, which subsequently determines how to sanitize by dictating that the choice between removal and replacement depends on the task’s reliance on factual integrity versus structural coherence; and (3) Combinatorial Interplay, which finally extends the protection scope by demonstrating that attributes form a semantic web of synergistic dependencies or antagonistic redundancies. Guided by these insights, we introduce an intent-driven local protection framework. By distilling a lightweight model Veilmind-4B to drive a dynamic extraction-sanitization-restoration pipeline, our approach reaches a low-leakage privacy point while preserving substantially higher response utility than existing privacy-oriented baselines, advancing the privacy-utility trade-off toward the Pareto frontier.
[AI-78] A Mathematical Theory of Prag matic Information
链接: https://arxiv.org/abs/2609.10986
作者: Kai Niu,Ping Zhang
类目: Information Theory (cs.IT); Artificial Intelligence (cs.AI); Robotics (cs.RO); Systems and Control (eess.SY)
备注: 152 pages, 18 figures
Abstract:We propose a pragmatic information theory unifying communication, control, and decision-making. Its core is the isoteleia mapping, formalizing equifinality: distinct semantic paths leading to the same optimal action are pragmatically equivalent. This induces a three-tier hierarchy of syntactic, semantic, and pragmatic information, each abstraction discarding task-irrelevant distinctions. We develop pragmatic entropy, up/down mutual information, channel capacity, and rate-distortion, and prove three coding theorems generalizing Shannon’s classical results. We introduce pragmatic value (VoI) and cost (CoI) of information as decision-theoretic duals to rate-distortion and capacity, respectively, and formulate a Lagrangian dual framework for cross-layer optimization. The pragmatic efficiency bound \mathcalE_p(\lambda)=\sup_R[\Phi_p®-\lambda,\mathrmCoI_p®] quantifies the maximum net utility any resource-constrained intelligent system can extract, thereby establishing a fundamental behavioral capacity limit—generalizing Shannon’s symbol-level capacity to goal-directed action. Extensions to continuous messages yield closed-form Gaussian expressions, while dynamic settings are addressed via a Bellman equation for sequential decision-making. This framework provides a rigorous foundation for task-oriented communication, networked control, autonomous systems, and embodied AI, shifting focus from symbol fidelity to the effectiveness of information in guiding actions, and offers a unified mathematical language for next-generation intelligent systems.
[AI-79] EGGROLL Unrolled: Understanding and Improving Low-Rank Evolution Strategies at Scale
链接: https://arxiv.org/abs/2609.10980
作者: Ege C. Kaya,Abolfazl Hashemi
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Optimization and Control (math.OC)
备注: 47 pages. 6 figures
Abstract:EGGROLL makes evolution strategies (ES) practical for LLMs by replacing dense Gaussian weight perturbations with low-rank Gaussian products, often of rank one. This choice is computationally attractive but geometrically severe: each rank-one perturbation lies in a zero-volume subset of the ambient matrix space, despite having identity covariance. We characterize the mean EGGROLL update field at finite rank and nonzero perturbation radii, then analyze the error of its finite-population estimator. The population field is obtained by applying an explicit resolvent to the gradient of the objective smoothed by the perturbations. We show that the resolvent can introduce a nonconservative component and can reverse the local stability of an optimum. EGGROLL is nevertheless exact on every quadratic objective at every rank and radius. For smooth objectives, its first local finite-rank correction is O(\sigma^2/r) , and nonasymptotic bounds control the resulting field error under smoothness assumptions. Under a local affine model, rank-one perturbations increase the variance of the gradient estimator by only \frac2(m+n+1)mn+1 relative to dense Gaussian ES, or 0.098% for a 4096\times4096 matrix. We then introduce LOO-ROLL, a leave-one-out estimator that preserves the finite-rank population field while replacing EGGROLL’s two antithetic evaluations per direction by one. At equal evaluation cost, LOO-ROLL halves estimator MSE in transformer blocks. At matched wall time across ten post-training settings and models up to 8B parameters, LOO-ROLL improves seven outcomes in individual paired tests, with no significant loss. On the GSM8K test set, accuracy increases from 38.1% to 63.0% at 0.6B and from 65.9% to 80.0% at 8B. Transformer measurements recover the predicted finite-rank variance, while the rank comparisons show no reproducible reward-based advantage for rank eight.
[AI-80] Decoupling Readiness from Release for Tail-Aware Scheduling of Agent ic LLM Workflows
链接: https://arxiv.org/abs/2609.10964
作者: Bochao Feng,Jianjiang Li,Haojie Wang,Lin Qiao,Yinghui Li,Yukun Yan,Jidong Zhai
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注:
Abstract:Agentic LLM workflows consist of sequences of model turns interleaved with tool interactions, so their end-to-end completion time depends not only on inference speed but also on when ready turns are released. Most runtimes release each turn immediately upon readiness. Under contention, this eager release policy can accumulate released but unfinished work; once submitted, those turns can no longer be reordered by the workflow-level policy, increasing tail latency. We present a tail-risk-aware turn release scheduling method that jointly decides which ready turn to release next and how much released but unfinished work to maintain. The method uses a mean–Conditional Value-at-Risk (CVaR) objective to capture the evolving tail risk of unfinished workflows, incorporates online estimates of turn work when prioritizing ready turns, and adapts the released work budget to observed queue pressure. We evaluate the method using real agent execution traces from software engineering tasks across multiple LLMs and workflow arrival rates. The method performs comparably to eager release under light load and substantially reduces the P95 of workflow flow time under contention, achieving up to a (3.50\times) speedup.
[AI-81] What a Random Draw from the MCP Registry Contains and What Tool-Use Benchmarks Contain Instead
链接: https://arxiv.org/abs/2609.10962
作者: Haseeb Mohammed Afsar
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注: 10 pages, 3 figures. Seeded, re-runnable pipeline and per-server outcomes: this https URL ; archived at doi: https://doi.org/10.5281/zenodo.21347997
Abstract:Studies of the Model Context Protocol (MCP) server ecosystem draw their samples in ways that quietly select for servers that work: reference sets, popularity lists, hand-curated frames, or pipelines that repair a server until it starts. We report what an unrepaired probability sample actually contains. From a 24,135-server registry census we draw 400 npm/stdio servers with a published seed and probe each one over the wire. Only 48.8% complete an initialize handshake, against 66.7% for a hand-curated frame measured with the same instrument, and the dominant failure is not missing credentials (13.3%) but servers that never start at all (37.5%). Among the 195 that do run, hard conformance is total: zero fatal JSON Schema violations across 2,766 advertised tools. Optional safety annotations are the real variance, and the tool-level omission rate on a random draw is 58.8% against 41.5% on the curated frame, so curation flatters this figure too. We then compare the tool descriptions these servers advertise against two tool-use benchmark corpora using one method held constant. Real MCP tools show 2.8% near-duplication at cosine 0.70, and all of it lies within single servers: cross-author near-duplication is 0.0% at every threshold tested. BFCL v4 shows 16.7%, of which 16.4 points lie between independently presented tasks. UltraTool shows 0.3%, cleaner than real tools, so this is a property of BFCL and not of synthetic corpora as a class. Separately, 68.8% of raw BFCL rows and 85.6% of raw UltraTool rows are exact name-plus-description repeats, against 0.4% for real MCP, so any statistic computed over these releases without global deduplication measures repetition rather than tools. All figures regenerate from released scripts and a published seed.
[AI-82] AUC Maximization from Biased Positive-unlabeled Data with Confidence
链接: https://arxiv.org/abs/2609.10928
作者: Atsutoshi Kumagai,Tomoharu Iwata,Hiroshi Takahashi,Taishi Nishiyama,Kazuki Adachi,Yasuhiro Fujiwara
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Machine Learning (stat.ML)
备注: 31 pages
Abstract:Maximizing the area under the receiver operating characteristic curve (AUC) is a standard approach to imbalanced binary classification. Although positive and negative data are required for maximizing the AUC, negative data are often difficult to collect in some real-world applications due to privacy concerns or the need for specialized expertise to annotate them. Thus, AUC maximization from positive and unlabeled (PU) data has been attracting attention. Existing methods assume that labeled positive data are unbiased samples from the true positive distribution. However, this ideal assumption is often violated in practice. In this paper, we propose a method to maximize the AUC from biased PU data. To address the bias, our key idea is to exploit \it confidence, i.e., the probability that an instance is positive, associated with the small number of labeled positive data. We derive an estimator of the AUC risk using biased PU data with confidence, enabling AUC maximization under such bias. We further show that the rewritten AUC risk induces a Bayes-optimal AUC ranking even when the available confidence is any strictly increasing transformation of the true posterior probability. We experimentally show the effectiveness of our method on eight real-world datasets.
[AI-83] ReactHuman: A Physics-Grounded Benchmark for Human-Like Reactive Decision-Making in Embodied Multimodal LLM s
链接: https://arxiv.org/abs/2609.10895
作者: Yizhan Li,Jianxin You,Mengyang Xiong,Yinhuan Chen,Zicheng Zhao,Dekun Wu,Dongqing Zhang,Bang Liu
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Reacting to sudden physical hazards (catching a slipping plate, dodging a falling knife) is both a meaningful test of embodied intelligence and a hard requirement for deploying multimodal large language models (MLLMs) as the decision coreof household robots. Existing evaluations, however, probe intuitive physics passively through question answering over videos, or target deliberate, long-horizon tasks such as navigation and rearrangement; none measure whether a model can turn physical understanding into immediate, safety-critical action. We introduce ReactHuman, the first physics-grounded benchmark for human-like reactive decision-making, in which the evaluated MLLM acts as the brain of a simulated humanoid facing sudden household hazards; it spans 17 event families and over 1,000 bit-for-bit reproducible scenes with exact, annotation-free ground truth derived from 240 Hz rigid-body simulation, including adversarial objects whose appearance contradicts their physics (a foam anvil, a steel apple). We further design a five-metric suite that scores each reaction along three axes: reasonable, safe, and physically grounded. We physically execute every committed plan so that decisions have observable consequences. With this harness we evaluate seven representative MLLMs. Results show that reactive safety is far from solved: models mishandle roughly one hazard in three, act from fixed dispositions rather than the observed scene, trust appearance over motion, and miss interception points at meter scale even when the chosen action is correct; none of these failures shrink with model scale. ReactHuman thus offers both a fine-grained diagnosis and a scalable training signal toward physically grounded, safety-aware embodied agents. The benchmark can be found here: this https URL
[AI-84] DriftNet: A Dual-Head Trajectory Transformer for Detecting and Localizing Prompt Injection in LLM Agents
链接: https://arxiv.org/abs/2609.10892
作者: Asif Pinjari,Mithun Paul Saint-Germain
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 15 pages, 8 figures, 11 tables. Companion detector paper to the AgentDrift benchmark ( arXiv:2609.06972 ); dataset at this https URL
Abstract:When an indirect prompt injection succeeds against an LLM agent, the compromise is visible in the agent’s own behavior: a benign prefix of tool calls, a poisoned observation, and a suffix of actions that serve the attacker. An operator needs three facts: where the attack entered, which steps it corrupted, and whether apparent poison was resisted. Existing systems return either a whole-trace verdict or a single unsafe index. We present DriftNet, a dual-head trajectory Transformer that reads a logged tool-call trajectory and answers all three questions in one forward pass: one head classifies the trajectory as compromised or not, and a second assigns every step one of four labels (benign, injection point, hijacked, failed injection). To our knowledge it is the first supervised detector to produce this joint output. A frozen sentence encoder and four identity-free world features embed each step; the trained trunk, under two million parameters and optimized with a class-weighted joint objective over both heads, needs no access to the agent’s model. On the task-disjoint split of the AgentDrift benchmark (12,536 trajectories, 71,024 labeled steps), with a 20-configuration sweep bounding hyperparameter sensitivity to 0.011 F1 and the test part evaluated exactly once, DriftNet reaches trajectory-level F1 of 0.983, exact injection-point recovery on 98.7% of attacked trajectories, hijacked-span IoU of 0.979, zero flags on 218 resisted attacks, and 2.9% flags on hard negatives. A surface baseline retrained on the identical split recovers 11.1% of partial hijacks and 17.1% of delayed executions; DriftNet reaches 98.6% and 93.2% while lowering every false-alarm rate. Reading all 26 residual errors shows that most misses trace to trajectories whose labeled injection observation carries no legible instruction, and we report the benchmark’s measured world-identity regularity alongside the results.
[AI-85] When Validation Stops Learning: Auditing Update Admission for Continual Embodied Agents
链接: https://arxiv.org/abs/2609.10873
作者: Qinzhen Ma,Ruihai Wu
类目: Artificial Intelligence (cs.AI)
备注: 9 pages, 2 figures
Abstract:Independent evaluation can reject harmful policy updates yet also prevent useful continual learning. We argue that update admission must be assessed through both error control and retained learning opportunities at a stated interaction budget. We identify a concrete failure: a range-based confidence gate cannot certify unchanged old-task behavior within otherwise substantial budgets. A standard paired-binomial construction reduces this burden when outcome disagreements are rare. We also specify certified historical-reference promotion and a round-level missed-opportunity metric. In a constructed one-step pushing diagnostic with 32 seeds, fresh paired checks admit 31.6% of a common update stream at 2,000 episodes per stage, versus zero for the range-based gate; unconditional replay nevertheless learns better in closed-loop runs. A separate learned-dynamics stress test distinguishes model bias from feedback-selection error. The contribution is an admission-audit protocol with analytical and synthetic evidence; physical-robot and VLA validation remain open.
[AI-86] No-Box Vulnerability Analysis: Description-only Detection of Indirect Prompt Injection Vulnerabilities in MCP Servers
链接: https://arxiv.org/abs/2609.10854
作者: Zehua Zhang,Jie Hu,Pratham Hegde,Aditya Maheshbhai Gabani,Souradip Nath,Yibo Liu,Siyu Liu,Hongkai Chen,Hulin Wang,Zhuoer Lyu,Chang Zhu,Divij Handa,Yan Shoshitaishvili,Tiffany Bao,Ruoyu Wang,Adam Doupe
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:Conventional vulnerability analysis relies on either system access or dynamic interaction, all of which may be unavailable to third-party analysts auditing closed-source, remotely hosted, critical in situ systems, or commercially gated software. Therefore, we propose a new paradigm of no-box vulnerability analysis in which neither access nor runtime interaction is available, and only functionality metadata is available. Such metadata defines the intended behavior of the system, including its inputs, outputs, and side effects, while constraining the space of implementations consistent with that behavior. We propose hypothesizing about vulnerabilities that exist across all possible implementations of a given system metadata, without observing or interacting with the target system. An analyst can later validate these hypotheses when additional access is available. We showcase the feasibility of no-box vulnerability analysis through implementing a prototype called MCPSEC, which audits Model Context Protocol (MCP) servers for indirect prompt injection vulnerabilities using only the tool metadata exposed at server registration time. We evaluate MCPSEC on 20 widely deployed MCP servers comprising 177 tools, among which human evaluators confirm 95 vulnerable tools. MCPSEC identified 143 tools as vulnerable, and for each vulnerable tool, it produced a hypothesized vulnerability along with exploitation technique. Using metadata alone, MCPSEC predicted 94 (98.9% recall) real verified vulnerabilities, compared against an LLM baseline with 80 (84.2% recall). Overall, our results introduce no-box vulnerability analysis as a new analysis paradigm and demonstrate its practical feasibility in realistic systems.
[AI-87] Counterfactual Marginalisation: Framework for Evaluating Robustness to Nuisance Variables MICCAI2026
链接: https://arxiv.org/abs/2609.10778
作者: Yasin Ibrahim,Hermione Warr,Robin J. Evans,Konstantinos Kamnitsas
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted at UNSURE Workshop, MICCAI 2026
Abstract:Machine learning models can achieve strong test performance while relying on demographic or acquisition-related shortcuts. We propose counterfactual (CF) marginalisation as a test-time evaluation procedure for assessing robustness of classification models to such variables. Given a CF image generator, we intervene on nuisance parent variables such as age or sex, generate CF versions of each test image, and average predictions over a target intervention distribution. This produces intervention-aware predictions that marginalise demographic effects while preserving patient-specific latent information. We use these predictions to define metrics for CF risk, calibration, stability and worst-case sensitivity. We demonstrate this framework’s utility for quantitative robustness evaluation.
[AI-88] Beyond Static Guarantees: Measuring the Static-Pass Dynamic-Fail Gap in Security-Sensitive and LLM -Generated Python Code
链接: https://arxiv.org/abs/2609.10762
作者: Jessica Pourleyli,Maitreyee Das Urmi,Glaucia Melo
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注: Accepted at CASCON 2026
Abstract:Advances in large language models (LLMs) fuel the quest for scalable methods to assess the security of generated and security-sensitive software. Static analysis is widely adopted as a scalable, reproducible, and inexpensive security gate, but cannot directly observe runtime exploit behaviour. Vulnerabilities dependent on adversarial inputs, execution context, or exploit chaining may evade static checks while remaining exploitable in practice, yet passing static analysis is often treated as evidence of secure behaviour. This paper introduces the Static-Pass Dynamic-Fail (SPDF) phenomenon and a three-stage agentic pipeline combining static scanning, LLM-driven Common Weakness Enumeration (CWE) reasoning, and autonomous exploit verification in isolated Docker containers. We evaluate 1,355 Python samples from SecurityEval, RedCode, and CyberNative datasets. Of the 654 samples producing no findings under the composite Bandit-Semgrep gate, the LLM detection stage identified 394 candidate vulnerabilities across 235 files. Dynamic verification confirmed or partially confirmed exploitability in 95 files, yielding an inclusive pipeline rate of 14.53% (roughly 1 in 7 statically clean samples). This rate represents the proportion of Bandit-Semgrep-clean samples for which the pipeline identified a candidate vulnerability and obtained runtime evidence supporting exploitability. Outcomes varied by dataset: among candidate file–CWE pairs, confirmed exploitability was 33.7% for RedCode, 28.6% for CyberNative, and 5.4% for SecurityEval. Several frequently confirmed classes, including CWE-338 and CWE-916, were flagged by neither Bandit nor Semgrep. These findings indicate that static-analysis success and runtime security are hierarchical layers of software assurance rather than interchangeable measures, and have the potential to reshape how AI-generated and security-sensitive code is evaluated.
[AI-89] Adaptive Margin Ordinal Loss: Penalizing Center-Class Hedging in Ordinal Classification
链接: https://arxiv.org/abs/2609.10752
作者: Manisha Kandel
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Standard cross-entropy loss causes neural networks trained on ordinal classification tasks to hedge predictions toward center classes, a failure mode we term \emphcenter-class hedging. This occurs because predicting the middle class minimizes expected symmetric loss, making it the path of least resistance regardless of the true label. Existing ordinal losses address related problems such as large-error penalization and rank consistency, but none directly suppresses center-class hedging as a function of where the true label lies relative to the ordinal center. We propose the Adaptive Margin Ordinal Loss (AMOL), a multiplicative weight applied to per-class loss terms of the form m(k,y) = 1 + \alpha \cdot (1 - |k-c|/c) \cdot (|y-c|/c) , where c is the center class, k is the candidate class, and y is the true label. The weight encodes a joint condition: it is large only when the candidate class is near center and the true label is far from center, collapsing to standard behavior otherwise. We further introduce the Center-Hedging Rate (CHR) as a diagnostic metric that directly quantifies this failure mode. Across four ordinal classification benchmarks and five random seeds, AMOL achieves the best or tied-best Quadratic Weighted Kappa (QWK) on all four datasets compared to cross-entropy, OLL, and SORD baselines. An asymmetric variant (AMOL-asym) eliminates center-class hedging entirely on the Abalone dataset ( \textCHR = 0.000 \pm 0.000 across all five seeds, n \approx 266 extreme-class test samples per run), compared to 0.074 \pm 0.005 for standard cross-entropy.
[AI-90] mporal and Multimodal Deep Learning for Cyberattack Detection in LEO Satellite Systems
链接: https://arxiv.org/abs/2609.10746
作者: Kyle Stein,Guillermo Francia III,Eman El-Sheikh,Hossain Shahriar
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:The growing reliance on Low-Earth Orbit (LEO) satellite communication systems has increased the need for intelligent methods capable of detecting cyberattacks across complex and dynamic space environments. Unlike conventional network intrusion detection, satellite systems generate heterogeneous information across radio-frequency (RF) links, onboard hardware, and orbital operations. However, many existing approaches either rely on terrestrial intrusion datasets or evaluate individual observations independently, limiting their ability to capture temporal attack behavior specific to LEO satellites. In this work, we conduct a systematic study of deep-learning-based cyberattack detection using the recently introduced satellite-specific UNSW-IoTSAT dataset. We investigate structured learning architectures that preserve hardware, orbital, and RF information, including a Subsystem-Fusion MLP and a hierarchical multimodal Transformer that models both cross-subsystem interactions and temporal evolution. We further evaluate leakage-resistant row-level and temporal settings, along with cross-satellite generalization, to characterize how model architecture and evaluation protocol influence satellite cyberattack detection. Experimental results demonstrate the value of structured multimodal modeling and rigorous evaluation, with the hierarchical Transformer achieving up to 91.66% accuracy and 85.63% macro F1 under the leakage-resistant evaluation protocol.
