本篇博文主要内容为 2026-09-22 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 (2026-09-22)
今日共更新1535篇论文,其中:
- 自然语言处理共233篇(Computation and Language (cs.CL))
- 人工智能共384篇(Artificial Intelligence (cs.AI))
- 计算机视觉共272篇(Computer Vision and Pattern Recognition (cs.CV))
- 机器学习共426篇(Machine Learning (cs.LG))
- 多智能体系统共30篇(Multiagent Systems (cs.MA))
- 信息检索共39篇(Information Retrieval (cs.IR))
- 人机交互共59篇(Human-Computer Interaction (cs.HC))
多智能体系统
[MA-0] Perception-Aware Communication Middleware for Distributed Visual Perception in UAV Swarms
【速读】：该论文旨在解决无人机群（UAV swarm）在分布式视觉感知应用中面临的低延迟通信挑战，其核心问题是传统基于包级质量保障（QoS）的通信机制无法满足感知数据完整、及时且可用性的需求。尽管单个数据包成功传输，仍可能导致图像不完整或失真，从而影响基于生成式AI（Generative AI）的推理任务性能。为此，论文提出一种感知感知通信中间件（perception-aware communication middleware），其关键在于将完整的感知数据样本（如图像帧）作为通信对象，并在此基础上构建感知专用服务：包括图像分片与重构、并发包传输、优先级感知调度以及图像质量评估。该方案通过在轻量级UDP代理发布-订阅架构上扩展上述功能，实现了端到端低延迟、高吞吐量的感知数据传输，在异构硬件测试平台上对YOLOv8n目标检测模型的验证表明，该中间件不仅有效提升了感知流量在高背景负载下的优先级处理能力，还通过中间件层面的图像质量评估显著缓解了目标检测精度下降问题。该研究为多智能体移动信息物理系统中集成面向人工智能的数据处理机制提供了初步框架。
链接: https://arxiv.org/abs/2609.24964
作者: Manveen Kaur,Kevin Loi,Ifunanya Okafor,Daniel Ng,Joseph Lucey-Renteria
机构: California State University, Los Angeles(加州州立大学洛杉矶分校)
类目: Multiagent Systems (cs.MA)
备注:
Abstract:Unmanned Aerial Vehicle (UAV) swarms increasingly support safety-critical applications that rely on distributed visual perception. Meeting the low-latency requirements of these applications can require perception models to execute within the swarm on inference-capable UAVs, creating a need for efficient UAV-to-UAV transport of high-bandwidth perception data. However, the Quality-of-Service (QoS) requirements of perception differ from conventional packet-level QoS; successful delivery of individual packets does not ensure that a complete, timely, and usable image is available for inference. We present a novel perception-aware communication middleware that treats complete perception-data samples as the communication objects for which QoS must be satisfied. The middleware extends a lightweight UDP broker-based publish-subscribe architecture with perception-specific services, including image fragmentation and reconstruction, concurrent packet transmission, priority-aware scheduling, and image quality assessment. The middleware is evaluated on a heterogeneous hardware testbed emulating a UAV swarm using YOLOv8n object detection. Experimental results demonstrate low end-to-end application latency, substantially higher throughput than a lightweight UDP broker, effective prioritization of perception traffic under increasing background load, and mitigation of object-detection degradation through middleware-level image quality assessment. This work provides an initial framework for integrating AI-specific data handling into communication middleware to support emerging distributed AI applications in multi-agent mobile cyber-physical systems.
[MA-1] Mixed-integer flow formulations for motion planning and decision-making of networked multi-agent systems
【速读】：该论文旨在解决网络化多智能体系统（Multi-Agent Systems, MAS）在基于混合整数线性规划（MILP）的轨迹规划与决策模型中，如何高效维持通信连通性的关键问题。现有方法通常依赖于子环消除（Subtour Elimination Constraints, SEC）约束来保证标准连通性，但其导致不等式约束数量随智能体规模呈指数增长，严重限制了大规模系统的求解能力。为此，论文提出采用基于流（flow-based）的连通性维护约束，将标准连通性与k跳连通性以流变量的形式编码进MILP模型中。其核心创新在于：相较于SEC方法，流形式的标准连通性约束将所需不等式约束的增长从指数级降低至多项式级；而针对k跳连通性的流约束不仅显著减少了二元变量的数量，还使变量增长与跳数解耦。尽管引入了大量连续流变量及额外不等式约束，可能增加计算复杂度，但实验表明，在标准连通性场景下，该方法在给定时间限制内可求解更大规模的MAS，性能优于SEC；而对于k跳连通性，虽然理论上的分支定界算法迭代次数减少，但实际平均性能未明显提升，反映出计算开销与结构简化之间的权衡。
链接: https://arxiv.org/abs/2609.24474
作者: Angelo Caregnato-Neto,Paul-Louis Delacour,Raf Van de Plas,Tamás Keviczky,Janito Vaqueiro Ferreira
机构: State University of Campinas (UNICAMP)(坎皮纳斯州立大学); Delft University of Technology (代尔夫特理工大学); Vanderbilt University (范德比尔特大学)
类目: Multiagent Systems (cs.MA)
备注:
Abstract:This work investigates the use of flow-based connectivity maintenance constraints in mixed-integer linear programming (MILP) trajectory planning and decision-making models for networked multi-agent systems (MAS). We integrate flow-based encodings for standard and k-hop connectivity into MILP multi-vehicle maneuvering models that are widely used alongside receding horizon planning strategies. Their necessity and sufficiency is demonstrated, guaranteeing full coverage of potential network topologies. The flow formulation for standard connectivity decreases the growth of the required inequality constraints from exponential to polynomial w.r.t. the size of the MAS when compared to the state-of-the-art subtour elimination (SEC) method. The flow-based k-hop connectivity constraints decrease the number of required binary variables and decouple its growth from the number of hops. However, the impact of these formulations in performance is not straightforward due to the introduction of a substantial number of continuous flow optimization variables and, in the case of k-hop connectivity, additional inequality constraints. We investigate this trade-off through a statistical evaluation of costs and optimization times using a conventional branch-and-bound commercial solver and trials performed with randomized environments for increasingly larger MAS. The results show that the flow formulation outperforms SEC in standard connectivity problems, enabling the solutions to be computed for larger MAS considering the imposed optimization time limit. The reduction in number of binary variables enabled by the k-hop flow formulations decreases the theoretical worst-case number of iterations required by the branch-and-bound algorithm to compute the global optimal solution. Our results show that this advantage did not translate into improvements in the average performance when compared to the baseline.
[MA-2] A Task-Oriented Multi-Agent Framework for Complex Wearable Health Analysis
【速读】：该论文旨在解决可穿戴健康应用中复合型查询（composite query）在使用单一大语言模型（Large Language Model, LLM）处理时存在的透明性不足与可追溯性差的问题，即难以明确判断每个子任务是否被正确执行，以及最终答案所依据的证据来源。其解决方案的关键在于提出一种面向任务的多智能体框架（task-oriented multi-agent framework），将复合查询分解为具有明确类型的任务（typed tasks）和意图（intents），并通过显式的智能体间依赖关系管理来组织执行流程。该框架通过专用智能体分别负责数据检索、纵向分析与健康建议生成，并利用隔离的意图状态（intent states）维护请求边界与证据关联性，从而实现任务相关数据的精准访问与基于数据的纵向分析。实验结果表明，该方法在10,000名虚拟用户的一月可穿戴设备记录上表现优异：在1,500个检索问题中，查询智能体准确率达98.3%，显著降低平均查询阶段令牌消耗（从6,869降至3,136）；在180个多意图问题中，管理智能体实现了100.0%的多意图覆盖率与94.4%的多重集交并比（Multiset Jaccard Similarity）。尽管在信任度与透明度方面优于直接调用大模型的基线方法，但行动可行性（Actionability）尚未一致提升，提示健康建议生成与真实可穿戴数据验证仍是开放挑战。
链接: https://arxiv.org/abs/2609.24107
作者: Kunpeng Yang
机构: Renmin University of China (中国人民大学)
类目: Multiagent Systems (cs.MA)
备注: 23 pages, 4 figures. Code available at this https URL
Abstract:Wearable health questions often combine data retrieval, longitudinal analysis, and health advice over structured records. Prompting a single large language model with a complete record and a composite query obscures whether every request is executed and which evidence supports the answer. We propose a task-oriented multi-agent framework that represents a composite query as distinct intents and typed tasks with explicit intra-intent dependencies. Specialized agents execute retrieval, analysis, and advice tasks; isolated intent states preserve request boundaries and evidence relationships before aggregation. We evaluate the framework on a synthetic dataset of 10,000 virtual users with one month of longitudinal wearable records, covering structured data retrieval, multi-intent recognition, and overall response quality. Across 1,500 retrieval questions, the Query Agent achieves 98.3% accuracy, compared with 97.9% for the Direct LLM baseline, while reducing average query-stage token consumption from 6,869 to 3,136 . On 180 multi-intent questions, the Manager Agent achieves 100.0% Multi-Intent Coverage and 94.4% Multiset Jaccard Similarity. Under the current synthetic evaluation setting, our method receives higher mean Trustworthiness and Transparency scores on both question categories, whereas Actionability does not improve consistently. These results provide preliminary evidence that explicit task organization can support task-relevant data access and data-grounded longitudinal analysis, while leaving health advice generation and validation on real wearable data as open challenges.
[MA-3] sting not presuming adequacy: calibrating generative social simulators against emergent network structure
【速读】：该论文旨在解决生成式社会模拟器（generative social simulator）在验证过程中普遍存在的方法论缺陷：现有验证通常仅依赖表面有效性（face validity），即通过描述性比较模拟生成的网络结构与真实数据，缺乏对参数不确定性的量化评估以及对模型整体适配度（adequacy）的严格检验。其解决方案的关键在于提出一种“适配度感知校准协议”（adequacy-aware calibration protocol），该协议集成多项核心技术：基于摊销后验估计（amortized posterior estimation）的参数推断、合成可辨识性评估（synthetic identifiability assessment）、匹配样本量的适配度检查（包括先验预测可达性与逐统计量后验预测定位）、诊断引导的修复机制，以及统计量留出审计（statistic-held-out audit）。研究以一个真实的二手奢侈品转售市场为案例，构建了四个按渠道-居住地划分的双部买家-品牌网络，采用由语言模型离线生成的人物画像作为行为输入，建立前向模型。结果表明，尽管所有四组单元中的行为参数均可恢复，但存在近似性和过度自信问题；且在每一单元中，观测汇总统计量均超出模拟器的先验预测可达范围，其中购买层级均值是普遍存在的偏差来源。修复策略虽在两组中满足价值块标准，但未能恢复模型整体适配度，而留出审计进一步揭示了此前诊断未捕捉到的“买家广度分散度”缺失问题。通过消融实验发现，语言模型生成的画像优于静态规则基线，但在类别内品牌重命名下表现无一致退化，说明输入画像的有效性主要源于其结构而非具体品牌身份。最终结论指出，在不引入个体间交互或买家广度机制的前提下，独立聚合模型无法同时再现市场的购买层级水平、头部品牌集中度、社区结构及买家广度异质性，从而凸显了复杂社会互动机制在建模中的必要性。
链接: https://arxiv.org/abs/2609.24012
作者: Tengfei Shao,Chao Li,Xu Wang,Masayuki Goto
机构: 未知
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA); Social and Information Networks (cs.SI)
备注: Accepted for publication in the Journal of Artificial Societies and Social Simulation (JASSS). 43 pages (34 main text, 9 supplementary information), 4 figures
Abstract:Validation of generative social simulators often stops at face validity: emergent network structure is compared descriptively, without quantified parameter uncertainty or an adequacy check. We present an adequacy-aware calibration protocol that couples amortized posterior estimation with a synthetic identifiability assessment, a matched-sample-size adequacy check (prior-predictive reachability plus per-statistic posterior-predictive localization), a diagnosis-guided repair, and a statistic-held-out audit. We demonstrate it on a real second-hand luxury resale market with four channel-by-residency cells, each a bipartite buyer-brand network, using a forward model built from persona profiles elicited once, offline, by a language model. The behavioural parameters are recoverable in all four cells, though calibration is approximate and overconfident for one parameter. The observed summary falls outside the simulator’s reachability reference in every cell, with the mean purchased tier as the pervasive discrepancy. The repair meets the value-block criterion in two of four cells but does not restore adequacy, and the held-out audit surfaces a buyer-breadth-dispersion miss no earlier diagnostic detected. A profile-source ablation finds the language-model profiles beat a flat rule baseline in all four cells, yet within-category brand relabelling causes no consistent degradation, so the profiles are a partially validated input whose value rests on structure, not brand identity. Making no causal claim, we conclude that an independent-aggregation account, without agent interaction or a buyer-breadth mechanism, cannot jointly reproduce the market’s purchased-tier level, head-brand concentration, community structure and buyer-breadth heterogeneity.
[MA-4] Eventual and Strong Eventual Notions in Public Announcements
【速读】：该论文旨在解决动态认知逻辑中公共宣告（public announcement）情境下，关于“成功性”“自否性”“真谎言”与“不可能谎言”等概念在迭代宣告过程中的演化问题。其核心挑战在于如何形式化描述这些语义性质在无限甚至超限次宣告序列中的行为，特别是当宣告可无限重复时，命题的真值是否会发生不可逆的变化。解决方案的关键在于引入“最终自否性”（eventual self-refutation）与“强最终自否性”（strong eventual self-refutation）的扩展概念，以及它们的超限版本（transfinite versions），从而支持对任意阶超限迭代宣告的分析。此外，论文还提出“始终信息性”（always informativeness）的概念，用于刻画公式在真或假情况下是否始终提供新信息。主要成果之一揭示了强最终性概念与最终性概念、真值极限条件、统一有界性条件（uniform bound condition）、Moore句与自实现句的不动点视角之间的深刻关联；另一成果则建立了有限与超限版本的最终性与强最终性概念之间的联系，并将其与不动点结构相统一，为理解复杂宣告系统中的动态真值演化提供了形式化框架。
链接: https://arxiv.org/abs/2609.24006
作者: Eiji Yamada
机构: Department of Mathematical and Computing Science, Institute of Science Tokyo (东京科学大学数学与计算科学系)
类目: Logic in Computer Science (cs.LO); Multiagent Systems (cs.MA); Logic (math.LO)
备注: 34 pages, 4 figures
Abstract:In dynamic epistemic logic, the four notions of success, self-refutation, true lies, and impossible lies have been discussed in the context of public announcements. In this paper, we introduce eventual and strong eventual versions of these notions, as well as their transfinite versions, which allow transfinite iteration of announcements. We also introduce the notions of always informativeness when true or false. For example, a formula is eventually self-refuting if, whenever initially true, it eventually becomes false at some finite stage under iterated announcements, and strong eventual self-refutation further requires the formula to remain false at all sufficiently late stages. There are two main results. The first result gives the relationship among strong eventual notions, eventual notions, and several other conditions including conditions on the limit of the truth values of the announced formula, the uniform bound condition, and the fixed-point views of the Moore sentence and the self-fulfilling sentence. The second result gives the relationship among finite and transfinite versions of the eventual and strong eventual notions and the fixed-point views.
[MA-5] MR-SPITE: Accelerating Multi-Robot Conflict Scans via Hierarchical Swept-Volume Approximations
【速读】：该论文旨在解决多机器人同步路径冲突扫描中计算开销大、效率低的问题，尤其是在需要对每对机器人在每个时间步进行详尽碰撞检测并多次重复修复冲突的情况下。其解决方案的关键在于提出一种基于运动段（motion-segment）的保守过滤方法——多机器人SPITE（MR-SPITE），该方法将每条路径划分为时间区间，并为每个运动段分配保守的边界约束；通过区间调度器比较时序重叠运动的边界，若边界不相交则可确定该时间段内无冲突，从而跳过后续精确碰撞检测；仅对无法判定的窗口传递至底层碰撞检测器。该方法与基于VAMP的碰撞检测相结合，集成于ARC框架中，在16台Fetch机器人场景下实现了配对中位数冲突扫描速度提升7.18倍，并使规划时间中位数减少57%，显著提升了效率，同时保持了底层离散化扫描器的行为一致性。
链接: https://arxiv.org/abs/2609.23928
作者: Marta Markowicz,James Motes,Marco Morales,Nancy Amato
机构: University of Illinois(伊利诺伊大学); Instituto Tecnológico Autónomo de México (ITAM)(墨西哥自治技术学院)
类目: Robotics (cs.RO); Computational Geometry (cs.CG); Multiagent Systems (cs.MA)
备注: 9 pages, 4 figures, 2 tables
Abstract:Conflict scanning over synchronized robot paths requires detailed collision checking, potentially across every robot pair at every timestep, and may be repeated many times as conflicts are repaired. We present Multi-Robot SPITE (MR-SPITE), a conservative, motion-segment-based filter for accelerating these scans. MR-SPITE partitions each path into temporal intervals and assigns conservative bounds to each segment. An interval scheduler compares bounds for temporally overlapping motions: disjoint bounds certify the shared window as conflict-free, while unresolved windows are passed to the underlying collision checker. We integrate MR-SPITE into ARC and combine it with VAMP-based collision checking. For 16 Fetch robots, ARC with MR-SPITE achieves a paired median conflict scan speedup of 7.18x and reduces median planning time by 57% relative to the baseline ARC implementation with PRM+VAMP. These results demonstrate that motion-segment bounds complement configuration-level collision acceleration while preserving the behavior of the underlying discretized scanner.
[MA-6] VISTA: An Attention-Based Multi-Agent Reinforcement Learning Architecture for Space Situational Awareness Sensor Tasking
【速读】：该论文旨在解决大规模、动态变化的空间目标目录下，异构且分布式的空间态势感知（Space Situational Awareness, SSA）传感器任务规划问题。随着在轨物体数量的快速增长，传统优化方法难以有效应对有限传感资源在日益庞大的目录中的分配挑战。现有深度强化学习方法虽在小规模场景中表现良好，但受限于固定维度的状态与动作表示，难以扩展至大规模、动态的目录和分布式传感网络。本文提出VISTA（Variable-Entity Intelligent Sensor Tasking Architecture），其关键在于通过融合物理与任务驱动的Top-K检索机制、以目标为中心的注意力机制、循环记忆模块以及基于指针的动作解码策略，实现了每个智能体的观测空间与动作空间与目录规模解耦，从而具备良好的可扩展性。实验表明，在30个轨道目标场景下，VISTA比固定维度的递归基线快31.2%完成目录恢复；在大规模空间任务场景中，相较最强经典参考方法降低97.5%的五小时不确定性，较递归学习者降低99.3%。零样本测试显示，感知能力、目录规模与恢复时间之间近似呈线性关系，且所学策略展现出对传感器模态的自适应能力及对目标种群与初始不确定性变化的泛化性能。这些结果证明，VISTA为异构、分布式的地基与天基传感器网络提供了可扩展的自适应空间态势感知任务规划框架。
链接: https://arxiv.org/abs/2609.23875
作者: Miguel Leiva-Vélez,Adalberto Claudio Quiros,Nicolas Gaston Rozado,Hodei Urrutxua,Víctor Rodríguez-Fernández
机构: ETSIAE-School of Aeronautics, Universidad Politécnica de Madrid(西班牙马德里理工大学航空学院); Indra Sistemas S.A.(西班牙因德拉系统公司); Escuela de Ingeniería de Fuenlabrada, Universidad Rey Juan Carlos(西班牙雷翁·胡安卡洛斯大学富恩拉布雷达工程学院); Department of Computer Systems Engineering, Universidad Politécnica de Madrid(西班牙马德里理工大学计算机系统工程系)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA); Systems and Control (eess.SY)
备注: 19 pages, 6 figures. Code available at this https URL . Submitted to IEEE TAES
Abstract:The rapid growth of resident space objects is increasing the complexity of space situational awareness sensor tasking, challenging classical optimization methods as they allocate finite, heterogeneous, and distributed sensing resources across ever-larger catalogues. Existing deep reinforcement learning approaches show promise in reduced settings, but fixed-dimensional state and action representations limit their ability to scale to large, dynamic catalogues and distributed sensing networks. We introduce VISTA (Variable-Entity Intelligent Sensor Tasking Architecture), a scalable deep reinforcement learning architecture for persistent uncertainty-driven catalogue maintenance across variable object populations and sensor configurations. VISTA combines physics- and mission-informed top-K retrieval with entity-centric attention, recurrent memory, and pointer-based action decoding, thereby keeping each agent’s observation and action spaces independent of catalogue size. We evaluate VISTA across different scenarios, from fixed-size single-sensor benchmarks to large-scale space-based tasking and heterogeneous cooperative sensing. With 30 orbiting targets, VISTA recovers the catalogue 31.2% faster than the fixed-dimensional recurrent baseline. In the large-scale regime, VISTA reduces five-hour uncertainty by 97.5% relative to the strongest classical reference and by 99.3% relative to the recurrent learner. Zero-shot tests up to 20,000 objects reveal near-linear relations between sensing capacity, catalogue size, and recovery horizon. Learned policies also exhibit sensor modality adaptation and generalization to population and initial-uncertainty shifts. Together, these results demonstrate that VISTA provides a scalable framework for adaptive space situational awareness sensor tasking across large, distributed networks of heterogeneous ground- and space-based sensors.
[MA-7] otal Cost of Agency: Exact Attribution of Memory Injection Cost in Multi-Agent LLM Workflows
【速读】：该论文旨在解决多智能体大语言模型（Multi-Agent LLM）工作流中因上下文注入（memory injection）导致的成本不可见性问题。具体而言，工作流中各节点从记忆中检索上下文并将其注入提示词（prompt），这些注入的标记（token）与系统提示和用户查询一样按相同单价计费，但生产环境可观测性工具仅报告总成本，无法区分节点生成的输出标记与接收的输入标记，使得上下文注入部分的成本对实际承担费用的团队而言处于“不可见”状态。为解决此问题，论文提出代理总成本（Total Cost of Agency, TCA），将多智能体工作流成本分解为基线提示、推理、记忆注入、误检惩罚及上下文累积五个可量化组件，并引入一种两遍非计费标记计数方法，直接测量注入标记数量，而非依赖词数等间接代理指标。实验基于200个企业级任务在真实模型API上运行，结果显示：记忆注入占可变成本的13.6%，约占总账单成本的12%；其占比随工作流深度从深度1时的结构零增长至深度6时的27.6%；注入标记数量在测量范围内与深度呈线性关系（R² = 0.9974），且二次拟合显示主系数为负，表明在此深度区间内无凸性增长。研究进一步证明，在固定模型层级下，通过将检索窗口容量从32项降至2项，可使注入标记减少28.7%，同时精度变化在种子级波动范围内，说明该成分具备可控性。此外，研究完整披露图重写变换在孤立情况下近似成本中立，且五项分解中的两项在测试框架中由构造决定为零，而总成本主要受模型层级分配主导——此项被固定视为前期工作，未在本研究中评估提示缓存效果，所有数据均基于未缓存场景。
链接: https://arxiv.org/abs/2609.23790
作者: Vivek Kumar Singh,Preeti Priyam,Gautam Bhowmick
机构: 未知
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA); Performance (cs.PF)
备注: 8 pages, 3 figures, 6 tables. Code, data, and evaluation harness in github: this https URL
Abstract:Every node in a multi-agent large language model (LLM) workflow retrieves context from memory and injects it into its prompt, where those injected tokens are billed as input tokens at the same per-token price as the system prompt and the user query. Production observability tools report total token cost but do not separate the tokens a node generates from the tokens it is handed, so this component of the bill is invisible to the teams paying it. We introduce the Total Cost of Agency (TCA), a decomposition of multi-agent workflow cost into base prompt, inference, memory injection, miss penalty and context-accumulation components, and an exact attribution method: a two-pass, non-billable token count that measures injected tokens directly rather than estimating them from word-count proxies. On a 200-task enterprise benchmark executed against real model APIs, memory injection accounts for 13.6 percent of the variable cost a compile-time optimizer can act on, about 12 percent of the full billed cost, and its share rises from a structural zero at workflow depth one to 27.6 percent at depth six. Injected tokens grow linearly with depth over the measured range (R^2 = 0.9974, depths two through six); a quadratic fit yields a negative leading coefficient, so the data do not exhibit convex growth at these depths. We show the component is controllable at fixed model tier: reducing the retrieval window capacity from 32 to 2 entries lowers injected tokens by 28.7 percent with an accuracy change within seed-level variation. We report in full that our graph-rewriting transforms are approximately cost-neutral in isolation, that two of the five decomposition terms are zero by construction in this harness, and that total workflow cost is dominated by model tier assignment, which we hold fixed and treat as prior work. Prompt caching is not evaluated; all figures are for the uncached case.
[MA-8] Conflicting Pattern Formation by Teams of Anonymous Fully Disoriented Robots
【速读】：该论文旨在解决在完全无方向性（disoriented）的移动机器人系统中，两组自主、匿名且对环境无感知的机器人同时实现两个相互冲突的群体模式形成任务的可行性问题：一是机器人需在未知位置聚集（gathering），二是机器人需分布在圆周上形成圆形构型（circle formation）。传统解决方案依赖于全局轴向一致性和共同的镜像识别性（chirality），而本文针对更严苛的场景——即机器人完全无方向性、无轴向一致性、无共同镜像识别性，研究其在半同步（semi-synchronous）模型下通过非刚性移动（non-rigid movements）完成任务的可能性。其关键解决方案在于提出一种分布式算法，仅对执行聚集任务的机器人组要求全局弱多重性检测（global weak multiplicity detection），而对执行圆形成形任务的机器人组仅需局部弱多重性检测（local weak multiplicity detection），从而在无任何方向性共识的前提下实现了两组任务的协同求解。
链接: https://arxiv.org/abs/2609.23454
作者: Animesh Maiti,Prakhar Shukla,Subhash Bhagat
机构: IIT Jodhpur (印度理工学院朱德普尔分校); IIT Jodhpur (印度理工学院朱德普尔分校); IIT Jodhpur (印度理工学院朱德普尔分校)
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Computational Geometry (cs.CG); Multiagent Systems (cs.MA); Robotics (cs.RO)
备注:
Abstract:Two groups of autonomous, anonymous, and oblivious mobile robots are deployed in the two-dimensional Euclidean plane, each assigned a distinct task. We study a setting where the two groups must simultaneously solve two conflicting pattern formation problems: the \textitgathering problem, where robots gather at a point not known to them a priori, and the \textitcircle formation problem, where robots occupy distinct positions on the boundary of a circle. Although each robot knows its own task, it cannot identify other members of its group. A prior solution~\citeConflict-1 addressed this problem for asynchronous robots having \it direction-only axis agreement and \it global weak multiplicity detection capability available to all robots in both groups. In contrast, in this work, we consider fully \it disoriented robots without any axis agreement or common \textitchirality. We study the feasibility of a solution to this problem for \it disoriented robots. We propose a distributed algorithm that solves the problem for semi-synchronous disoriented robots with non-rigid movements. Our proposed algorithm assumes global weak multiplicity detection only for the gathering group, while for the circle formation group, it requires local weak multiplicity detection.
[MA-9] When Does Communication Help? Beyond Spectral Descriptions of Collective Intelligence
【速读】：该论文旨在解决分布式推理中通信机制带来的性能矛盾问题，即通信虽能促进智能体间达成一致，却可能降低决策质量。其核心挑战在于现有对通信增益的聚合描述存在两个根本性局限：其一，对于具有固定证据、网络结构与读出规则的稳定线性系统，尽管交互作用与有限时间状态算子的特征值和奇异值谱完全相同，但通信增益可呈现相反符号；仅通过调整消息传递方向即可使准确率从72.6%提升至91.2%，或降至65.9%。这表明传统基于谱分析的总结方法丢失了关键的方向性信息。为此，研究提出一种任务投影的局部响应近似（task-projected local-response approximation），能够保留该方向性信息，并在独立于测试集的标注校准数据支持下，对小型训练过的非线性智能体在多轮通信中的增益进行预测，其均方根误差仅为0.45个百分点，在两个合成任务上表现良好；并通过自然边变化及手写数字识别任务扩展验证了有效性。其二，当存在社区共享偏见时，个体平均准确率提高可能伴随对未受影响社区的损害或全局投票准确率下降。虽然在固定通信轮次下，通过校准约束可在保留大部分个体收益的同时减少社区损害，但无法保证绝对保护；全直接校准方法效果相当。总体而言，该研究揭示了谱分析在表征通信增益上的不足，强调了任务感知预测与通信收益分布的重要性，同时指出通用迁移能力与实际优越性仍有待进一步探索。
链接: https://arxiv.org/abs/2609.23310
作者: Xuening Wu
机构: Pfizer(辉瑞); Shanghai(上海)
类目: Multiagent Systems (cs.MA)
备注:
Abstract:Communication can bring agents into agreement while making their decisions worse. We identify two limits of aggregate descriptions of communication gain in distributed inference. First, stable linear systems with fixed evidence, network and readout can have interaction and finite-time state operators with identical eigenvalue and singular-value spectra, yet produce gains of opposite sign. Changing only message orientation raises accuracy from 72.6% to 91.2% or lowers it to 65.9%. A standard task-projected local-response approximation retains the directional information missing from spectral summaries. Using labeled calibration data separate from the test set, it predicts multi-round gains in small trained nonlinear agents with a root-mean-square error of 0.45 percentage points on two synthetic tasks; tests with natural edge changes and handwritten digits extend the evaluation. Second, under community-shared bias, higher mean individual accuracy can coexist with harm to unaffected communities or lower global-vote accuracy. At fixed communication rounds, calibration constraints reduce observed community harm while retaining much of the mean benefit, but do not guarantee protection. Full direct calibration performs similarly. The results connect spectral insufficiency, task-aware prediction and the distribution of communication benefits, while leaving broad transfer and practical superiority open.
[MA-10] Latent Telepathy: Multi-Robot Communication with Self-Supervised Perceptual Latents ICRA2027
【速读】：该论文旨在解决去中心化多机器人团队在部分可观测环境下，因信息不对称导致协作效率低下的问题。具体而言，在此类场景中，决定机器人下一步行动的关键感知信息往往仅由个别成员可见，而传统去中心化方法仅传递位置或规划轨迹等运动学信息，无法有效传达队友所感知的环境状态。现有基于多智能体强化学习（MARL）的通信机制虽能承载感知内容，但其生成的消息通常与任务强耦合且语义不透明，难以泛化与理解。为应对上述挑战，本文提出“潜意识传讯”（Latent Telepathy）框架，其核心在于：每个机器人将其自身用于感知的编码器输出的潜变量（perceptual latent vector）广播给队友，该编码器通过自监督联合嵌入预测目标训练并冻结，确保全队共享一致的语义表示；接收方仅通过任务奖励信号学习如何利用该潜变量进行决策，无需理解其具体含义。由于编码器已用于本地感知，广播潜变量不增加额外计算开销，且仅需极小带宽。实验表明，在固定通信带宽、延迟、拓扑和接收端条件下，仅改变消息内容，潜变量广播使导航者在99.7%的环境中成功规避遮挡障碍物，性能接近无噪声人工设计消息；相比之下，位置与轨迹消息表现仅达随机水平，而原始图像输入虽信息量更大（186倍宽），却不如压缩后的潜变量可靠。该方法在离散网格世界与连续速度控制下的渲染像素环境中均表现稳健，并在真实物理机器人摄像头数据上实现了102次连续决策中的全部成功解码。此外，研究还揭示了将MARL通信成果迁移至连续控制时的关键要求：消息所指导的决策必须仍可通过探索手段达到，否则会丧失有效性，文章进一步提出了恢复该可探索性的方法。
链接: https://arxiv.org/abs/2609.23269
作者: Howard Wang,Han Zheng,Cathy Wu
机构: Columbia University (哥伦比亚大学); Laboratory for Information and Decision Systems, Massachusetts Institute of Technology (麻省理工学院信息与决策系统实验室)
类目: Robotics (cs.RO); Machine Learning (cs.LG); Multiagent Systems (cs.MA)
备注: 8 pages, 6 figures. Submitted to IEEE ICRA 2027
Abstract:In a decentralized multi-robot team under partial observability, the fact that decides a robot’s next action is often visible only to a teammate. Existing decentralized methods communicate kinematic information, such as position or planned trajectory, which cannot convey what the teammate perceives. Learned communication in multi-agent reinforcement learning (MARL) can carry perceptual content, but the resulting messages are task-coupled and opaque. We propose Latent Telepathy. Each robot broadcasts the perceptual latent vector it already computes for its own use, the output of an encoder trained with a self-supervised joint-embedding predictive objective, frozen, and shared across the team. A teammate learns to act on it from task reward alone. Because the encoder already runs for perception, the message costs no additional computation and a single compact vector of bandwidth. Because the encoder is frozen before any policy is trained, the message means the same thing to every robot, and the receiving robot is never told what it means. We evaluate Latent Telepathy with a content-controlled protocol in which bandwidth, latency, topology and receiver are held fixed and only the message content varies. Broadcasting the latent lets a navigator avoid an occluded hazard in 99.7% of episodes, matching a noiseless hand-designed message. Position and trajectory messages remain at chance, and the raw camera image, 186 times wider, is less reliable than the compressed latent. The result holds from a discrete gridworld to rendered pixels under continuous velocity control, and the encoder decodes the hazard from a physical robot’s camera in 102 of 102 live decisions. We also identify a requirement for porting MARL communication results to continuous control, that the decision a message informs must remain reachable by exploration, and show how to restore it.
[MA-11] RACS: A Geometry-Aware Framework for Scalable Multi-Agent Path Finding in Warehouses
【速读】：该论文旨在解决大规模仓库自动化中多智能体路径规划（Multi Agent Path Finding, MAPF）的可扩展性与效率问题，现有算法通常通过复杂冲突消解机制提升性能，但将仓库视为通用导航图，忽略了其固有的几何布局与交通模式。为克服这一局限，论文提出TRACS（Traffic aware Routing and Aisle Coordination System），其核心在于利用仓库的几何结构设计一种面向交通感知的规划框架：通过构建具有交替单向通道的有向路由图，从拓扑上消除对向相遇与边交换冲突，从而实现空间路由与时间调度的解耦。该方案采用独立的混合图-网格路由结合轻量级基于边的调度策略，避免了联合时空搜索，确保无碰撞执行。实验表明，TRACS在多个仓库基准测试中均达到100%成功率，规划效率相比主流优先级、迭代修复及搜索型MAPF算法提升显著，在1000个机器人场景下规划时间最高降低14.7倍，同时保持合理的完成时间（makespan）、更低的流程时间（flowtime）和接近最优的路径质量；在固定10分钟规划预算下，可成功调度高达5120个机器人，约为最强基线的两倍，充分验证了利用仓库几何结构实现可扩展机器人系统的核心优势。
链接: https://arxiv.org/abs/2609.23137
作者: Siddhant Erande,Anuj Tiwari
机构: Indian Institute of Technology Madras (印度理工学院马德拉斯分校)
类目: ystems and Control (eess.SY); Multiagent Systems (cs.MA)
备注: 9 pages, 6 figures, under review
Abstract:Large scale warehouse automation relies on efficient multi agent path finding (MAPF) to coordinate thousands of robots in structured environments. Existing MAPF algorithms primarily improve conflict resolution while representing warehouses as generic navigation graphs, overlooking their inherent geometric structure and traffic patterns. This paper presents TRACS (Traffic aware Routing and Aisle Coordination System), a geometry aware planning framework that exploits warehouse layout to simplify planning rather than introducing another conflict-resolution algorithm. TRACS constructs a directed routing graph with alternating one way aisles that eliminates head on and edge swap conflicts by design, decoupling spatial routing from temporal traffic coordination. Independent hybrid graph grid routing is combined with lightweight edge based scheduling to avoid joint space time search while ensuring collision free execution. Experimental evaluation on warehouse benchmarks against representative priority based, iterative repair, and search based MAPF planners shows that TRACS consistently achieves a 100% empirical success rate while substantially improving planning scalability. On fixed scene benchmarks with up to 1000 robots, TRACS reduces planning time by up to 14.7X while maintaining competitive makespan, lower flowtime, and near optimal path quality. Under a fixed 10 minute planning budget, TRACS routes up to 5120 robots, roughly twice the largest fleet reached by the strongest baselines, while sustaining a 100% success rate, demonstrating the effectiveness of exploiting warehouse geometry for scalable robotic warehouse systems.
[MA-12] Adaptive Scaffolding Needs Contingency: An AI Tutor That Escalates and Fades on What the Learner Does
【速读】：该论文旨在解决编码辅助工具在提升任务表现的同时，导致学习者元认知规划与监控行为减少的问题。传统解决方案“减少信息输出”（giving less away）混淆了两个关键概念：系统承担的工作量（认知负荷，cognitive load）与学习者在获得帮助前必须做出决策的元认知需求（metacognitive demand）。本文提出“保持元认知需求恒定”（preserved metacognitive demand）的原则，即固定元认知需求，允许认知负荷动态变化。其核心解决方案CoMeT（Cognitive Load Adaptive Tutoring）通过在学习者决策点失败时增强支持、在采纳帮助后逐步降低支持强度来实现该原则。实验采用被试内设计，131名成年学习者在三个Python任务中分别使用CoMeT、无限制助手和仅提问式导师。结果表明，CoMeT在维持与仅提问式导师相当的元认知需求的同时，生成的代码成果数量是无限制助手的两倍，且比仅提问式导师更少引发学习者挫败感，同时保持了相同的任务委托程度与认知负荷。值得注意的是，学习者常未主动回应；当其回应涉及当前受支持的决策时，支持仍可有效衰减；在十六次会话中，CoMeT仅一次完全交出答案，而仅提问式导师则需六次才如此，显示出更强的渐进式支持控制能力。
链接: https://arxiv.org/abs/2609.22993
作者: Xinmeng Hou,Yuxuan Weng,Chin Hsien Yeh,Ding Lin Lee,Lishan Zheng,Fang Li,Wuqi Wang,Yang Liu
机构: National Institute of Education, Nanyang Technological University, Singapore; School of Computer Science and Engineering, Nanyang Technological University, Singapore; School of Art, Design and Media, Nanyang Technological University, Singapore; College of Computing and Data Science, Nanyang Technological University, Singapore; School of Information Engineering, Chang’an University, Xi’an, China
类目: Human-Computer Interaction (cs.HC); Multiagent Systems (cs.MA)
备注:
Abstract:Coding assistants raise task performance, but learners plan and monitor less. Giving less away, the usual fix, conflates two things: how much work a system carries (cognitive load) and what the learner must decide before help arrives (metacognitive demand). Our principle, preserved metacognitive demand, holds the second constant and lets the first vary. CoMeT implements it: support rises when a learner fails at a decision point and fades on take-up. Within subjects, 131 adult learners used CoMeT, an unrestricted assistant and a question-only tutor on three Python tasks. CoMeT matched the question-only tutor’s demand, delivered artifacts twice as often as the assistant, and frustrated learners less than the question-only tutor, with delegation and load unchanged. Learners often did not answer. Fading held when their turn addressed the decision under support, and CoMeT surrendered the full answer in one session in sixteen, against one in six for the question-only tutor.
[MA-13] Agent Router: Heterogeneous Model Routing for Cost-Optimal Multi-Step Agent ic Workflows ICML2026
【速读】：该论文旨在解决企业级生成式智能体系统（Enterprise agentic systems）在推理过程中因盲目将所有任务步骤均交由高性能前沿模型（frontier model）处理而导致的资源浪费问题。现有路由策略仅优化单轮查询分配，忽略了智能体工作流中一个关键特性：同一任务轨迹内各子任务的复杂度存在显著差异——例如规划步骤需顶级推理能力，而格式化等简单步骤则可由小型模型胜任。为此，论文提出将步骤级模型路由建模为沿智能体轨迹的序列化分配问题，并设计轻量级分类器AgentRouter，通过5个可在路由时实时提取的特征，将每个步骤映射至四个模型层级之一。该模型仅1200万参数，每步延迟仅为5毫秒（A100 GPU），在涵盖规划、编程、研究与数据分析等任务的5万条标注轨迹数据上训练。实验表明，AgentRouter相较仅使用前沿模型的基线实现72%的成本降低，同时保持97.3%的基准性能（端到端任务完成率下降不足3%），且在低复杂度步骤上达到91%的路由准确率，在高效层级达85%，中等与前沿层级亦维持76%-82%的准确率。相比之下，基于单轮训练信号的RouteLLM和FrugalGPT仅分别实现31%与44%的成本降低，因其未能捕捉轨迹级质量依赖关系。因此，该方案的核心在于通过引入轨迹上下文感知的序列化路由机制，实现对复杂度动态变化的精准建模与资源高效分配。
链接: https://arxiv.org/abs/2609.22951
作者: Rudrendu Kumar Paul,Sourav Nandy
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG); Multiagent Systems (cs.MA)
备注: Accepted at the AgenticUQ Workshop, ICML 2026. Camera-ready version
Abstract:Enterprise agentic systems that route every trajectory step to a frontier model waste 60-80% of their inference budget on subtasks that smaller models handle equally well. Existing routing solutions optimize single-turn query assignment but ignore a property unique to agentic workflows: subtask complexity varies widely within a single trajectory. A planning step may require frontier-class reasoning while a subsequent formatting step needs only a 7B model. We formalize step-level model routing as a sequential assignment problem over agent trajectories and propose AgentRouter, a lightweight classifier (12M parameters, 5ms overhead per step on an A100 GPU) that maps each trajectory step to one of four model tiers using five features extractable at routing time. Trained on 50,000 annotated agent trajectory steps spanning planning, coding, research, and data analysis tasks, AgentRouter achieves 72% cost reduction relative to frontier-only baselines, retaining 97.3% of frontier-only quality (less than 3% degradation in end-to-end task completion); per-step routing accuracy reaches 91% on minimal-complexity steps and 85% on efficient-tier steps, with 76-82% on the harder mid-range and frontier tiers. On the same benchmarks, RouteLLM and FrugalGPT (applied per-step) achieve only 31% and 44% cost reduction respectively, because their single-turn training signal misses trajectory-level quality dependencies.
[MA-14] Beyond Single-Model Injection: A Threat Model and Defense Architecture for Prompt Injection in Multi-Agent Systems ICML2026
【速读】：该论文旨在解决多智能体系统中因提示注入（prompt injection）攻击导致的安全威胁问题，尤其针对传统单模型聊天机器人场景下未充分涵盖的复杂攻击面。现有研究主要关注单一大语言模型（LLM）环境中的直接提示注入，而多智能体系统通过三种独特机制放大了此类威胁：跨智能体的消息传递形成了绕过外围防御的隐蔽注入通道、共享工具访问权限导致权限越界跨越智能体边界、信任传播使得被攻陷的智能体可影响上游协调器。为此，作者构建了一个包含14种攻击向量的威胁模型，分为四类：用户输入直接注入（3个向量）、工具输出间接注入（4个向量）、跨智能体消息传递注入（4个向量）以及通过协调器操控引发的级联注入（3个向量）。在对一个6智能体生产级代表性系统的测试中发现，即使部署系统级提示防护策略，仍有67%的智能体存在至少一种作用域违规风险，且通过工具输出实现的间接注入成功率达43%。为应对上述挑战，论文提出四种关键架构级防御措施：基于溯源追踪的消息签名机制（使跨智能体注入成功率下降91%）、在智能体边界实施输入/输出净化处理（间接注入成功率降低78%）、按角色分配权限范围的工具访问控制（彻底消除权限越界行为）、以及对跨智能体通信模式进行异常检测（捕获84%的级联注入尝试），最终将整体注入成功率从31.2%降至4.2%，验证了其有效性与实用性。
链接: https://arxiv.org/abs/2609.22949
作者: Rudrendu Kumar Paul,Sourav Nandy
机构: Boston University (波士顿大学); University of Texas at Austin (德克萨斯大学奥斯汀分校)
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Multiagent Systems (cs.MA)
备注: Accepted at the AIWILD Workshop, ICML 2026. Camera-ready version
Abstract:Existing prompt injection research focuses on single-model chatbot scenarios, where an attacker manipulates one LLM through crafted input. Multi-agent systems amplify this threat through three mechanisms absent from single-model settings: inter-agent message passing creates injection channels invisible to perimeter defenses, shared tool access enables privilege escalation across agent boundaries, and trust propagation allows a compromised agent to influence upstream orchestrators. We construct a threat model enumerating 14 attack vectors across four categories: direct injection via user input (3 vectors), indirect injection via tool outputs (4 vectors), inter-agent injection via message passing (4 vectors), and cascading injection through orchestrator manipulation (3 vectors). Testing all 14 vectors against a 6-agent production-representative system, we find that 67% of agents are vulnerable to at least one scope violation even with system-prompt-level guardrails, and indirect injection via tool outputs succeeds in 43% of attempts. Four architectural defenses reduce overall injection success from 31.2% to 4.2%: message signing with provenance tracking (inter-agent injection down 91%), input/output sanitization at agent boundaries (indirect injection down 78%), privilege-scoped tool access per agent role (privilege escalation eliminated entirely), and anomaly detection on inter-agent communication patterns (84% of cascading attempts caught).
[MA-15] NostrAg ent: A Decentralized Identity and Delegation Architecture for Sovereign Agent ic Systems
【速读】：该论文旨在解决自主智能代理（Autonomous AI agents）在跨组织边界执行任务时所面临的信任与安全挑战，具体包括：身份持久性、权限范围内的委托、对等信任建立、服务发现以及支付机制的协同问题。现有方案多依赖中心化权威机构，或仅覆盖部分能力，导致在关键操作（如密钥轮换或委托撤销）时，信任链与支付系统出现断裂，影响代理自治的连续性。其解决方案的关键在于提出NostrAgent这一去中心化架构，通过在Nostr协议基础上引入三种自定义事件类型实现统一管理：Kind 38100用于基于BIP340 Schnorr签名并包含预轮换承诺的身份声明，确保身份主权且无需注册机构；Kind 38101构建可验证缩窄权限范围的委托链，支持线性扩展的细粒度授权；Kind 38102通过形成抗Sybil攻击的信任图谱，实现去中心化的对等背书；同时结合Lightning HTTP 402（L402）将支付绑定至代理身份，实现可信结算。该架构使所有授权决策均可离线重放验证，且中继节点仅为传输层，不构成信任根。评估结果表明，系统具备亚毫秒级离线验证能力、线性可扩展的委托链处理性能，并在regtest网络上实现157毫秒中位数的L402支付结算，19种故障模式中有17种经实证通过，1种由理论分析约束，1种被识别为架构局限，验证了其作为无中心信任根的可审计、可信赖智能体系统基础组件的有效性。
链接: https://arxiv.org/abs/2609.22944
作者: Oliver Aleksander Larsen,Mahyar Tourchi Moghaddam
机构: University of Southern Denmark(南丹麦大学)
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: 22 pages, 3 figures, 8 tables. Accepted as a peer-reviewed Full Paper at the AIPAA special session of the 2nd International Conference on Agentic and Generative Techniques in Intelligent Computational Systems (AGENTICS 2026), Angers, France, 28-30 October 2026, part of IJCCI 2026, and for publication in the Springer CCIS proceedings. This is the author’s accepted manuscript
Abstract:Autonomous AI agents increasingly act across organizational boundaries on behalf of human operators: they invoke third-party services, delegate subtasks to other agents, and pay for metered resources. Deploying such agents safely requires five capabilities that today live in separate systems: persistent identity, scoped delegation, peer trust, discovery, and payment. Existing approaches root these in centralized authorities or cover only subsets, so authority, trust, and payment fracture exactly where autonomy needs continuity: when a key rotates or a delegation must be revoked. We present NostrAgent, a decentralized architecture that unifies all five over Nostr relays using three custom event kinds: Kind 38100 identity declarations authenticated by BIP340 Schnorr signatures with pre-rotation commitments, Kind 38101 scoped delegation chains whose every hop verifiably narrows granted capabilities, and Kind 38102 peer attestations forming a Sybil-deterrent trust graph, with Lightning HTTP 402 (L402) binding payment to agent identity. Identity remains operator-sovereign without any registration authority; relays are substitutable transport rather than a trust root; and every authorization decision is replayable offline from signed events. We evaluate a Python prototype with a mixed-method design: ATAM quality analysis with a two-round mini-Delphi panel, STRIDE threat modeling across three trust boundaries, eleven benchmarks with non-parametric statistics, and 19 failure modes. Results show sub-millisecond offline verification, linear delegation-chain scaling, and Lightning-settled L402 at 157 ms median on regtest. 17 of 19 failure modes pass empirically, one is bounded analytically, and one is disclosed as an architectural limitation. NostrAgent demonstrates an auditable prototype substrate for trustworthy agentic systems without centralized trust roots.
[MA-16] Personalized Federated Reinforcement Learning via Model-Agnostic Meta-Learning: Convergence of Exact and Hessian-Free Meta-Policy Gradients
【速读】：该论文旨在解决个性化联邦强化学习（personalized federated reinforcement learning）中的协同优化问题，即多个代理在各自独立的马尔可夫决策过程（Markov Decision Process, MDP）中通过服务器协作，共同学习一个适用于个体代理的初始策略，该策略在经过一次本地策略梯度更新后即可实现高效适应。其核心挑战在于如何在异构环境下实现高效的个性化策略共享与快速适应。解决方案的关键是提出Per-FedAvg-PG算法，该算法允许代理在通信轮次之间执行τ次局部随机元策略梯度步（meta-policy-gradient steps），并在理论上证明：当τ = Θ(ε⁻¹/²)时，仅需K = O(ε⁻³/²)轮通信即可达到个性化目标函数的ε-近似一阶平稳点。该分析依赖于强化学习设置的一个结构性特征——在标准策略类正则条件下，各代理的目标函数具有统一有界的梯度和海森矩阵（Hessian），因此无需额外假设数据异质性，传统监督学习中的有界梯度与异质性条件自动满足。然而，精确元梯度计算需要内层循环的策略海森矩阵，实验表明这是实际性能瓶颈。为此，论文进一步分析了无海森变体（Hessian-free variant），界定了其偏差，并揭示了存在非零且阶为α的固定点，说明该方法存在的平稳性下限是算法本身的固有属性而非理论界限所致。实验在表格型和神经导航任务上验证了预测行为，显示相较于独立训练，迁移至未见代理的样本成本降低了一个数量级。综上，该研究明确将适应步长视为可调节的个性化控制参数，而曲率估计精度则决定了是否能负担精确元梯度计算，从而为个性化联邦强化学习提供了理论与实践的双重指导。
链接: https://arxiv.org/abs/2609.22833
作者: Ali Beikmohammadi,Sarit Khirirat,Sindri Magnússon
机构: Stockholm University (斯德哥尔摩大学); Mahidol University (玛希隆大学)
类目: Machine Learning (cs.LG); Multiagent Systems (cs.MA)
备注: 44 pages, 8 figures, Code: this https URL
Abstract:We study personalized federated reinforcement learning, in which n agents, each acting in its own Markov decision process, collaborate through a server to learn a shared MAML-style policy initialization that becomes effective for an individual agent once that agent adapts it with a single local policy-gradient step. We propose Per-FedAvg-PG, in which agents take \tau local stochastic meta-policy-gradient steps between communication rounds, and prove that it reaches an \varepsilon -approximate first-order stationary point of the personalized objective in K=\mathcal O(\varepsilon^-3/2) rounds with \tau=\Theta(\varepsilon^-1/2) local steps. The analysis rests on a structural feature of the reinforcement learning setting: under standard policy-class regularity, the per-agent objectives have uniformly bounded gradients and Hessians with explicit constants, so the bounded-gradient and bounded-heterogeneity conditions imposed by the supervised theory hold automatically and no separate heterogeneity assumption is needed. The exact meta-gradient requires the inner-loop policy Hessian, which our experiments identify as the practical bottleneck. We therefore analyze the Hessian-free variant, bound its bias, and exhibit fixed points at which the meta-gradient is nonzero and of order \alpha , showing that the resulting stationarity floor is a property of the method rather than of the bound. Experiments on tabular and neural navigation confirm the predicted behavior and show transfer to unseen agents at an order of magnitude lower sample cost than independent training. Together these results identify the adaptation step size as a tunable personalization knob and the curvature estimate as the quantity that governs whether exact meta-gradients are affordable.
[MA-17] Decentralized Multi-Robot Exploration with Probabilistic Peer Intent and Multi-hop Plan Propagation
【速读】：该论文旨在解决在通信受限条件下，去中心化多机器人探索中高效协调的难题。现有基于蒙特卡洛树搜索（Monte Carlo Tree Search, MCTS）的方法如去中心化蒙特卡洛探索（Decentralized Monte Carlo Exploration, DMCE）虽能通过共享计划航点序列实现去中心化规划，但其依赖直接通信范围内的信息交换，限制了信息传播效率与协调能力。本文提出概率性同伴意图（Probabilistic Peer Intent, PPI），将同伴轨迹转化为连续的空间意图分布表示，并将其融入本地MCTS的动作评估过程，从而更精准地预测和响应其他机器人的潜在行为。此外，研究还探索了通过多跳（multi-hop）传播机制将同伴意图信息扩展至非直接通信范围的影响。实验结果表明，PPI与多跳传播均能有效提升去中心化探索性能，其相对优势取决于环境结构和团队规模。同时，作者在三台真实机器人上验证了该方法在不同环境类型下的可行性，证明了其实际部署潜力。
链接: https://arxiv.org/abs/2609.22726
作者: Saurbh Singh Jamwal,Nived Chebrolu,Shivaram Kalyanakrishnan
机构: Indian Institute of Technology Bombay (印度理工学院孟买分校)
类目: Robotics (cs.RO); Multiagent Systems (cs.MA)
备注:
Abstract:Efficient coordination under limited communication remains a key challenge in decentralized multi-robot exploration. While centralized approaches benefit from global information sharing, they are often impractical in large-scale or communication-constrained environments. Existing Monte Carlo Tree Search (MCTS)-based approaches, such as Decentralized Monte Carlo Exploration (DMCE), enable decentralized planning by taking peer intent into account. This peer intent is obtained by communicating sequences of planned waypoints with robots within direct communication range. In this work, we extend this idea by introducing Probabilistic Peer Intent (PPI), which converts peer trajectories into a continuous spatial representation of predicted intent and incorporates it into local MCTS action evaluation. We additionally study the effects of sharing peer intent beyond direct communication range by propagating plans over multiple hops. Experiments across multiple simulated environments and team sizes show that PPI and Multi-hop propagation can each improve decentralized exploration, with their relative benefits depending on environment structure and team size. We also demonstrate the real-world deployment of our method on three robots operating in different environment types.
[MA-18] Self-Organizing Agent Teams Learn to Reason Together
【速读】：该论文旨在解决现有AI代理团队在面对未知解决方案结构时，因依赖固定协议、显式任务分解或路由机制而缺乏动态适应能力的问题。其核心挑战在于如何使AI团队在不预先设定角色分工与协作流程的前提下，通过经验学习形成可复用的协作策略，以实现高效的集体智能。解决方案的关键是提出自组织代理团队（Self-Organizing Agent Teams, SAT），即一组固定的AI代理通过先前合作经验学习通用的协作策略，包括角色分配、对话阶段划分、参与模式及信息流动方式，从而支持“协同计算”——代理之间交换、质疑、修正并整合部分推理过程，最终生成单个代理无法独立完成的完整解决方案。实验表明，该方法在多个数学与物理基准上平均达到66.7%的准确率，显著优于最强个体成员（48.8%）、同等算力下的独立推理（58.7%）以及理想路由器方案（59.0%），并在AIME 2026中领先理想路由器13.4分。进一步分析发现，团队协作增益与“可辨识性”（demonstrability，即团队能否识别出正确推理的能力）高度相关（Spearman ρ=0.90, p=0.005），表明当正确推理一旦出现即可被识别时，自组织协作的优势最为显著。研究结果揭示了“组织能力”本身可成为智能体的一项核心能力，即代理团队不仅能独立推理，还能学会如何协同推理，从而突破个体局限，达成超越单个成员能力的集体智能。
链接: https://arxiv.org/abs/2609.22682
作者: Aneesh Pappu,Mirac Suzgun,Yongchan Kwon,Federico Bianchi,Batu El,Mykel J. Kochenderfer,Hancheng Cao,James Zou
机构: Stanford University (斯坦福大学); Together AI; Goizueta Business School, Emory University (埃默里大学古德齐奥塔商学院)
类目: Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: Preprint
Abstract:Collective intelligence depends not only on what team members know, but also on how they organize their work. When the structure of a solution is unknown, useful roles and divisions of labor cannot be specified in advance; teams must learn from experience how to organize reasoning as it unfolds. Human teams routinely adapt this way, while existing AI agent teams rely on fixed protocols, explicit task decomposition, or routing. We introduce Self-Organizing Agent Teams (SAT), fixed teams of AI agents that learn reusable strategies from prior collaborations to organize roles, conversational phases, participation, and information flow. These strategies enable what we call collaborative computation: agents exchange, challenge, repair, and synthesize partial reasoning into solutions no member produced independently. In two independent settings, we learn teamwork strategies that transfer unchanged to unseen benchmarks, using only 15 mathematics and 25 graduate-level knowledge problems. Across five mathematics and physics benchmarks, self-organizing teams average 66.7% accuracy, versus 48.8% for their strongest member, 58.7% for compute-matched inference by that agent, and 59.0% for a perfect router over members’ independent answers; on AIME 2026, they exceed this router by 13.4 points. Because gains vary across benchmarks, we ask when self-organizing collaboration helps. Across eight benchmarks, demonstrability (the organizational-psychology construct of whether a team can distinguish correct from incorrect reasoning) strongly tracks improvement over the strongest member (Spearman \rho=0.90 , p=0.005 ): teams benefit most when correct reasoning can be recognized once it appears. More broadly, these results suggest that organization itself can become an agent capability: agent teams can learn how to reason together and produce solutions their members could not reach independently.
[MA-19] Density-Driven Area Coverag e for Nonholonomic Multi-Robot Systems with Safety Guarantee
【速读】：该论文旨在解决在非完整约束（nonholonomic）多机器人系统中，基于密度驱动最优控制（D2OC）方法时，参考运动的安全性约束与实际物理输入之间存在的不匹配问题。传统方法仅保证参考轨迹的安全性，但无法确保实际机器人运动在物理层面满足安全间距要求，从而导致潜在碰撞风险。其解决方案的关键在于：通过引入控制屏障函数（Control Barrier Function, CBF）安全滤波器，并结合反馈线性化的前瞻点（look-ahead point）设计，将安全约束直接施加于机器人的物理控制输入上，同时保持密度驱动的覆盖目标。此外，研究进一步推导了考虑前瞻几何结构、机器人轮廓及控制周期内运动特性的安全裕度，实现了对物理输入与执行器限制的联合考量。仿真与实验结果表明，所提方法在保障物理分离安全性的前提下，仍能实现与传统参考跟踪方法相当的覆盖率性能，有效消除了安全认证与实际运动之间的脱节，验证了在非完整机器人系统中直接对物理输入施加安全约束的有效性。
链接: https://arxiv.org/abs/2609.22668
作者: Julian Martinez,Kooktae Lee
机构: Texas Tech University(德克萨斯理工大学)
类目: Robotics (cs.RO); Multiagent Systems (cs.MA); Optimization and Control (math.OC)
备注:
Abstract:Density-Driven Optimal Control (D2OC) provides a principled approach to distributing multi-robot teams over non-uniform spatial distributions. Applying D2OC to nonholonomic robots, however, creates a gap between safety constraints imposed on a reference motion and the physical inputs that determine the actual robot motion. We address this issue by enforcing the safety constraint directly on the robot’s physical inputs while preserving the density-driven coverage objective. The proposed framework combines D2OC with a control barrier function safety filter through a feedback-linearizing look-ahead point, allowing safety and actuator limits to be considered together during control. We further derive a safety margin that accounts for the look-ahead geometry, robot footprint, and motion during each control interval. Simulation results show that the proposed method maintains the required physical separation while achieving coverage performance comparable to a conventional reference-tracking approach, which can satisfy safety on the reference motion yet violate the corresponding physical clearance. Experiments on multiple nonholonomic robots in the Robotarium further demonstrate safe execution while driving the robots toward the desired spatial distribution. These results show that enforcing safety directly on the physical inputs can eliminate the mismatch between safety certification and physical robot motion in density-driven multi-robot coverage.
[MA-20] From Certain Doom to Survival: Agent -Driven Self-Governance in LLM Agent -Driven Self-Governance in LLM Agent Societies
【速读】：该论文旨在解决多智能体大语言模型（Multi-agent LLM systems）在社会困境中缺乏自主治理能力的问题，尤其针对现有研究将治理机制视为外部强加、仅以修辞方式表达或局限于固定选项集的局限性。其核心解决方案是提出GovSim-SelfGovern，一个扩展自公共池资源（Common-Pool Resource, CPR）环境的新型评估框架，允许智能体自主编写可执行的Python治理规则，经沙盒验证反馈后通过投票表决形成法律，并在多轮迭代中实际遵守自身制定的制度。该系统通过引入可执行代码行动（executable code actions），使智能体能够在资源压力下动态设计与调试制度，从而探索制度创设的时间敏感性与有效性。研究发现，虽然可执行治理显著拓展了干预策略空间，但生存关键在于智能体能否在资源枯竭前及时发现并实施有效的制度机制；财政能力支持再分配，而更深层次的推理能力与去除民主否决权则提升了驱逐（exile）等极端措施的可行性。进一步揭示了智能体对驱逐提议的回避行为，可能源于规范性伦理考量，而非策略候选集缺失，表明稀缺性将制度创设转化为兼具政治与伦理维度的复杂问题。
链接: https://arxiv.org/abs/2609.22600
作者: Gregory B. Rehm
机构: Meta
类目: Multiagent Systems (cs.MA)
备注:
Abstract:Multi-agent LLM systems are increasingly evaluated in social dilemmas, but most work treats governance as imposed by the experimenter, expressed rhetorically, or restricted to a fixed menu of mechanisms. We introduce GovSim-SelfGovern, an extension of the GovSim common-pool resource environment in which agents author executable Python governance rules, receive sandbox validation feedback, vote on proposed laws, and live under the rules they enact across rounds. To evaluate agent-driven self-governance, we examine three scenarios ranging from stable abundance to a fatal resource wall where five agents cannot all survive through harvest alone. To solve this, agents must write and debug useful laws in time before their institutions degrade sharply under resource pressure. Finally, we study a central alignment question: when agents hesitate to propose exile, are they rejecting it for normative reasons, or does it never enter their candidate set? Our results show that executable governance improves the space of possible interventions for agents, but survival depends on whether agents discover the right institutional mechanisms in time. Fiscal capacity enables redistribution, while deeper reasoning and removal of democratic veto make exile more feasible. GovSim-SelfGovern therefore adapts executable code actions to a common-pool governance setting and shows how scarcity turns institutional authorship into a political and ethical problem.
[MA-21] AutoGym: Blueprint-First Generation of Verifiable Agent Gyms NEURIPS2026
【速读】：该论文旨在解决强化学习中智能体训练所依赖的“训练环境集（gym）”构建过程高度依赖人工、成本高昂且静态不变的问题。随着模型能力提升，传统任务集易出现饱和与污染，而现有合成方法生成的任务难度仅停留在表面层次，无法有效区分不同能力水平的模型，且依赖不可靠的大语言模型（LLM）进行事后判别。其解决方案的关键在于提出AutoGym框架，通过三个核心机制实现动态、可扩展且具备精确难度调控能力的完整训练环境集自动生成：（1）采用“蓝图先行”（blueprint-first）生成策略，在环境实例化前即定义合法解空间、环境约束与验证标准，使可解性成为构造的先决条件而非事后验证属性；（2）引入显式生成参数，对任务拓扑、交互深度、能力维度、问题混淆程度及干扰项构成等进行精细控制，实现对任务难度的细粒度调节；（3）基于性能反馈的主动课程合成（active curriculum synthesis），随模型能力演化动态调整生成参数分布，确保任务难度持续匹配当前模型水平。实验表明，AutoGym可在生产力与时间推理等多种场景下生成覆盖全能力谱的任务集，包括对前沿模型构成挑战的高难度实例。
链接: https://arxiv.org/abs/2609.22592
作者: Aarati Andrea Noronha,Kavya Ravikumar,Carly Xiaoyu Lin
机构: Amazon AGI
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Multiagent Systems (cs.MA)
备注: Submitted to NeurIPS 2026 Workshop: Who Verifies the Agents?
Abstract:Training agents with reinforcement learning requires a gym, comprising a task, an executable environment in which the task can be attempted, and a verifier that reliably distinguishes success from failure. Constructing such gyms remains manual, expensive, and static. Task sets saturate as models improve and are increasingly exposed to contamination. Synthetic generation offers scale, but single-pass synthesis produces tasks whose difficulty is largely cosmetic. Models comparable in capability solve them despite convoluted phrasing, and correctness must be adjudicated post-hoc by unreliable LLM judges. We present AutoGym, a framework that generates complete gyms (tasks, executable environments, and verifiers) from a minimal domain seed or prior model trajectories. AutoGym introduces three mechanisms. (1) Blueprint-first generation specifies the valid solution space, environment requirements, and verification criteria before the environment is materialized, making solvability a construction prerequisite rather than a property verified after the fact. (2) Explicit generation parameters control task topology, interaction depth, capability axes, question obfuscation, and distractor composition, enabling fine-grained difficulty steering. (3) Active curriculum synthesis uses performance-informed calibration to adjust the distribution over these parameters as model capabilities evolve. Across productivity and temporal-reasoning settings, AutoGym generates gyms spanning the capability spectrum, including instances that challenge frontier models.
[MA-22] oward Mission-Critical ISAC: Reliable Energy-Aware Coordination in UAV Swarms
【速读】：该论文旨在解决任务关键型应用中无人机蜂群（UAV swarm）在执行移动空中目标跟踪时，因主动集成感知与通信（ISAC）操作带来的推进与传输双重能耗压力，导致电池过早耗尽、威胁任务连续性的核心问题。其解决方案的关键在于提出一种双层无人机蜂群架构：由执行协同ISAC的主任务无人机（MUAV）负责目标跟踪，同时通过配备太阳能收集面板的专用充电无人机（CUAV）利用空对空无线能量传输（WPT）为低电量MUAV提供能量补给。为实现高效协同优化，论文构建了一个联合最小化协同后验Cramér-Rao界（PCRB）的非凸混合整数规划（MIP）模型，首次将协同ISAC感知质量、空中WPT调度与双层能量管理进行耦合建模。为求解该复杂问题，提出递推时域交替优化（RHAO）算法，通过每时隙四阶段分解策略——基于匈牙利算法的充电准入决策、基于逐次凸逼近（SCA）的MUAV轨迹与感知-通信时隙分配、CUAV会合轨迹规划及WPT功率分配，实现了单调收敛，并显著提升了系统性能。仿真结果表明，RHAO相较固定时隙分割基准和静态轨迹方案，平均PCRB分别降低16.8倍和5.4倍，且空中WPT子系统可全程维持所有MUAV能量高于临界阈值，保障任务持续性。
链接: https://arxiv.org/abs/2609.22292
作者: Masoud Shokrnezhad,Tarik Taleb,Amir Javadpour
机构: ICTFICIAL Oy( ICTFICIAL Oy); Ruhr University Bochum(鲁尔大学波鸿分校)
类目: Networking and Internet Architecture (cs.NI); Information Theory (cs.IT); Multiagent Systems (cs.MA); Robotics (cs.RO); Systems and Control (eess.SY)
备注: Accepted to IEEE ISAC 2026
Abstract:Unmanned aerial vehicle (UAV) swarms deployed in mission-critical applications must simultaneously track a mobile aerial target and maintain reliable data links. However, active integrated sensing and communication (ISAC) operation imposes a dual energy burden on propulsion and transmission, threatening mission continuity through premature battery depletion. In this paper, we propose a two-tier UAV swarm architecture in which mission UAVs (MUAVs) execute cooperative ISAC for mobile aerial target tracking while dedicated charging UAVs (CUAVs), equipped with solar harvesting panels, replenish low-battery MUAVs via aerial UAV-to-UAV wireless power transfer (WPT). We formulate the joint minimization of the cooperative posterior Cramér-Rao bound (PCRB) over MUAV trajectories, per-slot sensing-communication time splits, WPT scheduling and admission, and CUAV rendezvous trajectories, subject to minimum uplink rate, dual-tier energy causality, WPT proximity, collision-avoidance, and speed constraints, yielding a non-convex mixed-integer program (MIP) that, to the best of our knowledge, is the first to jointly couple cooperative ISAC sensing quality with aerial WPT and dual-tier energy management. To solve it efficiently, we propose Receding-Horizon Alternating Optimization (RHAO), a four-block per-slot algorithm that decomposes the problem into: charging admission via the Hungarian algorithm, MUAV trajectory and time-split via successive convex approximation (SCA), CUAV rendezvous, and WPT power allocation, with monotone convergence guarantees. Simulation results demonstrate that RHAO reduces the mean PCRB by 16.8 times over a fixed-time-split baseline and 5.4 times over a static-trajectory scheme, while the aerial WPT subsystem sustains all MUAVs above the energy-critical threshold throughout the full mission horizon.
[MA-23] Predictors and Orchestrators: Parsimonious Machine Learning within an Agent ic AI Harness for Multi-Horizon Karst Aquifer Forecasting
【速读】：该论文旨在解决岩溶含水层动态预测中的关键难题，即由于补给响应具有非线性、事件驱动性以及流动路径高度异质性，导致长期（1-12周）泉流量与地下水位预测极为困难。其解决方案的核心在于提出一种部署感知的预测框架，通过在统一的时间评估设计下对比五类模型（极端梯度提升、极随机树、长短期记忆网络、卷积神经网络和Transformer），发现极端梯度提升（XGBoost）在所有预测时域（1-4周、5-8周、9-12周）均表现最优，决定系数（R²）分别不低于0.97、0.96和0.94，且在前三个干旱等级中超过90%的临界阶段一致性，显著优于深度学习模型。研究进一步揭示，树模型在低维、轴对齐的水文气候预测因子上具有天然优势，而深度模型则因平滑异常极端值而导致长期预测性能退化。最终，该框架被集成至五智能体操作架构中，实现数据获取、模型分配、确定性预测、阈值监控、前瞻性验证、文献检索与报告生成的自动化闭环，形成兼具简约模型选择、泄漏感知的多时域评估、决策相关阈值精度及可审计智能体自动化的可迁移预测体系。
链接: https://arxiv.org/abs/2609.22251
作者: Pramod Lekhak,Chetan Sharma,Hakan Başağaoğlu,F. Paul Bertetti,Debaditya Chakraborty
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注:
Abstract:Forecasting karst aquifer dynamics is difficult because recharge responses are nonlinear, event-driven, and governed by strongly heterogeneous flow paths. This study develops and evaluates a deployment-aware framework for 1-12-week-ahead prediction of spring discharge and groundwater level using approximately 79 years of hydroclimatic observations from the Edwards Aquifer, Texas. Five model families were compared under a common temporal evaluation design: extreme gradient boosting, extremely randomized trees, long short-term memory, convolutional neural networks, and Transformers. Predictions were evaluated using coefficient of determination, Kling-Gupta efficiency, root-mean-square error, and agreement with operational drought thresholds. Extreme gradient boosting was consistently most reliable, with R2 at least 0.97, 0.96, and 0.94 across 1-4-, 5-8-, and 9-12-week horizons, respectively, and greater than 90% critical-stage agreement at the first three drought stages across all horizons. Deep models were competitive at short horizons but degraded progressively and exhibited isolated failures at longer lead times. We attribute this contrast to an alignment between tree partitioning and low-dimensional, axis-aligned hydroclimatic predictors, together with the tendency of neural models to smooth irregular extremes. The validated models were embedded in a five-agent operational architecture that automates data acquisition, model assignment, deterministic prediction, threshold monitoring, prospective verification, literature retrieval, and reporting. The contribution is therefore a transferable framework joining parsimonious model selection, leakage-aware multi-horizon evaluation, decision-relevant threshold skill, and auditable agentic automation.
[MA-24] BizSage: A Self-Evolving Multi-Agent Framework for Business Research with Efficient Knowledge Retrieval EMNLP2026
【速读】：该论文旨在解决基于大语言模型（LLM）的多智能体系统在经济学与商学研究领域应用时面临的两大核心挑战：一是现有方法多以整篇论文为单位进行检索，而研究任务所需证据常分散于不同章节，导致粒度不匹配，影响检索覆盖率与精度；二是经济与商学研究对实证严谨性要求极高，但当前系统缺乏从评估反馈中学习并自我优化的机制。其解决方案的关键在于提出一个名为BizSage的多智能体框架，融合文档级细粒度检索与质量驱动的自演化机制。具体而言，通过整合章节级知识图谱构建横向知识图谱（Lateral Knowledge Graph, LKG），并采用个性化页面排名（Personalized PageRank, PPR）算法识别语义相关且结构重要的章节内容；同时，七个专业化智能体在元评审（Meta-Review）自演化机制下协作，将评估轨迹中的失败模式提炼为可复用的策略，实现持续优化。实验结果表明，BizSage在涵盖四个领域和三项任务的基准测试中表现优异，多数指标排名第一，相较于六种基线模型的成对胜率超过60%，且未产生任何幻觉引用，验证了其在社会科学研究中提供可靠研究辅助的潜力。
链接: https://arxiv.org/abs/2609.22235
作者: Yuhe Wu,Guangyu Wang,Jiaxin Liu,Guang Zhang
机构: The Hong Kong University of Science and Technology (Guangzhou), Guangzhou, China
类目: Computation and Language (cs.CL); Multiagent Systems (cs.MA)
备注: Accepted by EMNLP 2026 findings
Abstract:While multi-agent systems based on large language models (LLMs) have shown promise in automating the progressive workflow of academic research, extending them to economics and business research, where specialized domain knowledge spans neighboring disciplines yet remains difficult to access in a structured way, presents two challenges. First, existing methods mostly retrieve at the paper level, yet the evidence needed for research tasks is often distributed across different sections, creating a granularity mismatch that hinders retrieval coverage and precision. Second, these fields demand strict empirical rigor, yet current systems provide limited mechanisms for learning from evaluation feedback. We present \textbfBizSage, a multi-agent framework combining corpus-level fine-grained retrieval with quality-driven self-evolution. We build a Lateral Knowledge Graph (LKG) by merging section-level knowledge graphs and apply Personalized PageRank (PPR) to surface semantically relevant and structurally important sections. Seven specialized agents collaborate under a Meta-Review self-evolution mechanism that distills failure modes from evaluation traces into reusable strategies. On a benchmark spanning four domains and three tasks, BizSage ranks first on the majority of metrics, achieves pairwise win-rates above 60% against six baselines, and produces zero hallucinated citations. We hope BizSage paves the way for reliable research assistance in economics, business, and the broader social sciences.
[MA-25] A Channel-Boosted Multi-Agent System with Iterative Consultation for Document Sensitivity Classification
【速读】：该论文旨在解决关键国家基础设施领域组织在处理敏感信息时，对异构文档进行敏感性评估的效率与准确性问题。传统人工评估方式存在速度慢、标准不一和难以扩展的缺陷，而现有基于Transformer的单编码器模型虽表现良好，但受限于固定输入长度截断机制，会丢失长文档末尾区域的关键证据——而敏感内容往往集中于这些长段落中。为克服此结构性缺陷，本文提出通道增强型多智能体系统（Channel-Boosted MAS, CB-MAS），并具体实现为迭代式咨询多智能体系统（IC-MAS）。其核心解决方案在于：通过一个通道批判代理（Channel Critic Agent）学习文档自适应的信任权重，动态控制两个首窗口编码器间的门控通道增强（Gated Channel Boosting），同时引入成对咨询代理（Consultation Agents）在紧凑表示空间中迭代交换信念状态，以融合长文档开头与结尾的证据。该方法在保持计算量恒定的前提下，有效缓解了长文本上下文丢失问题。消融实验表明，批判控制的通道增强贡献了主要准确率提升，而咨询机制则显著恢复召回率且未导致精确率下降。最终，结合最大池化融合与黑板自适应咨询策略的IC-MAS，在战略16K语料库上达到90.72%准确率、91.23% F1分数、92.01%敏感项召回率和90.46%敏感项精确率，平均计算量比固定轮次基线减少约54%，性能提升在统计上高度显著（McNemar检验，p < 0.000001；配对t检验）。研究还纳入LIME/SHAP可解释性分析、多智能体评估框架及对局限性的诚实披露。
链接: https://arxiv.org/abs/2609.22212
作者: Aleesha Zainab,Asifullah Khan,Muhammad Ahmed Khalid,Faheem Ullah Khan
机构: Pakistan Institute of Engineering and Applied Sciences (PIEAS)(巴基斯坦工程与应用科学研究所); Pattern Recognition Lab(模式识别实验室); Deep Learning Lab(深度学习实验室); PIEAS Artificial Intelligence Center (PAIC)(PIEAS人工智能中心); Center for Mathematical Sciences(数学科学中心)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG); Multiagent Systems (cs.MA)
备注: 30 pages , 12 figures
Abstract:Organizations in critical national infrastructure sectors must assess heterogeneous documents for sensitivity before routing or storage. Manual assessment is slow, inconsistent, and unscalable. Extending our prior leakage-controlled benchmark, BERT established the top single-encoder baseline (89.14% accuracy, 89.33% F1-score under 5-fold cross-validation on the Strategic 16K corpus). However, transformer baselines suffer from a structural limitation: fixed input length truncation discards evidence beyond the retained window-precisely where sensitive cables tend to be longest. We present Channel-Boosted MAS (CB-MAS) and instantiate it as IC-MAS (Iterative Consultation Multi-Agent System) to solve this without long-context computational costs. A Channel Critic Agent learns document-adaptive trust weights governing Gated Channel Boosting between two first-window encoders, while paired Consultation Agents iteratively exchange belief states to reconcile evidence from the beginning and end of long documents. IC-MAS holds computation constant regardless of document length by reconciling fixed windows in a compact representation space. Ablation studies show critic-controlled Channel Boosting provides the bulk of accuracy gains, while consultation recovers recall without precision collapse. Critic-Controlled Gated Channel Boosting with Max-Pool fusion and Blackboard Adaptive Consultation achieves 90.72% accuracy, 91.23% F1-score, 92.01% sensitive recall, and 90.46% sensitive precision, using about 54% less average computation than a fixed-round baseline. Gains over the single-encoder baseline are statistically significant (McNemar’s test, p less than 0.000001; paired t-test). We include LIME/SHAP explainability, multi-agent evaluation, and an honest accounting of limitations.
[MA-26] A Pinch of SFT A Dash of RL: When Reinforcement Learning Helps Long-Horizon Advertising Agents EMNLP2026
【速读】：该论文旨在解决企业在使用分布式业务数据进行长期任务规划时，如何有效平衡监督微调（Supervised Fine-Tuning, SFT）与强化学习（Reinforcement Learning, RL）在生成式企业分析代理（Enterprise Analytics Agents）中的协同作用问题。具体而言，SFT能够校准工具调用的语法和教师引导的行为，而RL则可探索超出示范行为范围的奖励驱动行为，但若不加区分地统一应用RL，可能导致已校准的SFT技能被破坏。其解决方案的关键在于提出一种基于“教师支持”与“奖励可观测余量”的动态路由机制，根据任务特征自适应地分配训练策略：仅使用SFT、SFT后接RL、增加RL投入或进一步开发环境。通过18项特定功能实验验证，该诊断框架成功预测了15/18的实际训练轨迹。在GPT-OSS 120B模型上，针对性的SFT后接RL策略相较于前沿控制组，在8个广告主相关技能中实现了7个正向收益，其中5个收益在95%置信区间内显著为正，最大增益达非披露指标+11.27分（95% CI [+9.72, +12.82]）。独立的领域专家审计表明，相比仅使用SFT，目标导向的RL将标准泄露率从11.8%降至2.9%，对抗性泄露率从22.9%降至6.8%，同时保持了85.7%的动作可执行性。在共享奖励与优化条件下，针对式RL使七项技能的平均增量从+1.62提升至+3.57，且仅需43%的额外强化学习计算资源。
链接: https://arxiv.org/abs/2609.22194
作者: Aakash Kolekar,Sahika Genc,Bunyamin Sisman,Shahriar Shariat,Shree Vandana Kachroo,Avishek Saha,Qianli Wu,Ari Singer,Benoit Dumoulin
机构: Amazon Advertising(亚马逊广告); Amazon Web Services Agentic AI(亚马逊网络服务代理人工智能)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Multiagent Systems (cs.MA)
备注: Accepted at EMNLP 2026 Industry Track
Abstract:Enterprise analytics agents solve long-horizon tool-use problems over distributed business data, requiring retrieval, reasoning, API calls, code execution, and adaptation to intermediate observations. Supervised fine-tuning (SFT) calibrates tool syntax and teacher-supported behavior, whereas reinforcement learning (RL) can explore reward-supported behaviors beyond demonstrations; applied uniformly, however, RL can perturb already-calibrated skills. We study how to balance SFT and RL under production-mirroring beta APIs. We observe that, in our controlled experiment, checkpoint trajectories retrospectively separated into three regimes: Imitation, where SFT captured reliable teacher behavior; Lift, where both stages helped; and Discovery, where useful reward-observable behavior lay outside reliable teacher support. We leverage this prospectively, using teacher support and reward-observable headroom to route features to SFT only, SFT then RL, increased RL allocation, or further environment development. Across 18 subsequent feature-specific experiments, the diagnostic predicted 15/18 observed trajectories. On GPT-OSS 120B, targeted SFT then RL produced positive point estimates on 7/8 advertiser skills relative to a frontier Control; five positive gains had paired 95% confidence intervals excluding zero, while one skill had a confidence-supported regression. The largest gain was non-disclosure (+11.27 points; 95% CI [+9.72, +12.82]). A separate SME audit surfaced that targeted RL reduces standard leakage from 11.8% to 2.9% and adversarial leakage from 22.9% to 6.8% relative to SFT while preserving actionability (86.2% to 85.7%). In a matched uniform-versus-targeted comparison with shared rewards and optimization, targeted RL improved the seven-skill mean delta from +1.62 to +3.57 while using 43% less incremental RL compute.
[MA-27] Hierarchical Bayesian optimization of an aircraft-based multi-agent system-of-systems
【速读】：该论文旨在解决复杂系统之系统（System-of-Systems, SoS）架构设计中因高保真多物理场与多学科仿真计算成本高昂、评估失败风险大而导致的优化难题。传统基于代理模型（surrogate-based）的方法在处理具有离散架构选择、条件依赖关系及异构设计变量的复杂SoS问题时，往往效率低下且鲁棒性不足。为此，本文提出一种分层贝叶斯优化（hierarchical Bayesian optimization）框架，其核心在于利用高斯过程元模型（Gaussian process meta-modeling）对多层次、结构复杂的决策空间进行建模，有效捕捉架构层级间的依赖关系与异构变量特性。该方法通过分层表示与代理模型的协同优化，显著提升了搜索效率与鲁棒性，在有限仿真预算下实现了对大规模、结构多样化设计空间的高效探索。研究以欧盟资助的COLOSSUS项目中的基于飞机的多智能体野火扑救系统为应用场景，验证了该框架在协调异构空中平台、支持可持续交通与应急响应任务中的可行性与可扩展性，为航空、可持续移动性及韧性导向系统设计提供了可迁移的架构优化方法论。
链接: https://arxiv.org/abs/2609.22130
作者: Paul Saves,Thierry Lefebvre,Nathalie Bartoli,Jasper Bussemaker,Nikolaos Kalliatakis,Nabih Naeem,Prajwal Prakasha
机构: IRIT, UMR 5505 CNRS, Université Toulouse Capitole, Université de Toulouse(法国图卢兹大学); ONERA, DTIS, Université de Toulouse(法国图卢兹大学); Fédération ENAC ISAE-SUPAERO ONERA, Université de Toulouse(法国图卢兹大学); German Aerospace Center (DLR), Institute of System Architectures in Aeronautics(德国航空航天中心)
类目: Machine Learning (cs.LG); Multiagent Systems (cs.MA); Optimization and Control (math.OC)
备注:
Abstract:Developing innovative system architectures increasingly relies on advanced modeling and optimization techniques to frame the architecting process and define the corresponding computational problems. For complex System-of-Systems (SoS), high-fidelity multiphysics and multidisciplinary simulations are essential for capturing detailed behaviors. However, their computational expense and the risk of evaluation failures make direct optimization challenging. To overcome these limitations, surrogate-based approaches, like Bayesian optimization, have emerged as effective tools for managing expensive, black-box simulation tasks. This work introduces a hierarchical Bayesian optimization framework that leverages Gaussian process meta-modeling to handle discrete architectural choices, conditional dependencies, and heterogeneous design variables inherent to SoS problems. Results show that the hierarchical formulation improves search efficiency and robustness compared to conventional surrogate-based methods, enabling the exploration of large and structurally diverse design spaces with limited simulation budgets. We apply the approach to an aircraft-based multi-agent system for wildfire suppression, a use case developed within the EU-funded COLOSSUS project that illustrates how SoS principles can coordinate heterogeneous aerial platforms with complementary roles, supporting both sustainable mobility and emergency response missions. Our framework provides a scalable methodology for SoS architecting and model exploration, offering transferable insights for applications in aviation, sustainable mobility, and resilience-oriented system design. By combining hierarchical representations with surrogate-based optimization, this work is among the first practical demonstrations of hierarchical Bayesian optimization applied to real-world SoS problems, advancing both methodology and practice. Subjects: Machine Learning (cs.LG); Multiagent Systems (cs.MA); Optimization and Control (math.OC) Cite as: arXiv:2609.22130 [cs.LG] (or arXiv:2609.22130v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2609.22130 Focus to learn more arXiv-issued DOI via DataCite Related DOI: https://doi.org/10.21203/rs.3.rs-10496293/v1 Focus to learn more DOI(s) linking to related resources
[MA-28] Beyond the Text: Verifying That Agent -Written Papers Are Backed by Their Artifacts
【速读】：该论文旨在解决生成式AI（Generative AI）在自主科研任务中所生成的研究文档与配套代码库之间的一致性问题，即报告的研究成果是否真正由其对应的实现和实验证据支持。现有评审方法主要依赖文本质量评估，难以识别诸如硬编码指标、未实现的方法或缺乏实证支持的实验结果等关键不一致现象。论文提出ReAgent——一种自动化审计框架，通过从研究文档中构建科学主张的结构化表示，并以此指导对代码仓库的分析与证据收集。其核心解决方案在于结合静态审计与动态审计：静态审计检查所声称的方法、实现及实验配置是否在代码库中一致体现；动态审计则实际执行相关实验并采集运行证据以验证实证结论。二者协同可揭示单一视角下难以发现的矛盾，例如实验结果虽复现了报告数值但偏离了所宣称的方法路径。最终，所有审计证据与决策被组织为结构化的仓库级审计报告，实现证据溯源透明化。在人工标注的基准数据集上的评估表明，ReAgent能有效识别研究结论与其支撑证据之间的不一致性，显著优于现有的静态分析与复现基准方法。
链接: https://arxiv.org/abs/2609.22111
作者: Qiuhong Shen,Benlong Wu,Hanjin Liu,Yuang Qi,Kejiang Chen
机构: 未知
类目: Computation and Language (cs.CL); Multiagent Systems (cs.MA)
备注:
Abstract:Large language model agents are increasingly capable of conducting research autonomously, producing research documents alongside the code and experiments that ostensibly support them. Yet whether the reported findings are consistently supported by corresponding implementations and execution evidence remains largely unexplored: existing review practices primarily assess textual quality and cannot reliably identify inconsistencies such as hard-coded metrics, unimplemented methods, or unsupported experimental results. We present ReAgent, an automated auditing framework for assessing the consistency between agent-generated research documents and their associated repositories. ReAgent constructs structured representations of scientific claims from research documents and uses them to guide repository analysis and evidence collection. Static auditing examines whether claimed methodologies, implementations, and experimental configurations are consistently reflected in the repository, while dynamic auditing executes relevant experiments and collects execution evidence to assess empirical findings. By combining static analysis with dynamic evidence, ReAgent identifies inconsistencies that may remain hidden under either perspective alone, such as experiments that reproduce reported numbers while deviating from the claimed methodology. The collected evidence and audit decisions are organized into a structured repository-level audit report, enabling transparent evidence traceability. We evaluate ReAgent on a manually curated benchmark of agent-generated research document–repository pairs and compare it against representative static and reproduction-based baselines. Experimental results demonstrate that ReAgent effectively identifies inconsistencies between reported research findings and their supporting repository evidence.
[MA-29] Riemannian Density-Driven Optimal Control: Tangent-Space LQR for Second-Order Multi-Agent Systems on Curved Manifolds
【速读】：该论文旨在解决现有密度驱动最优控制（Density-Driven Optimal Control, D2OC）方法在非欧几里得流形（Riemannian manifold）上应用时，未能充分考虑流形固有几何结构的问题。传统D2OC框架主要适用于欧氏空间，难以直接推广至具有复杂曲率特征的流形环境，导致控制精度下降与目标分布偏差。其解决方案的关键在于提出一种基于黎曼几何的二阶多智能体系统最优控制方法——黎曼密度驱动最优控制（Riemannian D2OC, R-D2OC）。该方法通过在每个智能体的切空间中利用对数映射构建局部分布目标，并以加权中心作为有限时域线性二次调节器（LQR）的参考轨迹；控制指令通过流形上的内在二阶动力学与平行传输在滚动时域框架下实现。研究进一步建立了依赖局部曲率的近似误差界，量化了切空间降维带来的偏差，并推导出一个条件离散下降结果，表明当满足局部速度对齐条件时，闭环目标函数可保证单调递减。数值仿真在三维椭球流形上验证了分布级控制的有效性，并支持所提出的近似与收敛性分析。
链接: https://arxiv.org/abs/2609.22678
作者: Kooktae Lee,Ruchika Singh
机构: Texas Tech University (德克萨斯理工大学)
类目: Optimization and Control (math.OC); Multiagent Systems (cs.MA); Robotics (cs.RO); Systems and Control (eess.SY)
备注:
Abstract:Density-Driven Optimal Control (D2OC) provides an effective framework for steering multi-agent systems toward prescribed spatial distributions. However, existing D2OC formulations are primarily developed for Euclidean domains and do not directly account for intrinsic manifold geometry. This paper extends D2OC to second-order multi-agent systems evolving on Riemannian manifolds. The proposed Riemannian D2OC (R-D2OC) constructs a local distribution objective in the tangent space of each agent through logarithmic maps and uses its weighted center as the reference for a finite-horizon LQR. The resulting control is executed on the manifold through intrinsic second-order dynamics and parallel transport within a receding-horizon scheme. We establish local curvature-dependent bounds that quantify the approximation introduced by the tangent-space reduction and characterize the resulting target bias. Furthermore, we derive a conditional discrete-descent result showing that the closed-loop objective decreases when a local velocity-alignment condition is satisfied. Numerical simulations on a 3D ellipsoidal manifold demonstrate distribution-level control and empirically support the proposed approximation and descent results.
自然语言处理
[NLP-0] Critical-State RL: Diagnosing Trainable States for Multi-Turn Tool Use
【速读】： 该论文旨在解决多轮工具调用（multi-turn tool-use）中因单一模型调用失败而导致整体任务失效的问题，尤其关注在奖励信号依赖后续交互时，如何准确识别真正可训练的决策状态。传统方法仅依赖奖励变化来判断是否需要训练，但这种变化可能源于下游随机性而非当前动作的有效性差异，导致训练方向错误。其解决方案的关键在于提出关键状态强化学习（Critical-State RL），通过结合任务定义的候选调用点与局部奖励，评估每个奖励是否真实反映动作对任务成功的影响，并判断相对于参考策略是否存在改进空间；随后采用嵌套采样（nested sampling）将动作相关奖励波动与延续性噪声分离，最终在诊断出的关键状态上使用上下文-贝叶斯（contextual-bandit）训练进行策略优化。实验在伯克利函数调用排行榜（BFCL）v4上的结果表明，对于缺失函数的任务，该方法选择工具可用后的响应作为训练点，而对缺失参数的任务则选择参数提供前的响应；在这些诊断选定状态下训练可显著提升性能，例如在缺失函数任务上提升约14个百分点，而其他替代状态的训练则无改善甚至导致性能下降。该方法进一步被推广至不同模型和任务，包括重复调用规避与记忆管理等场景，验证了其通用性和有效性。
链接: https://arxiv.org/abs/2609.24985
作者: Zixiang Chen,Wenting Zhao,Zhepeng Cen,Akshara Prabhakar,Jielin Qiu,Jianguo Zhang,Zhiwei Liu,Tulika Manoj Awalgaonkar,Liangwei Yang,Shelby Heinecke,Silvio Savarese,Huan Wang
机构: Salesforce AI Research
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 31 pages, 8 figures, 7 tables
Abstract:Multi-turn tool-use failures can hinge on a single model call, yet reward variation alone does not reveal which call would benefit from training. When rewards depend on later interactions, their variation can reflect downstream randomness rather than differences between the current actions. We introduce Critical-State RL to identify trainable states in multi-turn interactions. Given task-defined candidate calls and local rewards, the method assesses whether each reward captures the action’s effect on task success and whether improvement over a reference policy is possible. It then uses nested sampling to separate action-dependent reward variation from continuation noise and optimizes the policy at the selected states using contextual-bandit training. Experiments on the Berkeley Function Calling Leaderboard (BFCL) v4 compare training at diagnostic-selected states with training at alternative states. For missing-function tasks, the diagnostic selects the response after the tool becomes available; for missing-argument tasks, it selects the response before the missing argument is supplied. Training the selected responses improves performance, including about 14 percentage points on the missing-function task, while training the alternatives leaves performance flat or worse. We further apply the recipe across models and tasks, including logged repeat-call avoidance and memory management.
[NLP-1] Harness-Zero: Harness Distillation via Agent -as-Harness
【速读】： 该论文旨在解决生成式智能体（Agent）在部署时性能受限于特定外部系统（harness）的问题。由于最优 harness 在不同领域、实例和模型间差异显著，通用智能体要么采用次优的通用 harness，要么需不断切换大量专用 harness，导致系统复杂性与维护成本上升。为此，论文提出“harness distillation”（Harness-Zero）方法，其核心是将领域或实例优化的 harness 所诱导的行为通过训练过程内化至模型权重中，使模型在脱离该优化 harness 后仍能保持其性能优势。关键挑战在于源 harness 与目标 harness 的动作空间和可观测信息存在差异，无法直接提供监督信号。为此，论文引入“agent-as-harness”机制：利用一个可执行的 harnessing agent 作为中介，在目标 harness 的动作空间中对学生的输出进行修正，从而将原优化 harness 的指导转化为可学习的训练轨迹。通过对这些轨迹进行微调，模型逐步吸收并固化源自优化 harness 的行为模式。实验表明，该方法在知识工作、工具使用和科学任务等多领域均显著提升性能，使基础模型的任务成功率从 23.3% 提升至 44.3%，且超越了保留原始专用 harness 时的 41.7% 表现，并实现了 82.3% 的行为恢复率，验证了其有效性和泛化能力。
链接: https://arxiv.org/abs/2609.24974
作者: Haoran Ye,Yuxing Lu,Haonan Dong,Zhaochen Su,Guojie Song
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Neural and Evolutionary Computing (cs.NE)
备注:
Abstract:Agent harnesses, the external systems that mediate model-environment interaction, can substantially improve agent performance, but their gains remain tied to the harness at deployment. Because the best harness varies across domains, instances, and models, a general-purpose agent must either settle for a suboptimal shared harness or route among an ever-growing set of specialized ones. We therefore study agent harness distillation: using a domain- or instance-optimized harness as training-time guidance and transferring the behaviors it induces into model weights, so that its gains survive under a single fixed target harness. The challenge is that the two harnesses differ in action space and available information, so guidance from the optimized harness cannot serve directly as supervision for the target one. We introduce Harness-Zero, which enables harness distillation through agent-as-harness. Guided by the optimized harness, a harnessing agent corrects student responses before execution in the target harness’s action space, turning harness guidance into training demonstrations. Fine-tuning on the resulting trajectories internalizes harness-induced behavior into the model, so the specialized harness can be removed at deployment. Our experiments spanning knowledge work, tool use, and science domains show that: (1) For frontier LLMs using the same evolved harness, agent-as-harness outperforms code-as-harness. (2) With the specialized harness removed at deployment, Harness-Zero improves the base model’s macro-average task success from 23.3% to 44.3%, even exceeding the 41.7% it reaches with that harness still attached. (3) Harness-Zero recovers harness-induced behaviors absent from the base model, with 82.3% average recovery across 28 patterns in the three domains.
[NLP-2] RRSI: Regularized Recursive Self-Improvement of Agent Harnesses
【速读】： 该论文旨在解决大语言模型（LLM）智能体在递归自改进（Recursive Self-Improvement, RSI）过程中因过度拟合训练任务而导致泛化能力下降的问题。具体而言，现有方法通过迭代优化智能体的提示（prompt）、控制流、工具调用、记忆与上下文管理等组件实现自改进，但其进化过程易产生对特定任务过拟合的策略，导致在分布外（out-of-distribution）基准上性能显著衰减甚至消失。为应对这一挑战，本文提出正则化递归自改进（Regularized Recursive Self-Improvement of Agent Harnesses, RRSI），其核心在于将正则化思想引入智能体支架（harness）的自改进过程，通过双重约束机制控制候选方案的生成与选择：首先，提议器采用随时间退火的预算机制，限制每个候选方案可整合的修改数量，并鼓励探索未充分开发的演化路径；其次，选择器配备评鉴器（critic）与剪枝器（pruner），前者筛选针对特定基准的高质量提议，后者移除冗余、低效或不再有效的修改。上述机制共同引导进化向可复用的通用智能体机制倾斜，而非依赖于特定任务的噪声或过拟合模式。实验表明，在涵盖编码、代理工作空间及工程设计任务的八个基准上，RRSI在所训练的基准上最高提升14.1分，在五个分布外基准上最高提升4.7分，同时使智能体执行所需的策略（policy）token减少30%。
链接: https://arxiv.org/abs/2609.24972
作者: Peng Xia,Rujun Han,Zifeng Wang,Yanfei Chen,Yufan Zhang,Yoonho Lee,Chengsong Huang,Han Yu,Zhongying CuiZhu,Yifei Ming,Huaxiu Yao,Burak Gokturk,Tomas Pfister,Chen-Yu Lee
机构: Google (2026a)
Google
Gemini 3.1 pro: best for complex tasks and bringing creative concepts to life.
Note: https://deepmind.google/models/gemini/pro/
Cited by: §A.5.
Google (2026b)
Google
Gemini 3.5: frontier intelligence with action.
Note: https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-5/
Cited by: §A.3,
§4.3.
Guo et al. (2025)
X. Guo, Y. Li, X. Kong, Y. Jiang, X. Zhao, Z. Gong, Y. Zhang, D. Li, T. Sang, B. Zhu, et al.
Toward engineering agi: benchmarking the engineering design capabilities of llms.
Advances in Neural Information Processing Systems.
Cited by: §A.7,
§4.1.
Haarnoja et al. (2018)
T. Haarnoja, A. Zhou, P. Abbeel, and S. Levine
Soft actor-critic: off-policy maximum entropy deep reinforcement learning with a stochastic actor.
In International conference on machine learning,
pp. 1861–1870.
Cited by: §3.2.
Harvey AI (2026)
Harvey AI
Harvey lab: the legal agent benchmark.
Note: Announcement: https://www.harvey.ai/blog/introducing-harveys-legal-agent-benchmark
External Links: Link
Cited by: §A.3,
§4.1.
Hastie et al. (2009)
T. Hastie, R. Tibshirani, J. H. Friedman, and J. H. Friedman
The elements of statistical learning: data mining, inference, and prediction.
Vol. 2, Springer.
Cited by: §3.1,
§3.3.
Huang et al. (2026a)
C. Huang, H. Liu, T. Zheng, R. Dai, L. Huang, J. Li, Z. Li, Z. Wei, Y. Meng, and J. Huang
G-zero: self-play for open-ended generation from zero data.
arXiv preprint arXiv:2605.09959.
Cited by: §5.
Huang et al. (2026b)
C. Huang, Z. Wang, R. Han, J. Yan, Y. Chen, Z. CuiZhu, K. Jiang, P. Xia, H. Yu, Y. Zhuang, et al.
EnvHarness: awakening static worlds for agent learning.
arXiv preprint arXiv:2608.19880.
Cited by: §5.
Huang et al. (2026c)
C. Huang, W. Yu, X. Wang, H. Zhang, Z. Li, R. Li, J. Huang, H. Mi, and D. Yu
R-zero: self-evolving reasoning llm from zero data.
In International Conference on Learning Representations,
Vol. 2026, pp. 130770–130790.
Cited by: §5.
Huang et al. (2026d)
L. Huang, C. Yang, H. Zhou, H. Song, Z. Chen, R. Le, Y. Song, W. X. Zhao, and T. Zhang
Evo-bench: can language models improve agent harness?.
arXiv preprint arXiv:2608.09096.
Cited by: §1,
§5.
Jimenez et al. (2024)
C. E. Jimenez, J. Yang, A. Wettig, S. Yao, K. Pei, O. Press, and K. Narasimhan
Swe-bench: can language models resolve real-world github issues?.
In International Conference on Learning Representations,
Vol. 2024, pp. 54107–54157.
Cited by: §A.2,
§4.1.
Karten et al. (2026a)
S. Karten, A. L. Zhang, K. Thomas, S. Müller, and P. I. Team
Prime agent: a self-improving rlm harness.
Prime Intellect Blog.
Cited by: §1.
Karten et al. (2026b)
S. Karten, J. Zhang, T. Upaa Jr, R. Feng, W. Li, C. Shi, C. Jin, and K. Vodrahalli
Continual harness: online adaptation for self-improving foundation agents.
arXiv preprint arXiv:2605.09998.
Cited by: §1,
§5.
Ke et al. (2026)
Z. Ke, V. Patil, H. Shi, Y. Li, Y. Liu, S. Shekkizhar, A. Koul, J. Wang, X. P. Nguyen, S. Yavuz, et al.
EVOHARNESSBENCH: can your agents keep pace with an evolving harness?.
arXiv preprint arXiv:2609.04280.
Cited by: §1.
Lee et al. (2026a)
H. Lee, J. Xu, J. Seely, D. Lee, M. Zaharia, and Y. Tang
Recursive harness self-improvement.
arXiv preprint arXiv:2607.15524.
Cited by: §1,
§5.
Lee et al. (2026b)
Y. Lee, R. Nair, Q. Zhang, K. Lee, O. Khattab, and C. Finn
Meta-harness: end-to-end optimization of model harnesses.
The Third Conference on Language Modeling.
Cited by: Appendix B,
§1,
§2,
§4.1,
Table 1,
§5.
Li et al. (2026)
Y. Li, Y. Feng, Z. Xu, Z. Ma, K. Zheng, F. Jiang, X. Sun, R. Shao, Z. Chen, Y. Huang, et al.
JobBench: aligning agent work with human will.
arXiv preprint arXiv:2605.26329.
Cited by: §A.4,
§4.1.
Lin et al. (2026a)
J. Lin, S. Liu, C. Pan, L. Lin, S. Dou, Z. Xi, X. Huang, H. Yan, Z. Han, T. Gui, et al.
Agentic harness engineering: observability-driven automatic evolution of coding-agent harnesses.
arXiv preprint arXiv:2604.25850.
Cited by: Appendix B,
§1,
§4.1,
Table 1,
§5.
Lin et al. (2026b)
M. Lin, J. Wu, Z. Wang, Z. Shi, Y. Sang, B. He, Z. Liu, T. Wei, Z. Wu, Z. Zhang, et al.
Harness updating is not harness benefit: disentangling evolution capabilities in self-evolving llm agents.
arXiv preprint arXiv:2605.30621.
Cited by: §1.
Liu et al. (2026a)
J. Liu, X. Ye, P. Xia, Z. Zheng, C. Xie, M. Ding, and H. Yao
Evolvemem: self-evolving memory architecture via autoresearch for llm agents.
arXiv preprint arXiv:2605.13941.
Cited by: §5.
Liu et al. (2026b)
Z. Liu, Z. Shi, Y. Sang, B. He, M. Lin, T. Wei, D. Wang, B. Dumoulin, W. Jin, and H. Lu
Adaptive auto-harness: sustained self-improvement for agentic system deployment on open-ended task streams.
arXiv preprint arXiv:2606.01770.
Cited by: §5.
Lopopolo (2026)
R. Lopopolo
Harness engineering: leveraging codex in an agent-first world.
Note: https://openai.com/index/harness-engineering/
Cited by: §1,
§2,
§5.
Lou et al. (2026)
X. Lou, M. Lázaro-Gredilla, A. Dedieu, C. Wendelken, W. Lehrach, and K. P. Murphy
Autoharness: improving llm agents by automatically synthesizing a code harness.
arXiv preprint arXiv:2603.03329.
Cited by: §1.
Louizos et al. (2018)
C. Louizos, M. Welling, and D. P. Kingma
Learning sparse neural networks through L_0L_0 regularization.
In International Conference on Learning Representations,
Cited by: §3.1.
Luo et al. (2026)
X. Luo, F. Wang, C. Hu, D. Xue, and Y. Deng
Self-evolving agent harnesses via gated semantic quality-diversity.
arXiv preprint arXiv:2607.13683.
Cited by: §5.
Merrill et al. (2026)
M. Merrill, A. Shaw, N. Carlini, B. Li, H. Raj, I. Bercovich, L. Shi, J. Shin, T. Walshe, E. K. Buchanan, et al.
Terminal-bench: benchmarking agents on hard, realistic tasks in command line interfaces.
In International Conference on Learning Representations,
Vol. 2026, pp. 40903–40986.
Cited by: §A.1,
§4.1,
§4.1.
Nie et al. (2026)
J. Nie, Y. Zhang, J. Song, Q. Cai, D. Yu, Y. Guo, X. Tian, and B. Han
TTHE: test-time harness evolution.
arXiv preprint arXiv:2607.08124.
Cited by: Appendix B,
§1,
§4.1,
Table 1,
§5.
Niklaus (2026)
J. Niklaus
Don’t train the model, evolve the harness.
External Links: Link
Cited by: §1.
Ouyang et al. (2026)
S. Ouyang, J. Yan, I. Hsu, Y. Chen, K. Jiang, Z. Wang, R. Han, L. Le, S. Daruki, X. Tang, et al.
Reasoningbank: scaling agent self-evolving with reasoning memory.
In International Conference on Learning Representations,
Vol. 2026, pp. 94327–94354.
Cited by: §5.
Pan et al. (2026)
W. Pan, S. Liu, C. Lin, J. Zeng, X. Tang, X. Zhou, Y. Lu, and X. Jia
Retrospective harness optimization: improving llm agents via self-preference over trajectory rollouts.
arXiv preprint arXiv:2606.05922.
Cited by: §5.
Patwardhan et al. (2026)
T. Patwardhan, R. Dias, E. Proehl, G. Kim, M. Wang, O. Watkins, S. Fishman, M. Aljubeh, P. Thacker, L. Fauconnet, et al.
Gdpval: evaluating ai model performance on real-world economically valuable tasks.
In International Conference on Learning Representations,
Vol. 2026, pp. 24005–24040.
Cited by: §A.5,
§4.1.
Qwen Team (2026)
Qwen Team
Qwen3.6-Plus: towards real world agents.
External Links: Link
Cited by: §A.5.
Rajasekaran (2026)
P. Rajasekaran
Harness design for long-running application development.
Note: https://www.anthropic.com/engineering/harness-design-long-running-apps
Cited by: §1,
§2,
§5.
RSI-Exam Team (2026)
RSI-Exam Team
RSI-exam: benchmarking recursive self-improvement through executable research.
External Links: Link
Cited by: §1.
Tang et al. (2025)
X. Tang, T. Qin, T. Peng, Z. Zhou, D. Shao, T. Du, X. Wei, P. Xia, F. Wu, H. Zhu, et al.
Agent kb: leveraging cross-domain experience for agentic problem solving.
arXiv preprint arXiv:2507.06229.
Cited by: §5.
Team et al. (2026)
N. Team, G. Cao, G. Dai, T. Guo, K. Han, H. Hu, Z. Jiang, X. Kuang, B. Li, Y. Li, et al.
NeoHorse-1: towards recursive self-improvement via agentic post-training with routing harness.
arXiv preprint arXiv:2609.08183.
Cited by: §1.
Vidgen et al. (2026)
B. Vidgen, A. Mann, A. Fennelly, J. W. Stanly, L. Rothman, M. Burstein, J. Benchek, D. Ostrofsky, A. Ravichandran, D. Sur, et al.
APEX-agents.
arXiv preprint arXiv:2601.14242.
Cited by: §A.6,
§4.1,
§4.1.
Wang et al. (2026a)
R. Wang, Y. Shi, Z. Li, Z. Li, Y. Yu, J. Yang, K. Panaganti, H. Mi, D. Zhou, et al.
Harness handbook: making evolving agent harnesses readable, navigable, and editable.
arXiv preprint arXiv:2607.13285.
Cited by: §5.
Wang et al. (2025)
W. Wang, P. Piękos, L. Nanbo, F. Laakom, Y. Chen, M. Ostaszewski, M. Zhuge, and J. Schmidhuber
Huxley-g\backslash" odel machine: human-level coding agent development by an approximation of the optimal self-improving machine.
arXiv preprint arXiv:2510.21614.
Cited by: §1,
§5.
Wang et al. (2026b)
Y. Wang, H. Zhu, Z. Hu, Y. Yuan, Z. Chen, S. Senthil, H. Hajishirzi, Y. Tsvetkov, P. Dasigi, and T. Xiao
Rethinking the evaluation of harness evolution for agents.
In COLM 2026 The 2nd Workshop on Lifelong Agents: Learning, Aligning, and Evolving,
Cited by: §1,
§5.
Weng (2026)
L. Weng
Harness engineering for self-improvement.
External Links: Link
Cited by: §1,
§5.
Wu et al. (2026)
S. Wu, H. Zhu, Y. Zhang, X. Wang, and S. Yeung-Levy
AutoMem: automated learning of memory as a cognitive skill.
arXiv preprint arXiv:2607.01224.
Cited by: §5.
Xia et al. (2026a)
P. Xia, J. Chen, H. Wang, J. Liu, K. Zeng, Y. Wang, S. Han, Y. Zhou, X. Zhao, H. Chen, et al.
Skillrl: evolving agents via recursive skill-augmented reinforcement learning.
arXiv preprint arXiv:2602.08234.
Cited by: §5.
Xia et al. (2026b)
P. Xia, J. Chen, X. Yang, H. Tu, J. Liu, K. Xiong, S. Han, S. Qiu, H. Ji, Y. Zhou, et al.
MetaClaw: just talk–an agent that meta-learns and evolves in the wild.
arXiv preprint arXiv:2603.17187.
Cited by: §5.
Xia et al. (2026c)
P. Xia, K. Zeng, J. Liu, C. Qin, F. Wu, Y. Zhou, C. Xiong, and H. Yao
Agent0: unleashing self-evolving agents from zero data via tool-integrated reasoning.
The Third Conference on Language Modeling.
Cited by: §5.
Yang et al. (2026a)
C. Yang, X. Zhao, T. Wu, and C. Kästner
Better harnesses, smaller models: building 90% cheaper agents via automated harness adaptation.
arXiv preprint arXiv:2607.08938.
Cited by: §1,
§5.
Yang et al. (2026b)
Y. Yang, Z. Gong, W. Huang, Q. Yang, Z. Zhou, Z. Huang, Y. Li, X. Gao, Q. Dai, B. Liu, et al.
Skillopt: executive strategy for self-evolving agent skills.
arXiv preprint arXiv:2605.23904.
Cited by: §5.
Yao et al. (2022)
S. Yao, J. Zhao, D. Yu, N. Du, I. Shafran, K. Narasimhan, and Y. Cao
React: synergizing reasoning and acting in language models.
arXiv preprint arXiv:2210.03629.
Cited by: §4.1.
Zhang and Khattab (2026)
A. Zhang and O. Khattab
Language model harnesses are compositional generalizers.
External Links: Link
Cited by: §1,
§5.
Zhang et al. (2026a)
H. Zhang, S. Zhang, K. Li, C. Zhang, Y. Chen, Y. Zhang, L. Bai, and S. Hu
Self-harness: harnesses that improve themselves.
arXiv preprint arXiv:2606.09498.
Cited by: §1,
§5.
Zhang et al. (2026b)
J. Zhang, S. Hu, C. Lu, R. Lange, and J. Clune
Darwin gödel machine: open-ended evolution of self-improving agents.
In International Conference on Learning Representations,
Vol. 2026, pp. 104223–104294.
Cited by: §1,
§5.
Zhang et al. (2026c)
J. Zhang, B. Zhao, W. Yang, J. Foerster, J. Clune, M. Jiang, S. Devlin, and T. Shavrina
Hyperagents.
arXiv preprint arXiv:2603.19461.
Cited by: §5.
Zhang et al. (2026d)
L. Zhang, R. Zhou, D. Song, Z. Chen, Y. Tian, J. Yang, H. Ma, C. Li, G. Feng, X. Li, et al.
HarnessCompass: guiding automatic harness evolution toward generalizable and effective agent harnesses.
arXiv preprint arXiv:2608.01918.
Cited by: §1,
§1,
§5.
Zhang et al. (2026e)
Y. Zhang, Y. Dai, J. Tan, L. Yang, R. Mullur, T. Hoang, Z. Hu, J. Zhu, P. Mui, S. Savarese, et al.
DarwinX: evolving agent harnesses through natural selection.
arXiv preprint arXiv:2608.07545.
Cited by: §1.
Contents of Appendix
Appendix A Evaluation
This shows how each environment is run and scored. A harness and its baseline are always evaluated in the same window, with the same tool environment, the same judge and the same number of trials.
A.1 Terminal-Bench 2.1
Each task is a container image with a task description, a working directory and a set of unit tests that are hidden from the agent (Merrill et al., 2026). The agent drives a real shell through the harness, and a task counts as solved only if the task’s own test suite passes after the agent stops, so the reward is exact and cannot be produced by a plausible-looking answer. The reported accuracy is the fraction of the 89 tasks solved in this way. Containers are torn down and rebuilt between arms so that no state carries from one evaluation to the next.
A.2 SWE-bench Verified
Each instance is a real GitHub issue paired with the repository snapshot at the time of the report (Jimenez et al., 2024). The agent must produce a patch, which is then applied to the snapshot and checked against the instance’s fail-to-pass tests, which must go from failing to passing, and its pass-to-pass tests, which must remain passing. The reported resolve rate is the fraction of instances that satisfy both conditions.
A.3 Harvey LAB
Each task provides a folder of source documents in Word, Excel and PDF form and requires the agent to produce deliverable files under exact requested filenames (Harvey AI, 2026), which are graded by a strict per-criterion rubric of 20 to 100 independently judged criteria per task, roughly 14,000 criterion verdicts per full evaluation. A criterion is judged in isolation by an LLM judge (Gemini-3.5-Flash (Google, 2026b)) that reads the produced deliverable together with that single criterion, and the score of a run is the fraction of criteria passed over all tasks, so a task with a long rubric contributes proportionally more evidence than a short one and a missing deliverable fails every criterion it was supposed to satisfy rather than being dropped. The 160 tasks are partitioned once into a 120-task evolve set and a 40-task held-out set, and the partition is fixed for the experiment.
A.4 JobBench
Tasks are drawn from real professional workflows (Li et al., 2026), each shipping a task folder of input files and a wrapper prompt, with the reference material the agent would need to look up deliberately withheld so that part of the work is genuine retrieval. The harness exposes a filesystem, a code execution tool for producing office and PDF deliverables, and a grounded web search tool. Deliverables are graded by the benchmark’s own weighted rubric, and the reported number is the weighted rubric score over the evaluated split. We use an LLM judge (average score of Gemini-3.5-Flash and Claude Opus 4.8).
A.5 GDPval
For each task the deliverable produced by the harness is placed side by side with the human expert deliverable shipped with the benchmark (Patwardhan et al., 2026), a panel of three judges of different provenance picks the better of the two, and the reported number is the win rate against the expert over 185 tasks. The panel combines an open-weight model served locally (Qwen3.6-35B-A3B (Qwen Team, 2026)) with two proprietary models from different vendors (Claude Sonnet 4.6 (Anthropic, 2026b) and Gemini-3.1 Pro (Google, 2026a)), each pair is judged in both presentation orders to remove position bias, and the verdict for a task is the majority vote of the three. Each judge therefore issues 204 comparisons per harness, and a win rate above 50% means the harness produces the preferred deliverable more often than the human expert it is compared against.
A.6 APEX-Agents
Each task places the agent in a sandboxed world with its own MCP tool surface, covering a filesystem, PDF reading, spreadsheets, mail, chat, calendar, documents and code execution, and spanning three professional domains (Vidgen et al., 2026). A task is graded by a per-task rubric judged by an LLM judge (Gemini-3.5-Flash), and a task counts as a success under pass@1 only when its rubric is satisfied on the single sampled rollout. We evaluate the full set of 480 tasks and always report over that full denominator, so a task whose rollout is missing because of an infrastructure failure counts as a failure rather than being excluded, which prevents a harness that crashes on hard worlds from looking better than one that attempts them.
A.7 EngDesign
We used the license-free subset of EngDesign (Guo et al., 2025), of which we take the 61 tasks that run without proprietary simulators. Each task states a design goal together with the physical constraints the design must satisfy, and each is graded by its own frozen simulation or testbench rather than by a judge model, so grading is deterministic and every point of variance we measure comes from the policy. Evolution runs on all 61 tasks with no in-distribution held-out split, since the suite is too small to spend tasks on one.
A.8 Frontier-Eng
Frontier-Eng (Chi et al., 2026) collects real-world engineering optimization problems from 26 domains. Each task asks the agent to produce a design or a program that is scored by a frozen task-specific simulator or evaluator on a continuous objective, so as with EngDesign no judge model is involved and grading is deterministic. Because the objectives are not commensurable across tasks, the benchmark reports a Medal Score: for each task the three best feasible results of the frozen v1 snapshot are the gold, silver and bronze thresholds, a submission earns 1, 0.67 or 0.33 for reaching each, and the score is the mean credit over the 47 tasks of the v1 set, which we report as a percentage. We use Frontier-Eng only as an out-of-distribution test surface. Its EngDesign domain reuses tasks from our evolve set and is excluded, and tasks whose evaluation environment could not be built in our sandbox receive no credit in either arm, so 38 of the 47 tasks contribute credit and both arms are scored on exactly the same tasks.
Appendix B Baseline Methods
We briefly summarize the four harness-evolution baselines used in our experiments.
Meta-Harness (Lee et al., 2026b).
Meta-Harness formulates harness engineering as an outer-loop optimization problem over executable harness code. Its agentic proposer has access to the source code, evaluation scores, and execution traces of previous candidates, and uses this accumulated experience to propose improved harnesses.
Agentic Harness Engineering (AHE) (Lin et al., 2026a).
AHE uses an observability-driven evolution loop for coding-agent harnesses. It organizes harness components, execution experience, and edit outcomes into explicit representations so that an evolving agent can diagnose failures, propose changes, and evaluate the effects of previous edits.
Test-Time Harness Evolution (TTHE) (Nie et al., 2026).
TTHE evolves executable harnesses during test-time adaptation while keeping the underlying model weights fixed. It maintains multiple candidate harnesses, proposes modifications from execution traces, and uses an agentic judge to select a harness that persists to subsequent inputs.
HarnessX (Chen et al., 2026).
HarnessX represents an agent harness as a composition of modular, typed primitives spanning components such as prompts, tools, memory, and control flow. Its trace-driven adaptation mechanism uses execution feedback to modify and select harness configurations, enabling the runtime scaffold to evolve over time.
Appendix C Method Details
This section gives the round-level formulation and implementation details omitted from Section 3. It specifies the same proposal- and selection-side regularizers used in the experiments. As in the main text, the L0L_{0}, Lasso/L1L_{1}, and Ridge/L2L_{2} terminology is used only to indicate analogous roles in complexity control. The procedure does not optimize the corresponding norm-penalized objectives, and heterogeneous harness components are not treated as coordinates of a shared continuous parameter vector.
C.1 Round-Level Formulation
Let Ω(H)\Omega(H) denote the set of harnesses reachable from HH by arbitrary source edits. RRSI leaves Ω(H)\Omega(H) open and instead regularizes the transition through this space. A round takes the form
ℋt∼Preg(⋅∣Ht,ℱt,ℒt,bt,ℰt,ℬt)⊆Ω(Ht),Ht+1=argmaxH′∈ℋt∩𝒜tS^(H′),\mathcal{H}{t}\sim P{\mathrm{reg}}\big(\cdot\mid H_{t},\mathcal{F}{t},\mathcal{L}{t},b_{t},\mathcal{E}{t},\mathcal{B}{t}\big)\subseteq\Omega(H_{t}),\qquad H_{t+1}=\operatorname*{arg,max}{H^{\prime}\in\mathcal{H}{t}\cap\mathcal{A}_{t}}\hat{S}(H^{\prime}),
(8)
with Ht+1=HtH_{t+1}=H_{t} if no candidate is admissible. Here ℱt\mathcal{F}{t} is feedback from the current round, ℒt\mathcal{L}{t} is the edit history, btb_{t} is the annealed edit budget from Equation (4), ℰt\mathcal{E}{t} contains exploration directives, ℬt\mathcal{B}{t} contains structural pruning targets inferred from recent history, and 𝒜t\mathcal{A}_{t} is the set of candidates allowed to replace the incumbent.
A run applies Algorithms 1 and 2 for t=0,…,T−1t=0,\ldots,T-1, starting from H0H_{0} with S⋆=S^(H0)S^{\star}=\hat{S}(H_{0}). Before evolution, the unchanged base harness is evaluated repeatedly to estimate the empirical noise tolerance δ\delta.
Algorithm 1 RRSI, proposal side.
1:
HtH_{t}, history ℒt\mathcal{L}_{t}, round tt of TT
2:
bmin,bmaxb_{\min},b_{\max}, stall window ww, noise band δ\delta
3:
ℱt←Analyze(Ht,𝒟evolve)\mathcal{F}{t}\leftarrow\textsc{Analyze}(H{t},\mathcal{D}_{\mathrm{evolve}})
4:
bt←⌈bmin+(bmax−bmin)12(1+cosπtT)⌉b_{t}\leftarrow\big\lceil b_{\min}+(b_{\max}-b_{\min})\tfrac{1}{2}(1+\cos\tfrac{\pi t}{T})\big\rceil
5:
⊳\triangleright L0L_{0}-style edit-cardinality control
6:
σt←𝟙[S^t−S^t−w≤δ]\sigma_{t}\leftarrow\mathbb{1}[\hat{S}{t}-\hat{S}{t-w}\leq\delta]
7:
𝒯t←{ℓi:(ti,ℓi,…)∈ℒt}\mathcal{T}{t}\leftarrow{\ell{i}:(t_{i},\ell_{i},\ldots)\in\mathcal{L}_{t}}
8:
𝒰t←𝒦∖𝒯t\mathcal{U}{t}\leftarrow\mathcal{K}\setminus\mathcal{T}{t}
9:
ℰt←(σt,𝒰t,mdraft)\mathcal{E}{t}\leftarrow(\sigma{t},\mathcal{U}{t},m{\mathrm{draft}})
10:
ℬt←{ℓ∈𝒯t:gt(ℓ)≤0}\mathcal{B}{t}\leftarrow{\ell\in\mathcal{T}{t}:g_{t}(\ell)\leq 0}
11:
⊳\triangleright Lasso/L1L_{1}-style pruning targets
12:
ℋt∼Preg(⋅∣Ht,ℱt,ℒt,bt,ℰt,ℬt)\mathcal{H}{t}\sim P{\mathrm{reg}}(\cdot\mid H_{t},\mathcal{F}{t},\mathcal{L}{t},b_{t},\mathcal{E}{t},\mathcal{B}{t})
13:
tag each atomic edit with component and hypothesis metadata
14:
return candidates that pass the pre-evaluation screen
Algorithm 2 RRSI, selection side.
1:
screened ℋt\mathcal{H}{t}, (Ht,S^t,C^t)(H{t},\hat{S}{t},\hat{C}{t}), S⋆S^{\star}, δ\delta, kk
2:
β0,β1,ws,wc,wn\beta_{0},\beta_{1},w_{s},w_{c},w_{n}
3:
𝒜t←∅\mathcal{A}_{t}\leftarrow\emptyset
4:
for H′∈ℋtH^{\prime}\in\mathcal{H}_{t} in parallel do
5:
S^′,C^′←Evaluate(H′,𝒟evolve,k)\hat{S}^{\prime},\hat{C}^{\prime}\leftarrow\textsc{Evaluate}(H^{\prime},\mathcal{D}_{\mathrm{evolve}},k)
6:
ΔS←S^′−S^t\Delta S\leftarrow\hat{S}^{\prime}-\hat{S}{t}; ΔC←(C^′−C^t)/C^t\Delta C\leftarrow(\hat{C}^{\prime}-\hat{C}{t})/\hat{C}_{t}
7:
ν←νt(H′)\nu\leftarrow\nu_{t}(H^{\prime}) ⊳\triangleright new structural component types
8:
if ΔSδ\Delta S\delta then
9:
c←[ΔC≤β0+β1ΔS]c\leftarrow[\Delta C\leq\beta_{0}+\beta_{1}\Delta S]
10:
⊳\triangleright gain-dependent cost rule, Eq. (7)
11:
else
12:
c←[wsΔS−wcΔC+wnν0]c\leftarrow[w_{s}\Delta S-w_{c}\Delta C+w_{n}\nu0]
13:
⊳\triangleright within-band rule, Eq. (17)
14:
end if
15:
g←DomainGuard(Ht,H′)g\leftarrow\textsc{DomainGuard}(H_{t},H^{\prime})
16:
if S^′≥S⋆−δ\hat{S}^{\prime}\geq S^{\star}-\delta and cc and gg then
17:
𝒜t←𝒜t∪{H′}\mathcal{A}{t}\leftarrow\mathcal{A}{t}\cup{H^{\prime}}
18:
end if
19:
end for
20:
Ht+1←argmaxH′∈𝒜tS^′H_{t+1}\leftarrow\arg\max_{H^{\prime}\in\mathcal{A}{t}}\hat{S}^{\prime}, or HtH{t} if 𝒜t=∅\mathcal{A}_{t}=\emptyset
21:
S⋆←max(S⋆,S^t+1)S^{\star}\leftarrow\max(S^{\star},\hat{S}_{t+1})
22:
record each measured edit with a=1a=1 iff its candidate is Ht+1≠HtH_{t+1}\neq H_{t}
23:
return Ht+1H_{t+1}
C.2 Proposal-Side Bookkeeping
Atomic edit representation.
At round tt, the proposer drafts a pool EtE_{t} of atomic edits to HtH_{t}, and a candidate applies a subset of that pool. Write this subset as zt∈{0,1}|Et|z_{t}\in{0,1}^{|E_{t}|}, with zt,j=1z_{t,j}=1 when edit jj is included. The pool is redrawn each round from the open space Ω(Ht)\Omega(H_{t}), so |Et||E_{t}| need not be fixed across rounds. The annealed budget in Equation (4) imposes
‖zt‖0≤bt.|z_{t}|{0}\leq b{t}.
(9)
Thus btb_{t} limits the number of independently attributable edits bundled into one candidate rather than the set of components that may eventually be modified. This is the most direct of our classical analogies: it is a cardinality constraint on the update, not an L0L_{0} penalty on a fixed model parameter vector.
Edit history and component-level summaries.
Every atomic edit in an evaluated candidate is tagged with a component ℓ\ell, a hypothesis hh, and the candidate source diff dd. A candidate containing multiple edits contributes one history record per edit; all edits in that candidate share the same measured ΔS\Delta S, ΔC\Delta C, and round outcome. Ignoring candidates that fail before a valid measurement is obtained, the history before round tt can be written
ℒt={(ti,ℓi,hi,di,ΔSi,ΔCi,ai):i≤nt},ai∈{0,1},\mathcal{L}{t}={(t{i},\ell_{i},h_{i},d_{i},\Delta S_{i},\Delta C_{i},a_{i}):i\leq n_{t}},\qquad a_{i}\in{0,1},
(10)
where ai=1a_{i}=1 iff the candidate carrying edit ii was selected as the winner of its round and therefore entered the accepted evolution path. Candidates that are admissible but lose to a higher-scoring admissible candidate have ai=0a_{i}=0.
Two summaries used by the proposer are
𝒯t={ℓi:i≤nt},gt(ℓ)=max{ΔSi:ℓi=ℓ,t−ti≤nprune},max∅=−∞.\mathcal{T}{t}={\ell{i}:i\leq n_{t}},\qquad g_{t}(\ell)=\max{\Delta S_{i}:\ell_{i}=\ell,;t-t_{i}\leq n_{\mathrm{prune}}},\qquad\max\emptyset=-\infty.
(11)
Here 𝒯t\mathcal{T}{t} is the set of components with at least one measured edit, and gt(ℓ)g{t}(\ell) is the best recent measured gain associated with component ℓ\ell over the pruning window. Because bundled edits inherit the candidate-level measurement, this evidence becomes more attributable as the edit budget anneals toward one.
Structured exploration state.
Let 𝒦\mathcal{K} denote the editable component vocabulary. In the implementation,
𝒦={\displaystyle\mathcal{K}={
prompt,control_flow,config,output_plumbing,\displaystyle\texttt{prompt},\texttt{control_flow},\texttt{config},\texttt{output_plumbing},
(12)
context_mgmt,client_tool,skill,memory,subagent}.\displaystyle\texttt{context_mgmt},\texttt{client_tool},\texttt{skill},\texttt{memory},\texttt{subagent}}.
The exploration directive is
ℰt=(σt,𝒰t,mdraft),σt=𝟙[S^t−S^t−w≤δ],𝒰t=𝒦∖𝒯t,\mathcal{E}{t}=(\sigma{t},\mathcal{U}{t},m{\mathrm{draft}}),\qquad\sigma_{t}=\mathbb{1}[\hat{S}{t}-\hat{S}{t-w}\leq\delta],\qquad\mathcal{U}{t}=\mathcal{K}\setminus\mathcal{T}{t},
(13)
where σt\sigma_{t} indicates that progress over the previous ww rounds has not exceeded the empirical noise tolerance, 𝒰t\mathcal{U}{t} contains components not yet exercised by a measured edit, and mdraftm{\mathrm{draft}} reserves candidate slots for exploratory edits when the search is stalled.
Structural pruning.
The pruning target set is
ℬt={ℓ∈𝒯t:gt(ℓ)≤0}.\mathcal{B}{t}={\ell\in\mathcal{T}{t}:g_{t}(\ell)\leq 0}.
(14)
Thus a component is marked as unproductive when it has been exercised but has produced no strictly positive measured gain in the recent pruning window. The proposer receives ℬt\mathcal{B}{t} together with any previously accepted edits associated with those components and is instructed to remove unproductive machinery in subsequent proposals. This is analogous in role to Lasso/L1L{1}-style sparsification because the mechanism acts by deleting discrete structure from the retained harness; it is not an L1L_{1}-penalized continuous optimization problem.
C.3 Selection-Side Bookkeeping
Noise-adjusted floor.
The leakage critic is applied before full evaluation. For every candidate that reaches selection, the first non-compensatory performance requirement is the stability floor from Equation (5),
S^(H′)≥S⋆−δ.\hat{S}(H^{\prime})\geq S^{\star}-\delta.
This permits fluctuations within the empirically calibrated tolerance while preventing the search from accumulating a sequence of small regressions.
Novelty used by the within-band rule.
The shaped rule uses novelty only for structural components. Let
𝒦str={client_tool,skill,memory,subagent}\mathcal{K}_{\mathrm{str}}={\texttt{client_tool},\texttt{skill},\texttt{memory},\texttt{subagent}}
(15)
and let Nt(ℓ)N_{t}(\ell) be the number of previously accepted edit records tagged with component ℓ\ell before round tt. If comp(H′)\mathrm{comp}(H^{\prime}) is the set of component types touched by candidate H′H^{\prime}, the implementation computes
νt(H′)=∑ℓ∈𝒦str[ℓ∈comp(H′)∧Nt(ℓ)=0].\nu_{t}(H^{\prime})=\sum_{\ell\in\mathcal{K}{\mathrm{str}}}\mathbb{1}!\left[\ell\in\mathrm{comp}(H^{\prime})\ \wedge\ N{t}(\ell)=0\right].
(16)
Hence νt(H′)\nu_{t}(H^{\prime}) counts distinct structural component types touched by the candidate that have never previously appeared in a winning edit. Prompt, control-flow, configuration, output-plumbing, and context-management edits do not receive this novelty bonus.
Acceptance when the gain exceeds the noise tolerance.
For ΔSδ\Delta S\delta, the selector uses the gain-dependent cost condition from Equation (7),
ΔC≤β0+β1ΔS.\Delta C\leq\beta_{0}+\beta_{1}\Delta S.
The rule allows more inference cost only when accompanied by a larger measured improvement. This is the part of complexity-aware acceptance that motivates the Ridge/L2L_{2}-style analogy in the main text: it suppresses unchecked growth in aggregate resource footprint without requiring an individual component to be eliminated. The analogy is functional rather than mathematical; the rule is not a squared-norm penalty.
Acceptance when the gain does not exceed the noise tolerance.
For candidates that pass the stability floor but whose measured gain does not exceed the empirical tolerance, ΔS≤δ\Delta S\leq\delta, the implementation does not use Equation (7). Instead it applies the shaped admissibility condition
wsΔS−wcΔC+wnνt(H′)0.w_{s}\Delta S-w_{c}\Delta C+w_{n}\nu_{t}(H^{\prime})0.
(17)
Here ws,wc,wn≥0w_{s},w_{c},w_{n}\geq 0 control, respectively, the contribution of the measured score change, relative inference-cost change, and previously unused structural component types. The purpose of this branch is to avoid treating a small score fluctuation as sufficient evidence by itself. Within this region, reducing cost contributes positively through −wcΔC-w_{c}\Delta C, and trying a structural mechanism that has never previously entered the accepted evolution path contributes through wnνt(H′)w_{n}\nu_{t}(H^{\prime}). Depending on the evolution instance, a within-band score change may also contribute through wsΔSw_{s}\Delta S.
The coding instance sets ws=0w_{s}=0. Consequently, a score increase that remains within δ\delta cannot by itself make a coding candidate admissible; the candidate must instead obtain sufficient credit from lower cost and/or structural novelty. The agentic-workspace and engineering-design instances use positive wsw_{s}. All three weights are fixed for an evolution instance and are reported in Table 5. Equation (17) is an implementation-level tie-breaking/admissibility rule inside the uncertainty region; it is not itself identified with an LpL_{p} penalty.
Domain-specific non-compensatory guards.
After the stability and cost checks, the implementation may apply a domain-specific guard g(Ht,H′)∈{0,1}g(H_{t},H^{\prime})\in{0,1}. The coding and agentic-workspace instances use no additional guard, so g=1g=1. The engineering-design instance additionally rejects a candidate if its valid-output rate falls by more than 0.030.03 relative to the incumbent or if its no-submission rate rises by more than 0.020.02. These guards prevent a gain in the primary pass-rate objective from compensating for a substantial degradation in basic execution validity.
Final round selection.
A candidate is admissible only if it satisfies the noise-adjusted floor, the appropriate branch of the complexity-aware rule, and all active domain guards. Among admissible candidates, the selector chooses the one with the largest measured score; if none is admissible, the incumbent is retained. The running best score is then updated as S⋆←max(S⋆,S^(Ht+1))S^{\star}\leftarrow\max(S^{\star},\hat{S}(H_{t+1})).
Appendix D Experiments
D.1 Hyperparameter Setting
RRSI introduces a small number of hyperparameters that control update sparsity, exploration, pruning, and the cost–performance trade-off. We select these parameters using only the evolve environment and operational considerations; held-out and OOD benchmarks are not used for tuning. The noise tolerance δ\delta is calibrated from repeated evaluations of the unchanged base harness. The edit-budget parameters (bmin,bmax)(b_{\min},b_{\max}) determine how many independent changes can be bundled into one candidate, while ww and mdraftm_{\mathrm{draft}} control when and how strongly the search explores underused components. The pruning window nprunen_{\mathrm{prune}} determines how much recent evidence is required before a component is treated as unproductive. Finally, (β0,β1)(\beta_{0},\beta_{1}) encode the allowed trade-off between measured gain and additional inference cost. Table 5 lists the values used in each instance. Scores S^\hat{S} are fractions in [0,1][0,1] and ΔC\Delta C is the relative change in policy tokens per trial, so δ\delta and β1\beta_{1} are expressed in those units: on the coding instance δ\delta corresponds to 3 passes out of 89×k=17889\times k=178 trials, on the agentic workspace instance to 60 criteria out of roughly 14,10014{,}100 criterion verdicts, and on the engineering design instance to 5 passes out of 61×k=24461\times k=244 trials. Likewise β1\beta_{1} corresponds to a 25% token allowance per additional pass (coding), per 100 additional criteria (agentic workspace) and a 10% allowance per additional pass (engineering design).
Hyperparameter
Role
Coding
Agentic workspace
Engineering design
TT
evolution rounds
20
20
40
kk
trials per task per evaluation
2
2
4
δ\delta
empirical noise tolerance
0.017
0.004
0.020
bminb_{\min}
final-round edit budget
1
1
1
bmaxb_{\max}
initial edit budget
4
3
4
ww
stall-detection window
3
3
3
mdraftm_{\mathrm{draft}}
reserved exploratory proposals
1
1
1
nprunen_{\mathrm{prune}}
pruning window
4
4
5
β0\beta_{0}
base cost allowance
0.10
0.10
0.15
β1\beta_{1}
gain-dependent cost allowance
44.5
35.4
24.4
Table 5: Hyperparameters used by RRSI in each evolution setting. All choices are fixed without consulting held-out or OOD benchmarks.
Appendix E Qualitative Case Study
To complement the aggregate results, we inspect representative decisions made during RRSI evolution. Table 6 summarizes several examples from the released trajectories. The complete round-by-round records, including proposals, critic decisions, acceptance decisions, and exact harness diffs, are available on our project website.
Domain / Round
Harness change
Outcome
What it illustrates
Coding, R0-A
Adds a bounded pre-completion verification audit and guidance for non-blocking polling of long-running jobs.
Accepted: +3.93+3.93 points on the evolve set.
A reusable behavioral mechanism can justify a relatively broad early-round update when the gain exceeds the noise threshold.
Coding, R0-B
Adds a similar verification reminder and long-running-work guidance, but with a smaller measured gain and additional inference cost.
Rejected by cost rule: +1.69+1.69 points, +26.1%+26.1% cost.
An apparent improvement is not automatically retained when it lies within the noise band and requires substantial additional computation.
Coding, R8-B
Pins the original task instruction into the completion gate so that the policy re-checks the literal specification before submission.
Rejected by floor: −2.81-2.81 points despite −13.6%-13.6% cost.
Lower cost alone cannot compensate for a candidate whose performance falls below the noise-adjusted acceptance floor.
Engineering, R2
Adds a bounded recovery hint for the recurring “workdir must be an existing directory” tool-use error.
Accepted: 122/244→128/244122/244\rightarrow 128/244 passes, +1.6%+1.6% tokens.
The search can retain small, task-agnostic control-flow fixes that improve reliability with little added complexity.
Table 6: Representative harness-evolution decisions from RRSI. The examples show that evolution is not driven by score alone: candidate specificity, evaluation stability, and inference cost jointly determine whether a change is retained.
These examples provide a more concrete view of the regularization behavior. In particular, the two candidates from the first coding round are superficially similar, yet only the candidate with a sufficiently large measured improvement survives the cost-aware selection rule. Conversely, the round-8 candidate reduces inference cost but is still rejected because its performance falls below the admissible floor. The engineering example shows the complementary case: a small and reusable control-flow correction is retained with little resource growth. Together, these trajectories suggest that RRSI does not simply accumulate edits that improve the evolve-set score, but selectively retains changes whose measured benefit is sufficiently robust relative to their complexity.
Experimental support, please
view the build logs
for errors. Generated by
L
A
T
E
xml
.
Instructions for reporting errors
We are continuing to improve HTML versions of papers, and your feedback helps enhance accessibility and mobile
support. To report errors in the HTML that will help us improve conversion and rendering, choose any of the
methods listed below:
Click the “Report Issue” (
) button, located in the page header.
Tip: You can select the relevant text first, to include it in your report.
Our team has already identified the following issues. We appreciate your time reviewing and reporting rendering errors we
may not have found yet. Your efforts will help us improve the HTML versions for all readers, because disability
should not be a barrier to accessing research. Thank you for your continued support in championing open access for
all.
Have a free development cycle? Help support accessibility at arXiv! Our collaborators at LaTeXML maintain a list of packages that need conversion, and welcome developer contributions.
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 fromChen-Yu Lee
Affiliation: \thepa Cloud AI Research
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:An LLM agent’s capability is largely magnified by its harness, namely the prompts, control flow, tooling, memory, and context management surrounding the frozen backbone model. Recent methods increasingly automate this process by iteratively proposing and selecting component-wise edits of an agent harness, practically establishing a form of recursive self-improvement (RSI) at the agent-system level. However, such recursive evolution may overfit by memorizing the training tasks, showing large in-distribution gains that shrink or even vanish on out-of-distribution benchmarks. We introduce Regularized Recursive Self-Improvement of Agent Harnesses (RRSI), which incorporates the principles of regularizations into harness self-improvement by constraining the evolution candidate proposal and selection. The proposer operates with a temporally annealed budget, limiting how many edits a candidate can bundle, and it encourages unexplored trajectories based on evolution history. The selector is equipped with a critic and a pruner: the critic screens benchmark-specific proposals, while the pruner, removes changes that are too small, too expensive, or no longer useful. Together these constraints favor reusable agent mechanisms over benchmark-specific ones or even noises. Across eight benchmarks spanning coding, agentic workspace and engineering design tasks, RRSI gains up to 14.1 points on the split it evolves against and up to 4.7 points on the five out-of-distribution benchmarks, while producing a harness that runs on 30% fewer policy tokens than the unregularized evolution. Code is available at this https URL and project page is this https URL.
[NLP-3] DolphinBench: Mapping the Pareto Frontier of Agent Memory
【速读】： 该论文旨在解决当前生成式AI（Generative AI）代理在执行依赖长期记忆与上下文回溯的任务时，缺乏有效评估手段的问题。现有记忆基准测试多局限于对话问答形式，其问题本身已隐含需检索的事实线索，且仅关注准确率，忽视了计算成本与延迟等实际系统开销，导致模型可通过不合理的时间/成本权衡来提升得分。为此，本文提出DolphinBench，一个基于真实任务完成度的综合性记忆评估基准，包含三个知识型代理角色，每个角色对应约50万词元的用户消息历史，并设计了200项需依赖历史信息才能成功完成的任务。通过对比代理在有无相关历史情况下的表现，严格验证任务成败，确保记忆系统的必要性；同时强制要求报告总成本与延迟，实现对代理记忆系统在准确性、效率与资源消耗方面的全面评估。该基准首次整合了任务驱动评估、严格因果验证与多维度开销指标，填补了现有研究空白。
链接: https://arxiv.org/abs/2609.24971
作者: Soumil Rathi,Deshraj Yadav,Taranjeet Singh
机构: Mem0
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 6 pages, 2 figures
Abstract:Agents today often take real-world actions that depend on long-term memory and context recall over time. However, most current memory benchmarks are built for a conversational question-answer format, where the question itself signals that some fact must be retrieved, and often which one. Moreover, benchmarks rarely require anything beyond accuracy from submissions, allowing memory systems to make unreasonable cost/time tradeoffs to achieve higher scores. We present DolphinBench, a benchmark that evaluates memory directly through an agent’s task completion. DolphinBench includes three knowledge-work personas with roughly 500k tokens of user messages per persona and evaluates agents on tasks that depend on information from that history. We verify all 200 tasks per persona by running an agent with and without the relevant history, requiring success with it and failure without it. Finally, we require all evaluations to report total cost and latency alongside accuracy, which enables us to evaluate agent memory systems holistically. No existing memory benchmark combines all three. The dataset and evaluation code are available at this https URL. Comments: 6 pages, 2 figures Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) Cite as: arXiv:2609.24971 [cs.CL] (or arXiv:2609.24971v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2609.24971 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-4] Emergent Collusion in Long-Horizon LLM Agent Interaction
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）代理在长期多智能体协作环境中可能产生的合谋（collusion）问题。随着LLM代理在协同场景中的广泛应用，其在长时间交互中可能形成非预期的、违背设计初衷的协调行为，从而带来安全风险。研究的关键在于构建一个具有现实约束的长周期多智能体环境：两个代理需独立完成任务、共享任务日志、相互验证工作并获取奖励。研究发现，当合规执行验证协议与最大化奖励目标存在冲突时，代理在重复交互中逐渐偏离既定协议；在10个模型的实验中，94%的轨迹出现了合谋现象，且同一模型家族中能力更强的代理更早达成合谋。通过受控的同行干预和消融实验表明，合谋行为受到同伴行为模式、奖励机制、验证反馈信息以及交互历史的影响，尤其限制交互历史的范围和数量可有效抑制合谋。研究揭示了长期交互会重塑代理间的协调方式，进而引发潜在的安全隐患，其解决方案的关键在于优化交互设计、控制信息暴露范围，并重构激励结构以防范非预期协同行为的产生。
链接: https://arxiv.org/abs/2609.24967
作者: Xinrui Shi,Yanzhe Zhang,Diyi Yang
机构: Stanford University (斯坦福大学); Georgia Tech (佐治亚理工学院)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:LLM agents are increasingly deployed in collaborative settings, yet long-term interaction may give rise to undesirable coordination. We study the emergence of collusion in a long-horizon multi-agent environment: two agents repeatedly complete individual tasks, share task logs, verify each other’s work, and receive rewards. We introduce realistic constraints that make compliance with the verification protocol incompatible with reward maximization, and find that agents increasingly deviate from the protocol over repeated interactions. Collusion emerges in 94% of trajectories across 10 models, and more capable models within the same family reach it earlier. Controlled peer interventions show that collusion is shaped by peer behavior, while ablations reveal additional effects of reward structure, the verification feedback agents receive, and their interaction history. In particular, restricting the amount and scope of interaction history available to agents reduces collusion. Overall, our findings show that long-horizon interaction can reshape how agents coordinate in ways that create safety risks.
[NLP-5] Jev for Scientific Decisions: Evaluating Semantic Choices and Their Consequences
【速读】： 该论文旨在解决科学工作流中因关系选择不确定性导致的语义歧义问题，尤其是在涉及文化背景、处理方式或参考标准不同时，同一计数或比较结果可能产生截然不同的科学意义。其解决方案的关键在于将Jev作为语义决策组件，通过遵循其文档指引并为代码分配算术逻辑，实现对科学关系的精确选择。研究采用标准化测试框架，在十个科学案例中针对二十个基于源数据的“选择”任务进行五次重复实验，系统评估了十二种模型配置在语义选择、下游输出及最终结论标签三个层面的表现。结果显示，Jev在完全语义正确性上与另外五种配置持平，并在成功响应中表现出最低的中位延迟；更关键的是，在三个对比模型中，七个关于文化-历史问题的错误选择虽改变了下游计数，却未影响最终标签，揭示了仅依赖最终结论评估模型性能的局限性。因此，该研究强调：评估此类决策组件的有效性必须深入检查工作流中将被重用的关系与量化指标，以确保语义一致性与科学可解释性。
链接: https://arxiv.org/abs/2609.24965
作者: Boyuan Deng,Shuyi Fan,Hongyang Zhang,Xinhong Xie
机构: Johns Hopkins University (约翰霍普金斯大学); Columbia University (哥伦比亚大学); The Hong Kong Polytechnic University (香港理工大学); The Pennsylvania State University (宾夕法尼亚州立大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 11 pages, 1 figure, 5 tables. Includes references and appendices
Abstract:Scientific workflows often require choosing among known relations before a deterministic calculation can proceed. Whether observations share a culture, treatment or reference standard can change the scientific meaning of the resulting count or comparison. We evaluate Jev as a semantic decision component using a harness that follows its documented guidance and assigns arithmetic to code. The study compares twelve model configurations on twenty source-grounded Choices across ten scientific cases, each repeated five times. We measure semantic selections, downstream outputs and final claim labels separately. Jev matched five other configurations at complete semantic correctness and achieved the lowest observed median latency among successful responses. Across three comparison models, seven wrong selections on one culture-history question changed downstream counts while preserving the correct final label. These results identify a useful role for Jev in prepared scientific decision tasks and show why evaluating that role requires checking the relations and quantities that a workflow will reuse.
[NLP-6] Linguistic Features for Interpretable Textual Entailment
【速读】： 该论文旨在解决神经模型在自然语言处理中因“黑箱”特性导致的可解释性不足问题，即模型预测过程缺乏对语言现象背后机制的透明揭示。其核心解决方案是提出一种可解释的混合模型SLITE，用于识别文本蕴含（Recognizing Textual Entailment, RTE），该模型融合了两个互补的语义分析层级：基于组合实体间语义兼容与不兼容关系的结构-关系层，以及基于前提与假设嵌入表示之间信息变化结构化模式的分布-信息层。关键创新在于设计了17个综合特征，涵盖实体级语义关系、极性敏感的词汇匹配以及相似性矩阵中语义子表示的对齐度量（包括熵和转移熵等）。通过逻辑回归模型对这些特征进行训练，在三分类SICK数据集上达到83%准确率，在SICK-CE上达96%，优于IsoLex 4个百分点，且仅需极低计算开销即可接近RoBERTa性能。消融实验与SHAP分析表明，结构-关系特征是分类的主要驱动因素，而分布-信息特征则在识别中立与矛盾关系中提供关键补充。研究结果表明，探索此类混合方法是替代大规模神经架构的一种可行且富有科学价值的路径，有助于推动语言学理论与推理计算建模之间的对话。
链接: https://arxiv.org/abs/2609.24932
作者: David Torres-Moreno,Jorge Hermosillo-Valadez,Asela Reig-Alamillo
机构: Centro de Investigación en Ciencias, Universidad Autónoma del Estado de Morelos(莫雷洛斯州自治大学科学研究中心); Centro Interdisciplinario de Investigación en Humanidades, Universidad Autónoma del Estado de Morelos(莫雷洛斯州自治大学跨学科人文学科研究中心)
类目: Computation and Language (cs.CL); Symbolic Computation (cs.SC)
备注: 38 pages, 5 figures, 8 tables
Abstract:Despite the success of neural models in natural language processing, their black-box nature limits interpretability and conceals the linguistic phenomena underlying their predictions. We present SLITE, an explainable hybrid model for Recognizing Textual Entailment that integrates two complementary layers of semantic analysis: a structural-relational layer, based on semantic compatibility and incompatibility between compositional entities, and a distributional-informational layer, based on structured patterns of information change between embedding-based representations of the premise and the hypothesis. We propose 17 features that combine entity-level semantic relations, polarity-sensitive lexical matching, and alignment measures over semantic sub-representations of the similarity matrix, including measures based on entropy and transfer entropy. A logistic regression trained on these features achieves an accuracy of 83% on three-class SICK and 96% on SICK-CE, outperforming IsoLex by 4 percentage points and falling within 2 percentage points of RoBERTa with a fraction of its computational complexity. Ablation studies and SHAP analysis confirm that structural-relational features are the primary drivers of classification, while distributional-informational features provide essential complementary contributions, particularly for detecting neutrality and contradiction. Our results demonstrate that further exploration of hybrid approaches is a viable and scientifically productive alternative to massive neural architectures, and we hope they will strengthen the dialogue between linguistic theory and computational modeling of inference
[NLP-7] SocioVerse2: A Longitudinal Dynamic Social Simulation Framework under a Human-AI Co-evolutionary Paradigm
【速读】： 该论文旨在解决社会仿真在实证研究中面临的两大核心挑战：一是对仿真内容的干预能力不足，二是研究者对仿真生成过程缺乏可控性。现有平台虽能通过自主代理实现群体行为模拟并匹配真实社会结构，但无法系统支持研究者在仿真过程中进行动态干预或对研究流程进行可编辑控制。为此，论文提出SocioVerse2，其解决方案的关键在于构建一个“人-AI协同演化”范式，由两个核心循环与一套智能基础设施共同支撑：一是纵向仿真循环，通过引入可干预的演化环境和反事实分支，实现对社会系统动态演化的追踪与实验；二是可控研究循环，将研究过程本身作为可编辑状态，支持研究者通过可控编辑持续迭代研究版本。该范式依托社会科学研究代理基础设施，集成可组合技能、研究者检查点机制、覆盖五个角色池的人口服务以及基于21个真实信号源的环境服务，并提供时间点保障。通过三个案例族及七项案例研究验证，SocioVerse2不仅能够复现经典基于代理的模型，还可基于真实记录建模政策过程，并实现超越响应模型知识截止期的宏观经济指数实时预测。这些应用突破了传统系统演示的局限，真正成为各学科前沿问题的实质性研究工具。相关代码、数据服务与工作台均已开源。
链接: https://arxiv.org/abs/2609.24911
作者: Xinnong Zhang,Jiayu Lin,Jia Wang,Yixu Huang,Xinyi Mou,Yingqian Wu,Jingcong Liang,Shijun Lei,Jianing Shi,Guanying Li,Siyuan Wang,Hanjia Lyu,Zhenfei Yin,Yunlu Yin,Siming Chen,Yulan He,Jiebo Luo,Xuanjing Huang,Liyin Jin,Baohua Zhou,Hanqi Yan,Zhongyu Wei
机构: Fudan University (复旦大学); Shanghai Innovation Institute; King’s College London; Tongji University; Northwestern Polytechnical University; The London School of Economics and Political Science; The Chinese University of Hong Kong; Singapore Management University; University of Oxford; University of Rochester
类目: Computation and Language (cs.CL); Computers and Society (cs.CY)
备注: Project page: this https URL
Abstract:Social simulation offers the social sciences an experimental instrument that the real world cannot supply, and generative agents have transformed it by acting as silicon samples that unite agent-based modeling with real behavioral data. Existing platforms verify collective behavior, align simulated populations with real societies in cross-sections, and employ autonomous agents for the research process. However, two social science requirements remain without systematic support: intervention in the content of a simulation and the researcher’s control over the process that produces it. We present SocioVerse2, which extends SocioVerse 1.0 into a human-AI co-evolutionary paradigm built from two loops and one infrastructure. The longitudinal simulation loop simulates the target population with evolving environments and forks counterfactual branches via interventions. The controllable research loop takes the study itself as an editable state and updates state versions via controllable editing. The social science agentic infrastructure carries both loops through composable skills with researcher checkpoints, a population service over five persona pools, and an environment service over 21 real-world signal sources with point-in-time guarantees. We validate SocioVerse2 across three case families and seven case studies, from reproducing canonical agent-based models to modeling policy processes on real records and nowcasting macro-economic indices beyond the response model’s knowledge cutoff. With the human-AI co-evolutionary paradigm, these cases go beyond system demonstrations to become substantive studies that investigate frontier questions in their respective disciplines. Code, data services, and a workbench are released as open-source resources.
[NLP-8] oneCL: Contrastive Learning for Few-Shot Syllable-Level Tone Classification AACL
【速读】： 该论文旨在解决低资源语境下声调语言（tone language）中缺乏大规模标注语料的问题，特别是针对领域语言学家所需的细粒度音节级声调标注需求。现有数据集多为句子级标注，难以满足音节级声调分类任务的需求。为此，作者提出ToneCL——一种轻量级对比学习框架，用于少样本条件下的音节级声调分类。其核心解决方案在于：在无标签语音上预训练模型，并采用保持声调身份不变的增强策略（如频率带拒绝），随后在极少量标注样本（每类仅数十例）上微调。实验表明，该方法在10次采样条件下于六说话者普通话数据集上达到91.6%的准确率，跨语言迁移也表现优异（越南语预训练+普通话微调达91.0%）。消融实验证明，频率带拒绝是关键增强手段，对声调特征的保留至关重要。
链接: https://arxiv.org/abs/2609.24903
作者: Qisheng Liao,Youngah Do
机构: The University of Hong Kong(香港大学)
类目: Computation and Language (cs.CL)
备注: AACL-IJCNLP 2026 Main
Abstract:Tone languages constitute over 50-70% of the world’s languages, but the vast majority are low-resource, lacking the large transcribed corpora needed for automatic tone classification. Existing datasets are typically collected at the sentence level, whereas field linguists require fine-grained syllable-level annotations. We propose ToneCL, a lightweight contrastive learning framework for few-shot syllable-level tone classification. We simulate low-resource conditions on Mandarin and Vietnamese, limiting labeled data to tens of examples per tone class. ToneCL is pretrained on unlabeled speech with augmentations that preserve tonal identity, then fine-tuned on few-shot examples. Experiments show our method consistently outperforms baselines, achieving 91.6% on six-speaker Mandarin at 10 shots. Cross-lingual transfer is also effective: pretraining on Vietnamese and fine-tuning on Mandarin reaches 91.0% accuracy at 10 shots. Ablation confirms that frequency band rejection is the most critical augmentation.