[AI-91] CARTS: Contextual Autoregressive Rank Transcoding Steganography for Full-Capacity Keyed Text Encoding
链接: https://arxiv.org/abs/2609.10744
作者: Wissam Ghantous,Alexander V. Mantzaris
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Autoregressive language models can be used to transform a payload text into a stegotext of identical token length by preserving per-position rank information across contexts - a methodology we formalize as Contextual Autoregressive Rank Transcoding Steganography (CARTS). While the Calgacus construction of Norelli et al. demonstrated this phenomenon experimentally, no formal security analysis existed. This paper provides the first rigorous treatment of CARTS. We show its exact correctness under deterministic model assumptions, introduce a rank-coordinate representation in which keys act as bijections on rank-vector space, define relevant security notions and the computational problems naturally associated with the construction - context search, key collisions, message equivocation, and non-commutativity of the encoding maps - and study the theoretical relationships between them, including the characterization of message equivocation in terms of context search, and the tension between key collisions and message equivocation. An empirical study on Llama 3 8B confirms exact recovery of the original payload in all tested cases, finds no key collisions under random key generation, establishes that a hand-crafted collision is local rather than global, and finds no commuting key pairs - suggesting resistance to the attack vectors studied. This work opens a formally grounded research agenda for the constructive use of language models in cryptography and privacy-preserving communication.
[AI-92] owards a Deterministic Math Solver for Clinical Language Models
链接: https://arxiv.org/abs/2609.10728
作者: Felipe Ocampo Osorio,Sebastián Andrés Cajas Ordoñez,Maximin Lange,Rafi Al Attrach,Sahil Kapadia,Zakaria Laouabdia Sellami,Angelo Antonio Talio,Leo Anthony Celi
类目: Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注:
Abstract:Large language models are unreliable at arithmetic, which is a problem for clinical calculators where a single numerical error changes the recommendation. The standard response is to hardcode each calculator as a validated function, one at a time. We test an alternative: the model does not calculate. Instead, it writes case-specific Python that a restricted local executor runs as a deterministic solver, and the model’s task reduces to deciding how to use it. We evaluate this Program-Solve interface on MedCalc-Bench Verified (1,100 cases, 55 calculators) against direct model arithmetic and a hand-written 22-calculator library, using Qwen2.5-7B and Qwen2.5-32B-AWQ, after auditing the benchmark’s formulas against current clinical guidelines and flagging 16 of 55 with version, use or coefficient concerns. With formulas and gold variables supplied and both routes reading the whole note, handing off to the solver is not a reliable advantage at 7B (75.31% against 72.02%, a paired +3.29 points with a 95% calculator-cluster interval of [-3.49, 10.38]) but is one at 32B (90.53% against 83.47%, +7.05 [0.47, 14.60], clear of zero). The hand-written library is exact on its 440 supported cases but abstains elsewhere (40.0% overall). Adding an executor thus helps some open-weight models more than others even under matched formula, variable and note access, and is not a substitute for verified formulas or reliable variable extraction either way.
[AI-93] An Open Recipe for IMO Gold: Training Nemotron for Olympiad Mathematics
链接: https://arxiv.org/abs/2609.10712
作者: Ivan Moshkov,Stephen Ge,George Armstrong,Wei Du,Sadegh Mahdavi,Igor Gitman
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:We study how model post-training and test-time inference design affect natural-language proof generation for hard olympiad mathematics. Starting from Nemotron 3 Ultra, we train two specialist checkpoints using supervised fine-tuning and reinforcement learning, and evaluate checkpoint choice, verification, and refinement. Based on these findings, we present an open-model test-time-compute pipeline. The system operates entirely in natural language, with no formal prover, external tools, or internet access. Three Nemotron 3 Ultra checkpoints - the general-availability model and two post-trained specialists - power an iterative search that generates, verifies, and refines candidate proofs; a separate high-compute stage then selects each final submission. The system scored 30 out of 42 points at IMO 2026, reaching the gold-medal threshold. We release the two post-trained checkpoints as well as the training data, the training and inference code, the submitted solutions, and Nemotron-IMO-Bench, a new benchmark of 200 novel olympiad-level problems.
[AI-94] Architecting the Secure AI-SOC: A Neurosymbolic Framework for Pipeline Integrity and Threat Mitigation
链接: https://arxiv.org/abs/2609.10707
作者: Anna Gazani,Spyridon Kounoupidis,Panagiotis Katsaros,Nikolaos Kekatos,Grigorios Tsoumakas,Georgios Koutidis
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: 21 pages
Abstract:The integration of Large Language Models (LLMs) into Security Operations Centers (SOCs) streamlines threat intelligence but introduces critical vulnerabilities, notably indirect prompt injection via log poisoning. Adversaries exploit this vector to execute multistep ``promptware’’ kill chains by embedding malicious payloads within system logs to hijack the LLM’s operational logic. Securing this pipeline presents a dichotomy: deterministic defenses are computationally efficient yet semantically blind, while purely neural evaluations introduce prohibitive latency and probabilistic flaws. To address this, we propose a novel neurosymbolic defense-in-depth architecture that ensures end-to-end pipeline integrity. The primary layer employs customized SIEM decoders as a deterministic pre-filter, performing immediate structural sanitization to neutralize volumetric padding and signature-based injections at the ingestion edge. The secondary layer leverages NeMo Guardrails to enforce strict semantic boundaries through self-checking validation on the structured SIEM alerts prior to LLM processing. Furthermore, the framework integrates a closed-loop telemetry system, providing critical Human-in-the-Loop (HITL) visibility into thwarted attacks directly within the SOC dashboard. We present a comprehensive experimental evaluation mapped to the MITRE ATLAS taxonomy, assessing the framework against diverse prompt injections. Our results demonstrate that this synergistic approach effectively dismantles the promptware kill chain - bounding LLM stochasticity with verifiable constraints, and delivering a resilient, highly observable defense mechanism for next-generation AI-SOCs.
[AI-95] Quantifying the Memorization-to-Generalization Transition: Scaling Laws and Phase Structure in Grokking
链接: https://arxiv.org/abs/2609.10657
作者: Anish Kataria
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Neural networks trained past memorization frequently undergo a delayed transition to generalization, a phenomenon known as grokking. Despite theoretical progress on \emphwhy this transition occurs, the quantitative structure of \emphwhen it occurs in hyperparameter space remains uncharacterized. We map the memorization-to-generalization boundary across 384 configurations of two-hidden-layer MLPs on modular arithmetic, fitting a power-law scaling relation for generalization onset time: T_\mathrmgrok \propto H^-0.27, D^-2.04, \eta^-0.50, \lambda^-0.64 ( R^2 = 0.732 ; 0.821 with interactions). The exponent hierarchy reveals that data complexity ( D^-2.04 ) is the dominant driver of regime transition, not model capacity ( H^-0.27 ): doubling data accelerates generalization by \sim4\times , while doubling width yields only \sim1.2\times . A sharp phase boundary at weight decay \lambda \gtrsim 1.0 separates grokking from non-grokking configurations, and weight norm trajectories show monotonic compression during the transition, consistent with implicit regularization selecting low-complexity solutions. These results provide a quantitative foundation for predicting and controlling regime transitions in overparameterized networks.
[AI-96] Understanding LoRA Rank Trade-offs in Diffusion Model Fine-Tuning
链接: https://arxiv.org/abs/2609.10656
作者: Iman Khazrak,Narges Nejad,Mostafa M. Rezaee,Robert C. Green II
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 13 pages, 5 figures, 3 tables. Accepted at CSCE 2026
Abstract:Selecting LoRA rank for diffusion fine-tuning requires balancing quality and compute cost. We present a controlled study on CIFAR-10 using a DDPM U-Net with ranks 2,4,8,16,32, fixed optimization settings, and a reproducible local-folder pytorch-fid protocol. We report FID, trainable parameters, runtime, and GPU memory, then validate trends with extended-budget DDPM runs (20 epochs; ranks 4/8/16) and a Tiny DiT backbone (10 epochs; ranks 4/8/16). Results show moderate ranks are most efficient: rank 4 achieves the best DDPM FID (124.1380), rank 8 is close (124.2136), and higher ranks provide limited gains despite larger adaptation cost. These findings support small-to-moderate ranks as practical defaults under fixed training budgets.
[AI-97] A Multi-Stage Rule-Chaining Framework for Compositional and Interpretable Cognitive Reasoning WWW
链接: https://arxiv.org/abs/2609.10654
作者: Deblina Kar
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Code: this https URL Dataset: this https URL
Abstract:The Abstraction and Reasoning Corpus (ARC) benchmarks cognitive generalization, the ability to infer and apply abstract rules from limited examples. This paper presents a multi-stage rule-chaining framework that performs compositional reasoning across symbolic, structural, and conceptual levels. The framework integrates three complementary solvers: (1) a deterministic rule discovery module that induces atomic transformations through geometric, color, and object-based analysis; (2) a pattern-composition engine that reconstructs outputs via block merging, repetition, and spatial heuristics; and (3) a structural abstraction layer that infers hierarchical and nested relationships across grids. These solvers operate sequentially within a progressive fallback hierarchy, where each stage reuses prior reasoning traces to enhance interpretability and generalization. Training passed for 995 tasks out of 1000, further evaluated on 105 tasks out of 120 and solved 230 test tasks out of 240 ARC-AGI-2 tasks. The system achieved strong coverage across deterministic, compositional, and abstract categories, demonstrating an overall accuracy exceeding 95 percent. The proposed architecture bridges symbolic reasoning and pattern synthesis, providing interpretable insight into cognitive generalization. The results suggest that rule chaining and hierarchical composition can advance machine reasoning toward transparent, human-aligned abstraction without relying on task-specific tuning. Comments: Code: this https URL Dataset: this https URL Subjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG) Cite as: arXiv:2609.10654 [cs.AI] (or arXiv:2609.10654v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2609.10654 Focus to learn more arXiv-issued DOI via DataCite Submission history From: Deblina Kar [view email] [v1] Wed, 9 Sep 2026 15:24:42 UTC (589 KB) Full-text links: Access Paper: View a PDF of the paper titled A Multi-Stage Rule-Chaining Framework for Compositional and Interpretable Cognitive Reasoning, by Deblina KarView PDFHTML (experimental)TeX Source view license Current browse context: cs.AI prev | next new | recent | 2026-09 Change to browse by: cs cs.LG References Citations NASA ADSGoogle Scholar Semantic Scholar export BibTeX citation Loading… BibTeX formatted citation loading… Data provided by: Bookmark checked="checked"class=“labs-tab-input”> Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv’s community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) mathjaxToggle(); We gratefully acknowledge support from our major funders, member institutions, , and all contributors. About Help Contact Subscribe Copyright Privacy Accessibility Operational Status (opens in new tab) Major funding support from
[AI-98] Automating Quadratic Unconstrained Binary Optimization (QUBO) Formulation Generation from Natural Language ICML2026
链接: https://arxiv.org/abs/2609.10629
作者: Niloy Kumar Mondal,Md Rizwan Parvez
类目: Artificial Intelligence (cs.AI)
备注: Accepted at the ICML 2026 Workshop on AI as a Tool for Mathematics, Computer Science, and Machine Learning (AI4Research)
Abstract:Quadratic Unconstrained Binary Optimization (QUBO) is a central formulation for combinatorial optimization and has gained increasing attention due to its compatibility with quantum, hybrid quantum-classical, and quantum-inspired solvers. However, translating natural-language problem descriptions into correct QUBO formulations remains difficult, requiring the identification of binary variables, constraints, objective functions, penalty terms, and suitable penalty weights. This process is time-consuming and often demands substantial domain expertise. To address this challenge, we propose an end-to-end multi-agent framework that automatically generates QUBO formulations from natural-language problem descriptions, supported by structured or unstructured test cases. To evaluate its performance, We also introduce QUBOBench, a benchmark containing 100 combinatorial optimization problems across 12 application domains, curated from peer-reviewed literature, competitions, and canonical NP-hard problems. Experimental results show that our framework achieves 68% accuracy on QUBOBench, outperforming a direct single-call baseline by 22%. Further analysis identifies iterative self-repair as the most important component contributing to improved performance. The data and code are open-sourced at this https URL.
[AI-99] Probabilistic Focal Search: Accelerating Bounded-Suboptimal Search via Lower-Bound Advancement
链接: https://arxiv.org/abs/2609.10584
作者: Minh Vu Duc,Trung Le Huu,Hà Minh Hoàng,Trung Thanh Nguyen,Phuong Khanh Nguyen,Huynh Thi Thanh Binh
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Bounded-suboptimal search seeks a solution within a factor w of optimal while reducing search effort. Focal Search (FS) uses heuristic guidance within FOCAL, the frontier nodes eligible under the threshold w f_\min , but its deterministic policy may leave f_\min unchanged for many expansions. We introduce Probabilistic Focal Search (PFS), which follows the FS guided choice with probability p and expands a minimum- f OPEN node with probability 1-p . The latter branch encourages the lower bound to advance, enlarging FOCAL and admitting nodes that may lead to feasible solutions. By balancing guidance and lower-bound advancement, this mechanism can reduce time to a bounded solution when progress is limited by delayed FOCAL admission. As a secondary transfer experiment, we apply the same scheduler to Dynamic Potential Search, yielding Probabilistic Dynamic Potential Search (PDPS). We benchmark PFS against FS on N-Puzzle, Pancake Sorting, and the Traveling Salesperson Problem (TSP), and evaluate its anytime extension on the Generalized Covering TSP (GCTSP), using multiple w and p values. Across these benchmarks, the largest gains occur when long f_\min plateaus delay useful FOCAL admissions; in such settings, the probabilistic factor may reduce node expansions by about 90% or more (e.g., on N-Puzzle and TSP). For the anytime algorithm family, Anytime Probabilistic Focal Search (APFS) outperforms all tested algorithms in evaluating anytime methods on GCTSP. We also observe that the benefit is smaller when the deterministic search already advances efficiently (e.g., Pancake Sorting), indicating that the probabilistic factor is most useful when FOCAL admission is a search bottleneck. The PDPS transfer shows that the mechanism also transfers to potential guidance, although its common-success effects remain domain- and bound-dependent.
[AI-100] Generative Marketing Mix Modeling: A Causal Inference Framework Linking GEO and GEM to Business Impact
链接: https://arxiv.org/abs/2609.11915
作者: Masahiro Kato,Daiki Honma,Taka Kato
类目: Machine Learning (stat.ML); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Econometrics (econ.EM); Methodology (stat.ME)
备注:
Abstract:Generative artificial intelligence changes how firms reach customers, but standard marketing data do not record how often users see and notice a firm’s name in generated answers. We develop Generative Marketing Mix Modeling (GMMM) to estimate the causal effects of Generative Engine Optimization (GEO) and Generative Engine Marketing (GEM). For GEO, GMMM combines repeated generated answers with question counts, shares of use across generative systems, and notice probabilities. For GEM, it combines records of sponsored placements with notice probabilities. GMMM compares expected business responses under alternative treatment sequences and establishes sufficient conditions for identifying the resulting effects. We investigate the empirical performance of the proposed method using simulated answers to product recommendation in English and Japanese.
[AI-101] Physics-Informed Neural Networks to Infer the Perpendicular Energy Conductivity in the Scrape-Off Layer of Stellarator Devices
链接: https://arxiv.org/abs/2609.11628
作者: J. Gallego(1),P. Protopapas(2),A. Bustos(1),A. Alonso(3),S. Barquero(3),A. Baciero(3),I. Rivera(3),J. A. Moríñigo(1),R. Mayo-García(1) ((1) Departamento de Tecnología, CIEMAT, Spain, (2) Harvard John A. Paulson School of Engineering and Applied Sciences, USA, (3) Laboratorio Nacional de Fusión, CIEMAT, Spain)
类目: Plasma Physics (physics.plasm-ph); Artificial Intelligence (cs.AI)
备注: 18 pages, 10 figures
Abstract:In this work, we develop an inverse Physics-Informed Neural Network (PINN) framework to infer the dependence of the scrape-off layer (SOL) perpendicular heat conductivity on plasma density and temperature, \kappa_\perp(n,T) . The method combines radial profile measurements of electron density and temperature with the residual of a reduced one-dimensional SOL transport equation, so that the inferred conductivity is constrained by both the measurements and the underlying transport model. Three neural networks are trained simultaneously: two reconstruct the temperature and density profiles as functions of the radial coordinate and transported power, while a third represents the effective conductivity as a function of the local density and temperature. The framework is first validated using synthetic data generated from a prescribed conductivity function, allowing the inferred \kappa_\perp(n,T) to be compared directly with the ground truth. The model recovers the imposed functional dependence with errors below 10~% in the data-constrained region. Bootstrap resampling is shown to provide a practical indicator of prediction reliability and consistency. A scan in the number of plasma profiles used for training and the number of radial measurement positions per profile identifies a practical trade-off between reconstruction accuracy and data availability. Finally, the method is applied to an experimental dataset from the TJ-II stellarator obtained with the helium-beam diagnostic. This exploratory application provides an initial estimate of the effective SOL conductivity and illustrates the potential of inverse PINNs for extracting transport information from plasma edge measurements.
[AI-102] Investigating catastrophic forgetting in sound event classification
链接: https://arxiv.org/abs/2609.11447
作者: Riccardo Casciotti,Annamaria Mesaros
类目: Audio and Speech Processing (eess.AS); Artificial Intelligence (cs.AI); Sound (cs.SD)
备注: Accepted in MMSP2026
Abstract:This work investigates a number of approaches to prevent catastrophic forgetting in class incremental learning scenarios for sound event classification tasks. We analyze the problem using architectural and regularization approaches, using FSD50K and AudioSet datasets. We design incremental stages and solutions that selectively protect the kernels of the network from weight updates to prevent catastrophic forgetting, and a dynamic head solution that expands itself each time a new task is learned. The findings show that catastrophic forgetting mainly happens in deeper layers, in particular in the classifier head. For the studied in-domain sound classification problem, the solution that seems to alleviate catastrophic forgetting and is the most efficient is a full freezing of the feature extractor with a fine-tuning of the dynamic head classifier, showing little to no forgetting and great training stability, and a good balance between memory-stability and learning plasticity.
[AI-103] Generative Replay Mitigates Sample Starvation in Quantum Architecture Search
链接: https://arxiv.org/abs/2609.11248
作者: Akash Kundu,Amit Kumar Jaiswal,Sebastian Feld,Prayag Tiwari
类目: Quantum Physics (quant-ph); Artificial Intelligence (cs.AI); Emerging Technologies (cs.ET); Machine Learning (cs.LG)
备注: GenQAS: 38 pages, 7 figures, 2 tables and 1 algorithm in main text
Abstract:Reinforcement learning (RL) can automate quantum architecture search, but its scalability is limited when useful circuit trajectories become rare in the rapidly expanding search space. Existing replay mechanisms reuse observed transitions; the proposed learned model produces additional predicted one step transitions from real state-action seeds. Here we introduce GenQAS, a tensor network-guided RL framework that combines a fixed matrix product state warm-start with prioritized generative replay. A learned local transition model generates synthetic circuit transitions on demand and mixes them with real experience during Double Deep Q-Network updates. Under a random exploration analysis, near ground state circuits occupy a rapidly shrinking region of the accessible state space. We investigate whether real data anchored synthetic replay can improve the effective training signal in this regime. Across chemical Hamiltonian benchmarks from 6 to 12 qubits, GenQAS improves fixed-budget success probability and identifies compact circuits at competitive energy error. At 12 qubits, it improves final success probability by up to 7.0\times over passive replay. On a 15-qubit transverse field Ising model, GenQAS increases success probability from 12% to 21% . In a noisy 6-qubit BeH _2 transfer experiment, generative replay reduces the steps to chemical accuracy by 92.7% . These results show that generative replay can mitigate sample starvation in quantum architecture search and support more resource efficient circuit discovery.
[AI-104] X-RACE: XAI-assisted Recurrent neural network Attribution for Channel Estimation
链接: https://arxiv.org/abs/2609.11211
作者: Abdul Karim Gizzini,Yahia Medjahdi
类目: ignal Processing (eess.SP); Artificial Intelligence (cs.AI)
备注: This work has been submitted to the IEEE Transactions on Vehicular Technology (TVT) as a correspondence paper on 12/08/2026
Abstract:Deep learning models, notably Long Short-Term Memory (LSTM), have demonstrated promising performance in channel estimation for high-mobility vehicular environments. However, their black-box nature and architectural overhead limit trustworthiness and efficiency. Classical explainable AI (XAI) methods rely on costly iterative processes, offering only input-level filtering without addressing architectural fine-tuning. To overcome these limitations, this paper proposes the XAI-assisted Recurrent neural network Attribution for Channel Estimation (X-RACE) framework. X-RACE uses a low-complexity, one-shot dual-optimization strategy to simultaneously evaluate and prune irrelevant input subcarriers and internal hidden units. Furthermore, we propose novel temporal XAI metrics: Saturation Time, Importance Drift, and Relevance Contrast to characterize the LSTM’s learning dynamics and memory convergence. Extensive simulations demonstrate that X-RACE reduces inference complexity by at least 44.1% while improving or preserving Bit Error Rate (BER) performance, outperforming classical XAI schemes.