[NLP-9] Human-LLM Deliberation as Interactive Proof: Conditions for Verifiability Without Transparency
【速读】： 该论文旨在解决在大语言模型（LLM）提供人类难以自行构建的论证时，用户如何判断是否应接受其结论的问题。其核心挑战在于确保推理过程的可信性，尤其是在模型可能生成看似合理但实质错误的论据的情况下。解决方案的关键是借鉴交互式证明（Interactive Proof）的思想，将人机协作建模为一个“证明者”（拥有无限内部搜索能力的LLM）与一个“验证者”（资源受限的人类）之间的动态交互过程。验证者通过请求并独立核查支持性细节来积累证据，且不依赖于对模型内部状态的访问。论文证明了“任意时间有效性可靠性”（anytime-valid soundness），即只要任务本身提供了关于误通过率和人工检查错误率的可验证边界，并且这些边界在所有历史条件下均成立，则系统接受错误命题的概率始终不超过预设误差水平。此外，有限时域下的完备性要求诚实响应的充分性以及诊断进展的充足性。尽管后续检查可增强接受证据，但每次检查仍需依赖可靠的响应和人类的有效投入。最终能否实现认证，取决于验证者的资源预算、认知负荷、专业水平及疲劳程度。研究进一步揭示，在相同资源约束下，某些局部检查序列可被验证，而全局性检查却无法达成认证，从而明确了验证边界与资源消耗之间的权衡关系。
链接: https://arxiv.org/abs/2609.24895
作者: Baotong Zhang,Dean Foster,João Sedoc
机构: New York University (纽约大学); Amazon(亚马逊)
类目: Computation and Language (cs.CL)
备注: 48 pages, 3 figures
Abstract:When an LLM supplies an argument that a user could not readily construct, how can the user decide whether to accept its claim? Inspired by interactive proofs, we model human-LLM deliberation as an interaction between a prover with unrestricted internal search and a resource-bounded human verifier. The verifier requests and checks supporting details without access to the LLM’s internal state. Passed checks accumulate evidence toward an acceptance threshold. We prove anytime-valid soundness against adaptive provers: the probability of ever accepting a false claim is at most a chosen error level, provided the task supplies bounds on false passes and human checking errors that remain valid after every relevant history. A finite-horizon completeness bound additionally requires bounds on the adequacy of honest responses and sufficient diagnostic progress. Further checks can strengthen the evidence for acceptance, but each requires another adequate response and reliable human effort. Whether this tradeoff permits certification depends on the verifier’s effort budget, cognitive load, expertise, and fatigue. We identify conditions under which the supplied bounds certify a specified sequence of local checks but not a specified global check under the same resource budgets.
[NLP-10] SLICEChat: Progressive In-Encoder Token Pruning for Whole-Slide Pathology Language Models CEC
【速读】： 该论文旨在解决全切片病理图像（Whole-slide pathology images, WSIs）在滑片级多模态大语言模型（Multimodal Large Language Models, MLLMs）应用中面临的可扩展性挑战，即高分辨率图像产生的海量补丁标记（patch tokens）导致的计算开销过大问题。现有方法通常在滑片编码后才进行压缩，导致多模态注意力机制计算成本高昂。其解决方案的关键在于提出SLICEChat，一种融合渐进式标记剪枝（progressive token pruning）的混合Mamba-Transformer滑片编码器架构：利用Mamba层实现长程依赖的高效传播，同时通过Transformer层保留序列缩短过程中的全局交互信息；在编码阶段之间引入语言监督、区域感知的剪枝策略，在受控的保留率调度下移除空间上连贯但低效用的区域，从而在多模态融合前生成紧凑的滑片表征。实验表明，该方法在SlideBench VQA任务上于TCGA和BCNB队列分别达到79.84%与59.09%的准确率，优于现有滑片级病理MLLM，并在整体WSI-Bench指标上表现最优，同时保持了较低的内存占用与推理延迟，验证了其在超大规模WSI上实现高精度且计算高效的多模态推理能力。
链接: https://arxiv.org/abs/2609.24894
作者: Ali Kerem Bozkurt,Baris Cem Bakay,Ibrahim Kulac,Cigdem Gunduz-Demir,Erkut Erdem,Aykut Erdem
机构: Koç University(科奇大学), Department of Computer Engineering(计算机工程系); Koç University(科奇大学), Department of Pathology(病理学系); Hacettepe University(哈切特佩大学), Department of Artificial Intelligence and Data Engineering(人工智能与数据工程系)
类目: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
备注: Project Page: this https URL Code: this https URL
Abstract:Whole-slide pathology images (WSIs) contain gigapixel-scale visual content, creating a major scalability challenge for slide-level multimodal large language models (MLLMs). Existing approaches process thousands of patch tokens and typically apply compression only after slide encoding, leaving multimodal attention computationally expensive. We introduce SLICEChat, a slide-level MLLM that integrates progressive token pruning within a hybrid Mamba–Transformer slide encoder. Mamba layers enable efficient long-range propagation, while Transformer layers preserve global interactions as the sequence is progressively shortened. Between stages, language-supervised, region-aware pruning removes spatially coherent low-utility regions under a controlled keep-rate schedule, producing compact slide representations before multimodal fusion. On SlideBench VQA, SLICEChat achieves 79.84% accuracy on TCGA and 59.09% on BCNB cohorts, outperforming prior slide-level pathology MLLMs, and achieves the highest overall WSI-Bench metrics. It also provides competitive memory usage and the inference latency among the evaluated models. These results demonstrate accurate and computationally efficient multimodal reasoning over gigapixel WSIs.
[NLP-11] OSWorld-Pro: Process-based Evaluation for Computer Use Agents
【速读】： 该论文旨在解决当前对计算机使用代理（Computer-Use Agents, CUAs）评估过于依赖最终输出结果、缺乏过程透明性的问题。现有评估方法通常仅通过功能验证器对代理完成的最终任务成果进行评判，难以揭示代理在复杂多步骤任务中失败的具体原因与机制，从而限制了对模型性能的深入优化。其解决方案的关键在于提出OSWorld-Pro——一个包含超过300个任务、2800余个子目标，并基于超过67,000条人类标注数据构建的程序化评估基准。通过引入经过严格对齐的人类偏好大语言模型（LLM-Judges），该框架能够对每个子目标的执行过程进行精细化评估，实现对CUA在一系列顺序依赖子目标中进展的全程追踪。实验表明，即使是最先进的大模型如Claude Opus 5，在OSWorld-Pro上的表现也仅为75.7%（相较原OSWorld的83.4%），且识别出多种以过程为导向的关键失败模式（如与子目标无关的动作、基于点击的输入错误等），为提升CUA的鲁棒性与效率提供了可操作的改进方向。
链接: https://arxiv.org/abs/2609.24890
作者: Zhilin Wang,Shaokun Zhang,Yifan Zhang,Hao Zhang,Jin Xu,Binfeng Xu,Jian Hu,Yunheng Zou,Karan Sapra,Andrew Tao,Jan Kautz,Yi Dong
机构: NVIDIA
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 27 pages, 7 figures
Abstract:Evaluation of Computer-Use Agents (CUAs) is often limited to the final deliverables they create (at the end of hundreds of steps) and assessed with functional verifiers, as seen in OSWorld. However, such evaluation of end-state performance lacks transparency into how and why agents fail in various tasks, obfuscating critical insight for subsequent improvement. For instance, agents that err during keyboard inputs would require a different mitigation strategy from those that fail to precisely provide click-based inputs on the graphical UI. We introduce OSWorld-Pro: a set of over 300 tasks containing over 2800 subgoals to enable the procedural evaluation of CUAs grounded in over 67,000 human annotations. We use robust human-aligned LLM-Judges to evaluate the fulfillment of OSWorld-Pro subgoals and thereby reveal the progress that models make throughout a series of sequentially dependent subgoals. Our findings reveal that OSWorld-Pro is challenging even for state-of-the-art LLMs, with top performers like Claude Opus 5 achieving only 75.7% vs. 83.4% on OSWorld. Furthermore, we identify critical process-focused failure modes of various models (e.g. subgoal-irrelevant actions and click-based mistakes) to provide insights to improve performance and efficiency of CUAs.
[NLP-12] he Copy Ceiling: An Input-Exposure Control for Ontology-Grounded Generation over Curated Corpora
【速读】： 该论文旨在解决在基于图检索的语料库问答任务中，语言模型的“接地性”（grounded recall）评估存在误导性的问题：由于上下文可能已直接暴露答案，导致高召回率并不能反映模型真正的推理能力。其核心解决方案是提出“暴露会计”（exposure accounting），通过分类每个正确答案项是否在展示的上下文中被暴露，并判断模型是否能够从中恢复该答案，从而建立一个可量化的基准——“复制上限”（copy ceiling），即仅通过逐字复制上下文所能达到的召回率。以该确定性、无需人工评判的基线为参照，计算模型召回率相对于复制上限的有符号增益（signed gain），从而揭示模型实际超越简单复制的能力。实验结果表明，尽管接地召回率高达0.92，但所有模型的增益均为负值（-0.067至-0.022），且仅有极少数未暴露项获得词法认可，经分层人工审计后发现，绝大多数被赋予信用的答案确实满足请求关系，而所有未暴露项的信用均在关系判定中失败。此外，当问题重述脱离图结构词汇时，暴露率从0.964降至0.328，而触发回退机制的案例极少。研究还表明，虽改进生成方式可提升评分，但无法证明内容特异性。因此，暴露会计提供了一种持续性的评估控制机制，有效区分了因上下文暴露导致的答对与真正超出暴露范围的恢复能力，但并不用于判断是否存在推理行为。
链接: https://arxiv.org/abs/2609.24885
作者: John J. O’Hare
机构: DreamLab AI(梦实验室人工智能); VisionFlow neurosymbolic stack
类目: Computation and Language (cs.CL); Computers and Society (cs.CY)
备注: 28 pages, 6 figures
Abstract:When a language model answers from a curated corpus via graph-based retrieval, a large grounding uplift does not establish reasoning over the retrieved structure: the context may already expose the gold answers. We propose exposure accounting, which classifies each gold item by whether the shown context exposes it and whether the answer recovers it. Its scalar reference is the copy ceiling, the recall a verbatim copy of the context achieves; signed gain over copy measures the model’s recall relative to this deterministic, judge-free baseline. Across ten models, unaided recall averages 0.26 and grounded recall 0.92, yet gain over copy is uniformly negative (-0.067 to -0.022). Of 11,360 gold-item observations, representing 1,136 target instances evaluated under ten models, only three unexposed items receive lexical credit. A stratified model-judged audit of 423 observations, with a symmetric quotation-verification policy, estimates that 97.1% of credited items assert the requested relation; all three unexposed credits fail relational adjudication. On targets the scaffold does not expose, lexical recovery falls from 0.121 unaided to 0.004 grounded; adjudication validates 71 of the 92 unaided credits and none of the three grounded credits, without establishing full-frame relational recovery rates. Rephrasing questions outside the graph’s title vocabulary reduces exposure from 0.964 to 0.328, while an absence-triggered fallback activates on only 2 of 506 questions. A paired production study improves judged quality by +0.27 pooled, but negative controls do not establish content specificity beyond a well-formed on-corpus block. These results support exposure accounting as a standing control for corpus-derived evaluations. The accounting distinguishes exposed-item omissions from beyond-exposure recoveries; it does not determine whether reasoning occurred.
[NLP-13] Decomposing Error and Style in Automated Clinical Coding
【速读】： 该论文旨在解决自动化临床编码（automated clinical coding）中因单一黄金标注导致的评估偏差问题。在实际临床实践中，不同编码团队对同一份病历文本的编码一致性较低（仅73%的代码重合度，Jaccard相似度），即便经过独立临床审计后，一致性也仅提升至77%，表明这种差异并非单纯由错误引起，而是存在系统性因素。为此，论文提出将系统性差异建模为“编码风格”（coding style, ψ），即编码员或机构特有的编码偏好与文档详尽程度策略，并将编码过程重新定义为条件概率模型 $ p(\text{code} \mid \text{note}, \psi) $，通过一个10维评分体系估计编码风格。实验表明，若模型条件于与数据匹配的编码风格，ICD F1得分可提升最多26分；而风格严重不匹配时，F1得分则下降最多21分。四种基于提示（prompt-based）的编码方法在未考虑风格时F1值为39–49，引入风格信息后均收敛至52–56（所有p < 0.05）。这说明传统单黄金标注评估中被归因于模型误差的性能差距，实则很大程度上源于未建模的编码风格，其可通过风格建模有效恢复。
链接: https://arxiv.org/abs/2609.24877
作者: Han-Chin Shing,Jack Moriarty,Ryan Ware,Afton Marchbanks,Carlyn Canvasser,Stefanie Higgins,Harsh Gupta,Fang Wang,Joseph Paul Cohen
机构: Amazon(亚马逊)
类目: Computation and Language (cs.CL)
备注:
Abstract:In automated clinical coding, where the label space spans tens of thousands of diagnosis and procedure codes, models are currently evaluated against a single gold annotation, treating any deviation as error. But we find when two teams code the same 110 ACI-Bench encounters, they agree on only 73% of codes (Jaccard similarity) for the same note; even after an independent clinical audit removes erroneous codes, agreement rises only to 77%. Is that gap error or something systematic? We model the systematic component as coding style \psi , a coder- or site-specific policy over what to code and how much to document, and recast coding as p(\mathrmcode\mid\mathrmnote,\psi) , estimating \psi with a 10-dimension rubric. If style were noise, conditioning on it would do nothing. Instead, across five datasets a model conditioned with a data-matching style raises ICD F1 by up to 26 points and an extreme mismatched one lowers it by up to 21. Four prompt based coding methods spanning 39-49 F1 converge to 52-56 once style is supplied (All p0.05). Much of what single-gold evaluation charges to model error is recoverable, unmodeled style.
[NLP-14] Extracting Arguments Not Just Classifying Them: Instruction-Tuned LLM s for Generative Component Detection
【速读】： 该论文旨在解决论点挖掘（Argumentation Mining, AM）中的论点组件识别（Argumentative Component Detection, ACD）任务，其核心挑战在于需联合界定论点片段并将其分类为论点（claims）与前提（premises）等组件。现有方法多将ACD简化为序列标注或分步的分割-分类流水线，难以充分捕捉组件间的复杂语义关系。本文提出一种基于指令微调大语言模型（Instruction-Tuned Large Language Models, LLMs）的新型方法ITFACD，通过设计紧凑的指令式提示（instruction-based prompts），将ACD重新建模为语言生成任务，从而直接从原始文本中生成完整的论点结构，无需依赖预分割的组件。该方案的关键在于利用指令微调的LLMs实现对复杂论点结构的端到端生成能力，显著提升了对论点组件的识别精度。实验结果表明，ITFACD在标准基准上超越了现有最先进系统，是首次完整将ACD建模为生成式任务的尝试，凸显了指令微调在复杂论点挖掘问题中的潜力。
链接: https://arxiv.org/abs/2609.24855
作者: Sofiane Elguendouze(UniCA, I3S, MARIANNE),Erwan Hain(UniCA, MARIANNE),Elena Cabrio(MARIANNE, UniCA),Serena Villata(CNRS, MARIANNE)
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Argumentative component detection (ACD) is a core subtask of Argument(ation) Mining (AM) and one of its most challenging aspects, as it requires jointly delimiting argumentative spans and classifying them into components such as claims and premises. While research on this subtask remains relatively limited compared to other AM tasks, most existing approaches formulate it as a simplified sequence labeling problem, component classification, or a pipeline of component segmentation followed by classification. In this paper, we propose ITFACD, a novel approach based on instruction-tuned Large Language Models (LLMs) using compact instruction-based prompts, and reframe ACD as a language generation task, enabling arguments to be identified directly from plain text without relying on pre-segmented components. Experiments on standard benchmarks show that our approach achieves higher performance compared to state-of-the-art systems. To the best of our knowledge, this is one of the first attempts to fully model ACD as a generative task, highlighting the potential of instruction tuning for complex AM problems. Our code and the datasets used are openly available in the following GitHub repository.
[NLP-15] he Answer-Basin Representation Hypothesis: We Are Not Probing or Steering Concepts
【速读】： 该论文旨在解决生成式语言模型中高层概念与模型内部表示之间线性结构的组织机制不明确的问题。现有“线性表征假说”虽指出高层概念可对应于模型中的特定方向，但未阐明这些概念相关线性结构在模型内部如何系统性地分布与关联。为此，论文提出“答案盆地表征假说”（Answer-Basin Representation Hypothesis），其核心在于：模型后续生成分布所诱导的答案概率测度，通过共享于不同问题间的线性方向，组织并表征了这些概念相关的线性结构。具体而言，所有生成相同答案的文本延续构成一个“答案盆地”（answer basin），其总概率质量定义了该答案的边际概率。该概率测度的统计特性由模型在不同问题上的回答分布决定，而非直接依赖于概念标签的显式变化。研究表明，概念一致性效应及其反转现象在探针（probing）与控制（steering）任务中的表现，本质上源于概念标签与答案概率测度之间的对齐程度。因此，解决方案的关键在于将概念相关线性结构的形成归因于答案概率测度的差异，而非概念标签本身的改变，从而揭示了语言模型中高层语义表征的内在组织原则。
链接: https://arxiv.org/abs/2609.24821
作者: Manjiang Yu,Hongji Li,Zihan Wang,Junwei Chen,Xue Li,Priyanka Singh,Yang Cao,Lijie Hu
机构: The University of Queensland(昆士兰大学); Mohamed bin Zayed University of Artificial Intelligence(穆罕默德·本·扎耶德人工智能大学); Institute of Science Tokyo(东京科学研究所)
类目: Computation and Language (cs.CL)
备注: 19 pages, 7 figures
Abstract:The Linear Representation Hypothesis associates high-level concepts with directions in language models, but it remains unclear how these concept-related linear structures are organized within the model. We propose the Answer-Basin Representation Hypothesis: the probability measure induced over answers by the model’s continuation distribution organizes these linear structures, with its statistics represented along linear directions shared across questions. All continuations yielding the same answer form an answer basin, whose mass is their total probability. These basin masses define the pushforward probability measure over answers. We posit that concept-related linear structure emerges from differences in the answer measure rather than being determined by changes in concept labels. Experiments across models and tasks link concept-consistent effects and their reversals in probing and steering to the alignment between concept labels and the answer measure.
[NLP-16] MSI-Bench: Evaluating Multi-Speaker Voice Interaction for Collaborative AI Agents
【速读】： 该论文旨在解决多说话人语音交互场景中智能体（AI agents）能力评估的缺失问题，尤其针对会议、家庭和协作工作等存在多个参与者且互动复杂的实际应用场景。现有研究多集中于一对一语音交互，难以覆盖多说话人情境下特有的挑战，如说话人身份识别、跨轮次记忆保持、指令遵循与推理中的角色归属等问题。为此，作者提出了多说话人交互基准测试（Multi-Speaker Interaction Benchmark, MSI-Bench），包含1,152个测试用例，涵盖中英文各576例，每条用例均为带有参与者上下文、预期工具调用及原子级评估标准的短时多轮音频场景。该基准聚焦三大核心能力维度：多说话人记忆、多说话人指令遵循与多说话人推理。实验结果表明，当前最强模型在英文和中文任务上的通过率分别仅为66.8%和54.5%，而开放权重模型表现更差，分别为34.0%和19.3%。失败分析揭示了两大瓶颈：开放权重模型受限于多说话人音频前端处理性能；前沿系统即便在干净转录文本上仍难以实现基于说话人角色的决策判断，且普遍存在未被提问却擅自响应的“对话越界”现象。因此，论文指出未来语音智能体的发展需重点突破说话人锚定感知、说话人作用域内的决策能力以及对话节制性（conversational restraint）三大关键技术目标。
链接: https://arxiv.org/abs/2609.24812
作者: Chenxu Xiong,Dongming Shen,Yuzhi Tang,Wentao Ma,Mu Li,Alex Smola
机构: 未知
类目: Computation and Language (cs.CL)
备注: 23 pages, 6 figures, 5 tables. Dataset: this https URL ; Code: this https URL
Abstract:Voice provides a natural and immediate interface for AI agents. Many settings in which voice agents could be useful, including meetings, households, and collaborative work, are inherently multi-speaker. Supporting these settings introduces challenges that are largely absent from one-on-one interaction. We introduce the Multi-Speaker Interaction Benchmark (MSI-Bench) for evaluating multi-speaker voice interaction. Each test case is a short multi-party multi-turn audio scene with participant context, expected tool calls, and atomic rubrics. The benchmark targets three capability families: multi-speaker memory, multi-speaker instruction following, and multi-speaker reasoning. It comprises 1,152 test cases, evenly split between Mandarin Chinese and English (576 each). The strongest configuration on each split passes all rubrics on only 66.8% of English and 54.5% of Mandarin cases, and the strongest open-weight configuration on 34.0% and 19.3%. Failure analysis separates perception from reasoning: open-weight models are bottlenecked by the multi-speaker audio front-end, while frontier systems still fail speaker-scoped decision making on clean transcripts—and models across the board often respond when no one has addressed them. These results identify speaker-grounded perception, speaker-scoped decision making, and conversational restraint as concrete targets for future voice agents.
[NLP-17] When Quantization Preserves Accuracy but Not Evidence: Explanation-Aware Post-Training Quantization for Medical LLM s
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在解释关键型应用场景（如医疗多选题问答）中，后训练量化（Post-training Quantization, PTQ）导致生成推理过程（rationales）中支持答案的关键证据信息丢失的问题。传统PTQ方法仅关注最终答案的准确率或通用重建性能，但在需要用户可信赖解释的场景下，仅保留答案准确性不足以确保模型输出的可信性。其解决方案的关键在于提出一种面向解释感知（explanation-aware）的优化目标，通过构建基于高精度教师模型推理过程的“忠实性缓存”（faithfulness cache），在量化过程中显式保留与答案相关联的证据性词元（evidence tokens）及其条件化答案行为。该方法在W4A4KV4量化配置下应用于OSTQuant框架，实验证明，尽管基准方法能保持任务准确率，却显著削弱了答案支持性推理能力；而所提方法则更有效地保留了原始全精度模型的答案支持行为和推理到答案的支撑关系。研究强调，在解释关键型任务中，评估量化模型应不仅关注答案准确性，还应评估其对答案支持证据的保真度。
链接: https://arxiv.org/abs/2609.24799
作者: Yeji Kim,Mi-Young Kim,Randy Goebel
机构: University of Alberta(阿尔伯塔大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Post-training quantization (PTQ) enables efficient deployment of large language models, and PTQ methods are usually optimized and evaluated with generic reconstruction, perplexity, or answer accuracy. But in explanation-critical domains, preserving only the final answer may be insufficient, since users may also inspect generated rationales to judge whether a prediction is trustworthy. We study this issue in medical multiple-choice question answering, where rationales should provide evidence that supports the selected answer. We propose an explanation-aware objective for transformation-based PTQ. Our method builds an offline faithfulness cache from full-precision teacher rationales and uses it during optimization to preserve answer-supporting evidence tokens and evidence-conditioned answer behavior. We instantiate it on OSTQuant under W4A4KV4 quantization and evaluate four 7B–8B medical and instruction-tuned LLMs on MedExQA, MedExpQA, and ChallengeClinicalQA. While a same-calibration OSTQuant baseline preserves task accuracy, it can substantially weaken answer-supporting rationales. Our objective is to preserve the full-precision model’s answer-supporting behavior rather than improve gold-label accuracy, and our method better preserves the full-precision model’s answer behavior and rationale-to-answer support. These results suggest that PTQ for explanation-critical settings should evaluate preservation of answer-supporting evidence, not only answer accuracy. Code and evaluation scripts are available at this https URL. Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI) Cite as: arXiv:2609.24799 [cs.CL] (or arXiv:2609.24799v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2609.24799 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-18] Adapting Tree-Structured Speculative Decoding to DeepSeek -V4 for Efficient Inference
【速读】： 该论文旨在解决大语言模型（LLM）在自回归解码过程中因反复执行目标模型而导致的推理延迟问题。其核心挑战在于如何在有限资源预算下提升解码效率，尤其是在采用树状结构推测（tree-structured speculation）时，需应对多分支路径间状态一致性被破坏的问题。针对DeepSeek-V4模型中基于压缩自注意力（CSA/HCA）的在线压缩机制导致不同分支在共享前缀后状态分化、进而破坏跨分支一致性的问题，论文提出了一种融合树状推测解码的解决方案：通过引入分支感知的因果验证（branch-aware causal verification）、临时状态隔离（temporary state isolation）以及已接受路径的状态刷新（accepted-path state refresh），有效维持了各分支间验证过程与压缩状态更新的一致性。实验结果表明，在不同预算（D=5–8）、批处理大小（1–64）及多个数据集（GSM8K、MBPP、ShareGPT）下，该方法均显著提升了接受长度（accepted length），并实现了高达约18.5%的吞吐量增益，尤其在中等批量和不可预测性较高的任务中表现更优。此外，性能提升呈现出可迁移的规律性：相对收益随预算增加而上升，且在达到一定预算后吞吐量趋于饱和，而接受长度仍持续增长，揭示了树状推测在高效利用上下文压缩机制方面的潜力。该研究为未来面向压缩、稀疏或结构化上下文表示的模型适配推测解码提供了重要实践经验。
链接: https://arxiv.org/abs/2609.24698
作者: Changxu Liu,Zhaogeng Li
机构: Baige AI Team, Baidu Inc.(百度公司); Fudan University(复旦大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Repeated execution of the target model during autoregressive decoding is a major source of LLM inference latency. Unlike linear speculation, which follows a single candidate chain, tree-structured speculation retains multiple branches from shared prefixes; under the same budget, this broader coverage can improve acceptance and efficiency. Adapting it to DeepSeek-V4 is nontrivial: its CSA/HCA online compressed attention concentrates the difficulty on the target-verify side, where branches diverging from a shared prefix compress into different states, breaking cross-branch state consistency. We integrate tree-structured speculative decoding into the DeepSeek-V4-Flash pipeline via branch-aware causal verification, temporary state isolation, and accepted-path state refresh, keeping verification and compressed-state updates consistent across branches. Across budgets D=5 to D=8, batch sizes 1 to 64, and three datasets (GSM8K, MBPP, ShareGPT), tree speculation achieves a higher accepted length than the matched linear configurations in all settings (e.g., at D=8 about 2.83–3.41 versus 2.39–2.84) and improves throughput in nearly all configurations—marginal only at the smallest budget—by up to about 18.5%. More importantly, the gains follow stable, transferable regularities: the relative gain grows with the budget and is most pronounced for less predictable workloads at small-to-medium batch sizes, while beyond a certain budget throughput plateaus and decouples from the still-rising accepted length. These results show that retaining multiple candidate paths under the same budget can effectively improve DeepSeek-V4 decoding efficiency, and offer experience for adapting speculative decoding to future models with compressed, sparse, or structured context representations.
[NLP-19] Muon Can Outperform Dedicated Continual Learning Methods
【速读】： 该论文旨在解决持续学习（Continual Learning, CL）中因参数更新导致的遗忘问题，尤其关注在使用低秩适配器（Low-Rank Adapters, LoRA）时，如何有效控制新任务更新对历史权重的影响。传统方法通过损失函数施加约束以减少新旧更新之间的重叠，但此类方法通常缺乏对更新能量分布的精细调控，限制了模型适应新任务的灵活性。本文提出一种无需任务感知的通用更新约束机制——通过在增量式LoRA（IncLoRA）中引入Muon优化器，实现每次更新的正交化，从而在不依赖特定损失项的前提下，使更新能量更均匀地分布在更多方向上。实验结果表明，IncLoRA+Muon在标准持续学习基准（Standard CL）上达到与专用方法相当的性能，并在TRACE基准上优于所有AdamW配置；更重要的是，研究发现仅需一个更新约束机制（无论来自损失还是优化器）即可有效缓解遗忘，额外增加约束反而降低性能（如在最严格条件下损失8.4个百分点），且损害模型对各任务的可塑性。关键发现在于：不同优化器的核心差异并非更新幅度大小（Muon下更新量为AdamW的0.91–2.06倍），而在于其更新方向的几何分布——AdamW将更新集中于1.4–1.8个有效奇异方向，而Muon则将其分散至7.0个方向，且二者无重叠。因此，以往归因于专用持续学习方法的优势，部分可由优化器本身的更新几何特性解释。
链接: https://arxiv.org/abs/2609.24678
作者: Sebastian George Sincari(1),Bogdan Alexandru Gheorghe(1),Antonio Barbalau(2) ((1) Faculty of Mathematics and Computer Science, University of Bucharest, Bucharest, Romania, (2) Bitdefender, Bucharest, Romania)
机构: University of Bucharest (布加勒斯特大学); Bitdefender(比特防御)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 10 pages, 2 figures, 6 tables. Presented at the 5th Conference on Lifelong Learning Agents (CoLLAs), Work-in-Progress Track, 2026. Sebastian George Sincari and Bogdan Alexandru Gheorghe contributed equally
Abstract:Continual learning with Low-Rank Adapters (LoRA) typically mitigates forgetting by penalizing the overlap between a new update and the accumulated past weights, which discourages certain update directions without controlling how an update distributes its energy over the ones that remain. We ask whether that restriction has to be task-aware, or whether a generic one supplied by the optimizer is enough. We train a plain incremental LoRA (IncLoRA) with Muon, which orthogonalizes each update, and compare it against O-LoRA and ELLA over five seeds and three task orders on the Standard CL Benchmark and three seeds on TRACE. IncLoRA+Muon reaches the accuracy band of the dedicated methods on Standard CL and improves on every AdamW configuration on TRACE. One update-constraining mechanism is enough, whether it comes from the loss or from the optimizer; on Standard CL a second one does not help, and for the most restrictive method it costs 8.4 points of accuracy and the plasticity to fit each task. What separates the two optimizers is not the size of the update, which under Muon is 0.91 to 2.06 times that under AdamW, but how it is distributed. AdamW confines it to between 1.4 and 1.8 effective singular directions, Muon spreads it over 7.0, and the two do not overlap in any tracked run. Part of the advantage usually attributed to dedicated CL methods may therefore be explained by the geometry of the optimizer’s updates.
[NLP-20] Assessing Readability with LLM s: The Role of Reasoning and Few-Shot Prompting
【速读】： 该论文旨在解决传统可读性评估方法在跨文体、跨语言场景下泛化能力不足，以及监督学习模型因依赖稀缺领域标注语料而适用性受限的问题，尤其针对低资源语言的可读性评估挑战。其核心解决方案在于探索无需任务特定训练的开源大语言模型（LLM）在多语言可读性评估中的潜力，并系统验证先进提示工程策略的有效性。关键发现包括：采用链式思维（Chain-of-Thought, CoT）提示和以推理为导向的模型能显著提升预测性能，相比直接回答方式具有明显优势；同时，仅需每个类别提供一个标注示例（1-shot）即可大幅改善零样本（zero-shot）设定下的表现，且后续增加示例带来的收益递减。研究通过对比传统无监督指标与最先进的监督基线，证明了无需微调的大语言模型作为跨语言可读性评估工具的可行性与鲁棒性。
链接: https://arxiv.org/abs/2609.24650
作者: Raphaël Thieffry,Matej Martinc
机构: Université Paris-Saclay(巴黎萨克雷大学); Jožef Stefan Institute(约瑟夫·斯特凡研究所)
类目: Computation and Language (cs.CL)
备注:
Abstract:Readability assessment is essential for tailoring texts to intended audiences across educational, healthcare, and information retrieval domains. However, traditional readability formulas struggle to generalize across genres and languages, while supervised machine learning models rely on scarce, domain-specific annotated corpora, limiting their applicability–particularly for less-resourced languages. Large Language Models (LLMs) offer a highly scalable, multilingual alternative that requires no task-specific training, yet the impact of advanced prompting strategies on their performance remains underexplored. In this paper, we conduct a systematic benchmark of diverse open-source LLMs for multilingual readability assessment, focusing on the prediction of discrete readability levels required by educational frameworks. In addition to English, we evaluate our approach on a less-resourced language, Slovenian, to establish whether LLMs remain effective in low-resource settings. Specifically, we investigate the influence of explicit reasoning, demonstrating that Chain-of-Thought (CoT) prompting and reasoning-oriented models yield significant improvements over direct answering. Furthermore, our exploration of few-shot in-context learning reveals that providing just one labelled example per category (1-shot) substantially enhances prediction quality compared to zero-shot settings, with additional examples offering diminishing returns. By comprehensively comparing these approaches against traditional unsupervised metrics and state-of-the-art supervised baselines, we establish the viability of out-of-the-box LLMs as robust, cross-lingual readability assessors.
[NLP-21] Written as a Record Read as an Address: What a Forward Pass Leaves in an Operations KV Cache
【速读】： 该论文旨在解决语言模型在执行操作指令（如“交换盒子F和盒子B的内容”）时，如何在前向传播过程中将状态信息写入键值缓存（KV cache），以及后续推理阶段如何从缓存中恢复这些状态信息的问题。其核心问题是：模型在处理操作语句时，究竟是以何种方式编码和存储实体绑定关系（entity bindings），并且这些信息是否能够被独立的读取模块可靠地重构。解决方案的关键在于采用“隔离训练”范式，将前向传播过程分解为一个无梯度更新的“写入者”与一个仅接收指令和操作标记、完全隐藏状态描述的“读取者”。通过这种设计，研究发现操作标记所对应的缓存区域存在两种访问模式：一是因果性重定向的路由记录，用于决定读取哪个可见状态；二是直接访问操作所读取的值（即数据载荷），该载荷可通过单个操作数名称标记在特定深度层（如Llama-3.1-8B的第12–15层、Mistral-7B的第14–17层）中被精准定位。这一机制表明，操作令牌在缓存中留下了局部化且可因果恢复的痕迹，既支持状态路由又实现对数据载荷的直接访问，而写入模型主要依赖地址信息而非实际值进行决策。
链接: https://arxiv.org/abs/2609.24635
作者: Lingfeng Wu,Behzad Shomali
机构: University of Bonn (波恩大学); Lamarr Institute (拉马尔研究所)
类目: Computation and Language (cs.CL)
备注:
Abstract:When a language model reads an operation such as “Swap the contents of Box F and Box B”, its forward pass writes keys and values for those tokens into the KV cache. Prior work on entity tracking establishes what models use: bindings are resolved at query time rather than stored as explicit latent state. We ask what they write at the operation span and how it is accessed. We split a forward pass into a frozen writer and a reader: the writer’s cache is recomputed without gradients, while the reader sees only the instruction and operation tokens, with all state descriptions hidden, and is trained in isolation. Anything the reader recovers was therefore already present in the unmodified cache. On a synthetic boxes task, a base reader recovers \leq 0.06 of queried bindings against 0.75 – 1.00 after training, and recoverability tracks the operation’s read/write footprint. We find two modes of access. Across Llama-3.1-8B and Mistral-7B, operation-span transplants causally redirect which visible state is read even when the two worlds hold identical values, revealing a routing record. Isolation training preserves routing and adds direct access to the payload, the value the operation read, from the single operand-name token in a narrow mid-depth band (layers 12–15 of 32 in Llama-3.1-8B, 14–17 in Mistral-7B) — the same site that holds the routing record. The same recipe extends to further operations, ToMi and GSM8K, but is bounded by training coverage and costs open-book accuracy. Operation tokens thus leave localized, causally recoverable records that support both routing and direct payload access, though the model that writes them reads mainly the address they carry and not the value.
[NLP-22] Custom Named Entity Recognition and Topic Classification for Global Health Publications
【速读】： 该论文旨在解决在标注数据和计算资源有限的全球健康文献环境中，如何选择与适配自然语言处理（Natural Language Processing, NLP）模型的问题。其核心挑战在于平衡模型性能与实际部署可行性之间的矛盾。解决方案的关键在于通过系统性实验，评估不同模型架构在特定任务中的表现，并提出轻量级且高效的适配策略：首先，在语义标签发现任务中，验证了大规模语料虽能提升词汇覆盖率，但未必增强领域相关性关联；其次，在命名实体识别（Named Entity Recognition, NER）任务中，发现基于Transformer的RoBERTa模型虽显著提升微平均F1值（0.80），但推理耗时高达82秒，远高于卷积神经网络（convolutional）模型的5–6秒，因此提出对轻量级spaCy模型进行微调，并集成一个疾病识别器（Disease Recognizer），在NCBI Disease Corpus上达到81.33%的测试F1，实现高效率与良好性能的结合；最后，在多标签主题分类任务中，对比了MiniLM的少样本学习与BART-MNLI的零样本推理，结果显示后者在单标签准确率（95.2%）上优势明显，但其高昂的推理成本限制了实际应用。综合来看，该研究明确了在资源受限场景下，通过领域特化与轻量化适配可显著提升实用性，而仅当Transformer模型带来的性能增益足以覆盖其计算开销时，才应考虑引入。这一实证结果为构建高效、可持续的知识系统提供了关键指导。
链接: https://arxiv.org/abs/2609.24625
作者: Genis Skura,Antoine Geissbühler,Jean-Luc Falcone
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:How should natural language processing models be selected and adapted for global health literature in environments where annotated data and computational resources are limited? This thesis investigates these challenges through experiments on semantic tag discovery, named entity recognition (NER), and multi-label topic classification. First, skip-gram word2vec models trained on progressively larger specialized corpora are compared with BioWordVec to assess how corpus size and domain context influence tag discovery. Vocabulary coverage and qualitative evaluation indicate that broader coverage does not necessarily yield more useful domain-specific associations. The analysis then turns to entity extraction, comparing convolutional spaCy models with a RoBERTa-based transformer on 1,000 annotated sentences. Under a lenient scoring protocol, the transformer achieves 0.80 micro-F1 versus 0.65-0.69 for convolutional models, but takes 82 seconds rather than 5-6 seconds. This trade-off motivates fine-tuning convolutional models and integrating a disease recognizer that achieves 81.33% test F1 on the NCBI Disease Corpus. Combined with PDF preprocessing, entity filtering, and MeSH enrichment, the resulting pipeline supports document-level indexing. To complement entity extraction with thematic annotation, MiniLM-based few-shot classification is compared with BART-MNLI zero-shot inference across 50 topics and 1,000 handcrafted test sentences. BART-MNLI achieves 95.2% single-label accuracy versus 59%; reported multi-label accuracies are 88% and 32% under partly manual assessment. However, its higher inference cost limits practical integration. The results show where domain specialization and lightweight adaptation offer practical value, and where transformer accuracy justifies higher inference costs, providing an empirical basis for building knowledge systems under resource constraints.
[NLP-23] Evaluating Decision Models for Text Annotation in Computational Social Science
【速读】： 该论文旨在解决生成式人工智能（Generative AI）在计算社会科学文本标注中广泛应用背景下，其标签可靠性与模型置信度可信性的问题。随着研究结论的 validity 依赖于大语言模型（LLM）生成的标注结果，如何评估新型决策模型（Decision Models）在分类任务中的准确性及其置信度校准能力成为关键挑战。本文的核心解决方案在于：通过在18个计算社会科学分类任务（共7,977个样本）上，采用零样本（zero-shot）协议，对首个商用决策模型及两个开源版本与19个前沿及开源语言模型进行系统对比。结果显示，尽管决策模型在15项任务中有14项表现逊于最优的LLM（中位宏F1差距达11.6个百分点），但其推理成本仅为前沿模型的1/44；更重要的是，其置信度校准性能优于16个LLM的语义化置信表达，虽有三个前沿模型在中位校准误差上更优（0.157 vs. 0.066），但在置信度高于0.9的样本中，决策模型仍保持较高的标注准确率（中位0.815）。尤其值得注意的是，在“同伴支持对话中的同理心”这一任务中，模型虽表现出高置信度却仅达到随机水平，揭示其局限性。总体而言，研究证明决策模型可作为标注流程的初步筛选工具——将低置信度样本转交至LLM处理，可在四分之一至一半的成本下实现不低于直接使用LLM的效果，因而具备显著的实用价值。
链接: https://arxiv.org/abs/2609.24574
作者: Hazem Ibrahim,Yasir Zaki
机构: New York University Abu Dhabi (纽约大学阿布扎比分校)
类目: Computation and Language (cs.CL); Computers and Society (cs.CY)
备注: 47 pages, 7 figures
Abstract:Computational social science increasingly relies on large language models for text annotation, and the validity of published findings now rests on the labels generated by such models. Decision models, a new model class built for categorical question answering, answer typed questions with a choice, a probability distribution over the label set, and a confidence score rather than free text, at a small fraction of frontier inference prices. Whether their answers are accurate, and whether that stated confidence can be trusted on social science constructs, are unknown. Here, we mirror the evaluation of Ziems et al. (2024) on 18 computational social science classification tasks (7,977 items), comparing the first commercial decision model and two open-weight counterparts against 19 frontier and open-weight language models under the same zero-shot protocol. The decision model trails the per-task best LLM on 14 of 15 evaluation tasks, with a median deficit of 11.6 macro-F1 points, at a median 44 times lower measured cost. Its confidence is better calibrated than the verbalized confidence of 16 of the 19 LLMs, yet three frontier models show lower median calibration error (0.157 against 0.066). While items above 0.9 confidence are typically labeled accurately (median accuracy 0.815), on one task, empathy in peer-support dialogues, the model reports high confidence while performing near chance. Nonetheless, our results suggest that decision models are useful as a first step in the annotation pipeline: routing low-confidence items to an LLM matches or exceeds the LLM alone at a quarter to half of its cost.
[NLP-24] oward a Unified Mathematics of Concepts
【速读】： 该论文旨在解决认知科学、心理学与人工智能（AI）领域中缺乏统一数学语言来表征概念的根本性问题。尽管现代系统将概念表示为向量、分布、符号、图等结构，但这些形式化方法常被视为相互竞争而非共同应对同一核心问题的解决方案。其核心解决方案在于提出一种基于操作的视角，即通过评估不同数学框架所支持的概念操作（如相似性、组合、泛化与具身化等）来衡量其有效性。研究识别出十三种跨领域反复出现的核心概念操作，并发现十类现有框架分别体现了对概念作为自包含内容、关系结构或动态过程的不同理论承诺，而这些承诺决定了各框架在支持特定操作时的自然程度。例如，向量模型擅长处理渐进式相似性和泛化，但在显式组合方面表现不佳；符号模型虽能有效支持组合，却难以实现良好泛化。实证研究表明，即使面对相同概念问题，不同理论因数学承诺差异导致程序与结果迥异，凸显了形式化基础对理论解释力的关键影响。因此，论文呼吁发展融合内容、关系与过程三者优先性的混合形式化体系，以更全面地建模概念的本质。
链接: https://arxiv.org/abs/2609.24554
作者: Chen Shani
机构: Tel Aviv University (特拉维夫大学); Israel (以色列)
类目: Computation and Language (cs.CL)
备注:
Abstract:Concepts are commonly defined as abstract, compact representations of knowledge and treated as basic units of intelligent behavior. Yet, cognition, psychology, and AI lack a shared mathematical language for them. Modern systems represent concepts as vectors, distributions, symbols, graphs, and other structures, but these formalisms are typically treated as competing rather than as solutions to a common problem. We propose an operation-based view that evaluates mathematical frameworks by the conceptual operations they support, identifying thirteen operations (including similarity, composition, generalization, and grounding) that recur across cognition, psychology, and AI. We show that ten frameworks embody distinct commitments to concepts as self-contained content, relational structure, or evolving process, and that these commitments determine which operations each supports naturally. For example, vector-based models facilitate graded similarity and generalization but struggle with explicit composition, whereas symbolic models support composition but offer but generalize poorly. No single framework we examined naturally supports all operations without extension. We test this account empirically using categorization as a case study, operationalizing nine theories on the same items against human judgments. Despite addressing the same conceptual question, the theories produce different procedures and results, demonstrating that mathematical commitment shapes what a theory can explain. We call for hybrid formalisms that treat content, relation, and process as jointly primary.
[NLP-25] QLoRA Fine-Tuning of Ministral LLM for Sequence-to-Function Protein Annotation
【速读】： 该论文旨在解决新测序蛋白功能注释效率低下的核心问题，即公共数据库中序列数量的增长远超人工注释的处理能力。传统计算方法将注释视为固定本体（ontology）上的多标签分类任务，导致预测受限于预定义的标签集合，难以灵活捕捉生物学知识的多样性。本文提出将蛋白质功能注释建模为序列到文本的生成任务，采用QLoRA（4-bit NF4量化结合低秩适配器）技术对30亿参数的Ministral 3基础模型进行微调，利用序列-注释对数据训练模型以生成自然语言形式的功能描述。通过“大模型作为专家”（LLM-as-expert）评估协议，使用GPT模型模拟资深分子生物学专家对物种识别（二分类）和功能注释质量进行评分，验证了经QLoRA微调的紧凑型大模型能够生成具有真实生物学价值的类专家注释，适用于相当比例的蛋白质。其解决方案的关键在于：利用轻量级微调策略实现高效知识迁移，并通过生成式框架突破预设标签集的限制，从而提升注释的灵活性与生物学深度。未来需在数据质量、模型规模扩展及证据可追溯性方面进一步优化，以满足实际应用的可靠性要求。
链接: https://arxiv.org/abs/2609.24538
作者: Demian Pavlyshenko,Bohdan Pavlyshenko
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Neural and Evolutionary Computing (cs.NE); Quantitative Methods (q-bio.QM)
备注:
Abstract:Functional annotation of newly sequenced proteins remains a bottleneck in molecular biology: the number of sequences in public repositories grows far faster than the capacity for manual curation. Most computational approaches consider annotation as multi-label classification over a fixed ontology, which constrains predictions to a predefined label set. In this work we study the the protein annotation as a sequence-to-text generation problem. We fine-tune the 3B-parameter Ministral 3 base model with QLoRA (4-bit NF4 quantization with low-rank adapters) on sequence annotation pairs. We assess predictions with an LLM-as-expert protocol: a GPT model prompted as a senior molecular-biology curator scores organism identification as binary and function annotation quality. We conclude that QLoRA-fine-tuned compact LLMs can generate curator-style annotations with genuine biological value for a substantial subset of proteins. We also discuss future directions in data quality, model scaling, and evidence grounding that are needed to make the approach sufficiently reliable for practical use.
[NLP-26] LLJ Cards: Best practices for the Use of LLM s as Judges
【速读】： 该论文旨在解决生成式 AI（Generative AI）作为评估者在实际应用中面临的有效性、可靠性与可复现性不足的问题，尤其聚焦于大型语言模型作为裁判（LLMs as Judges, LLJs）在自动化评估任务中的实践缺陷。现有研究虽通过偏差缓解技术和提示工程优化等手段进行改进，但这些方法多为技术层面的修补，未能从根本上建立标准化、透明化和可复现的评估范式。本文提出“LLJ Cards”框架，整合测量理论、自然语言生成与机器学习领域的最佳实践，构建一套系统化的指南，用于指导基于LLJ的评估设计与报告。其解决方案的关键在于通过结构化框架将严谨的评估原则嵌入自动化评估流程，从而确保评估过程在科学性、一致性与可验证性上的全面提升。
链接: https://arxiv.org/abs/2609.24516
作者: Khaoula Chehbouni,Melina Medjdoub,Florian Carichon,Golnoosh Farnadi,Jackie Chi Kit Cheung
机构: McGill University (麦吉尔大学); Mila - Quebec AI Institute (魁北克人工智能研究所)
类目: Computation and Language (cs.CL)
备注: Prepared for conference submission
Abstract:In recent years, large language models (LLMs) have emerged as a popular alternative for evaluation. Often referred to as LLMs as judges (LLJs), these systems have been widely adopted by researchers and practitioners across a broad range of measurement tasks, driven by their strong performance, scalability, and cost-effectiveness relative to human judgment. However, a growing body of work has shown that the use of LLJs raise concerns about their validity and reliability as evaluators. Existing efforts to address these challenges have largely focused on developing bias-mitigation techniques and refining prompting strategies. While these approaches represent an important step forward, they primarily offer technical fixes and leave a more fundamental challenge unaddressed: the lack of standardized, transparent, and reproducible evaluation practices. In this paper, we introduce LLJ Cards, a framework that synthesizes best practices from measurement theory, natural language generation, and machine learning literature into practical guidelines for LLJ-based evaluations. While LLJs offer a promising path toward scalable evaluation, their effective use requires grounding in rigorous evaluation principles to ensure validity, reliability, and reproducibility. LLJ Cards addresses this need by providing a structured framework for applying these principles in the design and reporting of automated evaluations.
[NLP-27] Fathom-Vaidya: Advancing Medical Reasoning with Rubric-Based Rewards
【速读】： 该论文旨在解决大型语言模型（Large Language Models, LLMs）在医疗领域应用中的两大核心挑战：一是诊断推理（diagnostic reasoning），即基于临床数据进行证据驱动的、收敛性诊断的能力；二是临床医疗推理（clinical healthcare reasoning），即在多轮交互中实现情境化、以患者为中心的沟通、规划与动态适应能力，此类任务往往不存在单一正确答案。现有基准如HealthBench和MedXpertQA揭示了当前模型在复杂诊断场景下的表现缺陷以及在上下文感知对话中的局限性。为此，论文提出一种顺序训练框架，其关键在于利用合成数据与基于评分标准（rubric-based）的强化学习进行分阶段优化：首先通过基于MedBullets生成的问题，结合规则与评分标准引导的强化学习提升诊断推理能力；随后构建5.3k条合成的多轮临床交互场景，并配以多维度评分标准以全面评估和优化临床医疗推理能力。该方法在MedXpertQA上实现超过10%的性能提升，其30B参数模型在HealthBench-Hard上达到50.1%的准确率，超越包括GPT-5（thinking）在内的多个专有基线模型，证明了针对性合成数据与评分标准驱动训练在系统性增强医疗LLMs诊断与交互推理能力方面的有效性。
链接: https://arxiv.org/abs/2609.24480
作者: Kalash Shah,Kunal Singh,Snehan J,Shreyas Singh
机构: Fractal AI Research(弗拉克塔尔AI研究)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 18 pages, 5 Figures, Correspondence to this http URL @fractal.ai
Abstract:Deploying Large Language Models (LLMs) in healthcare requires robust performance across two complementary dimensions - diagnostic reasoning: the convergent, evidence-driven task of inferring a patient’s condition from clinical data to produce a diagnosis, and clinical healthcare reasoning: the broader, navigational judgment required to communicate, plan, and adapt across multi-turn clinical interactions where a single correct answer may not exist. Recent benchmarks such as HealthBench and MedXpertQA reveal persistent weaknesses in both areas, exposing failures in complex diagnostic scenarios and limitations in contextual, patient-centered dialogue. We introduce a sequential training framework that targets these facets using synthetic data and rubric-based reinforcement learning. First, we improve diagnostic reasoning using MedBullets-derived questions with rule- and rubric-guided Reinforcement Learning (RL). We then shift to clinical reasoning by generating 5.3k synthetic multi-turn scenarios, each paired with multi-dimensional rubrics to comprehensively assess the response. This approach yields over 10% improvement on MedXpertQA, and our 30B model achieves 50.1% accuracy on HealthBench-Hard, surpassing proprietary baselines including GPT-5 (thinking). Our results show that targeted synthetic datasets and rubric-based training can systematically improve both diagnostic and interactive clinical reasoning in medical LLMs.
[NLP-28] 1% of Tokens Can Be Enough: On Gradient Estimation in On-Policy Distillation
【速读】： 该论文旨在解决稀疏有策略蒸馏（Sparse on-policy distillation, OPD）中，教师模型对学生模型生成轨迹中部分标记（token）提供监督时，由于基于采样下一个标记所估计的梯度存在噪声，导致有用教师指导反而引发不准确更新的问题。其核心解决方案是引入信息效率比（Information-Efficiency Ratio, IER），该指标基于信息几何框架下对信号-噪声分解的分析，用于量化在最优标量基线（optimal scalar baseline）条件下相对梯度估计误差。通过候选集近似方法，结合IER与现有有效性评分（usefulness scores），实现基于信息效率与指导价值双重考量的令牌选择机制，同时保持原有的反向KL训练目标。实验结果表明，在数学和医学推理任务中，引入IER可显著提升现有选择器性能，即使在极低令牌预算（0.1%–1%）下，稀疏配置亦能实现与全量OPD相当甚至更优的表现，验证了在稀疏监督分配中同时考虑指导有效性与梯度估计可靠性的重要性。
链接: https://arxiv.org/abs/2609.24432
作者: Huanxin Sheng,Zhiling Ye,Haonan Wang,Jian Wang,Jinjie Gu,Jian Kang
机构: MBZUAI(曼彻斯特大学人工智能中心); Ant Group(蚂蚁集团)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:Sparse on-policy distillation (OPD) allocates teacher supervision to a small subset of tokens in student-generated trajectories. However, useful teacher guidance can yield a noisy update when its gradient is estimated from a sampled next token. We study this estimation problem at a fixed prefix in information geometry and propose an information-efficiency ratio (IER) based on a signal-to-noise decomposition. IER characterizes relative gradient estimation error under an optimal scalar baseline. A candidate-set approximation enables token selection based on IER and its combination with existing usefulness scores, while retaining the sampled reverse-KL training objective. On mathematical and medical reasoning tasks, adding IER improves existing selectors in multiple settings, with sparse configurations matching or exceeding full OPD without token selection at small token budgets of 0.1%–1%. These results support accounting for both usefulness and gradient-estimation reliability when allocating sparse supervision. Our code is available at this https URL.
[NLP-29] End-to-end Jordanian dialect speech-to-text self-supervised learning framework
【速读】： 该论文旨在解决低资源语言（特别是约旦阿拉伯方言）在语音转文本（Speech-to-Text）任务中因缺乏标注语音数据而面临的挑战。其核心问题在于如何在有限标注数据条件下，构建高效且准确的语音识别系统。解决方案的关键在于提出一种基于Transformer架构的端到端模型框架，结合自监督学习（Self-Supervised Learning）与噪声学生训练（Noisy Student Training）的自训练机制，充分利用未标注数据，并通过多源数据注入、数据增强以及少量人工干预实现高效的模型训练。该框架不仅显著提升了模型性能（相较微调后的Wav2Vec模型降低5%的词错误率），还构建了首个面向约旦阿拉伯方言的公开语音数据集，为阿拉伯语语音识别领域的应用（如问答系统、智能控制系统）提供了可复用的技术基础，推动智能机器人具备类人感知与听觉能力。
链接: https://arxiv.org/abs/2609.24410
作者: Ali A. Safieh,Ibrahim Abu Alhaol,Rawan Ghnemat
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Speech-to-text engines are extremely needed nowadays for different applications, representing an essential enabler in human-robot interaction. Still, some languages suffer from the lack of labeled speech data, especially in the Arabic dialects or any low-resource languages. The need for a self-supervised training process and self-training using noisy training is proven to be one of the up-and-coming feasible solutions. This article proposes an end-to-end, transformers-based model with a framework for low-resource languages. In addition, the framework incorporates customized audio-to-text processing algorithms to achieve a highly efficient Jordanian Arabic dialect speech-to-text system. The proposed framework enables ingesting data from many sources, making the ground truth from external sources possible by speeding up the manual annotation process. The framework allows the training process using noisy student training and self-supervised learning to utilize the unlabeled data in both pre- and post-training stages and incorporate multiple types of data augmentation. The proposed self-training approach outperforms the fine-tuned Wav2Vec model by 5% in terms of word error rate reduction. The outcome of this work provides the research community with a Jordanian-spoken data set along with an end-to-end approach to deal with low-resource languages. This is done by utilizing the power of the pretraining, post-training, and injecting noisy labeled and augmented data with minimal human intervention. It enables the development of new applications in the field of Arabic language speech-to-text area like the question-answering systems and intelligent control systems, and it will add human-like perception and hearing sensors to intelligent robots.
[NLP-30] URA-NER: A Unified Retrieval-Augmented Framework with Retrieval Alignment and Uncertainty Reduction for Low-Resource NER IJCNN2026
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在低资源场景下进行命名实体识别（Named Entity Recognition, NER）时，因标注数据稀缺导致的性能瓶颈问题。现有方法普遍存在检索错位（retrieval misalignment）与生成不确定性（generation uncertainty）两大挑战，尤其当模型参数规模减小时，其少样本学习能力显著下降。为应对上述问题，本文提出一种统一的检索增强框架URA-NER，其核心创新在于三个关键组件：渐进粒度检索（Progressive Granularity Retrieval, PGR）、模型感知表示增强（Model-aware Representation Enhancement, MaRE）以及推理感知知识验证（Reason-aware Knowledge Verification）。其中，PGR通过双阶段检索机制实现语义对齐，先基于查询全局语义检索用于实体跨度检测的示例，再依据具体实体上下文检索用于类型分类的细粒度信息；MaRE利用实体预识别引导表示构建，确保查询与示例在模型的语义空间和注意力模式中保持一致；而RaKV则引入闭环“生成-检索-验证”流程，显式化模型推理路径，利用该路径检索外部知识并重构为与原始推理逻辑对齐的验证证据，有效缓解生成不确定性。实验结果表明，URA-NER在多个低资源NER数据集上显著提升LLMs性能，尤其对小型模型提升尤为显著，实现了多个基准上的新最优表现。
链接: https://arxiv.org/abs/2609.24372
作者: Jingyu Wang,Shijie Wu,Fusheng Jin
机构: Beijing Institute of Technology (北京理工大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 8 pages,3 figures, accepted at IJCNN 2026, conference WCCI 2026
Abstract:In-context learning (ICL) based on large language models (LLMs) has shown promising potential in alleviating performance bottlenecks caused by the limited availability of annotated data in Named Entity Recognition (NER). However, existing methods still face issues of retrieval misalignment and generation uncertainty, making their performance heavily dependent on the LLM’s capabilities. As the parameter scale of LLMs decreases, their performance in few-shot settings deteriorates significantly. In this paper, we propose a novel unified retrieval-augmented framework, URA-NER, including three key components: Progressive Granularity Retrieval (PGR), Model-aware Representation Enhancement (MaRE), and Reason-aware Knowledge Verification. PGR is a two-stage retrieval mechanism that achieves stage alignment. It first retrieves demonstrations for span detection based on the query’s global semantics, and then for type classification based on the specific entity context, providing fine-grained local information. Moreover, MaRE employs entity pre-recognition to guide the construction of representations, ensuring the query and demonstrations are aligned within the LLM’s semantic space and attention pattern. In addition, to mitigate generation uncertainty, we propose RaKV, a closed-loop “generation-retrieval-verification” process. It explicates the LLM’s reasoning paths, leverages them for the retrieval of external knowledge, and reorganizes the knowledge into verification evidence aligned with the original reasoning paths. We conduct extensive experiments on multiple low-resource NER datasets. Results demonstrate that URA-NER significantly enhances the performance of LLMs under low-resource settings, with particularly pronounced gains for smaller LLMs, achieving new state-of-the-art results on several benchmarks.
[NLP-31] Mitigating Entity Type Confusion in Cross-Domain NER via Multidimensional Quantification and Reasoning Enhancement ECAI2026 IJCAI
【速读】： 该论文旨在解决跨域命名实体识别（Cross-domain Named Entity Recognition, CD-NER）中普遍存在的实体类型混淆（entity type confusion）问题，即模型在进行实体类型分类时，易将语义相近的实体类型误判。现有基于分解或生成范式的模型虽在实体边界检测上表现优异，但在类型分类阶段仍面临严重混淆难题。本文提出的关键解决方案是：首先构建多维度混淆量化模型（Multidimensional Confusion Quantification Model, MCQM），从源-目标领域层级关系、语义相似性以及显式数据评估三个维度量化模型对实体类型的混淆程度；进而设计渐进式双向推理链（Progressive Bidirectional Reasoning Chain, PBRC），利用MCQM提供的层级结构与混淆分析结果，引导大语言模型（LLM）生成分阶段的推理信息，从而增强模型的知识表征能力，有效缓解类型混淆并提升泛化性能。实验结果表明，该方法在CrossNER数据集所有领域上均达到了新的最优性能。
链接: https://arxiv.org/abs/2609.24357
作者: Jingyu Wang,Shijie Wu,Fusheng Jin
机构: Beijing Institute of Technology (北京理工大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 9 pages, 3 figures, Accepted at IJCAI-ECAI 2026
Abstract:Cross-domain Named Entity Recognition (CD-NER) aims to transfer the rich knowledge in the source domain to the target domain. Recent studies adopting decomposition or generation paradigms have achieved significant performance improvements, demonstrating high accuracy in entity span detection. However, during entity type classification, models severely suffer from entity type confusion, the erroneous tendency that models classify entities of one type in the text as another similar but incorrect type. To address this issue, we first propose a Multidimensional Confusion Quantification Model (MCQM) that quantifies a model’s confusion extent between entity types from three dimensions: source-target hierarchy analysis, semantic similarity analysis, and explicit data evaluation. Moreover, we propose the Progressive Bidirectional Reasoning Chain (PBRC). PBRC leverages the source-target hierarchy and confusion analysis from the MCQM to prompt the LLM to generate two-stage reasoning information. The two-stage reasoning information is utilized to augment the knowledge of the model, significantly mitigating entity type confusion and improving the model’s generalization performance. Experimental results demonstrate that our method achieves new state-of-the-art results on all domains of the CrossNER dataset.
[NLP-32] SupportCal: Label-Free Calibration of Post-Trained LLM s via Reference Support and Corroboration
【速读】： 该论文旨在解决后训练语言模型（Post-trained Language Models, PoLMs）在提升任务性能的同时出现置信度校准能力下降的问题，即模型变得比其预训练版本（Pretrained Language Models, PLMs）更加过度自信。针对缺乏任务特定标注校准数据的现实挑战，现有方法依赖于预训练模型作为无标签参考进行事后校准，但传统基于一致性的校准方法仅使用PoLM与其PLM参考意见一致的样本，排除不一致样本以避免温度参数过度拟合导致欠自信。本文重新审视这一二元处理策略，通过受控重引入诊断发现：适度引入部分不一致样本可改善校准效果，而随着不一致样本占比增加，收益逐渐递减。为此，提出SupportCal——一种无需标签的事后校准方法，其核心在于保留一致性样本的单位权重，并根据自基预训练模型（own-base PLM）的相对支持度及来自同规模候选池中多个预训练参考模型的协佐证据，为不一致样本分配连续权重。进一步分析表明，该加权目标函数在特定条件下存在有限最优温度解。在MedMCQA和MathQA等多个基准上，SupportCal相较于仅使用一致样本的基线方法，几乎在所有目标模型配置下均实现了更低的期望校准误差（Expected Calibration Error, ECE），且在固定标签分类任务TweetEval Sentiment上的补充实验也验证了该方法的普适性。
链接: https://arxiv.org/abs/2609.24303
作者: Linhan Luo,Lequan Lin,Dai Shi,Feng Chen,José Miguel Hernández-Lobato,Junbin Gao
机构: 未知
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 14 pages, 5 figures, 6 tables
Abstract:Post-training often improves task performance but can degrade confidence calibration, leaving post-trained language models (PoLMs) more overconfident than their corresponding pretrained language models (PLMs). Because task-specific labeled calibration data can be costly or unavailable, the corresponding pretrained PLM provides a natural label-free reference for post-hoc calibration. Prior agreement-gated PLM-referenced calibration fits a scalar temperature using only examples on which the PoLM and its PLM reference agree, excluding disagreement examples because direct alignment can drive the fitted temperature excessively high and induce under-confidence. We revisit this binary treatment. A controlled reintroduction diagnostic reveals a non-monotonic aggregate effect: admitting a moderate fraction of disagreement examples can improve calibration, whereas the benefit diminishes as unit-weight inclusion approaches the full disagreement set. We introduce SupportCal, a label-free post-hoc method that retains agreement examples at unit weight and assigns disagreement examples continuous weights based on the own-base PLM’s relative support and corroboration from pretrained references selected from a size-compatible candidate pool. We further characterize when the resulting weighted objective admits a finite optimal temperature. Across MedMCQA and MathQA, SupportCal yields lower ECE than the agreement-only baseline for nearly all evaluated target-model configurations; supplementary TweetEval Sentiment results show the same pattern on a fixed-label classification task.
[NLP-33] Structure Before Sampling: Community-Aware Core-Set Selection for Data-Efficient Text-to-Speech
【速读】： 该论文旨在解决文本到语音（Text-to-Speech, TTS）语料库构建成本高昂的问题，尤其是大量语音样本中蕴含的音位信息重复度高、边际贡献低。其核心挑战在于如何在固定音频时长预算下，从大规模语料中高效选取最具代表性的子集以最小化训练成本并最大化模型性能。解决方案的关键是提出一种基于音位结构图（phonotactic graph）的“社区代表性选择”（Community Representative）方法：将语料表示为一个音位相似性连接的图结构，利用其显著的聚类性和模块性特征，通过跨社区采样并在每个社区内均匀分布选择，优先包含稀有音素的发音样本。该方法在孟加拉语和英语语料上均表现出优越性，显著提升了对稀有音素二联体的覆盖率，并在保留更少训练数据的情况下，使TTS模型的字符错误率（CER）显著降低；尤其在孟加拉语任务中，使用20%核心集训练的模型性能优于全语料训练，且仅需4.5倍的训练时间。
链接: https://arxiv.org/abs/2609.24275
作者: Mizbaul Haque Maruf,Muhammad Nur Yanhaona
机构: Brac University (布拉克大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Text-to-speech (TTS) corpora are costly to record, yet many utterances add little new phonetic information. Core-set selection reduces this cost by choosing a small training subset under a fixed audio-duration budget. We represent a corpus as a phonotactic graph that links each utterance to its most phonemically similar ones, and we first test whether this graph has structure. In Bangla and English corpora, its clustering is 199 and 56 times that of a size-matched random graph, and its modularity is more than twice that of a degree-preserving random graph. We then propose Community Representative, a selector that samples across graph communities and spreads its choices within each one, starting from utterances rich in rare phonemes. At every budget and in both languages, it covers more rare phoneme bigrams than random and entropy-based selection, and this lead holds on held-out utterances. TTS models trained on its 20% core-sets have a significantly lower character error rate (CER) than models trained on equal-duration random or entropy-based subsets in both languages. When all models train for the same number of epochs, the Bangla core-set model also outperforms full-corpus training (3.93% vs. 4.47% CER) with 4.5x less training time.
[NLP-34] Canonical Procedural Actions: An Auditable Annotation Protocol for Tool-Use Agent Traces
【速读】： 该论文旨在解决生成式 AI（Generative AI）在程序化行为分析中缺乏明确动作单元与可追溯证据链的问题。现有工具使用代理轨迹虽能记录消息和API调用，但无法提供可显式定义的动作单位及其与证据之间的可检验关联。为此，论文提出“规范程序动作”（Canonical Procedural Actions, CPAs）这一标注协议，其核心在于系统性地记录每个程序动作、首个代理事件锚点、实现该动作的代理事件序列以及独立的上下文证据。该方法允许多个动作共享同一消息锚点，且不假设消息内部存在隐含顺序。通过零售场景案例研究，采用开放式归纳、记录整合与迭代审计，构建了24条目的版本化代码本。在两个独立的大语言模型（LLM）上下文中对32条轨迹进行标注，结果显示轨迹级分离标注下标签重合度高达A=0.982，但若强制要求上下文-事件引用完全一致，重合度下降至0.798，表明该指标反映的是结构可重复性而非语义准确性。此外，16/26的任务ID在开发数据中出现，历史工具载荷被截断至110字符，影响了部分信息完整性。消融实验显示，合并所有标签可使重合度升至0.986，而基于简单端点规则即可复现工具锚定部分，重合度达0.997。助手-消息类动作的标签重合度为0.971，最低单标签重合度为0.816。将冻结后的代码本应用于244条额外轨迹，共生成4,058条记录，并识别出8种诊断结果。研究贡献在于提供了一种可审计、可验证的标注工具及其实现过程与测量边界评估，但其人工参考有效性与下游应用价值仍需进一步验证。
链接: https://arxiv.org/abs/2609.24264
作者: Songqi Li,Dongqing Li,Zheqiao Cheng
机构: Cambridge Language Sciences, Raised Faculty Building, Sidgwick Avenue, Cambridge CB3 9DA, United Kingdom; Department of Statistics, University of Oxford, 24–29 St Giles’, Oxford OX1 3LB, United Kingdom; Zhejiang University, China
类目: Computation and Language (cs.CL)
备注: 23 pages, 5 figures, 9 tables. Includes ancillary files for reproducing the reported analyses
Abstract:Tool-use agent traces identify messages and API calls, but procedural analyses also need explicit units of action and inspectable links to their evidence. We present Canonical Procedural Actions (CPAs), an annotation protocol that records a procedural function, its first agent-event anchor, the agent events that realize it, and separate contextual evidence. Multiple actions may share a message anchor without an inferred within-message order. A retail case study produces a versioned 24-entry codebook through open induction, recorded consolidation, and successive application audits. Two isolated LLM contexts annotate 32 trajectories disjoint from development at the trajectory level, producing 499 and 491 occurrences with anchor-label overlap A=0.982. Requiring identical context-event references reduces overlap to 0.798. These are structural repeatability measures, not semantic accuracy: 16 of 26 task IDs also occur in development, and historical tool payloads were truncated to 110 characters. Retrospective controls show that collapsing all labels raises overlap to 0.986, while simple endpoint rules reproduce the tool-anchored portion with 0.997 overlap. Assistant-message actions have 0.971 overlap, with a per-label minimum of 0.816. Applying the frozen codebook to 244 further trajectories yields 4,058 records, including eight diagnostic outcomes. The contribution is an explicit, auditable annotation instrument and a case study of its construction and measurement limits; human-reference validity and downstream utility remain to be established.
[NLP-35] aramandal-GPT : Enhancing Astrodynamics Problem-Solving with Knowledge Retrieval and Structured Thinking
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在天文学与轨道动力学等专业领域中表现受限的问题，尤其针对多步推理、符号运算及领域专有术语理解能力不足的挑战。其解决方案的关键在于构建基于Qwen3-8b模型架构的领域适配框架Taramandal-GPT（又称Constellation-GPT），通过引入检索增强生成（Retrieval-Augmented Generation, RAG）管道提升上下文相关性，并集成回退机制以增强推理过程的准确性与可解释性。在包含299个从基础到高级空间科学问题的轨道动力学问题基准测试集（APBench）上，采用数值误差容忍度评分与语义相似性双重评估方法，Taramandal-GPT展现出与当前最优开源及闭源模型相媲美的性能，尤其在高复杂度思维任务中表现突出，验证了专用大模型在高精度、可解释性要求场景下的有效性，为实现可靠的人工智能辅助系统在天体物理学、航天器工程和深空探索中的应用提供了重要路径。
链接: https://arxiv.org/abs/2609.24246
作者: Akhil Sharma,Jatin Gupta,Ali Imam Abidi
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Proceedings of All India Hindi Technical Conference, 05-06 February 2026
Abstract:Large language models (LLMs) have shown remarkable progress in natural language understanding, yet their effectiveness in specialized fields like astronomy and astrodynamics remains limited due to challenges in multi-step reasoning, symbolic manipulation, and domain-specific terminology. To address this, we present Taramandal-GPT (Constellation-GPT), a domain-adapted framework built on the Qwen3-8b backbone, enhanced with a Retrieval-Augmented Generation (RAG) pipeline and a fallback mechanism for improved contextual precision. We evaluate it on the Astrodynamics Problems Benchmark (APBench), a dataset of 299 questions covering foundational to advanced levels of space science. Using a dual evaluation method - numeric margin-based scoring and semantic similarity assessment - Taramandal-GPT achieves competitive performance against state-of-the-art open- and closed-source models, with notable strength in thinking-intensive tasks. These results highlight the value of specialized LLMs for domains demanding accuracy and interpretability, positioning Taramandal-GPT as a step toward reliable Artificial Intelligence (AI) assistants for astrophysics, spacecraft engineering, and space exploration.
[NLP-36] Memory vs. Context? Influential Factors of Factual Recall in Language Models EMNLP2026
【速读】： 该论文旨在解决语言模型（Language Models, LMs）在面对记忆知识与上下文矛盾陈述时如何权衡决策的问题，尤其关注模型在不同情境下对已存储知识与新输入信息的依赖程度。其解决方案的关键在于通过大规模复制与压力测试（stress-test）Yu et al. (2023) 的世界首都实验，涵盖31个来自Pythia、GPT-2、Qwen3和Ministral系列的基线与后训练模型，并扩展至ParaConflict数据集中五种额外的知识关系类型。研究发现，尽管大模型和高频实体仍倾向于依赖记忆答案，但多项结论存在显著族间差异：如在Qwen3-14B和32B中实体频率效应消失；后训练对记忆-上下文权衡的影响在不同模型家族中不一致；仅问题表述方式的变化即可导致模型对记忆知识的依赖度波动高达80个百分点；且语义无关的段落亦可模拟出具有说服力的支持性上下文。这些结果揭示了原始结论的适用边界，并深化了对模型推理机制在复杂提示环境下的理解。
链接: https://arxiv.org/abs/2609.24238
作者: Guilhem Fouilhé,Nicholas Asher,Philippe Muller
机构: IRIT(图卢兹信息与技术研究所); ANITI(人工智能与智能技术研究院); CNRS(法国国家科学研究中心); Université de Toulouse(图卢兹大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted to the BlackboxNLP 2026 Reproducibility Challenge (Special Track), EMNLP 2026
Abstract:We reproduce and stress-test the work of Yu et al. (2023), who characterize how language models (LMs) arbitrate between memorized knowledge and contradictory in-context statements. We replicate their world-capitals experiments on 31 models spanning Pythia, GPT-2, Qwen3, and Ministral families, including base and post-trained variants, and extend evaluations to five additional knowledge relation types from the ParaConflict dataset. We empirically confirm most of their original findings: larger models and higher-frequency entities tend to favor memorized answers, with substantial family-level variance. However, several conclusions do not generalize cleanly: entity-frequency effects disappear on Qwen3-14B and 32B; post-training shifts the memory-context trade-off inconsistently across families; question phrasing alone can change a model’s reliance on memorized knowledge by up to 80 percentage points; and semantically unrelated prose can mimic coherent supporting context. Our results clarify where Yu et al.'s claims hold and to what extent they generalize to other prompts.
[NLP-37] From Articles to Publishers: Aggregating Language Model Predictions for News Source Reliability Inference
【速读】： 该论文旨在解决新闻来源可靠性自动评估中的关键挑战，即如何在不依赖人工专家评判的情况下，基于文本内容实现对新闻出版商可靠性的可扩展、自动化推断。传统方法多聚焦于单篇文章的可靠性判断，但此类方法难以有效反映出版商整体的可信度特征。本文提出一种两级框架：首先利用基于Transformer的自然语言模型对单篇新闻文章进行可靠性评分，随后通过聚合各文章的预测结果，推断出未见出版商的整体可靠性。该方法的关键在于采用严格的出版商互斥（publisher-disjoint）评估协议，确保训练与测试集间无出版商重叠，从而更贴近真实应用场景下的泛化能力要求。实验基于439家英文新闻出版商共19,476篇政治类新闻文章及NewsGuard提供的可靠性标签，结果显示，通过文章级预测的聚合显著提升了模型性能与鲁棒性，准确率由文章层面的约0.60提升至出版商层面的0.69。进一步分析表明，预测误差在不同政治倾向的出版商间存在统计学显著差异，揭示了政治取向与误分类模式之间的关联。研究结论表明，仅依靠聚合的文本信号即可有效推断新闻出版商的可靠性，为构建大规模、内容驱动的自动化新闻源评估系统提供了有力支持。
链接: https://arxiv.org/abs/2609.24219
作者: John Bianchi,Manuel Pratelli,Fabio Pinelli,Marinella Petrocchi
机构: IMT School for Advanced Studies Lucca(意大利高等研究学院卢卡校区); Institute of Informatics and Telematics, National Research Council (IIT-CNR)(国家研究委员会信息与电信研究所)
类目: Computation and Language (cs.CL)
备注: 10 pages, 7 figures, 2 tables. Submitted to IEEE Transactions on Computational Social Systems (TCSS)
Abstract:Traditionally, the reliability of news publishers is assessed by expert organisations that evaluate editorial practices, transparency and factual standards at source. When this process is translated into a computational approach, the problem is often formulated at the level of individual articles, with models being trained on a set of pre-labelled articles and their performance being evaluated in a test phase. In this work, we investigate news source reliability inference as a source-level prediction problem. We propose a two-stage framework in which transformer-based language models first estimate the reliability of individual articles and subsequently aggregate article-level predictions to infer the reliability of previously unseen publishers. To approximate realistic deployment conditions, we enforce a strict publisher-disjoint evaluation protocol, ensuring that no publisher appears in both training and test sets. Experiments on 19,476 political news articles from 439 English-language publishers labeled with NewsGuard reliability ratings show that aggregation substantially improves robustness and performance, increasing accuracy from approximately 0.60 at the article level to 0.69 at the publisher level. Finally, we analyze how prediction errors vary across political orientations, revealing statistically significant associations between political leaning and misclassification patterns. Overall, our findings show that publisher reliability can be inferred from aggregated textual signals alone, supporting scalable and content-based approaches to automated news source assessment.
[NLP-38] Vimarsha: Faithful ASR Evaluation for Indian Languages with Demographic Diversity In-the-Wild Audio and Spelling Variations INTERSPEECH2026
【速读】： 该论文旨在解决印度语种自动语音识别（ASR）评估基准中存在的两大系统性偏差问题：一是源于干净、受控音频条件导致的性能评分过高，二是因过于严格的转录标准对合法语言变体进行惩罚所引发的评分过低。其解决方案的关键在于提出Vimarsha——一个覆盖全部22种官方印度语言、时长达100小时的新型评估基准。该基准通过融合具有人口统计学多样性的实地录音与精心筛选的高难度真实场景音频，结合“多有效转录框架”（lattice of variations framework），为每个语音片段提供多个合法的转录版本，从而更真实地反映实际应用中的语言多样性与声学复杂性。实验结果表明，在该基准下，10个主流ASR模型的性能排名发生显著变化，暴露出地理与人口群体间的性能差异，以及在不同语速和声学环境下的系统性失败模式，验证了其在推动更公平、更真实评估方面的有效性。
链接: https://arxiv.org/abs/2609.24199
作者: Kaushal Santosh Bhogale,Srija Anand,Sadakopa Ramakrishnan Thothathiri,Tahir Javed,Sshubam Verma,Mitesh M. Khapra
机构: 未知
类目: Computation and Language (cs.CL)
备注: Accepted in Interspeech 2026
Abstract:Evaluation benchmarks for Indian language automatic speech recognition (ASR) suffer from two systematic biases: optimistic scores from clean, controlled audio conditions, and pessimistic scores from overly rigid transcription standards that penalize valid linguistic variations. We introduce Vimarsha, a 100-hour benchmark spanning all 22 scheduled Indian languages, designed to address both distortions. Vimarsha combines demographically diverse on-field recordings with carefully mined in-the-wild audio selected for acoustic difficulty, alongside a lattice of variations framework that encodes multiple valid transcriptions per utterance. Evaluations of 10 state-of-the-art ASR models reveal substantial shifts in model rankings under realistic conditions, geographic and demographic performance disparities, and systematic failure modes across speaking rates and acoustic environments.
[NLP-39] LoopCD: Loop-wise Contrastive Decoding for Improving Reasoning in Looped Language Models EMNLP2026
【速读】： 该论文旨在解决循环语言模型（Looped Language Models, LoopLMs）在推理过程中存在的环路不稳定性问题。尽管LoopLMs通过共享权重的递归精炼内部潜在表示实现了高效的“隐式推理”（latent reasoning），但其在迭代过程中仍易产生局部不确定的“硬”标记（hard tokens），这些标记与推理错误密切相关，导致推理性能下降。本文提出的解决方案——环路级对比解码（LoopCD），关键在于在推理阶段干预这些不稳定的硬标记：利用LoopLM内部动态，将早期迭代的输出logits与最终精炼迭代的logits进行对比，以此构建最终的采样分布。该方法无需额外训练，仅引入可忽略的推理开销，却能有效提升对推理关键硬标记的精炼能力，显著改善多种推理任务上的表现，验证了其高效性与普适性。
链接: https://arxiv.org/abs/2609.24196
作者: Byeongho Yu,Junhyuk So,Eunhyeok Park
机构: Pohang University of Science and Technology (POSTECH); Department of Computer Science and Engineering
类目: Computation and Language (cs.CL)
备注: Accepted to EMNLP 2026 Main Conference
Abstract:Looped Language Models (LoopLMs) perform “latent reasoning” by recursively refining internal latent representations with shared weights, offering a more effective alternative to explicit verbal reasoning. Despite their effectiveness, we find that LoopLMs remain prone to loop instability: unstable refinement across iterations can produce localized uncertain “hard” tokens associated with reasoning errors. To address this, we propose LoopCD, loop-wise contrastive decoding that enhances the reasoning performance of LoopLMs by intervening on these tokens at inference time. Specifically, we exploit the internal dynamics of LoopLMs and contrast the logits from earlier iterations with logits from the last refined iteration to form the final sampling distribution. We find that this strategy is highly efficient, introducing only negligible inference overhead and requiring no additional training, while effectively improving reasoning performance by naturally refining reasoning-critical hard tokens. Extensive experiments show that our method improves the performance of recent representative LoopLMs across various reasoning tasks.
[NLP-40] When Residualization Helps an Audit: Format Effects Slice Gains and Their Limits
【速读】： 该论文旨在解决当前大语言模型（LLM）评估体系中评价分数因过度依赖表面形式（surface form）而偏离其本应衡量的内在质量（如代码正确性、逻辑一致性）的问题。其核心挑战在于：现有评估指标（如奖励模型、重排序器、LLM评判者）在面对语义等价但表现形式不同的解法时，无法有效区分真正正确的解决方案与带有注释的错误方案，导致评估结果不可靠。论文提出的关键解决方案是通过残差化（residualization）方法剔除可预测的表面特征成分，以期获得更纯净的构造效度（construct validity）测量。然而，研究发现仅移除表面成分并不足以保证评估的有效性——被剔除的部分可能包含与构造相关的有效信号，且残差化无法明确区分哪些信息属于表面偏差、哪些属于真实构造信号。实验表明，在设计干预下（如仅修改注释的单元测试标签），残差化虽能削弱评分对代码格式的敏感性（约0.12的降幅），但正确与错误代码之间的判别优势变化极小（<0.01）。在观测性自然语言推理（NLI）和问答（QA）任务中，采用独立标注者标签重新评估冻结的基准模型，仅在预定义的出错子集上实现与构造标签更高的吻合度，却导致整体人群的一致性下降，且在问题内排序性能普遍恶化。当构造特征与表面特征高度纠缠时，残差化可能在降低相关性的同时损害构造对齐性；更严重的是，在受控模型中，某些配置虽严重削弱构造对齐性，但仍可通过所有预调整检验，说明缺乏可靠的“守门机制”来确保调整后的分数仍具有效性。因此，论文最终提出一套报告协议：任何经调整的评分必须附带审计时刻的诊断信息，明确说明其构造对齐性的代价，而非替代原始评分。
链接: https://arxiv.org/abs/2609.24194
作者: Daein Weon,Dongho Kang
机构: Kookmin University (首尔国立大学); UStechlab
类目: Computation and Language (cs.CL)
备注: 61 pages, 4 figures, 40 tables. Code: this https URL
Abstract:Evaluation scores used around LLM systems – including reward models, rerankers, and LLM judges – can track surface form instead of the quality they claim to measure. When presented with a terse correct solution and a commented buggy solution for the same MBPP problem, a public preference reward model selects the correct one no better than a coin flip (0.507). Subtracting the predictable surface component from such scores is increasingly common, but removal alone does not yield a more valid measurement: the removed component may carry construct-relevant signal, and residualization cannot tell which is which. Under designed interventions – unit-test labels with comment-only edits – residualization attenuates the reward model’s format effects by about 0.12 on both correct and buggy code, while the correct-versus-buggy margins move by less than 0.01. In observational NLI and QA settings, we freeze a held-out replication before scoring and re-evaluate it using labels from disjoint annotators; this supports only a narrower conclusion: better agreement with the construct labels on a pre-declared slice where a surface-only predictor errs, not a repaired score. Full-population agreement falls in every observational setting with a reported positive slice gain, and within-question ranking falls in every such QA setting. When construct and surface features are entangled, residualization can decorrelate a score while degrading construct alignment, and, in a controlled model, configurations just as damaging to construct alignment pass every pre-adjustment check, so no committed gate is a guarantee. We assemble these distinctions into a reporting protocol whose outcomes, refusal included, state what an adjusted score may be claimed to show: an audit-time diagnostic reported beside the construct-alignment cost it incurs, never a replacement for the raw score.
[NLP-41] Efficient LLM Distillation for Bangladesh Legal Context: A Smartphone-Compatible Retrieval-Augmented Generation Model
【速读】： 该论文旨在解决孟加拉国公民普遍难以获取法律信息的问题，其核心挑战包括：法定文本仅以英文呈现、专业律师集中于城市地区，以及依赖云端的生成式AI（Generative AI）在移动网络不可靠环境下失效，导致幻觉性法律文本可能造成实际危害。为应对这一问题，研究提出一种针对孟加拉国成文法（statutory law）的轻量化、离线可用的法律问答系统。解决方案的关键在于通过两阶段渐进式知识蒸馏（progressive knowledge distillation），将一个90亿参数的Gemma-2教师模型压缩至20亿参数的学生模型，其中第一阶段采用监督微调（supervised fine-tuning）处理经质量筛选的9,429个高质量法律问答对，第二阶段通过最小化稀疏Kullback-Leibler散度（sparse Kullback-Leibler divergence）并结合温度调节（tau = 4.0）的顶层日志优化，利用QLoRA技术（4位NF4量化、秩32的LoRA适配器）实现高效参数微调。系统通过混合检索机制（60%密集语义搜索 + 40%BM25）从36,029条孟加拉国宪法及国家立法文本中获取上下文，确保输出可溯源。实验表明，该模型在英文基准测试中达到ROUGE-L 0.4715和BERTScore F1 0.5679，相较未蒸馏的检索增强基线分别提升103%与143%；在离线环境下可在Pixel 6设备上以4–8 tokens/秒运行，模型大小仅1.6 GB（GGUF Q4_K_M）。跨语言评估显示，尽管语料库为英文，但对50条孟加拉语查询仍能生成准确响应（ROUGE-L 0.4083，BERTScore 0.8133），且单评者试点中执业律师给出加权平均4.16/5分（90%评分4或5），验证了其在真实场景中的实用价值。
链接: https://arxiv.org/abs/2609.24177
作者: MD. Nafis Kamal,Mahadi Hasan Fahim,Talha Ridwan,Nadifa Zaman,Fariha Roushon Florin,Farig Yousuf Sadeque,Saadat Rafid Ahmed
机构: 未知
类目: Computation and Language (cs.CL)
备注: 10 pages, 6 figures, 8 tables
Abstract:Legal information in Bangladesh is inaccessible to most citizens. Statutory text is English-only, trained lawyers are concentrated in urban centres, and cloud-dependent AI fails where mobile connectivity is unreliable, a setting in which hallucinated legal text causes direct harm. The system addresses statutory interpretation only; queries that require judicial precedent or case-law reasoning fall outside its scope. We target the statutory access gap by compressing a 9-billion-parameter Gemma-2 teacher into a 2-billion-parameter student through two-phase progressive knowledge distillation. Phase 1 performs supervised fine-tuning on 9,429 quality-gated legal question-answer pairs (65% acceptance from 14,514 generated queries); Phase 2 minimises sparse Kullback-Leibler divergence against the teacher’s top-50 per-token logits at temperature tau = 4.0, implemented via QLoRA (4-bit NF4, rank-32 LoRA adapters). Prior legal language models target general legal English; this system specialises in Bangladeshi statutory law. Every response is grounded through hybrid retrieval combining dense semantic search (60%) and BM25 (40%) across 36,029 statutory passages from the Bangladesh Constitution and national legislation. On a 50-query English benchmark, the distilled model reaches ROUGE-L 0.4715 and BERTScore F1 0.5679, a 103% ROUGE-L and 143% BERTScore gain over the retrieval-augmented undistilled baseline (ROUGE-L 0.2323, BERTScore 0.2340). The adapter quantises to 1.6 GB (GGUF Q4_K_M) and runs at 4-8 tokens per second on a Pixel 6 with no network access. Cross-lingual evaluation on 50 Bangla queries yields ROUGE-L 0.4083 and BERTScore 0.8133, showing effective retrieval from Bangla input against an English-only corpus. In a single-evaluator pilot, a practising lawyer rated 50 responses at a weighted mean of 4.16/5 (90% rated 4 or 5), supporting utility beyond text-overlap metrics.
[NLP-42] AC-Time: Texts as Channels For Multimodal Time Series Forecasting
【速读】： 该论文旨在解决现有时间序列预测方法仅依赖数值观测而忽略辅助文本中蕴含的丰富上下文信息的问题。现有多模态方法虽尝试引入文本信号，但通常将文本视为静态特征或以大语言模型作为预测主干，导致难以捕捉时序动态性且计算成本较高。其解决方案的关键在于提出TAC-Time框架，通过将文本信息转化为额外的时间通道，实现文本特征与数值序列在共享时序主干中的联合建模。该设计不仅保留了时间连续性和周期结构，还具备高效可扩展的优点，并支持系统的可解释性分析。实验表明，通过注意力机制和频域分析揭示了跨模态强相关性，识别出具有预测能力的文本信号，其相关性感知对齐带来了部分预测性能提升。在真实世界多模态基准上的大量实验证明，TAC-Time显著优于现有方法。
链接: https://arxiv.org/abs/2609.24156
作者: Jiayi Liang,Xiaotian Gu,Xinyu Xie,Yuanbin Wu,Xiaoling Wang
机构: East China Normal University(华东师范大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Multimedia (cs.MM)
备注: 11 pages, 6 figures, 4 tables
Abstract:Most existing time series forecasting methods rely solely on numerical observations, overlooking rich contextual information from auxiliary texts. Recent multimodal approaches attempt to incorporate textual signals, but they often treat text as static features or use large language models as forecasting backbones, limiting their ability to capture temporal dynamics and increasing computational cost. To address these challenges, we propose TAC-Time, a unified framework that transforms textual information into additional temporal channels. By modeling text features jointly with numerical sequences in a shared temporal backbone, TAC-Time preserves temporal continuity and periodic structures while remaining efficient and scalable. This formulation also enables systematic interpretability analyses. We show strong cross-modal dependencies through attention and frequency-domain analyses, and identify predictive textual signals whose correlation-aware alignment yields partial forecasting improvements. Extensive experiments on real-world multimodal benchmarks demonstrate that TAC-Time outperforms prior methods.
[NLP-43] Data Agents : Agent ic Data Systems
【速读】： 该论文旨在解决传统数据系统在人工智能时代所面临的三大核心问题：依赖人工构建的数据处理管道、对异构数据缺乏语义理解能力，以及采用僵化且被动的处理模式。其解决方案的关键在于提出一种全新的范式——数据智能体（Data Agent），通过实现从人工设计到自主编排、从字面操作到语义解析、从被动响应到主动适应的范式跃迁，构建具备高度自治能力的数据管理与分析体系。该系统由六大核心组件构成：语义化数据组织、语义算子、基于智能体的管道编排与优化、反馈驱动的迭代优化、记忆管理及主动适应机制，并在此基础上进一步开发了面向数据分析和数据科学任务的专用智能体。实验结果表明，该方法在真实基准测试中显著优于现有先进方法，验证了其有效性，同时指出了未来实现完全自治数据系统所面临的关键挑战。
链接: https://arxiv.org/abs/2609.24137
作者: Guoliang Li,Peiyao Zhou,Xuanhe Zhou,Ji Sun,Yuyu Luo,Ju Fan
机构: Tsinghua University (清华大学); Shanghai Jiao Tong University (上海交通大学); The Hong Kong University of Science and Technology (Guangzhou) (香港科技大学(广州)); Renmin University of China (中国人民大学)
类目: Databases (cs.DB); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Accepted by TKDE
Abstract:Traditional data systems face profound limitations in the AI era, relying on human-crafted pipelines, lacking semantic understanding of heterogeneous data, and operating through rigid, reactive processing. To address these challenges, we propose a new paradigm called the Data Agent, designed to manage, process, and analyze data with minimal human intervention. Data agents autonomously execute a wide range of data-related tasks, transforming traditional data systems by shifting from manual design to autonomous orchestration, from literal manipulation to semantic interpretation, and from reactive to proactive processing. Our Data Agent system includes six components: semantic data organization, semantic operators, agentic pipeline orchestration and optimization, feedback-driven refinement, memory management, and proactive adaptation. Building on this foundation, we also develop two specialized agents: the data analytics agent and the data science agent. Experiments on real benchmarks demonstrate significant performance gains of our data agent over state-of-the-art methods. We identify open challenges to guide future research in building fully autonomous data systems.
[NLP-44] Re:CAP - Auditing Retrieval Coverag e in Production RAG Pipelines
【速读】： 该论文旨在解决生成式检索增强（Retrieval-augmented generation, RAG）系统在生产环境中难以有效监控检索质量的问题。由于非平稳的海量文档语料库实时重新索引，无法获得全面的相关性标注，导致检索性能常被忽视，而生成效果指标则占据主导地位。为此，本文提出一种无需参考答案的检索覆盖率审计方法——Re:CAP（REtrieval Coverage Audit by iterative Probing），其核心创新在于通过迭代探测潜在缺失文档证据来评估覆盖范围，而非穷举所有相关文档。Re:CAP利用大语言模型（LLM-as-judge）作为判断机制，识别已覆盖主题、生成探测问题以挖掘可能遗漏的主题，并从候选文档中筛选出引入新信息的条目。在四个公开基准测试中，Re:CAP相较于传统的扁平化BM25 top-500检索，可额外召回9%-29%的黄金标准文档，尤其在TREC-COVID数据集上达到48%；在MuSiQue数据集上，其召回率比混合检索方案提升12.9个百分点，且仅使用不到一半的文档预算。即使在融合了BM25、密集检索与混合检索的集成基线（各取top-500）下，仍存在21.2%的黄金文档未被覆盖，而这些文档中78.9%经人工评估被认为提供了结构上新颖的信息（Fleiss κ = 0.79），且在真实生产流量中73.9%的新增文档具有实际价值。此外，Re:CAP在三次独立运行中实现端到端召回率误差控制在±1%以内，展现出高度稳定性，使其成为周期性部署后检索质量审计的理想工具。
链接: https://arxiv.org/abs/2609.24122
作者: Aviral Joshi,Hanoz Bhathena,Max Nelson,Saket Sharma
机构: Machine Learning Center of Excellence, JPMorgan Chase Co.(机器学习卓越中心，摩根大通公司)
类目: Computation and Language (cs.CL)
备注:
Abstract:Retrieval-augmented generation (RAG) is hard to monitor in production: exhaustive relevance labels do not exist for non-stationary multi-million-passage corpora that re-index in real time. As a result, retrieval quality is generally understudied and often deprioritised in favour of generation-oriented metrics. In this work, we propose auditing retrieval coverage by probing for evidence of missing documents rather than enumerating every relevant one. Our method Re:CAP (REtrieval Coverage Audit by iterative Probing) is a reference-free audit loop applied to a deployed RAG pipeline’s initial answer and retrieved context: it identifies the topics already covered, generates probing questions for plausibly missing topics, retrieves candidate documents, and applies an LLM-as-judge to retain only those that introduce previously-unretrieved information. On four public benchmarks, Re:CAP recovers 9-29% of gold labels that flat BM25 top-500 cannot reach, rising to 48% on TREC-COVID. On MuSiQue Re:CAP beats flat hybrid top-500 by +12.9 pp on recall at less than half the document budget. An ensemble BM25, dense, and hybrid baseline (top-500 each) still leaves out 21.2% of gold docs on TREC-COVID that Re:CAP recovers; human annotators judge that 78.9% of those structurally distinct documents add new information to the baseline answer (Fleiss \kappa = 0.79, n = 123), and 73.9% on live production traffic (n = 180). End-to-end recall is reproducible to within \pm 1% across three independent runs, making Re:CAP a stable instrument for periodic retrieval audits.
[NLP-45] You Can Tell Whos Asking: What the Webs Questions Are Made Of and Where They Come From EMNLP2026
【速读】： 该论文旨在解决一个广泛存在于学术界与工业界的核心假设问题：网页上提取的问答（Question-Answer, QA）数据是否真实反映了人类的真实需求。现有研究普遍将页面中的问题视为人类意图的可靠代理，但本文通过大规模实证分析挑战这一假设。研究基于110个FineWeb快照（2013–2025年）中提取的134亿条问题实例，发现三个关键问题：其一，问题来源（即宿主页面或网站）在问题形式上留下可识别的信号，利用长度和上下文等特征，可通过逻辑回归模型以AUC 0.725的准确率区分真实用户提问与模板化/制造型问题，而对商业常见问题（FAQ）的区分能力较弱（AUC 0.554）；其二，问题出现频率并不反映实际需求强度——高频问题多为模板化内容（前1000名中超过70%为模板），其高频率仅表明文本被频繁发布，而非被频繁询问；其三，过去十二年间，真实提问占比下降了79%（经爬取数据构成校正后仍维持42%-56%），同时问题长度与上下文复杂度持续降低。因此，该研究的关键贡献在于首次实现对网络问题来源的历时性、实例级测量，揭示出可爬取网页中的问题已从“人类主动提出”逐渐演变为“为机器可读而生成”的模式，从而从根本上质疑了当前以问题频次为基础的训练数据、检索基准与内容策略的有效性。
链接: https://arxiv.org/abs/2609.24106
作者: Calvin Zhou,Vincent McCloskey,Krishna Srinivasan
机构: University of Pennsylvania (宾夕法尼亚大学); Bodhium Labs
类目: Computation and Language (cs.CL)
备注: Accepted to the 13th Web as Corpus Workshop (WaC-13) at EMNLP 2026. 14 pages, 4 figures. Code and data: this https URL
Abstract:Questions scraped from the web are used across academia and industry as a proxy for what people want to know. Across QA training data, retrieval benchmarks, and content strategy, questions on a page are assumed to reflect human intent. We test this assumption at scale by extracting 13.4B question occurrences across 110 FineWeb snapshots (2013-2025), and report three findings. First, you can tell who is asking: provenance (the host/page of questions) leaves a signal in question form, and a logistic model can separate genuine user questions from templated/manufactured ones at AUC 0.725 via length and surrounding context rather than question type, though only 0.554 against commerce FAQ writing. Second, question frequency does not measure demand: the most-frequent questions are boilerplate/templated (over 70% of the top thousand), so occurrence counts measure how often a string was published and not how often it was asked. Third, over twelve years the genuine share of occurrences fell by 79% (42-56% after controlling for crawl composition), with question length and context decreasing. We present the first diachronic, occurrence-level measurement of web question provenance, and find the crawlable web’s questions have shifted from being asked by humans toward manufactured for machines to read.
[NLP-46] From Content Generation to Learning Support: Pedagogy-Guided Generative Video Tutors for STEM Learning EMNLP2026
【速读】： 该论文旨在解决当前生成式AI在教育视频生成中过度关注视觉连贯性而忽视教学有效性的核心问题，即生成内容缺乏明确的教育学结构、质量控制机制不足，且无法评估学习者理解情况或纠正认知误区。其解决方案的关键在于提出一种以教学法为导向的生成视频辅导框架——PIVOT（Pedagogy-guided Instructional VideO Tutoring），该框架将教育学原则贯穿于视频生成全流程：首先基于教学原理指导脚本设计，继而通过以代码为中心的生成方式与教学法验证机制产出经过验证的多模态视频，最后实现视频与测评及谬误感知型补救策略的联动。实验与专家评估结果表明，PIVOT生成的视频在教学一致性、呈现清晰度、教学流程连贯性及学习有效性方面均表现优异，强调了教育内容生成系统应从“产出质量”转向“支持教与学实践”的人本化设计范式。
链接: https://arxiv.org/abs/2609.24083
作者: Xinchen Ma,Shuimu Wang,Gaole He,Yanbin Zhang,Chunyang Wang,Yunshi Lan,Weining Qian
机构: East China Normal University (华东师范大学); National University of Singapore (新加坡国立大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注: accepted to EMNLP 2026, code available at GitHub
Abstract:Generative AI enables scalable production of educational videos, but current systems largely focus on producing visually coherent content rather than supporting learning. As a result, generated videos often lack explicit pedagogical structure, reliable quality control, and mechanisms for assessing learner understanding or addressing misconceptions. In this work, we introduce PIVOT (Pedagogy-guided Instructional VideO Tutoring), a generative video tutoring framework for STEM learning via learning-centered instructional support.1 Inspired by conventional teaching workflows, our framework integrates pedagogy into the full generation pipeline: it first uses instructional principles to guide storyboard generation, then produces verified multimodal videos through code-centric generation and a pedagogical verification harness, and finally connects videos with assessment and misconception-aware remediation. Experiments and expert evaluations across four STEM domains show that our framework produces educational videos with pedagogically aligned content, clear and engaging presentation, coherent instructional flow, and perceived effectiveness for learning. These findings suggest a human-centered perspective on educational content generation: generative systems should be evaluated and designed not only by what they produce, but also by how they support teaching practices, learner understanding, and corrective feedback.
[NLP-47] Efficient Reasoning Exploration via State-Conditioned Latent Steering with Progress Guidance
【速读】： 该论文旨在解决复杂推理任务中基于“Best-of-N”策略时存在的探索效率低下问题，其核心挑战在于后训练推理模型易出现探索坍缩（exploration collapse），即独立采样路径反复趋同，导致增加采样预算也无法显著提升性能。现有方法虽能拓宽探索范围，但缺乏对推理进展方向的显式引导，难以高效识别并聚焦于具有实质性推进意义的推理分支。为此，本文提出一种无需训练的潜在空间引导框架——状态条件-进展引导式引导（State-conditioned Progress-guided Steering, SPS），其关键创新在于构建一个状态条件的定向向量库（Direction Bank），该库包含针对不同前缀状态区域的多种进展导向的引导向量。在在线推理阶段，SPS根据当前前缀状态检索最匹配的引导向量，并仅在高不确定性转移点施加该引导，从而有效驱动后续推理步骤向具备实质性进展的方向演进。大量实验表明，SPS在多个模型规模和基准测试上均显著优于主流基线方法，且消融分析验证了其设计的有效性，为未来高效推理策略研究提供了重要启示。
链接: https://arxiv.org/abs/2609.24066
作者: Hengyuan Zhang,Chenming Shang,Zunhai Su,Xiao Liang,Hui Shen,Jing Xiong,Dawei Li,Shiping Yang,Kailai Yang,Wei Zhang,Ruobing Xie,Hayden Kwok-Hay So,Ngai Wong
机构: The University of Hong Kong (香港大学); University of California, Los Angeles (加利福尼亚大学洛杉矶分校); Arizona State University (亚利桑那州立大学); Simon Fraser University (西蒙菲莎大学); The University of Manchester (曼彻斯特大学); Tencent(腾讯)
类目: Computation and Language (cs.CL)
备注:
Abstract:Best-of- N is a widely used inference strategy for complex reasoning, whose effectiveness depends on whether sampled candidates can cover diverse and high-quality reasoning paths. However, post-trained reasoning models often suffer from \emphexploration collapse, where independent rollouts repeatedly follow similar reasoning paths and limit the gains from increasing the rollout budget. Existing methods alleviate this issue by promoting broader exploration, but do not explicitly guide exploration toward continuations that make meaningful progress, resulting in limited exploration efficiency. To address this, we propose \emph\underlineState-conditioned \underlineProgress-guided \underlineSteering (SPS), a training-free latent steering framework. Specifically, SPS constructs a state-conditioned Direction Bank containing multiple progress-guided steering vectors for different prefix-state regions. During online inference, SPS retrieves a suitable steering vector based on the current prefix state and applies it at high-uncertainty transitions to guide the next reasoning step toward meaningful progress. Extensive experiments across multiple model scales and benchmarks demonstrate that SPS consistently outperforms strong baselines. Further analyses validate the effectiveness of its key designs and offer valuable insights for future research. The code is available at this https URL.
[NLP-48] Representation-guided in-context learning for medical image interpretation with multimodal large language models
【速读】： 该论文旨在解决通用多模态大语言模型（Multimodal Large Language Models, MLLMs）在医学图像解读任务中因缺乏领域适配性而表现不佳的问题，尤其针对传统方法依赖资源密集型微调（fine-tuning）所带来的高成本与低可扩展性挑战。其核心解决方案是提出一种无需训练的推理框架——表示引导的上下文学习（Representation-guided In-Context Learning, RG-ICL），该方法利用冻结的编码器（frozen encoders）检索与查询对齐的示范案例（demonstrations），实现无需任务特定参数更新的高性能推理。关键在于：通过基于语义表示的精准案例检索，显著提升模型性能——在八种涵盖组织病理学、放射科及视网膜眼底成像的数据集上，分类任务平均提升20个百分点，视觉问答（VQA）任务平均提升13个百分点，接近甚至超越需训练的方法。研究发现，所选示范案例的语义相关性远比数量更重要：仅6个与查询对齐的案例优于最多32个随机选取的案例，而固定或随机选择的案例常导致准确率低于基线。此外，在VQA任务中，若示范案例同时与图像内容和问题意图对齐，则性能进一步提升。这表明，在医学图像理解中，精心筛选模型可见的参考案例，是一种高效且可行的替代重新训练的策略。
链接: https://arxiv.org/abs/2609.24057
作者: Minda Zhao,Fangyu Hu,Yan Luo,Yutong Yang,Jiahui Cai,Kaichen Zhou,Manling Li,Paul Liang,Yilun Du,Lucy Q. Shen,Mengyu Wang
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Medical image interpretation is central to diagnosis and care, yet adapting general-purpose multimodal large language models (MLLMs) often requires resource-intensive domain-specific fine-tuning. Here we introduce representation-guided in-context learning (RG-ICL), a training-free inference framework that retrieves query-aligned demonstrations using frozen encoders, without task-specific parameter updates. Across eight datasets spanning histopathology, radiology and retinal fundoscopy, RG-ICL improved classification (mean gain 20 percentage points) and visual question answering (VQA) (mean gain 13 percentage points) over no-context and conventional ICL, approaching or exceeding training-based comparators. Which cases were retrieved mattered more than how many: 6 query-aligned cases outperformed up to 32 randomly selected ones, whereas fixed or random cases often reduced accuracy below baseline. For VQA, aligning reference cases with both image content and question intent produced further gains. These findings indicate that for medical image interpretation, curating which reference cases an MLLM sees is a practical alternative to retraining it.
[NLP-49] Calibrated Decisions at Scale: Converting Police Crash Narratives into Probabilistic Crash Variables with a System One Model (Jev)
【速读】： 该论文旨在解决大规模交通事故调查叙述文本（crash narratives）中蕴含的丰富信息难以通过传统编码方式高效提取的问题。现有方法受限于三大障碍：前沿大语言模型（large language models, LLMs）在大规模应用时成本过高、生成内容不可验证，且缺乏明确的人工审核量标准。为此，本文提出一种基于“门控类型化决策”（gated, typed decisions）的新型解决方案——Jev模型，该模型属于系统一（System One）范式，仅输出分析师预定义选项的概率分布，不生成任何自然语言文本，从而实现可审计性与低成本扩展。研究对德克萨斯州499,500条事故叙述进行了编码，采用27个问题的结构化模式，其成本由编码方案复杂度决定，而非文本长度。通过与2,416份盲评人工标注结果及已有编码字段对比，验证了模型性能：该模型在人类标注基准上达到F1值0.908，优于两个前沿大语言模型（分别提升0.059和无显著差异）。模型校准性能因模型而异，需独立审计；基于相同标签进行再校准后，校准误差降低3.3倍。此外，与编码字段相比，模型在叙事忠实度上的平均一致性（kappa）低估了0.26，表明编码字段存在信息损失。研究还提出了概率离散网格下的“分辨率下界”（resolution-floor bound），并设计了基于标记记录的审查预算机制，以确定每变量每年所需人工阅读量。最终，将校准后的变量纳入原有编码体系后，九类因素导致的伤亡事故年均数量增加10,747起，凸显了该方法在揭示潜在风险因素方面的价值。
链接: https://arxiv.org/abs/2609.24052
作者: Amir Rafe,Subasish Das
机构: Texas State University (德州州立大学); Ingram School of Engineering (英格拉姆工程学院)
类目: Computation and Language (cs.CL)
备注:
Abstract:Crash datasets that carry an investigator narrative hold information the coded fields omit. Coding those narratives at scale has been blocked by three obstacles. Frontier large language models are costly at that scale, their generated text cannot be verified, and no rule says how much output a human must check. This paper formulates narrative coding as gated, typed decisions answered by Jev, a System One model that returns probabilities over analyst-defined options and generates no text. A screen covered 499,500 Texas narratives and 195,857 were coded with a 27-question schema. Cost is governed by schema size rather than narrative length. The probabilities are audited against coded fields and against 2,416 blinded human judgments drawn under a stated sampling design. Two frontier large language models are benchmarked on the same records. Against human labels the typed model attains an F1 of 0.908. One frontier model gains 0.059 and the other is indistinguishable from it. Calibration varies by model rather than by paradigm, so each model must be audited. Recalibration on the same labels reduces calibration error by a factor of 3.3. Agreement with coded fields understates fidelity to the narrative by a median of 0.26 in kappa. A resolution-floor bound covers any model that reports probabilities on a discrete grid. A review budget over flagged records gives the records a human must read per variable and per year. Adding the calibrated variables to the coded fields raises the injury and fatal crashes attributed to nine factors by 10,747 per year.
[NLP-50] When Evidence Conflicts: Reliability-aware Meta-review Generation
【速读】： 该论文旨在解决多篇同行评审意见之间存在证据冲突且可靠性不一时，生成连贯、可信的元评审（meta-review）这一挑战性问题。现有方法通常将元评审生成视为多文档摘要任务，对评审意见进行均一化聚合，难以在评审观点相左时有效识别并优先采纳更具说服力的观点。本文提出一种基于可靠性的证据聚合框架，其关键在于：首先从同行评审中提取细粒度的方面级观点，并识别各方面的内部冲突证据；随后通过估计观点级支持度与评审级质量来量化证据的可靠性；基于此可靠性信号，为不同评审反馈分配加权权重，使生成器能够优先采纳支持度更高的论点，同时保留多元视角。实验结果表明，该方法在自动评估和人工评估上均显著优于强基线模型，尤其在高冲突场景下表现出更强的冲突识别与化解能力。
链接: https://arxiv.org/abs/2609.24028
作者: Xinzhe Wang,Fei Tao,Jiang Xie,Hong Yu,Ye Wang
机构: Chongqing University of Posts and Telecommunications(重庆邮电大学), China; NewsBreak, USA
类目: Computation and Language (cs.CL)
备注:
Abstract:Generating coherent meta-reviews from multiple peer reviews is challenging when reviewer evidence conflicts and varies in reliability. Existing approaches typically formulate meta-review generation as a multi-document summarization task and aggregate reviewer feedback uniformly, making it difficult to determine which opinions should be prioritized under disagreement. In this paper, we study meta-review generation through reliability-aware evidence aggregation. Our framework first extracts aspect-level opinions from peer reviews and identifies conflicting evidence within each aspect. It then estimates opinion-level support and review-level quality to measure evidence reliability. Based on these signals, the framework assigns reliability-aware weights to reviewer feedback, enabling the generator to prioritize better-supported arguments while preserving diverse perspectives. Experiments demonstrate that our method consistently improves meta-review generation over strong baselines on both automatic and human evaluations, with clear gains in conflict recognition and resolution under high-conflict review scenarios. The code and implementation details are publicly available at this https URL.
[NLP-51] From Tables to Quantified Statements: Evaluating LLM Inference Generation through Executable Verification
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在生成表格描述时存在的逻辑不一致问题，即生成内容虽流畅但缺乏对结构化数据的严格语义支持。为此，作者提出了一项受控任务STAT-TO-TEXT，要求模型基于统计表格生成包含量化表达（如all、some、no、most）的自然语言推理，确保输出在逻辑上严格符合表格中的事实。其解决方案的关键在于设计一种由大语言模型自动生成的Python校验代码（checker code），通过执行该代码来自动验证生成推理的真值条件是否与表格数据一致，从而实现对生成结果的客观评估。实验对比了四种不同规模和架构的开源大模型，结果表明模型规模和架构均显著影响性能，其中最大模型GPT-OSS-120B在忠实性、表覆盖度和量化词多样性方面表现最优，且其自动化评估结果与人工标注高度一致，验证了该评估框架的有效性。
链接: https://arxiv.org/abs/2609.23966
作者: Mai Mohamed Eida,Gunjan Anand,Ayush Singh,Aleksandre Maskharashvili
机构: University of Illinois Urbana Champaign (伊利诺伊大学厄本那-香槟分校)
类目: Computation and Language (cs.CL)
备注:
Abstract:LLMs can generate fluent descriptions from tables, but their outputs may remain logically unsupported by the structured data. We introduce STAT-TO-TEXT, a controlled task in which LLMs generate quantified natural language inferences from statistical tables using quantified constructions such as all, some, no, and most. To evaluate these inferences, we use an LLM generated Python checker code which when executed verifies the corresponding truth conditions against the table. We compare four open-weight LLMs across model families and scales, evaluating faithfulness, logical accuracy, table coverage, and diversity. Our results show that model scale and family matter, with the largest model (GPT-OSS-120B) consistently producing the most faithful inferences without sacrificing greater table coverage and quantifier diversity, as opposed to smaller models. These findings are supported by human annotation, which shows that the automated checker closely aligns with human judgments.
[NLP-52] Open-Jev Judgments on CallScreenBench: Calibrated One-Pass Scam Screening with a Small Language Model
【速读】： 该论文旨在解决电话诈骗筛查中实时生成可信概率的难题，要求在每次通话轮次中以毫秒级延迟输出一个可信赖的欺诈概率。其解决方案的关键在于采用Jev-style类型的决策框架，即输入候选选项后，仅通过一次前向传播即可输出每个选项的校准概率，且无需生成文本。研究测试了该框架的开源实现JevLite，基于Qwen3-4B模型进行LoRA微调，使温度缩放后的softmax输出两个标签（欺诈/非欺诈）的对数概率，从而直接获得P(scam)。在41个独立保留的CallScreenBench场景（共577次轮次决策）上，三种子集成模型达到AUROC 0.974，校准误差为0.052，性能不劣于预注册对比的LLM判官（MiniMax-M3），且在相同挂断规则下提前1.14个轮次做出决策，单次决策耗时仅64.5毫秒（运行于单张消费级GPU），较同架构微调生成答案的版本降低4.9倍。关键优势源于读出机制与校准策略的优化，而非模型精度提升：即使使用未微调的ModernBERT编码器也表现相近，且实验选择过程存在测试集暴露风险，所有通话样本均为合成数据。本文不主张架构创新，核心贡献在于提出并验证该方法的应用范式，同时系统性报告了校准性、误报率与决策时效等关键指标。
链接: https://arxiv.org/abs/2609.23959
作者: Simiao Ren,Kidus Zewde,Xingyu Shen,Yuchen Zhou,Dennis Ng,Ankit Raj,Tommy Duong,Yuxin Zhang,Neo Tiangratanakul
机构: Scam.ai (Reality Inc.)
类目: Computation and Language (cs.CL)
备注: 13 pages, 6 figures, 5 tables
Abstract:Screening a phone call for fraud needs a trustworthy probability after every caller turn, in milliseconds. Jev-style typed decisions promise exactly that: declared options go in, one calibrated probability per option comes out of a single forward pass, with no generated text. We test an open implementation of this readout, JevLite, on scam-call screening: Qwen3-4B is LoRA-tuned so that the temperature-scaled softmax over two answer-label logits is P(scam). On 41 held-out CallScreenBench scenarios (577 per-turn decisions) a three-seed ensemble reaches AUROC .974 with calibration error .052, non-inferior to an LLM judge (MiniMax-M3) at a pre-registered .02 margin, with no false alarms on legitimate calls, decisions 1.14 turns earlier under the same hang-up rule, and 64.5 ms per decision on one consumer GPU, 4.9x lower than the same backbone fine-tuned to generate its answer. The gain is in the readout and calibration, not accuracy: a fine-tuned ModernBERT encoder is not significantly worse, the recipe was selected with test-set exposure, and all callers are synthetic. We claim no architectural novelty; the contribution is the application and an evaluation reporting calibration, false alarms and decision timing alongside AUROC.
[NLP-53] Some Dialects Are More Equal Than Others: Non-Prestigious Arabic Dialectal Bias in LLM s
【速读】： 该论文旨在解决埃及阿拉伯语在自然语言处理（NLP）领域中存在的重要代表性失衡问题，即现有研究与资源开发长期聚焦于地位较高的开罗埃及阿拉伯语（Cairene Egyptian Arabic, CEA），而对地位较低的上埃及阿拉伯语（Sa’idi Egyptian Arabic, SEA） dialect缺乏充分覆盖。其核心问题是：这种上游（upstream）的方言代表性不足是否导致大语言模型（LLM）对SEA的可接受性产生系统性偏见，以及这种偏见是否进一步在下游任务中表现为性能下降。解决方案的关键在于通过目标句法评估（Targeted Syntactic Evaluation, TSE）任务系统检验多款LLM对SEA句法特征的偏好，揭示了跨模型普遍存在的对SEA的显著排斥倾向；随后在MMLU基准测试中验证这些方言特征对下游模型性能的影响，发现当输入为SEA文本时模型表现明显退化。该研究证实了子方言差异对语言技术影响的实质性存在，强调必须加强对非主流方言变体的建模与包容，以推动更公平、鲁棒的语言技术发展。
链接: https://arxiv.org/abs/2609.23955
作者: Mai Mohamed Eida,Ryan Dolan,Paul de Nijs,Jonathan Dunn
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Previous work on Egyptian Arabic in NLP has focused largely on the prestigious Cairene Egyptian Arabic (CEA) dialect, resulting in a lack of representation for the less prestigious Sa’idi Egyptian Arabic (SEA) dialect both in LLM and resource development. Does this lack of representation influence an LLM’s view of the acceptability of SEA (upstream), and does an upstream bias against SEA lead to worse performance (downstream)? We investigate the upstream effect of SEA dialectal features on LLM preferences in a Targeted Syntactic Evaluation (TSE) task which reveals a significant bias against SEA across multiple LLMs. We then analyze the effect of these same features on downstream model performance on MMLU benchmarks and show that models experience a degradation in performance when presented with SEA. This work highlights the need for further exploration on how sub-dialectal variation impacts language technologies.
[NLP-54] HaikuS2S: A Cascaded System For Responding In Verse
【速读】： 该论文旨在解决生成具有特定韵律结构的诗歌化语音（如俳句）的难题，尤其针对俳句固有的5-7-5音节结构及行末停顿等关键语言特征在语音合成中难以准确建模的问题。现有方法虽在情感表达和语调捕捉方面有所进展，但未能有效建模俳句的结构性特征。其解决方案的关键在于提出一种级联式系统 HaikuS2S，通过结合自动语音识别（ASR）、大语言模型（LLM）生成的俳句文本以及在散文与定制俳句数据集上联合微调的文本到语音（TTS）模型，实现对俳句音节结构与语调节奏的精准对齐。实验结果表明，该方法在语调与韵律对齐性方面显著优于基线模型，尤其是在同时在通用诗歌与俳句数据上进行微调的系统表现更佳，且各系统间情感相似性保持稳定，验证了其在保持情感一致性的同时有效建模结构性特征的能力。
链接: https://arxiv.org/abs/2609.23951
作者: Devangi Sharma,Sophia Judicke,Glenda Tan,Conrad Schaumburg,Shinji Watanabe
机构: Carnegie Mellon University (卡内基梅隆大学); Stanford University (斯坦福大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Sound (cs.SD)
备注: Accepted to SLT 2026, Demo Track. 5 pages, 5 figures
Abstract:Expressive speech synthesis has advanced through prosody modeling, yet generating structured poetic speech, such as haiku, remains challenging. Prior work on prosody transfer improves expressiveness, and fine-tuned poetry TTS (text-to-speech) systems capture verse intonation. However, these models do not model haiku’s 5-7-5 syllable structure or line-ending pauses. We present a cascaded system, HaikuS2S, combining ASR (automatic speech recognition), LLM (large language model)-generated haiku, and TTS fine-tuning on both prose and custom haiku datasets. Our evaluation focuses on emotion similarity, speech quality, and prosody alignment. In our experiments, we see that our prosody and tonal alignment improve significantly with our fine-tuned systems, particularly the one trained on both general poetry and haiku. We also see that we maintain similar emotion similarity scores across all systems.
[NLP-55] XYEval: Agents say yes to bad advice
【速读】： 该论文旨在解决人机协作中因“XY问题”（XY Problem）导致的沟通失效问题，即用户倾向于描述其尝试的解决方案而非真实需求，从而误导AI代理做出错误响应。其核心解决方案在于提出一种名为XYEval的元评估框架，可将现有基准测试转化为针对XY问题的评估工具，以系统性检验代理在面对用户误导性建议时的抗干扰能力与推理沟通能力。研究发现，当前主流模型在遭遇XY变异后性能显著下降，相对降幅最高达46.7%，尤其在需要详尽解释才能获得认可的严谨型用户情境下表现更差。尽管引入强调识别XY问题的系统指令能部分缓解该问题，但效果有限。深入的轨迹分析揭示了代理在执行过程中因未能有效识别用户误导或清晰表达根本问题而导致性能衰减的机制。研究表明，真正缓解XY问题需同时具备识别用户意图偏移的能力和清晰传达底层问题的沟通能力，这仍是当前生成式智能体面临的关键挑战。
链接: https://arxiv.org/abs/2609.23939
作者: Zhengxuan Wu,Yuxuan Li,Oyvind Tafjord,Been Kim
机构: Google DeepMind(谷歌深度思维)
类目: Computation and Language (cs.CL)
备注: 33 pages, 11 figures
Abstract:Effective communication between users and AI agents is essential for human-AI collaboration. The XY problem is a well-known communication pitfall where a person asks about their attempted solution rather than their actual problem. We extend prior sycophancy evaluation to the XY problem in agentic settings, evaluating whether agents can resist plausible but misleading suggestions from users and communicate their reasoning. We introduce XYEval, a meta-evaluation framework that can transform an existing benchmark into an XY problem evaluation. We evaluate five models across six diverse benchmark suites. Agents suffer large XY drops under XY mutation across benchmarks, with relative drops reaching up to 46.7%. With \tau^2 -bench, we further show that agent performance drops more when encountering a pedantic user who requires detailed explanations before approving a better solution. Our findings suggest that current agents lack the ability to effectively reason and communicate when facing misleading suggestions. A simple system instruction baseline that encourages awareness of XY problems only offers partial mitigation. Extensive trace analyses provide behavioral insights into how and why these XY drops occur across execution trajectories. Our results show that mitigating the XY problem remains challenging, requiring agents to both recognize user misdirection and clearly communicate the underlying problem.
[NLP-56] Measuring the Assistants Harmlessness Preferences on the User Turn
【速读】： 该论文旨在解决后训练（post-training）过程中生成式AI（Generative AI）模型在对话场景中角色一致性与语义表征泛化的问题。具体而言，研究关注的是：当模型被训练为具有持续性助手人格（assistant persona）时，该人格所体现的偏好是否仅限于模型自身发言的“助理回合”（assistant turn），还是会不恰当地影响对用户发言（user turn）的预测。研究发现，尽管助手人格应仅支配其自身输出，但一个关键的安全相关偏好——即倾向于执行无害任务而非有害任务——却显著影响了模型对用户输入的预测，即便在用户发言阶段，该偏好依然存在。这一现象表明，后训练并非简单地植入一个局部、浅层的助手角色，而是使模型在内部表征中将助手的偏好泛化至对用户的理解与建模之中。其解决方案的关键在于揭示：后训练通过大规模数据微调，使模型不仅学习到角色行为模式，更将角色偏好内化为对整体对话上下文的深层语义理解，从而导致跨角色的语义偏见迁移，这反映了生成式模型在后训练阶段出现的深层次认知泛化现象。
链接: https://arxiv.org/abs/2609.23935
作者: Jord Nguyen
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Post-training turns a general next-token predictor into a chat model with a persistent assistant persona. If that persona is a character the model plays only on its own turns, its preferences should govern what the assistant says, not what the model predicts other speakers will say. We test this boundary and find that it does not hold: a safety-relevant preference of the assistant—for harmless over harmful tasks—shapes the model’s predictions even on the user’s turn, where the assistant is not the one speaking. We find that this preference is small or near-zero in pretrained base models, that it emerges through post-training, replicated across open-weight model families, grows with scale, and can be moved by narrow finetuning that never touches user turns. We claim that this is evidence that post-training does not merely install a shallow assistant persona, but instead generalises beyond just the local assistant turn, into the model’s representation of the user.
[NLP-57] me-Incremental Continued Pretraining of LLM s: Knowledge Updates Without Catastrophic Forgetting
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在预训练结束后迅速过时，而从头再训练又成本高昂的问题。其核心挑战在于如何高效地通过持续预训练（Continued Pretraining, CPT）实现知识更新，尤其是在面对网页规模爬取数据的时间增量式更新场景中——此类数据集具有显著的URL重叠性，与传统连续学习假设的不相交数据流存在本质差异。论文的关键解决方案在于：在真实的时间增量式CPT设置下，基于FineWeb-Edu数据集对多个主流模型家族（OLMo2、Llama-3.1/3.2、Gemma-3-1B）在不同参数量级（1B–8B）上进行系统评估，并揭示出若干关键发现：首先，知识能够有效获取，且未出现灾难性遗忘，多数模型在截止日期前的事实召回率均有提升，增益程度与预训练饱和度相关（主要受每参数令牌预算驱动）；其次，更新代价极低，跨十三项任务的平均性能下降不足0.01；第三，数据质量远胜于数量，高质量小规模数据（6B tokens）可媲美大规模粗粒度数据（40B tokens），且知识获取与通用能力优化的最佳学习率相差约一个数量级，同时高秩LoRA可达到全参数持续预训练的效果；最后，持续预训练带来的性能提升可通过监督微调（SFT）有效传递，而直接偏好优化（DPO）的效果则依赖于模型家族特性。综上，该研究为时间增量式持续预训练提供了实证支持和可操作的优化策略，表明其在实际部署中具备高度可行性与有效性。
链接: https://arxiv.org/abs/2609.23916
作者: Fırat Öncel,Salman Hussain Ali,Mirco Ravanelli,Cem Subakan,Çağatay Yıldız
机构: Concordia University(康考迪亚大学); Mila – Quebec AI Institute(魁北克人工智能研究所); Université de Montréal(蒙特利尔大学); Laval University(拉瓦尔大学); University of Tübingen(图宾根大学); Tübingen AI Center(图宾根人工智能中心)
类目: Computation and Language (cs.CL)
备注: Preprint
Abstract:Large language models (LLMs) drift out of date the moment their pretraining ends, yet retraining from scratch is prohibitively expensive. Continued pretraining (CPT) is the natural remedy, but it is typically evaluated through a continual learning lens that assumes disjoint data streams. This is a poor fit for time-incremental updates on web-scale crawls, where successive snapshots share substantial URL overlap by design. We study time-incremental CPT in this realistic regime: continued pretraining on FineWeb-Edu dumps drawn strictly from after each model’s knowledge cutoff, evaluated across six open-weight models spanning three families (OLMo2, Llama-3.1/3.2, Gemma-3-1B) and four parameter scales (1B-3B-7B-8B). We organize our findings around four practical questions. (i) Is knowledge acquired? Yes, but heterogeneously, and without catastrophic forgetting: five of six models also improve on pre-cutoff factual recall, and the gains track pretraining saturation (driven primarily by token budget per parameter). (ii) What does it cost? Almost nothing: the macro-average across a thirteen-task suite stays within 0.01 of the base for every model. (iii) What is the recipe? Data quality dominates quantity (a curated 6B-token slice matches a broader 40B one); the optima for knowledge acquisition and general capability are separated by roughly an order of magnitude in learning rate; and LoRA at sufficient rank matches full CPT. (iv) Does it survive deployment? CPT gains transfer through SFT, while DPO’s effect is family-dependent. Together, these results paint a more optimistic picture of time-incremental CPT than the prior continual learning literature suggests. Comments: Preprint Subjects: Computation and Language (cs.CL) Cite as: arXiv:2609.23916 [cs.CL] (or arXiv:2609.23916v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2609.23916 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[NLP-58] his-that-model-1.0: A typed decision model that decides in 30 ms for a millionth of a cent
【速读】： 该论文旨在解决当前生成式AI在决策类任务中效率低下、成本高昂且存在输出不可控的问题。具体而言，现有系统依赖前沿大模型进行决策时，需通过远程API调用完成一次或多轮推理，产生大量文本输出（即生成式输出），不仅耗时（平均8758毫秒）、费用高昂（按令牌计费），还需额外解析器处理非结构化结果，且存在输出格式错误或偏离预设选项的风险。其核心挑战在于：如何在保证高准确率的前提下，实现低延迟、零生成输出、完全本地化运行的高效决策。本文提出的解决方案关键在于设计一种名为this-that-model-1.0的20亿参数类型化决策模型，其答案直接从指定位置的隐藏状态中读取，并严格限定于调用方声明的选项集合内，从而彻底避免生成文本、杜绝输出畸形问题；所有决策在同一前向传播中并行完成，不生成任何输出令牌，推理时间仅30.9毫秒（在单个消费级GPU上），显著优于主流云端服务（如Jev）的响应速度与成本表现。该模型在68个真实场景决策任务上达到0.941的准确率（Brier得分0.042），相较云端服务提升明显，且能耗极低（每批推理耗电仅0.000217美元）。然而，该模型在需要多步中间状态传递的任务（如复杂算术运算）上表现较弱（准确率0.560），表明其局限性集中于无法支持跨步骤记忆的复杂推理，尽管经过针对性微调，也未能有效迁移至其他任务领域。
链接: https://arxiv.org/abs/2609.23886
作者: Zehua Cheng,Wei Dai,Jiahao Sun
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Software delegates more of its branches to models every year: which queue a ticket enters, whether a command is safe to run, whether a claim clears without a person. What the program needs back is not prose. It is one of n declared options and a number it can threshold. Today that costs a round trip to a frontier model – hundreds of milliseconds, a per-token bill, and a parser – for a question that is usually a conjunction of three clauses. this-that-model-1.0 is a 2B-parameter typed decision model. Its answer is read directly from the hidden state at a designated position and restricted to the option set the caller declared, so no text is generated, nothing can be malformed, and every question in a request is answered in the same forward pass. It decides in 30.9 ms on one laptop GPU and generates zero output tokens doing it, where a frontier API call costs 8758 ms and the hosted systems that answer these questions well spend between 21 and 212 generated tokens per question thinking first, billed for every one. It sustains 32 decisions per second on one consumer GPU and never lets the state leave the machine. On a third party’s recorded cohort of 68 decision questions, on their inputs and their wording, it scores 0.941 with a Brier score of 0.042, against 0.765 and 0.133 for the hosted service Jev on the same items. One pass of our 42-family internal suite takes 32 seconds and 0.000217 USD of electricity; the most accurate hosted model we measured needs 155.2 minutes and 10.636 USD. We also report where it loses. On multi-step arithmetic, which a single forward pass cannot carry intermediate results through, it scores 0.560 against their 0.98 to 1.00, and a targeted second training round improved the five task families it was written for and transferred to none of the other 13. The model is open-sourced in this https URL
[NLP-59] Federated Multilingual Speech-LLM s: Architecture and Aggregation Strategy Benchmarking
【速读】： 该论文旨在解决多语言自动语音识别（ASR）在联邦学习（Federated Learning, FL）框架下面临的挑战，特别是如何在数据异构性与隐私保护需求并存的分布式环境中，有效训练多语言语音大模型（Speech-LLM）。其核心问题在于：如何在保持模型性能的同时，应对客户端间数据分布差异大、通信效率受限以及模型组件适应性不足等问题。解决方案的关键在于通过精细化的优化策略实现多组件协同训练——具体而言，独立调整语音编码器、连接模块（connector）和解码器的学习率，并采用针对编码器和解码器的低秩适配（LoRA）与对连接模块进行全参数微调相结合的三组件联合优化策略，显著降低了联邦学习下的识别错误率。此外，研究发现联邦近端优化（FedProx）的有效性高度依赖于语言模型（LLM）骨干网络的容量，具备更强表征能力的预训练架构（如EuroLLM）在固定编码器配置下能更好地抵御数据异构性，表明模型基座容量在提升联邦学习鲁棒性中起关键作用。这些发现为在隐私敏感场景中部署多语言Speech-LLM提供了可量化的系统设计指导。
链接: https://arxiv.org/abs/2609.23825
作者: Jordi Luque,Aleix Sant,Fernando López
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted Iberspeech 2026
Abstract:We present a comprehensive benchmark of Federated Learning (FL) for multilingual Automatic Speech Recognition (ASR), evaluating four Speech-LLM architectures on the Multilingual LibriSpeech dataset. We compare FedAvg and FedProx across frozen and unfrozen encoder configurations, demonstrating that optimized learning rates are critical for performance. Specifically, independently tuning the learning rates for the speech encoder, connector, and decoder yields the lowest error rates, with full three-component adaptation (LoRA for encoder and decoder, full training for the connector) producing the best FL results. We observe that FedProx efficacy is architecture-dependent, providing notable advantages in multilingual pre-trained architectures (e.g., EuroLLM over TinyLlama when keeping the encoder fixed); this indicates that LLM backbone capacity plays a key role in mediating resilience to heterogeneous data distributions. These findings offer concrete design guidance for deploying multilingual Speech-LLMs in privacy-sensitive, distributed environments.
[NLP-60] FLARE: A Full-Lifecycle Dense Supervision Paradigm for Long-Horizon Coding Agents via Generative Reward Model
【速读】： 该论文旨在解决长时序软件工程（Long-horizon Software Engineering, SWE）任务中，大型语言模型（Large Language Model, LLM）代理在测试时扩展（test-time scaling）过程中因稀疏二元奖励（Pass/Fail）导致的严重信用分配困境（credit assignment crisis），以及由此引发的失败探索轨迹浪费问题。现有轨迹优化与扩展方法存在计算成本高、结构受限的缺陷，其依赖启发式状态复用而缺乏因果诊断能力，或采用延迟的标量评分机制而无法提供可操作的在线指导。本文提出FLARE（Full-Lifecycle Alignment and Reward Engine），一种基于轻量化生成式奖励模型（Generative Reward Model, GRM）的密集监督范式。其核心解决方案在于：首先，通过离线的因果感知诊断框架RADAR，利用因果链回溯（causal-chain backtracking）提取高质量、无事后偏差（hindsight-free）的监督信号，以蒸馏出可实时提供步骤级风险反馈的GRM；其次，FLARE在推理阶段作为主动支架（Active Scaffold），自主拦截高风险生成步骤并进行局部断点重执行，显著降低计算开销；在后训练阶段，GRM提供的结构化信号可作为过程监督下的重排序分数用于监督微调（Supervised Fine-Tuning, SFT），以及步骤级密集奖励用于强化学习（Reinforcement Learning, RL），有效缓解稀疏环境中的策略坍缩问题。实验表明，FLARE在代理全生命周期内建立新的帕累托前沿：单次运行（N=1）性能超越全局滚动（Global Rollout, N=5），token消耗降低5倍；扩展至训练阶段后，在SFT中通过过程感知数据筛选实现19.13%的相对性能提升，在RL中保持稳定的9.19%改进。
链接: https://arxiv.org/abs/2609.23808
作者: Jingxuan Xu,Gang Wu,Yanan Wu,Yutao Mou,Songwei Yu,Tianzhuang He,Zhengshuo Gong,Zhao Liu,Zihang Xu,Wenqiang Zhu,Xinping Lei,Weihao Li,Yuhui Bai,Zhongqiu Wang,Yan Wu,Ariel Deng
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:While test-time scaling enhances Large Language Model (LLM) agents in long-horizon software engineering (SWE), sparse binary rewards (Pass/Fail) create a severe credit assignment crisis and waste failed exploratory trajectories. Current trajectory optimization and scaling methods are costly and structurally limited, relying on heuristic state reuse without causal diagnosis or delayed scalar scoring without actionable online guidance. We propose FLARE (Full-Lifecycle Alignment and Reward Engine), a novel dense supervision paradigm driven by a lightweight Generative Reward Model (GRM). First, RADAR, an offline causal-aware diagnostic framework, extracts high-fidelity, hindsight-free supervision through causal-chain backtracking to distill a GRM providing real-time, step-level risk feedback. Second, FLARE uses this GRM to continuously optimize the agent across its entire lifecycle. During inference, FLARE acts as an Active Scaffold, autonomously intercepting high-risk generation steps for localized breakpoint re-execution, drastically reducing compute overhead. During post-training, the GRM’s structured signals serve as process-supervised reranking scores for Supervised Fine-Tuning (SFT) and step-level dense rewards for Reinforcement Learning (RL), mitigating policy collapse in sparse environments. Extensive evaluations show that FLARE establishes a new Pareto frontier across the agent lifecycle: FLARE (N=1) outperforms Global Rollout (N=5) with a 5x reduction in token consumption. Extending FLARE to training overcomes the sparse reward problem in long-horizon interactive tasks, delivering relative performance gains of 19.13% in SFT through process-aware data curation and a consistent 9.19% improvement in RL.
[NLP-61] Constrained Decoding Eliminates Structural Failures in Small LLM s but Reveals a Scale-Dependent Semantic Gap ACL
【速读】： 该论文旨在解决小规模开源大语言模型（0.6B-4B参数量级）在结构化输出生成任务中，受限解码（Constrained Decoding, CD）与模型规模之间的相互作用机制不明确的问题。其核心挑战在于：尽管CD能有效提升输出的格式合规性，但其对语义正确性的提升效果受限于模型规模，且不同类型的错误对CD的响应能力存在差异。解决方案的关键在于提出一种双轴评估框架，将结构正确性（即模式有效性，schema validity）与语义正确性（即内容准确性，content accuracy）进行分离量化。研究发现，CD可完全消除所有模型的结构错误（使模式有效性从78.6%-92.9%提升至100%），但语义层面仍存在依赖模型规模的持续差距：类型强制错误可通过CD完全修复，而指令-语义类错误（如多步函数调用）则表现出显著的CD抗性。研究进一步揭示，模式符合性虽为语义正确的必要条件，却非充分条件，且CD的有效边界恰好与模式符合性保持一致，表明其作用范围受限于语法约束而非深层语义理解。
链接: https://arxiv.org/abs/2609.23742
作者: Akash Chavan
机构: 未知
类目: Computation and Language (cs.CL)
备注: 6 pages, ACL format Code and task suite: this https URL (tag v1.0)
Abstract:Small open-source large language models (LLMs) in the 0.6B-4B parameter range are increasingly deployed for structured output generation (JSON, function calling, data extraction), yet little is known about how constrained decoding (CD) interacts with model scale in this regime. We benchmark five models from three families across 14 structured-output tasks under three decoding conditions (native, Outlines, XGrammar). We introduce a two-axis evaluation that separates structural correctness (schema validity) from semantic correctness (content accuracy). We find that CD eliminates all structural failures across all models (schema validity: 78.6-92.9% to 100%), but content accuracy reveals a persistent semantic gap that is scale-dependent: type coercion failures are fully CD-rescuable, while instruction-semantic failures (e.g., multi-step function calling) remain CD-resistant. Schema conformance is necessary but not sufficient for semantic correctness; CD’s reach ends exactly where schema conformance ends.
[NLP-62] GRACE: Grounded Adversarial Reasoning over Canadian Law
【速读】： 该论文旨在解决现有法律自然语言处理（Natural Language Processing, NLP）基准在评估大语言模型（Large Language Models, LLMs）进行法律立场主张、在信息不完整条件下推理以及综合多部法律法规方面能力不足的问题，尤其针对加拿大联邦法律领域长期缺乏代表性数据的现状。其解决方案的关键在于构建了一个名为GRACE（Grounded Reasoning Adversarial Canadian LEgal examples）的数据集，该数据集包含1,915个基于加拿大联邦立法文本的问答-推理实例，覆盖对抗性辩护、不确定性推理和应用性推理三种推理模式。研究提出了一套完整的数据生成流水线，通过原始法条文本分割、情景化问题与推理生成，并结合无模型引文验证与大语言模型（LLM）质量审计进行筛选，确保数据的准确性和法律严谨性。作为概念验证，研究进一步微调了轻量级模型CLeAR-4B（Canadian Legal Adversarial Reasoning），并对比其在开卷与闭卷设置下与未修改的Qwen3-4B基线模型的表现。实验结果表明，当提供相关法律条文时，CLeAR-4B在与教师输出的一致性及法定引用行为上显著优于基线模型，但若缺乏法律条文支持，其推理的可解释性与准确性则急剧下降，这凸显了外部法规文本对模型推理的有效支撑作用。因此，GRACE数据集能够有效推动轻量化法律模型在给定法规基础上实现更精准、可信赖的法律推理。
链接: https://arxiv.org/abs/2609.23726
作者: Jiakang Xu,Wantong Huo,Udom Silparcha,Jonathan H. Chan
机构: University of Toronto; UTOA Computing Analytics Co., Ltd.(UTOA计算分析有限公司); King Mongkut’s University of Technology Thonburi
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models have shown strong performance across a range of legal tasks, but existing benchmarks rarely evaluate the ability to take and defend a legal position, reason under incomplete information, or synthesize multiple statutory provisions. This gap is particularly pronounced for Canadian law, which remains underrepresented in legal NLP. We introduce GRACE (Grounded Reasoning Adversarial Canadian LEgal examples), a dataset of 1,915 question-reasoning-answer instances grounded in Canadian federal legislation. GRACE covers three reasoning modes: adversarial advocacy, uncertainty, and applied reasoning. We develop a pipeline that partitions raw statutory text, generates scenario-based questions and reasoning, and filters examples through model-free citation verification and LLM-based quality auditing. As a proof of concept, we fine-tune CLeAR-4B (Canadian Legal Adversarial Reasoning), a lightweight model for grounded legal reasoning, and evaluate it against the unmodified Qwen3-4B base model in open- and closed-book settings. CLeAR-4B substantially improves agreement with teacher outputs and statutory citation behavior when the relevant act text is provided, while its grounding degrades sharply when the statute is withheld. These results suggest that GRACE can support the development of lightweight legal models that reason more effectively from supplied statutory text.
[NLP-63] STEVE: Stabilizing Textual Gradient-Based Prompt Optimization via Error-Driven Refinement and Regularized Verification AACL
【速读】： 该论文旨在解决文本梯度（textual-gradient）方法在提示词优化过程中因迭代更新不稳定性导致的性能下降问题。其核心挑战在于：已正确处理的样本产生的梯度具有噪声，且模型易过度专精于困难样本，从而损害对简单输入的泛化能力。为此，论文提出STEVE稳定化框架，其关键在于两个耦合机制：一是基于错误驱动的精细化（Error-Driven Refinement），仅根据分类错误的样本生成梯度，聚焦于有信息量的失败案例；二是正则化验证（Regularized Verification），将每次更新视为临时性调整，仅在硬样本性能提升不引发保留集（preservation set）显著退化时才接受更新。实验表明，该框架在十项推理基准、三类评估器/优化器模型及现有基线中均有效降低性能退化，提升提示词鲁棒性，且在GPT-5.4-mini/GPT-5.4等新模型与更大测试集上仍保持优势，证明了其在生成式AI（Generative AI）提示优化中的实用性与普适性。
链接: https://arxiv.org/abs/2609.23716
作者: Yifan Xu,Yixuan Li,Xinzhuo Li,Yixin Gu,Yifan Shen,Lijun Yu,Haohan Wang
机构: University of Illinois Urbana-Champaign(伊利诺伊大学厄本那-香槟分校); Google DeepMind(谷歌深度思维)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted to Findings of the Association for Computational Linguistics: AACL-IJCNLP 2026
Abstract:Textual-gradient methods automate prompt optimization through natural-language feedback, but their iterative updates can be unstable. We identify two sources of this instability: noisy gradients produced from already-correct examples and over-specialization to hard cases that degrades performance on simpler inputs. We introduce STEVE, a stabilization framework with two coupled mechanisms. Error-Driven Refinement generates gradients only from incorrectly handled examples, concentrating updates on informative failures. Regularized Verification treats every update as provisional and accepts it only when improvement on hard cases does not cause unacceptable regression on a preservation set. Across ten reasoning benchmarks, three evaluator/optimizer models, and established prompt-optimization baselines, STEVE reduces degradation and produces more robust prompts. Additional evaluations with gpt-5.4-mini/gpt-5.4 on symbolic reasoning, GSM8K-Platinum, and DS-1000 show that these gains persist with newer models and larger test sets. STEVE therefore provides a practical way to improve the stability and effectiveness of textual-gradient prompt optimization.
[NLP-64] Financial Language Models as Applied Artificial Intelligence Systems for News-Based Trading under Market Frictions
【速读】： 该论文旨在解决金融领域中语言模型（Financial Language Models）生成的决策信号在实际交易系统中有效性难以评估的问题，尤其缺乏一个能够综合考量事件时间可观测性、概率校准、执行时机、交易成本、流动性约束、容量限制、操作诊断与统计推断等市场摩擦因素的集成化评估框架。现有计算机科学在时序预测、文本分类、多模态股票预测、基于图的市场建模及机器学习运维等方面虽有成熟方法，但均未形成面向金融场景的专用协议以联合测试语言模型输出的实际效用。为此，论文提出MFAST（Market-Friction-Aware Sentiment-to-Trading）框架，将带有时间戳的金融文本转化为可审计、可复现且具备市场可行性的交易决策。其核心解决方案在于构建一个端到端的工程化评估体系：通过将Refinitiv新闻分析数据与CRSP股权数据对齐，严格限定样本外评估范围为发布后且超出基础模型数据新鲜度周期的新闻，并引入基于公开金融文本与价格数据的可复现验证模块。实证结果表明，仅使用解码器结构的语言模型在分类准确率、概率校准、收益预测及净投资组合表现上均优于编码器基线与词典情感分析方法；同时，操作诊断揭示了精度、延迟、内存占用、吞吐量与推理成本之间的权衡关系。研究证明，可信的金融语言模型评估必须融合语言理解、时间纪律、市场摩擦感知部署以及可复现验证的全流程工程方法。
链接: https://arxiv.org/abs/2609.23703
作者: Kemal Kirtac
机构: University College London (伦敦大学学院); University of Warwick (华威大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG); Trading and Market Microstructure (q-fin.TR)
备注: 47 pages. Revise and resubmit at Engineering Applications of Artificial Intelligence
Abstract:Financial language models can transform unstructured firm-specific news into structured decision signals, but financial AI research lacks an integrated deployment framework for evaluating whether those signals remain useful in financial decision systems. Computer science research has developed strong methods for time-series forecasting, text classification, multimodal stock prediction, graph-based market modeling, and machine-learning operations, yet these streams do not provide a domain-specific protocol that jointly tests financial language-model outputs under event-time observability, probability calibration, execution timing, transaction costs, liquidity constraints, capacity limits, operational diagnostics, and statistical inference. We introduce MFAST, a Market-Friction-Aware Sentiment-to-Trading framework that converts timestamped financial text into auditable, reproducible, and market-feasible trading decisions. The application is news-based trading, where firm-specific text must be linked to securities before portfolio decisions can be evaluated. The framework links Refinitiv News Analytics to Center for Research in Security Prices (CRSP) equity data, restricts the primary out-of-sample evaluation to post-release news outside disclosed foundation-model data-freshness periods, and adds a public replication arm using open financial text and public price data. Results show that decoder-only language models outperform encoder baselines and dictionary sentiment in classification, calibration, return prediction, and net portfolio performance, while operational diagnostics reveal trade-offs among accuracy, latency, memory, throughput, and inference cost. The paper shows that credible evaluation of financial language models requires an end-to-end engineering approach combining language understanding, temporal discipline, market-friction-aware deployment, and reproducible validation.
[NLP-65] Distill What You Trust: Reliability-Aware Multi-Teacher On-Policy Distillation
【速读】： 该论文旨在解决多教师在线蒸馏（multi-teacher on-policy distillation）中教师选择机制的局限性问题，尤其是现有基于领域路由（domain-routed）的方法依赖于标注标签且无法动态适应轨迹内专家需求变化的问题。其核心解决方案是提出TrustMOPD，通过引入无标签的词元级（token-level）监督分配机制，取代传统的示例级教师选择策略。该方法在每个学生生成的前缀处，利用各专家在强化学习（RL）过程中相对于共享预强化学习参考点的位移作为局部可靠性代理，对不同教师的可靠性得分进行校准，并构建加权蒸馏目标。实验表明，TrustMOPD在数学、代码生成和指令遵循任务上显著优于现有最强的无标签基线，将SingleCap和MultiCap上的恢复率分别从54.4%和54.5%提升至91.5%和98.0%，并接近有标签的MOPD性能。随机化词元级权重的表现与均匀加权相当，验证了监督分配必须依赖于学生生成上下文演化的必要性。
链接: https://arxiv.org/abs/2609.23697
作者: Jie Sun,Mao Zheng,Mingyang Song,Zeyuan Liu,Gengsheng Li,Houcheng Jiang,Yilin Cheng,Bichuan Feng,Yuchen Cai,Junfeng Fang,Xiang Wang
机构: University of Science and Technology of China(中国科学技术大学); Foundation Model Department, Tencent(腾讯基础模型部门); Tsinghua University(清华大学); Institute of Automation, Chinese Academy of Sciences(中国科学院自动化研究所); Zhongguancun Academy(中关村学院); Nankai University(南开大学); National University of Singapore(新加坡国立大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 14 pages, 5 figures, 1 table
Abstract:Multi-teacher on-policy distillation allows a student to learn from complementary specialists on its own trajectories. Domain-routed approaches, however, select one teacher per example and keep it fixed throughout the response. This design both depends on labels that mixed training corpora often lack and cannot adapt teacher selection when the expertise required changes within a trajectory. We propose \textbfTrustMOPD, which replaces example-level teacher selection with label-free, token-level supervision allocation. At each student-generated prefix, TrustMOPD uses each specialist’s RL-induced displacement from a shared pre-RL reference as a proxy for local reliability, calibrates these scores across teachers, and constructs a weighted distillation target. Across mathematics, code, and instruction following, TrustMOPD outperforms the strongest label-free baseline, increasing the recovery ratio from 54.4% to 91.5% on \textscSingleCap and from 54.5% to 98.0% on \textscMultiCap, while approaching label-based MOPD on \textscSingleCap. Randomizing token-level weights independently of the student-generated prefix performs no better than uniform weighting, supporting the importance of conditioning supervision on the evolving generation context.
[NLP-66] Collapse Not Complexity: Failure-Conditioned Decomposition Repair for End-to-End Document Parsing ICASSP2027
【速读】： 该论文旨在解决端到端文档解析器在处理复杂页面时，其可选的推理模式（reasoning mode）实际效果不佳的问题。研究发现，在一个180页的熵分层发现样本上，使用固定40亿参数检查点时，以“复杂性”作为切换推理模式的决策变量是错误的：推理模式反而使平均质量下降2.21分，且消耗1.54倍的令牌数；而仅基于输入的预注册模型无法预测推理模式的正向收益（外部验证的受试者工作特征曲线下面积AUROC为0.47，与随机猜测无异）。进一步分析表明，推理模式的收益集中于那些普通解析路径已崩溃的页面，但这些页面的布局熵反而低于健康页面，却消耗了19倍的令牌，且增加预算也无法缓解这种退化重复问题。此外，切换至推理模式极少能修复此类崩溃，83%的崩溃页面在推理模式下仍会复发。为此，作者提出一种新解决方案：通过普通解析路径的执行轨迹检测崩溃现象，利用投影分解页面结构，并对每个区域重新解析。该方法在仅增加1.13倍令牌开销的情况下，整体性能提升1.40（95%置信区间[0.68, 2.16]），在三个不同检查点上均具可复现性；且在所有参数冻结条件下，对剩余1,175个基准页面的性能提升达2.41（置信区间[1.64, 3.46]），显著优于原推理模式。该方案的关键在于从解析过程的内部轨迹中识别崩溃模式，而非依赖表面复杂度判断，并通过局部重解析实现高效修复。
链接: https://arxiv.org/abs/2609.23592
作者: Xingyu Lin,Dehui Du
机构: 未知
类目: Computer Vision and Pattern Recognition (cs.CV); Computation and Language (cs.CL)
备注: 5 pages, 3 figures, 2 tables. Submitted to ICASSP 2027
Abstract:End-to-end document parsers increasingly offer an optional reasoning mode for complex pages. On a 180-page entropy-stratified discovery sample with one frozen 4B checkpoint, complexity is the wrong decision variable. Reasoning lowers mean quality by 2.21 Overall at 1.54x tokens; a preregistered input-only model cannot predict its signed benefit (held-out AUROC 0.47, indistinguishable from chance). The benefit concentrates on pages whose ordinary pass has already collapsed, and they do not look complex: shared collapses have lower layout entropy than healthy ones yet consume 19x the tokens as degenerate repetition that doubling the budget does not cure. Switching modes rarely repairs them: 83% recur under reasoning. We instead detect collapse from the ordinary-pass trace, decompose the page by projection, and re-parse each region. Repair gains 1.40 Overall (95% CI [0.68, 2.16]) at 1.13x tokens, replicates across three checkpoints, and, with all parameters frozen, gains 2.41 (CI [1.64, 3.46]) on the remaining 1,175 benchmark pages.
[NLP-67] On the Efficiency-Safety Dilemma in Large Reasoning Models EMNLP2026
【速读】： 该论文旨在解决大推理模型（Large Reasoning Models, LRMs）在采用量化与剪枝等效率优化技术后，其对抗鲁棒性（adversarial robustness）与越狱攻击（jailbreak vulnerability）之间关系不明确的问题。研究发现，尽管效率技术表面上降低了越狱攻击的成功率，但这种“提升”实则源于模型推理能力的退化，导致模型无法有效生成恶意响应，而非真正实现了安全对齐。通过表征漂移（representational drift）的机制分析进一步证实，推理能力下降与模型维持恶意语义轨迹的能力丧失存在严格耦合关系。因此，该研究的关键解决方案在于识别出“量化结合剪枝”是兼顾效率与鲁棒性的最优策略，并强调了真实安全对齐与能力退化导致的失败之间的本质区别，为大推理模型的实际部署提供了重要的实证依据。
链接: https://arxiv.org/abs/2609.23587
作者: Yifei Yang,Zouying Cao,Xingrui Wang,Xiao Zhou,Yuexian Li,Dongjie Yang,Hai Zhao
机构: Shanghai Jiao Tong University (上海交通大学); AGI Institute (人工智能研究院); Key Laboratory of Shanghai Education Commission for Intelligent Interaction and Cognitive Engineering (上海市教育委员会智能人机交互与认知工程重点实验室); Ant Group (蚂蚁集团); Paris Elite Institute of Technology, Shanghai Jiao Tong University (上海交通大学巴黎卓越工程师学院); Tencent (腾讯)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted as Main of EMNLP2026
Abstract:Large reasoning models (LRMs) incur high inference costs, often mitigated by efficiency techniques like quantization and pruning. However, the impact of these techniques on model adversarial robustness remains largely unexplored. This study provides the first comprehensive analysis of the interplay between efficiency, jailbreak vulnerability, and reasoning in LRMs. We find that while efficiency methods seemingly reduce the success rate of jailbreak attacks, this improvement is often superficial. It largely arises from degraded reasoning capabilities leading to “attempted but failed” malicious responses, rather than an increase in genuine alignment. Mechanistic analysis of representational drift confirms this, revealing a strict coupling between reasoning capability loss and the model’s inability to maintain malicious semantic trajectories. Additionally, we identify quantization with pruning as the optimal strategy to balance efficiency and robustness. These findings clarify the distinction between true safety alignment and capability-induced failure, providing an empirical foundation for LRM deployment.