[AI-105] Exploring Second-Order Pattern Recognition in Speaker Recognition ICASSP2027
链接: https://arxiv.org/abs/2609.11182
作者: Yanze Xu,Wenwu Wang,Mark D. Plumbley
类目: Audio and Speech Processing (eess.AS); Artificial Intelligence (cs.AI)
备注: Submit to ICASSP 2027
Abstract:In classical pattern recognition tasks, neural networks are trained to recognise human-defined patterns for model inputs. Some Explainable AI (XAI) methods can explain other latent patterns that underlie the network’s recognition of inputs as human-defined patterns; in this work, we call these latent patterns second-order patterns, and we propose to discover them. To this end, we apply a hierarchical clustering algorithm to analyse whether representations learned by a speaker recognition network from utterances naturally form hierarchical clusters. Each resulting cluster represents a second-order pattern that characterises how the network recognises some known utterances as speaker identities. All the resulting second-order patterns are then semantically interpreted using the existing Hierarchical Cluster-Class Matching (HCCM) method. Furthermore, we propose a new task, second-order pattern recognition, to identify which discovered second-order patterns characterising known utterances are exhibited by an unseen utterance. To achieve this, we design the Hierarchical Cluster Navigation and Assignment (HCNA) method. HCNA recognises a known second-order pattern as applying to an unseen utterance when the unseen utterance’s network representation lies within the extrapolation space of the cluster regarded as that second-order pattern. Our experiments show that the extrapolation mechanism introduced by HCNA substantially improves performance on the second-order pattern recognition task. Comments: Submit to ICASSP 2027 Subjects: Audio and Speech Processing (eess.AS); Artificial Intelligence (cs.AI) Cite as: arXiv:2609.11182 [eess.AS] (or arXiv:2609.11182v1 [eess.AS] for this version) https://doi.org/10.48550/arXiv.2609.11182 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-106] Less can be More: What Aspects of Speech Drive End-of-Turn Detection
链接: https://arxiv.org/abs/2609.11066
作者: Rini Sharon,Manickavela A,Kadri Hacioglu,Andreas Stolcke
类目: Audio and Speech Processing (eess.AS); Artificial Intelligence (cs.AI); Sound (cs.SD)
备注:
Abstract:In conversational AI, detecting when a speaker has finished talking is crucial for natural turn taking. While recent work incorporates semantics, the relative contribution of different modalities remains unclear. We present a controlled ablation of acoustic, prosodic, and semantic signals for streaming end of turn detection using a lightweight trimodal classifier. Under identical training conditions, the acoustic prosodic combination achieves the best balance of accuracy and latency, achieving utterance F1 of 0.93 with 7.8% false alarms at 400ms median latency. Adding text increases premature detections without improving performance. Feature space analysis confirms that prosodic features have the strongest class separability, while text representations overlap substantially. These findings suggest that turn-taking is primarily conveyed through intonation and silence patterns rather than semantic completeness, enabling faster and more reliable systems without expensive text inference.
[AI-107] A machine-checked proof of the Dong-Yang classification of optimal (n4) binary codes for BSCs
链接: https://arxiv.org/abs/2609.10579
作者: Shenghao Yang,Yanyan Dong
类目: History and Overview (math.HO); Artificial Intelligence (cs.AI); Information Theory (cs.IT)
备注:
Abstract:We present a machine-checked Lean~4 formalization of Dong and Yang’s classification of optimal finite-length (n,4) binary block codes for binary symmetric channels. The formalization was developed mainly by feeding the paper’s proofs to an AI tool. To establish correctness, the authors verified the main theorem statements in Lean and the accepted axioms. This note discusses the corrections and simplifications made to the AI-generated formalization, and records discrepancies found in the paper during the formalization. The Lean code is available at this https URL.
[AI-108] Beyond Verified Answers: Solver-Informed Self-Distillation for Bootstrapping Operations Research Language Models
链接: https://arxiv.org/abs/2609.09957
作者: Rui Zhu,Minglong Cao,Chenyu Zhou,Jianghao Lin,Dongdong Ge
类目: Optimization and Control (math.OC); Artificial Intelligence (cs.AI)
备注:
Abstract:Modern large language models (LLMs) can translate natural-language descriptions into operations research (OR) formulations. Post-training techniques including reinforcement learning and on-policy self-distillation have further improved this capability. However, three limitations remain in training LLMs for OR formulations. First, training commonly relies on synthetic formulations validated by human experts or stronger models, constraining scalable supervision. Second, credit assignment is either coarse or costly: outcome rewards score an entire trajectory without locating the responsible modeling decision, whereas process-level supervision requires an additional evaluator. Third, privileged self-distillation can induce style mismatch by using solver context unavailable at deployment. We find that a model can improve from solver-artifact feedback generated by its own rollouts, making self-distillation a practical, evaluator-free source of dense supervision. Therefore, we propose SOLID: Solver-Informed On-Policy LearnIng through Self-Distillation, a novel framework for self-improving OR language models without verified answers or external evaluators. SOLID executes candidate programs from multiple rollouts, clusters their objectives, and selects a majority-group artifact as a pseudo-reference. The model then performs updates using group-relative advantages and dense self-supervision signals. Across multiple OR benchmarks, SOLID improves solution accuracy for both general-purpose and OR-tuned models over outcome-only group-relative training. These results show that solver artifacts can support scalable self-improvement without trusted answers.
机器学习
[LG-0] From Protocols to Evidence: Bounded Claims for AI in Service of the Common Good
链接: https://arxiv.org/abs/2609.11910
作者: Nitesh V. Chawla,Paulo Benanti
类目: Machine Learning (cs.LG)
*备注:
Abstract:Artificial Intelligence does more than create a governance problem. It can also reveal where institutions have already failed to provide responsiveness, belonging, care, and accountability. Once deployed, AI becomes an intervention in those conditions. It can repair, compound, substitute for, or conceal the failures it encounters. Responsible AI must therefore evaluate both the system and the institutional rupture into which it is introduced. The move from principles to protocols is already underway. The EU AI Act, NIST AI RMF, ISO/IEC 42001, and assurance practices translate commitments into roles, requirements, records, oversight, and assessment. The harder questions are what these protocols actually establish, whose power they leave untouched, and where measurement must stop. Pope Leo XIV’s Magnifica Humanitas provides a broader moral frame centered on dignity, technological power, and the common good. Drawing on that frame, we develop a rupture test that links institutional baselines to system evaluation. We distinguish evidence-bounded deployment, which limits claims to what has actually been evaluated, from measurement-bounded governance, which records constraints that favorable evidence cannot override. Within those limits, RISE AI provides an architecture for making bounded, evidence-based claims about Responsibility, Inclusivity, Safety, and Empowerment. Responsible AI requires better engineering, institutional repair, and continued moral and political judgment.
[LG-1] ART: A Modular Tool for Technique-Aware Audio-to-Tablature Guitar Transcription
链接: https://arxiv.org/abs/2609.11904
作者: Akshaj Gupta,Hwi Joo Park,Andrea Guzman,Shamak Gowda,Samhita Konduri,Jiachen Lian,Robin Netzorg,Gopala Anumanchipalli
类目: Machine Learning (cs.LG)
*备注: ISMIR 2026
Abstract:Automatic Music Transcription (AMT) for guitar remains limited by three challenges: existing systems often fail to capture expressive techniques such as slides, bends, and percussive hits; they often assign notes to incorrect string-fret combinations; and they are typically trained on clean recordings, limiting their generalization to noisy real-world audio. To address these challenges, we propose TART, a modular four-stage audio-to-tablature pipeline consisting of (1) an audio-to-MIDI transcription model, (2) an expressive technique classifier, (3) an audio-conditioned T5 encoder-decoder for string-fret assignment, and (4) an automated tablature generator. We evaluate TART in a zero-shot setting on GuitarSet, EGDB, and two augmented benchmarks, Noisy GuitarSet and Noisy EGDB. Averaged across these four benchmarks, TART achieves 81.35% audio-to-MIDI F50 (+6.67 points over the best prior baseline), 71.8% string-fret Tab F1 (+8.5 points over the best prior baseline), and 54.08% end-to-end Tab F1. To our knowledge, TART is the first framework to generate guitar tablature with both fingering and expressive technique annotations directly from guitar audio.
[LG-2] CausalArena: Benchmarking Causal Discovery in the Foundation Model Era
链接: https://arxiv.org/abs/2609.11897
作者: Zi-Rong Li,Si-Yang Liu,Tian-Zuo Wang,Han-Jia Ye
类目: Machine Learning (cs.LG)
*备注: 47 pages, 19 figures
Abstract:Causal discovery aims to uncover causal structures from data and is fundamental to scientific reasoning and intervention-based decision making. Its evaluation relies heavily on structural causal models (SCMs), which specify a causal graph together with the mechanisms that generate data, yet existing studies differ substantially in graph families, mechanisms, and evaluation protocols. The emergence of causal discovery foundation models (CDFMs) further complicates evaluation: performance may reflect not only causal discovery ability, but also overlap between pretraining environments and test SCMs, making results on fixed synthetic benchmarks difficult to interpret. We introduce CausalArena, a unified and evolvable benchmark for causal discovery under a common protocol. Synthetic SCMs supply controlled breadth over structures and mechanisms; semantic operational SCMs provide human-auditable, semantically grounded environments beyond standard synthetic generators; and formula-grounded SCMs test discovery under explicit scientific mechanisms. Public real-world datasets provide an additional external-validity check. Experiments across classical, neural, and pretrained methods reveal substantial ranking shifts across SCM families and protocols, showing that strong performance in one benchmark regime does not reliably transfer to others. These results highlight benchmark diversity and pretraining–evaluation overlap as central challenges for evaluating causal discovery in the foundation model era.
[LG-3] AdamX: Cosine similarity meets gradient descent
链接: https://arxiv.org/abs/2609.11867
作者: Francisco Caldas,Ruben Belo,Cláudia Soares
类目: Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注:
Abstract:We introduce AdamX, a first-order optimizer that incorporates cosine similarity as an adaptive mechanism for controlling update magnitudes. The proposed method is scalable, model-agnostic, and straightforward to integrate into existing training pipelines. We further introduce a variance rectification scheme that promotes smoother optimization during the early stages of training. Overall, we provide empirical evidence that AdamX achieves competitive convergence rates across a range of benchmark datasets and architectures. Performance is evaluated in terms of the number of epochs required to reach predefined performance thresholds under a fixed hyperparameter budget. Code and Experiments available at: this https URL.
[LG-4] Dynamic language model representations for multi-objective reaction optimisation
链接: https://arxiv.org/abs/2609.11790
作者: Joshua W. Sin,David Ming Segura,Bojana Ranković,Siu Lun Chau,Marius D. R. Lutz,Andrea Anelli,Ryan P. Burwood,Kurt Püntener,Maximilian J. Notheis,Raphael Bigler,Philippe Schwaller
类目: Machine Learning (cs.LG)
*备注:
Abstract:Optimising chemical reactions across multiple objectives, such as yield, selectivity, and safety, is central to chemical synthesis, and model-driven approaches depend critically on how reaction components are represented. Established featurisations are either chemically uninformative, as with one-hot encodings, or, as with molecular descriptors, do not readily extend across chemically distinct components. For structurally and functionally diverse components, it is therefore unclear what a shared representation should contain. Constructing such a representation is itself a challenging research undertaking that must be revisited for each new reaction system. Here we bypass this step by learning the reaction representation dynamically from text. Textual descriptions of reaction conditions are encoded by a fine-tuned language model trained jointly with Gaussian process surrogates, yielding task-adaptive representations within a multi-objective Bayesian optimisation loop. Across nickel- and palladium-catalysed cross-couplings in both sequential and parallel experimentation regimes, this approach reaches optimisation convergence in fewer experiments than descriptor libraries or one-hot encoding. Applied prospectively to a palladium-catalysed cyanation spanning mixed ligand denticity and heterogeneous additives, and to a three-objective asymmetric hydrogenation across chiral iridium and ruthenium catalyst families, two rounds of high-throughput experimentation (192 reactions, under 3% of each design space) delivered conditions translating directly to gram scale in 94% and 84% isolated yield, the latter at 99.6% enantiomeric excess.
[LG-5] Predicting Privacy Leakage from Weight Spectral Density
链接: https://arxiv.org/abs/2609.11780
作者: Richard J. Preen,Jim Smith
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR); Neural and Evolutionary Computing (cs.NE)
*备注:
Abstract:Membership inference attacks (MIAs) are widely used to audit the privacy disclosure risk of machine learning models, however current state-of-the-art attacks require training computationally expensive shadow models, making large-scale privacy evaluation impractical. In this work, we investigate whether inexpensive spectral metrics derived from the heavy-tailed self-regularisation framework can serve as proxies for MIA vulnerability. We evaluate several WeightWatcher spectral metrics on image and tabular classification tasks and compare their relationship with MIA privacy leakage against conventional measures of generalisation. Across datasets, stable rank exhibits a strong positive correlation with overall MIA success, while Log alpha-Norm shows a consistent negative correlation with MIA vulnerability at the low false-positive regime. These associations are observed to be stronger than those obtained using the generalisation gap. The results indicate that neural network spectra may contain information about privacy leakage that is not fully captured by conventional measures of overfitting, motivating spectral analysis as a promising direction for scalable privacy auditing.
[LG-6] Differentially Private EEG Feature Anonymization: A Privacy-Utility Case Study in Clinical Neurophysiology
链接: https://arxiv.org/abs/2609.11777
作者: Noman Sadiq,Mohsen Toorani
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注: 27 pages, 11 figures, 7 tables
Abstract:Clinical electroencephalography (EEG) data are valuable for healthcare research and for developing artificial intelligence (AI)-based clinical decision-support systems, but EEG recordings and derived features may contain sensitive patient-specific information. This creates privacy risks when data are reused, analyzed, or shared across clinical and research environments. Conventional anonymization methods are often insufficient for high-dimensional biomedical signals, since removing direct identifiers does not necessarily prevent re-identification, linkage, or inference risks. At the same time, strong privacy protection may distort clinically relevant signal characteristics and reduce data utility. This paper studies subject-level differential privacy for protecting clinical EEG-derived feature representations using Gaussian and Laplace perturbations. The proposed framework considers three deployment scenarios: client-side anonymization, centralized server-side anonymization, and decentralized local training. Following EEG preprocessing and feature extraction, Gaussian and Laplace perturbations are applied to the resulting patient-level EEG feature representations. The Laplace experiments evaluate the implemented noise scales, while the scales required for formal full-vector calibration are derived separately. The effects of both perturbations are assessed using statistical utility measures and a downstream machine-learning-based utility check. The results show that differentially private perturbation can be integrated into EEG processing workflows, but the selected mechanism, privacy parameters, and sensitivity calibration strongly influence data utility. The study highlights the practical privacy-utility trade-off in DP-based EEG feature anonymization and the challenges of preserving downstream utility in small and imbalanced clinical EEG datasets.
[LG-7] Building py-kvcache: A Performance Characterization of External KV Caching for vLLM with NVMe SSDs
链接: https://arxiv.org/abs/2609.11744
作者: Joseph Kanichai,Tiziano De Matteis,Animesh Trivedi
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注:
Abstract:Prefix caching can reduce the time to first token (TTFT) of long-context LLM requests by reusing previously computed key-value (KV) states, but for short prefixes or fast GPUs, recomputation can be faster than loading from an external cache. We characterize this tradeoff in vLLM across GPU, CPU, and NVMe tiers using synthetic workloads, long-context benchmarks, production traces, and find that cache performance depends on transfer granularity, intermediate memory use, and when transfers enter the request schedule, not only on device bandwidth. These findings motivate py-kvcache, a vLLM KV Offload connector with asynchronous direct I/O, bounded shared staging, and scheduler-aware preloading, which starts disk reads while requests are still waiting, overlapping with compute. At 80k tokens, py-kvcache loading from disk is 2.0x faster than LMCache, with preloading contributing 1.34x. With GPU, CPU, and disk caching enabled, it is 1.23x faster than LMCache and within approximately 4% of the native vLLM KV Offload implementation. LongBench and SCBench show that these benefits extend to irregular prefix chains and multi-turn workloads. Bailian trace replays improve TTFT on a weaker GPU, but on an H100 the average request falls below the break-even point and GPU memory alone retains enough prefixes. External KV caching should therefore be treated as a setup specific admission decision. The py-kvcacheimplementation is available at: this https URL.
[LG-8] Reflex-Informed Neuromuscular Reinforcement Learning for Muscle-Driven Locomotion
链接: https://arxiv.org/abs/2609.11733
作者: Jian Zhou,Xingyu Zhang,Rui Ma,Yu Cao,Shane Xie,Zhi-qiang Zhang
类目: Robotics (cs.RO); Graphics (cs.GR); Machine Learning (cs.LG)
*备注: 28 pages, 13 figures, and 9 tables
Abstract:Muscle-driven locomotion provides a physically grounded approach to generating realistic human movement. However, achieving both physiological plausibility and adaptability to changes in musculoskeletal capacity and external disturbances remains a fundamental challenge. To address this limitation, we propose a Reflex-Informed Neuromuscular Reinforcement Learning framework for muscle-driven locomotion. Within this framework, a fixed phase-dependent reflex controller serves as the underlying neuromuscular control mechanism, while the reinforcement learning policy produces four biomechanically meaningful residual parameters to modulate key reflex gains and thresholds associated with hip swing, knee support, and ankle propulsion according to the current state. Experimental results demonstrate that the proposed framework generates physiologically plausible locomotion with improved kinematic accuracy and dynamic consistency, as well as better bilateral symmetry and stride-to-stride consistency under nominal walking conditions. The learned policy remains robust under muscle weakness and external perturbations without retraining.
[LG-9] Learnware and AI Model Management System
链接: https://arxiv.org/abs/2609.11656
作者: Zhi-Hua Zhou
类目: Machine Learning (cs.LG)
*备注:
Abstract:The transition from file storage to database management systems transformed stored data into managed resources. AI now faces an analogous transition from AI model storage to AI model management. Existing model pools essentially serve as \textitAI model storage systems. What is needed instead are \textitAI model management systems that enable models trained by different developers, for different tasks, with different data, and under different objectives to be identified, reused, and even assembled to address future user tasks. Because AI model developers are generally unwilling to share their training data, such systems should operate without accessing the training data of model developers and, ideally, without accessing raw data of future users. This requirement poses a fundamental challenge: the functionality of a modern AI model may not be fully understood even by the developer who trained it. How, then, can a system identify which models are useful for a given user task, let alone assemble models developed independently for different purposes? At first glance, this objective may appear unattainable. It becomes possible, however, by upgrading the basic unit of management from a machine learning model to a \textitlearnware. \textitLearnware = Model + Specification. The specification, whose assignment transforms a trained model into a learnware, is generated with the help of a machine learning process without disclosing the training data of the developer and has a theoretically established data-preservation property. The \textitLearnware Dock System (LDS) provides a path toward powerful AI model management systems. Because specifications are generated according to a published reference and are comparable across models, they can also serve as an AI model \textitcollaboration protocol through which independently developed models, including intelligent agents, can collaborate.
[LG-10] Musec: MomentUm SpEctral Clipping for Stable Muon-type Training
链接: https://arxiv.org/abs/2609.11655
作者: Zhuanghua Liu,Menglian Wang,Luo Luo
类目: Machine Learning (cs.LG)
*备注:
Abstract:Muon has emerged as a highly effective optimizer for large language model training, often achieving superior convergence and performance compared with the widely adopted Adam and AdamW optimizers. Nevertheless, Muon is prone to training instability due to its spectral flattening, manifested by loss spikes and unbounded growth of model weights. Existing approaches primarily rely on weight or attention-logit clipping, which require architecture-specific modifications and do not directly address instability across all model components. We propose MomentUm SpEctral Clipping (Musec), which replaces Muon’s spectral flattening with spectral clipping: rather than setting all singular values of the momentum matrix to approximately one, Musec clips singular values that exceed a threshold while preserving the underlying spectral structure of the momentum. Our strategy provides an optimizer-level, architecture-agnostic mechanism for stabilizing Muon training. We further develop Soft Musec, an efficient implementation that uses a smooth spectral saturation function approximated by coupled Newton-Schulz iterations. Theoretically, we establish convergence guarantees for Musec in nonconvex nonsmooth stochastic optimization. To the best of our knowledge, this is the first convergence guarantee for Muon-type methods in the nonconvex nonsmooth setting. We provide empirical studies to show that Soft Musec consistently improves training stability over existing Muon variants across a wide range of learning rates and model sizes. Notably, Soft Musec remains stable in settings where existing Muon variants diverge, while matching their performance under well-tuned configurations.
[LG-11] RDDMPI: Residual Denoising Diffusion Model for Probabilistic Multivariate Time Series Imputation
链接: https://arxiv.org/abs/2609.11648
作者: Ramiro Valdes Jara,David Chapman,Adam Meyers
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:Multivariate time series imputation (MTSI) aims to recover missing values in temporal data composed of multiple interdependent variables. This problem is central to real-world applications such as healthcare monitoring, traffic networks, and energy systems. Recent diffusion-based approaches have shown strong potential for probabilistic imputation by learning to generate missing values through iterative denoising. However, most existing approaches perform diffusion directly in the original data space, requiring the denoising network to simultaneously capture global structure, temporal dynamics, and stochastic variability. This makes the generative task unnecessarily complex, especially when modern deterministic imputers can already provide accurate initial reconstructions. To address this limitation, we propose RDDMPI, a conditional residual diffusion framework that operates directly in residual space. Instead of modeling the full missing signal directly, we reformulate probabilistic imputation as a baseline-residual decomposition, where a pretrained model captures the dominant signal and a diffusion process models the residual uncertainty. To better exploit deterministic guidance, \model conditions the reverse denoising process on both the baseline-completed signal and its latent representation, while a reliability-aware conditioning mechanism adaptively controls the influence of baseline information during residual generation. This formulation simplifies the diffusion learning objective, enabling it to focus on structured correction terms rather than reconstructing the full signal. Experiments on multiple benchmark datasets demonstrate that RDDMPI consistently improves both reconstruction accuracy and uncertainty quantification.