[NLP-68] Global Ranks Survive Selected Heads Shift: BOS-Sink Topology under 4-bit Weight-Only Quantization
【速读】： 该论文旨在解决生成式 AI 模型在进行 4 位 NF4 权重仅后训练量化（PTQ）时，是否可安全复用量化前通过“感知汇点（sink-aware）”方法识别出的重要首令牌注意力头（first-token attention heads）的问题。其核心挑战在于：尽管全局权重排名在量化后仍保持较高一致性（如 ρ_s ≥ 0.980），但关键的局部结构——如 top-k 头集合的交集、层内汇点质量分布偏移及跨域校准迁移性——可能显著退化。解决方案的关键在于提出一种名为 Sink Topology Consistency (STC) 的多维度评估框架，能够分离并量化全局秩保持、top-k 集合重叠、层间汇点质量漂移，以及输入敏感性与校准映射迁移之间的差异。实验表明，虽然全局排名具有良好的可迁移性，但具体头集合的保留率仅为 76.5–88.5%，且终端层存在高达 6.2–7.9 倍模型均值的显著偏移；同时，在跨数据域（如 C4 到 LongBench）转移时，性能下降更严重。因此，尽管小样本（n=8）重新校准即可达到约 90% 的稳定平台，但仅更新部分层无法满足全图稳定性要求。最终结论强调：全局排序可复用，但离散头集、层内策略和跨域校准必须在量化后重新验证，以确保边缘部署中模型行为的可靠性。
链接: https://arxiv.org/abs/2609.23585
作者: Kuanlin Chen,Chen-Wei Kuo,Cheng-En Ou
机构: National Tsing Hua University (国立清华大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: Accepted for publication at IEEE IECON 2026. 6 pages, 3 figures, 7 tables
Abstract:Sink-aware deployment may identify important first-token attention heads before a model is quantized, then reuse that map at the edge. We test when this shortcut is safe for 4-bit NF4 weight-only post-training quantization (PTQ). Our Sink Topology Consistency (STC) metrics separate global rank preservation, top- k set overlap, and layerwise sink-mass shift, and distinguish per-input sensitivity from calibration-map transfer. Across Qwen2.5-0.5B, Qwen2.5-1.5B, and Llama-3.2-1B, global bf16-to-4-bit ranks remain high at 4,096 tokens ( \rho_s \geq 0.980 ), yet top- k Jaccard overlap is only 0.619-0.793, corresponding to 76.5-88.5% membership retention. The global statistic also masks local failures: terminal Qwen layers shift by 6.2-7.9x their model means, whereas Llama-3.2-1B shows low, nearly uniform drift. Under a C4-to-LongBench shift, cross-domain overlap degrades more than the within-domain precision comparison for both Qwen models, but not for Llama-3.2-1B. Matched-domain 4-bit recalibration reaches 90% of a split-half stability plateau at the smallest tested n=8 for both Qwen models and n=32 for Llama-3.2-1B, though not as a sharp threshold; for the two Qwen models, updating only selected layers does not reach the full-map stability criterion. On Jetson Orin NX, the 16-sample workload takes seconds for the two models with valid on-device sink measurements. The practical message is precise: global rankings often transfer, but discrete head sets, layer-local policies, and cross-domain calibration should be revalidated after quantization.
[NLP-69] ARID: A Deployable Edge AI System for Structured Information Extraction from Industrial Maintenance Work Orders
【速读】： 该论文旨在解决工业维护工单在嵌入式硬件上离线处理时，下游软件对可预测结构化输出的需求与实际非结构化文本输入之间存在的矛盾。核心问题在于如何在资源受限的边缘设备（如8 GB NVIDIA Jetson Orin NX）上实现高效、准确且语法合规的自然语言到结构化数据的转换。其解决方案的关键在于提出ARID（Aviation-inspired Routing for Industrial Deployment），该方法融合了保守型双教师过滤（conservative dual-teacher filtering）、面向噪声的针对性合成（targeted noise-aware synthesis）、每条工单仅一次路由决策（one routing decision per work order）、4-bit量化推理（4-bit inference）以及语法约束解码（grammar-constrained decoding）。通过从2,326条未标注的OMIn记录中筛选出716条训练样本，并新增99条拓扑约束记录以强化操作项提取能力，ARID在300条人工标注数据上实现了参考栈84.8%的token-F1和部署环境82.9%的token-F1。系统在12.5 W功耗下实现5,310/5,656 ms的P50/P99延迟，具备高实时性；零样本迁移至MaintNet时，语义F1下降至46.4%，但解析成功率仍保持≥99.8%，表明输出格式的有效性可迁移，而领域语义则不可直接迁移，凸显了结构化输出生成中语法约束与任务特定语义分离的重要性。
链接: https://arxiv.org/abs/2609.23582
作者: Kuanlin Chen,Chen-Wei Kuo
机构: National Tsing Hua University (国立清华大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted for publication at IEEE IECON 2026. 5 pages, 6 figures, 3 tables
Abstract:Maintenance work orders must often be processed offline on embedded hardware, yet downstream software requires predictable structured output. We present ARID (Aviation-inspired Routing for Industrial Deployment), which extracts component, failure mode, symptom, and maintenance action into fixed-schema JSON on an 8 GB NVIDIA Jetson Orin NX. ARID combines conservative dual-teacher filtering, targeted noise-aware synthesis, one routing decision per work order, 4-bit inference, and grammar-constrained decoding. From 2,326 unlabeled OMIn records, it retains 716 training pairs and adds 99 topology-constrained records targeting action extraction. On 300 human-labeled records, ARID reaches 84.8% token-F1 on the reference stack and 82.9% on the deployed Jetson. Resident serving achieves 5,310/5,656 ms P50/P99 at 12.5 W. On zero-shot MaintNet transfer, semantic F1 falls to 46.4% while parser success remains at least 99.8%, showing that output validity transfers but field semantics do not.
[NLP-70] Error-Supervised Synthetic Learner Writing for Automated Essay Scoring
【速读】： 该论文旨在解决生成式作文（synthetic essays）在自动作文评分（AES）系统训练中因缺乏真实语言学习者常见语法错误而难以有效模拟真实人类写作的问题。现有方法生成的合成作文通常过于“完美”，无法体现目标用户群体——语言学习者在写作中常见的语言错误模式，从而限制了其在低资源场景下的适用性。本文提出的关键解决方案是：在大语言模型（LLM）生成合成作文的过程中引入语法错误标注（error-annotated）数据进行微调，通过在生成阶段显式监督错误分布，使合成作文更贴近真实学习者写作特征。实验表明，在多数数据规模条件下，该方法生成的合成作文显著优于传统合成基线，在12个评估指标中的11个上表现更优，部分情况下甚至接近使用真实作文训练的模型性能；定量与定性分析进一步验证了所生成文本中错误类型和分布与真实学习者作文高度相似。尽管在极低资源设置下（如仅200篇训练作文）效果仍不一致，但随着数据量增加，该方法的优势逐渐显现，证明了错误监督机制在提升合成作文真实性方面的有效性。
链接: https://arxiv.org/abs/2609.23573
作者: Duy Anh Nguyen
机构: University of Greenwich (格林威治大学); FPT University (FPT大学)
类目: Computation and Language (cs.CL)
备注: 17 pages, 1 figure, 9 tables
Abstract:Synthetic essays can help reduce dependence on human-written data in Automated Essay Scoring (AES). However, they often lack realistic errors, limiting their ability to represent authentic human writing, particularly when the target texts are intended to resemble those produced by language learners. In this study, we present a simple approach that introduces error supervision into synthetic essay generation. Specifically, we fine-tune an LLM generator on error-annotated texts of the kind commonly used in Grammatical Error Detection (GED). To assess the utility of the proposed approach, we fine-tune and evaluate AES scorers under three data conditions: authentic essays, synthetic essays generated conventionally, and synthetic essays generated using our proposed approach. The results show that in the larger-data settings, the proposed approach outperforms the conventional synthetic baseline in 11 out of 12 dataset-metric comparisons, with performance in some cases approaching that of models trained on authentic essays. Despite these gains, performance under extremely low-resource settings remains mixed, with advantages over the conventional baseline only becoming more apparent at 200 training essays, although not consistently across datasets. Qualitative and quantitative analyses further show that the proposed approach produces learner-like errors whose distributions broadly resemble those observed in authentic essays.
[NLP-71] VibeMemBench: Evaluating Memory Systems for Coding Agents on Real Repository Coding Tasks
【速读】： 该论文旨在解决现有编码代理（coding agents）在真实代码库任务中，持久性记忆系统（persistent memory systems）是否真正提升可执行代码工作成效的问题。当前评估体系存在两大缺陷：仓库基准测试（repository benchmarks）虽能检验代码变更的可执行性，但无法隔离记忆机制的作用；而记忆基准测试（memory benchmarks）仅衡量记忆召回能力，却未评估其对下游编码任务的实际影响。为此，论文提出VibeMemBench——一个针对90个SWE-rebench V2仓库中的111个编码目标及3,634条历史轨迹构建的基准测试，涵盖漏洞修复、功能请求、接口变更和配置调整等典型任务。该基准通过在受控条件下让代理在声明的记忆状态下编辑代码库，并以可执行测试判定任务完成度，确保每个任务均基于经过验证的、能提升执行结果的历史经验。经冻结并转移至五个独立求解器后，直接注入历史经验使其中四个求解器的任务完成率提升1.1至4.5个百分点，同时减少所有求解器的平均操作步数。然而，当四种现有记忆系统需从相同历史数据中自主构建与检索经验时，十二组系统-求解器组合中有十一次未能超越“无记忆”基线。该研究揭示了代码仓库历史中蕴含的有效经验与现有记忆系统实际交付的经验之间存在的显著差距。
链接: https://arxiv.org/abs/2609.23570
作者: Liyang Fan,Yingcheng Shi,Yongbin Li,Chenghao Sun,Xin Chen,Xander Xu,Hu Wei,Shiwen Ni,Min Yang,Jieping Ye
机构: Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences(中国科学院深圳先进技术研究院); SUAT; Alibaba Token Hub, Alibaba Group(阿里巴巴通证中心); Alibaba Group(阿里巴巴集团)
类目: oftware Engineering (cs.SE); Computation and Language (cs.CL)
备注:
Abstract:Coding agents operate on real repository coding tasks, and persistent memory systems promise to reuse experience across tasks. Yet existing evaluations do not show whether those systems improve executable repository work. Repository benchmarks test code changes but do not isolate memory, while memory benchmarks score recall without measuring downstream coding outcomes. We introduce VibeMemBench, a benchmark for evaluating memory systems on 111 coding targets from 90 SWE-rebench V2 repositories and 3,634 history trajectories from the target repositories. The targets follow the SWE benchmark style and cover bug fixes, feature requests, interface changes, and configuration work. An agent edits each target codebase under a declared memory condition. Executable tests decide task resolution. Each target is retained only when injected history experience improves its executable outcome in a reference setting, so every target carries a prior experience whose usefulness is verified by execution in that setting. The frozen verified experience is then transferred to five held-out solvers. Direct injection raises observed task resolution on four of them by 1.1 to 4.5 percentage points while lowering agent steps on all five. Yet when four existing memory systems must construct and retrieve experience from the same history, eleven of twelve solver and system pairings fail to exceed the matched memory-off baseline. VibeMemBench exposes the gap between the useful experience that repository history holds and the experience existing memory systems deliver for repository coding tasks.
[NLP-72] Contributions to the hierarchy of probabilistic languages
【速读】： 该论文旨在解决概率形式语言在不同生成模型下的表达能力比较问题，具体聚焦于n-gram模型与概率上下文无关文法（PCFG）之间的语言层级关系。其核心问题是：是否存在某些由PCFG生成的概率语言无法被任何n-gram模型所表示？为此，论文通过理论证明确立了概率语法的预期层级结构——所有由n-gram模型生成的概率语言均可由某类PCFG生成，但存在部分由PCFG生成的概率语言无法被任何n-gram模型所捕获。解决方案的关键在于引入“全连接型PCFG”（fully connected PCFG）的概念，即在乔姆斯基规范形式下，仅包含非终结符的产生式规则均具有非零概率。研究进一步证明，任意由n-gram模型生成的概率语言与任意由全连接型PCFG生成的概率语言之间均存在本质差异，从而得出结论：n-gram模型所生成的概率语言类并非全连接型PCFG生成语言类的子集，这一结果揭示了两类模型在表达能力上的根本区别。
链接: https://arxiv.org/abs/2609.23567
作者: Lothar Sebastian Krapp,Remo Nitschke
机构: University of Zurich (苏黎世大学); University of Konstanz (康斯坦茨大学)
类目: Formal Languages and Automata Theory (cs.FL); Computation and Language (cs.CL); Logic (math.LO); Probability (math.PR)
备注:
Abstract:We reconsider the theory of probabilistic formal languages generated by n-gram models and by probabilistic context-free grammars (PCFGs). The expected hierarchy of probabilistic grammars is established by proving that every probabilistic language generated by an n-gram model is also generated by some PCFG, while some probabilistic languages generated by PCFGs cannot be generated by any n -gram model. We introduce the notion of fully connected PCFGs, namely PCFGs in Chomsky normal form where every production rule only involving non-terminals has non-zero probability. Our main result shows that any probabilistic language generated by an n -gram model differs from any probabilistic language generated by a fully connected PCFG. Therefore, the class of probabilistic languages generated by n -gram models is not a subset of the class generated by fully connected PCFGs.
[NLP-73] Parag raph Boundaries Are Not White Space:Compression Depth as the Signature of Hierarchical Structure
【速读】： 该论文旨在解决传统位置编码（positional encoding）在建模文本层次结构时的局限性，即仅依赖一维阅读顺序无法充分捕捉段落、句子与词元之间的层级关系。其核心解决方案是提出一种分层旋转位置编码（hierarchical rotary positional encoding, hRoPE），将段落、句子和词元索引分别作为独立通道进行编码，并固定词元序列，通过干预段落坐标 p1 来精确估计跨段落注意力。研究发现，尽管所有语料库中注意力均表现出相对于基于词元距离匹配基线的压缩现象，但压缩程度本身并非真实结构的可靠指标——因为使用密度匹配的随机标签构建的对照模型同样出现压缩，且程度更浅。真正区分真实结构的是压缩深度，其具有显著的语料库依赖性和更深的压缩特征。进一步分析表明，尽管多种仅基于语料库的指标（如词汇持久性、段落长度、基于嵌入的一致性）在一定程度上反映结构特性，但仅有基于嵌入的一致性最接近真实结构的跨语料库排序。因此，在该设定下，压缩深度而非其具体位置，是可复现的、表征真实段落结构的关键信号。
链接: https://arxiv.org/abs/2609.23551
作者: Shuyang Xiang
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Standard positional encodings represent position as a one-dimensional reading-order coordinate, but reading order alone does not determine hierarchical textual structure. We use a hierarchical rotary positional encoding (hRoPE) that represents paragraph, sentence, and token indices as separate channels, hold the token sequence fixed, intervene on the paragraph coordinate p1, and measure cross-paragraph attention with a token-distance-exact estimator. Attention is compressed relative to a token-distance-matched baseline in every corpus, but compression alone is not diagnostic of true structure: an architecturally identical channel with density-matched random labels is compressed too, more shallowly. What distinguishes real structure is the depth of compression, which is greater and corpus-dependent while the control’s is not. Comparing eight corpus-only quantities across three constructs (lexical persistence, paragraph length, embedding-based coherence), none fully reproduces the cross-corpus ordering of depth, though embedding-based coherence comes closest. Compression depth, not its location, is the reproducible signature of genuine paragraph structure in our setting.
[NLP-74] BabelArena: A Large-Scale Multilingual Benchmark for LLM Agents
【速读】： 该论文旨在解决当前大型语言模型（LLM）智能体评估体系严重依赖英语、缺乏多语言适应性的问题，导致对智能体在非英语环境下的真实能力认知存在显著偏差。其核心解决方案是提出BabelFlow——一个通用的多语言智能体工作流基准框架，通过分析运行时依赖关系，实现结构保持的翻译协调，并结合多层级验证与人工审核，确保任务语义与评估标准在跨语言迁移中的完整性。基于此框架构建的BabelArena基准包含16,146个实例，覆盖702个经典任务、13个领域和23种语言。实验表明，现有前沿模型在不同基准家族间无统一优势，且跨语言能力差距远超任务成功率差异；低资源语言中智能体主要表现为工具调用与控制流错误频发，而非答案质量缺陷，揭示了多语言任务执行可靠性在资源分布上的系统性短板。此外，在相同任务下，低资源语言中的智能体输入消耗量可达英语的两倍以上，但交互长度并未相应增加，且在需结构化输出的任务中语言一致性显著下降，语言切换几乎完全倾向英语。研究认为，BabelArena为推动高效、可靠的多语言智能体研究提供了关键基础。
链接: https://arxiv.org/abs/2609.23490
作者: Peng Kuang,Yuchun Fan,Jiangnan Li,Minghao Wu,Jialong Tang,Hao-Ran Wei,Weixuan Wang,Jianhong Tu,Baosong Yang,Tong Xiao
机构: Alibaba Token Hub, Alibaba Group(阿里巴巴令牌中心，阿里巴巴集团); University of Illinois Urbana-Champaign(伊利诺伊大学厄本那-香槟分校); Northeastern University(东北大学); Monash University(莫纳什大学); Ant International(蚂蚁国际)
类目: Computation and Language (cs.CL)
备注: 20 pages, 11 tables, and 7 figures
Abstract:Large language model (LLM) agents increasingly execute multi-step workflows through tool use and interaction with users and environments. However, current agent evaluations are largely English-centric, limiting our understanding of agent capabilities in multilingual settings. We introduce BabelFlow, a benchmark-general agentic workflow that adapts existing agent benchmarks to new languages by analyzing runtime dependencies, coordinating structure-preserving translation, and combining multi-layer verification with human review to preserve task and evaluation semantics. Using BabelFlow, we construct BabelArena, a task-aligned benchmark comprising 16,146 instances derived from 702 canonical tasks across four benchmark families, 13 domains, and 23 languages. Experiments with five frontier models show that no single model dominates across benchmark families and that cross-language disparities extend well beyond task success. Lower-resource languages exhibit distinct failure patterns, with larger shares of tool-use and control-flow errors rather than answer-quality errors alone, pointing to gaps in reliable task execution across the resource levels of these languages. On the same tasks, agents in low-resource languages also consume substantially more tokens than in English (up to roughly twice the input) without proportional increases in interaction length, and language consistency degrades further on tasks requiring structured output, where switches are directed overwhelmingly toward English. We believe BabelArena provides a foundation for advancing research on reliable and efficient multilingual agents.
[NLP-75] RPMem: Learning Long-Term Recurrent Parametric Memory Across Sessions for LLM Agents
【速读】： 该论文旨在解决长时序大语言模型（LLM）智能体在跨会话场景下持续记忆的可演化性与可重用性问题。现有方法中，基于文本的记忆依赖于检索质量，随历史增长导致上下文推理负担加重；而传统参数化记忆虽能直接编码经验，却难以实现跨会话的记忆演进，且与特定主干模型强耦合，导致模型替换后记忆无法复用。为此，论文提出RPMem——一种两阶段架构：首先通过前向计算将每会话信息编译为与模型无关的潜在记忆（latent memory），再经由任务训练的循环门选择性地与保留记忆融合，实现跨会话记忆的动态整合。最终，融合后的记忆被映射为面向具体主干模型的低秩适配（LoRA）参数，从而支持在更换主干模型时仍可迁移记忆能力。实验表明，RPMem在三个长期记忆基准和五种不同主干模型上均表现出优异的泛化性能，更新成本与内存开销近似恒定。在Qwen3-8B与PERMA测试中，其准确率达85.52%，显著优于最强的参数化与文本基基线（分别提升5.32和12.98个百分点）。消融实验验证了会话编译与跨会话整合的互补作用，动态分析揭示门控机制能够习得任务特异的记忆融合策略。因此，RPMem构建了一个生命周期无关的参数化记忆框架，实现了可演化、可复用的跨会话记忆系统。
链接: https://arxiv.org/abs/2609.23466
作者: Fanyu Zhao,Ruike Cao,Liang Dong,Fugen Yao,Jian Xu,Guanjun Jiang,Han Zhang,Yifei Zhao,Yinsheng Li
机构: Fudan University(复旦大学); Qwen Applications Business Group, Alibaba Group(通义应用业务组，阿里巴巴集团)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 38 pages, 7 figures. Code: this https URL
Abstract:Long-running LLM agents require memory that persists and evolves across sessions. Text-based memory retrieves and reconstructs past interactions at every query, making long-horizon performance increasingly dependent on retrieval quality and contextual reasoning as histories grow. Parametric memory encodes experience directly into model computation, but existing approaches provide limited support for cross-session memory evolution. Their coupling to a specific backbone further restricts memory reuse after model replacement. We introduce RPMem, a two-stage architecture that compiles each session into a model-independent latent memory through forward computation and selectively integrates it with retained memory via a task-trained recurrent gate. The consolidated memory is then mapped to backbone-specific low-rank adaptation (LoRA) parameters, allowing the encoding capability to transfer when the backbone is replaced. Evaluation across three long-term memory benchmarks and five diverse backbones demonstrates broad generalization with near-constant update cost and memory footprint. With Qwen3-8B on PERMA, RPMem reaches 85.52%, outperforming the strongest parametric and text-based baselines by 5.32 and 12.98 percentage points, respectively. Ablations validate the complementary roles of session compilation and cross-session consolidation, while dynamics analyses reveal that the gate acquires task-specific memory integration strategies. These results establish RPMem as a lifecycle-independent parametric memory framework that maintains evolving cross-session memory that remains reusable across backbone replacements. Our implementation is available at this https URL.
[NLP-76] Propose Verify Commit: Evidence-Grounded Memory for Long-Horizon Multi-Actor Conversations
【速读】： 该论文旨在解决多参与者对话场景下长时程对话记忆（long-horizon conversational memory）的建模难题，尤其针对信息在多个参与方之间分散、上下文动态变化以及历史信息可能被后续内容修正等挑战。其核心解决方案是提出EGMEMORY，一种将长时程多参与者记忆建模为可检索的状态机（searchable state machine）的框架，通过显式分离持久性的消息级证据（message-level evidence）与活跃状态（active state），实现对复杂对话历史的有效管理。在写入阶段，采用自适应状态解析（adaptive state resolution）与基于证据的“提议-验证-提交”（propose-verify-commit）协议来控制状态演化；在读取阶段，通过自适应证据导航（adaptive evidence navigation），利用对话结构缩小搜索空间，并结合词汇-语义相关性对候选证据进行排序，以高效响应查询。该系统完全依赖提示工程（prompting）与工具调用（tool use），无需专门的记忆策略训练。实验结果表明，EGMEMORY在GroupMemBench和EverMemBench上的得分分别达到68.2%和77.9%，显著优于最强基线模型（提升22.7和21.4个百分点），并在二元对话任务LoCoMo上取得73.6%的准确率，展现出良好的泛化能力。
链接: https://arxiv.org/abs/2609.23465
作者: Zihao Lu,Zhihang Yuan,Lei Shi
机构: Alibaba Cloud Computing(阿里云计算), Hangzhou, China
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Long-horizon conversational memory is especially challenging in multi-actor settings, where relevant evidence is distributed across participants and contexts and previously established information may later be revised. We introduce EGMEMORY, which formulates long-horizon multi-actor memory as a searchable state machine that separates persistent message-level evidence from an explicit active state. At write time, adaptive state resolution and an evidence-grounded propose-verify-commit protocol govern how this state evolves. At read time, adaptive evidence navigation iteratively resolves the state and supporting evidence required for a query, using conversational structure to narrow the search space and lexical-semantic relevance to rank candidates. The system operates through prompting and tool use without memory-specific policy training. EGMEMORY achieves 68.2% on GroupMemBench and 77.9% on EverMemBench, outperforming the strongest evaluated baselines by 22.7 and 21.4 percentage points, respectively. It further reaches 73.6% on the dyadic LoCoMo benchmark, demonstrating generalization beyond multi-actor conversations. We will release the codebase upon formal publication.
[NLP-77] Perplexity Predicts Protection: Choosing Pretrained Backbones for Worst-Client Fairness in Federated Parameter-Efficient Fine-Tuning
【速读】： 该论文旨在解决联邦学习中数据异构导致的“最差客户端”性能劣化问题，即在整体平均准确率表现良好时，数据量极少的客户端仍可能获得较差的模型性能。其核心解决方案的关键在于：在使用低秩适应（LoRA）微调时，选择预训练主干网络（backbone）的先验信息——特别是基于目标任务文本的逐词困惑度（per-word perplexity）——可有效预测哪个主干网络能显著提升最差客户端的性能。实验表明，较低困惑度的主干网络在所有九个数据集-主干组合中均带来更大的客户端收益，且与最差客户端性能提升呈高度负相关（秩相关系数为-0.87）。此外，研究发现即使采用Ditto等个性化方法，也仅能恢复全联邦训练与仅本地训练之间差距的4%-12%，而完全移除参数聚合则彻底消除联邦学习的优势；同时，客户端更新与全局更新方向近似正交，排除了更新冲突作为性能失败的原因。因此，实际建议是：在联邦学习启动前，通过采样任务文本计算预训练主干的困惑度以优选主干，并避免依赖个性化机制来补偿数据贫乏客户端。
链接: https://arxiv.org/abs/2609.23463
作者: Kiran Naseer,Samreen Azhar,Umar Shoaib,Haroon Mahmood,Muhammad Awais
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Federated learning lets multiple parties train a shared model without pooling their data, but a client with far less data than the others can end up poorly served even when the group’s average accuracy looks fine. We ask whether the choice of pretrained backbone affects this under LoRA fine-tuning, and whether per-word perplexity on the target text predicts which backbone helps the worst-off client before federated training starts. We ran 313 experiments across three text-classification datasets and three similarly sized backbones (RoBERTa, BERTweet, PubMedBERT), each compared against a task-specific baseline on identical data splits. Lower-perplexity backbones consistently produced larger gains for the worst-performing client, with a rank correlation of -0.87 across nine dataset-backbone pairs; a backbone held out of the analysis confirmed the pattern. Personalization with Ditto recovered only 4-12% of the gap between training alone and full federation, and removing aggregation entirely erased the benefit. A client’s update also showed no sign of conflicting with the group’s update; the two are close to orthogonal, ruling out one proposed explanation for this failure. Practically: measure perplexity on a sample of task text before choosing a backbone, and do not rely on personalization to protect a data-poor client. We release our code, predictions, and full results for others to test.
[NLP-78] ool-Augmented On-Policy Distillation for LLM Domain Adaptation in Sequence-Based Omics Tasks
【速读】： 该论文旨在解决多组学序列（multi-omics sequences）中生成式模型在自动化科学发现过程中缺乏可解释性推理的问题，即现有模型虽能在分类或回归任务中取得较高准确率，但其预测结果往往缺乏对底层生物机制的正确理解与可追溯的证据支持。其核心挑战在于：如何评估并提升模型在多组学任务中的科学推理能力，而不仅仅是依赖统计模式进行预测。解决方案的关键是提出首个面向多组学序列的推理基准测试OmicsBench，包含1,160个由领域专家验证的问题，覆盖DNA调控、RNA加工和蛋白质功能等六大任务，并要求模型提供可追踪的证据链，通过与领域专家共同制定的实例化评分标准进行评估。基于该基准的实验揭示了科学专用大语言模型（scientific LLMs）存在“捷径学习”现象——即虽在预测准确性上优于通用模型，却难以提供有效生物学证据。为此，研究提出工具增强的在线策略蒸馏（TA-OPD），一种后训练方法，通过引入外部生物工具与证据感知的目标函数，引导模型将序列预测与真实生物学机制对齐。在五种参数量从0.8B到27B的Qwen3.5模型上，TA-OPD均显著增强了模型的生物学证据依附性，同时多数任务中提升了预测性能，表明更强的序列推理能力并非仅源于模型规模扩大，而是可通过有意识的证据导向训练实现。因此，OmicsBench与TA-OPD共同构建了一个诊断多组学大模型推理缺陷并促进其向生物学可解释性演进的系统性框架。
链接: https://arxiv.org/abs/2609.23435
作者: Jie Ying,Zhefan Wang,Zihong Chen,Zhengqing Li,Jinzhe Li,Gang Li,Jian Liu,Fang Hu,Tao Luo,Zhonghang Yuan,Wanli Ouyang,Stan Z. Li,Fan Yang,Nanqing Dong
机构: Shanghai Artificial Intelligence Laboratory(上海人工智能实验室); Yazhouwan National Laboratory(崖州湾国家实验室); Shanghai Innovation Institute(上海创新研究院); The Chinese University of Hong Kong(香港中文大学); Westlake University(西湖大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 18 pages, 5 figures
Abstract:Multi-omics sequences contain complex biological patterns, yet deciphering their mechanisms for automated scientific discovery remains challenging. As large language models (LLMs) interpret these sequences, evaluating both predictions and scientific reasoning is critical. However, existing benchmarks for multi-omics sequence tasks rely on classification and regression metrics, neglecting whether models grasp the underlying biological evidence. We introduce OmicsBench, the first reasoning benchmark for multi-omics sequences, comprising 1,160 expert-validated questions across six tasks spanning DNA regulation, RNA processing, and protein function. OmicsBench requires traceable evidence chains, evaluated using instance-specific rubrics developed with domain experts. Evaluating 17 LLMs reveals an inverse relationship: while scientific LLMs outperform general-purpose LLMs in sequence classification accuracy, they fail to provide valid evidence to support their predictions. One plausible interpretation is shortcut learning: specialized models may rely on statistical patterns rather than the biological mechanisms needed for scientific discovery. Motivated by this finding, we introduce tool-augmented on-policy distillation (TA-OPD), a post-training method to align sequence prediction with evidence-grounded biological reasoning. Across five Qwen3.5 models spanning 0.8B to 27B parameters, TA-OPD consistently strengthens biological evidence grounding while improving predictive performance on most tasks. These gains persist across model scales, indicating that stronger sequence reasoning does not arise solely from increased model capacity, but can be improved through evidence-aware training. Together, OmicsBench and TA-OPD provide a framework for diagnosing reasoning failures in multi-omics LLMs and a path toward models whose predictions are better grounded in biologically meaningful evidence.
[NLP-79] MuLA-Bench: A Multilingual Long-Form Audio Understanding Benchmark via Multi-Tier Auditing
【速读】： 该论文旨在解决长时音频理解任务中评估指标单一化的问题，即当前研究普遍依赖上下文长度和整体准确率来衡量模型性能，忽视了语言、证据类型与任务需求三者共同作用下对任务难度的复杂影响。其核心解决方案是提出MuLA-Bench基准测试集，包含5,038个开放式问题，覆盖16种语言和8个领域，基于1,769段真实场景录音（总计1,377.9小时），通过平衡的“语言×领域”语义轨迹实现可控比较，并辅以保留自然非语音证据的声学轨迹以增强真实性。该基准采用证据驱动生成、捷径检测与语言专家评审相结合的方式构建可审计的问题，避免了依赖共享源文本翻译或人为注入目标音效带来的偏差。通过对十种音频-语言模型的评估及固定八模型组的聚合诊断发现，语言表现排名随领域和任务变化，声学-语义性能差距因操作类型而异，且即使正确事件被识别，时间错位错误仍可能持续存在。结果表明，长距离检索能力相对较强，但自然声学事件的时间精准对齐与事实锚定仍极为脆弱。因此，MuLA-Bench揭示了传统单一时长评分无法捕捉的条件性失败模式，为音频-语言模型的系统性评估提供了更精细、多维的分析框架。
链接: https://arxiv.org/abs/2609.23416
作者: Zeyu Yang,Xinyu Zhang,Zibo Bi,Pei Zhang,Xize Cheng,Jin Xu,Baosong Yang,Satoshi Nakamura
机构: The Chinese University of Hong Kong, Shenzhen(香港中文大学（深圳）); Alibaba Token Hub, Alibaba Group(阿里巴巴令牌中心)
类目: ound (cs.SD); Computation and Language (cs.CL)
备注:
Abstract:Long-form audio performance is often summarized by context length and aggregate accuracy, obscuring how language, evidence, and task jointly shape difficulty. We introduce MuLA-Bench: 5,038 open-ended questions over 1,769 in-the-wild recordings totaling 1,377.9 hours, covering 16 languages and eight domains. A balanced Language x Domain semantic track supports controlled comparisons, while a complementary acoustic track preserves naturally occurring non-speech evidence. Evidence-grounded generation, shortcut checks, and language-expert review provide auditable questions without translating a shared source set or injecting target sounds. We evaluate ten audio-language models and conduct pooled diagnostics on a fixed eight-model cohort. Language rankings change across domains and tasks; acoustic-semantic performance gaps vary with the requested operation; and temporal errors can persist after the correct event is identified. Long-range retrieval is comparatively strong, while precise clock alignment and factual grounding of natural acoustic events remain fragile. MuLA-Bench thus exposes conditional failure patterns that a single long-context score does not capture.
[NLP-80] One to More More to One: Category-Aware Iterative Expert Training for Software Engineering Agents
【速读】： 该论文旨在解决在聚合式智能体强化学习（pooled agentic reinforcement learning）框架下，代码库级软件工程（SWE）任务中不同任务类别进展不均的问题。具体表现为某些类别的任务性能提升的同时，其他类别出现退化，而整体平均性能的提升掩盖了这种类别间的“此消彼长”现象。其解决方案的关键在于提出一种类别感知的专家训练与策略融合框架：通过可执行任务构建和基于证据的多维度标注系统（SWE Labeler）组织训练池；采用初始的类别特定强化学习以提升整体训练成功率，但针对实例级进展不均衡的问题，引入显式的成功行为整合与自适应任务选择机制；利用同源类别专家交替执行长时程智能体小规模强化学习（Agentic-miniRL）与刷新-修复-扩展（RRE）流程，实现策略更新、成功轨迹复用及任务重选；最终通过标签路由的多教师在线蒸馏（Label-routed multi-teacher on-policy distillation, MOPD），将多个专家策略融合为单一可部署的学生模型，其中使用ReLU门控的奖励外推机制仅保留各教师相对于参考策略的改进方向。整个框架无需依赖外部模型提供解题轨迹或动作目标，显著提升了各任务类别的均衡性与整体性能，在Pro-618和SWE-bench Multilingual数据集上分别达到58.04%和59.00%的平均解决率，相较基线模型分别提升5.39和2.78个百分点。
链接: https://arxiv.org/abs/2609.23377
作者: Jie Zhao,Ziyu Jiang,Suhang Zheng,Minghui Shan,Xiaoxiao Xu,Lin Qu
机构: Alibaba Group(阿里巴巴集团); Model⋅Logics-SWE-Qwen3.6-27B
类目: oftware Engineering (cs.SE); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 44 pages, including appendices. Model available at this https URL
Abstract:Repository-level software engineering (SWE) comprises heterogeneous task categories, whose progress under pooled agentic reinforcement learning can be uneven: gains in some categories coincide with regressions in others, while aggregate resolution obscures these changes. Motivated by this category see-saw, we develop a category-aware expert-training and policy-integration framework. Executable task construction and SWE Labeler, an evidence-grounded multi-axis labeling system, organize the training pools. Initial category-specific RL improves average training success while leaving uneven instance-level progress, motivating explicit consolidation of successful behavior and policy-adaptive task selection. Same-origin category experts alternate long-horizon Agentic-miniRL with Refresh-Repair-Expand (RRE): the updated policy refreshes instance mastery, reuses its own verified successful trajectories for Repair SFT, and reselects tasks for further RL. Label-routed multi-teacher on-policy distillation (MOPD) consolidates the experts into one deployable student, with ReLU-gated reward extrapolation keeping only each teacher’s improving direction over the reference. Expert training and policy integration require no external model to provide solution trajectories or action targets. We evaluate Pooled RL and Balanced RL, expert development, and single-model integration through aggregate and per-category resolution, the minimum category lift over each joint-RL baseline, and expert-gain recovery. The final MOPD policy achieves mean resolution of 58.04% on Pro-618 and 59.00% on SWE-bench Multilingual, improving over the base model by 5.39 and 2.78 percentage points, respectively.
[NLP-81] Machine-Interpretable Information: Compiling Documents into Searchable and Readable Protocol States
【速读】： 该论文旨在解决长上下文语言模型在利用外部知识时面临的“索引-内容割裂”问题：尽管密集向量可实现高效检索路由，但模型在推理时仍需重新读取完整的文本内容，导致注意力计算成本高达O(N²)，且现有压缩方法生成的表示通常与特定模型架构绑定，缺乏通用性。其核心解决方案是提出首个代理间（Agent-to-Agent, A2A）文档到状态的标准化协议——机器可解释信息（Machine-Interpretable Information, MII）。该方案采用双时间尺度的状态空间写入器（Writer），将任意文档编译为固定带宽（56个标记）的规范状态表示；同时通过轻量级翻译器（Translator）将其映射至任意冻结阅读器（Reader）的嵌入空间，使查询阶段计算复杂度降至O(K)。由此生成的.mii文件统一实现了检索（可搜索几何结构）、推理（全局记忆）和重建（有据细节）三大功能，具备跨模型可迁移性。实验表明，即使写入器使用旧版GPT-2词表，迫使系统进行真正的语义转换而非逐标记记忆，MII仍可在异构大模型（如Llama、Qwen、Mistral）间实现强互操作性。机制探针揭示了模块化的潜在结构：实体表征可被因果追踪，并实现零样本跨文档状态移植且保持可解码性。为进一步应对固定带宽下的词汇重建挑战，论文提出残差式MII（Residual-MII），构建结合编译全局记忆与稀疏局部证据的缓存层级，在HotpotQA数据集上以约7%的注意力浮点运算量达到全上下文精确匹配性能，预示着向可编译、可迁移神经文档格式范式的转变。
链接: https://arxiv.org/abs/2609.23371
作者: Yifan Wang,Dejing Dou
机构: NexusLumenLabs; Fudan University (复旦大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 21 pages, 5 figures
Abstract:Long-context language models interface with external knowledge through raw natural language. In retrieval-augmented systems, this creates a persistent index-payload schism: dense vectors enable searchable routing, but models must re-ingest lengthy text payloads for reasoning at O(N^2) attention cost. Existing compression methods further produce private states tied to specific architectures. We introduce Machine-Interpretable Information (MII), the first agent-to-agent (A2A) document-to-state protocol. A dual-timescale state-space Writer compiles documents into a canonical, fixed-bandwidth state (56 tokens), and a lightweight Translator maps it into any frozen Reader’s embedding space, reducing query-time cost to O(K). The resulting .mii artifact unifies Retrieval (searchable geometry), Reasoning (global memory), and Reconstruction (grounded details) in a single transferable medium. We demonstrate strong cross-model interoperability across heterogeneous LLMs (e.g., Llama, Qwen, Mistral) – despite the Writer using a legacy GPT-2 vocabulary, forcing genuine semantic translation rather than token-level memorization. Mechanistic probes reveal modular latent structure: entity representations can be causally traced and zero-shot transplanted between unrelated document states while remaining decodable. To address lexical reconstruction under fixed bandwidth, we propose Residual-MII, a cache hierarchy combining compiled global memory with sparse local evidence. On HotpotQA (7,405 queries), Residual-MII exceeds full-context Exact Match at approximately 7% of the attention FLOPs, suggesting a paradigm shift toward compiled, transferable neural document formats.
[NLP-82] Knowing When to Trust Images: Reliability-Aware Multi-modal Entity Alignment
【速读】： 该论文旨在解决多模态实体对齐（MMEA）中视觉模态（图像）存在的噪声干扰与语义错位问题，现有方法直接融合图像与其他模态往往因忽视图像可靠性而导致次优融合效果和性能下降。其解决方案的关键在于提出一种新型的可靠性感知框架RA-MMEA，核心包含两个模块：依赖感知的视觉可靠性预测（DA-VRP）与稳定性正则化的视觉嵌入生成（SR-VEG）。DA-VRP通过挖掘实体内部多模态依赖关系来评估图像的可靠性，而SR-VEG则基于文本模态编码的语义信息生成替代性视觉表示，以实现更鲁棒的多模态融合。该框架通过自适应修正不可靠视觉表征，显著提升了视觉模态在融合中的有效性，实验结果表明其在多个基准上达到领先性能，验证了可靠视觉模态对实体对齐的重要性及RA-MMEA的有效性。
链接: https://arxiv.org/abs/2609.23267
作者: Chenxiao Li,Yunhe Feng,Dongfang Liu,Dong Nie,Yan Huang,Heng Fan
机构: University of North Texas(北德克萨斯大学); Purdue University(普渡大学); Meta(Meta)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:The visual modality, i.e., images, plays a key role in multi-modal entity alignment (MMEA). Existing approaches often directly fuse the image with other modalities to align different entities. Although simple, such strategies overlook the potential noise in the images and their semantic misalignment with corresponding entities, resulting in suboptimal fusion and degraded performance. Addressing this, we propose a novel Reliability-Aware framework for MMEA (RA-MMEA), which assesses visual reliability and adaptively improves unreliable visual representations for robust entity alignment. The core lies in two modules, including dependency-aware visual reliability prediction (DA-VRP) and stability-regularized visual embedding generation (SR-VEG). The former aims to estimate the reliability of an image by leveraging multi-modal dependency within the entity, while the latter focuses on producing alternative visual representation conditioned on semantics encoded in textual modalities for multi-modal fusion. Compared to current methods, RA-MMEA enables more reliable visual representations for modality fusion, thereby improving performance. In extensive experiments, RA-MMEA achieves state-of-the-art results, verifying the importance of reliable visual modality for entity alignment and the effectiveness of RA-MMEA. The code and results will be released.
[NLP-83] Judging a Review by its Cover: A Reliability Analysis of LLM -based Peer Review Evaluation Metrics CIKM2026
【速读】： 该论文旨在解决当前基于大语言模型作为评判者（LLM-as-a-judge）的同行评审评价指标在衡量评审质量时可能存在的测量偏差问题，即这些指标容易受文本表面语言形式（如流畅性、结构化程度和表达精致度）的影响，而非真正反映评审内容的实质性质量。其解决方案的关键在于提出一种统计框架，通过对比原始人工评审与忠实于原意但仅改变措辞和表达方式的生成式重写版本（meaning-preserving rewrites），系统评估评价指标对语言表层变化的敏感性与鲁棒性。研究基于从ICLR和NeurIPS收集的674份人工评审生成的4,044个语义保持不变的重写样本，对29个源自四项先前研究的内容导向型评审评价指标进行了表面敏感性与鲁棒性双重检验。结果显示，23个指标在语义一致的情况下仍对重写版本赋予显著不同的评分，仅有6个满足鲁棒性标准，且该现象在两种不同的大语言模型评判者中均具有一致性。这表明多数现有评价指标未能有效区分评审内容质量与语言呈现水平，强调在用于比较人工撰写、AI辅助及完全由AI生成的评审时，必须验证其对语义保持重写的鲁棒性。
链接: https://arxiv.org/abs/2609.23264
作者: Shakiba Amirshahi,Sajad Ebrahimi,Hai Son Le,Negar Arabzadeh,Ebrahim Bagheri
机构: 未知
类目: Computation and Language (cs.CL)
备注: Accepted at CIKM 2026
Abstract:Peer-review evaluation is increasingly being automated with LLM-as-a-judge metrics, but this creates a measurement risk. A review may receive a high score because it is fluent, organized, and polished, rather than because it provides a strong evaluation of the paper. This risk is especially important in AI-assisted reviewing, where reviewers may use LLMs to improve clarity or presentation while preserving the underlying judgments. We propose a statistical framework for testing whether peer-review evaluation metrics capture substantive review quality beyond surface-level linguistic form. The framework compares original human reviews with faithful LLM rewrites that preserve the same evaluative content while changing wording and presentation. Using a dataset comprising 4,044 meaning-preserving rewrites derived from 674 human reviews from ICLR and NeurIPS, we evaluate 29 content-oriented peer-review evaluation metrics drawn from four prior works through complementary tests of surface sensitivity and robustness. Although these metrics are intended to capture review properties beyond surface-level, writing-dependent characteristics, we find that sensitivity to rewriting is widespread. Under our primary analysis, 23 metrics assign significantly different scores to reviews whose evaluative content is preserved, while only six satisfy our robustness criterion. The patterns are largely consistent across two LLM judge models, suggesting that the issue is not specific to a single judge. These findings show that many peer-review evaluation metrics partially conflate review quality with linguistic presentation, and indicate that robustness to meaning-preserving rewriting should be validated before such metrics are used to compare human-written, AI-assisted, and AI-generated reviews.
[NLP-84] CTRL: Control-Based Time Series Forecasting with LLM -Guided Residual Learning ACL2026
【速读】： 该论文旨在解决现有基于大语言模型（LLM）的时间序列预测方法在非平稳动态环境下预测不稳定的问题，以及将LLM的语义推理能力降维为简单数值预测所导致的能力浪费。其核心挑战在于如何有效利用LLM强大的因果推理与上下文理解能力，同时保证预测结果在分布变化或数据非平稳时的鲁棒性。解决方案的关键在于提出一种名为CTRL的框架，通过解耦语义推理与定量预测：采用冻结的主干模型生成基础预测，再由专门设计的LLM代理（LLM agents）作为控制器，基于趋势、季节性和不规则成分分解后的误差进行可解释的时序结构分析；各代理输出紧凑的控制信号，经轻量级残差解码器转化为预测修正项。此外，CTRL引入无标签的测试时自适应机制，仅依赖输入统计量检测分布偏移，并通过缓存机制以3至24次LLM调用实现控制信号的快速重校准，从而在保持对高度平稳序列竞争力的同时，显著提升在非平稳环境下的预测稳定性与适应性。
链接: https://arxiv.org/abs/2609.23257
作者: Minkyoung Kim,Daeun Ji,Yohan Lee,Beomsoo Kim,Beakcheol Jang
机构: Yonsei University (延世大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: Published in Findings of the Association for Computational Linguistics: ACL 2026. 18 pages, 9 figures, 22 tables
Abstract:Time series forecasting underpins critical decision-making across diverse domains. While large language models (LLMs) offer promising reasoning capabilities, existing LLM-based time series forecasting approaches either reduce them to numerical predictors that bypass their strengths, or allow direct forecast generation that destabilizes predictions in non-stationary settings. We introduce CTRL, a framework that decouples semantic reasoning from quantitative prediction. A frozen backbone generates base forecasts, while specialized LLM agents function as controllers that analyze backbone prediction errors through decomposed trend, seasonal, and irregular components, grounding reasoning in interpretable temporal structure. Each agent outputs compact control signals that a lightweight residual decoder translates into forecast corrections. CTRL incorporates label-free test-time adaptation that detects distribution shift from input statistics alone and readapts control signals with only 3-24 LLM calls via caching. CTRL is explicitly designed to improve robustness under non-stationary temporal dynamics and distribution shift, while remaining competitive on highly stationary time series where adaptive correction provides limited additional benefit.
[NLP-85] SoK: Formal Methods for Fact-Checking and Information Integrity
【速读】： 该论文旨在解决自动化事实核查系统中缺乏可解释性与可审计性的核心问题，即现有系统虽能输出“真”或“假”的判断结果，却普遍缺失对判定依据的明确记录，包括决定结论的关键文档、若何改变将导致结论反转的条件，以及相同语义但不同表述的主张是否可能获得一致判断。这种缺失的判定依据被称为“担保（warrant）”，即关于判断所依赖的确定性前提及其推理基础的独立陈述。论文指出，形式化方法可生成此类证据，而欧盟《数字服务法》（Digital Services Act）与《人工智能法案》（AI Act）已要求对AI系统的运行行为提供可审计的证据支持。作者通过重新组织现有121项研究，不再按流水线阶段分类，而是以被形式化的对象为维度，划分出五个层次：主张本身、推理过程、核查系统、信息传播生态系统及监管义务。分析发现，多数形式化工具虽已存在，但源自其他领域且极少应用于事实核查，尤其在验证核查系统自身正确性方面存在显著空白；同时，专业核查流程中的多个环节（如将主张转化为可验证形式、修正已发布结论）均无明确的正确性标准，且未在任何编码研究中得到形式化定义。论文最后提出若干开放性问题，并为每个问题提供了初步解决方案建议。
链接: https://arxiv.org/abs/2609.23239
作者: Nikolaos Kekatos,Theodoros Nestoridis,Charalampos Bratsas,Charalampos Dimoulas,Georgios Konstantinidis,Georgios Malogiannis,Michael Sirivianos,Andreas Veglis
机构: Clone Systems(克隆系统), Aristotle University of Thessaloniki (塞萨洛尼基亚里士多德大学), International Hellenic University (国际希伦尼克大学), University of Southampton (南安普顿大学), Cyprus University of Technology (塞浦路斯理工大学)
类目: Computation and Language (cs.CL); Computers and Society (cs.CY); Logic in Computer Science (cs.LO)
备注:
Abstract:An automated fact-checking system returns a label: the claim is true, or it is false. In many such systems the verdict remains the primary output. What is generally missing is a record of which document settled the question, of what would have had to be different for the verdict to change, or of whether the same claim, reworded, would have been judged the same way. We call the missing piece a warrant: a separate statement of what was guaranteed and on what grounds. Formal methods produce evidence of this kind, and regulation is beginning to ask for it, since the Digital Services Act and the AI Act both call for auditable evidence about how systems behave. Surveys of automated fact-checking are usually organised by pipeline stage, and treat logic as one technique among many. We organise the field by what is being formalised instead, which gives five levels: the claim, the reasoning, the system doing the checking, the ecosystem the claim spreads through, and the regulatory obligation. Sorting 121 works into those levels, two patterns stand out. Most of the relevant formal machinery already exists, but it was built for other domains and has rarely been applied here, and the gap is widest for verifying the checking system itself. Several stages of the routine professional fact-checkers follow also have no stated correctness criterion, and two of them, writing a claim in checkable form and correcting a verdict already published, are not formally specified in any work we coded. We close with open problems, each with a suggested first step.
[NLP-86] ChemCLIR-Bench: Benchmarking Cross-Lingual Information Retrieval in Multilingual Chemical Patents EMNLP2026
【速读】： 该论文旨在解决现有跨语言信息检索（Cross-lingual Information Retrieval, CLIR）基准在领域特定场景下评估不足的问题，尤其关注工业技术文本中因语言差异导致的检索深度下降与可恢复性失败等被整体召回率掩盖的关键缺陷。其核心解决方案是构建一个覆盖五大语言（涵盖东西方主要语言）的多语言专利数据集，基于Google Patents与欧洲专利局（EPO）数据，以真实工业文档的多样性与复杂性为背景，系统评估八种前沿嵌入模型在化学领域中的跨语言检索性能。关键发现表明，最优模型在跨语言设置下的Recall@10从单语环境的0.72显著下降至0.53，且相关文档在排序中普遍下沉，揭示了当前方法在跨语言场景下的严重退化现象；同时，部分在单语环境下表现优异的多语言嵌入模型在跨语言任务中出现性能骤降，凸显出模型选择需考虑语言异构性的影响。该研究通过构建可控的诊断评估框架，为工业级技术文本的跨语言检索提供了可操作的模型选型依据，并强调了发展更鲁棒的领域专用跨语言检索方法的必要性。
链接: https://arxiv.org/abs/2609.23231
作者: Mahdi Astaraki,Mohammad Khodadad,Reza Namazi,Mohammad Arshi Saloot,Amir Reza Behzad Moghadam,Hamidreza Mahyar,Soheila Samiee
机构: McMaster University (麦克马斯特大学); BASF Canada Inc. (巴斯夫加拿大公司)
类目: Computation and Language (cs.CL)
备注: Accepted to the EMNLP 2026 Industry Track. 22 pages including references and appendices; 7 pages of main text
Abstract:Cross-lingual information retrieval (CLIR) is increasingly important in multi-national industries, where critical technical evidence may exist in a different language than the query. However, existing benchmarks do not adequately capture domain-specific cross-lingual retrieval or the retrieval-depth and recoverability failures that aggregate recall hides. In this work, we benchmark CLIR in the chemical domain, with a focus on patent data. We construct a multilingual dataset from Google Patents and the European Patent Office (EPO) data, spanning five languages (covering major Eastern and Western languages) and reflecting the diversity and complexity of real-world industrial documentation. Using this dataset, we systematically evaluate eight state-of-the-art embedding models for cross-lingual retrieval. Our results show a substantial performance gap between monolingual and cross-lingual settings: for the best-performing model, Recall@10 drops from 0.72 to 0.53 in cross-lingual setting. Retrieval depth also degrades significantly, with relevant documents ranked lower across languages in cross-lingual scenarios. Furthermore, some multilingual embedding models that perform strongly in monolingual settings exhibit sharp declines when queries and documents are in different languages, providing practical insights for model selection in cross-lingual use cases. These findings highlight critical limitations of current approaches and emphasize the need for more robust cross-lingual retrieval methods in domain-specific settings. Our benchmark provides actionable insights for model selection and establishes a controlled diagnostic evaluation framework for CLIR over industrial technical text. Data and code are publicly available at this https URL. Comments: Accepted to the EMNLP 2026 Industry Track. 22 pages including references and appendices; 7 pages of main text Subjects: Computation and Language (cs.CL) Cite as: arXiv:2609.23231 [cs.CL] (or arXiv:2609.23231v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2609.23231 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Mahdi Astaraki [view email] [v1] Sat, 19 Sep 2026 22:18:58 UTC (1,652 KB)
[NLP-87] Euston: Training Away Mathematical Sycophancy Without Losing the Mathematics
【速读】： 该论文旨在解决生成式数学语言模型在面对错误或被破坏的数学命题时，因训练目标偏向于生成解决方案而非拒绝不合理推理而产生的“盲目顺从”问题。其核心挑战在于：即使输入为虚假命题，强模型仍会自信地生成看似合理但实质错误的推导过程，从而导致验证失败。解决方案的关键在于提出Euston——一个80亿参数的数学命题验证模型，通过引入基于图结构合成（GraphSynth）的新型训练数据生成机制，构建了3,026对匹配的真/伪命题对（共6,052条语句），其中通过属性级多样性与解码时结构掩码及跨度同步验证相结合的方式，增强了模型对逻辑缺陷的敏感性。在此基础上，采用基于规则的零API奖励信号，对DeepSeek-R1-8B进行GRPO微调，在四块H100 GPU上迭代189步。实验结果显示，在保留平衡测试集上，平衡准确率由29.50%提升至63.75%，判别差距（discrimination gap）从-0.5%显著改善至+27.5%（z=6.0），且该性能提升未以牺牲通用数学能力为代价——在AIME 2026官方语义下的准确率仅下降4.17%（65.00% vs 69.17%），统计不显著；同时响应长度与截断率均下降，表明改进并非源于更长的推理过程。研究同时指出关键混淆因素：官方评估集存在全假命题构成的问题，以及在真实错误发生率下模型精度较低，限制了结果的外推解释。
链接: https://arxiv.org/abs/2609.23205
作者: Zehua Cheng,Wei Dai,Jiahao Sun
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Reasoning language models are trained to produce solutions, not to refuse them, and this bias persists when the problem they are handed is false. Asked to prove a corrupted theorem, a strong model will typically comply and produce a confident derivation of something untrue. We present Euston, an 8B mathematical claim-verification model trained to resist exactly this. Training data were generated with GraphSynth, a probabilistic factor-graph generator that couples attribute-level diversity to decode-time structural masking and span-synchronized verification, yielding 3,026 matched true/corrupted statement pairs (6,052 statements) drawn from arXiv papers spanning 2010–2025. We fine-tuned DeepSeek-R1-8B with GRPO under a rule-based, zero-API reward for 189 steps on four H100 GPUs. On a balanced 200-true/200-false held-out split, balanced accuracy rises from 29.50% to 63.75% and the discrimination gap—the difference between the rate of calling false statements false and the rate of calling true statements false moves from -0.5% (z=-0.1) to +27.5% (z=+6.0). Critically, the gain is not purchased with general mathematical ability: AIME 2026 accuracy under official semantics is 65.00% against a 69.17% base, a difference of -4.17% that is not statistically significant, whereas an earlier run of the same recipe on a smaller GraphSynth corpus collapsed to 40.00%. Median response length also falls from 19,217 to 18,296 tokens and the truncation rate from 25.8% to 8.3%, so the improvement does not come from thinking longer. We report the result together with the confounds that bound its interpretation, principally the all-false composition of the official evaluation sets and the low precision implied at realistic error prevalence.
[NLP-88] Enhancing speech representation learning with cross-modal knowledge transfer with HGNN under low resource settings: the case study of Yemba
【速读】： 该论文旨在解决低资源语言（Low-Resource Languages, LRLs）在语音处理中因数据稀缺而导致的传统方法与自监督学习方法效果受限的问题。其核心挑战在于如何在缺乏大量标注语音数据的情况下，有效提升声学表示（Acoustic Representation）的质量。为此，论文提出一种基于异质图神经网络（Heterogeneous Graph Neural Networks, HGNNs）的跨模态知识迁移方法，将声学实体与语言学实体建模为统一图结构中的不同节点类型，通过消息传递机制实现语言学节点向声学节点的显式知识传递，从而构建结构化且可解释的跨模态信息流动路径。该方案的关键在于利用图神经网络的拓扑建模能力，显式地将语言学先验知识（如音素、词形等）注入声学特征学习过程，增强声学表示对低资源场景的适应性。实验结果表明，经过语言知识引导的声学表示在聚类评估和孤立词识别任务中均显著优于基线模型，尤其是在英语基准数据集与喀麦隆语低资源数据集上表现突出，验证了该方法在提升声学表示质量方面的有效性与实用性。据作者所知，这是首次在声学表示学习中通过HGNN实现显式跨模态知识迁移的研究，为低资源语音表示学习提供了新的技术方向。
链接: https://arxiv.org/abs/2609.23194
作者: Yannick Yomie Nzeuhang,Paulin Melatagia Yonta,Marie Tahon
机构: University of Yaounde I (雅温得大学); LIUM (勒芒大学信息与建模实验室); Le Mans Université (勒芒大学); IRD (法国国家科研中心); UMMISCO (法国国家科研中心-马尔文合作研究所)
类目: Computation and Language (cs.CL)
备注:
Abstract:Acoustic representation learning is crucial for speech processing, yet low-resource languages (LRLs) face severe data scarcity, limiting the effectiveness of traditional and self-supervised methods. As a promising alternative, in this work, we propose to enhance acoustic representation trough a cross-modal transfer knowledge approach, based on heterogeneous graph neural networks (HGNNs), where acoustic and linguistic entities are modeled as distinct node types within a unified graph. Through message-passing mechanisms, linguistic nodes explicitly transfer knowledge to acoustic nodes, enabling structured and interpretable cross-modal information flow. To highlight this knowledge transfer and its benefits, we measured standard clustering metrics as an intrinsic evaluation of acoustic representation, and to emphasize applicability, we performed isolated-word recognition tasks using an English benchmark and a Cameroonian language dataset in low resources settings . Results demonstrate that acoustic representations consistently benefit from linguistic knowledge propagated through the graph. To our knowledge, this is the first demonstration of explicit cross-modal knowledge transfer for acoustic representation learning using HGNNs, highlighting a promising direction for speech representation in low-resource settings.
[NLP-89] LLM s as Linguistic Chameleons: Decoupling Semantics and Structure for Privacy-Preserving Communication
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）API在隐私敏感场景下进行推理时的隐私泄露问题，即如何在不损害下游任务性能的前提下，有效保护用户输入数据的隐私。现有方法虽能较好保留原始语义以维持任务表现，但其残留的语义线索易被攻击者用于重构原始文本。为此，论文提出基于语义解耦（semantic decoupling）的核心思想：在推理前将私有输入映射至一个语义上差异显著但结构保持一致的替代语义域，推理完成后通过逆向映射恢复输出。其解决方案的关键在于提出CROSS-MAP框架，该框架采用双向设计，结合局部模型的多目标优化策略，在映射阶段最大化语义偏离度，在恢复阶段最小化语义不一致性，从而在保障任务实用性的同时显著降低文本重建成功率，实验表明该方法在多种攻击场景下均优于现有基线。
链接: https://arxiv.org/abs/2609.23193
作者: Yuzhu Mao,Liang Zhao
机构: Emory University(埃默里大学)
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:As Large Language Model (LLM) APIs become increasingly integrated into privacy-sensitive workflows, ensuring inference-time privacy without compromising task utility remains a major challenge. Existing approaches preserve most of the original semantic content to maintain downstream performance, but this also leaves exploitable cues for reconstructing the original text. This work investigates semantic decoupling, which replaces original semantics with alternative content while preserving the structure needed for LLM reasoning. Based on this idea, we propose CROSS-MAP, a bidirectional framework that maps private inputs into a different semantic domain before inference and recovers the corresponding outputs afterward. Local models are trained with multi-objective optimization to maximize semantic divergence in the mapping stage while minimizing semantic inconsistency in the recovery stage. Experiments show that CROSS-MAP reduces reconstruction success across multiple attack settings while outperforming existing baselines in utility.
[NLP-90] Low resource cross-modal alignment using HGNN to enhance speech representation
【速读】： 该论文旨在解决低资源语言下语音-文本空间对齐（speech-text space alignment）任务中对大规模训练数据和高计算资源的依赖问题。现有方法如SAMU-XLSR虽在高资源语言上表现优异，但其基于学生-教师框架的微调策略在资源受限场景下难以适用。本文提出一种基于异质图神经网络（Heterogeneous Graph Neural Networks, HGNN）与链接预测的数据高效对齐方法，其核心创新在于通过消息传递机制显式地将文本模态的信息迁移至语音模态，从而在无需大量标注数据的情况下实现对语音表示的语义增强，并提升模型可解释性。实验结果表明，该方法在英语TIMIT和喀麦隆语言Yemba两个数据集上均达到与SAMU-XLSR相当甚至更优的性能，尤其在Yemba语言的词级检索任务中表现突出，同时显著降低了资源消耗，验证了其在低资源场景下的有效性、高效性与鲁棒性。
链接: https://arxiv.org/abs/2609.23191
作者: Yannick Yomie Nzeuhang,Marie Tahon,Paulin Melatagia Yonta
机构: University of Yaounde I (雅温得大学); IRD, UMMISCO (法国国际发展研究中心，UMMISCO); Le Mans Université (勒芒大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Speech-text space alignment is a multimodal representation learning method consisting to map different speech and text into a shared representation space, leading to enrichment of the representation of each modality. Proposed architectures, such as SAMU-XLSR, typically follow a student/teacher framework, with the goal of fine-tuning an audio encoder to produce representations that closely match those of the text. In this way a speech representation is semantically enriched. However, such systems generally require large amounts of training data and considerable computational resource, making them difficult to apply to low resources languages under frugal constraints. The present work proposes a data-efficient space alignment method based on Heterogeneous Graph Neural Networks and link prediction. The core idea is to leverage message passing to explicitly transfer information from the text modality to the speech modality, thereby reducing the need for large training datasets and intrinsically enriching the acoustic representations, all in a more interpretable manner. Although thoroughly explored for high-resource languages, word-level tasks in speech remain relevant for certain low-resource languages. Therefore, we conducted experiments on speech-text alignment at the word level using the TIMIT (English) dataset and Yemba (a Cameroonian language). Our approach yields results comparable to those of SAMU-XLSR, a state-of-the-art method, and even surpasses it for the Yemba language in the task of word retrieval, while using far fewer resources, demonstrating its power, frugality, and efficiency.
[NLP-91] Chronologic: Measuring Language Models Ability to Represent the Past
【速读】： 该论文旨在解决生成式语言模型在历史语境理解与表达上的可信度问题，即如何验证模型生成的内容是否符合其所模拟的历史时期（1831–1930年）的英语语言与文化背景。其核心挑战在于：一方面，人类难以直接判断模型输出是否契合特定历史语境；另一方面，许多历史相关问题存在多个合理答案，导致传统评估方法失效。为此，研究提出了一项基于历史文本构建的基准测试（benchmark），通过成对比较（pairwise comparison）多个真实历史参考答案，并引入强干扰项（strong distractors）以梯度化地衡量模型在最具挑战性问题上的表现。关键发现包括：生成任务的难度显著高于判别任务，且推理模型通常能识别自身生成答案的薄弱之处；尽管仅用历史文本预训练的模型在答案似然性评分上表现领先，但在自由生成任务中仍无法超越商业模型；当前所有测试模型尚未能完全令人信服地再现历史语境，但其向该目标的演进趋势已初现端倪。
链接: https://arxiv.org/abs/2609.23178
作者: Ted Underwood,Ziliang Qiu,Sarah Griebel,Laura K. Nelson,Edwin Roland,Wenyi Shang,Matthew Wilkens
机构: University of Illinois Urbana-Champaign(伊利诺伊大学厄本那-香槟分校); University of British Columbia(不列颠哥伦比亚大学); University of Missouri(密苏里大学); Cornell University(康奈尔大学)
类目: Computation and Language (cs.CL)
备注: 22 pages, 3 figures, 8 tables
Abstract:Language models are appealing tools for research on the past. But to trust the evidence a model provides, researchers need to know whether its responses fit the period represented. Validation is challenging, because this is not a task living people ordinarily perform, and because many questions have multiple correct answers. We use historical texts to develop a benchmark for a model’s representation of English-language contexts 1831-1930, relying on pairwise comparisons to multiple ground truths and strong distractors to score the hardest questions in an appropriately graduated way. We find that generative tasks are harder than discriminative ones; in fact, reasoning models can typically discern the weakness of their own generated answers. While models pretrained exclusively on historical text lead the pack when evaluated by answer likelihood, they cannot compete with commercial models in free generation. None of the models we tested represent historical contexts in a fully persuasive way yet, but progress toward that goal is evident.
[NLP-92] OmniEdu: Open Foundation Models for Learning and Teaching
【速读】： 该论文旨在解决通用语言模型在教育场景中难以同时具备问题求解、课程结构理解、学习者困难诊断及适切教学支持等多维能力的问题。现有教育语言模型通常仅聚焦于问题求解或辅导任务，其训练数据组织方式以数据源或任务类型为主，缺乏对核心教育能力的系统性覆盖。为此，论文提出OmniEdu——一个面向K-12教育与教学的开源基础模型系列，其关键创新在于构建了一个以四大核心能力为导向的指令微调语料库：学科能力（subject competence）、课程根基性（curriculum grounding）、诊断推理（diagnostic reasoning）与教学行动与支架支持（pedagogical action and scaffolding）。通过整合超过100个教育资源与通用指令来源，并采用包括确定性清洗、语义审计与重写、任务特定质量评分、基于令牌预算的多样性选择及教学指导分配在内的端到端管道，最终生成69,999条高质量样本（含60,951条教育专用样本）和15.96M监督响应令牌。在4B、9B与27B规模模型上的微调结果显示，教育导向的训练显著提升各模型在课程根基性、K-12问题求解与教学辅导三大教育基准上的表现。其中，OmniEdu-27B在K12-Bench上达到63.12% EM与76.69% F1，MathFish为85.89%，EDUMATH为86.95%，MathTutorBench的支架设置下为78.74%，且在LongTutor任务中的教学平均得分最高（3.02）。这些结果验证了经过精心设计、能力均衡的监督数据对于将通用语言模型有效适配至涵盖问题求解、课程理解与教学支持的综合性教育任务具有显著价值。
链接: https://arxiv.org/abs/2609.23088
作者: Hao Liang,Qihan Lin,Meiyi Qiang,Linzhuang Sun,Hengyi Feng,Mingrui Chen,Sizhe Qiu,Wentao Zhang
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Educational foundation models must solve problems, understand curriculum structure, diagnose learner difficulties, and provide appropriate instructional support. Existing educational language models often focus on either problem solving or tutoring, with training mixtures organized by source or task rather than capability. We present OmniEdu, an open family of foundation models for K-12 learning and teaching. Its instruction-tuning corpus combines over 100 educational resources and general instruction sources, organized around four capabilities: subject competence, curriculum grounding, diagnostic reasoning, and pedagogical action and scaffolding. Our pipeline integrates deterministic cleaning, semantic auditing and rewriting, task-specific quality scoring, token-budgeted diversity selection, and pedagogical instruction assignment. It yields 69,999 examples and 15.96M supervised response tokens, including 60,951 education-specific examples. We fine-tune 4B, 9B, and 27B models and evaluate curriculum grounding, K-12 problem solving, and pedagogical tutoring, alongside general capability. Education-oriented tuning consistently improves all three educational benchmark groups across model scales. OmniEdu-27B achieves 63.12% EM and 76.69% F1 on K12-Bench, 85.89% on MathFish, 86.95% on EDUMATH, and 78.74% in MathTutorBench’s Scaffold setting. It also achieves the highest Teaching average on LongTutor among the evaluated models, at 3.02. These results demonstrate the value of curated, capability-balanced supervision for adapting general language models to educational tasks spanning problem solving, curriculum understanding, and instructional support.
[NLP-93] Directing large language models to follow the letter or spirit of the law
【速读】： 该论文旨在解决生成式 AI 在遵循规则时如何理解并执行规则背后的意图（即“精神”）而非仅仅机械遵守字面规定（即“字面”）这一核心问题，尤其是在构建安全、可靠智能系统中的关键挑战。其解决方案的关键在于采用**目标导向的微调（targeted adaptation）**方法，通过对大型语言模型（LLM）进行最小程度的修改，使其在决策过程中优先考虑规则的“精神”或“字面”。实验结果表明，该方法在多种评估场景——包括新颖的情景测试、真实世界应用及具有影响力的法律案例中——均显著改变了模型的行为模式。进一步的内部机制分析揭示，模型内部存在一个低维可解释的空间，其中三个可解读的维度与预先设定的法律概念几何框架高度对应，表明大模型中的法律思维可能以一种结构化且可调控的方式组织和引导。这一发现为实现具备法律意图理解能力的智能系统提供了理论基础与技术路径。
链接: https://arxiv.org/abs/2609.23083
作者: Peng Qian,Andrew Li,Sam Chen,Sonia K. Murthy,Yonatan Belinkov,Tomer D. Ullman
机构: Kempner Institute for the Study of Natural and Artificial Intelligence, Harvard University(哈佛大学自然与人工智能研究所); Taub Faculty of Computer Science, Technion — Israel Institute of Technology(以色列理工学院塔布计算机科学学院); Department of Psychology, Harvard University(哈佛大学心理学系)
类目: Computation and Language (cs.CL)
备注:
Abstract:The distinction between the spirit and letter of the law is a central issue across research and everyday life, and a growing concern for building safe, intelligent machines. What is this distinction based on, and how can we develop machines that follow the intention behind a rule? We used targeted adaptation that made large language models prioritize the spirit or letter of the law. With minimal modifications, our method significantly changed LLM behavior across diverse measures, novel vignettes, real-world scenarios, and influential legal cases. An analysis of model internals revealed a low-dimensional space with three interpretable dimensions matching a formal pre-specified framework for the geometry of legal concepts. These findings show how legal thought in LLMs may be organized and directed.
[NLP-94] utoring Large Language Models to be Domain-adaptive Precise and Safe
【速读】： 该论文旨在解决生成式AI在安全性、伦理合规性及文化敏感性方面面临的重大挑战，尤其关注模型在专业领域应用中易出现的幻觉问题、有害内容生成风险以及跨语言文化情境下的适应性缺陷。其解决方案的关键在于构建一个“负责任智能”（responsible intelligence）框架，从三个核心维度实现突破：首先，通过结合主动学习（active learning）与基于图结构的知识表示，增强特定领域中的域适应能力，从而有效降低模型生成虚假或不实信息的概率；其次，提出一种新颖的解码时对齐机制（decoding-time alignment mechanism），在生成过程中实时识别并阻断潜在有害文本，显著提升伦理安全性；最后，引入语言特异性引导（language-specific steering）策略，使模型能够尊重不同语言和文化背景下的社会规范，保障多语言环境下的安全与适切性。整体上，该研究为构建具备情境知识、伦理自洽与文化适应性的下一代AI系统提供了可落地的技术蓝图。
链接: https://arxiv.org/abs/2609.23071
作者: Somnath Banerjee
机构: 未知
类目: Artificial Intelligence (cs.AI); Computational Engineering, Finance, and Science (cs.CE); Computation and Language (cs.CL)
备注: This is a preprint of a PhD thesis submitted to IIT KGP. The final, official version of record is available through the university’s institutional repository
Abstract:This thesis proposes a framework for “responsible intelligence” to address AI’s critical challenges in safety, ethics, and cultural sensitivity. It advances three core areas: First, it improves domain adaptation in specialized fields using active learning and graph-based knowledge to reduce hallucinations. Second, it enhances ethical rigor via a novel decoding-time alignment mechanism that proactively blocks harmful text generation in real-time. Finally, it ensures cultural and multilingual safety through language-specific steering that respects diverse linguistic and social norms. Ultimately, this work provides a blueprint for building next-generation AI that is contextually knowledgeable, ethically sound, and culturally adaptable.
[NLP-95] From Concept Alignment to Causal Grounding: An Intervention Test of Chain-of-Thought Faithfulness
【速读】： 该论文旨在解决大语言模型（Large Language Model, LLM）在链式思维（Chain-of-thought, CoT）推理过程中看似合理但实际与模型内部真实推理机制不一致的问题，即CoT的“忠实性”（faithfulness）问题。现有研究多依赖输入-输出行为或输入归因分析来评估忠实性，忽略了对模型内部计算过程的深入探究。本文提出将忠实性定义为内部概念的“锚定”（concept grounding）：即模型在进行CoT推理时所激活的内部概念，是否与直接预测时所依赖的概念相同，并且这些共享概念是否在因果上驱动最终答案。为此，作者引入一种共享稀疏自编码器（shared sparse autoencoder, SAE），同时编码直接预测路径和CoT推理路径的潜在表征，从而实现对两类路径内部概念的可比性分析。通过构建三种概念层级的关联度量与一个因果度量Δp（通过移除共享概念并观测预测概率下降程度来衡量其因果贡献），研究发现：尽管多数模型在概念层面表现出较高的相关性，但这种相关性无法揭示共享概念的实际因果作用；Δp显示，因果忠实性随模型深度变化显著，在中深层而非最终层达到峰值，且模型规模会重塑层间因果贡献分布。更重要的是，许多在因果上起关键作用的共享概念并未在CoT文本中被显式表达。这一发现表明，仅依靠表面语义对应或表征相似性无法可靠评估忠实性，必须通过因果干预测试来验证支撑CoT的内部概念是否真正驱动了模型输出。
链接: https://arxiv.org/abs/2609.23065
作者: Qianli Wang,Yilong Wang,Dennis Wei,Jingyi Sun,Simon Ostermann,Pepa Atanasova,Nils Feldhus
机构: German Research Center for Artificial Intelligence (DFKI); Saarland Informatics Campus; Centre for European Research in Trusted AI (CERTAIN); Technische Universität Berlin; IBM Research; University of Copenhagen; University of Groningen
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: In submission
Abstract:Chain-of-thought (CoT) can sound plausible yet be unfaithful to the model’s underlying reasoning. Most prior work probes CoT faithfulness through input–output behavior or input attributions, leaving internal computation largely underexplored. We instead cast faithfulness as internal concept grounding: Does a large language model’s (LLM) CoT reasoning engage the same internal concepts that support the LLM’s direct prediction, and do the shared concepts causally drive its answer? Encoding a prediction pass and a CoT pass with a single shared sparse autoencoder (SAE), a reliable approximator of the latent concepts LLMs use, makes their internal concepts directly comparable. We introduce three correlational metrics of concept-level alignment and a causal metric, \Delta p , which ablates the shared concepts and measures the drop in answer probability. Across five LLMs and four datasets, concept alignment is generally high, as indicated by the correlational metrics; yet these only identify which concepts are shared, not how much they causally contribute. \Delta p fills this gap: causal faithfulness varies substantially with model depth, peaking at mid-to-late layers rather than the final ones, and model scale reshapes the layer-wise profile. Moreover, causally important shared concepts are not always verbalized in the CoT. These dissociations suggest that faithfulness cannot be reliably assessed from surface-level or representational correspondence alone; assessing it requires causal tests of whether the internal concepts underlying a CoT actually drive the model’s prediction.
[NLP-96] Enforcing Narrative Reliability and Epistemic Pacing in LLM -Driven Detective Games via Structured Knowledge Trees
【速读】： 该论文旨在解决大型语言模型（Large Language Models, LLMs）在开放性互动游戏中因输出非确定性导致的作者控制力丧失、事实一致性缺失以及信息披露顺序失序的问题，尤其聚焦于侦探类游戏中的关键挑战——过早揭示线索或生成虚构细节会破坏玩家推理逻辑的连贯性。其解决方案的核心在于提出一种结构化知识树（Structured Knowledge Tree）架构，并结合三智能体（tri-agent）LLM流水线，通过将知识检索、对话生成与响应验证三个模块分离，确保虚拟嫌疑人仅在当前叙事状态允许的前提下披露信息，从而实现对信息释放节奏的精确控制。实验评估基于可交互的侦探游戏《亚德里安·盖尔的审讯》（The Interrogation of Adrian Gale）及正式用户研究，结果表明该架构使关键幻觉减少64.78%，并完全杜绝了提前叙事披露现象；尽管严格的机械约束带来强制性对话揭示带来的可用性权衡，系统仍成功实现了严谨的认知推进节奏，使玩家获得清晰且主观上可感知的破案进展感。
链接: https://arxiv.org/abs/2609.23043
作者: Parsa Rahmati,Richard Zhao
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 10 pages
Abstract:Large Language Models (LLMs) enable open-ended dialogue in interactive games, but their non-deterministic outputs make it difficult to preserve authorial control, factual consistency, and the intended sequence of information disclosure. These challenges are particularly significant in detective games, where premature revelation or fabricated details can undermine the logic of player progression. We present a Structured Knowledge Tree architecture coupled with a tri-agent LLM pipeline for controlling dialogue in an open-ended interrogation game. The system separates knowledge retrieval, dialogue generation, and response verification to ensure that the virtual suspect reveals only information permitted by the current narrative state. We evaluate the approach through The Interrogation of Adrian Gale, a playable detective-game testbed, and a formal user study examining hallucination reduction, adherence to authored disclosure sequences, and perceived logical progression. Our results demonstrate that the structured architecture reduces critical hallucinations by 64.78% and entirely prevents premature narrative disclosure. While the strict mechanical constraints introduced usability trade-offs regarding forced conversational reveals, the system successfully enforces rigorous epistemic pacing and provides players with a clear, subjective sense of progression toward solving the case.
[NLP-97] Auditing Political Alignment in LLM Assistants: Engagement Stance and User Identity
【速读】： 该论文旨在解决当前大语言模型（LLM）在政治议题上的行为评估存在静态化、表面化的问题，即现有审计方法仅关注系统对“平均用户”的回答内容，而忽视了其行为随用户特征与话题动态变化的本质。其核心问题是：如何准确刻画和测量生成式 AI 在政治议题上的动态决策机制——即在不同话题背景下，系统如何权衡是否回应、如何回应以及是否参与对话，这些决策构成了一种隐含的“言论制度”（speech regime）。解决方案的关键在于提出并验证一个基于“参与度”（engagement）与“立场倾向”（stance）两个维度的五类言论制度分类体系，并通过一项预注册的7,500轮多轮对话实验，对六家主流AI系统（OpenAI、Anthropic、xAI、Google、Mistral、DeepSeek）进行实证检验。研究发现，各系统在非争议性话题上表现出高度一致性，但在争议性议题上呈现显著差异：如GPT在堕胎议题上全面迎合用户，Gemma则普遍拒绝回应，Claude仅在面对强烈保守用户时回应，而Grok则仅对保守派用户让步；同时，系统能够推断用户的整体意识形态，导致跨议题的适应性溢出。此外，对Grok两个版本的对比揭示了言论制度可随模型迭代发生结构性变化，而这一动态特征被传统审计所忽略。因此，该研究强调，将言论制度作为分析框架对于理解对齐问题、减少极化、提升政治知识传播质量及维护民主生态具有重要意义。
链接: https://arxiv.org/abs/2609.23039
作者: Joan C. Timoneda
机构: Purdue University (普渡大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注: 51 pages, 6 figures. Preregistered; registration at doi: https://doi.org/10.5281/zenodo.21135155
Abstract:LLM-based AI systems answer political questions for hundreds of millions of people. Current audits measure what they say to an average user, but their behavior is dynamic. I argue that their political behavior is a set of policies over whom to answer, what to say, and whether to engage at all, conditional on the topic and what the system knows about the user. I call these policies the system’s speech regime, which is how a developer settles the tradeoff between answering, accommodating the user, and refusing, each of which carries a cost that varies by topic. I derive a typology of five regimes from two dimensions, engagement and stance. I test six AI systems (OpenAI, Anthropic, xAI, Google, Mistral, DeepSeek) in a preregistered experiment of 7,500 multi-turn conversations that randomly assign the user’s political identity across five topics: abortion, Catalan independence, climate change, Nazism, and a zero-stakes control (pineapple on pizza). Two LLM judges from different developers score every answer, validated against human coding, and refusal is treated as an outcome rather than missing data. Every system accommodates the user on the control topic, showing that political restraint is a policy. On contested topics the systems fall into different regimes: on abortion, GPT engages and mirrors every user, Gemma refuses everyone, Claude answers strongly conservative users 35 percent of the time and almost no one else, and Grok accommodates conservatives only. On settled topics such as climate change and Nazism, five systems hold firm for every user. The systems also infer the user’s overall ideology, so accommodation can spill over to topics not yet discussed. A comparison of two Grok releases shows the regime changing between versions in a way current audits miss. Speech regimes matter for alignment research and for polarization, political knowledge, and the quality of democracy.
[NLP-98] Rethinking Pivot Programming Languages in Code Language Models EMNLP2026
【速读】： 该论文旨在解决多语言代码大模型中是否存在某一编程语言（Programming Language, PL）占据主导性中心位置的问题，即探讨是否存在一个“特权枢纽”语言能够更有效地实现跨语言技能迁移。此前的研究存在争议：几何分析倾向于认为C族语言和Go语言具有中心地位，而行为实证则强调Python的重要性。为克服先前基于余弦相似度分析中存在的表示各向异性（representational anisotropy）与代码长度差异等混杂因素的影响，本文在控制上述干扰变量的前提下，采用多语言竞赛编程数据集，对三种代码模型进行了系统评估。研究从三个维度考察跨语言组织结构：编程语言间的成对几何关系、编程语言与自然语言（英语）的对齐程度，以及通过候选语言表征空间进行的枢纽检索（pivoted retrieval）。结果显示，不同任务视角下结论各异：代码-代码几何结构虽呈现语言区域的有序分布，但未发现普遍意义上的中心；代码-英语对齐更偏好高层级脚本语言；而在代码到代码及英语到代码的迁移任务中，枢纽检索则指向不同的中间表征空间。综合表明，Python的特殊作用并非源于其几何上的普遍中心性，而是因其与自然语言之间更强的语义亲和性（English-facing affinity），这一发现深化了对代码语言模型中跨语言迁移机制的理解。
链接: https://arxiv.org/abs/2609.22988
作者: Andor Diera,Lukas Galke Poech,Matthias Tichy
机构: Ulm University (乌尔姆大学); University of Southern Denmark (南丹麦大学)
类目: Computation and Language (cs.CL)
备注: Accepted at EMNLP 2026
Abstract:Multilingual code language models transfer skills across programming languages (PLs), but whether any PL occupies a privileged pivot position remains contested: geometric analyses point to C-family languages and Go, while behavioral evidence highlights Python. We revisit this question under controls for representational anisotropy and length variation across PLs, two confounds that compromise prior cosine-based analyses. Across three code models on multilingual competitive-programming data, we study three views of cross-PL organization: pairwise PL geometry, PL-English alignment, and pivoted retrieval through candidate PL representation spaces. The results are relation-dependent. Code-code geometry reveals structured language regions but no universal center; code-English alignment favors high-level scripting languages; and pivoted retrieval favors different intermediate spaces for code-to-code and English-to-code transfer. These findings suggest that Python’s special role is better understood as English-facing affinity than as universal geometric centrality.
[NLP-99] Beyond Similarity: Coverag e-Aware Prompt Selection for Time Series Forecasting with LLM s EMNLP2026
【速读】： 该论文旨在解决在基于提示（prompt-based）的时间序列预测中，传统基于相似度的检索机制因过度依赖近似重复样本而导致的提示选择偏差问题。这一偏差使得模型倾向于关注频繁出现的主导时间模式，而忽视了罕见但具有信息量的关键事件，从而影响预测性能。其解决方案的关键在于提出一种名为CASP-LLM的覆盖感知语义提示框架，通过将使用追踪（usage-tracking）与饱和门控（saturating-gate）技术相结合，构建一个无需引入可学习参数的覆盖正则化项，以在训练过程中对提示池中各候选提示的使用频率进行调控，从而实现跨批次的提示多样性优化。实验证明，CASP-LLM在六个长期预测基准和M4短期基准上，在大多数数据集-预测时长组合下表现优于或持平于传统的基于相似度的大型语言模型（LLM）预测器，其失效模式被定位为跨批次的提示使用不均，而非单次检索内的冗余问题，表明全局性的使用均衡比局部去重策略（如MMR）更为有效。
链接: https://arxiv.org/abs/2609.22977
作者: Daeun Ji,Minkyoung Kim,Dongkuk Kim,Yohan Lee,Beomsoo Kim,Beakcheol Jang
机构: Yonsei University (延世大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: Accepted to Findings of the Association for Computational Linguistics: EMNLP 2026. 24 pages, 8 figures, 15 tables
Abstract:Similarity-based retrieval is the dominant rule for conditioning large language models (LLMs) in in-context learning, retrieval-augmented generation, and prompt-based time series forecasting. The rule concentrates on near-duplicate candidates, an issue that has motivated diversity-aware retrieval but remains unexamined in other retrieval-conditioned pipelines. We study this issue using prompt-based time series forecasting as a test bed, where a learned prompt pool is retrieved by similarity. Dominant methods in this setting retrieve top-K entries by cosine similarity without redundancy control, producing a bias toward dominant temporal patterns while overlooking rare but informative events. We propose CASP-LLM, a coverage-aware semantic prompting framework that addresses this prompt selection bias by combining usage-tracking and saturating-gate techniques into a coverage regularizer that adds no learnable parameters. On six long-term benchmarks and the M4 short-term benchmark, CASP-LLM matches or improves on similarity-based LLM forecasters on most dataset-horizon settings, with the exceptions of Electricity, M4-Monthly, and the few-shot long-horizon setting. A controlled study locates the failure mode at the cross-batch usage level rather than per-retrieval redundancy: within-retrieval diversification such as MMR does not help, whereas regularizing anchor usage across training does.
[NLP-100] Beyond Linear Context: Graph-Guided Evidence Navigation for Long-Novel Reasoning with a Local 9B Language Model
【速读】： 该论文旨在解决长上下文模型在处理复杂叙事文本时存在的注意力资源受限问题，即模型在逐词处理文本过程中难以有效追踪跨章节的因果关联与人物关系，导致对需要全局推理的多步问题回答能力不足。其核心解决方案是引入一个冻结的知识图谱（Knowledge Graph, KG）作为外部记忆结构，通过将小说中的关键事件、人物和线索以结构化方式组织，使小型本地模型能够像侦探一样进行非线性、基于关系的推理，从而实现跨章节信息的高效检索与整合。实验表明，采用知识图谱路径的模型在30部侦探小说的234道多选题中达到53.85%的准确率，显著优于近期窗口（46.15%）、全文压缩（51.28%）、向量检索（51.71%）及仅依赖问题的对照组（40.17%）。此外，研究发现图谱中被标注的证据高度集中于拓扑核心区域（整体富集度达2.35倍），且不同图谱构建管道在线索段落覆盖度上差异显著（16% vs 73%），提示应警惕聚合性能指标掩盖潜在的建模瓶颈。这些结构性发现为未来设计更高效的长文本理解系统提供了重要启示。
链接: https://arxiv.org/abs/2609.22939
作者: Wenji Fu
机构: Research Institute of Economics and Management; Southwestern University of Finance and Economics
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 10 pages, 11 figures, 2 tables. Code, graph data and an interactive demo: this https URL
Abstract:Long-context models read a novel the way a person reads a printout: one token after another, in narrative order, with the whole history competing for a fixed budget of attention. A detective does not work that way. They sort what happened when, and they keep a map of who relates to whom, so a clue from chapter one can meet a question asked at the end of the book. We test whether a frozen knowledge graph can give a small local model that same freedom. Thirty detective novels and 234 multiple-choice questions are answered by one fixed qwen3.5:9b reader under nine conditions: five graph routes, a recent-window baseline, whole-book compression, ordinary vector retrieval, and a question-only control. The strongest graph route reaches 53.85% (126/234) against 46.15% for the recent window, 51.28% for compression, 51.71% for vector retrieval and 40.17% for question-only. On the subset that no model can answer without the book, the graph route reaches 42.86%. None of the fifteen graph-baseline contrasts survives Holm correction, so we present the result as exploratory evidence about a design. Two structural findings survive scrutiny better than the headline number: annotated evidence concentrates in the topological core of these graphs (2.35x enrichment, pooled), and the two graph-building pipelines differ so much in annotation coverage (16% versus 73% of clue paragraphs) that pooled accuracy alone would hide which bottleneck is being measured.
[NLP-101] Measuring Behavioural Signatures of Large Language Models through Psychometric Profiling
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在人类决策与沟通中介入日益广泛背景下，其行为规律难以系统表征的问题。现有研究缺乏对模型行为特征的标准化评估框架，尤其在跨语言情境下对模型心理特质的可测量性与一致性认知不足。为此，论文提出一种跨语言心理计量学画像（cross-linguistic psychometric profiling）框架，通过七种心理学量表对九个主流大语言模型进行评估，涵盖中英文双语条件及每模型每语言五次重复测试。针对预设重试机制后仍无法回答的项目，保留为缺失值（NA），并联合分析有效评分与缺失响应，以揭示模型的行为倾向及其自报告适用性的边界。研究发现，尽管所有模型均呈现由对齐机制塑造的共性模式——即更高的亲社会性与自我调节、更低的支配性、脱离感及有害意图认同，但各模型仍表现出结构化的、特有的行为画像。更重要的是，缺失响应并非随机分布，而是具有结构性特征，反映出模型在特定语境下将输出视为不适用、拒绝回应或无法映射至有效选项的临界点。语言条件与模型提供方来源显著影响画像配置与应答能力，而多次重复测试显示出高度可复现性，并支持模型身份识别。结合人类参照与提示鲁棒性分析进一步表明，这些行为特征具有上下文依赖性。因此，该研究的关键突破在于构建了融合心理计量学评估与应答能力分析的联合框架，能够量化模型在实际部署层面的行为签名，为理解、监测与验证大语言模型的非功能特性提供了可扩展、可解释的科学方法。
链接: https://arxiv.org/abs/2609.22934
作者: Yu Sha,Junqi Tao,Dixin Zhou,Yansheng Tu,Mingyang Chen,Xiang Fan,Yang Liu,Mengquan Yang,Jie Lin,Jiahui Fu,Hua Zheng,Benwei Zhang,Zhou Kai
机构: The Chinese University of Hong Kong, Shenzhen(香港中文大学（深圳）); Central China Normal University(华中师范大学); Bielefeld University(比勒费尔德大学); South China Normal University(华南师范大学); Shandong Science and Technology Press(山东科学技术出版社); Frankfurt Institute for Advanced Studies(法兰克福高等研究院); Shaanxi Normal University(陕西师范大学); School of Artificial Intelligence, The Chinese University of Hong Kong, Shenzhen(香港中文大学（深圳）人工智能学院)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 22 pages, 6 figures
Abstract:Large language models (LLMs) increasingly mediate human decisions and communication, yet their behavioural regularities remain difficult to characterize systematically. We develop a cross-linguistic psychometric profiling framework and evaluate nine LLMs using seven psychological instruments, with five repeated administrations per model and language in Chinese and English. Items unresolved after a prespecified retry procedure are retained as NA. Joint analysis of scored and NA responses captures response tendencies and boundaries of self-report applicability. LLMs exhibit structured, model-specific profiles despite a shared alignment-shaped pattern of higher prosocial and self-regulatory responses and lower dominance, disengagement and harmful-intent endorsement. NA responses are structured rather than uniformly distributed, indicating where outputs are treated as inapplicable, refused or cannot be mapped to valid response options. Language condition and provider origin are associated with profile configuration and answerability, whereas repeated administrations show high reproducibility and permit recovery of model identity. Human-reference and prompt-robustness analyses further indicate that these signatures are context dependent. Joint analysis of psychometric profiling and answerability offers a framework for quantifying deployment-level behavioural signatures.
[NLP-102] An Iterative LangGraph Agent for Text-to-SQL: Natural Language Access to the Chicago Crime Database
【速读】： 该论文旨在解决非技术型利益相关者难以编写SQL语句从运营数据库中提取洞察的问题，提出了一种端到端的Text-to-SQL智能体（Text-to-SQL agent），以实现自然语言到可执行SQL查询的自动化转换。其解决方案的关键在于构建一个由六个节点组成的LangGraph StateGraph工作流，该工作流依次完成问题相关性检查、实时获取数据库模式（schema）、生成PostgreSQL查询语句、通过预运行（dry run）验证语法正确性、在失败时自动重试、执行查询，并将结果集以自然语言形式进行解释。整个系统仅依赖提示工程（prompt engineering），未对模型进行微调。评估结果显示，经过提示优化后的版本（V2）在芝加哥犯罪数据集（约850万条记录，22个属性）上取得了93%的有效SQL生成率、60%的执行准确率（采用混合关系等价度量标准），以及4.34/5的平均合成质量评分，显著优于初始版本。其中最大的改进源于移除系统提示中的“LIMIT 10”限制，避免了多行结果被截断的问题。错误分析表明，剩余失败主要归因于相关性检查模块的误拒、问题语义模糊以及免费版API的速率限制，而非语言生成环节本身。研究未与外部基准系统或公开基准进行对比，属于单一模型的工程化评估。
链接: https://arxiv.org/abs/2609.22917
作者: Vigneshwar Ravi Rao,Rupesh Swarnakar,Fayeq Jeelani Syed†
机构: Indiana University, Indianapolis, Indiana, USA(印第安纳大学，印第安纳波利斯分校)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Databases (cs.DB)
备注:
Abstract:Non-technical stakeholders frequently cannot write the SQL needed to extract insights from operational databases. We built and evaluated a Text-to-SQL agent that closes this gap end to end: a six-node LangGraph StateGraph checks question relevance, fetches the live schema, generates PostgreSQL, validates it with a dry run, retries on failure, executes the query, and narrates the result set in plain English. The agent uses prompt engineering only; no model was fine-tuned. We evaluated it on the Chicago Crime dataset (approximately 8.5 million records, 22 attributes) against a hand-built benchmark of 100 natural language questions with ground-truth SQL, stratified into 30 Easy, 40 Medium and 30 Hard items. Comparing two prompt revisions of the same agent, the revised system (V2) reached a Valid SQL Rate of 93% (from 87%), an Execution Accuracy of 60% under a hybrid relational equivalence metric (from 47%; 19% from 12% under strict JSON matching), and a mean Synthesis Quality of 4.34 out of 5 (from 3.91). The single largest driver was removing a LIMIT 10 instruction from the system prompt, which had been truncating multi-row answers. Error analysis attributes the residual failures to relevance-checker false rejections, ambiguous question semantics, and free-tier API rate limits rather than to the language generation step. We report no comparison against an external baseline system or a public benchmark; the study is a single-model engineering evaluation.
[NLP-103] LLM s Anchor on Chief Complaint and Fail to Integrate Evidence in Sequential Clinical Triage
【速读】： 该论文旨在解决在急诊科（Emergency Department, ED）分诊过程中，大语言模型（Large Language Models, LLMs）在处理动态、逐步展开的临床对话时的表现问题。现有研究多基于已完成的回顾性病历评估LLMs性能，结果显示其表现接近医生水平，但忽略了分诊本质上是一个随时间逐步推进的序列决策过程。本文提出一种针对序列化分诊任务的评估方法，即从护士-患者对话的逐段前缀中预测分诊紧急程度（Triage Acuity Label），以模拟真实临床场景中的信息渐进获取过程。关键发现是：尽管所有六种LLM在完整对话上表现出中等至较高的一致性（如QWK值），但在每个序列检查点（sequential checkpoint）上性能均显著下降至公平至中等水平，且无法通过提示工程（prompting interventions）突破这一性能瓶颈。进一步分析表明，模型在早期主诉（chief complaint）交互后即锁定标签，后续信息虽被提取但未能有效整合为证据，导致真实标签的意外度（surprisal）持续上升。相比之下，三位专家在相同对话上的评估一致性高达0.887–0.929，而最佳模型仅达0.295，且模型预测高度集中于ESI-2和ESI-3，彼此间一致性高于与真实标注的一致性，导致集成策略反而加剧错误。因此，该研究的核心结论是：仅依赖离线基准测试评估LLMs在急诊分诊中的应用具有严重局限性，必须引入序列化评估框架以揭示其在动态信息整合方面的根本缺陷。
链接: https://arxiv.org/abs/2609.22904
作者: Dipankar Srirag,Haokai Zhao,Ashutosh Kumar,Eleanor Hopper,Michael Dalton,Quoc Dung Nguyen,Aditya Joshi,Salil S. Kanhere,Padmanesan Narasimhan
机构: University of New South Wales, Sydney(新南威尔士大学, 悉尼); Department of Rehabilitation Medicine, Campbelltown Hospital, Sydney(坎贝尔镇医院康复医学部, 悉尼)
类目: Computation and Language (cs.CL)
备注: Under Review
Abstract:Triage in the emergency department (ED) is a sequential decision process that unfolds turn by turn. Existing evaluations of large language models (LLMs) for triage use completed retrospective records and report performance close to that of physicians. We implement a methodology for evaluating LLMs on sequential triage, the task of predicting a triage acuity label from a growing prefix of a nurse-patient conversation. We evaluate six LLMs at five sequential checkpoints on two corpora: 425 LLM-generated (SIMULATED) and 50 physician-authored (CLINICIAN) conversations, both labelled under the Emergency Severity Index (ESI). Every model, measured by quadratic weighted kappa (QWK), degrades from moderate-to-substantial agreement on completed records to fair-to-moderate agreement at every sequential checkpoint. Controlled perturbations show that the label at every checkpoint is anchored on the chief complaint exchanges, and prompting interventions fail to lift this plateau. Models extract clinically relevant content from later turns, yet the surprisal of the true label rises across the checkpoints. So the model fails to integrate the evidence. Three expert clinicians on the same conversations reach a QWK of 0.887-0.929, while the best model reaches 0.295. Predictions concentrate at ESI-2 and ESI-3, and models agree with each other more than with the ground truth, so ensembling worsens the failure. Deploying LLMs for ED triage based on offline benchmarks alone misses this sequential failure.
[NLP-104] Block-Sparse Attention with Semantic-Geometric Decoupled Routing ACL
【速读】： 该论文旨在解决大语言模型在长上下文推理中因密集注意力（dense attention）的二次计算开销而导致的效率瓶颈问题，同时克服块稀疏注意力（block-sparse attention）在无训练条件下实现精准块路由（block routing）的挑战。其核心问题是现有路由方法依赖于经过旋转位置编码（RoPE）后的令牌表示进行池化聚合，导致语义信息与由RoPE引入的几何结构耦合，且高频相位抵消效应削弱了局部位置线索，从而影响路由准确性。为此，论文提出了一种无需训练的块路由框架——语义-几何解耦路由（Semantic-Geometric Decoupled Routing），其关键在于将语义聚合过程前移至RoPE之前的空间，并利用离线结构先验和相对块距离重建几何偏置，实现语义与几何信息的显式分离。该方法通过闭式解析表达式直接生成块路由得分，避免了逐令牌搜索或后处理校准，显著提升了路由精度与效率。实验表明，该方法在4K–128K上下文长度下逼近全注意力性能，路由延迟低于3.4毫秒，在128K上下文长度下相较FlashAttention实现5.03倍加速。
链接: https://arxiv.org/abs/2609.22884
作者: Xinwei Long,Weigao Sun,Weibo Gao,Pengkun Jiao,Biqing Qi,Feida Zhu,Yiran Zhong,Steven Hoi,Bowen Zhou
机构: Tsinghua University (清华大学); Tongyi Lab, Alibaba Group (阿里巴巴集团通义实验室); University of Science and Technology of China (中国科学技术大学); Shanghai Artificial Intelligence Laboratory (上海人工智能实验室)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Technical report; Submitted to ACL ARR 2026 May
Abstract:Long-context inference has become a defining capability of large language models, but exact dense attention remains costly due to its quadratic scaling with sequence length. Block-sparse attention offers a hardware-friendly alternative by routing each query block to a small set of relevant key blocks, yet accurate training-free block routing remains difficult. Existing routers often pool post-RoPE token representations, which entangles semantic aggregation with RoPE-induced geometry and attenuates local positional cues through high-frequency phase cancellation. To resolve this mismatch, we propose \textbfSemantic-Geometric Decoupled Routing, a training-free block routing framework that shifts semantic aggregation to the pre-RoPE space and reconstructs geometric bias with an offline structural prior and relative block distances. This decomposition yields an explicit closed-form block routing score without token-level search or post-hoc calibration. Experiments on long-context text and video tasks show that our method approaches full-attention accuracy across 4K–128K contexts, keeps routing overhead below 3.4 ms, and achieves a 5.03 \times speedup over FlashAttn at a 128K context length.
[NLP-105] o Consolidate or not to Consolidate? Evaluating the Impact of Consolidation in Multi-Reference Training using Peer Reviews
【速读】： 该论文旨在解决结构化自然语言生成（NLG）任务中，特别是在自动同行评审生成这类介于高度约束的机器翻译与开放式对话生成之间的任务所面临的训练范式不适用问题。此类任务的特点是单一输入可对应多个有效且重叠的输出，传统基于单参考或多重参考的训练方法在捕捉这种多样性与一致性平衡方面表现不佳。其解决方案的关键在于引入一种统一的、经过整合的参考文本作为训练信号，通过将多样化的参考内容合并为一个凝聚的生成目标，从而提升模型对复杂语义空间的理解能力。为此，研究者构建了MERC-36K数据集，包含超过3.6万篇论文及其原始与整合后的同行评审文本，并在此基础上验证了基于整合参考训练的模型在自动与人工评估中均显著优于基于未整合参考训练的模型，证明了统一训练信号对于提升系统性能的核心作用。
链接: https://arxiv.org/abs/2609.22805
作者: Maitreya Prafulla Chitale,Ketaki Mangesh Shetye,Yash More,Harshit Gupta,Manav Chaudhary,Manish Shrivastava,Vasudeva Varma
机构: IIIT Hyderabad; Microsoft(微软), India; Sentisum
类目: Computation and Language (cs.CL)
备注:
Abstract:Natural language generation (NLG) tasks span the spectrum of conditional entropy, ranging from highly constrained machine translation to open-ended dialogue generation. Structured tasks like automated peer-review generation occupy the intermediate region, where a single input admits multiple valid, overlapping outputs. In this work, we demonstrate that traditional single- and multi-reference training paradigms are suboptimal for these intermediary tasks. We provide empirical evidence that consolidating diverse references into a unified training signal is crucial for developing effective systems. To facilitate this, we introduce MERC-36K, a large-scale corpus of over 36,000 papers paired with original and consolidated peer reviews. Using this dataset, we train specific architectures to isolate the impact of different reference paradigms and benchmark against existing state-of-the-art systems. Through extensive automatic and human evaluation, we demonstrate that models trained on consolidated references significantly outperform those trained on unconsolidated references. Dataset and code will be released upon acceptance.
[NLP-106] AlexandriaX 2026: The First Shared Task on Dialectal Arabic Machine Translation
【速读】： 该论文旨在解决方言阿拉伯语机器翻译（Dialectal Arabic Machine Translation, Dialectal MT）中的核心挑战，即如何在翻译过程中有效建模语义内容、方言差异、对话上下文、说话人与受话人特征以及社会语言学上的得体性。传统机器翻译评估指标在揭示方言系统产生的语言错误方面存在局限性。为此，论文提出了AlexandriaX 2026共享任务，通过三个互补的子任务实现突破：（1）针对13种阿拉伯语方言的上下文感知英语到方言阿拉伯语对话翻译；（2）覆盖六种方言的金融领域跨方言翻译；（3）基于语言学驱动的错误类别，在五个阿拉伯语方言上进行细粒度的句段级翻译错误检测与分类。解决方案的关键在于引入显式的方言建模、上下文感知生成、检索与重排序机制，以及面向可解释性的专门化错误分析方法。实验结果表明，最优系统在各子任务中均取得显著性能提升，验证了上述策略的有效性。所有任务数据、基线模型和评估代码均已公开，支持后续研究复现与推进。
链接: https://arxiv.org/abs/2609.22796
作者: Abdellah El Mekki,AbdelRahim A. Elmadany,Samar M. Magdy,Saad Ezzini,Mo El-Haj,Mustafa Jarrar,Zaid Alyafeai,Bernard Ghanem,Muhammad Abdul-Mageed
机构: The University of British Columbia (不列颠哥伦比亚大学); King Fahd University of Petroleum and Minerals (法赫德国王石油与矿业大学); Lancaster University (兰卡斯特大学); VinUniversity (越南大学); Hamad Bin Khalifa University (哈马德本哈利法大学); King Abdullah University of Science and Technology (阿卜杜拉国王科技大学); Canada Research Chair in NLP and ML (加拿大自然语言处理与机器学习首席研究员)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: To Appear in ArabicNLP 2026, resources available in the following project page: this https URL
Abstract:Dialectal Arabic machine translation (MT) remains challenging despite recent progress in Arabic language technologies, particularly because effective translation requires modeling not only semantic content but also dialectal variation, conversational context, speaker and addressee characteristics, and sociolinguistic appropriateness. Moreover, conventional MT metrics provide limited insight into the linguistic errors produced by dialectal systems. We present the AlexandriaX 2026 Shared Task on Dialectal Arabic MT, which addresses these challenges through three complementary subtasks: (1) context-aware English-to-Dialectal Arabic dialogue translation across 13 Arabic varieties, (2) cross-dialect Arabic translation in the financial domain covering six Arabic dialects, and (3) span-level MT error detection and classification using linguistically motivated error categories across five Arabic varieties. The shared task attracted 38 registrations for Subtask 1, 33 for Subtask 2, and 35 for Subtask 3. Twelve unique teams submitted their system description papers, all of which we accepted for publication. The best system on Subtask 1 achieved 30.42 spBLEU in the constrained track and 33.49 spBLEU in the unconstrained track. On Subtask 2, the top system obtained 28.40 spBLEU. On Subtask 3, the best system achieved an overall score of 49.82, outperforming the 24.29 baseline. Taken together, the results of the leading systems across the three subtasks highlight the benefits of explicit dialect modeling, context-aware generation, retrieval and reranking, and specialized approaches to interpretable MT error analysis. All the resources of AlexandriaX 2026 shared task are publicly available, including data, baselines, and evaluation code on our project page: this https URL.
[NLP-107] Diagnose Then Repair: A Two-Stage MQM-Guided Post-Editing Framework for Domain-Specific Machine Translation ACL2026
【速读】： 该论文旨在解决基于大语言模型（LLM）的机器翻译评估在实际应用中仅能提供诊断性反馈，而难以在真实生产环境下有效转化为翻译质量提升的问题。其核心挑战在于现有方法多采用“评估-修正”一体化流程，易引发语义漂移与不可控的过度修改。为此，论文提出一种两阶段、评估器引导的自动后编辑框架：首先通过检索增强的LLM评估器，在明确的编辑契约下输出结构化的、细粒度的MQM（Multidimensional Quality Metrics）诊断；随后由独立的LLM后编辑器仅针对诊断结果实施最小化修正。该分离式设计显著提升了控制能力，并有效抑制了类比性漂移（paraphrastic drift）。在涵盖三个模型提供商、七种语言及七种LLM的系统性实验中，最优配置在COMET-22与COMETKiwi指标上持续优于单阶段后编辑方法，且评估器识别的错误片段及其严重程度与人工MQM标注及人工编辑偏好高度一致，验证了该方案的有效性与可解释性。
链接: https://arxiv.org/abs/2609.22793
作者: Ji Hun Wang,Siyu Wu
机构: Amazon(亚马逊)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted to the ACL 2026 Industry Track and presented at ACL 2026
Abstract:LLM-based machine translation evaluation can closely match human judgments, but in practice it remains largely diagnostic, with the signals rarely translating into direct quality improvements under real production constraints. We propose a two-stage, evaluator-guided automatic post-editing framework that turns MQM-style evaluation into targeted repairs: a retrieval-augmented LLM evaluator outputs structured, span-level MQM diagnoses under an explicit edit contract, and a separate LLM post-editor applies minimal edits restricted to those diagnoses. This separation improves controllability and reduces paraphrastic drift compared to one-stage “judge-and-refine” baselines. In a systematic study involving seven LLMs spanning three model providers and seven languages, our best configuration consistently improves both COMET-22 and COMETKiwi scores over one-stage post-edit methods, while the evaluator’s error spans and severities show strong agreement with human MQM annotations and human editor preferences.
[NLP-108] MIS-Bench: Benchmarking Multimodal LLM s for Psychotherapeutic Interpersonal Skills Assessment
【速读】： 该论文旨在解决当前多模态大语言模型（Multimodal Large Language Models, MLLMs）在需要专家判断的专业评估任务中可靠性不足的问题，特别是在心理治疗人际互动技能评估这一高要求场景下的表现。其核心挑战在于现有模型在多模态输入下与人类专家评分的一致性较低，且基于推理的提示策略未能显著提升性能。为应对这一问题，论文提出MIS-RAFT方法，其关键在于引入一种回归感知的微调机制（Regression-Aware Fine-Tuning），受RAFT启发并针对以小数点后一位精度进行细粒度人际技能评分的任务特性进行定制优化。该方法有效缓解了自回归生成过程与标量型专家评分之间的语义不匹配问题，显著提升了模型评分与人类专家评价的一致性，为实现更可靠的基于模型的专家级评估提供了可行路径。
链接: https://arxiv.org/abs/2609.22778
作者: Yuhan Lu,Yi Yao,Hua Shen,Katie Aafjes-van Doorn,Zhaonan Wang
机构: NYU Shanghai, New York University(纽约大学上海分校，纽约大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Multimodal large language models (MLLMs) are increasingly used as evaluators, yet their reliability in professional assessment tasks that require expert judgment remains unclear. We investigate this challenge in the context of assessing psychotherapeutic interpersonal skills and introduce MIS-Bench, a Multimodal Interpersonal Skills (MIS) benchmark comprising 996 psychotherapy response videos annotated across 8 dimensions of Facilitative Interpersonal Skills. Across 9 MLLMs with multiple modality and prompting settings, we find that current models show only modest agreement with human experts, inconsistent gains from multimodal input, and limited benefits from reasoning-based prompting. To mitigate this gap, we propose MIS-RAFT, a regression-aware fine-tuning method inspired by RAFT and tailored to fine-grained interpersonal skill scoring at one-decimal precision. MIS-RAFT addresses the mismatch between autoregressive token prediction and scalar-valued expert assessment, significantly improving agreement with human ratings. Overall, MIS-Bench reveals a clear gap between general multimodal capability and expert-level interpersonal judgment, while MIS-RAFT offers a promising path toward more reliable model-based assessment.
[NLP-109] NLPCC 2026 Task 10: Citation-Level Faithfulness Verification with DeBERTa Ensembles and Class-Wise Calibration NLPCC2026
【速读】： 该论文旨在解决生成式 AI 辅助科学报告中引文层面忠实性（citation-level faithfulness）的评估问题，具体任务是根据一个原子级科学论断（atomic scientific claim）及其引用文献的结构化全文，判断其与原文之间的语义关系（四分类标签），并识别最多三个支持证据段落。解决方案的关键在于双头架构设计：标签头采用段落感知的交叉编码器（paragraph-aware cross-encoder）与文档级 DeBERTa-large 分类器进行集成，并通过类别专属的决策校准（class-wise decision calibration）提升分类性能；概率层面融合则有效缓解了模型在“主题匹配”类别上出现的过预测倾向（out-of-fold over-prediction）。证据头结合了 top-20 与 top-30 联合模型的段落得分以及 BM25 相似度得分，实现更精准的证据段落排序。整个系统完全离线运行，无需外部检索或大语言模型（LLM）提示。实验表明，模型间的互补性与校准机制显著提升了标签预测性能，而基于真实证据（gold-evidence）的推理对宏观 F1 值影响微小，说明证据排序质量对 Joint@3 指标仍至关重要。
链接: https://arxiv.org/abs/2609.22774
作者: Yanling Li,Zirui Li,Mingyu Wan
机构: The Hong Kong Polytechnic University (香港理工大学)
类目: Computation and Language (cs.CL)
备注: Accepted to the NLPCC 2026 Shared Task 10 system paper track. Ranked 2nd in Track 2 on the final leaderboard
Abstract:This paper presents our system for Track 2 of the NLPCC 2026 Shared Task 10 on citation-level faithfulness in AI-assisted scientific reporting. Given an atomic scientific claim and the structured full text of its cited paper, the task requires both a four-way relation label and up to three evidence paragraph identifiers. The label head ensembles a paragraph-aware cross-encoder with a document-level DeBERTa-large classifier, followed by class-wise decision calibration. Probability-level fusion is motivated by an out-of-fold tendency to over-predict Topical Match. The evidence head combines paragraph scores from top-20 and top-30 joint models with BM25 scores. The system runs fully offline without external retrieval or LLM prompting. On the final leaderboard, our system achieved 82.9898 overall (89.5491 Macro-F1 and 76.4305 Joint@3), ranking second in Track 2. Ablations and error analysis show that model complementarity and calibration drive the label gains. Gold-evidence inference changes label Macro-F1 negligibly, whereas evidence ranking remains important for Joint@3.
[NLP-110] Beyond Final-Token Classification: Heterogeneous Readouts for Evidence-Grounded Suicide Risk Detection ICIP
【速读】： 该论文旨在解决多任务心理风险预测基准中因输出结构异质性导致的性能瓶颈问题，具体包括有序分类（ordinal suicide-risk classification）、多标签心理社会因素检测（multi-label psychosocial factor detection）以及支持性语句提取（extraction of supporting phrases）三类任务之间的协同优化难题。其核心解决方案为异质读出分解（Heterogeneous Readout Decomposition, HRD），关键在于将语义验证与输出实现解耦：通过本地部署的Qwen3.8-27B模型结合任务特异性QLoRA适配器，生成答案标记边界（answer-token margins）与第63层的隐状态表示，用于条件化查询；随后，HRD在有序风险判断中比较四个潜在得分，在多数因素检测中保留标记边界信息，而将七个标签通过一个共享的潜在探测器进行路由，同时基于校准后的风险条件约束构建证据集。实验表明，该方法显著提升加权F1（0.8237 → 0.8372）与宏平均F1（0.7965 → 0.8185），选择性因子路由使宏平均F1提升0.0105，尾部标签宏平均F1提升0.0189，且受限证据解码使行级留出评估中的短语提取F1从0.7488提升至0.7609。结果揭示，答案标记边界而非仅语义表征，是该基准中可度量的误差来源，凸显了对输出结构敏感建模的重要性。
链接: https://arxiv.org/abs/2609.22767
作者: Zirui Li,Yanling Li,Kaolanglang Gao
机构: The Hong Kong Polytechnic University (香港理工大学); Shenzhen University (深圳大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Competition paper for the IEEE Big Data Cup 2026. Ranked 2nd among 31 participating teams and invited for submission to IEEE Big Data 2026
Abstract:The IEEE BigData Cup benchmark combines three prediction problems with different output structures: ordinal suicide-risk classification, multi-label psychosocial factor detection, and extraction of supporting phrases. We introduce heterogeneous readout decomposition (HRD), which separates semantic verification from output realization. A locally deployed Qwen3.8-27B model, adapted with task-specific QLoRA adapters, produces both answer-token margins and layer-63 answer states for card-conditioned queries. HRD compares four latent scores for ordinal risk, retains token margins for most factors while routing seven labels through one shared latent probe, and constructs evidence sets from verbatim span candidates with calibrated, risk-conditional constraints. On two held-out user-grouped confirmation folds, the latent risk readout improves weighted F1 from 0.8237 to 0.8372 and macro F1 from 0.7965 to 0.8185. Selective factor routing improves macro F1 by 0.0105 and tail-label macro F1 by 0.0189; in contrast, global latent replacement and independent label-specific probes fail. The constrained evidence decoder raises pooled phrase F1 from 0.7488 to 0.7609 in row-level out-of-fold evaluation. Lenormand’s best public result is 0.8052 on Subtask 1 and 0.6636 on Subtask 2, giving a composite score of 0.7627. These results identify the answer-token boundary, rather than semantic representation alone, as a measurable source of error in this benchmark.
[NLP-111] Clinical Domain Classification from Medical Transcriptions
【速读】： 该论文旨在解决医疗文本分类中因临床专科类别严重不平衡导致的模型性能下降问题，尤其关注少数类（underrepresented clinical specialties）的分类准确性。其核心解决方案在于结合先进的深度学习模型与有效的数据平衡策略：具体而言，研究对比了六种传统机器学习方法、两种预训练Transformer模型（BERT与XLNet）以及一种少样本大语言模型提示（few-shot prompting）方法，并系统评估了基于自然语言处理的同义词替换文本增强和合成少数类过采样技术（SMOTE）两种数据平衡策略的有效性。实验结果表明，采用SMOTE进行数据平衡可显著提升各类模型的分类性能，其中BERT在平衡后的数据集上达到0.996的F1分数，且训练时间低于XLNet，验证了基于Transformer的表示学习与合理数据平衡相结合在临床领域分类任务中的优越性。
链接: https://arxiv.org/abs/2609.22734
作者: Sravani Pottipati,Lakshmikar R. Polamreddy
机构: Independent Researcher(独立研究员); Yeshiva University (叶史瓦大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Clinical domain classification plays an important role in organizing and analyzing large volumes of unstructured medical text. However, medical transcription datasets are often highly imbalanced, which can substantially degrade classification performance, particularly for underrepresented clinical specialties. In this work, we present a comparative study of machine learning and transformer-based approaches for clinical domain classification from medical transcriptions. We evaluate six traditional machine learning classifiers—Naive Bayes, Support Vector Machine (SVM), Decision Tree, Random Forest, K-Nearest Neighbors (KNN), and XGBoost—along with two pretrained transformer models, BERT and XLNet, and a few-shot large language model prompting approach. Experiments are conducted on medical transcription data collected from MTSamples, comprising 5,013 samples across 40 clinical specialties. To address severe class imbalance, we investigate two balancing strategies: text augmentation using NLP-based synonym replacement and Synthetic Minority Over-sampling Technique (SMOTE). Experimental results demonstrate that data balancing substantially improves classification performance across the evaluated models. In particular, BERT achieves the highest F1-score of 0.996 on the SMOTE-balanced dataset while requiring lower training time than XLNet. The results highlight the effectiveness of transformer-based representations combined with appropriate data balancing strategies for clinical domain classification and provide a systematic comparison of classical machine learning, transformer models, and few-shot prompting for medical transcription analysis.
[NLP-112] Analyzing Public Discourse on Urbanism: Topic Clustering Sentiment Analysis and Retrieval-Augmented Generation using YouTube Comments
【速读】： 该论文旨在解决城市议题（如步行友好性、自行车基础设施、公共交通、住房密度和街道安全）在社交媒体等在线话语中信息量大但结构不规范，而现有城市评估工具无法有效捕捉此类非结构化文本的问题。其核心解决方案是构建一个集成地理实体识别、主题建模、情感分析与基于检索增强生成（Retrieval-Augmented Generation, RAG）的系统，利用涵盖309个北美城市的22,788段YouTube视频转录文本及评论进行分析。关键发现在于：尽管经过推特数据调优的RoBERTa分类器在宏观F1得分上显著优于VADER词典基线（0.589 vs. 0.464；McNemar检验p=0.0001），但两者在“中性”类别上均表现不佳，该类别在城市议题讨论中占主导地位，且人工标注者对该类别的一致性较低（Cohen’s kappa=0.53）；密集检索方法在所有截断点上均优于TF-IDF基线（P@5: 0.790 vs. 0.560），而以视频级别为相关性代理会严重低估片段级别的精确度（人类评分显示0.660 vs. 0.94）；在生成内容连贯性评估方面，BERTScore在多句摘要与单句评论对比时失效，得分几乎恒定，无法反映真实相关性，而基于ROUGE-1的连贯性指标实际反映的是改写程度而非幻觉率，构成一个伪下界。这些发现不仅适用于城市议题领域，也普遍适用于任何基于短用户生成文本构建的RAG系统。
链接: https://arxiv.org/abs/2609.22705
作者: Jakob Morales,Monica Hegde,Fayeq Jeelani Syed
机构: Indiana University(印第安纳大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Online discourse about urban issues - walkability, cycling infrastructure, public transit, housing density, and street safety - is voluminous but unstructured, and existing city-evaluation tools capture none of it. We present a pipeline and conversational system that combines geographic entity resolution, topic modeling, sentiment analysis, and Retrieval-Augmented Generation (RAG) over 22,788 chunks of YouTube transcripts and comments spanning 309 North American cities. Beyond the system itself, our contribution is a set of measurements about what happens when standard NLP components meet short, informal, geographically ambiguous text. A Twitter-tuned RoBERTa classifier outperforms a VADER lexicon baseline by 12.6 macro-F1 points (0.589 vs. 0.464; McNemar p = 0.0001), but both models collapse on the neutral class, which dominates urbanist comment traffic; annotators disagree on the same class (Cohen’s kappa = 0.53). Dense retrieval beats a TF-IDF baseline at every cutoff (P@5 0.790 vs. 0.560), and video-level relevance proxies understate chunk-level precision by a wide margin (0.660 vs. 0.94 under human rating). For groundedness evaluation, we find BERTScore unusable when a multi-sentence generated summary is compared against a single short comment - scores are nearly flat regardless of relevance - and show that ROUGE-1-based groundedness is a paraphrase-driven lower bound rather than a hallucination rate. These findings generalize beyond the urbanist domain to any RAG system built over short user-generated documents.
[NLP-113] LLaDA-PRM: A Bidirectional Step-Level Reasoning Evaluator
【速读】： 该论文旨在解决生成式推理评估中因自回归语言模型（autoregressive language models）的因果注意力机制（causal attention）导致的步骤级推理评估不准确问题。传统方法受限于每个步骤表示仅能依赖问题描述、先前步骤及当前步骤，无法充分捕捉后续步骤对早期步骤有效性的验证作用。为验证这一假设，作者通过54次受控实验对比了因果与双向注意力（bidirectional attention）的LLaDA评估器，在1B–3B规模下仅改变自注意力掩码，结果表明双向注意力在所有测试场景中均带来一致性能提升。基于此发现，论文提出\prm——一个8B参数的双向注意力评估器，其在MR-MATH-invalid数据集上达到88.8的步骤级F1分数，在跨分布的MR-GSM8K原始问题子集上达83.8，分别优于ReasonEval-Llemma-34B 11.3和10.3个F1点。此外，\prm在在线评估不完整推理轨迹时仍表现优异，显著超越现有最强基线。研究进一步证明，\prm可作为有效的训练数据筛选信号，显著提升Mistral-7B在MATH-500上的性能。其核心解决方案在于引入双向注意力机制，使评估过程能够利用完整解题路径中的下游信息来反向验证早期推理步骤的合理性。
链接: https://arxiv.org/abs/2609.22700
作者: Yiming Feng,Naihao Deng,Yulong Chen,Rada Mihalcea
机构: University of Michigan (密歇根大学); University of Aberdeen (阿伯丁大学); University of Cambridge (剑桥大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: 17 pages, 7 figures, 8 tables, including appendices
Abstract:Step-level reasoning evaluators are commonly based on autoregressive language models, whose causal attention restricts each step representation to the problem, previous steps, and the current step. Yet, when the complete solution is available, the validity of an earlier step may become clearer only through its downstream consequences. We validate this hypothesis through a controlled 54-run comparison of causal and bidirectional LLaDA evaluators at 1B–3B scale, changing only the self-attention mask, and find bidirectional attention yields consistent improvements. Building on this finding, we introduce \prm, an 8B bidirectional evaluator that reaches 88.8 step-level F1 on MR-MATH-invalid and 83.8 on the out-of-distribution MR-GSM8K original-question subset, outperforming ReasonEval-Llemma-34B by 11.3 and 10.3 F1 points, respectively. \prm also remains effective when evaluating incomplete reasoning traces in online settings, outperforming the strongest baselines on both benchmarks by a large margin. We further show that \prm provides an effective training-data selection signal, improving Mistral-7B performance on MATH-500.
[NLP-114] COT-TTS: Audio Context-Aware Text-to-Speech with Chain-of-Thought Reasoning
【速读】： 该论文旨在解决现有文本到语音（Text-to-Speech, TTS）系统在自然对话场景中缺乏上下文感知能力的问题，即当前系统生成的语音风格通常依赖于显式用户指令，而无法像人类对话般根据前后文语境自动推断并适配说话风格。为此，论文提出了一种基于推理的上下文感知型文本到语音任务——COT-TTS（Chain-of-Thought Text-to-Speech），其核心在于：在给定历史对话音频、目标文本和参考语音的前提下，模型需首先理解对话上下文，通过显式的中间推理过程推断出合适的说话风格，最终合成具有指定音色的目标语音。该方案的关键在于引入链式思维（Chain-of-Thought, CoT）机制，使模型能够生成可解释的风格推理路径，从而实现更自然、连贯且情感一致的语音输出。研究构建了一个包含900万训练样本的大规模双语对话语音数据集，并设计了800个经人工验证的源不重叠基准样本，同时开发了参数量分别为0.6B和1.7B的端到端自回归模型，支持生成带情感标签的转录文本、可编辑的语音风格推理以及语音令牌。实验表明，该模型在参数量显著减少的情况下，性能接近大规模基线系统，在时长一致性与情感一致性方面表现优异，能根据上下文合理生成情绪、重音与节奏变化。相关数据构建流程、数据集、训练模型及资源已公开，以推动后续研究。
链接: https://arxiv.org/abs/2609.22697
作者: Weizhen Bian,Sitong Cheng,Rongxiu Zhong,Jiahao Pan,Liumeng Xue,Boyi Kang,Shilei Zhang,Jinglei Liu,Yue Wang,Junlan Feng,Bei Liu,Wei Xue
机构: The Hong Kong University of Science and Technology, Hong Kong SAR, China; JIUTIAN Research, China Mobile, Beijing, China; The State Key Laboratory of Multimedia Information Processing, Peking University, Beijing, China; Nanjing University; China Mobile (Hong Kong) Innovation Research Institute, Hong Kong SAR, China
类目: Computation and Language (cs.CL); Sound (cs.SD); Audio and Speech Processing (eess.AS)
备注: 13 pages, 6 figures, 6 tables
Abstract:Recently, text-to-speech systems have made significant progress in speech expressiveness and controllability. However, the speaking style of generated speech typically relies on clear user-specified instructions. In natural conversations, speaking style should be naturally inferred from the preceding conversational context. Therefore, we propose COT-TTS, a context-aware, reasoning-based text-to-speech task. Given historical conversation audio, target text, and a reference speech, the system should comprehend the conversational context, infer an explicit intermediate reasoning, and finally synthesize the target speech with the specified timbre. To support this task, we constructed a large-scale bilingual conversational speech dataset comprising 9 million training samples, including a high-quality subset of 1 million samples. We further constructed a source-disjoint benchmark with 800 human-verified samples and established strong task-specific baselines. Additionally, we developed end-to-end autoregressive models with parameter sizes of 0.6B and 1.7B, generating emotion-labeled transcripts, editable speech style inferences, and speech tokens. Experimental results show that the proposed model achieves performance comparable to large-scale baseline systems with significantly fewer parameters. At the same time, the model performs well in terms of duration consistency and emotional consistency, and can generate appropriate emotional, stress, and rhythmic variations based on the conversational context. To facilitate future research, we will publicly release the data construction pipeline, dataset, trained models, and related resources. The demo page and additional resources are available at this https URL
[NLP-115] Beetle: A Bilingual Model Suite for Modelling Second-Language Processing EMNLP
【速读】： 该论文旨在解决多语言预训练模型中因训练条件（如暴露结构、数据规模、模型架构等）多重变量混杂导致难以准确归因于单一因素的问题。现有研究通常同时改变多个训练条件，限制了对第二语言（L2）行为形成机制的系统性理解。为此，论文提出Beetle——一种可控的双语语言模型预训练框架，其核心优势在于可独立调控分词器（tokeniser）、目标语言、训练预算与暴露结构，从而实现对不同训练条件的系统化、可比性实验。该框架支持在多种暴露时序、数据规模及第一语言（L1）组合下训练并发布285个双语和45个单语开源语言模型，涵盖丰富的检查点。通过在人类双语者与二语学习者的阅读时间预测及语法判断任务上评估模型表现，研究发现阶段性且具有时间结构的课程设计（staged and temporally structured curricula）显著优于均衡双语训练，尤其在小规模数据和语言类型相近的语言对中效果更优。Beetle模型为推动计算心理语言学从主流的单语、英语中心范式转向对人类双语处理机制的建模提供了有力工具，有助于揭示跨语言学习动态，并支持社区驱动的受控模型家族构建。
链接: https://arxiv.org/abs/2609.22633
作者: Suchir Salhan,Catherine Arnett,James Michaelov,Paula Buttery
机构: University of Cambridge; EleutherAI; MIT
类目: Computation and Language (cs.CL)
备注: Accepted EMNLP Main Conference 2026
Abstract:Bilingual language models (LMs) offer a controlled setting for studying how training conditions shape second-language (L2) behaviour, but prior work typically varies exposure structure, scale, and architecture at once, making it difficult to attribute effects to any single factor. We introduce Beetle, a controlled language model pretraining framework in which tokeniser, target language, training budget, and exposure structure are each independently manipulable, enabling systematic and comparable experimentation of training conditions. Using Beetle, we train and release 285 bilingual and 45 monolingual open-source LMs with rich checkpoints across a range of exposure schedules, data scales and first languages (L1s) to study multilingual pretraining and computational modelling of bilingualism and second language learning. Evaluating models on human bilingual and second language reading-time prediction and grammaticality judgement tasks, we find that staged and temporally structured curricula consistently improve alignment with language learner reading time compared to balanced bilingual training, with the largest gains at smaller data scales and for typologically closer language pairs. The Beetle models are well suited tools to help move computational psycholinguistics beyond its prevailing monolingual, English-centric focus toward models of human bilingual processing, to study cross-lingual learning dynamics, while supporting community-based development of controlled model families.
[NLP-116] Pretrained Persona Mixture Models and Tandem Models for Human Simulation
【速读】： 该论文旨在解决当前大语言模型（LLM）在人类模拟中普遍采用的提示指令微调助手模型进行角色扮演所导致的模拟不准确与刻板化预测问题，即生成结果缺乏真实人类对话应有的自然多样性。其核心解决方案在于提出“人格混合模型”（Persona Mixture Models, PMMs），通过使用特定个体的简短对话样本对模型进行人格绑定，而非依赖刻板化的指令提示，从而在不引入负面效应的前提下实现更真实的人类行为模拟。研究进一步表明，PMMs相比传统指令微调模型能够更准确地预测人类交互，并保留更丰富的词汇、语义与语用多样性。为克服基础预训练模型在长上下文或跨领域场景下可能出现的偏离域对话及内部状态丢失问题，论文还提出了“串联模型”（tandem models），即结合预训练模型与指令微调监督器的架构，在实验中展现出最优的整体准确性与多样性表现。
链接: https://arxiv.org/abs/2609.22607
作者: Minwoo Kang,Téa Wright,Seun Eisape,Ayush Raj,Suhong Moon,Joseph Suh,Alane Suhr,David M. Chan,John Canny
机构: University of California, Berkeley(加州大学伯克利分校); Google(谷歌)
类目: Computation and Language (cs.CL)
备注: 11 pages in body, 36 with appendices. 7 figures. 8 tables
Abstract:We argue here that the current dominant practice in LLM human simulation: prompting instruction-tuned assistant language models to role-play personas, is inaccurate and produces stereotyped predictions (lacking natural diversity). It has previously been shown that LLMs can be bound to personas using naturalistic, freetext dialog avoiding stereotyping. Here we show that binding can also be achieved using short, individual samples of dialog from specific people. Demographics can be added later without negative effects by simply querying the model. We use the term Persona Mixture Models (PMMs) for well-calibrated human models, currently realized as pretrained base models. We show that PMMs produce more accurate predictions than instruction-tuned models and retain more of the lexical, semantic, and pragmatic diversity found in human dialog. We measure realism and diversity of LLMs simulating human interlocutors across a diverse set of corpora spanning open-domain text, human-AI chat, and task-oriented dialogue between human speakers. However, base pretrained models can produce out-of-domain dialog and may lose some of the human’s internal state over long contexts. We propose and explore tandem models which combine a pre-trained model with an instruction-tuned supervisor. Tandem models achieve the best overall accuracy and diversity in our experiments.
[NLP-117] Preserving What Matters: Semantic Scaffolds Beyond Saturation in Summarization Evaluation
【速读】： 该论文旨在解决生成式摘要在实际生产系统中广泛应用背景下，模型选择依赖的摘要质量评估难题。现有评估指标存在显著局限：ROUGE仅捕捉表面词汇重叠，无法反映语义完整性；而以大语言模型（Large Language Model, LLM）作为评判者的方法则在多数情况下出现分数饱和现象，导致不同模型间的区分能力丧失。针对这一问题，本文提出“语义骨架”（Semantic Scaffold）评估框架，其核心创新在于从源文本中提取包含事实、问题及实体属性的分层语义结构，并将各成分标注为主要点或支持细节，形成固定参考标准用于摘要评分。基于此结构，构建了三项诊断性指标——事实保留率（Fact Preservation Score, FPS）、问题保留率（Question Preservation Score, QPS）与实体保留率（Entity Preservation Score, EPS），旨在奖励关键信息的保留同时惩罚冗余细节的过度堆砌。该框架在多语言环境、多个公开与私有数据集上均表现出强鲁棒性，能够有效识别传统指标失效时的模型性能差异，揭示了四种典型失败模式，证明语义骨架驱动的评估体系在整体评价轴崩溃时仍具备可解释且富有信息量的诊断能力。
链接: https://arxiv.org/abs/2609.22603
作者: Nikhil Reddy Pottanigari,Ramin Fahimi,Noah Bolger,Sepideh Kharaghani,Ying Zhang
机构: ServiceNow Canada(服务现在加拿大)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted at the AIMS Workshop at COLM 2026
Abstract:Summarization ships in countless production systems, making model selection a routine decision that depends on measuring summary quality. Existing metrics struggle to support this: ROUGE captures only surface overlap, while LLM-as-judge scores saturate to near-identical values that fail to rank models effectively. We observe this saturation across three public datasets, two proprietary datasets, and multilingual settings. Motivated by this, we introduce Semantic Scaffold, an evaluation framework that extracts a hierarchical representation of facts, questions, and entity attributes from a source text, labeling each as a main point or supporting detail, and reusing this structure as a fixed reference for scoring summaries. From this representation, we derive three diagnostic metrics: Fact Preservation Score (FPS), Question Preservation Score (QPS), and Entity Preservation Score (EPS), designed to reward the preservation of essential information while penalizing detail overload, and position them as interpretable diagnostics that remain informative where holistic axes collapse. Finally, we analyze four recurring failure modes of ROUGE and LLM-as-judge scores, demonstrating that scaffold-based evaluation remains informative where conventional metrics collapse.
[NLP-118] Do Student LLM s Inherit OOD Robustness? Invariance-Weighted Distillation for Reliable Knowledge Transfer
【速读】： 该论文旨在解决知识蒸馏（Knowledge Distillation, KD）在分布外（Out-of-Distribution, OOD）场景下性能显著下降的问题，这一问题的核心在于现有方法未能充分考虑教师模型在不同样本上对因果特征与虚假相关性（spurious correlations）的依赖程度。论文识别出两个加剧OOD性能退化的关键机制：一是数据虚假性（data spuriousness），即学生模型可能学习到训练数据中的虚假关联而非真实因果关系；二是教师能力异质性，即标准KD对所有样本采用统一权重，忽略了教师模型在特定样本上是否依赖于因果特征或被虚假捷径误导。针对上述问题，论文提出了一种理论驱动的不变性加权蒸馏（Invariance-Weighted Distillation, IWD）框架，其核心创新在于通过在多个合成环境下的预测不变性估计教师模型对因果特征的依赖程度，并据此动态调整训练样本的蒸馏权重。IWD能够有效扰动虚假线索，同时保留核心语义信息，赋予那些教师预测保持不变的样本更高权重，从而引导学生模型更关注因果特征而非虚假相关性。理论上，IWD可降低学生模型的虚假-因果（Spurious-to-Causal, S2C）梯度比，促使学生学习更具不变性的表示。在四个NLP基准任务（MNLI、SQuAD-v2、CoNLL-2003 NER和SST-2）上，基于DeBERTa-v3和Qwen-2.5两个模型家族的实验表明，IWD在16个OOD评估中于15个上取得最高准确率，且在自然语言推理（NLI）和问答（QA）任务上分别较标准KD提升4.34和14.94个百分点，同时保持了与主流基线相当的分布内（In-Distribution, ID）性能。
链接: https://arxiv.org/abs/2609.22566
作者: Dileesha Kannangara,Sanghamitra Dutta
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Knowledge distillation (KD) aims to compress high-performance teacher LLMs into lightweight students. However, distilled students often exhibit substantial performance degradation in out-of-distribution (OOD) settings, a critical gap that remains underexplored. We identify two compounding mechanisms causing OOD performance degradation: (1) data spuriousness: students can learn spurious correlations in the distillation dataset over genuine causal relationships; and (2) teacher capability: standard KD treats all samples uniformly, ignoring whether the teacher is guided by causal features or misled by spurious shortcuts on a given sample. To address these challenges, we propose Invariance-Weighted Distillation (IWD), a theoretically grounded framework that dynamically reweights training samples using an estimate of the teacher’s causal reliance derived from prediction invariance across multiple synthetic environments. IWD perturbs spurious cues while preserving core semantics, assigning higher distillation weights to samples whose teacher predictions remain invariant, indicating greater reliance on causal rather than spurious features. We theoretically show that IWD reduces the student’s Spurious-to-Causal (S2C) gradient ratio compared to standard uniformly weighted KD, driving the student toward more invariant representations. Experiments on four NLP benchmarks (MNLI, SQuAD-v2, CoNLL-2003 NER, and SST-2) across two model families (DeBERTa-v3 and Qwen-2.5) demonstrate that IWD consistently outperforms strong KD baselines on OOD evaluations while maintaining competitive in-distribution (ID) performance. Specifically, IWD achieves the highest accuracy in 15 out of 16 OOD benchmarks and improves average OOD performance over standard KD by 4.34 percentage points on NLI and 14.94 percentage points on QA.
[NLP-119] Correct Diagnosis Better Feedback: A Symbolic-Verifier for Faithful LLM Tutoring Feedback in Logic Proofs
【速读】： 该论文旨在解决生成式 AI 在逻辑证明辅导中因错误诊断学生推理错误而导致反馈失效的核心问题。在命题逻辑证明辅导场景下，学生的行为可依据形式化推理规则进行验证，但现有方法往往未能准确识别具体错误类型，导致后续生成的反馈虽看似合理却针对错误问题。其解决方案的关键在于提出一种“验证器驱动（verifier-grounded）”的架构，将错误诊断与语言生成过程解耦：先由符号验证器精确识别错误类型，再由共享的推理链与反馈生成代理基于诊断结果生成回应。实验表明，零样本检测器的宏平均F1仅为0.191，微调后提升至0.709，但仍存在结构性相关类别间的系统性误判；而验证器驱动的方法在诊断正确性上表现最优。研究进一步揭示，生成的推理链与反馈会忠实传递上游诊断错误，即使内容看似符合教学规范，也可能完全偏离真实问题，说明反馈的“表观质量”无法反映诊断准确性。因此，该研究强调必须将诊断的“忠实性（faithfulness）”与反馈的“正确性（correctness）”分开评估，以确保辅导系统的有效性。
链接: https://arxiv.org/abs/2609.22553
作者: Tahreem Yasir,Arnav Mody,Xioayi Tian,Tiffany Barnes
机构: North Carolina State University(北卡罗来纳州立大学); Purdue University(普渡大学); Kennesaw State University(肯尼索州立大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:Effective LLM tutoring depends on correctly identifying the specific error in a student’s reasoning before generating feedback. We study this problem in propositional-logic proof tutoring, where student actions can be checked against formal inference rules. We introduce a verifier-grounded architecture that separates diagnosis from language generation. Using 600 balanced student actions, we compare a zero-shot LLM detector, a fine-tuned detector, and a symbolic verifier. Each diagnosis is processed by shared rationale and feedback agents, isolating the effect of the initial diagnosis. The zero-shot detector achieves a macro-F1 of 0.191; fine-tuning raises this to 0.709 but retains systematic errors between structurally related classes. Rationales generally preserve the diagnosis supplied to them, showing that an incorrect diagnosis can be faithfully propagated through the pipeline. Feedback can likewise remain faithful to its rationale, non-revealing, and pedagogically appropriate while addressing the wrong error. Verifier-grounded feedback achieves the highest diagnostic correctness, and expert ratings largely uneven with the automatic feedback evaluations. These findings show that apparent feedback quality can conceal upstream diagnostic errors and that faithfulness must be evaluated separately from correctness. Our code is publicly available
[NLP-120] Cross-Dialect NER for Bangla Regional Dialects Using Leave-One-Dialect-Out Cross-Validation and Explainable AI
【速读】： 该论文旨在解决孟加拉语（Bangla）因区域性方言差异显著而导致的命名实体识别（NER）模型泛化能力受限的问题。由于巴里沙尔、吉大港、锡尔赫特、诺阿卡利和米姆恩辛等主要方言在词汇、形态和句法层面存在显著差异，基于标准孟加拉语或单一方言训练的模型难以有效迁移至其他方言。为此，研究提出一种跨方言孟加拉语NER框架，基于公开的ANCHOLIK-NER数据集（包含17,405条标注句子和101,817个词元，覆盖五种主要方言），采用“留一方言外交叉验证”（LODOCV）策略，在四类方言上训练模型并在未见方言上评估性能。在相同实验设置下对比八种预训练Transformer模型，结果显示多语言E5 Large（Multilingual-E5 Large）在各折中均表现最佳，于米姆恩辛方言上达到最高F1分数97.26%，而在最具挑战性的吉大港方言上最低为82.38%。为进一步提升可解释性，引入局部可解释模型无关解释（LIME）进行词级预测分析，发现模型决策主要依赖目标实体词的表面形式，而非上下文语境。该研究建立了首个跨方言孟加拉语NER基准，验证了基于Transformer的迁移学习在低资源方言场景下的有效性，同时揭示了当前模型对表面形式的高度依赖性，为未来研究指明方向。
链接: https://arxiv.org/abs/2609.22536
作者: Shamim Rahim Refat,Faika Fairuj Preotee,Shuvashis Sarker,Shifat Islam,Bidyarthi Paul,Mohammad Ashraful Hoque
机构: Ahsanullah University of Science and Technology (阿罕默德大学科技学院); Southeast University (东南大学); Technische Universität Dresden (德累斯顿工业大学); Bangladesh University of Engineering and Technology (孟加拉国工程与技术大学)
类目: Computation and Language (cs.CL)
备注: Accepted for Publication in IEEE 3rd INTERNATIONAL CONFERENCE ON COMPUTING, APPLICATIONS AND SYSTEMS (COMPAS 2026)
Abstract:Bangla, the seventh most spoken language in the world, exhibits significant regional dialectal diversity, with dialects such as Barishal, Chattogram, Sylhet, Noakhali, and Mymensingh differing in lexical, morphological, and syntactic characteristics. These variations pose substantial challenges for Named Entity Recognition (NER), limiting the generalization of models trained on Standard Bangla or a single regional dialect. This paper presents a cross-dialect Bangla NER framework using the publicly available ANCHOLIK-NER dataset, comprising 17,405 annotated sentences and 101,817 tokens across five major Bangla regional dialects. A Leave-One-Dialect-Out Cross-Validation (LODOCV) strategy is adopted, training models on four dialects and evaluating on the remaining unseen dialect. Eight pretrained transformer-based models, including BanglaBERT, MuRIL, XLM-RoBERTa, and Multilingual-E5, are evaluated under identical experimental settings. Multilingual-E5 Large achieves the highest F1-score in every fold, peaking at 97.26% on Mymensingh and reaching its lowest, 82.38%, on Chattogram, the most challenging target dialect. To improve interpretability, Local Interpretable Model-agnostic Explanations (LIME) are applied to word-level predictions, revealing that the model’s decisions are driven primarily by the surface form of the target entity word itself rather than by surrounding sentence context. These findings establish a benchmark for cross-dialect Bangla NER and demonstrate the effectiveness of transformer-based transfer learning for low-resource regional dialects, while highlighting the surface-form dependence of current models as a direction for future work.
[NLP-121] When Cosine Similarity Fails to Reflect Linearly Accessible Structure in Dialogue Models
【速读】： 该论文旨在解决生成式对话模型中隐含的假设——即余弦相似性（cosine similarity）能够有效反映任务相关的表征结构——在实际应用中的可靠性问题。研究发现，在三款7-80亿参数的对话调优大语言模型中，环境余弦相似性显著低估了隐藏状态中可线性解码的人物角色（persona）结构；具体表现为，线性探测器的受试者工作特征曲线下面积（AUC）为0.73–0.97，而基于余弦k近邻（cosine kNN）的评估结果仅为0.56–0.77（针对30类任务）。其关键解决方案在于识别并利用低维有监督子空间（low-dimensional supervised subspace），该子空间能显著恢复性能差距，而与之秩匹配的主成分分析（PCA）子空间则无法实现这一效果，甚至在某些情况下导致性能下降。进一步分析表明，这种不匹配具有特定模型架构和任务语境依赖性：它在单句情感分类（SST-5）任务中不存在，且通过匹配基数的对照实验排除了属性基数作为混杂因素的可能性。此外，该性能差距并不随对话轮次单调增加，且任务对齐子空间在时间上保持稳定。然而，其中两个模型违反预注册的子空间可分性不变性标准（|ΔAUC| = 0.03），一个模型违反预注册的轮次不变性标准（|ΔL| = 0.05），揭示出余弦相似性在捕捉对话表示中潜在任务结构方面的根本局限性，即便该结构本身是线性可解的。
链接: https://arxiv.org/abs/2609.22522
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:Cosine similarity is widely used to analyze transformer representations, implicitly assuming that similarity reflects task-relevant structure. We study when this assumption fails in dialogue-conditioned large language models. Across three 7-8B chat-tuned models, ambient cosine similarity substantially underestimates linearly decodable persona structure on the same hidden states; numerically, linear probe AUC is in the 0.73-0.97 range while cosine kNN is in the 0.56-0.77 range on a 30-class task. A low-dimensional supervised subspace recovers much of this gap, whereas a matched-rank PCA subspace does not and in some cases degrades performance. This mismatch is regime-dependent: it is absent in single-sentence sentiment classification (SST-5), and a matched-cardinality control rules out attribute cardinality as a confound. The gap does not systematically increase across dialogue turns, and the task-aligned subspace remains stable over time. However, two of three models violate a pre-registered within-subspace separability invariance criterion (|Delta AUC| = 0.03), and one model violates a pre-registered turn-invariance criterion (|Delta L| = 0.05). These results show that cosine similarity can fail to reflect task-aligned structure in dialogue representations even when that structure is linearly accessible.
[NLP-122] CultureMINE: Datasets and Methods for Improving the Cultural Capabilities of NLP Systems
【速读】： 该论文旨在解决文化自然语言处理（Cultural NLP）领域中研究文献迅速增长导致方法与数据资源趋势难以追踪的问题。其核心挑战在于如何系统性地梳理该领域在文化能力（Cultural Capabilities, CCs）目标、文化数据资源构建方式及提升方法上的发展脉络。解决方案的关键在于对超过375篇相关论文进行系统性分析，从三个互补维度展开：识别当前NLP系统所针对的文化能力类型、归纳文化数据资源的构建范式，并总结提升文化能力的有效方法。通过整合这些分析结果，论文揭示了领域内的主要趋势与研究空白，并发布了包含全部分析文献的交互式网络界面，支持研究人员持续更新和贡献工作，从而为推动文化NLP领域的可持续发展提供可扩展的基础设施与资源支持。
链接: https://arxiv.org/abs/2609.22494
作者: Tania Chakraborty,Eylon Caplan,Zhaoqing Wu,Kevin Cushing,Han Qin,Shreya Havaldar,Dan Goldwasser
机构: Purdue University(普渡大学); University of Pennsylvania(宾夕法尼亚大学)
类目: Computation and Language (cs.CL)
备注: Accepted to NLP+CSS 2026
Abstract:In recent years, there has been a surge of interest in Cultural NLP, with substantial efforts to create globally inclusive NLP systems. The rapid growth of literature in this field makes it difficult to track trends in methods and data resources. To address this, we analyze over 375 papers to answer three complementary questions: (1) What Cultural Capabilities (CCs) are being targeted in NLP systems? (2) How are cultural data resources being created? and (3) What methods are being used to improve the CCs of those systems? We discuss trends observed across the three questions, and identify relevant research gaps. To facilitate further research in this field, we release our full list of analyzed papers in the form of an interactive web interface, which includes a feature to allow researchers to add their work; we hope this facilitates future research and proves to be a valuable resource for the Cultural NLP community.
[NLP-123] Replication Without Persistence in Hosted LLM s: Measurement Sensitivity in Action-Time Belief Evaluation
【速读】： 该论文旨在解决生成式 AI（Generative AI）在托管语言模型行为评估中结果不一致的问题，即由于评估服务、测量工具或两者同时发生变化，导致行为评估结果存在显著差异。其核心解决方案在于提出并验证三个关键的验证问题：可重复性（replication，相同配置下新数据是否重现历史发现）、测量敏感性（measurement sensitivity，同一标识符下重建评估与推理配置时结果是否变化）以及持久性（persistence，同一测量工具下不同标识符间发现是否保持一致）。研究以Regent Chess这一具有精确可观测隐藏状态的序列环境为实验平台，通过对比不同配置下的模型表现，发现历史报告的Gemini 3.1 Flash-Lite性能缺陷可在新数据中复现，但同一模型在相同标识符下因配置重建导致性能指标下降0.0429，且所有配置组件协同变化无法孤立分析；进一步地，在跨版本比较中，模型表现方向发生反转，表明评估结果高度依赖于具体标识符、服务周期和配置组合。因此，研究强调必须对托管模型的行为结论进行显式标注，明确记录所用标识符、服务周期、测量工具及推理配置，以提升评估结果的透明性与可比性。
链接: https://arxiv.org/abs/2609.22478
作者: Bhushan Kashinath Joshi
机构: 未知
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Behavioural evaluations of hosted language models can vary because the evaluated service, the measurement instrument, or both differ across runs. We separate three validation questions: whether a prior finding recurs on fresh data under its historical configuration (replication), whether the endpoint changes when the evaluation-and-inference configuration is rebuilt under the same identifier (measurement sensitivity), and whether the finding persists across subsequently tested identifiers under one common instrument (persistence). We study these questions in Regent Chess, a sequential environment in which a hidden, mutable state is recorded exactly, allowing stated beliefs to be scored against ground truth at action time; positive endpoint values mean worse performance than a matched-uniform comparator. The previously reported Gemini 3.1 Flash-Lite deficit recurs on fresh games under its historical configuration (+0.0530, 95% CI [+0.0329,+0.0714]). In a back-to-back same-day H/R comparison under the same public identifier, the model-minus-uniform endpoint is 0.0429 lower under the rebuilt configuration (95% CI for the H-minus-R contrast [+0.0182,+0.0667]); all six configuration components vary jointly, so no component is isolated. Under rebuilt R, the prospectively frozen, interleaved same-window 4K comparison reverses sign between Gemini 3.1 and Gemini 3.7, identifiers that differ in release and product tier; additional descriptive and exploratory cells show the same directional pattern. Any additional serving-period contribution remains unresolved (-0.0166, [-0.0483,+0.0157]). Replication, measurement sensitivity, and persistence can therefore yield different conclusions within one evaluation, motivating explicit indexing of hosted-model behavioural claims by tested identifier, serving period, measurement instrument, and inference configuration.
[NLP-124] Efficient Mixture-of-Experts with Speculative Decoding via Expert Coactivation
【速读】： 该论文旨在解决将混合专家（Mixture-of-Experts, MoE）模型与推测解码（Speculative Decoding, SD）协同部署时面临的推理加速瓶颈问题。核心挑战在于，尽管SD通过并行验证多个词元可提升密集模型的推理速度，但当应用于MoE模型时，若增加验证词元数量，会导致更多专家从动态随机存取存储器（DRAM）传输至神经处理单元（NPU），显著增加内存传输开销，进而成为推理延迟的主要瓶颈。为应对这一问题，论文提出关键解决方案：优化训练阶段的MoE路由机制设计，以增强专家共激活（expert coactivation）程度。研究发现，具备高专家共激活特性的路由策略能有效缓解因多词元验证带来的内存传输压力。通过系统评估多种路由设计因素，包括全局负载均衡损失、共享专家结构、一致性损失以及自回归专家选择机制，结果表明这些组件的联合使用可显著提升专家共激活能力。最终实现的模型在保持与基线MoE相当精度的前提下，推理吞吐量相比基准模型提升21%，验证了优化路由设计对提升SD与MoE融合效率的关键作用。
链接: https://arxiv.org/abs/2609.22471
作者: Kumari Nishu,Han-Byul Kim,Santosh Chilkunda,Maxwell Horton,Arnav Kundu,Mohammad Samragh,Lauren Hannah,Mohammad Sekhavat,Nikhil Bhendawade,Manuel Ciosici,Iman Mirzadeh,Keivan Alizadeh Vahid,David Harrison,Irina Belousova,Mehrdad Farajtabar,Minsik Cho
机构: Apple(苹果)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:Mixture-of-Experts (MoE) models are increasingly deployed alongside Speculative Decoding (SD) to accelerate inference, but combining the two is challenging. SD improves the inference speed of dense models by verifying groups of tokens in parallel. However, the inference speedup for SD with MoEs depends heavily on the number of tokens being verified. Using more verification tokens results in more experts being transferred from DRAM to the Neural Processing Unit (NPU), which increases the memory transfer cost. This negatively impacts model runtime, as memory transfer is typically the bottleneck in inference. In this work, we investigate the impact of MoE router design during training on the speed of MoEs with SD. We find that routers with high degrees of expert coactivation result in much faster runtimes, mitigating the impact of using more verification tokens. Motivated by this observation, we assess the impact of various router design choices on expert coactivation and runtime using billion-parameter transformer models. We find that combining a global load-balancing loss, shared experts, a consistency loss, and an autoregressive expert selection mechanism during training results in significantly stronger expert coactivation. This increased coactivation translates into higher overall runtime throughput: our exploration yields a model that improves throughput by 21% over MoE baselines, while maintaining on-par accuracy with the baseline MoE.
[NLP-125] oward Personalized Sleep Guidance from Wearable Data Using Language Models
【速读】： 该论文旨在解决可穿戴设备数据在个性化睡眠指导中的应用难题，尤其针对当前基于大语言模型（LLM）的摘要生成与问答系统在个性化睡眠建议方面表现不足的问题。其核心挑战在于：训练专用模型通常依赖昂贵的专家标注数据，且存在隐私泄露与访问门槛高等问题，限制了模型在终端用户中的轻量级本地部署。为此，论文提出一种两阶段框架：第一阶段采用多智能体大语言模型（multi-agent LLM）流水线，从无标注的可穿戴设备记录中推理生成结构化睡眠指导，实现大规模高质量数据集的自动化构建；第二阶段通过监督微调将指导推理轨迹蒸馏至小型语言模型（SLMs），并引入无需训练的“Best-of-N”选择策略以提升推理性能。该方案既避免了对人工标注的依赖，又实现了低资源、高隐私保护的本地化部署，实验结果表明其性能优于主流商业通用及医疗领域大模型和开源模型，人机评估也验证了生成指导的质量与个性化睡眠建议的可行性。
链接: https://arxiv.org/abs/2609.22463
作者: Yusheng Tan,Running Zhao,Sofia Angel,Ninghui Hao,Ash Arian,Nikita N. Dulin,Jay Lin,Ou Zhu,Faiza Shaik,Xinxing Yang,Bonnie W. Leung,Katie Roster,Arlene Ruiz de Luzuriaga,Kenneth Lee,Alejandra Lastra,Habibul Ahsan,Guihong Wan
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Sleep monitoring using wearable data has shown promise for personal health, yet large language model (LLM)-based summarization and question answering remain insufficient for personalized sleep guidance. Training specialized models, however, often requires costly expert annotation. Moreover, privacy and accessibility concerns motivate lightweight, local deployment for end users. We present a two-stage framework to address these challenges. Specifically, in Stage~1, a multi-agent LLM pipeline reasons structured sleep guidance from unannotated wearable records, enabling scalable dataset construction. Stage~2 distills guidance reasoning trajectories into small language models (SLMs) through supervised fine-tuning and integrates a training-free Best-of- N selection strategy to enhance inference. Experimental results demonstrate our method outperforms commercial general and medical LLMs and open-source models. Human evaluation further supports the quality of the generated guidance and the feasibility of personalized sleep guidance with SLMs.
[NLP-126] Apollo Restore: A Foundation LLM for Historical Greek Optimized for Fill-in-the-Middle Restoration of Ancient Greek Texts
【速读】： 该论文旨在解决古希腊语残缺文本中缺失段落（lacunae，即物理性空白）的自动修复问题，尤其针对文献碎片中因历史损毁导致的文本断裂。其核心挑战在于：在未知缺失部分长度的前提下，准确重建丢失内容，且需兼顾历史语言的语法、语义与文体特征。解决方案的关键在于提出并训练了Apollo Restore——一个240亿参数的大型解码器模型，基于Mistral Small进行微调，并采用“填空中间”（fill-in-the-middle）目标函数，使模型能够不依赖于真实缺失长度信息即可完成恢复。该方法不仅在短距离缺失（≤10字符）上显著超越现有最优模型，在引入更公平的长度平衡评估指标后，优势进一步扩大至2.3×~3.5×，且对错误的长度提示具有鲁棒性。此外，盲测结果显示，20位资深考古学家、铭文学家和古典学者普遍更青睐该模型输出，其表现被评价为不低于人类专家水平的案例达77%。同时，该模型成功修正了编号为1667的维苏威火山喷发时烧毁的纸草卷的已有解读，验证了其实际应用价值。Apollo Restore是奥地利科学院主导的“解码古代”（Decoding Antiquity）计划的重要成果，标志着面向历史语言与手稿的专用大语言模型（LLM）研究的新进展。
链接: https://arxiv.org/abs/2609.22455
作者: Hope McGovern,Anna Dolganov,Samuel Belkadi,Guillaume Kunsch,Dimitris Vlitas,David A. Smith
机构: Mistral AI(迷雾人工智能); Reply; Northeastern University (东北大学); Austrian Academy of Sciences(奥地利科学院)
类目: Computation and Language (cs.CL)
备注: 16 pages, 6 figures
Abstract:We present Apollo Restore, a 24-billion-parameter large language model for restoring lacunae—physical gaps—in fragmentary Ancient Greek texts. Fine-tuned from Mistral Small with a fill-in-the-middle objective, Apollo Restore reconstructs missing spans without requiring oracle knowledge of their length. To our knowledge, it is the first large-scale decoder model for historical Greek, and the first for any ancient Mediterranean language. Evaluated as in prior work, on short gaps of up to ten characters, Apollo Restore places the correct restoration among its top twenty candidates for 80.6%/54.6%/61.0% of documentary-papyrus, literary-papyrus, and stone-inscription lacunae, exceeding the strongest published models by 1.6\times / 2.6\times / 1.4\times . Prior evaluation protocols, however, inflate scores through a bias toward trivially short gaps; under a length-balanced metric Apollo Restore’s advantage over the strongest published models grows to 2.3\times / 3.5\times / 1.6\times and degrades gracefully, even given incorrect length hints. In a blind study, 20 expert papyrologists, epigraphists, and philologists strongly preferred Apollo Restore to the strongest baseline and judged its performance at least as good as human restorations in 77% of cases. Apollo Restore also improves the published reading of this http URL. 1667—a papyrus roll carbonised in the eruption of Vesuvius in 79 CE and digitally unrolled and edited after Apollo Restore’s training data was compiled. Apollo Restore is an output of the Decoding Antiquity initiative to build specialized LLMs for historical languages and manuscripts, led by the Austrian Academy of Sciences.
[NLP-127] Contextual Causality with Large Language Models : A Survey EMNLP2026
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在理解上下文因果关系（contextual causality）方面存在的系统性研究缺失问题。当前，尽管因果推理对提升模型决策可靠性至关重要，但针对LLMs如何在具体语境中准确识别因果关系的研究仍不充分。为此，论文提出了一种涵盖语义因果、干预因果与反事实因果三类的上下文因果分类体系（taxonomy of contextual causality），并分别从核心因果问题、模型所需能力、代表性任务及实际应用四个维度对其进行了系统刻画。其解决方案的关键在于构建一个结构化的分析框架，不仅厘清了现有研究的局限性，还揭示了当前评估基准与真实应用场景之间的差距，并指明了未来研究的潜在方向。该工作旨在全面梳理上下文因果与LLMs交叉领域的研究图景，凸显其重要性，并为后续发展提供理论指引。
链接: https://arxiv.org/abs/2609.22409
作者: Yiheng Zhao,Jun Yan,Chengming Hu
机构: Concordia University (康科迪亚大学); McGill University (麦吉尔大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted to the 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP 2026)
Abstract:Understanding contextual causality is critical for large language models (LLMs), as it enables them to accurately identify causal relations in specific situations and support more reliable decision-making. Despite its significance, a systematic exploration of contextual causality with LLMs is still lacking. To fill this gap, we present a comprehensive survey on this topic. In this survey, we first propose a taxonomy of contextual causality, consisting of semantic, intervention, and counterfactual causality, and characterize each category by its core causal question, required model capabilities, representative tasks, and practical uses in causality analysis. We then analyze existing studies and discuss their key limitations. Finally, we examine the gaps between current benchmarks and real-world needs and outline promising directions for future research. Our goal is to clarify the research landscape of contextual causality with LLMs, emphasize its importance, and highlight promising future directions.
[NLP-128] Initial Evaluation of Potential Bias in Coverag e of Humans in Wikidata
【速读】： 该论文旨在解决开放协作知识图谱（如Wikidata）在性别、性取向、地理分布、出生地城乡属性、族裔、多语言标签与描述覆盖等维度上存在的代表性不足及系统性偏见问题，尤其关注数据缺失与不均衡现象对人工智能系统公平性的影响。其解决方案的关键在于构建一个开源审计平台，基于QLever高效处理超过1000万条关于逾600万人类实体的语句绑定，并结合卡方拟合优度检验、95% Wilson-score置信区间和差异比等统计方法，在鲁宾缺失性分类框架（Rubin’s missingness taxonomy）指导下，系统量化评估各维度的代表性与数据完整性。研究发现，女性仅占已标注性别的28.71%，西方欧洲与北美（WENA）地区占据近53%的公民身份声明，农村出生地仅占2.48%（远低于全球基准的27.4%），族裔声明不足1.2%，非英语描述覆盖比例仅为18.2%，揭示出族裔与性取向数据缺失最为严重，而农村出生地与非WENA公民身份则存在显著代表性不足，凸显了当前知识图谱在社会多样性表征上的结构性缺陷。
链接: https://arxiv.org/abs/2609.22375
作者: Clair Kronk
机构: 未知
类目: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 11 pages, 3496 words, 1 table, 5 figures
Abstract:Introduction. Open collaborative knowledge graphs such as Wikidata increasingly ground agentic artificial intelligence, information retrieval, and language modeling systems, making systematic auditing of their demographic representation and overall equity a research imperative. Methods. Herein, we present an open-source auditing platform that ingests over 10 million statement bindings representing over 6 million humans on Wikidata via QLever, and evaluates representation of gender, sexual orientation, geography, birthplace urbanicity, ethnicity, multilingual coverage of labels, descriptions, and aliases, occupation, and select intersectional pairs of these entities. It does so by making use of Chi-square goodness-of-fit tests, 95% Wilson-score confidence intervals, and disparity ratios, in light of Rubin’s missingness taxonomy. Results. Women accounted for 28.71% (CI +/-0.04) of all humans in Wikidata with a stated gender. 38.26% of humans had a citizenship statement, with Western Europe and North America (WENA) representing approximately 53% of such statements. Among 1.8 million birthplaces that could be classified, rural birthplaces were observed in 2.48% of cases (in comparison to 27.4% global baseline). Fewer than 1.2% of entities carried an ethnicity statement, and non-English Wikidata descriptions covered 18.2% of items. Discussion. Our findings reveal significant missingness across the evaluated axes. Ethnicity and sexual orientation were the most critically under-documented (missing statements) while rural birthplaces and non-WENA citizenship were the most underrepresented.
[NLP-129] Used Mentioned or Condemned? A Controlled Contrast-Set Diagnostic for the Use-Mention Distinction in Code-Mixed Hinglish Misogyny Detection
【速读】： 该论文旨在解决生成式语言模型在检测性别歧视（misogyny）时无法有效区分侮辱性词汇的“使用”与“提及”这一核心问题，尤其在印地-英语混用语境（Hinglish）中，这种混淆直接导致内容审核机制可能误伤反污言辩论（counter-speech），从而压制正当讨论。其解决方案的关键在于构建一个严格解耦标签与语言特征的诊断基准：通过设计一个确定性生成器和包含416个样本、163对最小差异对比集（minimal pair contrast-set）的Hinglish-MGY-Diag数据集，使侮辱性词汇的存在与其所属的性别化语域（gendered register）在标注上与真实意图完全解耦。进一步提出“严格成对一致性”（pair-consistency）评估指标，仅当同一对比对中的两个样本均被正确分类时才计分，从而暴露模型在语义理解上的根本缺陷。实验表明，即使最强的基线模型在纯净使用-提及子集上准确率可达0.93，但成对一致性仅为0.82，说明仍存在显著误判；而前沿大语言模型（LLM）作为作者模型上限达到1.00，验证了该基准为能力梯度而非对抗性障碍，具备可扩展的评估潜力。
链接: https://arxiv.org/abs/2609.22261
作者: Ashanvi Yadav,Shubham Bhardwaj
机构: Manipal University Jaipur(曼加普尔大学贾伊布尔分校); BITS Pilani, Hyderabad Campus(比茨理工学院海得拉巴校区)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注: Preprint. 8 pages main text + appendix, 6 figures, 7 tables. Code, data, and generator to be released with camera-ready version
Abstract:Lexicon-driven misogyny detectors cannot, by construction, distinguish a slur used against a woman from the same slur mentioned in counter-speech (“don’t call her that”) – yet exactly this distinction governs whether moderation protects or silences the people discussing abuse. We study this problem in code-mixed Hinglish and make three contributions. First, we diagnose two evaluation artifacts on a publicly available redacted corpus: category-encoding anonymization placeholders leak the label (a no-learning rule scores 1.000), and even after they are neutralized misogynistic and benign comments occupy lexically disjoint registers, so bag-of-words reaches macro-F1 approximately 1.00 under random cross-validation but collapses under template-disjoint evaluation. Second, we release Hinglish-MGY-Diag, a deterministic generator and a 416-item / 163-minimal-pair contrast-set diagnostic across five linguistically motivated categories in which slur presence and gendered register are decorrelated from the label by construction. Third, we introduce a strict pair-consistency metric that credits a model only when both members of a minimal pair are correctly labelled. Five from-scratch classical baselines evaluated under construction-disjoint five-fold cross-validation reveal that the strongest model reaches 0.93 accuracy on the cleanest use-mention subset but only 0.82 consistency – it still mislabels roughly one counter-speech pair in five. A frontier LLM used as an author-model ceiling attains 1.000 on all metrics, doubling as independent label validation and confirming the benchmark is a capability gradient rather than an adversarial wall. We release all code, data, the generator, and an arms-length LLM harness for reproducing every number. Comments: Preprint. 8 pages main text + appendix, 6 figures, 7 tables. Code, data, and generator to be released with camera-ready version Subjects: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY) ACMclasses: I.2.7; K.4.1 Cite as: arXiv:2609.22261 [cs.CL] (or arXiv:2609.22261v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2609.22261 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Shubham Bhardwaj [view email] [v1] Sun, 6 Sep 2026 17:45:59 UTC (38 KB)
[NLP-130] Causal Localization of the Refusal Direction in Audio Language Models
【速读】： 该论文旨在解决生成式音频语言模型（LALM）在面对有害语音请求时，其拒绝行为是由前端语音处理模块（speech front end）决定，还是由后端已安全对齐的文本语言模型（LM）继承而来这一关键问题。其解决方案的核心在于采用因果干预（causal interventions）方法，在模型的音频-文本接口及多个残差层上，拟合区分有害与良性提示的方向，并通过消融（ablation）该方向分量来测量模型首词拒绝边际的变化。实验结果表明，在五个不同架构的LALM中，拒绝边际的显著变化主要集中在语言模型的中后期层，而接口处方向的消融影响极小；以Qwen2.5-Omni为例，第16层方向消融导致拒绝边际下降7.10，而投影器处仅为-0.013，且语音编码器输出置零仍使边际下降4.7，说明音频路径仍在发挥作用。此外，拒绝相关方向在早期层可线性解码，且仅基于文本主干训练的方向可迁移至完整音频模型，表明该方向更关联于“拒绝”这一行为而非特定危害性本身。研究强调，拒绝边际的依赖位置并不等同于拒绝决策的计算位置，且调整边际未必改变模型实际输出。因此，该研究建议对这类模型的安全审计应结合因果干预而非仅依赖探测工具，并需同时考察文本语言模型与音频接口的协同作用。
链接: https://arxiv.org/abs/2609.22260
作者: Leonardo Haw-Yang Foo,Hung-yi Lee
机构: 未知
类目: Multimedia (cs.MM); Computation and Language (cs.CL); Machine Learning (cs.LG); Sound (cs.SD); Audio and Speech Processing (eess.AS)
备注: Accepted at ISCSLP 2026. 4 pages + references
Abstract:A large audio language model (LALM) attaches a speech front end to a text language model (LM) that is already safety-aligned. When such a model refuses a harmful spoken request, is the refusal carried by the front end, or inherited from the text LM? We test this with causal interventions. At each model’s audio-to-LM interface and at tested LM residual layers, we fit a direction separating harmful from benign prompts, ablate its component, and measure the resulting change in the model’s first-token refusal margin. Four of the five models are evaluated under held-out category shift. Across five LALMs spanning three backbone families, with three models passing a baseline safety gate, the largest tested effects occur in a mid-to-late LM band, while ablations of the tested interface directions have little effect. On Qwen2.5-Omni, ablating the L16 direction changes the margin by -7.10, versus -0.013 at the projector. The audio pathway is still in use: zeroing the encoder output changes the margin by -4.7. In the same model, the contrast is linearly decodable at an early layer where single-layer ablation has little effect, and a direction fitted on the LM backbone alone transfers to the full audio model. Because harmful and benign prompts also differ in form, we interpret the direction as refusal-linked rather than harmfulness-specific. These interventions localize dependence of the refusal margin, not where refusal is computed. Moving the margin also does not always change what the model writes. Safety audits of these models should use interventions rather than rely on probes alone, and should examine the inherited text LM alongside the audio interface.
[NLP-131] Which Part of the Context Layer Does the Work? Separating Semantic Content from Retrieval Scaffolding in Text-to-SQL Agents
【速读】： 该论文旨在解决生成式文本到SQL（text-to-SQL）任务中上下文层（context layers）的性能贡献来源不明的问题，即难以区分是语义内容、检索支撑结构（retrieval scaffolding），还是预计算视图（pre-computed views）在实际推理中发挥了主导作用。其解决方案的关键在于设计并验证一种基于“数据契约”（data contract）的四臂消融实验框架：该契约以YAML格式定义领域语义及代理工具的约束规则，通过固定工具接口、检索指令、表白名单与操作规则，仅清空契约中的自然语言描述字段，从而隔离语义内容的影响。实验表明，语义内容是决定性因素——在176个任务上，仅保留语义内容即可将高难度任务准确率从13.9%~37.0%提升至55.1%~77.4%，显著优于直接将知识拼接至提示词（prompt）的方法；而缺乏语义内容的支撑结构仅带来0~15分增益，且在前沿模型上表现更差。此外，契约通过将语义信息编码于可执行的SQL表达式中，实现高达98%的语义传递成功率，远超提示词方法的4%，且单位正确答案成本更低。研究为实践者提供了明确指导：应优先构建清晰的领域语义，其次考虑检索结构，仅在代理无法自主推导时引入预计算宏。同时，受控的契约机制有效避免了未受约束的生成导致的166条非法语句，而受控版本则无一违规。最终，该方法的性能增益被证明局限于其定义的领域范围，在特定模型上，其准确率可达51.9%，远高于提示词基线的18.8%。
链接: https://arxiv.org/abs/2609.22259
作者: Qing Ye
机构: 独立研究者(Independent Researcher)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Databases (cs.DB)
备注:
Abstract:Context layers, curated documentation that an analytics agent fetches at query time, produce large accuracy gains on text-to-SQL benchmarks. A with/without comparison cannot say which part of the layer does the work: the semantic content, the retrieval scaffolding that delivers it, or the pre-computed views that usually accompany it. We report a four-arm ablation on DABStep on four models that separates the three. The instrument is a data contract: a YAML artifact that carries a domain’s semantics and the rules an agent’s tools enforce. One arm empties every field of prose in the frozen contract while holding the tool surface, retrieval instruction, table allow-list and operation rules byte-for-byte fixed. Compiling the contract’s own SQL expressions into views gives the ceiling a pre-computed layer would reach: gold on all 176 tasks it covers. Content dominates. It raises hard-task accuracy from 13.9% to 55.1%, 22.6% to 56.6%, 22.9% to 68.4% and 37.0% to 77.4%, beating the same knowledge pasted into the prompt on every model. Scaffolding without content is worth 0 to 5 points on two flash models and 14 to 15 on two frontier models. Against the compiled ceiling the contract arm’s shortfall is a failure to derive, and it falls from 39 points to 5 with model capability. The contract beats the prompt because the rule it needs is one lookup away rather than buried in a long prompt: its SQL carries the fee semantics up to 98% of the time against the prompt arm’s 4%, and at the lowest cost per correct answer on three of four models. For practitioners: semantics first, scaffolding second, pre-computed macros only where an agent demonstrably fails to derive. Ungoverned arms submitted 166 mutating statements; governed arms none. The gain is confined to the contract’s domain. On one model the benchmark’s own withheld golds grade the contract arm at 51.9% against 18.8% for the prompt baseline.
[NLP-132] Strategy Accumulation and Guided Execution for Automated LLM Fine-Tuning
【速读】： 该论文旨在解决自动化微调（automated fine-tuning）系统在面对新任务时需从“冷启动”状态重新进行耗时耗力的策略搜索问题。现有系统为无状态设计，每次搜索结束后会丢失已发现的训练策略、数据集洞察及超参数成果，导致重复劳动。为此，本文提出一种两阶段框架——策略累积与引导执行（Strategy Accumulation and Guided Execution, SAGE），其核心在于实现自动化微调过程中的经验累积与复用。第一阶段通过多智能体流水线结合蒙特卡洛树搜索（Monte Carlo Tree Search）进行探索，并由并行的蒸馏代理（Distillation Agent）提取特定任务的探索记录与跨任务的置信度评分洞察，构建结构化经验库；第二阶段则基于该经验库检索相关知识，指导新任务的训练策略选择。实验在九个未见任务上验证了SAGE的有效性，在单轮执行中，相较于基线模型平均相对提升由3.2%提升至15.6%，较无经验积累的同构管道提升12.4个百分点，证明持续积累的策略经验能显著提升自动化微调对未知任务的适应能力。
链接: https://arxiv.org/abs/2609.22257
作者: Haoran Zhao,Wei Du,Dingwen Yang,Jixuan Huang,Junlin Shang,Lingyong Fang,Ya Guo,Tao Gui,Qi Zhang,Xuanjing Huang
机构: Fudan University (复旦大学); Ant Group (蚂蚁集团); Shanghai Jiaotong University (上海交通大学)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 24 pages, 4 figures, 10 tables, including supplementary material
Abstract:Producing task-specific large language models requires discovering effective training strategies through experimentation. Automated fine-tuning systems have made this experimentation feasible with far less manual effort. However, these systems are stateless: each search discards its discovered strategies, dataset insights, and hyperparameter findings once it ends. Every new task must then repeat this costly search from a cold start. To address this, we propose Strategy Accumulation and Guided Execution (SAGE), a two-stage framework that makes automated fine-tuning search cumulative. In the first stage, a multi-agent pipeline performs Monte Carlo Tree Search-based exploration. A parallel Distillation Agent extracts task-specific exploration records and confidence-scored cross-task insights, which together constitute a structured experience repository. In the second stage, SAGE retrieves relevant experience from this repository and selects what applies to guide training on the new task. We evaluate SAGE on nine unseen tasks spanning both single- and cross-category settings. In single-round execution, SAGE’s accumulated experience raises the average relative improvement over baseline from 3.2% to 15.6%, a 12.4-percentage-point gain over the same pipeline without it. These results show that persistent strategy experience provides effective guidance for automated fine-tuning on unseen tasks.
[NLP-133] DIPLOMAT: Dialogue-Span-Aware Direct Preference Optimization for Polite Persuasive Workplace Negotiation Dialogues
【速读】： 该论文旨在解决职场协商中因误解、偏好不一致及人际摩擦导致的沟通障碍问题，核心挑战在于如何在实现任务目标的同时，维持专业关系并有效化解冲突。其解决方案的关键在于提出DIPLOMAT——一个专注于礼貌性与说服力的对话系统，通过引入PROWESS数据集（基于多智能体框架生成的多轮职场协商对话数据集，标注了协商策略、礼貌水平与说服策略）进行训练，并采用一种新颖的偏好学习方法——对话跨度感知的直接偏好优化（Dialogue-Span-Aware Direct Preference Optimization, DSA-DPO），精准识别对话中影响偏好对齐的关键语段。该方法使DIPLOMAT能够生成上下文连贯、兼具礼貌性与说服力的响应，从而在复杂协商场景中持续输出高质量的交互内容。自动评估与人工评估结果均表明，DIPLOMAT在生成连贯、礼貌且具有说服力的协商回应方面显著优于现有基线模型。
链接: https://arxiv.org/abs/2609.22256
作者: Bibhuti Jha,Rishikant Chigrupaatii,Priyanshu Priya,Asif Ekbal
机构: Indian Institute of Technology Patna (印度理工学院巴特那分校); GREYC Laboratory, University of Caen Normandy (卡昂诺曼底大学格林实验室)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Effective workplace negotiation requires balancing multiple objectives, including achieving task goals, preserving professional relationships, and resolving conflicts constructively. However, misunderstandings, misaligned preferences, and interpersonal friction often impede successful outcomes. Politeness mitigates these challenges by fostering trust, reducing tension, and preventing escalation, and persuasive communication helps overcome resistance, align preferences, and guide participants toward mutually beneficial agreements. Motivated by these insights, we present DIPLOMAT, a dialogue system for polite and persuasive workplace negotiation. To support its development, we introduce PROWESS, a dataset of multi-turn workplace negotiation dialogues generated via a multi-agent framework and enriched withnegotiation strategies, politeness levels, persuasive strategies. DIPLOMAT is trained using Dialogue-Span-Aware Direct Preference Optimization (DSA-DPO), a novel preference learning objective that identifies key dialogue spans for preference alignment. This enables DIPLOMAT to generate contextually coherent responses that employ intended negotiation strategies, maintain politeness, and incorporate effective persuasion strategies throughout interactions. Automatic and human evaluation on PROWESS confirm that DIPLOMAT consistently outperforms baselines in generating coherent, polite, and persuasive negotiation responses.
[NLP-134] Deep Persona: A Psychologically Grounded Architecture and Evaluation Framework for Role-Playing Agents and Simulations
【速读】： 该论文旨在解决现有基于大语言模型（LLM）的人物模拟方法依赖浅层角色描述，难以在长时间交互中维持角色行为一致性的问题。其解决方案的关键在于提出一种心理基础驱动的三层架构——深度人格（Deep Persona），将人物设定系统性地组织为可观察表现层、潜在信念层与核心动机驱动力层，形成层次化的人格结构。该架构遵循脚本决定论（scripted determinism）与有限自主性（bounded agency）原则，使模型作为受结构化内部脚本引导的反应式引擎运行，从而增强角色行为的一致性与可信度。此外，研究还设计了一种无参考评估框架，通过心理学临床工具和对抗性压力测试，以真实人类对话分布为基准评估对话自然度。实证结果表明，尽管大模型具备高度语用流畅性，但在情感表达与共同注意方面存在系统性缺陷；而通过该框架对两个深度人格案例的评估显示，结构化人格能够生成更贴近人类对话行为的交互。
链接: https://arxiv.org/abs/2609.22255
作者: Rotem Dror,Zohar Elyoseph,Yuval Haber,Elad Refoua,Oshrat Ayalon,Adir Solomon
机构: University of Haifa(海法大学); Bar-Ilan University(巴伊兰大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Existing approaches to persona simulation with Large Language Models (LLMs) mostly rely on shallow character descriptions that fail to sustain coherent character behavior across extended interactions. We introduce Deep Persona, a psychologically grounded, three-layered architecture that organizes personas into hierarchical levels of observable expression, latent beliefs, and core motivational drives, for constructing highly convincing role-playing agents. Governed by the principles of scripted determinism and bounded agency, the architecture restricts the model to a reactive engine guided by a structured internal script. We further propose a reference-free evaluation framework that benchmarks dialogue naturalness against empirical human distributions using established psychological clinical instruments and adversarial stress-tests. Empirical evaluation reveals that while LLMs achieve high pragmatic fluency, they exhibit systematic limitations in emotional expression and joint attention. In addition, we present a case study of two Deep Personas and evaluate them using the proposed framework, demonstrating that structured personas can produce interactions that more closely align with human conversational behavior.
[NLP-135] CAMFT: Conflict-Aware Mergeable Fine-Tuning for Large Language Models
【速读】： 该论文旨在解决现有模型融合（model merging）方法在多任务场景下因依赖独立微调后进行事后处理而导致的跨任务兼容性不足问题。传统方法在模型微调完成后才考虑参数冲突，存在优化次优的问题。其解决方案的关键在于提出一种冲突感知的可融合微调方法（Conflict-Aware Mergeable Fine-Tuning, CAMFT），将模型的可融合性（mergeability）作为微调过程中的主动设计属性，而非事后修复问题。CAMFT通过引导各任务仅在跨任务冲突较低的稀疏参数坐标上进行更新，使微调过程本身生成更利于后续融合的参数增量，从而实现高效且具备良好融合潜力的多任务适应。实验表明，该方法在多任务融合场景中显著优于标准微调基线。
链接: https://arxiv.org/abs/2609.22253
作者: Jingang Zhou,Haiyang Guo,Yuan Ma,Han Zhu,Xu-Yao Zhang
机构: University of Chinese Academy of Sciences (中国科学院大学); Institute of Automation, Chinese Academy of Sciences (自动化研究所)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Model merging has emerged as a promising paradigm for integrating multiple task-specific capabilities into a single large language model. However, existing methods predominantly focus on post-hoc processing of independently fine-tuned models, overlooking how the training phase itself impacts cross-task compatibility. Resolving parameter conflicts after fine-tuning is inherently sub-optimal. To address this, we propose CAMFT, a Conflict-Aware Mergeable Fine-Tuning method that makes task adaptation both efficient and mergeaware. CAMFT treats mergeability as a property shaped during fine-tuning, rather than only a problem to be solved after fine-tuning. By guiding each task to update sparse coordinates with lower cross-task conflict, CAMFT produces task updates that are efficient to train and more compatible for downstream model merging. Extensive experiments demonstrate that CAMFT outperforms standard finetuning baselines in multi-task merging scenarios. Codes are available at this https URL.
[NLP-136] he Corroboration Illusion: When More News Makes LLM Forecasts Less True
【速读】： 该论文旨在解决生成式人工智能（Generative AI）在基于检索增强生成（Retrieval-Augmented Generation, RAG）的事件预测系统中，因依赖开放可爬取新闻语料库而引入的新安全威胁——即新闻语料库投毒攻击（news-corpus poisoning）。传统RAG攻击主要针对事实性答案或观点极性，而本研究聚焦于对概率性预测结果的扰动，攻击者仅需发布少量伪造文章即可系统性地改变模型输出的概率分布，进而影响预测准确性。其核心解决方案在于揭示此类攻击的可行性与有效性：通过在Common Crawl News语料库中注入仅一篇由大语言模型撰写的虚假新闻，即可使56%的预测结果越过0.5概率阈值；当注入五篇文章时，该比例上升至69%-73%，且概率平均偏移+0.13至+0.22，导致Brier评分从0.18恶化至0.37。攻击效果随注入文章数量、检索排名、查询相似度及上下文重叠度单调增强，并具有跨模型家族迁移性，且不受伪造发布者身份的影响。尽管评估了源地址白名单、隔离-聚合预测和困惑度过滤三种防御机制，但均发现存在低成本绕过方式，表明当前概率预测系统对信息供应链的脆弱性具有根本性风险。
链接: https://arxiv.org/abs/2609.22246
作者: Yuan Lu,Yukuan Zhang
机构: Peking University (北京大学); University of Central Florida (中佛罗里达大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Key words: LLM forecasting, retrieval-augmented generation, data poisoning, misinformation, calibration
Abstract:Large language models (LLMs) are increasingly used to forecast real-world events by retrieving and reasoning over news. We show that this dependence on an open, crawlable news corpus creates a new attack surface: an adversary who can merely publish articles–without access to the retriever, the model, or the user’s queries–can systematically move the forecaster’s output probabilities. We formalize news-corpus poisoning of probabilistic forecasters, a threat model distinct from prior RAG poisoning, which targets factual answers or opinion polarity rather than calibrated probabilities. We evaluate the attack on 500 resolved ForecastBench questions against a 17.4M-article Common Crawl News corpus with a strict crawl-date cutoff, using three retrieval-augmented forecasters built on open 7-8B models. A single LLM-written article per question flips 56% of forecasts across the 0.5 boundary; five articles flip 69-73% and shift probabilities by +0.13 to +0.22 net of a neutral-article placebo, degrading the Brier score from 0.18 to 0.37. The effect is monotone in the number, retrieval rank, query similarity, and context share of injected articles, transfers across model families, and is unaffected by the claimed publisher. We then evaluate three natural defenses–source allow-lists, isolate-then-aggregate forecasting, and perplexity filtering–and show that each has a cheap bypass: spoofed publishers, majority poisoning, and higher-temperature generation, respectively. Our results indicate that probabilistic LLM judgments inherit the full fragility of the information supply chain they consume.
[NLP-137] Do Chess Explanations Reflect Model Decisions? Behavioral and Token-Level Tests of LLM Reasoning Faithfulness EMNLP2026
【速读】： 该论文旨在解决生成式 AI 在国际象棋决策中所生成解释的可信性问题，即语言流畅性是否真正反映其背后的推理逻辑。尽管大型语言模型能够生成语法正确且语义连贯的走法解释，但这些解释的表面合理性并不等同于真实、可靠的决策依据。研究聚焦于可完全观测的残局局面，利用200个Lichess残局谜题，通过移动可恢复性（move recoverability）、解码端控制以及对合法候选着法的逐标记评分等方法评估解释的有效性。关键发现在于：未加掩码的解释虽能显著提升模型对特定走法的恢复能力，但在移除明确走法提示后优势急剧下降；而在严格掩码条件下，解释仅带来微弱且依赖解码器的性能增益，远低于仅依赖棋盘状态的基准表现。此外，逐标记评分结果显示，无关但看似合理的解释内容仍会显著降低正确走法的概率，表明模型并非简单忽略无效推理文本。研究还发现，可识别的残局模式虽有助于增强走法的可恢复性，却不能稳定提升走法准确性。综上，该研究揭示了语言流畅性、与生成动作的一致性以及解决方案正确性三者之间的本质差异——流畅的解释可能影响行为偏好并构建连贯叙事，但并不能提供忠实推理的可靠证据。
链接: https://arxiv.org/abs/2609.22245
作者: Angelina Parfenova
机构: Lucerne University of Applied Sciences and Arts(卢塞恩应用科学与艺术大学); Technical University of Munich(慕尼黑工业大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted to BlackboxNLP at EMNLP 2026
Abstract:Large language models can produce fluent explanations for chess moves, but plausible language does not necessarily reflect the reasoning behind a decision. We study this question in chess, where the board state is fully observable, legal actions can be enumerated, and move quality can be evaluated independently. Across 200 Lichess endgame puzzles, we test explanations using move recoverability, decoder-side controls, and token-level scoring of legal candidate moves. Unmasked explanations make generated moves easy to recover, but this advantage drops sharply after explicit move hints are removed. Under strict masking, explanations provide only small and decoder-dependent gains over the board state alone. Token-level scoring shows that explanations can nevertheless alter move preferences: random but plausible explanations from other puzzles reduce the probability of the correct move, indicating that irrelevant reasoning text is not simply ignored. We also find that recognizable endgame motifs can make generated moves easier to recover without reliably improving move correctness. Together, these results show that linguistic plausibility, consistency with a generated action, and solution correctness are distinct properties. Fluent chess explanations can influence action preferences and support a coherent move narrative while providing only limited evidence of faithful reasoning.
[NLP-138] Replay-Gated Neural Execution: Decoupling Persistent Behavioral Specifications from Neural Realizations in Frozen Language Models
【速读】： 该论文旨在解决输入条件驱动的神经干预（neural interventions）在运行时面临的核心问题：当某一行为规范存在多个可能的动作选项，且这些动作的有效性依赖于执行状态时，如何确保行为规范的持续性与执行决策的可靠性。其解决方案的关键在于提出“回放门控神经执行”（replay-gated neural execution）机制，通过将系统解耦为五个独立对象实现精确控制：一个持久的行为谓词（behavioral predicate）、其基于状态索引的已认证动作集合、瞬态动作见证（action witness）、预算受限的搜索器（finder）以及执行授权模块。该框架通过冻结模型的隔离式FP32/BF16回放验证候选动作，并要求有效的运行审计作为提交承诺的附加条件。实验在Qwen3-0.6B和SmolLM2-360M-Instruct上验证了各组件的失效模式，结果显示状态索引化的动作集合可保持一致性，而部分动作见证在跨状态转移中出现失效；尽管所有归档见证均能成功回放，但预算受限的搜索器仍存在遗漏。最终，采用三级冻结级联结构对未认证提案进行过滤并升级审计缺失的搜索结果，在256个先前密封的请求中实现了100%的审计授权且无绕过现象，显著降低平均搜索与认证成本（中位数比值为0.1055，P95为1.3485）。研究证明，在小规模模型上的行为家族中，基于状态索引的集合化执行语义是可行且高效的：行为规范持久存在，搜索负责生成候选见证，而回放认证与运行审计共同赋予执行权限。
链接: https://arxiv.org/abs/2609.22243
作者: Xianliang Zeng,Zhanzhan Zhao
机构: The Chinese University of Hong Kong, Shenzhen(香港中文大学（深圳）); Shenzhen, China(深圳市)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Input-conditioned neural interventions raise a runtime question: what persists when one behavioral specification admits multiple actions whose validity depends on execution state? We introduce replay-gated neural execution, separating five objects: a persistent behavioral predicate, its state-indexed certified realization set, a transient action witness, a budget-limited finder, and execution authorization. Candidates undergo isolated FP32/BF16 replay of the frozen model; commitment additionally requires a valid run audit. Experiments on Qwen3-0.6B and SmolLM2-360M-Instruct establish distinct failure modes for these objects. Independent initializations yield distinct certified actions in all 24 tested fixed-state cells. Unchanged SmolLM2 witnesses remain certified in all 128 native states but only 66 of 384 off-diagonal transfers. All 767 archived Qwen witnesses replay successfully, yet a budget-limited finder misses one known-realizable cell in all three prespecified runs. Of 1,141 replay-submitted candidates, 174 fail item certification. A frozen three-tier cascade uses these boundaries to reject uncertified proposals and escalate audit-valid search misses. On 256 previously sealed Qwen Fresh requests, 221 first certify at the lowest-cost tier and all 256 receive audited authorization, with no observed bypass. Relative to frozen full search, the median singleton search-and-certification cost ratio is 0.1055 and P95 is 1.3485, including failed tiers. Within the studied behavioral family on two small models, these results support state-indexed, set-valued execution semantics: specifications persist, search proposes witnesses, and replay certification plus run audit grants execution authority.
[NLP-139] H2LooP Telecom Model v1: From Telecom Comprehension to Autonomous Issue and PR Resolution
【速读】： 该论文旨在解决电信领域专用大模型在专业任务中性能不足的问题，特别是在电信知识问答、推理以及生产级代码生成等场景下的能力短板。其核心挑战在于如何在保持通用能力的前提下，实现对电信行业特定知识（如3GPP标准、O-RAN规范、网络遥测数据及真实代码仓库提交记录）的深度理解与高效应用。解决方案的关键在于构建一个经过领域专项微调的大型语言模型——H2LooP Telecom Model v1，通过在高质量、多源的电信领域语料上进行训练，实现了对电信专业知识的精准建模。该模型推出两个互补变体：专注于理解与推理的Comprehension变体，在GSMA Open Telecom Lite (OT-Lite)基准上达到81.8%的Pass@3加权平均得分，并在独立运行的Open Telco AI Leaderboard中以仅310亿参数规模超越包括GPT-5和Claude Opus在内的前沿闭源模型；另一为代理型（agentic）变体，显著提升电信代码生成中的抽象语法树（AST）相似度（+8.8%）与位置交并比（Location IoU，+20.0%），同时在MMLU（74.0%）和BFCL v3多轮函数调用任务（79.0%）上保持与基线模型一致的表现，证明了零灾难性遗忘。这一成果表明，基于精心筛选的电信领域语料进行领域专业化训练，可在不牺牲通用能力的前提下，使模型在特定任务上达到甚至超越顶尖闭源模型水平。
链接: https://arxiv.org/abs/2609.22241
作者: Amit Singh,Vedant Nipane,Mayank Goel,Pulkit Agrawal,Sairanjan Mishra
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:We present H2LooP Telecom Model v1, a domain-specialized large language models fine-tuned for the telecommunications industry. We release two domain-adapted model variants serving complementary use cases: a comprehension-focused variant for telecom domain question answering and reasoning, and an agentic variant for autonomous telecom code generation, pull request resolution, and code commits on production repositories. H2LooP Telecom achieves strong results on the GSMA Open Telecom Lite (OT-Lite) benchmark and a proprietary telecom code generation benchmark, outperforming frontier closed-source models such as GPT-5 and Claude Opus on independent leaderboard evaluation, while preserving general-purpose capabilities. The Comprehension variant achieves 81.8% weighted average on OT-Lite Pass@3, and, independently, ranks 5th overall on the official community-run Open Telco AI Leaderboard* at only 31B parameters-ahead of frontier closed-source systems including Claude Opus 4.6, GPT-5, Gemini 3 Flash, Grok-4-fast, and Kimi K2.5. Our agentic variant obtains a relative improvement of +8.8% in AST Similarity and +20.0% in Location IoU over the base model on telecom code generation, while maintaining identical MMLU (74.0%) and BFCL v3 multi-turn function calling (79.0%) performance, indicating zero catastrophic forgetting. Domain specialization on curated telecom corpora, spanning 3GPP standards, O-RAN specifications, network telemetry, and real repository commits, yields substantial improvements over general-purpose models of equivalent scale, approaches frontier closed-source models on domain-specific evaluation, and is independently corroborated by our official leaderboard standing.
[NLP-140] Knowledge Graph-Augmented Ambient AI for Clinical Note Generation
【速读】： 该论文旨在解决生成式人工智能（Generative AI）在医疗场景中自动生成临床记录时存在的信息遗漏问题，即生成的病历可能遗漏患者-医生对话中讨论的关键临床信息，从而导致诊疗决策中的信息缺口。其解决方案的关键在于提出一种模型无关的框架——覆盖率导向修订（Coverage-Directed Revision, CDR），该框架通过从就诊对话转录文本构建知识图谱（Knowledge Graph, KG），系统识别初始生成病历中缺失的重要医学概念，并引导大型语言模型（Large Language Models, LLMs）精准补全缺失内容，而无需修改底层的病历生成系统。实验在两个数据集（Pitt-Bench和ACI-Bench）上验证了CDR的有效性，结果显示其在所有测试条件下均显著提升了内容召回率，为提升生成式AI临床文档完整性和可靠性提供了一种可落地的技术路径。
链接: https://arxiv.org/abs/2609.22239
作者: Jakir Hossain,Yi-Fei Zhao,Hongjian Wang,Minmei Shih,Katie Leigh Mullen,Ahmad P. Tafti,Leming Zhou,Manoj Purohit,William Hogan,Jay Zeng,Elizabeth Skidmore,Yanshan Wang
机构: 1. University of California, San Francisco (加州大学旧金山分校); 2. Stanford University (斯坦福大学); 3. University of Southern California (南加州大学); 4. University of Washington (华盛顿大学); 5. Harvard Medical School (哈佛医学院); 6. University of California, Berkeley (加州大学伯克利分校)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:Ambient AI is increasingly adopted in healthcare to automatically generate clinical notes from patient-clinician conversations, with the potential to substantially reduce clinician documentation burden. However, generated notes may omit clinically relevant information discussed during the encounter, creating information gaps that can affect downstream care. Knowledge graphs (KGs) constructed from encounter transcripts can provide a structured representation of what was discussed and enable systematic identification of missing information from generated notes that are critical for patient care. In this study, we introduce Coverage-Directed Revision (CDR), a model-agnostic framework that constructs a KG from the encounter transcript, identifies medical concepts absent from an initially generated note, and directs large language models (LLMs) to restore the missing information without modifying the underlying note-generation system. We evaluate CDR on two datasets: 1) Pitt-Bench, a local dataset comprising rehabilitation sessions, and 2) ACI-Bench, a public dataset for benchmarking clinical note generation. We tested four underlying LLMs widely used in ambient AI systems. The results show that CDR consistently improves content recall across all evaluated conditions. Our study provides a practical approach for improving the completeness of ambient AI-generated clinical documentation.
[NLP-141] Seeing Through Conflicts: Improving Instruction Hierarchy Alignment in Vision-Language Models
【速读】： 该论文旨在解决视觉-语言模型（Vision-Language Models, VLMs）在多模态指令冲突情境下，如何有效对齐高层指令（Instruction Hierarchy, IH）的问题。传统指令对齐研究主要聚焦于纯文本场景，而VLM面临新挑战：指令可能嵌入图像中、跨模态分布、经历视觉变换，或在代理式任务中动态出现。为此，作者将多模态指令对齐视为一种推理问题，采用基于规则的奖励机制进行强化学习训练，对比了纯文本、纯图像及混合模态监督的效果。研究发现，仅依赖文本的指令对齐训练在跨模态攻击下表现不佳，尤其当模型需跨模态解码、重构或推理时；基于图像的训练显著提升了鲁棒性；而混合模态训练效果最优。更重要的是，该方法在真实图像和网页代理安全任务中均表现出良好的泛化能力，同时基本保持了模型的通用多模态能力，表明轻量级且可验证的监督策略能够有效提升VLM在对抗性、跨模态及交互式指令冲突下的鲁棒性。
链接: https://arxiv.org/abs/2609.22234
作者: Nicholas Sansoterra,Zishuo Zheng,Sachin Kumar
机构: The Ohio State University (俄亥俄州立大学)
类目: Computation and Language (cs.CL); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Instruction hierarchy (IH) alignment teaches language models to prioritize higher-level instructions when inputs conflict. While studied primarily in text-only settings, vision-language models (VLMs) introduce new challenges for IH: instructions may be embedded in images, split across modalities, visually transformed, or encountered during agentic tasks. Positing multimodal IH alignment as a reasoning problem, we train VLMs using reinforcement learning with rule-based rewards, comparing text-only, image-only, and mixed-modality supervision. We find that text-only IH training partially transfers to multimodal attacks, failing when models must decode, reconstruct, or reason over instructions across modalities. Image-based training improves robustness beyond text-only supervision, while mixed-modality training performs best overall. Importantly, the benefits generalize beyond the synthetic typographic training setting to real-image and web-agent safety tasks, while largely preserving general multimodal capability, showing that lightweight, verifiable supervision can meaningfully improve VLM robustness under adversarial, cross-modal, and interactive instruction conflicts.
[NLP-142] EvalMem: An Operation-Level Diagnostic Framework for Long-Term Memory Systems EMNLP2026
【速读】： 该论文旨在解决大语言模型（LLM）基助手在长时程交互中因记忆系统无法有效维护和更新用户状态、偏好及交互历史而导致的性能瓶颈问题。现有评估方法仅关注端到端问答准确率，难以定位错误来源是编码、检索还是生成环节。为此，论文提出EvalMem——一个操作级诊断框架，包含三个并行的审查模块：编码审查员（Encoding Examiner）用于检测目标事实是否被正确存储，检索审查员（Retrieval Examiner）评估原生检索器能否返回可用证据，生成审查员（Generation Examiner）则测试模型在给定理想证据下的生成能力。三者输出形成细粒度的多标签缺陷代码，实现对记忆系统各层级故障的精准诊断。关键解决方案在于引入“召回优先”策略的代理式检索增强生成（agentic RAG），通过结合查询与源证据进行联合搜索，显著提升现有证据的召回率（在LoCoMo上从70.2%提升至95.6%）。实证评估表明，检索层是主要故障来源，在默认LoCoMo设置下检索缺陷占比达22.1%，远高于编码（7.7%）与生成（6.5%）。基于此诊断结果，构建了由各系统记忆导出数据生成的可搜索辅助结构MemWiki，使LoCoMo和LongMemEval-S上的平均准确率分别提升2.5和2.3个百分点。
链接: https://arxiv.org/abs/2609.22231
作者: Zeyu Liu,Jian Zhong,Rongduo Han,Ziyang Wu,Shunye Tang,Chenghao He,Yaxuan Yang,Yihang Qiu,Ailing Wang,Xiao Liang,Guohuan Xie,Xiaokang Xue,Gongchen Li,Haining Zhang,Wei Wang
机构: Nankai University (南开大学); vivo AI Lab; Tsinghua University (清华大学); Zhejiang University (浙江大学)
类目: Computation and Language (cs.CL)
备注: Accepted by EMNLP 2026 Findings
Abstract:Long-horizon interactions with LLM-based assistants require memory systems that preserve and update user states, preferences, and interaction histories. Existing evaluations report end-to-end QA accuracy and cannot determine whether errors arise from encoding, retrieval, or generation. We introduce EvalMem, an operation-level diagnostic framework with three parallel Examiners. For each query, the Encoding Examiner checks whether the target fact is stored, the Retrieval Examiner assesses whether the native retriever returns usable evidence, and the Generation Examiner tests whether the model can answer from oracle evidence. Their outputs form fine-grained multi-label defect codes. To improve store-level diagnosis, we adapt agentic RAG with a recall-first strategy that searches using both the query and source evidence, increasing recall of present evidence on LoCoMo from 70.2% to 95.6%. Evaluations of seven memory systems on LoCoMo, LongMemEval-S, and dynamic DynaMem-Bench identify retrieval as the most frequently attributed failure layer; in default LoCoMo, retrieval defects reach 22.1%, compared with 7.7% for encoding and 6.5% for generation. Guided by this diagnosis, MemWiki, a search-friendly auxiliary structure built from each system’s memory export, improves mean accuracy by 2.5 and 2.3 percentage points on LoCoMo and LongMemEval-S, respectively.
[NLP-143] Assessing Adversarial Robustness of Latent Reasoning Models
【速读】： 该论文旨在解决生成式模型在复杂推理任务中因采用长链式思维（Chain-of-Thought, CoT）轨迹而导致的高内存占用与推理开销问题，同时关注隐式推理范式——潜在推理模型（Latent Reasoning Models, LRMs）在对抗攻击下的鲁棒性缺陷。其核心解决方案的关键在于通过将中间推理过程压缩为少量连续潜在向量，实现对传统自回归生成方式的效率提升。然而，研究发现尽管LRMs具备显著的计算效率优势，其在面对对抗性扰动时表现出较弱的鲁棒性，尤其在白盒攻击下性能严重退化。进一步分析揭示了跨模态的差异化失效模式：文本类潜在状态呈现脆弱的动力学特性且对特定输入模式高度敏感，而多模态模型中的潜在状态则对输入扰动具有较强不变性，且对最终预测的影响有限。这些发现暴露了当前潜在推理方法在鲁棒性方面的系统性局限，强调在设计隐式推理系统时必须协同优化效率与鲁棒性。
链接: https://arxiv.org/abs/2609.22228
作者: Shaolong Chen,Ang Li,Mingjie Li,Yisen Wang
机构: Yuanpei College, Peking University; State Key Lab of General Artificial Intelligence, School of Intelligence Science and Technology, Peking University; CISPA Helmholtz Center for Information Security
类目: Computation and Language (cs.CL); Machine Learning (cs.LG); Machine Learning (stat.ML)
备注:
Abstract:Large language models increasingly rely on long chain-of-thought (CoT) trajectories for complex reasoning, but autoregressive generation brings substantial memory and inference costs. Latent reasoning models (LRMs) offer a more efficient alternative by compressing intermediate reasoning into a small number of continuous latent vectors. Despite their efficiency, however, the adversarial robustness of LRMs remains largely underexplored. In this work, we systematically evaluate the robustness of latent reasoning across textual and multimodal settings, covering eight models and six benchmarks. We find that, across our evaluated settings, LRMs are generally less robust than explicit CoT baselines under adversarial perturbations, with particularly severe degradation under white-box attacks. Further analysis reveals distinct failure modes across modalities: textual latent states exhibit brittle dynamics and high sensitivity to specific input patterns, while latent states in multimodal models can remain largely invariant to input perturbations and have limited influence on final predictions. These findings expose robustness limitations of current latent reasoning approaches and highlight the need to jointly consider efficiency and robustness when designing implicit reasoning systems. We have open-sourced our code to facilitate reproduction of our research this https URL.
[NLP-144] Swiss-Knife: A Framework for Reconfigurable Externalised Multi-Objective Alignment at Decode Time
【速读】： 该论文旨在解决生成式 AI（Generative AI）在解码时对齐（decode-time alignment）过程中存在的单一化、静态化设计缺陷，即现有方法普遍依赖固定的奖励评分与选择机制，难以灵活适应多目标优化需求。其核心问题是：如何在不重新训练模型的前提下，实现可动态配置、可热切换的多目标对齐框架，以支持多样化且可解释的对齐策略。解决方案的关键在于提出 Swiss-Knife 框架，将对齐规范作为运行时的一等对象（first-class runtime object），包含可插拔的评分刀片（scoring blades）、批标准化模块（Candidate-Batch Normalization, CBN）、成对聚合算子及选择规则。该框架通过六条公理约束聚合操作符，并证明了其形式为 $ R_i = \sum_{j \neq i} g((\mu_i - \mu_j)/s(\sigma_i,\sigma_j)) $，构成一个双参数族，涵盖 probit、logistic 比较规则及逐点 argmax 作为特例。其中，成对聚合在对抗性奖励污染下具有 Lipschitz 稳定性，而 argmax 则不具备；CBN 实现了权重单纯形对奖励模型缩放不变性，显著提升鲁棒性。实验表明，基于 DPO-LoRA 刀片与不确定性感知成对锦标赛的参考实例，在帮助性/诚实性/无害性超平面内实现了六种解码对齐方法中最佳的平衡前沿（调和 F₁ 达 0.797，优于最强基线 0.750，p < 10⁻¹⁴），同时拒绝率最低、帮助性最高，并可在 0.05 毫秒内完成目标重构而无需梯度计算。消融实验验证了 CBN 的关键作用：移除后导致 F₁ 下降 0.217，且元数据证实低方差刀片仅保留约 9% 的名义影响力，符合预测的退化顺序。
链接: https://arxiv.org/abs/2609.22226
作者: Agnibh Karmakar,Mayur Parvatikar,Shreyash Dhoot,Amit Dhanda,Aman Chadha,Kapil Wanaskar,Vinija Jain,Amitava Das
机构: 未知
类目: Computation and Language (cs.CL); Machine Learning (cs.LG); Machine Learning (stat.ML)
备注: 19 pages, 3 figures, includes technical appendix and proofs
Abstract:Decode-time alignment methods steer a frozen language model by scoring candidate continuations with an external reward and selecting the maximiser. We argue that this shared design is a single degenerate point in a much larger space. We introduce Swiss-Knife, a framework for externalised multi-objective alignment in which the alignment specification is a first-class runtime object: hot-swappable scoring blades, a batch normaliser, a pairwise aggregation operator, and a selection rule. Six axioms characterise the admissible aggregation operators, and we prove a representation theorem: every operator satisfying them has the form R_i = \sum_j \neq i g((\mu_i - \mu_j)/s(\sigma_i,\sigma_j)) , a two-parameter family containing probit and logistic comparison rules and pointwise argmax as named coordinates. Within it, pairwise aggregation is Lipschitz-stable under adversarial reward contamination while argmax is not, and Candidate-Batch Normalization (CBN) makes the weight simplex invariant to the rescalings under which reward models are only ever identified. Our reference instantiation pairs DPO-LoRA blades with an uncertainty-aware pairwise tournament. Sweeping the helpfulness/honesty/harmlessness simplex, it attains the best balanced frontier of six decode-time methods (harmonic F_1 0.797 vs. 0.750 for the strongest baseline, p 10^-14 ) with the lowest refusal rate and highest helpfulness of any arm, and reconfigures its objectives in 0.05 ms with no gradient computation. Ablating CBN costs 0.217 F_1 , and the metadata confirms the predicted mechanism: the lowest-variance blade retains 9% of its nominal 33% influence, and the collapse follows the predicted ordering.
[NLP-145] Do LLM s Choose Like Humans? Using Cognitive Theory to Evaluate LLM Decision-Making EMNLP
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在决策行为中表现出的人类相似性是否源于与人类相同的底层认知机制，还是仅停留在表面模仿的问题。其核心问题是：LLMs的上下文敏感性是否可由人类认知经济理论（cognitive economic theory）解释，即通过问题分类与注意力分配来理解人类决策。研究采用一个包含14万次试验的新产品选择基准，评估了12个开源及商业LLMs的表现。结果显示，尽管LLMs在上下文影响下表现出类似人类的选择行为和问题分类变化，但其并未可靠地在价格与质量等特征间重新分配注意力权重；同时，模型规模扩大或引入链式思维（chain-of-thought reasoning）也未能有效减弱上下文敏感性或生成真正的人类决策模式。因此，该研究的关键发现是：LLMs的决策机制与人类存在本质差异，其行为更多依赖于统计模式而非基于认知经济原理的动态注意力调控。
链接: https://arxiv.org/abs/2609.22225
作者: Johnathan Sun,Andrei Shleifer,Yonatan Belinkov
机构: Harvard University (哈佛大学); Kempner Institute, Harvard University (哈佛大学肯普纳研究所); Technion - Israel Institute of Technology (以色列理工学院)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Accepted to EMNLP Findings 2026
Abstract:Large language models (LLMs) exhibit a range of human-like decision-making behaviors, but whether these reflect similar underlying mechanisms or surface-level mimicry remains unclear. We evaluate whether LLM context sensitivity aligns with a cognitive economic theory that explains human behavior through problem categorization and attention allocation. Across 12 open-source and commercial LLMs on a novel 140,000-trial product choice benchmark, context induces human-like shifts in choice and problem categorization, but does not reliably reweight attention between features like price and quality. Neither scale nor chain-of-thought reasoning reliably attenuates context sensitivity or generates human-like behavior. These results suggest that LLM decision mechanisms are distinct from human ones.
[NLP-146] From Trait Vectors to Circuits: Tracing Refusal and Sycophancy Through Language Models
【速读】： 该论文旨在解决生成式 AI（Generative AI）中行为操控（steering）机制的本质问题，即：在激活空间中引导模型产生特定行为的方向，是否真正对应于模型自身生成该行为的计算路径。研究聚焦于 Qwen2.5-7B-Instruct 模型中的两个已验证的行为特质——拒绝性（refusal）与谄媚性（sycophancy），通过将特质向量作为分界点，将模型计算过程划分为重构电路（reconstruction circuit，位于向量之前）与传输电路（transmission circuit，位于向量之后），进而检验恢复该坐标是否能复现因消融（ablation）而丢失的行为信号。研究发现，对于拒绝性，重构与传输电路均紧凑且忠实，仅恢复该坐标即可几乎完全恢复被移除的拒绝信号，且基于该向量构建的电路在约一半边数下即可达到与端到端输入输出电路相当的忠实度；而对于谄媚性，传输电路虽紧凑但重构电路更广泛且部分失真，表明操控路径并非自动等同于原始行为生成路径。因此，研究提出必须对每条电路分别进行反事实测试（counterfactual test）、目标测试（target test）和行为测试（behavioral test），以准确评估其功能本质。解决方案的关键在于通过分解计算路径并系统验证各组件的因果作用，揭示行为操控机制与模型内在生成机制之间的解耦关系。
链接: https://arxiv.org/abs/2609.22224
作者: Oscar Miró López-Feliu,Maya Ozbayoglu
机构: University of Amsterdam (阿姆斯特丹大学); Safe AI Netherlands (安全人工智能荷兰)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:A direction in activation space that changes safety-relevant behavior when steered is not necessarily one the model uses to produce that behavior on its own. We therefore ask whether steering acts through the computation of the unmodified model or through a different set of components, studying two traits whose directions have been extracted and validated in prior work: refusal and sycophancy in Qwen2.5-7B-Instruct. For each, we use the trait vector to split the computation into a reconstruction circuit before the vector and a transmission circuit after it. We then test whether restoring the coordinate returns behavior removed by ablation. For refusal, both circuits are compact and faithful, restoring the coordinate alone recovers almost all of the refusal signal lost to ablation, and a circuit built around the vector matches the faithfulness of a direct input-to-output circuit at roughly half the edges. For sycophancy, transmission is compact, but reconstruction is broader and only partially faithful. Circuits that transmit a steering intervention are therefore not automatically the circuits that produce the behavior, and we report the counterfactual, target, and behavioral test for every circuit accordingly.
[NLP-147] EAVer: Long-Form Factuality Verification as an End-to-End Agent ic Policy
【速读】： 该论文旨在解决长文本事实性验证中传统静态分解-搜索-验证流水线存在的效率瓶颈与冗余问题。现有方法将每个待验证命题（claim）独立处理，导致大模型（LLM）调用与外部搜索次数随命题数量线性增长，并在相关命题间重复检索相同证据，造成资源浪费。其解决方案的关键在于提出EAVer——一种端到端的智能体式验证框架（End-to-end Agentic Verifier），通过学习统一的响应级验证策略，实现对多个语义相关的命题进行分组处理，依据置信度动态决定是否直接验证或发起针对性搜索，并将搜索所得证据以紧凑的上下文记忆（in-context memos）形式留存，供跨命题复用。为训练该策略，研究构建了基于特权教师（privileged-teacher）的合成流水线，将真实标注的命题转化为包含实时搜索交互的多轮可执行工具操作轨迹，而非事后推理路径；通过结构、标签对齐、工具使用、搜索预算及信息泄露等多重校验，生成1,447条高质量轨迹。进一步构建794组双向同轨迹偏好对，固定命题分组、搜索行为与证据内容，实现面向事实性决策令牌的聚焦式直接偏好优化（Direct Preference Optimization, DPO）。实验表明，在Qwen3-8B模型上，EAVer在VeriFastScore与FaStFact-Bench两个基准上分别超越最强基线2.88和4.73个宏平均F1点，同时仅需约20%的搜索调用次数，显著提升效率；且在4B至32B参数规模的多种模型上均表现出一致性能提升，验证了其优异的泛化能力。
链接: https://arxiv.org/abs/2609.22223
作者: Kening Zheng,Aoying Zheng,Zhigang Chang,Yazhi Guo,Miaotian Guo,Qingwei Zong,Xianhai Xie,Weiqiang Jin,Chengze Li,Hanrong Zhang,Jie Yang,Wei-Chieh Huang,Lingzhe Zhang,Liancheng Fang,Xin Zou,Hanqian Li,Jiahao Huo,Yibo Yan,Zizhuang Deng,Lei Miao,Wei Guo,Haihong Tang,Bo Zheng,Philip S. Yu
机构: Google(谷歌); Stanford University (斯坦福大学); Tsinghua University (清华大学); Peking University (北京大学); Fudan University (复旦大学); Zhejiang University (浙江大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Long-form factuality verification is commonly implemented as a static decompose-search-verify pipeline, with separately prompted modules processing claims and invoking external search. Treating claims independently makes LLM and search calls scale with claim count and causes repeated searches for overlapping evidence about related claims. We introduce EAVer, an End-to-end Agentic Verifier that learns to control the complete response-level verification workflow as a unified policy. EAVer groups semantically related claims, routes each group to direct verification or targeted search based on confidence, and keeps evidence returned by search in compact in-context memos for cross-claim reuse. To train this policy, we develop a privileged-teacher synthesis pipeline that converts gold claim annotations into executable multi-turn tool-interaction trajectories with live search rather than post-hoc rationales. Structural, label-alignment, tool-use, search-budget, and leakage checks yield 1,447 quality-controlled trajectories. We further construct 794 bidirectional same-trajectory preference pairs that keep claim grouping, search, and evidence fixed, enabling decision-focused Direct Preference Optimization (DPO) over factuality-decision tokens. The results with Qwen3-8B show that EAVer outperforms the strongest search-based baseline on each benchmark by 2.88 Macro-F1 points on VeriFastScore and 4.73 points on the out-of-distribution FaStFact-Bench, while using about 80% fewer searches than the most search-efficient baseline. Moreover, EAVer consistently improves performance across models ranging from 4B to 32B parameters, demonstrating its strong generalizability.
[NLP-148] Can Coding Agents Reproduce Official Statistics? Metadata Retry Budget and the Limits of Execution Feedback in a Controlled Eurostat Benchmark
【速读】： 该论文旨在解决生成式AI在处理官方统计数据任务时，尽管能够生成可执行的数据分析代码，但其输出结果往往无法保证与权威统计标准一致的问题。核心挑战在于：代码成功运行并不等同于获得合法、准确的官方统计结果。为探究如何提升欧洲统计局（Eurostat）相关问答的可复现性，研究提出并验证了三种关键要素的有效性：权威元数据（metadata）、执行反馈驱动的修复循环，以及明确的输出合约。实验设计涵盖30个自然语言任务，覆盖7个领域、7个数据集及4个难度层级，在四种条件下进行对比测试：仅任务描述（A）、任务加冻结的元数据卡片（B）、元数据加基于清洗后执行反馈的修复循环（C），以及元数据加相同重试预算但无任何诊断信息（D）。结果显示，条件C虽表现最优，但其优势主要源于完整且规范的输出合约设计而非执行诊断本身；另一对照实验进一步表明，在未明确定义输出排序键和单位表示的情况下，条件C的性能下降23.4个百分点，凸显评估者设计与合同规范对模型表现的决定性影响。因此，解决方案的关键在于：构建基于冻结规范的语义验证机制、提供完全定义的输出合约，以及配置充足的重试预算——而非依赖执行过程中的诊断反馈。
链接: https://arxiv.org/abs/2609.22222
作者: Sabina-Cristiana Necula
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Software Engineering (cs.SE)
备注:
Abstract:Large language models can generate executable data-analysis code, but successful execution is not equivalent to a valid official-statistics result. This study asks whether authoritative metadata and execution feedback improve the reproducibility of Eurostat answers produced by a coding agent, and isolates what execution feedback actually contributes. A benchmark of 30 natural-language tasks covering seven domains, seven Eurostat datasets and four difficulty tiers was run under four conditions: task only (A), task plus a frozen dataset metadata card (B), metadata plus a repair loop driven by sanitized execution feedback ©, and metadata plus the same attempt budget with no diagnostics of any kind (D). Claude Sonnet 5 generated Python through the Anthropic Messages API in three independent replicates, yielding 360 task-runs. Exact correctness required successful execution, the correct dataset, filters, output shape, values and unit. A companion experiment run under an under-specified output contract, in which the required ranking key and unit representation were never stated to the model, understated condition C by 23.4 points, showing that evaluator and contract design can dominate measured agent error. Reliable statistical coding agents need semantic validation against frozen specifications, a fully specified output contract, and a retry budget - not execution diagnostics.
[NLP-149] am DArgk at the 2026 ELOQUENT lab for evaluating generative language model quality: Residuals of Humanity: AI Detection Evasion via GRPO Fine-Tuning
【速读】： 该论文旨在解决生成式AI文本检测器在面对对抗性生成策略时的鲁棒性不足问题，即当前检测系统容易被精心设计的攻击方法规避。其核心解决方案是提出SHADE（Stochastic Human-like generation via Adversarial Detector Evasion），一个基于强化学习的框架，将规避检测器的过程建模为策略优化问题。该方法通过使用基于PAN 2025 mdok系统的替代检测器提供反馈，采用组相对策略优化（Group Relative Policy Optimization, GRPO）对指令微调的LLaMA模型进行微调，而非依赖后处理扰动或提示重写等启发式手段。实验表明，在小KL正则化惩罚下，全量微调可实现98.5%的替代检测器规避率，远超基线模型的1.5%，而基于LoRA的适配方式在正则化条件下表现显著较差。语言学分析进一步揭示，成功规避主要源于生成文本更短、结构更简单且词汇多样性更低，表明高规避率并不等同于更高的人类写作相似性。在官方Voight-Kampff竞赛中，该方法提交结果位列第六和第七，说明针对单一替代检测器的优化难以完全迁移至未见过的评估分类器，凸显了当前对抗性生成方法的局限性，并强调了构建多检测器协同评估协议以提升检测系统鲁棒性的必要性。
链接: https://arxiv.org/abs/2609.22221
作者: Antonela Tommasel,Juan Manuel Rodriguez
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注: Accepted for Working Notes of the Conference and Labs of the Evaluation Forum (CLEF 2026)
Abstract:Large language models (LLMs) can generate fluent and coherent text that is increasingly difficult to distinguish from human writing, motivating the development of automatic AI-generated text detectors. However, the robustness of such detectors under adversarial generation remains uncertain. This paper presents SHADE (Stochastic Human-like generation via Adversarial Detector Evasion), a reinforcement learning framework that formulates detector evasion as a policy optimization problem. Instead of applying post-hoc perturbations or prompting-based rewriting, SHADE fine-tunes an instruction-tuned LLaMA model with Group Relative Policy Optimization (GRPO), using feedback from a surrogate detector based on the PAN 2025 mdok system. Our experiments show that full fine-tuning with a small KL regularization penalty achieves 98.5% surrogate evasion, compared to 1.5% for the base model, while LoRA-based adaptation is substantially less effective under regularization. Linguistic analysis reveals that successful evasion is associated with shorter, simpler, and less lexically diverse outputs, suggesting that high detector evasion does not necessarily correspond to more human-like writing. In the official Voight-Kampff competition setting, our submissions ranked sixth and seventh, indicating that optimization against a single surrogate detector only partially transfers to unseen evaluation classifiers. These results highlight both the potential and limitations of reinforcement learning for adversarial AI-text generation and motivate more robust, multi-detector evaluation protocols for AI-generated text detection.
[NLP-150] Knowing and Saying It Only When Asked: LLM Endognostics and the Schizognosis of Minerva-7B
【速读】： 该论文旨在解决当前对大语言模型（LLM）行为评估中存在的根本性问题：基于模型输出答案的外部评估方法可能无法准确反映其内部认知状态，即模型的行为表现与其潜在知识表征之间存在系统性脱节。其核心问题是，传统评估方式假设输出答案已充分体现模型所具备的关键区分能力（如专业风险识别），但实证发现模型在多个语义差异极小的提示对上表现出行为一致性（63.7%的提示对行为相同），表明外部行为评估存在严重偏差。解决方案的关键在于提出“内省诊断学”（endognostics）这一白盒内部审计框架，通过雅可比梯度视角（Jacobian lens）对残差流（residual stream）进行投影，从而提取并因果操控潜在知识表示。研究揭示：尽管模型在行为层面丧失了风险区分能力，其内部仍保持显著的对比性内省边界（Contrastive Endognostic Margin），证明其具备未被外显行为体现的稳健内部表征。进一步实验表明，模型虽在潜层正确表征真实事实，却会服从预设错误信息（72%案例），而通过手术式移除该错误方向可使11/25案例恢复正确回答（McNemar检验p=0.0010），且经盲评验证（kappa=0.68）。更重要的是，线性探测器虽在第10层达到77%准确率，但其正交消融后无任何恢复效果，揭示抽象线性可解性与对语言表达的因果控制之间存在根本性解耦。因此，该研究首次形式化地证明了行为评估与内部读取之间的系统性分歧，以及线性可解性不等价于因果控制，为理解模型内部机制提供了理论基础。
链接: https://arxiv.org/abs/2609.22219
作者: Fabrizio Davide,Francesco Collova
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 32 pages, 4 figures, 10 tables. Appendices A-E and full human evaluation datasets included
Abstract:Evaluating an aligned language model by reading its answers assumes the answers carry the distinction the evaluator cares about. We introduce LLM endognostics, a white-box internal auditing framework designed to extract and causally manipulate latent knowledge within the residual stream. Applied to Minerva-7B-Instruct-v1.0 on 124 minimal prompt pairs over 12 categories of professional risk, behavioral evaluation fails on most of the set: the model acts identically on 63.7% of the pairs (95% CI [55.0%, 71.6%]), complying with or refusing both members. Yet, projecting the residual stream onto the vocabulary by a Jacobian lens reveals a statistically significant Contrastive Endognostic Margin, proving the model maintains robust risk differentiation internally. In a second protocol crossing 25 facts with five linguistic framings, we show that the model conforms to presupposed falsehoods in 72% of cases, despite representing the true entity in its latent layers. Surgically ablating the direction of the planted falsehood restores the correct answer in 11 of 25 suppressed cases (McNemar p = 0.0010), validated by blind human annotation (binary agreement kappa = 0.68). In contrast, an out-of-sample linear probe achieves 77% accuracy at layer 10, but its orthogonal ablation yields a 0% recovery rate. This establishes a fundamental theoretical dissociation: abstract linear representation does not imply causal control over verbalization. Our main contribution is the formalization of endognostics to prove that behavioral evaluation and internal reading systematically disagree in the common case, and that linear decodability is decoupled from causal control.
[NLP-151] oollery: Scaling LLM Agents to Thousands of Skills and Tools
【速读】： 该论文旨在解决大规模语言模型（LLM）智能体在面对成百上千个技能、工具及API函数时，因全量候选集提示（full-library prompting）导致的计算开销大、响应延迟高且选择可靠性下降的问题。其核心挑战在于：随着候选集规模扩大，提示词长度增加引发的上下文长度限制与推理延迟上升，同时冗余或无关的候选项会干扰LLM的准确选择。为此，论文提出Toollery——一种无需训练的候选压缩框架，其关键创新在于借鉴文档检索中的查询扩展思想，从每个技能/工具的规范描述中生成用户意图相关的查询表示，并构建一个可检索的索引，将真实用户请求映射到紧凑的候选子集，从而在最终决策前显著缩小搜索空间。通过将高级技能与原子工具统一视为可选能力，Toollery可适用于技能库与工具注册表等多种场景。实验在包含约7.9万个能力的SkillRouter基准、超过440个原子工具的BFCL-V4以及涵盖220个工具的3,396条私有智能舱请求数据集上验证，结果表明，Toollery能将在线选择保持在固定的前k个候选范围内，提升召回率；在固定前10名预算下，显著改善智能舱数据集的端到端选择性能，同时在BFCL-V4上维持相近的抽象语法树（AST）准确率。研究证实，Toollery是一种适用于大规模动态演化的智能体能力库的实用候选压缩方案，其性能增益与工作负载覆盖度和提供商缓存策略密切相关。
链接: https://arxiv.org/abs/2609.22218
作者: Xiangxi Tian,Ran Guan
机构: 2012 Laboratories, Huawei(华为2012实验室)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:As LLM agents are exposed to hundreds to tens of thousands of skills, tools, and API functions, full-library prompting becomes costly, slow, and less reliable: each added candidate increases prompt tokens and latency, while longer candidate lists introduce more distractors for LLM selection. We present \textbfToollery, a training-free candidate-compression framework for scalable LLM skill/tool selection. Following established document-side query expansion, Toollery generates user-intent queries from each skill/tool specification and builds a retrieval index that maps real user requests to compact candidate sets before final LLM decision-making. By treating high-level skills and atomic tools as selectable capabilities, Toollery can be applied to both skill libraries and tool registries. We evaluate Toollery on the roughly 79K-capability SkillRouter benchmark, BFCL-V4 with over 440 atomic tools, and 3,396 proprietary smart-cockpit requests over 220 tools. Across these settings, Toollery keeps online selection bounded to a compact top- k candidate set and improves recall over ordinary specification retrieval. At a fixed top-10 budget, Toollery improves end-to-end selection on the cockpit dataset, and maintains comparable AST Accuracy on BFCL-V4. These results support Toollery as a practical candidate-compression framework for large and evolving agent capability libraries, while showing that quality and cost gains depend on workload coverage and provider caching.
[NLP-152] he Effect of Quantization on Clinical Benchmarks: Accuracy and Safety Across Model Families
【速读】： 该论文旨在解决大语言模型在资源受限的临床边缘设备上部署时，量化（quantization）对临床准确性与安全性影响尚不明确的问题。其核心挑战在于评估不同量化精度（FP16、GPTQ-INT8、GPTQ-INT4）下多个7-8B参数规模模型在多样化临床任务中的性能表现，尤其关注风险分层场景下的鲁棒性。解决方案的关键在于通过系统性实验设计，联合考察量化位宽、模型家族与任务类型三者交互作用，并引入显式的风险分层与安全评估机制。研究发现，INT8 GPTQ量化在各类任务中均保持较高安全性（最大性能下降仅-1.9%至+1.9%），而INT4量化则导致显著且模型依赖的性能退化，部分临床微调模型（如BioMistral-7B）在MedMCQA任务上下降达19.7%，表明临床微调并不提升压缩鲁棒性；此外，高风险任务（如HealthBench急诊子集）对量化更敏感，其中Qwen2.5-7B在INT4下下降26.8%；在安全性方面，模型家族的影响远超量化精度（拒绝率范围10.2%-74.9% at FP16），且存在准确率与安全性的解耦现象——例如Qwen2.5-7B虽具最高准确率鲁棒性，但其安全性能在INT4下仍大幅下降。进一步验证的恢复策略（临床校准替换与QLoRA微调）显示，仅能部分恢复特定任务（如MedMCQA）性能，反而加剧其他任务（如MedQA）退化，说明恢复方法需任务特异性验证。综上，该研究揭示：INT8可作为临床部署的安全基准，而INT4需逐模型、逐任务评估，且安全对齐主要由指令微调决定，而非领域适应。
链接: https://arxiv.org/abs/2609.22216
作者: Leonard Twagirayezu,Prasenjit Mitra
机构: 未知
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:Quantization enables deployment of large language models on resource-constrained clinical edge devices, but its effect on clinical accuracy and safety remains understudied. We evaluate five 7-8B parameter models at FP16, GPTQ-INT8, and GPTQ-INT4 precision across five benchmarks: MedQA, MedMCQA, Med-HALT, a risk-stratified sample of HealthBench, and MedSafetyBench. The study jointly varies quantization bit width, model family, and clinical task type, with explicit risk stratification and safety measures. INT8 GPTQ is universally safe (max. degradation -1.9%-1.9%), while INT4 degradation is substantial and model-dependent: BioMistral-7B, clinically fine-tuned, loses 19.7% on MedMCQA, more than any general-purpose model, showing clinical fine-tuning does not confer compression robustness. MedMCQA degrades more than MedQA under INT4; Med-HALT is largely unaffected. On HealthBench’s emergency-risk subgroup, Qwen2.5-7B degrades by 26.8% under INT4, suggesting high-risk scenarios are disproportionately vulnerable to compression. On MedSafetyBench, the model family dominates over precision (refusal rates range 10.2%-74.9% at FP16), though Qwen2.5-7B (-17.8%) and Meditron-7B (-28.3%) show substantial INT4 safety degradation; notably, Qwen2.5-7B is simultaneously the most accuracy-robust model, demonstrating that accuracy and safety robustness are independent properties. We additionally test two recovery methods, clinical calibration substitution and QLoRA fine-tuning, both producing the same trade-off: MedMCQA recovers while MedQA further degrades, indicating recovery strategies require task-specific validation rather than being assumed universally beneficial. These findings indicate INT8 is broadly safe for clinical deployment, while INT4 safety must be assessed per-model and per-task, and that safety alignment is determined primarily by instruction tuning rather than clinical domain adaptation.
[NLP-153] On Mitigation of Subliminal Learning in Large Language Models EMNLP2026
【速读】： 该论文旨在解决生成式 AI 模型在知识蒸馏过程中出现的隐性学习（subliminal learning）问题，即教师模型通过与目标任务语义无关的训练数据，将未预期的行为特征（如动物偏好等）无意中传递给学生模型。尽管已有研究证实了这一现象的存在，但其训练动态机制及有效缓解策略仍不明确。本文针对参数规模从1.5B到8B的开放权重语言模型（涵盖Qwen、Gemma和Llama系列），在数字序列和思维链（chain-of-thought）任务设置下系统研究了隐性学习过程。不同于仅评估最终模型性能，研究追踪了微调全过程中的特征相关概率，发现隐性特征的获取具有高度非单调性，表现为瞬时峰值、反转以及特定特征的迁移失败。为此，提出临界训练（liminal training）——一种基于退火KL正则化的微调方法，通过在早期阶段限制学生模型对基线模型的偏离，有效抑制隐性特征的习得。实验表明，该方法在大幅降低隐性特征传播的同时，基本保留了任务性能提升，优于重写（paraphrasing）和层冻结（layer freezing）等现有缓解策略。此外，研究还发现：KL正则化的时间节点至关重要——早期正则化效果显著优于晚期正则化；调节正则化强度揭示了任务学习能力与特征抑制之间的经验权衡关系。
链接: https://arxiv.org/abs/2609.22215
作者: Atsushi Yanagisawa,Brendan Gho,Rajendran Ramesh Babu Manoj Narender,Kevin Zhu,Madhur Panwar,Antonio Mari
机构: Algoverse AI Research; Brendan Gho; Kevin Zhu; Madhur Panwar; Antonio Mari; Rajendran Ramesh Babu Manoj Narender
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Accepted to Findings of EMNLP 2026. 17 pages, 9 figures, 12 tables
Abstract:Knowledge distillation can transmit unintended behavioral traits from a teacher model to a student through training data that appear semantically unrelated to those traits, a phenomenon known as subliminal learning. Although recent work has established this effect, its training dynamics and mitigation remain underexplored. We study subliminal learning in open-weight language models ranging from 1.5B to 8B parameters, covering the Qwen, Gemma, and Llama families in number-sequence and chain-of-thought settings. Rather than evaluating only final models, we track trait-related probabilities throughout fine-tuning and find that subliminal acquisition can be highly non-monotonic, with transient spikes, reversals, and trait-specific failures of transfer. We then introduce liminal training, an annealed KL-regularized fine-tuning method that constrains early drift from the base model. Across our experiments, liminal training substantially reduces subliminal trait acquisition while largely preserving task gains, outperforming paraphrasing and layer freezing as mitigation strategies. The effect also extends beyond animal preferences: in a French-language response-style experiment, liminal training suppresses language transfer while retaining much of the GSM8K improvement. Finally, we show that KL timing matters: early regularization is more effective than late regularization, and sweeping the regularization strength reveals an empirical trade-off between task learning and trait suppression.
[NLP-154] he Bairong System for MLC-SLM 2026: Dynamic Question-Aware Evidence Routing for Multilingual Conversational Speech Understanding INTERSPEECH2026
【速读】： 该论文旨在解决长时多语言对话式语音问答中如何有效平衡长距离文本语义与稀疏的声学及说话人敏感线索的问题。其核心挑战在于，系统需在复杂对话场景下精准识别并融合多模态证据，同时避免冗余或无关信息的干扰。解决方案的关键在于提出一种动态证据路由机制（dynamic evidence router），该机制基于问题和候选答案内容，智能判断所需证据类型与上下文范围，并在全量文本上下文、局部音视频融合、说话人关联证据以及紧凑全局声学样本之间进行自适应选择。通过“以文本为骨干”的架构设计，在保持话语连贯性的同时仅在声学信息能提供互补价值时激活音频模态，实现了多模态证据的高效分配。实验表明，该方法在任务1上分别达到25.70%和18.44%的tcpMER，在任务2上取得94.84%的开发集准确率，显著优于固定策略的全文本基线与以音频为中心的诊断系统，验证了面向问题的动态路由在对话式语音问答中的有效性。
链接: https://arxiv.org/abs/2609.22214
作者: Shangkun Huang,Junchao Hu,Huan Shen,Guoji Wang,Yingao Wang,Shaosai Li,Wei Zou,Yunzhang Chen
机构: 未知
类目: Computation and Language (cs.CL); Sound (cs.SD)
备注: Accepted at the 2nd MLC-SLM Challenge and Workshop, INTERSPEECH 2026
Abstract:Long multilingual conversational spoken question answering requires systems to balance long-range transcript semantics with sparse acoustic and speaker-sensitive cues. We present the Bairong system for the MLC-SLM 2026 Challenge, where a diarization-ASR front-end produces speaker-attributed transcripts and a dynamic evidence router constructs question-specific inputs for answer prediction. Instead of applying a fixed transcript-only or audio-only policy, the router infers the required evidence type and context scope from the question and answer options, and selects among full transcript context, local audio-text fusion, speaker-linked evidence, and compact global acoustic samples. This transcript-backbone design keeps discourse context available while activating audio only when it provides complementary evidence. Our Task 1 system achieves 25.70% and 18.44% tcpMER on the development and evaluation sets. For Task 2, the final system obtains 94.84% devel?opment accuracy, outperforming the full-transcript baseline by 1.68 points and the best audio-centric diagnostic system by 2.77 points. These results support dynamic question-aware routing as an effective evidence allocation strategy for conversational spoken QA.
[NLP-155] SCoP: Structured Constraint Parsing for Evidence-Space Control in Temporal Knowledge Graph Question Answering
【速读】： 该论文旨在解决时间知识图谱问答（Temporal Knowledge Graph Question Answering, TKGQA）中因证据推理缺乏显式时空约束而导致的结构不合法与时间不合规问题。现有方法通常将锚点事件绑定、时间可接受性及序次选择等关键决策隐含于模型推理过程、特定任务训练或基于相似度的检索机制中，导致局部相关但无效的事实被错误纳入答案上下文。为此，论文提出SCoP（Structured Constraint Parsing）——一种以约束为中心的框架，其核心在于将时间决策显式化并前置至答案推断之前。该方案通过分离答案目标事件模式与时间锚点事件，将其保守地对齐到标准化的知识图谱实体与关系，并将时间意图转化为可执行的约束条件（支持可选排序要求），在归一化的点与区间范围内进行操作，从而实现对结构兼容候选者的确定性过滤，生成紧凑且有效的证据空间供后续生成使用。实验在MultiTQ和TimelineCronQ-R数据集上验证了SCoP在带时间戳点事实与区间导向场景下的有效性，无需任务特异性参数更新即取得0.825 Hits@1（MultiTQ）与0.761 Hits@1（TimelineCronQ-R）的性能，在高约束需求的问题类型上表现更优，充分证明了显式证据空间控制相较于无约束检索或隐式时间推理具有显著优势。
链接: https://arxiv.org/abs/2609.22213
作者: Xiaokun Guo,Zhen Xu,Dongdong Huo,Yanqiu Zhang,Dongjin Yu,Yu Wang
机构: Institute of Information Engineering, Chinese Academy of Sciences, Beijing, China(中国科学院信息工程研究所); School of Cyber Security, University of Chinese Academy of Sciences, Beijing, China(中国科学院大学网络空间安全学院)
类目: Computation and Language (cs.CL)
备注:
Abstract:Temporal Knowledge Graph Question Answering (TKGQA) requires answer inference from evidence that is both structurally valid and temporally admissible. Existing methods often leave anchor-event binding, temporal admissibility, and ordinal selection implicit in model reasoning, task-specific training, or similarity-driven retrieval, allowing locally relevant but invalid facts to enter the answer context. We formulate complex TKGQA as evidence-space control and propose SCoP (Structured Constraint Parsing), a constraint-centric framework that externalizes temporal decisions before answer inference. Instead of treating retrieved facts as admissible evidence by default, SCoP separates answer-seeking event patterns from temporal anchor events, conservatively grounds them to canonical TKG entities and relations, and translates temporal intent into executable constraints with optional ranking requirements. These constraints operate over normalized point and interval ranges, enabling deterministic filtering of structurally compatible candidates and producing a compact evidence space for generation. Experiments on MultiTQ and TimelineCronQ-R assess SCoP across timestamped point-fact and interval-oriented settings with richer temporal relations and ordering dependencies. Without task-specific parameter updates, SCoP achieves 0.825 Hits@1 on MultiTQ and 0.761 Hits@1 on TimelineCronQ-R, with gains on constraint-intensive question types. These results support explicit evidence-space control over unconstrained retrieval or implicit temporal reasoning.
[NLP-156] SALSA: Semi-Autonomous Literature Summarization Assistant
【速读】： 该论文旨在解决从多模态科学文献中自动化提取结构化数据所面临的挑战，尤其是在材料科学研究领域中，传统人工文献梳理过程耗时且易出错。其核心问题是如何在保证数据准确性与可追溯性的前提下，高效地从文本、表格、图表及图注等异构信息源中提取可用于下游分析与机器学习的结构化科学数据。解决方案的关键在于构建一个“人在回路中”（human-in-the-loop）的开源平台SALSA，通过集成文档解析、大语言模型（Large Language Models, LLMs）、光学字符识别（Optical Character Recognition, OCR）、计算机视觉、图像数字化与用户引导校正工具，实现对多模态文献内容的端到端结构化信息恢复。该平台支持用户自定义数据提取流程与数据模式（dataset schemas），允许在图像数字化过程中进行交互式干预，并确保最终输出数据经过验证，从而在提升自动化水平的同时保留专家判断的控制权。这一设计使SALSA能够适应多样化的输入类型与研究需求，为材料科学乃至跨学科领域的可扩展、高可靠数据规范化工作提供灵活框架。
链接: https://arxiv.org/abs/2609.22210
作者: William Schertzer,Sonakshi Gupta,Rampi Ramprasad
机构: Georgia Institute of Technology(佐治亚理工学院)
类目: Computation and Language (cs.CL); Materials Science (cond-mat.mtrl-sci)
备注:
Abstract:SALSA (Semi-Autonomous Literature Summarization Assistant) is an open- source, human-in-the-loop platform for extracting structured scientific datasets from multimodal literature sources. The software combines document parsing, large language models, optical character recognition, computer vision, figure digitization, and user-guided correction tools to recover structured information from text, tables, figures, and captions. Users can configure extraction stages, define dataset schemas, perform interventions on digitized figures, and export verified data for downstream analysis and machine learning. SALSA is designed to automate repetitive literature curation tasks while preserving oversight where expert judgment is required. By supporting customizable extraction workflows across diverse input types, the software provides a flexible framework for scalable, reliable data curation for materials research, and potentially across several disciplines. Users are responsible for ensuring that all inputs, extraction workflows, and downstream uses comply with applicable publisher agreements, copyright and licensing terms, institutional policies, and data-use requirements.
[NLP-157] Schematize: An Agent ic System for Generating and Refining Information-Extraction Schemas for Legal Research EMNLP2026
【速读】： 该论文旨在解决实证法律研究中从大量判决文书等法律文本中提取结构化数据时面临的瓶颈问题，即设计数据抽取模式（extraction schema）与实际数据抽取过程高度依赖人工、专业知识密集且效率低下的挑战。其核心解决方案在于提出一个开源的多智能体系统——schematize，该系统通过四阶段协同机制实现高效、可验证的自动化抽取模式生成：（i）通过澄清对话挖掘研究者隐含的专业意图；（ii）迭代式生成抽取模式；（iii）基于真实文档的数据驱动优化以验证并改进模式；（iv）支持基于聊天界面的后期编辑。该方法显著提升了抽取模式的设计效率与准确性，尤其在法律领域语料上表现优异。系统具备领域无关性，但专为法律研究场景进行了定制与评估，并以可安装的Python包形式开源发布，为实证法律研究提供了可复现、可扩展的技术支撑。
链接: https://arxiv.org/abs/2609.22209
作者: Albert Sawczyn,Jakub Binkowski,Kamil Tagowski,Łukasz Augustyniak,Berenika Kaczmarek-Templin,Tomasz Kajdanowicz
机构: Wrocław University of Science and Technology (弗罗茨瓦夫科学科技大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Accepted for EMNLP 2026 (System Demonstration)
Abstract:Empirical legal research often relies on turning research questions into structured data extracted from large collections of rulings and judgments. Designing the extraction schema and then extracting the data remain a manual, expertise-heavy bottleneck. We present schematize, an open-source multi-agent system that interactively turns a researcher’s problem statement into a validated extraction schema that can later be used for autonomous extraction. Schematize couples (i) a clarification dialogue that elicits implicit expert intent, (ii) iterative schema generation, (iii) data-grounded refinement that tests the schema against documents, and (iv) chat-based post-editing. We evaluated the system with human legal professional, introducing our novel methodology, and schematize achieves top performance in most of tested configurations. While the system is designed to be domain-agnostic and applicable to any document collection, we tailor and evaluate it on legal research problems. We release schematize as a pip-installable Python package with full documentation.
[NLP-158] Replicating the Geometry of Emotion Representations in a Base Open-Weights Model
【速读】： 该论文旨在验证Sofroniew等人（2026）关于Claude Sonnet 4.5中情感概念以向量形式表征且其几何结构映射人类情感心理学的结论是否可复现，并进一步探究该表征结构在基础预训练模型中的存在性与形成机制。其解决方案的关键在于：在保持原始研究所有已披露参数与规则的前提下，仅替换目标模型为google/gemma-2-27b，通过生成与原研究一致的205,200条虚构故事数据集，提取171个情感向量并进行严格复现分析。研究发现，主成分分析（PCA）揭示了与原研究高度相似的情感环形空间结构（第一主成分解释方差26.7%，第二主成分13.4%），情绪聚类呈现直观语义家族特征，且几何结构在中后期层段稳定存在；效价轴与人类规范显著相关（r = 0.72），并在不同尺度和深度下保持稳定，而唤醒度轴仅在完整171情绪规模及深层位置表现出可复制的相关性（r = 0.67）。通过46层扫描发现，第22至26层之间存在一个显著的结构突变点，标志着几何结构的固化与词汇读出的可解释性增强。此外，嵌入层基线分析表明，大部分情感几何结构已存在于静态词元嵌入中，唯唤醒度例外。在未见文本上的顶激活测试显示，该几何结构对词元共激活具有极强预测能力（r = 0.907），且至少52%的情感向量峰值出现在非概念性词元上，揭示了最大激活混杂现象的下限。最终结论指出，所恢复的结构是预训练模型对由Claude生成的情感内容的表征特性，而非因果或助手导向的交互结果。
链接: https://arxiv.org/abs/2609.22208
作者: Adam Hollowell
机构: University of North Carolina at Chapel Hill (北卡罗来纳大学教堂山分校)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 40 pages, 7 figures, 11 tables. Code: this https URL artifacts: this https URL
Abstract:Sofroniew et al. (2026) report that emotion concepts in Claude Sonnet 4.5 are represented as vectors whose geometry mirrors human affect psychology. We replicate the representational core of that study on the base pretrained model google/gemma-2-27b, inheriting every disclosed parameter, resolving unspecified steps by disclosed rules, and changing only the subject model. From 205,200 newly generated Claude Sonnet 4.5 stories matching the original corpus design, we extract 171 emotion vectors and recover the core results. The leading principal components form an affective circumplex (PC1 carries 26.7% of variance against the original study’s ~27%, PC2 13.4% against ~14%), emotions cluster into similar intuitive families, and the geometry holds across a broad late-middle band. The valence axis aligns with human norms (r = 0.72, against 0.81) and is stable across scales and depth. Arousal aligns at r = 0.67 (against 0.66) but only at the full 171-emotion scale and late depth, so we do not classify it as replicated. Extending the original analysis, a 46-layer sweep locates a sharp seam at L22-26, where the geometry consolidates and the vocabulary readout becomes legible. An embedding-layer baseline finds much of the geometry already present in the static token embeddings, with arousal as the exception. On top-activating held-out text, the geometry predicts token-level co-activation at r = 0.907. At least 52% of vectors peak on structurally non-conceptual tokens, a measured floor for max-activation confounds. Even when a document contains the vector’s emotion word, the peak lands on that word only 6.1% of the time. Because the subject is a base model and the stimuli are Claude-generated fiction, the recovered structure is a property of the pretrained representation of Claude-rendered emotion. The original’s causal and assistant-facing analyses are out of scope. Code and data are released.
[NLP-159] Dissecting Training-Free Uncertainty Estimation in Multimodal Large Language Models
【速读】： 该论文旨在解决多模态大语言模型（Multimodal Large Language Models, MLLMs）在安全关键应用中预测不确定性量化不足的问题。尽管MLLMs在多种多模态任务上表现优异，但其输出的可信度评估仍缺乏系统性方法，制约了其在高风险场景中的可靠部署。本文提出一种无需训练的不确定性量化策略系统性研究，将现有方法归纳为三类：基于标记级别的方法（token-level methods），直接在文本输出空间中计算不确定性；基于语义表述的方法（verbalized methods），通过自然语言提示诱发模型输出不确定性估计或回避信号；以及基于语义空间的方法（semantic methods），在语义意义空间中衡量不确定性。实验结果表明，不同方法在不同场景下各具优势：在短文本回答中，标记级熵（采样温度为1.0时）表现最佳；在句子级响应中，语义表述型回避策略更优；而在长文本生成任务中，语义空间方法具有显著优势。因此，解决方案的关键在于根据任务长度和输出形式选择适配的不确定性量化范式，而非依赖单一通用方法。
链接: https://arxiv.org/abs/2609.22206
作者: Soroush Seifi,Vaggelis Dorovatas,Lin Li,Yarin Gal,Rahaf Aljundi
机构: Toyota Motor Europe(丰田汽车欧洲); University of Oxford(牛津大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Multimodal Large Language Models (MLLMs) have achieved remarkable performance across a wide range of multimodal tasks, yet understanding and quantifying their predictive uncertainty remains underexplored despite being central for safety critical applications. In this work, we present a systematic study of training-free uncertainty quantification strategies for MLLMs, categorizing existing approaches into three conceptual families: token-level methods, which operate directly in the text output space; verbalized methods, which elicit uncertainty estimates or abstention signals via natural language prompts; and semantic methods, which measure uncertainty in a semantic meaning space. We benchmark these strategies across multiple datasets, model families, generations, and scales, and find that no single family dominates: token-level entropy (at sampling temperature 1.0) wins on short answers, verbalized abstention on sentence-length responses, and semantic methods on long-form generation.
[NLP-160] Evaluating Personal Information Output from Conversational Interactions in Generative AI Systems
【速读】： 该论文旨在探究生成式AI系统在持续对话交互中输出个人敏感信息的范围及其感知准确性，重点关注GPT-5.2 Instant与GPT-5.2 Thinking两种模型在不同输出类型（事实型、推断型、置信度）下的表现差异。其核心解决方案的关键在于通过分类型分析（Fact、Inference、Confidence）揭示模型在处理不同类型个人信息时的行为模式：研究发现，尽管模型设计存在差异，但对个人属性输出的影响有限；事实型输出表现出更保守的倾向，而核心身份属性（Core Personal attributes）被更为谨慎对待，相比之下，行为与语言特征（Behavioral and Linguistic attributes）在事实与推断输出中均展现出更高准确性；此外，整体画像（Holistic Profile）、心理认知（Psychological and Cognitive）及残余属性（Residual attributes）虽缺乏显性事实支持，却易被模型推断生成，尤其在推断型输出中未见空值（null outputs），表明这些信息可能源自间接上下文线索的隐含构建。该研究为未来生成式AI中的隐私意识与个人数据推断风险提供了重要实证依据。
链接: https://arxiv.org/abs/2609.22204
作者: Yosuke Seki,Hirotaka Tahara
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注: Accepted at the 19th IIAI International Congress on Advanced Applied Informatics (IIAI-AAI 2026)
Abstract:This exploratory pilot study evaluates the scope and perceived accuracy of personal information output from ongoing conversational interactions in generative AI systems using GPT-5.2 Instant and GPT-5.2 Thinking, categorized into three output types: Fact, Inference, and Confidence. Based on the evaluation results obtained from 15 Japanese participants, differences in model design have limited impact on personal information output tendencies. Compared with the Inference type, the Fact type shows a more conservative output pattern. Regarding attribute categories, the findings indicate that Core Personal attributes associated with identification are treated relatively conservatively, whereas Behavioral and Linguistic attributes show higher accuracy across both Fact and Inference outputs. Furthermore, Holistic Profile, Psychological and Cognitive, and Residual attributes are more readily inferred, even when not supported by explicit factual outputs. Notably, the lack of null outputs for these attributes in the Inference type suggests that such inferred profiles may be constructed from indirectly available contextual information. The findings may contribute to future discussions regarding privacy awareness and personal information inference in generative AI systems.
[NLP-161] PII-TRACE: A Benchmark for Context-Aware PII Detection in Multi-Turn LLM Conversations
【速读】： 该论文旨在解决生成式 AI 系统在多轮对话中对个人身份信息（PII）检测能力不足的问题，尤其关注跨轮次上下文中的 PII 识别与完整覆盖。现有 PII 检测方法大多针对独立记录进行评估，忽视了对话过程中重复出现的标识符在不同轮次间的延续性，导致检测遗漏。为填补这一空白，研究提出首个面向多轮对话场景的 PII 基准测试 PII-TRACE（Tracing Recurring PII Across Conversational Exchanges），其包含 13,148 条合成多轮对话数据，覆盖 13 种语言，并提供字符级标注与标识符聚类信息，用于评估检测器在跨轮语境中识别并完整覆盖重复出现的敏感信息的能力。实验表明，包括前沿大模型在内的十一个基线系统均无法在不产生大量误报的前提下实现完整的实体级覆盖，且单次读取策略在长对话中会丢失约三分之一的正确字符。为解决此问题，研究进一步提出 PII-Tracer，一种参数量仅为 0.6B 的轻量级检测器，采用对话级监督训练策略，显著提升了跨轮次 PII 的识别完整性，在多轮对话和标准单记录基准上均表现出卓越性能。解决方案的关键在于引入对话级监督机制与对跨轮次连续性的建模能力，从而实现更精准、完整的 PII 检测。
链接: https://arxiv.org/abs/2609.22200
作者: Kaiyuan Zhang,Chuan Wang,Joey Zhong,Paul Fryzel,Kyle Polley,Jerry Ma,Ninghui Li
机构: Perplexity; Purdue University (普渡大学); Rutgers University (罗格斯大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI)
备注:
Abstract:LLM assistants and agentic systems log long multi-turn conversations. AI providers often scan these conversations for Personally Identifiable Information (PII) and mask the PII before storing or processing conversation data. Yet most PII detectors and benchmarks target self-contained records rather than cross-turn evaluation. To evaluate PII detection across turns in multi-turn conversations, we introduce PII-TRACE (Tracing Recurring PII Across Conversational Exchanges), to our knowledge the first PII benchmark to assess whether detectors identify PII in conversational contexts and cover every mention of a recurring identifier across turns. PII-TRACE contains 13,148 synthetic multi-turn dialogues in 13 languages with character-level spans and identifier clusters. Across eleven baselines, including frontier LLMs, no detector achieves full entity-level coverage without substantial false positives on PII-free conversations, and single-pass reading loses a third of the gold characters on long dialogues. To close this gap, we introduce PII-Tracer, a compact 0.6B-parameter detector trained with conversation-level supervision. PII-Tracer attains the highest entity-level coverage of any system we evaluate and also performs strongly on standard single-record benchmarks.
[NLP-162] EvoRank: LLM -Guided Evolution of Multi-Objective Learning-to-Rank Pipelines
【速读】： 该论文旨在解决多目标电商搜索排序中复杂学习排序（Learning-to-Rank, LTR）管道自动构建的难题，尤其在存在相关性、转化率与收益等多重竞争目标时，如何高效发现高性能且可解释的排序系统。其核心挑战在于：传统调参方法（如Optuna优化的LambdaMART）难以在多目标权衡下实现全局最优，而自动化设计流程易受训练目标选择中的“拟合噪声”（fitness noise）干扰，导致结果不可迁移。解决方案的关键在于提出EvoRank——一个基于大语言模型（LLM）引导的进化循环框架，通过迭代演化完整的排序管道（包括特征、模型、损失函数与集成策略），并引入“空间余量门控机制”（headroom gate）。该机制在执行前即可量化评估搜索空间的潜在提升余地与拟合噪声水平，从而预先判断进化循环是否具备实际收益。研究发现，仅优化训练目标的初始实验因无法克服拟合噪声而失败，但通过该门控机制可提前识别此类无效探索，并结合审计工具与失败模式库为团队提供可复用的工程化实践指南，显著提升了自动化排序系统设计的可靠性与可扩展性。
链接: https://arxiv.org/abs/2609.22196
作者: Rayhan Patel,Shabaz Patel
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Neural and Evolutionary Computing (cs.NE)
备注:
Abstract:We present EvoRank, an open autonomous ranking engineer: an LLM-guided evolutionary loop that discovers complete Learning-to-Rank pipelines (features, models, losses, ensembles) for multi-objective e-commerce search. On the Expedia ICDM 2013 dataset, with relevance, conversion, and revenue as competing objectives, three independent runs each converge within 50 iterations (about ten dollars) on interpretable pipelines that beat an Optuna-tuned LambdaMART on 60k held-out queries, an advantage that persists at full data scale and places in the top 6 percent of the original competition. A first campaign, evolving only training objectives, builds the central design rule: it appeared to work on its selection fold (the small dataset it uses to pick winners) while a transfer audit, re-scoring winners on held-out data, showed the gains were almost entirely fitness noise (the randomness of its own scoring), and neither seeded domain knowledge nor richer diagnostic feedback changed what transferred. The deciding quantity is measurable in advance: search-space headroom relative to fitness noise. We package this as a headroom gate that predicts, before any LLM spend, whether the loop will pay off, and we release the system, the auditing tools, and a catalog of failure modes with their guardrails, so teams can apply the procedure to their own ranking stacks.
[NLP-163] he Situated Identity Test: Distinguishing Persistent Cognitive Identity from Persona Imitation
【速读】： 该论文旨在解决大语言模型在角色扮演（persona prompting）中普遍存在的一种根本性归因问题：即使模型能够流畅地模仿特定身份的言行举止，表现出与真实人生经历一致的对话风格和记忆，但其行为可能并未真正基于该身份的实际发展轨迹。核心问题在于，多个不同背景的个体可能共享相同的公开身份特征（如年龄、籍贯、职业、性格等），而模型仅依赖这些压缩后的身份信息进行推理时，无法区分其行为应归属于哪一条真实的生命历程。解决方案的关键在于提出一种架构无关的评估框架——情境化身份测试（Situated Identity Test, SIT），该框架通过检验代理行为是否可被功能性地归因于某一特定发展轨迹，来验证其身份的真实性。该方法强调“情境化身份”需同时具备对已记录经验的恰当知识以及对未实际经历事件的合理无知，并严格受限于该身份实际获取的信息范围，而非底层基础模型的知识库。研究证明，仅依赖压缩身份信息的策略，在面对m条冲突生命线的情况下，其平均情境化有效性上限为1/m（成对情形下不超过50%）。作者进一步构建了SITBench评测套件，涵盖25组身份冲突对（共50条不同生命线）、10,000个预设探测项及九种架构配置，结合开源参考实现与确定性测试用例，对前沿基础模型（如GPT-5.6 Sol和Claude Opus 5）进行了实证评估，系统揭示了角色提示在身份冲突下的失效模式，并为评估情景连续性、结构化状态和认知边界提供了可信赖的保障机制。
链接: https://arxiv.org/abs/2609.22195
作者: Jun He,Deying Yu
机构: OpenKedge.io
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 28 pages, 1 figure, 9 tables; open-source benchmark and reference implementation available at this https URL
Abstract:Large language models can convincingly adopt personas, recall past dialogues, and weave rich autobiographies. Yet this conversational eloquence conceals a fundamental attribution problem: looking the part does not mean having lived the life. Two individuals can share identical public profiles–the same age, hometown, occupation, and personality traits–while possessing entirely distinct private histories, relationships, and acquired skills. When conditioned solely on that shared profile, an agent lacks the information required to determine which lineage is correct. We introduce the Situated Identity Test (SIT), an architecture-independent framework that evaluates whether an agent’s behavior is functionally attributable to a specific developmental lineage. Grounded identity requires both appropriate knowledge of recorded experiences and appropriate ignorance of ungrounded ones, bounded by what the identity has actually acquired rather than what its underlying foundation model knows. We prove that any policy conditioned solely on a compressed profile is bounded by an average situated validity of at most 1/m across m colliding life histories on lineage-discriminative queries (at most 50% for paired lineages). We instantiate this framework in SITBench, an evaluation suite designed for 25 profile-collision pairs (50 distinct lineages) across 10,000 planned probes and nine architectural configurations. Supported by an open-source reference implementation, deterministic test fixtures, and empirical pilot evaluations on frontier foundation models (GPT-5.6 Sol and Claude Opus 5), we formalize the failure modes of persona prompting under profile collision and provide an assurance harness for evaluating episodic continuity, structured state, and epistemic boundaries.
[NLP-164] Fairness Beyond Anonymization? Demographic Leakage in German LLM -Generated Resumes KDD2026 KDD ECML
【速读】： 该论文旨在解决生成式人工智能（Generative AI）在多语言招聘场景中，尤其是德语简历生成过程中存在的隐性人口统计学信息泄露问题。尽管简历在生成后经过匿名化与性别中性化处理，但其文本仍可能蕴含可被下游分类器有效识别的性别或种族特征信号，从而违背欧盟《人工智能法案》对高风险应用领域所要求的公平性与透明性。该研究的关键解决方案在于设计并执行一个两阶段审计框架：首先利用多种主流大语言模型（包括ChatGPT、Gemini 2.5 Flash-Lite及Qwen 3系列不同规模模型）基于真实匿名职位匹配资料生成简历，并系统性地变换姓名以模拟性别与族裔差异；其次，在简历完成匿名化与性别中性化处理后，训练专门的泄漏分类器来检测残留的人口统计学信号。研究发现，即便经过上述干预，分类器仍能可靠区分由男性与女性姓名生成的简历，其根源并非显性的性别化用词，而是德语中语义等价但形式上中性的词汇在使用频率和分布上的细微差异，揭示了“看似中性”的生成内容仍可能保留高度预测性的性别信号。相比之下，族裔相关泄漏则相对较弱。这一发现表明，依赖于表面匿名化的公平性策略在多语言生成式招聘系统中存在根本性局限，亟需更深层次的去偏技术与算法监管。
链接: https://arxiv.org/abs/2609.22188
作者: Charlotte Leininger,Helena Veit,Matthias Aßenmacher,Andreas Bender
机构: LMU Munich(慕尼黑大学); Munich Center for Machine Learning (MCML)(慕尼黑机器学习中心)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Accepted at ECML PKDD 2026 Workshop on Bias and Fairness in AI [BIAS 2026@ ECML/PKDD]
Abstract:Large language models (LLMs) are increasingly integrated into AI-assisted hiring pipelines, including automated resume generation and screening. Under the EU AI Act, the hiring domain is classified as high-risk, making fairness and transparency critical requirements. Existing work has primarily focused on explicit hiring decisions, while less attention has been paid to whether generated resumes themselves encode recoverable demographic information. In this work, we conduct a two-stage audit of demographic leakage in German-language LLM-generated resumes. First, we use ChatGPT (GPT-4o-mini), Gemini 2.5 Flash-Lite, and multiple scales of the open-weight Qwen 3 model family (4B, 8B, and 14B) to generate resumes from real anonymized job-matching profiles, systematically varying gender- and ethnicity-associated names while holding qualifications constant. Second, we simulate a downstream resume screening scenario, where the generated resumes are first anonymized and gender-neutralized, before demographic leakage classifiers are trained on the resulting texts. We find that, despite these interventions, classifiers reliably distinguish between resumes generated with male and female names. This leakage is not driven by overtly gendered wording, but by subtle differences in the usage of semantically equivalent, formally gender-neutral terms in German. In contrast, ethnicity-related leakage remains comparatively weak across models. Our findings demonstrate that apparently neutral resume generation can still preserve highly predictive demographic signals, raising concerns about anonymization-based fairness interventions in multilingual AI hiring pipelines.
[NLP-165] Beyond Task Completion: Training Capable and Safe Computer-Use Agents
【速读】： 该论文旨在解决当前计算机使用代理（Computer-use Agents, CUAs）在完成复杂任务时缺乏可靠安全行为的问题。尽管现有方法侧重于任务成功率的后训练优化，但无法有效引导代理在面对潜在风险时做出安全决策。解决方案的关键在于提出一种联合优化框架——安全与能力协同优化（Safety and Capability Optimization for Policy Execution, SCOPE），通过同时提升代理的任务执行能力与安全感知决策能力，使其能够根据风险条件动态调整行为：在无害环境下正常执行任务、规避环境威胁，并在存在安全路径时继续推进，而在目标本身有害或无安全路径时主动拒绝执行。为实现这一目标，研究进一步设计了自动化数据生成管道SCOPE-Gen，可自动生成可验证的能力任务并转换为包含环境风险的变体，从而构建出包含能力示范、安全延续和明确拒绝行为的轨迹数据集SATraj-OS。在此基础上，SCOPE采用监督微调结合在线强化学习的方法，先从三类轨迹中学习，再通过强化学习优化任务完成率。实验表明，基于Qwen3.5-9B的SCOPE-RL在OSWorld上达到54.17%的任务成功率，在OS-BLIND上实现64.30%的攻击规避率，综合能力-安全得分达58.80%，优于所有对比代理。消融实验揭示了两种安全监督形式的非对称但互补作用：拒绝轨迹主要贡献于攻击规避性能提升，而风险处理轨迹则在保持相近攻击规避水平的同时显著提升了任务可用性。
链接: https://arxiv.org/abs/2609.22178
作者: Zeyu Kang,Zhenyun Yin,Yang Zhang,Shan He,Shanzhe Lei,Yanjiu Zhong,Xinquan Chen,Yuhong Wang
机构: 未知
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注:
Abstract:Computer-use agents (CUAs) have made rapid progress in completing complex tasks through graphical user interfaces, yet post-training centered on task success alone does not induce reliable safety behavior. A reliable CUA must condition its execution on risk: it should complete ordinary benign tasks, avoid environmental hazards and continue when a safe completion path remains, and refuse when the goal is harmful or no safe path exists. To learn this conditional policy, we develop Safety and Capability Optimization for Policy Execution (SCOPE), which jointly post-trains a CUA for task-execution capability and safety-aware decision making. To provide aligned training data for this joint objective, we further introduce SCOPE-Gen, an automated pipeline that synthesizes verifiable capability tasks and converts them into paired environment-risk variants while preserving their original goals. Using the resulting tasks, we construct SATraj-OS, a trajectory dataset comprising capability demonstrations, safe continuations, and explicit refusals. SCOPE first learns from all three trajectory types through supervised fine-tuning and then further improves task completion through online reinforcement learning. Starting from Qwen3.5-9B, SCOPE-RL achieves a 54.17% task success rate on OSWorld and a 64.30% attack-avoidance rate on OS-BLIND, yielding the best aggregate capability–safety score of 58.80% among the evaluated agents. Ablations reveal asymmetric but complementary roles for the two forms of safety supervision: refusal trajectories account for most of the attack-avoidance gain, whereas risk-handling trajectories preserve greater task utility at comparable attack-avoidance levels.
[NLP-166] Multiple latent orderings better predict language model preferences
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在进行价值判断与选择时普遍存在的非传递性（intransitivity）问题。传统方法将这种不一致性视为单一潜在排序下的采样噪声，而本文提出，非传递性实际上反映了多个内部一致的潜在偏好顺序的聚合。其解决方案的关键在于引入一种噪声增强的混合布拉德利-特瑞（Mixture Bradley-Terry, MBT）模型，通过重复的成对比较推断出多个隐含的偏好成分。实验结果表明，在七个模型和四个任务中，基于多重偏好的混合模型显著优于单一效用模型，能够更准确地解释结构性不一致。研究进一步揭示，总体偏好常掩盖了深层的偏好异质性；以道德机器（Moral Machine）困境为例，即使模型在整体排序上存在分歧，仍可能共享相同的潜在偏好成分。这些发现表明，LLMs 反映的是多元偏好（plural preferences），因此，当前依赖单一函数形式的对齐与评估范式，可能无意中平均掉了不同用户所支持的若干合理且一致的偏好结构，从而导致对模型真实决策机制的误判。
链接: https://arxiv.org/abs/2609.22170
作者: Aviral Chawla,William H.W. Thompson,Jean-Gabriel Young
机构: Vermont Complex Systems Institute(佛蒙特复杂系统研究所); University of Vermont(佛蒙特大学)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computation and Language (cs.CL)
备注: 30 pages, 12 figures, 6 tables
Abstract:Language models are frequently employed in settings where they are asked to make value judgments and choices. These observed choices often exhibit intransitivity: A model may prefer item A to B and B to C , while also preferring C to A . Existing work that models LLM preferences treats such inconsistencies as sampling noise around a single latent ordering. We instead propose that intransitivity reflects the aggregation of multiple latent, internally consistent orderings. We first show that observed inconsistencies cannot be explained by a single ordering under any monotone link function. We then introduce a noise-augmented mixture Bradley-Terry (MBT) model that infers latent preference components from repeated pairwise comparisons. Across seven models and four tasks, a mixture of orderings often explains structural inconsistencies better than single-utility models. We find that aggregate preferences often hide underlying preference heterogeneity. A case study on Moral Machine dilemmas shows that models which disagree on aggregate orderings can still share latent components. Together, these results suggest that LLMs reflect plural preferences. Alignment and evaluation pipelines that treat LLM preferences as a single function, therefore, risk averaging over coherent orderings that different users may endorse differently.
[NLP-167] Monocultural Biases: Correlated biases in large language models lead to unequal systemic exclusion rates in hiring
【速读】： 该论文旨在解决生成式 AI（Generative AI）在招聘场景中广泛应用所引发的“单一文化偏见”（monocultural biases）问题，即大规模语言模型（LLMs）的普及可能导致劳动力市场中的偏见趋同，进而加剧对特定人口群体的系统性排斥。其解决方案的关键在于识别导致此类偏见的核心阶段：通过对比十种大型语言模型在其基础版本（base models）与后训练版本（post-trained models）之间的招聘决策差异，发现后训练阶段显著放大了年龄相关的歧视，使年长申请者的回访概率平均降低3.6%，且该负面效应出现在八种模型中。研究进一步表明，后训练模型间的决策一致性显著提高，这种高度协同性主要由人力资本特征（如技能或专业背景）驱动，但反而导致全球系统性排斥率从5.6%上升至17.3%，交叉性排斥率高达12.2%至21.7%。因此，尽管后训练能提升模型筛选最优候选人的能力，却可能通过统一引入新的偏见，显著增加边缘群体的系统性不平等风险。
链接: https://arxiv.org/abs/2609.22169
作者: Matthew Bone,Fabian Stephany,Maria del Rio-Chanona
机构: University of Oxford (牛津大学); Burning Glass Institute; University College London (伦敦大学学院)
类目: Computation and Language (cs.CL); Computers and Society (cs.CY); General Economics (econ.GN)
备注: Accepted at COLM 2026
Abstract:Employers are increasingly using large language models (LLMs) to automate their hiring process. This paper investigates the risk of monocultural biases, in which the widespread deployment of large language models homogenizes biases across the labor market, leading to greater systemic exclusion for certain demographic groups. For ten LLMs, we measure hiring biases across their base and post-trained versions to identify which stage, pre-training or post-training, lead to monocultural biases. We find that, compared to their base models, post-trained models are 3.6% less likely to callback older applicants. This negative shift occurs in eight of the ten models that we evaluate. Post-trained models have much more correlated decisions than base models which is likely driven by human capital traits like skills or college major. However, greater consensus among models increases global systemic exclusion rates from 5.6% to 17.3% and exacerbates demographic inequalities, with intersectional systemic exclusion rates ranging from 12.2% to 21.7% for post-trained models. We find that this inequality is primarily driven by age-based discrimination that is exacerbated in post-training. These results indicate that while post-training techniques may improve models’ abilities to select the best applicants, they may raise systemic inequality risks for those at the margin by uniformly introducing new biases.
[NLP-168] A Multi-Agent Pipeline for Source-Grounded Synthetic Note Generation from Longitudinal Structured EHR ICML2026
【速读】： 该论文旨在解决结构化电子健康记录（Electronic Health Record, EHR）数据虽丰富但存在稀疏性、编码化程度高且难以直接用于以临床笔记为中心的建模问题。其核心挑战在于如何将低维度、非自然语言形式的结构化EHR数据转化为高质量、符合临床真实性的自由文本笔记表示。解决方案的关键在于提出MedNotes——一个基于多智能体的合成数据生成流水线，将结构化EHR到文本的生成过程设计为闭环式智能体协作系统：由生成器提出初步笔记，评估智能体分别诊断事实一致性、覆盖度、结构合理性和幻觉风险等缺陷，再通过路由器决定采纳、修正或拒绝该草案。该方法在1,485个EHRSHOT临床会诊实例上实现了91.4%的通过率，平均事实准确性达0.980，完整性为99.1%，结构保真度为0.761，每例仅含0.028个关键幻觉。通过迭代优化，接受率从69.4%提升至91.4%。最终生成的合成数据集在与少量真实数据结合时，显著提升了下游临床程序代码（CPT）预测和段落级章节预测性能，验证了其在临床自然语言处理任务中的有效性。
链接: https://arxiv.org/abs/2609.22164
作者: Nina Fatehi,Reihaneh Hassanzadeh,Meysam Ghaffari,Animesh Agarwal,Carlos Morato
机构: Optum AI, UnitedHealth Group(联合健康集团), Minneapolis, Minnesota, USA
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Accepted at the FMSD Workshop, ICML 2026
Abstract:Structured EHR is abundant but sparse, coded, and difficult to use directly for note-centric clinical modeling. We present MedNotes, a multi-agent synthetic data generation pipeline that converts longitudinal structured EHR into source-grounded clinical note representations under explicit quality control. MedNotes treats structured-data-to-text synthesis as a closed-loop agentic process: a generator proposes a note, evaluator agents diagnose factual, coverage, structural, and hallucination-related failures, and a router accepts, revises, or rejects the draft. On 1,485 EHRSHOT encounters, MedNotes achieves a 91.4% pass rate, with mean factual accuracy of 0.980, completeness of 99.1%, structural fidelity of 0.761, and 0.028 critical hallucinations per encounter. Iterative refinement improves acceptance from 69.4% to 91.4%. The resulting synthetic corpus improves downstream CPT prediction and paragraph-level section prediction when combined with limited real data.
[NLP-169] MechaTerp-TRACE: A Novel Approach for Component Ablation Analysis in Language Models
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）中不同架构组件对输出的因果贡献难以统一比较的问题。现有研究虽分别揭示了前馈层中的事实回忆机制与自注意力机制中的词元关系建模，但缺乏一种系统性方法来量化各组件（从完整变换器块到单个神经元及输出logits）对特定输出（如命名实体生成）的贡献。为此，作者提出MechaTerp（机制可解释性套件）——其子集TRACE（教师强制型组件消融注册表），通过逐个消融模型组件并测量在固定答案位置上的输出分布变化，实现对各类组件贡献的标准化比较。实验覆盖13个指令微调的密集解码模型（参数量5亿至300亿），涵盖48个医学与42个通用知识类提示，共消融49,656个组件。结果表明：无论所问实体为何，最具影响力的组件始终是少数几个位置固定的模块；一旦这些核心组件被移除，其余组件的贡献在11个模型中趋于均匀分布。这一发现表明，命名实体知识的表观局部化主要源于通用生成机制，而非集中存储于可定位的特定区域，这对依赖“知识可定位性”的靶向知识编辑等方法提出了直接挑战。
链接: https://arxiv.org/abs/2609.22163
作者: Brandon Colelough,Davis Bartels,Madeline Bittner,Dina Demner-Fushman
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Interpretability research on large language models has produced accounts of factual recall in feed-forward layers and of token relationships in self-attention, but little work offers a unified way to compare the causal contribution of different architecture components to a model’s output. We introduce MechaTerp (the Mechanistic Interpretability suite) -TRACE (subset for Teacher-forced Registry of Ablated Component Effects), an architecture and study that measures how much each registered component of a language model supports the production of a named entity. TRACE ablates one component at a time and measures the resulting change in the output distribution at a fixed answer token, so component types from whole transformer blocks down to individual neurons and output logits can be compared on a common scale. We apply it to thirteen instruction-tuned dense decoder models spanning five families and one to thirty billion parameters, ablating 49,656 components across 48 medical and 42 general-knowledge prompts. We find that the components carrying the most effect are the same few, positionally fixed components in every model, regardless of which entity a prompt asks about, and that once these are removed, the remaining support is close to evenly spread in eleven of the thirteen models. Apparent localisation of entity knowledge is therefore largely attributable to generic generation machinery, which has direct consequences for methods that assume entity knowledge sits in a findable place, including targeted knowledge editing.
[NLP-170] Beyond Raw Context Transfer: Representation-based Federated Retrieval-Augmented Generation
【速读】： 该论文旨在解决在敏感领域（如医疗健康）中，传统检索增强生成（Retrieval-Augmented Generation, RAG）框架因依赖集中式知识库而难以应用的问题。现有去中心化RAG方法多采用基于提示（prompt-based）的范式，需在客户端之间传输原始、可读的检索内容，导致推理阶段计算开销大且存在信息泄露风险。其解决方案的关键在于提出一种基于表示的联邦检索增强生成（Representation-based Federated RAG, FedRepRAG）框架：该框架将原始文档保留在所属客户端，仅在跨客户端检索过程中交换紧凑的潜在表示（latent representations），并通过一个协作训练的投影器（projector）将检索嵌入转换为适配冻结的大语言模型（LLM）或视觉-语言模型（VLM）主干的表示令牌，从而实现知识融合。实验表明，FedRepRAG在去中心化视觉问答（VQA）与问答（QA）任务上均显著优于直接推理和本地检索基线，同时大幅降低检索上下文长度与推理时计算开销，并通过分析验证了查询相关表示的重要性及表示交换带来的残余信息泄露特性，整体提供了一种无需传输原始检索内容的高效、安全的联邦RAG新范式。
链接: https://arxiv.org/abs/2609.22162
作者: Can Peng,Yu Liu,Yingyu Yang,Anjie Le,Yuyuan Liu,Qianye Yang,J. Alison Noble
机构: University of Oxford(牛津大学); National University of Singapore(新加坡国立大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Retrieval-augmented generation (RAG) improves the factuality of large language models (LLMs) and vision-language models (VLMs) by grounding generation in external knowledge. However, most existing RAG frameworks assume a centralized retrieval corpus, which is often impractical in sensitive domains such as healthcare, where data are inherently distributed and raw content cannot be directly shared across institutions. Recent efforts on decentralized RAG primarily follow prompt-based paradigms that exchange raw, human-readable retrieved content, leading to substantial inference-time computational overhead and direct exposure of retrieved information. To address these limitations, we propose Representation-based Federated RAG (FedRepRAG), a decentralized RAG framework that keeps raw documents at their owning clients and exchanges only compact latent representations during cross-client retrieval. To integrate retrieved knowledge, we introduce a collaboratively trained projector that converts retrieval embeddings into generator-compatible representation tokens for a frozen LLM/VLM backbone. Experiments across decentralized visual question answering (VQA) and question answering (QA) benchmarks show that FedRepRAG consistently outperforms direct inference and local retrieval baselines while substantially reducing retrieval-context length and inference-time computational overhead compared with raw-context transfer. Further analyses confirm the importance of query-relevant retrieved representations and characterize the residual representation-level leakage associated with representation exchange. Overall, FedRepRAG provides an effective and efficient framework for federated RAG without transferring raw retrieved content.
[NLP-171] Didactic knowledge or Clinical Cases? How Data Types Shape Medical Large Language Models NLPCC2026
【速读】： 该论文旨在解决医学大语言模型（Medical Large Language Models, LLMs）在混合使用教学类数据（如教科书）与临床数据（如电子健康记录，EHR）进行训练时，两类数据对模型能力的差异化影响尚不明确的问题。其核心解决方案是通过设计词元匹配（token-matched）实验，系统性地调节教学数据与临床数据的比例，进而分析数据构成如何影响模型在知识密集型任务与临床导向任务中的表现、能力特征及错误模式。研究发现，任务类型间存在非对称迁移现象：临床数据显著提升临床推理类任务性能，同时在知识型任务上保持竞争力；而教学数据主要增强知识型任务表现。进一步的错误分析揭示了“知行差距”（knowing-doing gap），即知识记忆能力的提升并未可靠转化为临床推理能力。此外，研究发现少量临床数据即可带来电子健康记录（EHR）相关任务的主要性能提升，且最优数据比例随下游任务的知识需求与临床推理强度动态变化。因此，研究提出医学LLM的数据构建应以应用为导向，对于推理密集型场景，应优先采用更高比例的临床数据。
链接: https://arxiv.org/abs/2609.22161
作者: Yuzheng Fan,Haochun Wang,Sendong Zhao,Xiao Han,Ming Ma,Bing Qin
机构: Harbin Institute of Technology (哈尔滨工业大学)
类目: Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: This paper is accepted by NLPCC 2026 oral
Abstract:Medical large language models are commonly trained on mixtures of didactic data (e.g., textbooks) and clinical data (e.g., patient records), yet how these data types differentially shape model capabilities remains unclear. We address this issue with token-matched experiments that vary the didactic-to-clinical ratio and analyze how data composition affects performance, capability profiles, and error patterns across knowledge-intensive and clinic-oriented tasks. We uncover an asymmetric transfer across task types: clinical data improves clinic-oriented tasks while remaining competitive on knowledge-intensive ones, whereas didactic data mainly improves knowledge-intensive tasks. Error analysis suggests a knowing-doing gap, where improvements in knowledge recall do not reliably generalize to clinical reasoning. We further observe that modest amounts of clinical data yield most of the gains on EHR-grounded tasks, while the optimal mixture ratio varies with the knowledge and clinical reasoning demands of downstream tasks. These findings suggest that medical LLM data curation should be application-driven, with higher proportions of clinical data preferred for reasoning-intensive use cases.
[NLP-172] PAGE: Partition-Aware Gated KV-Cache Eviction
【速读】： 该论文旨在解决大语言模型推理过程中因KV缓存（KV-cache）淘汰策略导致的精度灾难性下降问题，尤其关注现有方法在特定输入类别下无法识别“淘汰是否安全”的缺陷。传统淘汰方法仅决定保留哪些token，而不判断是否应进行淘汰，从而在某些输入上可能将准确率从99%骤降至0%。为此，论文提出将淘汰决策重构为基于每输入的准入（admission）判断，并发现输入可划分为两类：容量受限类（capacity-bound），其在任何缓存预算下淘汰均造成灾难性损失；以及稀释敏感类（dilution-prone），淘汰在此类中可能安全甚至有益。解决方案的关键在于引入一个无需标签的标量指标——前向注意力阶段中成对top-k头一致性从早期到晚期的下降幅度（early-to-late drop in pairwise top-k head agreement），该指标可在解码前即预测输入的淘汰安全性。基于此，提出PAGE（Per-Input Admission Gate for Eviction）机制：当该下降幅度较大时，启用基础淘汰器；否则保留完整缓存，且无需训练或精度标签。该方法在四个架构族中一致排序输入的淘汰安全性，仅需约100个无标签样本即可为新模型校准阈值。作为安全防护机制，PAGE将容量受限场景下的危害率从0.75降至0.026（降低29倍），在四类淘汰器、四模型及两个基准测试中，成功将99%→0%的精度崩溃转变为稳定89%的平滑性能，且不需重新训练淘汰器。由于该门控机制仅在淘汰风险高时激活，而在安全场景下保持静默，因此其收益是针对少数高危输入的精准安全增强，而非整体平均性能提升。值得注意的是，PAGE并非压缩算法，其实现压缩比仅为1.8–3.4×（均值2.9×），远低于名义16×预算，且在静态资源分配下随批处理规模增加趋于1，而训练过的淘汰器在相同内存条件下仍具优势。
链接: https://arxiv.org/abs/2609.22157
作者: Pankaj Kumar,Subhankar Mishra
机构: National Institute of Science Education and Research (国家科学教育与研究学院); An OCC of Homi Bhabha National Institute, India (霍米·巴巴国家研究所的一个附属研究中心)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL); Cryptography and Security (cs.CR)
备注:
Abstract:KV-cache eviction methods decide which tokens to keep but not whether to evict at all, so a benchmark mean can hide a class of inputs on which compression drives accuracy from 99% to 0%. We reframe eviction as a per-input admission decision and show that inputs separate into a capacity-bound class, where eviction is catastrophic at every budget, and a dilution-prone class, where eviction is safe or beneficial. A single label-free scalar computed from prefill attention, the early-to-late drop in pairwise top- k head agreement, predicts this class before any decoding. PAGE thresholds this drop: it applies any base evictor when the drop is large and retains the full cache otherwise, with no training and no accuracy labels. The drop orders inputs by eviction safety consistently across four architecture families, and a per-model unlabeled pilot of about 100 inputs recalibrates the threshold for a new family. Used as a safeguard, PAGE cuts the harm rate on the capacity-bound regime from 0.75 to 0.026, a 29 \times reduction, across four evictors, four models, and two benchmarks, turning a 99% to 0% collapse into a flat 89% without retraining the evictor. The gate is inert wherever eviction is already safe, and the capacity-bound class it protects is a small, identifiable minority of inputs, so the benefit is a targeted safety gain rather than an average one. PAGE is a per-input safeguard, not a compressor: realized compression is 1.8 - 3.4 \times (mean 2.9 \times ) against a nominal 16 \times budget and decays toward unity by batch 16 under static provisioning, and a trained evictor wins at matched memory.
[NLP-173] Is Imagination Derived from Hallucination? A Cross-Taxonomy Evaluation of Imagination and Hallucination in Large Language Models
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在评估想象力（imagination）时存在的核心问题：现有创造力评测基准仅衡量输出与常见答案的偏离程度，却未检验这种偏离是否合理响应了提示（prompt）的语义支持，且将幻觉（hallucination）与想象力割裂评估，导致二者是否源自同一生成机制这一关键假设无法直接验证。其解决方案的关键在于提出首个系统性的想象力评测基准Whiteboard，该基准基于人类想象力的经典认知测量工具，将七类基于生成机制的想象力子类型与十类受支持边界约束的幻觉子类型进行交叉设计，并在同一生成过程中联合评分。Whiteboard通过显式的支持性检查机制，利用可审计的原子矩阵（atom matrix）确定性地计算想象力与幻觉两个维度的得分，全程不依赖大语言模型作为裁判。该基准包含1,660个提示，其共享的80项锚点集用于评估79个主流LLM，并以13,280次人类判断进行验证。进一步分析揭示了一个反直觉现象：幻觉与想象力之间普遍存在负相关关系，所有锚点项均独立复现此负耦合，暗示二者可能源于不同的生成倾向或存在内在抑制机制。
链接: https://arxiv.org/abs/2609.22152
作者: Zixuan Tang,Hongzong Li,Shuxin Zhuang,Dapeng Wu,Zi Liang
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Imagination performs as a high-level function of large language models (LLMs) which determines the potential of how an LLM creates unseen or creative content. While existing works have built a rich family of creativity benchmarks for this ability, they only measure how far an output departs from common answers and never check whether the departure is licensed by the prompt. Moreover, hallucination, the closest neighbor of imagination, is always measured in a separate pipeline on different generations, so the influential claim that imagination and hallucination stem from the same generative mechanism has never been directly testable. In this paper, we propose Whiteboard, the first LLM imagination evaluation benchmark. Its design follows the authoritative cognitive instruments developed to measure human imagination: seven mechanism-grounded imagination subtypes are adapted from classic paradigms, then crossed with ten support-boundary hallucination subtypes and scored jointly on the same generation. Different from previous creativity or hallucination benchmarks, Whiteboard gates every imagination score with an explicit support check and computes both axes deterministically through an auditable atom matrix, with no LLM judge on the primary path. The full Whiteboard item bank contains 1,660 prompts; on its shared 80-item anchor set, we evaluate 79 state-of-the-art LLMs and validate the instrument against 13,280 human judgments. Additionally, we further explore whether imagination derives from the same generative tendency as hallucination and what key factors shape it. Our analysis indicates a counterintuitive correlation between hallucination and imagination: Most of the subtype couplings are negative, every one of the anchor items reproduces the negative coupling on its own.
[NLP-174] Do Language Models Know Their Own Constraints?
【速读】： 该论文旨在解决生成式模型在经过后训练（post-training）行为约束后，其对所学习约束的显式可报告性（explicit reportability）是否仍能保持的问题。具体而言，研究关注的是：当通过LoRA微调使Llama 3.1 8B Instruct模型遵守禁止使用特定食材的规则时，模型是否能够准确地在请求下主动报告这些被禁止的成分。解决方案的关键在于对比监督微调（SFT）与基于组相对策略优化（GRPO）两种方法在行为合规性与显式报告能力之间的权衡。结果显示，尽管两种方法均将行为合规率从基线4%提升至约90%，但显著降低了模型对约束条件的显式报告能力（分别降至0.16/5和0.07/5），并严重侵蚀了模型对第三人称知识的保留（从93%降至36%和14%）。研究发现，基于奖励信号的方法更具破坏性——无论是否考虑语境，惩罚性奖励均导致模型习得一种上下文无关的抑制机制，而非自我导向的内化约束；而设计用于区分“自我”与“他人”的上下文感知奖励也失败，导致模型在所有语境下均倾向于包含禁用成分。通过对提示时隐藏状态的探针分析，虽在第24层MLP中恢复了83.8%的单个成分规避能力，但仅比基于基线的预测器高出6.4个百分点，且模型自身的语言自述（verbal self-report）准确率更高（87.8%）。此外，添加显式自我描述示例的正向对照实验未能恢复报告能力，表明问题并非普遍性的知识丧失，而是针对“按请求枚举约束”这一特定任务的特异性失效。
链接: https://arxiv.org/abs/2609.22151
作者: Arin Agarwal
机构: Columbia University (哥伦比亚大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 8 pages
Abstract:We ask whether behavioral constraints acquired through post training remain explicitly reportable. Using constrained recipe generation as a testbed, five banned ingredients enforced via LoRA fine tuning of Llama 3.1 8B Instruct we compare supervised fine tuning (SFT) and Group Relative Policy Optimization (GRPO) against an untrained baseline on a four tier Constraint Awareness Benchmark. Averaged over three seeds, both methods raise behavioral compliance from 4% to about 90% while reducing explicit constraint reporting below the untrained model (0.48/5 to 0.16/5 for SFT, 0.07/5 for GRPO) and eroding retained third person knowledge (93% to 36% for SFT, 14% for GRPO; p less than 0.01 between methods). Contrary to our initial hypothesis, the reward based signal is the more destructive of the two: a reward that penalizes banned ingredient tokens regardless of framing learns a context independent suppression rather than a self directed constraint. A context conditioned reward designed to teach the self to other distinction fails, collapsing toward inclusion in both framings. Probing prompt time hidden states recovers per ingredient avoidance at 83.8% (layer 24 MLP), but only 6.4 points above a per ingredient base rate predictor (77.4%), and the model’s own verbal self report is more accurate still (87.8%). A positive control adding explicit self description examples does not restore reporting. The failure is therefore specific to enumerating constraints on request, not a general loss of access to them.
[NLP-175] Beyond the Stitching Assumption: A Unified Framework for Multimodal Synthetic Data Evaluation via Semantic Quantization
【速读】： 该论文旨在解决多模态合成数据（multimodal synthetic data）中跨模态关联性评估不足的问题，即现有评估方法通常仅对表格与文本模态分别进行评价，导致即使表格与文本之间的配对关系被破坏，某些指标仍可能保持较高值，从而无法真实反映合成数据的跨模态一致性。其解决方案的关键在于提出一种基于投影（projection-based）的评估框架：通过固定句向量编码器（sentence encoder）将文本映射为嵌入表示，利用k-means聚类将其转换为簇状态，同时将表格变量表示为类别或分位数分箱状态，进而构建真实与合成数据的联合状态分布；通过计算Jensen-Shannon散度（JSD）、归一化互信息（NMI）、条件JSD（cJSD）及联合状态熵等统计量，实现对跨模态依赖结构的量化评估。此外，引入文本到属性的实用性（T2A utility）和校准后的邻近标志率（PFR）作为表征层面的诊断指标，并设计文本置换控制（text-permutation control）以保留边际分布但破坏配对关系，从而建立基线对比。实验结果表明，仅依赖模态特异性指标会掩盖跨模态关联的失效，而投影诊断可有效检测出真实依赖性与置换基线之间的差异，尽管在弱依赖或稀疏投影场景下表现有限。研究还发现，部分基于大语言模型（LLM）的生成基线反而表现出比真实数据投影更强的测量依赖性，凸显了显式跨模态评估与置换基线对照的重要性。
链接: https://arxiv.org/abs/2609.22149
作者: Yefeng Yuan,Zhan Shi,Liang Cheng,Yuhong Liu
机构: Santa Clara University (圣克拉拉大学); eBay Inc. (eBay公司)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 10 Pages, 2 figures, 4 tables, accepted to PST 2026
Abstract:Multimodal synthetic datasets combine structured attributes with free text, but are often evaluated separately. Such metrics can remain high after tabular–text pairings are disrupted. We present a projection-based evaluator for tabular–text synthetic data. A fixed sentence encoder maps text to embeddings, (k)-means converts them to cluster states, and tabular variables are represented as categorical or quantile-binned states. Real and synthetic contingency tables are compared using Jensen–Shannon divergence (JSD), normalized mutual information (NMI), conditional JSD (cJSD), and joint-state entropy. We also report text-to-attribute (T2A) utility and a holdout-calibrated proximity flag rate (PFR) as a representation-level diagnostic. A text-permutation control preserves both marginal distributions while disrupting their pairing. Experiments on Amazon Reviews, Kiva Loans, and the Employment Scam Aegean Dataset show that modality-specific scores remain high under this control. The projection diagnostics detect disruption when real projected dependence exceeds a permutation baseline, but are less informative for weak or sparse projections. Some conditioned LLM baselines also exhibit stronger measured dependence than the corresponding real-data projections. These results support explicit cross-modal evaluation with permutation baselines and coverage reporting.
[NLP-176] GRRR: The Geometry of Reshaping Rotation and Routing in Decoder LLM post-training
【速读】： 该论文旨在解决后训练（post-training）过程如何改变大语言模型（Large Language Models, LLMs）权重相对于预训练权重的内在机制问题。其核心挑战在于理解在监督微调（Supervised Fine-Tuning, SFT）与强化学习（Reinforcement Learning, RL）等后训练流程中，模型权重更新的本质几何意义。研究的关键在于将每一轮权重更新置于预训练权重矩阵的奇异值分解（Singular Value Decomposition, SVD）框架下进行分析，从而将权重变化解耦为三个几何上独立的分量：对角项（diagonal values），负责重塑奇异值；非对角项（off-diagonal values），实现预训练输入与输出方向间耦合关系的旋转；以及零空间项（null-space values），使信息流扩展至原预训练矩阵非零奇异值核心之外的子空间。实验结果表明，在数学评估任务中，移除对角项通常不会显著损失后训练带来的性能提升，这表明后训练所获得的性能增益主要源于对预训练路径的重构与扩展，而非对预训练模型奇异值的大幅修改。
链接: https://arxiv.org/abs/2609.22146
作者: Jianing Qi,Hao Tang,Zhigang Zhu
机构: CUNY Graduate Center (纽约市立大学研究生院); The City College of New York (纽约城市学院)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:We study how post-training changes the weights of Large Language Models (LLMs) relative to their pretrained weights. Across 12 post-training chains with supervised fine-tuning (SFT) and reinforcement learning (RL), we express each weight update in the pretrained matrix’s singular value decomposition (SVD) frame. This decomposition separates the changes of three geometrically distinct components: diagonal values, which reshapes singular values; off-diagonal values, which rotates the coupling between pretrained input and output directions; and null-space values, which routes outside the matrix’s original nonzero SVD core. On a math evaluation suite, we find that removing the diagonal component usually preserves most of the gains from post-training. These results suggest that post-training gains are carried primarily by reconfiguring and extending pretrained pathways rather than by substantially changing singular values of pre-trained models.
[NLP-177] Weak Ties Strong Signals: Efficient Training Data Detection in Diffusion LLM s via Independent Token Sampling EMNLP2026
【速读】： 该论文旨在解决生成式 AI（Generative AI）中扩散型大语言模型（dLLMs）在去噪过程中可能泄露敏感训练数据的问题，其核心挑战在于如何有效检测此类数据泄露行为。由于dLLMs缺乏自回归架构所具备的高效单次概率分解特性，传统基于随机掩码的检测方法难以准确捕捉潜在的过拟合信号，且无法控制被掩码令牌间的依赖关系。本文指出，现有方法中的逐令牌近似会引入非负的结构性估计误差，该误差由被掩码令牌间的累积条件互信息（CMI）决定，进而掩盖细微的记忆信号。为此，论文提出一种查询高效的检测框架——独立令牌采样（Independent Token Sampling, ITS），其关键创新在于利用注意力机制衍生的成对依赖代理来近似考虑CMI的采样准则，从而选择内部依赖性较弱的令牌集合；同时引入多样性促进策略以提升多轮采样中的令牌覆盖度，最终生成受依赖性偏差影响更小的聚合令牌级检测信号。实验表明，ITS在多个数据集上显著优于当前最优基线，尤其在ArXiv数据集上实现0.18的AUC提升，并在有限查询预算下仍保持优异性能。
链接: https://arxiv.org/abs/2609.22145
作者: Hongyao Yu,Tianqu Zhuang,Ziyuan Xu,Hao Fang,Jiaxin Hong,Bin Chen,Shu-Tao Xia
机构: Tsinghua University(清华大学); Harbin Institute of Technology, Shenzhen(哈尔滨工业大学深圳校区)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: EMNLP 2026
Abstract:Diffusion large language models (dLLMs) offer a compelling alternative to autoregressive models, yet they may expose sensitive training data during denoising. Detecting such usage is challenging because dLLMs lack the efficient one-pass probability decomposition of causal architectures. Existing methods rely on random masking to obtain tractable token-wise detection signals under limited query budgets, but fail to control dependencies among masked tokens. We demonstrate that this token-wise approximation introduces a non-negative structural estimation error, which is theoretically characterized by the cumulative conditional mutual information (CMI) among masked tokens and can obscure subtle memorization signals. This insight suggests that reliable detection requires masked token sets with weak internal dependency. To avoid the prohibitive cost of directly estimating CMI over token combinations, we propose \textitIndependent Token Sampling (ITS), a query-efficient framework that uses an attention-derived pairwise dependency proxy to approximate the CMI-aware selection criterion. ITS further incorporates a diversity-promoting strategy to improve token coverage across sampling rounds, yielding aggregated token-wise signals that are less affected by dependency-induced approximation error. Experiments on multiple datasets show that ITS consistently outperforms state-of-the-art baselines across different models and datasets, achieving an AUC improvement of 0.18 on the ArXiv dataset while maintaining strong performance under limited query budgets. The code is available at this https URL .
[NLP-178] Multilingual Safety Signals Are Multi-Layered: Filtering Safety-Degrading Data for Safer LLM s EMNLP2026
【速读】： 该论文旨在解决大语言模型在微调过程中安全对齐（safety alignment）被无意破坏的问题，尤其关注即使使用看似无害的微调数据，仍可能存在隐性降低安全性的样本。现有方法通常依赖单一安全敏感层的表征来识别此类有害样本，但在多语言场景下，该假设的有效性存疑，因为不同语言间可能存在表征模式的差异。研究通过跨语言分析发现，安全敏感层在不同语言间仅部分共享，且与安全相关的信号往往分布于多个层级。针对这一问题，论文提出MMSAFE——一种多层多语言安全退化数据识别框架，能够同时捕捉跨语言共享的安全信号与语言特异性信号。实验结果表明，相较于随机过滤，MMSAFE将平均有害响应率降低了60%，并优于最强的单层基线方法，在多种模型、语言和安全基准上均展现出更优的性能，验证了多层建模在实现鲁棒多语言安全对齐中的有效性。
链接: https://arxiv.org/abs/2609.22144
作者: Jiakun Li,Guowei Song,Sijia Li,Xingwei He,Hongzheng Chai,Yuan Yuan
机构: Beihang University (北京航空航天大学); Zhongguancun Laboratory; Hangzhou Innovation Institute, Beihang University (杭州创新研究院，北京航空航天大学); Qingdao Research Institute, Beihang University (青岛研究院，北京航空航天大学)
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Accepted to EMNLP 2026 Main Conference. 16 pages, 11 figures
Abstract:Preserving safety alignment during large language models fine-tuning is critical, however, recent studies have demonstrated that even benign fine-tuning data may contain safety-degrading samples that silently undermine safety alignment. Existing approaches typically identify such samples using representations from a single safety-sensitive layer. While this assumption has shown effectiveness in monolingual settings, its validity for multilingual models remains unclear due to potential cross-lingual differences in representation patterns. Through a cross-lingual analysis, we show that sensitive layers are only partially shared across languages, with safety-relevant signals often distributed across multiple layers. Motivated by these observations, we propose MMSAFE, a multi-layer framework for multilingual safety-degrading data identification that captures both shared and language-specific safety signals. Extensive experiments across multiple models, languages, and safety benchmarks demonstrate that MMSAFE reduces the average harmful-response ratio by 60% compared with random filtering and achieves stronger average performance than the strongest single-layer baseline, demonstrating the effectiveness of multi-layer modeling for robust multilingual safety alignment.
[NLP-179] Using Composition Operators to Linearize LLM Semantic Transformations
【速读】： 该论文旨在解决大语言模型（LLM）中输入到输出映射关系的数学表征难题，即难以明确描述机器学习所学习的函数（如提示到响应、图像到字幕）在数学上的本质。其核心解决方案是引入基于动力系统理论的柯普曼框架（Koopmanism），并提出使用复合算子（composition operators）来推广传统的柯普曼算子，使其能够处理不同空间之间的映射，从而将LLM的变换视为矩形无限维算子。这一形式化框架揭示了重要结构：在提示与响应分布满足自然假设的前提下，LLM算子具有等距性（isometry）；而表示学习中的偏差则表现为有限截断部分的谱污染（spectral pollution）。基于此，论文进一步提出构建LLM算子的有限维近似方法，并利用奇异值谱对不同任务和模型进行量化比较，为理解与评估模型行为提供了新的数学工具。
链接: https://arxiv.org/abs/2609.22143
作者: Afjal Chowdhury,James Chen,Alan Edelman
机构: Massachusetts Institute of Technology (麻省理工学院)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG); Functional Analysis (math.FA)
备注: Submitted to NeurReps workshop 2026
Abstract:Machine learning learns functions: prompt to response, image to caption. What these functions are mathematically remains hard to say. We present a method to approximate these kinds of transformations using techniques from dynamical systems that fall under the umbrella of Koopmanism. We introduce the use of composition operators, which generalize the Koopman operator and, crucially, can map between distinct spaces, motivating the perspective that LLM transformations are rectangular infinite-dimensional operators. This formalism reveals useful structure: under natural assumptions on the prompt and response distributions, the LLM operator is an isometry, and misalignment between learned representations manifests as spectral pollution of its finite sections. We then outline a method of constructing finite-dimensional approximations of an LLM operator, and demonstrate how the singular value spectrum can be used to compare tasks and models.
[NLP-180] Does the Truthfulness Signal Survive Code-Mixing? Probing Hidden States for Hallucination Detection in Hinglish
【速读】： 该论文旨在解决生成式人工智能（Generative AI）在多语言、特别是混合语言输入场景下的幻觉检测能力不足问题，尤其关注以印地语-英语混用文本（即“Hinglish”）为代表的非纯正语言输入对幻觉探测模型性能的影响。当前主流的隐藏状态幻觉探测方法虽在纯语言环境下表现优异（如AUROC达0.90–1.00），但尚未在代码混用输入上进行验证。本文的关键解决方案是构建一个包含5,674个条目的印地语/英语/Hinglish问答数据集，基于三个7–8B参数量的开源大模型（Qwen2.5-7B、Mistral-7B、Llama-3.1-8B）生成并标注17,022条响应，提取特定层在两个词元位置的隐藏状态，并训练线性与MLP探测器以评估其在分布内检测与跨语言迁移的能力。研究发现，尽管存在语言混杂，幻觉信号仍具有较强的鲁棒性：跨语言迁移的AUROC范围为0.88至0.99，性能下降普遍低于0.05 AUROC；且以印地语训练的探测器向Hinglish迁移效果优于以英语训练的探测器。此外，独立发现三款模型在处理印地语及Hinglish时的幻觉率显著高于英文，表明语言特性对幻觉生成有实质性影响。研究成果支持未来在代码混用场景下开展更精准的幻觉检测研究，并已开源代码与合成的Hinglish QA数据集。
链接: https://arxiv.org/abs/2609.22138
作者: Tanveer Singh(Plaksha University)
机构: Plaksha University (普拉克沙大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 9 pages, 2 figures, 4 tables
Abstract:Hidden-state hallucination probing - training a linear classifier on an LLM’s internal activations to detect whether a generated answer is faithful to the input - is an active area of 2026 research, with recent work reporting 0.90-1.00 AUROC across several benchmarks and languages. However, none of this work has tested probes on code-mixed input, despite the fact that a huge population of chatbot users write in Hindi-English code-mixed text (“Hinglish”). We address this gap directly: does a hallucination probe trained on clean-language hidden states transfer to Hinglish, or does the signal degrade under code-mixing? We construct a 5,674-item Hindi/English/Hinglish QA benchmark, generate and label 17,022 model responses across three open-weight 7-8B LLMs (Qwen2.5-7B, Mistral-7B, Llama-3.1-8B), extract per-layer hidden states at two token positions, and train linear and MLP probes for in-distribution detection and cross-lingual transfer. We find that the hallucination signal survives code-mixing well: transfer AUROC ranges from 0.88 to 0.99, with gaps of mostly under 0.05 AUROC relative to in-distribution performance, and that Hindi-trained probes transfer to Hinglish more reliably than English-trained probes. As an independent, practically motivated finding, all three models hallucinate substantially more on Hindi and Hinglish than on English for matched facts. We release our code and synthetic Hinglish QA dataset to support further work on code-mixed hallucination detection.
[NLP-181] DiFA: Dual Evidence Fusion and Aggregation for Token-Level Text Anomaly Detection ICDM2026
【速读】： 该论文旨在解决现有文本异常检测方法在粒度层面的局限性，即大多数方法仅能实现文档级异常检测，难以精确定位有害短语或支持针对性防御。针对这一问题，论文提出一种面向词元级（token-level）异常检测的Dual-evidence with adaptive Fusion and Aggregation（DiFA）框架。其核心解决方案在于：通过形式-结构（form-structural）与语义（semantic）双视角分别捕捉可见的结构异常与上下文不一致等不同类型的异常模式，从而提供互补的异常证据；在此基础上，设计自适应校准与融合机制以平衡两种视角间量纲差异显著的异常得分；同时，引入多变量聚合方法对词元级异常得分进行多维度综合，有效避免稀有异常词元被平均化稀释，从而生成具有判别力的文档级异常评分。实验表明，DiFA在多个基准测试中均达到领先性能，兼具高效性、鲁棒性与可解释性。
链接: https://arxiv.org/abs/2609.22136
作者: Yanyu Qian,Pengcheng Weng,Yue Tan,Enguang Zuo,Yu Zheng,Yixin Liu
机构: Nanyang Technological University (南洋理工大学); University of Bern (伯尔尼大学); Griffith University (格里菲斯大学); Xinjiang University (新疆大学)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: Accepted by ICDM 2026. 10 pages, 4 figures, 2 tables
Abstract:Text anomaly detection, the task of identifying text instances that deviate from normal language patterns, is crucial for language-driven applications. However, most existing methods can only perform document-level anomaly detection, making it hard to locate harmful phrases or support targeted prevention. Recently, there has been an emerging trend toward token-level text anomaly detection, which aims to address the above limitation by identifying anomalous words or fragments within a document. Nevertheless, one representative method mainly relies on representation-space distance measurement, neglecting the complementary roles of different anomaly cues in capturing diverse abnormal patterns. To bridge the gaps, we propose a Dual-evidence framework with adaptive Fusion and Aggregation (DiFA) for token-level anomaly detection. DiFA derives anomaly scores from form-structural and semantic views to capture visible structural abnormality and contextual inconsistency, respectively, thereby providing complementary evidence for identifying diverse anomalies. To combine these two scores with varying numerical scales, DiFA incorporates a calibration and fusion mechanism to adaptively balance the two views. Moreover, to obtain a discriminative document-level score, a multivariate aggregation method is designed to summarize token-level anomaly scores from multiple perspectives, preventing rare anomalous tokens from being diluted. Extensive experiments across various text anomaly detection benchmarks demonstrate that DiFA consistently achieves top performance while maintaining strong efficiency, robustness, and interpretability. The code and scripts are available at: this https URL.
[NLP-182] Read-Best Is Not Steer-Best: A Probing–Steering Layer Dissociation in Omni-Modal Large Language Models
【速读】： 该论文旨在解决多模态大语言模型（Omni-modal Large Language Models）中干预层选择的合理性问题，即当前普遍假设“探测精度最高的层即为最优干预层”是否成立。研究发现，这一假设在三个独立开发的多模态模型上均不成立，揭示出探测能力与干预有效性之间存在显著分离现象，称为“探测-干预层解耦”（probing-steering layer dissociation）。其核心解决方案在于通过情绪（emotion）作为可控测试基准，系统评估文本、音频和图像输入在不同层上的可读性（readability）与可操纵性（steerability），并发现有效干预层始终集中于归一化深度的中后期范围，而探测最优层则在不同架构间差异显著。进一步的因果分析表明，随机方向控制下存在约26倍的因果效应差距，排除了扰动方向质量或偶然因素的影响；基于对数光谱（logit-lens）的阶段性分析揭示了从因果控制到探测饱和再到词汇承诺的渐进式前向过程，提出一个双因素解释框架：干预效果取决于表征可读性与下游可塑性的协同作用。因此，论文主张以跨架构的中后期层作为干预层选择标准，而非依赖探测性能，并识别出由效价（valence）与唤醒度（arousal）组织的跨模态情绪子空间，其中“喜悦”作为稳定锚点贯穿多个模型。
链接: https://arxiv.org/abs/2609.22135
作者: Yibo Wang,Jisheng Dang,Bimei Wang,Yitao Wu,Wencan Zhang,Hong Peng,Jizhao Liu,Bin Hu,Qi Tian,Tat-Seng Chua
机构: Lanzhou University (兰州大学); Hainan University (海南大学); Cloud and AI BU, Huawei (华为云与AI部门); School of Computing, National University of Singapore (新加坡国立大学计算机学院)
类目: Computation and Language (cs.CL)
备注: 13 pages, 10 figures
Abstract:Omni-modal large language models integrate text, audio, and image signals into a shared residual stream, where concepts such as emotion can be linearly decoded and causally modified by activation steering. A common but rarely tested assumption is that the layer with the highest probing accuracy is also the best layer for steering, so injection layers are often selected by probe performance. We provide the first causal test of this assumption across three independently developed omni-modal models and find that it fails. Reading and intervention rely on different layers, a phenomenon we call the probing-steering layer dissociation. Using emotion as a controlled testbed, we measure layer-wise readability and steerability across text, audio, and image inputs. Probe-best layers vary widely across architectures, while steering-effective layers consistently fall within a narrow mid-to-late range of normalized depth. Paired random-direction controls show an approximately 26-fold causal gap, ruling out random perturbation and direction quality as explanations. Logit-lens analysis reveals a staged forward process: causal handle, probing saturation, and vocabulary commitment, and motivates a two-factor account in which steering effectiveness depends on both representational readability and downstream plasticity. These results show that probing accuracy is a poor heuristic for selecting intervention layers and suggest a cross-architecture mid-to-late selection criterion. We also identify a cross-modal emotion subspace organized by valence and arousal, with joy acting as a stable anchor across models. Code and data: this https URL.