[LG-12] A Dataset and Model for Imputing Water Surface Elevation on a Large and Extremely Sparse Spatiotemporal Graph
链接: https://arxiv.org/abs/2609.11580
作者: Ruben Cartuyvels,Karim Douch,Gabriele Bertoli,Mounia El Baz,Artemis Vrettou,Sébastien Lefèvre,Diego Fernandez Prieto
类目: Machine Learning (cs.LG)
*备注:
Abstract:Continuous monitoring of water surface elevation across river networks is critical for flood forecasting, water resource management, and understanding the global water cycle. Yet, the scarcity of in situ gauges across much of the globe constrains the development of reliable modeling frameworks. Satellite altimetry has the potential to alleviate this problem but its use is currently hindered by sparse temporal coverage. To this end, we introduce AmazonSWE, a dataset for training and evaluating large-scale spatiotemporal graph imputation methods that integrates processed satellite altimetry measurements from a range of sources, including the recent wide-swath SWOT sensor. The dataset covers over 19K river sections and 10 years (2016-2026) in the Amazon river basin, with in situ gauges held out for evaluation. Besides contributing a novel real-world use case with the potential for societal impact, AmazonSWE introduces significant technical challenges: with fewer than 1% of sections observed per day, the dataset is far sparser than existing imputation benchmarks, and its directed acyclic river topology is both structurally different from and larger than graphs in existing datasets. We show that prior spatiotemporal graph imputation methods are not adapted to this topology, scale and sparsity, and propose a simple bidirectional selective state space model that outperforms them by sampling connected subgraphs and flattening space and time into a single token sequence with topology-aware positional encodings. Compared to the state-of-the-art published method for SWOT-based WSE densification, which integrates statistics with physical modeling, our model reduces RMSE against in situ gauges by 18-39%, while producing predictions for every river section rather than only those with sufficient nearby satellite coverage.
[LG-13] Particle GFlowNets: Rethinking Generative Marginalization Models UAI2026
链接: https://arxiv.org/abs/2609.11538
作者: Tiago da Silva,Diego Mesquita,Salem Lahlou
类目: Machine Learning (cs.LG)
*备注: Accepted at UAI 2026
Abstract:Generative Marginalization Models (MaMs) have been recently introduced as efficient neural sampling models for any-order autoregressive modelling of discrete distributions. By learning both the marginal and conditional probabilities of a persistent-block Gibbs sampler, MaMs enable fast posterior evaluation with a single neural network forward pass. While prior work has considered MaMs to be distinct from Generative Flow Networks (GFlowNets), a well-established paradigm for inference in discrete stochastic models, we show that they are equivalent. Then, we also extend MaMs’ sampling strategy to non-autoregressive generative processes. In particular, we describe an automatic criterion for full-state rejuvenation of the Gibbs sampler, derived from the Gelman-Rubin statistic, which plays a key role in speeding up learning convergence. Our experiments show that our method, called Particle GFlowNets, markedly accelerates training in large combinatorial spaces.
[LG-14] Generalized Score Matching for Parameter Estimation on Convex Domains
链接: https://arxiv.org/abs/2609.11521
作者: Nishanth Shetty,Saisuchith Mahajan,Chandra Sekhar Seelamantula
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:Maximum likelihood (ML) estimation is a principled and statistically efficient approach for learning probabilistic models. However, for unnormalized models, ML estimation requires evaluating the partition function and differentiating through it, which may not always be tractable. Score matching provides a practically viable alternative that circumvents this obstacle by fitting the score in a way that eliminates dependence on the normalizing constant. We derive the generalized score matching objective on a convex subset of \mathbbR^d constructively starting from Minimum Probability Flow (MPF) learning, and show how classical score matching as well as domain-adapted variants for non-negative data arise naturally within the proposed framework. We show that the resulting objective is a \it proper local scoring rule of second-order, which provides the theoretical guarantee that the true density is recovered when the objective is minimized. Furthermore, for a model belonging to the exponential family, we establish convexity of the objective together with consistency of the finite-sample estimator under standard regularity conditions. Our derivation sheds new light on the scope and applicability of generalized score matching in various problem settings. We compare generalized score matching-based estimators on constrained domains, where the partition function is analytically intractable. We provide experimental results on parameter estimation for model densities belonging to the exponential family defined over convex subsets of \mathbbR^d , and a generative modeling use-case to demonstrate broader applicability of the proposed generalized score matching framework.
[LG-15] DeFiFlowBench: Benchmarking and Improving Safe Executability in Natural-Language DeFi Workflow Synthesis
链接: https://arxiv.org/abs/2609.11504
作者: Abhinav Rajeev Kumar,Harshit Arora,Varun Singh,Manikandan Nanjappan
类目: Machine Learning (cs.LG); Software Engineering (cs.SE)
*备注: Code and benchmark: this https URL
Abstract:A structurally valid DeFi workflow can still authorize a costly trade. We introduce DeFiFlowBench, a benchmark of 207 team-authored prompts for natural-language DeFi workflow synthesis. It measures graph coverage, configuration completeness, and declared safety predicates, then tests supported trade configurations on a local EVM. Direct, constrained, and few-shot prompting produce 14-19 unsafe held-out executions per configuration under a fixed 5% price-impact cap. A slippage bound derived from a quote does not prevent the price impact of the order itself. We propose Koan-Safe, which combines a prompt-only intent parser, a replaceable generator, and structural repair with default safety parameters. On 75 held-out workflow prompts, its hybrid variant scores 0.67 on the static safety proxy, compared with 0.33 for the best baseline. Koan-Safe records no unsafe executions on the saved benchmark outputs. A matched-candidate ablation produces 14-17 unsafe executions when enforcement is disabled. Additional tests expose the limits of default injection: permissive existing thresholds can still authorize unsafe trades. A separately evaluated policy cap addresses this failure on a 36-case diagnostic grid. These results support explicit trade protections and execution-based evaluation, while distinguishing declared safety from a general guarantee.
[LG-16] Combining Synthetic and Real Data for Low-Resource Historical OCR: A Manchu Case Study
链接: https://arxiv.org/abs/2609.11495
作者: Yan Hon Michael Chung,Hanlin Wang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Manchu, now critically endangered, was one of the principal languages of the Qing empire (1636-1912), and its extensive archival record is increasingly digitized but remains difficult to search and analyze at scale. Previous work showed that vision-language models (VLMs) trained only on synthetic Manchu word images can reach 87.4% word accuracy on real Qing manuscripts and prints, leaving a substantial synthetic-to-real gap. This study examines how synthetic and real historical training data should be combined for low-resource OCR. Using 60,000 synthetic and 20,306 real historical word images, we evaluate three pretrained VLMs and a compact convolutional recurrent neural network (CRNN) under four regimes: synthetic-only, real-only, joint synthetic-real, and sequential synthetic-to-real training, following a common checkpoint-selection and archival evaluation protocol. Introducing real training images raises the leading configurations to between 95.09% and 96.28% word accuracy, while no synthetic-only configuration exceeds 87.92%. Synthetic supplementation substantially improves all three VLMs, whereas its marginal effect for the CRNN is sensitive to the training objective. Joint and sequential training yield broadly similar archival accuracy under the tested practical pipelines. A compact CRNN also reaches the leading performance range once real images are available, showing that model scale alone does not determine recognition accuracy. Finally, complementary errors among strong recognizers allow voting to raise accuracy to 98.27% without additional training, while an eighteenth-century Manchu dictionary provides a principled rule for adjudicating disagreements.
[LG-17] Deep operator learning for efficient sampling from invariant measures of stochastic differential equations
链接: https://arxiv.org/abs/2609.11376
作者: Lin Guo,Li Lei,Jingtong Zhang
类目: Numerical Analysis (math.NA); Machine Learning (cs.LG)
*备注:
Abstract:We introduce an amortized neural sampler that combines operator learning with flow methods for sampling. It maps SDE coefficient functions to pushforwards from a reference measure to the invariant measures, enabling efficient sampling across families of stochastic differential equations. Our framework shifts traditional sampling cost to an initial training phase, after which new SDE instances require only one encoder pass and a few ODE solver steps, independent of mixing time. To handle problems in high dimensions, we use Lagrangian trajectory sensors for the coefficient functions and cross attention in the architecture. We also theoretically establish the expressivity and resolution invariance of our framework. Experiments on 1D and 2D SDE families show competitive accuracy with substantial speedups over MCMC in regimes with slow mixing, transfer across sensor counts, and demonstration results on a 64D interacting particle SDE where traditional grid approaches are infeasible.
[LG-18] Local Robustness Quantification for Naive Bayes Classifiers and Generative Forests: a General Approach
链接: https://arxiv.org/abs/2609.11366
作者: Adrián Detavernier,Jasper De Bock
类目: Machine Learning (cs.LG)
*备注:
Abstract:We provide methods for calculating the robustness of the predictions of two types of generative classifiers whose underlying distribution is a Probabilistic Graphical Model (PGM): naive Bayes classifiers and generative forests (a probabilistic extension of random forests). Following the paradigm of robustness quantification, we define the robustness of a prediction as the extent to which the distribution of the classifier can be perturbed without changing this prediction. We consider perturbations obtained by varying the local models of the PGMs within general neighborhoods and focus in particular on epsilon-contamination, total variation distance and chi-squared divergence balls. We test our methods on benchmark datasets, demonstrate that the robustness value of a prediction serves as an indicator for its trustworthiness and compare our approach with other such indicators.
[LG-19] Estimating Inconsistency Response Surfaces under Uncertainty in Cyber-Physical System Development ICDM2026
链接: https://arxiv.org/abs/2609.11331
作者: Johannes Mäkelburg,Tim Schwabe,Maribel Acosta
类目: Machine Learning (cs.LG); Software Engineering (cs.SE); Systems and Control (eess.SY)
*备注: Extended version of the paper accepted at IEEE ICDM 2026; 10 pages + appendix, 11 figures, 3 tables
Abstract:Cyber-Physical Systems (CPS) are commonly represented through multiple interconnected models. During development, CPS consistency requires that shared model elements remain compatible across these models. Uncertainty, for example, due to sensor noise or model abstraction, changes the admissible values of model elements and can introduce inconsistencies, i.e., situations in which models can no longer be jointly satisfied. While existing approaches can determine consistency for a given uncertainty configuration, they provide limited support for systematically exploring, analyzing, and explaining inconsistency across large uncertainty spaces. We address this challenge by reformulating inconsistency as an intervention response modeling problem. Using Saltelli sampling and multi-fidelity Monte Carlo estimation, we generate intervention-response datasets and train a surrogate model that directly predicts inconsistency from the propagated uncertainty geometry. Experiments on 48 scenarios and 10 CPS domains show that the surrogate matches Monte Carlo estimates while reducing evaluation time from milliseconds to microseconds, enabling orders-of-magnitude more response-surface evaluations within fixed computational budgets. Building on the learned response surfaces, we perform sensitivity analysis to identify dominant uncertainty drivers and introduce a gradient-based consistency recourse method to determine minimal uncertainty interventions that restore consistency. The results show that inconsistency under uncertainty can be effectively learned, analyzed, and repaired through response-surface modeling, providing a scalable foundation for uncertainty-aware consistency management in CPS development.
[LG-20] A Dynamic Fusion Large Language Model for Traffic Flow Prediction
链接: https://arxiv.org/abs/2609.11314
作者: Xue Qiu,Jianli Xiao
类目: Machine Learning (cs.LG)
*备注: Accepted by WISA 2026
Abstract:Traffic flow prediction is a core supporting technology for intelligent transportation systems. It uses historical data to infer future traffic dynamics in specific areas, thereby helping to alleviate congestion and improve resource allocation efficiency. Traditional neural networks struggle to break through accuracy limits due to their reliance on singular feature modeling, while large language models (LLMs) suffer from insufficient capture of spatial topological information and mining spatiotemporal correlation. This study proposes a Dynamic Fusion Large Language Model (DF-LLM) for traffic flow prediction. The model incorporates three core components: spatiotemporal embedding module, spatiotemporal fusion module, and LLM backbone. The spatiotemporal embedding module enables synergistic representation of multi-scale spatiotemporal features. The spatiotemporal fusion module integrates spatial topology and dynamic dependencies via graph convolution. The LLM backbone adopts a differentiated parameter adaptation strategy to balance training efficiency and traffic data adaptability. Additionally, it introduces a context aggregation attention module to strengthens global dependencies. More importantly, the LLM backbone takes the residual connections to mitigate the gradient vanishing in deep networks. Experiments show that DF-LLM has achieved better performance by comparing the metrics on all the four datasets.
[LG-21] Polyhedral Geometry of Time-to-First-Spike Neural Networks
链接: https://arxiv.org/abs/2609.11227
作者: Manjot Singh,Guido Montúfar,Gitta Kutyniok
类目: Machine Learning (cs.LG); Combinatorics (math.CO)
*备注:
Abstract:We study the expressivity of spiking neural networks, which provide a natural framework for asynchronous, event-driven computation complementary to conventional feedforward neural networks. We consider the time-to-first-spike model in a setting for which the input-output map is continuous and piecewise linear, with affine pieces governed by causal feasibility constraints that determine which presynaptic spikes occur before a neuron fires. We first show that each neuron’s firing time admits a maxout-like representation with exponentially many, highly constrained affine pieces. We then formalize causal regions as polyhedral regions with fixed causal sets and derive upper and lower bounds on the maximal number of causal regions in both shallow and multilayer feedforward spiking networks. Our theoretical and experimental results show that spiking networks can generate richer partitions of the input space than conventional feedforward ReLU networks.
[LG-22] Legible Failures: Detecting and Repairing In-Context Binding Errors
链接: https://arxiv.org/abs/2609.11216
作者: Manas Venkata Sai Ravulapalli,Samrath Singh Chadha,Abhinav M. Hari
类目: Machine Learning (cs.LG)
*备注:
Abstract:A wrong answer does not show whether the model lacked the needed information or held it and failed to use it. On an entity-obligation binding task, a language model can emit an incorrect prompt-supplied binding while a linear probe can recover the correct one from its frozen hidden state. We measure how often this occurs across 16 public checkpoints, each evaluated with three seeds. We fit a probe on a training fold, select its layer on a validation fold, and report results on a disjoint test fold. On the trials each model gets wrong, probe accuracy exceeds the strict present-obligation baseline, 1/K = 0.125, by +0.196 (95% CI [+0.101, +0.296], bootstrapped over models). A query-entity counterfactual rules out token presence and recency. A score built from the sign of probe-output disagreement improves failure detection over the model’s own confidence by +0.079 AUROC (95% CI [+0.036, +0.126]). Raw probe confidence gives no measurable improvement over model confidence. Steering the residual stream toward the probe-decoded binding, with no gold label, raises accuracy on all eight models tested by a mean of +0.168 (95% CI [+0.066, +0.280]). Where recent studies report that probe-detected errors are resistant to interventions, we find that in-context binding is a setting in which probes are actionable.
[LG-23] Diversity of EML-type operators
链接: https://arxiv.org/abs/2609.11210
作者: Andrzej Odrzywołek
类目: ymbolic Computation (cs.SC); Machine Learning (cs.LG); Logic (math.LO)
*备注: 25 pages, 2 figures, see also the TNG Big Techday conference recording at this https URL . Wolfram Mathematica implementation of a Goldstern-type single operator in the Appendix. Follow-up to arXiv:2603.21852
Abstract:The discovery of the EML operator, sufficient to evaluate the standard explicit purely transcendental elementary functions, has led to considerable interest and discussion across multiple scientific disciplines. However, most authors have focused on the binary EML itself, while numerous similar variants with slightly different properties are now known. This article attempts to close this gap by enumerating and classifying them. We also take this opportunity to clarify common misconceptions related to the EML operator. The principal goal, symbolic regression within an architecture as close as possible to proven neural networks which combine matrix multiplication with a single univariate non-linear activation function, remains beyond reach. Instead, we propose a Möbius layer, with rational functions replacing matrix operations, and showcase the recently discovered activation function eml(x,1/x), which allows exp(x) and ln(x) to be recovered separately, and hence all elementary functions to be evaluated within a rational generalization of the neural network.
[LG-24] Convex Optimization with Nested Evolving Feasible Sets (CONES) under Time-Varying Loss Functions
链接: https://arxiv.org/abs/2609.11207
作者: Rahul Vaze
类目: Machine Learning (cs.LG); Data Structures and Algorithms (cs.DS); Optimization and Control (math.OC)
*备注:
Abstract:Convex Optimization with Nested Evolving Feasible Sets (CONES) was introduced in \citeCONESVaze where the objective function (f) remains fixed but the feasible region evolves over time as a nested sequence (S_1 \supseteq S_2 \supseteq \cdots \supseteq S_T). The goal of an online algorithm is to simultaneously minimize the regret with respect to hindsight static optimal benchmark and the total movement cost M_\cA(T) while ensuring feasibility at all times. CONES is an optimization-oriented generalization of the well-known \emphnested convex body chasing (NCBC). In this paper, we extend CONES to allow for loss functions f_t’ s to also change over time. When all loss functions are convex, we show that the projected proximal algorithm achieves O(T^1-\beta), O(T^\beta) simultaneous regret and movement cost, respectively, for any \beta \in [0,1) , over a time horizon of T . We also show that any \it weakly adaptive online algorithm with O(T^\beta) regret has a movement cost of \Omega\left(T^\frac1-\beta2\right) for any \beta \in [0,1) . When all loss functions are strongly convex, we show that the projected proximal algorithm simultaneously achieves O(1) regret and a movement cost of O(\log T) . To complement this, we show that any online algorithm with sublinear \it anytime regret has a movement cost of \Omega\left(\log T\right) .
[LG-25] Hierarchical Clustering Can Jointly Satisfy Richness Consistency and Scale Invariance
链接: https://arxiv.org/abs/2609.11173
作者: Daichi Kuroda,Maximilien Dreveton,Matthias Grossglauser,Patrick Thiran
类目: Machine Learning (cs.LG); Methodology (stat.ME); Machine Learning (stat.ML)
*备注: 51 pages, 3 figures
Abstract:Despite its ubiquity, clustering lacks a universally accepted definition of what is a cluster. Kleinberg’s Impossibility Theorem formalizes this difficulty by showing that no flat clustering method can simultaneously satisfy three natural axioms: scale invariance, richness, and consistency. In this paper, we ask whether this impossibility persists when the output is a hierarchy rather than a single partition. We show that, in contrast to the flat clustering setting, the hierarchical analog of these axioms are jointly satisfiable. In fact, there exist uncountably many hierarchical clustering methods satisfying these axioms, which we call admissible. We explicitly construct several admissible methods, including methods based on well-separated clusters and a non-binary version of single linkage. For certain pairs of admissible methods, the hierarchy produced by one always refines that produced by the other. This refinement relation defines a partial order on the class of admissible methods. This partially ordered set has no greatest element and contains uncountably many pairwise incompatible maximal elements, revealing substantial diversity among admissible methods. Nevertheless, this diversity is constrained: every admissible method contains a hierarchy of sufficiently well-separated clusters, and every finite collection of admissible methods shares such a nontrivial common backbone.
[LG-26] Semi-Tensor Product-Based Multi-Term Randomized T-SVD and Its Visual Applications
链接: https://arxiv.org/abs/2609.11168
作者: Xingchen Xiao(1),Feng Zhang(1),Wenjin Qin(1),Jianjun Wang(1) ((1) School of Mathematics and Statistics, Southwest University, Chongqing, China)
类目: Machine Learning (cs.LG)
*备注: 47 pages, 13 figures, 3 tables
Abstract:Tensor singular value decomposition (T-SVD), which is built upon the tensor-tensor product (t-product), has emerged as a powerful tool for processing high-dimensional visual data such as color images and videos. However, the standard t-product imposes strict dimensional compatibility constraints. Although extensions based on the semi-tensor product (STP) relax this restriction, their single-term formulations still suffer from limited approximation accuracy. Moreover, these deterministic methods incur high computational costs when processing large-scale tensor data. To address these issues, this paper introduces a novel semi-tensor product for third-order tensors under the t-product framework induced by arbitrary invertible linear transforms. The resulting tensor semi-tensor product breaks the rigid dimension matching requirement of the standard t-product, while retaining the closed-form property of T-SVD. Based on this construction, we develop a multi-term semi-tensor product singular value decomposition (MSTP-SVD), which integrates multiple orthogonal decomposition terms to significantly improve low-rank approximation accuracy compared with single-term schemes. To reduce the computational cost of multi-term modeling, we incorporate randomized projection and power iteration techniques into the MSTP-SVD framework, yielding an accelerated multi-term randomized semi-tensor product SVD (MRSTP-SVD) algorithm that achieves a balance between reconstruction accuracy and computational efficiency. Experiments on image and video compression and completion tasks demonstrate the effectiveness of the proposed method.
[LG-27] When does a spectral prior help graph learning? Connectivity-loss estimation under road-network disruptions
链接: https://arxiv.org/abs/2609.11166
作者: Van-Truong Le
类目: Machine Learning (cs.LG)
*备注: 24 pages, 7 figures, 7 tables. Code and data: doi: https://doi.org/10.5281/zenodo.22307723
Abstract:Rapid evaluation of many simultaneous road-link disruptions requires a practical compromise between exact spectral recomputation and local approximation. We estimate relative algebraic-connectivity loss after multi-edge deletion using graph neural networks (GNNs) that learn a bounded correction to a first-order Fiedler sensitivity. The study considers independent, spatially clustered, and edge-betweenness-targeted failures, with graph-disjoint synthetic splits and zero-shot transfer to 13 OpenStreetMap (OSM) areas in six countries. GCN, GraphSAGE, and edge-aware MPNN backbones are compared with analytical baselines. In expanded OSM tests, residual GCN improves spatial-failure MAE by 0.0391 (95% hierarchical interval 0.0151-0.0662), while residual GraphSAGE improves targeted-failure MAE by 0.0257 (0.0095-0.0446). Second-order perturbation improves first-order MAE by only 0.0028-0.0053. Correction slopes decrease under targeted transfer, indicating residual shrinkage around systematic prior error. Leave-one-country-out OSM-to-OSM transfer is mixed: residual GCN improves targeted-failure MAE by 0.0622 (0.0169-0.1153) but worsens the spatial point estimate. Sparse scaling extends to 20,000 nodes and separates one-time spectral setup from amortized screening cost. These results characterize the spectral residual as a useful but domain-sensitive inductive bias for structural connectivity screening. Code, cached networks, and reproducibility artifacts are archived at doi:https://doi.org/10.5281/zenodo.22307723.