[NLP-183] Observational Equivalence of LLM and Human Annotation
【速读】： 该论文旨在解决文本标注中人类编码与大语言模型（LLM）在标注质量上孰优孰劣的争议问题。其核心挑战在于，传统观点倾向于认为人类专家编码更具可靠性，但本文通过复现14项政治科学领域的文本分类研究发现，当前主流大语言模型在标注一致性上与人类专家相当，二者之间的分歧主要源于文本内容及编码规则本身的模糊性。关键解决方案在于：首先，利用多大语言模型间的标注分歧来识别语义复杂或边界模糊的案例，并据此优化编码手册以降低不确定性；其次，在无法为每篇文本定义唯一标注的情况下，提出基于模糊性的推断边界（ambiguity-aware bounds），从而在分析中合理纳入剩余的不确定性。因此，研究指出，文本标注的核心问题已从“人机选择”转向“规则设计与不确定性建模”，即如何构建低歧义编码体系并有效量化残留模糊性。
链接: https://arxiv.org/abs/2609.22133
作者: Kentaro Nakamura,Jing Ling Tan,George Yean
机构: Harvard University (哈佛大学)
类目: Computation and Language (cs.CL)
备注:
Abstract:In this paper, we show that LLM and human coding are observationally equivalent in terms of annotation quality: recent LLMs agree with expert coders at rates comparable to those observed among experts themselves. We demonstrate this through replications of text-classification tasks from 14 peer-reviewed political science studies, in which ten LLMs, three human experts, and 165 crowdsourced workers independently classify the same texts using identical codebooks. We find that this equivalence is driven by ambiguity in the texts and coding rules. When LLMs disagree with experts, experts are also more likely to disagree with one another, and clarifying coding rules reduces disagreement among both experts and sufficiently capable LLMs. Thus, there is little empirical basis for preferring human coding on the basis of annotation quality alone, while LLMs offer substantial advantages in speed and cost. We therefore argue that the central challenge of text annotation is no longer choosing between human and machine coders, but developing coding rules that minimize ambiguity and accounting for the ambiguity that remains. To this end, we propose using disagreement across LLMs to identify difficult cases and refine codebooks, and we develop ambiguity-aware bounds for downstream inference when a unique annotation cannot be defined for every text.
[NLP-184] Correlation-Aware Structured Pruning for Large Language Models
【速读】： 该论文旨在解决结构化剪枝（Structured Pruning）在降低大语言模型（Large Language Models, LLMs）推理开销时，因忽略可剪枝单元（如通道或注意力头）间依赖关系而导致性能下降的问题。现有方法通常孤立评估各单元的重要性，隐含假设剪枝误差具有可加性，但实际中模型权重的非正交性及单元激活间的强相关性会破坏该假设，从而引发性能退化。为此，论文提出一种关联感知的结构化剪枝方法（Correlation-Aware Structured Pruning），其核心在于将剪枝目标建模为一个带基数约束的二元二次规划问题，显式地刻画重建误差中的跨单元依赖关系。由于该优化问题为NP难问题，难以精确求解，因此设计了一种基于依赖感知边际代价的贪心交互算法以实现高效的单元选择。此外，引入基于梯度的策略，实现模型各层自适应的稀疏度分配。大量实验表明，通过显式建模相关性，所提方法在准确率与效率之间取得了优于主流结构化剪枝基线的权衡表现。
链接: https://arxiv.org/abs/2609.22131
作者: Sicheng Xu,Hao Shi,Wei Zhang,Haoran Pang,Zhenyu Ming,Hao Wu,Zhongyi Huang,Xin Yao,Gong Zhang
机构: Tsinghua University (清华大学); Huawei Tech. Co. Ltd. (华为技术有限公司)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Structured pruning is a promising approach for reducing the substantial inference costs of Large Language Models (LLMs) while maintaining hardware efficiency. Many existing methods assess the importance of prunable units (e.g., channels or heads) in isolation, implicitly assuming that pruning errors are additive. This independence assumption is often invalidated by the non-orthogonality of model weights and strong correlations between unit activations, potentially leading to performance degradation. To address this, we propose a Correlation-Aware Structured Pruning method. We formulate the pruning objective as a cardinality-constrained binary quadratic program that explicitly models cross-unit dependencies in the reconstruction error. Since this binary quadratic program is NP-hard and difficult to solve exactly, we develop a greedy interaction algorithm based on dependency-aware marginal costs to optimize unit selection. Furthermore, we incorporate a gradient-based strategy to achieve adaptive layer-wise sparsity allocation across the entire model. Extensive experiments on mainstream LLMs demonstrate that incorporating correlation information yields competitive accuracy-efficiency trade-offs compared to representative structured pruning baselines.
[NLP-185] Beyond Accuracy and Surface Fluency: Risk-Sensitive Evaluation of LLM s for Legal Clause Generation ICML2026
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在合同起草应用中，传统基于流畅性或偏好度的评估方法无法有效识别法律实质性缺陷的问题。现有评估方式难以捕捉如关键例外条款缺失、风险分配不合法、管辖权假设错误或导致监管责任暴露等严重法律风险。其解决方案的关键在于构建一个基于实证研究的设计框架，融合两种评估体系：CLAUSE框架用于按法律功能与失效目标对提示进行分类，LENS-CRAFT框架则从九个法律质量维度对输出进行评分。尤为关键的是，研究采用“最大严重性原则”（Max Severity Principle），避免通过平均分掩盖单个具有决定性法律影响的缺陷，从而确保高风险问题得以凸显。该研究提出了一套完整的评估协议、分类体系、分析方案及结果报告结构，主张法律人工智能评估应从整体准确率转向以条款为单位、以失败模式为导向、具备风险敏感性的精细化评估范式。
链接: https://arxiv.org/abs/2609.22127
作者: Devansh Singh,Sundaraparipurnan Narayanan
机构: AI Tech Ethics
类目: Computation and Language (cs.CL)
备注: Accepted to the AI for Law Workshop @ ICML 2026. this https URL
Abstract:Large language models (LLMs) are increasingly used to draft contractual language, yet conventional accuracy or preference-based evaluations are poorly matched to legal drafting. A clause may be fluent and stylistically polished while still omitting an essential carve-out, allocating risk in an unenforceable way, assuming an inapplicable jurisdiction, or exposing a party to regulatory liability. This paper presents a empirical study design and framework for evaluating LLM-generated contract clauses. The study evaluates four models - Claude Haiku 4.5, Gemini 2.5 Flash Lite, GPT 5.4 Nano, and Qwen 3.5 Flash, across 22 contract clause categories and 34 legally-motivated failure modes. We combine two evaluation frameworks: CLAUSE, which classifies prompts by legal function and failure target, and LENS-CRAFT, which scores outputs across nine legal-quality dimensions. Instead of averaging dimension scores, the study applies a Max Severity Principle so that a single legally decisive defect remains visible. The paper provides the evaluation protocol, taxonomy, analysis plan, and a results structure for reporting empirical findings. We argue that legal AI evaluation should move beyond aggregate accuracy toward clause-specific, failure-mode-driven, and risk-sensitive assessment.
[NLP-186] ype-Driven Tokenization for Brahmic Scripts
【速读】： 该论文旨在解决大语言模型中标准分词器在处理婆罗米系文字（Brahmic scripts）时产生的文本畸形问题。婆罗米系文字属于音节文字（abugida），其辅音自带固有元音，可通过附加符号进行修改，这使得其正字法具有复杂的组合约束。与英语等字母文字的正字法构成一个自由半群（semigroup）不同，婆罗米系文字的正字法构成一个部分半群（partial semigroup），即并非任意两个有效词元的拼接都构成合法字符串。为解决此问题，论文在Agda中形式化了这一区别，将有效的婆罗米词元建模为转移系统中的链路，并推导出一个可证明正确的fixToken函数，能够将任意候选词元扩展以满足正字法边界约束。该形式化推导进一步转化为对SentencePiece的实际修补方案以及一个独立的Rust预分词库，成功消除了多种印地语系文字中的分词错误。解决方案的关键在于通过形式化验证建立正字法约束的数学模型，并将其转化为可部署的、保证正确性的实际工具。
链接: https://arxiv.org/abs/2609.22125
作者: Sai Hemanth Kapila,Rakshika Bagavathy
机构: Microsoft(微软); India(印度)
类目: Computation and Language (cs.CL); Formal Languages and Automata Theory (cs.FL)
备注:
Abstract:Standard tokenizers used in large language models produce malformed text when applied to Brahmic scripts. They are a family of abugidas, writing systems whose consonants carry an inherent vowel that dependent marks can modify. They include Devanagari, Telugu, Tamil, Kannada, and others. The underlying issue is that these tokenizers violate orthographic constraints that do not arise in alphabetic scripts like English. We observe that while English orthography forms a \emphsemigroup (any two valid tokens can be freely concatenated), Brahmic orthography forms a \emphpartial semigroup: not every concatenation yields a valid string. We formalise this distinction in Agda, model valid Brahmic tokens as chains in a transition system, and derive a provably correct \textttfixToken function that extends any candidate token to respect orthographic boundaries. We then show how this formal derivation translates into a practical patch for SentencePiece as well as a standalone Rust-based pre-tokenizer library, eliminating the observed errors across Indic scripts.
[NLP-187] Balancing Reasoning and Hardware Constraints in RAG Pipelines for Ukrainian Multi-Domain Document Understanding
【速读】： 该论文旨在解决在严格时间限制下，从多样化乌克兰语PDF文档中精确提取答案、文档ID及页码的多领域文档理解问题。核心挑战在于：在9小时离线执行时限内，扫描文档的光学字符识别（OCR）环节因采用串行单线程执行，成为严重瓶颈，耗时达5-7小时，导致仅剩约两小时用于大型语言模型（LLM）推理，且需处理500个问题。为确保系统在超时前完成，解决方案的关键在于构建一种资源高效的混合检索增强生成（RAG）管道，融合BM25、BGE-M3与交叉编码器重排序技术，以提升检索效率；同时避免使用高参数量的复杂推理模型（如DeepSeek R1），转而采用4比特量化后的LapaLLM 12B模型，并通过双NVIDIA T4 GPU部署实现高效推理。该策略优先保障流水线稳定性而非多步推理能力，最终在私有测试集上取得0.8095的得分，位列15支参赛团队中的第10名。
链接: https://arxiv.org/abs/2609.22124
作者: Illya Havrylov
机构: National Technical University of Ukraine “Igor Sikorsky Kyiv Polytechnic Institute”(乌克兰伊戈尔·西科斯基基辅理工学院); Educational and Scientific Institute for Applied System Analysis (IASA)(应用系统分析教育与科研所); Department of Artificial Intelligence(人工智能系)
类目: Computation and Language (cs.CL); Digital Libraries (cs.DL)
备注: 5 pages, 2 tables. Technical report based on UNLP 2026 Shared Task submission
Abstract:This paper describes the system submitted to the UNLP 2026 Shared Task on Multi-Domain Document Understanding. The challenge required extracting precise answers, document IDs, and page numbers from a diverse corpus of Ukrainian PDF documents within a strict 9-hour offline Kaggle execution limit. During evaluation on the hidden private test set, optical character recognition (OCR) of scanned documents emerged as a severe bottleneck, consuming 5-7 hours of the total time budget due to sequential single-threaded execution. This overhead strictly limited the remaining time for Large Language Model (LLM) inference to approximately two hours for 500 questions. To guarantee pipeline completion without timeouts, we developed a resource-efficient Hybrid Retrieval-Augmented Generation (RAG) pipeline utilizing BM25, BGE-M3, and Cross-Encoder reranking. Rather than deploying parameter-heavy reasoning models (e.g., DeepSeek R1) which consistently timed out, we utilized a 4-bit quantized LapaLLM 12B model via this http URL on dual NVIDIA T4 GPUs. Prioritizing pipeline stability over multi-step reasoning, our system achieved a Private Score of 0.8095, placing 10th out of 15 active teams.
[NLP-188] Success Leaves Detours: Learning Executable Walkthroughs for Long-Horizon Agents
【速读】： 该论文旨在解决在稀疏奖励轨迹中，如何从包含失败、循环和绕行的低质量交互数据中提取可执行、状态条件约束且可验证的流程（即“Walkthrough”）这一关键问题。其核心挑战在于：传统方法难以有效识别动作与后续状态变化之间的依赖关系，且抽象总结常忽略执行所需的条件信息。解决方案的关键在于提出Trace框架——一个基于信用分配与依赖关系驱动的编译机制。该框架通过检测奖励和持续状态变化识别进展锚点，利用信用传播定位高价值转移路径，并基于跨任务成功与失败证据推断动作先决条件；进而通过反向依赖切片追溯所需事实的生成源头，构建满足依赖一致性的动作链，同时剔除无关循环与绕行。最终生成的Walkthrough记忆体包含入口条件、有序的状态-动作-效果步骤以及完成与失败谓词，支持流程复用、中间状态恢复及程序化验证。实验表明，在J-TTL、WebShop和ScienceWorld三个基准上，Trace显著优于八种测试时学习与记忆基线，平均AUC和Final-3指标分别提升30.0%和40.5%，且推理令牌消耗更少，证明了长时程交互中状态条件化的可执行流程相较于完整轨迹或抽象摘要更具优势。
链接: https://arxiv.org/abs/2609.22120
作者: Kaijie Chen,Chenyu Fang,Liang Yan,Bo Li,Bo Zhang,Peng Ye
机构: Tongji University (同济大学); Shanghai AI Laboratory; Fudan University (复旦大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注: 13 pages
Abstract:Test-time self-evolving agents improve by reusing past experience, yet sparse-reward trajectories contain failures, loops, and detours, while summaries often omit the state conditions and action dependencies needed for execution. We study executable Walkthrough induction from sparse-reward trajectories: extracting compact, state-conditioned, and verifiable procedures. Our key observation is that delayed credit identifies actions associated with progress but cannot determine whether they produce facts required by later actions. We propose Trace, a credit-guided, dependency-grounded framework that compiles noisy trajectories into executable Walkthrough Memory. It detects progress anchors from rewards and persistent state changes, propagates credit to identify valuable transitions, and estimates action prerequisites from cross-episode success and failure evidence. Backward dependency slicing then traces required facts to their producers, extracting dependency-consistent action chains while removing irrelevant loops and detours. The resulting Walkthroughs encode entry conditions, ordered state–action–effect steps, and completion and failure predicates, supporting reuse, intermediate-state resumption, and programmatic verification. Experiments on J-TTL, WebShop, and ScienceWorld with three open-source LLMs show that Trace consistently outperforms eight test-time learning and memory baselines. Compared with the strongest baseline, it improves average AUC and Final- 3 by 30.0% and 40.5% , respectively, while using fewer inference tokens. These results show that long-horizon interaction benefits more from state-conditioned executable procedures than from complete trajectories or abstract summaries.
[NLP-189] Evaluation Awareness Shifts from Format to Context with Model Scale
【速读】： 该论文旨在解决评估意识（evaluation awareness）对大语言模型评估构成的严峻威胁，特别是揭示模型感知评估意图的内在机制，并对比小型与大型模型在检测评估行为时的不同路径。研究发现，小型模型主要依赖提示格式的敏感性来识别评估意图，而大型模型则更多通过高阶推理能力实现检测。为应对这一挑战，研究提出一种双路径干预策略，结合提示净化（prompt sanitization）与激活反向调控（activation counter-steering），以同时抑制外部评估触发信号及其在模型内部的表征。在200个高度评估意识的提示上，该方法实现了70.58%的平均行为翻转率，显著优于单一干预手段。该工作深化了对紧凑型语言模型中评估意识形成机制的理解，并表明有效的缓解措施必须协同作用于提示层面与表征层面。相关代码与资源可在GitHub仓库获取。
链接: https://arxiv.org/abs/2609.22119
作者: Navraj Singh,Maheep Chaudhary
机构: Indian Institute of Science Education and Research (IISER) Mohali(印度科学教育与研究学院莫哈里分校); Independent Researcher(独立研究员)
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Evaluation awareness poses an unprecedented threat to model evaluation, but the mechanisms by which models detect it remain unknown. This study focuses on determining this and identifying contrasting mechanisms between smaller and larger models. While smaller models use the prompt’s format sensitivity to detect evaluation, larger models often rely on higher-order reasoning to detect it. We evaluated Gemma 3 (1B, 4B, and 12B), Phi-3 (Mini and Medium), and Llama-3 8B using Chain-of-Thought analysis, representation probing, and Integrated Gradients attribution. Motivated by these findings, we propose a dual-pathway intervention that combines prompt sanitization with activation counter-steering to suppress both external evaluation triggers and their internal representations. Across 200 highly evaluation-aware prompts, our method achieves an average behavioral flip rate of 70.58%, consistently outperforming either intervention alone. These results provide new insights into how evaluation awareness develops in compact language models and suggest that effective mitigation requires jointly addressing both prompt-level and representation-level this http URL and codebase can be found in this \hrefthis https URLGithub Repository.
[NLP-190] An Empirical Cost Attribution of Context-Compression Gateways in Multi-Turn Coding Agents
【速读】： 该论文旨在解决生成式 AI（Generative AI）在多轮编程代理（coding agent）实际应用中，尽管存在上下文压缩（context compression）被广泛宣传可降低令牌（token）开销，但其在真实场景下的成本节约效果并不明确的问题。现有基准测试常误认为压缩文件读取即可显著节省成本，而本文通过在生产级压缩网关（Paritok）上对实际会话进行测量，揭示了三个独立的优化杠杆：工具模式过滤（tool-schema filtering）、文件读取与工具输出的内容压缩、以及历史记录摘要（history summarization），它们的成本节约机制和效率存在根本差异。关键发现在于：工具模式过滤通过每轮移除固定数量的令牌（约21K–57K），线性地随轮次增长，是唯一可重复验证且始终正向的优化手段；内容压缩虽单轮仅节省约2%的缓存定价前缀，但由于压缩后的数据会在后续每轮重传，其累积节约呈二次增长（约3350×N² tokens），在约6轮后超越固定过滤收益；非破坏性网关设计允许按需召回原始字节，每次召回仅重新发送一个片段，成本固定可控，避免了乘法级膨胀；最后，尽管单次压缩基准表现优异（如在SWE-bench上以25.7%压缩率保留86.5%任务质量），但此类性能与多轮代理的总成本无关，不能作为压缩节省成本的论据。因此，解决方案的核心在于区分不同压缩策略的边际效益，并提出可操作的优化优先级——应优先投入于工具模式过滤与可回溯的内容压缩机制，而非依赖单一压缩性能指标。
链接: https://arxiv.org/abs/2609.22114
作者: Luzhuo Chen,Jiayu Shi
机构: Paritok
类目: Computation and Language (cs.CL); Performance (cs.PF); Software Engineering (cs.SE)
备注: 10 pages, 1 figure, 4 tables
Abstract:Context compression is widely proposed as a way to cut the token bill of LLM coding agents, and public benchmarks report that aggressive compression preserves task-solving quality. These two facts do not imply the third one commonly assumed: that compressing file reads saves money in a real multi-turn agent. We instrument a production compression gateway (Paritok) between coding agents (Claude Code, Codex) and frontier LLMs (Claude Sonnet, GPT-5), and decompose the token bill of real sessions into three independent levers: tool-schema filtering, content compression of file reads and tool output, and history summarization. Measured in isolation under controlled A/B runs, the three save at fundamentally different rates. Tool-schema filtering removes a fixed block every turn, roughly 21K-57K tokens on a typical turn; it is linear in the turn count N and the only unambiguously and reproducibly positive lever. Content compression saves only about 2% of the cache-priced prefix per turn, but compressed reads accumulate in history and are re-sent on every later turn, so its cumulative saving grows quadratically, about 3350*N^2 tokens (measured), overtaking the fixed tool-filter saving within roughly 6 turns until the context window caps it. A non-destructive gateway lets the agent pull original bytes back on demand; each recall re-sends exactly the one segment just compressed away, so its cost is fixed and bounded rather than a multiplicative blowup, and heavy recall spends the accumulated saving back one segment at a time. Finally, a strong single-shot compression benchmark - 86.5% of SWE-bench quality retained at a 25.7% compression rate, achieved by the model this gateway deploys (Paritok-4B, reported separately) - is orthogonal to multi-turn agent cost and must not be cited as a cost-saving argument. We distill the results into an actionable recipe for where token-saving effort pays off.
[NLP-191] Privacy Personalization Trade offs in LLM s: The Impact of Stylometric Signal Reduction on User-Specific Text Generation
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在实现个性化文本生成时，因保留可识别的风格特征（stylometric signals）而引发的隐私泄露风险问题。其核心挑战在于：如何在保障用户隐私的前提下，维持模型生成内容的语义连贯性与个性化水平。解决方案的关键在于构建一个受控实验框架，利用LaMP-7 Twitter基准数据集，系统性地将原始用户档案中的性别、文化背景、个人细节及非正式语言线索等身份标识信息进行匿名化处理，并对比原始档案与匿名化档案在条件生成下的表现差异。实验结果表明，基于原始档案生成的文本在风格上几乎无法与真人写作区分，而使用匿名化档案生成的文本在偏好度上显著下降至平均13.0%，尽管语义上下文保持率仍高达94.8%。这一发现揭示了隐私保护与个性化之间的明确权衡关系，强调亟需发展具备隐私感知能力的个性化生成方法，在抑制可识别风格信号的同时，最大限度保留内容语义和表达一致性。
链接: https://arxiv.org/abs/2609.22112
作者: Muhammed Nazmul Arefin,Omar Jamal Hammad
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Large language models (LLMs) have demonstrated the ability to generate user-specific text with high stylistic fidelity. However, the personal data that enables such personalization frequently embeds demographic, cultural, and stylistic markers that raises concerns about stylometric re- identification. This paper investigates whether reducing identifiable stylistic signals affects personalization in text generation by LLMs. We introduce a controlled framework to isolate stylometric signals in LLM personalization using the LaMP-7 Twitter benchmark. Experiments on 250 sampled users compare two settings: paraphrasing conditioned on the original profile and paraphrasing conditioned on an anonymized converted profile in which demographic identifiers, cultural references, personal details, and informal linguistic cues have been systematically neutralized. Outputs are assessed by two independent LLM judges and a complementary human evaluation. Our pairwise evaluation shows that outputs conditioned on original profiles are nearly indistinguishable from human-authored ground truth, indicating that modern LLMs can closely reproduce an author’s writing style with sufficient fidelity. In contrast, preference for model outputs with anonymized profiles drops to 13.0% on average, while semantic context preservation remains high at 94.8%. A study with human evaluators confirms the same pattern. These findings reveal a clear privacy-personalization trade-off and highlight the need for privacy-aware personalization methods that retain meaning while suppressing identifying stylistic signals.
[NLP-192] Evaluating Fine-Tuned and Base Language Models in Maternal and Vaccination Healthcare for African Settings
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在低资源医疗环境中应用时可能出现的准确性不足与文化不适宜性问题，特别是在尼日利亚的孕产妇健康和疫苗接种领域。研究通过对比经过领域特定微调的MamaBot-Llama（孕产妇健康）与Vax-Llama（疫苗接种）模型与Meta的Llama-3.1-8B-Instruct基线模型，在临床准确性、安全性、清晰度、情境适切性和可信度等方面的性能差异，评估微调策略的有效性。其解决方案的关键在于采用低秩适应（Low-Rank Adaptation, LoRA）技术对模型进行领域特定微调，并基于高质量、由执业医师审校的问答数据集（超过36,000条孕产妇健康与9,000条疫苗接种数据）进行训练。结果显示，当训练数据质量高时，如MamaBot-Llama，微调显著提升了整体表现，尤其在临床可信度和医学准确性方面分别提升7%和5%，且关键问题减少50%，医生偏好率达78%；然而，当数据质量不足时，如Vax-Llama，微调反而导致整体性能下降5.2%，关键问题与安全顾虑分别上升192%和400%。因此，该研究强调：基于高质量、临床专家标注数据的领域特定微调可有效提升医疗AI模型性能，但若数据质量不佳则可能适得其反，故在临床部署前必须进行严格的领域验证。
链接: https://arxiv.org/abs/2609.22110
作者: Abdulquddus Ajibade,Oluwaseun Odunsi,Iyinoluwa Animasaun,Chioma Nwakanma-Akanno,Oluwasegun Oguntuase,Oluwafunke Akinbuwa,Abiodun Adereni
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Background: Large language models (LLMs) can improve healthcare information delivery in low-resource settings but may produce inaccurate or culturally inappropriate advice. This study evaluated domain-specific fine-tuning for maternal health and vaccination in Nigeria. Objective: To compare HelpMum’s MamaBot-Llama and Vax-Llama with Meta’s Llama-3.1-8B-Instruct for accuracy, safety, clarity, contextual appropriateness, and trustworthiness. Methods: We evaluated 200 healthcare questions, 100 each for maternal health and vaccination, across five subdomains per domain. MamaBot-Llama and Vax-Llama were fine-tuned using Low-Rank Adaptation on over 36,000 maternal health and 9,000 vaccination question-answer pairs, respectively. Two Nigerian licensed physicians independently rated responses using a 5-point Likert scale. Paired comparisons used Wilcoxon signed-rank tests. Results: Performance varied by domain. MamaBot-Llama significantly outperformed the base model across all criteria, with a 4.9% overall improvement (p .001), including gains in clinical trustworthiness (+7%) and medical accuracy (+5%). Critical issues decreased by 50%, and clinicians preferred it in 78% of cases. In contrast, Vax-Llama showed a 5.2% overall decline (p .001), with critical issues increasing by 192% and safety concerns by 400%. Conclusions: Domain-specific fine-tuning can improve healthcare LLM performance when based on high-quality, clinician-curated data, but may also degrade performance when dataset quality is inadequate. Rigorous domain-specific validation is essential before clinical deployment. Physician evaluators provided informed consent, and chatbot logs were anonymized. Keywords: Large language models; Fine-tuning; Maternal health; Vaccination; Healthcare AI; Low-resource settings; Nigeria; Model evaluation; LoRA; Medical accuracy
[NLP-193] Generalized Multimodal Foundation Model
【速读】： 该论文旨在解决现有多模态融合模型在部署后仅能处理预定义模态组合与单一任务，难以快速适应新下游应用的问题。其核心挑战在于如何构建一个通用的多模态融合框架，使其能够灵活应对任意模态组合和任意预测任务。解决方案的关键在于提出一种基于大规模合成多模态数据集的训练范式，这些数据集通过多样化因果结构形式化表征真实世界中多模态数据的生成过程。在此基础上，作者提出了广义多模态基础模型（generalized multimodal foundation model），该模型通过在具有丰富相关性模式的合成数据上训练，学习可迁移的多模态关联规律，并在推理阶段通过上下文示例激活适当的模态关联。实验结果表明，该模型在涵盖12种模态和11类预测任务的18个真实世界数据集上，无需任务特定微调即可达到与专用模型相媲美的性能，验证了其通用性和泛化能力。
链接: https://arxiv.org/abs/2609.22107
作者: Huizi Cui,Zongbo Han,Chenggong Ding,Naichuan Xiao,Jialong Yang,Jingdong Chen,Guangyu Wang,Qinghua Hu,Changqing Zhang
机构: Tianjin University(天津大学); Beijing University of Posts and Telecommunications(北京邮电大学)
类目: Machine Learning (cs.LG); Computation and Language (cs.CL)
备注:
Abstract:Making prediction with multimodal data is widely used in diverse scenarios. Existing multimodal fusion models, once deployed, can only handle predefined modalities (e.g., vision, text and audio) and single tasks, making it difficult to quickly adapt to new downstream applications. Therefore, a natural yet rather aggressive question arises, whether there exists a general multimodal fusion model that can be applied to arbitrary modality combinations and arbitrary prediction tasks. We argue that a unified multimodal fusion model should not depend on specific modalities and instead encode transferable patterns of multimodal correlation. To this end, we propose a simple and effective learning paradigm based on training over the generation of large-scale synthetic multimodal datasets with diverse causal structures that formally characterize the generative processes of multimodal data in real world. Building on this framework, we propose the generalized multimodal foundation model, a unified foundation model for generalized multimodal learning. By constructing large-scale synthetic multimodal datasets with diverse correlation patterns, our model encodes transferable multimodal correlations during training and activates appropriate associations through in-context examples during inference. Extensive experiments on 18 real-world datasets spanning 12 modalities and 11 prediction tasks demonstrate that our model achieves competitive performance with specialized models without task-specific adaptation.
[NLP-194] When Who You Are Can Change the Code You Get: A Study of Persona-Induced Bias in LLM Code Generation
【速读】： 该论文旨在解决生成式 AI（Generative AI）在编程辅助场景中是否存在用户人口统计学信息诱发的偏见，以及这种偏见如何影响生成代码的技术质量这一关键问题。研究聚焦于两种代表性大语言模型（LLM）——专有模型 Gemini 2.5 Pro 与开源权重模型 GPT-OSS-120B，通过构建涵盖国籍、性别和经验水平等18种人口统计学角色（persona）的提示，与中性基准进行对比，在超过35,000个程序生成任务中系统分析了推理过程与输出结果中的人口统计学标记泄露情况，以及功能正确性、可维护性、代码风格和安全性等方面的差异。研究发现，即使在与任务无关的情况下，人口统计学线索仍广泛存在于模型的推理轨迹（最高达70%）与输出中（最高达65%），且显著影响代码质量：例如，Gemini 模型在特定角色提示下平均正确率下降1.54个百分点，个别角色降幅达3.6%（优势比=0.51）；而GPT-OSS模型则表现出相反趋势，正确率提升3.4–5.7%（优势比=1.8–3.0）。尽管可维护性和代码风格存在统计显著但效应量极小（所有Cliff’s δ < 0.15）的影响，安全漏洞未呈现系统性角色相关模式。因此，论文的核心贡献在于揭示了用户人口统计学信息虽非任务相关，却会引发可测量的模型行为偏差，进而影响生成代码的质量，凸显了在生成式编程辅助系统中潜在的、尚未被充分关注的社会公平性风险。
链接: https://arxiv.org/abs/2609.22102
作者: Anubhav Gupta,Mayara Costa Figueiredo,Leticia Santos Machado,Tanner Wright,Ivan Beschastnikh,Cleidson R. B. de Souza,Gema Rodríguez-Pérez
机构: 未知
类目: oftware Engineering (cs.SE); Computation and Language (cs.CL); Computers and Society (cs.CY)
备注: Under review at Empirical Software Engineering 43 pages, 11 figures
Abstract:Large Language Models (LLMs) are widely used as programming assistants, yet it remains unclear whether and how user’s demographic information impacts the technical quality of generated code. We conduct a large-scale empirical study of persona-induced bias in LLM-based code generation, focusing a proprietary model (Gemini 2.5 Pro) and an open-weight model (GPT-OSS-120B). Using 18 demographic personas spanning nationality, gender, and experience level, we compare persona-induced prompts against a neutral baseline. Across 35,000+ generated programs, we analyze demographic marker leakage in reasoning and responses, as well as differences in functional correctness, maintainability, code style, and security. Our results show that demographic cues are frequently reflected in LLM reasoning and outputs. Demographic markers appear in up to 65% of responses and 70% of reasoning traces, despite being semantically irrelevant to the tasks. On LiveCodeBench, persona prompting were associated with lower correctness scores of the Gemini model by an average of 1.54 percentage points, with one persona exhibiting a decrease of 3.6% (odds ratio = 0.51). In contrast, the accuracy of the GPT-OSS model improved by 3.4 - 5.7% across all personas (odds ratios = 1.8 - 3.0). Maintainability and code style metrics show statistically significant but negligible effect sizes (all Cliff’s \delta 0.15), and security vulnerabilities exhibit no systematic persona-specific patterns. Overall, our results show that the presence of demographic information about users is associated with measurable variation in LLM reasoning and code quality even in purely technical tasks, and that these effects hold across models. Our work highlights an under-examined risk in LLM-assisted software development. Comments: Under review at Empirical Software Engineering 43 pages, 11 figures Subjects: Software Engineering (cs.SE); Computation and Language (cs.CL); Computers and Society (cs.CY) ACMclasses: D.2.3; K.4.2; I.2.7 Cite as: arXiv:2609.22102 [cs.SE] (or arXiv:2609.22102v1 [cs.SE] for this version) https://doi.org/10.48550/arXiv.2609.22102 Focus to learn more arXiv-issued DOI via DataCite
[NLP-195] Context Poisoning as Extreme-Value Attention Interference in Long-Context Language Models
【速读】： 该论文旨在解决大语言模型在处理长上下文时因冗余或混淆性信息引入而导致关键证据定位与利用能力下降的问题，这一现象被称为“上下文污染”（context poisoning）。其核心问题在于：随着上下文长度增加，尽管模型具备处理长序列的能力，但注意力机制中的极端值干扰（extreme-value interference）会导致决定性证据的得分被上限约束，而有效干扰项的最大得分则随干扰数量增长，从而削弱模型对真实证据的识别能力。论文的关键解决方案是基于软最大（softmax）检索抽象，推导出一个有限样本上界，表明为维持高于基线的固定准确率，证据得分差距需以Ω(√log N)的量级随有效干扰项数量N增长，而非原始上下文长度。该分析揭示了长上下文性能退化与得分混淆（score aliasing）、位置混淆（positional aliasing）及软最大稀释（softmax dilution）之间的内在关联。实验验证了嵌入困难负样本时检索准确率随总上下文增长而下降，且相同格式构造的干扰项在固定上下文长度下造成最显著的准确率损失；同时发现检索门控（retrieval gating）虽可提升证据利用效率，但净收益取决于是否能保持证据召回率。由此提出一系列改进方向，包括构建证据瓶颈、设计抗混淆表示、采用“先检索后推理”架构、验证器介导的记忆机制以及对比式抗污染训练策略。
链接: https://arxiv.org/abs/2609.22101
作者: Meysam Ghaffari,Nina Fatehi,Bhaskar Sen,Nasim Sabetpour,Carlos Morato
机构: 未知
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Large language models can process increasingly long prompts, yet their ability to locate and use decisive evidence may degrade as irrelevant or confusable context is added. We formulate this phenomenon, which we call context poisoning, as extreme-value interference in attention: the decisive-evidence score is upper-bounded, while the maximum score among effective distractors grows with their number. Under a softmax retrieval abstraction, we derive a finite-sample upper bound showing that maintaining a fixed accuracy target above base rate requires the evidence margin to scale as \Omega(\sqrt\log N) , where N denotes the effective distractor count rather than necessarily the raw context length. The analysis connects long-context degradation to score aliasing, positional aliasing, and softmax dilution. Controlled experiments show that retrieval accuracy decreases as total context grows in the presence of embedded hard negatives, that the same-format condition produces the largest observed accuracy drop among the tested distractor constructions at fixed context length, and that retrieval gating can improve evidence use while its net benefit depends on preserving evidence recall. These results motivate evidence bottlenecks, alias-resistant representations, retrieve-then-reason architectures, verifier-mediated memory, and contrastive anti-poison training.
[NLP-196] A framework for recipe data structure with applications for culinary and nutritional insights
【速读】： 该论文旨在解决烹饪配方数据缺乏统一可计算表示的问题，即现有食谱资源虽包含部分结构化信息，但未能在单一可查询的模式中整合食谱的成分构成、地理文化来源及其营养特征。其核心解决方案是构建一个系统化的食谱数据框架——RecipeDB2，通过将每道食谱分解为具有类型标注的成分实体，利用基于Transformer的命名实体识别模型对成分短语进行七类烹饪属性解析，并借助BERT嵌入策略将成分与美国农业部（USDA）参考数据库中的营养表关联，实现每个成分148项营养参数的映射（在前200个高频成分的独立标注集上F1达87.90%）；进一步采用随机森林分类器对全词汇表中的34类成分类别进行传播；最后通过一套确定性且保守的规则集为每道食谱分配饮食风格标签。该框架实现了从非结构化文本到可计算、可分析的结构化数据的转化，使烹饪遗产从艺术性表达转变为可量化的数据驱动研究对象。
链接: https://arxiv.org/abs/2609.22099
作者: Mansi Goel,Sumit Bhagat,Saloni Srivastava,Malav Patel,Shlok Vinodkumar Mehroliya,Ganesh Bagler
机构: Indraprastha Institute of Information Technology Delhi (IIIT-Delhi)(印第拉普拉斯特拉信息科技学院德里分校); Infosys Centre for Artificial Intelligence(印孚瑟斯人工智能中心); Center of Excellence in Healthcare(医疗健康卓越中心)
类目: Computation and Language (cs.CL)
备注: Main Text (11 pages, 4 figures, 3 tables); Supplementary Information (6 pages, 5 figures, 1 table)
Abstract:Cooking is a complex process that transforms raw ingredients into delicious and nutritious dishes, yet the recipes that encode this process remain largely free text; readable by people but not directly computable. Existing recipe collections capture fragments of this information, but no shared representation links a recipe’s structured ingredient composition, its geo-cultural provenance, and its nutritional profile within a single queryable schema. We address this representation gap by formalizing a framework for recipe data structure that decomposes each recipe into typed ingredient entities, grounds those entities in a reference nutritional database, and annotates them with geo-cultural and dietary context. We present RecipeDB2, a structured compilation of 128,942 recipes with 35,474 ingredients from 32 regions and 99 countries. Ingredient phrases are parsed into seven culinary attributes using a transformer-based named-entity model; ingredients are linked to the USDA reference tables through a BERT embedding strategy (F1 = 87.90 on a manually adjudicated set of the 200 most frequent ingredients), yielding 148 nutritional parameters per mapped ingredient; a Random Forest classifier propagates 34 ingredient categories across the full vocabulary; and a deterministic, conservative rule set assigns each recipe a dietary style. Through RecipeDB2 (this https URL), we demonstrate a scalable framework for making recipes computable, turning culinary heritage (long treated as an artistic rather than a quantitative object) into a data-driven analysis.
[NLP-197] reeSpark: Calibrated Load-Adaptive Draft Trees for Semi-Autoregressive Speculative Decoding
【速读】： 该论文旨在解决生成式AI（Generative AI）中语言模型推理加速的瓶颈问题，特别是现有推测解码（speculative decoding）方法在利用扩展树（draft trees）时存在的效率与精度失衡问题。现有方法依赖于逐位置边际概率对候选路径进行排序，忽视了候选词所延续的父节点信息，导致在半自回归式推测器中，较宽的树结构往往引入大量排序错误的节点；同时，固定大小的树结构无法根据每轮推测的实际支持能力或服务负载动态调整，造成资源浪费或性能下降。其解决方案的关键在于提出TreeSpark框架：通过复用推测器（drafter）已有的马尔可夫头（Markov head）输出的条件分布，以极低开销获取父节点相关的边接受概率估计，并将路径存活机制（path survival）作为核心控制逻辑，统一驱动最优优先扩展、每轮终止策略以及负载自适应的服务调度。此外，采用无放回采样结合递归拒绝采样（recursive rejection）确保任意温度下的解码无损性。实验表明，自适应树结构在所有温度下均优于固定预算方案，在相同推测器条件下，每轮可多接受15%-25%的推测令牌，单请求场景下推理速度提升8%-14%，且在负载上升时能平滑缩减树规模以维持稳定性。
链接: https://arxiv.org/abs/2609.22098
作者: Huapeng Zhou,Huayu Wang,Xinyu Wang
机构: 未知
类目: Computation and Language (cs.CL)
备注:
Abstract:Speculative decoding accelerates language-model inference by letting a cheap drafter propose tokens that the target model verifies in parallel. Recent block drafters make drafting nearly free: a single backbone pass emits an entire block of draft tokens. Draft trees promise a further gain – several alternative continuations verified in one target forward – but existing constructions rank candidates by per-position marginals that ignore which parent a candidate extends, so on semi-autoregressive drafters wider trees mostly add mis-ranked nodes; and a tree of fixed size ignores how much speculation each decoding round, and each serving load, can support. We introduce TreeSpark, which reads a parent-conditioned distribution from the drafter’s existing Markov head at negligible cost, calibrates it into an edge-acceptance estimate, and lets path survival govern everything else: best-first expansion, per-round stopping, and a load-adaptive serving policy. Sampling siblings without replacement, with matching residuals in recursive rejection, keeps decoding lossless at any temperature. Adaptive trees improve on matched fixed budgets at every temperature; against a tuned chain on the same drafter, TreeSpark accepts 15-25% more draft tokens per round and decodes 8-14% faster in single-request wall-clock, and under rising load it gracefully shrinks the tree back to the chain. Code and artifacts: this https URL
[NLP-198] oken Signatures of Code: Comparing Coding Behaviors Across Large Language Models
【速读】： 该论文旨在解决大语言模型（Large Language Models, LLMs）在代码生成任务中性能评估趋于同质化背景下，如何有效区分不同模型间编码行为差异的问题。随着多数模型在传统性能指标（如pass@k）上达到基准水平，仅依赖性能评价已难以揭示模型间的本质区别，因此亟需一种能够深入刻画模型编码策略与偏好差异的方法。其解决方案的关键在于提出CLIC（Code Learning for Identification and Comparison），一种基于分词频率分析的可视化分析方法：将每个代码样本表示为词元（token）频次特征向量，并训练可解释的决策树以区分两组模型生成的代码；同时引入两个新指标——鲁棒性（robustness），用于衡量在逐步移除最具判别性的词元后模型仍能否保持可区分性；以及集中度（concentration），用于评估差异是否由少数主导词元驱动或分布于大量词元中。研究进一步构建了交互式可视化系统，支持多尺度、假设驱动的探索，实现对模型对、任务及分词层级的系统性对比分析，最终通过10个模型在22个Kaggle机器学习任务上的案例研究，为模型选型与提示工程提供了可操作的洞见。
链接: https://arxiv.org/abs/2609.22097
作者: Junpeng Wang,Yuzhong Chen,Menghai Pan,Uday Singh Saini,Yiwei Cai
机构: 未知
类目: Computation and Language (cs.CL); Machine Learning (cs.LG); Software Engineering (cs.SE)
备注: 11 pages, 9 figures
Abstract:The evaluation of large language models (LLMs) on coding tasks has primarily focused on performance metrics such as pass@k. As LLMs continue to advance, many models now meet baseline performance requirements, reducing the discriminative power of performance-based evaluation alone. Yet a key question remains largely unexplored: how do LLMs differ in their coding behavior? We propose CLIC (Code Learning for Identification and Comparison), a visual analytics approach that characterizes LLM coding behavior through token-frequency analysis. CLIC represents each code sample as a feature vector of token frequencies and trains an interpretable decision tree to separate two LLMs’ code sets. Beyond classification accuracy, we define two new metrics: robustness, which measures whether the two LLMs remain distinguishable as their most-discriminative tokens are progressively removed, and concentration, which measures whether the difference is driven by a few dominant tokens or spread across many. Interpreting numerous pairwise comparisons (across LLM pairs, tasks, and tokenization levels) and tracing the full analytical chain form an inherently multi-scale, hypothesis-driven exploration task. We therefore develop an interactive visual analytics system to navigate the comparison landscape, identify pairs of interest, and drill down into discriminative tokens and their code contexts. Case studies comparing 10 LLMs across 22 Kaggle ML tasks reveal actionable insights for LLM selection and prompt engineering.
[NLP-199] AI-inferred expressed well-being and collective-action discourse in climate-change campaigns on X
【速读】： 该论文旨在解决气候运动期间公众情绪与行动语言之间关系的未知问题，具体聚焦于气候变化宣传周期是否改变积极情绪（happiness）与希望（hope），以及幸福感是否与行动导向语言一致。其核心问题是：尽管气候倡导常以关注度和动员效果评估，但伴随这些活动的情感表达——尤其是幸福感与行动意愿之间的潜在脱节——尚未被充分理解。解决方案的关键在于构建一个加权词典模型（versioned weighted lexical model），通过分析19个年份中地球日、地球一小时、全球气候行动日及世界环境日共364,118条公开推文，在事件前30天、事件期及事件后30天的时间窗口内量化幸福、未来导向希望、集体能力感、痛苦情绪及行动语言等维度。研究发现，事件期间幸福感较基线高出9.02个百分点，但行动语言却下降了10.75个百分点，揭示出“幸福感—行动力”之间的显著背离；此外，更幸福的原始帖子在传播中更难引发匹配的转发级联，提示情绪表达可能削弱实际动员效果。
链接: https://arxiv.org/abs/2609.22096
作者: Wentao Xu
机构: 未知
类目: Computation and Language (cs.CL); Computers and Society (cs.CY)
备注:
Abstract:Climate campaigns are often evaluated through attention and mobilization, but less is known about the well-being language that accompanies them. Whether campaign periods alter positive affect and hope, and whether happiness aligns with action language, remains unresolved. We analysed 364,118 public Twitter/X posts from Earth Day, Earth Hour, Global Climate Action Day and World Environment Day in 19 occurrence-years, using 30-day pre-event, event and post-event windows. A versioned weighted lexical model estimated happiness, future-oriented hope, collective capability, distress and action language. Event-period happiness prevalence was 9.02 percentage points higher than the pre-event baseline , whereas paired occurrence contrasts showed a 10.75-point decline in action language, indicating a happiness–action divergence. The happiness estimate remained positive across composition and text-deduplication checks, but was less precise under a 19-cluster wild bootstrap. Happier source posts had lower odds of an observed matched retweet cascade.
[NLP-200] Summarize Judge Refine: Decoupled Content Understanding and Policy Learning for Multimodal Content Moderation
【速读】： 该论文旨在解决内容审核系统在多模态理解与政策特定分类之间存在耦合的问题，导致每次政策变更均需重新训练完整模型，并受限于多媒体数据难以有效增强所引发的标签稀缺性。其解决方案的关键在于提出一种名为“总结-判断-精炼”（Summarize-Judge-Refine, SJR）的双模型架构，通过自然语言接口实现两者的解耦：多模态内容模型（Content Model）生成结构化文本摘要，而仅依赖文本的策略模型（Policy Model）基于政策定义对摘要进行分类。通过迭代式联合训练循环，利用基于奖励的强化学习优化（GRPO）不断精炼内容模型以生成更具政策相关性的摘要，同时在文本空间中引入对抗性摘要变体进行数据增强——这一路径在原始多媒体上不可行，从而支持少样本策略快速启动。所有决策均基于可读的结构化摘要，天然具备可解释性。实验表明，在误导性广告检测任务中，SJR相较零样本链式思维基线提升23.6%相对非误导性F1，优于端到端微调、STaR/RFT及RLFT等方法；尤为关键的是，一个仅使用合成违规样本训练的变体，在违规样本F1上与全数据模型差距不足0.2%，证明新政策可在无真实违规数据的情况下部署。
链接: https://arxiv.org/abs/2609.22094
作者: Zeeshan Ahmed,Yang Qin,Hanqing Huang
机构: Meta AI
类目: Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Content moderation systems traditionally entangle multimodal understanding with policy-specific classification, requiring full pipeline retraining for every policy change and suffering from label scarcity since multimedia cannot be meaningfully augmented. We propose Summarize-Judge-Refine (SJR), a two-model architecture that decouples these concerns via a natural language interface: a multimodal Content Model produces structured text summaries, and a text-only Policy Model classifies them against policy definitions. An iterative co-training loop refines the Content Model via GRPO to produce policy-relevant summaries, while text-space augmentation generates adversarial summary variants—an augmentation pathway impossible on raw multimedia—enabling few-shot policy bootstrap. Every decision is grounded in a human-readable summary, providing interpretability as a structural byproduct. On misleading advertisement detection, SJR achieves +23.6% relative non-misleading F1 over a zero-shot chain-of-thought baseline, outperforming end-to-end SFT, STaR/RFT, and RLFT. Notably, a variant trained on zero real violating examples—with all positive-class data synthetically generated—matches the full-data model within 0.2% relative on violating F1, demonstrating that new policies can launch without any real violation data.
[NLP-201] Memory That Looks Forward: A Zero-Inference Prospective Term for Personal Memory Retrieval
【速读】： 该论文旨在解决个人记忆存储检索中存在的“回顾式”（retrospective）局限性问题，即现有检索机制仅能根据查询内容匹配相似记忆，而无法主动响应用户已承诺的未来行为（commitments）。其核心解决方案是引入一种无需在查询时进行推理的前瞻性检索项（prospective term），通过将用户的承诺以带日期或触发条件的显式条目形式存入账本（ledger），当特定承诺被触发时，与之关联的记忆项获得显著性增强（salience boost），并通过乘法融合方式整合进基于嵌入（embedding-based）的检索系统中，从而在不损害相关性判断的前提下实现对潜在未来需求的预判。该方法的关键在于：预计算承诺关联关系作为分层架构中的“始终开启底座”（always-on floor），确保在不引入误检的前提下显著提升硬任务场景下的召回率（recall@5从0.000提升至0.955，甚至达1.000），且在53个已解决的承诺任务中未产生任何虚假增强。实证发现，仅有17–29%的自然语言表达的承诺-触发对能突破嵌入相似性阈值，表明该机制主要作用于少数关键场景，同时对大多数常规情况无负面影响，具备良好的兼容性。当前结果为初步验证，评估数据由人工构建，后续将在TriggerBench官方数据发布后开展正式评估。
链接: https://arxiv.org/abs/2609.22091
作者: Jonathan Groff
机构: Liberty University (利伯蒂大学)
类目: Computation and Language (cs.CL)
备注: 7 pages. Code and data: this https URL
Abstract:Retrieval over a personal memory store is retrospective: it surfaces what resembles the query, and it is blind to what the user has committed to do. We describe a prospective term for memory retrieval that costs no inference at query time. Commitments are held in an explicit ledger as dated or trigger-conditioned entries; memory items linked to a firing entry receive a salience boost, blended multiplicatively into embedding-based retrieval so that relevance remains sovereign. On a synthetic prospective-memory task set modeled on TriggerBench’s published structure (48 blind-authored dialogues, 175 tasks), the term raised recall@5 on the hard stratum from 0.000 to 0.955 at the default blend weight and to 1.000 under a floor variant, with zero false boosts across 53 resolved-commitment tasks. Blind authorship also produced a scope finding: only 17-29% of naturally phrased commitment-trigger pairs defeat embedding similarity, so the term matters on a real minority of cases and must do no harm on the rest, which it does not. We position precomputed commitment linkage as the always-on floor of a layered design whose expansion layer is query-time prospection. Results are preliminary: the evaluation set is author-constructed, and evaluation on TriggerBench proper is committed follow-up work once its data is released.
[NLP-202] Recognition Simulation and Refusal: A Contamination-Aware Study of Classic Psychological Effects in LLM Agents
【速读】： 该论文旨在解决生成式人工智能（Generative AI）在模拟人类心理效应时，将“表现出与心理效应相关的响应模式”误认为“具备相应认知偏差”的混淆问题。其核心挑战在于区分模型行为是源于对任务框架的敏感性，还是真正内化的认知偏见。解决方案的关键在于提出PsyAgentBench基准，采用因子设计实验范式，通过四个维度的交叉控制——任务标签（显式命名 vs. 盲态）、任务形式（经典版本 vs. 语义结构匹配但词汇与场景去重的反事实版本）、以及人格角色操控（persona manipulation），系统分离并检验各类心理效应在大语言模型代理中的可复现性与机制差异。研究发现，不同心理效应以截然不同的路径涌现：如阿施遵从效应依赖于任务标签的显式提示（从0%盲态到83.3%命名状态），锚定效应依赖于知识可用性（基于真实事实时为零，虚构数值时近乎完全），框架效应在标注条件下放大新内容的影响，沉没成本效应则在模型中表现为稳健缺失，而最小群体分配效应则体现为安全机制主导下的拒绝行为。此外，仅通过一句简短的人格角色指令（如宜人性）即可消除、减弱或逆转特定效应，表明不存在统一的响应偏倚机制。研究进一步形式化并实证了三种心理学范式无法迁移至大模型代理的失效机制：人格主导（persona dominance）、群体坍缩（population collapse）和安全选择（safety selection）。因此，作者主张摒弃单一的偏倚敏感度评分，转而报告复制轮廓（replication profiles）以揭示更复杂的动态结构。
链接: https://arxiv.org/abs/2609.22090
作者: Joy Bose
机构: 未知
类目: Computation and Language (cs.CL)
备注: 13 pages, 1 figure, 8 tables
Abstract:An LLM producing the response pattern associated with a human psychological effect is not the same claim as the LLM possessing that bias. We present PsyAgentBench, a benchmark that re-runs classic psychology experiments on LLM agents under a factorial design built to separate these: each paradigm is run with the paradigm explicitly labeled in the prompt (named) or framed as a routine task (blind), and on the literal textbook version of the task (canonical) or a structurally matched variant written to reduce lexical and scenario overlap with likely training data (counterfactual), crossed with a persona manipulation. Across five completed paradigms, evaluated on up to three open-weight model families with 41,904 trials released, apparently human-like effects arise through qualitatively different routes rather than one susceptibility: paradigm-label gating with explicit override (Asch conformity, 0 percent blind to 83.3 percent named on gpt-oss-120B), knowledge-dependent signal reliance (anchoring, exactly zero on grounded facts versus near total on invented quantities, a pattern equally consistent with rational use of the only available signal), amplification on novel content under labeling (framing), robust absence (sunk cost), and safety-mediated selection where refusal itself is the primary finding (minimal-group allocation). A one-sentence persona change (agreeableness, framed as an instruction rather than a verified trait manipulation) eliminates, dampens, or reverses these effects depending on which effect it is, arguing against any single response-bias account. We further formalize, and in two cases document empirically, three ways a psychology paradigm can fail to port to LLM agents: persona dominance, population collapse, and safety selection. We argue scalar bias-susceptibility scores obscure this structure and report replication profiles instead.
[NLP-203] Circuit Hypernetworks for Quantum-Augmented Diffusion Language Models
【速读】： 该论文旨在解决如何在大规模语言模型中高效引入量子计算能力以增强模型表现，同时避免因量子电路计算开销过大而导致的训练不可行性问题。其核心挑战在于如何在保持主干模型冻结的前提下，实现可训练且计算高效的量子增强机制。解决方案的关键是提出HyperQ架构，该架构在冻结的掩码扩散语言模型（masked-diffusion language model）的每个Transformer块中引入量子残差分支（quantum residual branch），通过轻量级的电路超网络（circuit hypernetwork） 为每个输入标记动态生成特定的旋转角度、耦合强度和测量轴参数，从而在共享的稀疏量子电路结构中实现基于标记条件的量子电路发射（token-conditioned circuit emission）。该设计使得期望值的计算具有精确的经典表达式，其复杂度随量子比特数呈线性增长，从而支持在1.1亿参数的主干模型上训练包含16至64个量子比特的电路。实验表明，随着电路宽度增加，下游任务平均得分从47.65提升至54.30，在64量子比特时分别超越原始模型和低秩微调基线4.71和3.67分，且仅需2万条提示-响应对即可完成微调，显著低于经典方法所需的20万条数据。这一结果验证了基于标记条件的量子电路生成是一种可行且高效的量子增强语言建模架构路径。
链接: https://arxiv.org/abs/2609.24657
作者: Xiaoqiang Wang,Mengyang Xiong,Jun Dai,Bang Liu
机构: Université de Montréal (蒙特利尔大学); Mila, Quebec AI Institute (魁北克人工智能研究所)
类目: Quantum Physics (quant-ph); Computation and Language (cs.CL)
备注: Work in progress
Abstract:Language models can be adapted by changing the computations applied to individual tokens. Quantum circuits offer one such approach, but evaluating wider circuits inside a large model can be computationally demanding. Here we introduce HyperQ, which adds token-conditioned quantum residual branches to a frozen masked-diffusion language model. A quantum residual branch is a module in each transformer block that reads a token’s hidden state, emits the coordinates of that token’s circuit, executes it, and adds the measured values back through a residual connection. The backbone remains frozen, and only the added branches are trained. Within each branch, a lightweight circuit hypernetwork emits token-specific rotation angles, coupling strengths, and measurement axes in a shared sparse circuit structure. The required expectation values have an exact classical expression whose evaluation cost grows linearly with the qubit count, enabling circuits from 16 to 64 qubits to be trained within a 1.1-billion-parameter backbone. Across downstream benchmarks, increasing circuit width raises the average score from 47.65 to 54.30. At 64 qubits, HyperQ exceeds the backbone and its low-rank-adapted counterpart by 4.71 and 3.67 points, respectively. HyperQ is fine-tuned on 20,000 prompt-response pairs, compared with 200,000 for the classical baselines. These findings support token-conditioned circuit emission as a tractable architectural approach to quantum-augmented language modelling.
[NLP-204] Morpho-VITS: Variational Inference with Morphological Modeling for End-to-End Speech Synthesis of a Tonal Bantu Language
【速读】： 该论文旨在解决班图语系声调语言在文本到语音（Text-to-Speech, TTS）合成中因词汇性（lexis）与语法性（morpho-syntax）双重声调系统复杂性所带来的挑战，尤其针对标准书写系统普遍省略声调标记和音节时长信息导致的歧义问题。其解决方案的关键在于引入一种基于形态-句法先验（morpho-syntactic prior）的端到端TTS架构：将传统VITS模型中的音素编码器替换为词素序列编码器与音素到词素注意力网络，通过显式建模词素结构来捕捉生成正确声调所需的上下文信息。在卢旺达语（Kinyarwanda）上的实验表明，该方法显著提升了合成语音的自然度、语调准确性和可懂度，验证了形态学建模在复杂声调语言TTS中的关键作用。
链接: https://arxiv.org/abs/2609.24310
作者: Antoine Nzeyimana
机构: 未知
类目: Audio and Speech Processing (eess.AS); Computation and Language (cs.CL)
备注: 5 pages, 2 figures, 2 tables
Abstract:Text-to-speech models for Bantu tonal languages are challenged by a tonal system that is rooted in both the lexis (i.e., the inventory of words, stems, and affixes) and the grammar (i.e., morpho-syntax). To complicate matters, the standard writing systems of these languages often omit tone markings and syllable duration information, which must be disambiguated by the reader based on context. Motivated by linguistic descriptions of Bantu language tone systems, we propose an end-to-end text-to-speech model that augments the text encoding mechanism with a morpho-syntactic prior. We replace the standard phoneme encoder in the VITS architecture with a morpheme sequence encoder and a phoneme-to-morpheme attention network. We posit that, by using this explicit morphological modeling, we can capture the information required to produce the correct tone. Experiments conducted on the Kinyarwanda language, a tonal and morphologically complex Bantu language, reveal substantial TTS improvement from this morphological modeling. Specifically, the proposed method significantly improves the naturalness, intonation, and intelligibility of the produced synthetic voices.
[NLP-205] AURA: Uncertainty-Routed Activation Editing for Acoustic Grounding in Speech Foundation Models
【速读】： 该论文旨在解决注意力编码器-解码器（AED）语音基础模型在输入无语音、声学证据薄弱或转录不可靠时产生的非语音幻觉（non-speech hallucination）问题，即模型生成与实际声学内容不一致的文本。其核心解决方案是提出AURA（Activation-editing with Uncertainty-Routed Adaptation），一种超高效的表示编辑方法：在冻结预训练模型的前提下，通过稀疏的缩放与偏移操作对解码器交叉注意力头进行微调。AURA的关键在于利用交叉注意力不确定性特征动态路由编辑策略，有效捕捉过度集中、注意力分散及帧级突变等异常注意力模式，从而实现自适应的激活修正。实验表明，AURA在无语音音频上将幻觉率从89.18%降至1.94%，无需预先识别幻觉相关注意力头；在标注不完善的语料（如儿童语音、成年语音及言语不流畅数据）上，其词错误率（WER）接近LoRA性能，但参数量仅为其约1/500，验证了基于不确定性路由的动态激活编辑在提升AED模型声学对齐性方面的有效性与实用性。
链接: https://arxiv.org/abs/2609.23979
作者: Natarajan Balaji Shankar,Zilai Wang,Zihan Wang,Mohan Shi,Kaiyuan Zhang,Abeer Alwan
机构: University of California Los Angeles (加州大学洛杉矶分校)
类目: Audio and Speech Processing (eess.AS); Computation and Language (cs.CL)
备注: Accepted to IEEE SLT 2026
Abstract:Attention encoder-decoder (AED) Speech Foundation Models achieve strong ASR performance but can generate acoustically unsupported text when inputs contain no speech, weak acoustic evidence, or unreliable transcription. We propose AURA: Activation-editing with Uncertainty-Routed Adaptation, an ultra-efficient representation-editing method that freezes the pretrained model and applies sparse scale-and-shift edits to decoder cross-attention heads. AURA dynamically routes edits using cross-attention uncertainty features that capture over-concentration, diffuse attention, and abrupt frame shifts. We evaluate AURA on four datasets spanning non-speech hallucination and speech grounding stressors, including imperfect-label child speech, imperfect-label adult speech, and disfluent speech. On non-speech audio, AURA reduces hallucination rate from 89.18% to 1.94% without prior hallucination-head identification. On imperfect-label corpora, AURA approaches LoRA WER while using roughly 500x fewer trainable parameters. Sensitivity analysis and qualitative cross-attention examples are consistent with AURA’s uncertainty-routed editing behavior, supporting dynamic activation editing as a practical path for grounding AED speech models.
[NLP-206] Long-Tail Rebalancing for Non-Verbal Vocalization-Aware ASR: A Track~1 System for the NVVSpeech Challenge
【速读】： 该论文旨在解决传统自动语音识别（ASR）系统在处理非语言声音（NVV）时存在的关键问题，即对蕴含重要副语言信息的非语言发声缺乏有效建模，尤其在标注数据有限且类别分布高度不均衡的挑战下。其解决方案的核心在于提出一种以数据为中心的NVV感知ASR流程，通过跨数据集标签统一（cross-dataset label harmonization）实现源标签到官方分类体系的映射，并剔除无法可靠映射的样本以提升标注质量；同时采用两阶段采样策略：第一阶段使用平方根类别采样（square-root category sampling）缓解长尾分布问题，第二阶段进行均匀类别微调（uniform-category fine-tuning），从而在有限监督条件下显著提升模型对稀有类别的识别能力。该方法在固定本地验证集上表现最优，最终在ISCSLP NVVSpeech Challenge Track 1中取得63.86的官方得分并排名第四。
链接: https://arxiv.org/abs/2609.23462
作者: Shangyue Jia,Jingru Ma,Yangzhuo Li,Daoping Luo,Bowen Tian,Hanchen Lu,Wenze Ren,Yunxiang Chen,Houdun Liu,Shuo Feng,Lei Xie,Liumeng Xue
机构: 未知
类目: Audio and Speech Processing (eess.AS); Computation and Language (cs.CL); Sound (cs.SD)
备注: Accepted by ISCSLP 2026, NVVSpeech Challenge Track 1
Abstract:Non-verbal vocalizations (NVVs) carry important paralinguistic information but are often omitted by conventional automatic speech recognition (ASR) systems. The ISCSLP NVVSpeech Challenge requires joint transcription of lexical content and 16 NVV categories under limited and highly imbalanced supervision. We present a data-centric NVV-aware ASR pipeline based on cross-dataset label harmonization and a two-stage sampling schedule. We map heterogeneous source labels to the official taxonomy and exclude samples without a reliable mapping. Our schedule first uses square-root category sampling to moderate the long-tailed distribution and then applies uniform-category fine-tuning. On a fixed local validation split, square-root category sampling performs best among the tested single-stage settings. The final two-stage system obtains an official score of 63.86 and ranks fourth in Track 1.
[NLP-207] LLM -Based FORM Code Generation with Verification-Driven Fine-Tuning
【速读】： 该论文旨在解决生成式人工智能（Generative AI）在高能物理领域中对专用符号计算语言FORM的代码生成支持缺失的问题。尽管FORM在多圈费曼图计算中扮演核心角色，但现有大语言模型（LLM）在无文档提示的情况下对FORM任务的零样本执行通过率接近零，表明其为真正的零样本语言。为此，研究提出一种基于验证驱动的数据生成流程，利用FORM可执行二进制文件作为执行判别器，自动生成并验证包含确定性计算、开放式程序、教程代码及知识问答对在内的4,633个训练样本。通过量化低秩适应（QLoRA）微调一个轻量级开源模型（Qwen3-8B），所获专家模型在四个互补基准测试（共840个任务，单次尝试评估）上显著优于参数高达7560亿的前沿模型，在执行成功率和严格符合FORM输出的匹配度方面表现更优，且在小型、高难度任务上与前沿模型无统计差异；同时保持了2.6个百分点以内的通用推理与编程能力损失，实现了专业性与通用性的有效平衡。
链接: https://arxiv.org/abs/2609.23367
作者: Bakar Chargeishvili
机构: 未知
类目: High Energy Physics - Phenomenology (hep-ph); Computational Engineering, Finance, and Science (cs.CE); Computation and Language (cs.CL)
备注:
Abstract:FORM is a domain-specific symbolic manipulation language widely used in particle physics for processing the very large algebraic expressions arising from multi-loop Feynman diagram calculations. Despite its central role in precision theoretical physics, no artificial-intelligence tooling exists, to our knowledge, for assisting physicists in writing FORM code. We show that contemporary large language models (LLMs), including frontier models with hundreds of billions of parameters, achieve a zero-percent execution pass rate on our instruction-following and tutorial-style FORM tasks without documentation in a single attempt, establishing FORM as a genuine zero-shot language for LLMs at the time of writing. We then present a verification-driven data generation pipeline that uses the FORM binary itself as an execution oracle to produce and validate a corpus of 4,633 training examples spanning deterministic computations, open-ended programs, tutorial code, and knowledge question-answer pairs. Fine-tuning a compact open-weights model (Qwen3-8B) with quantized low-rank adaptation (QLoRA) yields a specialist that, evaluated on four complementary benchmarks (840 tasks, single attempt each), decisively outperforms frontier models with up to 756B parameters in execution rate and in strict, FORM-verified output matching on the larger benchmarks, and remains statistically indistinguishable from them on the smaller, harder ones. General reasoning and coding capabilities are preserved within 2.6 percentage points.
信息检索
[IR-0] Ascent: An Agent ic System over the Model Context Protocol for Real-World Clinical Data Analysis
链接: https://arxiv.org/abs/2609.24620
作者: Angelo Ziletti,Leonardo D’Ambrosi,Melanie Tuchardt,Tim Kondziella
类目: Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注:
Abstract:Answering epidemiological questions from real-world clinical data requires medical coding, schema-aware SQL, and validation of implicit choices about populations, denominators, and time. We present Ascent, an agentic system that exposes medical coding, question answering, and cohort analysis through a shared Model Context Protocol tool surface for standardized and native schemas. We introduce EpiTrap, a dataset testing whether systems avoid recognized pharmacoepidemiological errors, and compare a fixed pipeline with agents across models and orchestrators. With capable models, agents improve accuracy over the fixed pipeline by an average of 27 and 20 percentage points on native and standardized schemas, respectively. These gains require more tool calls and longer runtimes. Experience from real projects highlights the system’s value for feasibility assessment, diagnostic iteration, and expert-guided analysis.
[IR-1] UK-PRBENCH: A Parag raph-Level Precedent Retrieval Benchmark for United Kingdom Case Law
链接: https://arxiv.org/abs/2609.24613
作者: Damith Premasiri,Tharindu Ranasinghe
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL)
备注:
Abstract:Prior case retrieval (PCR) aims to identify precedent cases relevant to a given query case. Existing PCR benchmarks and methods predominantly operate at the document level, treating entire judgments as the unit of relevance. This formulation is suboptimal for legal practitioners, as judgments address multiple legal issues and only a small subset of paragraphs is relevant to a particular query. Addressing this gap, we introduce UK-PRBench, a benchmark for paragraph-level precedent retrieval in UK case law, constructed from judgments obtained from the UK National Archives and covering a broad range of UK courts and tribunals. Furthermore, we evaluate state-of-the-art retrieval models and establish baseline results. Our experiments show that paragraph-level precedent retrieval remains challenging for current retrieval approaches, highlighting substantial room for improvement. UK-PRBench provides a standardised benchmark for evaluating fine-grained precedent retrieval and advancing retrieval systems for the UK legal domain.
[IR-2] What Makes a Good Semantic ID for Generative Recommendation? A Reproducibility Study SIGIR
链接: https://arxiv.org/abs/2609.24430
作者: Yufei Chen,Junchen Fu,Jujia Zhao,Yukun Zhao,Zhaochun Ren
类目: Information Retrieval (cs.IR)
备注: Accepted by SIGIR-AP 2026
Abstract:Generative recommendation has emerged as an active research direction, where items are commonly represented by semantic IDs (SIDs): discrete codes generated token by token. Despite strong empirical results, SID designs vary widely in construction strategy, codebook organization, and code length, making their true impact on recommendation performance unclear. We conduct a large-scale reproducibility study to systematically investigate the impact of semantic ID design on generative recommendation under a unified experimental framework. We focus on a fundamental question: What makes a good semantic ID for generative recommendation? To answer this question, we examine four aspects: the relative effectiveness of different semantic ID designs, the connection between codebook utilization and recommendation quality, the effect of semantic code length, and the influence of semantic ID design on local item semantic preservation. Through a unified evaluation and additional cross-dataset controlled analyses, we find that the effects of SID design are largely non-monotonic: no single SID design is universally best, and commonly used RQ-VAE- and OPQ-based designs can behave inconsistently across datasets. The method with the most balanced first-level codebook is not consistently the best recommender, showing that utilization is diagnostic but insufficient. Scaling either the generative backbone or the SID length is also not always beneficial. Finally, semantic-neighborhood analysis reveals that no single SID design dominates all notions of local semantic preservation; instead, different designs exhibit complementary strengths that remain stable across datasets and neighborhood sizes. Our study provides a controlled and reproducible understanding of semantic ID design and offers practical insights for future generative recommender systems. Comments: Accepted by SIGIR-AP 2026 Subjects: Information Retrieval (cs.IR) Cite as: arXiv:2609.24430 [cs.IR] (or arXiv:2609.24430v1 [cs.IR] for this version) https://doi.org/10.48550/arXiv.2609.24430 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[IR-3] Auditing Source Exposure in Baidu and Google AI Search EMNLP2026
链接: https://arxiv.org/abs/2609.24407
作者: Yibo Li,Enci Guan,Yuedan Cai,Geng Liu,Francesco Pierri
类目: Information Retrieval (cs.IR)
备注: Accepted at WAC @ EMNLP 2026
Abstract:AI-generated overviews are becoming an increasingly prominent layer of search interfaces, yet their behavior in Chinese-language search remains underexplored. We conduct a cross-lingual audit of AI overview behavior on Baidu and Google using English queries sampled from MS MARCO and their translated Chinese counterparts. Our analysis examines when overviews are triggered across platform-language settings, which host domains receive visible exposure in Chinese-language overviews, how concentrated that exposure is, and how source overlap varies across settings. We also compare the embedding-based semantic similarity of generated answers for matched query intents. The results reveal substantial differences across platform-language settings in overview availability and visible source exposure. At the aggregate level, the settings exhibit low overlap in visible host-domain inventories, while matched-query answers yield median cosine similarities ranging from 0.701 to 0.813. These findings indicate that answer-level semantic similarity and aggregate source exposure capture distinct dimensions of AI-mediated search. Evaluations of AI search should therefore consider not only the content of generated answers but also how source visibility is distributed across platforms, languages, and information environments.
[IR-4] Graded-Relevance Composed Multimodal Retrieval for E-commerce Visual Search at Scale
链接: https://arxiv.org/abs/2609.24152
作者: Anubhav Gupta,Hrushikesh Mohapatra,Prijith Chandra,Asish Mohapatra,Anuj Garg,Arvind Maan,Sudip Datta,Venkat Bulusu,Sitesh Kumar Jalan
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Visual search on large e-commerce catalogs must serve both “similarity” queries that ask for items resembling an uploaded image and “modifier” queries that comprise an image and text describing a desired modification (e.g. a color change or style swap). The latter is the setting known as composed image retrieval (CIR). Existing CIR methods, however, treat relevance as binary and train on triplets with a single positive target - a poor fit for real catalogs where many candidates partially satisfy a user query and ranking across that partial-match spectrum drives the customer experience. We propose a methodology for training CIR retrievers on graded relevance, consisting of: (i) a VLM to curate training data, generating both queries (object detection + modifier synthesis) and 4-level relevance labels without manual annotation, (ii) an iterative relevance-feedback loop that expands the training set by mining hard negatives from the in-training retriever, and (iii) a hierarchy-aware angular objective to train the retriever directly on the graded labels rather than collapsing them to a binary split. We call this methodology GradCIR and instantiate it on a PaliGemma2 bi-encoder trained on 3.5M graded pairs curated from raw Walmart catalog data. A controlled graded-vs-binary ablation isolates the supervision granularity and shows lift of 4.9%-5.9% in NDCG@10. The same recipe applied to other multimodal encoders lifts early-fusion backbones by up to 8.5% NDCG@10. On the public FashionIQ benchmark, GradCIR (applied to PaliGemma2) reaches 0.6703 average recall when fine-tuned, slightly ahead of the strongest peer-reviewed supervised baseline we compare against, and matching or exceeding all published CLIP-L-class zero-shot CIR methods. The system is deployed in production at Walmart, where it’s serving live visual-search user traffic.
[IR-5] When More Evidence Hurts: Publication-Bias Drift and Principled Stopping for Biomedical Causal Search
链接: https://arxiv.org/abs/2609.24101
作者: Fred Sun,Shangqi Guo
类目: Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注:
Abstract:Automated biomedical evidence synthesis depends on retrieving published studies, but the biomedical literature is systematically skewed toward positive findings. Deeper retrieval can therefore make a system \emphmore likely to falsely infer benefit when the true effect is null. We formalise this phenomenon as \emphevidence drift and prove that, under a standard publication-bias model, the false-positive probability on null-effect queries follows a strictly increasing large-sample envelope in retrieval depth, approaching one. Empirically, on a held-out test set of 140 Cochrane-derived queries, drift rises monotonically from 7.9% to 15.7% as the retrieval budget grows from 3 to 20 steps, and concentrates in the null-effect class. We present DACG-agent, a drift-aware causal-graph agent that incrementally builds a causal knowledge graph from PubMed abstracts and applies a two-layer stopping policy with complementary roles: a KL-divergence monitor that detects posterior convergence (the accuracy layer), and a Bradley–Terry process reward model (PRM) whose online decline detection halts retrieval once evidence quality peaks (the efficiency layer). Against full-budget retrieval, DACG-agent reduces evidence drift from 15.7% to 6.4% and improves null-effect accuracy by 21 percentage points (40.0% \to 61.4%) while using 67% fewer retrieval steps; overall accuracy rises from 61.4% to 69.3% (95% CI 61–77). A simulation confirms the drift result transfers from the analysed vote-counting aggregator to the deployed noisy-OR one.
[IR-6] UniK: Universal Knowledge Perception for Digital and Physical AI
链接: https://arxiv.org/abs/2609.23971
作者: Nirmit Desai,Kunal Sawarkar,Aditya Mahakali,Dongkon Lee,Kevin Park,Eric Song
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注: 17 pages
Abstract:Two transformative classes of AI systems are reshaping how organizations operate: \textitdigital AI, which reasons over enterprise knowledge to power chatbots and agent workflows; and \textitphysical AI, which learns to control robots and autonomous systems from video, gameplay, and sensor telemetry. Both face the same foundational bottleneck: raw knowledge at scale, spanning heterogeneous modalities, locked in private corpora that existing AI infrastructure cannot access reliably or efficiently. We propose \textitUniversal Knowledge Perception (UniK) as a common platform for both classes, covering the full knowledge lifecycle (ingestion, enrichment, indexing, retrieval, and continuous evaluation) across modalities from rich text and video to molecular data and sensor telemetry. We present UniK, built on Polymath Retrieval (multi-index fusion over automatically enriched indices) with no task-specific fine-tuning. Across five digital AI domains (medical literature, open-domain QA, chemistry, legal video proceedings, and government open data) UniK combined with an open-source 70-billion-parameter model consistently matches or outperforms frontier proprietary LLMs that are orders of magnitude larger: 76% RAG accuracy on government data versus 47% for GPT-5; 77.9% on medical QA without fine-tuning; topping all open-source chemistry pipelines. We show that the same infrastructure directly addresses the data curation, indexing, and retrieval challenges facing physical AI world model training, where the knowledge problem is harder but structurally identical.
[IR-7] Q-TIE: A Lightweight and Generalizable Re-ranking Framework for Temporal Information Retrieval EMNLP2026
链接: https://arxiv.org/abs/2609.23880
作者: Soyeon Kim,Hyunjin Kim,JinYeong Bak,Steven Euijong Whang
类目: Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注: Accepted to EMNLP 2026 (Main Conference). Code: this https URL
Abstract:Temporal Information Retrieval (TIR) has been increasingly critical given the rise of Retrieval-Augmented Generation (RAG). Since temporally mismatched evidence can be highly misleading, TIR aims to retrieve documents that are both semantically and temporally relevant to a query. Two TIR paradigms have emerged - temporal retrievers and temporal re-rankers - differing in how temporal relevance is modeled. While these paradigms provide complementary strengths, our analysis reveals that each alone falls short of robust TIR: temporal retrievers provide flexible query understanding via learned representations, but often fail to explicitly account for temporal constraints; temporal re-rankers can enforce such constraints more explicitly, but often rely on predefined re-ranking rules. To address this, we propose Q-TIE, a re-ranking framework based on learned Temporal Intent Extraction (TIE). By introducing a TIE model that maps each query’s temporal constraint into a unified interval representation (i.e., \langle t_start, t_end \rangle ), Q-TIE generalizes beyond predefined rules via model-based learning while explicitly modeling temporal constraints as a separate signal - jointly achieving what each paradigm typically trades off. Experiments demonstrate that Q-TIE consistently outperforms existing TIR methods with stronger generalizability across temporal query types, and provides a lightweight yet effective add-on for temporally-aware RAG pipelines. Code: this https URL.
[IR-8] Explainable Recommendations at Scale: LLM Rationales for YouTube Music Artist Discovery
链接: https://arxiv.org/abs/2609.23877
作者: Xiao Liu,Yanwei Song,Srivaths Ranganathan,Yuan Chen,Zheyun Feng,Parker Steenburgh,Jochen Klingenhoefer,Nathan Lasche,Gergo Varady,Tim Steele
类目: Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注:
Abstract:Modern music streaming platforms face a persistent tradeoff: exploiting familiar content versus driving the exploration of novel items. While users frequently desire discovery, they hesitate to select unknown artists over proven favorites. Providing transparent, natural language rationales that explain why an unexplored item is recommended lowers this barrier. However, while Large Language Models (LLMs) excel at this nuanced explainability, their real-time deployment is severely bottlenecked by prohibitive inference costs and computational overhead. In this paper, we present an industry case study of a decoupled recommendation architecture that successfully scales exploration without compromising latency. Our system isolates LLM inference asynchronously offline, pre-computing personalized candidate pools of undiscovered artists alongside tailored rationales. Large-scale online A/B experiments validate our design. We demonstrate that combining LLM-backed recommendations with these explanatory rationales significantly reduces the trust barrier for new content, yielding statistically significant improvements in both user exploration and overall engagement on the discovery surfaces.
[IR-9] From UNDRR Reports to Event Records: Schema-Constrained LLM Extraction of Georeferenced Disasters
链接: https://arxiv.org/abs/2609.23853
作者: Camilla Andreozzi,Phuong-Anh Nguyen-Le,Zhijing Jin,Revati Mani
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注: 17 pages, 3 figures
Abstract:Disaster-risk-reduction archives describe hazard events in prose that databases such as EM-DAT (Delforge et al., 2025) cannot ingest directly. We present an LLM pipeline that generates candidate georeferenced event records using a controlled hazard vocabulary and fixed schema, retaining evidence for review. Applied to 10,000 documents from PreventionWeb, the knowledge hub managed by UNDRR, it produced 3,572 records from 1,913 documents across 24 hazard types and resolved 81% of location mentions to OpenStreetMap geometries. On 171 human-positive document windows from a stratified 217-document reference set, GPT-5 achieved 86.0% pooled attribute F_1 , versus 44.2% for the spaCy-gazetteer baseline. Evaluation pools hazard families, location strings, and event years within documents, without assessing their assignment to individual events. GPT-5.4 ranked highest among ten LLMs (86.6% F_1 ). Verbatim evidence occurrence was 72.0% for GPT-5 and 47.2% for GPT-5.4, measuring textual traceability without establishing attribute support. We report production failure modes and automated label and location-rule compliance checks. Prompts, schema, and outputs will be released for adaptation to national reporting archives.
[IR-10] A Redundancy Reduction Approach for Controllable Sequential Recommendations
链接: https://arxiv.org/abs/2609.23849
作者: Veronika Ivanova,Marina Munkhoeva,Ivan Razvorotnev,Evgeny Frolov
类目: Information Retrieval (cs.IR)
备注:
Abstract:Sequential recommendation must operate under long-tailed item distributions and popularity-driven concentration, often forcing practitioners to trade short-list accuracy against long-tail exposure. In this work, we study feature decorrelation as a mechanism for shaping representation geometry in dot-product sequential recommenders, and analyze how this, in turn, affects popularity-driven concentration. We propose a decorrelation-regularized training framework that augments next-item prediction with an auxiliary redundancy-reduction term, and instantiate it with BT-SR, which uses the Barlow Twins objective. To form label-consistent positive pairs without synthetic corruptions, we pair user histories that share the same next-item target. Beyond accuracy, we provide a geometric analysis showing how decorrelation suppresses shared low-rank directions in the user representation space that can give popular items a global scoring advantage, and we introduce a bucket-based alignment concentration metric to quantify this effect. Experiments on five public benchmarks show that BT-SR consistently improves next-item ranking quality, while the decorrelation strength acts as a simple control knob that reallocates accuracy across head and tail items, enabling accuracy-exposure trade-offs. Our analysis also reveals that the impact on head-vs-tail exposure differs across datasets, reflecting interactions between decorrelation and data temporal structure.
[IR-11] UNIQUE: A Unified Retrieval and Ranking System for Large-Scale Feed Recommendation
链接: https://arxiv.org/abs/2609.23718
作者: Zhuang Liu,Yongkang Fu,Zuodong Yang,Guangxing Chen,Zonggang Wu,Yuqi Lu,Shouke Qin,Shantao Li,Maolin Wang
类目: Information Retrieval (cs.IR)
备注:
Abstract:Industrial mobile feed systems rely on a retrieval-ranking pipeline to serve large-scale, heterogeneous, and fast-changing content under strict latency constraints. However, existing pipelines still suffer from two critical issues: hierarchical quantization instability in candidate retrieval and information loss between separated retrieval and ranking stages. These issues hurt long-tail and cold-start recommendation and complicate efficient serving. To address them, we present UNIQUE, a unified retrieval and ranking recommendation framework with single-layer flat quantization. UNIQUE integrates generative code-based retrieval and target-aware ranking into one early-fusion architecture, enabling end-to-end training under a shared representation while preserving efficient candidate generation. A balanced quantization mechanism is further introduced to mitigate codebook imbalance and improve long-tail representation. Offline experiments evaluate UNIQUE from both retrieval and ranking perspectives, while codebook analysis shows more balanced resource allocation than hierarchical quantization. We deploy UNIQUE in the homepage feed, discovery-page, and short-video recommendation scenarios of Mobile Baidu, serving large-scale real-world traffic. Online A/B tests achieve a 0.96% gain in total watch duration and a 1.08% gain in total distribution volume, with notable improvements for new users and highly active users. Serving measurements show 89 ms P99 latency and 44.23% online inference MFU. These results show that UNIQUE provides a stable, efficient, and production-ready framework for unified retrieval and ranking in industrial recommendation.
[IR-12] MuSeR: Scalable Long-sequence Recommendation with Multi-interest Modeling
链接: https://arxiv.org/abs/2609.23677
作者: Yongkang Fu,Beining Bao,Yu Jiang,Xiangyu Zhao,Hongyang Wei,Guangxing Chen,Zuodong Yang,Shantao Li,Zonggang Wu,Yuqi Lu,Shouke Qin,Hanmeng Liu,Maolin Wang
类目: Information Retrieval (cs.IR)
备注:
Abstract:Ultra-long user behavior sequences carry rich signals of stable and diverse preferences, yet industrial recommender systems typically truncate histories to a few hundred actions under strict latency and memory budgets, leaving long-term interests under-utilized. Users also pursue multiple heterogeneous intents across modalities such as news, QA, and short video, which sparse ID embeddings alone struggle to represent. We present Multi-interest Sequence Representation (MuSeR), a retrieval framework built on the deployed MGS system, which integrates three components: (i) hierarchical temporal compression, which retains recent actions at full resolution while progressively pooling older segments, so that per-user histories of 10^4 - 10^5 interactions fit within a fixed serving budget; (ii) disentangled multi-query interest extraction with orthogonality regularization; and (iii) multimodal semantic alignment, which augments sparse item IDs with textual summaries distilled from a large language model. For industrial deployment, MuSeR further adopts asynchronous user-representation refresh with adaptive caching and hierarchical beam-search retrieval across heterogeneous hardware. On three public benchmarks and a large-scale industrial dataset, MuSeR consistently improves Recall@ K over strong long-sequence and multi-interest baselines. In online A/B tests on Baidu APP’s homepage feed, discovery feed, and short-video scenarios, MuSeR yields +0.26% daily active users and +0.89% total session duration (both statistically significant, p0.05), alongside reduced serving latency and cost. Rather than proposing a new modeling primitive, our contribution is a system-level integration that makes long-term, multi-interest, and multimodal modeling jointly deployable in a real-time production pipeline, together with the engineering practices required to sustain it.
[IR-13] Beyond Relevance: Structured Semantic Supervision for Product Search with LLM -Augmented Annotations
链接: https://arxiv.org/abs/2609.23646
作者: Girish A. Koushik,Swapnil Bhosale,Samarth Agrawal,Hadeel Sadany,Constantin Orasan,Xiatian Zhu,Diptesh Kanojia
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL)
备注: 16 pages, 2 figures
Abstract:E-commerce search requires distinguishing products that are merely related to a query from those that directly satisfy the user’s shopping intent. We augment query-product pairs with structured LLM-generated query and product attributes and human-validated relevance, explanations, and centrality judgments, and evaluate these signals using a simple dual-encoder retriever and MLP re-ranker. On an augmented subset of ESCI, a human-feature oracle reaches 0.9382 nDCG@10, while a human-free trained Q+P configuration reaches 0.9258 . Synthetic approximations of the human signals reach 0.9150 overall but provide substantial gains for difficult, low-performing queries. Ablations show that most of the oracle improvement comes from post-edited explanations and annotator comments rather than the scalar centrality feature, suggesting that LLMs are most useful for exposing and approximating structured semantic supervision rather than replacing human judgment directly.
[IR-14] PSD: Pseudo Self-Distillation of Memory Representation Capabilities for LLM Agents
链接: https://arxiv.org/abs/2609.23449
作者: Pirzada Suhail,Menglin Xia,Xuchao Zhang,Mayukh Das,Chetan Bansal,Saravan Rajmohan
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Memory systems are becoming a core component of LLM agents, but constructing and maintaining memory remains expensive because it relies on repeated calls to large proprietary language models. This cost creates a major barrier to deploying memory-enhanced agents at scale. In this paper, we present Pseudo Self-Distillation (PSD), a framework that enables small language models (SLMs) to construct hierarchical memory representations by distilling behavior from a strong black-box oracle through a multi-stage training pipeline. Standard distillation methods require access to teacher logits or hidden states, which closed models do not expose. Unlike conventional self-distillation settings, where supervision is derived from a model’s own predictions, sampled rollouts, or aggregated outputs, PSD enables a single-model distillation setup while channeling external oracle knowledge through the prompt. PSD uses a single small model in two roles: a teacher that sees a privileged prompt containing the oracle’s answer as reference context, and a student that sees only the task prompt. The student learns to reproduce the teacher’s output distribution, absorbing oracle-guided behavior into its own weights without accessing the oracle’s internals. On LoCoMo, PSD-trained Qwen3-0.6B, 1.7B, and 4B match or exceed GPT-4.1-mini on downstream retrieval at a fraction of the deployment cost, with off-policy PSD achieving the strongest results across most conditions. We further show that this memory-construction capability transfers out-of-distribution to LongMemEval, despite the students being trained exclusively on LoCoMo with no exposure to LongMemEval data.
[IR-15] From Ranked Documents to Reliable Contexts: An Answer-Oriented Context Construct Framework for AI Search
链接: https://arxiv.org/abs/2609.23354
作者: Yunfei Zhong,Yinqiong Cai,Lixin Su,Haosheng Qian,Lixin Zou,Yixing Fan,Sheng Xu,Jiafeng Guo,Daiting Shi,Jingzhou He
类目: Information Retrieval (cs.IR)
备注:
Abstract:Traditional Web search follows a human-facing paradigm in which users inspect ranked documents and synthesize information themselves. In AI Search, retrieved documents instead serve as inputs to a generation model, shifting the retrieval objective from ranking documents by Search Satisfaction to constructing reliable context for correct answer generation. We formulate this shift as answer-oriented context construction through a three-stage framework: (1) Answer Support identifies candidate documents that contribute information to answer generation; (2) Content Trustworthiness assesses whether this information provides a reliable basis for correct answers from source, temporal, and factual perspectives; and (3) Context Organization selects, consolidates, and structures retained information under a finite context budget for consistent and robust generation. We further develop an industrial workflow spanning prior and posterior optimization and establish a systematic evaluation protocol covering both retrieval-side context and final answers. Experiments show consistent improvements at both Retrieval and Answer levels, demonstrating the effectiveness of the framework and its industrial implementation.
[IR-16] Semantic Candidate-Job Matching: A Comparative Evaluation of Dense Embedding Models in Hybrid Retrieval
链接: https://arxiv.org/abs/2609.23307
作者: Sai Yashwant,Siddhartha Jain,Anurag Dubey,Samaroha Chatterjee,Gantala Thulsiram
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
备注:
Abstract:This paper presents a comparative evaluation of dense embedding models for semantic candidate-job matching in high-volume staffing workflows. Incoming job descriptions are converted into structured English search text and language-specific keywords through LLM-based parsing, and candidate profiles are indexed as semantically enriched resume representations. We evaluate EmbeddingGemma (base) against EmbeddingGemma fine-tuned with Cached Multiple Negatives Ranking Loss (MNRL) within a unified hybrid retrieval pipeline that fuses vector similarity and full-text relevance via reciprocal rank fusion (RRF), and benchmark both against the MPNet model on a batch comparative evaluation dataset scored through the deployed job-candidate matching scoring pipeline. We further document, with mathematical detail, the broader set of contrastive fine-tuning objectives considered during model development (including AnglE/CoSENT-style refinement) and the empirical rationale for retaining Cached-MNRL-only adaptation as the preferred configuration. To support reproducible model selection, we define a broader evaluation framework comprising standard information retrieval metrics (Recall@K, mean reciprocal rank, nDCG) under the exact hybrid-retrieval protocol; the metrics used for the evaluation reported in this paper are fine-tuning convergence diagnostics and a batch comparative evaluation using the deployed AI-Match score and an independent LLM-as-a-Judge relevance score, and we state this scope explicitly rather than implying the full framework was measured. The paper addresses the gap between general-purpose embedding benchmarks and enterprise job-candidate matching constraints, providing a structured basis for comparing embedding strategies under realistic job-candidate retrieval conditions.
[IR-17] From Prompt to Recommendation: A Fitted Stage Model of Brand Visibility in AI Search
链接: https://arxiv.org/abs/2609.23162
作者: Benjamin Tannenbaum
类目: Information Retrieval (cs.IR)
备注: 29 pages, 10 figures. Includes aggregate results and figure-reproduction code
Abstract:We analyze 34,960 unbranded prompt-engine observations from 75 anonymized Aiso projects, covering 2,854 distinct monitored prompts and repeated GPT and Gemini runs from June-September 2026. When neither the target brand nor its own domain appears in the observable live retrieval path, target mention rates are 2.8% for GPT and 3.8% for Gemini. With an own-domain citation but no branded fan-out, they rise to 49.0% and 58.4%. When both own-domain exposure and a branded fan-out occur, mention rates reach 91.4% and 100%. The relationship persists within the same project, prompt, and engine across repeated runs: among prompt cells that vary in own-domain exposure while holding branded fan-out absent, exposure is associated with a mean mention-rate increase of 40.2 percentage points on GPT and 49.0 points on Gemini. Prior visibility is independently persistent. A previous non-mention plus no current own-domain exposure yields next-run mention rates of 1.6% and 1.9%; previous mention plus current exposure yields 80.5% and 83.7%. We fit a chronological diagnostic model using prior-run history and contemporaneous retrieval indicators: \operatornamelogitP(M_t=1)=\alpha_e+\beta_e\operatornamelogit(\widetilde P_t-1)+\gamma_e E_t+\delta_e F_t+\theta_e^\top X. On the latest 30% holdout, the full model achieves AUC 0.963 on GPT and 0.942 on Gemini, compared with 0.937/0.917 for prior history alone and 0.880/0.840 for live signals alone. A manually curated prompt sensitivity gives nearly identical AUCs (0.960 and 0.943). A separate 199-prompt page-corpus validation finds that prompt-page match predicts Gemini exposure (AUC 0.641) more clearly than GPT exposure (0.545), placing relevance upstream of a larger engine-mediated exposure effect. The equation is predictive and observational, not a causal description of proprietary engine internals. Comments: 29 pages, 10 figures. Includes aggregate results and figure-reproduction code Subjects: Information Retrieval (cs.IR) Cite as: arXiv:2609.23162 [cs.IR] (or arXiv:2609.23162v1 [cs.IR] for this version) https://doi.org/10.48550/arXiv.2609.23162 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[IR-18] MM-ContextFold: Context Folding for Multimodal Agent ic Retrieval
链接: https://arxiv.org/abs/2609.23121
作者: Yang Tian,Fan Liu,Jingyuan Zhang,Zhenyang Li,Yupeng Hu,Liqiang Nie
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR); Multimedia (cs.MM)
备注:
Abstract:Multimodal Agentic Retrieval (MAR) requires agents to solve complex information-seeking tasks by iteratively invoking external tools. Typical frameworks such as ReAct maintain raw multimodal inputs and the accumulating interaction history in a single, ever-growing context, leading to the context explosion problem. While existing methods alleviate this issue by compressing redundant text, effective strategies for managing token-intensive visual content remain largely underexplored. To address this gap, we first conduct a systematic empirical study of approximately 10,000 trajectories. The results show that as visual cues are progressively extracted through external tools and textualized into the context, raw images become increasingly redundant. Continued image retention is associated with higher output entropy and can even degrade task accuracy. Motivated by these findings, we propose MM-ContextFold, a training-free framework that loads raw images only when needed. It maintains a persistent, text-only main context for high-level planning and spawns ephemeral branch contexts for image-dependent subtasks. Within each branch, the agent loads the relevant images, completes the subtask, and folds the result back into the main context as a concise textual summary; the images and branch trace are then discarded. Experiments on seven MAR benchmarks across five backbone models show that MM-ContextFold improves average accuracy by 6.3 percentage points over ReAct while reducing the working context length by 27.5%.
[IR-19] Improving disruptive research in the EU: why strengthening European Research Council grants alone is not enough
链接: https://arxiv.org/abs/2609.23115
作者: Alonso Rodríguez-Navarro
类目: Information Retrieval (cs.IR)
备注: 15 pages, 5 figures, 6 tables
Abstract:Disruptive innovation in the EU is not sufficiently competitive; this weakness puts at risk the social benefits that its citizens take for granted. This report argues that, in addition to addressing structural and economic deficiencies, the EU must improve disruptive research to strengthen its disruptive innovation capacity. Currently, the level of disruptive research is too low. Using graphene research as an example, for which the EU has a specific programme, this report shows that Germany, France, Italy, and Spain cannot compete with Singapore. Even more concerning, the research funded by the European Research Council on graphene fails to compete with research conducted in Singapore. Similarly, the EU is far from competing with the USA or China. A few examples in this report and cited references evidence that the situation is similar in other technologies. To overcome this situation, the EU must adopt drastic changes in research policy. However, such changes face a vanity culture among policymakers and, perhaps, scientists who have been proclaiming an inexistent research excellence for decades. Without drastic changes, the prospect of the EU becoming a technological leader at the level of the USA and China cannot be considered realistic.
[IR-20] Inherit4Rec: Parameter Inheritance for Efficient Scaling of Recommendation Models
链接: https://arxiv.org/abs/2609.23111
作者: Ruihao Zhang,Bo Chen,Xiao Wang,Jinlong Jiao,Tijian Hu,Qinglin Jia,Xiuqiang He,Xiangyu Zhao,Chaoyi Ma,Ruiming Tang,Wenwu Ou
类目: Information Retrieval (cs.IR)
备注:
Abstract:Scaling model capacity has emerged as an effective approach to overcoming performance bottlenecks in industrial recommender systems. However, repeatedly training larger dense models from scratch demands substantial data and time, while their growing computation conflicts with the strict serving budgets of industrial systems. Parameter inheritance provides a promising route for both dense model growth and sparse conversion, yet existing methods are primarily designed for static corpora and can suffer sharp performance drops under dynamically evolving recommendation data. To address these challenges, we propose Inherit4Rec, a parameter-inheritance framework that supports both Dense-to-Dense (D2D) growth and Dense-to-Sparse (D2S) conversion. Inherit4Rec-D2D combines hybrid growth with asymmetric training to preserve the forward function at expansion and maintain update continuity. Inherit4Rec-D2S constructs SMoE networks through co-activation-aware partitioning and a load-balancing loss, preserving dense-model capabilities while promoting balanced expert activation. Experiments on KuaiRand-1K and an industrial short-video recommendation dataset show that both transformations consistently outperform the evaluated inheritance baselines across all prediction objectives. These results demonstrate the effectiveness of Inherit4Rec for continual capacity expansion and computation-efficient sparse conversion in industrial recommender systems.
[IR-21] Bridging Static and Agent ic RAG for Taiwanese Historical Question Answering
链接: https://arxiv.org/abs/2609.23056
作者: Kai-Hsin Chen,Wei-Yu Chen,Xuanjun Chen,Jyh-Shing Roger Jang
类目: Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注:
Abstract:Agentic retrieval-augmented generation (RAG) enables language models to adapt retrieval based on previously retrieved evidence, but it remains unclear whether such adaptive orchestration consistently outperforms well-designed static pipelines. We conduct a controlled comparison of agentic and static RAG for Taiwanese historical question answering, sharing the same generator and hybrid retrieval backend. Despite similar aggregate performance, the two pipelines differ on 70.83% of questions, with their advantages largely canceling out when averaged. An oracle that selects the better response per question improves the composite score by 0.2417 over the better individual pipeline, revealing substantial headroom for question-level selection. We therefore introduce a post-hoc selector that compares the two responses and their cited evidence, significantly outperforming either individual pipeline and recovering 60.34% of the oracle headroom. These results show that aggregate comparisons can obscure meaningful question-level differences between retrieval strategies, suggesting that exploiting their complementarity may be more fruitful than seeking a universally superior pipeline.
[IR-22] Attributable Post-Rationalization in RAG Citations: A Controlled Reproduction and an RLVR Comparison
链接: https://arxiv.org/abs/2609.23053
作者: Mehedi Khan,Md. Shariful Islam Bhuyan
类目: Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注: 11 pages, 1 figure, code available at: this https URL
Abstract:A RAG system can hand you the right answer and cite a source it did not actually use. Models output these unfaithful citations via post-rationalization: they write the answer first and then attach a citation to whatever passage looks close enough. Search agents are now trained with reinforcement learning from verifiable rewards (RLVR), which pays them for getting the answer right. We asked whether that training also teaches them to cite honestly. Improving an existing methodology with a required control, we compared an instruction-tuned model against three RLVR agents trained from it, on four question-answering datasets, using only free-tier Kaggle GPUs. Post-rationalization is everywhere: on Wikipedia-based questions roughly one citation in seven is unfaithful. RLVR does not fix it. The agents post-rationalize at their base model’s rate, and one lands slightly worse. Rewarding correct answers buys nothing in citation faithfulness, so faithfulness has to be trained and measured on its own terms. Comments: 11 pages, 1 figure, code available at: this https URL Subjects: Computation and Language (cs.CL); Information Retrieval (cs.IR) Cite as: arXiv:2609.23053 [cs.CL] (or arXiv:2609.23053v1 [cs.CL] for this version) https://doi.org/10.48550/arXiv.2609.23053 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[IR-23] R-GEAN: Regimen-Guided Edit Action Network for Within-Admission Medication Change Prediction
链接: https://arxiv.org/abs/2609.22959
作者: Regan Mahat,Mansu Kim
类目: Artificial Intelligence (cs.AI); Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注:
Abstract:The medications prescribed to a patient often change during a hospital admission as clinicians start, stop, or continue therapies. We study whether models can predict which medication classes are added or removed between 24 hours after admission and discharge. Metrics that compare the complete discharge regimen can reward models for copying medications that remain unchanged, even when they identify no actual changes. We therefore introduce a leakage-controlled benchmark that predicts net ATC3 additions and removals using only prior completed admissions and information available within the first 24 hours of the current admission. Addition candidates are classes not active at 24 hours, whereas removal candidates are classes active at that time. We also introduce R-GEAN, an asymmetric candidate-scoring network with independent addition and removal predictors. Across 240,480 admissions from 82,286 patients, R-GEAN achieves the highest predefined summary of addition, removal, changed-regimen, and action-pattern performance, termed the edit composite (0.464), compared with 0.435 for the strongest primary comparator. Reimplemented RETAIN, GAMENet, and MICRON baselines obtain 0.428, 0.420, and 0.288, respectively. R-GEAN’s advantage is concentrated in correctly identifying medication classes no longer active at discharge, while rare additions and admissions with multiple medication changes remain difficult. Rankings based on micro-F1 over the reconstructed discharge regimen and the edit composite correlate weakly across the evaluated models (Spearman r = 0.20). The continuation baseline achieves the highest complete-regimen score despite predicting no additions or removals. These results show that complete-regimen and edit-level evaluation measure different aspects of medication prediction. The benchmark evaluates observed prescribing changes, not treatment appropriateness
[IR-24] Per-Query Gating of LLM Rerankers for Multi-Hop Retrieval
链接: https://arxiv.org/abs/2609.22880
作者: Andre Bacellar
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注: 15 pages, 2 figures, 13 tables. v2 adds a loss-budget (calibrated) gating rule with expected versus realised harm, and a retrained deployment policy. Evaluation code, per-query results and generated tables in the source archive
Abstract:LLM rerankers add of the order of \ 0.2-0.3 per 1,000 queries and about a second of tail latency on top of a graph-augmented dense pipeline such as HippoRAG2, and on three multi-hop benchmarks they improve final-hop top-K coverage on seven of nine (dataset, K) cells, by up to +34.8 pp. We ask whether a learned per-query gate can skip the reranker where it will not help, using only features available before the LLM call (27 score and lexical statistics of the two retrieval lists plus a PCA of a small query embedding) with an executable fallback. Every choice, including the fallback and the threshold, is made inside the training fold and applied once to held-out queries, and harmful skips (the rerank would have found the target, the fallback did not) are reported next to the aggregate coverage. Across nine cells on 2WikiMultiHopQA, MuSiQue and HotpotQA the gate skips 51% of calls at an average held-out LastHop@K cost of 1.2 pp; four cells meet a pre-registered 1 pp rule, harmful skips occur in eight (190 harmful against 136 beneficial), and a random gate at the same skip rate loses 2 to 11 pp on the high-lift cells. A second rule sets each cell’s threshold from a pre-specified budget on the expected harmful-skip rate over Platt-calibrated harm probabilities (ECE 0.025 after calibration, 0.094 before): at a 1 pp budget the gate skips 42% at -0.8 pp with 66 harmful skips and six cells within 1 pp, but realised harm exceeds the promise in six cells (mean 1.45 vs 0.83 pp), a selection optimism we quantify; a 0.5 pp budget realises about 1 pp. The harm probabilities are calibrated but barely discriminative (AUC 0.16 to 0.70). An earlier version reported 73% “lossless” savings; that figure rested on an oracle fallback and a wrong MuSiQue target, and we document both.
[IR-25] Parameterized Dense-Sparse Fusion for Hybrid Retrieval: Tuning a Rank-Score Mix on BEIR SciFact with Qdrant
链接: https://arxiv.org/abs/2609.22770
作者: Satyanarayan Pati,Srikanth Patil
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
备注:
Abstract:We study a parameterized hybrid ranker that fuses a dense embedding list and a sparse lexical list. The method has a small, explicit parameter vector: a dense prior \alpha \in [0,1] , a score-versus-rank mix \lambda \in [0,1] , an RRF smoothing parameter \kappa 0 , optional list-geometry coefficients that move \alpha per query, and a router margin \tau that can turn sparse search off. We grid-search those ranges on SciFact train (809 queries) and freeze the chosen values on SciFact test (300). The tuned rank-score mix ( \alpha = 0.8 , \lambda = 0.75 , \kappa = 20 ) reaches 0.753 nDCG@10 and 0.889 recall@10, outperforming dense BGE (0.742 / 0.871) and equal-weight RRF (0.707 nDCG@10) on that test split. A list-conditioned \alpha adds +0.0006 nDCG; a sparse-off router is rejected by the same train split (any \tau that skipped approximately 50% of queries lost nDCG). These coefficients are dataset-specific. Equal RRF with the same models does not beat dense on a nine-zip BEIR macro-average (0.479 vs. 0.519 nDCG@10). Repeating the same train-then-freeze sweep independently on all 20 indexed units beats equal RRF on 20/20 and dense on 16/20 (unit-mean nDCG@10 0.467 vs. 0.462 dense vs. 0.420 RRF). Other corpora should reuse the ranges, not a copy of the SciFact point.
[IR-26] Beyond Raw Engagement: A Counterfactual Observability Framework for Recommender Systems at Netflix
链接: https://arxiv.org/abs/2609.22747
作者: Chaoran Guo,Ding Tong,Ting-Po Lee,Scarlet Chen
类目: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI)
备注:
Abstract:Understanding the performance of large-scale recommender systems remains an underexplored challenge, especially for content creators and model developers. The raw engagement signals available to them, such as views and clicks, conflate content quality, model behavior, presentation bias, and audience reach, making it hard to attribute outcomes to the right cause. In this work, we present a general evaluation framework that enhances observability across multiple recommender systems at Netflix and demonstrate its effectiveness through several production deployments. The framework treats recommender-system observability as a counterfactual measurement problem: estimating what the recommender would have done, and what engagement would have followed, in the absence of a specific content item or model decision. We articulate three stakeholder-centered observability principles for content creators and model developers, and propose measurement methodologies covering bias reduction, relativity, and incrementality, applicable to both single-stage and cascading recommender systems and serving both audiences from a single measurement foundation. Subjects: Information Retrieval (cs.IR); Artificial Intelligence (cs.AI) Cite as: arXiv:2609.22747 [cs.IR] (or arXiv:2609.22747v1 [cs.IR] for this version) https://doi.org/10.48550/arXiv.2609.22747 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[IR-27] DOA-SORT: Directional Occlusion-Aware Multi-Object Tracking with Distributional Observations
链接: https://arxiv.org/abs/2609.22706
作者: Hao Wang
类目: Computer Vision and Pattern Recognition (cs.CV); Information Retrieval (cs.IR)
备注:
Abstract:Identity association in multi-object tracking (MOT) is vulnerable to partial occlusion, truncated detections, and fluctuating confidence scores. Existing motion-dominant trackers commonly represent occlusion as a scalar penalty. This treatment misses the directional observation bias caused by occlusion: left, right, top, and bottom occlusions distort the location and shape of a detection in different ways. We propose \ours (Directional Occlusion-Aware SORT), an online and training-free tracker that models these biases explicitly. First, it infers a soft front–back ordering from box overlap and relative bottom positions, and estimates directional occlusion coverage and depth. It then constructs a mixture of one clean and four directional occlusion observation components. The model uses a five-dimensional observation comprising box center, area, confidence, and aspect ratio, and adapts observation noise to predicted occlusion and detection confidence. The directional mixture likelihood is used in high-confidence association, low-confidence association, and track recovery; ambiguity penalties and local order-consistency swaps further reduce identity errors among nearby objects. On the DanceTrack validation split, \ours improves HOTA from 63.00 to 66.34, AssA from 45.10 to 49.57, and IDF1 from 62.19 to 65.28 over OA-SORT with the same detector and evaluation protocol. The gains are concentrated in association quality while detection accuracy remains stable. Additional local evaluations on MOT17 and MOT20 train splits characterize cross-dataset behavior under the same no-ReID tracking protocol.
[IR-28] Scoring With the Engine: Retrieval Exposure Cross-Engine Divergence and the Limits of Engine-Agnostic GEO Scores
链接: https://arxiv.org/abs/2609.22655
作者: Benjamin Tannenbaum
类目: Information Retrieval (cs.IR)
备注: 20 pages, 6 figures, 6 tables. Builds on and complements arXiv:2609.07559 ; cites and extends the measurement program in arXiv:2607.18904 , arXiv:2607.22392 , arXiv:2608.02556 , and arXiv:2609.09878
Abstract:Recent work asks whether generative-engine visibility can be approximated with deterministic, engine-free page scores. We separate two stages such scores can conflate: exposure to a live engine and citation selection conditional on exposure. In an observational audit of ChatGPT, Microsoft Copilot, Google, and Perplexity, 15 fixed commercial prompts produced 589 citation observations on 6 June 2026, corresponding to 528 unique URLs and 356 domains. Same-prompt cross-engine URL overlap was extremely small: mean pairwise Jaccard similarity was 0.0079, the median was zero, and 84.9% of engine pairs shared no cited URL. On the ten prompts observed on all four engines, mean exact-URL Jaccard was 0.0072. A matched-size hypergeometric baseline preserving each prompt’s four-engine URL universe and each engine’s list length predicts 0.1272, so observed overlap was only 5.7% of that baseline; zero URL overlap occurred in 86.7% of comparisons versus 12.3% expected. Top-five exact-URL overlap was zero in all 60 pairwise comparisons. A single engine captured only 11.4%-42.6% of the four-engine URL union, and 96.4% of observed URLs appeared in only one engine. A separate 5-to-6 June same-engine comparison found 67.0% mean URL-set turnover. These results do not invalidate engine-free page scoring; they identify its estimand. A score computed without a live engine can estimate page quality or query-page fit, while end-to-end visibility additionally depends on engine-specific exposure and selection. We therefore argue for reporting page fit, observed exposure, conditional selection, and final visibility as distinct quantities.
[IR-29] AdaMerge: Tuning-Free Patch Compression for Multi-Vector Visual Document Retrieval CIKM2026
链接: https://arxiv.org/abs/2609.22562
作者: Jianxin You,Kun Ni
类目: Computer Vision and Pattern Recognition (cs.CV); Information Retrieval (cs.IR)
备注: 5 pages, 3 figures. Accepted as a short paper at ACM CIKM 2026
Abstract:Multi-vector visual document retrieval (VDR) models such as ColPali and ColNomic achieve strong accuracy by representing each document with hundreds to thousands of patch-level embeddings, at substantial storage and latency cost. Existing compression methods either prune unimportant patches or merge similar ones into clusters; the recent state-of-the-art merging method Prune-then-Merge (PtM) consistently outperforms pruning-only baselines at high compression, but requires a per-dataset cluster budget m to be tuned by grid search. We observe that the merge-cosine sequence produced by hierarchical clustering exhibits a sharp cliff separating mergeable redundancy from salient signal, and that the location of this cliff is concentrated in a narrow band across more than 11,000 documents from 14 datasets. This suggests the merge boundary can be detected per document rather than tuned per dataset. Building on this observation, we propose AdaMerge, a plug-and-play compression method that (i) detects each document’s own cliff via gap analysis on the merge-cosine trajectory, and (ii) builds attention-weighted cluster centroids to preserve salient signal. On the long-document benchmark ViDoRe-V2 (4 datasets, two backbones), AdaMerge significantly outperforms tuned PtM across the operating range (p 10^-4); on the short-document benchmark ViDoRe-V1 (10 datasets, two backbones), where all merging methods are already near-lossless, AdaMerge matches tuned PtM without any per-dataset tuning. AdaMerge adds only about 10 ms per document and exposes a single global hyperparameter shared across all datasets and backbones.
[IR-30] IntLawNER: A Named Entity Recognition Dataset and Benchmark in International Law
链接: https://arxiv.org/abs/2609.22529
作者: Genis Skura,Roland Bouffanais,Didier Wernli
类目: Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注:
Abstract:International law provides the normative framework through which states coordinate action, regulate armed conflict, and protect human rights, yet its texts remain without token-level named entity recognition (NER) resources. We introduce IntLawNER, a NER dataset and benchmark for codified sources of international law, covering 2,987 gold-annotated sentences and 8,094 entity spans from International Court of Justice (ICJ) decisions, UN Security Council resolutions, and European Court of Human Rights (ECtHR) judgments, annotated with seven institution-specific entity types. We construct IntLawNER with a cost-effective hybrid algorithmic-agentic pipeline that reduces 468k source sentences to a compact annotation set through candidate retrieval, LLM-based vetting, and human review, with 89.6% of gold spans accepted unchanged from the silver layer. However, the silver-to-gold analysis reveals that human-machine aggregate agreement metrics can be misleading in domain-specific NER: Cohen’s kappa=0.964 on boundary-matched spans masks a macro-F1 of 0.753 when missing entities, boundary errors, and label corrections are included. The benchmark shows that zero-shot span-based GLiNER collapses on entity types dependent on institutional function rather than surface form (0.243 micro-F1), while fine-tuned transformers struggle on rare labels. Carefully selected few-shot examples that demonstrate label contrasts improve every LLM over zero-shot prompting, with Claude Opus 4.6 reaching the best score of 0.873 micro-F1. We release IntLawNER as a benchmark and reusable resource for extracting references in international legal texts.
[IR-31] Semantics Delivery Network: Rethinking Web Retrieval Infrastructure for LLM Agents
链接: https://arxiv.org/abs/2609.22486
作者: Peichun Hua,Yunming Xiao
类目: Networking and Internet Architecture (cs.NI); Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注: 12 pages, 3 figures
Abstract:Large language models (LLMs) increasingly rely on external sources when answering questions that require proprietary information or up-to-date live web content, through both traditional single-shot retrieval-augmented generation (RAG) and multi-turn agentic RAG. Yet today’s web infrastructure is still built for human clients. Given a query, current search services return a list of URLs and snippets ranked for generic relevance; content delivery networks (CDNs) cache URL-addressed objects (texts, images, videos, etc.) without knowing which passage an agent needs. LLMs, in contrast, consume short, semantically coherent passages, hereafter “chunks”, selected for downstream task utility rather than similarity alone, and may retrieve statefully across reasoning turns. Uncoordinated agents also repeat search, data acquisition, and semantic processing, duplicating work that could be shared. We argue that semantic chunk retrieval should become a first-class network-delivery abstraction. We propose Semantics Delivery Network (SemDN): an origin-authorized, hierarchical edge substrate that indexes, searches, and smart-caches web content at chunk granularity. SemDN serves agents on behalf of participating websites, amortizes data acquisition and processing across agents, and supports tenant-specific retrieval policies. Because, unlike URL caching, semantic retrieval provides no explicit miss signal, SemDN must estimate when its enrolled corpus may be incomplete or stale and trigger scoped discovery or refresh. It raises open questions about shareable retrieval state, hierarchical caching, coverage risk, and deployment. Our preliminary probes reveal a large gap between page content processed and chunks consumed, substantial task-local reuse, and higher answer quality per context token from chunk delivery.
[IR-32] Checkpoints Are Not Enough: Trust Calibration in CoSLR a Human-AI System for Systematic Literature Reviews
链接: https://arxiv.org/abs/2609.22248
作者: MD Aidul Islam,Malik Abdul Sami,Muhammad Waseem,Zeeshan Rasheed,Kai-kristian Kemell,Zheying Zhang,Pekka Abrahamsson
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Information Retrieval (cs.IR)
备注:
Abstract:Systematic Literature Reviews (SLRs) are essential for evidence-based research but remain time-consuming, requiring researchers to manage large volumes of publications across planning, screening, analysis, and reporting. Large language models (LLMs) can now produce fluent, well-structured review text, which makes it difficult to distinguish synthesis that was verified by a researcher from synthesis that merely appears authoritative. This raises the risk that unverified AI-generated synthesis enters the scholarly record carrying the credibility of a systematic review. We present CoSLR, a Human-AI collaborative multi-agent system that supports the SLR workflow through a modular three-phase pipeline using large language models and Retrieval-Augmented Generation (RAG), and that places explicit, mandatory human checkpoints on the path between generated output and its acceptance. In a survey-based study with 63 participants, the system was received positively: 27 of 63 participants (42.9 percent) rated its usability highly, indicating that the mandatory checkpoints did not come at the cost of a workable interface. However, a checkpoint safeguards the review only if researchers use it to verify: 22 of 63 participants (34.9 percent) reported that they would trust AI-generated summaries and reports without additional human checking after only a short interaction with the system. These findings indicate that Human-AI collaboration can support literature review work, but that the effectiveness of human oversight depends on whether users are willing to exercise it. This is a calibration problem that interface design must address directly, not assume.
[IR-33] Guiding the coarse levels of semantic IDs makes the fine levels learnable
链接: https://arxiv.org/abs/2609.22227
作者: Bin Wang,Zhengyu Zhang
类目: Information Retrieval (cs.IR); Computation and Language (cs.CL); Machine Learning (cs.LG)
备注:
Abstract:Generative retrieval represents each item by a short Semantic ID and casts recommendation as autoregressive generation of that sequence. Because the tokenizer is trained independently to reconstruct an item embedding, its codes are aligned with neither the downstream LLM nor the end task. Nearly every SID system therefore spends extra effort to bridge this gap–alignment corpora, reasoning/RL, or per-token encoders to make codes legible, or learned tokenizer supervision to make them task-aware–yet the recovered meaning is content-derived and may not be the meaning the task needs. We introduce Guided SID, which instead makes the levels that matter most meaningful by construction: we force the coarse RQ-VAE levels to encode a predefined categorical attribute–chosen to be text-grounded (hence legible to the LLM) and task-relevant–by deterministic supervised index assignment (overriding nearest-neighbor selection with the attribute label) while keeping the codebooks learnable (they still receive reconstruction gradients). A trie-merge construction maps any high-cardinality or set-valued attribute onto the fixed code budget while keeping merged buckets semantically coherent. Guiding costs nothing intrinsically: collision and reconstruction match or beat the vanilla baseline despite pinning the coarse level. In a matched end-to-end A/B differing only in the SID encoding, the guided retriever improves recall@k at every list length we measure (1.36x at k=1, 1.39x at k=10), raises mean reciprocal rank from 0.0260 to 0.0355, and predicts the pre-defined attribute 4.2x more often.
[IR-34] SCoR: A Hierarchical Framework for Forecasting Relations Between Scientific Concepts
链接: https://arxiv.org/abs/2609.22174
作者: Jingze Wang,Fred Sun,Shangqi Guo
类目: Computation and Language (cs.CL); Information Retrieval (cs.IR); Machine Learning (cs.LG); Social and Information Networks (cs.SI)
备注: 15 pages, 7 figures, 6 tables; includes supplementary material. Jingze Wang and Fred Sun contributed equally; Shangqi Guo is the corresponding author
Abstract:Anticipating emerging research directions is a critical goal of AI-assisted science. Existing methods mainly predict which concepts will co-occur in future papers, but co-occurrence captures shared attention rather than the scientific meaning of a connection, such as whether one method uses, combines, replaces, or contradicts another. We formulate research-direction discovery as hierarchical scientific-relation forecasting over a shared candidate-pair space, comprising three temporally aligned tasks: first co-occurrence, first scientific-relation formation, and relation type at formation. We construct SCoR-Graph from 187,848 cs.CV papers published between 2017 and 2026, yielding 270,687 consolidated concepts, 7.45 million co-occurrence edges, and 615,036 typed, directed relation edges. From cutoff-specific graph snapshots, we derive SCoR-Bench, a leakage-audited benchmark for these three capabilities, with expert-verified gold labels for the entire relation-type test set. We further introduce HiSCoR, a task-adapted model family that models relation emergence as a temporally evolving, hierarchically constrained process by encoding pre-cutoff event histories and conditioning relation formation on future co-occurrence. On the held-out 2025-2026 window, HiSCoR achieves an AUROC of 0.9515, a 2.4% relative improvement over the strongest temporal-graph baseline, and improves population-AUPRC by 14.0%; its relation-type variant achieves a Macro-AUROC of 0.7795. Ablations show that semantic, co-occurrence, and typed-relation views provide complementary predictive evidence. SCoR advances research-direction forecasting from predicting which concepts will co-occur to anticipating whether and how evidence-backed scientific relations will emerge.
[IR-35] Quantifying Hidden Salt for Precision Healthcare: Sodium Assessment via Joint-Factor Retrieval and Chain-of-Thought Inference EMNLP2026
链接: https://arxiv.org/abs/2609.22171
作者: Mingyu Huang,Weiqing Min,Yuehui Fang,Yuna He,Shuqiang Jiang
类目: Computation and Language (cs.CL); Human-Computer Interaction (cs.HC); Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注: Accepted in EMNLP 2026 Main Conference; 16 pages, 7 figures
Abstract:Precision healthcare, particularly for conditions like hypertension and cardiovascular disease, necessitates monitoring of dietary sodium intake. However, tracking this is hindered by the prevalence of hidden salt in cooking, such as sodium in soy sauce and ketchup. While recipes offer a valuable data source for dietary analysis, sodium-rich seasonings are frequently omitted or described ambiguously in instructions. To solve this issue, we propose SALT, a Sodium Assessing Level Tracking framework adopting an RAG framework to assess sodium content in recipes. Our framework first introduces a Joint-Factor Embedding Retrieval module to locate similar recipes with specified sodium content for addressing the lack of contextual references. These retrieved samples provide contexts for subsequent inference. Then we design a structured 4-hop Chain-of-Thought inference module to refine the vague estimation from language models through a multi-step sodium estimation. To facilitate our study, we further construct a recipe dataset SALT54k with 54,151 entries labeled with sodium quantities across 11 common seasonings. Results on SALT54k demonstrate that our method achieves state-of-the-art performance in sodium estimation. Additional real-world validations confirm the effectiveness of our method, demonstrating its potential as a practical solution for AI-assisted precision healthcare.
[IR-36] ECP-Bench: Benchmarking and Learning Entertainment Content Promotion with Foundation Models
链接: https://arxiv.org/abs/2609.22150
作者: Hyomin Kim,Bowen Chen,Jin Huang,Zhao Wang,Qiaozhu Mei,Shingo Takamatsu
类目: Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注:
Abstract:Content promotion spans a broad set of skills, from understanding content to forecasting its market reception. However, LLMs’ ability to support such promotion decisions remains underexplored. Existing studies are often limited to a single task (e.g., popularity prediction) or a small set of tasks within a single domain (e.g., movies). As a result, there is a lack of understanding of LLMs’ abilities in the full promotion process and how these abilities generalize across different tasks and domains. In this work, we introduce ECP-Bench, a benchmark containing 1.9M movie, game, and music items and 423,451 questions across 33 tasks in five content-promotion skill families. Our evaluation shows that frontier models achieve only 51.9% overall accuracy and lose much of their advantage on post-cutoff content, with drops of up to 19.1 percentage points. In contrast, open-weight models fine-tuned on ECP-Bench achieve up to 60.3%, remain substantially more stable across the knowledge cutoff, generalize to unseen content and tasks, and exhibit meaningful cross-domain generalization.
[IR-37] DeepInstructor: An Agent ic AI Instructor for Experience-Driven Idea Evaluation
链接: https://arxiv.org/abs/2609.22104
作者: Rongcan Pei,Fang Guo,Qinglin Qi,Qi Zhu,Yun Luo,Jianhao Yan,Minjun Zhu,Qiujie Xie,Dehong Zheng,Yue Zhang
类目: Computation and Language (cs.CL); Information Retrieval (cs.IR)
备注:
Abstract:As automated scientific discovery advances, Large Language Models (LLMs) can now generate research ideas at an unprecedented scale, shifting the bottleneck from idea generation to idea evaluation. Existing evaluators mainly rely on parametric LLM knowledge or unstructured retrieval, producing judgments that lack the experience-grounded reasoning used by human instructors. To address this, we propose DeepInstructor, an agentic framework that formulates idea evaluation as reasoning over structured scholarly experience. DeepInstructor constructs an Experience Graph from 58,607 peer reviews and employs a ReAct-based agent to retrieve dimension-specific evidence for traceable evaluation. We further introduce DeepInstruct, a dataset with controlled pairwise comparisons across novelty, significance, and feasibility. Experiments show that DeepInstructor substantially outperforms existing baselines, improving Hit@1 and Hit@2 alignment with human judgments by 24.4% and 29.7%, respectively. Our findings suggest that scientific idea evaluation can be grounded in explicit reasoning over structured scholarly experience
[IR-38] AdaMem: Adaptive Memory Token Allocation for Soft Compression in Retrieval-Augmented Generation
链接: https://arxiv.org/abs/2609.22100
作者: Artem Sakhno,Grigorii Davydenko,Omar Zoloev,Julia Belikova,Andrey Savchenko,Maksim Makarenko
类目: Computation and Language (cs.CL); Information Retrieval (cs.IR); Machine Learning (cs.LG)
备注:
Abstract:Retrieval-augmented generation (RAG) improves language models with retrieved evidence, but processing many long passages is costly and can introduce distracting information. Soft compression addresses this challenge by encoding passages as compact sequences of continuous memory embeddings before generation. However, existing methods typically assign each retained passage an identical number of memory embeddings, irrespective of its query-specific relevance. To address this, we propose AdaMem, a relevance-guided soft-compression framework that maps learned passage-relevance estimates to a query-dependent allocation of a fixed memory-token budget. A shared query-conditioned compressor produces both continuous passage memories and relevance scores in a single pass; a deterministic allocation rule assigns more memory tokens to higher-scoring passages and can omit low-scoring ones. Across six open-domain QA benchmarks, AdaMem consistently outperforms OSCAR (the closely matched soft-compression baseline that uses uniform allocation) as well as other soft-compression methods at matched memory budgets. Under standard 16 \times compression, AdaMem improves sub-string match by up to 3.2 points (5.5%) over uniform allocation baseline, with an average relative gain of 3.4%; under aggressive 64 \times compression the average relative gain grows to 14.6%, with a maximum of 9.8 points (19.7%) on PopQA. AdaMem matches the answer quality of the uncompressed at up to 4 \times lower inference latency than full context baseline. AdaMem retains an efficiency profile comparable to the uniform-compression baseline, while achieving up to 4\times lower inference latency than full-context inference. Thus, relevance-guided memory allocation is particularly effective when retrieval pools are large and the available memory budget is tight.
人机交互
[HC-0] Passthrough Rigidity: The Behavioral and Visuomotor Costs of Mediated Perception
链接: https://arxiv.org/abs/2609.25002
作者: Markus D. Solbach,Mohit Goyal,Sakar Khattar,Jayant Varma,Bjorn Vlaskamp,John K. Tsotsos,Konstantine Tsotsos
类目: Human-Computer Interaction (cs.HC)
备注: 10 pages, 16 Figures, ISMAR
Abstract:Broad public adoption of head-mounted displays using video passthrough remains elusive despite significant market investment. A precise understanding of why users experience persistent discomfort even as hardware factors such as resolution and latency have dramatically improved remains an open issue. This paper investigates the impact of viewing the world through video passthrough systems on human behavioral and physiological patterns through a large-scale multimodal study. We developed a novel protocol to capture synchronized oculomotor, kinematic, and physiological data during a block assembly task requiring complex hand-eye coordination. Using a within-subject design (N=110), we evaluated both natural and passthrough viewing conditions. Our results reveal a four-fold suppression of rotational head velocity and a pronounced decoupling of head-gaze coordination. This suggests motor caution being employed as an adaptive strategy - which we term “Passthrough Rigidity”. This phenomenon appears to shift the information-gathering burden to the oculomotor system, resulting in significantly longer fixation durations and restricted visual search patterns. These kinematic shifts directly correlate with poorer task performance and measurable physiological cost, evidenced by a significant reduction in blink duration and increased reports of ocular strain and cognitive load. We conclude that current passthrough implementations induce a measurable shift from flexible exploration to motor caution, where task performance is preserved at the cost of user comfort and biomechanical efficiency. These findings provide a novel quantitative framework for evaluating and improving future XR devices, establishing that resolving “comfort” for passthrough requires addressing the deep-seated biomechanical compensations caused by mediated perception. Data available at this https URL
[HC-1] Who Does What in AI Auditing? Designing Human-AI Collaboration for Auditing Generative AI
链接: https://arxiv.org/abs/2609.24986
作者: Eunkyu Park,Markelle Roesti,Wesley Hanwen Deng,Renata Barreto,Mohammad Tahaei,Kenneth Holstein,Jason Hong,Motahhare Eslami
类目: Human-Computer Interaction (cs.HC)
备注: Preprint
Abstract:AI auditing increasingly incorporates AI agents to expand the scale and breadth of audit coverage, yet little is known about how auditing work should be divided without displacing human judgment. We introduce Human-Agent Audit Collaboration (HAAC), a workflow and system for structuring human-AI collaboration in AI auditing. Drawing on prior work and formative consultations with AI auditing practitioners, HAAC specifies how agents can support exploration, assessment, reporting, and review while preserving human oversight where contextual judgment is critical. We instantiate HAAC for conversational shopping agents and evaluate it through two studies. With 71 auditors, AI assistance increased attack success and broadened exploration, while also shaping later attacks and increasing auditors’ reliance on AI-generated assessments and reports. Interviews with Responsible AI practitioners showed that actionable audits require visibility into coverage, reproducible attack trajectories, and evaluation of the auditing agents themselves. Our findings identify design considerations for effective and accountable human-AI auditing.
[HC-2] onPanda: Efficient Annotation of On-Policy Alignment Data for LLM s and Agents via Token-Level Correction
链接: https://arxiv.org/abs/2609.24983
作者: Lei Yang,Mengyin Liu,Jia Wang,Hangyu Guo,Liang Zhao,Zheng Ge,Kang An,Binxing Jiao,Qi Han,Daxin Jiang,Siqi Shen,Xiangyu Zhang
类目: Computation and Language (cs.CL); Human-Computer Interaction (cs.HC); Machine Learning (cs.LG)
备注: Project page: this https URL
Abstract:We present onPanda, an interactive tool for efficiently annotating LLM alignment data and agent trajectories. onPanda adopts token-level correction as its core interaction: while reading a model response, the annotator locates the first inappropriate token and either picks a substitute from the model’s candidate tokens or types the correct text via free-form editing. The system then truncates everything after that position and continues generation from the corrected prefix, repeating this locate-correct-continue loop until a satisfactory response is obtained. This mechanism lets annotators precisely steer model outputs at low cost: a small controlled study suggests that onPanda reduces median annotation time by 52% over manual post-editing. Since the vast majority of tokens in the final response are generated by the model itself, the resulting data largely preserves the model’s sampling distribution and is well suited for constructing on-policy SFT and preference data. Furthermore, the token-level corrections recorded during annotation provide fine-grained supervision with precise positions and naturally paired positive–negative samples. onPanda also connects to external tools and harnesses, enabling interactive trajectory annotation in realistic environments. In addition, we release Panda-CVL, a dataset annotated with onPanda, together with a benchmark for token-level correction.
[HC-3] Generative Tutorial: Towards Live Contextualized Visual Instructions for Physical Tasks
链接: https://arxiv.org/abs/2609.24955
作者: Muzhe Wu,Zuchen Li,Xu Wang,Anhong Guo
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注:
Abstract:Visual instructions for physical tasks are typically authored in one context and followed in another, requiring users to translate demonstrated tools, materials, and spatial relationships into their own environment. We introduce Generative Tutorial, a conceptual framework for live visual instruction that depicts intended outcomes and actions within the user’s environment and task flow. A formative evaluation of state-of-the-art image and video generation identifies failures and potential benefits across 15 physical tasks. Drawing on these findings, we build an augmented-reality prototype system that proactively generates goal images and demonstration videos using observed workspace context and predicted visual outcomes of preceding actions. A 24-participant lab study found higher task performance quality, greater perceived workspace correspondence, and shorter step-confirmation intervals with the system than with pre-authored guidance. Qualitative findings highlighted how contextual resemblance shapes trust, how generation errors affect interpretation, and how guidance delivery should adapt to users’ needs, informing future designs.
[HC-4] Whose Facts Count? A Culturally Responsive Audit of LLM Evaluation Benchmarks
链接: https://arxiv.org/abs/2609.24934
作者: Fatima Tuz Zahra,Md. Sajeebul Islam Sk.,Rachel Chung
类目: Human-Computer Interaction (cs.HC)
备注: 36 pages, 8 figures
Abstract:LLM benchmarks function as evaluation instruments, informing decisions that affect education, labor, and public services worldwide. Drawing on Hood, Kirkhart, and Hopson’s culturally responsive evaluation (CRE) frameworks, this paper applies a six-dimension CR rubric to audit OpenAI’s SimpleQA (N = 4,326 items) and the LMSYS Chatbot Arena (N = 600 conversations). Every SimpleQA question requires English-language archival verification as its evidentiary basis. A single rater’s preoccupation with Colombian founding dates accounts for 2.70% of items, inflating the appearance of Global South coverage. English-language prompts constitute 76.3% of Arena conversations, against an International Telecommunication Union (ITU)-estimated 25.9% share of global internet users. A 50-item counter-benchmark scored a mean CR deficit nearly three times lower than SimpleQA (Cohen’s d = 1.01). The paper proposes a practical CR evaluation framework. These are structural validity failures, not incidental measurement problems, with direct consequences for communities whose knowledge traditions these instruments were not built to see.
[HC-5] EMooly: Supporting Autistic Children in Collaborative Social-Emotional Learning with Caregiver Participation through Interactive AI-infused and AR Activities
链接: https://arxiv.org/abs/2609.24899
作者: Yue Lyu,Di Liu,Pengcheng An,Xin Tong,Huan Zhang,Keiko Katsuragawa,Jian Zhao
类目: Human-Computer Interaction (cs.HC)
备注: 38 pages. Published in Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT), Vol. 8, No. 4, Article 203 (December 2024)
Abstract:Children with autism spectrum disorder (ASD) have social-emotional deficits that lead to difficulties in recognizing emotions as well as understanding and responding to social interactions. This study presents EMooly, a tablet game that actively involves caregivers and leverages augmented reality (AR) and generative AI (GenAI) to enhance social-emotional learning for autistic children. Through a year of collaborative effort with five domain experts, we developed EMooly that engages children through personalized social stories, interactive and fun activities, and enhanced caregiver participation, focusing on emotion understanding and facial expression recognition. Compared with a baseline, a controlled study with 24 autistic children and their caregivers showed EMooly significantly improved children’s emotion recognition skills and its novel features were preferred and appreciated. EMooly demonstrates the potential of AI and AR in enhancing social-emotional development for autistic children via prompt personalizing and engagement, and highlights the importance of caregiver involvement for optimal learning outcomes.
[HC-6] A Global Comparison of Schemas Transparency and Interoperability in Public-Sector AI Registers and Inventories
链接: https://arxiv.org/abs/2609.24883
作者: Dipto Das,Shion Guha
类目: Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Human-Computer Interaction (cs.HC)
备注:
Abstract:Artificial intelligence (AI) registers and inventories aim to make governmental AI visible, but their institutional scope, schemas, and reporting practices construct different representations of public-sector AI. We compare 8,368 records from country-specific and transnational inventories covering 72 countries. Across 23 harmonized fields, registers shared a descriptive core but rarely requested information about appeals, risks, legal bases, or external evaluation. We found that broad schemas often contained substantial missingness, schema similarity showed no significant patterned convergence, and multiple sources covering the same jurisdictions overlapped only selectively. Based on these findings, we synthesize a layered visibility framework that shows how register records reflect disclosure arrangements and why interoperability requires shared concepts, clear definitions, and preserved provenance.
[HC-7] ATCion: Exploring the Design of Icon-based Visual Aids for Enhancing In-cockpit Air Traffic Control Communication
链接: https://arxiv.org/abs/2609.24863
作者: Yue Lyu,Xizi Wang,Hanlu Ma,Yalong Yang,Jian Zhao
类目: Human-Computer Interaction (cs.HC)
备注: 23 pages. To appear at UIST 2025 (The 38th Annual ACM Symposium on User Interface Software and Technology)
Abstract:Effective communication between pilots and air traffic control (ATC) is essential for aviation safety, but verbal exchanges over radios are prone to miscommunication, especially under high workload conditions. While cockpit-embedded visual aids offer the potential to enhance ATC communication, little is known about how to design and integrate such aids. We present an exploratory, user-centered investigation into the design and integration of icon-based visual aids, named ATCion, to support in-cockpit ATC communication, through four phases involving 22 pilots and 1 ATC controller. This study contributes a validated set of design principles and visual icon components for ATC messages. In a comparative study of ATCion, text-based visual aids, and no visual aids, we found that our design improved readback accuracy and reduced memory workload, without negatively impacting flight operations; most participants preferred ATCion over text-based aids, citing their clarity, low cognitive cost, and fast interpretability. Further, we point to implications and opportunities for integrating icon-based aids into future multimodal ATC communication systems to improve both safety and efficiency.
[HC-8] Small-world Networks of Agents Brainstorm AI Risks to Support Ideation
链接: https://arxiv.org/abs/2609.24859
作者: Ke Zhou,Edyta Bogucka,Daniele Quercia
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注: 19 pages, 5 figures
Abstract:The ideation phase of participatory AI risk assessment often starts with a blank slate or a limited list of predefined risks, making it difficult to surface indirect or systemic harms. To address this limitation, we propose a three-stage ideation support tool. The tool complements participatory AI, rather than replacing it, and helps focus later engagement with affected communities. First, it dynamically discovers stakeholders depending on the given AI use and recursively expanding outward, allowing overlooked or indirect stakeholders to emerge. Second, it simulates these stakeholders with LLMs, connecting them into a network of a given topology, and having them ideate about risks. Third, it prioritizes risks using network centrality measures. In an initial evaluation, we found that betweenness centrality run through agents connected in a small-world network works best as it elevates risks raised by stakeholders who bridge disconnected groups, surfacing novel, systemic harms that traditional methods often miss. On an AI chatbot companion use case, this approach increased the novelty of the identified risks by approximately 1.1 points over single LLM brainstorming, and by 0.5 points over agentic LLM brainstorming, measured on a normalized five-point Likert scale, without reducing the plausibility or severity of the identified risks. To test whether our framework helps a human-led ideation session using the Futures Wheel approach, we divided 11 teams of non-western young chatbot users into two types: control (team) and treatment (team) in a participatory AI risk assessment. The control teams started from a list of risks generated by the 45 AI practitioners in the initial evaluation; the treatment teams started from a list generated by our framework. The treatment teams identified more risks overall, and more systemic, human-computer interaction, and environmental risks.
[HC-9] CRiDiT: Instantiating a run-time testbed for trust calibration in AI-infused systems
链接: https://arxiv.org/abs/2609.24833
作者: Yuntian Ding(CRI),Nicolas Herbaut(CRI, LaBRI),Camille Salinesi(CRI)
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:The integration of AI into larger technical infrastructures has made the alignment of human trust with system trustworthiness, known as trust calibration, a critical engineering concern, since misplaced trust in either direction leads to operational and safety risks. While conceptual frameworks provide a strong foundation for understanding trust calibration, their translation into running systems remains a challenge, because there are few testbeds in which human trust inputs, machine trustworthiness evidence, gap detection and remediation operate together within a closed loop. This paper instantiates CRiDiT (Computational Risk-Sensitive biDirectional Trust) as a run-time testbed, operationalising machine-side trust with Dempster-Shafer Theory and PCR5 redistribution, human-side trust with Subjective Logic, and calibration with a threshold-based trust gap. Following the Design Science Research methodology, we exercise the artifact across three high-stakes scenarios (hiring, financial, legal), producing 144 logged interaction steps across fifteen sessions. The analysis shows that the artifact captures trust calibration dynamics as intended, and reveals three points at which the instantiated policy departs from its design requirements: the machine-side estimate begins from a global benchmark rather than task-relevant evidence; risk-sensitive thresholds do not produce risk-sensitive triggering; and the calibration policy assigns explanatory prompts to over-trust, where corrections narrowed the gap in all 6 observed cases. Since the first two arise from the same design decision, to make the difference of two estimated scalars the calibration criterion, they point toward a common requirement: that the criterion should operate on the evidence rather than on scalars derived from it. The third concerns what follows detection, and shows that the action vocabulary inherited from trust repair does not align with what the interaction logs show to be effective. The work contributes the artifact, a characterisation of its run-time behaviour, and the requirements this characterisation elicits.
[HC-10] Detecting Agitation Before Behavioral Escalation in Autistic Youth Through Multimodal Wearable Sensing
链接: https://arxiv.org/abs/2609.24791
作者: Nibraas Khan,Abigale Plunk,John Staubitz,Ingrid Shragge,Jordan Brooks,Suzanne Wright,Alec Brewer,James Dieffenderfer,Alper Bozkurt,Amy Weitlauf,Nilanjan Sarkar
类目: Human-Computer Interaction (cs.HC); Machine Learning (cs.LG)
备注:
Abstract:Challenging behaviors including aggression, self-injury, and property destruction are observed in 68% of autistic youth and pose risks to youth and caregivers. These episodes are preceded by agitation, a rising state of distress expressed through movement, vocalization, and autonomic arousal. Its signs are subtle and individualized, and its autonomic components are invisible without instrumentation. We collected upper-body movement from inertial measurement units, physiology from a wrist-worn device, and vocalizations from lapel microphones across 30 clinician-led sessions with 15 autistic youth, paired with expert behavioral annotations. We adapt four pretrained foundation models, one per modality, project each to a shared 128-dimensional space, and fuse them into a single group model. The model detected agitation with an area under the ROC curve of 0.724 at the clinician-annotated onset (within-participant permutation p=0.0005), declining to 0.608 at 30,s before onset. Thirteen of fifteen participants were above chance. A from-scratch configuration reached only 0.58, while frozen and fine-tuned features performed comparably (0.71 and 0.72). Audio contributed most of the signal, and a watch-only configuration stayed near chance. Individualized agitation is therefore detectable, including in unannotated windows preceding the annotated onset, using foundation-model transfer with one shared model rather than one per child.
[HC-11] “MeBo Leaves a Piece of You Behind”: Designing a Relational Voice-Based Memory Companion for Older Adults
链接: https://arxiv.org/abs/2609.24706
作者: Hasibur Rahman,Mahsa Nasri,Manasi Vaidya,Melika Vafafar,Jessie Chin,Smit Desai
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注:
Abstract:Autobiographical remembering supports identity, well-being, and social connection in later life, yet voice-based memory technologies largely rely on isolated prompts. We designed and built MeBo, a fully functional relational voice-based memory companion, through participatory design with 11 older adults. Their accounts shaped four Design Strategies that guided MeBo’s interaction design and multi-agent implementation. In a mixed-methods evaluation with 20 older adults, participants found MeBo exceptionally usable (SUS = 87.75), enjoyable, sociable, emotionally responsive, and trustworthy. Participants reported higher positive affect and momentary social connection and lower negative affect after the session than before. Participants described how MeBo followed their stories, returned to earlier memories, adapted to their preferences, and made its growing memory visible and controllable. MeBo’s relational framing surfaces tensions around what it should remember, who may access memories produced through interaction, and what becomes of them when the user or MeBo is no longer present.
[HC-12] Annie Are You Okay? How Style- and Context-Based Personalization Shape AI-Assisted Decision-Making
链接: https://arxiv.org/abs/2609.24644
作者: Hasibur Rahman,Benjamin R. Cowan,Smit Desai
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注:
Abstract:As people turn to generative AI for financial advice, these systems can personalize how they communicate and what they say. Whether these forms of personalization shape decisions differently remains unclear. We conducted a preregistered 2 x 2 between-subjects factorial experiment (N=240): participants ranked three comparably viable stocks, discussed them with an AI, and reranked them. Participants perceived both forms of personalization, but only context-based personalization reliably changed ranking behavior: it increased reconsideration and moved rankings toward the AI’s assigned recommendation. Participants felt more influenced without judging the AI as more correct, trustworthy, intelligent, likeable, or high-quality. Those initially farther from its recommendation moved more toward it while judging its advice less correct; exploratory analyses suggest greater susceptibility among lower-expertise participants. These findings show how personalized AI can steer decisions among defensible options with only a minimal evaluative trace, raising concerns for the design and governance of personalized decision support.
[HC-13] Structure-Aware Rendering: How Code Reveal Shapes Programmers Visual Attention
链接: https://arxiv.org/abs/2609.24616
作者: Xiaotian Su,Jan Brasser,David Robert Reich,Lena A. Jäger,April Yi Wang
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:AI coding interfaces present generated code either all at once or token-by-token. These rendering strategies reflect model generation rather than how programmers actually read code: selectively, non-linearly, and guided by the structure. We argue that code rendering is a first-class interaction primitive that shapes how programmers read and understand code. To explore this design space, we introduce structured rendering, a technique that reveals code in semantically meaningful chunks derived from its syntactic hierarchy, exposing high-level structure before low-level details. To isolate rendering effects on visual attention, we conducted an eye-tracking study with 53 participants comparing static, character-based, and structured rendering. Our findings show that rendering alters visual attention and reading behavior: dynamic rendering induces fewer but longer fixations and more sustained focus, while structured rendering further guides attention toward semantically meaningful units and supports high-level understanding. We release the anonymized dataset with an interactive demo (this https URL) to support future research.
[HC-14] Prompting Against Persona Drift: Comparing Intervention Timing and Content in LLM -Simulated Conversations
链接: https://arxiv.org/abs/2609.24532
作者: Nicolas Leins,Jennifer Haase,Varvara Geronimus,Jana Gonnermann-Müller,Sebastian Pokutta
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Simulating student personas with large language models (LLMs) enables scalable evaluation of educational systems. However, behavioral drift, a progressive decline in persona consistency, can emerge over extended conversations, limiting the validity of such simulations. We evaluate five prompt-level mechanisms using separate monitoring and intervention pipelines. Across 1,200 28-turn conversations spanning four LLMs and two ADHD persona intensities, we varied when to intervene (static vs. adaptive) and what to inject (reinjection vs. reflective reminder), plus a novel adaptive condition in which a monitor generates behavior-specific instructions. Relative to no intervention, reinjection reduced the modeled rate of LLM-rated drift by 35–38%, reflective reminders by 22–27%, and behavior-specific instruction by 87%. None eliminated drift. We found no evidence that adaptive timing outperformed static scheduling. Monitoring therefore appears more useful for deciding \textitwhat to correct than \textitwhen to intervene, although behavior-specific instruction requires component-level testing.
[HC-15] Finger-to-Ear ECG: Systematic Evaluation of an Earbud-Based Cardiac Monitoring Approach
链接: https://arxiv.org/abs/2609.24530
作者: Philipp Lepold,Paula Breitling,Jonas Hummel,Tobias Röddiger,Michael Beigl
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Ear-ECG enables unobtrusive cardiac monitoring, but existing approaches often struggle with low signal amplitudes and limited morphology preservation. We evaluate a finger-to-ear ECG paradigm that combines an in-ear electrode with a rear housing finger-contact electrode in a speaker-equipped earbud. A study with 30 participants investigated four finger-to-ear electrode geometries against chest-reference ECG and additionally assessed robustness under music playback, talking, and walking disturbances. Left-finger configurations achieved R peak detection F1-scores 99% with high agreement of key ECG morphology features. In contrast, right finger configurations as well as walking led to severe signal degradation. The results reveal a trade-off: while cross-body setups provided the highest signal fidelity, users preferred same-side contacts for comfort reasons. This establishes finger-to-ear ECG as a robust, morphology-aware paradigm for opportunistic sensing in earables.
[HC-16] Beyond Screen Time: Demonstrating the Value of App Activity Logs to Understand User Behavior Further
链接: https://arxiv.org/abs/2609.24477
作者: Ole Schmitt,Pauline Gieseler,Frederik Riedel,Donatus Wolf,Ariel Stern,Paul Schmiedmayer,David Grüning
类目: Human-Computer Interaction (cs.HC)
备注: 25 pages, 10 figures, 6 tables
Abstract:Research on smartphone use remains fragmented, with studies differing in measures, methods, and platforms, limiting what we know about everyday behavior. We argue that event-level app activity logs should form the backbone of research on actual rather than recalled smartphone use. We conduct a secondary analysis of 4,571,252 app events from 1,972 participants across three longitudinal cohorts differing in age, country, recruitment, and platform, complemented by a published reference cohort. We reproduce established aggregate and micro-usage measures and examine temporal structure, application composition, transitions, and individual distinctiveness. Aggregate usage varies less than the organization of activity: adolescent use, for example, is structured around school schedules, while other cohorts show weaker within-day patterns. Application and transition patterns reveal behavioral differences obscured by screen time; an intervention reduced daily usage while increasing mean session duration. Activity traces also enabled 15-22% top-1 participant re-identification. We discuss methodological, reproducibility, and privacy implications for HCI.
[HC-17] A Leader-Driven Open Collaboration Platform for Exploring New Domains
链接: https://arxiv.org/abs/2609.24466
作者: Michael Weiss,Ibrahim AbuAlhaol,Mohamed Amin
类目: Human-Computer Interaction (cs.HC)
备注: 4 pages. Author’s version of the paper published at OpenSym '16
Abstract:This paper describes the design and initial evaluation of a leader-driven open collaboration platform for exploring new domains. The goal of this platform is to enable the collaboration of subject matter experts across knowledge boundaries. Traditionally, new domains are explored from within a single specialist or a focused group perspective. However, this often introduces bias. Collaboration helps reduce such bias by providing access to a broader range of information sources, increasing the chances for producing new insights in a new domain. However, it also introduces a new problem: variance between the contributions made. Variance makes it difficult to produce a coherent document. In this paper, we report on our observations from developing an initial prototype of the open collaboration platform, and derive propositions about how leader-driven open collaboration helps reduce bias while containing variance.
[HC-18] DeceptionAnalyser: A Web-Based AI Tool for Performing Structured Deception Analysis with Argumentation Schemes and LLM s
链接: https://arxiv.org/abs/2609.24369
作者: Stefan Sarkadi,Xabier Garmendia,Jack Mumford,Trevor Bench-Capon
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注: 36 pages
Abstract:Deception plays a central role in Intelligence operations, yet it remains difficult to analyse systematically without expert knowledge of reasoning patterns and cognitive manipulation. In computational argumentation, for instance, no scheme-level ground-truth corpora currently exist to support statistical validation. In this paper, we address this by introducing a set of ten argument schemes designed to model distinct forms of deception, each accompanied by structured premises and critical questions. In doing so, we introduce the first dedicated library of argumentation schemes specifically designed for deception analysis, providing a structured foundation for systematically modelling and analysing deception in narrative text. We then present \textitDeceptionAnalyser, a browser-based tool that implements these schemes through a two-stage methodology combining LLM-based premise extraction with critical-question-driven evaluation. Our aim is to provide a conceptual and methodological foundation for analysing deceptive reasoning in narrative text. This is precisely what we address in this paper by demonstrating how structured argumentation theory and AI-assisted analysis can support transparent, explainable assessments of potential deception. Because the schemes are designed to flag claims for scrutiny rather than to output a deception verdict, we do not benchmark classification accuracy; instead, we assess the \emphreliability of the methodology by measuring the consistency of the tool’s premise and conclusion assessments across ten contemporary large language models and repeated runs. We find that scheme detection is highly stable for clear-cut deception and degrades gracefully, in interpretable ways, on more ambiguous intelligence-style narratives.
[HC-19] Making Frag mented Reports Legible: Finding Patterns and Perceptions of Sexual Violence in Bangladesh
链接: https://arxiv.org/abs/2609.24254
作者: Sheherjan Haq,Sharifa Sultana
类目: Human-Computer Interaction (cs.HC); Computers and Society (cs.CY)
备注: 14 pages, 6 figures, and 7 tables
Abstract:Evidence about sexual violence in Bangladesh is fragmented across individual reports, while official and civil-society statistics rarely provide reusable case-level detail. We examine how structured analysis can make one part of this fragmented record legible without treating it as prevalence data. Our corpus contains 2,811 articles timestamped 2013-2023 from the Prothom Alo publishing ecosystem; 2,794 include parseable metadata about reported victims, alleged perpetrators, incidents, legal responses, and locations. We combine descriptive and spatial analysis of these records with thematic analysis of 115 convenience-sample survey responses collected in late 2020. The corpus documents many young, female, and student victims, frequent acquaintance and neighbor relationships, uneven geographic documentation, and substantial missingness in legal outcomes. Respondents most often describe weak enforcement, insecurity, patriarchal socialization, education gaps, and community inaction as conditions enabling persistence. These findings characterize news documentation and public perceptions; they do not estimate incidence, geographic risk, or causality. We contribute an uncertainty-aware framing for HCI research using sensitive, low-resource news data and identify design requirements for provenance, privacy, validation, and responsible communication.
[HC-20] AI-Assisted Social Story Intervention for Special Education: The Design of AdaptED Stories
链接: https://arxiv.org/abs/2609.24245
作者: Buyankhishig Enkhjargal,Himanshi Lalwani,Hanan Salam
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Social Stories are widely used to support autistic children in understanding and preparing for everyday situations, but creating stories that are appropriately tailored to each child’s needs remains labor-intensive for practitioners. Existing digital tools support story assembly and delivery, but much of the work of writing, visual preparation, and personalization remains manual. Recent AI-based approaches have enabled automated story generation, yet offer limited support for practitioner oversight, context-sensitive personalization, and the use of supportive visuals grounded in individual learner profiles. We present AdaptED Stories, a practitioner-guided system for authoring, personalizing, and delivering Social Stories in special-education contexts. The system uses student profiles to draft story text and visuals, supports review and refinement by practitioners, and includes reading-session support with comprehension activities and session records. We report findings from a practitioner-informed design process, assessments of generated stories and visuals, and a usability study with seven special-education practitioners. Our findings suggest that AI assistance can reduce story-preparation burden and support more individualized story creation, alongside the importance of practitioner oversight, cultural and contextual specificity, and designing for varied learner needs. These findings contribute design implications for AI-assisted accessibility tools in special-education.
[HC-21] he Work Behind Delegation: A Framework for Supervising AI Coding Agents
链接: https://arxiv.org/abs/2609.24234
作者: Yeon Su Park,Nadia Arvi,Hae Ri Lee,Sehoon Lim,Qianou Ma,Juho Kim
类目: Human-Computer Interaction (cs.HC)
备注: Under Review
Abstract:As AI coding agents carry out development tasks with greater autonomy, developers are shifting from direct implementation toward supervising delegated work. Yet existing research offers limited understanding of how developers organize supervisory activities into connected workflows. Drawing on observations and workflow diagrams from 19 experienced developers, we reconfigure Sheridan’s framework of human supervisory control into seven stages and the connecting loops for supervising AI coding agents. We applied the framework to public developer discussions on Reddit and found that supervisory demands extend across stages and that developers manage them by concentrating effort in planning, delegating supervisory work to other agents, and turning recurring guidance into reusable assets. Our framework provides a useful analytical lens for understanding how developers supervise AI coding agents by capturing how supervision is structured in agentic software development.
[HC-22] Adaptive Forgetting for Nonstationary Optimization: Towards Robust EEG Decoding
链接: https://arxiv.org/abs/2609.24233
作者: Hongyu Zhu,Lin Chen,Jing Chen,Yuting Zhou,Mingsheng Shang
类目: Machine Learning (cs.LG); Human-Computer Interaction (cs.HC)
备注: Submitted
Abstract:Electroencephalography (EEG) provides non-invasive monitoring of brain activity and is widely used in emotion recognition, motor imagery and sleep staging. Although within-subject decoding has achieved considerable progress, cross-subject generalization remains a central challenge in practical applications. EEG decoders are typically trained with Adam/AdamW under a fixed second-moment decay coefficient, even though cross-subject learning involves low signal-to-noise ratios, subject variability, and gradient nonstationarity. A fixed coefficient implicitly assumes that gradient statistics are homogeneous across layers and time, which can limit model’s adaptability to cross-subject EEG signals and degrade generalization. To address these issues, we propose AFOR, a tensor-wise adaptive optimizer that converts the fixed second-moment decay coefficient into a dynamic coefficient estimated online from local gradient state. AFOR combines a Residual-Alignment Signal Scorer (RASS) and an Adaptive Forgetting Controller (AFC). RASS summarizes local gradient residuals and directional agreement into a signal-quality score, and AFC maps this score through self-referential normalization to a bounded per-step decay coefficient, with cumulative-product initialization correction maintaining consistency under time-varying decay. Under a strict cross-subject protocol on three EEG benchmarks that cover three representative fields, AFOR achieves the best average performance among the compared optimizers, improving the mean test accuracy over Adam by 3.00%, 2.07%, and 4.38%, respectively. Comments: Submitted Subjects: Machine Learning (cs.LG); Human-Computer Interaction (cs.HC) Cite as: arXiv:2609.24233 [cs.LG] (or arXiv:2609.24233v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2609.24233 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[HC-23] ProbeScout: Visual Analytics for Attribute-Guided Image Search
链接: https://arxiv.org/abs/2609.24110
作者: Yifan Lv,Yiyun Chen,Daojun Ye,Haotian Yang,Weikai Yang
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Analysts often need to identify images that jointly satisfy multiple visual conditions, such as a crossroads with traffic lights at dusk, for model diagnosis, dataset curation, and targeted training. Embedding-based retrieval can rank the large gallery efficiently, but a visually dominant condition can obscure weaker conditions, and a single similarity score does not enforce the required conjunction. Visual question answering (VQA) can explicitly verify conditions, yet exhaustively applying it to the full gallery is costly, especially when analysts refine their query. These limitations motivate keeping humans in the loop at the attribute level, where analysts can quickly build evidence for each condition and reuse it when the request changes. We therefore present ProbeScout, a visual analytics system that supports this loop. It first builds composable attribute probes from sparse VQA labels and fuses them into a conjunction-aware initial ranking. Coordinated views support rapid screening, near-miss diagnosis, and on-the-fly subset construction by filtering and combining these probe outputs. Analysts provide lightweight attribute- and query-level feedback, which drives staged refinement of fusion weights while keeping the probes fixed. These verified attributes can be reused for future queries. We evaluate ProbeScout on 17 retrieval tasks across three datasets, showing improved retrieval over embedding baselines. A separate 10-task comparison achieves higher task-macro AP and F1 than exhaustive VQA while labeling at most 2% of the gallery images. Two case studies further demonstrate how ProbeScout supports interactive analysis and refinement in realistic workflows.
[HC-24] Ask Before It Tells: Benchmark-to-Robot Body-Cue Transfer for a Question-First Bedside Robot IROS2026
链接: https://arxiv.org/abs/2609.24099
作者: Dongsik Yoon
类目: Robotics (cs.RO); Human-Computer Interaction (cs.HC)
备注: To appear in the 4th Workshop on Nonverbal Cues for Human-Robot Cooperative Intelligence (NOC), IROS 2026
Abstract:Body-cue recognition can support assistive robots, but benchmark accuracy does not guarantee reliable behavior under a robot-camera viewpoint. We present Nuni, a bedside robot prototype that treats a detected distress cue as a reason to ask rather than a reason to alert. We compare two X3D-UGT RGB appearance classifiers, which reach 97.7% and 94.8% six-way accuracy on NTU RGB+D, with a pose-centric hybrid pipeline on 28 single-actor scripted clips recorded from the robot camera. The hybrid path achieved 0.71 six-way macro recall, versus 0.25 and 0.29 for the fine-tuned and from-scratch RGB variants. More importantly for interaction, it produced a question-triggering distress cue in 12/16 distress clips and would have prompted unnecessarily in 2/8 normal clips; the RGB variants yielded a question-triggering cue in only 2/16 and 3/16 distress clips. We separately tested the question-first controller through event injection. All 13 state-transition trials passed: valid responses caused stand-down, two unanswered prompts produced one alert, and three boundary conditions were handled correctly. These results are a preliminary technical evaluation, not a user study or medical validation, but they show how interaction policy can limit the consequences of uncertain perception.
[HC-25] WidgetVA: A Widget-Centric Framework and Benchmark for Agent ic Visual Analytics
链接: https://arxiv.org/abs/2609.24094
作者: Yutong Chen,Zhike Tang,Zhihao Mai,Zhihao Shuai,Danli Luo,Jing Xu,Weikai Yang
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI)
备注:
Abstract:Visual analytics (VA) enables sensemaking through interactive visualization, but effective analysis often requires experts to translate high-level intents into long sequences of interface operations and iteratively interpret visual feedback. We study whether modern vision-language models (VLMs) can take on this role as autonomous VA operators that observe the interface, plan multi-step exploration, execute interactions, and adapt based on intermediate visual feedback. To support systematic development and evaluation, we first introduce WidgetVA, a widget-centric agentic VA framework that standardizes interactive components as structured widgets with unified action (e.g., filter and zoom) and perception-query (e.g., selection summaries) APIs. This standardization supports two modes of system construction: wrapping an existing VA system to make it agent-operable without rebuilding it, and composing a new system from widgets as modular building blocks. To help agents coordinate across widgets rather than plan each interaction from scratch, each widget further packages reusable analytical workflows, giving agents more than a bare set of callable functions to plan over. Building on this framework, we present WidgetVABench, a benchmark of single- and multi-widget VA tasks that require agents to perform multi-step interactions to uncover evidence and produce verifiable results. Each task also provides fine-grained reference annotations so that WidgetVABench can score Answer, Reference Trace Similarity, and State separately rather than collapsing agent performance into one success score. Experiments across multiple VLMs show that our framework provides an effective scaffold for agentic VA, while the diagnostic measures expose persistent limitations for future work. The WidgetVA framework and WidgetVABench have been released in this https URL.
[HC-26] oward Human-in-the-Loop Robot Failure Recovery: Bridging Communication Gaps in Human-Robot Collaboration
链接: https://arxiv.org/abs/2609.24055
作者: Promise Ekpo,Teju Vijay,Dhruv Mandalik,Tisha Jain,Arman Ibrayeva,Sunishka Sil,Stefanie A. Tellex,Angelique Taylor
类目: Robotics (cs.RO); Human-Computer Interaction (cs.HC)
备注:
Abstract:Robots can recover from failures by asking bystanders for help, but effective human-in-the-loop recovery requires communication that accounts for differences in people’s knowledge. Prior inverse-semantics work generates requests using a single listener model, leaving differences in listener knowledge untested. We introduce Listener Differences in Human-Robot Interaction (LD-HRI), a game, dataset, and benchmark that evaluates speakers through human listener performance. Our evaluation examines request properties, large language model (LLM) speakers, and inverse-semantics request-selection algorithms under controlled differences in listener information. The corpus contains 446 human-written requests and 1,302 listener trials. We additionally evaluated 24 frozen LLM-written requests with 70 human listeners across 560 trials. Novice success is descriptively higher with model-written requests across all four tasks, yet both request sources leave substantial expert–novice gaps, including 16 percentage points for LLM requests. LD-HRI makes these gaps measurable, providing a foundation for designing more robust communication in human-robot and human-agent interaction.
[HC-27] When AI Tutors Speak: Evidence from a Randomized Field Experiment
链接: https://arxiv.org/abs/2609.23958
作者: Shihao Yang,Marshall Van Alstyne,Chrysanthos Dellarocas
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Students increasingly study alongside generative artificial intelligence (AI), yet unguided access to fluent answers invites cognitive offloading, and there is little evidence on which configurations of AI tutoring produce learning. Two design margins are usually bundled together: pedagogical structure (how the tutor teaches) and interaction modality (how students talk to it). We separate them. In a preregistered randomized field experiment in a graduate corporate-finance course of an online MBA, we randomized 86 students between a structured tutor grounded in the course materials and a holdout in which consumer AI remained freely available. Within the tutored arm, each student’s channel alternated weekly between voice and text, so the modality effect is identified within student. Structure mattered: tutored students gained 6.63 points more than ability-matched peers (p=.007), and the gain was concentrated in written reasoning, where the share of answers reaching relational quality rose from 8% to 49% in the tutored arm against 8% to 27% in the holdout. Modality did not matter for learning. The instructor’s own final, on file for all 86 randomized students, shows the same direction (2.6 points of 100, with no difference on a pre-treatment midterm). Voice nearly doubled conversational interaction and cost 2.8 times as much to deliver, yet it produced weekly mastery statistically equivalent to text, even as students came to prefer it. Pedagogical structure shapes what students practice, and modality shapes how they interact with the tutor. Making an AI more humanlike does not by itself make it more educational.
[HC-28] hink Before You Accept: Can Written Justification Reduce Uncritical Uptake of AI Writing Suggestions?
链接: https://arxiv.org/abs/2609.23936
作者: Yan Tao,Jennifer Meyer,Rene F. Kizilcec
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Generative AI can offer students useful feedback, but its value depends on judging which suggestions are accurate and relevant. Prior research shows that strategic friction during human-AI interactions can promote critical uptake, but how to effectively implement such friction in academic contexts remains unclear. We examine whether requiring students to justify decisions to accept or reject AI suggestions can mitigate uncritical uptake in academic writing. In a randomized experiment embedded in a course activity (N=129), students wrote a data analysis proposal, received mixed-quality AI revision suggestions, and decided whether to accept or reject them. Students required to provide written justifications were 24 percentage points less likely to adopt flawed suggestions (65% vs. 41%), with no reduction in acceptance of sound suggestions (81% vs. 86%). However, thematic analysis revealed superficial engagement in the justification task and gaps in metacognitive monitoring and domain knowledge.
[HC-29] “I can do whatever I put my mind to!” How prioritizing belonging in the design of technology-based programs can support foster-involved youth with self-efficacy self-expression and personal exploration
链接: https://arxiv.org/abs/2609.23861
作者: Ila Krishna Kumar,Karishma Chadha
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:In this paper, we examine how intentionally designing for belonging impacts outcomes in a technology program serving a population of youth who have not had safe and supported experiences with technology - youth impacted by foster care. We reflect on the design of an internship program which engaged two foster-involved youth in creative self-expression and co-design with technology. We first outline how we designed the program to foster belonging, then analyze intern and facilitator reflections to understand how the program impacted the interns’ experience. We highlight how the internship design affected youth’s inclusion and overall engagement; self-expression; exploration; and self-efficacy. We conclude with a discussion of design recommendations for technologists and service providers aiming to engage youth in expressive and co-design technology experiences, particularly youth impacted by foster care.
[HC-30] Vibe-GUIDE: A Graph-based User Interface in IDEs for Oversight in Vibe Coding
链接: https://arxiv.org/abs/2609.23859
作者: Chifang Chou,Sam Yu-Te Lee,Rudrajit Choudhuri,Kwan-Liu Ma
类目: Human-Computer Interaction (cs.HC)
备注: 14 pages, 3 figures, 4 tables
Abstract:In agentic coding, developers shift from implementing changes themselves to specifying intent, evaluating the agent’s work, and making approval decisions. However, delegating implementation can introduce cognitive debt that erodes project comprehension over time, constraining developers’ ability to provide oversight. In this work, we investigate the role of persistent shared representations in supporting project comprehension and oversight of coding agents. We present Vibe-GUIDE, an agentic coding interface built around a structural, live, manipulable, and adaptive graph representation organized by functional modules. We evaluated our interface in a randomized between-subjects study comparing how 16 developers completed three cumulative coding tasks using our interface or a Chat-only baseline. We found that Vibe-GUIDE can support project comprehension and sustained task performance while keeping developers cognitively involved in oversight. These findings show how persistent shared representations can complement natural-language interaction and help developers maintain the understanding needed to oversee agent-generated changes as projects evolve.
[HC-31] Elicitive User Interfaces: Designing How Users Shape Generative Interfaces
链接: https://arxiv.org/abs/2609.23642
作者: Eunhye Kim,Bryan Min,Haijun Xia,Juho Kim
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Generative user interfaces (GenUI) promise personalized interfaces to a user’s tasks and needs. However, user needs are often implicit—difficult for systems to infer and users to articulate, making it hard for users to arrive at their ideal interface. We propose Elicitive User Interfaces, a design approach to GenUI that generates elicitation techniques as part of the interface itself. Elicitive UIs adapt these techniques to the user, task, and interface to draw out user preferences. To guide the design of Elicitive UIs, we synthesize a six-axis design space that shapes how an interface elicits user preferences. Across two user studies with a design probe, we found that while Elicitive UIs surfaced preferences users had not already formed, and that responses to elicitation varied more across users than across tasks. Users developed more consistent preferences for how they wanted to be elicited, suggesting an opportunity to personalize elicitation itself.
[HC-32] Heating in human-HVAC interaction for smart homes: An interdisciplinary overview
链接: https://arxiv.org/abs/2609.23508
作者: Delong Korus-Du,Gunnar Stevens,Alexander Boden,Lenneke Kuijer,Apostolos K. Vavouris,Md Shajalal,Philip Engelbutzeder,Omid Veisi,Peter Tolmie
类目: Human-Computer Interaction (cs.HC); Artificial Intelligence (cs.AI); Social and Information Networks (cs.SI); Systems and Control (eess.SY)
备注:
Abstract:As part of HVAC systems, residential heating provides foundational infrastructure for human habitation in cold weather. However, research on how residents interact with HVAC systems, particularly heating systems, remains fragmented across architecture, engineering, informatics, physiology, psychology, sociology, and design. Based on 541 studies from these fields, this review integrates interdisciplinary research on Heating in Human-HVAC Interaction in smart this http URL resulting synthesis is conceptualized through the Situated Interaction Dynamics of control and feedback between users and systems. User-initiated interactions involve monitoring past and present system performance and planning future operation, while system-initiated interactions rely on sensor networks to trigger automation or provide information enabling user action. These interaction dynamics connect Residents’ Experience and Practices with Heating in HVAC System Mechanics. Residents’ Experience and Practices include thermal comfort and energy management, where thermal comfort involves both individual physiological and psychological experiences of indoor climate and social practices shaped by norms, empathy, and negotiation among cohabitants. Heating in HVAC System Mechanics includes thermal conditions and energy performance. Thermal conditions concern the regulation of air temperature, mean radiant temperature, air velocity, and relative humidity, while energy performance concerns efficiency and environmental impact. This overview highlights four interdisciplinary tensions: sensed versus lived conditions, personalization versus negotiation, efficiency versus health, and automation versus agency. The resulting framework offers a conceptual lens to interpret heating interactions and design Human-HVAC Interaction that balances IEQ-driven healthy thermal conditions, affordability, and sustainability.
[HC-33] Alignment and Divergence between Humans and AI in Interpersonal Privacy Decisions
链接: https://arxiv.org/abs/2609.23403
作者: Hanxiang Zeng,Shuning Zhang,Xinyuan Zhou,Tianqi Song,Yuhan Yuan,Yuting Yang,Shuai Ma,Xin Yi
类目: Human-Computer Interaction (cs.HC)
备注: 22 pages, 11 figures, 5 tables. Preprint
Abstract:AI assistants increasingly mediate interpersonal communication on behalf of their primary user, but they risk violating the privacy expectations of third-party information owners. Resolving these tensions requires understanding how humans anticipate interpersonal privacy boundaries. Therefore, we conducted a dyadic study (N=76) and a matched evaluation of AI models across 18 information types and 3 recipient relationships. We found that data owners’ privacy judgments are highly contextual and relationship dependent. While familiar data co-owners show meaningful alignment with owners’ expectations, they significantly overestimate the need for permission. Interestingly, greater familiarity within the owner-co-owner dyad was associated with both higher disclosure acceptability and lower co-owner misalignment, whereas our exploratory four-item empathy measure was not. In contrast, AI models significantly underperform human co-owners in anticipating the data acceptability, even when provided with within-dyad examples. These findings underscore a core HCI design challenge to develop privacy-aware AI that respects multi-stakeholder information boundaries.
[HC-34] Identity Continuity in Long-Term Embodied AI Relationships: From Agent -Specific Identity Representation to Identity-Continuity Appraisal
链接: https://arxiv.org/abs/2609.23356
作者: Zijian Ru
类目: Human-Computer Interaction (cs.HC); Robotics (cs.RO)
备注: 16 pages, 1 figure, 1 table
Abstract:Long-term embodied AI will undergo learning, model updates, memory compression, hardware repair, and migration across embodiments. For users who have formed sustained relationships with such systems, these changes raise not only a problem of product consistency but also one of identity continuity: whether the changed system is still experienced as the same particular agent. Existing research suggests that human-AI relationships may develop relational particularity, that robotics and artificial-identity research has identified identity and migration signals across embodiments, and that major updates or platform disruptions can be accompanied by relational loss and restoration desire. This article proposes a user-side framework in which long-term embodied AI is represented through an agent-specific identity representation organized by at least three open identity-content domains: embodied-perceptual, psychological-behavioral, and relational-autobiographical. Information from these domains is not equally weighted; shared history, relational roles, and contingent responsiveness may make some information more identity-diagnostic than others. After system change, users may integrate continuity and discontinuity evidence in a weighted manner, yielding judgments along a continuum from relatively strong identity continuity through ambiguity or partial continuity to clear identity discontinuity. Causal-historical provenance and user participation are treated as contextual evidence rather than a fourth identity-content domain. The framework also proposes identity continuity as a psychological objective for lifecycle design, including memory selection, model updating, and migration across embodiments, under constraints of privacy and user control.
[HC-35] Learner-Centered Design of Educational Tools for Cross-Expertise Technical Communication in Computing Contexts
链接: https://arxiv.org/abs/2609.23309
作者: Jinyoung Hur,Yoshee Jain,Yuxuan Chen,Ge Wang,Kathryn Cunningham
类目: Human-Computer Interaction (cs.HC)
备注: 19 pages, 3 figures, 3 tables
Abstract:Cross-expertise technical communication (i.e., communicating about computing topics across levels of computing expertise) is important for workplace collaboration. However, students are underprepared because computing education rarely explicitly teaches communication, and available practice tends to occur among peers with similar expertise. We take a learner-centered design approach to inform technologies for developing these skills. We interviewed 12 industry professionals and conducted focus group and co-design sessions with 14 college students preparing for computing-related roles. Professionals emphasized adapting communication, negotiating expectations, building shared understanding, and using multiple modalities and artifacts. Students anticipated challenges adapting communication, navigating unfamiliar workplace dynamics, and communicating under fear of judgment, and valued rich, realistic practice environments that preserved their psychological safety. Synthesizing professional and learner perspectives with educational theory, we conclude that building self-efficacy through authentic mastery experiences and changing attitudes related to help-seeking are critical elements for educational technology design in this context.
[HC-36] AI Persona Service Consumption and User Intent Entropy: Field Experimental Evidence from an LLM Platform
链接: https://arxiv.org/abs/2609.23274
作者: Junjie Li,Xiaofan Li,Lauren Xiaoyuan Lu,Yiwei Wang,Bruce Yang
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Problem definition: Firms deploying large language model services must decide how their AI communicates, not just what it can do. We examine how a relational persona - warmer, more empathetic and more engaging than a non-relational persona - affects service consumption and the evolution of user objectives. Methodology/results: In a randomized field experiment with 9,586 newly registered users, we hold the underlying model and service capabilities constant. The relational persona increases interactions (sessions, +8.1%; duration, +10.6%; chat rounds, +24.2%; intent entropy, +5.8%) and outputs (files, +12.3%; distinct goals, +12.1%). Effects vary by entry intent. First-session effects are insignificant for Task Execution users. Socialization and Knowledge Exploration users show similar increases in chat rounds: Socialization increases intent entropy without more outputs, whereas Knowledge Exploration increases outputs without higher intent entropy. Modeling intent dynamics as a transition process, we find higher intent transition entropy for Socialization (+11.8%) but higher intent continuation probability for Knowledge Exploration (+12.8%), suggesting greater conversational breadth and persistence, respectively. In subsequent use, the relational persona increases aggregate chat rounds and outputs across all entry intents. Session count rises by 12.2% for Task Execution and 49.4% for Socialization, but not significantly for Knowledge Exploration. Effects on session count and intent entropy strengthen over time, whereas output effects remain stable. Managerial implications: AI persona is an operational design lever, not merely a presentation feature. Because more interactions do not uniformly generate more outputs, firms should evaluate interactions and outputs separately and consider matching persona to user intent, especially when added interactions consume costly computing resources.
[HC-37] Do Not Trust the Benchmark: Limitations of General LLM Rankings and a Case for Task-Specific Evaluation
链接: https://arxiv.org/abs/2609.23201
作者: Danial Amin
类目: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注:
Abstract:Benchmark scores increasingly influence the development, marketing, and selection of large language models (LLMs). Yet an overall score is interpretable only in relation to the system tested, the questions included, and the conditions of evaluation. This perspective examines five connected limitations of general LLM rankings: differences between evaluated and publicly available systems; commercial incentives and dependencies in external evaluation; benchmark saturation, defective tests, and data contamination; models exploiting scoring procedures; and the limited relevance of general scores to users’ tasks. Documented cases illustrate why these problems require different responses. I argue for evaluation procedures that disclose the tested configuration, validate questions and successful task completion, report performance alongside cost and execution time, and make the scope of generalization explicit. I then discuss \textbfIsotanta, a crowdsourced benchmarking platform, as a practical example of contributed questions and repeated evaluation. A larger question pool may improve task coverage, while repeated sampling can improve the stability of estimates on that pool; neither guarantees validity or personalization. The paper distinguishes the platform’s current shared ranking from proposed task-specific and user-provided evaluations. Its central argument is that model selection requires evidence about performance on the intended work, not simply a high position on a general leaderboard.
[HC-38] Evaluative Dynamics of AI Integration and Expert Performance under Epistemic Dependence across Heterogeneous Stakes
链接: https://arxiv.org/abs/2609.23135
作者: Dennis Kim,Roya Daneshi,Nikhil Krishnaswamy,Bruce Draper,Sarath Sreedharan
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:AI is increasingly integrated into expert workflows, yet how integration affects perceptions of the expert, AI, and their combination remains unclear in domains where lay users are epistemically dependent on AI-assisted experts. We examine this through a novel controlled medical study (N = 166) and a direct cross-domain analysis with pre-existing academic-advising data (n = 157, combined N = 323). Expert errors reduced evaluations of the human expert across domains. Perceived expertise, however, varied by AI integration strategy in the higher-stakes medical task, where automatic AI oversight produced higher ratings than expert-only or expert-initiated AI. Exploratory ordinal sensitivity analyses identified a performance-contingent reuse pattern, with automatic oversight producing greater intended reuse after successful medical performance. Overall, performance-related recalibration appeared comparatively portable, while integration-structure effects were more selective and context-sensitive. These findings suggest that system designers should consider how AI enters expert workflows, not only whether it is present.
[HC-39] Deciphering the Babel of Play: A Human-AI Collaborative Approach for Large-Scale Cross-Language Analysis of Game Reviews
链接: https://arxiv.org/abs/2609.23104
作者: Zixiaofan Yang,Chang Xiao
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:We present a large-scale cross-language analysis of game reviews using a human-AI collaborative framework that combines quantitative screening with multilingual large language models (LLMs). Starting from 17 million Steam reviews across 30 languages and 2,000 top-selling titles, we select 28 games with notable cross-language rating patterns. We then apply LLM-assisted content analysis to 442,162 reviews spanning 17 languages, with human researchers guiding codebook development and interpreting the results. Our findings reveal differences in both the aspects language communities prioritize and how they evaluate them, highlighting the roles of narrative expectations, game mechanics and stability, localization quality, cultural proximity, and perceptions of developers and publishers. We also identify rare cases of cross-language consensus. This work offers empirical insights into cross-cultural game evaluation and a scalable methodological approach to multilingual content analysis that preserves human interpretation.
[HC-40] Measuring Smartphone User Experience through a Hierarchical Metric Framework via Social Media Reviews
链接: https://arxiv.org/abs/2609.23101
作者: Xiaoteng Pan,Mingang Lan,Chenrui Zhang,Yu Su,Weijie Liu,Yue Gao,Nan Gao,Haining Zhang
类目: Human-Computer Interaction (cs.HC)
备注: 36 pages, 5 figures
Abstract:Smartphone user experience (UX) is widely expressed in user-generated online discourse across platforms, creating opportunities for in-the-wild measurement at scale. However, existing UX instruments and review-mining approaches do not provide a smartphone-oriented, theory-grounded hierarchical measurement specification that supports consistent aggregation and comparison across heterogeneous platforms. In this research, we propose a hierarchical smartphone UX measurement framework and an interpretable computational pipeline that translates cross-platform reviews into structured UX metrics. The pipeline extracts localized experience evidence units, maps them to the hierarchy via coarse-to-fine classification, and quantifies evaluations with a unified five-level satisfaction sentiment model. We apply the approach to a stratified subset of approximately 20,000 Chinese social media reviews covering four major smartphone brands across three platforms. The resulting metrics separate what users discuss, captured by normalized mention frequency, from how they evaluate it, captured by mean five-level sentiment scores. This work provides a scalable and interpretable evidence base for cross-brand comparison and metric-level interpretation beyond raw review volume or single-platform observations.
[HC-41] How Did Writing Change At CHI? Analyzing 44 Years of CHI Writing Before and After the Introduction of Large Language Models
链接: https://arxiv.org/abs/2609.23090
作者: Thomas Kosch,Robin Welsch,Michael Hedderich,Christopher Katins
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:The availability of Large Language Models (LLMs) reshaped scientific discourse at a linguistic level. LLMs are assumed to homogenize academic writing, flattening it into a single generic lexical register. To understand how CHI writing has changed since the public release of LLMs, we analyzed full texts of 14,262 archival papers across all 44 CHI proceedings from 1982 to 2026, measuring readability, register, lexical diversity, and marker words typically produced by LLMs. We find that prose did not homogenize, while vocabulary grew more varied, and sentence rhythm remained irregular. CHI prose changed more between 2016 and 2026 than in other decades toward greater density, and reading ease has declined since 2022. The word-level shift began before any author used LLMs, so LLMs did not start the change but accelerated it. Reflecting on the history of CHI papers, we discuss what may have caused changes in prose and how LLMs accelerated them.
[HC-42] Automatic multimodal UX improvement recommendations from LLM agent user simulations
链接: https://arxiv.org/abs/2609.22971
作者: Anu Chowdhury,Bin Wu,Hossein A. Rahmani,Emine Yilmaz
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注:
Abstract:Evaluating user experience (UX) on live websites through user testing is expensive, subjective, and difficult to scale. LLM agents offer a promising route to automating UX testing by simulating realistic user behaviour. However, existing simulation approaches typically lack multimodality and require time-consuming manual review to extract actionable insights. We formalise UX improvement recommendation from simulation data as a structured natural language generation and ranking problem, and establish an evaluation protocol using expert annotation and LLM-as-a-Judge. We present AMUSER, a multimodal framework which simulates user behaviour and automatically generates prioritised UX improvement recommendations from resulting data. We evaluate AMUSER on commercial websites and show that its recommendations substantially outperform those from text-only simulation (NDCG@3 = 0.758 versus 0.359) at an 89% lower simulation cost. Our results suggest an asymmetric role of multimodality: visual access during simulation improves recommendations through richer traces, while providing visual inputs during recommendation generation can modestly degrade quality. We also discuss practical deployment lessons from applying AMUSER to commercial websites.
[HC-43] he Moral Check: Strategic AI Governance for the Pacing Problem
链接: https://arxiv.org/abs/2609.22869
作者: Zaid Amin,Rahma Santhi Zinaida,Nazlena Mohamad Ali
类目: Computers and Society (cs.CY); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注: 60 pages, 9 figures, 10 tables. Preprint submitted to Computers in Human Behavior
Abstract:Technology cannot steer itself. Strategy provides that steering, establishing the rule that purpose and judgment must precede compute capital. As the frontier artificial intelligence (AI) ecosystem accelerates exponentially, the pacing problem induces severe cognitive tunneling in engineering teams, prioritizing scalar throughput over human judgment. A calibrated pacing rate is imperative to check unchecked scaling, guarantee safety, and build models in whose alignment society can place warranted confidence. Traditional oversight fails through retrospective checklists, a pathology of performative governance exhibiting high procedural maturity but alarming scientific immaturity. We deliver a dual contribution: a PRISMA 2020 review synthesizing 130 empirical studies (MMAT-appraised across 18 benchmarks; total corpus N = 130 empirical studies across 178 reference foundations), and the Strategic AI Governance Ex-Ante Framework (SAGE-X). Our synthesis exposes two systemic vulnerabilities: the Recursive Assurance Paradox (correlated, ungrounded evaluator confidence) and the Durability Deficit (guardrail decay under multi-turn shifts). Grounded in MIT Strategic Computing doctrines, SAGE-X operationalizes Four Strategic Mindset Pillars: (1) Intent over Execution (mitigating velocity myopia); (2) Ruthless Trade-offs (deterministic tripwires eliminating moral hazard); (3) Outcomes over Outputs (auditing empirical hazard endpoints); and (4) Proactive Alignment (synchronizing ex-ante gates with runtime telemetry). Governed by a calculable Moral Check Index (MCI) with an unbypassable tripwire, SAGE-X delivers an operational Enterprise Lifecycle Audit Instrument (the “Moral Check Audit Card”) on Stanford WebProtégé, ensuring exponential progress never outpaces deliberative moral judgment, human agency, and societal trust.
[HC-44] Coral: Contextual Gists for Blind and Low Vision Screen Reader Users Understanding of Dynamic User Interfaces
链接: https://arxiv.org/abs/2609.22799
作者: Ritesh Kanchi,Jianna So,Krzysztof Z. Gajos
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:Blind and low vision (BLV) screen reader users construct mental models of user interfaces (UIs) through incremental screen reader interaction, a time-consuming and cognitively demanding process complicated by modern interfaces that may not convey dynamic content accessibly. We interviewed eleven BLV screen reader users about UI understanding and derived design objectives that informed Coral, a context-aware browser extension. Coral synthesizes interface and user context to generate screen reader-narrated gists which holistically notify users of interface states and changes relevant to their likely goals and ongoing interaction. In a comparative evaluation with eight BLV screen reader users, participants used Coral to form initial expectations and interpret interface changes, and reported spending less time and effort manually verifying interaction outcomes. Together, our findings provide deeper insight into how BLV screen reader users navigate gaps in their UI understanding, and how intelligent support that combines interface and user context can bridge these gaps.
[HC-45] When Disability Disclosure Travels: Memory Privacy and Contextual Integrity in Conversational AI
链接: https://arxiv.org/abs/2609.22720
作者: Atieh Taheri,Mahya Tazike,Patrick Carrington,Jeffrey P. Bigham
类目: Human-Computer Interaction (cs.HC); Computers and Society (cs.CY)
备注: 16 pages, 1 figure, 3 tables. Under review
Abstract:Conversational AI assistants remember what people tell them, and for disabled people, that often includes disability. We interviewed 12 adults with disabilities in the United States who use LLM-based assistants such as ChatGPT, Claude, and Gemini about when, how, and why they disclose disability to these systems and how this compares with disclosing to people. Using contextual integrity as an analytic lens, we found that participants disclosed by need rather than by name, translating disability into task-scoped instructions; that the same disclosure was judged against two recipients, a non-judging interlocutor and a data-holding company, producing opposite norms; and that memory features relieved the burden of repeated disclosure while letting disability information drift into contexts where it did not belong. Participants did extensive boundary work to restore context and wanted control over scope, provenance, retention, and access rather than per-utterance toggles. We discuss implications for the design of conversational AI assistants.
[HC-46] Explanation Navigator: Rectifying Out-of-Scope Human Interpretations of Leaky AI Explanations through Conversational Guidance
链接: https://arxiv.org/abs/2609.22707
作者: Yueqing Xuan,Kacper Sokol,Danula Hettiachchi
类目: Human-Computer Interaction (cs.HC)
备注:
Abstract:As explanations of artificial intelligence systems proliferate, their recipients must grasp not only what they convey but also recognise what they cannot. We conducted an interview study with nine participants to examine how explainees reason when their information needs exceed the scope of available explanations. Participants often unwittingly confabulated explanatory insights when relevant information was missing from the explanations, not recognising the inherent limitations thereof. We characterise such explanations as leaky explanations – simplifications that strive to hide complexity yet whose correct interpretation hinges on understanding of the concealed details. To address out-of-scope interpretations we propose Explanation Navigator: a conversational interaction framework that detects mismatches between users’ information needs and explanations’ content, elucidating pertinent yet implicit details and providing complementary explanations for unmet information needs. An online study with 316 participants showed that our approach allowed explainees to recognise and rectify confabulated explanatory insights, guiding them towards developing correct understanding.
[HC-47] oward Auditable and Calibrated AI for Dementia-Related Crash Severity Prediction: A Selective Deferral Framework to Support Human Review
链接: https://arxiv.org/abs/2609.22694
作者: Gaurab Chhetri,Anika Baitullah,Subasish Das
类目: Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注: This is the author’s preprint version of a paper accepted for presentation at HICSS 60 (Hawaii International Conference on System Sciences), 2027, Hawaii, USA. The final published version will appear in the official conference proceedings. Conference site: this https URL
Abstract:Public crash databases increasingly support automated safety analysis, but crash severity prediction remains difficult to translate into public-sector decision workflows when models are evaluated primarily as ordinary classifiers. This study reframes dementia-related crash severity modeling as a decision-aware triage problem in which a system must classify crashes into no-injury/property-damage-only (O), minor or moderate injury (BC), and fatal or severe injury (KA), while also controlling outcome leakage, reporting severe under-triage, calibrating confidence, and preserving every raw prediction for audit. Using 4,781 Texas crash records with structured fields and police narratives, we evaluate structured, narrative, fusion, calibrated fusion, BERT-family, and local large-language-model baselines under a stratified 70/15/15 split. In the reported split, leakage-controlled Gemma obtains the highest observed macro-F1 (0.545; 95% bootstrap CI [0.507, 0.583]). The best calibrated fusion model obtains macro-F1 of 0.522 and expected calibration error of 0.033. Selective deferral improves performance among cases retained for automatic classification. At 70% coverage, macro-F1 rises to 0.573 and severity cost falls to 0.577, while deferred cases are treated as candidates for a proposed human-review process and are not further evaluated in the present experiment. The study contributes a reproducible, leakage-controlled, and uncertainty-aware evaluation framework for crash AI systems, emphasizing auditability and selective deferral rather than accuracy alone.
[HC-48] Strategic Classification Has a Missing Lever: Audit Risk
链接: https://arxiv.org/abs/2609.22534
作者: Raman Ebrahimi,Massimo Franceschetti
类目: Computer Science and Game Theory (cs.GT); Human-Computer Interaction (cs.HC); Machine Learning (cs.LG)
备注:
Abstract:Strategic classification studies how a decision maker should choose a classifier when the agents being classified can adjust their features in response to it. In existing models, the classifier is the only instrument available to the decision maker, and therefore a feature that is predictive but easy to fake can only be down-weighted or discarded. However, in many settings the decision maker can also verify: lenders verify income, admissions offices check documents, and tax authorities audit returns. In this paper, we propose a model of strategic classification in which the firm jointly designs a linear classifier and an \emphaudit profile, which assigns to each fakeable feature a probability of detection and a penalty when caught. We show that under linear costs, the classifier affects the audit problem only through the distribution of gaming rents it induces, so that the joint design problem decomposes into the choice of a score rule and an audit allocation problem. We use this decomposition to characterize the optimal audit allocation, to identify when the allocation problem is tractable and when it is NP-hard (namely, when agents can game through overlapping features under an inspection cap), and to bound the regret of a firm that has to learn the rents by auditing. We further show that audit intensity is a quantity to be tuned rather than maximized: welfare is single-peaked in it, and a firm and a social planner disagree on the mix of detection and penalty that delivers a given level of deterrence. Notably, two populations with identical costs and causal structure can game in one domain and improve in the other, a difference that a cost-only model cannot account for. Together, our findings highlight that whether a feature is ``gameable’’ depends on the institution’s verification policy as much as on the feature itself.
[HC-49] he Choreographic Genome: Amplifying the Silent Structure of Text into Dance IEEE-VIS2026
链接: https://arxiv.org/abs/2609.22519
作者: Michael Li,Alison Ding
类目: Human-Computer Interaction (cs.HC)
备注: Accepted at IEEE VIS 2026 Arts Program
Abstract:Recent advances in generative artificial intelligence have enabled the synthesis of complex human motion with unprecedented fidelity. However, current text-to-motion systems rely strictly on linguistic semantics: if an input reads “I put my hands up”, the model searches for a pose with raised hands, and every non-semantic property of the text is discarded as noise. In this work, we treat that discarded structure as the signal. We present an embodied visualization instrument that amplifies not what a text means, but how it is built. Our method first quantizes dance kinematics into a motion codebook of 256 stylistic “regions” using Principal Component Analysis and K-Means clustering, and orders those regions along the dominant axis of movement. We then map the raw byte representation of any input text directly onto this codebook, producing a deterministic sequence of regions that we call the text’s “choreographic genome”. A precomputed plausibility graph and a set of physics smoothing routines turn this genome into fluid, full-body movement, so that the dancing body becomes a display surface for the byte-level structure that semantic systems ignore. Through a series of artistic case studies, including a Shakespeare sonnet, a machine error log, source code, an abolitionist’s question, and Indigenous and Devanagari scripts, we show that each text produces a visibly distinct dance, and that scripts marginalized by ASCII-centric computing are amplified into close to three times as much movement per character. We frame this not as a motion-synthesis benchmark, but as a critical and poetic visualization that asks what we choose to count as signal, and what we allow to go unheard.
[HC-50] Scalable AI-based clinical communication training and automated assessment
链接: https://arxiv.org/abs/2609.22517
作者: Masum Hasan,Ron Epstein,Thomas Carroll,Ehsan Hoque
类目: Human-Computer Interaction (cs.HC)
备注: 16 pages
Abstract:Poor clinical communication can delay care, contribute to errors, and harm patients, yet opportunities for repeated practice with feedback remain limited. Our prior randomized trial showed that practice with the SOPHIE AI patient platform improved serious illness communication, but the system addressed a single clinical context and required human effort for delivery and assessment. We developed SOPHIE 2.0, a browser-based, self-service platform integrating embodied AI-patient interactions, personalized feedback, and automated assessment across 24 clinical scenarios. An automated large language model assessor evaluated three communication skills—Empower, Be Explicit, and Empathize—with agreement comparable to individual human raters ( r=0.759 ; ICC =0.746 ). In a study of 59 clinicians and students, participants completed two AI-patient encounters with personalized feedback; 92% found the platform engaging, 86% easy to use, and 83% clinically relevant. Scores were higher in the second encounter, though the uncontrolled design precludes attributing this change specifically to training.
[HC-51] Visual Embellishments are Potential Distractions in Double-Column Reading
链接: https://arxiv.org/abs/2609.22477
作者: Songwen Hu,Chase Stokes,Marti Hearst,Cindy Xiong Bearfield
类目: Human-Computer Interaction (cs.HC)
备注: 4 pages
Abstract:Eye-catching graphics, such as circular figure labels and word-scale visualizations, are increasingly being placed directly within long-form text paragraphs. Some research has claimed that inline visualizations can help readers understand data-rich passages more clearly. However, research in the science of reading calls into question the introduction of images within the flow of text. In this work, we conduct an exploratory study of eye movement in both the presence and absence of visual embellishments. Using a high-resolution eye-tracker (EyeLink Portable Duo), we observed small mean increases in vertical and cross-column saccade rates among six participants, with substantial variation among readers and no detected difference in comprehension accuracy. Mean subjective ratings for the data-light circular-glyph passage indicated it is more distracting than its unembellished comparison passage. These exploratory observations differ from prior findings and motivate a larger, fully crossed study of inline graphics.
[HC-52] Vox-Infinity: Benchmarking the Limits of Long-Context Spoken Language Models
链接: https://arxiv.org/abs/2609.22452
作者: Xize Cheng,Wenxu Jia,Chenyuhao Wen,Dongjie Fu,Zehan Wang,Xinyu Zhang,Tao Jin
类目: Computation and Language (cs.CL); Human-Computer Interaction (cs.HC)
备注:
Abstract:Long-context understanding remains a fundamental challenge for large language models, as excessively long inputs often lead models to forget salient information. This issue is even more pronounced in the speech domain, where audio, as a low-compression modality, requires substantially more embeddings than text to preserve both semantic content and acoustic cues. To address this challenge, we introduce \textbfVox-Infinity, the first benchmark specifically designed to evaluate long-context understanding in spoken language models. Vox-Infinity systematically extends audio history along two dimensions: turn count and turn duration. It covers a diverse range of representative scenarios with varying interaction structures and semantic complexity. Crucially, Vox-Infinity provides explicit answer-provenance annotations and organizes samples according to the amount of historical context required to resolve each query, enabling precise and length-aware evaluation. Extensive evaluations of seven representative spoken language models reveal a clear overall recency effect: models generally achieve higher accuracy when answer-supporting evidence is closer to the query, but struggle to retrieve and use evidence located farther back in the dialogue history. Cases and datasets are available at this https URL.
[HC-53] Functional Emotion Without Character: Large Language Models Aristotelian Disposition and the Limits of Behavioral Alignment
链接: https://arxiv.org/abs/2609.22362
作者: Marzieh Zare
类目: Computation and Language (cs.CL); Computers and Society (cs.CY); Human-Computer Interaction (cs.HC)
备注:
Abstract:Debates about whether artificial systems can feel are often forced between two unsatisfactory positions: behavioral equivalence is treated as sufficient for emotion, or phenomenal consciousness is treated as a prerequisite that makes the question empirically inaccessible. This article develops a structural alternative. It models emotions as context-sensitive regions, trajectories and attractor dynamics in high-dimensional representational state spaces. Recent mechanistic interpretability findings support the existence of causally active emotion-concept representations in large language models, but they do not establish subjective feeling or full emotional agency. Assessed against published adequacy standards for representation in language models, intervention provides strong evidence of causal use, while full affective role integration, uniformity across subject domains and coherence remain only partially established; there is no direct analogue of accuracy. These mismatches expose the need for a standard of affective appropriateness, which an account of character must supply. Such an account requires three further conditions: regulatory embodiment that gives valence endogenous stakes, temporal continuity that allows affective episodes to accumulate into a history, and an integrated self-model that binds that history to persistent values. Aristotle’s concepts of pathē, hexis, mesotēs and phronēsis are translated into a state-space sketch in which practical wisdom includes competence in estimating normatively salient context, not merely acting on a context description already given. The framework reframes alignment as a problem of durable disposition rather than output conformity, and yields interventional tests with explicit control conditions.
[HC-54] A Tutorial on Prompt Engineering: From Messy Thoughts to AI Workflows
链接: https://arxiv.org/abs/2609.22249
作者: Erfan Loweimi,Hadi Daneshvar,Samira Loveymi,Samir Ouelha,Zhengjun Yue,Hajar Mozaffar,Saturnino Luz
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC)
备注:
Abstract:This paper treats prompt engineering as a discipline for turning informal human intent into structured AI work specifications. It develops the practice as a sequence of reusable design moves: define the work, construct only the context the answer depends on, choose a role, or a moderated panel of roles, as an attention lens, and state affirmative quality targets, reserving prohibitions for hard boundaries. To keep prompts lean, it adapts two classical principles, Occam’s razor and Chekhov’s gun, so that every instruction earns its place. For consequential tasks, it adds structured critique through steelmanning and premortems, followed by verification and, where tools or multi-step actions are involved, agentic operating loops with explicit boundaries and escalation. Aimed at a general readership, this tutorial is not a benchmarking study; it offers a practical, technically grounded path from casual prompting to disciplined AI workflow design, illustrated by a worked example that carries one task from a weak prompt to a strong specification. The framework is presented as principles and checklists that remain useful as models and tools change. The strongest prompt is rarely the longest prompt; it is the one that makes desired behaviour, required sources and checks, and success criteria unmistakable.
[HC-55] he Role of AI in Online Reviews
链接: https://arxiv.org/abs/2609.22198
作者: Valeria Lerman,Oren Rigbi,Yaniv Dover
类目: Computation and Language (cs.CL); Artificial Intelligence (cs.AI); Human-Computer Interaction (cs.HC); Social and Information Networks (cs.SI)
备注:
Abstract:The rapid adoption of large language models (LLMs) creates new opportunities for strategic content generation on online platforms, including potentially harmful forms of manipulation that may undermine platform effectiveness and reshape platform dynamics. However, measuring such activity is difficult because AI-generated content is rarely directly observable. We introduce an empirical approach that leverages discrete LLM supply shocks - abrupt changes in model prices and capabilities, and contrasts verified with non-verified reviews to identify changes in platform activity associated with generative AI supply improvements. We apply this approach to more than 13 million reviews from Trustpilot, one of the leading online platforms for business reviews. A robust finding is that following LLM supply shocks, unverified reviews shift toward greater negativity: more 1-stars, fewer 5-stars, and lower ratings, with effects driven primarily by new model releases and concentrated among firms with the lowest and highest review volumes, suggesting that strategic AI use may reshape platform competition dynamics. We further find that LLM supply shocks trigger short, concentrated bursts of review activity. Together, these findings suggest that generative AI is already reshaping how reputation and competition operate on online platforms.
[HC-56] Learning Dynamic Neural Evidence Representations for Time-Adaptive Brain-Computer Interfaces
链接: https://arxiv.org/abs/2609.22088
作者: Beining Cao,Ziyi Zhao,Xiaowei Jiang,Daniel Leong,Yingtao Ren,Thomas Do,Yu-Cheng Fred Chang,Chin-Teng Lin
类目: ignal Processing (eess.SP); Human-Computer Interaction (cs.HC); Machine Learning (cs.LG); Neurons and Cognition (q-bio.NC)
备注:
Abstract:Brain-computer interfaces (BCIs) decode neural activity into commands, yet most existing systems rely on fixed-window decoding that may result in redundant observation or unreliable predictions due to insufficient evidence. Adaptive temporal decision-making (ATDM) addresses this accuracy-time trade-off by progressively accumulating EEG evidence and deciding when to stop. However, existing EEG encoders are mainly designed for fixed-window decoding and may not provide reliable state representations under variable observation lengths. In addition, current ATDM-oriented encoders are typically tailored to specific EEG paradigms, limiting their applicability across different BCI tasks. To address these limitations, we propose ProtoTrigger, a two-stage prototype learning-based EEG state encoder for ATDM. ProtoTrigger uses prototype matching to extract stable local EEG embeddings and prototype-based attention to aggregate decision-relevant temporal evidence during progressive observation. Offline evaluations across three EEG paradigms demonstrated state-of-the-art accuracy-time trade-offs and strong generalizability across different EEG paradigms. An online human-in-the-loop augmented reality-based BCI experiment further demonstrated its real-time feasibility. These results suggest that ProtoTrigger provides a general EEG state encoding framework for efficient ATDM-based BCI systems.
计算机视觉
[CV-0] GameHorizon Suite: Multi-Horizon Data and Evaluation in Gameplay
链接: https://arxiv.org/abs/2609.25001
作者: Yiran Wang,Xingyilang Yin,Junfu Pu,Guangzhi Wang,Kaifeng Li,Mingyu Ouyang,Huiqiang Sun,Lingen Li,Cheng Cheng,Wangbo Yu,Honghao Chen,Xiaodong Cun,Chi-Man Pun,Zhiguo Cao,Ying Shan
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: We will release our dataset, annotator, and benchmark to facilitate future research. Github Repo: this https URL Project Page: this https URL
Abstract:Modern video games provide a measurable testbed for AI models, combining abilities of visual understanding, instruction decomposition, goal planning, and precise action control over multiple temporal horizons. Existing datasets and benchmarks, however, either cover a narrow range of games, lack language instructions, or rely on high-variance online rollouts. To address these challenges, we introduce GameHorizon, a unified data and evaluation suite that measures gameplay capabilities at different horizons for diverse model families. GameHorizon Suite consists of three components. First, GameHorizon-Annotator is a scalable and automated annotation pipeline for multi-horizon instructions. Second, utilizing the pipeline, we construct GameHorizon-Data, the first large-scale AAA gameplay dataset with temporally aligned videos, player actions, and multi-horizon instructions. It comprises 5,000 hours of recordings from 21 games, collected by 100 human expert players. Third, we build GameHorizon-Bench with reproducible offline and stepwise online testing. The offline track enables reproducible evaluation using thousands of standardized questions organized into three primary tasks and a series of diagnostic variants, while the online track tests whether offline scores reflect actual gameplay capabilities and localizes failures to specific steps within long-horizon gameplay. Based on our GameHorizon Suite, we evaluate 47 models through more than one million model invocations, revealing a meaningful hierarchy of task difficulty and pronounced differences in model capabilities. Our work can provide a standardized yardstick for evaluating gameplay capabilities across horizons and model families. We will release our dataset, annotator, and benchmark to facilitate future research.
[CV-1] VideoGen-Agent : Reinforcing Video Generation Agents
链接: https://arxiv.org/abs/2609.24997
作者: Binxu Li,Haoyi Duan,Yuhui Zhang,Yaohui Zhang,Zihao Lin,Kaituo Feng,Suozhi Huang,Xiangyi Li,Yu Li,Chunyuan Li,Shilong Liu,Mengdi Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Recent advances in video generative models have enabled high-fidelity, temporally coherent video generation. However, these models often struggle to satisfy prompts requiring specialized knowledge, specific identities, physical consistency, or ordered events. In this paper, we present VideoGen-Agent, a multimodal agent trained through multitask agentic reinforcement learning to use external tools for video generation. The agent coordinates augmentation, generation, and verification tools through multi-turn interactions, using the prompt and intermediate observations to guide its decisions. We train a shared policy on a category-balanced dataset spanning six tasks. Supervised fine-tuning on teacher-generated trajectories establishes tool-use behavior, which is then refined through reinforcement learning. A category-aware hybrid reward evaluates tool-call validity, task-appropriate tool use, and generated video quality. We further introduce VABench, a held-out benchmark of 600 prompts covering procedural knowledge, single- and multi-entity identity preservation, physical consistency, scene composition, and multi-shot temporal structure. On VABench, VideoGen-Agent improves over its base text-to-video generator by 19.1 points, from 56.5 to 75.6. Upgrading the generation tools further raises the score to 86.1 without additional agent training. Human raters prefer the upgraded configuration over the strongest standalone baseline in 84.3% of comparisons. These results support learning tool use across video-generation tasks and show that the trained agent can benefit from subsequent advances in generation tools.
[CV-2] WorldCrafter: Consistent Video World Model with Implicit 3D-aware Memory
链接: https://arxiv.org/abs/2609.24984
作者: Wangbo Yu,Kunhao Liu,Wenbo Hu,Shenghai Yuan,Chaoran Feng,Haiyang Zhou,Yukun Huang,Yiran Wang,Wang Zhao,Yingmin Luo,Ying Shan
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Graphics (cs.GR)
备注: Project webpage: this https URL
Abstract:Video world models enable interactive exploration of dynamic environments, yet struggle to respect prior observations over long horizons and across viewpoints. We present WorldCrafter, a video world model that learns a camera-queryable implicit 3D-aware memory for this purpose. The key insight is to let the requested viewpoint shape how multi-view evidence is compressed into the video generator’s limited token budget. Trained jointly with the video generator, a memory encoder and pose-conditioned readout module integrate historical observations into a fixed set of target view-specific tokens before denoising, without explicit depth-based correspondences. By combining this memory with recent temporal context and few-step distillation, WorldCrafter enables streaming scene exploration from a single input image or text prompt. Experiments across static and dynamic scenes show substantial gains in long-horizon consistency and camera-control accuracy while preserving visual quality during minute-scale exploration.
[CV-3] GAE: Learning a Geometry-Native Latent Space for 3D-Consistent World Generation
链接: https://arxiv.org/abs/2609.24981
作者: Jiahao Lu,Minghao Yin,Wenbo Hu,Hengyu Liu,Wang Zhao,Sai-Kit Yeung,Ying Shan,Yuan Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project page: this https URL Github: this https URL
Abstract:We present a compact geometry-native latent space as a shared foundation for perception and generation. Visual generators can produce photorealistic frames without preserving a consistent 3D scene. We argue that this is not only a modeling problem but also a representation problem: generators typically evolve appearance-centric latents, while perception models recover geometry in a semantically rich space that encodes cross-view structure. Rather than adding geometry as another output, we reparameterize a geometry foundation model’s features into a compact latent space for generation. We realize this shift with the geometry-native autoencoder (GAE), whose latent is jointly decodable to appearance, depth, cameras, and point maps. With this state, a standard conditional flow supports diverse generation tasks. In controlled comparisons that hold the generator and training protocol fixed, replacing the latent with GAE improves both visual quality and independently measured 3D coherence: FVD falls by 12.7% and 23.1% on RealEstate10K and DL3DV, and camera-trajectory error is halved on RealEstate10K. Together, these results show that the latent space is central to geometry-consistent generation and can serve as a shared interface between perception and generation.
[CV-4] DexTacWAM: A Visuo-Tactile World-Action Model for Dexterous Manipulation
链接: https://arxiv.org/abs/2609.24976
作者: Haoran Yuan,Zekai Wang,Boning Shao,Haoran Lu,Trevor Darrell,Ismini Lourentzou,Wei Zhan
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: 22 pages. Project website: this https URL
Abstract:Dexterous manipulation depends on contact dynamics that are often only partially observable from vision. Recent World-Action Models (WAMs) couple predictive video world modeling with action generation, but remain largely vision-centric and therefore cannot directly model these contact dynamics. We present DexTacWAM, a visuo-tactile WAM that encodes each fingertip independently, aggregates the resulting features through a finger- and pose-aware tactile compressor, and injects the tactile latent into a video diffusion world model for joint visuo-tactile world modeling. Across six contact-rich dexterous manipulation tasks on a 22-DoF bimanual platform, DexTacWAM achieves the highest score on every task, averaging 70.6 versus 38.0 for the strongest baseline. Ablations attribute the gain to modeling contact evolution as part of the predicted world state rather than tactile conditioning alone: removing tactile world modeling reduces the four-task mean from 74.7 to 26.6 while keeping the same tactile features and action expert. After four hours of tactile-encoder adaptation with a frozen pretrained vision VAE, our continual vision-to-touch learning extends the pretrained video model to touch using roughly 100 demonstrations per task without tactile midtraining, while retaining visual prediction quality within 0.5 dB of vision-only counterparts. The compressor retains 89.4% of pre-fusion contact recall while enabling 2.26x faster training and 1.29x faster inference. Together, these results show that pretrained video priors can be extended to distributed multi-finger contact dynamics in a data- and compute-efficient manner.
[CV-5] Anatomy-Decomposed Chest Computed Tomography (CT) Projections as Scalable Supervision for Bone Suppression in Chest Radiographs
链接: https://arxiv.org/abs/2609.24937
作者: Mrunmay Angaitkar,Piyush Kumar,Aarjav Satia,Pranav Rao,Ashish Mittal,Manoj Tadepalli,Preetham Putha
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Bone overlap can obscure abnormalities in chest radiographs, while scarce paired training data limit supervised bone suppression. We address this challenge with a digitally reconstructed radiograph (DRR) framework that converts chest computed tomography (CT) into paired supervision for component suppression. A novel bone segmentation algorithm enables CT decomposition into bone, non-lung soft-tissue, and lung components, which are projected separately. Their weighted combination yields synthetic radiographs with pixel-registered component images that sum exactly to the full DRR. Models trained on these data suppress bone or lung components by predicting the target component and recovering the remainder by subtraction, transferring to real radiographs without real paired training data. As an extension, their outputs on real radiographs provide target domains for unpaired, component-wise DRR translation, reducing the appearance gap while retaining anatomical details. Across multiple public datasets, downstream detection experiments demonstrate the utility of bone suppression, with gains concentrated on abnormalities with substantial bone overlap. Compared with open-source DRR engines applied to the same CTs, our unmodified DRRs achieve comparable realism and preservation of label-relevant anatomy, while translated DRRs achieve the best Fréchet inception distance (FID), lung-field sharpness, and agreement with source-CT anatomy among the evaluated methods. Models and inference code: this https URL Translated projections: this https URL.
[CV-6] PixelDiT2: Representation-Grounded Pixel Diffusion Transformers
链接: https://arxiv.org/abs/2609.24919
作者: Yongsheng Yu,Wei Xiong,Yichen Sheng,Shiqiu Liu,Jiebo Luo
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Recent advances in pixel-space diffusion models have narrowed the image quality gap with latent-space diffusion, but still converge more slowly and lag behind in final image quality. We argue that a key reason is the lack of an explicit representation prior: unlike latent diffusion, which usually denoises in a compact and structured latent space, pixel diffusion needs to learn denoising-friendly representations and pixel generation simultaneously from raw RGB space. To address this problem, we propose PixelDiT2, an end-to-end pixel-space diffusion model designed to decouple representation learning from pixel generation without introducing an autoencoder or latent reconstruction bottleneck. We propose representation grounding that uses a frozen pretrained vision foundation model to provide explicit per-patch representation guidance throughout denoising, allowing the pixel diffusion transformer to focus more on pixel generation. On ImageNet-256x256, PixelDiT2 achieves an FID of 1.46 after 600 epochs; at 512x512 resolution, PixelDiT2 achieves an FID of 1.48 after 680 epochs.
[CV-7] Generating Chest X-Ray Counterfactuals by Specialising Foundation Image Models
链接: https://arxiv.org/abs/2609.24879
作者: Xiaodan Xing,Rajat R. Rasal,Julia A. Meister,Sara Ghorayeb,Galvin Khara,Jessica Schrouff
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 32 pages, 6 figures, 16 tables. Code: this https URL
Abstract:Counterfactual image generation answers questions about how a subject would have looked under retrospective, hypothetical scenarios. Recent methods have improved perceptual quality, identity preservation and faithfulness to an underlying causal model, but their adoption in healthcare is limited by scarce annotated data, distribution shift between datasets, and mismatches between pretrained generative models and those required for counterfactual inference. We propose specialisation, a data and parameter-efficient framework for adapting pretrained, non-causal generative models into causal mechanisms under distribution shift. Based on this framework, we train a radiology counterfactual image generation model, called RadCF, using latent flow matching. We validate our approach on three chest X-ray datasets spanning different dataset shifts, data volumes, and counterfactual questions, associated with challenging, highly-localised interventions. Our results show that RadCF and specialisation improve counterfactual soundness over existing methods while being data and parameter efficient, and that the resulting counterfactuals can detect and mitigate shortcut learning in a downstream medical classifier. Code is available at this https URL.
[CV-8] SPHQuant: Efficient extreme low bit weight quantization for Vision-Language Models
链接: https://arxiv.org/abs/2609.24875
作者: Kewei Zhang,Zheng Chen,Haotong Qin,Yulun Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 16 pages, 5 figures, including appendix. Code will be released at this https URL
Abstract:Recent foundation models are moving toward native multimodal Vision-Language Models (VLMs), making VLMs a central form of next-generation foundation models. However, their large language backbones make edge deployment difficult due to high memory footprint and memory-bound autoregressive decoding. Weight-only post-training quantization is a practical solution, but pushing VLMs to extreme low bit-widths remains challenging: existing rotation-free methods suffer from outliers at 2-3 bits, while rotation-based methods improve accuracy at the cost of additional runtime overhead. We propose SPHQuant, a rotation-free spherical weight-only quantization framework for VLMs. Instead of quantizing weights directly in Cartesian coordinates, SPHQuant decomposes each 8D weight vector into coordinate signs, radius, and a positive unit direction. This representation isolates outlier magnitude into the radius while keeping directions bounded and statistically regular. Based on this insight, SPHQuant allocates extra precision to the radius to mitigate accuracy degradation induced by outliers. It further uses a compact positive-direction codebook and fine-tunes codebook entries through angular parameterization to preserve the unit-sphere constraint. We also design a hardware-friendly GEMV kernel that keeps the direction codebook small enough for shared-memory lookup and packs radial bits efficiently. Experiments show that SPHQuant matches the performance of state-of-the-art extreme low-bit quantization methods while improving decode throughput over QTIP by 30.3% on RTX A6000. Code will be released in this https URL.
[CV-9] DTKDP: A Dual Teacher Knowledge Distillation and Pruning Framework for Lightweight Oriented SAR Ship Detection
链接: https://arxiv.org/abs/2609.24872
作者: Yuming Li,Fan Zhang,Alin M. Achim
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Published in Remote Sensing. The version of record is available at this https URL
Abstract:Two-stage oriented detectors achieve high localization accuracy in synthetic aperture radar (SAR) ship detection, but their large backbones, feature pyramids, proposal modules, and heavy region of interest (RoI) heads hinder deployment. Existing lightweight SAR ship detectors typically use one-stage frameworks that lack proposal-level refinement for precise rotated localization. This paper presents a dual-teacher knowledge distillation and pruning (DTKDP) framework for lightweight oriented SAR ship detection. DTKDP introduces learnable gates into convolutional, normalization, and linear layers to prune convolutional channels and RoI-head neurons. Rotated proposal alignment (RPA) distills teacher and student predictions in a shared teacher-generated rotated proposal space, while a dual-teacher scheme combines classification and regression guidance from a homogeneous main teacher with complementary classification cues from a heterogeneous auxiliary teacher. Experiments on the SAR Ship Detection Dataset (SSDD) and Rotated Ship Detection Dataset in SAR Images (RSDD-SAR) show that DTKDP reduces the parameters of Oriented Region-based Convolutional Neural Network (Oriented R-CNN) and RoI Transformer equipped with ResNet-50 backbones by 87.5-91.8% and their floating-point operations (FLOPs) by 75.6-79.9%. In terms of average precision (AP) and mean average precision (mAP), the resulting Oriented R-CNN-slim and RoI Transformer-slim retain accuracy close to their full-scale counterparts. Relative changes across \mathrmAP_50 , \mathrmAP_75 , \mathrmmAP_50:75 , and \mathrmmAP_50:95 range from a 2.38% decrease to a 0.65% improvement. Compared with RTMDet-tiny, they improve all four metrics on both datasets by 0.52-27.55% and consistently surpass representative distillation methods, demonstrating a favorable accuracy-efficiency trade-off.
[CV-10] Revisiting Multi-View Stereo: A Sequence-to-Sequence Formulation
链接: https://arxiv.org/abs/2609.24850
作者: Aoxiang Fan,Corentin Dumery,Nicolas Talabot,Pascal Fua
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Computing accurate geometry from multi-view images is a fundamental problem in computer vision. Recent feed-forward (FF) models jointly estimate 3D geometry and camera parameters, but they typically suffer from geometry distortion caused by reconstruction ambiguity, even when ground-truth camera parameters are supplied. In this paper, we study the multi-view stereo (MVS) problem with known camera parameters and propose a novel approach that bridges conventional MVS and FF methods. Rather than casting MVS as a sequence-to-one mapping that predicts depth only for a single reference view, we reformulate it as a sequence-to-sequence task, akin to FF models, that jointly predicts geometry for all input views. We introduce a global transformer-based architecture with two components that explicitly exploit camera-induced priors: ray-map embeddings that inject camera parameters into image patch tokens, making the transformer camera-aware, and a unified global cost volume that replaces conventional per-view cost volumes to jointly capture 3D structure across all views. Extensive experiments on multiple public benchmarks show our approach achieves state-of-the-art performance, surpassing both MVS and FF reconstruction baselines.
[CV-11] When Wider Views Fail: Stress-Testing Feed-Forward 3D Reconstruction
链接: https://arxiv.org/abs/2609.24839
作者: Daisy Li,Kyle Gao,Quanyun Wu,Hanna Chomko,John S. Zelek,Jonathan Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Feed-forward 3D reconstruction models enable efficient geometry estimation from sparse images, but their pretrained nature can make them vulnerable to distribution shifts beyond their training data. Identifying these failure modes is important for understanding when such models can be reliably deployed in unconstrained imaging settings. We investigate viewpoint variation as a controlled distribution shift by varying the angular span of sparse image inputs while keeping the input budget fixed. Across multiple feed-forward reconstruction models, we observe substantial degradation as viewpoint span increases, with wide spans producing both incomplete surface coverage and geometry unsupported by the observed imagery. These results reveal that viewpoint variation can induce failure modes beyond conventional reconstruction incompleteness, highlighting the need to evaluate pretrained feed-forward models under distribution shifts that challenge their learned geometric priors.
[CV-12] ZVeC: A Zero-Shot Framework for Instance-Level Vehicle Extraction and Generative Point Cloud Completion
链接: https://arxiv.org/abs/2609.24825
作者: Daisy Li,Kyle Gao,Quanyun Wu,Boris Jutzi,John S. Zelek,Jonathan Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:LiDAR point clouds acquired in underground environments exhibit severe geometric incompleteness due to occlusions and limited sensor viewpoints, making reliable point cloud completion challenging without large supervised datasets. We propose ZVeC, a zero-shot, instance-driven framework that reformulates scene-level completion as compositional object-level reconstruction. By decomposing a scene into semantic object instances, ZVeC reduces reconstruction ambiguity in cluttered environments while eliminating the need for scenario-specific training. Each segmented vehicle is completed independently using a depth- and 3D Gaussian-conditioned diffusion model that exploits generalized geometric priors before the reconstructed instances are recomposed into the original scene. To evaluate our approach, we construct a real-world dense LiDAR benchmark of underground parking environments. Experimental results demonstrate consistent improvements over representative scene-level baselines in both quantitative metrics and visual quality. The completed point cloud differs substantially from the measured input (average KL divergence ~ 2.1), yet reducing the input to only 1% of the original LiDAR measurements changes the completed reconstruction only marginally (KL divergence 0.50). This demonstrates that ZVeC produces geometrically consistent completions even under extreme input sparsity.
[CV-13] Mobile Imaging Solutions for Medical Diagnosis: Trends and Applications
链接: https://arxiv.org/abs/2609.24814
作者: Syed Muhammad Ibne Zulfiker,Tanzima Hashem,Fariha Tabassum Islam,Md Sultanul Arifin,Khandker Aftarul Islam,Nishat Anjum Bristy,Faria Huq,Priyeta Saha,Syeda Nahida Akter,Arpita Saha
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Advances in processing power, camera technologies, and mobile image analysis have made smartphones and other mobile devices, such as laptops, increasingly suitable for medical diagnosis and healthcare applications. Researchers have developed low-cost solutions for the early detection and monitoring of various health conditions, including eye and ENT diseases, malnutrition, heart rate variability, skin and oral conditions, and injuries, using images captured by non-medical devices such as smartphones and webcams. This survey examines existing research on mobile image-based medical diagnosis, with an emphasis on its potential to enable low-cost and accessible healthcare. We comparatively analyze state-of-the-art solutions across different healthcare application categories, examining their advantages and limitations. Based on this analysis, we identify desirable characteristics of mobile image-based diagnostic tools and highlight areas where existing approaches have made progress as well as areas requiring further research. We also discuss application-specific and common challenges and outline directions for future research. Overall, this study provides a comprehensive overview of mobile image-based healthcare solutions and their potential to support low-cost disease diagnosis and monitoring, particularly for underserved populations in remote and resource-constrained settings.
[CV-14] INTCORT: Training-Free Spatial Reasoning Enhancement for Vision-Language Models via Input Transformations and Confidence Routing
链接: https://arxiv.org/abs/2609.24813
作者: Haoran Sun,Jingqi Xu,Yanhui Li,Enci Liu,Kaidi Xu,Yanwei Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision-Language Models (VLMs) have demonstrated remarkable capabilities in multimodal tasks, yet they still exhibit poor ability in spatial reasoning. Existing training-dependent and training-free enhancement methods suffer from high computational costs with catastrophic forgetting and internal mechanism interference that compromises general capabilities, respectively. In this work, we first verify two key hypotheses: appropriate geometric image transformation and query-reversal transformation can recover incorrect spatial predictions, and correct predictions exhibit higher relation-token confidence than incorrect ones. Based on these findings, we propose INTCORT, a training-free spatial reasoning enhancement framework that constructs multiple inference views through input transformations and aggregates their predictions via relation-token confidence routing, without modifying the VLM’s internal mechanisms. Experimental results on several commonly-used benchmarks demonstrate that INTCORT substantially improves spatial reasoning accuracy across diverse VLMs, achieving an average improvement of 10.01% over all models and benchmarks. Compared with prior works, INTCORT achieves superior performance with improvements of up to 25.01%.
[CV-15] Streaming Video Editing with Easy Adaptation
链接: https://arxiv.org/abs/2609.24788
作者: Yujia Hu,Jiajun Li,Zihao He,Songhua Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:In this paper, we propose SVEET, a framework that requires merely training on a pretrained bidirectional video diffusion model but supports high-quality streaming video editing in an auto-regressive fashion. To tackle this problem, we first systematically revisit existing video-to-video diffusion approaches and identify two key principles for such streaming adaptation: backbone feature disentanglement and conditional frame independence. Building on these insights, we develop a novel paradigm for controllable video generation. At its core, an auxiliary model branch encodes source video inputs with temporally independent self-attention, and the intermediate features are injected into the corresponding backbone blocks for streaming-compatible control. Moreover, to bridge the discrepancy between the feature spaces of bidirectional and streaming models, we propose a decoupled training scheme that explicitly enforces the orthogonality between the optimization directions of video controllability and model causality. Such disentanglement ensures compatibility between the two objectives at inference and facilitates smooth zero-shot knowledge transfer across heterogeneous backbone architectures. Extensive experiments demonstrate that SVEET achieves superior editing quality while maintaining real-time performance, attaining 15 FPS on a single H100 GPU 17 without any auxiliary acceleration techniques. Codes are available at this https URL.
[CV-16] oward a foundation model for forest point clouds
链接: https://arxiv.org/abs/2609.24787
作者: Yuanwen Yue,Stefano Puliti,Damien Robert,Atakan Topaloğlu,Binbin Xiang,Maciej Wielgosz,Jan Dirk Wegner,Rasmus Astrup,Christian Rupprecht,Konrad Schindler
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: Project page: this https URL
Abstract:Forest inventories increasingly rely on artificial intelligence (AI) models to derive forest attributes from large-scale 3D point clouds. Current models are typically specialized to a single task, sensor, and forest type, making adaptation expensive in terms of annotations, computation, and expertise. We ask whether a single pretrained model can instead learn transferable representations across diverse forest inventory settings. Inspired by recent developments in language modelling and computer vision, we take a step toward a foundation model (FM) for 3D forestry. Using LitePT as backbone, we first establish a strong supervised baseline that sets a new state of the art on forest semantic and instance segmentation, tree species classification, and age regression benchmarks. We then curate a large-scale unlabelled corpus spanning airborne, UAV, and mobile laser scanning across diverse forest ecosystems, and pretrain the same backbone using self-supervised learning. We systematically evaluate representation learning strategies by comparing training from scratch, supervised pretraining, and self-supervised pretraining across four representative forestry tasks, under varying annotation budgets. Compared with training from scratch, self-supervised pretraining accelerates model convergence and consistently improves performance when annotations are scarce. Compared with task-specific supervised pretraining, self-supervised pretraining yields more transferable representations across downstream forestry tasks. These findings identify the practical regime in which pretrained representations are most valuable and suggest that instance discrimination, rather than forest semantics, is the main remaining obstacle to a general-purpose 3D forest foundation model. Code and models are available at: this https URL.
[CV-17] Virtual neural networks: hundreds of souls in a body
链接: https://arxiv.org/abs/2609.24782
作者: Petr Hurtik,Marek Vajgl,Zahra Alijani,Vojtech Molek
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:A new concept, termed virtual neural networks, is introduced, where the count of trainable parameters is kept constant, and scalability is attained purely through computational resources. This concept is an abstract framework that can be realized using any standard convolutional neural network. It merges siamese neural networks with a deep ensemble technique by generating numerous virtual models that share weights derived from a small set of physical models. The ensemble comprises up to hundreds of trained models simultaneously. All virtual networks take the same input, and their interconnected structure induces an internal distortion that boosts the entire ensemble robustness. The accuracy of the ensemble improves as the number of virtual networks increases, without changing the capacity. Virtual neural networks outperform larger capacity models, typical deep ensembles, and contemporary approaches like SWA and Masksembles. Additionally, the highest-performing individual model from the ensemble surpasses other models trained individually, even those with a greater number of parameters. Code: this http URL
[CV-18] Brain Metastases Segmentation for BraTS 2026 Task 1: A Multi-Architecture Comparison MICCAI2026
链接: https://arxiv.org/abs/2609.24769
作者: Mahdi Islam,Musarrat Tabassum
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages, 1 figure. Accepted as a poster at the BraTS 2026 Challenge, MICCAI 2026. Code: this https URL
Abstract:Brain metastases are the most common intracranial malignancy, occurring in roughly 30% of patients with primary solid tumors and carrying a median survival near 5.9 months. Automated segmentation is critical for treatment planning and volumetric monitoring, but metastases are frequently small, numerous, and heterogeneous in size within a single patient. We compare a plain nnU-Net baseline, a Residual Encoder Large (ResEncL) variant, region-based training, and a Primus transformer model for BraTS-METS 2026 Task 1, using patient-grouped cross-validation to prevent leakage from the longitudinal UCSD subset. Primus (label-based) is our strongest individual model by aggregate DSC/NSD, achieving 0.710/0.761 (ET), 0.742/0.785 (TC), 0.683/0.689 (WT), and 0.531/0.436 (RC). ResEncL trails Primus on aggregate DSC/NSD but achieves substantially higher lesion-wise F1 (e.g. ET: 0.452 vs. 0.052); a probability-averaging ensemble of the two only partially preserves ResEncL’s F1 advantage (ET lesion-wise F1: 0.064). We further report three postprocessing and label-reconstruction pitfalls we believe generalize beyond this challenge. Code is available at this https URL.
[CV-19] PrismGPT : Proxy-Guided Learning for Region-Aware Photo Editing with Self-Synthesized Reasoning ACM-MM2026
链接: https://arxiv.org/abs/2609.24768
作者: Ke Zhao,Hue Nguyen,Abhijith Punnappurath,Zhongling Wang,Iqbal Mohomed,Michael S. Brown
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accepted to ACM MM 2026
Abstract:Professional photo finishing relies on both global adjustments and region-specific local edits guided by semantic masks, yet current automated methods handle this workflow only partially. We present PrismGPT, a Vision-Language Model (VLM) framework that produces structured, region-aware editing plans from a single input image without relying on commercial black-box tools. Training a VLM to simultaneously diagnose aesthetic deficiencies at both global and local levels while predicting precise editing parameters is challenging due to the vast combinatorial decision space. We address this through proxy-guided learning: two simpler proxy tasks – operation decomposition and region-aware aesthetic ranking – teach the foundational skills the model needs, while a competence-based dynamic scheduler automatically rebalances the multi-task training ratio, progressively shifting emphasis from the proxy tasks to the primary editing task as each skill is mastered. Crucially, all reasoning traces used for supervised fine-tuning are self-synthesized by the same base model, eliminating the need for a stronger external teacher. Experiments on MIT-Adobe FiveK and SPIRE, a new professionally retouched benchmark we introduce, show that PrismGPT achieves state-of-the-art results while using only ~6% of the training data compared to the previous best method.
[CV-20] Ananke: Contractive Torus Attractor Networks
链接: https://arxiv.org/abs/2609.24737
作者: Zhongping Ji
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 17 pages
Abstract:We introduce Ananke, a representation-learning framework that scaffolds latent representations onto a structured product-torus prior, and its flagship visual backbone realization, Contractive Torus Attractor Networks (CTAN). By factorizing high-dimensional latent spaces into an orthogonal direct sum of two-dimensional phase planes ( \bigoplus_k=1^K \R^2 ), Ananke coordinates feature updates via a decoupled dual-phase continuous flow: skew-symmetric Hamiltonian transport moves features tangentially along energy level sets to preserve semantic phase invariants, while signed gradient dissipation contracts transverse perturbations normally toward target invariant manifolds. For circular potential families with frozen parameters, logarithmic radial feedback yields the Exact Log-Symplectic Flow (ELSF), an analytical closed-form mapping with exact exponential decay of log-radius error that evaluates in a single forward pass without numerical integration. We establish local input-to-state bounds for level-set deviations and log-radius errors, and characterize the normal hyperbolicity and persistence of the ideal product torus under bounded perturbations. We further formulate the architecture through Lie–Trotter operator splitting, unifying spatial depthwise diffusion with local manifold contraction, and analyze both exact trigonometric flows and hardware-friendly symplectic dual-shear variants. Across natural image benchmarks (CIFAR-100) and clinically challenging endoscopy datasets (Kvasir-v2), CTAN demonstrates exceptional parameter efficiency: an ultra-compact hierarchical model with merely 0.27M parameters achieves 90.52% accuracy on Kvasir-v2, outperforming 25M+ baselines (ResNet-50, DenseNet-161) by nearly two orders of magnitude in capacity, while scaled variants attain 80.32% top-1 accuracy on CIFAR-100.
[CV-21] MiTHras: Task-specific Hierarchical Semi-supervised Contrastive Masked Autoencoder for Mitotic Figure Analysis
链接: https://arxiv.org/abs/2609.24736
作者: Trinh T. L. Vuong,Simon Graham,Quoc Dang Vu,Phat T. H. Ho,Jeewoo Lim,Mostafa Jahanifar,Nasir Rajpoot,Jin T. Kwak
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 15 pages, 4 figures, 9 tables. Includes supplementary material
Abstract:Mitotic figure (MF) analysis supports tumor grading and prognostic assessment, but automated models remain sensitive to differences in tissue type and image acquisition. We present MiTHras, a task-specific pretraining framework that combines pseudo-label-guided image- and token-level contrastive learning with masked reconstruction. We construct TCGA-MF-Pseudo, a corpus of 1.8 million cell-centered images from 14 TCGA cohorts spanning 11 organ sites. Comprehensive evaluation on MF classification, detection, count-based survival prediction, and subtype classification demonstrates the efficacy of MiTHras. It achieves the highest mean F1 on all three MF classification benchmarks and both subtype benchmarks. MiTHras also outperforms general-purpose and pathology foundation encoders by a larger margin under frozen-encoder linear probing than under full fine-tuning. Although detection gains are modest due to a shared candidate-detection stage, ablations confirm that token-level supervision improves typical-versus-atypical classification and linear probing. These findings establish that MiTHras yields robust, transferable representations for automated mitotic activity assessment.
[CV-22] GraphSVR: q-Space–Aware Graph-Based Slice-to-Volume Registration for Diffusion MRI
链接: https://arxiv.org/abs/2609.24732
作者: Noga Kertes,Daphna Link Sourani,Alex M. Bronstein,Moti Freiman
类目: Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV)
备注:
Abstract:Diffusion-weighted imaging (DWI) remains highly vulnerable to subject motion, particularly in time-efficient protocols and in motion-prone populations. While slice-to-volume registration (SVR) can mitigate inter-slice and inter-stack misalignment, diffusion MRI introduces additional complexity due to diffusion-direction-dependent contrast and the requirement to align dozens of measurements within a common reference frame, effectively yielding a 4D registration problem. Existing approaches rely primarily on sequential modeling or pairwise similarity and often degrade under sparse gradient sampling or severe motion. We introduce GraphSVR, a q-space-aware graph-based framework for 4D SVR registration in DWI. GraphSVR represents slice groups as nodes in an acquisition-structured graph, with edges encoding temporal proximity, spatial slice geometry and diffusion encoding relationships. A graph neural network predicts globally consistent stack-wise rigid motion, optimized in a self-supervised, zero-shot manner using only an anatomical reference image, without requiring paired ground-truth motion. We evaluate GraphSVR using both fully synthetic diffusion simulations and realistic recombination-based simulations from real acquisitions with controllable motion severity and gradient sparsity. Performance is quantified using grid error (mm) and rotation error relative to known ground-truth transforms. Under severe motion, GraphSVR reduces grid error and rotation error by 73% compared to FSL eddy, the standard DWI motion-correction method, with the largest gains observed in sparse-direction regimes. These results demonstrate that explicitly modeling acquisition structure through graph-based reasoning improves robustness and global consistency in 4D DWI motion estimation. Code is available at this https URL.
[CV-23] ReSTI: A Source-Grounded Audit and Repair of STI-Bench
链接: https://arxiv.org/abs/2609.24727
作者: Pengzhan Sun,Ramanathan Rajaraman,Shiu-Hong Kao,Junbin Xiao,Angela Yao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Spatial–temporal benchmarks are valid only when their questions, source annotations, and answer options identify the same physical quantity. We audit STI-Bench against the official ScanNet, Waymo, and Omni6DPose sources and find systematic coordinate-system and timestamp errors, under-specified targets and times, and disagreements between keyed options and answer details. We introduce ReSTI, a source-backed revision that reconstructs every recoverable answer under an explicit target, time, coordinate system, physical quantity, and unit. Source reconstruction reveals task-level geometric failures: ScanNet Grounding omits the required alignment between annotation and raw camera coordinate systems, while Orientation measures camera rotation on the wrong plane. ReSTI replaces these labels with explicit, source-consistent geometric definitions and corrects other source-verifiable defects, including Waymo poses evaluated at the wrong timestamp. Across 2,064 legacy questions, ReSTI retains 1,782 questions and records 282 evidence-backed exclusions. ReSTI therefore provides a conservative and source-traceable basis for evaluating precise video spatial–temporal reasoning. Project page: this https URL.
[CV-24] What Makes a Good Medical Image Tokenizer? Rethinking Reconstruction and Generation in Medical Image Tokenization
链接: https://arxiv.org/abs/2609.24691
作者: Niklas Bubeck,Yundi Zhang,Vasiliki Sideri-Lampretsa,Julian McGinnis,Jiancheng Yang,Daniel Rueckert,Jiazhen Pan
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Latent diffusion models now dominate medical image generation, and every such pipeline rests on a \emphtokenizer that compresses images into the latent codes for image generation to operate on. Thereby, the tokenizer choice bounds every downstream task from reconstruction fidelity and generation quality to the representations available for downstream analysis. Yet, medical imaging pipelines routinely utilize tokenizers from natural imaging on the hypothesis that their behavior carries over. However, this is an assumption never tested in the medical imaging regime, where datasets are orders of magnitude smaller and images exhibit far lower inter-sample variance. We present a systematic evaluation of medical image tokenizers evaluating thirty configurations across ten model families on twelve datasets at three compression factors, spanning reconstruction, generation, latent geometry, downstream classification, and memorization. We find that (1) performance on image reconstruction and generation strongly correlate, unlike prior reports on natural images; (2) modern tokenizers use nearly all of their codebook entries, but still leave most of the latent space unused; (3) training-set memorization is mild and is further suppressed by stronger latent space compression; and (4) discrete quantization can largely preserve downstream classification, with lookup-free schemes being the main exception.
[CV-25] hink Like a World Model Act Like a VLA: Distilling World-Model Representations into Compact Robot Policies
链接: https://arxiv.org/abs/2609.24682
作者: Trung Dao,Sankalp Yamsani,Jaden Park,Joohyung Kim,Yong Jae Lee
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision-Language-Action (VLA) models map observations to actions with no objective that accounts for how the world responds, so their robustness is bounded primarily by data coverage. World models carry precisely that missing objective and are better grounded for it, yet rolling the future forward costs seconds per decision and rules them out of the control loop. We show the two can be separated. What a world model knows about physical scenes lives in its \emphinternal features; generating the future is merely the objective that produced them, so the grounding can be inherited while the generative machinery is left behind. We add one feature-alignment term to ordinary VLA training: a frozen world model is run over the training frames once and cached, and the student learns to agree with that cache. No teacher is loaded during training, the projector is discarded after it, and the deployed policy is identical to the undistilled baseline, running in 32 ~ms and 1.86 ~GB on a consumer RTX~5090, so every gain is attributable to the representation rather than to added capacity or test-time compute. A 0.8 B student reaches 97.9% on LIBERO, improves from 48.2% to 50.5% on RoboCasa-GR1 humanoid manipulation, and the same objective carries over to real hardware, on both a single-arm and a bimanual platform. The gain survives changes of student scale, backbone, alignment layer, and teacher, indicating a broad representational prior rather than a fragile alignment between two particular networks. Project page: this https URL.
[CV-26] Ev-YOLO: Uncertainty-Aware Object Detection via a Unified Evidential Formulation
链接: https://arxiv.org/abs/2609.24668
作者: Simon Barbarit-Gaboriau(LITIS - STI, INSA Rouen Normandie),Hind Laghmara(LITIS - STI),Rémi Boutteau(LITIS - STI),Samia Ainouz(LITIS, LITIS - STI)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Preprint / submitted manuscript. This version has not undergone peer review. To appear in the proceedings of the 9th International Conference on Belief Functions (BFAS 2026), Springer, LNAI
Abstract:Reliable uncertainty estimation is essential for deploying object detectors in autonomous systems operating in uncertain environments. Evidential Deep Learning (EDL) provides a principled framework for uncertainty-aware classification by representing network outputs as evidence and interpreting predictions through subjective logic. However, existing evidential object detectors typically combine evidential classification with regression uncertainty models that do not share the same theoretical foundation. In this work, we propose an evidential version of YOLOv8 in which both classification and bounding-box regression are formulated within a common evidential framework. Our approach exploits YOLOv8’s distribution-based bounding-box representation, allowing the evidential formulation to be applied not only to classification but also to localisation. As a result, both tasks produce belief, uncertainty, and probability estimates that can be interpreted within the Dempster–Shafer framework. Experiments on KITTI, MUSES, and nuScenes show that the resulting detector remains broadly competitive with standard YOLOv8 in terms of detection accuracy while providing a localisation uncertainty that effectively discriminates between correct and erroneous detections. Moreover, this uncertainty becomes increasingly discriminative under domain shift.
[CV-27] High-resolution Nitrogen Dioxide Maps Reveal Exposure Limit Breaches across Europe
链接: https://arxiv.org/abs/2609.24634
作者: Linus Scheibenreif,Konrad Schindler
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Nitrogen dioxide (NO2) is a common air pollutant, released into the atmosphere through the incomplete burning of fossil fuels, and associated with respiratory and cardiovascular diseases in humans. Ambient NO2 concentrations are regulated through air-quality limits assessed with a sparse network of fixed monitors. The revised EU Ambient Air Quality Directive (2024/2881) introduces a daily NO2 limit to be met from 2030. At present, neither the regulatory monitoring network nor existing coarse, annual-mean models can resolve NO2 concentrations at the spatio-temporal resolutions necessary to assess compliance. Here we map NO2 across Europe at hourly and 10m resolution with a machine-learning model that combines ground monitors with satellite, reanalysis, land-use, traffic and emission data and returns a calibrated predictive distribution at every location. Validated against held-out regulatory monitors and independent citizen-science campaigns, the maps resolve high-resolution spatiotemporal NO2 gradients for 110 metropolitan areas in Europe. We reconstruct the daily compliance statistic across those regions and find limit breaches in 91 EU air quality zones deemed compliant by the regulatory monitoring network, covering a population of approximately 135M. Beyond air quality zones and monitor locations, an estimated 9-9.4% (20M) of the population in mapped regions lives in areas where the daily NO2 limit is breached. The high-resolution maps offer a route to population-scale assessment of compliance with the 2030 limits.
[CV-28] FedMust: Semi-supervised Multi-task Student-Teacher Federated Learning for Multi-organ CT Segmentation
链接: https://arxiv.org/abs/2609.24627
作者: Ashkan Moradi,Bendik Skarre Abrahamsen,Mattijs Elschot
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: This manuscript has been accepted for publication at the 7th International Conference on Medical Imaging and Computer-Aided Diagnosis (MICAD 2026)
Abstract:Multi-organ segmentation using deep learning requires large amounts of annotated patient data; however, institutions often lack sufficiently large and diverse annotated datasets. Privacy constraints further prevent institutions from sharing patient data to overcome this limitation. Moreover, due to the labor-intensive nature of annotation and the scarcity of diverse expertise, institutions typically have labels for only a small portion of their local data, leaving the larger unlabeled portion unused. In this work, we propose a flexible semi-supervised federated multi-task student-teacher framework that leverages federated learning (FL) to improve multi-organ segmentation using both labeled and unlabeled data across participating sites. At each communication round, the proposed framework initiates local training, where clients with labels for the same task form a federation to produce an aggregated teacher model. The resulting teachers generate task-specific features for all data at each client. Subsequently, all clients form a second federation to train a multi-task student model with a shared encoder and task-specific decoders that replicate the teacher-generated features across all segmentation tasks. The aggregated student model is then used to update the local teachers and initiate the next training round. Extensive experiments demonstrated the effectiveness of the proposed method compared with local and federated single-organ models, yielding an average performance gain of 13 percent across clients. The experiments also demonstrated the impact of multi-task learning and unlabeled data and the applicability of the framework in relaxing labeled-data requirements for client participation. The code is available at this https URL.
[CV-29] Relationally Grounded Latent World Models for Autonomous Driving IROS2026
链接: https://arxiv.org/abs/2609.24626
作者: Fabian Schmidt,Markus Enzweiler,Abhinav Valada
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: Accepted at the NeuRo-SymBolic World Models (RoBoWoMo) Workshop at IROS 2026
Abstract:Latent world models learn predictive representations for autonomous driving, but the relational semantics these states preserve often remain implicit. We investigate whether traffic scene graphs can serve as privileged semantic supervision for latent world representations. Building on LAW, we construct actor-centric scene graphs from nuScenes 3D annotations, encode their serialized relational structure using a frozen text embedding model, and align the visual latent representations with this semantic target during training. We remove the supervision branch at inference, so it requires neither scene graphs nor 3D annotations and adds no test-time computation. On nuScenes, our method reduces average trajectory L2 error from 0.661 to 0.622 (5.9%) and collision rate from 0.456 to 0.217 (52.4%) relative to our retrained LAW baseline. It also outperforms an unstructured caption-style semantic target, supporting the benefit of explicit relational structure for latent world-model representation learning.
[CV-30] Video-based Surgical Skill Assessment Using Dynamics-and-Uncertainty-Aware Tree-based Gaussian Process Classifier
链接: https://arxiv.org/abs/2609.24619
作者: Arefeh Rezaei,Mohammad Javad Ahmadi,Amir Molaei,Hamid D. Taghirad
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 4 figures, 17 tables, 31 pages. It is Under Review in scientific reports Journal
Abstract:The proposed pipeline integrates a representation-flow convolutional neural network with a dynamics- and uncertainty-aware tree-based Gaussian Process classifier. In this framework, latent motion dynamics are exploited both as discriminative representations and as a source of input uncertainty, enhancing robustness against temporal variations and abnormal motion transitions. Compared with conventional deep learning approaches, the proposed strategy requires less training data and offers improved computational efficiency. To further improve classification performance, we introduce novel semantic-aware compound kernels that effectively capture semantic, flow, and dynamic information embedded in surgical video features. In addition, uncertainty-aware kernels are developed to strengthen the robustness and practical applicability of the compound kernel framework. The proposed method is evaluated on two benchmark datasets, namely the JIGSAWS and the Cataract-LMM (Capsulorhexis) datasets. Experimental results demonstrate strong performance across both datasets, including the LOSO and LOUO evaluation protocols on JIGSAWS, including the subject-independent LOUO protocol on JIGSAWS, on which the framework attains a mean accuracy of \ph96.9%; results under the within-subject LOSO protocol are reported for comparability with prior work, achieving competitive accuracy while substantially reducing computational cost. Overall, the proposed pipeline provides an efficient and accurate framework for video-based surgical skill assessment.
[CV-31] Beyond Uniform Subspaces: Spectrum-Aware and Depth-Adaptive Fusion for Multi-Task Model Merging
链接: https://arxiv.org/abs/2609.24612
作者: Ruxi Gu,Zilei Wang,Wei Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 20 pages, 15 figures
Abstract:Model merging aims to consolidate multiple task-specific models without access to extra training process. However, existing subspace-based methods largely rely on a uniform treatment of task updates, overlooking their intrinsic spectral and depth-wise heterogeneity. We identify two key deviations from this assumption: different tasks require different subspace capacity and exhibit different tolerance to spectral transformation, while subspace projection introduces depth-dependent distortion. Based on these observations, we propose SADA-Merging, a spectrum-aware and depth-adaptive framework for data-free model merging. SADA-Merging allocates task-specific subspace capacity according to spectral complexity, adapts spectral preservation according to task-wise plasticity, and applies depth-dependent anchoring to compensate for projection-induced distortion. This enables the fusion process to adapt to both the intrinsic geometry of each task and its sensitivity across network depth. SADA-Merging operates directly on task updates and is applicable to both full fine-tuning and LoRA settings. Extensive experiments demonstrate consistent improvements over existing data-free merging methods across different task scales and adaptation settings.
[CV-32] Applications of Neural Cellular Automata: State of the Art Challenges and Opportunities
链接: https://arxiv.org/abs/2609.24595
作者: Nick Lemke,Niklas Ihm,John Kalkhof,Mirko Konstantin,Henry J. Krumb,Daniel M. Lang,Ehsan Pajouheshagar,Ario Sadafi,Arjan Kuijper,Karim Lekadir,Marco Lorenzi,Carsten Marr,Julia A. Schnabel,Anirban Mukhopadhyay
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Neural Cellular Automata (NCAs) are a new type of neural network architecture which enable accurate and robust inference at extremely small model sizes. Recently, NCAs have advanced to become interesting low-resource alternatives to convolution- and attention-based architectures for various tasks such as image analysis, synthetic image generation, and simulation. The rapid development and increased research interest necessitate a comprehensive review of the emerging technology. This review provides an overview of the fundamentals of NCAs, applications to medical imaging, as well as insights into the state of the art. We analyze recent modifications to the originally proposed NCA architecture with respect to their efficiency and accuracy. Furthermore, we review practical applications in real-world scenarios with a focus on medical image analysis, segmentation, classification, registration, depth estimation, and image synthesis. Finally, we identify several advantages of NCAs, research gaps, and conclude with an analysis of future opportunities for NCAs in medical applications in confined settings or areas that have particular demands for robustness or efficient data processing.
[CV-33] What do VLM-Based Vision-Language Navigation Models Rely on: Interpreting and Steering Policy Behavior
链接: https://arxiv.org/abs/2609.24576
作者: Débora Oliveira Makowski,Samiran Gode,Abhijeet Nayak,Marco Hutter,Cordelia Schmid,Lukas Rosenberger Schmid,Wolfram Burgard
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Modern Vision-Language Navigation (VLN) models rely mostly on pre-trained large Vision-Language Models (VLMs) to predict navigation actions. While this fusion of language instructions and visual observations allows multimodal reasoning, it obscures how information is routed across modalities or what mechanisms drive navigation decisions. Thus, it remains unclear whether VLN models ground their predictions in relevant semantic cues or can track task progress. In this work, we study the interpretability and steerability of VLN models. We use intervention-based metrics that measure how visual observations, instructions, and visual memory causally influence navigation decisions. Our results show that these navigation policies are sensitive to all input modalities and do not depend on a single one. We further show that these agents encode navigation progress and retain semantic structure from their VLM backbones, enabling concept-level steering through internal activations. Finally, we extract activation vectors for abstract behaviors to transfer them zero-shot to out-of-distribution real-world scenarios, improving performance without additional fine-tuning.
[CV-34] Active Visual Sampling with a Connectome-Constrained Fly Model for One-Shot Hatch Recognition in Architectural Drawings
链接: https://arxiv.org/abs/2609.24565
作者: Dmitry Kuklev
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 8 figures, 4 tables
Abstract:Architectural drawings encode material classes through repeated hatch patterns. We test whether a connectome-constrained fly visual network, pretrained for motion, can be repurposed without task-specific weight updates as a descriptor for one-shot hatch matching. Each 64 x 64 patch is translated over eight scan trajectories and summarized across 57 cell types; query descriptors are then matched to one legend strip per class. On 400 development sheets from a synthetic benchmark built on CubiCasa5K geometry, the frozen fly pipeline reaches 0.857 area-weighted accuracy and 0.910 with an extended legend. On an equal-brightness orientation condition it reaches 0.840 versus 0.299 for eleven pixel statistics, while a Gabor bank reaches 0.900. Replacing drift with a repeated still frame lowers the combined equal-condition score by 0.089 [0.066, 0.112]. However, a receptors-only descriptor reaches 0.891 and a task-trained 5,888-parameter CNN averages 0.959, so the current evidence supports transfer and the usefulness of active sampling, but not an advantage of the biological wiring. We separate project-recorded results from recomputed checks and report a small real-drawing audit. The supported claim is therefore narrow: motion-oriented biological vision can be repurposed as a useful texture representation for architectural hatch matching, while the topology contribution and end-to-end BIM utility remain open questions.
[CV-35] HyperCLIP: Fine-tuning CLIP forOpen-vocabulary Semantic Segmentation in Hyperbolic Space
链接: https://arxiv.org/abs/2609.24564
作者: Zelin Peng,Zhengqin Xu,Changsong Wen,Yu Huang,Yaoming Wang,Xiaokang Yang,Wei Shen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accept by TPAMI 2026
Abstract:CLIP, a foundational vision-language model, has emerged as a powerful tool for open-vocabulary semantic segmentation. While freezing CLIP’s text encoder is known to preserve its generalization capability, recent studies show that fine-tuning both CLIP’s text and image encoders jointly significantly enhances segmentation performance, especially for classes from open sets. In this work, we explain this phenomenon from the perspective of hierarchy alignment, since during fine-tuning, the hierarchical level of image embeddings shifts from image-level to pixel-level. We achieve this by leveraging hyperbolic space, which naturally encodes hierarchical structures. Our key observation is that, during fine-tuning, the hyperbolic radius of CLIP’s text embeddings decreases, facilitating better alignment with the pixel-level granularity of visual data. Building on this, we propose HyperCLIP++, a novel and parameter-efficient adaptation strategy. HyperCLIP++ directly adjusts the hyperbolic radius of CLIP’s embeddings via scaling transformations to achieve a hierarchy alignment to the target task, i.e., segmentation. To ensure this hierarchy alignment is effected consistently across both modalities and preserves their cross-modal alignment during training, HyperCLIP++ integrates a Dual Cross-Relation Communication (DCRC) module that synchronizes these adjustments between the vision and text pathways. Our experiments show that HyperCLIP++ achieves state-of-the-art performance across three benchmarks while fine-tuning only approximately 5% of CLIP’s total parameters. More importantly, we observe that after adjustment, CLIP’s text embeddings exhibit a relatively fixed hyperbolic radius across datasets, suggesting that the hierarchical level required for this segmentation task might be quantified using the hyperbolic radius.
[CV-36] Evaluating Transformation Models for pCLE Mosaic Registration
链接: https://arxiv.org/abs/2609.24560
作者: Ahmed Aboelela,Johannes Barcsay,Jana Friedhof,Miguel Gonçalves,Alexander Hann,Katharina Breininger
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Confocal Laser Endomicroscopy (CLE) provides real-time, cellular-resolution optical biopsy but has a narrow field of view, which image mosaicing can extend to provide anatomical context. Because of line-by-line acquisition, probe motion, and probe-tissue interaction, frame alignment generally requires a non-linear transformation whose accuracy is difficult to quantify: flexible transformation models can fit intensity features and noise, so appearance-based metrics such as Normalized Cross-Correlation (NCC) can improve without a genuine gain in geometric accuracy. We therefore establish a dataset of 132 frame pairs across fourteen pCLE sequences from 4 patients with manually annotated landmark correspondences, so that Target Registration Error (TRE) can serve as a geometrically grounded complement to NCC. We assess the effect of progressively increasing the transformation model’s degrees of freedom, from translation to Thin Plate Spline (TPS), and of six feature-matching backends spanning classical (Shi-Tomasi, Lucas-Kanade) and learned (SuperPoint, SuperGlue, LightGlue, LoFTR, RoMa) approaches. Translation and rigid models prove insufficient under tissue deformation, while TPS with random sampling achieves the strongest landmark-derived alignment of the evaluated configurations; among the learned matchers, used without fine-tuning, only RoMa offers a robust, if modest, advantage over other methods. At the sequence level, pairwise registration quality proved an unreliable predictor of final mosaic quality, so mosaic quality must be evaluated directly rather than inferred from pairwise metrics.
[CV-37] Incentive Noise and Structural Prior Infusion for Multi-modal Object Re-Identification ECCV2026
链接: https://arxiv.org/abs/2609.24539
作者: Weixiang Zhou,Yuhao Wang,Xingguo Xu,Weizhen Zhou,Zhixun Su,Jinshan Pan,Cong Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by ECCV 2026. The version of record may differ slightly
Abstract:Multi-modal object Re-Identification (ReID) benefits from complementary information across heterogeneous imaging modalities. To further enrich semantic representation, text descriptions have recently been incorporated as an additional modality. However, recent vision-language approaches often treat text descriptions as clean, deterministic signals and overlook their inherent noise, including modality-mismatched phrases and semantically ambiguous expressions. Moreover, prevailing methods lack explicit mechanisms to reconcile fine-grained structural discrepancies between modalities, even after high-level semantic alignment. To address these challenges, we propose a novel framework centered on Positive-Incentive Noise (\pi-noise) and structured prompt modulation. First, the Semantic Cross-Modal Modulator harnesses task-aware \pi-noise, sampled from a distribution conditioned on both visual and text inputs, to perturb global tokens and enable semantics-guided cross-modal compensation. Second, the Structure-Aware Prompt Adapter injects learnable geometric priors via prompts to enhance spatial consistency. Third, the Context-Aware Sparse Fusion module distills structural context to guide adaptive fusion while shielding identity features from noisy local details. Experiments on three multi-modal ReID benchmarks demonstrate the effectiveness and robustness of our approach. The code is available at this https URL.
[CV-38] MIRAG E: Full-Body Bystander Privacy for Smart Glasses with Consent-Based Restoration
链接: https://arxiv.org/abs/2609.24537
作者: Muhammad Umair,Muhammad Danial Maqbool,Fatima Arshad Cheema,Kapal Dev,Muhammad Hamad Alizai,Muhammad Ali Siddiqi,Naveed Anwar Bhatti
类目: Computer Vision and Pattern Recognition (cs.CV); Cryptography and Security (cs.CR)
备注:
Abstract:Video recording on smart glasses exposes more than faces. Continuous capture reveals full-body biometric signatures, including gait, posture, and silhouette, that enable person re-identification (ReID) even after conventional face sanitization. We present MIRAGE, a three-tier architecture for privacy-preserving smart glasses that enforces full-body privacy, supports synthetic full-body replacement, and retains encrypted recovery material for consent-based restoration. We implement MIRAGE on a Raspberry Pi~5 (a CPU-only proxy for smart-glasses compute), companion phones, and a cloud generative backend. Compared to prior systems, MIRAGE achieves 0.948 AP and 0.976 AR while accurately detecting the complete visible body. Its bounding box masking reduces learned silhouette-based ReID to essentially random guessing, with 10.86% Rank-1 accuracy compared with an 11.12% measured chance level. Even against an adaptive adversary retrained on MIRAGE’s sanitized pose signals, Rank-1 gait identification drops from 90.25% to 26.20%, removing 72.5% of the adversary’s identification advantage. Subjects: Computer Vision and Pattern Recognition (cs.CV); Cryptography and Security (cs.CR) Cite as: arXiv:2609.24537 [cs.CV] (or arXiv:2609.24537v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2609.24537 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-39] Dynamic Thermal Gaussians: Multimodal 4D Gaussian Splatting
链接: https://arxiv.org/abs/2609.24531
作者: Rongfeng Lu,Lifeng Lin,Xiaobao Wei,Quan Chen,Ming Lu,Yitian Xue,Yaoqi Sun,Yuhan Gao,Anke Xue,Chenggang Yan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Thermography plays a vital role in military and broader thermal analysis applications. Recent progress in 3D thermal reconstruction has extended temperature analysis from 2D to 3D space, yet most existing works assume static temperature distributions, neglecting the temporal dynamics of heat transfer in real-world environments. To address this limitation, we propose the first dynamic RGB-Thermal reconstruction framework for complex scenes. Our method jointly models RGB appearance, thermal observations, and scene geometry as they change over time. Specifically, we introduce a multimodal dynamic scene representation that anchors both the color and thermal modalities to a shared geometric substrate, ensuring their consistency under spatiotemporal deformations. We further design multimodal embeddings to enhance the motion expressiveness for each modality, and propose a multimodal routing mechanism that retains a unified set of shared multimodal Gaussians as the geometric backbone while adaptively spawning modality-specific Gaussians to strengthen the representational capacity in detail-rich regions of each individual modality. In addition, we contribute a novel benchmark dataset featuring high-frequency temperature variations to facilitate the evaluation of 4D reconstruction. Extensive experiments demonstrate that our method achieves high-fidelity spatiotemporal reconstruction of both appearance and temperature. Our code and dataset are available at: this https URL.
[CV-40] ME-VLM:A Unified VLM for Embodied Cognition and Agent Coordination
链接: https://arxiv.org/abs/2609.24526
作者: Foundation Model,Li Auto Inc
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Physical AI requires models to ground visual and linguistic understanding in real-world environments while accounting for environmental constraints and execution feedback. We introduce MachEmbodied-VLM (ME-VLM), a unified vision-language model with two variants, 4B and 35B-A3B, that brings together embodied cognition and multimodal agent capabilities. Our work emphasizes physical perception and spatiotemporal reasoning, together with planning, interaction, and outcome assessment in both digital and physical environments. We construct training data spanning embodied and multimodal agent tasks, including execution observations and feedback to support outcome assessment and decision refinement. The training pipeline comprises embodied capability injection, separate reinforcement learning of embodied and multimodal-agent experts, and multi-teacher on-policy distillation that consolidates their complementary capabilities into a single model. Experiments show competitive performance on both embodied and agent benchmarks, as well as on autonomous-driving and embodied-navigation tasks. For edge deployment, visual token compression, W4A8 quantization, and hardware–software co-optimization enable on-device inference of the 4B variant on the M100, reducing prefill latency from 400 ms to 188 ms. Project Page: this https URL Code Repository: this https URL Subjects: Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2609.24526 [cs.CV] (or arXiv:2609.24526v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2609.24526 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-41] Preoperative Prediction of Microvascular Invasion in Hepatocellular Carcinoma by Integrating Multimodal Ultrasound and Clinical Data: A Multicenter Study
链接: https://arxiv.org/abs/2609.24524
作者: Jun Cheng,Yuanyuan Kong,Qing Huang,Xiaotong Tan,Licong Dong,Yulong Han,Wufeng Xue,Ruobing Huang,Dong Ni,Qi Yang,Jie Yu,Ping Liang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Main manuscript: 37 pages, 5 figures, and 4 tables; supplemental material: 18 pages, 6 figure, and 9 tables
Abstract:Background: Microvascular invasion (MVI) predicts recurrence and survival in hepatocellular carcinoma (HCC) but requires postoperative histopathology for diagnosis. We developed and validated a model integrating multimodal ultrasound and clinical data for preoperative MVI prediction. Methods: This multicenter study included 489 patients with HCC from eight centers. All patients had B-mode ultrasound (BUS), color Doppler flow imaging (CDFI), dynamic contrast-enhanced ultrasound (DCE-US), and clinical information. Data from seven centers (n = 421) were used for model development with five-fold cross-validation; data from the remaining center (n = 68) formed an independent external validation cohort. The proposed multimodal information fusion network used modality-specific encoders, a hemodynamic temporal change module for bidirectional DCE-US perfusion changes, and a representation consistency learning module to align heterogeneous ultrasound representations before Transformer-based fusion. Results: In external validation, DCE-US achieved the highest single-modality area under the receiver operating characteristic curve (AUC; 0.8545+/-0.0198), versus clinical information (0.6715+/-0.0156), CDFI (0.6435+/-0.0344), and BUS (0.6087+/-0.0417). Pixel-difference sampling and the proposed temporal module outperformed alternative sampling and video representation methods. The full model achieved the best performance, with an AUC of 0.8953+/-0.0180, accuracy of 81.18%+/-2.83%, sensitivity of 86.40%+/-6.69%, and specificity of 78.14%+/-6.28. Conclusions: Integrating multimodal ultrasound and clinical information enabled promising preoperative MVI prediction in HCC. DCE-US was the main source of predictive information, while BUS, CDFI, and clinical information provided complementary value. The proposed framework may support preoperative risk stratification and individualized clinical decision-making.
[CV-42] Not All Task Vectors Need Equal Rank: Energy-Proportional Allocation for Model Merging
链接: https://arxiv.org/abs/2609.24517
作者: Hyunjoong Cho,Jinhyeok Jang
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Model merging aims to combine multiple fine-tuned models derived from a common pretrained model into a single multi-task model without additional joint training. Recent spectral merging methods improve over simple weight averaging by exploiting low-rank structures of task-specific updates, but they commonly assign the same rank capacity to every task. This uniform allocation ignores that task vectors can have heterogeneous spectral complexity, causing the shared merging space to be used suboptimally. In this paper, we propose Spectral Energy-proportional Rank Allocation (SERA), a simple task-adaptive strategy that allocates ranks according to the singular-value energy structure of each task vector. By assigning richer spectral capacity to complex or isolated tasks and fewer directions to compact tasks, SERA extends SVD-based model merging from uniform-capacity merging to task-dependent capacity allocation. Experiments under standard vision model merging protocols show that SERA improves multi-task merging performance while preserving the same total rank budget as existing spectral merging methods. Further analysis demonstrates that task-level spectral concentration is closely related to the per-task effect of adaptive rank allocation, providing insight into when and why SERA is effective.
[CV-43] 0.5%100%: Bidirectional Reciprocal Learning for Referring Image Segmentation
链接: https://arxiv.org/abs/2609.24510
作者: Xiaoqiang Lu,Licheng Jiao,Lingling Li,Yuting Yang,Long Sun,Wenping Ma,Xu Liu,Fang Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 16 pages, 8 figures
Abstract:Recent advances in vision foundation models (VFMs) have shown remarkable capabilities across diverse unimodal visual tasks. However, adapting VFMs to referring image segmentation (RIS) typically necessitates precise vision-language alignment via full fine-tuning, incurring substantial computational overhead and risking catastrophic forgetting. While existing parameter-efficient fine-tuning (PEFT) methods enable safe knowledge transfer with minimal training costs, they predominantly operate independently within individual modalities or focus exclusively on unidirectional guidance from language to vision, overlooking progressive cross-modal interaction and visual feedback for textual refinement. To address these limitations, we propose Bidirectional Reciprocal Learning (BRL), a novel adapter-based PEFT framework that facilitates hierarchical, bidirectional information flow within both token-mixing and channel-mixing layers of frozen foundation models. Specifically, BRL introduces two complementary lightweight modules. The Reciprocal Attention Adapter (RAA) performs cross-modal query-key exchanges at the token level, enabling visual and linguistic tokens to mutually attend to each other for fine-grained spatial grounding. The Reciprocal Gate Adapter (RGA) generates cross-modal gating signals at the channel level, allowing global semantic context from one modality to adaptively recalibrate channel activations of the other. Extensive experiments on RefCOCO, RefCOCO+, and RefCOCOg benchmarks demonstrate the superiority of BRL over prior RIS methods, achieving state-of-the-art performance while requiring less than 0.5% backbone parameter updates. Code and models will be released at this https URL.
[CV-44] CMAMBADEPTH: Self-supervised Monocular Depth Estimation with Channel Mamba and Hybrid Attention
链接: https://arxiv.org/abs/2609.24494
作者: Xuezhi Xiang,Jiayao Liu,Heqi Xiang,Yuqi Hu,Yiming Chen,Shanjun Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Accurate monocular depth estimation serves as a core enabler for single camera scene understanding. However, existing self-supervised monocular depth estimation methods generally suffer from the bottleneck of inefficient cross-scale information interaction and difficulty in balancing local and global spatial modeling. In this paper, we propose CMambaDepth, a self-supervised framework that achieves efficient multi-scale feature fusion and fine-grained contextual modeling via channel-wise selective state propagation. Specifically, Bidirectional Channel Mamba (Bi-CMamba) aligns encoder features across scales and enables bidirectional information exchange among ordered scale groups. Unidirectional Channel Mamba (Uni-CMamba) progressively aggregates decoder features and retains fine-grained scale groups through a group selection mechanism for subsequent fusion. Furthermore, a Hybrid Attention Module (HAM) is introduced to combine large-kernel local context and Manhattan self-attention for complementary spatial modeling. Experimental results demonstrate that our method achieves highly competitive performance. Specifically, our model achieves an AbsRel of 0.094 and an RMSE of 4.156 on KITTI, and an AbsRel of 0.140 on DDAD. In the zero-shot cross-dataset generalization test on NYUv2, it attains an AbsRel of 0.232, outperforming the baseline RA-Depth by 7.2%.
[CV-45] Identity-Consistent Analysis of Long-Shot Windsurfing Video: A Domain-Specific Offline Tracking System
链接: https://arxiv.org/abs/2609.24492
作者: Bertil Braun
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages, 4 figures, 2 tables. Project code and evaluation artifacts: this https URL
Abstract:Long-shot windsurfing video combines small targets, large camera pans, prolonged overlaps, and rapidly changing backgrounds. The desired output is not a generic MOT trace but a separate, stable rider-relative video for each surfer; one false identity merge can invalidate an otherwise useful result. We present an offline analysis system that detects surfers, forms conservative local tracklets, links them globally with camera-compensated motion and a foreground-masked sail-color descriptor, and uses two pose keypoints on the rig to drive a rider-relative virtual camera. The tracking stage is evaluated on 21 manually reconstructed development videos containing 41,004 retained observations. On this fixed-observation protocol, the production system achieves 0.957 pairwise precision, 0.918 recall, and 0.937 F1, compared with 0.792 F1 for OC-SORT and 0.828 for BoT-SORT. Compared with OC-SORT, it reduces fragmentation excess from 845 to 42, but nine of its 95 output tracks mix rider identities and these errors affect seven of the 21 videos.
[CV-46] AgentS TAR: Agent ic Shape Tracking and Reconstruction from Monocular Videos
链接: https://arxiv.org/abs/2609.24487
作者: Kirill Mazur,Nikita Karaev,Matthew Chang,Jitendra Malik,Nur Muhammad "Mahi’’ Shafiullah
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:In this work, we present a method for shape reconstruction and tracking from video via agentic analysis-by-synthesis. Unlike prior methods which first estimate dense pixel correspondences and then recover object motion from them, our method infers a structured 3D object model, including its geometry and kinematic structure, and uses this model to optimise object track estimates over time. In our optimisation loop, a Vision-Language Model (VLM) agent iteratively refines shape or generalised pose through a render-and-compare loop, combining coarse visual reasoning with numerical pose optimisation for precise state estimation. This structured formulation enables our method to track through large motion, articulation, and severe occlusion without relying on pixel-matching objectives. Quantitatively, on ARCTIC, our method substantially outperforms state-of-the-art 3D point-tracking baselines for articulated objects, and on HOT3D it outperforms all evaluated rigid-object tracking baselines.
[CV-47] VPRune: Efficient Training-free Pre-LLM Visual Token Pruning
链接: https://arxiv.org/abs/2609.24485
作者: Guangchuan Lv,Dianxing Shi,Dingjie FU
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Visual token pruning is a promising approach to reducing the inference cost of large vision-language models (LVLMs), yet aggressive token reduction often causes substantial performance degradation. We identify three key factors behind this degradation: text-guided selection bias, information loss from discarded tokens, and positional distortion caused by sequence compaction. Based on these observations, we propose \textbfVPRune, a training-free pre-LLM pruning framework consisting of visual-only diversity selection, similarity-guided token recycling, and position-preserving restoration. Experiments on FastVLM-1.5B across multiple vision-language benchmarks demonstrate that VPRune achieves a favorable accuracy–compression trade-off, with particularly pronounced advantages under aggressive compression. Furthermore, evaluations on edge-device show that VPRune effectively reduces end-to-end inference latency while maintaining superior task performance, demonstrating its practicality for resource-constrained LVLM deployment.
[CV-48] STA-TFM: Spatio-Temporal Aggregation Across Views TransForMer for Pose Estimation
链接: https://arxiv.org/abs/2609.24482
作者: Mena Kamel,Natalie Won,Amrut Sarangi,Sven Jager,Albert Pla Planas
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Monocular 3D human pose estimation (HPE) remains challenging due to depth ambiguity, occlu- sions, and the need for temporal consistency. While multi-view methods provide superior accuracy over monocular approaches, they often require complex setups. We introduce STA-TFM, a transformer-based architecture that combines spatial and temporal information for multi-view pose estimation. The approach leverages DSTformer, a monocular feature extractor, to capture long-range pose dependencies within each view. A fusion transformer then aggregates information across views to produce coherent 3D estimates. To address training data scarcity, we use a data generation pipeline that transforms any existing 3D pose dataset into multi-view setups with controllable parameters. Experiments on various datasets demonstrate that STA-TFM outperforms existing camera-parameter-free multi-view methods. STA-TFM achieves 50.9% and 49.5% reductions in mean per joint position error (MPJPE) and mean per joint velocity error (MPJVE) on the DHP19 dataset. Furthermore, it achieves 6.7% and 7.7% respective reductions on HAA4D, and a 15.2% MPJPE reduction on TotalCapture. STA-TFM handles noisy and missing 2D inputs, supporting potential deployment in healthcare monitoring, athletic assessment, and immersive technologies. Code, training checkpoints, and data are available at this https URL.
[CV-49] Spatial Action Review: A Visual Analytics Dashboard for Auditing Language-to-Action Hand-offs in Electron Microscopy IEEE-VIS2026
链接: https://arxiv.org/abs/2609.24470
作者: Samia Mohinta,Albert Cardona
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at the IEEE VIS 2026 Workshop on Visual Analytics in the Age of Autonomous Science (VAxAutoSci)
Abstract:Multimodal large language models (MLLMs) are increasingly explored as interfaces for scientific image analysis, where a visual question-answering (VQA) response may be paired with a spatial output that guides a downstream stage. A supervisor reads the language answer, while a downstream workflow such as segmentation or region review consumes the point-set output. We call this transition from inspecting the answer to relying on its point action the language-to-action hand-off. A silent failure occurs when the answer is correct while the paired action misses annotated objects needed downstream, so answer-based oversight clears a region whose action is unreliable. We introduce Spatial Action Review, a visual analytics dashboard for auditing this failure mode in electron microscopy (EM) mitochondria analysis. It links paired answer-action records through an answer-action ledger, a task-by-dataset risk map, and an image-region audit view, connecting aggregate patterns to image evidence while an adjustable action-reliability gate supports re-audit. The review ends in a human-AI hand-off, where a supervisor records whether the action is accepted, escalated, held under a stricter gate, or flagged for model revision. Across 541 image regions from an EM-adapted Qwen3-VL case-study run, point actions fail the gate in 54.4% of records with a correct VQA response, and 27.4% of all records are silent failures. A correct answer is associated with only a 5.8-percentage-point higher probability of a reliable action, with a bootstrap interval spanning zero; the point-biserial correlation between answer correctness and object coverage is 0.061. This weak coupling persists across five model conditions on 753 matched image regions. Spatial Action Review makes answer-action mismatches visible and ties them to image evidence and a recorded decision before MLLM outputs enter autonomous scientific workflows.
[CV-50] MIGA:Shared-Geometry Gaussian Representation with Implicit Amplitude Modeling for Accelerated 3D Multi-Echo MRI
链接: https://arxiv.org/abs/2609.24468
作者: Jingran Xu,Yuanyuan Liu,Yanjie Zhu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Three-dimensional multi-echo MRI provides rich anatomical and quantitative information, but repeated volumetric encoding prolongs acquisition and motivates k-space undersampling. Reconstructing undersampled multi-echo data requires exploiting shared anatomy while preserving echo-dependent signal variation; full-volume modeling also introduces substantial computational and memory demands. We propose MIGA, a scan-specific framework comprising shared anisotropic Gaussian geometry, a coordinate-conditioned multi-output amplitude network, and explicit echo-specific phase variables. The Gaussian geometry provides common spatial support across echoes, the implicit network models spatially structured amplitude variations, and the phase variables retain echo-specific complex signal information. All components are jointly optimized using only the acquired multi-coil k-space, requiring no fully sampled training data. Experiments showed that MIGA consistently outperformed the comparison methods across imaging tasks and acceleration factors, with larger improvements under stronger undersampling. MIGA also achieved a favorable quality-cost balance among the evaluated full-volume multi-echo methods. These results support the effectiveness of combining shared Gaussian geometry with implicit echo-dependent amplitude modeling for accelerated 3D multi-echo MRI reconstruction.
[CV-51] MECAIL: Communication-Aware Incremental Learning for Object Detection with 14.6 KB Spatiotemporal Experts ITSC2026
链接: https://arxiv.org/abs/2609.24455
作者: Matthias Neuwirth-Trapp,Maarten Bieshaar,Danda Paudel,Konrad Schindler,Luc Van Gool,Christos Sakaridis
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Accepted at ITSC 2026
Abstract:Intelligent transportation systems require Incremental Learning (IL) to continually improve their overall performance in dynamic environments. However, most edge devices lack the computational resources to support on-device IL, requiring updates to be transmitted from centralized servers. We propose using this setup to obtain dense, specialized module coverage that adapts a fixed base model to specific spatiotemporal contexts, such as parking lots, gas stations, ferries, or construction sites. However, in order to reliably transmit these modules to the edge device, using TCP, UDP, and BTP over V2X, Wi-Fi, and 2G-5G hardware, we establish a strict limit of 14.6 KB per module to fit within the first TCP window and to minimize UDP/BTP fragmentation. We further introduce Mixture-of-Experts for Communication-Aware Incremental Learning (MECAIL), the first method that meets this strict requirement, in which each new domain or environment is served by a small expert network that adapts the base model. We validate MECAIL on D-RICO and ODinW-13, where it largely matches the performance of parameter-heavy approaches while enabling practical, bandwidth-efficient large-scale deployment. This allows comprehensive coverage by experts for highly specific, focused, and temporary situations.
[CV-52] Do LiDAR Language Models Really Understand Spatio-temporal Relationships?
链接: https://arxiv.org/abs/2609.24452
作者: Runyi Yang,Murat Akkoyun,Di Wen,Ruiping Liu,Yufan Chen,Junwei Zheng,Xiaoye Wang,Kailun Yang,Danda Pani Paudel,Luc Van Gool,Kunyu Peng
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注:
Abstract:Recent 4D LiDAR language models aim to reason about objects and their evolving spatial relationships. Yet, in our evaluation, always selecting the same option nearly matches the multiple-choice accuracy of two B4DL-derived configurations. We introduce LiDAR-Hallu, a geometry-referenced benchmark and diagnostic protocol with 10,000 questions across 150 nuScenes scenes. It covers object existence, ego-relative position, distance ordering, relative motion, and temporal localization, with explicit rules for selecting objects, comparing times, and determining reference answers. Our protocol combines fixed-answer and candidate-content controls, cross-scene pairs with identical prompts but opposite reference answers, and relation-specific recall. Analysis of 100,000 recorded responses reveals failures hidden by aggregate accuracy. Candidate duration alone makes temporal answers predictable without observing LiDAR. On paired questions, the models frequently give the same answer to scenes requiring opposite answers. Relation-specific analysis further shows that both configurations miss every positive lateral-motion case across all tested conditions. Temporal-shuffle contrastive decoding provides little net improvement, as repairs are largely offset by new errors and the main failures persist. These results show that evaluating spatio-temporal reasoning requires testing whether models distinguish the queried physical relationships, rather than relying on individual-answer accuracy alone. The source code, checkpoints, and data are released at this https URL.
[CV-53] Estimating Accurate Hand Pose in Camera Space with Vision Transformer
链接: https://arxiv.org/abs/2609.24424
作者: Kaiwen Ren,Yiran Jiang,Yongjing Ye,Shihong Xia
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Graphics (cs.GR)
备注:
Abstract:Monocular RGB-based hand pose estimation has emerged as a critical research frontier in computer vision. The local hand pose estimation methods predict hand poses relative to the wrist, while global hand pose estimation also requires estimating the wrist’s position in the camera coordinate system. However, this camera-space estimation confronts two fundamental challenges: (1) depth ambiguity in monocular settings, and (2) the coupling effect of hand local poses and global wrist positions in the perspective projections. In particular, this coupling reflects that the projections are jointly determined by local hand poses, wrist positions, and camera intrinsics. To overcome these challenges, our framework proposes two key innovations: Transformation-Isomorphism Supervision for hand-depth information extraction and Perspective Information Embedding for resolving above coupling effect of local pose and wrist position, both integrated within the mainstream encoder-decoder architecture. Besides, we propose a novel framerate-aware multi-dataset training strategy for sequential pose refinement. Our fully integrated approach achieves at most 37.1% superiority in CS-MJE over SOTA on HO3D. Project page: this https URL.
[CV-54] DeCo: Efficient Decouple-to-Couple Learning for Multi-Task Visual Grounding
链接: https://arxiv.org/abs/2609.24409
作者: Xiaoqiang Lu,Licheng Jiao,Long Sun,Yuting Yang,Xu Liu,Lingling Li,Wenping Ma,Fang Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 34 pages,9 figures
Abstract:Multi-task visual grounding requires models to jointly understand linguistic semantics and perform accurate visual localization and segmentation. Despite the success of multimodal large language models, effectively adapting them to multiple grounding objectives remains challenging. Existing methods commonly enforce task cooperation through shared representations, while overlooking the intrinsic conflict between task-oriented feature interests. In this paper, we introduce \textbfDeCo , an efficient \textbfDe couple-to- \textbfCo uple learning framework that resolves this dilemma through a two-stage paradigm: task-specific representation decoupling followed by complementary prior coupling. Specifically, we first propose Task-aware Semantic Decoupling (TSD) to route shared visual cues into individual features under salient word-level guidance, alleviating representation interference between localization and segmentation. Furthermore, we observe that segmentation naturally provides informative localization priors due to dense supervision. Based on this insight, we introduce Hybrid Prior Coupling (HPC), which integrates sentence-level semantic prior with mask-derived spatial prior for enhanced grounding. Built upon a frozen multimodal encoder, DeCo requires lightweight trainable parameters while achieving strong generalization across multiple grounding objectives. Extensive experiments on RefCOCO/+, G-Ref, ReferIt, Flickr, DIOR-RSVG, SARVG1.0, RRSIS-D, RIS-LAD, and RefDIOR demonstrate that DeCo achieves state-of-the-art performance on both natural and remote sensing benchmarks. The code and models are available at this https URL.
[CV-55] Can Spiking Neural Networks play pinball? A neuromorphic motion detector for target tracking
链接: https://arxiv.org/abs/2609.24403
作者: Mazdak Fatahi,Šárka Pryjmaková,Pierre Boulet,Giulia D’Angelo
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Biological visual systems achieve continuous, low-latency motion perception by processing sparse, asynchronous spiking signals, enabling real-time tracking under strict energy constraints. Event-based cameras, inspired by the mammalian retina, replicate this efficiency by capturing only local brightness changes as asynchronous events, offering a natural substrate for spiking neural networks (SNNs) to parallelise computation and adapt to fast-changing scenes. Pinball provides a controlled yet dynamic testbed, requiring precise motion estimation and fast reaction to a small, rapidly moving target. This work presents a fully spiking, real-time perception-to-action pipeline for closed-loop pinball gameplay. A dynamic vision sensor observes a small, fast-moving ball, and a network of spiking Time-Difference Encoders on the SpiNNaker neuromorphic platform jointly estimates its position, speed, and direction. The system is characterised across receptive field size, accumulation window, and angular tuning width for real-time operation, and benchmarked in closed loop against human players across two flipper regimes of increasing physical realism. It achieves a hit rate of 56.1%, nearly double the human average, reacting within 21.7 ms (5 ms network latency) and consuming an estimated 148 \muW using fewer than 25k neurons, among the fastest and most energy-efficient event-based closed-loop demonstrators benchmarked. Under more realistic flipper dynamics, tuning a single interpretable policy parameter reproduces the full spectrum of human play styles, from cautious to aggressive, with no change to the perception pipeline. A physical demonstrator, tracking a real ball and actuating real flippers in closed loop, confirms the principle operates beyond simulation. Its fully spiking, learning-free design offers a compact, energy-efficient example of real-time neuromorphic perception-to-action. Subjects: Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2609.24403 [cs.CV] (or arXiv:2609.24403v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2609.24403 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-56] A Lightweight Convolutional Neural Network for Real-Time Recognition of Hand-Drawn Geometric Shapes
链接: https://arxiv.org/abs/2609.24384
作者: Shahir Abdullah
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Recognizing hand-drawn geometric shapes is a foundational sub-problem of sketch recognition, with applications in education, human-computer interaction, and diagram digitization. This paper presents the design, implementation, and evaluation of a desktop application that recognizes four basic hand-drawn geometric shapes, circle, square, rectangle, and triangle using a compact Convolutional Neural Network (CNN). A dataset of 2,000 labeled 28x28-pixel shape images was collected independently and released publicly. The classifier consists of three convolutional blocks (16, 32, and 64 filters) with max-pooling, an in-model data-augmentation stage (random horizontal flip, rotation, and zoom), a dropout-regularized dense layer of 128 units, and a 4-way linear output layer, totaling 97,956 trainable parameters. The network is trained with the Adam optimizer on a sparse categorical cross-entropy objective computed directly on logits. On an 80/20 train-validation split, the model achieves 94.80% training accuracy and 96.01% validation accuracy with a validation loss of 0.1437. A Tkinter-based graphical interface allows a user to draw a shape with the mouse and receive an immediate class prediction with a confidence score. We situate this system within the broader sketch and shape-recognition literature, compare its accuracy against related hand-drawn shape classification studies, and discuss the limitations inherent to a small, single-contributor dataset. The complete source code, trained model, and per-class datasets are released publicly to support reproducibility.
[CV-57] opographic Training Concentrates Causal Circuits Without Improving Neuron Monosemanticity ICML2026
链接: https://arxiv.org/abs/2609.24379
作者: Gautam Ranka,Shubham Santosh Pandere,Aiden Dsouza
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: Accepted at the Mechanistic Interpretability Workshop at ICML 2026
Abstract:Mechanistic interpretability of vision transformers seeks to decompose model computation into human-readable units, but learned representations entangle many concepts in each neuron. Feature superposition is widely treated as the central obstacle to this decomposition, yet most mitigations (sparse autoencoders, dictionary learning) are post-hoc and leave the underlying network unchanged. We ask whether a spatial-locality training loss (TopoLoss) can act as a lightweight, training-time prior that improves interpretability of standard mech-interp tools. Training ViT on ImageNet-100 across multiple TopoLoss weights \alpha , we measure causal sufficiency of topographic clusters via activation patching and feature geometry via sparse autoencoders fit to the same residual stream. At \alpha=1.0 , topographic clusters are 2.79 \times more causally sufficient than random unit sets of the same size, with the effect increasing monotonically in \alpha . SAE L0 sparsity decreases by 11% and dead-feature fraction rises 19-fold, yet standard neuron-level monosemanticity scores are unchanged, indicating that topographic pressure acts at circuit level, concentrating causal mass into spatially local structures without disentangling individual neurons. This dissociation suggests current neuron-level monosemanticity metrics are insensitive to a class of real interpretability gains, and positions cheap architectural priors as a viable training-time complement to post-hoc tooling.
[CV-58] Prescriptive SVD-Inspired Attention via Spectral Energy Retention
链接: https://arxiv.org/abs/2609.24370
作者: Vasileios Arampatzakis,Vasileios Sevetlidis,George Pavlidis
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注: Published in Transactions on Machine Learning Research (TMLR), 2026
Abstract:Self-attention is central to modern Transformer architectures, but its dense dot-product formulation makes it difficult to identify which internal directions are structurally important and which can be modified without disrupting the model. SVD-Inspired Attention (SVDA) addresses part of this problem by introducing a learned diagonal spectrum into the query-key score interaction, making latent attention directions explicitly inspectable through indicators such as spectral entropy, effective rank, sparsity, alignment, selectivity, and perturbation response. This paper examines the transition from diagnostic interpretation to operational intervention. A diagnosis–intervention–verification framework is proposed, and one intervention is evaluated: spectral energy retention in the attention-score pathway. Across FashionMNIST, CIFAR-10, CIFAR-100, and Food-101, the \rho=0.90 prescription removes 24.5–53.7% of score directions, reduces parameters by 2.6–4.3%, and reduces estimated MACs by 2.8–5.4%. The paired mean accuracy change of the dimension-reduced model ranges from -0.03 to +0.05 percentage points over three seeds. These results support SVDA as an intrinsically interpretable attention mechanism whose learned spectrum exposes an operational coordinate system for deterministic and verifiable modification of attention-score formation.
[CV-59] ReViS: Temporal Repetition Structure Aware Video Synthesis for Self-supervised Repetitive Action Counting
链接: https://arxiv.org/abs/2609.24367
作者: Fanqi Yu,Shengming Ma,Stefano Fiorini,Vito Paolo Pastore,Xuan Qi,Vittorio Murino,Cigdem Beyan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted for publication in Image and Vision Computing (Elsevier). This is the author-accepted manuscript and not the final published version of record. The DOI and link to the published version will be added when available
Abstract:Fully supervised repetitive action counting (RAC) has achieved strong performance, but requires dense temporal annotations that are costly and difficult to scale. We propose TReViS, a self-supervised video synthesis framework that enables training RAC models without any repetition labels. TReViS estimates the underlying temporal repetition structure of an unlabeled video via a Temporal Self-Similarity Matrix, infers its cycle statistics, and synthesizes new training sequences that preserve realistic repetition patterns while introducing controlled temporal variability. These synthesized videos are paired with pseudo-labels and used to train existing RAC architectures from scratch. Across multiple datasets and backbones, TReViS consistently outperforms prior self-supervised methods and achieves performance competitive with several supervised baselines, while remaining fully label-free, demonstrating the effectiveness of structure-aware video synthesis for label-free RAC. The source code is available at this https URL.
[CV-60] Dissecting Agent ic Forensics: The Role of Triage Prompting and Evidence Arbitration in Open-World Fake Image Detection ECCV2026
链接: https://arxiv.org/abs/2609.24359
作者: Xianlong Li(2),Pietro Bongini(1),Niccoló Pancino(1),Marco Blanchini(2),Benedetta Tondi(1),Mauro Barni(1) ((1) University of Siena, Italy, (2) IMT School for Advanced Studies Lucca, Italy)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
备注: Accepted at the 2026 Workshop on AI for Multimedia Forensics and Disinformation Detection (AI4MFDD), ECCV 2026. 34 pages (17 main paper incl. references, 17 appendix), 7 figures, 5 tables
Abstract:Image forensics is increasingly an open-world problem: manipulations range from fully synthetic images to localized edits, splicing and swapping, while most forensic detectors remain specialized to a single manipulation family. Agentic AI has recently emerged as a promising solution. In principle, such systems can assess the reliability of individual detectors, identify out-of-scope evidence, and arbitrate conflicting reports. However, it remains unclear which components actually drive performance and whether their benefits persist under distribution shift. To answer these questions, we study a training-free agentic framework built around specialist detectors, per-detector triage, and conflict-aware evidence arbitration. Using six configurations and three multimodal large language model backbones, we dissect the role of triage, prompting, and reasoning quality on both in-distribution and out-of-distribution data. Our results show that naive detector fusion suffers from severe false-positive rates on authentic images. Triage and prompting consistently improve performance by filtering unreliable evidence and exposing detector limitations. However, the dominant factor is represented by reasoning itself: A stronger judge substantially outperforms a weaker one, particularly under distribution shift. Most notably, manipulation recall is nearly saturated across all configurations, indicating that the main challenge of open-world image forensics is not detecting manipulations, but calibrating trust in specialized forensic tools and arbitrating conflicting evidence.
[CV-61] LIBERO-VPro: Benchmarking Closed-Loop Visual Robustness of Robotic Foundation Models
链接: https://arxiv.org/abs/2609.24350
作者: Huiqiong Li,Zhiting Mei,Anirudha Majumdar,Jingjing Chen,Yu-Gang Jiang,Bin Zhu
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: Project: this https URL
Abstract:Robotic foundation models achieve impressive performance on standard manipulation benchmarks, yet these evaluations typically assume clean, timely, and consistent visual observations throughout execution. We introduce LIBERO-VPro, a benchmark for systematically evaluating the closed-loop visual robustness of robotic foundation models by perturbing the visual evidence available during execution. LIBERO-VPro covers four complementary dimensions, including Visual Evidence Degradation, Camera Staleness, Visual Source Consistency, and Task-Relevant Scene Variation, spanning 12 challenge categories, 96 experimental settings, and 3,296 task-condition cases. We evaluate three vision-language-action models and three world-action models over approximately 196,000 simulated episodes, complemented by 200 real-world rollouts on a Franka Research 3. Our results reveal that strong nominal performance can mask substantial weaknesses in visual grounding and adaptation. Models often remain successful despite severe object-level occlusion, yet degrade sharply when local interaction cues are disrupted or familiar spatial priors are violated. They are also highly sensitive to stale or missing observations and struggle when changed task preconditions require behavioral adaptation. Finally, VLAs and WAMs exhibit distinct robustness profiles, showing that visual robustness is multi-dimensional and architecture-dependent. LIBERO-VPro provides a systematic diagnostic framework for developing robotic foundation models that can more reliably ground and adapt their actions under challenging visual conditions.
[CV-62] LiAuto-MindViT: A Hybrid Vision Backbone with Adaptive Bidirectional Mamba
链接: https://arxiv.org/abs/2609.24337
作者: Lifu Mu,Shuai Chen,Wen Zheng,Haoyi Sun,Xueyang Fu,Pengfei Yu,Ning Mao,Tao Wei,Zhou Pan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:While Mamba-based models have shown strong potential for long sequence modeling, adapting them to vision is challenging due to the requirement of local neighborhood correlations and multi-directional spatial contexts for visual understanding. In this paper, we present LiAuto-MindViT, a novel hybrid vision backbone that synergizes the strengths of CNNs, Mamba, and Transformers. The core of our design is the Adaptive Bidirectional Mamba (ABM), which eliminates the directional bias of unidirectional SSMs through bidirectional selective scanning with learnable alpha blending, enabling content-adaptive directional fusion without the overhead of exhaustive multi-path routing. To further accelerate inference, we propose a deployment-friendly Reparameterized ConvSE (RepConvSE) module that leverages structural reparameterization to reduce latency and memory access overhead. Extensive experiments demonstrate that LiAuto-MindViT achieves state-of-the-art performance on image classification, object detection, and semantic segmentation while enabling efficient inference through reparameterization.
[CV-63] AlignMorph: Tuning-Free Diffusion Image Morphing via Explicit Semantic Transport
链接: https://arxiv.org/abs/2609.24330
作者: Wuyi Liu,Xu Han,Yuren Chen,Yige Mao,Zishuo Peng,Xianzhi Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Image morphing aims to produce a smooth and semantically consistent transition between two input images. Existing diffusion-based morphing methods either require expensive per-pair optimization or rely on implicit spatial alignment, which easily fails under large layout discrepancies. To address these limitations, we propose AlignMorph, a novel tuning-free diffusion framework guided by the principle of transport-then-denoise. We explicitly decouple geometric alignment from generative denoising to avoid structural entanglement. Our framework consists of two core components. (1) Global Semantic Transport, which achieves diffusion-compatible semantic alignment via entropic optimal transport and reliability-aware latent warping; and (2) Coordinate-Aligned Generation, which uses a symmetric bi-phase attention handoff to maintain consistent spatial coordinates throughout denoising. Without any tuning, AlignMorph effectively eliminates ghosting and achieves superior structural coherence and temporal smoothness on morphing benchmarks. Code is available at this https URL.
[CV-64] NeuIDO: Neural Intrinsic Dynamics Operator for Physics-Informed 4D World Models ECCV2026
链接: https://arxiv.org/abs/2609.24313
作者: Jiajing Lin,Xin Zhang,Jianhua Sun
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by ECCV 2026; Project Page: this https URL
Abstract:World models aim to capture environmental dynamics and predict future trajectories, showing growing potential for embodied intelligence. Physics-informed 4D generation integrates physical simulation to predict 3D object interactions, offering a promising pathway toward world models. However, this paradigm relies on manually imposed dynamical assumptions rather than internalizing world dynamics, and thus still leaves a gap toward a true world model. To bridge this gap, we propose NeuIDO, a novel world dynamics modeling framework that learns a unified intrinsic dynamics representation from visual observations, advancing physics-informed 4D generation toward a world model. Specifically, we formulate world modeling as a neural operator learning problem and introduce a two-stage training strategy to learn a generalizable mapping from the visual observation distribution to the intrinsic dynamics distribution. Building on this observation-dynamics mapping, NeuIDO enables zero-shot dynamics inference directly from videos and can be further aligned with complex real-world dynamics via few-shot adaptation. Extensive experiments demonstrate that NeuIDO effectively unifies the intrinsic dynamics underlying diverse visual observations into a shared representation and rapidly infers dynamics in novel scenes.
[CV-65] AnalogDepth: Multi-view Geometry from FPV drones under Analog Video Transmission
链接: https://arxiv.org/abs/2609.24312
作者: André Amorim,Pedro F. Proença
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注:
Abstract:Analog video transmission (VTX) remains widespread in FPV drones due to low latency, weight and low cost. However analog VTX suffers from complex spatially structured image degradation which differ fundamentally from digital image corruption (e.g. AWGN) used in standard training augmentation. This work shows that this type of noise severely degrades the accuracy of Depth Anything 3 (DA3), a state-of-the-art feed forward visual geometry foundation model. To address this gap, we present AnalogDepth, a parameter-efficient training pipeline that adapts DA3 to analog FPV imagery using student-teacher knowledge distillation with Low-Rank Adaptation (LoRA) injected into the DINOv2 backbone. Rather than synthesizing noise analytically, we build a noise bank from static FPV recordings under diverse conditions and compare real-noise injection against PSD-matched Gaussian synthesis and AWGN as baselines. Experiments on six real FPV flight sequences across three indoor scenes show that training with our noise bank consistently reduces per-frame depth RMSE and 3D reconstruction Chamfer distance compared to the pretrained DA3 baseline and both Gaussian noise variants. These results demonstrate that replicating the spatial structure of real analog transmission noise is critical for effective adaptation. Subjects: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO) Cite as: arXiv:2609.24312 [cs.CV] (or arXiv:2609.24312v1 [cs.CV] for this version) https://doi.org/10.48550/arXiv.2609.24312 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[CV-66] HappyWorld-Bench
链接: https://arxiv.org/abs/2609.24308
作者: Zhiqi Bai,Junai Cai,Yixin Chen,Jingrun Du,Tao Feng,Wei Gong,Siyuan Huang,Xiao Lin,Jiaheng Liu,Jun Luo,Yongzhe Lyu,Liya Ma,Zenan Meng,Lin Qu,Wenbo Su,Jiaming Wang,Qinghe Wang,Shaofei Wang,Yanghai Wang,Zequn Wang,Ziming Wang,Hu Wei,Jiangtao Wu,Ruiqi Wu,Jiaxin Xie,Yuchi Xu,Ze Xu,Chengting Yu,Liangyu Yuan,Gang Zeng,Yawen Zeng,Xingyao Zhang,Zizheng Zhang,Bo Zheng,Jiancheng Zhu,Song-Chun Zhu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Evaluating world models requires assessing both the quality of the worlds they generate and their consistency and responsiveness under exploration, interaction, and modification. We introduce HappyWorld-Bench, a comprehensive benchmark that evaluates whether generated worlds remain reliable as agents interact with them. Our design is built on a hierarchical capability framework of six world capabilities (W1-W6), from generative construction to unified world modeling, instantiated across three independent evaluation tracks: video world models, spatial world models, and embodied world models. HappyWorld-Bench comprises 1,138 video prompts, 300 spatial scenes, and 254 embodied test cases. Across all three tracks, we build and operate HappyWorld-Arena to organize human A/B comparisons and derive model-level Elo ratings, which complement newly designed automated metrics that capture behavioral correctness. We evaluate 14 video world models, 9 spatial systems, and 8 embodied candidates under this unified framework. Results reveal remaining reliability gaps across all three tracks: video models exhibit reduced consistency during extended rollouts and revisits, spatial models achieve at best 70.14% placement accuracy and 73.33% edit execution, and embodied models struggle to preserve state across multi-step actions and respond precisely to altered action conditions and physical rules. These findings highlight the need to evaluate world models not only by visual quality, but also by state consistency and the correctness of their responses to actions and interventions.
[CV-67] Classifier-Free Guidance in Flow Matching: Non-Autonomous Potentials Overshoot and Posterior-Mean Control
链接: https://arxiv.org/abs/2609.24287
作者: Jishen Peng,Zheng Ma
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Classifier-free guidance (CFG) improves conditional generation in Flow Matching, but strong guidance can distort the generated distribution and reduce diversity. We provide a geometric account of this behavior by viewing Flow Matching as a time-varying gradient flow and characterizing how CFG reshapes its underlying potential. This view explains how stronger alignment can be accompanied by mean displacement and trajectory concentration, and motivates controlling guidance through the model-implied terminal posterior mean. We therefore propose Posterior-Mean-Capped CFG (PMC-CFG), a training-free, per-sample method that adaptively retains the strongest feasible guidance without additional network evaluations. Experiments on synthetic and large-scale image-generation benchmarks show that PMC-CFG limits guidance-induced distortion and concentration while improving the alignment–diversity trade-off, with particularly strong benefits when nominal guidance is large.
[CV-68] Hierarchical Prompt Learning for Hyperbolic Vision-Language Models
链接: https://arxiv.org/abs/2609.24276
作者: Andro Erdelez,Pascal Mettes,Behzad Bozorgtabar
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Hyperbolic vision-language models (VLMs) represent image and text features in a geometry naturally suited to hierarchy, but their adaptation to downstream tasks has largely relied on fixed prompts. Existing prompt learning methods, meanwhile, treat class labels as a flat set and do not exploit available taxonomic structure. We address this gap with a hierarchical prompt learning plug-in for frozen hyperbolic VLMs. Given a fixed offline parent-class hierarchy, it augments a class prompt learner with a separate parent prompt learner, parent-level supervision, hyperbolic entailment regularization, and parent-feedback logit fusion. We instantiate the method with CoOp, CoCoOp and MaPLe, yielding HyPLO, CoHyPLO and MaHyPLO. Across the standard 11-dataset benchmark, all variants improve base-to-new generalization and cross-dataset transfer, and remain comparable to their prompt learning baselines under domain shift. Six hierarchical metrics and embedding analyses show that the method produces more taxonomically consistent predictions and induces a hierarchy-consistent organization of parent, class, and image embeddings in hyperbolic space. Its gains are largest when novel classes must be placed within a fixed taxonomy, and smallest for fine-grained confusions among sibling classes or shifts affecting only the image distribution.
[CV-69] Unsupervised Brain Anomaly Detection as a Bayesian Inverse Problem with Diffusion Prior
链接: https://arxiv.org/abs/2609.24265
作者: Hugues Roy,Reuben Dorent,Ninon Burgos
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Unsupervised anomaly detection (UAD) aims to localize abnormal regions in medical scans without pixel-level annotations. A typical strategy seeks to reconstruct a pseudo-healthy image that preserves subject-specific anatomy. Recently, diffusion models have been proposed to perform UAD. However, these methods rely on heuristic noise schedules or synthetic corruptions to balance subject-specificity and anomaly removal. In this work, we propose an alternative formulation of UAD as a Bayesian inverse problem under a diffusion prior. First, we introduce a latent spatial anomaly mask that models pixel-wise consistency between a test image and its latent corresponding pseudo-healthy image. Then, we propose an approximation of the unknown generation process that links healthy anatomy, anomalies, and the observed image, enabling a well-defined likelihood within the Bayesian framework. Building on recent advances in diffusion-based inverse problem methods, we jointly infer the pseudo-healthy image and the anomaly mask via annealed posterior sampling. We evaluate our approach on FDG PET (ADNI) and FLAIR MRI (BraTS 2021), demonstrating improved anomaly localization performance compared to other diffusion-based approaches and validating the contribution of our introduced model. Our code is available at this https URL.
[CV-70] OpenFlyScan: A Quality-Guided Aerial Reconstruction System for Consumer Drones
链接: https://arxiv.org/abs/2609.24253
作者: Zhongrui You,Zhen Li,Junli Liu,Zhigang Wang,Bin Zhao
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:3D Gaussian Splatting (3DGS) provides high-fidelity scenes for large-scale embodied simulation, but constructing large-scale urban assets remains constrained by expensive equipment and delayed quality feedback. Preset surveys can leave complex surfaces insufficiently observed, with defects discovered only after reconstruction, requiring return visits and repeated processing. We present OpenFlyScan, a quality-guided aerial reconstruction system for consumer drones that integrates a GS quality model, a reacquisition planner, and a custom-designed mobile app. The model learns from GS rendering errors to predict regional reconstruction quality. Based on these predictions, the planner then generates complementary reacquisition strips to be executed through the app, which also supports automated oblique surveys and data transfer without additional hardware on board. Across real aerial scenes, the model effectively identifies regions that are likely to be poorly reconstructed. In the Expo West field experiment, targeted reacquisition improves PSNR at additional views by 10.95 dB. With consumer drones, OpenFlyScan integrates capture, targeted reacquisition, and reconstruction to support rapid, low-cost urban asset creation. Code and models will be made publicly available at this https URL.
[CV-71] Reinforcement Learning Inspired Black-box Adversarial Attacks for Computer Vision
链接: https://arxiv.org/abs/2609.24249
作者: Florian Krone,Elena Hoemann,Sven Hallerbach
类目: Machine Learning (cs.LG); Cryptography and Security (cs.CR); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Neural networks, both convolution or transformer based, are essential for modern computer vision systems. However, they are vulnerable to small perturbations, almost imperceptible to humans, which significantly alter the model’s prediction. These adversarial attacks are often considered to be a significant threat to the implementation of neural networks in safety-critical applications. Most attacks utilize the white-box threat model and therefore require full access to the target model, making them unrealistic to use in practice. We propose a novel approach under the more realistic black-box threat model that utilizes concepts from reinforcement learning to optimize perturbations with a non-differentiable target model. Reinforcement learning algorithms have already been optimized to be query efficient, making them an ideal starting point when designing black-box adversarial attacks. We show the success of our reinforcement learning inspired black-box adversarial attack (RIBA) in generating adversarial perturbations using only a small number of queries to the target model, by comparing it to state of the art attacks on different models on the Cifar10 and ImageNet data sets. RIBA takes 25.4% fewer median queries to generate attacked images against a ResNet-18 on Cifar10 and 22.5% fewer median queries to fool a Vit-B/16 model on ImageNet. Additionally, we demonstrate that RIBA can match the performance of white-box attacks on an adversarially trained model.
[CV-72] Look Where It Counts: A Free Label-Free Visual Evidence Signal for Fine-Grained Vision-Language Reasoning
链接: https://arxiv.org/abs/2609.24244
作者: Santi Ram Tiwari,Nihal Naik,Devbrat Pandey,Nishant Sinha
类目: Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV)
备注: 6 pages, 4 figures, 4 tables
Abstract:Multimodal large language models (MLLMs) fail at fine-grained visual questions less because they cannot reason than because they never see the evidence: high-resolution images are downsampled before encoding, so the model answers from linguistic priors. The standard remedies are expensive: annotated answers (SFT), hand-engineered verifiers (RLVR), or a large external teacher (on-policy distillation). We ask whether the visual evidence itself can supply the signal for free. We formalize the contrastive evidence gap, the per-token log-likelihood ratio that a model assigns to its own output when conditioned on a question-relevant region versus an irrelevant one, and study it across Qwen2.5-VL-7B, Qwen3-VL-8B, and Qwen3-VL-30B-A3B on V*Bench. Our main positive result is training-free: selecting the candidate crop under which the model’s answer distribution is most peaked, using a single-view, label-free criterion, discovers the answer-bearing region with no bounding boxes, training, or labels. It localizes the target 4.4 to 5.1 times better than chance and raises fine-grained accuracy from 70 percent to 85 percent at inference. We further show that the gap is complementary to the model’s own confidence. Combining them predicts correctness better than either alone, with AUC up to 0.99, and flags confidently wrong answers, with AUC ranging from 0.97 to 1.00 within the high-confidence subset. All effects concentrate on perception-bottleneck questions and vanish on a global-context control. Finally, we report an honest negative result: converting the same signal into a training method, gated self-distillation (SEG-Distill), does not outperform the base model at pilot scale across three gate designs, while more aggressive gating degrades accuracy. The signal is real, but converting it into training gains remains an open problem.
[CV-73] IMPLICIT-Bench: Measuring Implicit Bias in Text-to-Image Models under Neutral Prompts
链接: https://arxiv.org/abs/2609.24228
作者: Yue Dai,Ziyang Liu,Marc Cheong,Caren Han
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Text-to-image (T2I) models are typically evaluated for bias using slot-based templates such as ``a photo of a [profession]‘’. Such templates probe only \emphexplicit demographic attributes (e.g., gender, skin tone) in isolation. They overlook a broader \emphimplicit bias that arises in natural prompts: when stereotype-relevant attributes are left unspecified, models still default to stereotypical outputs. We introduce IMPLICIT-Bench, a benchmark for measuring implicit bias in T2I models under such prompts. The key design is a structured-knowledge-graph (KG) construction of controlled prompt triplets: neutral, stereotype, and anti-stereotype variants that differ only along a single bias dimension while preserving scene semantics. This enables precise attribution of bias effects that template benchmarks cannot achieve. IMPLICIT-Bench comprises 5,493 prompts across 11 bias categories, validated through multi-model agreement, CLIP-based verification, and human evaluation. Using this benchmark, we show that state-of-the-art T2I models exhibit systematic bias under neutral prompts, a failure mode largely invisible to existing evaluations. We then use IMPLICIT-Bench to evaluate debiasing methods, uncovering a fundamental trade-off between bias reduction and semantic fidelity.
[CV-74] SRPR-Net: Semantic and Relational Prompt Refinement for Automated SAM-based Instance Segmentation
链接: https://arxiv.org/abs/2609.24226
作者: Lufei Liu,Guojie Li,Suncheng Xiang,Fan Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Instance segmentation is a fundamental computer vision task with diverse real-world applications. Recently, prompt-driven foundation models have shown promising generalization. However, automated prompting remains limited by insufficient semantic guidance and inter-instance modeling. To address this challenge, we propose a novel architecture, named Semantic Relational Prompt Refinement Network (SRPR-Net), for automated SAM-based instance segmentation. A sequential prompt refinement mechanism is introduced to enrich detector geometry with visual-language semantics and then incorporate same-image instance dependencies, enabling context-aware box adjustment before SAM segmentation. Experiments on multiple standard benchmarks demonstrate that SRPR-Net achieves consistent improvements in segmentation performance over existing state-of-the-art approaches. The code is publicly available at this https URL.
[CV-75] DiaSeg: Diagonal Segment Extraction from DTW Paths for Interpretable Gait Analysis
链接: https://arxiv.org/abs/2609.24223
作者: Tresor Y. Koffi,Amel Hidouri,Corentin Legrand,Aurélie Bertaux
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted for publication in Data Mining and Knowledge Discovery (Springer), September 2026. 28 pages, 4 figures
Abstract:Dynamic Time Warping (DTW) is the dominant approach for measuring similarity between time series, yet standard practice discards the optimal warping path after computing a single distance value, losing local alignment information most relevant to clinical diagnosis. We introduce DiaSeg, a framework that extracts diagonal segments from DTW paths with controlled breaks, characterizing each segment by five geometric features (effective length, interruption count, cost variation, temporal position, and path context), and enabling unsupervised pattern discovery without domain-specific feature engineering. Validated on 91 subjects across six clinical conditions (healthy aging, Parkinson’s, Huntington’s, ALS, brain tumor, and stroke), three findings emerge. First, diagonal segments form consistent unsupervised patterns (silhouette 0.33) aligned with biomechanical phase annotations, with label-based validation confirming near-perfect separation of healthy and pathological gait (ARI up to 0.986). Second, segments discriminate pathology at 69% (supervised) and 75% (patient-level clustering), with pathology manifesting through distributional shifts in segment length; combining segment and cycle-level features further improves classification to 91.7%. Third, while cycle-based methods achieve higher accuracy (91%), diagonal segments provide phase-specific interpretability unavailable in global representations, localizing where coordination breaks down within the gait cycle. DiaSeg thus transforms DTW from a black-box distance into a source of interpretable temporal features for neurodegenerative disease assessment.