[LG-28] Bidirectional Multimodal Fusion of Sky Images and Time-Series for Solar Forecasting with Large Language Models
链接: https://arxiv.org/abs/2609.11135
作者: Ken Chen,Maneesha Perera,Wei Wang,Sachith Seneviratne,Hansani Weeratunge,Saman Halgamuge
类目: Machine Learning (cs.LG)
*备注:
Abstract:Short-term photovoltaic (PV) power and global horizontal irradiance (GHI) forecasts are essential for effective dispatch, reserve scheduling, and grid operations. At these forecasting horizons, errors are predominantly driven by cloud induced ramps: relying solely on historical numerical data may struggle to anticipate an incoming cloud, making ground-based sky images a crucial complementary physical signal. Furthermore, forecast performance is highly sensitive to location and local observing conditions, creating a strong need for site-specific data that are often scarce. Recently, large language models (LLMs) have demonstrated competitive performance and high data efficiency in time-series forecasting. Despite their success, existing LLM-based forecasting methods remain predominantly unimodal, relying primarily on historical numerical time-series data. Effectively incorporating sky imagery into an LLM-based forecasting framework remains under-explored and an open challenge. In this paper, we propose SolCloudLLM, an LLM-based multimodal forecasting framework. SolCloudLLM aligns sky-image patches with time-series patches and fuses their corresponding representations through bidirectional multimodal fusion, yielding a unified representation that is subsequently mapped into the embedding space of an LLM. Extensive experiments on the SIRTA and SKIPP’D datasets demonstrate that SolCloudLLM consistently outperforms the best baseline methods in MSE across all forecasting horizons, achieving a maximum relative MSE reduction of 25.4%. Stratified analysis further indicates that the benefits of multimodal fusion are concentrated primarily under cloudy conditions. Notably, SolCloudLLM achieves the best performance in nearly all few-shot settings, whereas other deep learning baselines experience substantial performance degradation and are frequently outperformed by the non-learning physical method.
[LG-29] Phase-Decoupled Model-Calibrated Power Control for Disaggregated LLM Serving
链接: https://arxiv.org/abs/2609.11133
作者: Jae Gon Kim,Donghoon Yoo,Hanyul Ryu,Sungho Ha,Juyeon Lee,Soojung Ryu
类目: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC)
*备注:
Abstract:Datacenter GPU power is the binding constraint on LLM serving capacity, and production serving has shifted to prefill/decode (PD) disaggregation. Deploying NVIDIA’s Max-Q inference profile on a disaggregated B200 system, we found its realized gain modest (+8.6% tokens/J), model-dependent, and carrying a mean end-to-end latency cost (+5.2%) that throughput-only evaluation does not surface; the profile also applies one setting to prefill and decode GPUs that operate in opposite hardware regimes. We hypothesize that the optimal power setting is a property of the deployed (model, quantization, engine, hardware) combination rather than of the GPU class, that each lane warrants its own profile, and that converting SLO headroom into energy safely requires latency-gated calibration under a runtime SLO guard rather than a fixed recipe. We present a phase-decoupled, model-calibrated controller: the prefill lane runs under an SM-clock window whose floor is a latency guarantee by construction, and the decode lane under a power cap placed by automatic calibration just above a measured throughput/latency cliff. Because a disaggregated decode lane draws flat, memory-bound power, the cap binds continuously, the reactive-overshoot weakness that led POLCA to reject capping is absent, and the GPU’s own power manager retains throughput under the cap. On an 8x B200 node serving Qwen3-Coder-480B (FP8) under agentic load, our balanced mode delivers +20.4% tokens/J at +3.5% mean e2e versus +8.6% at +5.2% for Max-Q, a Pareto improvement on both axes. On Qwen3-235B-A22B (NVFP4) every operating mode meets the ITL-p99 SLO in every repetition; both vendor profiles miss it. A decode-actuator A/B shows the calibrated cap beats static clock locks, and a three-day sustained run saves 32.3% of a lane pair’s electricity. Both models are MoE; a dense model recovers roughly 5x less, so we scope our claims to MoE serving.
[LG-30] How Wrong Can a Good Predictor Be? Diverging Updates with Vanishing Predictive KL
链接: https://arxiv.org/abs/2609.11132
作者: Qifu Wen,Shuaijun Liu,Zihan Zhou,Xi Zeng,Ningxin Su
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 28 pages, 3 figures, 8 tables
Abstract:Accurate posterior prediction need not require accurate approximation of Bayesian updates. We prove that an unbounded gap between the update maps can coexist with vanishing predictive KL for every fixed finite K\ge2 in a stationary symmetric Gaussian HMM. Exact Bayesian mixing and an explicit deterministic radial filter act on the same K-1 belief coordinates. As q\to0^+ , their separation in centered logits in the worst case grows at least linearly in the natural confidence scale L_K(q) , while their categorical D_\mathrmKL(\mathrmexact|\mathrmradial) vanishes at the same explicit witness. Along stationary HMM trajectories, the expected terminal KL between filtered posteriors also converges to zero at H(q)=\lceil-\log(q)/c\rceil+1 . Typical blocks without switches drive both filters into a common confidence cone, where softmax curvature suppresses their disagreement; a single Gaussian maximal event controls adaptive noise. A sweep with equally spaced Gaussians over K\in\2,4,8\ illustrates the opposing trends, and binary controls at long horizons compare saturating and nonsaturating recurrences. The result isolates two missing links between internal update gaps and predictive cost: the contribution of separating states to expected loss and decoder sensitivity. Thus even an unbounded internal update gap does not by itself certify predictive failure. The construction is fixed in K and does not provide a universal criterion for when compression is harmless or characterize when internal gaps must incur task loss.
[LG-31] HERALD: High-Fidelity Exemplar Retrieval with Adaptive Landmark Distillation for Heterophily-Aware Graph Condensation
链接: https://arxiv.org/abs/2609.11123
作者: Sujan Chakraborty,Priyanka Saha,Saptarshi Bej
类目: Machine Learning (cs.LG)
*备注:
Abstract:Graph condensation aims to produce a small surrogate graph that preserves the downstream node-classification performance of a much larger original graph. Existing methods rely on Weisfeiler-Lehman neighbourhood aggregation or gradient-based distribution matching, both of which assume that adjacent nodes share the same label, an assumption that breaks down under heterophily. We propose HERALD (High-fidelity Exemplar Retrieval with Adaptive Landmark Distillation), a gradient-free graph condensation framework that adapts the node scoring and feature selection in the condensation pipeline to the graph’s measured heterophily. HERALD selects features via a joint Fisher-discriminability and activation-density criterion that down-weights aggregated representations on heterophilic graphs, and scores nodes by a weighted combination of prototype representativeness, decision-boundary proximity, and Local Intrinsic Dimensionality (LID), where the weights are driven by a smooth sigmoid function of the heterophily ratio. Nodes are then assembled into a condensed subgraph through score-ordered BFS expansion, Personalised PageRank pruning, and class rebalancing, all at an identical storage budget to BONSAI, enabling direct comparison. Experiments on eight benchmark datasets spanning homophilic and heterophilic settings show that HERALD matches or outperforms state-of-the-art condensers on heterophilic graphs and remains competitive on homophilic ones across four GNN architectures.
[LG-32] EMMI: Edge Multi-Modal Intelligence for Communication-Efficient MLLM Inference via Fused Representation Compression
链接: https://arxiv.org/abs/2609.11058
作者: Motahare Mounesan,Irfan Khan
类目: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC)
*备注:
Abstract:Recent advances in multimodal large language mod- els (MLLMs) have opened new opportunities for edge intelligence by enabling reasoning across heterogeneous sensor modalities, such as vision, text, and telemetry data. However, deploying these capabilities on resource-constrained edge platforms remains challenging due to the substantial computational, memory, and communication demands of modern MLLMs. Rather than transmitting raw sensor observations or partitioning neural networks at intermediate layers, Edge Multi-Modal Intelligence (EMMI) communicates a compact representation between edge devices and server resources, enabling communication-efficient edge MLLM inference. To achieve this, EMMI performs modality-specific encoding, cross-modal representation fusion, and learned compression at the edge, transmitting only a compact latent representation to server-side resources for high-capacity MLLM reasoning. This representation-centric design reduces communication overhead, preserves local data privacy, and provides a fixed-size interface between heterogeneous edge devices and server-side MLLMs. Evaluation on a representative multimodal benchmark demonstrates that EMMI can reduce the communication payload by 32x while maintaining comparable downstream accuracy, resulting in up to a 3.4x reduction in estimated end-to-end inference latency under bandwidth-constrained edge conditions.
[LG-33] A variational physics-informed graph neural network for heterogeneous solid mechanics
链接: https://arxiv.org/abs/2609.10983
作者: Aashay Rajan Yadav,Amiya Prakash Das,Ratna Kumar Annabattula
类目: Numerical Analysis (math.NA); Machine Learning (cs.LG)
*备注:
Abstract:Stress localization in heterogeneous solids is governed by the bimaterial interface, where the displacement field remains C^0 -continuous, while in-plane stresses jump due to the stiffness mismatch. Coordinate-based physics-informed neural networks (PINNs) represent this jump via a prescribed regularization width or a weighted interface penalty, making their accuracy sensitive to how phase-contrast changes are handled. This work presents a variational, label-free physics-informed graph neural network (PI-GNN) in which the heterogeneity is carried by the discretization rather than by the trial field. The solver operates on a conforming adaptive mesh graph, assigns constitutive behavior per element, and minimizes the discrete total potential energy as a single unweighted objective in which only first derivatives appear. The discrete energy on piecewise-linear elements coincides with the finite element (FE) Ritz functional. Dirichlet conditions are enforced by construction, with no penalty term, no interface weight, and no prescribed transition width. Using one fixed architecture, optimizer, and loss across small-strain elasticity and finite-strain Neo-Hookean hyperelasticity in two and three dimensions, the von Mises error remains below 3.58% across a stiffness-contrast sweep spanning (E_\mathrminc/E_\mathrmmat\in[10^-2,10^2]) , where a strong-form PINN degrades to 5.58% , and its displacement error reaches 7.66% against 0.49% for the PI-GNN. A trained network halves the ( \sigma_xx ) error of an energy-based PINN ( 5.01% versus 10.94% ). Training cost exceeds a single FE solve by more than an order of magnitude, so the construction is a variationally consistent, penalty-free interface representation for parametric surrogates and inverse identification rather than a replacement for a one-off FE analysis.
[LG-34] hompson Sampling for Non-Monotone Convex Ridge Bandits: Monotonicity Is Not Needed for Polynomial Regret
链接: https://arxiv.org/abs/2609.10981
作者: Xuan Li
类目: Machine Learning (cs.LG)
*备注:
Abstract:Bakhtiari, Lattimore and Szepesvári (COLT 2025) proved that Thompson sampling (TS) has Bayesian regret \tilde O(d^5/2\sqrt n) for bandit convex optimisation with convex \emphmonotone ridge losses f(x)=\ell(\ipx\theta) , and asked whether monotonicity of the link is necessary. We give a qualitative negative answer. For every prior on [0,1] -valued, 1 -Lipschitz convex ridge losses with an arbitrary convex, possibly non-monotone, link, and for any fixed measurable selection of minimisers, exact-posterior TS has Bayesian regret O\big((d+1)^4\sqrtdn,\log(e+nd\max\1,\diam K)\big)=\tilde O(d^9/2\sqrt n) . The monotone proof relies on a single-removal John-ellipsoid dichotomy; we show by an explicit twelve-point configuration that this dichotomy fails for non-monotone links, and replace it by an O(d^2) cardinality bound for ``uninformative’’ configurations. The bound uses a Boolean rounding argument: a 0 - 1 matrix within 1/(4r) in max-norm of a rank- r matrix has rank at most 2r-1 . We construct d(d+1) uninformative losses, showing that the cardinality bound is tight up to constants in the large-diameter-to-gap regime, and give a self-contained information-ratio-to-regret transfer that is uniform over fixed measurable selections. Whether the d^5/2 dependence of the monotone case can be retained remains open.
[LG-35] Phases in a class of associative memories via hidden neurons
链接: https://arxiv.org/abs/2609.10976
作者: Toshihiro Ota,Masato Taki
类目: Machine Learning (cs.LG); Disordered Systems and Neural Networks (cond-mat.dis-nn); Neural and Evolutionary Computing (cs.NE); Machine Learning (stat.ML)
*备注: 43 pages, 5 figures
Abstract:Associative memory in the Hopfield network is attractor dynamics in a disordered many-body system, and higher-order and exponential extensions turn its retrieval update into softmax attention. The polynomial and exponential regimes have been analyzed by different methods, with no common architecture in which to ask what fixes the storage scale. In this paper we study the bipartite architecture of Krotov and Hopfield, which we call the class H , whose model is fixed by a Lagrangian for each layer, taking the hidden neurons as the order parameter of retrieval. At polynomial load the replica method yields the replica-symmetric phase diagrams and closed-form capacities, and the crosstalk moment is common to Ising and spherical visible neurons, so their differences come from the visible entropy. With a softmax hidden layer the load is exponential, and a copy representation maps the thermodynamics onto random-energy-model counting, with paramagnetic, condensed, and frozen phases. Heating destabilizes retrieval by quantized reassignments of attention, and typical Gaussian patterns remain metastable at every load. The regimes differ in their crosstalk statistics, central-limit at polynomial load and large-deviation at exponential load, and the class H splits retrieval into two roles, the visible Lagrangian fixing stability and the hidden one the storage scale, two axes that may also guide the design of new Lagrangians.
[LG-36] When More Is Not Better: Component Anti-Synergy in a P300 Speller
链接: https://arxiv.org/abs/2609.10961
作者: Lucas Yang,Rui Liu,Fusheng Wang
类目: Machine Learning (cs.LG)
*备注:
Abstract:P300 brain-computer interface (BCI) spellers can provide hands-free communication for people with severe motor impairments. Modern pipelines combine multiple individually promising components, often assuming that ‘more-is-better’. We tested this assumption using a four-component full-factorial experiment varying the inclusion of Euclidean Alignment (EA), xDAWN spatial filtering, subject calibration, and language model priors on a public P300 dataset. Performance was evaluated using accuracy, repetitions, and information transfer rate (ITR) with mixed-effects models. Results show that the value of components is conditional rather than additive. Calibration was the strongest singular contributor, while EA compensated for its absence in zero-calibration settings. Adding independently useful components could also reduce performance, revealing component anti-synergy. Contrary to conventional wisdom, LM support was not universally beneficial: its effect depends strongly on the strength of the underlying EEG pipeline, while results from a larger LM showed a similar pattern. Together, these findings challenge maximal ‘all-on’ pipeline design and highlight the value of selecting spatial and language-support components according to the quality of available EEG evidence.
[LG-37] Measuring the Value of World-Model Updates: A Counterfactual Utility Protocol for Continual Adaptation
链接: https://arxiv.org/abs/2609.10954
作者: Anqi Peter Li,Kaden Kim
类目: Machine Learning (cs.LG)
*备注: 13 pages, 2 figures, under review at the CWM workshop
Abstract:Continual world models must decide whether new data justify changing the model. Fixed replay schedules and prediction-error triggers specify when to update, but neither reveals the value of an individual update: one deployment run cannot show how the same model would have performed at that moment had it held its parameters. We introduce the fork ledger, which branches a deployment stream at pre-registered decision points into matched update and hold continuations under common random numbers. It evaluates both continuations on the same episodes and records \Delta R = R_\mathrmupdate - R_\mathrmhold . Always applying one fixed update mechanism lowers return on all three simulated control tasks: CartPole ( -144.0 ; checkpoint-bootstrap 95% CI [-185.4,-116.1] , against a converged return near 650 ), Walker ( -82.8 ; [-101.1,-61.7] ) and Cheetah ( -18.6 ; [-29.0,-6.6] ). Divergence is an outcome of applying the update, so the estimand counts every attempted fork; restricted to the 693 of 720 that did not collapse, CartPole and Walker are unchanged in sign ( -113.4 and -82.1 ) and Cheetah becomes unresolved ( -3.9 ; [-17.5,+13.0] ). The task is the unit of inference: each contributes 240 attempted forks over five pretrained checkpoints crossed with two drift directions. The ledger makes counterfactual utility observable for a fixed mechanism, allowing triggers to be judged by the updates they select rather than by surprise detection alone.
[LG-38] Empirical Evaluation of Data Poisoning Attacks in Supervised Learning
链接: https://arxiv.org/abs/2609.10952
作者: Toshif Khan(Minot State University),Muhammad Abusaqer(Minot State University)
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注: Presented at the 58th Midwest Instruction and Computing Symposium (MICS 2026), Eau Claire, WI, March 27 to 28, 2026. 15 pages, 4 figures, 3 tables
Abstract:Data poisoning corrupts training data to degrade a model or to plant attacker-controlled behavior. This study evaluates two representative training-time attacks, label flipping and backdoor poisoning, on MNIST and Fashion-MNIST with three baseline classifiers: Logistic Regression, Linear SVM, and Random Forest. Clean training is compared with poisoning rates of 5%, 10%, and 20% using clean-test accuracy, macro-precision, macro-recall, macro-F1, and, for backdoors, attack success rate. Label flipping caused clear degradation, largest for Logistic Regression and Linear SVM, while Random Forest stayed comparatively stable. Backdoor poisoning reached attack success rates from 0.9667 to 1.0000 on both datasets and all three models while often keeping clean-test performance near baseline. The results separate indiscriminate poisoning, which shows up in standard metrics, from targeted backdoor poisoning, which stays comparatively stealthy while embedding highly effective malicious behavior, and they support security-oriented evaluation beyond conventional clean-test metrics.
[LG-39] sting Between the Test Cases: Proving End-to-End Steering in Conditions You Never Drove
链接: https://arxiv.org/abs/2609.10951
作者: Menuka Ghalan,Charles Rodgers,Zachary D. Asher
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: 10 pages, 7 figures, 2 tables
Abstract:AI-based automated vehicle testing is challenging because a model that passes every test condition can still fail in the real world. Formal verification offers a way to directly address this gap. On a simulated highway and an arterial road we trained two small end-to-end steering networks each in CARLA, one on clear conditions alone and one on clear, fog, night and low sun. All four models were driven against a 2.19 ft lane-departure budget. Without driving again, we used bound propagation, a formal method that reads the trained weights, to compute how far steering can drift at every disturbance strength between two captured images. One calculation covers more than a campaign could drive: on the arterial it spans 133 poses, where ten intensities each would be 10^133 combinations, in minutes on one GPU. Not only did formal verification find conditions that broke the clear-trained policy without simulation testing, it provided some preliminary evidence for potential failures between the test cases. Our overall conclusion is that formal verification is a viable complement to simulation, and could be adopted as a part of verification and validation for automated driving.
[LG-40] ObstaDiff: Generalizable Diffusion Policy Learning via Obstacle-aware Representations
链接: https://arxiv.org/abs/2609.10918
作者: Jiawen Wang,Kevin Yao,Khalid Jawed
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: Accepted to the 10th Conference on Robot Learning (CoRL 2026), Austin, TX, USA. 16 pages, 5 figures
Abstract:Imitation learning has achieved impressive results in robotic manipulation, yet most existing approaches assume clean backgrounds and lack explicit mechanisms for obstacle-aware motion generation. Extending such policies to cluttered, real-world scenes with unstructured obstacles remains a key generalization challenge. We present ObstaDiff, a decomposed diffusion-policy framework with a lightweight obstacle-aware visual encoder. ObstaDiff extracts a structured target-obstacle-background representation, enabling the downstream alignment policy to generate end-effector trajectories toward a target-centered bottleneck pose while reasoning about surrounding obstacles. We evaluate ObstaDiff on 61 real-robot greenhouse trials per method (366 executions in total). ObstaDiff achieves 75.41% average task success and 8.20% average obstacle collision rate, outperforming representative imitation-learning baselines and improving generalization in cluttered agricultural scenes.
[LG-41] Relatively Smart II: Tractable or Semi-Supervised Instance-Optimal Learning
链接: https://arxiv.org/abs/2609.10886
作者: Shaddin Dughmi,Alireza F. Pour
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:We continue the study of relatively smart learning, introduced by Dughmi and Pour (2026), which asks a supervised learner to compete, marginal by marginal, with every distribution-fixed error guarantee soundly certifiable from unlabeled data. They showed that the One-Inclusion Graph (OIG) learner is relatively smart with a quadratic sample-complexity blowup, and that no relatively smart learner can do better, leaving open whether ERM or another natural or tractable learner achieves comparable guarantees. They also left open whether the blowup can be restricted to unlabeled data. Our firs results shows that ERM—and in fact any proper consistent learner—is relatively smart for binary classification in the distribution-free setting. We show that a small certifiable error with m samples implies a similarly small error on the uniform distribution over a random sample of size O(m^2) , yielding a cover of size at most 2^m+1 on that sample. This suffices to control the error of proper consistent learners with O(m^2) samples. We then show that semi-supervised relatively smart learning is information-theoretically possible with a quadratic blowup only in unlabeled sample complexity and no blowup in labeled sample complexity. The learner uses a natural generalization of OIG to a leave-most-out transductive problem, where labels of part of a finite pool are revealed and the remaining labels are predicted. Finally, this label efficiency comes at a cost in simplicity and tractability. If the hypothesis class is accessed only through an agnostic ERM oracle, any semi-supervised relatively smart learner with substantially sub-quadratic labeled-sample blowup requires super-polynomially many oracle calls. This holds even when the marginal is given explicitly, and thus also yields an intractability result for distribution-fixed learning that may be of independent interest. Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML) Cite as: arXiv:2609.10886 [cs.LG] (or arXiv:2609.10886v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2609.10886 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Alireza F. Pour [view email] [v1] Wed, 9 Sep 2026 22:45:18 UTC (113 KB)
[LG-42] Learning Orthogonal Multi-Index Models Beyond Small Initialization: Incremental Learning Competitive Dynamics and Symmetry
链接: https://arxiv.org/abs/2609.10879
作者: Mo Zhou,Weihang Xu,Simon S. Du,Maryam Fazel
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 102 pages
Abstract:Recent work has identified incremental learning in shallow networks trained on single-index and multi-index models. However, existing analyses often rely on simplifying settings, such as small initialization, correlation loss, or layer-wise training. These choices reduce neuron interactions and leave some feature learning dynamics under standard initialization unexplored. We study training dynamics for polynomial-width two-layer networks learning orthogonal multi-index targets under standard initialization using polynomially many samples. We first prove that incremental learning still occurs: the loss decreases sequentially according to the Hermite expansion of the target, with lower-order components learned before higher-order components recover the individual target directions. In this standard initialization regime, training also shows a competitive reallocation of parameter mass: after the total mass fits the target mean and stabilizes, mass shifts into the target subspace and then concentrates on aligned neurons. Our theoretical analysis uses slightly modified gradient flow, while vanilla gradient descent empirically exhibits the same qualitative dynamics. Technically, we introduce a symmetry-based finite-width approximation via symmetrized networks, rather than comparing directly with an infinite-width limit. This yields better control of approximation errors and may be of independent interest.
[LG-43] Certifying Lower Bounds for Risk-Sensitive Reinforcement Learning under Adversarial State Perturbations
链接: https://arxiv.org/abs/2609.10866
作者: Tong Li,Saunak Kumar Panda,Yisha Xiang
类目: Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注: 26 pages, 3 figures
Abstract:Reinforcement learning (RL) agents deployed in real-world environments are often vulnerable to adversarial perturbations in state observations, creating risks in safety-critical applications. Certification methods can improve robustness against adversarial perturbations by providing lower bounds on expected cumulative rewards. Existing certification methods, however, mainly focus on risk-neutral objectives. In this paper, we extend certification methods to risk-sensitive objectives by establishing lower bounds on the exponential utility of cumulative rewards under l_p -norm-bounded state adversarial perturbations ( 1\leq p \infty ). By introducing a \phi -divergence relaxation of the perturbation set, we formulate the risk-sensitive certification problem as a convex optimization and derive its dual to obtain a tractable approximation of the certified lower bound. We further propose an empirical method that improves certified lower bounds by selecting the training risk-aversion parameter \beta independently of the risk level used during evaluation. Experiments on both OpenAI Gym environments and a machine replacement problem show that, compared to risk-neutral training, risk-averse training generally yields policies with higher certified lower bounds, particularly under larger perturbation budgets. Moreover, under both risk-neutral and risk-averse evaluation settings, increasing risk aversion during training leads to non-monotonic certification performance, where certified lower bounds initially improve but eventually decrease due to overly conservative policies.
[LG-44] Flow Duality and Source Geometry for Categorical Generation
链接: https://arxiv.org/abs/2609.10863
作者: Etrit Haxholli
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:Continuous and discrete flow matching are usually treated as separate constructions. This paper identifies a duality between them: projecting continuous convex-interpolant paths with one-hot targets through a position-wise argmax yields discrete convex-interpolant paths. The result requires source laws with appropriate coordinate symmetry and boundary regularity, and it makes the continuous source distribution an explicit design choice for categorical generation. We derive the induced discrete interpolation behavior for Gaussian, bounded-uniform, and centered negative-exponential sources, showing that different source geometries lead to qualitatively different transition timing and vocabulary-size dependence. Small visual diagnostics and a short language-modeling pilot suggest that these source-design effects can also appear in learned transports and early generative quality.
[LG-45] Processing and classifying bird songs using wavelet techniques and supervised learning
链接: https://arxiv.org/abs/2609.10826
作者: Laura Lucia Dominguez Barrios,Fidel Aniano Causil Barrios,Alex Rodrigo dos Santos Sousa,Mariana Rodrigues Motta
类目: Machine Learning (cs.LG); Methodology (stat.ME)
*备注:
Abstract:This study proposes an integrated framework for the processing and classification of invasive bird species vocalizations within natural soundscapes, characterized by high levels of environmental noise. We address the challenge of signal degradation by employing a Bayesian wavelet shrinkage methodology based on the Epanechnikov kernel prior, which offers a closed form decision rule and high computational efficiency for processing large bioacoustic datasets. The methodology was applied to recordings of three species obtained from the iNaturalist platform: \textitEuphonia violacea, \textitLeiothrix lutea, and \textitPasser domesticus. After signal denoising, we extracted a comprehensive set of features, including Mel-Frequency Cepstral Coefficients (MFCCs) and spectral indices such as entropy and zero-crossing rate. Several supervised learning models: Random Forest, Multinomial Logistic Regression and Support Vector Machine (SVM) were evaluated across different feature dimensionalities. Our results demonstrate that the proposed wavelet based preprocessing significantly enhances classification performance, with the SVM model achieving the highest accuracy (up to 0.9398) under a 10-dimensional MFCC configuration. This research provides a robust statistical tool for automated ecological monitoring and the management of biological invasions.
[LG-46] DR-LabStack: Design and Implementation of a Clinician-Facing Web System for Diabetic Retinopathy Prediction
链接: https://arxiv.org/abs/2609.10796
作者: Yingfan Xu,Tieming Liu,Ye Liang
类目: Machine Learning (cs.LG); Software Engineering (cs.SE)
*备注:
Abstract:Pretrained diabetic retinopathy (DR) prediction models differ in their input fields, serialization formats, preprocessing requirements, and output semantics. Making these models accessible through a common clinical interface therefore requires explicit coordination between the user interface and the inference service. We designed and implemented DR-LabStack, a React-Flask web system integrating four externally developed pretrained models: RuleFit, Pruned RuleFit, Elaborative XGBoost, and Two-level Ensemble. A shared form retrieves ordered model features, renders model-specific numerical and categorical controls, and constructs a positional input vector. Backend adapters load heterogeneous artifacts and apply the ensemble’s accompanying scaler, while a common JSON response supports binary classification display alongside method and source information. Functional evaluation on September 8, 2026 used copied application files and real model artifacts in a documented isolated environment. All four models loaded and exposed their 14-, 6-, 8-, and 25-field contracts. Sixty-two Flask test-client requests characterized service behavior; 12 limited-vector checks confirmed invocation-path and threshold consistency. Twenty-four browser-component scenarios with mocked transport verified input ordering and result rendering and characterized input-validation behavior. The resulting system demonstrates a reusable interaction and serving workflow for heterogeneous DR models. The contribution is web-system design, integration, and software functionality; clinical effectiveness and clinician usability require separate evaluation.
[LG-47] Composable CXL Memory as a Kubernetes-Native Shared Memory for LLM Serving
链接: https://arxiv.org/abs/2609.10790
作者: Hongjian Fan,Kevin Zhang,David Habinsky,Sean Dykstra
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注: 23 pages, 9 figures, 9 tables
Abstract:We present a Kubernetes Dynamic Resource Allocation (DRA) driver that makes composable CXL memory a schedulable cluster resource, and evaluate the resulting shared-memory tier for cross-node KV-cache reuse in LLM serving. The driver composes CXL regions on demand, materializes them as DAX devices on each participating host, and injects them into pods under a single Container Device Interface (CDI) name so that pods on different nodes access the same physical region. A shared-memory connector for vLLM/llm-d uses that region as a KV-cache tier with a slot directory embedded inside the shared medium, which eliminates the need for an external metadata service. On a two-node cluster with a 512,GiB CXL appliance and Qwen2.5-7B-Instruct, cross-node prefix reuse reduces TTFT by 5.5 \times --36.6 \times at an external hit rate of 95.4–99.5,%, while node-local tiers (GPU prefix caching, CPU-DRAM offload) fall back to full recompute. The sharing gap, defined as the latency ratio between cross-node and same-node reuse, is 1–4%, indicating that cross-node reuse incurs little additional latency relative to same-node reuse on our testbed. Both replicas run full engines; the study demonstrates memory disaggregation rather than prefill/decode disaggregation. We report this as a feasibility study rather than a performance evaluation.
[LG-48] From Connectivity to Rewards: Dense Reward Learning with Directed State Graphs
链接: https://arxiv.org/abs/2609.10781
作者: Shuyuan Zhang,Zihan Wang,Xiao-Wen Chang,Doina Precup
类目: Machine Learning (cs.LG)
*备注:
Abstract:The integration of graphs with Goal-Conditioned Hierarchical Reinforcement Learning (GCHRL) has received increasing attention, as graphs naturally encode task hierarchies for effective subgoal sampling. However, existing methods often overlook intrinsic connectivity information, failing to fully leverage the underlying topology for efficient learning. Most graph-based GCHRL methods use the graph as a stochastic sampling tool rather than as an environmental model that encodes connectivity and state-accessibility information. This limitation is particularly acute in quasimetric environments, where the inherent asymmetry of state transitions poses a fundamental challenge to stable policy learning and robust path planning. In this paper, we address these problems by introducing a state connectivity model designed to predict pairwise state connectivity strength in asymmetric environments. We transform these connectivity strengths into scalar auxiliary dense rewards, providing continuous guidance across multiple hierarchical levels. We demonstrate that our proposed framework, Graph-Guided Quasimetric Dense Reward (G2QDR), can theoretically be integrated into any existing GCHRL architecture, and the state connectivity model is efficiently implemented via a neural network trained on a directed state graph generated during exploration. Empirical results across a wide range of sparse reward environments indicate that, in general, G2QDR can enhance the performance of baseline GCHRL approaches with acceptable computational overhead.
[LG-49] A Bellm an Optimality Equation for Plasticity
链接: https://arxiv.org/abs/2609.10776
作者: Jeremy Lucas,Doina Precup
类目: Machine Learning (cs.LG)
*备注:
Abstract:In continual reinforcement learning, carefully managing the stability-plasticity tradeoff remains a core challenge. Recent work by Abel et al. (2025) formalized this dilemma by defining plasticity as the generalized directed information from an agent’s observations to its actions, and empowerment as the generalized directed information from its actions to its observations. This formulation successfully reframes the traditional stability-plasticity tradeoff as an empowerment-plasticity tradeoff. However, while extensive literature exists on optimizing for empowerment, there is currently no research addressing the optimization of plasticity under this new definition. This paper presents preliminary work toward optimizing plasticity within Markov decision processes. We show that there exists a Bellman optimality equation for optimizing plasticity similar to previous work for empowerment.
[LG-50] SynCo: Synthetic Community-Aware Attributed Graph Generator for Graph Neural Network Benchmarking
链接: https://arxiv.org/abs/2609.10742
作者: Guilherme Henrique Messias,Mariana Caravanti de Souza,Sylvia Iasulaitis,Alan Demétrius Baria Valejo
类目: ocial and Information Networks (cs.SI); Machine Learning (cs.LG)
*备注: 15 pages
Abstract:Graph Neural Networks (GNNs) are powerful models for handling attributed graphs in tasks such as classification, link prediction, and community detection, as they enable the aggregation of information from both structural and semantic sources. However, progress in community detection is hindered by the lack of high-quality datasets, since ground-truth community labels are often unavailable and most algorithms proposed in recent literature rely on the same benchmark datasets for model training and evaluation. To address this issue, attributed random graph generators are commonly employed to create synthetic graphs for assessing the strengths and limitations of GNN-based models. Nevertheless, most existing generators rely heavily on power-law degree distributions, despite recent evidence indicating that scale-free networks are rare, particularly in social network contexts. Moreover, state-of-the-art attributed graph generators provide limited flexibility, as they do not allow users to construct communities with varying densities, degree distributions, and sub-community structures. To overcome these limitations, we introduce the Synthetic Community-Aware Attributed Graph Generator (SynCo), a graph generation algorithm that allows users to control the node degree distribution and sub-community structure. We evaluate SynCo across three different tasks: graph mimicking, hyperparameter evaluation, and node clustering tuning. The results show that our model outperforms state-of-the-art approaches in synthetic graph generation and data augmentation, while preserving the original distributions of duplicated and augmented datasets, as confirmed by statistical tests well know in literature. We also demonstrate the ability of SynCo to generate nodes in large scale, up to 2.1 million nodes.
[LG-51] Conformal Calibration Transfer ICML2026
链接: https://arxiv.org/abs/2609.10737
作者: Achref Doula
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: Accepted at the 43rd International Conference on Machine Learning (ICML 2026)
Abstract:Conformal prediction converts point predictions into set-valued predictions with coverage guarantees under exchangeability between calibration and deployment data. We study conformal calibration transfer, where this requirement fails because labeled calibration is available only in a source space, while prediction sets are needed in a target space linked to the source through unlabeled paired observations (e.g., paired modalities or sensor changes). We propose Transported Conformal Calibration (TCC): we transport labeled source calibration into the target space using the paired data, and then correct residual post-transport mismatch using only unlabeled target inputs. We instantiate this correction with two complementary methods: TCC-KS, which uses a label-free uncertainty surrogate to detect mismatch and adjust calibration conservatively, and weighted-TCC, which reweights transported calibration toward the target domain for improved efficiency when weights are stable. We provide finite-sample target-domain coverage guarantees that adapt to an observable measure of mismatch. Across CIFAR-100-C, Tiny-ImageNet-C, and SEN12MS, we show reliable target-domain coverage transfer without labeled target calibration data, with label-free diagnostics that predict when correction is needed.
[LG-52] GEOSTEER: Geodesic Optimization for Activation Steering in Large Language Models
链接: https://arxiv.org/abs/2609.10658
作者: Xuan Cuong Ngo,Hao Vo,Ngan Le
类目: Machine Learning (cs.LG)
*备注:
Abstract:Activation steering provides a lightweight way to control large language models (LLMs) by modifying their hidden activations at inference time. Among these approaches, norm-preserving steering aims to change model behavior without altering the activation norm, reducing the risk of representation collapse and degradation. However, existing norm-preserving methods are limited by predefined steering trajectories and by their reliance on one-step updates, which may fail to capture the complex structure of activation distributions. We propose GeoSteer, an optimization-based method for norm-preserving activation steering. GeoSteer formulates steering as a Riemannian optimization problem and updates activations through a sequence of small geodesic steps on the representation manifold. To avoid fixed steering directions, GeoSteer learns a nonlinear activation-space objective that distinguishes desired from undesired activations, and uses this function to adaptively guide each steering step. This multistep formulation yields smoother, more stable, and more consistent steering behavior while preserving the activation norm. Across TruthfulQA, RealToxicityPrompts, and UltraFeedback benchmarks, GeoSteer consistently improves over state-of-the-art activation steering baselines. These results suggest that norm-preserving steering can be made more effective by replacing predefined one-step edits with adaptive, geometry-aware optimization.
[LG-53] Byzantine-Robust Federated Fire Detection with a Rotating Coordinator
链接: https://arxiv.org/abs/2609.10647
作者: Georgia Argyrou,Aymen Bahrouny,Hedi Fendriy,Alexander Jung
类目: Machine Learning (cs.LG)
*备注:
Abstract:We study the application of federated learning (FL) to indoor fire detection. Such fire-detection systems use edge cameras that record sensitive footage which cannot easily be collected at a central server. Existing federated solutions leave three practical obstacles unaddressed: limited uplink bandwidth, Byzantine (malicious or faulty) clients, and unconditional trust in a single, permanently fixed aggregation server. Our main contributions address all three. In particular, we provide (i) a curated indoor fire-detection dataset assembled from eight public sources; (ii) an edge-deployable detector whose model updates are compressed up to 10 time with only a small loss in balanced accuracy; and (iii) a semi-decentralized Byzantine-robust FL method that combines history-aware aggregation with a rotating coordinator, evicting stealthy attacks that per-round filters miss while removing the fixed-server single point of failure. On the held-out test set the rotating-coordinator method matches its fixed-server counterpart in accuracy and detection speed, and a physically distributed six-node cloud deployment confirms feasibility.
[LG-54] Zero-shot rib design: merging training-free generative prior with topology optimization
链接: https://arxiv.org/abs/2609.10643
作者: Yongmin Kwon,Namwoo Kang
类目: Machine Learning (cs.LG); Applied Physics (physics.app-ph)
*备注:
Abstract:Natural load-bearing patterns such as leaf venation, trabecular bone, and spider webs achieve high stiffness per unit mass, yet classical topology optimizers rarely reach such geometries, and few let engineers express structural design intent through natural language. This work treats a frozen text-to-image diffusion model as a training-free source of design knowledge and distills it into the physics loop of density-based topology optimization via score distillation sampling, so that a text prompt becomes an explicit, machine-interpretable representation of engineer intent. The prompt-induced generative gradient and the finite element sensitivity are combined at every iteration, letting physics decide which prompt-induced features survive. In 245 primary SDS runs spanning four geometric domains and two physics regimes, 38 of 49 prompt–domain combinations achieved statistically significant compliance reductions (up to -31.5% mechanical and -23.0% thermoelastic), outperforming gradient-based baselines. Cross-domain morphological analysis identifies a recurring structural signature of improvement: in most domains the generative prior suppresses dead-end branches in the rib skeleton, with endpoint–compliance correlation r = +0.56 to +0.99 . A Heaviside projection with \beta -continuation resolves a pronounced intermediate-density tendency in this diffusion–physics coupling ( 42.6% to 3% ), and an automated skeleton-based pipeline converts optimized density fields into \revcandidate geometry ready for computer-aided design. By retargeting the generative prior across domains, loading conditions, and physics objectives through a change of text prompt, with each new problem’s physics setup specified separately, the framework uses a pretrained generative model as a reusable, training-free prior for engineering design.
[LG-55] Numbat: Building and Verifying a Self-Contained Machine-Learning Stack
链接: https://arxiv.org/abs/2609.10632
作者: Thang Tran(CloudKites AI Lab, New South Wales, Australia),Lan Dang(Monash Business School, Monash University, Victoria, Australia)
类目: oftware Engineering (cs.SE); Machine Learning (cs.LG)
*备注: 19 pages, 4 figures, 4 tables. Companion to arXiv:2608.24267
Abstract:Machine-learning systems are built almost exclusively on a few large Python-orchestrated frameworks, and they inherit those stacks’ engineering costs: environments of hundreds of version-coupled packages, separate export toolchains for deployment, and the split between the language research is written in and the language products ship in. We report on the construction and verification of numbat, a machine-learning stack written in one general-purpose language (Zig) with no third-party runtime dependencies. The stack spans tensor computation, automatic differentiation, neural-network modules, mixed precision, multi-GPU training, data loading and monitoring; an SDK exposes it behind a stable, additively versioned C ABI of over 1,400 entry points, with bindings for six languages; and its clinical domain planes encode regulatory requirements as executable acceptance gates rather than documentation. Verifying such a stack is the harder half of building it: a defective training run rarely fails, it converges quietly to a slightly worse model. We treat a widely used reference implementation as an executable specification and verify against it at five levels, from operator gradient checks to an automated trajectory gate against a same-machine reference run - the arrangement our companion study formalizes as a trajectory-level differential oracle. The protocol surfaced ten silent recipe divergences, which we catalog with mechanisms and symptoms. As the acceptance test, we train a 25.9M-parameter detector of the YOLOv8m class from random initialization on COCO 2017 for the full 500-epoch schedule: the exported weights score 0.4956 mAP50-95 under the official protocol, scored by the reference stack’s own validator (published endpoint 0.502), with single-GPU step time at parity on identical hardware. Weights, per-epoch metrics and the full run manifest are released.
[LG-56] From Cycle Space to Cycle Manifold: Limits and Achievability of Blind False Data Injection Attacks
链接: https://arxiv.org/abs/2609.10631
作者: Xin Li,Chenhan Xiao,Jonathan Cohen,Aviad Elyashar,Yang Weng,Rami Puzis
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注: 11 pages
Abstract:A false data injection attack (FDIA) can change the estimated grid state while evading a residual-based bad data detector (BDD). Existing blind attacks learn a low-rank measurement subspace, but this algebraic view does not state the physical grid constraints that make an attack stealthy or the minimum information needed to recover the complete attack space. Under the connected direct-current (DC) branch-flow model, we show that the residual-sensitive subspace of the noiseless orthogonal test is exactly the weighted cycle space. Its orthogonal complement is therefore the complete stealthy attack space, making weighted cycle-space knowledge both necessary and sufficient for complete blind FDIA. This space identifies the topology only up to 2-isomorphism and the relative cycle-edge parameters only up to one scale per biconnected component; bridge parameters are neither identified nor required. We then formulate a computationally unconstrained benchmark and a tractable measurement-only reconstruction method. Experiments on IEEE systems compare BDD bypass rate at a 95% nominal-acceptance threshold against state impact. As a compact alternating-current (AC) extension, we characterize feasible branch P/Q measurements by a cycle manifold and demonstrate topology-assisted manifold fitting and measurement generation on a graphics processing unit (GPU). In the lossless fixed-voltage small-angle limit, the normal space of the active-power slice reduces to the DC weighted cycle space.