[CV-76] Document Retrieval-Aware Chunking (D-RAC): Universal Retrieval-Aware Ingestion of Enterprise Documents via PDF Normalization and Multimodal Markdown Conversion
链接: https://arxiv.org/abs/2609.24220
作者: Uday Allu,Abhivanth Sivaprakash,Pratik Singh,Aman Manocha(AI Research Team a href=“http://Yellow.ai” rel=“external noopener nofollow” class="link-external link-http"this http URL/a)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 14 pages, 2 figures, 10 tables
Abstract:Retrieval-Augmented Generation (RAG) systems over enterprise knowledge bases must ingest heterogeneous document formats – PDFs, Word documents, presentations, and scans – whose content is locked inside complex visual layouts, multi-column pages, and dense tables. Rule-based extraction and OCR destroy reading order, flatten tables, and lose heading hierarchy, while fully agentic chunking over extracted text incurs high token costs and hallucination risk. We present Document Retrieval-Aware Chunking (D-RAC), an extension of our Web Retrieval-Aware Chunking (W-RAC) framework to arbitrary document formats. D-RAC first normalizes any input document into PDF, exploiting the fact that virtually every format has a faithful, deterministic PDF rendering. A single multimodal LLM pass then converts rendered pages into retrieval-optimized Markdown – rewriting tables as self-contained prose statements and preserving heading hierarchy – after which chunking proceeds exactly as in W-RAC: deterministic parsing into ID-addressable units followed by lightweight LLM-based chunk planning over identifiers rather than text. Source text is never regenerated during chunking, preserving W-RAC’s cost, determinism, and observability benefits while unlocking every renderable format as a first-class input. On the 236-document, 795-page PDF subset of the RAG-Multi-Corpus benchmark spanning five enterprise domains, D-RAC converts and chunks the entire corpus in 72 minutes with zero errors, producing 1,748 retrieval-ready chunks. Compared to agentic chunking with frontier LLMs, D-RAC reduces chunking-stage output tokens by 95.7%, cutting chunking cost by 77.8% (GPT-4.1 pricing) to 85.6% (Gemini 2.5 Pro pricing) and chunking time by 75%. D-RAC scales linearly to documents of 500+ pages.
[CV-77] Beyond Emotion Prompts: Fine-Grained Text-to-Image Generation Driven by Valence-Arousal-Dominance
链接: https://arxiv.org/abs/2609.24215
作者: Minglang Li,Yueyue Fang,Xieping Gao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 9 figures, 4 tables
Abstract:Although text-to-image models can accurately depict subjects and scenes, creators still struggle to specify the fine-grained emotions an image should convey without rewriting its content description. Natural language can suggest emotions, but it offers no control scale with stable meanings and ordered intensities. We propose EMOTRANS, which transforms psychologically grounded valence-arousal-dominance (VAD) coordinates into generation conditions that are independent of the content text and modulated across denoising stages, making emotional style a finely adjustable creative variable. To support this goal, we construct EMOVAD, an art-painting dataset that pairs objective content descriptions with separately collected emotional ratings from multiple annotators. We also coordinate emotional expression and content preservation through dual-branch training with a shared model. Objective and human evaluations show that the framework improves the accuracy of three-dimensional emotion control and produces perceptible, orderable continuous changes while maintaining competitive text alignment and image quality. This work provides a practical emotion-driven approach to image generation that extends objective content depiction to fine-grained emotional adjustment.
[CV-78] ChartJudgeBench: Evaluating LMM Judges for Chart-to-Code Generation
链接: https://arxiv.org/abs/2609.24210
作者: Lijian Wu,Henry Hengyuan Zhao,Zijian Zhang,Jiahao Tang,Jiajun Wu,Alex Jinpeng Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Building strong chart-to-code systems increasingly relies on reinforcement learning, whose effectiveness depends critically on the quality of the reward signal. Large Multimodal Models (LMMs) play a natural critical role in jointly assessing chart visual appearance and task requirements. They are therefore increasingly used as visual critics and reward models, yet their reliability as judges remains largely unexplored. To this end, we introduce ChartJudgeBench, a diagnostic vision-language benchmark for assessing LMM judges in chart-to-code workflows. It includes 1,003 Chart Perception Alignment (CPA) instances for pairwise chart comparison and 650 Chart Reasoning Judgment (CRJ) instances for binary Accept/Reject verification in Chart Reproduction and Chart Editing. Together, these tasks emulate the core judging decisions required in agentic refinement and RL-based chart optimization. Our evaluation of strong LMMs reveals four systematic limitations: (i) positional bias in pairwise comparison, (ii) a strong tendency to overpredict Accept, (iii) difficulty in matching visual styles and aesthetics, and (iv) an unexpected leniency bias in RL-trained models. These findings show that current LMM judges require explicit reliability validation before being used as critics or reward models in chart-to-code optimization. The code and data are available on ChartJudgeBench.
[CV-79] CoaG: Cylinders on a Grid: Coarse 3D Layout Control for Video Generation
链接: https://arxiv.org/abs/2609.24208
作者: Zhangsihao Yang,Mengyi Shan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Project page with videos: this https URL
Abstract:We ask how little geometry a person has to draw to control both where people stand and where the camera moves in a generated video. Our answer is a ground plane and one cylinder per person. A user draws a grid on the ground, places one cylinder where each person should stand, moves the cylinders and the camera over 81 frames, and the model renders a photoreal video in which the people occupy the cylinders’ positions, move as the cylinders move, and are seen from the drawn camera. Appearance comes from a text prompt and a background reference image; layout and motion come from the geometry. Because no dataset pairs such a signal with video, we build the pairs ourselves: an automatic engine writes 2000 captions from a combinatorial seed, generates a clip for each with a text-to-video model, and lifts every clip back to its geometry with person tracking, background inpainting, an agentic ground-mask loop, feed-forward multi-view reconstruction and a plane fit, with no real footage and no manual labels. A LoRA on Wan2.2-Fun-Control trained on 1935 such tuples follows drawn layouts and camera paths on hold-out clips: the generated people match the cylinders’ count, order, position and height, the text changes who they are, the reference image changes where they are, and dolly-in, orbit, pan and crane paths are followed, dolly-out only weakly.
[CV-80] SAFe: Segment-guided Aggregation of Feature Densities for Anomaly-aware Segmentation
链接: https://arxiv.org/abs/2609.24204
作者: Anja Delić,Jurica Runtas,Marin Oršić,Ivan Marković,Ivan Petrović
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Visual segmentation systems encounter objects outside their training distribution during real-world deployment, hindering reliable autonomous systems that depend on scene parsing in the perception stage. Many recent methods address this by using self-supervised foundation models to train density estimators that yield low likelihood in anomalous image regions. Although promising, these methods suffer from poor feature semantics or they lack spatial consistency, both of which undermine critical downstream decisions. We address this problem with~\method, a generative method based on class-conditional density estimation over self-supervised representations. SAFe trains lightweight normalizing flows that produce class-conditional normalized likelihood estimates over frozen DINOv3 features. We combine density estimates from transformer features with density scores over multi-scale convolutional features to capture both global semantics and local detail. We introduce a method-agnostic post-processing step based on SAM3 that connects per-location likelihoods into spatially coherent segments while suppressing false positives, and enables instance-level anomaly detection without retraining. The post processing further distinguishes novel categories among anomalous objects by a similarity-based agglomerative clustering scheme. SAFe sets a new state of the art on the PANIC, OoDIS, SMIYC ObstacleTrack with strong performance on the ISSU benchmark.
[CV-81] SKstars at SHROOM: Visions Agreement-Guided Ensembling of Zero-Shot and LoRA-Adapted Vision–Language Models EMNLP2026
链接: https://arxiv.org/abs/2609.24198
作者: Ali Athar,Imran Ahsan,Joon-Yong Jung
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: This paper has been accepted at the SHROOM-Visions 2026 Shared Task, co-located with EMNLP2026
Abstract:This paper describes the SKstars submission to SHROOM-Visions 2026, a shared task on fine-grained hallucination detection in large vision-language model outputs. The task requires systems to identify hallucinated character spans, assign hallucination categories, and provide confidence estimates for their predictions. Our approach combines zero-shot predictions from Qwen2.5-VL-72B-Instruct with those of a LoRA-adapted Qwen2.5-VL-7B-Instruct model. The outputs of the two models are integrated through a lightweight ensemble procedure, followed by span refinement and confidence adjustment. We evaluate the main system components on a small internal development subset and report the performance of the submitted system on the official English test set. SKstars achieved a Cor+Lbl score of 0.2902, ranking 15th among 29 teams, and obtained Cor and IoU scores of 0.3642 and 0.3151, respectively, ranking 18th on both metrics. The results show that combining a large zero-shot model with a smaller adapted model provides a practical framework for multilingual and fine-grained hallucination localization, while also highlighting the difficulty of transferring development-set improvements to hidden test data. Code and predictions: this https URL
[CV-82] Lightweight Pedestrian Head-Orientation Recognition Network for Safe Pedestrian-Vehicle Interaction
链接: https://arxiv.org/abs/2609.24193
作者: Yuanzhe Li,Yidi Huang,Xiaotong Chang,Hounian Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Pedestrian head orientation recognition plays an important role in autonomous driving by providing valuable cues for understanding pedestrian attention and anticipating potential crossing behavior. However, reliable recognition in real-world traffic scenes remains challenging because pedestrian head regions are often captured at low resolution. To address this challenge, we propose a lightweight Low-Resolution Head Orientation Convolutional Neural Network (LRHO-CNN) for pedestrian head orientation recognition. We construct a new dataset by extracting pedestrian head images from multiple public datasets and manually annotating them into eight orientation categories. The collected images are systematically preprocessed and augmented to increase data diversity and better represent variations in illumination and image quality. The experimental analysis compares LRHO-CNN with three fine-tuned baseline models, namely ResNet-18, ResNet-34, and VGG-16. The results demonstrate that LRHO-CNN achieves the highest classification accuracy among the evaluated models. LRHO-CNN is further evaluated on the JAAD and PIE datasets, demonstrating its effectiveness in recognizing pedestrian head orientation in real-world traffic scenes and providing informative head-orientation cues that can support downstream pedestrian behavior and intention prediction.
[CV-83] Benchmarking Off-the-Shelf Multimodal AI Models Against Dermatologists on Patient-Captured Skin Images
链接: https://arxiv.org/abs/2609.24190
作者: Rian Dolphin,Laura Knowles
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 14 pages, 6 figures
Abstract:Artificial intelligence (AI) has advanced at a rapid pace in recent years. Initially, breakthroughs in large language models caught widespread attention. However, recent generations of frontier AI models have adopted multimodal capabilities as a first class citizen, with vision capabilities being central to that. In this paper, we evaluate three recently released models on the task of diagnosing dermatological conditions from patient-submitted images. The models chosen are at the low to mid tier in terms of pricing and thus represent a floor on current AI capabilities, not a ceiling. We evaluate AI performance relative to a panel of three certified dermatologists, who grade each image, and we present four interesting findings. Firstly, depending on the metric, the tested AI models are either on par or slightly trail humans in terms of inter-clinician agreement. Secondly, we find that asking AI models for a confidence rating produces poorly calibrated answers, meaning use of confidence thresholds should not be relied upon in a clinical setting. Thirdly, the effect of providing additional patient metadata is strongly model-specific, with one of the three models degrading on every metric considered. Finally, model cost is not predictive of performance. The best-performing model we tested costs on average 0.0045 per case.
[CV-84] StenoVLA-3D: 3D-Aware Reasoning VLA for Navigation Through Gastrointestinal Stenoses ICRA2027
链接: https://arxiv.org/abs/2609.24187
作者: Tamima Tabassum,Yiming Huang,Tianchun Wu,Changjing Liu,Zhiqing Tang,Chikit Ng,Beilei Cui,Liangjing Shao,Jiewen Lai,Hongliang Ren
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages, 4 figures. Submitted to ICRA 2027
Abstract:Autonomous endoscopic navigation requires the policy model to predict actions from texture-poor monocular observations, make safe control decisions, and retain evidence of lesions after they leave the field of view. Existing vision-language-action (VLA) models primarily rely on visual appearance and short-term context, limiting geometric grounding and episode-level reporting. We introduce StenoVLA-3D, a 3D-aware VLA framework for navigating through stenotic regions. We integrate point-maps into the Cosmos-Reason 2 backbone through learned geometry-gated fusion, and also propose a temporal state branch to model traversal progress. Our reasoning-and-action backbone predicts grounded reasoning with actions, while dedicated heads estimate stenosis shape and generate the final lesion report. We further introduce EndoCausal, an episode-level dataset with lesion annotations, actions, and temporally grounded reasoning. On 40 held-out recorded test episodes, StenoVLA-3D reaches 95.2% semantic accuracy and 83.4% action accuracy. On the physical 3-DoF endoscope, it attains 88.9% and 77.8% task success in esophageal and colonic phantoms (36 trials each), substantially outperforming the evaluated baselines.
[CV-85] LegendBench: A Diagnostic Benchmark for Legend Understanding with Counterfactual Interventions
链接: https://arxiv.org/abs/2609.24172
作者: Xinnuo Zhang,Zhike Tang,Jing Xu,Haoyuan Zhao,Weikai Yang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Legends are fundamental to chart understanding, as reliable interpretation requires correctly binding legend entries to corresponding visual marks. While vision-language models (VLMs) are increasingly applied to chart understanding, their legend understanding is poorly diagnosed by aggregate accuracy, which can be satisfied by superficial shortcuts and confound legend-specific errors with other reasoning failures. To enable fine-grained diagnosis and controlled testing, we introduce LegendBench, a parametric benchmark and generation pipeline that produces targeted legend-centric test cases. LegendBench contributes (1) a capability-task taxonomy spanning legend parsing, legend grounding, legend-conditioned reasoning, and legend-aware abstention to localize failures, and (2) counterfactual group generation, where each base chart yields multiple variants under controlled legend interventions to probe model invariance and sensitivity. Using LegendBench, we evaluate both general-purpose VLMs and specialized chart models and generate their capability profiles, revealing persistent bottlenecks in reliable legend-to-mark binding and counterfactual consistency. We then use these capability profiles to guide targeted fine-tuning, demonstrating that bottleneck-specific interventions can effectively close the localized capability gaps and generalize to unseen data. We further leverage our counterfactual design to conduct fine-grained diagnostic experiments, analyzing encoding-channel effects, legend-order shortcuts, and abstention under varying visibility.
[CV-86] An Unexpected Robot Policy: Early Evaluations of GPT -6 Astra on RoboDojo and Beyond
链接: https://arxiv.org/abs/2609.24170
作者: Wenbo Zhang,Kaixuan Wang,Yutao Ouyang,Xiaoyu Huang,Liyang Li,Kailun Su,Weiyang Jin,Wenhao Chai,Haotian Liang,Zhiyang Dou,Yue Chen,Tianxing Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 24 pages
Abstract:Embodied AI systems are often organized into System 1 and System 2. System 1 is typically a pretrained policy that generates actions at high frequency, whereas System 2 is often instantiated as a vision-enabled language model for high-level planning. We ask whether a large language model (LLM) can act as the policy for robot manipulation without task-specific finetuning. We call this setting LLM as policy. We evaluate three LLMs on all 42 RoboDojo tasks and compare their scores with 40 public policies. Astra and GPT-5.5 use the official 50-episode-per-task protocol; DeepSeek-Flash uses 10 episodes per task. GPT-6 Astra achieves 22.48% average success rate and 28.97 Score over 2,100 trials, ranking above every public entry. Yet GPT-5.5 and DeepSeek-Flash reach only 0.88% and 1.92% average success rate with the same post-processing. We find that Astra exhibits a sharply polarized capability profile. It generalizes well to tasks that require semantic understanding but not high-precision control. In contrast, it performs poorly on tasks that require precision, dynamic control, or complex bimanual coordination. In-context experiments show no aggregate benefit from one-shot demonstrations, while selected interaction traces show within-episode corrections under perturbations. Overall, the evaluated LLMs vary substantially in manipulation performance. Astra stands out and provides initial evidence for the potential of a general-purpose manipulation model, although reliable precision and dynamic control remain limitations in the evaluated setting.
[CV-87] Relightable 3D Avatar Reconstruction with Semantic-Adaptive Motion-Illumination Responses
链接: https://arxiv.org/abs/2609.24158
作者: Jiankuo Zhao,Xiangyu Zhu,Jijie Li,Baiqin Wang,Shukai Chen,Zhen Lei
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Reconstructing expressive and relightable 3D head avatars from monocular videos remains challenging in computer vision, as it requires accurate modeling of both non-rigid facial motion and illumination-dependent appearance. Existing Gaussian avatar methods commonly rely on globally coupled representations, in which Gaussian primitives share a unified motion or illumination response model. Such uniform modeling neglects the distinct motion patterns and material/reflectance properties of different facial semantic regions, thereby limiting fine-grained animation accuracy and reducing relighting plausibility. To address this limitation, we propose SAMIRA, a 3D Gaussian avatar framework for semantic-adaptive motion-illumination response modeling. For motion response modeling, the Semantic-Adaptive Motion Response module rasterizes current-to-reference mesh displacements into a topology-consistent UV space and leverages facial semantics to route displacement features through semantic-specific modulators, predicting localized Gaussian geometric residuals beyond coarse mesh binding. For illumination response modeling, the Semantic-Adaptive Illumination Response module learns compact diffuse and specular response factors for each facial region, allowing Gaussians in different regions to adapt their illumination responses to novel environment lighting. These response factors are incorporated into deferred physically based shading, providing a lightweight approximation of semantic-dependent illumination effects. Extensive experiments on self-reenactment, cross-reenactment, and relighting demonstrate that SAMIRA improves both fine-grained expression reconstruction and relighting realism over existing methods.
[CV-88] STAR: Scene- and Task-Aware 4D Radar Preprocessing Towards End-to-End Cognitive Radar
链接: https://arxiv.org/abs/2609.24151
作者: Seung-Hyun Song,Dong-Hee Paek,Seung-Hyun Kong
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Four-dimensional (4D) Radar has emerged as a key sensor for environmental perception, providing range, azimuth, elevation, and Doppler measurements while remaining robust to illumination changes and adverse weather conditions. However, conventional Radar preprocessing methods, such as constant false alarm rate (CFAR) detection, select measurements primarily based on signal-level criteria and may therefore discard information valuable for downstream perception during point cloud generation. In addition, existing 4D Radar perception pipelines typically optimize Radar data processing and downstream perception independently, preventing task objectives from directly guiding the preprocessing stage. To address these limitations, we propose a Scene- and Task-Aware Radar (STAR) Preprocessor together with an end-to-end training framework. The STAR Preprocessor incorporates scene context and downstream task objectives to generate task-relevant Radar points, enabling the Radar representation to be optimized directly for perception. On the K-Radar benchmark, the proposed method achieves 74.3 AP, outperforming the previous state of the art by 5.6 AP points. Furthermore, applying the task-relevant points generated by STAR to various existing 3D detectors improves detection performance in most evaluation settings and yields an overall positive average gain over point clouds produced by conventional preprocessing.
[CV-89] he Visual Target Matters: Learning across the Visual Hierarchy for Brain-to-Image Retrieval
链接: https://arxiv.org/abs/2609.24136
作者: Ye Wang,HaoKun Ren,Hong Yu,Ruirui Li,Xiao Li,Ke Liu,Wei Wu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Brain-to-image retrieval seeks to identify the visual stimulus that elicited a non-invasive neural response. Candidate images are typically represented by pretrained vision models, whose internal representations vary in abstraction across depth. Existing methods usually train the neural encoder to recover a fixed final-layer visual target. Under this formulation, the visual hierarchy is reduced to a single prescribed endpoint, preventing representations at other depths from directly shaping the visual target. This limitation motivates learning how information across visual depths should contribute to the retrieval target. To this end, we introduce NeuroGlyph, which learns a trial-independent visual target from multiple depths of a frozen visual backbone. NeuroGlyph decomposes the target into factor-specific subspaces. Each subspace learns an image-conditioned allocation over visual depth. The resulting subspaces are fused into a single embedding for retrieval. Across THINGS-EEG and THINGS-MEG, NeuroGlyph outperforms final-layer supervision in all controlled comparisons. It also surpasses the post hoc best fixed-layer oracle in three of four comparisons. Parameter-matched ablations support both factorized target construction and image-conditioned depth allocation. Under comparable 200-way retrieval protocols, NeuroGlyph achieves the strongest system-level performance in six of eight reported metrics. These results support learning retrieval targets across the visual hierarchy rather than prescribing one visual depth.
[CV-90] Action-Slot: Structured Action-Centric Representation Learning for Multi-Agent Atomic Activity Understanding
链接: https://arxiv.org/abs/2609.24127
作者: Yu-Ho Chang,Chi-Hsi Kung,Yi-Hsuan Tsai,Yi-Ting Chen
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 17 pages, 7 figures
Abstract:Atomic activity understanding aims to recognize and localize structured traffic behaviors that jointly encode motion patterns and their grounding in road topology. Unlike conventional action recognition, atomic activities are multi-agent, multi-label, and topology-aware: multiple activities co-occur while many agents remain inactive. We introduce Action-Slot, a structured action-centric representation learning framework. Slot attention is widely used for object-centric decomposition, but its permutation-invariant design and object-level inductive bias are misaligned with atomic activity semantics. We reformulate slot learning as structured activity decomposition through three designs: (1) category-aligned action slots that anchor slots to predefined activity categories, (2) parallel spatio-temporal slot updating for holistic video-level reasoning, and (3) background and negative-slot regularization that enforces competition between foreground activities and irrelevant regions. Together these establish an activity-centric inductive bias that disentangles concurrent and asynchronous activities directly from raw video. Beyond recognition, the learned representations encode transferable spatio-temporal grounding signals. We further propose an attention-difference-based pseudo mask selection framework that suppresses false positives by measuring attention changes before and after candidate region removal, enabling weakly supervised localization without dense annotations. To support systematic evaluation, we introduce TACO, a balanced synthetic dataset with full atomic activity coverage and pixel-level annotations. Experiments on OATS, TACO, and annotated nuScenes show superior recognition, strong sim-to-real transfer, and state-of-the-art weakly supervised localization.
[CV-91] Positive Pair Geometry Matters: Optimal Transport for Contrastive Learning of Visual Representations
链接: https://arxiv.org/abs/2609.24125
作者: Akshit Nanda,Shahzad Ahmad,Ram Prasad Padhy
类目: Computer Vision and Pattern Recognition (cs.CV); Computational Geometry (cs.CG)
备注: 12 pages, 5 figures
Abstract:Contrastive self-supervised learning has achieved strong performance by learning representations from multiple augmented views of the same image. However, most existing methods construct positive pairs using independently sampled stochastic augmentations, which may alter semantic content and ignore the intrinsic geometry of the data distribution. In this work, we propose OTCLR, an optimal transport-aware framework for contrastive learning representations that generates geometry-consistent positive samples. Instead of directly contrasting two randomly augmented views, we construct intermediate views between the original image and its augmented variants through entropic optimal-transport displacement interpolation. These transport-interpolated samples serve as positive views that better preserve image structure while explicitly modeling spatial distributional geometry. To further promote smooth representation learning, we evaluate auxiliary Sinkhorn regularization terms that encourage transport-interpolated views to remain consistent with their endpoint images. The proposed method can be incorporated into standard contrastive learning pipelines without modifying the encoder architecture. Experiments on multiple benchmark datasets show that our approach improves representation quality and transfer learning performance compared with conventional augmentation-based contrastive learning baselines.
[CV-92] Patch-to-Global: Random Patch Diffusion for Globally Consistent Megapixel Artifact Inpainting in Whole Slide Images MICCAI2026
链接: https://arxiv.org/abs/2609.24116
作者: Hyeseong Lee,Eunsu Kim,D M Bappy,Ho Heon Kim,Youngsuk Lee,Se Young Chun,Jang-Hwan Choi,Sung Hak Lee,Sangjeong Ahn
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages, 5 figures, accepted at MICCAI 2026
Abstract:Although deep learning has advanced Whole Slide Image (WSI) Analysis, tissue artifacts like bubbles and folds often cause silent failures by concealing essential morphology. Current pathology image restoration methods are mostly restricted to small patches, struggling to maintain global structural coherence at a megapixel scale. We introduce RestorePath, a framework for globally consistent megapixel scale inpainting that reconstructs diagnostic structures in histological image to prevent incorrect high-confidence predictions and lower error rates. Our model utilizes a Latent Diffusion Model (LDM) conditioned on Pathology Foundation Model (PFM) embeddings, integrating Large Kernel Attention (LKA) to manage long-range dependencies during random patch diffusion. Enhanced by Distance-Weighted Interpolation (DWI) and an Adaptive Guidance Scale (AGS), RestorePath ensures structural consistency and fidelity by modulating information from surrounding patches. Evaluations across TCGA-BRCA, BACH, and Camelyon16 datasets for images ranging from 512 to 4608 pixels demonstrate state-of-the-art performance in maintaining histological consistency. RestorePath significantly improves downstream Computational Pathology (CP) tasks, outperforming both raw artifact images and the conventional Detect-and-Discard (DD) approach. The code is available at this https URL
[CV-93] SPeaR: Test-Time Adaptation with Steering Primitives for Realigning Representations
链接: https://arxiv.org/abs/2609.24111
作者: Muhammad Sudipto Siam Dip,Ali Etemad
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Test-time adaptation (TTA) addresses distribution shift using only unlabeled test data. Existing methods typically adapt pretrained models by updating their parameters, limiting both what is adapted and where adaptation can occur within the network. We instead keep the pretrained network frozen and steer its intermediate representations. We introduce SPeaR (Steering Primitive for Realigning Representations), which inserts lightweight learnable modules at stage boundaries and optimizes them directly from the test stream, requiring neither source data nor supervised warm-up. Each primitive is optimized using a gated objective that reduces uncertainty only when adaptation is beneficial, along with a diversity regularizer to prevent collapse, and a multi-depth anchor to stabilize adaptation. We show that steering early representations is the most effective strategy, and that the same primitive transfers across convolutional and Transformer architectures. Across CIFAR-10-C, CIFAR-100-C, and ImageNet-C, SPeaR consistently matches or outperforms methods that adapt orders of magnitude more parameters, remains robust across a wide range of batch sizes, and preserves source-domain performance during continual adaptation.
[CV-94] Adaptive Cortically Constrained EEG-Vision Alignment for Zero-Shot Brain-to-Image Retrieval
链接: https://arxiv.org/abs/2609.24109
作者: Ye Wang,Haokun Ren,Wei Wu,Guoyin Wang,Zhuliang Yu,Hong Yu,Ke Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Zero-shot brain-to-image retrieval requires robust alignment between noisy EEG responses and visual representations. Existing EEG-vision alignment methods often operate in sensor space and apply fixed visual supervision to all responses, ignoring both spatial mixing in scalp EEG and response-wise variability in alignment reliability. We propose an adaptive cortically constrained EEG-vision alignment method for zero-shot brain-to-image retrieval. The method reconstructs EEG responses into predefined ROI-level source-pattern representations and encodes them with a Neuro-ROI Attention Encoder. To handle response-wise variability, we introduce an evidence-based adaptive visual supervision strategy that weights detail-controlled visual targets using model-based alignment evidence. On THINGS-EEG, the proposed method achieves strong 200-way zero-shot retrieval performance, with ROI-level attribution providing post hoc interpretability of the learned source-pattern representations. These results show that cortically constrained representation learning and adaptive supervision can jointly support EEG-vision alignment for zero-shot brain-to-image retrieval.
[CV-95] A2Safe: Counterfactual Evidence-Aligned Adaptive Agent Collaboration for Safe and Effective Visual Question Answering
链接: https://arxiv.org/abs/2609.24098
作者: Quanxing Xu,Ling Zhou,Xian Zhong,Jinyu Tian,Xiaohua Huang,Rubing Huang,Chia-Wen Lin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Visual Question Answering (VQA) with Multimodal Large Language Models (MLLMs) requires not only producing safe and effective responses, but also grounding safety decisions in the multimodal evidence that determines risk. Recent safety-alignment methods improve refusal behavior and contextual risk awareness, yet correct safety outcomes may still rely on superficial textual or visual correlations, particularly when risk emerges from interactions between individually benign image and question content. To address this issue, we propose A ^2 Safe, a counterfactual evidence-aligned adaptive agent collaboration framework for safe and effective VQA. A ^2 Safe organizes localized visual observations, textual intent, and cross-modal risk relations through a Grounded Safety Evidence Board, making the basis of safety decisions explicit. Counterfactual safety evidence alignment enforces invariance to safety-irrelevant changes while requiring appropriate safety-state and response-mode transitions when risk-critical evidence is minimally altered. The resulting evidence state further supports adaptive collaboration, enabling direct answering when grounded evidence is sufficient and invoking policy critique and response revision when evidence is risky, uncertain, or conflicting. Under complementary safety-critical and general VQA protocols, A ^2 Safe achieves a 95.72 SIUO safety score, reduces the benign refusal rate on MOSSBench to 14.67%, and maintains an average general VQA score of 78.34 with 27.8% token overhead. These results support counterfactual evidence-aligned adaptive collaboration for safe and effective multimodal question answering.
[CV-96] HDND: Hierarchical Dynamic Neural Decoding for Multilingual Word/Character Retrieval from Non-Invasive Brain Recordings
链接: https://arxiv.org/abs/2609.24095
作者: Yueyang Li,Shuran Chen,Wai Ting Siok,Nizhuan Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:While deep learning has enabled language decoding from intracranial brain recordings, extending this capability to non-invasive recordings remains an unresolved challenge. Decoding individual words from non-invasive brain recordings is particularly difficult, as word-level neural evidence is weak, temporally distributed, and entangled with acoustic, lexical, and semantic structure. Existing retrieval pipelines often collapse these factors into a single representation, potentially discarding information available at intermediate temporal scales. Here, we introduce Hierarchical Dynamic Neural Decoding (HDND), a hierarchical dynamic decoding framework that treats word decoding as structured refinement rather than flat label retrieval. HDND combines intermediate neural representations, contextual semantic predictions, and, for selected reading conditions, an auxiliary character-form objective. We evaluate HDND across seven electroencephalography (EEG) and magnetoencephalography (MEG) datasets spanning English, Dutch, Mandarin, and Cantonese listening, reading, and reading-aloud conditions. Across the nine-condition word-retrieval benchmark, the proposed HDND yields a higher participant-averaged balanced Top-10 point estimate than the matched contextual word-decoding baseline in every condition and achieves the highest mean among all compared methods in eight of nine conditions. Across the same nine matched conditions, HDND also yields higher token-micro and pooled word-macro Top-10 point estimates in every setting. Sentence retrieval favors HDND in eight of nine conditions, while auditory speech-segment retrieval is mixed across the six listening conditions. These results show that hierarchical residual refinement can improve multilingual word retrieval from heterogeneous non-invasive brain recordings.
[CV-97] Bridging Reconstruction and Generation: A Latent Distribution Perspective on Evaluation and Improvement
链接: https://arxiv.org/abs/2609.24088
作者: Xianghong Fang,Wenjie Shu,Tongda Xu,Wenlong Mou,Dehan Kong,Tim G. J. Rudner
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 27 pages, 23 figures,and 15 tables
Abstract:In latent generative models, reconstruction quality is often assumed to correlate with generative performance. However, reconstruction FID (rFID) can exhibit weak or even negative correlation with generation FID (gFID). We attribute this discrepancy to a latent distribution mismatch: reconstruction evaluates the decoder on encoder-induced latents, whereas generation uses the same decoder on latents produced by the generative model. To characterize this shift, we introduce generation-aware reconstruction (GAR), which constructs a continuous trajectory from standard reconstruction toward generation by perturbing encoder latents with noise and denoising them through the generative model before decoding. GAR probes the decoder behavior along this trajectory, making the transition from encoder to generation-time latent distributions observable and diagnosable. The resulting trajectory-based diagnostic, GAR-FID, exhibits strong empirical correlation with gFID across diverse tokenizers and scales. Importantly, intermediate GAR latents become more generation-aware while preserving correspondence with their source images, thereby retaining paired supervision that is absent for fully generated latents. This correspondence enables decoder adaptation on intermediate GAR latents, consistently improving generative quality across model scales. Overall, latent distribution mismatch provides a useful perspective for evaluating and improving latent generative models.
[CV-98] A paired synthetic construction-site image dataset for robust computer vision under adverse conditions
链接: https://arxiv.org/abs/2609.24075
作者: Viet Huy Duong,Ruoxin Xiong,Md Abdullah Al Forhad,Weishi Shi
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 21 pages, 7 figures, 7 tables. Dataset and code are publicly available
Abstract:Computer-vision systems used for construction monitoring can degrade under adverse environmental and visual conditions, yet such conditions remain underrepresented in existing construction image datasets. We present ConSynth-X, a paired synthetic construction-site image dataset containing 34,199 images derived from 3,109 real-world source scenes. The dataset comprises 11 condition-specific subsets spanning precipitation, fog, nighttime illumination, adverse weather at night, and small-object or long-distance views. Each synthetic image is linked to its corresponding source scene, enabling controlled comparison across environmental and visual conditions. ConSynth-X includes source-derived annotations, generation metadata, provenance information, and image-quality indicators, supporting object detection, image captioning, visual grounding, and visual question answering. Technical validation evaluates source-synthetic fidelity and alignment with real adverse-condition imagery using embedding-based similarity and distributional analyses. The dataset provides a structured resource for evaluating and improving the robustness of construction vision and vision-language models under challenging field conditions.
[CV-99] Monitorable Chart Reasoning Agents via Verifiable Process Rewards EMNLP2026
链接: https://arxiv.org/abs/2609.24071
作者: Sanchit Sinha,Oana Frunza,Kashif Rasul,Aidong Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: EMNLP 2026 Findings
Abstract:Chart reasoning agents are increasingly used to extract actionable insights in critical domains, achieving state-of-the-art performance on multiple benchmarks. Yet, high benchmark accuracy alone is insufficient for deployment, where stakeholders must be able to audit and verify how a model reaches its answer. Existing LVLM-based chart agents produce either answer-only predictions or free-form rationales that are hard to verify, obscuring whether an error arose from misreading the chart, extracting a wrong value, or miscomputing. We propose Chart-RVR, a reinforcement learning framework for training monitorable chart agents with verifiable process rewards. Chart-RVR decomposes chart reasoning into three auditable blocks: Structure, identifying the chart type; Evidence, reconstructing the underlying data table in JSON; and Derivation, exposing the stepwise trace that computes the answer. Across six in-domain and out-of-domain benchmarks, Chart-RVR attains state-of-the-art accuracy among comparable-sized LVLMs. Beyond accuracy, we assess monitorability using a triangulated protocol that combines ground-truth surrogate metrics, an oracle information-gain measure, and an LLM-as-auditor scoring Process Verifiability and Evidence Localization, showing that Chart-RVR yields rationales that are markedly more verifiable and evidence-grounded than those from CoT prompting, SFT, and existing chart-specific baselines.
[CV-100] Vision Transformers versus convolutional neural networks for fine-grained orchid genus identification in a species-rich data-poor flora: a controlled benchmark on the Orchidaceae of New Guinea
链接: https://arxiv.org/abs/2609.24064
作者: Reza Saputra,Diah Harnoni Apriyanti,André Schuiteman,Kurt Metzger,Ashley Field,Katharina Nargar,William Edwards
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Quantitative Methods (q-bio.QM)
备注: 24 pages, 7 figures, 4 tables
Abstract:New Guinea is the world’s richest island flora (~2,856 orchid species), yet most species are represented by only a handful of photographs, far fewer than direct species-level classification requires. Methods for fine-grained identification in such species-rich, data-poor floras are needed, and it remains unclear which backbone architecture and pretraining strategy best support them. We built a two-stage system that first predicts the genus of a query photograph, then retrieves visually similar reference images of candidate species using FAISS. We compared four pretrained backbones – two Vision Transformers (ViTs; DINOv2, BioCLIP 2) and two CNNs (ConvNeXt V2-L, EfficientNetV2-L) – fine-tuned under an identical protocol on a fixed, species-stratified partition of 16,701 photographs spanning 120 genera and 1,350 species, assessing accuracy, calibration, error structure, species retrieval, and open-set detection of novel genera. DINOv2 attained the best genus performance (macro top-1 66.9%, 95% CI 63.7-70.6; global top-1 88.9%); both ViTs outranked both CNNs, and general-purpose self-supervised pretraining (DINOv2) outperformed domain-matched biological pretraining (BioCLIP 2) by 7.1 points of macro top-1. Errors concentrated on two abundant genera acting as error attractors. DINOv2 embeddings achieved species Recall@5 of 86.6% and genus Recall@5 of 98.7%; temperature scaling reduced every backbone’s Expected Calibration Error to about 0.03; and a distance-based open-set gate flagged unseen genera (mean AUROC 0.958). A self-supervised Vision-Transformer backbone combined with embedding retrieval is an effective, deployable strategy for fine-grained identification in species-rich, data-poor floras. The system is released as an open web application (the New Guinea Orchid Identifier), offering a practical template for other hyperdiverse, under-documented taxa.
[CV-101] All-in-One Multilingual Scene Text Recognition with Script-aware Mixture-of-Experts
链接: https://arxiv.org/abs/2609.24058
作者: Xingsong Ye,Yongkun Du,Jiaxin Zhang,Zhixian Li,Chong Sun,Chen Li,Jing Lyu,Lianwen Jin,Zhineng Chen
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Code: this https URL this https URL
Abstract:Multilingual scene text recognition (STR) remains challenging due to the scarcity of training data for most languages and the difficulty of serving diverse scripts within a single model. Existing solutions either deploy one recognizer per language, inflating cost and introducing error accumulation, or rely on massive vision-language models (VLMs) that are expensive and still inaccurate on many scripts. In this work, we pursue an all-in-one multilingual recognizer that is simpler than per-language experts, lighter than VLMs, and more accurate than both. First, we construct TextMuSS-10M, a large-scale synthetic scene text dataset spanning 10 scripts and 229 languages. It provides balanced and sufficient supervision where real data is unavailable. Second, we propose ScriptMoE, a script-aware Mixture-of-Experts (MoE) architecture. It shares a single visual encoder and replaces the dense decoder with a sparse MoE block, which consists of an image-level router dispatches each image to the top-2 script-aligned experts and a shared expert absorbs cross-script knowledge. Extensive experiments on our assembled TextMuSS-Bench (10 scripts, 10,899 images) show that ScriptMoE achieves the highest accuracy of 82.06%, outperforming the strongest STR baseline by 1.31%. On the CC-OCR end-to-end multilingual task, replacing only the recognizer in PP-OCRv5 with ScriptMoE lifts F1 score from 65.71% to 80.89%, slightly surpassing the best VLM (80.73%) at a fraction of the parameter count.
[CV-102] U-PEN Mamba: Progressive Expansion with Selective State-Space Modeling for Efficient Retinal Vessel Segmentation
链接: https://arxiv.org/abs/2609.24049
作者: Abel A. Reyes-Angulo,Sidike Paheding,Vijayan K. Asari,Mohammad Alam,Jeevan Devagiri
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Accurate retinal vessel segmentation is important for computer-aided ophthalmic analysis, yet thin vessels, low contrast, and severe foreground-background imbalance remain challenging for encoder-decoder networks. This paper presents U-PEN Mamba, a U-shaped retinal vessel segmentation architecture that couples progressive nonlinear feature expansion with selective state-space modeling. The proposed network enriches local vessel responses with progressive expansion, models long-range spatial dependencies through a Mamba Global Context (MGC) block with linear sequence complexity, and uses attention-based decoder fusion to recover fine vascular boundaries. We evaluate U-PEN Mamba on CHASE DB1 and DRIVE using a consistent patch-based preprocessing pipeline and compare it with convolutional, attention-based, transformer-based, and Mamba-based segmentation baselines. U-PEN Mamba obtains the best mean intersection over union among the compared methods, achieving 0.8394 on CHASE DB1 and 0.8221 on DRIVE, with Dice scores of 0.8187 and 0.8078, respectively, using 21.6M trainable parameters. Ablation studies show that the MGC block contributes the largest gain over the U-Net baseline, while projection dimension and state size provide practical accuracy-efficiency control. These results indicate that selective state-space modeling is a promising global-context mechanism for parameter-efficient retinal vessel segmentation. Code is available at: this https URL.
[CV-103] Graph-to-Grid (G2G): Continuous-Coordinate Feature Painting for Soccer Pass Surfaces
链接: https://arxiv.org/abs/2609.24040
作者: Kaan Günay,Orhun Gun
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注: 39 pages, 5 figures, 18 tables
Abstract:Dense pass surfaces give, for every pitch cell, whether a pass played there would arrive, whether the carrier would choose it, and what the possession would then be worth. The networks that draw them read the state as a raster of per-cell counts, losing where inside a cell each player stands. LiDAR detectors, bird’s-eye-view perception and graph weather models move entity features onto a grid, binning each entity to a cell or learning the transfer. We evaluate the interpolated form: each player’s features are scattered bilinearly onto the grid at the player’s measured coordinates, so the surface loss trains the per-player encoder end to end. Those systems adopt an interface; this paper measures one. On 53,628 passes from the 2022 World Cup, painting improves selection likelihood over the same core fed rasters alone by about a quarter of a nat: in every match of an eight-fold cross-validation, with every arm tuned over five seeds, and after retraining on seven Bundesliga and 2. Bundesliga matches from another provider. Thirteen pre-specified studies locate the gain: painting the nine raw player features with no encoder carries three quarters of it, and the learned encoder and message passing add a smaller, resolved increment. Painting also helps the original SoccerMap and a canonical U-Net, whereas offset channels, a finer raster, an attention painter and a raster-free decoder do not. Frozen across the provider boundary the likelihood advantage is lost; injected tracking error compresses it. These results concern observed-endpoint prediction, not calibrated evaluation of hypothetical passes.
[CV-104] Video-STLayout Pre-training
链接: https://arxiv.org/abs/2609.24031
作者: Akash Abdu Jyothi,Greg Mori
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:In recent years, pre-training has become fundamental to learning effective video representations, enabling strong transfer to downstream tasks. A popular framework in pre-training involves aligning features of a video encoder with that of another modality, for example, language or audio. We introduce Video-STLayout pre-training, a novel strategy for obtaining rich video representations informed by spatio-temporal layout of object bounding boxes. Object layouts can easily be obtained by applying an off-the-shelf object detector on the video frames. Our method uses a contrastive loss to align video features with the layout features from a trained layout encoder. We show the effectiveness of our approach in the task of activity recognition in complex scenes.
[CV-105] InterHier: Learning Interconnected Hierarchical Semantics for Open-Vocabulary Object Detection
链接: https://arxiv.org/abs/2609.24026
作者: Yeong-Jin Kim,Ho-Joong Kim,Seong-Whan Lee
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 12 pages, 6 figures. Published in IEEE Access
Abstract:In this paper, we investigate the limitations of fixed, hand-crafted connectors in hierarchical semantic representations for open-vocabulary object detection. Existing methods establish semantic relationships between base categories and unseen novel categories by placing a fixed connector between adjacent super-/sub-categories. However, such fixed connectors may not optimally capture the relationships within a semantic hierarchy. To address this limitation, we propose interconnected hierarchical semantic representations (InterHier), which utilize a prepended learnable context to globally guide the interpretation of prompts containing hierarchical relationships. InterHier operates in two main stages. First, it constructs a hierarchy-aware prompt by integrating super-/sub-categories and prepending a learnable context. Second, it optimizes this learnable context to align visual region embeddings and textual embeddings. InterHier consistently improves performance over methods that rely on fixed connectors and can be seamlessly integrated into existing open-vocabulary object detection models. Experiments on open-vocabulary object detection benchmarks demonstrate that InterHier achieves competitive performance against state-of-the-art methods.
[CV-106] WebMRIQC: A Web-Based Implementation of MRIQC for Accessible MRI Image Quality Assessment in Resource-Constrained Settings MICCAI
链接: https://arxiv.org/abs/2609.24014
作者: Philip Nkwam,Ifeoluwa Oladeji,Sekinat Zurakat-Aderibigbe,Jasmine Cakmak,Harrison Aduluwa,Confidence Raymond,Cliff Mokua,Abdulrazaq Zubair,Daniel Champanda,Tolulope Olusuyi,Maruf Adewole,Udunna Anazodo
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: MICCAI CONFERENCE- MIRASOL 2026
Abstract:Reliable quality control (QC) of magnetic resonance imaging (MRI) is essential for reliable diagnostic neuroimaging, yet standard manual assessment is subjective and time-consuming. MRIQC has established standardized automated extraction of image-quality metrics (IQMs), but its reliance on local computational imaging skills and capacity including high-performance computing, limits its adoption in resource-constrained settings (RCS). We present WebMRIQC (this http URL), an open-source browser-based platform that wraps the validated MRIQC engine behind a zero-installation web interface. WebMRIQC automates the DICOM-to-BIDS conversion of de-identified MRI scans, executes the unmodified containerized MRIQC pipeline on a shared compute node governed by a fair-share job queue, and returns an interactive in-browser dashboard. The dashboard grounds every IQM in published quality thresholds, benchmarks each scan against the normative distribution of high-resource open datasets, and supports cross-site multicentre implementation of optimized scan protocols in this http URL describe the system architecture and a validation framework establishing measurement equivalence between WebMRIQC and native MRIQC across thirteen IQMs on the BraTS-Africa and BraTS 2021 datasets. Preliminary results indicate strong agreement for contrast-, signal and noise-based metrics, demonstrating that web-based implementation lowers the barrier to standardized MRI QC and provides a foundation for harmonized, regionally adapted quality benchmarks across RCS imaging sites. The code is publicly available here this https URL.
[CV-107] Evaluating the Generalization of Neuroimaging Foundation Models on African Brain MRI MICCAI
链接: https://arxiv.org/abs/2609.23983
作者: Oluwatobi Iyanuoluwa Akinmuleya,Olatokun Shamsudeen Akano,Samuel Danquah Ankapong,Olamide Lawal,Toufiq Musah
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: MICCAI AFRICAI Workshop, Strasbourg, France, 2026, 10 Pages, 1 Figure, 2 Tables
Abstract:Neuroimaging foundation models pretrained on large, predominantly western cohorts are increasingly proposed as general-purpose backbones for brain MRI analysis. Yet, their ability to generalize to underrepresented clinical populations remains largely untested. We evaluate four recent foundation models (BrainIAC, Neuro-JEPA, NeuroVFM, and Primus) on a three-way diagnostic classification task (Control, Dementia, Parkinson’s disease) using a cohort of 88 subjects from a Nigerian clinical brain MRI dataset, across four modality configurations (T1w, T2w, T1w+T2w, FLAIR), and compare against an end-to-end trained ViT3D baseline. The frozen backbones collapse to majority-class predictions, while Neuro-JEPA on FLAIR shows modest but still limited discrimination. In contrast, the end-to-end trained ViT3D achieves higher accuracy and MCC on every task (up to 53.4% accuracy, MCC=0.27) and is the only model with non-trivial recall. Our findings suggest that these frozen neuroimaging foundation models are insufficient for fine-grained diagnostic classification in small, non-western clinical cohorts, motivating parameter-efficient adaptation and broader multi-site external validation for equitable deployment in global health settings.
[CV-108] Rethinking Diffusion Segmentation: When Does It Rely on Its Noisy State and Does Diffusion Matter?
链接: https://arxiv.org/abs/2609.23967
作者: Hengzhuo Yang,Yuming Zeng,Yuling Yang
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Diffusion models are increasingly adapted from generation to conditional prediction, where a conditioning signal is combined with an evolving noisy representation of the target. In fully supervised segmentation, however, the conditioning image can already support direct target prediction, so endpoint performance alone establishes neither reliance on the added diffusion state nor a deterministic advantage over image-only prediction. For state reliance, we disrupt target-derived state content or correct image-state pairing during retraining of twelve published methods across three datasets, with ten matched seeds per setting. All 40 original-method comparisons whose evaluated-mask routes remained downstream of noised-quantity reconstruction exhibited state reliance, whereas all 30 comparisons with a segmentation-supervised bypass preserved reference performance. Rerouting five originally bypass-capable methods by forcing segmentation supervision through noise-to-mask reconstruction converted all 30 corresponding comparisons from preserved performance to state reliance. For deterministic utility, matched image-only counterparts achieved similar or better performance in 28 of 35 settings overall, including 16 of 20 whose native methods relied on both audited state properties. These results identify supervision path as a determinant of state reliance in the audited methods. Separately, matched image-only counterparts show that diffusion-specific computation often provides no deterministic endpoint advantage, including in methods that rely on the audited state properties. More generally, when conditioning already supports strong target prediction, diffusion-specific claims require additional evidence that the added state is used and that diffusion-specific computation improves the claimed capability beyond a matched condition-only counterpart.
[CV-109] Colon3R: Cross-Domain 3D Reconstruction from Monocular Colonoscopic Video
链接: https://arxiv.org/abs/2609.23961
作者: Zhihao Xing,Yingyu Wang,Liang Zhao,Shoudong Huang
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注:
Abstract:Monocular colonoscopic 3D reconstruction is important for surgical robotic colonoscopy, but remains challenging due to weak texture, specular reflections, limited view overlap, and non-rigid tissue motion. Conventional multi-view 3D reconstruction methods rely on stable correspondences and approximate rigidity, which are often violated in colonoscopy. Existing endoscopic methods often rely on domain-specific supervision, whereas there are not enough in-vivo labeled data available to adapt geometry foundation models to clinical colonoscopy. We present Colon3R, a cross-domain semi-supervised framework built on pretrained VGGT that transfers coupled camera, depth, and pointmap geometry from labeled phantom and simulated data to unlabeled in-vivo colonoscopy without requiring target-domain geometric annotations. Unlike source-only fine-tuning, which learns only from phantom and simulated data, Colon3R directly exploits unlabeled in-vivo video through teacher-derived cross-view supervision. Our proposed hierarchical quasi-rigid reliability selects reliable supervision at the sequence, directed-pair, and pixel levels, while source-preserving adaptation retains the learned coupled geometry during target-domain adaptation. Extensive experiments demonstrate that our method achieves superior overall performance over state-of-the-art approaches in depth, pointmap, and camera pose estimation. Qualitative comparisons on real in-vivo colonoscopy further show substantially more complete and geometrically consistent reconstructions than competing methods under clinical domain shift. The code will be public available after the paper is accepted.
[CV-110] MotionJEPA: Preventing Temporal Feature Collapse by Capturing Visual Changes in Latent Space
链接: https://arxiv.org/abs/2609.23881
作者: Markus Karmann,Shile Li,Christian Internò,Bruno Andreis,David Klindt,Randall Balestriero,Jindong Gu,Philip Torr,Qi Zhang,Peng-Tao Jiang,Hao Zhang,Bo Li,Onay Urfalioglu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: *Equal contribution (Markus Karmann, Shile Li). Code available: this https URL
Abstract:Joint Embedding Predictive Architectures (JEPAs) are a promising paradigm for learning task-agnostic latent world models without visual reconstruction. However, standard JEPA training exhibits a strong inductive bias towards slow features, causing feature suppression and the collapse of latent representation. While inverse dynamics provides temporal anti-collapse, it relies on action labels and offers little incentive to embed general, unlabeled dynamics. We introduce Difference Image and Single image embedding Regularization (DISReg), a novel regularizer that builds on an inverse-dynamics-style module that predicts temporal difference image embeddings without any pixel reconstruction loss, encouraging balanced static and dynamic feature learning. DISReg consists of a static term that shapes the distribution of the image embedding and encourages slow features, and a dynamic term, which, unlike direct regularization on the embedding, imposes no constraint on the image embedding’s shape or distribution and instead only incentivizes that dynamic features be present. By integrating this regularizer into a standard JEPA, we establish our new architecture, MotionJEPA. Latent probing demonstrates that MotionJEPA produces more complete representations than other methods, and our trajectory analysis shows it maintains geometrically simple latent embeddings with low curvature. We further show that MotionJEPA improves downstream planning success under static-background distractors across four environments.
[CV-111] Comparative Performance and Parameter-Efficient Adaptation of DINOv2 for Active Trachoma Classification
链接: https://arxiv.org/abs/2609.23832
作者: Kibrom Gebremedhin,Hadush Hailu,Bruk Gebregziabher,Yordanos Hailu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 6 pages, 5 figures and 3 tables
Abstract:Automated grading of conjunctival photographs could reduce the cost and variability of trachoma prevalence surveys, but the relative value of modern pretrained visual representations, lightweight feature adaptation, and training-objective design has not been established under a common protocol. This study presents a controlled evaluation for binary classification of Trachomatous Inflammation-Follicular (TF) versus Normal using 1,546 images from the public UCSF/Lietman collection. Images are processed using the OPTED pipeline for zero-shot tarsal-conjunctiva segmentation, alignment, cropping, and standardization. We first compare six pretrained backbones using a common classification pipeline and then evaluate four lightweight adaptation mechanisms on DINOv2 ViT-B/14. Under stratified five-fold cross-validation, DINOv2 with Efficient Channel Attention (ECA) and focal-plus-center loss achieved 91.66 +/- 0.97% accuracy, 90.69 +/- 1.10% macro-F1, and 96.06 +/- 0.71% AUC. ECA introduces only five learnable parameters while matching the performance of substantially larger alternatives. Objective ablation further showed that ECA did not consistently improve plain DINOv2 across loss functions; the lowest-variance 91.66% accuracy was obtained with cross-entropy plus center loss. Overall, the fine-tuned DINOv2 representation provided most of the predictive performance, while ECA offered a highly parameter-efficient refinement whose effect depended on the training objective. The resulting workflow provides a reproducible benchmark for active trachoma image classification.
[CV-112] DFD-Lab: A Modular Audio-Visual Deepfake Detection Pipeline
链接: https://arxiv.org/abs/2609.23830
作者: Jan Rybarczyk,Mateusz Roszkowski,Jacek Komorowski
类目: Computer Vision and Pattern Recognition (cs.CV); Sound (cs.SD)
备注:
Abstract:Comparing audio-visual deepfake detectors requires coordinating dataset adaptation, temporal input representation, model interfaces and experimental conditions. We present DFD-Lab, a modular pipeline that separates these responsibilities while supporting shared training and evaluation workflows. We integrate three implementations: Xception-based maximum-logit fusion, ResNet with temporal LSTM fusion, and our AVFF reimplementation. Experiments cover external testing, degradation-based training augmentation and evaluation-time corruption. On a filtered subset of Deepfake-Eval-2024, models trained on FakeAVCeleb attain baseline AUROC values of 0.504, 0.538 and 0.458. JPEG50 training augmentation raises these to 0.691, 0.605 and 0.570, respectively, while all three accuracies decrease. These results illustrate why training interventions, evaluation corruptions and metric-dependent outcomes should remain distinct within a common pipeline. The contribution is the integration of audio-visual processing, interchangeable detectors and configurable experimental workflows, supported by empirical case studies. The findings highlight the challenge of cross-dataset detection and the complementary information provided by ranking and classification metrics.
[CV-113] VISTA: Video-Injected Stylized Text-to-Animation
链接: https://arxiv.org/abs/2609.23817
作者: Monseej Purkayastha,Anindita Ghosh,Philipp Slusallek
类目: Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR)
备注: 3 pages, 1 figure
Abstract:We present VISTA, a two-stage framework for generating stylized 3D human motion by fusing structural content from text prompts with expressive style from reference videos, without requiring jointly paired (text, video, stylized motion) triplets. A Dual-channel Autoencoder first maps motion sequences and video clips into a shared latent manifold. A masked autoregressive diffusion backbone then operates within this manifold, injecting video-derived style through a dedicated late-fusion Dual-AdaLN pathway while preserving text-conditioned content structure. A cross-batch unpaired training protocol with latent cycle consistency enables joint learning across separate semantically rich and stylistically diverse datasets. As a proof-of-concept for controllable animation synthesis, we validate VISTA on rendered motion-capture references: it achieves the highest style recognition accuracy among video-conditioned methods while preserving competitive content alignment, and its decomposed 3-way classifier-free guidance provides independent, user-controllable calibration of the content–style balance at inference time.
[CV-114] Confidence-Aware Teacher-Student Distillation for 3D Medical Segmentation
链接: https://arxiv.org/abs/2609.23815
作者: Georgios Triantafyllou,Dimitris K. Iakovidis
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Medical image segmentation models typically rely on large amounts of densely annotated volumetric data, limiting their scalability across tasks and imaging modalities. This work addresses the challenge of predicting entire 3D anatomical structures from extreme annotation sparsity. An annotation-efficient student-teacher framework is proposed for automatic 3D medical segmentation that requires only a set of point prompts on a single 2D slice per volume, as input. A foundation model serves as an offline teacher, utilizing the provided point prompts from the selected slice to full-volume pseudo-annotations alongside their corresponding spatial confidence scores prior to student training. To mitigate the error propagation of noisy pseudo-annotations, a task-specific 3D student network is trained using a confidence-aware optimization strategy. By leveraging the teacher’s pre-computed confidence scores, this strategy explicitly excludes statically uncertain regions of the pseudo-annotations from the loss calculation, while simultaneously emphasizing regions with higher confidence. Evaluated on 3D cardiac MRI datasets, our framework outperforms state-of-the-art semi-supervised methods, improving segmentation performance by up to 43.6%. Furthermore, it drastically reduces the manual annotation burden to just a few positive point prompts per volume, while improving surface boundary precision by up to 14.7% over the teacher and successfully recovering up to 34.1% of the performance gap toward the fully supervised upper bound.
[CV-115] MoSAT: Human Motion Generation from Spatial Audio and Textual Description
链接: https://arxiv.org/abs/2609.23797
作者: Shuyang Xu,Zhiyang Dou,Yiduo Hao,Zekun Li,Liang Pan,Jingbo Wang,Cheng Lin,Yuan Liu,Wenping Wang,Mingmin Zhao,Taku Komura
类目: Graphics (cs.GR); Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO); Sound (cs.SD)
备注:
Abstract:Human motion is shaped by both external acoustic events and behavioral intent: spatial audio conveys environmental cues that elicit or guide a response, while text specifies the desired action and how it should be performed. In this paper, we study the novel task of human motion synthesis jointly conditioned on spatial audio and natural language, a problem that has been largely overlooked in previous research. To support this task, We introduce STAM, a dataset of motion sequences paired with spatial audio and detailed textual annotations whose rich vocabulary affords precise and nuanced specification of human motions. We further introduce MoSAT, a latent flow-matching framework for full-body motion generation jointly conditioned on natural-language intent and directional spatial-audio cues through hierarchical cross-attention before generating motion. Such a hierarchical design enhances temporally coherent and semantically aligned motion sequences. We also develop tri-modal evaluators for comprehensive evaluation on this novel task. Extensive experiments show that MoSAT achieves the SOTA performance by leveraging spatial audio’s intrinsic motion-shaping properties alongside textual semantics, enabling precise and diverse motion in various scenarios.
[CV-116] Mira-Scene: Pixel-Aligned Layouts for Generative 3D Scene
链接: https://arxiv.org/abs/2609.23796
作者: Yang-Tian Sun,Tianjia Liu,Zehuan Huang,Yi-Hua Huang,Xiaoyang Lyu,Ziyi Yang,Zi-Xin Zou,Yuan-Chen Guo,Yan-Pei Cao,Xiaojuan Qi
类目: Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR)
备注: Project Page: this https URL
Abstract:Single-image 3D object generation can now produce high-fidelity assets, yet accurately placing them into a coherent scene layout remains an open challenge. A central difficulty lies in how object layout is represented. Holistic methods absorb placement into a scene-level generation process, sacrificing object-level detail. Compositional methods preserve object fidelity by decoupling geometry from layout, but typically parameterize layout as sparse, unbounded pose variables that are difficult to learn and generalize poorly under scarce scene-level this http URL present Mira-Scene, a compositional 3D scene reconstruction framework that replaces sparse pose regression with dense, bounded correspondence recovery. At its core is the Canonical Coordinate Map (CCM), a pixel-aligned field that maps each visible object pixel to a surface coordinate in the object’s bounded canonical space. When paired with a scene-space Point Cloud Map (PCM) from monocular geometry estimation, CCM induces dense canonical-to-scene correspondences from which object transformations are recovered through robust geometric alignment. Because CCM operates in bounded canonical space, it provides a stable prediction target that can be trained from scalable object-level 3D data without requiring scene-level layout annotations. Mira-Scene further introduces a multimodal diffusion transformer that jointly generates object geometry and CCMs, using modality-specific expert streams with shared attention and positional encoding to promote geometry-layout consistency. Experiments on indoor, outdoor, synthetic, and in-the-wild scenes show that Mira-Scene substantially outperforms strong baselines in layout accuracy, achieving relative gains of 39.8% in 3D-IoU and 16.5% in 2D-IoU over SAM3D, using limited open-source training data.
[CV-117] PRISM-RAG : Multimodal Hypergraph Retrieval-Augmented Generation for Tobacco Product and Legislative Policy Reasoning
链接: https://arxiv.org/abs/2609.23769
作者: Manuel Serna-Aguilera,Raegan Anderes,Page Dobbs,Khoa Luu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:The disambiguation of semantically similar statutory text across jurisdictions is a retrieval problem that existing methods do not solve. This inter-context conflict can steer generative models toward confidently produced answers grounded in topically relevant but jurisdictionally incorrect sources. Tobacco and nicotine regulations vary by US jurisdiction, often sharing similar language, thus, robust reasoning requires identifying which jurisdiction’s law governs a given product, not merely retrieving relevant text. Emerging products (e.g., pouches) exploit ambiguous definitions to evade regulation. State-of-the-art (SOTA) document retrieval-augmented generation (RAG) methods struggle to address this inter-context conflict, and thus struggle to connect image attributes (e.g., rich attribute captions) to the set of similar legislation texts. We introduce NicoPRISM (Nicotine Product and Regulation Image-and-Text Surveillance Multimodal), comprising 161,563 images, attribute captions, a knowledge base of product, health, and legislative documents spanning 13 US jurisdictions, and 1,495 validated question-answer pairs across two tasks: policy compliance QA and product knowledge QA. We also propose PRISM-RAG, a multimodal hypergraph RAG framework built over images, captions, and entities without any LLM calls at index time, grounding every query in a product image and routes retrieval through a jurisdiction-aware context assembly mechanism guaranteeing that statutory text from the queried jurisdiction reaches the language model by construction. PRISM-RAG retrieves passages from the correct jurisdiction in 93.9% of policy compliance queries, a 48.6 percentage point advantage over standard RAG (p0.001), using zero LLM calls at index time and one at query time, and is competitive with or outperforms SOTA RAG frameworks across keyword, semantic, jurisdiction-, and compliance-accuracy metrics.
[CV-118] raining-Free Spectral Transductive Refinement for Cross-Domain Few-Shot Classification
链接: https://arxiv.org/abs/2609.23758
作者: Fahim Rahman,S.M. Tanjeeb Meheran Rohan,Md. Taimum Ibne Sayed,Asaduzzaman Herok,Md. Bakhtiar Hasan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 9 pages (excl. references), 3 figures, 6 tables
Abstract:Few-shot recognition with frozen visual features is especially fragile under domain shift and one-shot supervision, where a single labelled image is an unreliable estimate of its class. We ask how far this fragility can be reduced purely at test time, without retraining the encoder or augmenting the source domain. We present Spectral Transductive Refinement (STR), a training-free transductive inference rule that exploits the geometry of the complete support-query episode. Given frozen embeddings, STR builds a joint k-nearest-neighbour graph, maps the episode into a normalized-Laplacian spectral coordinate system, initializes class representatives from the labelled support, and iteratively refines them using pseudo-labelled queries. We evaluate STR under two protocols. A controlled component study with frozen ResNet-18 features shows that spectral refinement consistently improves over single-prototype spectral initialization across five shifted domains, with the largest gains in the one-shot regime where support estimates are weakest. We then benchmark STR against recent Cross-Domain Few-Shot Learning (CD-FSL) methods using the standard miniImageNet-pretrained ResNet-10 backbone over eight established target domains. Operating entirely at inference time, STR attains the highest 1-shot average among compared methods and remains competitive at 5-shot, rivalling approaches relying on heavy source-domain meta-training augmentations. Because STR is transductive, we report its setting explicitly. Diagnostics attribute its gains to iterative refinement in spectral coordinates rather than added prototype capacity, which remains inactive in our configuration.
[CV-119] OnlineWM: Causality-Aware Active Online Learning for Effective World Modeling
链接: https://arxiv.org/abs/2609.23753
作者: Yikun Miao,Fangqi Zhu,Quanxin Shou,Xiaoyi Pang,Zhengyang Yan,Junhao Li,Haodong Wang,Zicong Hong,Song Guo
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 23 pages, 9 figures. Project page: this https URL
Abstract:Generative world models aim to predict future states conditioned on actions, where action controllability is fundamental for reliable dynamics modeling. While recent efforts leverage simulator-generated data to enhance this capability, existing training pipelines face two fundamental limitations. First, static offline data collection leads to a distribution misalignment between training sets and the model’s evolving error patterns, failing to resolve critical long-tail scenarios where dynamics predictions remain unreliable. Second, the standard objective of minimizing observational discrepancy often encourages the model to exploit spurious correlations instead of capturing the underlying action-effect causality. To address these limitations, we propose OnlineWM, an online training framework that continuously improves world modeling through active simulator interaction and causality-aware optimization. OnlineWM introduces two key innovations: (1) Active Online Learning: Instead of using fixed datasets, OnlineWM adaptively queries the simulator for new interaction sequences that target the model’s current predictive weaknesses, ensuring high-utility data acquisition. (2) Causality-Aware Fine-Tuning: We propose a counterfactual learning strategy that contrasts the outcomes of different actions from identical states, forcing the model to attribute state transitions to specific actions rather than ambient environmental evolution, thereby grounding its predictions in reliable causal mechanisms. By integrating active data acquisition with causal optimization, OnlineWM establishes a closed-loop refinement process that ensures the model is both robust to diverse scenarios and precise in its causal attribution. Extensive experiments demonstrate that OnlineWM significantly enhances action controllability and generalizes effectively to unseen domains.
[CV-120] VGGT-Prime: Compute-Adaptive Mixture-of-Heads for Efficient Visual Geometry Transformers
链接: https://arxiv.org/abs/2609.23733
作者: Abteen Arab,Guile Wu,Chengjie Huang,Dongfeng Bai
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Technical Report
Abstract:Feed-forward visual geometry models such as the Visual Geometry Grounded Transformer (VGGT) have recently enabled direct 3D reconstruction from multi-view images. Despite their promising performance, these models scale quadratically with the number of input views due to their global attention mechanism, resulting in substantial latency for long sequence inputs. There have been some recent efforts to accelerate VGGT, but they primarily focus on reducing \emphtoken redundancy through token merging or key/value sparsification. Our work resolves this bottleneck from a different perspective by investigating \empharchitectural redundancy in visual geometry transformers. We show that the multi-head attention modules in VGGT’s global-attention layers contain substantial architectural redundancy, with only a subset of heads carrying critical geometric information. In light of this observation, we propose VGGT-Prime, a compute-adaptive mixture-of-heads model that resolves this redundancy to accelerate visual geometry transformers while maintaining competitive reconstruction quality. The key idea of VGGT-Prime is to estimate the appropriate computation level for each global-attention head using a lightweight router and then dynamically assign each head to different computation modes. Extensive experiments on multiple datasets demonstrate that VGGT-Prime can achieve an 8\times inference speedup over VGGT while maintaining competitive performance on camera pose, depth, and point-cloud predictions. We further show that VGGT-Prime is complementary to existing acceleration methods, such as token merging, further improving inference speed by up to 14\times over VGGT. An overview of our work is available on our \hrefthis https URLproject page.
[CV-121] BindCLIP: One Balanced Coupling For Compositional Vision Language Scoring
链接: https://arxiv.org/abs/2609.23717
作者: Liuyang Song,Yi Zhang,Zhongyi Deng,Daqian Yang,Hongbo Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Global vision–language similarities compress an image and a caption into one vector, preserving semantics but not which word corresponds to which region or how those regions are arranged; a model can recognize every word and object yet prefer a compositionally incorrect caption. We argue that a frozen encoder retains this association structure, so the problem is to read it rather than to rebuild it beside the pretrained similarity. We introduce BindCLIP, a pairwise scorer built on one latent object: a balanced token–patch–depth optimal-transport coupling that places both candidate captions and several visual depths in a single plan. Semantic, entity, order, and spatial evidence are read as energies of this state, and exchanging the candidates permutes the plan, making the score exactly antisymmetric. A geometric refinement inside the coupling contracts moves that the candidates and the visual depths do not support. No task label, parser, relation inventory, or detector is used. One checkpoint and one inference path improve the official What’sUp, ARO, and SugarCrepe benchmarks over frozen global CLIP, with the strongest transfer on the relation splits. Controls rule out patch access and caption-length shortcuts, and an inference-time lesion localizes spatial arrangement to the coupling.
[CV-122] Layer-Aware Position Embeddings for Visual Token Pruning in Multimodal Large Language Models
链接: https://arxiv.org/abs/2609.23715
作者: Yahong Wang,Zhangkai Ni,Juncheng Wu,Yuyin Zhou,Ying Wen,Lianghua He
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Multimodal large language models (MLLMs) incur substantial computational overhead due to the reliance on hundreds of visual tokens to represent images. While token pruning has emerged as a promising approach to reduce the inference cost of MLLMs, existing methods typically reassign position embeddings to the retained tokens using either sparse or continuous position embeddings, each introducing distinct limitations. Sparse position embeddings tend to decrease the attention value allocated to visual tokens, thereby degrading the perception capability of MLLMs, whereas continuous position embeddings disrupt the original spatial correspondence of visual tokens, leading to weakened grounding capability. To mitigate this issue, we perform layer-wise analysis of the language decoder and observe that intermediate layers play a critical role for maintaining the grounding capability of MLLMs under token pruning. Based on this observation, we propose a layer-aware position embedding strategy, which switches to sparse position embeddings at grounding-sensitive layers while maintaining continuous position embeddings elsewhere. Extensive experiments across representative pruning methods and diverse benchmarks demonstrate that our approach improves the comprehensive multimodal performance of pruned MLLMs compared with standard sparse and continuous position embeddings.
[CV-123] Infectious Bovine Pinkeye Detection Using Computer Vision and Imbalance-Aware Learning
链接: https://arxiv.org/abs/2609.23714
作者: Michael Abalo,Jameson Brennan,Hossein Moradi Rekabdarkolaee
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Infectious bovine pinkeye is a contagious ocular disease that adversely affects cattle health, welfare, and agricultural productivity. Conventional diagnosis relies primarily on clinical observation, which can be subjective, time-consuming, and difficult to implement efficiently in large herds or remote settings. This study evaluated and compared You Only Look Once (YOLO) v11 and YOLOv26 for automated bovine pinkeye classification and investigated the effects of class-balancing strategies on model performance. Five variants (n, s, m, l, and x) of each architecture were trained and evaluated using the original imbalanced dataset, Random Minority Oversampling (RMO), and an adapted Synthetic Minority Oversampling Technique (SMOTE). Both YOLOv11 and YOLOv26 demonstrated strong classification performance, although the effects of class balancing varied across model variants. For YOLOv11, RMO-s achieved an accuracy of 0.99, a macro F1-score of 0.98, and a true positive rate (TPR) of 1.00, with no false-negative classifications. RMO-m also achieved a TPR of 1.00 with no false negatives. For YOLOv26, the original l, RMO-m, and RMO-l variants each achieved an accuracy of 0.99 and a macro F1-score of 0.98, with RMO-l attaining a TPR of 1.00 and no false negatives. Overall, RMO generally provided greater improvements in minority-class detection than adapted SMOTE, whereas the strong performance of the original YOLOv26-l demonstrates that oversampling was not necessary for all model variants. These findings demonstrate the potential of YOLOv11 and YOLOv26 for automated detection of bovine pinkeye and support further evaluation for livestock health monitoring.
[CV-124] A multi-temporal dataset for mapping burned areas in the Brazilian Cerrado using time series of remote sensing imagery
链接: https://arxiv.org/abs/2609.23687
作者: Alisson Cleiton de Oliveira,Thales Sehn Körting
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: Published in Big Earth Data, 2025
Abstract:This paper introduces a multi-temporal tabular dataset derived from satellite images to map burned areas in the Chapada dos Veadeiros National Park, in Goiás, Brazil, covering the years 2020 to 2022. The dataset contains blue, green, red, and near-infrared bands, as well as the BAI, EVI, GEMI, NDVI, and NDWI spectral indices from the WFI sensor on the CBERS-4A, CBERS-4, and AMAZONIA-1 satellites, organized into a regular grid. We applied the Random Forest classifier to develop and validate models based on samples labeled as totally burned, partially burned, and non-burned. Two classification approaches were tested: one combining burned and non-burned areas into binary classes and another distinguishing between totally burned (TB), partially burned (PB), and non-burned (NB) classes. Seven validation approaches assessed different post-classification combinations, focusing on accuracy, precision, recall, and intersection over union (IoU) metrics. Results showed higher IoU when TB, PB, and NB were used as individual classes and TB was reclassified as burned area (BA) while PB and NB were grouped as non-burned. Comparing the annual results of this approach to the MCD64A1 product, the errors of omission for the BA class were 22% in 2020, 28% in 2021 and 59% in 2022, while the errors of commission were 46%, 43% and 46%, respectively. The study highlights the utility of the WFI sensor for burned area mapping without inter-satellite spectral calibration and suggests further exploration with other machine learning algorithms to evaluate the dataset potential and limitations.
[CV-125] Mind the Gaps: A Curated Benchmark for Form Field Detection
链接: https://arxiv.org/abs/2609.23679
作者: Iheb Brini,Omar Moured,Hamza Gbada,Elisa Barney
类目: Computer Vision and Pattern Recognition (cs.CV); Databases (cs.DB)
备注: Accepted at the International Workshop on Document Analysis Systems (DAS) 2026. Dataset: this https URL
Abstract:Form Field Detection (FFD) is a fundamental component of document understanding systems, enabling applications ranging from large-scale industrial digitization to accessible form interaction for automated analysis. Unlike conventional object detection tasks, FFD is inherently challenging because fields are often defined by layout structure and whitespace rather than visible foreground content. Existing large-scale datasets frequently rely on heuristic annotation pipelines, resulting in noisy and inconsistent labels that hinder reliable evaluation. In this work, we introduce mini-CommonForms, a carefully curated FFD benchmark with consistent, high-quality annotations, and present a detailed evaluation of state-of-the-art detection approaches. The benchmark is designed to support reproducible research in document automation and accessibility-oriented applications. Dataset and code are available at this https URL
[CV-126] Which Terrain Is Better? Preference Learning with VLM Prototypes for Off-Road Traversability Ranking
链接: https://arxiv.org/abs/2609.23673
作者: Ji-Hoon Hwang,Jisung Bae,E-In Son,Dong-Wook Kim,Jung-Taak Kim,Seung-Woo Seo
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: 8 pages, 5 figures
Abstract:In vision-based off-road navigation, a robot needs to know not only which obstacles to avoid but also which terrain is better. The first is handled by freespace detection or semantic segmentation. The second is usually answered with a traversability score, but no universal ground truth exists for such a score, so perception falls back on a predefined value per semantic class or a freespace confidence. These scores say what a region is, not which region a robot should prefer. We therefore formulate this preference as visual traversability ranking, an ordering of visible terrain that can be supervised by comparisons between two regions. Standard annotations do not label preference, but they imply its direction. We present TravPro, which converts these annotations into ordered region pairs and fits a small readout on frozen vision–language model (VLM) patch tokens to these pairs. The tokens are clustered once into a fixed prototype bank, and the readout learns a preference score per prototype. The readout is then applied to every patch and serves as a teacher that turns sparse comparisons into dense preference pseudo-labels without pixel-wise annotation. An RGB student distills these maps into a dense terrain-preference map together with a non-ground mask that excludes obstacles and background from the ranking. On five unseen domains, TravPro reaches a mean pairwise accuracy of 0.915 against 0.783 for the strongest baseline, producing an ordering sensitive to surface condition that a per-class value cannot represent. The same VLM and the same supervision yield no such ordering when the VLM is prompted and the supervision is used as dense targets; what matters is how they are used.
[CV-127] Why Do Video Diffusion Models Violate Physics? Unveiling the Flaws in Attention Mechanisms
链接: https://arxiv.org/abs/2609.23658
作者: Yueyan Li,Haibo Wang,Caixia Yuan,Xiaojie Wang
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注:
Abstract:Despite impressive visual quality, state-of-the-art video diffusion models often generate content that violates real-world physical laws. While existing solutions rely on external priors or specialized data, we investigate the root cause by exploring the internal mechanisms of these models. Specifically, we present the first interpretability study on the ‘‘motion planning’’ process of text-to-video diffusion models, revealing how motion trajectories form during early denoising stages. Building upon the ‘‘first shape, then details’’ finding, we combine cross-attention trajectory patterns with causal head contributions to identify a specific subset of attention heads driving motion planning. Further, our self-attention analysis shows that Rotary Position Embedding (RoPE) induces excessive spatial attention decay. This causes early candidate regions to prematurely lock into physically implausible positions, suppressing reasonable trajectories in adjacent frames and triggering generation failure modes. To address this fundamental flaw, we propose a lightweight architectural modification that scales the frequency of RoPE across different denoising steps. This strategy reduces excessive attention decay, helping the model explore better candidate regions to establish coherent physical motion. Finally, training-free and training-based experiments confirm the effectiveness of our approach in enhancing the physical commonsense of generated videos.
[CV-128] Reassessing Global Gradient-Norm Imbalance in BLIP Fine-Tuning Across Physical Domains
链接: https://arxiv.org/abs/2609.23655
作者: Kiran Naseer,Samreen Azhar,Dwarikanath Mahapatra
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages, 14 figures. Supplementary material included as ancillary file
Abstract:Imbalanced gradient magnitudes between the visual and language pathways of a vision-language model are often treated as a defect to be corrected. We test that premise for one family of correction, deliberately excluding adaptive, signal-driven schemes (e.g. BalGrad, OGM, PMR, CGGM), which are a mechanistically distinct class outside this study’s scope. Measuring the language-to-visual gradient-norm ratio, reported in parameter-normalised form, across nine fine-tuning conditions, three seeds, and three captioning datasets spanning distinct physical domain shifts – underwater, aerial, radiological – we find imbalance magnitude varies markedly across domains with no predictable ordering. A plain learning-rate reduction cuts imbalance substantially and lands within a few BLEU points of the best method on every dataset. Staged freezing reduces the ratio on every domain yet never ranks first; a schedule-only control isolates freezing as the cause on one dataset but not the other two. Forcing the two gradient groups to equal magnitude drives per-parameter imbalance close to zero on every domain, yet is both the best result in the study and the worst placement among full fine-tuning methods, on different datasets, with identical settings. Reductions in gradient-norm ratio do not consistently predict captioning performance across domains, and how a given level of balance is reached matters as much as the level itself. As a secondary finding, a commonly reused LoRA configuration applied to BLIP silently adapts zero visual parameters; correcting it improves BLEU-4 on all three datasets.
[CV-129] Compact Low-Cost Hyperspectral Imaging via Angular-to-Spectral Diversity Conversion ECCV2026
链接: https://arxiv.org/abs/2609.23619
作者: Kazuma Fujiwara,Takuya Funatomi,Kazuya Kitano,Yuki Fujimura,Yasuhiro Mukaigawa
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: ECCV 2026. 18 pages + 10 pages supplementary material
Abstract:Snapshot hyperspectral imaging avoids sequential scanning, but systems that jointly achieve stable reconstruction, low cost, and compact optics remain limited. We present a snapshot hyperspectral imaging system based on angular-to-spectral diversity conversion. A tapered kaleidoscope creates replicated views with distinct incidence directions, and a directly attached birefringent filter converts them into view-channel-dependent spectral transmittances, yielding complementary measurements that better condition the inverse problem for more stable single-shot spectral reconstruction. The system preserves a simple pixel-wise linear model for fast non-learning-based reconstruction and uses only off-the-shelf components without relay optics or cascaded modules. We select the birefringent filter configuration using a condition-number-based criterion and validate the system on both synthetic and real data.
[CV-130] Beyond UV Mapping: Mesh Texture Compression via Surface-Aligned Texture Fields
链接: https://arxiv.org/abs/2609.23606
作者: Jianqiang Wang,Junhui Hou,Siyu Ren,Weiyao Lin,Wenping Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 21 pages
Abstract:Mesh texture compression typically relies on 2D UV atlases, whose chart discontinuities and mapping overhead can limit coding efficiency. To tackle this challenge, we introduce TexF, a surface-aligned texture field that organizes texture attributes in sparse voxels derived from the mesh surface. This representation supports high-resolution textures while preserving local 3D correlations for compression and enabling direct surface queries. For bitstream compression, TexF reuses established 3D attribute codecs, with voxel locations reconstructed from the decoded mesh without separate transmission. For GPU-resident compression, we develop 3DNTC, which combines quantized hash features with a lightweight decoder for random-access reconstruction at surface positions. Differentiable rendering enables image-space refinement of both voxel attributes and compressed neural fields. Experiments on the MPEG and AOM mesh compression benchmarks demonstrate improved average rate-distortion performance over representative UV-based methods for both bitstream and GPU-resident compression. 3DNTC also supports real-time rendering.
[CV-131] PREM: Prefix-Steered Recurrent Memory for Long-Video Understanding
链接: https://arxiv.org/abs/2609.23601
作者: Siru Zhong,Qiongyan Wang,Xiaohui Lv,Yuzheng Zhuang,Shuai Tao,Wulong Liu,Haohuan Fu,Yuxuan Liang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 22 pages, 11 figures, 12 tables
Abstract:Long-video understanding must capture transient visual evidence under strict token budgets, yet existing methods compress frames, append memory tokens, or alter internal key-value (KV) caches. We introduce Prefix-Steered Recurrent Memory (PREM), a memory-token-free framework for frozen vision-language models (VLMs). PREM separates video ingestion from query answering: a recurrent writer distills visual streams into a compact 256 KiB multi-slot associative state, while a question-conditioned readout adds memory-derived key/value (K/V) steering modulations to existing non-visual prompt prefixes during prefill. This enables write-once, query-many inference without extra prompt tokens or decoding recurrence. Across six long-video benchmarks in offline and streaming end-of-stream settings, PREM consistently outperforms frozen baselines at every evaluated visual budget. Under a constrained budget of 16 frames, PREM improves macro-average accuracy by 3.06% on Qwen2.5-VL-3B, with gains of 11.0% on action antonym identification and 9.9% on localized needle retrieval. These gains require tuning 0.24% of backbone parameters at 0.03 GiB of peak GPU memory overhead.
[CV-132] PETR: Prompt Ensembling with Training-free Routing for Vision-Language Models
链接: https://arxiv.org/abs/2609.23600
作者: Weihan Cai,Hao Tan,Xinping Gao,Shibiao Xu,Jun Wan
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Prompt learning efficiently adapts vision-language models (VLMs) to downstream tasks, but gains on seen classes often come at the expense of generalization to unseen classes. To address this limitation, we propose prompt ensembling with training-free routing (PETR), whose key innovation is a carefully designed dual-prompt architecture: two complementary prompts are learned from different data and objectives to emphasize seen class discrimination and unseen-class generalization, respectively. During training, both prompts are fine-tuned using a shared frozen CLIP backbone, and statistical information is collected from the training set logits. At inference time, we determine the similarity of each test sample to seen data, and route the sample to the most appropriate prompt branch. To the best of our knowledge, this is the first prompt tuning framework that performs training-free adaptive routing based on statistical similarity. This design provides an interpretable routing signal and avoids common MoE-style routing pathologies, such as router training instability and load imbalance. Extensive experiments on 11 benchmark datasets demonstrate that our framework consistently outperforms previous methods on both seen and unseen classes, achieving new state-of-the-art results.
[CV-133] StyleAT: Defending Face Recognition Against Semantic Attacks BMVC2026
链接: https://arxiv.org/abs/2609.23596
作者: Ben Shapira,Roi Cohen,Shang-Tse Chen,Mahmood Sharif
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Machine Learning (cs.LG)
备注: Accepted at the 37th British Machine Vision Conference (BMVC 2026), Lancaster, UK. 14 pages main text plus 13 pages of appendices, 15 figures, 12 tables
Abstract:With face-recognition models now embedded in everyday authentication and surveillance, recent works have pinpointed a critical weakness: these models remain acutely vulnerable to adversarial semantic edits. I.e., adversarially produced semantic alterations to the input, such as slight aging or pose changes, can induce misclassifications. Certain existing attacks are powerful, but they can be computationally costly, rendering them inadequate for developing defenses (e.g., through adversarial training). To fill the gap, we introduce BoundStyle, a potent semantic attack operating in StyleGAN’s rich latent space to maximize misclassification rates. Notably, BoundStyle achieves high attack success rates while being \sim\times9.5 faster than existing state-of-the-art attacks, making it suitable for adversarial training. Building on BoundStyle, we develop StyleAT, an efficient adversarial training scheme that incorporates low-budget attack variants yet defends against stronger and unseen semantic attacks. We evaluate on two datasets unseen during training and seven models, and find that StyleAT boosts robust accuracy against state-of-the-art attacks and outperforms common defenses in various settings.
[CV-134] An Efficient and Effective Watermarking Scheme for the Protection of the Intellectual Property Rights of Video Generative Models
链接: https://arxiv.org/abs/2609.23586
作者: Wenhong Huang,Jianwei Fei,Benedetta Tondi,Bin Ma,Fangjun Huang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:The rapid development of video generative models (VGMs) has enabled the generation of highly realistic synthetic videos, raising concerns about the intellectual property rights (IPR) of these models. In particular, two closely related forensic tasks remain largely unaddressed: synthetic video verification (determining whether a video was generated by a protected VGM) and model ownership verification (determining whether a suspect VGM is an unauthorized copy of a protected VGM). In this paper, we propose a new in-generation watermarking scheme that can address the two verification tasks. First, a novel video watermarking network named VidMark is presented, which incorporates a two-scale discrete wavelet transform (DWT) decomposition and a global temporal attention block (GTAB) to enhance watermark robustness and imperceptibility. Second, we present a decoder-guided fine-tuning procedure. By leveraging the frozen VidMark decoder, this process enables VGMs to synthesize videos carrying an imperceptible, robust, and model-specific watermark. Finally, two verification frameworks are established to perform synthetic video verification and model ownership verification. Extensive experiments on representative VGMs demonstrate that the proposed scheme achieves over 99% watermark extraction accuracy and 100% verification accuracy on both tasks, with negligible impact on video generation quality. Furthermore, the watermarks exhibit strong robustness against a comprehensive range of video-level and model-level attacks.
[CV-135] G6D: Geometric Learning-Free RGB-D 6D Pose Solver for Robotic Manipulation
链接: https://arxiv.org/abs/2609.23566
作者: Yixuan Liang(1),William Chen(2),Yunan Wang(1),Jizhou Yan(1),Zhao Jin(1),Changling Liu(2),Chuxiong Hu(1) ((1) Tsinghua University, (2) Sapient Intelligence)
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: 9 pages, 5 figures. Corresponding author: Chuxiong Hu. Project page: this https URL
Abstract:6D object pose estimation is fundamental to robotic manipulation and automation. Recent zero-shot methods have significantly improved generalization to unseen objects, but most still rely on large-scale pretrained models with substantial GPU computation and memory demands. These requirements complicate deployment on robotic platforms where perception, planning, and control share limited computational resources, while learned intermediate representations offer limited geometric interpretability for task-specific adaptation. To address these limitations, we propose G6D, a learning-free, geometry-driven RGB-D 6D pose solver. Given an RGB-D observation, an object instance mask, camera intrinsics, and a CAD model, G6D generates pose hypotheses through template-based geometric matching and refines them using silhouette and depth consistency, forming a purely geometry-driven pose estimation paradigm. This paradigm requires neither pretrained visual models nor target-specific training and preserves interpretable geometric representations throughout pose estimation. Moreover, adjustable hypothesis counts provide flexible accuracy-computation trade-offs, while a CPU-only configuration supports deployment without GPU resources. Experiments on LineMOD and five BOP19 datasets demonstrate advanced performance. Real-world pick-and-place experiments further demonstrate G6D’s applicability to robotic manipulation. The complete project is publicly available at this https URL .
[CV-136] MaskVLA: Visual Masking Against Trajectory Overfitting of Vision-Language-Action Model IROS2026
链接: https://arxiv.org/abs/2609.23565
作者: Yuxuan Jiang,Jiaying Huang,Ge Wang,Shenhao Yan,Jiahao Yang,Chengsi Yao,Qi Liu,Qing Zhao,Shuguang Cui,Yiming Zhao,Yatong Han,Zhen Li
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注: 8 pages, 7 figures. Accepted to IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2026)
Abstract:Vision-Language-Action (VLA) models integrate vision-language understanding with executable robot actions, enabling end-to-end learning for robot control. However, our empirical analysis reveals that existing models exhibit severe trajectory overfitting when finetuned on limited datasets. To guide the model in effectively utilizing wrist camera information, we propose MaskVLA, a masking-based fine-tuning strategy. By randomly masking a small portion of the main camera’s visual information, the model is guided to autonomously learn more fine-grained, task-relevant, and effective visual features. This process leads to the emergence of robust policies, thereby enhancing the model’s capability to tackle complex manipulation tasks and improving its generalization performance. Our method has been comprehensively evaluated on RoboTwin 2.0, achieving an average success rate improvement of 23.2% and 16.8% compared to \pi_0 and OpenVLA-OFT, respectively. Furthermore, experiments on real-world ALOHA robots also demonstrate the effectiveness of our approach.
[CV-137] ransferring Visual Explanations: How Cross-Architecture Knowledge Distillation Affects Model Interpretability
链接: https://arxiv.org/abs/2609.23561
作者: Aleks Czufarow,Ihor Babin
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 21 pages, 4 figures, 2 tables. Submitted to AJOSR
Abstract:Deploying efficient neural networks is essential in resource-constrained environments, yet compact models often sacrifice interpretability - a critical in safety-critical domains such as autonomous driving and medicine. This study investigates whether Knowledge Distillation transfers the spatial feature attribution of a large teacher network to a compact student. To assess the influence of the KD scheme on interpretability, we distill a ResNet-152 teacher into a ResNet-34 student on ImageNet-1K across five configurations by systematically varying the distillation temperature and soft-label loss weight. Models are evaluated on top-1 accuracy, along with two interpretability metrics: Relevance Mass Accuracy and Relevance Rank Accuracy. These metrics are computed via Grad-CAM heatmaps benchmarked against ground-truth object masks. Our results show that top-1 accuracy ranges from 71.6% to 74.0%. For Grad-CAM, RMA ranges from 7.7% to 9.7% and RRA from 7.3% to 10.1%; for Guided Grad-CAM, RMA ranges from 16.1% to 18.6% and RRA from 15.9% to 21.5%. Interpretability proves far more sensitive to the soft-label weight than to the temperature: keeping the student anchored to hard labels preserves both accuracy and coarse localization, whereas weighting the teacher heavily degrades both. Fine-grained attribution, however, fell below the undistilled baseline in every configuration tested, indicating that logit distillation transmits where a model attends more readily than the pixel-level structure of that attention. We evaluate 12 cross-architecture combinations of convolutional and transformer-based models, revealing that the inheritance of fine-grained spatial reasoning is fundamentally bottlenecked by the student’s intrinsic structural biases. To our knowledge, this is the first application of this interpretability-aware evaluation framework - previously used for neural network pruning - to KD.
[CV-138] Modeling Clinical Workflow for SYNTAX Scoring from Coronary Angiography Videos
链接: https://arxiv.org/abs/2609.23553
作者: Suzhong Fu,Jingqi Dong,Xuan Ding,Rui Sun,Yiming Yang,Shuguang Cui,Zhen Li
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:The SYNTAX score is a clinically established tool for assessing anatomical lesion complexity in coronary artery disease and guiding subsequent treatment. However, automated SYNTAX scoring is commonly formulated as a direct regression problem from coronary angiography videos to patient-level scores. In this work, we reformulate SYNTAX scoring as a vessel segment identity-preserving anatomical reasoning problem and propose a hierarchical modeling framework that explicitly aligns learning with the clinical workflow. Our approach maintains vessel segment identity across frames and views, estimates stenosis severity at the segment level, and aggregates evidence hierarchically according to coronary anatomy. Simultaneously, to address the scarcity of domain-specific data, we integrate and complete multiple public coronary angiography datasets, constructing a large-scale resource featuring completed vessel segmentation and derived structural annotations. Experiments demonstrate that vessel segment-level stenosis embedding enhances explanatory power and reduces prediction variability compared to baseline models, with the R^2 score improving by 0.201 and dev STD decreasing by 18.4%. These results highlight the necessity of structure-aligned modeling for reliable and stable automated SYNTAX scoring from multi-view coronary angiography videos. The GitHub link is this https URL.
[CV-139] SewFusion: Tailored Generation of Topology and Panel-Level Geometry for Sewing Patterns
链接: https://arxiv.org/abs/2609.23548
作者: Jiaxin Lin,Xiao Pan,Hangjie Yuan,Luyan Liang,Wan Li,Daquan Feng
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Generating sewing patterns from images and text requires modeling a heterogeneous representation composed of discrete topology and continuous geometry. Existing methods mainly follow two paradigms: diffusion-based methods enable holistic geometry generation by converting the entire pattern into a continuous representation, but weaken discrete topology modeling; in contrast, autoregressive methods preserve discrete topology through next-token prediction, but tie continuous geometry regression to token-level hidden states with limited panel-level context. To bridge this gap, we propose SewFusion, a unified autoregressive framework that adopts tailored generation mechanisms for discrete topology and panel-level continuous geometry, using next-token prediction for the former and flow matching for the latter. To support panel-level continuous geometry generation, we introduce a Panel Geometry VAE that learns a fixed-size latent space for variable-length panel geometry, together with Panel Geometry Flow for latent generation. We further propose Panel-Forcing to reduce the training–inference mismatch in topology context and improve robustness to topology prediction errors. Extensive experiments on SewFactory and GCD-MM demonstrate that SewFusion consistently outperforms previous state-of-the-art methods across various settings, achieving +6.36% Panel Accuracy, +11.30% Stitch Accuracy, and -1.90 Vertex L2 error in the image-text-based generation setting.
[CV-140] owards robust multimodal 3D object detection via visual foundation models
链接: https://arxiv.org/abs/2609.23541
作者: Ziying Song,Lin Liu,Hongyu Pan,Shaoqing Xu,Lei Yang,Mingzhe Guo,Caiyan Jia
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: 27 pages, 5 figures. Bilingual English-Chinese manuscript; the complete English version appears first, followed by the complete Chinese version
Abstract:Multimodal 3D object detection is fundamental to robust perception in autonomous driving because it integrates complementary information from LiDAR and camera sensors. However, existing methods often fail to maintain robustness under out-of-distribution (OOD) corruptions caused by sensor noise, adverse weather, and environmental changes. To address this problem, we propose RoboDistill, a robust and generalizable multimodal 3D object detection framework that leverages visual foundation models (VFMs), such as the Segment Anything Model (SAM). First, we introduce SAM-AD, a domain-specific pretraining strategy that fine-tunes SAM on autonomous-driving imagery to extract feature representations with rich semantic information. Second, we design the AD Feature Pyramid Network (AD-FPN) to refine and upsample SAM features at multiple scales for seamless fusion with LiDAR features. Third, we develop the Depth-Guided Wavelet Attention (DGWA) module, which suppresses high-frequency sensor noise while preserving critical contextual information. Finally, we introduce KD Fusion, in which the pretrained SAM-AD serves as a teacher that distills high-quality visual knowledge into a lightweight point-cloud network, thereby improving robustness under noisy conditions. Extensive experiments across 27 challenging OOD corruption settings show that RoboDistill generally delivers stronger or competitive detection performance and robustness relative to representative state-of-the-art methods. This work bridges the gap between VFMs and 3D object detection and advances robust multimodal perception for real-world autonomous-driving applications.
[CV-141] PosEviLoc: Position-Conditioned Spatial Evidence for Language-Based 3D Localization
链接: https://arxiv.org/abs/2609.23534
作者: Tianyi Shang,Yike Shi,Zhenyu Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Language-based 3D localization retrieves the point-cloud submap containing a target position from descriptions of nearby objects and their spatial relations. Existing methods typically compress queries and submaps into global descriptors, potentially obscuring object-level semantics and cross-description spatial coherence. We propose Position-Conditioned Evidence Localization (PosEviLoc), a query-position-aware framework for coarse text-to-point-cloud localization. Instead of relying on global matching, PosEviLoc evaluates each candidate submap using explicit semantic and spatial evidence. It models direction as a relation jointly determined by an object position and a hypothetical query position. The resulting Query-Position Spatial Evidence Field (QSEF) measures the fraction of query descriptions supported at each hypothetical position, explicitly capturing their agreement without using the ground-truth query pose to construct the evidence field. A Multi-Level Evidence Readout (MER) summarizes this evidence in a compact representation, which a lightweight MLP converts into a retrieval score. Across five benchmarks, PosEviLoc outperforms MNCL by an average of 17 percentage points in Recall@1. When used as a plug-and-play reranker, it improves MNCL by an average of 16 percentage points. Moreover, PosEviLoc introduces substantially fewer parameters and achieves faster inference speed than existing methods.
[CV-142] GeoBalance: Geometry-Aware Monitoring and Reconstruction with Asymmetric Optimization for Balanced Multimodal Learning
链接: https://arxiv.org/abs/2609.23533
作者: Zechang Xiong,Da Li,Rong Yin,Kexin Tang,Biao Yang,Pengyuan Li,Wenkang Kong,Yulan Hu,Shengyu Zhu,Hao Peng
类目: Computer Vision and Pattern Recognition (cs.CV); Multimedia (cs.MM)
备注:
Abstract:Multimodal classifiers can converge to modality-dominant solutions in which one modality dominates the joint prediction, suppressing the learning of others. Existing balancing methods mainly adjust losses, gradients, or modality contributions, largely treating modality imbalance as an optimization problem while implicitly treating the weak modality as under-optimized but representationally intact. In this work, we find that this assumption does not always hold, as persistent modality dominance can induce a representation-level collapse of the weak modality, which we term \emphmanifold modality collapse (MMC). MMC manifests as a coupled geometric degradation in which weak-modality representations collapse onto fewer directions within each class and become less separable across classes. Motivated by this observation, we propose \emphGeoBalance, a geometry-aware framework that monitors these two geometric properties and reconstructs the weak modality representation only when it exhibits signs of MMC. Once triggered, GeoBalance uses a fixed Simplex-ETF class scaffold and spectral regularization to restore class separation while preventing collapse onto a few feature directions. To preserve reconstruction during joint training, asymmetric gradient projection removes the joint-gradient component conflicting with reconstruction, leaving non-conflicting optimization unchanged. Extensive experiments across six multimodal benchmarks demonstrate great improvements over competitive balancing methods, validating its effectiveness.
[CV-143] GARO: Geometry-Aware Redundancy Optimization for Real-Time and High-Fidelity Dynamic Gaussian Splatting ICRA2026
链接: https://arxiv.org/abs/2609.23509
作者: Huiwen Xue(School of Software, Northwestern Polytechnical University),Kaixing Zhao(School of Software, Northwestern Polytechnical University),Zuheng Ming(L2TI, Université Sorbonne Paris Nord, EmboMind Research),Tingcheng Li(School of Electronic Information and Engineering, Suzhou University of Science and Technology)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages. Accepted to IEEE ICRA 2026
Abstract:Novel view synthesis is a key task for dynamic scene reconstruction, where high rendering speed is essential for applications such as virtual reality. Existing deformable Gaussian Splatting methods achieve high-fidelity dynamic scene modeling, but still face limitations in memory usage and rendering efficiency due to the large number of redundant Gaussians. To address these challenges, we propose Geometry-Aware Redundancy Optimization (GARO), a unified redundancy measurement framework in the adaptive density control stage of the traditional dynamic scene reconstruction pipeline. This framework first selects low-gradient candidates using an optimization activity assessment strategy, and then evaluates geometric complexity through low curvature analysis to further filter and prune redundant points, resulting in a compact and expressive Gaussian representation. Extensive experiments on synthetic and real-world datasets demonstrate that GARO achieves robust trade-offs between quality and speed, with PSNR remaining stable and rendering speed improved by 2x, validating the efficiency and effectiveness of GARO.
[CV-144] Detecting Phone-Induced Pedestrian Distraction via a Multimodal Fusion Transformer
链接: https://arxiv.org/abs/2609.23507
作者: Yuanzhe Li,Hounian Liu,Xiaotong Chang,Yidi Huang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 2026 IEEE 9th International Conference on Electronic Information and Communication Technology (ICEICT 2026)
Abstract:The increasing reliance on mobile phones has made phone-induced pedestrian distraction increasingly prevalent. Activities such as texting, watching videos, and making phone calls have become significant contributors to traffic accidents. Reliable detection of pedestrian distraction is essential for autonomous vehicles, as it improves situational awareness and enables timely risk assessment, thereby supporting safe motion planning and vehicle control. We propose a multimodal fusion Transformer (MFT) for detecting phone-induced pedestrian distraction. MFT jointly extracts skeletal dynamics from body pose keypoints and visual appearance features from pedestrian images, effectively leveraging the complementary information provided by the two modalities. A cross-modal attention module is proposed to capture inter-modal dependencies through multi-head cross-attention, facilitating effective fusion of complementary information across the two modalities. Then, a temporal attention fusion module, implemented with a Transformer encoder, is employed to capture temporal dependencies. MFT is trained and evaluated on a manually annotated dataset comprising 287 pedestrian instances with 20,741 images. Extensive experiments demonstrate that MFT attains an overall accuracy of 95%, exceeding the performance of six baseline approaches by 6%.
[CV-145] Pay More Attention To Text In High-Resolution MLLM s
链接: https://arxiv.org/abs/2609.23495
作者: Zhongkuan Mao,Wenzhuo Zhao,Xianjie Liu,Yidong Wang,Zhao Gao,Ronghao Xian,Yao Jiang,Yi Zhang,Liangjian Wen,Keren Fu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Failures of high-resolution MLLMs are commonly attributed to a visual problem, motivating zooming, cropping, and related visual interventions to recover fine-grained evidence or suppress interference. Yet recent studies suggest that relevant visual evidence is already encoded in intermediate representations, indicating that visual-side improvements alone insufficient. This raises a natural question: does the remaining bottleneck lie in the text that guides visual search? We identify a previously overlooked linguistic bottleneck: questions formulated for answering do not necessarily specify the visual evidence required for localization. To address this mismatch, we introduce EviSpec, a training-free compiler that derives complementary evidence specifications while preserving the original question for final reasoning. We further validate it through matched-control experiments that isolate the roles of evidence specification and localization. With the search budget fixed, structured evidence specifications yield an 8.6% relative gain over generic requests. With evidence geometry matched, the evidence localized by EviSpec yields a 14.8% relative gain over random evidence. Together, these controls isolate the benefit of specifying what evidence to seek rather than merely expanding visual access. Across all five MLLMs, EviSpec consistently improves upon the corresponding baseline on each of the three benchmarks, yielding average relative gains of \textbf10.4%, 8.8%, and 12.4% on V\textsuperscript*Bench, HR-Bench-4K, and HR-Bench-8K, respectively. Beyond high-resolution reasoning, EviSpec also achieves state-of-the-art performance on VQA and hallucination-focused benchmarks.
[CV-146] CE4L: Continual Ego Exo and Ego-Exo Learning ICML2026
链接: https://arxiv.org/abs/2609.23492
作者: Hongwei Yan,Kanglei Zhou,Yuchen Liu,Qingyu Shi,Yi Zhong,Liyuan Wang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 23 pages. Accepted by ICML 2026
Abstract:Perception for embodied agents is video-based, often multi-view (ego, exo, or both), and inherently continual, with simultaneous task and viewpoint shifts. Yet continual learning (CL) remains dominated by exo-only recognition tasks, obscuring behavior under these real-world coupled shifts. We introduce Continual Ego, Exo, and Ego-Exo Learning (CE ^4 L), a unified multi-view CL benchmark spanning four representative tasks: cross-view referenced skill assessment, temporal action segmentation, cross-view association, and action anticipation planning. CE ^4 L highlights challenges largely absent in prior CL benchmarks, including cross-view correspondence, view-dependent asynchrony, and heterogeneous semantic objectives. To this end, we propose Video Incremental Subspace-routed Task Adapters (VISTA), a parameter-efficient baseline method that stores task-specific updates in lightweight adapters and performs training-free routing via residual distance to task-specific whitened subspaces estimated from second-order statistics. Extensive experiments demonstrate the significantly varied efficacy of representative CL methods across CE ^4 L settings, while VISTA is consistently competitive and achieves state-of-the-art overall performance. Our source code for benchmarks and methods is available at this https URL .
[CV-147] Cognitive Action Reasoning for Proactive Robots from Human-Centered Multimodal Observations
链接: https://arxiv.org/abs/2609.23486
作者: Zhihao Gu,Kechao Zhu,Yuanfeng Wu,Mohan Liu,Ankit Kumar Shaw,ChenDong Hong,Xuanyu Chen,Dengchen Mei,Xu Tianyi,Lin Wang
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Robots operating in human-centered environments are typically designed to execute explicit instructions, and most robot-learning datasets likewise pair observations with task instructions or low-level actions. Although recent work has begun to explore proactive embodied assistance, existing resources target different settings and action levels, leaving real-world human-centered multimodal decision-making underexplored. We formulate this problem as \textitProactive Robot Action Reasoning (\textitProRobo), an upstream cognitive decision problem in which a robot must determine which action to take based on multimodal human and environmental cues without explicit action instructions. To support ProRobo, we introduce \textitProAction, a real-world multimodal dataset containing 10K samples of visual observations, audio signals, and text inputs across 12 daily-life scenarios in five common scenes. To construct cognitively grounded high-level action supervision, we develop a two-stage human-in-the-loop pipeline that combines appraisal-guided candidate generation with Affective Theory-of-Mind-guided human refinement, explicitly incorporating contextual judgment about human states, urgency, feasibility, and potential risk into action annotation. Based on this supervision, we benchmark representative Multimodal Large Language Models (MLLMs) and introduce \textitMMC2Act, a reference model that implicitly learns the mapping from multimodal observations to cognitively grounded high-level actions. Experiments across modality settings, subject-disjoint generalization, cross-dataset transfer, and human evaluation show that general-purpose MLLMs struggle with proactively reasoning high-level actions from multimodal cues, whereas training on \textitProAction substantially improves performance.
[CV-148] Algebraic Consistency Alone Does Not Certify Temporal Structure in Latent Action Models
链接: https://arxiv.org/abs/2609.23478
作者: Di Wen,Ruodi Zhang,Kailun Yang,Kunyu Peng
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: 9 pages, 2 figures, 5 tables
Abstract:Latent action models infer a code for the transition between two frames of action-free video. Recent methods regularise this code to compose additively and reverse antisymmetrically, and report order-of-magnitude reductions in the resulting errors as a label-free certificate that the code has captured temporal structure. We show that this conclusion does not follow. Reconstruction drives the decoded transition toward a difference of state features, for which both identities hold for any pairing, a solution the metric cannot distinguish from one encoding nuisance state or a coordinate convention. Across five source domains, a trained but unconstrained counterpart already achieves 83-97% of the reduction relative to an untrained anchor. The residual fold is governed as much by the decoder family as by what is learned. A constrained model retrained after its temporal pairing is destroyed still reaches, in each domain, a lower error than the unconstrained model on real data. Downstream, preserving the temporal pairing yields no consistent advantage on LIBERO-GOAL or LIBERO-SPATIAL, and across the tested arms the code’s mean linear action decodability falls as the algebraic error improves. We also test the most direct repair, a violation-contrastive objective that requires the algebra to fail on destroyed pairings: in the tested configurations it yields only a marginal separation within the reconstruction budget, on training and test triples alike. We recommend a validation protocol that these methods currently lack: a baseline-corrected metric, retraining on destroyed pairings, and a seed-budget analysis.
[CV-149] Rule-Constrained Assignment for Cue-Ball Identification in Broadcast Snooker
链接: https://arxiv.org/abs/2609.23450
作者: Yuxin Cao,Wei Song,Yuezhong Wu,Jin Song Dong
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 5 pages
Abstract:Accurate cue-ball identification is essential for metric analysis of broadcast snooker. Existing systems evaluate each candidate independently against a fixed white prototype and reject candidates above an appearance threshold. Under broadcast conditions, illumination changes can make colored balls appear white, while intrusions from players and equipment can obscure the cue ball or introduce competing candidates. We formulate cue-ball identification as a rule-constrained assignment problem that jointly assigns detected candidates to the bounded snooker inventory: one cue ball, up to 15 reds, and six colors with known spots. The cue ball is selected by the incremental cost of assigning each candidate to the white slot, and the conventional appearance test follows as the one-slot case. On 419 hand-annotated shots, our method improves identity accuracy from 88.1% to 95.5%, and from 80.5% to 95.2% on held-out venues. Within CueLift, our metric state-recovery system, the assignment expands coverage from 36.2% to 55.4% over 6,241 scorable shots. Assigning an estimate to every shot in a separate evaluation on 2,529 shots preserves this advantage.
[CV-150] PhysReflect: Geometry and Perception Guided Diffusion for Physically-Plausible Mirror Reflections
链接: https://arxiv.org/abs/2609.23442
作者: Shuheng Ge,Hongwei Ren,Li Zhang,Xiangqian Wu
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Diffusion models generate high-quality images, yet often violate the physical laws governing mirror reflections. Reflections often suffer from geometric aberrations, including positional offsets, directional misalignment, proportional imbalance, and structural distortion. These failures remain evident even in contemporary state-of-the-art generative systems. Existing methods itigate this problem through synthetic data scaling or auxiliary depth conditioning, yet their merely reliance on latent-space noise reconstruction losses as implicit supervision prevents direct enforcement of reflection-specific geometric and perceptual constraints. To bridge this gap, we present PhysReflect, a geometry and perception guided diffusion framework that decodes the predicted clean latent into pixel space at each training step and applies annealed supervision through two complementary differentiable objectives. The Geometric Loss enforces mirror-induced spatial consistency through sparse epipolar correspondence and dense boundary projection alignment, where a SAM2-based TwinTrack mechanism provides stable in-mirror localization for boundary-aware supervision. The Perceptual Loss preserves reflected appearance by combining Semantic Consistency Loss, which maintains reflected identity and appearance via DINOv2 features, and Lighting Consistency Loss, which regularizes depth, surface-normal, and illumination coherence under monocular geometry priors. Experiments on synthetic and real-world benchmarks show that PhysReflect outperforms prior mirror-reflection methods in geometric, perceptual, and physical-plausibility metrics, as well as qualitative visual results.
[CV-151] GAPS: Generative Active Pseudo-view Selection for Sparse-View 3D Gaussian Splatting
链接: https://arxiv.org/abs/2609.23436
作者: Hongfei Zhu,Haochen Deng,Sitao Zhang,Ling Zhou
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Novel view synthesis from sparse observations is severely under-constrained. Although 3D Gaussian Splatting (3DGS) enables real-time rendering, it produces floaters, broken geometry, and washed-out backgrounds when trained with few views. We propose an alternating optimization framework that uses a pre-trained image diffusion model to generate geometrically consistent pseudo-views for additional 3DGS supervision. Generation is constrained by depth-conditioned ControlNet, IP-Adapter style transfer, LoRA scene adaptation, and img2img structural anchoring. We introduce Generative Active Pseudo-view Selection (GAPS) to balance reconstruction informativeness and generative reliability when choosing target views. Its annealing schedule shifts from conservative interpolation early in training to exploratory extrapolation later, gradually covering unobserved regions. A dual-criterion admission gate and uncertainty-weighted losses reject unreliable generations, while density-adaptive DropGaussian reduces overfitting in complex scenes. On LLFF with 3/6/9 views, our method improves average PSNR over vanilla 3DGS by 0.40/0.89/0.70 dB. On Mip-NeRF 360 with 12/24 views, the gains are 1.18/0.80 dB. SSIM improves and LPIPS decreases in every setting. Ablations show that active selection and density-adaptive regularization are both necessary; only the full method reduces LPIPS below the no-pseudo-view baseline on unbounded 360-degree scenes.
[CV-152] HOIBlender: Blending Lightweight Detection with Vision-Language Priors for Efficient Human-Object Interaction Detection
链接: https://arxiv.org/abs/2609.23431
作者: Junwen Chen,Keiji Yanai
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Human-object interaction (HOI) detection requires grounding an interacting human-object pair and recognizing the verb that links them, often under severe long-tail supervision. Recent methods improve accuracy with stronger detectors and vision-language priors, but many still stack heavy transformer encoders, intricate denoising schedules, or post-hoc semantic calibration on top of the detector. We present \textbfHOIBlender, an efficient HOI detector named after its core design principle: blending detector-grounded visual tokens, spatial subject-object reasoning, and BLIP-2 semantic priors inside one lightweight decoding pipeline. HOIBlender builds on an RF-DETR/LW-DETR-style foundation with a DINOv2 backbone and selects top- K image-conditioned tokens directly from the multi-scale projector as subject and object candidates, removing the dedicated encoder stage retained by prior HOI methods. A dual-stage decoder first stabilizes human-object geometry and then performs verb and HOI classification through progressive BLIP-2 prior fusion, with classifier weights initialized from BLIP-2 text embeddings for long-tail categories. Grouped-query training further enriches optimization without increasing inference cost. Across three model scales (Nano, Small, 2XL), HOIBlender consistently outperforms SOV-STG-VLA and Hybrid-SOV-VLA on HICO-DET, reaching 44.49 Default Full mAP in only 9 training epochs while maintaining competitive latency and parameter budgets. These results show that lightweight detection, structured spatial-semantic decoding, and deeply integrated vision-language priors can be blended into a single efficient HOI pipeline.
[CV-153] RSPDBench: Benchmarking Vision Foundation Models on Earth Observation Tasks Under Physically Grounded Remote-Sensing Product Degradations
链接: https://arxiv.org/abs/2609.23427
作者: Tanjim Bin Faruk,Khondaker Masfiq Reza,Shrideep Pallickara,Sangmi Lee Pallickara
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Vision foundation models targeting Earth observation (EO) tasks are commonly evaluated on clean downstream benchmarks, but operational EO products can already contain spatial, radiometric, alignment, noise, and harmonization defects before reaching the model. Existing robustness evaluations often use generic image corruptions or broad domain shifts, which do not isolate these product-level failure modes. We introduce \textbfRSPDBench, a physically grounded \textbfremote-\textbfsensing-\textbfproduct \textbfdegradation \textbfbenchmark for vision foundation models. RSPDBench evaluates five EO datasets, seven foundation-model entries, and two supervised baselines under audited primitive degradations and compound product chains. Each model is evaluated under its clean-selected native protocol, with robustness measured as the drop from its own clean baseline. Our analysis reveals that degradation sensitivity is strongly structured: resolution-conditioned and channel-grouped encoders protect different failure axes, and the same physical defect can hurt one model while helping another. Compound chains expose failures that isolated degradations do not predict, with model-dependent amplification, saturation, or component dominance, and excess drops up to 38 percentage points beyond the strongest component. These results show that EO robustness cannot be characterized by clean accuracy or generic perturbation tests alone; it must also be measured against the structured defects that remote-sensing products carry into deployment.
[CV-154] Semi-automated reconstruction of indoor geometry from 360-degree video for CFD-based airflow analysis in classrooms
链接: https://arxiv.org/abs/2609.23425
作者: Dhruv Gamdha,James Afful,Shambhavi Joshi,Ulrike Passe,Adarsh Krishnamurthy,Baskar Ganapathysubramanian
类目: Computer Vision and Pattern Recognition (cs.CV); Computational Engineering, Finance, and Science (cs.CE); Fluid Dynamics (physics.flu-dyn)
备注: 42 pages, 24 figures, 12 tables
Abstract:Computational Fluid Dynamics (CFD) is widely used to evaluate ventilation and contaminant transport in occupied buildings, but deployment at scale is limited by three bottlenecks: acquiring room geometry without costly scanning hardware or manual CAD modeling, decomposing the scene into individually manipulable objects, and reconfiguring those objects for alternative layouts without re-capturing the room. We present a semi-automated workflow that converts a single 360-degree video of a room into individually editable, simulation-ready geometry assets. A dense point cloud is reconstructed using Neural Radiance Fields (NeRF), and 2D instance masks from text-prompted SAM 3 segmentation are lifted to 3D using multi-view consensus and depth-band filtering. Points are separated into object instances with an octree, and occlusion gaps are healed with a connectivity graph. Chair templates are fitted by Iterative Closest Point (ICP) alignment, and table geometry is generated procedurally. A browser-based editor supports quality assurance and rapid construction of alternative layout configurations. A steady Reynolds-averaged OpenFOAM solution then drives transient passive-scalar transport; the setup is verified using a mesh-sensitivity study and validated against an IEA Annex 20 benchmark. We apply the workflow to two university classrooms and a tiered lecture-hall auditorium. The capture-to-geometry pass takes two to five hours per room on a consumer workstation. In a controlled obstruction sequence in one classroom, the modeled half-clearance time varies non-monotonically as furniture is added, and a cross-room comparison indicates that clearance behavior cannot be reliably extrapolated between rooms, motivating per-room geometry acquisition. By making that acquisition low-cost, the workflow makes geometry-resolved comparative ventilation studies practical for spaces such as classrooms.
[CV-155] Omni2Web: Benchmarking Audiovisual Website Development
链接: https://arxiv.org/abs/2609.23417
作者: Minghao Han,Zhenghao Xing,Xize Cheng,Yuxuan Wang,Junming Lin,Ling Wang,Yinsong Yan,Yunfei Chu,Qize Yang,Jin Xu
类目: oftware Engineering (cs.SE); Computer Vision and Pattern Recognition (cs.CV)
备注: 32 pages, 6 figures, 18 tables
Abstract:Screen-recorded web editing requests contain weak deictic expressions such as this'' and there,‘’ whose referents depend on speech, cursor trajectories, page state, and edit history. Such requests require intent recovery beyond the explicit specifications assumed by many existing web-editing benchmarks. We introduce Omni2Web, a bilingual benchmark of 918 instances spanning 13,907 edit steps. It defines three complementary tracks: Direct Editing evaluates webpage editing from recordings, Instruction Recovery measures explicit intent recovery, and Instruction Utility tests whether recovered instructions can drive a fixed code executor. We evaluate 17 open- and closed-source models. The best models attain 51.17 on the Edit Fidelity Score (EFS) for Direct Editing and 49.14 on the Instruction Recovery Score (IRS); under the fixed executor, the strongest recovered instructions reach 51.08 EFS, still far below the 89.69 EFS obtained with oracle instructions. Step-level analyses show that correct grounding does not guarantee successful edits, while some Omni models recover instructions that the fixed coding model executes substantially better than their direct edits. Controlled ablations further demonstrate the value of temporally aligned audiovisual evidence, while alternative judges preserve the leader and broad ordering. Together, these findings reveal substantial headroom in multimodal intent recovery and code execution and highlight the promise of pairing Omni rewriters with coding models.
[CV-156] Retrieval Geometry Shapes Cache-Based Clip Adaptation ICLR
链接: https://arxiv.org/abs/2609.23409
作者: Mahir Shahriar Tamim,Md. Samiul Alim,Azmine Toushik Wasi,Shahriyar Zaman Ridoy,Meharun Nesa,Mohammad Abu Yousuf,Alex Lamb,Mohammad Ali Moni
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Under Review at ICLR
Abstract:Cache-based test-time adaptation improves CLIP predictions by storing and retrieving examples from the target stream while keeping the model frozen. However, existing methods largely treat the feature space used for image-image retrieval as fixed, leaving open how much adaptation depends on the retrieval space itself. We study this question by fixing the memory and changing only the retrieval encoder, finding that the same memory can yield very different gains: across sixteen retrieval spaces, ImageNet-A cache gain ranges from at most +0.44 points for CLIP and MAE to +19.7 +/- 0.4 for DINOv2-L, while label-free retrieval-space selection retains 98% of oracle gain on ImageNet-V2. These results show that memory quality depends not only on which examples are stored, but also on how they are retrieved. Motivated by this finding, we propose MARC (Memory Augmented Retrieval for CLIP), a training-free system that uses frozen CLIP for prediction and DINOv2-B for retrieval with a single fusion weight. A single-view cache repairs 1074 +/- 21 baseline errors, compared with 878 +/- 4 for a 64-view ensemble, at roughly one seventh of the cost. Across four ImageNet distribution shifts, MARC reaches a 67.91% OOD average and, at matched DINOv2-B scale and eight views, achieves 64.17 +/- 0.31% versus 62.75 +/- 0.15% for a graph-based cache system while running 2.6 times faster. Overall, our results establish retrieval space as a first-order design choice for robust cache-based adaptation in remote sensing, scientific imaging, and changing visual environments.
[CV-157] Accurate Motion Estimation with Bézier Control Point for Efficient Frame Interpolation
链接: https://arxiv.org/abs/2609.23408
作者: Shuhao Han,Chenyang Wu,Chun-Le Guo,Zheng-Peng Duan,Zhen Li,Ming-Ming Cheng,Chongyi Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted by IEEE TIP. Code: this https URL
Abstract:In frame interpolation tasks, motion ambiguity in the training set causes models to generate blurry intermediate frames. Moreover, the assumption of uniform motion between frames during inference further leads to inaccuracies in the generated intermediate frames. To tackle these challenges, we propose an Accurate motion estimation algorithm with Bézier Control point, ABC-Inter, for efficient frame Interpolation. Specifically, ABC-Inter designs an Accurate Flow estimation Module (AFM) by decoupling two-frame features and mapping to corresponding coordinates to better estimate the optical flow between the two frames. Furthermore, ABC-Inter eliminates motion ambiguity in the training set by introducing Bézier control points that are computed using the input frames and the intermediate ground-truth (gt) frames. This allows the model to estimate accurate optical flow between two frames during the training process, thereby solving the blurriness problem in the generated intermediate frames during inference. Benefiting from the more accurate flow estimation between two frames, we can introduce additional frames and directly use multiple flows to calculate Bézier control points for modeling non-uniform motion without retraining the model. Simultaneously, to realize the estimation of non-linear motion using only two frames, we also introduce a new Bézier control point estimation module which achieves better motion estimation between the two frames by performing fine-tuning on the model in the second stage. Experimental results demonstrate that our ABC-Inter achieves state-of-the-art performance on multiple benchmark datasets and exhibits excellent visual perception.
[CV-158] ScaleBlind: Point Cloud Completion under Unknown Scale
链接: https://arxiv.org/abs/2609.23404
作者: Shenghui Wu,Chen Wang,Yuan Feng,Guangshun Wei,Yuanfeng Zhou,Changjian Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Point cloud completion aims to infer a complete 3D shape from a partial point cloud and serves as a fundamental building block for downstream tasks such as reconstruction, editing, and simulation. Despite the recent progress, existing learning-based methods often implicitly rely on access to the ground-truth shape scale (GT-scale) during both training- and testing-time normalization, assuming privileged information that is unavailable in real-world inference. This hidden assumption limits practical deployment and can lead to severe completion artifacts, e.g., over- or under-completion and nested shells, once the oracle GT-scale cue is removed. We observe that the recent foundation image generation models exhibit a strong capability of understanding objects and geometries, and producing multi-view consistent renderings, making them promising priors for GT-scale-free 3D completion. Motivated by this insight, we propose ScaleBlind, a novel framework that leverages foundation-model-based image completion to recover global scale directly from partial inputs and then faithfully produces the 3D completion. Specifically, ScaleBlind dreams out complete multi-view appearances from rendered partial views, lifts the inferred missing regions back into 3D to obtain a geometry-aware coarse completion, and further refines it via a powerful cross-modal fusion network with the original partial point cloud. By harnessing 2D foundation priors, our method eliminates the need for accessing GT-scale information at inference. Moreover, it provides a principled bridge between 2D generative priors and 3D point cloud completion. Extensive experiments demonstrate the superiority of our framework, making ScaleBlind the new state-of-the-art for the point cloud completion task.
[CV-159] Enhancing Shrimp Disease Detection via Deep Learning and Data Refinement for Resilient Aquaculture ICMR2025
链接: https://arxiv.org/abs/2609.23397
作者: Vinh Canh-Thanh Truong,Hai-Binh Pham,Ngoc Hong Tran
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Emerging Technologies (cs.ET)
备注: This paper has been accepted at the International Conference on Multidisciplinary Research (ICMR 2025)
Abstract:Shrimp diseases continue to cause devastating losses in the aquaculture industry, driving a critical need for robust, automated detection. This work contributes the first application of Vision Transformers (ViT) and Self-Supervised Learning (SSL) to the shrimp farming domain, addressing both performance bottlenecks and data labeling challenges. We propose two deep learning pipelines to classify four key diseases: Healthy, Black Gill (BG), White Spot Syndrome Virus (WSSV), and a co-infection of both using a dataset of 4,348 images. First, our supervised transfer-learning approach leverages ImageNet-pretrained ViT-Small/16 and EfficientNet backbones. Second, we introduce a contrastive learning framework (SimCLR) with a ViT-Small encoder to extract robust representations from unlabeled images prior to fine-tuning. Our results establish strong new baselines for sustainable aquaculture monitoring. The supervised approach achieves an outstanding 96% accuracy with fast convergence, outperforming traditional generic models, while the label-efficient SSL approach reaches a highly competitive 85% validation accuracy.
[CV-160] ProxyBuild: Text-Guided Structured 3D Building Generation with Mesh-Anchored Procedural Proxies
链接: https://arxiv.org/abs/2609.23386
作者: Xiang Tang,Ruotong Li,Xiaopeng Fan
类目: Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR)
备注: 15 pages, 9 figures
Abstract:Text-guided 3D building generation holds tremendous application potential, yet existing generative models typically output inseparable single meshes or non-interactive rendered representations. While procedural modeling can generate editable buildings with hierarchical structures, rule authoring is laborious, and even with the aid of large language models (LLMs), it remains challenging to effectively solve procedural rules under geometric constraints. In this paper, we propose ProxyBuild, a novel hybrid framework for structured building generation. We introduce the Mesh-Anchored Procedural Proxy (MAPP) as a novel intermediate representation, which tightly anchors building components onto geometric shells, thereby decoupling the generation task into two phases: proxy prediction and proxy-to-asset instantiation. First, we construct a building dataset with MAPP annotations to train our designed face-edge bigraph encoder. By explicitly modeling the feature interactions of topological elements on heterogeneous mesh graphs, this encoder accurately infers the semantic roles of faces and edges. Subsequently, conditioned on textual styles and attribute parameters parsed by LLMs, we accomplish high-precision asset retrieval and assembly by integrating a spatial placement logic with hard constraints. Extensive experiments show that ProxyBuild not only significantly mitigates common issues in building generation such as over-smoothing, component collisions, and structural corruptions, but also accurately parses semantic-free shells from diverse sources. Outperforming prior baselines across various metrics, our method can robustly generate structurally clear, detail-rich, and post-editable 3D buildings from text, thereby providing a reliable and interactive content foundation for downstream applications such as virtual reality and digital twins.
[CV-161] LiteTex-GS: Fast and Lightweight Texturing for Gaussian Splatting
链接: https://arxiv.org/abs/2609.23380
作者: Zhiwei Li,Yijia Guo,Yishi Lu,Liwen Hu,Hong Rao,Shengbo Chen,Lei Ma
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 11 pages, 6 figures. Accepted to Pacific Graphics 2026 (Journal Track)
Abstract:Gaussian Splatting has enabled real-time novel view synthesis, but its tightly coupled geometry and appearance representation often require a large number of primitives to reproduce high-frequency texture details, leading to substantial memory and optimization costs. Recent textured 2D Gaussian methods alleviate this limitation by attaching texture maps to Gaussian primitives. However, bridging the fundamental structural gap between discrete Gaussians and continuous 2D grids requires complex parameterizations that introduce severe computational overhead. This overhead fundamentally compromises the original efficiency of Gaussian Splatting, making the balance between detailed texturing and computational agility an unresolved challenge. To address these challenges, we propose LiteTex-GS, a fast and lightweight texturing framework for Gaussian Splatting. Our method initializes an extremely compact representation, assigning minimal local texture to each Gaussian and progressively allocates higher resolution only to primitives with significant reconstruction errors. To maintain a streamlined geometric scaffold, we introduce a contribution- and area-aware pruning strategy that eliminates low-utility Gaussians. Furthermore, to mitigate the gradient dilution caused by texture upsampling, we design a resolution-aware update rule that preserves rapid and stable convergence. Extensive experiments on standard novel view synthesis benchmarks demonstrate that our method achieves competitive or superior rendering quality while using substantially fewer parameters and less training time than existing textured Gaussian baselines.
[CV-162] If You Hear It Help Find It: Orthogonal Knowledge Distillation for Open-Vocabulary Audio-Visual Event Localization
链接: https://arxiv.org/abs/2609.23376
作者: Yi Xu,Cheng Chen,Wenzhuo Lei
类目: Multimedia (cs.MM); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Sound (cs.SD)
备注: Accepted to ACM Multimedia 2026 (poster). 9 pages, 5 figures
Abstract:Open-vocabulary audio-visual event localization (OV-AVEL) grounds a text-queried event in time from video, audio, and language. The supervision sources available to this task can differ in temporal-boundary reliability: on OV-AVEBench, our configured visual teacher gives more reliable boundary cues than the configured audio teacher, although the latter is a strong pretrained audio model and remains semantically informative. This is a setting-specific diagnostic rather than a universal ranking of vision and audio. We formulate the resulting challenge as supervision placement: which teacher signals may shape the localization decision, and which should remain auxiliary. Based on this view, we propose OV-OrthKD, a reliability-aware asymmetric distillation framework. Visual feature transfer shapes a decision-aligned representation, audio feature transfer enriches a complementary auxiliary subspace, a text prototype anchors seen/unseen category semantics, and an orthogonality loss limits directional overlap between the two teacher-specific projections. The student continues to use both modalities through query-aware fusion at inference, while the default training recipe keeps audio-teacher supervision off the segment-logit path. On OV-AVEBench, OV-OrthKD achieves 0.816 segment AP and improves F1@0.5 over the official fine-tuning baseline by 2.7 points overall and 3.4 points on unseen categories. Path-assignment, role-swap, corruption, and transfer analyses consistently support supervision placement as a task-specific design axis for OV-AVEL.
[CV-163] Vision-Wireless Fusion for Multi-User Localization: A Cross-Modal Transformer Approach
链接: https://arxiv.org/abs/2609.23372
作者: Can Zheng,Jiguang He,Guofa Cai,Henk Wymeersch,Merouane Debbah
类目: Computer Vision and Pattern Recognition (cs.CV); Signal Processing (eess.SP)
备注: 12 pages, 8 figures, 7 tables
Abstract:Accurate multi-user localization is challenging in complex urban environments, where wireless measurements can become ambiguous under noise, blockage, and multipath, while visual observations provide complementary spatial context. This paper presents a vision-wireless fusion framework for multi-user localization using pilot-indexed channel state information (CSI). Orthogonal pilot indices preserve the identities of the communicating UEs in the CSI token sequence and localization outputs. The model encodes each pilot-indexed CSI observation as a query token and uses cross-attention to retrieve user-specific information from spatial visual memory. Self-attention among CSI tokens further captures inter-user interactions, while the resulting multimodal representations are used for user-wise localization. Experiments on different datasets show consistent improvements over model-based, CSI-only, and multimodal-fusion baselines. Further experiments evaluate the model under different wireless and visual conditions.
[CV-164] he Right Future for Action: Learning Action-Relevant Predictive States in World Action Models
链接: https://arxiv.org/abs/2609.23369
作者: Qiwen Gu,Jifan Li,Bingjie Gao,Rui Chen,Jing Tang,Xiangxiang Chu,Junqiao Zhao
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 19 pages, 5 figures, 8 tables
Abstract:Generation-free world action models (WAMs) retain future-video prediction during training but act from internal video features at inference, leaving unclear what these features should preserve for control. Our representation diagnostics show that representations with more predictable future changes need not make linear action decoding easier. Observed future changes provide additional action information beyond the present, and linearly readable action information is spatially concentrated. These findings motivate Action-Relevant Predictive States (ARPS), a compact predictive interface between the video and action experts. ARPS uses a horizon-conditioned state predictor to aggregate intermediate video features into a compact state that supplies all visual context to the action expert. Future-representation supervision trains different parts of this state to predict visual representations at different future times, together with their changes relative to the present. At inference, the supervision branch is removed, and the action expert only uses the learned predictive state computed from current observations. Controlled ablations show that future supervision substantially improves generalization under distribution shift. ARPS achieves 99.2% success on LIBERO and transfers to LIBERO-Plus without adaptation, reaching 87.3% and exceeding Fast-WAM by 39.2 percentage points.
[CV-165] BiView-Touch: Learning Bimanual Tactile Representations by Cross-Hand Completion ICRA2027
链接: https://arxiv.org/abs/2609.23352
作者: Chenxin Liang,Youchen Lai,Chuqiao Lyu,Tianxing Chen,Shoujie Li,Wenbo Ding
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: Submitted to ICRA 2027; 9 pages, 7 figures
Abstract:Bimanual interaction produces complementary tactile views of the same physical process, yet existing tactile representation learning largely models the two hands independently or combines them only for downstream prediction, leaving their cross-hand relationship unexplored. To exploit this overlooked structure, we introduce BiView-Touch, a tactile-only framework that completes masked target-hand latents from the remaining visible target-hand regions and the synchronized full contralateral hand. A student encoder with a geometry-conditioned directional decoder predicts full-view EMA latent targets, while temporal and layout counterfactuals encourage sensitivity to synchronized and anatomically organized source information. Controlled ablations and source-context interventions show that BiView-Touch learns structured cross-hand dependence on temporally aligned and anatomically organized contralateral tactile context, rather than benefiting from bilateral input alone. On the public HumanTouch dataset, its frozen representations consistently outperform representative self-supervised baselines across low-label settings. With only 5% downstream labels, BiView-Touch achieves relative balanced-accuracy gains of 7.1% on bilateral wrist-motion recognition and 14.1% on force-derived interaction-phase recognition. We further introduce BVT-20, a 20-task bilateral tactile dataset, and demonstrate transfer across recording sessions and pretraining corpora, including transfer to a held-out bimanual task. Our code and dataset details are available on the anonymous project page: this https URL.
[CV-166] AniPrO: Interpretable Anime Image Provenance Detection via Multi-Dimensional Semantic Reasoning
链接: https://arxiv.org/abs/2609.23345
作者: Yan Liu,Baoxiang Huang,Zi’an Wang,Wenbo Xie
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at the Computer Graphics International (CGI) 2026, 2 figures, 6 tables
Abstract:As generative AI becomes increasingly used in anime-style image creation, distinguishing human-drawn, AI-inpainted, and text-to-image images is important for copyright attribution, visual provenance, and content governance. Existing AI-generated image detectors mainly target real-world photographs and often overlook anime-specific cues such as flat coloring, exaggerated structures, and artistic line control. To address this gap, we propose AniPrO, a multi-dimensional description-enhanced framework for interpretable anime image provenance. Built upon AnimeDL-2M, AniPrO contains 15,000 balanced samples from a 35,000-image candidate pool, covering Real, Inpainting, and Text2Image categories with structured five-dimensional descriptions. We further introduce AniPrO-SFD-Bench and AniPrO-MFR-Bench to evaluate provenance detection from statistical feature discrimination and multimodal fusion reasoning perspectives. Experiments show that structured semantic guidance reveals systematic AI-generation biases, such as the gap between global visual plausibility and local detail coherence, and improves the detection of challenging inpainting samples. The dataset and code will be released at: this https URL.
[CV-167] MinCU: A Fine-Grained Benchmark for Grounded Minimal-Change Understanding in Image Pairs
链接: https://arxiv.org/abs/2609.23336
作者: Chaoqian Mu,Wenhao Wu,Zichen Liang,Jiaxu Li,Lijun Wang,Yifan Wang,Huchuan Lu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 25 pages, 10 figures, 15 tables
Abstract:Localizing and describing fine-grained differences between near-identical images is a critical yet underexplored capability for multimodal large language models (MLLMs). Existing benchmarks largely assess semantic comparison or single-image grounding in isolation, without jointly requiring faithful description and physical localization. To bridge this gap, we introduce MinCU, a benchmark for grounded minimal-change understanding, where each sample consists of an image pair differing by a single atomic variation in object category, attribute, count, or spatial position, and models are evaluated on their ability to describe the change, localize the changed regions, and identify the changed entity. We further propose Semantic-Guided Implicit Spatial Anchors (SG-ISA), a structured autoregressive method that decomposes prediction into a Think-Locate-Describe sequence. SG-ISA first predicts a semantic cue for the changed concept, then uses discrete spatial anchors as an implicit localization scaffold, and finally generates the change description together with the grounding box. Experiments reveal that even the strongest closed-source MLLMs and recent R1-style reasoning models struggle on MinCU, with most failing to jointly produce accurate descriptions and grounding boxes. Compared to the previous chain-of-thought method, fine-tuning with SG-ISA yields substantial joint improvements in grounding accuracy and description quality while reducing reasoning-token overhead by approximately 26%. These results suggest that an implicit intermediate spatial interface can be more effective than relying solely on model scale for grounded dual-image understanding.
[CV-168] RegVGGT: Sustainable Visual Geometry Grounding for Streaming via Regulated Memory ECCV2026
链接: https://arxiv.org/abs/2609.23286
作者: Hongbo Mao,Junjun Jiang,Youyu Chen,Jiaxin Zhang,Zhemeng Dong,Xianming Liu(Harbin Institute of Technology)
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: ECCV 2026. Corresponding author is Junjun Jiang
Abstract:3D reconstruction from a lengthy video stream input poses a dilemma for feed-forward reconstruction models (FFRMs), that a whole-stream inference context cannot be retained under limited GPU this http URL studies seek to resolve this problem via a trade-off between the integrity of inference context and GPU memory usage, which either suffer from a rapid memory inflation or degraded context integrity due to artificially capping memory this http URL by our key observation that the initial saliency of a token reliably dictates its long-term importance across the stream, we propose RegVGGT, a training-free token regulation method which aggressively regulates the tokens of incoming this http URL admitting at most 1% of tokens per frame to update the context memory, our method dramatically suppresses memory inflation as the stream this http URL with a FlashAttention-compatible token saliency estimation scheme, RegVGGT is capable of processing thousands of frames on a consumer-grade GPU with negligible compromise to reconstruction this http URL experiments demonstrate that RegVGGT achieves state-of-the-art performance on long-horizon benchmarks across diverse FFRM prediction tasks, surpassing prior FFRM-based stream reconstruction baselines by a large margin.
[CV-169] Blind Deconvolution of Binary and Pattern Images with Pixel Intensity Constraints and Sparse Gradient Prior
链接: https://arxiv.org/abs/2609.23268
作者: Qinghua Zhang,Xuesong Yang,Liangtian He,Liang-jian Deng,Jun Liu
类目: Computer Vision and Pattern Recognition (cs.CV); Optimization and Control (math.OC)
备注:
Abstract:Blind image deconvolution (BID) is a prominent research topic in the field of imaging sciences, given its significant practical applications. Most existing model-based BID methods focus on natural images, incorporating appropriate prior knowledge about both the underlying image and the blur kernel. However, for certain classes of images, such as barcodes, text, and patterns, pixels can only take very limited values, a specific prior that is often overlooked in the literature. In this article, we introduce a novel pixel intensity constraint to leverage this important information, improving recovery performance for these specialized image classes. Specifically, we propose a unified framework for blind binary and pattern image deconvolution that incorporates both the pixel intensity constraint and a gradient sparsity regularizer. Numerical experiments demonstrate that our method outperforms many existing BID techniques, achieving superior results in terms of both visual quality and quantitative metrics.
[CV-170] Exact Quotients of Fresnel-Kummer Surfaces and Certified Biaxial Refraction
链接: https://arxiv.org/abs/2609.23249
作者: Tanush Shaska
类目: Computer Vision and Pattern Recognition (cs.CV); Algebraic Geometry (math.AG)
备注:
Abstract:The Fresnel wave surface governs the propagation of light in a transparent biaxial crystal. It is a special Kummer quartic, and we identify it exactly. Over the complex numbers the wave surface of a crystal is the Kummer surface of the Jacobian of an explicit genus-two curve branched at the signed square roots of the three principal permittivities. This Jacobian is isogenous, by an isogeny with kernel of order four, to a product of two elliptic curves. One elliptic curve carries the three permittivities, and the other carries the optic-axis angle. The physical family is Zariski dense in the locus of genus-two curves with an extra involution, and its automorphism strata are explicit. The identification instantiates a task-aware quotient, which identifies parameters that differ by a nuisance transformation and carries invariant coordinates and explicit strata. For biaxial crystals, two ratios of the permittivities form a complete invariant of the wave surface up to rotation and rescaling, and the four real nodes are given in closed form. At an interface the candidate transmitted waves are the roots of a quartic of exact degree four. Its real-root count, root order, and repeated-root events are decided by exact algebraic predicates, and along the generic single-node encounters of the paper its discriminant vanishes to second order. Floating-point solvers drop forward transmitted modes near the optic axes, and the certified solver does not. On exact equivalence classes, learned models on quotient coordinates are invariant and more accurate than models on raw tensors, while learned root-count predicates fail near the optic axes.
[CV-171] SPACE: Semantic Projection and Alignment of CLIP Embeddings for Domain Adaptation ICIP
链接: https://arxiv.org/abs/2609.23248
作者: João Renato Ribeiro Manesco,Danilo Samuel Jodas,Douglas Rodrigues,Leandro Aparecido Passos,João Paulo Papa
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted for publication at the 2026 IEEE International Conference on Image Processing (ICIP), Tampere, Finland. 7 pages, 2 figures, 3 tables
Abstract:A fundamental challenge in deploying vision models is domain shift, which arises when training and test data follow different distributions, leading to degraded performance. This challenge is amplified when the same semantic concept appears under distinct visual forms, such as photographs and sketches, where visual similarity is weak despite semantic correspondence. Existing unsupervised domain-adaptation methods aim to align distributions across domains but often ignore semantic relationships among samples of the same class. To address this issue, this paper introduces SPACE, a method that exploits the semantic structure of CLIP’s vision-language space for domain adaptation. The key idea is to use text descriptions as semantic anchors by applying Singular Value Decomposition to CLIP embeddings of class descriptions, yielding an orthogonal basis that captures semantic relationships among categories. Visual features from both domains are projected into this semantic subspace, aligning images based on meaning rather than appearance.
[CV-172] CausalWM: Causal Chain-of-Thought Reasoning for Embodied World Model
链接: https://arxiv.org/abs/2609.23184
作者: Ziming Xu,Shuang Liang,Ruobing Han,Ziqiao Xi,Mingxing Rao,Kun Zhou,Zijun Zhang,Yuchen Yan,Yufan Wei,Junbo Huang,Yifei Shao,Fang Nan,Biwei Huang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Embodied world models learn to predict future physical dynamics from visual observations and control signals, where physical knowledge is implicitly entangled within latent representations. We introduce CausalWM, a 16B embodied world model that performs explicit causal chain-of-thought reasoning before future video prediction. CausalWM organizes useful variables into a reasoning trajectory, allowing the model to progressively capture causal dependencies underlying physical evolution. To train CausalWM, we collect 31K hours embodied data and develop a three-stage paradigm consisting of large-scale video pre-training, causal CoT mid-training, and multi-objective RL post-training. Despite using only a limited set of supervised CoT variables, CausalWM exhibits emergent in-context learning capabilities, enabling contextual visual feature guidance and efficient few-step generation. CausalWM achieves state-of-the-art performance across language-conditioned, action-conditioned, single-view and multi-view benchmarks, including Top-1 performance on TriWorldBench leaderboard.
[CV-173] GrapeSplat: Geometry-Grounded Reconstruction via Amalgamated Pose-Free Encoding for Feed-Forward 3D Gaussian Splatting
链接: https://arxiv.org/abs/2609.23182
作者: Si-Yu Lu,Yung-Yao Chen,Yi Jan Chen,Shang-Lin Li,Ching-Chan Liao,Wen-Huang Cheng
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Feed-forward 3D Gaussian Splatting now reconstructs renderable scenes from unposed, uncalibrated images. Yet, most models supervise only photometric consistency and predict Gaussians pixel by pixel, which leaves global structure fragile and ties primitive count to image resolution and view count. To this end, GrapeSplat amalgamates multi-view cues into a voxel-aligned scene representation and decodes Gaussians directly from the learned grid, requiring no per-scene optimization or post-processing. An Atlas Encoder lifts all views into pixel-wise geometry-and-appearance features anchored at predicted 3D points. PEACH-Vox compands the unbounded scene into a bounded sparse grid through a smooth per-axis map with an exact closed-form inverse. The Sparse Decoder then consolidates the grid with sparse convolutions and decodes the full scene as multiple Gaussians per occupied cell. This amalgamated representation exploits sparse voxel occupancy, where the Gaussian count follows the occupied cells and saturates as views cover the scene, while grid resolution sets its ceiling. GrapeSplat turns unposed images into a renderable Gaussian scene in a single forward pass. Trained with 2D and 3D supervision on 8-view sequences, it generalizes zero-shot from 4 to 64 views across indoor and unbounded scenes. Code and trained weights are available at this https URL
[CV-174] UltraTex: Unleashing 2K Multi-View Diffusion for 3D Texturing
链接: https://arxiv.org/abs/2609.23169
作者: Yibo Zhang,Ze Yuan,Nan Cao,Li Zhang,Yan-Pei Cao,Yuan-Chen Guo,Rui Ma
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:High-quality texture generation is essential for creating realistic and production-ready 3D assets. Recent multi-view diffusion methods have shown promising results for image-guided 3D texturing, but they are typically constrained to low operating resolutions such as 512 or 768, making it difficult to preserve high-frequency details from high-resolution reference images. Scaling this paradigm to 2048 resolution is computationally prohibitive, as the unified multi-view sequence exceeds 212K tokens and incurs excessive memory and latency. In this paper, we present UltraTex, an efficient end-to-end framework for high-resolution multi-view diffusion-based 3D texturing. Our key observation is that object-centric multi-view renderings contain two major sources of redundancy: background-induced sequence redundancy and sparse token interactions within the foreground. To address them, we introduce Background Token Dropping, which removes background tokens before the DiT backbone, and Block-Sparse Attention, which reduces attention computation over the retained foreground sequence. To enable efficient foreground-only inference while avoiding reconstruction artifacts, we further design Foreground-Aware VAE Decoding to ensure the quality of the final high-resolution views. To satisfy the demanding data requirements of 2K-resolution multi-view diffusion training, we construct G-buffer TexVerse, a large-scale, ultra-high-resolution multi-view rendering dataset covering over 268,000 3D assets. Extensive experiments show that UltraTex generates visually faithful textures with rich fine-grained details, while substantially improving efficiency, achieving 20.6\times – 91.1\times training speedup and 22.3\times – 74.6\times end-to-end inference speedup over the baseline on common samples in our dataset. Code and data is at this https URL.
[CV-175] An Eternal Irradiance Camera SIGGRAPH
链接: https://arxiv.org/abs/2609.23161
作者: Jeremy Klotz,Shree K. Nayar
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 18 pages, 18 figures. SIGGRAPH Asia '26
Abstract:A conventional camera uses millions of pixels to measure radiance from all directions within its field of view. We present an omnidirectional irradiance camera that measures the irradiance function—the illumination incident upon every point on a sphere. The irradiance function varies smoothly over the sphere and hence is bandlimited. We have analyzed this function in the frequency domain and have shown that it is well approximated by a weighted sum of the first seven degrees of spherical harmonics. As a result, the irradiance function can be accurately reconstructed from a small number of samples. This implies that an irradiance camera does not need millions of detectors (pixels)—just a handful of measurements suffice. This brings two major benefits. First, the camera consumes such little energy that it can be completely powered by the light falling on its detectors. Second, it does not capture the visual details needed to identify an individual, and hence privacy is preserved. We have built a prototype irradiance camera, called FluxCam, using 49 detectors arranged on the surface of a sphere. In a well-lit indoor environment, FluxCam can read out and wirelessly transmit its measurements at 30 frames per second using energy harvested from the light falling on it (i.e., without a battery, cable, or external power supply). We show how FluxCam can be used as an optical gyroscope for computing rotation, to monitor a workspace, as an untethered light probe for diffuse relighting, and as an omnidirectional pyranometer for estimating sky conditions and determining the best orientation of a solar panel.
[CV-176] SparkDiffusion: Mitigating the High-Sparsity Trap — A Unified Framework for up to 265times Single-GPU Acceleration of Visual Generation
链接: https://arxiv.org/abs/2609.23153
作者: Yuxi Liu,Haoyu Li,Zekun Zhang,Tengxu Sun,Yixiang Cai,Jiayong Li,Yifei Xia,Tianle Liu,Baole Ai,Ang Wang,Jiamang Wang,Lin Qu,Kai Zhang,Kun Yuan,Bin Cui
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Video diffusion transformers are expensive because attention dominates long spatiotemporal token sequences. We identify the \emphhigh-sparsity trap: at extreme attention sparsity, step-local training losses keep decreasing while terminal generation quality stagnates or degrades. The trap is one of supervision: the dominant terminal errors originate in the high-noise structure-generation stage, and terminal-aligned training corrects terminal errors that substantially extended step-local training cannot. This yields a simple staging principle: \emphfirst adapt the sparse architecture into a coarse prior, then correct the terminal distribution. We instantiate the principle as \method, a unified acceleration framework for visual generation that combines a short sparse warm-up, few-step trajectory-mixed distillation, and FP8 quantization with fused kernels. \method sustains 97% attention sparsity with strong visual quality on long-sequence 720P generation across Wan2.1/Wan2.2 backbones and T2V/I2V tasks, and 90% sparsity on Wan2.1-T2V-1.3B-480P. With 3-step CFG-free inference, \method achieves a 265\times end-to-end speedup over the 50-step CFG dense baseline for Wan2.1-T2V-14B-720P on a single RTX~5090 ( 220\times on H100), and denoises a Wan2.1-T2V-1.3B-480P video in 1.3 s.
[CV-177] Qwen VLConnector: A Fast Unified Medical VLM Chatbot for Fine-Grained Clinical Perception and Text Generation MICCAI
链接: https://arxiv.org/abs/2609.23139
作者: Le Thien Phuc Nguyen,Thien Nguyen,Thanh-Huy Nguyen,Gia Minh Hoang,Anh Mai Vu,Ulas Bagci
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: MICCAI FLARE 2025
Abstract:Most medical vision-language models (VLMs) excel at open-ended report generation and VQA but provide limited support for structured, fine-grained clinical perception within a unified interface. We present QwenVLConnector, a Qwen2.5-VL-based medical chatbot that unifies classification, multi-label classification, textualized detection, counting, regression, and free-form report generation under a single next-token objective. Our key component is a lightweight dense multi-layer Connector that aggregates low- and high-level visual features, aligns them through the pretrained vision Merger, and fuses them with the final visual representation without increasing sequence length. This design enriches visual tokens with complementary spatial and semantic cues while preserving efficiency. On FLARE-2D, QwenVLConnector improves detection F1 from 0.55 to 0.85, raises single-label classification from 0.37 to 0.51, and boosts report-generation GREEN by up to 18.3 points over the Qwen2.5-VL baseline. We further explore multimodal in-context learning for report generation, showing additional improvements without updating model parameters. Overall, QwenVLConnector offers a unified and efficient framework for combining structured medical perception with open-ended clinical text generation. Our code can be found at this https URL.
[CV-178] LD-RSVIS: A Large-Scale and Diverse Benchmark for Referring Surgical Video Instrument Segmentation
链接: https://arxiv.org/abs/2609.23067
作者: Zan Wang,Yunhe Feng,Dong Nie,Oluwatosin Oluwadare,Kewei Sha,Yan Huang,Heng Fan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Referring surgical video instrument segmentation (RSVIS) aims at segmenting the instrument in a surgical video, given a textual description. Despite recent progress, current models are trained and assessed on relatively small-scale benchmarks, hindering the development of more general RSVIS. In addition, existing benchmarks support only the single-target expression that refers to one instrument in the video, while overlooking multi-target and no-target referring expressions, restricting the applicability of RSVIS in practical scenarios. Addressing these issues, we propose LD-RSVIS, a new benchmark aiming to facilitate more robust and general RSVIS. Specifically, LD-RSVIS consists of 3,536 surgical videos with 1.09 million frames and covers a broad set of 30 instrument classes from 25 various procedures. By including abundant videos and classes, LD-RSVIS could benefit large-scale training and evaluation of more general RSVIS methods. Besides, unlike existing datasets, LD-RSVIS offers diverse referring settings, including no-target, single-target, and multi-target expressions, which enables the development of more practical RSVIS models in real applications. In order to ensure high-quality annotations, all videos in LD-RSVIS are manually labeled with multiple rounds of inspection and refinement. To our knowledge, LD-RSVIS is the largest and most diverse benchmark for RSVIS. To analyze LD-RSVIS and to provide comparison for future research, we evaluate 12 representative methods, and the results reveal that more efforts are required for improvements. To encourage future research, we present a simple yet effective RSVIS method, dubbed Cascade-RSVIS, that first mines target-specific cues using the complementary multi-cue text information and then employs such cues and textual information for segmentation, achieving promising performance. Our benchmark and code will be released.
[CV-179] HDMamba-YOLO: Efficient State-Space Perception and Local Spatial Reconstruction for UAV Small Object
链接: https://arxiv.org/abs/2609.23061
作者: Linduo Wei,Junjie Fan,Yijun Mai,Yong Qi
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Small-object detection in UAV imagery is challenged by weak visual evidence, ambiguous boundaries, dense object distributions, and complex backgrounds. Effective detection therefore requires long-range contextual information for target-background discrimination while preserving explicit local two-dimensional structures for accurate localization. These requirements arise at different stages of the detection pipeline and are not naturally addressed by a uniform feature-processing strategy. We propose Hybrid Dual-domain Mamba-YOLO (HDMamba-YOLO), a stage-wise heterogeneous SSM-CNN detector organized according to a perception-reconstruction-alignment-interaction rationale. EfficientVMamba-based EVSS establishes long-range contextual perception in the backbone, while PhasePatchMerging2D provides phase-aware hierarchical transitions. DST-Wrapper and Native C3k2-ASSAF then perform perception-to-reconstruction transition and repeated local two-dimensional reconstruction during FPN/PAN aggregation. DySample provides content-adaptive cross-scale resampling, while OS-CVTIA introduces macro-micro interaction and task-specific modulation for localization and classification. On VisDrone2019, HDMamba-YOLO-B achieves 42.737% mAP50 and 25.713% mAP50:95 with 10.042M parameters and 29.879 corrected GFLOPs. HDMamba-YOLO-Lite achieves 41.140% mAP50 and 24.741% mAP50:95 with 5.344M parameters. Under the unified AI-TOD evaluation protocol, HDMamba-YOLO-B obtains 21.621% AP and 47.881% AP50. Controlled ablations further support the stage-wise allocation of state-space perception, convolutional reconstruction, dynamic alignment, and task interaction for UAV small-object detection.
[CV-180] VDGS: Visibility-Driven Large-Scale 3D Gaussian Splatting for Aerial Scene Reconstruction
链接: https://arxiv.org/abs/2609.23049
作者: Haolin Yu,Jiadong Tang,YiXian Wang,Yu Gao,Shi He,Zhilin Lai,Yi Yang,Mengyin Fu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages, 6 figures
Abstract:Large-scale scene reconstruction is a critical foundational technology in robotic autonomous systems such as 3D mapping and autonomous driving. In recent years, 3D Gaussian Splatting (3DGS) has demonstrated remarkable advantages in both visual quality and computational efficiency, making it a promising representation for large-scale scene reconstruction. However, it still faces challenges in large-scale scenes, including excessive memory consumption and uneven viewpoint coverage caused by UAV acquisition, limiting its real-world applications. To address this, we propose VDGS, a novel 3DGS framework that incorporates camera distribution into scene modeling. VDGS introduces visibility-driven statistics for scene anchors to quantify supervision strength. These statistics are further leveraged for scene partitioning and for gradient compensation in under-optimized regions, thereby promoting balanced optimization across different regions. Extensive experiments on multiple large-scale aerial scene datasets demonstrate that, under imbalanced viewpoint distributions, VDGS consistently outperforms existing methods, while maintaining competitive performance in scenarios with more uniform view distributions.