[LG-57] SoK: Privacy Attacks on Machine Learning via Explainable AI
链接: https://arxiv.org/abs/2609.10627
作者: Abdullah Caglar Oksuz,Anisa Halimi,Erman Ayday
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注:
Abstract:Machine learning explanations reveal model behavior beyond predictions, creating attack surfaces for model confidentiality and data privacy. We systematize 25 studies that exploit explanations for model extraction, membership inference, and model inversion, treating attribute inference as partial inversion. Existing work is often labeled only black- or white-box, obscuring substantial differences in what explanation signal reaches an adversary. We therefore separate model knowledge from explanation acquisition and identify five paths: target-released, attacker-derived, secondary disclosure, privileged access, and released global artifacts. Across these paths, explanations reduce extraction cost, expose membership signals through explanation statistics, recourse distance, and explanation-guided robustness, and support spatial or algebraic reconstruction of private inputs. We compare system and threat models, explanation signals, auxiliary knowledge, target models, modalities, query budgets, evaluation metrics, reported performance, and defenses. Our analysis shows that no explanation family is uniformly unsafe and no defense is uniformly effective. Risk depends on which signal is exposed, how it is acquired, which asset is targeted, and what the attacker already knows. We argue that explanation privacy should therefore be evaluated as an end-to-end disclosure problem, with defenses matched to the acquisition path and protected asset.
[LG-58] Understanding In-Context Multimodal Jailbreaks via Posterior Reweighting
链接: https://arxiv.org/abs/2609.10613
作者: Xu Zhang,Dev Mistry,Xiang Xu,Ren Wang
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注:
Abstract:In-context learning (ICL) jailbreaks reveal a critical vulnerability in multimodal large language models (MLLMs): harmful demonstrations in the prompt can induce unsafe outputs without modifying model parameters. Despite extensive empirical evidence, existing work lacks a principled understanding of why such jailbreaks reliably succeed or how their effectiveness scales with context composition. We propose a posterior reweighting framework that models a safety-aligned MLLM as implicitly operating over competing behavioral modes, and interprets in-context demonstrations as inference-time evidence that dynamically shifts the model’s posterior preference between safe and harmful behaviors. This view formalizes jailbreak as a process of evidence accumulation, yielding predictive scaling laws with respect to demonstration count, harmful ratio, adversarial strength, and semantic diversity. Guided by this framework, we introduce a posterior-aware inference-time defense that adaptively injects benign counter-evidence based on estimated risk, effectively suppressing harmful posterior drift while preserving model utility. Compared to existing in-context defenses, our method achieves a significantly improved robustness-utility trade-off under a fixed intervention budget. Together, our results establish posterior reweighting as a unifying and predictive framework for understanding and mitigating ICL jailbreak in MLLMs.
[LG-59] PEARL: A Task-Aware Framework for Evaluating Differentially Private Synthetic Educational Data
链接: https://arxiv.org/abs/2609.10612
作者: Xianghui Meng,Yujing Zhang,Jionghao Lin
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注:
Abstract:Personalized learning systems rely on real learner data, including performance, behavior, and demographic information, but these data are highly privacy-sensitive. Differentially private (DP) synthetic data can support system development and educational research while reducing exposure of individual learners. Existing evaluations, however, assess privacy and predictive usefulness separately, without determining whether synthetic learner data remain usable for the intended personalized learning task. We introduce PEARL (Privacy-Equivalence Audit and Release Ledger), which approves a DP synthetic educational dataset only when it passes all required checks of validity, privacy protection, predictive usefulness, and suitability for the intended educational task, while recording why each rejected dataset fails. Across 96 study settings, each defined by a dataset, data-generation method, privacy budget, and random seed, only 12 produced synthetic datasets that passed all applicable PEARL checks. Many privacy-protected datasets were rejected for omitting important outcome groups, such as withdrawn students, or for failing to preserve the order of learning activities. Fairness analysis further showed that some datasets passing privacy and predictive-usefulness checks still yielded unequal at-risk prediction performance across groups defined by disability and socioeconomic background. Moreover, Deep Knowledge Tracing and Self-Attentive Knowledge Tracing learned no meaningful next-response patterns from any tested synthetic knowledge-tracing dataset, showing that privacy protection alone does not guarantee usefulness for dropout prediction, knowledge tracing, or adaptive tutoring.
[LG-60] Black-Box Membership Inference via Word-Level Probability Estimation
链接: https://arxiv.org/abs/2609.10611
作者: Shengjie Niu,Yeheng Ge,Jian Huang
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 17 pages, 5 figures, and 17 tables
Abstract:Membership inference attacks (MIAs) have emerged as critical tools for auditing privacy risks in large language models (LLMs), aiming to determine whether a given text was included in a model’s training corpus. However, most existing MIAs require access to per-token logits or probabilities, making them inapplicable in practice to proprietary LLMs that expose only textual continuations. To address this underexplored setting, we propose Word-level Probability MIA (WPMIA), a statistically principled MIA for strict black-box privacy auditing. WPMIA estimates word-level generation probabilities via Monte Carlo sampling with local kernel smoothing, then aggregates these estimates into a sequence-level likelihood estimator. Furthermore, WPMIA constructs the likelihood conditioned on different prefixes, thereby amplifying the distributional differences between members and non-members. We evaluate WPMIA across various open-source LLMs and find that it consistently outperforms existing black-box baselines. Importantly, we also evaluate WPMIA on modern proprietary LLMs, including GPT-5-Chat, Gemini-2.5-Flash, and Claude-4.5-Haiku, achieving an average TPR@5%FPR of 42.0 across these models. These results offer a sound foundation for future research on strict black-box membership inference. Code is available at \hrefthis https URLthis https URL.
[LG-61] On the Relation between Code Quality and Machine Learning Performance: A Large-scale Empirical Study
链接: https://arxiv.org/abs/2609.10610
作者: Marius Mignard(CRIStAL),Steven Costiou(CRIStAL),Anne Etien(CRIStAL, EVREF)
类目: oftware Engineering (cs.SE); Machine Learning (cs.LG)
*备注:
Abstract:Context: Computational notebooks are the standard environment for machine learning (ML) development. Within the ML community, model performance is often the primary considered metric, and code quality is treated as a secondary concern. This prioritization relies on a largely untested assumption that code quality and ML performance are unrelated. Practitioners also reuse existing code that may come from notebooks selected through social signals (popularity, author expertise) whose reliability as quality proxies has never been assessed. Objective: We empirically investigated the relationship between code quality and ML performance in notebooks, and evaluated whether popularity and author expertise give indication on code quality or performance. Method: We conducted a large-scale empirical study of 265,363 Python notebooks submitted to Kaggle competitions. We assessed code quality with two static analysis tools: Pylint, capturing general Python code quality, and SonarQube, configured with a profile of 34 rules targeting data-science and ML-specific practices. Results: The relationship between code quality and performance depends on the notion of quality considered. General Python code quality is decoupled from ML performance, showing negligible or non-significant correlations across all observations. In contrast, ML-specific violations exhibit a consistent, small negative association with performance that persists across all observations. The popularity of a notebook does not give information on the code quality or performance. Code expertise provides no information on quality or performance, but competition expertise correlates with better performance, fewer ML-specific violations, and slightly more Python errors and refactoring violations.
[LG-62] Adaptive Diffusion Freezing: Privacy-preserving Diffusion Models Against Membership Inference Attacks
链接: https://arxiv.org/abs/2609.10608
作者: Jialu Guo,Xiao Han,Junjie Wu
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注: 16 pages, 8 figures. Accepted by the Proceedings of the 2026 ACM SIGSAC Conference on Computer and Communications Security
Abstract:Diffusion models have achieved remarkable success in generative tasks across various areas, however their training process raises significant privacy concerns, particularly under membership inference attacks (MIAs). Prior studies on privacy-preserving of diffusion models fail to balance privacy, utility, and efficiency. To address this gap, we propose a novel framework of privacy-preserving diffusion models, Adaptive Diffusion Freezing (ADF), which can defend against MIAs with better trade-off. By leveraging cross-timestep adaptive freezing training, ADF explicitly control the participation of different data subsets across diffusion timesteps via a mask matrix, which reduces the over-memorization and leads to more uniform model behaviors between member and nonmember samples. To construct a freezing mask matrix that effectively reduce membership leakage without unnecessarily harming generation quality, we introduce a pretraining-based risk-aware freezing policy to estimate MIA risk based on memorization tendency, and suppress the contribution of the subset-timestep pairs with higher risk. Evaluations on multiple datasets demonstrate that ADF provides effective defense performance as well as state-of-the-art privacy-utility-efficiency trade-off performance compared to various baselines.
[LG-63] An Empirical Measurement of Jailbreaking Evaluators
链接: https://arxiv.org/abs/2609.10594
作者: Yujie Mu
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注: 12 pages
Abstract:Expert evaluation of jailbreak responses is costly and difficult to scale, so the community increasingly relies on automated evaluators to determine whether an attack succeeds. However, jailbreak studies typically validate their chosen evaluator independently, repeatedly spending resources on similar evaluation efforts while making results across papers difficult to compare. Different evaluators also encode different definitions of jailbreak success, meaning that reported attack strength and apparent progress can depend substantially on which evaluator is used. We systematically compare six evaluators that recur in recent jailbreak attack and defense research: HarmBench, JailbreakBench, JailbreakRadar, StrongReject, JADES, and JailMeter. To our knowledge, no prior study has evaluated all six on the same human-labeled data under a controlled setup. We evaluate them on JailbreakQR and JailMeter-Eva, using human judgments as the reference, and measure agreement with humans, error types, and consistency across attack families. For evaluators that require a general-purpose LLM judge, we use a shared backbone to control for model-specific variation. We found that JADES exhibits the best overall performance, while HarmBench and StrongReject also demonstrate good performance.
[LG-64] Halo: Improving forecast accuracy through heteroscedastic estimation
链接: https://arxiv.org/abs/2609.10589
作者: Adam Cataldo
类目: Machine Learning (cs.LG)
*备注:
Abstract:Heteroscedastic forecasting, where a network estimates a scale parameter alongside a location parameter, is normally motivated by uncertainty quantification. This paper shows it also improves the point estimate, in contrast to reported negative results for heteroscedastic estimation outside time series. Halo is a modification that reuses an existing deep forecaster’s architecture, giving it a second output for the scale of its implied distribution and training it under the matching negative log likelihood. Adapting three state-of-the-art models — a transformer, a graph network paired with a variational autoencoder, and a single-layer convolutional network — under both Gaussian and Laplacian losses demonstrates the phenomenon. On the five electricity price markets of a standard forecasting benchmark, Halo improves MSE and MAE in 28 of 30 model-market-metric comparisons, cutting average MSE by 2.6% to 16.5% and average MAE by 1.7% to 11.0%. Two findings emerge: (1) whether the scale estimate comes from a second projection head or from a full parallel network matters far less than whether the network estimates scale, and (2) the improvement holds under the hyperparameters already tuned for the point-estimate baseline, so retuning is optional.
[LG-65] Optimizing AI Inference Across the Deployment Stack
链接: https://arxiv.org/abs/2609.10550
作者: Tejinder Singh,John Pflueger,Jeebak Mitra,Robert Lincourt,Mitchell Markow,Bhavesh A. Patel
类目: oftware Engineering (cs.SE); Machine Learning (cs.LG)
*备注: 29 pages, 12 figures
Abstract:AI deployment performance is shaped not by model architecture alone, but by interactions among compression, compiler transformations, and serving policies. Published benchmarks often report latency and throughput under incomparable conditions, limiting their use for deployment decisions. This paper presents a unified analytical treatment of inference optimization across the deployment stack. We introduce a three-layer taxonomy covering model-level techniques such as quantization, pruning, and distillation; compiler transformations such as graph fusion, layout optimization, and kernel autotuning; and system policies such as dynamic batching, admission control, and memory tiering. We formulate deployment as a constrained multi-objective optimization problem over accuracy, latency, throughput, memory footprint, and energy, and analyze a deployment-ranking functional with Pareto monotonicity and scale invariance. Roofline models show how memory-bandwidth hierarchies bound performance across precision regimes, while queuing models explain how service-time changes amplify response time under load. To improve comparability, we propose an evidence protocol that separates measured, derived, and analytical claims; limits numerical comparison to within-paper results; and requires reporting of hardware, software versions, batch semantics, and thermal state. We synthesize evidence from edge platforms, including Jetson AGX Orin and five inference frameworks; data center GPUs, including A100 and H100 with three LLM serving engines; and quantization studies across the Llama-3.1 family. The synthesis shows that deployment outcomes are governed by cross-layer interactions that no single-layer analysis can predict. We conclude with a constraint-aware selection procedure and open problems in compiler-serving co-optimization, cross-hardware performance prediction, and standardized energy reporting.
[LG-66] MUC-FL: Block-Wise Marginal Utility Contribution for Communication-Efficient Federated Learning
链接: https://arxiv.org/abs/2609.10545
作者: Akshay Mhatre,Vikram Karthick,Deepti Gupta,Jia Zou
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注:
Abstract:Federated Learning (FL) enables distributed model training without centralizing data but suffers from high communication overhead. To address this, we propose Block-Wise Marginal Utility Contribution (MUC), a framework that selectively transmits only the most impactful data blocks based on their contribution to model performance. To evaluate our framework, we apply it to a multimodal dataset integrated from multiple MIMIC clinical datasets and show that only 24 out of 1,135 candidate blocks (1.76%) carry meaningful improvement signals, enabling a potential communication reduction of 45-50% while maintaining or improving model quality. Our deduplication-based block selection achieves a macro F1 score of 0.8566 compared to 0.8155 for standard federated optimization, demonstrating that selective transmission can improve performance, particularly in underrepresented classes.
[LG-67] Evaluating Time-Series Foundation Models and Multimodal Dietary Context for CGM Forecasting
链接: https://arxiv.org/abs/2609.11872
作者: Bowen Zhang,Hsiu-Wen Cheng,Hongyu Yang,Evie L. Shen,Joleen Vansomphone,Yuna Li,Kerry Zhou,Zitian Qu,Suning Zhao,Xiangning Deng,Hua Zhou,Jin J. Zhou
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Continuous glucose monitoring (CGM) provides high-frequency measurements of glucose dynamics and enables short-term glucose forecasting for diabetes management. Although time-series foundation models have shown strong general forecasting ability, their effectiveness for CGM prediction and the added value of multimodal dietary context remain unclear. We conduct a comprehensive empirical study using eight public CGM datasets spanning Type 1 diabetes, Type 2 diabetes, and non-diabetes populations. Under a unified protocol across multiple context lengths and prediction horizons, zero-shot foundation models did not consistently outperform strong task-specific baselines such as Elastic Net and PatchTST. In contrast, lightweight fine-tuning substantially improved forecasting performance. For example, fine-tuned Chronos-Bolt reduced RMSE by 6.5%-18.4% in the T1D cohort and by 8.6%-18.2% in the non-diabetes/T2D cohort, with comparable improvements in both in-distribution and out-of-distribution test settings. We further evaluate multimodal dietary context using CGMacros, which provides temporally aligned CGM signals, food images, and macronutrient records. A residual-based fusion framework reduced overall RMSE by approximately 3% and postprandial RMSE by approximately 15% relative to the CGM-only baseline. Moreover, Chronos-based CGM representations were more strongly correlated with observed postprandial glucose increments than representations from LSTM and CatBoost, even after those models incorporated additional dietary modalities, suggesting that pretrained temporal representations better preserve meal-induced excursion patterns. These findings show that foundation models require CGM-specific adaptation for reliable forecasting and that dietary context provides clinically meaningful signals beyond CGM alone, especially during postprandial periods.
[LG-68] Near-Optimal Reinforcement Learning with Multi-Step Transition Lookahead
链接: https://arxiv.org/abs/2609.11807
作者: Corentin Pla,Hugo Richard,Marc Abeille,Vianney Perchet
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:We study reinforcement learning (RL) with transition look-ahead, where the agent may observe which states would be visited upon playing any sequence of \ell actions before deciding its course of action. Although look-ahead can substantially improve achievable performance, it is known that optimal planning with multi-step transition look-ahead is NP-hard, but this hardness was established using discount factors arbitrarily close to one. It was therefore unknown whether the problem remains hard for any discount factor, and whether near-optimal planning can nevertheless be performed efficiently. We resolve both questions. First, we show that for every fixed rational discount factor ( \gamma\in(0,1) ), exact planning remains NP-hard. Second, we introduce a randomized polynomial-time approximation scheme for every fixed look-ahead depth. We then extend our approach to unknown transitions and stochastic rewards using optimism and variance-adaptive confidence bounds. The resulting algorithm achieves cumulative regret whose leading term matches classical tabular discounted RL up to logarithmic factors. Thus, although exact planning with transition look-ahead is NP-hard, efficient near-optimal planning and learning remain possible.
[LG-69] Sparsity Regularized and Robust Mean Variance Portfolio Selection Under Ellipsoidal Uncertainty
链接: https://arxiv.org/abs/2609.11749
作者: Deniz Akkaya,Emre Can Yayla,Buse Şen,Mustafa Ç. Pınar
类目: Optimization and Control (math.OC); Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:We investigate mean-variance portfolio selection with an \ell_0 -penalty to promote sparsity in asset allocations. Uncertainty in the mean return vector is incorporated through an ellipsoidal uncertainty set, yielding a robust sparse optimization framework. We characterize the structure of both local and global minimizers and exploit these properties in the risk minimization and return maximization formulations. Building on this structural insight, we develop a branch-and-bound algorithm tailored to the resulting robust sparse portfolio problems, together with a new pruning rule that can discard exponentially many candidate portfolios in a single step. Extensive computational experiments on real market data, together with comparisons against a mixed-integer second-order cone programming solver, demonstrate the effectiveness and competitiveness of the proposed approach.
[LG-70] Learning structural balance of graphs from quantum spectral features
链接: https://arxiv.org/abs/2609.11736
作者: Stefano Scali,Oleksandr Kyriienko
类目: Quantum Physics (quant-ph); Disordered Systems and Neural Networks (cond-mat.dis-nn); Machine Learning (cs.LG); Social and Information Networks (cs.SI)
*备注: 12 pages, 6 figures
Abstract:We develop a quantum approach to spectral feature extraction from the density of states (DOS) of a problem-dependent Hamiltonian, and apply it to machine learning on signed graphs. We propose to embed a signed graph as an Ising model instance with positive and negative interactions, and use the standardized moments of the Ising DOS as features for learning. We show that these moments count signed closed walks, are switching-invariant, and are size-free by construction. As a benchmark, we target learning the frustration index, an NP-hard measure of structural balance that can be labeled exactly at moderate size. At zero field, the models can be sampled classically, allowing the quantum extraction procedure to be certified against exact ground truth. We propose DOS-QPE, a phase estimation on a purified maximally mixed probe, which samples the spectral density with orders of magnitude fewer shots than Hadamard test-based trace sampling and feeds the resulting features directly into classically trained models. On 1.4\times10^5 labeled graphs the exact DOS determines the frustration index, and five moments recover it with a mean error of 0.4, well below one sign flip. Beyond zero field, the underlying trace-estimation problem is DQC1-complete, providing access to spectral features for which no efficient classical sampling method is known. Our work opens routes towards quantum applications in social network balance analysis, spin-glass studies, correlation clustering, and protein-interaction networks.
[LG-71] Generalization Analysis of Distributed Kernel-based Robust Gradient Descent Algorithms
链接: https://arxiv.org/abs/2609.11712
作者: Jun-Yi Meng,Zheng-Chu Guo,Yuan Mao
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Operator Algebras (math.OA); Probability (math.PR)
*备注: 40 pages, 4 figures
Abstract:In this paper, we investigate the generalization performance of distributed gradient descent algorithms in a reproducing kernel Hilbert space under a robust loss function l_\sigma . By exploiting the spectral characterization of gradient descent together with the intrinsic properties of robust loss functions, we establish optimal learning rates for the distributed kernel-based robust gradient descent (DKRGD) algorithm with an appropriately chosen scale parameter \sigma . The proposed parameter choice of \sigma simultaneously alleviates the saturation phenomenon and guarantees statistical robustness. A key technical contribution is a novel error analysis that provides substantially sharper bounds for products of operators, thereby significantly relaxing existing restrictions on the maximum number of local machines while retaining optimal learning rates. Finally, we develop a communication-efficient strategy that further improves the convergence performance of DKRGD.
[LG-72] Geospatial Foundation Models Capture Health-Relevant Dimensions of Place Beyond Conventional Social Risk Indices
链接: https://arxiv.org/abs/2609.11689
作者: Nathaniel Hendrix,Carl Y. Zhang,Chris Heitzig,Andrew Bazemore,David H. Rehkopf
类目: Applications (stat.AP); Machine Learning (cs.LG)
*备注:
Abstract:Area-based social risk indices summarize residents’ socioeconomic conditions but incompletely capture physical features of place that may affect health. We evaluated whether numerical representations of physical place produced by four geospatial foundation model families from 2022 satellite data explained residual variance in tract-level associations between the Area Deprivation Index, Social Deprivation Index, and Social Vulnerability Index with health outcomes. We used LightGBM to predict variables from the American Community Survey and 40 chronic disease and health-behavior outcomes from CDC PLACES across 82,646 census tracts in the contiguous United States, evaluating performance across 10 held-out states. Among survey variables, models were moderately predictive of some variables including housing type (R-squared up to 0.54) but weak for disability, unemployment, and income disparity. For health outcomes, models explained up to 54% of variance left unexplained by social risk indices, with the largest gains for annual checkups, arthritis, and high blood pressure. Mean total variance explained by geospatial foundation models across the 40 health-related outcomes increased from 0.31 in the smallest tract-size decile to 0.39 in the largest. Geospatial foundation models capture health-relevant features of place not represented by conventional social risk indices and may usefully augment them in epidemiological analyses.
[LG-73] Identifiability of Nonnegative Tensor Decompositions via Positive Scattering
链接: https://arxiv.org/abs/2609.11606
作者: Haoming Wang,Ming Yuan
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Combinatorics (math.CO); Statistics Theory (math.ST)
*备注:
Abstract:Identifiability of tensor decompositions is often established through linear-algebraic conditions on the factor families. For nonnegative decompositions, however, positivity provides additional information that is not captured by dimension and independence alone: nonnegative terms cannot cancel, and their supports constrain competing decompositions. We introduce a positive scattering term that quantifies this additional source of identifiability and combine it with the dimension budget underlying the Lovitz–Petrov generalization of Kruskal’s theorem. For every subset of components, we obtain two sufficient conditions: a threshold of 2|S|-2 guarantees minimality and nonnegative rank, while the stronger threshold 2|S|-1 guarantees uniqueness among nonnegative decompositions of the same length. The key result is a positive splitting inequality for irreducible exchanges of nonnegative rank-one tensors, which combines the dimension constraint with support-induced geometric rigidity. Although the scattering term is defined through an optimization over intermediate factor spaces, we show that its mode costs are exactly 0 , 1 , or +\infty , yielding an exact activation characterization in terms of graph connectivity. The resulting criterion can strictly certify sparse nonnegative tensor decompositions beyond the reach of Kruskal and Lovitz–Petrov conditions, including examples for which those conditions fail even after reshaping. In the matrix case, the two criteria reduce respectively to full-rank factorization and two-sided separability.
[LG-74] A distribution-free certification framework for trustworthy crash-severity prediction
链接: https://arxiv.org/abs/2609.11592
作者: Amir Rafe,Subasish Das
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Crash-severity models inform screening, dispatch and site prioritization, yet are deployed without a finite-sample statement of what one prediction means. Off-the-shelf guarantees fail here, because the features that make crash severity distinctive defeat them: the KABCO outcome is ordinal, the recorded label is a field assessment agreeing with medical severity about half the time, erring in a structured way, and deployment crosses jurisdictions and years calibration never saw. We develop a certification layer that wraps any severity model unmodified, with distribution-free guarantees using this structure: contiguous ordinal sets that read as “B or worse”; per-class validity for any pre-declared partition, with an oracle efficiency characterization; transfer of coverage to unobserved true severity through a declared reporting band, with a worst-case sharpness result; a one-sided certificate under deployment shift; and severity-weighted risk control. The guarantees compose with an attributable slack budget. The same analysis bounds what certification can achieve. A certified set’s informativeness is governed by a functional of the true law that no base model can evade and that cannot be lower-bounded distribution-free; given a declared misreporting channel identified from record-linkage data, a nonvacuous lower bound on that floor becomes computable. On 5.2 million Texas records across seven base models spanning four decades, the layer attaches identical validity and certifies, on the vulnerable road users, a model-independent floor on set width that no base model beats, separating it from a remainder that stays bounded but distribution-free unidentifiable. The framework is released as an open-source package with theorem-level tests.
[LG-75] Risk-Averse Decision Making with Multi-Level Reliability Guarantees
链接: https://arxiv.org/abs/2609.11524
作者: Amirmohammad Farzaneh,Osvaldo Simeone
类目: Machine Learning (stat.ML); Information Theory (cs.IT); Machine Learning (cs.LG)
*备注:
Abstract:Many applications in engineering, including wireless broadcasting, require designs that provide performance certificates at different target outage levels. This paper studies the problem of maximizing the weighted average of such certificates in the presence of uncertainty about the true system state. The problem is shown to be equivalent to an optimization over nested prediction sets, connecting to the literature on conformal prediction and extending prior art on single-level risk-averse decision making. Furthermore, we derive a dual formulation that decouples optimization across input values. Numerical experiments on a diversity-based wireless transmission system illustrate the cost of enforcing multi-level certificates with a single shared policy and trace the Pareto trade-off between multiple reliability levels.
[LG-76] Improving the Sensitivity of Gravitational Wave Detection with Weighted Conformal Prediction
链接: https://arxiv.org/abs/2609.11401
作者: Ann-Kristin Malz,Gregory Ashton,Nicolo Colombo
类目: General Relativity and Quantum Cosmology (gr-qc); Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:In the last decade, kilometre-scale interferometric gravitational-wave detectors have observed hundreds of compact binary mergers, the majority of which are binary black holes. However, the data are noise-dominated, and multiple independent search algorithms (pipelines) are used to enhance sensitivity and improve robustness. Rather than the standard approach of selecting the most significant pipeline output, we combine the outputs from all pipelines using a conformal prediction-based framework to provide statistically rigorous confidence estimates for candidate events. While combining pipelines improves sensitivity and ranking robustness, it requires a principled statistical framework that remains valid as data properties evolve across observing runs. A key challenge is distribution shifts between simulated datasets used for training and calibration and the real, unlabelled, observations used for testing, which can invalidate coverage guarantees and bias confidence estimates. In this work, we address this challenge by incorporating likelihood-ratio reweighting into our conformal prediction framework to account for covariate shift. Using mock datasets containing simulated signals, we demonstrate that weighted conformal prediction restores well-calibrated coverage under covariate shift and increases the confidence of events near the detection threshold, recovering true signals that would otherwise be missed.
[LG-77] A Two-Mirror Faceted Projection System for EUV Lithography
链接: https://arxiv.org/abs/2609.11299
作者: Vasiliy A. Es’kin,Egor V. Ivanov,Olga V. Martynova
类目: Optics (physics.optics); Machine Learning (cs.LG); Applied Physics (physics.app-ph); Classical Physics (physics.class-ph); Computational Physics (physics.comp-ph)
*备注:
Abstract:We propose an all-reflective two-mirror projection system for extreme ultraviolet (EUV) lithography operating at exposure wavelengths of 13.5 ~nm (Mo/Si) and 11.2 ~nm (Ru/Be), delivering a fourfold ( 4\times ) demagnification of the periodic mask pattern at a numerical aperture approaching unity ( \mathrmNA_\max \approx 0.993 ). In contrast to conventional EUV projection objectives that incorporate 6–10 aspheric mirrors with an overall optical throughput of less than 15% , the proposed design redirects each accepted discrete spatial diffraction order scattered by the mask onto the wafer via a dedicated pair of planar mirror facets. The number of reflections is strictly fixed at two for all accepted orders, retaining 50 – 60% of the power leaving the mask in each accepted order. We derive a spatial geometry providing rigorous optical path length equalization across all diffraction orders, thereby removing order-dependent propagation phase shifts. Individually optimized 30-bilayer Bragg multilayer coatings are designed for each facet using the transfer matrix method combined with global evolutionary optimization algorithms. The architecture is generalized to a three-dimensional vector formulation with a two-dimensionally periodic mask. Utilizing inverse lithography technology, Fourier parameterization, and a differentiable electromagnetic modal waveguide solver, we solve the synthesis problem for binary absorber masks (La absorber on a Ru/Be/Sr multilayer mirror). We demonstrate simulated aerial images of sub-10-nm features on the wafer (isolated peaks with a full width at half maximum (FWHM) of approximately 5.4 ~nm and line pairs with a critical dimension of 6 ~nm) and find that the two peaks remain resolved for the tested wafer defocus values from 0 to 5 ~nm along the z -axis.
[LG-78] A Hilbert-Valued Functional Decomposition Framework for Explaining Time-Dependent Outputs
链接: https://arxiv.org/abs/2609.11295
作者: Sophie Hanna Langbein,Niklas Koenen,Marvin N. Wright,Julia Herbinger
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Feature-based explanations quantify features’ influence on model predictions, but are primarily designed for scalar outputs. In many applications, however, outputs are functional or multivariate, such as time-dependent trajectories in demand forecasting. Consequently, existing approaches typically explain each output location independently, ignoring dependencies across the output components. We address this limitation by developing a unified framework for feature-based explanations of time-dependent outputs. Specifically, we generalize functional decomposition to Hilbert-valued prediction functions and extend an existing feature-based explanation framework to this setting. Our framework introduces kernel-based output representations that enable time-dependency-aware explanations at multiple levels of temporal granularity, including time-specific, time-resolved, and time-aggregated, while providing a unified view in which existing methods arise as special cases. We validate our framework on synthetic and real-world data, including intraday financial market volatility prediction and energy demand forecasting.
[LG-79] Rethinking Radiomap Blind Prediction with Limited Environment and Configuration Representations
链接: https://arxiv.org/abs/2609.11255
作者: Xiaojie Li,Yu Han,Han Fang,Shangqing Liu,Shi Jin,Chao-Kai Wen
类目: ignal Processing (eess.SP); Machine Learning (cs.LG)
*备注: This paper has been accepted for presentation at IEEE Globecom 2026
Abstract:Radiomap blind prediction infers radiomaps from observable representations of the propagation environment and base station (BS) configuration without field measurements. These representations are inherently incomplete and cannot uniquely determine the target radiomap. Under squared loss, we identify the conditional-mean radiomap as the population-optimal deterministic target and decompose domain risk into target-approximation error and irreducible uncertainty. The train-test risk gap motivates propagation priors as cross-domain guidance, although their partial or simplified forms may bias the attainable predictor. We therefore propose RadioDecomp, which treats a prior-guided predictor as a correctable base and uses deterministic residual refinement to learn its remaining predictable discrepancy. We instantiate RadioDecomp as RadioLSR (LoS-Shadow-Residual). Experiments under cross-configuration and cross-environment settings show that RadioLSR is especially effective for cross-configuration generalization and provides overall gains over a controlled monolithic counterpart under cross-environment generalization.
[LG-80] Coherent Floquet quantum reservoirs for molecular property prediction
链接: https://arxiv.org/abs/2609.11071
作者: Luofei Wang,Da Zhang,Congren Wang,Yiming Li,Yuxiao Yang,Xuan Zhang,Xuefeng Cui,Zhang-Qi Yin
类目: Quantum Physics (quant-ph); Machine Learning (cs.LG)
*备注: 15 pages, 10 figures
Abstract:Quantum reservoir computing (QRC) uses quantum dynamics to represent input histories for prediction through a trained classical readout. Discrete time crystals (DTCs) exhibit robust subharmonic responses under periodic driving, and previous work has used their dynamics to construct DTC-QRC. Here we construct a DTC-based reservoir architecture to predict molecular properties from structural and dynamical observations. Coherent Floquet evolution processes local molecular graph events and surface-hopping frames, while controlled reset regulates the contribution of earlier inputs. Measurements at the end of each input sequence yield a feature vector of fixed dimension. Trained classical decoders use this vector for inhibitor-activity and blood–brain-barrier permeability classification and electronic-gap forecasting, while the reservoir parameters remain fixed during training. With matched input lengths and output widths, DTC-QRC outperforms echo-state networks on long-prefix graph classification and the studied ethene gap forecasting tasks. Dephasing lowers performance in both applications, consistent with a role for coherent propagation. Experiments on the Quafu superconducting quantum cloud platform show that pair observables retain task information under device noise. The architecture provides a common framework for molecular screening and time-resolved property prediction using quantum reservoir computing.
[LG-81] he Platonic brain bridge hypothesis: human brain networks as an architectural prior for omni models
链接: https://arxiv.org/abs/2609.10947
作者: Pengfei Zhang,Biao Tian,Xiangang Li,Li Liu
类目: Neurons and Cognition (q-bio.NC); Machine Learning (cs.LG)
*备注: 42 pages, 8 figures; includes 13 pages of Supplementary Information
Abstract:We propose the Platonic brain bridge hypothesis: omni models, which process video, audio and text jointly like the brain, converge on brain-like representations, and the correspondence is bidirectional. From model to brain, brain-likeness of seven omni models is stable across participants, and our encoding models on their internal hidden states rank first on the Algonauts 2025 out-of-distribution leaderboard. From brain to model, three contributions follow. Brain-MoE gives seven cortical networks one brain-pretrained expert each and raises held-out accuracy in all 15 model-benchmark pairs by 6.42 percentage points on average. Brain-AVQA builds questions from video clips labelled by the most responsive brain network; the real network-to-expert map exceeds shuffled maps in-domain on all three models. Brain-Scope uses sparse autoencoders to localize the correspondence to a small subset whose removal weakens brain prediction in all three bases tested. Human brain networks are therefore a usable architectural prior for omni models.
[LG-82] scDEFT: A deep learning framework for drug-effect prediction and counterfactual reasoning
链接: https://arxiv.org/abs/2609.10831
作者: Murthy Devarakonda
类目: Quantitative Methods (q-bio.QM); Machine Learning (cs.LG)
*备注: 21 Pages, 5 Figures, 3 Tables, and a Supplementary Notes section for overflow material. Approximately 3000 words (excluding Figures, Tables, and Captions)
Abstract:Longitudinal single cell atlases now capture matched pre treatment and post treatment states from responders and non responders, presenting an opportunity to mechanistically explain why two patients on the same drug diverge. We introduce scDEFT (single cell Drug EFfect Transducer), which treats a drug as a conditioning operator on cell representations, enabling prediction and explanation. In scDEFT, feature wise linear modulation produces drug conditioned cell latents, learned under abundant per cell supervision and then frozen. Two independent heads aggregate those latents over shared transcriptional neighborhoods to predict drug induced state change and responder status. A backward stage ranks the latent dimensions by how strongly they separate responders from non responders and maps them to genes under a cell composition control. On a harmonized inflammatory bowel disease atlas of 1.16 million cells, three cohorts and two drug classes, scDEFT predicts state change at 45% of the baseline to reproducibility ceiling headroom and stratifies responders before treatment at AUROC 0.70, where standard predictors remain at chance. These predictions and the drivers behind them support target and co target nomination, patient stratification, and counterfactual prediction of unseen drug cohort effects.
[LG-83] Weighted Empirical Risk Minimization for Machine Learning under Long-Range Dependence: Exact Pathwise Rates and Learning-Error Geometry
链接: https://arxiv.org/abs/2609.10767
作者: Elina Moldavskaya
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 42 pages, 5 figures
Abstract:We develop an exact almost-sure learning theory for smooth parametric models trained by regularly weighted empirical risk minimization on long-range dependent data. The training observations are generated from a fixed finite window of a stationary Gaussian sequence, and the sample weights are regularly varying. If the loss gradient at the population minimizer has Wiener-chaos rank m and a nonzero low-frequency coefficient, then, in the long-memory interior regime, the finite-lag score reduces on the iterated-logarithm scale to a single weighted Hermite chaos. This yields an almost-sure Bahadur representation, an exact limsup law for the learned parameter, and, for m\ge2 , the functional cluster set of the complete learning trajectory. The polynomial learning exponent is determined by the memory parameter and the chaos rank and is invariant under the admissible power weighting, whereas the sharp pathwise constant and cluster geometry depend on the weights. In the rank-one case, global optimization over the admissible power exponents shows that every optimizer is positive. Time-series prediction and classification examples illustrate the results.
[LG-84] A Quantum-Inspired Dequantization Method for Diagonally Weighted Matrix Functions: Application to Learning with Optimized Random Features
链接: https://arxiv.org/abs/2609.10729
作者: Natsuto Isogai,Mio Murao,Hayata Yamasaki
类目: Quantum Physics (quant-ph); Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 18 pages, 1 figure
Abstract:Quantum-inspired classical algorithms have dequantized several quantum machine learning routines by replacing quantum linear-algebra subroutines with classical counterparts. However, the sampler based on quantum singular value transformation (QSVT) for learning with optimized random features is not covered by existing dequantization frameworks, because the matrix to be inverted is not itself available through sampling access. In this work, we develop a classical algorithm to address this type of quantum-advantage candidate. Our method samples heavy indices, reduces the transformation to a small principal block, and outputs a sparse classical representation with operator-norm guarantees. Applying this method dequantizes the sampler for optimized random features, giving a classical sampler with prescribed accuracy and polynomially related runtime. These results show that the factorization underlying a quantum block encoding can itself provide sufficient classical structure even when sampling-and-query access to the composite matrix is unavailable.
[LG-85] Sequence-Informed Geometric Evaluation of RNA 3D Structures
链接: https://arxiv.org/abs/2609.10644
作者: Andrea Zerio,Yighua Yao,Alessandro Micheli,Roland G. Huber,Mile Sikic,Samir Bhatt,Andres R. Masegosa,Yuangang Pan
类目: Biomolecules (q-bio.BM); Machine Learning (cs.LG)
*备注:
Abstract:Computational RNA structure pipelines generate many candidate conformations for the same sequence. Reliable evaluation therefore requires more than recognising plausible geometry, it requires determining whether that geometry is compatible with the sequence. We introduce SIRGE, a sequence-informed geometric evaluator that conditions structural representations on nucleotide embeddings from a pretrained RNA language model. Early results show that SIRGE outperforms established evaluators in Kendall-- \tau alignment, Top-1 selection, and Top-3 ranking. Controlled comparisons further show that sequence conditioning corrects errors made by an otherwise matched geometric model and improves target-level rank structure. These findings provide initial evidence that pretrained sequence representations supply ranking information that complements geometric reasoning.
[LG-86] A Station-Based Evaluation of Machine Learning-based Weather Forecasting Models in Northern Norway
链接: https://arxiv.org/abs/2609.10564
作者: Siyan Chen,Lars Uebbing,Eirik Mikal Samuelsen,Georgios Leontidis,Arnt-Børre Salberg,Sébastien Lefèvre,Robert Jenssen,Kristoffer Wickstrøm
类目: Atmospheric and Oceanic Physics (physics.ao-ph); Machine Learning (cs.LG)
*备注: 19 pages, 10 figures, 8 tables
Abstract:Recent machine learning weather prediction (MLWP) models have demonstrated remarkable forecasting skill on global reanalysis-based benchmarks. However, their performance remains unclear in challenging environments such as Northern Norway, where narrow fjords and rapidly changing weather result in highly variable local wind conditions. In this case study, we evaluate FourCastNet3 (FCN3), GraphCast, and ECMWF High Resolution Forecast (HRES) for wind speed forecasting using multi-year station observations from Northern Norway, focusing on their relative performance, generalization beyond the training period, and performance under high-wind conditions. Our results show that HRES slightly outperforms FCN3 and GraphCast, with an overall RMSE of 2.89 \mathrmm,s^-1 , compared to 2.96 \mathrmm,s^-1 for FCN3 and 2.94 \mathrmm,s^-1 for GraphCast. Notably, the MLWP models maintain comparable performance beyond their respective training periods, with no clear evidence of noticeable degradation. FCN3 performs best under high-wind conditions, although all models substantially underestimate strong winds. Our findings suggest that MLWP has become competitive with NWP for local wind, but further refinements are still needed to capture complex terrain better.
[LG-87] Supply Chain Analytics: A Data-Driven Approach
链接: https://arxiv.org/abs/2609.10563
作者: Elioth Sanabria
类目: Optimization and Control (math.OC); Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: Draft chapters / working book manuscript, 101 pages
Abstract:Modern supply chain networks increasingly rely on real-time data to navigate structural uncertainties, market volatility, and operational disruptions. This manuscript bridges the gap between statistical data-driven learning and robust decision-making frameworks in logistics and operations management. We present a comprehensive, mathematically rigorous treatment of supply chain analytics, moving from empirical demand forecasting to optimal inventory and network control under uncertainty. Key topics explored include sample minimization, dynamic programming recursions for time-varying inventory replenishment, network fulfillment frameworks, and advanced distributionally robust optimization (DRO) via transport theory to hedge against rare events. By integrating predictive statistical models with prescriptive control algorithms, such as column generation for vehicle routing and non-homogeneous queueing regimes, this text provides the foundational tools necessary for designing resilient, data-driven automated systems. It serves as both a theoretical blueprint and an algorithmic guide for researchers and practitioners operating at the intersection of machine learning, mathematical optimization, and applied probability.
[LG-88] EVTradeMatch: A Mobility-Aware Multi-Objective Matching Framework for EV–EV Energy Trading
链接: https://arxiv.org/abs/2609.10551
作者: Md. Mahfujur Rahman,Alistair Barros,Raja Jurdak,Darshika Koggalahewa
类目: Optimization and Control (math.OC); Machine Learning (cs.LG); Systems and Control (eess.SY)
*备注:
Abstract:Peer-to-peer energy trading among electric vehicles (EVs) can improve charging flexibility under limited charging infrastructure, but effective EV–EV trading requires coordinated provider–consumer matching under journey-specific conditions. This paper proposes EVTradeMatch, a prediction-guided multi-objective optimization framework for mobility-aware EV–EV energy trading. Building on the EVNextTrade study, a prior learning-to-rank model for charging-node recommendation, we define charging-node suitability as a prediction-derived score reflecting the appropriateness of assigning a provider–consumer pair to a candidate charging node based on mobility, energy, and contextual trading features. This score is used as a guidance signal and as an explicit optimization objective rather than as a hard selection rule. The EV–EV matching problem is formulated as a multi-objective mixed-integer linear program that maximizes matching coverage, transferred energy, and charging-node suitability while minimizing mobility cost under spatial, temporal, one-to-one matching, and charging-node capacity constraints. To approximate Pareto-efficient solutions in wide-area dynamic settings, we develop a tailored non-dominated sorting genetic algorithm II (NSGA-II). Experimental results show that EVTradeMatch improves transferred energy by 74.2–82.8% and charging-node suitability by 8.3–84.4% compared with proximity- and auction-based state-of-the-art methods, while improving matching coverage by 3.74–25.07 percentage points. Balanced NSGA-II solutions achieve 53.07 \pm 0.76% matching coverage and transfer 1701.94 \pm 17.03 kWh, with higher mobility cost as an explicit trade-off against travel-minimizing methods. Pareto-front analysis shows that the framework supports flexible selection among high-coverage, high-energy, low-mobility-cost, and high-suitability solutions according to operational priorities.
附件下载