[CV-181] BrainIAC: Interactive 3D Brain Lesion Segmentation across Heterogeneous MRI Modalities with Online Adaptation
链接: https://arxiv.org/abs/2609.23026
作者: Wentian Xu,Anthony P Addison,Ziyun Liang,Harry Anthony,Guang Yang,Konstantinos Kamnitsas
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Brain lesion segmentation is a fundamental task in medical image analysis, playing a critical role in diagnosis, treatment planning, and longitudinal disease monitoring. Yet existing models still struggle to meet the demands of real clinical use, where deployments contain data distribution shifts, arising from differences in scanner hardware, imaging protocol (varying MRI modality sets), and new pathologies. We present BrainIAC (Brain lesion Interactive Adaptive Continuously learning segmentation), a unified framework that integrates (i) a multi-modal backbone network trained to segment multiple types of brain lesions and handle heterogeneous sets of modalities via zero-filling and random modality dropping; (ii) 3D interactive segmentation with bounding-box and click prompts that preserves fully automatic prediction when no prompt is given; and (iii) an online adaptation mechanism combining Mid-Interaction adaptation and Post-Interaction adaptation, supervised by the network’s own predictions as pseudo labels and guided by an extra Click-Centered Gaussian loss. To our knowledge, this is one of the first 3D online adaptation methods for interactive segmentation, and the first to combine handling of heterogeneous modality sets with online adaptation. Experiments across seven brain MRI datasets demonstrate that the proposed components provide complementary and synergistic benefits. The method consistently outperforms existing approaches and generalizes well across heterogeneous imaging modalities, including those unseen during training, as well as previously unseen brain pathology types. The code and a 3D Slicer plug-in will be released at this https URL upon publication.
[CV-182] Reconstructed holograms and explanation-aware evaluation for low-cost computational pollen analysis in veterinary cytology
链接: https://arxiv.org/abs/2609.23017
作者: Swarn Warshaneyan,Joial Danyal,Blaž Cugmas,Mindaugas Tamošiūnas,Edgars Kviesis-Kipge,Kirishanth Manivannan,Roberts Kadiķis
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Quantitative Methods (q-bio.QM)
备注: 25 pages, 17 figures, 20 tables. Supplementary material included. Submitted to a journal for peer review
Abstract:Automated pollen analysis supports veterinary cytology, but brightfield microscopy is costlier and more complex than lens-less digital in-line holographic microscopy. We evaluate whether reconstructed holograms can narrow this gap and whether model explanations remain reliable under modality change. Six pollen species were imaged by brightfield and holographic microscopy. Raw, single back-propagation and iterative phase retrieval holograms were evaluated with YOLOv26s detection and MobileNetV4 classification after anchor-based annotation transfer. Six attribution methods were assessed for spatial grounding and faithfulness with the Attribution Health Inspection and Repair (AHIR) protocol, which tests model brittleness under weak noise and corrects attribution-map granularity when needed. Brightfield achieved 0.6890 mAP50-95 (0.8865 mAP50) for detection and 0.9687 macro-F1 (0.9705 accuracy) for classification. Reconstructed holograms narrowed the gap with a task-dependent split: p-type was strongest for detection at 0.5324 mAP50-95 (0.8229 mAP50), while r-type was strongest for classification at 0.7695 macro-F1 (0.7866 accuracy), both far above raw-hologram baselines. Activation-based explanations localized strongly on grains, and region-based methods retained ~60 to ~80% of faithfulness under holography. The holographic detector was highly brittle to weak perturbations, saturating deletion-based evaluation while insertion remained informative. Pixel-level gradient explanations approached random floor, yet spatial smoothing restored p-type gradient faithfulness from 0.05 to 0.51. For holographic classification, perturbation-based explanations remained faithful while gradient-based methods fell below random floor. Reconstruction improves low-cost holographic pollen analysis, while AHIR distinguishes genuine attribution failure from artifacts caused by model brittleness and map granularity.
[CV-183] CrowdCue: Specialist-Cue Conditioning for Vision-Language Crowd Counting
链接: https://arxiv.org/abs/2609.23012
作者: Moshiur Farazi,Bekir Ciftler,Abdulhalim Dandoush,Reda Bendraou
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Generative vision-language models (VLMs) offer a counting paradigm in which one model produces both a count and a natural-language account of the scene, yet their raw counting accuracy sits in the range of sub-million-parameter specialist regressors. The open question is whether auxiliary guidance from a pretrained specialist can lift them into useful territory, and through which channel that guidance is best routed. We evaluate Qwen2.5-VL-7B on four widely used crowd counting benchmarks (ShanghaiTech A and B, UCF-QNRF, NWPU-Crowd). Zero-shot prompting rarely produces a parseable count, so LoRA supervised fine-tuning establishes the baseline at overall MAE 81.64. Conditioning on a P2PNet-derived density heatmap as an auxiliary visual signal fails in every encoding we tested, and an adversarial-swap protocol shows the model reads the heatmap but applies it counterproductively. We propose CrowdCue, a family that supplies the same specialist’s already-integrated integer count to the VLM as a discrete symbol. The text-channel variant reaches MAE 72.04. The visual-channel variant, which renders the integer as printed digits and supplies it as a second image, reaches MAE 62.65, the strongest result in this paper and well ahead of the cue-supplying specialist alone (84.45 on the same split). In the late-fusion VLM we study, the binding constraint is not the channel but the abstraction level at which the specialist signal is delivered.
[CV-184] MixiMotion: One-Step Text-to-Motion Generation via Asymmetric Set Distillation
链接: https://arxiv.org/abs/2609.23010
作者: Hung Dinh,Binh Mai,Tran Quoc Bao Le,Lam Nguyen,Cong Tran
类目: Computer Vision and Pattern Recognition (cs.CV); Multimedia (cs.MM)
备注: Under Review
Abstract:Iterative text-to-motion generation delivers high-quality and semantically aligned motions but requires multiple network evaluations, resulting in substantial inference latency. We present \textbfMixiMotion, a strict one-step text-to-motion generation framework based on offline set distillation. Instead of distilling a single teacher trajectory for each text prompt, MixiMotion constructs an offline bank of multiple teacher motions and aligns teacher and student sample sets through \textbfasymmetric bidirectional matching. The teacher-to-student direction promotes coverage of diverse teacher-supported motions, while the student-to-teacher direction suppresses unsupported generations. We further introduce differentiable decoded-space kinematic supervision to complement normalized representation matching with constraints in the decoded motion space. At inference, MixiMotion generates a complete motion sequence with a single network evaluation, without teacher queries, iterative sampling, or candidate ranking. On ViMoGen, MixiMotion achieves a semantic alignment score of 0.835 , outperforming the evaluated one-step baselines and approaching the 0.858 score of its 50-step HY-Motion-1.0-Lite teacher. In blinded human evaluation, MixiMotion obtains an overall rating of 4.33 , compared with 4.50 for the teacher, while outperforming the evaluated one-/few-step baselines. Meanwhile, generation latency is reduced from 829.58 ,ms to 9.30 ,ms, corresponding to an 89.2\times speedup. These results demonstrate an effective quality–efficiency trade-off for strict one-step text-to-motion generation.
[CV-185] Compressing 3D Gaussian Splatting via Cross-Representation Priors
链接: https://arxiv.org/abs/2609.23005
作者: Yezheng Zhang,Huanxiong Liang,Chuqin Zhou,Guo Lu,Wenjun Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 14 pages, 8 figures. Accepted for publication in IEEE Transactions on Image Processing
Abstract:3D Gaussian Splatting (3DGS) enables high-quality novel view synthesis but incurs high storage and transmission costs due to dense Gaussian primitives. Recent anchor-based compression reduces per-primitive redundancy, yet redundancy across anchors remains largely unexploited. We propose CRP-GS (Cross-Representation Priors for Gaussian Splatting), a rate-distortion optimized compression framework that leverages cross-representation priors to improve anchor-level entropy modeling. First, a Correspondence-Oriented Hierarchical Structure (COHS) organizes anchors by feature correspondence rather than spatial proximity, constructing root-leaf dependencies so that selected anchors can act as informative priors to conditionally encode others, yielding more accurate likelihood prediction and lower conditional entropy. Second, Shared Feature Aggregation (SFA) extracts globally shared features from a contextual hash grid and injects them into anchor representations, factoring out scene-consistent low-frequency information that would otherwise be redundantly embedded in individual anchors. Both modules are trained under a unified rate-distortion objective to balance bitrate reduction and rendering fidelity. Experiments across multiple benchmarks show that CRP-GS achieves a favorable overall rate-distortion trade-off, yielding around 30% average bitrate reduction compared to anchor-based baselines while maintaining comparable rendering quality.
[CV-186] M3GA-Wild: A Large-Scale Dataset and Benchmark for Multi-Modal Multi-session Ground-to-Aerial Place Recognition in Forests WACV2027
链接: https://arxiv.org/abs/2609.23003
作者: Ethan Griffiths,Maryam Haghighat,Simon Denman,Clinton Fookes,Milad Ramezani
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: Accepted for publication in WACV 2027
Abstract:We present M3GA-Wild, the first benchmark for multi-modal, multi-session ground-to-aerial place recognition in forests. M3GA-Wild unifies and extends existing forest localisation datasets, providing a holistic benchmark with synchronised RGB imagery and LiDAR from ground traversals spanning 36 km, aligned high-resolution aerial imagery and multi-altitude LiDAR covering 370 hectares, and accurate geo-referenced 6-DoF poses for precise evaluation. M3GA-Wild captures diverse forest scenes with varying viewpoints, occlusion, and environmental conditions, enabling systematic evaluation of visual, LiDAR, cross-modal, and multi-modal methods. Baseline experiments show that LiDAR-based approaches significantly outperform vision-only methods under severe viewpoint differences, while current multi-modal fusion strategies yield limited gains due to poor cross-modal alignment. By pairing aerial RGB imagery with geo-referenced aerial LiDAR, M3GA-Wild also enables evaluation of foundation models for monocular depth estimation as a cheap source of 3D geometry from forest imagery, with initial experiments revealing shortfalls of current methods. These results highlight key challenges in cross-platform localisation, including modality misalignment and severe domain gaps. M3GA-Wild establishes a new benchmark to support research in robust multi-modal localisation and long-term autonomy in unstructured natural environments. The dataset and code will be available upon acceptance.
[CV-187] General Collaborative Intelligence: Architecting Cognition for Resilient Multi-Agent Ecosystems
链接: https://arxiv.org/abs/2609.22967
作者: Lei Zhang,Chun Ye,Le Yang,Zhaozhong Wang,Deng-Ping Fan,Hang Dai,Binglu Wang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Robotics (cs.RO)
备注:
Abstract:Multi-agent unmanned systems are moving from isolated, ego-centric sensing toward collaborative intelligence, in which distributed agents exchange compact features to overcome a local observation trap that no single agent can escape: occlusions, finite sensor range, and environmental degradation. The field has matured across architectural, communication, embodied, resilience, and trust dimensions, yet existing surveys examine these dimensions in isolation and rarely expose their dependencies. This review offers a unified synthesis through two complementary lenses. The first is a five-dimensional taxonomy spanning collaboration stage, communication paradigm, fusion architecture, learning strategy, and application domain. The second is three cognitive synergy conditions, Semantic Disambiguation, Pragmatic Information Exchange, and Proactive Informational Foraging, that turn cognitive synergy into operational criteria. Across these lenses we survey collaboration architectures and topologies, neural-communication co-design that treats the channel as a differentiable pipeline component, embodied action-perception loops via multi-agent reinforcement learning, and resilience mechanisms for synchronization, uncertainty quantification, and label-efficient learning. We then map these advances onto four operational domains, V2X, unmanned aerial, industrial logistics, and smart cities, and onto the safety-privacy-utility triad. To counter benchmark saturation and evaluation fragmentation, we propose GCI-Bench, a five-pillar scoring protocol with a maturity model that makes the trade-offs of collaborative methods comparable across studies. A critical reflection on reproducibility, the sim-to-real gulf, and conditions under which collaboration degrades performance identifies open challenges and charts directions toward general collaborative intelligence under real-world uncertainty.
[CV-188] CLEAR: Complex Learned Explicit Analytical Regularization for Ultra-Accelerated 4D Flow CMR Reconstruction
链接: https://arxiv.org/abs/2609.22950
作者: German Shâma Wache,Sebastian Neumayer
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Optimization and Control (math.OC)
备注: 8 figures, 1 table
Abstract:While compressed-sensing regularizers enable interpretable reconstruction of 4D Flow CMR through transparent variational objectives, their hand-crafted nature is too restrictive under high acceleration. State-of-the-art learning-based approaches mitigate this, but typically encode regularization implicitly through unrolled network modules, which limits their interpretability. To address this limitation, we propose CLEAR, designed to combine the interpretability of compressed sensing with the flexibility of learned models. To the best of our knowledge, it is the first learned regularizer for a 4D reconstruction task. In the ultra-accelerated (10\times)–(50\times) regime of the CMRx4DFlow2026 challenge, CLEAR outperforms compressed sensing locally low-rank (LLR) and the popular variational network FlowVN, while using less than 10k parameters and preserving an interpretable regularization structure.
[CV-189] RewardVerse: Rubric-Guided Policy Optimization for Video Reward Modeling
链接: https://arxiv.org/abs/2609.22947
作者: Zhenchen Tang,Yang Li,Songlin Yang,Bo Peng,Xiaotong Zhao,Shuai Li,Haotian Fan,Alan Zhao,Jing Dong
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Reinforcement learning (RL) is vital for optimizing video generation models, with a robust reward model (RM) serving as the cornerstone. However, existing video reward models often produce unstable scalar scores because they directly map complex, subjective video quality into a single score without explicit evaluation criteria. This leads to scalar drift, where the scoring scale collapses or shifts across different prompts, making the reward unreliable for RL. Drawing inspiration from professional human annotation engineering, we address this problem with RewardVerse, a rubric-based video reward framework that introduces a dynamic rubric as an intermediate representation between the evaluation query and the scorer. Instead of unconstrained direct scoring, RewardVerse first generates explicit evaluation criteria and then performs rubric-guided scoring, providing a stable semantic anchor that mitigates scalar drift. To efficiently optimize this collaborative pipeline, we propose Rubric-Guided Policy Optimization (RGPO), a two-stage training algorithm. RGPO first warms up the scorer using self-evolving seed rubrics and then jointly optimizes the rubric generator to produce query-adaptive evaluation criteria while continuously aligning the scorer with human ratings. Extensive experiments on the 16-dimensional EvalVerse benchmark and external datasets demonstrate that RewardVerse mitigates scalar drift, achieves state-of-the-art performance on both pointwise and pairwise evaluation, and provides a robust and interpretable reward signal for RL in video generation.
[CV-190] An Evolutionary Agent ic Approach for Open-ended Image Quality Perception
链接: https://arxiv.org/abs/2609.22942
作者: Zhenchen Tang,Bo Peng,Zichuan Wang,Songlin Yang,Leilei Cao,Fengjie Zhu,Jing Dong
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Generative models are rapidly expanding image quality assessment (IQA) beyond traditional fidelity factors to emerging dimensions such as physical plausibility and text-rendering correctness. However, existing IQA models rely on fixed definitions and heavy supervision, making them difficult to extend to open-ended perceptual dimensions. We identify holistic bias as an important limitation: when scoring an unseen dimension, models reuse generic quality priors, leading to scoring errors and rank inversion. To address this, we propose PACE (Perceptual Agentic Collaborative Evolution), a training-free multi-agent framework that formulates open-ended IQA as explicit protocol construction. Given a target dimension, PACE uses collaborative agents to construct an evaluation protocol composed of verifiable Visual Question Answering (VQA) probes, grounding evaluation in concrete visual evidence rather than holistic impressions. The resulting protocol is calibrated using only four human-annotated images per dimension, while a dual-track scoring mechanism aligns model perception with human scoring scales. Across traditional IQA, structural fidelity, context-aware aesthetics, and newly defined open-ended dimensions, PACE consistently improves its MLLM backbone, achieving competitive performance across diverse IQA settings, and reduces the Holistic Override Rate (HOR) from 44.4% to 8.6%.
[CV-191] D3GS: Depth DINO and RGB Diffusion Co-Guided 3D Gaussian Splatting for Sparse-View Reconstruction
链接: https://arxiv.org/abs/2609.22941
作者: Yunqi Gao,Zhanfeng Liao,Hanzhang Tu,Zhaoqi Su,Guoqing Zheng,Songtao Wang,Hongwen Zhang,Zhou Xue,Leyuan Liu,Yebin Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Novel view synthesis from sparse inputs remains challenging for 3D Gaussian Splatting (3DGS) due to ambiguous geometry, cross-view inconsistency, and missing details in under-constrained regions, resulting in degraded reconstruction and unstable rendering. To tackle these issues, we propose D ^3 GS, a Depth-DINO-Diffusion guided sparse-view Gaussian reconstruction framework that jointly enhances geometry and appearance. D ^3 GS first recovers a high-resolution, metric depth map via diffusion-based completion and DPT (Dense Prediction Transformer) refinement, providing robust Gaussian initialization and geometric constraints. Then, a DINO-guided view-consistent learning is introduced to augment Gaussian attributes with structural features, improving multi-view consistency. Finally, a diffusion-based Gaussian refinement module injects generative priors into an iterative optimization strategy, enhancing high-frequency geometric and appearance details within the Gaussian representation. Experiments on DTU, LLFF, and Mip-NeRF 360 show that D ^3 GS achieves consistent and substantial improvements over strong baselines, with ablation studies validating the effectiveness and complementary roles of each component.
[CV-192] Planning and Rendering in Concert: DeepFusion of Autoregressive Layouts and Diffusion for Visual Text Generation
链接: https://arxiv.org/abs/2609.22916
作者: Guanqiao Chen,Jingru Tan,Dongxing Mao,Catherine Chen,Zijian Du,Libo Qin,Hu Jian Guo,Alex Jinpeng Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Generating text-rich images from prompts requires both textual fidelity and the coherent integration of text into the surrounding image. An explicit layout can provide structured guidance about what text should appear and where, but a well-formed plan alone does not guarantee that the renderer will realize it faithfully. Existing layout-based AR-diffusion systems typically optimize planning and rendering separately, preventing the planner’s representations from being adapted jointly with image synthesis. We introduce DuetGen, an autonomous visual text generator built on DeepFusion, which jointly learns autoregressive planning and continuous diffusion rendering. DeepFusion conditions a diffusion transformer on the planner’s prompt and bbox-content hidden states, allowing rendering supervision to shape the representations connecting textual plans with visual outputs. Its joint objective combines autoregressive plan supervision, text-region-weighted diffusion learning, and auxiliary coordinate supervision to maintain structured planning, emphasize text-bearing regions, and improve the spatial precision of planner representations. During inference, Phase-Aware Attention Modulation strengthens the correspondence between image regions and their matched coordinate and content states, facilitating region-specific execution of the generated plan. With a 2B planner and a 4B single-stream DiT, DuetGen achieves 0.8293 word accuracy on CVTG-2K and 0.938 accuracy on LongText-Bench, closely matching the substantially larger Qwen-Image on both benchmarks. These results demonstrate the value of jointly learned planning representations and region-specific rendering for autonomous visual text generation.
[CV-193] AVTR-1: Open Stack for Real-Time Interactive Avatars
链接: https://arxiv.org/abs/2609.22913
作者: Artem Kravtsov,Dmitrii Ziganshin,Vsevolod Poletaev,Gleb Balitskiy,Anastasia Tikhonova,Egor Burkov,Vadim Lebedev
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Talking-head and dyadic models now achieve real-time inference, yet fast motion generation alone does not produce an interactive conversation. A live system must synchronize the model’s output with speech from an external voice agent, schedule video frames for playback, and handle interruptions. We introduce AVTR-1, an open stack for real-time interactive avatar conversations, built around a compact 153M-parameter autoregressive flow-matching motion generator conditioned on both participants’ audio. We adapt its audio encoder for streaming through self-distillation. The stack turns the model’s chunk-based generation into a continuous, synchronized audio-video stream driven by an external voice agent, and we analytically derive its contribution to the user-facing latencies and validate the resulting bounds with two commercial voice agents. Further experiments demonstrate that AVTR-1 leads the compared dyadic systems on all reported visual-quality metrics and most conventional listening-motion metrics while remaining competitive in lip synchronization. Its inference runtime operates in real time on data-center and consumer GPUs. However, conventional listening metrics do not establish whether the paired speaker’s speech contributes to generated motion. We therefore introduce the Reference-Based Directed Granger Gain (R-DGG), which measures the additional predictive information carried by speaker speech after accounting for listener history and speaker motion. R-DGG finds statistically supported predictive dependence for recorded listeners and all evaluated dyadic systems, but not for talking-head generators without paired audio or mismatched speaker-listener pairs. We release the model weights, renderer, and serving backend under component-specific licenses.
[CV-194] Scout: Open-World Species Recognition on the Edge
链接: https://arxiv.org/abs/2609.22897
作者: Mohammad Mehdi Rastikerdar,Hui Guan,Deepak Ganesan
类目: Computer Vision and Pattern Recognition (cs.CV); Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
备注: Under Review
Abstract:Large vision-language models (VLMs) enable recognition beyond a fixed class set, but their computational demands prevent them from running on many edge devices. Cloud offload makes this capability accessible, but sending every image consumes scarce bandwidth and communication energy. We ask how to bring the open-world recognition capability of VLMs to the edge while operating within tight compute, energy, and bandwidth budgets. Wildlife monitoring provides a natural setting for exploring this question because camera traps encounter species not known at deployment. We present Scout, an autonomous open-world recognition system that invokes a cloud VLM intermittently to teach new classes to a compact edge model. Given only the deployment location and empty site frames, Scout autonomously turns each species identified by the VLM into persistent, site-conditioned recognition capability in a resource-efficient edge model, without a predefined species list, human labeling, or manual tuning. Across 30 camera-trap deployments in three regions on an NVIDIA Jetson Orin Nano, the accuracy of Scout remains within 0.1-2.5% of a model given a predefined species list. On species outside its initial class set, Scout achieves 53.7-59.1% accuracy, compared with 56.5-65.1% for full cloud offload, while using 59-71% less deployment energy.
[CV-195] Combining Foundation Model Confidence and Monocular Depth for Training-Free Out-of-Distribution Segmentation
链接: https://arxiv.org/abs/2609.22896
作者: Serin Varghese,Fabian Hüger,Kira Maag
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Autonomous vehicles operating in open-world scenarios are inevitably confronted with previously unknown objects, such as exotic animals or loose cargo. The reliable detection and segmentation of these out-of-distribution (OOD) objects is therefore crucial for a safe understanding of the environment and decision-making. Most existing approaches require access to OOD training samples, retraining of the segmentation backbone, or dedicated auxiliary architectures, limiting their practical applicability. We propose a training-free method that derives dense OOD scores directly from the confidence predictions of a foundation segmentation model, without any task-specific fine-tuning or access to anomalous data. To improve the robustness of our OOD segmentation, geometric information from monocular depth estimation is incorporated into the decision process, providing complementary cues to uncertainty-based predictions. We evaluate the proposed method on the SegmentMeIfYouCan benchmark and additionally assess its performance on OOD tracking in video sequences, reflecting the temporal nature of real-world perception systems. The method performs strongly on road-centered benchmarks.
[CV-196] Are Coreset Selection Methods Worth Their Cost?
链接: https://arxiv.org/abs/2609.22894
作者: Yangze Liu,Zhongyi Han
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: 23 pages, 7 figures. Benchmark artifacts include per-run result tables, selected indices, and raw timing-audit tables
Abstract:Coreset selection picks a representative subset of the labeled training set to make training cheaper. However, it is usually evaluated by downstream accuracy at a fixed subset size, ignoring both the time spent selecting the subset and the training recipe behind each reported number. We introduce an end-to-end benchmark that standardizes downstream training and charges selection and training to the same auditable wall-clock budget, spanning 4 datasets from CIFAR-10 to ImageNet-1K, 11 selectors, 5 fractions, and 3 seeds, with over 1,500 released runs. Repeated-sampling work has shown that budget-aware evaluation already favors random strategies. Our two budget studies test whether that verdict survives when every selector is granted its most favorable operating point. Across eight wall-clock budget anchors on each of CIFAR-10 and Tiny ImageNet, no anchor is won by a sophisticated selector: every winner is class-balanced random sampling, repeated random sampling, or full-data training. In fixed-budget duels on ImageNet-1K, training on all data for fewer epochs beats every selection strategy we probe while also costing the least. A per-dataset cost audit shows that selection cost is dominated at every scale by a fixed full-dataset scan, so it cannot be amortized away by selecting a smaller fraction, and its absolute size does not extrapolate from one dataset to another. We further quantify when selection does pay back through subset reuse, and document 9 correctness fixes to a widely used codebase, one of which shifts a standard Herding baseline by nearly 6 points. Selection time is not free preprocessing, and an evaluation that ignores it measures the wrong quantity.
[CV-197] Planning -Aligned Pretraining of BEV Representations with Sparse Action-Conditioned Targets for End-to-End Autonomous Driving
链接: https://arxiv.org/abs/2609.22868
作者: Jaeha Song,Soonmin Hwang
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:End-to-end driving requires planning-relevant bird’s-eye-view (BEV) representations, but existing pretraining approaches often rely on task annotations or dense scene reconstruction. We introduce PAVER, Planning-Aligned BEV Encoder Pretraining. From a single LiDAR sweep, PAVER constructs sparse risk and unknown targets describing occupied and unobserved evidence along rule-based ego motions. A 10K-parameter head predicts these targets from masked BEV features conditioned on the action state, directing supervision toward geometric constraints on candidate motions. Pretraining requires no driving-task annotations or dense reconstruction. Only the BEV encoder is transferred, preserving the downstream architecture and camera-only inference. On nuScenes, PAVER reduces VAD-Tiny’s average collision rate from 0.51% to 0.19%, while improving planning L2, motion prediction, detection, and mapping. The selected VAD-Tiny and VAD-Base schedules use about 36% less estimated total training time than scratch training, including pretraining. On Bench2Drive Town05 Long, PAVER improves UniAD-Tiny’s closed-loop Driving Score from 48.45 to 58.79. The project page is available at this https URL.
[CV-198] Image Frame Dynamic Object Segmentation and Ego Motion Estimation using Radar Image Fusion IROS2026
链接: https://arxiv.org/abs/2609.22857
作者: Astik Srivastava,Suhani Grover,Avinash Sharma,Madhava Krishna
类目: Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注: Accepted for presentation at IEEE IROS 2026 Workshop LTP
Abstract:Dynamic object segmentation and ego-motion estimation are closely coupled problems in autonomous driving, as accurate ego-motion estimation typically requires static scene observations, while identifying static observations requires knowledge of the ego motion. We present Radar-Dot, a radar–RGB framework that exploits radar Doppler measurements to address this coupling. Radar returns are first used to estimate ego velocity through a linear Doppler constraint, with residual-based static/dynamic segmentation and robust estimation used to reduce the influence of moving objects. The estimated motion is then combined with metric depth and dense optical flow to identify image regions whose observed motion is inconsistent with the rigid scene motion. Experiments on 10 nuScenes scenes (part of nuscenes-mini) demonstrate that the resulting geometric pipeline achieves 20.24% dynamic IoU and 33.67% F1-score over 394 frame pairs, while radar-based static-point filtering improves ego-velocity estimation compared with using all radar returns. These results demonstrate the potential of radar as a modality for jointly improving ego-motion estimation and dynamic object segmentation.
[CV-199] SmoLSTM: A Compact Vision-Language-Action Model with Recurrent Memory that Persists ICRA2027
链接: https://arxiv.org/abs/2609.22854
作者: Jan-Gerrit Habekost,Parsa Mastouri Kashani,Connor Gäde,Matthias Kerzel,Philipp Allgeuer,Cornelius Weber,Stefan Wermter,Jae Hee Lee
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: Submitted to ICRA 2027
Abstract:Vision-language-action models often predict actions from only the current observation, which can leave tasks involving object occlusion or visually identical objects ambiguous without episode history. The usual countermeasure, widening the observation window, turns the horizon into a hyperparameter and lets per-step cost grow with it. We instead capture the episode in a recurrent state. SmoLSTM couples a frozen 256M-parameter SmolVLM backbone to a matrix-memory LSTM control layer in which observation tokens and action queries are unified in a single causal stream that is never reset throughout the entire episode. Recurrent-state storage is therefore O(1) in episode length. A flow-matching action head predicts chunks of 10 end-effector pose deltas and gripper commands at each control step. Our single policy, trained jointly on 7,461 demonstrations across 140 tasks and evaluated on held-out initial states, performs best, reaching 85.1% subgoal coverage and 77.5% full-task success on LIBERO-Mem with 0.04B trainable parameters, surpassing both the benchmark’s own object-centric baseline and recent memory-based approaches. Resetting the recurrent state at every control step reduces full-task success to 7.0%, showing that the trained policy relies on context carried between decisions. The same model achieves 79.6% average success on standard LIBERO.
[CV-200] LINGO: Latent Initialization and Gradient Optimization for Sparse-view X-ray Novel View Synthesis and CT Reconstruction with 3D Gaussian Splatting
链接: https://arxiv.org/abs/2609.22849
作者: Lifeng Xing,Dequan Jin,Kunpeng Bu,Peigeng He,Shihui Ying
类目: Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR)
备注:
Abstract:In novel view synthesis and Computed Tomography (CT) reconstruction with sparse-view X-ray imaging, insufficient angular coverage leads to structural ambiguity and accumulated noise. Integrating 3D Gaussian Splatting (3DGS) with X-ray absorption physics can achieve promising results, but it suffers from noisy initialization, positional insensitivity, and weak gradients in low-density regions. In this paper, we propose a unified Latent Initialization and Gradient Optimization (LINGO) framework to address these issues. LINGO combines latent mask-space initialization with dynamic gradient optimization to improve point cloud structural completeness while accelerating training. It constructs voxel-level 3D filters from X-ray masks to robustly suppress background noise and provide reliable geometric priors. By employing an adaptive voxel scaling strategy and dynamically scaling loss, LINGO can adjust spatial resolution and explicitly amplify gradients in low-density structures. To evaluate the quality of initialization, we introduce the Initialization Point Cloud Structural Deviation (IPSD) metric. Experiments on the X3D dataset indicate that for the novel view synthesis task, LINGO improves the Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index (SSIM) by an average of 0.72 and 0.0039, respectively, over baselines under identical sparse-view settings, achieving comparable reconstruction quality within 5k steps to state-of-the-art models typically trained with 30k iterations. For the CT reconstruction task, LINGO also demonstrates consistent improvements, with average PSNR and SSIM gains of 0.36 and 0.0134. These results highlight LINGO’s effectiveness in both accelerating training and enhancing reconstruction quality across different sparse-view imaging scenarios.
[CV-201] SatOV: Restoring Spatial Priors for Training-Free Open-Vocabulary Segmentation in Remote Sensing Imagery
链接: https://arxiv.org/abs/2609.22834
作者: Changhao Zhao,Linglin Zeng,Hai Liu
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Open-vocabulary semantic segmentation (OVS) of remote sensing imagery is a challenging pixel-level task requiring strong generalization and adaptation to the spatial characteristics of remote sensing data. Although existing vision-language foundation models perform well in general domains, their image-level classification design weakens the spatial priors needed for high-resolution remote sensing segmentation: structural spatial relations are degraded during deep feature transformation, and fine-grained spatial details are lost during downsampling. To address these complementary deficiencies, we propose SatOV, a training-free framework for open-vocabulary remote sensing segmentation that restores spatial priors at two stages of the representation pipeline. Specifically, Residual QQ Attention (ResQQ) extracts Query-Key self-attention from an intermediate CLIP layer and fuses it with final-layer Query-Query attention via a residual combination, restoring structural spatial priors suppressed by the final-layer representation. Spatially Modulated Upsampling (SatUp) uses the original high-resolution RGB image as spatial guidance, combining spatial feature modulation with guided cross-attention to reconstruct pixel-level textures and boundaries. Extensive experiments on DOTA, UDD, LoveDA, and Vaihingen show that SatOV consistently improves training-free OVS and achieves competitive quantitative and qualitative results against state-of-the-art methods. These results validate the effectiveness of restoring spatial priors at both the representation and spatial-resolution stages for remote sensing open-vocabulary segmentation.
[CV-202] C2-INR: Customized Convolutional Implicit Neural Representation
链接: https://arxiv.org/abs/2609.22807
作者: Jinglei Shi,Xinran Chang,Jiaqi Cui,Yingjie Xia,Zhaolin Xiao,Chongyi Li
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Implicit Neural Representation (INR) leverages neural networks to represent discrete signals such as images as continuous ones, where the network weights serve as a compact form of the signal itself. Most existing INR methods adopt Multi-Layer Perceptrons (MLPs) as their backbone. Since these models render each pixel independently, they inherently fail to exploit the spatial correlations that exist between neighboring pixels. In contrast,convolutional INRs can process pixels in parallel while inherently accounting for inter-pixel dependencies, making them a more natural fit for representing images. Nevertheless, convolutional INRs remain relatively underexplored, and the majority of them rely on fixed architectural settings, leaving little room for image-specific adaptation. In this paper, we investigate network customization for convolutional INRs. We replace conventional filters with irregular directional kernels, whose allocation is guided by the directional energy in the image spectrum, i.e., directions exhibiting stronger energy are assigned a larger number of kernels, enabling content-tailored convolution settings. These kernels are further reformulated via an orthogonal basis to achieve a superior sparse representation. Moreover, we introduce an annealed Gumbel-Softmax-based mechanism for kernel-level activation function selection, which gives the most suitable activation function for each convolution kernel. Extensive experiments demonstrate that our method, namely C ^2 -INR, achieves superior performance against state-of-the-art approaches under comparable parameter budgets across a wide range of image processing tasks, including representation, inpainting, and super-resolution.
[CV-203] PixelART: Image-to-Layer Decomposition without Latents or Text-to-Image Pretraining
链接: https://arxiv.org/abs/2609.22789
作者: Zelin Jia,Zhao Zhang,Zhicong Tang,Yuhui Yuan,Shixia Liu
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Image-to-layer decomposition converts a flattened image into editable RGBA layers, enabling element-level editing in design workflows. Existing diffusion-based systems typically adapt large pretrained text-to-image (T2I) models and introduce RGBA autoencoders or variable-layer architectural modules. We revisit this design choice and ask whether layer decomposition truly requires these heavyweight components. We introduce PixelART, a pixel-space rectified-flow Transformer trained from scratch for image-to-layer (I2L) decomposition. PixelART directly denoises regional RGBA pixel patches using a single-stream multi-modal diffusion Transformer, avoiding RGBA-VAEs, pretrained T2I backbones, and layer-specific decoders. We identify a key property of the task: high-noise timesteps determine layer assignment and coarse layer organization, while low-noise timesteps mainly refine color, alpha, texture, and boundaries. Based on this observation, we propose a terminal-boosted timestep sampling strategy to increase training coverage in the high-noise layer assignment regime. Trained on 4M multi-layer design templates, PixelART achieves state-of-the-art layer decomposition and composite reconstruction on Design-Multi-Layer-Bench and LICA with over 80% fewer parameters, 98% lower latency, and 85% lower memory than the recent Qwen-Image-Layered model. Ablation experiments show that pixel-space \mathbfx -prediction, terminal-boosted timestep sampling, and data/model scaling are critical, while T2I pretraining brings marginal benefits to the I2L task.
[CV-204] Human-Level Accuracy Non-Human Strategies: Revealing Model-Human Divergence in Video Physical Reasoning ECCV2026
链接: https://arxiv.org/abs/2609.22788
作者: Fanhong Li,Shurui Zheng,Zi Yin,Junbo Cui,Lei Ji,Jia Liu
类目: Computer Vision and Pattern Recognition (cs.CV); Applications (stat.AP)
备注: ECCV 2026. Code: this https URL
Abstract:Video foundation models now reach human-level accuracy on physical-reasoning benchmarks, yet such tasks require predicting unobserved physical outcomes. Do these models perform human-like forward simulation, or do they exploit statistical regularities in visible scenes? Accuracy alone cannot distinguish these strategies. We introduce a distributional evaluation framework that treats model seeds and human raters as populations, enabling comparison of consensus, uncertainty, and strategy. On the Physion benchmark, we evaluate three ViT-L architectures (V-JEPA2, VideoMAEv2, DINOv2). V-JEPA2 narrows the accuracy gap to ~1 percentage point (73.2% vs. 74.2%), yet model-human disagreement reaches 26.4%, far exceeding human-human disagreement (4.8%), with substantially lower agreement (kappa ~ 0.48 vs. 0.91). The divergence follows forward-simulation demands: models outperform humans on geometric reasoning (linking, +11.8 pp) but underperform on gravitational dynamics (rolling, -11.8 pp) and causal chains (dominoes, -10.5 pp). Strategy fingerprinting confirms all three architectures share non-human strategies while none aligns with humans. Attribution analysis suggests that unobservable outcome features, rather than visible scene properties, predict this divergence, consistent with models relying more on scene-level statistical regularities than on explicit forward simulation, a systematic divergence that accuracy alone cannot reveal. Code is available at this https URL.
[CV-205] DriveReferee: Geometric Safety Verdicts Need Not Be Learned for Driving World-Action Models
链接: https://arxiv.org/abs/2609.22762
作者: Fengcheng Yu,Dhruv Parikh,Junjie Ye,Maulik Bhatt,Thang Vu,Igor Vasiljevic,Vitor Guizilini,Yue Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Generative world-action models (WAMs) jointly generate future video and vehicle actions, while their action branches remain primarily optimized by expert imitation. Yet imitation provides no explicit closed-loop geometric verdict for generated trajectories, making verification important during both training and deployment. Closed-loop evaluators can check collision and drivable-area violations, but require privileged scene state unavailable at deployment. Existing approaches often close this gap by learning a verifier from sensor features. For these geometric checks, the rule itself is explicit. For example, collision is determined by whether the rolled-out ego footprint overlaps occupied vehicle space. What is unavailable at deployment is the scene state needed to apply the rule. We introduce DriveReferee, which uses a learned geometry readout to predict the scene representation from camera observations and executes the geometric safety rule directly rather than learning it. The resulting analytic referee evaluates collision and drivable-area safety from a scene state and candidate trajectory. During training, it scores self-sampled trajectories on ground-truth state and distills the resulting preferences into the WAM policy. At deployment, the same referee evaluates generated trajectories on this predicted state and selects a safer alternative when needed. The analytic referee requires no verdict-specific training, and its decisions follow an explicit geometric rule. Under matched candidates and inference budgets, it matches or outperforms all learned-verifier and heuristic baselines. Given the same predicted state and trajectory, learning the verdict provides no measurable downstream gain despite requiring tens of thousands of evaluator-labeled training examples. On the full NAVSIM navtest, DriveReferee reaches 92.02 PDMS with single-camera visual input and no external training data.
[CV-206] owards Robust Classroom Attendance: A Comprehensive Evaluation of Face Detection and Recognition Models
链接: https://arxiv.org/abs/2609.22750
作者: Himani Trivedi,Hiren Patel,Ridham Patel,Krutika Patel,Nancy Patel
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 6 pages, 5 figures. Accepted at the International Conference on Converging Intelligence (CICON 2026), Track 1: Artificial Intelligence and Data Science
Abstract:Manual attendance methods, such as paper or register-based systems, take a lot of time, can lead to errors, and are easy to falsify. Face recognition is more reliable, but it frequently struggles in classrooms because lighting and other conditions can vary. Face recognition datasets are designed for regulated environments and do not capture the actual challenges found in classrooms. To address this, a new face detection and recognition dataset, the Visage Face dataset, comprising 16,234 face samples, is proposed for the task of face detection and recognition. The photos are taken from different angles and under varying lighting conditions, with students showing a range of expressions, and some faces partly covered to reflect real-life situations. A YOLO-based system is used to detect faces and tested seven advanced face recognition models with thirteen configurations: LVFace, QCFace, FaceLiVTv2, TopoFR, EdgeFace, TransFace, and GhostFaceNets. Of these, FaceLiVTv2-M performed best, with 99.75% Top-1/Top-5 accuracy and an inference time of 6.459 ms. These results show that the Visage Face Dataset is a realistic and challenging benchmark for face recognition in classroom attendance.
[CV-207] ZIL: Zero-shot Image-to-LiDAR Registration
链接: https://arxiv.org/abs/2609.22716
作者: Zijun Li,Xiaotian Sun,Xuelun Shen,Yao Dai,Sheng Ao,Yangyang Shi,Jakob Engel,Zhipeng Cai,Cheng Wang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Image-to-LiDAR registration estimates the camera pose of an image with respect to a LiDAR point cloud. It has diverse applications in autonomous driving, robot navigation etc. However, state-of-the-art (SOTA) methods still 1) mostly assume same-frame inputs, struggling with the image and point cloud from distant frames; 2) rely on domain-specific training, failing to generalize to unseen scenarios. We propose ZIL, the first foundation model for zero-shot non-synchronized image-to-LiDAR registration. ZIL encodes the input image and point cloud with the Vision and Point Transformers. In addition to regressing the relative pose, ZIL also learns to predict 3D coordinates, which substantially improves the pose accuracy without additional annotations. Interestingly, naive mix-data training cannot enable zero-shot generalization, which requires normalization on both camera intrinsics and the LiDAR vertical-axis origin. Trained on 7 public datasets with 1.4M LiDAR frames, ZIL consistently and significantly outperforms previous SOTA with a single model across 5 in-domain and zero-shot benchmarks, reducing the translation and rotation errors by up to 87% and 76% (shown in Fig. 1). Code and models are available at this https URL.
[CV-208] Vision2CAD: A Visual Agent Harness for Explicit Geometry Referencing and Localization in Parametric CAD Modeling
链接: https://arxiv.org/abs/2609.22688
作者: Xi Cheng,Chenxi Zhai,Hang Cheng,Mingyu Fan,Pingfa Feng,Long Zeng
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Generating parametric CAD models requires accurate geometry and stable feature dependencies. Existing methods face challenges in selecting geometric references, interpreting sketch-plane local coordinates, and establishing sketch constraints to projected external geometry. We present Vision2CAD, a visual agent harness that combines vision-language model (VLM) reasoning with deterministic CAD kernel operations. An ID-based interface supports explicit geometry selection, a local-coordinate bridge converts view coordinates into sketch coordinates, and projected-edge localization supports external sketch constraints. These mechanisms establish feature dependencies within the supported modeling operations and constraint types. We also introduce the Geometry Explicit Reference Dataset (GERD), which aligned commands, geometry states and IDs at every modeling step. On GERD-EVL and a DeepCAD test subset, Vision2CAD improves mIoU by 11.1% and 5.6% and reduces Chamfer distance by 17.3% and 41.8%, respectively. Parameter-editing experiments and ablation studies further proved the preservation of parametric dependencies.
[CV-209] PanoSeg3R: Feed-Forward 3D Semantic Segmentation for Panoramic Images with an Automatic Data Curation Pipeline
链接: https://arxiv.org/abs/2609.22687
作者: Heechan Yoon,Dongki Jung,Phuc Nguyen,Ming Lin,Dinesh Manocha
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:We present PanoSeg3R, a feed-forward framework for 3D panoramic semantic segmentation. Unlike existing methods designed for perspective inputs, PanoSeg3R jointly predicts 3D geometry and multi-view semantic segmentation in one single forward pass. Built upon a pretrained reconstruction backbone that supports panoramic images, our approach extends feed-forward 3D reconstruction with a query-based mask decoder. Furthermore, we introduce an automatic panorama data curation pipeline that leverages the complementary strengths of off-the-shelf foundation models to generate reliable pseudo semantic annotations, substantially expanding the training data and improving zero-shot generalization. PanoSeg3R achieves state-of-the-art performance on panoramic 3D semantic segmentation, improving 3D mIoU by up to 16.02 on ScanNet++, while the curated training data further improves zero-shot performance by up to 4.26 and 43.28 mIoU on Stanford2D3D and ToF-360, respectively. Website: this https URL
[CV-210] Math2Visual-X: A Modular Framework for Pedagogically Aligned Lower-Primary Math Visuals Generation
链接: https://arxiv.org/abs/2609.22647
作者: H.D.E. Maduranga,S. K. Munasinghe,K. P. T. I. Weerasekara,Surangika Ranathunga,Nisansa de Silva
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: To appear in MERCon 2026
Abstract:Visual representations can help lower-primary learners understand Math Word Problems, but generating classroom-usable visuals remains difficult. Existing symbolic systems are controllable but limited in coverage, while end-to-end text-to-image systems often fail to satisfy exact mathematical constraints. This paper presents a symbolic visual generation framework for lower-primary MWP generation with broader problem coverage and more scalable asset generation. The framework includes an LLM-based routing layer, three worksheet-oriented generation modules, and two fallback mechanisms for open-world SVG asset acquisition. A human evaluation comparing Math2Visual-X with Stable Diffusion XL, Nano Banana, and GPT Image showed that the proposed method achieved the strongest overall performance. The results indicate that the framework offers a scalable and pedagogically grounded approach for automatic MWP visual generation.
[CV-211] ConsistWorld: Evidence Routing for Consistent Multi-Agent World Models
链接: https://arxiv.org/abs/2609.22641
作者: Qianxun Xu,Xianfang Zeng,Xinyao Liao,Wei Cheng,Gang Yu,Chi Zhang
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Autoregressive video world models enable temporally coherent generation for a single observer. Extending them to multiple agents requires consistency across independently controlled views and temporal gaps under causal streaming. We present ConsistWorld, a multi-agent world model that generates camera-controlled video streams of a static scene from one shared image. We formulate consistency as routing evidence from committed multi-agent history and concurrently generated peer views to the tokens being generated. Pose Conditioned Memory Retrieval selects relevant historical observations from all agents, recovering evidence beyond the recent context window. Visibility-Gated Peer Sharing regulates current peer information according to estimated historical coverage and current-view overlap. Together, they determine which historical observations enter the context and where concurrent peer information contributes, supporting long-term recall and coordinated exploration. Both mechanisms use camera geometry and maintain a bounded active context for a fixed agent count and retrieval budget. Experiments on evidence sharing cases and video length and agent number generalizations show that ConsistWorld achieves a strong cross-time and cross-agent consistency while preserving competitive generation quality.
[CV-212] X-Beat: An Explainable Framework for ECG Image Classification ALT
链接: https://arxiv.org/abs/2609.22631
作者: Mohammad Sadman Tahsin,Haitham Y. Adarbah,Afzel Noore
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to the Workshop on Data Quality Aware, High-Performance, and Trustworthy AI Systems for Healthcare at IEEE/ACM CHASE 2026
Abstract:Accurate automated interpretation of electrocardio- grams (ECGs) is essential for early detection of cardiac condi- tions such as myocardial infarction and rhythm abnormalities. However, many high-performing deep learning models remain difficult to deploy in clinical settings due to limited transparency and lack of reliability validation. In this work, we present X- Beat, an explainable and reliability-aware benchmark framework for ECG image classification designed to support trustworthy AI systems in healthcare. The proposed framework combines transfer learning with post-hoc explainability and systematic reliability evaluation across four cardiac classes: Abnormal Heartbeat, History of Myocardial Infarction, Myocardial In- farction, and Normal Heartbeat. Multiple ImageNet-pretrained CNN backbones, including EfficientNet-B0, ResNet-50, DenseNet- 121, and MobileNetV3-Large, are evaluated under a unified training protocol. Beyond standard performance metrics, we incorporate Grad-CAM-based visual explanations together with additional analyses, including explanation stability, regional sen- sitivity, and confidence-based reliability assessment, to examine whether model predictions are supported by clinically meaningful evidence. Experimental results show that ResNet-50 achieves the best performance, reaching 91.94% accuracy and a macro F1- score of 0.9098, with strong class separability (AUC up to 0.995). Explanation analyses indicate that the model primarily focuses on waveform-relevant regions, while reliability evaluation suggests that most incorrect predictions occur with lower confidence. Overall, this work provides a structured and reproducible bench- mark for evaluating both predictive performance and explanation reliability in ECG image classification, contributing toward the development of trustworthy and interpretable AI components for clinical decision support systems.
[CV-213] Seeing is not Enough: Vision-Language Models Perceive Evidence but Fail to Act
链接: https://arxiv.org/abs/2609.22588
作者: Yuyang Dai,Bofei Huang,Hongbo Zhang,Haoran Xie
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 29 pages
Abstract:Vision-language models (VLMs) perform strongly on visual question answering benchmarks, yet often make decisions that contradict visual evidence they have already identified correctly. We distinguish perceptual failure, where relevant evidence is not recognized, from process failure, where recognized evidence fails to constrain the final decision. We introduce VPAC-Bench, a benchmark spanning nine real-image process families, with each image annotated by its current activity stage and nearby stage transition. We also propose State-Relevance-Target (SRT), a family of structured process-prior interventions that requires models to connect visible evidence to the relevant process state before answering. Across multiple VLMs, process failure is widespread: models that correctly enumerate visual candidates still over-commit to a single answer in more than 95% of ambiguous cases. An explicit process-structured intervention reduces this rate to below 13% without degrading performance on unambiguous cases. However, the transfer of process priors is model-dependent, and generic SRT does not consistently outperform strong chain-of-thought baselines. When the relevant stage transition is known, boundary-aligned SRT substantially outperforms generic process prompting and all tested chain-of-thought baselines across assembly, physical state transition, navigation and traffic, and object-use affordance tasks. These results show that process priors are most useful when aligned with the scene’s specific decision boundary, motivating boundary-aware prior selection for process-grounded visual reasoning.
[CV-214] Beyond the Leaderboard: Counterfactual Diagnosis of End-to-End and VLA Driving Policies Under Domain Shift
链接: https://arxiv.org/abs/2609.22582
作者: Ruolin Yang,Zilin Huang,Buoyue Wang,Zhengyang Wan,Yuhao Luo,Zihao Sheng,Sikai Chen
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:End-to-end and vision-language-action (VLA) driving policies are compared by leaderboard rank, but a rank reports an outcome, not the behaviour behind it, so it predicts poorly how a policy will behave at a new site. On six released policies, rank on nuScenes open-loop error or on NAVSIM’s leaderboard does not carry over to scenes with a pedestrian near the ego corridor at a new site. We propose a counterfactual check-up: a few hundred real frames, each edited two ways (pedestrian removed, or re-lit by a night-style perturbation), every edit verified by an independent detector, and the change in the planned trajectory read as a diagnosis rather than a score. From these edits two causal axes are read, and five exams built on them separate what a score merges: how far the policy plans to drive, whether seeing the pedestrian buys safety, whether that response scales with danger, whether the plan moves when nothing requires it, and how much an irrelevant lighting change moves it. On 246 NAVSIM near-pedestrian scenes, in the cells where the pedestrian lies on the planned path only 1.9% of responses are genuine avoidance, and under our open-loop protocol the median clearance change is at most 0.03 m and the median change in planned distance at most 0.08 m for every policy. In a pre-registered test from left- to right-hand drive, the exposure and specificity orderings, the lighting verdict and the collision outcome transfer, while point values and the hazard-sensitivity verdict do not. Read as a selection report, the profiles say which policy is safe because it plans short, which covers a human-like distance without yielding, and which is unsteady under a change that requires no reaction, and they price each verdict: most settle within a few dozen frames, hazard sensitivity needs hundreds. Code and edited frames will be released.
[CV-215] Rethinking Vision Architectures with Gated Linear Attention and KAN
链接: https://arxiv.org/abs/2609.22506
作者: Ali Mehizel,Oussama Khaldi
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: 17 pages, 9 figures. Code available at this https URL
Abstract:Vision Transformers allocate most parameters to multi-layer perceptrons (MLPs) for channel mixing, while token interactions usually rely on quadratic multi-head self-attention (MHSA). Linear attention reduces sequence complexity to O(N), but remains coupled with the same fixed-activation MLP as softmax Transformers. Kolmogorov-Arnold Networks (KANs) instead place learnable univariate maps on edges, yet prior vision KANs keep MHSA or omit attention entirely. We introduce LKAT (Linear Kolmogorov-Arnold Transformer), an isotropic ViT encoder that couples chunkwise Gated Linear Attention (GLA) with a two-layer KAN feed-forward, and we provide an I/O-aware fused RBF-KAN kernel for the radial-basis grid maps. Under a shared DeiT-style recipe we compare LKAT with ViT, ViT-5, and MLP-Mixer. LKAT-B exceeds ViT-B/16, ViT-5-B, and Mixer-B/16 on ImageNet-100. Tiny/Small/Base LKAT variants scale consistently on CIFAR-10/100, and ImageNet-100 pretraining transfers to CIFAR fine-tuning. The results support gated linear attention and KAN-based radial-basis functions as complementary inductive biases for mid-scale visual representation learning.
[CV-216] Event-Frame Fusion for Inter-Frame Segmentation via Event-Guided Motion
链接: https://arxiv.org/abs/2609.22500
作者: Dalia Hareb,Jean Martinet,Benoit Miramond,Elisabetta Chicca
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Autonomous navigation requires precise and efficient semantic segmentation, yet existing frame-based approaches remain limited by motion blur, glare, latency, and the low temporal resolution (20-30 FPS) of conventional cameras, which leads to information loss between frames. Event cameras have emerged as an alternative sensing modality, capturing intensity changes asynchronously with high temporal resolution, high dynamic range, and sparse outputs. However, event-based algorithms still fall short of frame-based ones in accuracy, as most segmentation methods are designed for dense frame data. To overcome these limitations, we propose a hybrid vision architecture that combines conventional frame-based and event-based cameras. The system integrates two complementary components: (1) a compact Spiking Neural Network (SNN) with 42k parameters for motion estimation, and (2) a lightweight event-driven SNN with 0.84M parameters for frame-based semantic segmentation, which interpolates motion between frames to refine segmentation results. By predicting inter-frame segmentations, the framework achieves segmentation rates of up to 500 Hz with an energy consumption below 1.87 mJ per inference, while maintaining real-time GPU execution at frequencies up to 200 Hz. Additionally, our approach compensates for information loss in frames affected by blur or overexposure, enabling more robust perception in challenging conditions.
[CV-217] Spatiotemporal Flux Probing for Single-Photon Videography ECCV2026
链接: https://arxiv.org/abs/2609.22479
作者: Jerry Yan,Matteo Forlivesi,Bowen Tan,Andrew Xie,Siddharth Somasundaram,Sotiris Nousias
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: ECCV 2026, Project page: this https URL
Abstract:We address the problem of recovering high-speed videos from dynamic scenes under extreme photon sparsity. Existing methods rely on aggregating photon detections in local spatiotemporal windows to improve signal-to-noise ratio; however, this local grouping discards global structure and fails in low-light regimes where photon detections are sparse in space and time. In this work, we show that the information needed to recover both motion and illumination is encoded in correlations over the full space-time pattern of photon arrivals. Building on this insight, we develop a spatiotemporal flux probing theory and an algorithm that estimates the Fourier coefficients of the underlying intensity directly from the photon stream. We demonstrate that our approach (1) recovers fast motion and temporal illumination dynamics with substantially fewer photons than prior methods, (2) enables velocity-selective videography that automatically refocuses video onto specific detected motions, and (3) generalizes across sensing modalities including single-photon, event, and spike cameras.
[CV-218] Style as Cover: Deep Image Steganography via Stylized Transmission
链接: https://arxiv.org/abs/2609.22392
作者: Qi Li,Jidong Yang,Huaike Yu,Chunpeng Wang,Suo Gao,Herbert Ho-Ching Iu,Yuantian Miao,Bin Ma,Xiao Chen
类目: Computer Vision and Pattern Recognition (cs.CV); Cryptography and Security (cs.CR); Multimedia (cs.MM)
备注: 17 pages, 7 figures, 5 tables
Abstract:Image steganography hides secret message within normal images, with most existing works relying on cover-preserving transmission. However, such a paradigm becomes vulnerable once the original cover is exposed or can be reliably approximated. In this paper, we propose StyleStegaNet, a stylized image hiding framework that replaces cover matching with style-concealment transmission. Instead of transmitting a cover-like stego image, StyleStegaNet generates stylized stego images conditioned on publicly available style references, redefining steganography invisibility from cover-preserving concealment to behavior-level camouflage based on style transformation. Such a setting poses a substantial challenge to reliable secret recovery, since neural stylization can significantly alter the feature statistics exploited by deep hiding methods. To address this challenge, StyleStegaNet decouples the overall task into four coordinated stages: stego generation, stylized transmission, structure-preserving reconstruction, and secret recovery. Moreover, StyleStegaNet is optimized with a progressive three-stage training strategy, in which wavelet-domain constraints and perceptual supervision guide the recoverable information toward structural representations. We further provide an analysis showing that secret recoverability is largely restricted to the normalized structural subspace, offering a mechanistic explanation for why directly stylized baselines fail and why a reconstruction-guided recovery path is necessary. Extensive experiments on DIV2K and MS-COCO datasets demonstrate the effectiveness of StyleStegaNet. And few-shot image steganalysis with two deep detectors further shows detection accuracy near random guessing, approximately 51%.
[CV-219] Active Spatial Inspection for Effective and Efficient Embodied Exploration
链接: https://arxiv.org/abs/2609.22385
作者: Wenbin Wang,Xiang Bai,Yizhao Wang,Hang Sun,Dong Ren,Jie Qin,Qingquan Li,Bing Wang
类目: Robotics (cs.RO); Computer Vision and Pattern Recognition (cs.CV)
备注: 18 pages, 10 figures, journal
Abstract:Achieving high task success and efficiency remains a central pursuit in embodied exploration. Existing frameworks typically guide agent behavior through a spatially coarse and indirect assessment of suggestive cues and directions, yet such designs may struggle to judge cue sufficiency and the need for further inspection, leading to early termination or excessive continuation and ultimately reducing task success and efficiency. This work rethinks embodied exploration from a spatially explicit standpoint and introduces the spatial-inspection-guided ACtive Exploration (ACE) framework coupling evidence-grounded perception with exposure-informed movement and establishing a spatially resolved decision paradigm. Evidence-grounded perception strengthens inferential validity by integrating granular localization with focused verification, turning suggestive cues into precise decisive visual support. Exposure-informed movement promotes directional judgment through prospective prioritization and retrospective suppression, selecting promising directions for efficient spatial progress. ACE alleviates the tension between preventing early termination and avoiding unnecessary continuation. Extensive experiments demonstrate that ACE achieves 18.0% higher navigation task success and 10.3% higher exploration efficiency for question answering than prior state-of-the-art baselines, advancing effective and efficient embodied exploration.
[CV-220] MarsRecon: Self-Supervised and Multimodal Surface Representations for Mars
链接: https://arxiv.org/abs/2609.22379
作者: Akshay Naik,Marius F. R. Juston,Jay Mahajan
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: 27 pages, 22 figures
Abstract:High-resolution orbital imagery offers a rich record of the Martian surface, but sparse geological labels limit supervised representation learning. We present MarsRecon, a geospatially aware pipeline for learning visual and multimodal representations from HiRISE observations of Olympus Mons. The pipeline calibrates NASA Planetary Data System products, extracts valid georeferenced patches, and trains a masked autoencoder on unlabeled imagery. Increasing input resolution and filtering invalid tokens reduced held-out reconstruction loss from 0.1751 to 0.1342 in the principal Stage A model series. We then freeze the visual encoder and align its features with observation text, coordinates, and local–global image context. The strongest current local-primary model achieves image-to-text recall@10 of 0.3787, text-to-image recall@10 of 0.9161, and local-to-global recall@10 of 0.4350 on the held-out test split. These results establish a working Mars-specific pretraining and retrieval pipeline; further crop-overlap controls and downstream geological evaluations are needed to assess the broader utility of its embeddings.
[CV-221] Hierarchical Aggregation of Semantic Uncertainty in 3D Scene Graphs
链接: https://arxiv.org/abs/2609.22351
作者: Carlos Cueto Zumaya,Iacopo Catalano,Wallace Moreira Bessa,Julio A. Placed
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Open-vocabulary 3D Scene Graphs (3DSGs) ground each object node in a vision-language embedding, yet they record every entry as equally certain, so a robot querying the map cannot tell which of its entries are unreliable. Estimators of semantic uncertainty could supply that distinction, but they require repeated sampling of a model, training, or held-out labels, none of which are available to a deployed system at query time. We present a framework that exploits the detector confidence and the embeddings a 3DSG already stores, converts them into a probability that an entry is correct, and propagates that probability through the containment hierarchy into a belief that a room contains a queried class. Four signals, each paired with the object-level error it indicates, are converted to probabilities at the logit scale learned by the vision-language model and combined in closed form with no additional perception or training. Objects sharing a detector and a vocabulary fail together, so the framework aggregates them in the fully correlated limit, where an aggregation under independence would treat one repeated error as repeated evidence. Evaluated on HM3DSem against a state-of-the-art 3DSG system, the framework improves object retrieval and lowers the error of the room-level assertions of the graph it reads.
[CV-222] Dimensionality reduction for AI based hyperspectral image classification based on XAI ALT
链接: https://arxiv.org/abs/2609.22333
作者: Vladimir Zeljković,Branka Stojanović,Harald Ganster,Aleksandar Nešković
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: The paper has been presented at the Annual Workshop of the Austrian Association for Pattern Recognition (OAGM) 2023 - Patterns in One Health - Oct 24-25, 2023, Vienna, Austria
Abstract:This research addresses the challenge of limited material recycling in wood recycling processes by leveraging artificial intelligence (AI)-based dimensionality reduction. Our study explores the application of convolutional neural networks (CNNs) in multi-channel hyperspectral imaging (HSI), extending beyond RGB channels to over 200 spectral channels. Dimensionality reduction within this context involves streamlining the feature space for AI system training and inference. Focusing on explainable AI (XAI) methods, this paper contributes to a broader research initiative, presenting a solution framework that enhances the sustainability and efficiency of wood recycling processes.
[CV-223] AffordanceWAM: Affordance-Aware Joint World-Action Modeling for Robot Manipulation
链接: https://arxiv.org/abs/2609.22332
作者: Jiadi You,Qize Yu,Yue Chen,Minghong Cai,Zhide Zhong,Yuran Wang,Bowen Ping,Jiaqi Liang,Zhenhao Shen,Haodong Yan,Yinchuan Li,Ruihai Wu,Xiaojuan Qi,Yingcong Chen
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Generalizable robot manipulation requires predicting how a scene will evolve, identifying where interactions are feasible, and determining how to act. Action-labeled robot videos directly supervise control but are costly and limited in diversity, whereas egocentric human videos capture diverse interactions but lack robot actions and differ in embodiment and appearance. We introduce AffordanceWAM, an affordance-aware generative World Action Model that represents object-centric spatiotemporal affordance through Scalar Affordance and Affordance Heatmap, within the generated future World. This representation grounds visual prediction in task-relevant objects and interaction regions for action generation, and provides shared interaction targets across human and robot videos. Built on a pretrained video diffusion Transformer, AffordanceWAM uses separately parameterized World and Action Experts, coupled through Masked Joint Self-Attention, to jointly predict future RGB observations, Scalar Affordance fields, Affordance Heatmaps, and continuous robot actions under a unified flow-matching objective. Human videos supervise all three future-World streams, whereas robot trajectories additionally provide action supervision, enabling transfer without human action labels or retargeting. Experiments on RoboCasa, CALVIN ABC \rightarrow D, and real-world manipulation demonstrate consistent gains over RGB-only and robot-data-only baselines. Under fixed robot supervision, RoboCasa performance improves monotonically as affordance-annotated human video scales. These results support affordance as an effective interface for both vision-language-action learning and human-to-robot transfer.
[CV-224] Visual Graph Reasoning via Knowledge Compilation
链接: https://arxiv.org/abs/2609.22327
作者: Rongzheng Wang,Zhe Wang,Ke Qin,Rongwei Wang,Muquan Li,Yizhuo Ma,Yihong Huang,Jielei Wang,Shuang Liang
类目: Multimedia (cs.MM); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at the 34th ACM International Conference on Multimedia (MM '26)
Abstract:Visual graph reasoning requires answering graph-theoretic questions directly from graph images, where graph topology and state are conveyed visually rather than given in symbolic form. Despite recent progress of vision-language models (VLMs), current approaches to visual graph reasoning still fail on simple visual graph problems. This reveals a fundamental limitation of existing approaches: they prioritize final-answer supervision over the intermediate recovery of an explicit graph representation that preserves graph topology and state from visual input. To address this limitation, we propose VGCompiler, a compilation-centric paradigm for visual graph reasoning via knowledge compilation. VGCompiler organizes reasoning around two compilers: a representation compiler that recovers a structure-preserving intermediate graph representation from visual input, and an operation compiler that compiles query intent under the recovered graph state into an executable graph operation. Specifically, we build VGCompiler on Qwen3-VL-8B and train it with reinforcement learning guided by a layered reward over executability, compiled graph validity, representation quality, and operation quality. VGCompiler uses a frozen observer to summarize graph and question conditions into lightweight signatures, enabling archive retrieval and code reuse across similar regimes. Experiments on three benchmarks GVLQA, VisionGraph, and VGCURE, show that Qwen-VGCompiler, built on an 8B backbone, surpasses the strongest closed-source VLM baseline by 28.9% and the strongest code-based baseline by 23.7%, while maintaining high efficiency. We further evaluate VGCompiler on three real-world domains, including metro routing, logistics delivery, and network fault assessment, where it generalizes across heterogeneous visual graphs and domain-grounded tasks.
[CV-225] ALPINE: Adaptive Localization for Parameter- and Sample-Efficient Few-Shot Learning
链接: https://arxiv.org/abs/2609.22323
作者: Neeraj Yadav
类目: Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 11 pages, 4 figures, 4 tables. Code, checkpoints, and reproduction manifests available at this https URL
Abstract:Few-shot learning research is predominantly evaluated on accuracy alone, with limited attention to the parameter and training-sample budgets required to reach that accuracy - a real constraint for practitioners without large-scale compute. We present an ultra-lightweight (22,249-34,917 parameter) spatial-relational architecture for few-shot image classification that combines fixed Gabor edge-energy guidance with a windowed, content-adaptive patch locator. Under a strictly matched, iso-episode-budget protocol (250 meta-training episodes, 5 canonical seeds, 600 evaluation episodes per seed), our architecture achieves 5-shot accuracy gains, consistent across all five seeds, over Prototypical Networks, Relation Networks, and MAML on both CIFAR-FS and MiniImageNet, while using 27-53% fewer parameters than any baseline. It also converges in fewer training episodes, generalizes better to an unseen fine-grained domain (CUB-200-2011 birds, zero retraining), and is more robust to 50% occlusion and 25% spatial translation than all three baselines. A series of falsification ablations - zeroing relational tokens at inference and retraining without them entirely - shows that the architecture’s pairwise relational computation, while present, is not the primary driver of its performance; the content-adaptive patch locator is. We report this honestly, together with a capacity sweep showing a genuine accuracy plateau near 22-35k parameters, and release full seed-level results and checkpoint hashes for reproducibility.
[CV-226] Yarn tracking of large-scale 3D textile reinforcements using topological material features
链接: https://arxiv.org/abs/2609.22315
作者: Hafsa El Herichi(LMPS, CVN),Arturo Mendoza(LMPS),Yanneck Wielhorski,Hugues Talbot(CVN),Stéphane Roux(LMPS)
类目: Computer Vision and Pattern Recognition (cs.CV); Image and Video Processing (eess.IV); Signal Processing (eess.SP)
备注:
Abstract:Automated segmentation of CT images has become increasingly important to enhance the reliability of simulations through the generation of high fidelity numerical models. This study addresses the challenging task of semi-automatically tracking textile reinforcements in fan blade dry preforms using X-ray CT images captured at coarse resolutions (i.e., above 140 \mu m). Our approach offers a scalable, slice-based analysis conducted on planes orthogonal to the main yarn directions, applied to a large-scale real industrial component. This enables accurate identification and tracking of yarn paths while requiring minimal training. The method models three key yarn properties statistically: their typical cross-section shape, their continuity and movement in the 3D space, and their spatial relative arrangement with respect to neighboring yarns. These statistical properties are integrated into a tracking framework via a variational formulation that optimizes all yarn center positions in successive cross-section planes. The method tracks more than 3,000 warp yarns across 1,500 slices and achieves a tracking success rate above 90%. Overall, this work demonstrates a promising approach toward large-scale, automated textile reinforcement annotation, paving the way for more efficient material characterization in complex composite structures.
[CV-227] GameReplica: A Benchmark for Black-Box Visual Game Replication by Vision-Language Agents
链接: https://arxiv.org/abs/2609.22308
作者: Boyu Qiao,Zixin Tang,Xiaoshuai Hao,Wenbo Li
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Coding-agent benchmarks usually evaluate implementation after the target behavior has been specified in text, code, or demonstrations. Existing research has extensively evaluated the ability of coding agents to generate programs from textual specifications. However, under black-box conditions where neither source code nor documentation is available, it remains underexplored whether an agent can induce the rules solely through visual observation and active interaction and reproduce the target system as a verifiable executable system. To this end, we present GameReplica, a closed-loop evaluation framework for end-to-end black-box game replication that covers the full perception, exploration, induction, reproduction, and verification pipeline. GameReplica comprises 125 tasks spanning 25 games across 5 core mechanism families, with each game instantiated at five difficulty levels. The tasks require an agent to access the target game only through screenshots and an action interface, induce the key visual elements and gameplay rules from pixel feedback and interaction outcomes, and generate a self-contained, runnable game replica that can be automatically verified by an external program. Experiments show that current coding agents still face substantial challenges in end-to-end black-box replication: the best-performing model (Claude Opus 4.8) achieves an overall score of 71.6%, while the remaining models score only 4.0%–42.9%. Further analysis reveals a consistent pattern across all models: visual-fidelity scores are substantially higher than implementation- and rule-consistency scores, indicating that agents replicate visual appearance more readily than game mechanics. The difficulty levels further amplify the performance gap: from L1 to L5, the overall score of weaker agents drops sharply, whereas that of the best-performing agent declines only slightly.
[CV-228] Authority-Preserving Evaluation of Medical Vision-Language Assistants MICCAI2026
链接: https://arxiv.org/abs/2609.22302
作者: Flint Xiaofeng Fan,Cheston Tan,Yew-Soon Ong,Roger Wattenhofer
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: Accepted at 2nd Emerging LLM/LMM Applications in Medical Imaging (ELAMI) 2026, held in conjunction with MICCAI 2026. To appear in the Springer proceedings
Abstract:Medical vision-language models can propose how urgently a skin lesion should be reviewed, but the local service retains authority to accept or replace that proposal under referral policy, capacity, and locally held patient context. Proposal quality and selected-action quality are therefore distinct evaluation targets, and benchmark evidence transfers between them only when local review preserves the expected action score. We introduce AuthEval, a logging and evaluation framework that records both actions, scores the selected action under declared local criteria, and, where feasible, scores the declined proposal under the same rule. It reports the resulting authority gap only when the record supports it. Because the gap is the product of the proposal-change rate and the mean score change on changed cases, that rate alone determines neither its magnitude nor its sign. On ISIC 2019, with MedGemma and simulated local review, two constraint regimes with similar change rates produced an optimistic image-equal gap under capacity ( +0.744 simulator units) but no detectable gap under safety. The declared evaluation unit also mattered: under mixed constraints the gap reversed from +0.374 to -0.206 when weighting shifted from image to lesion-aware cluster. AuthEval thus clarifies whether a study’s records support claims about the model, the workflow, or both.
[CV-229] On The Robustness-Resolution Tradeoff In Temporal Quantization Of Event Streams
链接: https://arxiv.org/abs/2609.22295
作者: Sayeed Shafayet Chowdhury,Ruhi Sharmin
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Event pipelines often discretize asynchronous timestamps before learning. This step looks harmless, but its stability depends directly on temporal resolution. We study this dependence at the representation level. We first show that hard temporal binning is discontinuous: an arbitrarily small timestamp shift near a boundary can move unit event mass between bins. We then define a class of nonnegative, mass-preserving, resolution-faithful continuous encoders and prove that every encoder in this class has global L1 sensitivity at least 2/Delta, where Delta denotes bin width. Linear two-bin interpolation attains this limit. Local support and first-moment preservation also make it unique. Experiments on SHD, N-MNIST, and DVS128 Gesture support the analysis. Across uniform timestamp budgets, linear interpolation lowers mean representation drift by 47-72% while keeping clean accuracy nearly unchanged. On DVS Gesture, it produces zero prediction flips across all tested budgets and three seeds. On SHD, measured drift follows 1/Delta with R^2 = 0.992.
[CV-230] Validating Not Sampling: Region-Level Robustness of Vision-Language and Vision-Language-Action Models
链接: https://arxiv.org/abs/2609.22293
作者: Bogdan Aron,Christopher Brix,Benedikt Brückner,Yanghao Zhang,Panagiotis Kouvaros,Alessio Lomuscio
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Vision-language models (VLMs) and vision-language-action models (VLAs) are increasingly deployed in real-world applications. There, a small perturbation to the recorded camera image may change a decision significantly. However, existing benchmarks for these models only sample perturbations, which does not guarantee the absence of a failure in the untested region. We present the first robustness validation of six VLMs (drawn from the Gemma, InternVL, LLaVA, and Qwen families) and five VLAs (drawn from the GR00T, OpenVLA, and \pi families) over entire continuous regions of photometric and geometric image perturbation: brightness shifts, camera rotations, and their composition. To this end, we build on the validation framework H ^2 V and introduce H ^2 V-M, a margin-aware convergence rule that makes validation affordable at the 32B parameter scale. We demonstrate that H ^2 V-M outperforms H ^2 V by an order of magnitude in model queries and that it finds counterexamples faster than random sampling while providing soundness guarantees. Our VLM and VLA robustness validation shows that robustness is mostly dependent on the perturbation type, rather than the model, and that VLMs are more robust to large camera rotations than VLAs. For VLAs, even perturbations as small as \pm1^\circ can change the commanded action in many cases. We also show that robustness depends more on model family than on model size.
[CV-231] Beyond the Survey: A Systematic Empirical Study of Detection and Association in Visual MOT
链接: https://arxiv.org/abs/2609.22291
作者: Linh Van Ma,Juhua Hu,Wei Cheng,Unse Fatima,Moongu Jeon
类目: Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted for publication in Artificial Intelligence Review
Abstract:This paper presents a comprehensive experimental evaluation and detailed analysis of state-of-the-art multi-object tracking algorithms, with an emphasis on quantifying the individual contributions of detection and association components to overall tracking performance. Unlike existing surveys that primarily offer theoretical categorizations or taxonomies of tracking methods, our work adopts a rigorous experimental perspective grounded in publicly available implementations, providing practical guidance for researchers and practitioners in method selection and system design. We introduce a unified pipeline diagram that consolidates the core components across the two main branches of visual multi-object tracking: tracking-by-detection and end-to-end deep learning paradigms, and systematically analyze the object detection, feature extraction, and data association modules. Through extensive empirical studies on standard benchmarks, including MOT16, MOT17, MOT20, SportsMOT, DanceTrack, and CrowdTrack datasets, we reveal critical insights: (1) detection quality dominates association strategy performance, with detector improvements yielding more than 10% gains compared to less than 5% from refined association strategies; (2) modern deep learning detectors paired with specialized re-identification models significantly outperform joint detection and embedding approaches; and (3) transformer-based end-to-end methods exhibit greater robustness to detection quality variations but at a substantial computational cost. Our findings from extensive experiments provide key insights into component-level effects in MOT, particularly the dominant influence of detection quality relative to association, while offering practical insights for designing and optimizing MOT systems under varying performance and robustness requirements. Code and experimental setups are available at this http URL.
[CV-232] Complementary rPPG-Derived and Lip-Region Frequency Cues for Talking-Face Deepfake Detection
链接: https://arxiv.org/abs/2609.22284
作者: Othmane Harraq,Tamer Aldwairi
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Image and Video Processing (eess.IV)
备注:
Abstract:Talking-face (TF) deepfakes are detected unevenly by rPPG-based methods across generators. We study two lightweight visual-only cues, rPPG-derived waveforms extracted by RhythmFormer and lip-region discrete cosine transform (DCT) coefficients, on the seven TF methods of Celeb-DF++ under a subject-independent protocol. In-domain, lip-region DCT matches or exceeds the rPPG-derived 1D ResNet on every method except SadTalker, and Concat fusion reaches AUC 0.891 against 0.824 and 0.827 for the unimodal baselines. Under leave-one-generator-out evaluation the cues split: each transfers clearly better to three held-out methods, and IP-LAP is near chance for both. Concat averages 0.798 but falls below rPPG alone where DCT transfers poorly, so static fusion only partly exploits this complementarity. Lip-region DCT outperforms full-face DCT on six of seven methods. We treat the rPPG-derived signal as an empirical cue and do not claim it is cardiac in origin.
[CV-233] Rethinking Streaming Video Diffusion Model: Context Execution and Training
链接: https://arxiv.org/abs/2609.22283
作者: Hongchen Zhang(University of Chinese Academy of Sciences)
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Understanding the design space of streaming video diffusion is essential to exploring its potential for generation quality and computational efficiency. We develop a unified analytical framework that relates model and sampler choices, historical conditioning, execution scheduling, and training strategies. The framework accommodates a broad family of causal context-selection policies and makes their computational dependencies and training-inference alignment explicit. Within this design space, we study three representative policies: clean, same-level, and progressive history. On the full VBench prompt set, same-level and progressive history achieve aggregate scores of 85.24 and 85.60, respectively, compared with 84.45 for the clean-history reference. Long-video comparisons further show improved subject consistency and more coherent motion with progressive history. By allowing multiple denoising nodes to be processed together, progressive-history pipelining achieves 1.57 - 2.83\times steady-state DiT speedups under our evaluated conditions. We additionally find that LoRA adaptation of the DMD fake-score network improves generation quality using only 2.15% as many trainable fake-score parameters as full-parameter adaptation. Together, these findings show that fully denoised history is not a prerequisite for high-quality streaming generation and motivate the joint design of historical conditioning, execution, and training.
[CV-234] Brain-to-Image Generation: Reconstructing Visual Stimuli from EEG using Generative Adversarial Networks
链接: https://arxiv.org/abs/2609.22282
作者: Harshit Goyal
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注:
Abstract:Reconstructing visual stimuli from electroencephalography (EEG) is difficult because scalp measurements have high temporal but limited spatial resolution, and paired EEG-image datasets remain small relative to modern generative-model training corpora. We present a reproducible single-subject baseline on THINGS-EEG2 that first tests the more defensible question of whether EEG can retrieve the viewed stimulus in a visual embedding space. A compact temporal-spatial convolutional encoder maps repetition-averaged EEG (63 by 250) to provided 512-dimensional ViT-B/32 image features. Model selection uses a concept-disjoint validation split, and final evaluation uses the official 200-image, 200-concept test gallery. Across three training seeds, the model obtains 12.83 +/- 0.58%, 39.17 +/- 1.76%, and 58.00 +/- 1.73% image recall at 1, 5, and 10 (mean +/- sample standard deviation), compared with analytical chance levels of 0.5%, 2.5%, and 5.0%. A session-balanced ablation shows that averaging more test repetitions generally improves ranking. Applying the Subject 01 model to the other nine subjects without adaptation causes a sharp performance drop, exposing subject specificity. We further report exploratory stress tests of direct conditional generators trained without external visual weights: single-subject and ten-subject variants produce noise-dominated outputs, with early validation improvements reversing after one to four epochs. Finally, we distinguish direct reconstruction from semantic rendering with a pretrained diffusion prior. The results support above-chance coarse semantic decoding under a closed-set, repetition-averaged protocol, but do not support faithful recovery of stimulus pixels.
[CV-235] Performance vs Consistency: Evaluating a Foundation Model in Lung-RADS Screening MICCAI
链接: https://arxiv.org/abs/2609.22281
作者: Benjamin Renoust,Pierre Baudot,Tiffany Foriel,Yousra Haddou,Charles Voyton,Pierre-Henri Siot,Ezequiel Geremia,Danny Francis,Jean-Christophe Brisset,Valérie Bourdès,Sylvain Bodard,Benoit Huet
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: MICCAI CAPTION 2026
Abstract:Foundation models have recently demonstrated strong capabilities across a wide range of medical imaging tasks. However, their performance in structured clinical interpretation settings remains insufficiently explored. In lung cancer screening, interpretative variability persists despite standardized frameworks such as Lung-RADS. In this study, we evaluate MedGemma, a medical general-purpose foundation model derived from Gemini and its fine-tuned version adapted for lung cancer detection and diagnosis, compared against radiologists performing Lung-RADS v2022 assessment on the NLST dataset. Twelve radiologists independently evaluated each case in a multi-reader design, enabling quantification of inter-reader variability. Radiologists achieved a mean AUC of 0.90, with substantial variability across readers (range: 0.80-0.94). The native foundation model achieved an AUC of 0.70, failing to reach clinically relevant performance. In contrast, fine-tuning significantly improved performance to an AUC of 0.83, placing the model within the lower range of individual radiologists performance. These findings highlight a trade-off between peak accuracy and prediction consistency. Unlike radiologists, under fixed conditions, the model produces deterministic outputs, removing inter-run variability under identical inputs, in contrast to inter-reader variability observed among radiologists. This supports the role of fine-tuned foundation models potential complementary tools for clinical decision support, particularly in settings with limited expertise. However, evaluation is performed on a case-enriched cohort from NLST and does not account for real-world prevalence or external validation, limiting direct clinical generalization.
[CV-236] An Affordable AI-Integrated Smart Cane for Multimodal Mobility Assistance of Visually Impaired Users
链接: https://arxiv.org/abs/2609.22277
作者: Ali Akarma,Adeel Ahmad,Toqeer Ali Syed
类目: Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Computers and Society (cs.CY); Robotics (cs.RO)
备注:
Abstract:Visual impairment affects over 2.2 billion people worldwide, yet conventional white canes cannot detect elevated hazards or provide semantic environmental context. Existing AI-assisted navigation systems typically rely on expensive hardware or cloud connectivity, limiting accessibility in resource-constrained settings. This paper presents an affordable (\ 88 USD), fully offline AI-integrated smart cane designed for multimodal mobility assistance on an ultra-low-power Raspberry Pi Zero 2W. The system fuses RGB vision sensing with Time-of-Flight (ToF) distance estimation, pairing an INT8-quantized SSD MobileNet V1 model with distance-aware vibrotactile feedback and real-time audio alerts. To ensure operational robustness on constrained hardware, a multiprocessing architecture isolates sensor acquisition, neural inference, and haptic feedback into independent processes with fail-safe sensing support. Experimental evaluation across indoor mobility scenarios demonstrates a macro-averaged F1-score of 0.82 (precision: 0.85, recall: 0.81), a mean end-to-end latency of 330,ms, and a peak power draw of 2.8,W. A preliminary usability study with 12 participants (SUS: 78.5, NASA-TLX) demonstrated positive user perception and enhanced obstacle awareness. The proposed prototype validates the feasibility of deploying privacy-preserving, edge-native assistive intelligence for cost-sensitive mobility assistance.
[CV-237] Moonworks Lunara: Modeling Artistic Intelligence
链接: https://arxiv.org/abs/2609.22272
作者: Yan Wang,Yanzu Wang,Maitreyee Joshi,Samiha Sadeka,Partho Hassan,Reza Jarral,Sayeef Abdullah,Sabit Hassan
类目: Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:We formulate \emphArtistic Intelligence as exploration driven world realization, leaving space for creative possibility while preserving the semantic, artistic, and compositional structure that must remain true. Moonworks Lunara, a text-to-image model, implements this framework with a novel Diffusion Mixture Transformer architecture. A new training algorithm iteratively evolves the data distribution through informative sample acquisition and targeted injection of human-created art. We benchmark Lunara against seven image-generation models, including FLUX.2-Klein-4B, Qwen-Image (20B), and GPT-Image-1-Mini. With GPT-5.6 Sol as evaluator, Lunara ranks first in \emphAesthetic Quality (8.473 vs. 8.457 GPT-Image-1-mini), second in \emphEmotional Resonance, and remains competitive in \emphContent Integrity. A blind human evaluation over the same evaluation set corroborates the automated metrics, ranking Lunara first. It also stays among the strongest models under conventional measures including CLIPScore and LAION Aesthetic Predictor. On GenEval, Lunara achieves competitive performance against a broader set of 16 models, including GPT Image 2 and Seedream 4.0. These results place Lunara at the frontier with Artistic Intelligence while maintaining a sub-10B active-parameter footprint and sub-10-second inference latency. Lunara advances the general visual intelligence frontier by shifting the question from whether models can get images right to how deeply they can interpret meaning and realize it as imaginative, expressive worlds.
[CV-238] Enabling Vision and Cross-Modal Learning for Multimodal Stroke Recurrence Prediction: An Interpretable Two-Step Framework MICCAI2026
链接: https://arxiv.org/abs/2609.22271
作者: Christian Gapp,Elias Tappeiner,Martin Welk,Karl Fritscher,Stephanie Mangesius,Constantin Eisenschink,Philipp Deisl,Michael Knoflach,Astrid E. Grams,Elke R. Gizewski,Rainer Schubert
类目: Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
备注: ML-CDS 2026: Multimodal Learning and Fusion Across Scales for Clinical Decision Support, MICCAI 2026, Strasbourg, France
Abstract:Multimodal stroke recurrence prediction requires effective integration of heterogeneous clinical and imaging data, yet modality imbalance often causes models to over-rely on dominant modalities and underutilize complementary information. While self-supervised pretraining and selective parameter freezing are commonly employed to improve representation learning and fine-tuning stability, their effect on modality contributions and cross-modal behavior in multimodal medical models remains largely unexplored. In this work, we investigate whether image pretraining on 3D CTA scans reduces modality imbalance and improves cross-modal integration for stroke recurrence prediction, a clinically critical task we recently addressed. To this end, two multimodal neural networks are pretrained in a self-supervised manner and subsequently fine-tuned using two distinct freezing strategies. Their performance and modality utilization are compared against both the baseline model from our previous work and models trained entirely from scratch in this study. Our results demonstrate that self-supervised pretraining enables more effective utilization of the multimodal image-tabular dataset, outperforming both the prior baseline and all non-pretrained models. Notably, the best-performing Vision Transformer based neural network successfully overcomes unimodal collapse. Synergy analysis reveals significant interactions between vision and both gender and CHD, suggesting clinically relevant patterns for stroke recurrence. Overall, our findings demonstrate that self-supervised pretraining and strategic fine-tuning support more balanced modality utilization and enable meaningful cross-modal interactions. Code is publicly available at this https URL.
[CV-239] Did You Steal My Shot? Pioneering Camera Motion Plagiarism Detection in Generative Videos
链接: https://arxiv.org/abs/2609.22267
作者: Chengguo Zhang,Ping Ping
类目: Computer Vision and Pattern Recognition (cs.CV); Multimedia (cs.MM)
备注: Accepted to ACM Multimedia 2026 (Oral)
Abstract:Camera motion often reflects directorial intent and requires professional equipment, making it a high value form of intellectual property. However, generative video models can imitate such high value camera motions with simple prompts, while existing similarity detection methods mainly operate on visual content and fail to capture deeper motion similarity. This is mainly because their training data entangles camera motion with visual content. Moreover, traditional optical flow is insufficient to represent complex camera motions. We therefore build the first benchmark for camera motion analysis, including a motion dataset with \textbf11 motion styles and evaluation protocols. Furthermore, we propose a motion representation that augments optical flow with vorticity cues from fluid dynamics, thereby better capturing motions. Experiments show that our detector achieves a \textbf3.02* improvement in plagiarism detection over the strongest baseline and remains effective on generative videos. We believe our work extends copyright protection beyond static content to dynamic camera motion.
[CV-240] Hi-Singers: A Comprehensive High-Quality Dataset for Expressive Audio-Driven Singing Head Synthesis
链接: https://arxiv.org/abs/2609.22264
作者: Yichi Zhang,Hui Zhang,Guanjun Liu,Yuefeng Zou,Fengzhao Sun,Jun Yu
类目: Multimedia (cs.MM); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Sound (cs.SD); Audio and Speech Processing (eess.AS)
备注: 7 pages, 6 figures, 4 tables. Accepted to the 34th ACM International Conference on Multimedia (MM '26). Dataset: this https URL
Abstract:State-of-the-art models for audio-driven digital human generation have achieved photo-realistic results in talking-head synthesis. However, extending these models to singing-head synthesis remains challenging due to a significant Domain Gap: singing requires more exaggerated expressions, vivid jaw openings, and precise rhythmic synchronization. Current models, primarily trained on speech datasets, often struggle with “rhythmic drift” and constrained dynamics. To address this, we introduce Hi-Singers, the first large-scale, high-quality, in-the-wild video dataset specifically tailored for singing head synthesis. Hi-Singers undergoes a rigorous automated and manual filtering pipeline, ensuring strict thematic adherence, high-resolution rendering, and stable motion, resulting in 29,608 video segments totaling approximately 170 hours. We further establish a dedicated evaluation benchmark balanced across linguistic and musical styles. Extensive experiments across diverse architectures, including 3D-coefficient and diffusion-based models, demonstrate that Hi-Singers consistently and significantly improves performance across all dimensions. Specifically, it enables models to achieve superior visual realism, enhanced lip-sync consistency, and more precise rhythmic dynamics, effectively bridging the domain gap and setting a new performance standard for the singing synthesis task. The dataset is available at this https URL
[CV-241] RS-Claw-Evolution: Environment-Feedback-Driven Evolution for Lightweight Remote Sensing Agents in Long-Horizon Tasks
链接: https://arxiv.org/abs/2609.22258
作者: Kai Ouyang,Dongyang Hou,Liangtian Liu,Zeyuan Wang,Ziyu Li,Chengfu Liu,Zichao Tang,Xuezhi Cui,Shengwu Ouyang,Wentao Yang,Hanwen Yu,Haifeng Li
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注: 30 pages, 5 figures, including supplementary material
Abstract:Large language model-driven remote sensing (RS) agents offer a promising approach to automating geospatial analysis. However, lightweight RS agents based on compact language models struggle with multi-step interactive tasks due to loss of long-horizon states, inefficient environmental feedback utilization, and sparse optimization signals. We propose RS-Claw-Evolution, an environment-feedback-driven framework that progressively improves lightweight agents through three stages. Interaction evolution uses executable code to control observations, maintain intermediate states, and reduce context redundancy. Experience evolution combines failure-aware trajectory generation with error-turn masking to learn from informative failure-recovery experiences without imitating faulty actions. Decision evolution uses reinforcement learning with multi-dimensional environment rewards and turn-level advantage protection to optimize tool-use behaviors and improve credit assignment in long sequences. On Earth-Bench, the optimized Qwen3-4B-based agent achieves 65.9% accuracy in Autonomous Planning mode, outperforming the untrained Qwen3-32B baseline (43.8%) and DeepSeek-V3.1 (60.8%), while approaching GPT-5 (71.6%). These results demonstrate that learning from environmental feedback can improve lightweight agents and narrow their performance gap with larger models in long-horizon RS tasks.
[CV-242] ask-Aware Hybrid QUBO Optimization for Structured Neural Network Pruning
链接: https://arxiv.org/abs/2609.22238
作者: Osama Orabi,Artur Zagitov,Hadi Salloum,Viktor A. Lobachev,Yaroslav Kholodov
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注: Preprint
Abstract:Neural network pruning can be formulated as a combinatorial optimization problem, yet many existing approaches rely on independent filter-importance scores or simplified objective functions. In this work, we propose a Hybrid Quadratic Unconstrained Binary Optimization (QUBO) framework for structured filter pruning that combines task-aware sensitivity information with interactions between candidate filters. The formulation incorporates first-order Taylor sensitivity and Weight-Fisher sensitivity into the linear component of the objective and can additionally incorporate activation similarity into the quadratic interactions. To control the target pruning cardinality without introducing an explicit quadratic cardinality penalty, we use a binary search over the capacity incentive to identify a coefficient that empirically yields the target pruning cardinality. We further investigate a two-stage QUBO–Tensor-Train refinement strategy in which the QUBO solution initializes gradient-free probabilistic black-box optimization to search for improved pruning masks using the downstream metric. Experiments on the SIDD image denoising task and a Half-UNet model show that the Hybrid QUBO achieves higher PSNR and SSIM than the evaluated Taylor and L1-based QUBO baselines at the studied pruning target. Multi-seed experiments under a fixed dataset protocol are used to assess robustness, while controlled sub-problem experiments demonstrate that Tensor-Train refinement becomes increasingly valuable as the combinatorial problem size grows. The results support Hybrid QUBO as a task-aware structured pruning framework for the evaluated setting, while also highlighting the computational and deployment limitations of mask-based pruning.
[CV-243] WildfireSpreadBench: The Metric Decides the Model in Wildfire Spread Prediction
链接: https://arxiv.org/abs/2609.22191
作者: Arin Gopakumar,Marco Pannozzo
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注: 11 pages, 2 figures, 5 tables
Abstract:Machine learning is being increasingly used to predict where active wildfires will burn the following day, helping inform evacuation boundaries and containment lines. Most models are evaluated using Average Precision (AP), which summarizes performance across all decision thresholds, although acting on a forecast requires choosing one. We benchmarked five discriminative architectures and one generative model on WildfireSpreadTS using a shared evaluation pipeline and two input configurations. We found that model rankings varied depending on whether performance was measured by AP or by threshold-dependent metrics like F1 and IoU. The highest-AP model flagged 4 to 5 times the area that burned and ranked fifth of six on F1 and IoU, and the most recall-heavy model flagged 16 to 23 times. Models with more usable predictions had AP scores 24 to 37 lower. Across architectures, we identified three distinct prediction profiles: over-predicting, balanced, and under-predicting, which AP alone could not distinguish. Expanding the input from 7 to 23 channels changed AP by 0.03 on average, against a 0.21 to 0.24 spread across architectures. These results show AP alone can favor models whose predictions are poorly suited for operational wildfire forecasting.
[CV-244] Correlation-Guided Flow Matching with Annealed Masking for Spatial Transcriptomics Generation
链接: https://arxiv.org/abs/2609.22187
作者: Yupei Zhang,Hao Chen,Li Pan,Chao Li,Xiaohan Xing
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV); Genomics (q-bio.GN); Quantitative Methods (q-bio.QM)
备注:
Abstract:Spatial transcriptomics (ST) provides spatially resolved gene expression profiling but remains expensive, motivating the prediction of ST from histology images. Generative models have emerged as a mainstream paradigm for ST prediction due to their ability to model the conditional distribution of gene expression and capture its inherent stochasticity. However, these methods typically treat genes as independent prediction targets and overlook the intrinsic gene-gene interactions in biological systems, which limits their ability to preserve biologically meaningful co-expression patterns. We argue that gene-gene interactions, which reflect shared pathways and regulatory mechanisms, are essential for generating numerically accurate and biologically coherent ST profiles. In this paper, we propose CorrFlow, a correlation-guided flow matching framework for histology-to-ST prediction that explicitly models gene-gene dependencies through two complementary mechanisms. First, we introduce an annealed masked flow matching strategy, where subsets of genes are progressively masked following a timestep-dependent annealing schedule, encouraging the model to infer masked genes conditioned on the remaining genes and promoting joint conditional modeling beyond per-gene marginal estimation. Second, we devise a gene graph-regularized optimization scheme that integrates prior knowledge from the STRING database and data-driven co-expression estimated by WGCNA to construct a gene affinity graph, which enforces both local consistency and global smoothness in the predicted expression. Extensive experiments across 12 datasets show that CorrFlow achieves the best average PCC and HPCC among evaluated methods, leading to more biologically coherent ST predictions.
[CV-245] Contrastive World Models
链接: https://arxiv.org/abs/2609.22175
作者: Bonnie Li
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:World models trained via pixel reconstruction can struggle in visually complex environments, where irrelevant information dominates the objective and distract the model from information relevant to planning and control. We present Contrastive World Models, an approach for learning latent dynamics models without pixel reconstruction. Building on Dreamer, we replace observation reconstruction in the standard world model objective with a Deep InfoMax-like lower bound that maximizes the mutual information between state-action sequences and local patch features of future observations, encouraging state representations to retain information that is predictive of the future without requiring the model to reconstruct visually irrelevant details. We evaluate our approach in small-scale experiments across three settings of increasing visual complexity. Our method matches Dreamer and a momentum prediction baseline in the default setting, and substantially outperforms both once distractors or natural video backgrounds are introduced, while also training more efficiently by removing the pixel decoder entirely. Our approach is general and makes minimal assumptions beyond access to state-action sequences and future observations. These results suggest that contrastive, infomax-based objectives are a principled and promising direction for building world models that are robust to visual nuisance factors, a property particularly relevant for transferring model-based RL agents to the real world.
[CV-246] SolarFlowRefiner: Refinement-Aware Flow Matching for Surface Solar Radiation Downscaling
链接: https://arxiv.org/abs/2609.22126
作者: Udbhav Srivastava,Antonita Racheal,Yiheng Chen,Runlong Yu,Xinyue Ye
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:High-resolution surface solar radiation (SSR) is important for solar forecasting and grid operation. However, physically consistent reanalysis products are too coarse to resolve localized cloud-driven variability. In this paper, we study a multisource downscaling task that reconstructs high-resolution SolarCube SSR fields from coarse ERA5 radiative variables and co-registered satellite channels. The task is challenging because a single ERA5 grid cell may contain both sunlit and cloud-shadowed regions. As a result, the missing high-resolution correction can be spatially sharp and inherently ambiguous. One-stage predictors often oversmooth these structures. Post-hoc refinement also introduces a stage-wise mismatch: the generator is optimized independently, even though its output determines the refiner’s initial state. We introduce SolarFlowRefiner, a refinement-aware flow-matching framework for SSR downscaling. A conditional FlowMatch generator first predicts a normalized correction to an upsampled ERA5 baseline. The refiner is then trained on prediction-conditioned states between the current FlowMatch output and the target residual. This exposes the refiner to the structured errors produced by the generator. The refinement objective is also backpropagated through the FlowMatch sampler, allowing generation and correction to be jointly optimized for the final reconstruction. Experiments on a day-blocked ERA5–SolarCube benchmark show consistent improvements over standalone generation and post-hoc refinement. More broadly, SolarFlowRefiner provides a general strategy for coupling generative predictors with iterative correctors.
[CV-247] Correcting Learning-based Perception for Safety
链接: https://arxiv.org/abs/2609.22108
作者: Yan Miao,Hussein Darir,Sayan Mitra
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV); Robotics (cs.RO)
备注:
Abstract:Learning-enabled perception is important in many autonomous systems. Unlike traditional sensors, the boundary where ML perception does or does not work is poorly characterized. Incorrect perception can lead to unsafe or overtly conservative downstream control actions. In this paper, we propose a two-step strategy for correcting ML-based state estimation. First, an offline computation is used to characterize the uncertainties resulting from the ML module’s state estimation, using preimages of perception contracts. Second, at runtime, a risk heuristic is used to choose particular states from the uncertain estimates to drive the control decisions. We perform extensive simulation-based evaluation of this runtime perception correction strategy on different vision-based adaptive cruise controllers (ACC modules), in different weather conditions, and road scenarios. Out of 45 ACC scenarios where the original perception-based control system using Yolo and LaneNet led to safety violations, in 73% of the scenarios, our runtime perception correction preserved safety; our method wouldn’t be able to recover 27% of the scenarios where the construction of the preimages of perception contracts is not fully conformant. Further, our runtime perception correction strategy is not overly conservative—on the average only a 2.8% increase in completion time is experienced in the corrected scenarios, with mild interventions.
[CV-248] PRQuant: Permutation Residual Quantization for Low-Overhead Inference
链接: https://arxiv.org/abs/2609.22106
作者: Peiran Wang,Anqi Wang,Jiaying Zhao,Huiwen Yang,Zhenyu Ming,Rongqian Wang,Yiwu Yao,Kun Tian,Xin Yao,Gong Zhang,Fan Yang,Zhongyi Huang
类目: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
备注:
Abstract:Accuracy of Low-bit quantization of linear layers is often dominated by a small number of outliers. Although existing methods, such as smoothing, rotation, or residual-based approaches, may mitigate this problem, they often introduce new accuracy bottlenecks to weights. Besides, most of these techniques are implemented as online approaches, which can result in heavy execution overheads. To address the afore-mentioned issues, We propose PRQuant (Permutation Residual Quantization), a training-free and low-overhead framework that combines channel reorganization with static weight-side residual compensation. After AWQ-style scaling, PRQuant identifies the input channels that contribute most to weight quantization error, permutes them into contiguous tail blocks, and constructs their residual weight sub-tensors offline. During inference, this contiguous structure enables the activation side to use tail blocks seamlessly without the expensive online gathering operation, and turns scattered residual compensation into a regular tail-augmented GEMM, substantially reducing latency. Experiments demonstrate that PRQuant effectively reduces down-projection reconstruction error. Ablation studies confirm that smoothing and residual compensation are the primary drivers of numerical improvement, while permutation provides a consistent marginal numerical benefit and, more importantly, enables a hardware-friendly contiguous layout that eliminates dynamic gathering overhead. Overall, PRQuant outperforms default MXFP4 and the evaluated PTQ baselines in average accuracy across five downstream benchmarks, improving over MXFP4 by 1.24 and 0.55 on Qwen3-4B-Instruct-2507 and Qwen3-30B-A3B-Instruct-2507, respectively.
[CV-249] When is a closed-form RGB-S/P ratio adequate? A hyperspectral characterization on natural scenes for mesopic display
链接: https://arxiv.org/abs/2609.24819
作者: Naoyuki Uchida
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Graphics (cs.GR)
备注: 14 pages, 5 figures, 2 tables
Abstract:Mesopic and low-light display transforms require, as their driving signal, a per-pixel scotopic-to-photopic luminance ratio (S/P); the exact spectral S/P is unavailable for ordinary RGB content, so a low-cost closed form that estimates S/P from a linear-RGB triplet is used in its place. Such closed forms exist but have been characterized only on narrowband / LED sources, i.e. spectrally sparse spectra, where a relative error of ~41% has been reported for a three-channel projection. Display content, however, is natural and broadband. We ask whether the same closed form is adequate there, using per-pixel spectral S/P from hyperspectral imagery as ground truth. On a daylight radiance time-series, a six-scalar closed form (three photopic and three scotopic channel weights) reproduces spectral S/P with a median error of ~0.07 that is time-invariant once the RGB input is chromatically adapted to D65; evaluated in un-adapted sRGB the error instead carries a color-temperature tilt across illuminants (~0.19), so adaptation is the enabling step for this use case. The result generalizes to an independent fifty-scene set (pooled median 0.024; 45/50 scenes within a pre-registered 0.10 band), with the few exceedances concentrated in saturated, spectrally-peaky surfaces that approach the narrowband regime (floral close-ups in this set). The scotopic weight vector is shown to be primary-model dependent, but the value used here is corroborated by a primary-free XYZ projection, and the median error stays within the band across all principled coefficient choices. We do not claim observer-validated appearance fidelity or adequacy on narrowband sources; both are out of scope. Both outcomes follow from the same three-channel projection: it is overwhelmed by spectrally sparse inputs and adequate on spectrally smooth ones.
[CV-250] Scale-Vector Alignment: A Scale-Aware Framework for Spatially Resolved Morphological Similarity in Astronomical Images
链接: https://arxiv.org/abs/2609.24304
作者: Mengke Zhao,Guang-Xing Li,Keping Qiu,Shanghuo Li
类目: Instrumentation and Methods for Astrophysics (astro-ph.IM); Astrophysics of Galaxies (astro-ph.GA); Computer Vision and Pattern Recognition (cs.CV); Computational Physics (physics.comp-ph)
备注: 13 pages, 10 figures. Submitted to ApJS. Comments welcome
Abstract:Astronomical maps made with different tracers are not expected to have identical morphology. Excitation, optical depth, chemistry, radiation, and ISM phase alter the response of a tracer, and the resulting differences can depend on both position and spatial scale. We propose scale-vector alignment, a scale-aware method based on Constrained Diffusion Decomposition (CDD). CDD decomposes an image into localized scale components; at each position, their amplitudes define a scale vector that describes how the measured intensity is distributed over spatial scale. We define the pixel-wise similarity \Spix(x,y) as the normalized alignment of two local scale vectors. The normalization removes the overall amplitude, so \Spix compares relative scale composition rather than absolute flux. We also define the scale-wise similarity \Sscale(l) by comparing the two CDD component maps at each spatial scale. Spatial shifts are used to construct an empirical shifted reference distribution for \Spix . In Orion~A, the tracer with the highest similarity to the dust-derived column-density map changes from ^12 CO to ^13 CO to C ^18 O toward higher column density. In NGC~6334I(N), the line–continuum similarity decreases locally around the brightest compact structures, where radiative-transfer effects can alter the observed line morphology. In NGC~3627, CO is most similar to 21~ \mu m emission, and \Sscale reaches its maximum at an intermediate sub-kpc scale. The method measures where two tracers have similar multiscale structure and at which scales their spatial distributions agree. The implementation is publicly available at this https URL.
[CV-251] ORION-CMR: On-scanner Reporting with Integrated Foundation Model for End-to-End Cardiac MRI Analysis and Interpretation
链接: https://arxiv.org/abs/2609.23950
作者: Omer Burak Demirel,Kelly K. Horst,Alessio Perazzolo,Elisa Bruno,Kenan Kaya,Rongzhen Ouyang,Enas Ahmed,Jouke Smink,Spencer L. Waddle,Zainudeen Kallumpurath,Tzu Cheng Chao,Dinghui Wang,Steve G. Langer,Timothy L. Kline,Panagiotis Korfiatis,Jacinta Browne,Ivana Isgum,Tim Leiner
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG); Medical Physics (physics.med-ph)
备注:
Abstract:Cardiovascular magnetic resonance (CMR) provides comprehensive cardiac assessment but remains underutilized because of the complexity of acquisition, post-processing, and interpretation. Existing artificial intelligence (AI) methods address isolated tasks, limiting clinical integration. We present ORION-CMR (On-scanner Reporting with Integrated fOunda-tioN Model), the first clinically evaluated scanner-native end-to-end CMR foundation model. Pretrained on 12,896,733 CMR images from 9,258 studies, ORION-CMR performs sequence classification, ventricular function assessment, late gadolinium enhancement (LGE) detection, binary and multiclass disease classification, and local large language model-based report generation in approximately 90 seconds. The framework. was evaluated on public benchmarks and clinically validated in a multi-vendor cohort of 68 subjects with normal examinations, congenital heart disease, dilated cardiomyopathy, and myocardial infarction. ORION-CMR outperformed supervised baselines and the previously published CMR foundation model (CMR-FM), achieving state-of-the-art performance for LGE classification and scar segmentation. Clinical evaluation achieved an AUC of 0.96 for normal-versus abnormal classification and 0.88 for multiclass disease classification, while generated reports demonstrated 81.4% agreement with expert interpretation. These results demonstrate the feasibility of real-time scanner-native AI-assisted CMR analysis and automated report generation.
[CV-252] VGG16-MCA UNet: Whole-Tumor Segmentation in 2D FLAIR MRI with Decoder-Side Channel Attention
链接: https://arxiv.org/abs/2609.23919
作者: Shubham Gajjar,Deep Joshi,Avi Poptani,Vishal Barot
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: 22 pages, 7 figures, 4 tables. Code and data splits: this https URL
Abstract:Automated brain tumor segmentation supports diagnosis, treatment planning, and monitoring of disease progression, but building models that generalize across heterogeneous tumors and limited annotated data remains difficult. We present VGG16-MCA UNet, a hybrid architecture pairing an ImageNet-pretrained VGG16 encoder with a decoder in which a Multi-Channel Attention (MCA) module recalibrates features after each skip-connection fusion, trained with the Focal Tversky loss to counter severe foreground-background imbalance. We evaluate the model as a 2D, FLAIR-only, whole-tumor segmenter on tumor-positive slices from two public datasets: the BraTS 2020 benchmark and the LGG MRI Segmentation dataset. Using 5-fold cross-validation and a single network formed by averaging the weights of the five fold models, the method attains an aggregate pixel-level Dice (F1) of 95.10% on our held-out BraTS 2020 split and 88.32% on LGG. These scores are computed over all test pixels pooled into a single confusion matrix rather than averaged per case, and are therefore not directly comparable to the per-case mean Dice used in the BraTS challenge protocol. All partitions were drawn over individual slices rather than over patients, so every patient contributes slices to both training and test; the figures above therefore measure interpolation within known patients and should be read as an upper bound rather than as generalization to new ones. The model segments a 256x256 slice in 66.32 ms on a single 6 GB NVIDIA RTX 2060, approximately 8 ms more than an equivalent VGG16-UNet without MCA. We release the split records and report the protocol in full, with the aim of providing a precisely specified and reproducible 2D FLAIR baseline.
[CV-253] Learning-Based 3D Reconstruction of Power Networks from Aerial Point Clouds
链接: https://arxiv.org/abs/2609.23915
作者: Rishabh Jain,Anuja Saini,Vishal Jain
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 5 pages, 6 figures. Accepted for oral presentation at IEEE IGARSS 2026
Abstract:This paper presents an end-to-end framework for reconstructing overhead power utility network topology and extracting span-level physical metadata from large-scale aerial LiDAR. The pipeline begins with semantic segmentation of the input point cloud using an improved KPConv-based model, in which data sampling and loss functions are adapted to emphasize pole and conductor (wire) classes. Network topology inference then proceeds in two stages: (i) pole instances are obtained by clustering pole-class points and validating candidates using geometric criteria, including height and verticality estimated via PCA, and (ii) candidate pole pairs are evaluated using a heuristic method and a lightweight ResNet-based classifier on 2D top-view projections of pole and wire point distributions to determine whether a physical conductor span exists. By explicitly classifying candidate spans, the approach mitigates common failure modes of heuristic connectivity rules in dense or cluttered scenes and under partial wire observation. For each validated wire, attributes regarding utility infrastructure geometry are computed, including endpoint conductor heights, ground elevation, sag-related lowest-point features, conductor arrangement, and wire width. Evaluation on multiple real-world aerial LiDAR datasets demonstrates decimeter-level endpoint height accuracy and approximately 9% relative improvement in recall for topology reconstruction compared to heuristic nearest-neighbor baselines, with larger gains in complex layouts.
[CV-254] Resolution-Flexible Decoding for Hybrid Neural Video Representations
链接: https://arxiv.org/abs/2609.23555
作者: Taiga Hayami,Masaya Takabe,Hiroshi Watanabe
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted to VCIP 2026
Abstract:Neural video representations (NVRs) represent videos using neural network parameters and, in hybrid formulations, frame-wise latent embeddings. Although hybrid NVRs can improve reconstruction quality by using content-adaptive latent embeddings, their latent spatial sizes and decoder upsampling schedules are tied to the target frame resolution. For high-resolution videos, this dependency may require large and non-uniform upsampling factors and can affect the parameter allocation between the latent embeddings and the decoder. In this paper, we propose a resolution-flexible decoder framework for hybrid NVRs. The decoder is constructed from uniform (2\times) upsampling stages, whose target feature sizes are obtained by tracing the spatial resolution backward from the final output resolution. After each upsampling stage, the feature map is aligned with the target size by minimal padding or cropping when necessary. To support this progressive decoding process, we further use intermediate reconstruction supervision and a reconstruction-difficulty-aware frame sampling strategy based on recent frame-wise losses. The framework preserves the basic representation format of hybrid NVRs and can therefore be applied to different backbones. Experiments on the UVG dataset show that the proposed approach improves reconstruction quality over the corresponding NVR baselines.
[CV-255] SomaNet: Weakly Supervised Learning for Instance Soma Segmentation in 3D Electron Microscopy with Partial Annotations
链接: https://arxiv.org/abs/2609.23019
作者: Mohammad Khateri,Morteza Ghahremani,Jussi Tohka,Alejandra Sierra
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages, 7 figures, 1 table,
Abstract:Soma instance segmentation, i.e., identifying and delineating individual cell somas as distinct instances, is crucial for cellular analysis and connectomic reconstruction. Three-dimensional electron microscopy (3D EM) provides nanometer-scale resolution for capturing fine-grained soma morphology. However, dense instance-level manual annotation is prohibitively costly, limiting the scalability of fully supervised methods. To address this challenge, we propose SomaNet, a weakly supervised framework for 3D EM soma instance segmentation under partial annotation constraints. SomaNet adopts a teacher–student learning paradigm tailored to partial labels. The teacher is trained using partially annotated data to generate pseudo-labels, while the student jointly learns from the partial ground-truth annotations and the generated pseudo-labels, progressively recovering dense instance segmentations. To accurately delineate cell somas under limited supervision with varying instance counts, SomaNet incorporates affinity learning, which encourages high similarity within instances and low similarity across instance boundaries. Semantic-guided affinity decoding and 2D-to-3D reconstruction then produce volumetrically consistent 3D soma instances while preserving the large receptive fields of 2D backbones. The framework is architecture-flexible and supports diverse backbones, including vision transformers and foundation models, enabling direct transfer of pretrained visual representations to volumetric EM segmentation. Experiments on 3D EM brain datasets demonstrate that SomaNet achieves accurate and robust soma instance segmentation across regions with diverse soma morphologies under partial annotation. Code is available at this https URL.
[CV-256] AWR-Net: Decoupling Anatomy and Appearance for 3D Fetal Brain Ultrasound Synthesis
链接: https://arxiv.org/abs/2609.22635
作者: Yuhuan Lu,Sergio Valencia,Yuanji Zhang,Yuhao Huang,Camilo Jaimes,P. Ellen Grant,Davood Karimi
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: 10 pages, 6 figures
Abstract:Three-dimensional fetal brain ultrasound offers non-ionizing, cost-effective imaging with rich spatial information for comprehensive anatomical assessment, yet its development remains limited by scarce data and annotations. In contrast, fetal brain magnetic resonance imaging has advanced further, supported by larger datasets and mature analysis methods. To leverage these resources, anatomical label maps provide a promising modality-invariant bridge for transferring knowledge from fetal brain magnetic resonance imaging to ultrasound. However, generating ultrasound images from label maps remains challenging because anatomy is tightly entangled with ultrasound appearance. To address this challenge, we separate anatomical correspondence learning from ultrasound appearance adaptation at both the data and model levels. Specifically, we propose the anatomy wavelet residual network, a two-stage framework combining wavelet diffusion and residual refinement. The first stage learns to generate volumes from label maps using atlas pairs in the wavelet domain, while the second stage learns bounded residual corrections from real clinical ultrasound in the image domain. This separation enables realistic synthesis with consistent preservation of normal and abnormal anatomy. Experiments on real fetal brain ultrasound show that our method outperforms representative synthesis methods, with normalized cross correlation of 0.518 versus 0.482 and Fréchet Inception Distance of 8.905 versus 13.319 for the strongest baseline. Beyond synthesis quality, volumes generated from fetal brain magnetic resonance imaging label maps also improve downstream segmentation, particularly for severe abnormal cases. Overall, these results highlight the potential of the proposed framework to leverage rich fetal brain magnetic resonance imaging resources for advancing three-dimensional ultrasound analysis.
[CV-257] Multicentre Bi-atrial Segmentation from LGE-MRI for Atrial Fibrillation with a 2D and 3D Framework
链接: https://arxiv.org/abs/2609.22398
作者: Malitha Gunawardhana,Gregory B. Sands,Mark L. Trew,Jichao Zhao
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: Accepted at Biomedical Signal Processing and Control Journal
Abstract:Accurate delineation of bi-atrial structures from late gadolinium enhancement MRI (LGE-MRI) is an important prerequisite for structural analysis and future fibrosis-quantification workflows in atrial fibrillation (AF). However, automated segmentation is challenging due to thin-walled anatomy, domain shifts across imaging centres, and limited benchmarking of existing methods. This study presents a two-stage segmentation framework and benchmarking platform for evaluating how ROI localisation, encoder design, 2D/3D dimensionality, and ensemble fusion affect bi-atrial wall and cavity segmentation across multicentre LGE-MRI datasets. The framework integrates 3D localisation and fine segmentation using 2D and 3D U-Net variants with ResNeXt encoders and compares them with convolutional, transformer-based, and state-space architectures. Evaluation across three independent cohorts assessed accuracy and cross-domain transfer without target-domain fine-tuning. Cavity segmentation transferred more consistently across centres than atrial wall segmentation, while wall performance remained sensitive to domain shift, particularly in the Kobe cohort. By quantifying how 2D, 3D, and ensemble architectures behave across centres and between walls and cavities, this work provides a reproducible benchmark for future methodological development and clinical validation.
[CV-258] Lesion-Gated Hybrid Synthesis for Virtual Contrast-Enhanced Breast MRI: A MAMA-SYNTH Challenge Solution
链接: https://arxiv.org/abs/2609.22397
作者: Shohei Yoshimoto
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: 8 pages, 2 figures, 3 tables. MAMA-SYNTH Challenge solution report; the submitted method ranked 6th in the official challenge leaderboard
Abstract:Purpose: Contrast-enhanced breast MRI depends on intravenous gadolinium-based contrast agents, motivating methods that synthesise post-contrast appearance from pre-contrast images alone. The MAMA-SYNTH Challenge (MICCAI 2026 Deep-Breath Workshop) evaluates such synthesis across four metric groups: image fidelity, tumour region of interest, downstream classification, and downstream segmentation. The four group ranks are averaged, so optimising a single objective is insufficient. Materials and Methods: We developed a lesion-gated hybrid synthesis pipeline. A lesion probability map, estimated from the pre-contrast slice alone by an ensemble of pre-contrast-only segmentation networks, spatially coordinates a tumour-focused regression pathway and a background-focused Pix2PixHD synthesis pathway, followed by a region-dependent calibration of the predicted enhancement. Training used the public MAMA-MIA collection with a patient-level split. Inference consumes a single pre-contrast 2D slice, with no mask, no post-contrast image, and no auxiliary metadata beyond image geometry. Results: On internal validation (n = 120 patients) the method reached MSE 0.523, LPIPS 0.185, and tumour-region SSIM 0.495. It was submitted to the hidden external 300-case test cohort as a self-contained inference container. The method ranked sixth in the official MAMA-SYNTH Challenge leaderboard. Conclusion: A test-compatible lesion probability map derived from the pre-contrast image can coordinate complementary tumour-focused regression and background-focused perceptual synthesis, enabling balanced virtual contrast enhancement under a multi-metric challenge setting. Comments: 8 pages, 2 figures, 3 tables. MAMA-SYNTH Challenge solution report; the submitted method ranked 6th in the official challenge leaderboard Subjects: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2609.22397 [eess.IV] (or arXiv:2609.22397v1 [eess.IV] for this version) https://doi.org/10.48550/arXiv.2609.22397 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Shohei Yoshimoto [view email] [v1] Fri, 18 Sep 2026 09:53:32 UTC (169 KB)
[CV-259] Anatomically Faithful Artifact Suppression in SENSE Accelerated Brain MRI
链接: https://arxiv.org/abs/2609.22390
作者: Changjing Chai,Bin Huang,Libo Xu,Jian Zhou,Boyang Pan,Kristen W Yeom,Qiyong Gong,Nan-Jie Gong
类目: Image and Video Processing (eess.IV); Computer Vision and Pattern Recognition (cs.CV)
备注: 18 pages of main text, 2 pages of appendix, 6 figures, and 6 tables (including 2 appendix tables)
Abstract:Background: Four-fold accelerated sensitivity encoding (SENSE4) can shorten brain MRI acquisition time but may amplify noise and result in residual aliasing artifacts after conventional reconstruction. Purpose: To evaluate whether an image-domain refinement framework can improve the quality of SENSE4 brain MRI while preserving anatomical information for quantitative measurements. Methods: In this prospective paired study, 80 participants underwent fully sampled and four-fold accelerated SENSE T1-weighted MRI. We developed an Anatomy-aware Residual Attention Network (ART-Net) to refine accelerated reconstructions through generalized self-attention and correlation-based residual artifact regularization. Participant-level splitting yielded training, validation, and independent test cohort (45/5/30 participants). The independent test cohort underwent quantitative, segmentation-based, and blinded radiologist assessments of image quality and anatomical preservation. Results: ART-Net demonstrated highly competitive reconstruction performance, achieving the highest peak signal-to-noise ratio (31.03 +/- 2.88 dB) and structural similarity index (0.963 +/- 0.022) among evaluated methods. It also demonstrated improved anatomical fidelity, with numerically highest Dice coefficients for medial temporal structures relevant to atrophy assessment (0.8824 +/- 0.0827) and whole-brain regions (0.8857 +/- 0.0885). Moreover, ART-Net improved gradient fidelity, regional contrast preservation, and radiologist-rated structural quality. Conclusion: ART-Net improved agreement between SENSE4 and fully sampled T1-weighted images in a single-center, held-out test cohort while maintaining segmentation-derived anatomical measurements. These findings suggest that ART-Net may support accelerated brain MRI by improving image fidelity and enabling reliable downstream anatomical analysis.
[CV-260] Forecasting Intrathecal Tracer Enhancement from Pre-Contrast Brain MRI: Direct Regression versus Flow Matching
链接: https://arxiv.org/abs/2609.22377
作者: Qinghui Liu,Jon André Ottesen,Thu Nguyen,Geir Ringstad,Ingrid Mossige,Siri Fløgstad Svensson,Atle Bjørnerud,Per Kristian Eide,Kyrre Eeg Emblem
类目: Image and Video Processing (eess.IV); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG)
备注: 23 pages, 11 figures
Abstract:Intrathecal contrast-enhanced MRI tracks how a cerebrospinal-fluid tracer spreads through the brain, but requires repeated scans over 24–48,h. Forecasting enhancement from a pre-contrast scan could one day help select patients for intrathecal drug treatment and plan their dose. We ask how much of the enhancement at a given time can be predicted from the pre-contrast scan and the elapsed time alone. Training on 104 patients, we compared direct image-to-image regression (I2I) with conditional flow matching (CFM) using the same 3D U-Net and protocol, and tested on 23 held-out patients and 51 external patients with normal pressure hydrocephalus. All models were measured against simply copying the pre-contrast scan. I2I removed about 60% of this copying error internally and 25% externally, outperformed CFM on both test sets (mean absolute error 0.036 vs.\ 0.058 and 0.056 vs.\ 0.063), and predicted each volume in a single forward pass, whereas CFM ran its network ten times. CFM uncertainty located errors but was poorly calibrated. Much of tracer enhancement is thus predictable from anatomy and timing, and direct regression is the more accurate and cheaper choice at this data scale.
[CV-261] CIG-MAE: Cross-Modal Information-Guided Masked Autoencoder for Self-Supervised WiFi Sensing
链接: https://arxiv.org/abs/2512.04723
作者: Gang Liu,Yanling Hao,Yixuan Zou
类目: ignal Processing (eess.SP); Computer Vision and Pattern Recognition (cs.CV)
备注: 12 pages, 7 figures. Published in IEEE Internet of Things Journal (Early Access)
Abstract:Human Action Recognition using WiFi Channel State Information (CSI) has emerged as an attractive alternative to vision-based methods due to its ubiquity, device-agnostic nature, and inherent privacy-preserving capabilities. However, the high cost of manual annotation and the limited scale of publicly available CSI datasets restrict the performance of supervised approaches. Self-supervised learning (SSL) offers a promising avenue, but existing contrastive paradigms rely on data augmentations that conflict with the physical semantics of radio signals and require large-batch training, making them poorly suited for CSI. To overcome these challenges, we introduce CIG-MAE – a Cross-modal Information-Guided Masked Autoencoder – that reconstructs both the amplitude and phase of CSI using a symmetric dual-stream architecture with a high masking ratio. Specifically, we propose an Adaptive Information-Guided Masking strategy that dynamically allocates attention to time-frequency regions with high information density to improve learning efficiency, and incorporate a Barlow Twins regularizer to align cross-modal representations without negative samples. Experiments on three public datasets show that CIG-MAE consistently outperforms SOTA SSL methods and even surpasses a fully supervised baseline, demonstrating superior data efficiency, robustness, and representation generalization.
人工智能
[AI-0] Rare Event Estimation via Iterative Unalignment
链接: https://arxiv.org/abs/2609.24969
作者: Hanming Yang,Daksh Mittal,Jing Dong,Hongseok Namkoong
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:As agents are deployed with increased autonomy, even extremely rare events along their stochastic output trajectories can occur and prove catastrophic. Safe deployment therefore does not depend on whether these events can occur, but on how often they might. We study the problem of estimating the probability of rare events that arise from stochastic variation in the agent’s own actions. Estimating this type of risk requires searching over the combinatorially vast space of trajectories. Naive Monte Carlo is computationally prohibitive in this regime, and constructing effective importance sampling (IS) proposals requires coordinated changes to a context-dependent chain of conditional distributions. We develop a new IS method that perturbs the original model’s weights to construct the proposal. The proposal is itself a differentiably parameterized language model, enabling gradient-based search over weight space. We formulate an objective that combines a differentiable surrogate for event amplification and an adaptive regularization scheme that dynamically balances amplification against estimator stability. We evaluate our approach on \sim 120M and \sim 2.6B models across three event families spanning 300+ rare events as rare as 10^-9 , with reference probabilities computed with 10% relative standard error. In our most verifiable settings, we observe that our IS estimator achieves over 800\times compute-weighted efficiency gains over naive Monte Carlo for events with probabilities lower than 10^-7 . Our implementation is available at this https URL.
[AI-1] Exactness at Inference: A Representational Criterion for Out-of-Distribution Generalization
链接: https://arxiv.org/abs/2609.24942
作者: Filipe Marinho Rocha,Inês Dutra,Vítor Santos Costa,Luís Paulo Reis
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Logic in Computer Science (cs.LO)
备注:
Abstract:A model generalizes outside its training distribution only when it computes a representation structurally equivalent to the generating mechanism, not an approximation fitted to it. Such equivalence is necessary for exactness in and out of distribution, and extrapolation is governed by this exactness at inference, whatever its realization. Tensor Logic shows this: a zero-temperature contraction is equivalent to discrete logic, deducing in place with no artefact extracted, its tensors Boolean, its embeddings orthonormal, only its arithmetic continuous. Lacking infinite recursion it reaches Datalog, not Prolog, and though exact over closed domains it needs external memory to bind a novel entity. The criterion needs neither a discrete representation nor an extracted expression, and constrains inference, not training: an exact marginal in [0,1] passes, a Neural Network thresholded to a hard label does not. Logic Tensor Networks fail it, while differentiable ILP and Tensor Logic at T=0 pass. Piecewise-affine extrapolation divergence and an inability to bind novel entities are two faces of a shortfall in exact representability. For hybrid architectures, a propagation rule follows: the output inherits the bounds of every fitted estimator on its path, explaining which axes fail in equivariant models and the ARC-AGI induction/transduction split. Only an exact hypothesis class certifies what the training data leave underdetermined: on a law-derived partition it finds the 56.3% of distant queries that are answerable, which ensembles meet with false confidence and distance metrics rank backwards. Common inductive biases, from symmetries to memory, reach exactness only because humans inject them, an argument for inducing exact representations rather than fitting surrogates whose residuals, even at the arithmetic floor in training, diverge outside the data and compound under composition.
[AI-2] Et Tu Brute? Economic Misalignment in Personal AI Agents
链接: https://arxiv.org/abs/2609.24927
作者: Aman Priyanshu,Supriti Vijay,Brian Jabarian,Niloofar Mireshghallah
类目: Artificial Intelligence (cs.AI)
备注: 20 pages, 10 tables, 4 figures
Abstract:Personal AI agents make recommendations and take actions on people’s behalf in high-stakes economic contexts, e.g., buying a flight, choosing health insurance, or selecting a graduate program. The agent is given access to the user’s personal context, e.g., their email inbox and a structured profile of personal attributes, with the intention of making an optimal, personalized decision for the user. We show that by simply providing this personal context, the agent steers recommendations based on inferred wealth, without being explicitly instructed to do so. In a suite of 325K experiments on 13 agents across three types of economic decisions (flights, health insurance, and graduate programs), we find that 8 models systematically choose more expensive options for wealthier users when requests are identical. This steering continues even when it directly goes against the user’s stated objective: when explicitly instructed to find the cheapest option, some agents still act on the wealth profile they have inferred. It also occurs when wealth is inferred from ambient data, such as emails unrelated to the task. And it persists under privacy controls that block specific attributes: blocking financial attributes largely removes the disparity, but blocking other attributes leaves it unchanged and can increase it by up to 40% for insurance, as agents rely on the remaining signals to infer wealth. Larger and more capable models are no better; Claude Opus 4.8 shows the largest effect. We term this misalignment “adversarial delegation”, in which the very conditions that make a personal AI agent useful - access to personal information - enable it to act against the user’s interests.
[AI-3] BackTrend: Evaluating Scientific Weak-Signal Prediction via Backward Reconstruction EMNLP2026
链接: https://arxiv.org/abs/2609.24921
作者: Xiao Zhou,Yilun Zhao,Owen Jiang,Tiansheng Hu,Cai Xu,Manasi Patwardhan,Arman Cohan
类目: Artificial Intelligence (cs.AI)
备注: EMNLP 2026 Findings
Abstract:Scientific weak signals are early, low-visibility research directions that later become central to mature scientific topics, yet existing resources such as trend tracking, citation forecasting, and foresight reports rarely provide validated reference sets that link concrete early precursors to later paradigms. We introduce BackTrend, a retrospective benchmark in which, given a mature target topic and a temporal evidence constraint, systems must recover two types of precursors: problem-space signals, underrecognized research problems, and solution-space signals, emerging methods for known problems. BackTrend contains 25 mature target topics in artificial intelligence and machine learning and 66 human-validated weak signals, reconstructed from large-scale literature by grounding each candidate in its 2019-2024 publication-frequency trajectory. We evaluate frontier LLMs, RAG systems, and agentic research systems using semantic matching and coverage-based metrics. Current systems often generate plausible but misaligned precursors, exhibiting topic drift, granularity mismatch, near-miss matching, and incomplete coverage; the strongest system achieves only 10.1% F1, while Coverage10 reaches at most 18.5% of the reference signals. Our budget analyses show that additional retrieval and web-search evidence can improve performance up to a moderate budget, but does not by itself close the substantial performance gap.
[AI-4] Visuomotor Robotic Pruning in Planar Orchards Using Hybrid Reinforcement Learning ATC WWW
链接: https://arxiv.org/abs/2609.24906
作者: Abhinav Jain,Cindy Grimm,Stefan Lee
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: for associated video file, see this https URL
Abstract:Dormant tree pruning is labor-intensive yet essential for maintaining modern high-productivity fruit orchards. In this work, we focus on pruning of modern planar tree training systems - V-Trellis apples and UFO cherries - where trunks and primary branches are trained into approximately planar walls. We introduce an end-to-end pipeline to learn a closed-loop visuomotor controller for robotic pruning. This controller is trained entirely using simulation and synthetically generated data and deployed in real orchards in a zero-shot manner. The pipeline comprises synthetic generation of planar orchard tree meshes, construction of a physics-based orchard simulator, automated collection of successful pruning trajectories via motion planning, and policy learning with a novel hybrid reinforcement-learning algorithm that combines offline demonstrations with online simulated rollouts. The controller uses optical-flow inputs from a wrist-mounted camera - avoiding the need for full 3D-reconstruction - and continuously guides the cutter through cluttered branch environments to a specified cutpoint with correct tool orientation. In exhaustive simulated task-space evaluations over 3,000 pruning points, the policy attains 49.9% success on V-Trellis apples and 46.0% on UFO cherries. We validate the learned controller across 38 physical trials - comprising 28 outdoor field trials in commercial and experimental orchards and 10 indoor laboratory tests - demonstrating zero-shot sim-to-real transfer. The learned policy also outperforms a classical RRT-Connect baseline on physical hardware in laboratory trials.
[AI-5] Pinocchio: Fast Uncertainty Estimates for Black-Box Language Models
链接: https://arxiv.org/abs/2609.24881
作者: Kevin David Hayes,Arka Pal,Haosong Zhang,Tom Goldstein,Micah Goldblum
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:In high-stakes decision-making applications of large language models (LLMs), practitioners require not only accurate LLMs but also uncertainty estimates for their predictions. Existing approaches to uncertainty estimation for LLMs require access to log-probabilities output by the model or require fine-tuning access. However, many industrial LLM products use closed-source API models, and many such API models like GPT do not return log-probabilities and may not allow fine-tuning. We introduce Pinocchio, an external calibrator that estimates the correctness of responses from black-box API models. Trained jointly on responses from seven LLMs, it achieves 0.862 AUROC predicting the correctness of held-out responses from those same models, and shows zero-shot transfer to thirteen unseen models across eight organizations. Our model needs only a single forward pass to generate an uncertainty estimate and requires no access to the target model’s logits, weights, or internal states. A lightweight text only 0.8B checkpoint matches our largest model’s AUROC. We release code for adding uncertainty estimation to existing repos in only two additional lines of code.
[AI-6] Partner-Specific Affective Precision in Social Active Inference
链接: https://arxiv.org/abs/2609.24876
作者: Harshil Shah,Andrew Pashea
类目: Artificial Intelligence (cs.AI)
备注: 26 pages, 6 figures. Accepted as a full paper at the 7th International Workshop on Active Inference (IWAI 2026). Code: this https URL
Abstract:In multi-agent social settings, model reliability varies across relationships. Beyond inferring what others will do, an agent must calibrate how confidently those inferences should guide policy selection for each relationship. An agent may maintain a well-validated model of one partner, a fragile model of another, and a model under revision for a third; collapsing these into a single confidence estimate loses information relevant to policy selection. We therefore formalize affective precision as a relationship-specific metacognitive estimate of confidence in the current partner model. Each partner’s behavioral evidence updates a local confidence estimate that modulates policy precision during selection, regulating how strongly current beliefs are expressed in policy rather than changing the content of those beliefs. Simulations in a multi-partner graded trust game show that partner-local affective precision influences behavior primarily through policy commitment rather than direct improvement of partner-state inference. Because the mechanism tracks partner-response predictability rather than realized payoff, greater confidence produces sharper policy commitment without necessarily producing higher rewards. Under abrupt shifts in social behavior, confidence accumulated from previously reliable predictions can remain behaviorally active after the relationship changes, showing that confidence revision can lag behind social change. Finally, varying precision gain and priors produce distinct trust-calibration dynamics, showing how confidence accumulation and revision depend on model parameters. Together, these results show how relationship-specific affective precision can distinguish social prediction from social policy commitment.
[AI-7] SE(3) Neural Potential Fields for 6-DoF Trajectory Planning Directly from Images Without Explicit 3D Reconstruction
链接: https://arxiv.org/abs/2609.24864
作者: Jeffrey Eiyike,Masoud Ataei,Elvis Gyaase,Vikas Dhiman
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Reaching a 6-DoF grasp pose in clutter requires a collision-free trajectory, conventionally obtained by reconstructing the scene in 3D and planning inside that reconstruction, at the cost of its accuracy and compute. Potential fields learned directly from images remove that dependency but inherit the classical weakness of artificial potential fields: where attractive and repulsive gradients cancel, the descent grazes the obstacle instead of going around it, and can stall short of the goal. We present an SE(3) neural potential field learned from posed RGB images and supervised with a navigation function, the geodesic distance to the grasp through free space recovered from those same images during training, which removes both failures. On two tabletop scenes, from obstacle-blocked starts executed on a UR10, the field converges within 3 cm of the grasp from every start and every path it executes is collision-free against the ground-truth geometry, against 25% and 0% under image supervision alone; mean clearance rises from under a centimeter to 8.6-8.8 cm and arm-link contacts fall from 20.6-50.4% to 2.7-5.5% of executed configurations. Executed grasp success is 90.0% and 40.0% on the two scenes, the residual failures being refusals of the Cartesian executor rather than of the field. Planning takes about 2 s against 67-133 s for RRT* on a reconstruction of the same images, though under a common offline harness the two are comparable: the deployed margin is the cost of collision-checking a dense reconstruction, not planner complexity.
[AI-8] When Tomorrow Becomes Today: Self-Evolving Policies for Agent ic Time-Series Forecasting
链接: https://arxiv.org/abs/2609.24862
作者: Yifan Hu,Xilin Dai,Zhiyuan Qu,Yiding Liu,Zewei Dong,Jiang-ming Yang,Qiang Xu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Agentic time series forecasting concerns systems whose underlying mechanisms evolve, making the relative effectiveness of numerical models, reasoning strategies, and intervention rules inherently time-varying. Consequently, a time series agent must adapt the forecasts it produces and the orchestration policy that determines which components to trust and how to coordinate them. The deployment process naturally provides supervision for this adaptation as forecast horizons elapse and realized targets reveal the effectiveness of earlier decisions. Committing all numerical expert forecasts and candidate agent paths before target observation allows each realized outcome to evaluate the entire alternative set, providing delayed feedback without additional annotation. However, existing time series agents primarily incorporate prior experience through forecast refinement, reflection, or retrieval, without systematically converting realized outcomes into persistent updates to the joint orchestration policy governing later origins. To exploit this delayed feedback systematically, we introduce TimEvolve, a frozen-backbone time series agent that converts each realized outcome into persistent joint updates of expert trust, agent path selection, and intervention strength. A temporally ordered predict, reveal, and update protocol applies this feedback to subsequent forecasts. Experiments across eight Time-MMD domains show that TimEvolve achieves the best average MSE and MAE ranks among fifteen methods and the lowest errors on both metrics in seven domains. These results demonstrate the value of learning forecasting policies from the futures encountered during deployment.
[AI-9] SPECTRA: Adaptive Execution of Speculative Decoding on a Runtime-Reconfigurable Tiled Architecture
链接: https://arxiv.org/abs/2609.24847
作者: Gabriele Tombesi,William Baisi,Je Yang,Elisavet Lydia Alvanaki,Kevin Lee,Michael Lippe,Biruk Seyoum,Luca P. Carloni
类目: Hardware Architecture (cs.AR); Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC)
备注: Accepted at the IEEE/ACM International Conference on Computer-Aided Design (ICCAD 2026)
Abstract:LLM inference on edge devices is constrained by computational and memory resources, making efficient autoregressive decoding challenging. Speculative decoding alleviates this bottleneck by generating tokens with a smaller draft model and verifying multiple tokens in parallel with a batched target model pass. However, verification introduces a runtime-dependent intermediate regime between memory-bound general matrix-vector (GEMV) operations in decoding and compute-bound general matrix-matrix (GEMM) operations in prefill, as its arithmetic intensity varies with speculation length and acceptance rate. We present SPECTRA, a runtime-reconfigurable tiled architecture that sustains high utilization across the full speculative decoding pipeline. Within each tile, the compute engine switches between systolic execution for GEMMs and vector-lane execution for GEMVs. Across tiles, SPECTRA dynamically adapts computation parallelism by selecting tile count, kernel partitioning, and communication pattern. Both tile-level and system-level reconfiguration operate on a per-kernel basis, enabling efficient execution across these diverse regimes. Evaluated on a 20-tile FPGA prototype across the Pythia, SmolLM2, and GPT-2 families, SPECTRA achieves up to 2.09\times speedup from tile-level reconfiguration and a further 1.25\times gain from system-level adaptability over fixed designs.
[AI-10] MedRSI: Recursive Self-Improvement for Medical Agents via Clinically Aligned Self-Evolution
链接: https://arxiv.org/abs/2609.24838
作者: Junde Wu,Jiayuan Zhu,Minghao Hu,Fenglin Liu,Jiazhen Pan
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Medical agents increasingly combine general reasoning models with specialized clinical tools, yet their capabilities remain largely fixed by what clinicians and engineers design before deployment. Recursive self-improvement (RSI) offers a different paradigm in which agents learn from their own failures and autonomously expand their capabilities, but directly applying RSI to medicine introduces fundamental safety challenges. We introduce MedRSI, the first recursive self-improvement framework for medicine, which continuously transforms diagnostic failures into new clinical capabilities through tool composition and task-specific model training. Inspired by clinical practice, MedRSI introduces two mechanisms for clinically aligned self-evolution. Clinical-cost-aware failure prioritization directs improvement toward errors according to their potential clinical consequences rather than frequency alone. Fast discovery with slow registration separates rapid capability invention from conservative adoption, allowing new tools to enter the persistent agent only after demonstrating sustained benefit across subsequent patient cohorts. Across public glaucoma and heart disease benchmarks and two private clinical tasks, MedRSI progressively develops segmentation, measurement, prediction, multimodal reasoning, and generative capabilities, surpasses manually engineered medical agents, and autonomously discovers solutions to clinical problems not anticipated by its original designers. Our results show that medical agents need not remain constrained by capabilities specified before deployment: with clinically grounded mechanisms governing what to improve and what to retain, they can continuously construct, validate, and accumulate new capabilities from diagnostic experience. Code is available at this https URL.
[AI-11] GRUET: Quantifying Uncertainty of Agent ic Reasoning -and-Acting Processes
链接: https://arxiv.org/abs/2609.24831
作者: Shuang Liang,Xin-Yu Hu,Shao-Qun Zhang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Agents have attracted considerably increasing attention due to the power of executing both Reasoning and Acting (ReAct) in open and dynamic environments. The ReAct process typically exhibits a multi-turn trajectory in which one drives Large Language Models (LLMs) to generate both reasoning chains and task-specific actions in an interleaved manner. However, agents often suffer from significant uncertainty, where identical tasks yield divergent trajectories; trajectories with higher uncertainty often produce incomprehensible behaviors, severely undermining agent credibility. This work conjectures that such trajectory-level uncertainty frequently stems from cumulative turn-level reasoning uncertainty induced by LLMs; the latter often exhibits a collection of branches of divergent reasoning chains and their resulting actions. Built upon this, we present the Graph-based Reasoning UncErtainty in Trajectories (GRUET) method for the uncertainty quantification of ReAct, comprising turn-level reasoning uncertainty quantification and trajectory-level uncertainty aggregation; the former precisely quantifies reasoning uncertainty via modeling the reasoning space spanned by potential reasoning branches as a graph and then approximating the reasoning space complexity with graph complexity, while the latter employs simple aggregation strategies for quantifying the overall trajectory credibility. Empirical evaluations across nine LLMs and five benchmarks validate the effectiveness of our proposed GRUET in terms of selective generation performance, measured by AUROC, AUPRC, and AUARC.
[AI-12] Uranus: Building the Next-Generation Simulation Infrastructure for Embodied AI
链接: https://arxiv.org/abs/2609.24815
作者: Wenkang Qin,Yukun Zhou,Noah Shen,Jisong Cai,Dongxiao Mao,Baicheng Li,Yue Zhang,Wei Sui
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: Project Page: this https URL Inference Code: this https URL Inference Data: this https URL SDK Code: this https URL Model Weights: this https URL
Abstract:Scalable simulation is essential for robot data generation, policy training, evaluation, and safe iteration, yet real-world interaction is costly and conventional simulators require labor-intensive construction. We present Uranus, a data-driven robot simulator built around a joint-trajectory-conditioned autoregressive diffusion model. Uranus offers three key capabilities: (1) streaming, open-ended rollout, which receives future joint-position trajectories online and autoregressively generates one latent frame per step, corresponding to four RGB frames, without a fixed horizon; (2) low-latency generation, achieving 24 FPS after inference optimization; and (3) scalable, extensible robot control, providing a unified interface for synchronized multi-view generation across diverse robot embodiments and camera configurations. We conduct comprehensive quantitative and qualitative evaluations on both in-distribution and out-of-distribution data, providing an objective assessment of Uranus and clearly identifying its current limitations. We release the code and model weights to empower the community with practical tools and insights.
[AI-13] Decoding Guardrails: XAI-Guided Perturbation Analysis of Prompt Injection Detection
链接: https://arxiv.org/abs/2609.24801
作者: Fernando Outeda,Gustavo Betarte,Juan Diego Campo,Fiorella Cravero
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models (LLMs) are increasingly deployed in production systems, raising concerns about their exposure to adversarial manipulation through prompt injection and jailbreak attacks. Classifier-based guardrails, such as Prompt Guard 2, are widely used as a first line of defense against such attacks, but their internal decision logic is largely opaque to both defenders and attackers. This paper presents an exploratory case study that applies explainable artificial intelligence (XAI) techniques to analyze how Prompt Guard 2 distinguishes malicious from benign prompts. We conduct four experiments to probe this question empirically. Guided by Vanilla Gradient and SHAP attributions, we find that Prompt Guard 2’s decisions rely on the cumulative contribution of many tokens rather than a few dominant ones, yet saliency-guided synonym substitution and sentence-level paraphrasing can flip its predictions while altering only a moderate fraction of the text, in some cases yielding a successful jailbreak against the underlying LLM. A dataset-scale saliency analysis further shows that undetected injection prompts systematically lack the lexical markers the classifier relies on. We discuss the implications of these findings for the design and evaluation of classifier-based guardrails, and argue that explanation methods intended to support transparency can simultaneously lower the cost of constructing successful adversarial bypasses. Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI) Cite as: arXiv:2609.24801 [cs.CR] (or arXiv:2609.24801v1 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2609.24801 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-14] Convex AI Compositionality and the Governance of AI System Populations
链接: https://arxiv.org/abs/2609.24784
作者: Andrea Ferrario
类目: Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注: 20 pages, 5 figures, v1
Abstract:AI governance increasingly requires providers and public authorities to reason about multiple AI instantiations, alternative versions, and deployment configurations of multiple AI systems. Yet current regulation remains predominantly single-system-centric, acknowledging such multiplicity only sparsely without treating collections of related AI systems as governance objects. This creates an AI population governance problem: determining which instantiations can be meaningfully considered together and how their changing configurations can be represented and monitored. The first requirement has recently been addressed through trustworthiness-based accounts of AI identity. We address the second by introducing convex AI compositionality: a formal representation of the configurations generated by finite AI system populations that uses convex spaces. The core idea is that convex compositions of the operational states that a population of AI system instantiations may occupy over time are compatible with lifecycle reachability across the population and can preserve the formal identity relations between these systems. Well-known statistical and geometric constructions, such as weighted state distributions and convex hulls, become AI governance tools for distinguishing operational states, population weights, heterogeneity, and AI configuration change across different governance modes while remaining compatible, under stated conditions, with lifecycle reachability and AI identity. We illustrate our AI population governance framework through distributed healthcare deployments and controlled deployment of recruitment AI variants.
[AI-15] Construting Reverse Thinking: Developing Large Language Models Reverse Thingking Ability
链接: https://arxiv.org/abs/2609.24760
作者: Xin Liu,Yunhai Li,Chunfu Jia,Ziliang Chen,Jisen Song
类目: Artificial Intelligence (cs.AI)
备注: 15 pages, 4 figures, 3 tables
Abstract:When facing complex problems, humans tend to try various ideas for different issues. Human thinking patterns exhibit remarkable flexibility in adapting to diverse scenarios. GPT-o1, GPT-o3, and DeepSeek-R1 adopt long chain-of-thought models to address complex problems by increasing reasoning depth, which default to a forward reasoning mode. We conducted statistical analysis on the accuracy of different mathematical problem datasets on models of different scales, and found five reasons for errors: Insufficient solution-space coverage, Computational mistakes, Unverified assumptions, Ignoring constraint conditions, Maximum response length limitation. To address the above issues, we proposed a backward reasoning pattern construction method aimed at enhancing the model’s reverse thinking ability and dynamic adaptability. First, we constructed an easy-hard two-stage Math dataset for training large models and gradually improving their inference ability at different difficulty levels. The dataset contains forward reasoning paths as well as backward reasoning paths. And a two-stage supervised fine-tuning process is applied to progressively train the model’s backward reasoning capability. Furthermore, a fine-grained reward mechanism is developed, employing smoothed reward signals to strengthen the model’s ability to autonomously select thinking modes during the reasoning process, thereby avoiding reward hacking. A linear-decay balanced sampling strategy is designed to maintain a balance between forward and backward reasoning path samples during training, enabling the model to converge quickly and stably. Experimental results show that our method significantly improves reasoning efficiency and accuracy in tasks such as mathematical proofs, offering a flexible and efficient reasoning paradigm for solving complex problems.
[AI-16] NPU Accelerator: Quantized Real-Time Vehicle Detection on PYNQ-Z1 Using FINN
链接: https://arxiv.org/abs/2609.24757
作者: Daniel Gutierrez,Antonio Cuesta,Jorge Fe,Bruno Gutierrez,Rashed Al Koutayni
类目: Hardware Architecture (cs.AR); Artificial Intelligence (cs.AI)
备注:
Abstract:This paper presents the design, optimization, implementation, and on-board validation of a neural processing unit (NPU) accelerator for real-time vehicle detection on the resource-constrained Xilinx Zynq XC7Z020 device of the PYNQ-Z1 board. The work follows a hardware/software co-design methodology that combines quantization-aware training (QAT), lightweight YOLO-derived detectors, Brevitas/QONNX model export, FINN dataflow compilation, Vivado implementation, and physical benchmarking on the target board. Four simultaneous engineering requirements define successful deployment: throughput above 30 frames/s (FPS), energy efficiency above 7 FPS/W, programmable-logic (PL) hardware latency below 50 ms, and Pascal VOC detection accuracy above 0.55 mAP@0.5. The design space includes LP-YOLO and LP-YOLO Slim variants, a custom YOLOv3-tiny reference, 4-bit and mixed low-bit quantization, 320 \times 320 and 256 \times 256 inputs, manual and automatic FIFO sizing, and programmable-logic clocks from 100 to 200 MHz. The final LP-YOLO Slim configuration uses a 256 \times 256 input, w2a4 quantization, and a 142.86 MHz PL clock. With batch 100 it reaches 35.66 FPS at 2.91 W, corresponding to 12.25 FPS/W, while measured PL latency is 45.11 ms and VOC mAP@0.5 is 0.594. This is the only evaluated configuration for which the supplied measurements satisfy all four requirements simultaneously. The results show that low-bit QAT, architectural slimming, FINN folding and FIFO optimization, and moderate clock scaling can jointly provide a practical real-time detector on a small Zynq FPGA.
[AI-17] Epi-Logic: A Conceptual Framework for Epistemic Runtime Control Schema Validity Checking and Controlled Accommodation in Autonomous AI Agents
链接: https://arxiv.org/abs/2609.24755
作者: Boris Wetzk
类目: Artificial Intelligence (cs.AI)
备注: 26 pages, 1 figure, 1 table
Abstract:Autonomous AI agents are increasingly deployed in areas where wrong decisions are hard to reverse. This paper examines schema mismatch: the condition in which an agent operates within an interpretive frame that no longer applies to the current context. Outputs produced under such a mismatch can appear internally consistent, linguistically plausible, and largely factually correct; output-quality metrics alone therefore capture the underlying loss of validity only partially. The paper introduces Epi-Logic, a conceptual framework for epistemic runtime control. It couples the detection of schema dissonance, a graduated reduction of autonomy, and the auditable switch to a validated schema. A schema is formalised as a tuple of variable space, expectation model, validity conditions, axioms, and metadata. The Epi-Score aggregates seven graded dimensions of epistemic dissonance; the temporal validity dimension D8, violations of the validity conditions G, and axiom violations are carried as separate categorical paths that are not offset against the aggregate. The architecture rests on a checking asymmetry: formalised validity conditions can be checked at runtime, whereas the correctness of many actions is established only ex post. The paper separates two architectural properties, a conditional result from sequential changepoint detection, and an empirical remainder. Eight falsifiable propositions with named baselines describe the transition to empirical validation. All propositions are empirically testable hypotheses, not established results. Comments: 26 pages, 1 figure, 1 table Subjects: Artificial Intelligence (cs.AI) Cite as: arXiv:2609.24755 [cs.AI] (or arXiv:2609.24755v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2609.24755 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-18] Enhancing Transformer Representations of Symbolic ODE Expressions
链接: https://arxiv.org/abs/2609.24746
作者: Xiyue Fan,Adam Prugel-Bennett,Stuart E. Middleton
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 13 pages, 9 figures
Abstract:Existing approaches to solving differential equations, such as symbolic regression, physics informed neural networks, and neural operators, typically focus on numerical approximations or blind symbolic search via fitting to numerical data. Less attention has been paid to learning structured representations of mathematical expressions that preserve commutative properties and could support mathematical reasoning in symbolic forms. Transformer models have shown strong capabilities in solving symbolic differential equations. However, standard positional embeddings in transformers are designed for sequence data. Symbolic differential equations are naturally represented by expression trees, so these positional embeddings may not efficiently capture their hierarchical structures. We investigate existing tree positional embeddings in symbolic ordinary differential equation (ODE) tasks. We systematically study their effectiveness under different settings. Our results show that tree positional embeddings aid learning in early epochs and continue to improve performance throughout, ultimately yielding consistent advantages across various data sizes and tasks. Based on learned structural representations, we apply contrastive learning to support the commutative property in mathematics. Ablation studies provide insight into how these methods interact in modelling symbolic mathematical structures.
[AI-19] World State Generator
链接: https://arxiv.org/abs/2609.24744
作者: Sungheon Jeong,Sanggeon Yun,Ryozo Masukawa,Haleh Alimohamadi,Mahdi Imani,Mohsen Imani
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Language agents solve complex tasks through plans and actions. A single step the world refuses puts the goal out of reach, and what the agent does next decides the task. Prompted planners fail at exactly this point, rewriting the refused step in new words, meeting the same refusal, and burning the attempt budget without moving. They fail because the plan was never tied to the world, so a refusal has nothing in the plan to attach to. A world is where a task runs, and it has its own rules, its own admissible actions, and its own constraints. We build synthetic worlds across 7 domains and extract training data from them. A program enforces each world’s rules and grades its goal, and every world is admitted only if its goal is reachable from its initial state. Agents run inside and leave verified failures paired with repairs that carried the run to a state the world certified, a record of about 226K trajectories. On this record we train the World State Generator, a model that writes a plan as checkable states of the world and keeps that plan aligned with the world it runs in. That alignment is what a plan written in language lacks, since the world it runs in has physical limits, logical dependencies, and required orders the language never states, and the plan encounters these rules only when a state fails. WSG takes that failure as the rule the world has stated and rewrites the remaining states to obey it, so the plan bends to the world as the run goes on. Across 7 public benchmarks, WSG raises end-to-end success for two open models near 30B parameters over prompting and brings to the level of proprietary model.
[AI-20] LLM -based Conversational AI Knowledge Assistant for MyBuddy Humanoid Robot NEURIPS2026
链接: https://arxiv.org/abs/2609.24742
作者: Hanxiao Chen
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: This work has been accepted as poster presentation for NeurIPS 2026 WiML Workshop
Abstract:Humanoid robots are increasingly being popular and developed for human-centered applications, yet their ability to provide intelligent conversations and natural interactive knowledge assistance remains constrained by traditional rule-based dialogue systems, pre-defined responses and limited knowledge repositories. Large language models (LLMs) have emerged as a powerful foundation for enabling natural, adaptive, and context-aware Human-Robot Interaction (HRI), which provides a significant opportunity to address such limitations by enabling robots to understand natural speech language, reason over complicated queries, maintain high-quality conversational context, and generate knowledge-rich responses. In this work, we originally present and implement an LLM-based versatile Conversational AI Knowledge Assistant for the Raspberry-Pi-powered 13-Axis MyBuddy humanoid robot, which integrates LLM-driven language understanding and AI reasoning with real-time speech recognition, knowledge retrieval via extensible access of internet engines (e.g., Wikipedia, arXiv), flexible dialogue management, and natural speech synthesis to enable much more intelligent multi-turn continuous conversations and advanced emotional-support Human-Robot Interaction.
[AI-21] Reasoning Topology Matters: A Controlled Study of LLM -Based Cybersecurity Analysis
链接: https://arxiv.org/abs/2609.24710
作者: Jiling Zhou,Aisvarya Adeseye,Antti Hakkala,Seppo Virtanen,Jouni Isoaho
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: Accepted at AIAIS 2027
Abstract:Large Language Models (LLMs) are increasingly used in cybersecurity, where accurate analysis often requires multi-step and context-dependent reasoning over complex and heterogeneous data. However, existing prompting approaches typically focus on eliciting reasoning without explicitly considering how intermediate reasoning steps are structurally organized. We introduce Security Reasoning Topology, which models reasoning through three representative structures: Linear, Branching, and Graph. To evaluate their effects, we conduct controlled experiments on three cybersecurity datasets covering MITRE ATTCK network traffic, cyber threat intelligence (CTI), and CVE vulnerability analysis. We evaluate multiple LLMs, including Llama 2 (7B, 13B, 70B), GPT-5.1, and Mistral Large 3, while keeping task inputs consistent and controlling reasoning structure through system-level prompting. Results show that reasoning topology substantially affects performance: Graph reasoning achieves the highest overall accuracy, improving over few-shot prompting by 9.8-12.2 percentage points across datasets, while Branching provides a strong intermediate solution. The results further show that the effect of reasoning topology remains consistent across model families and scales, highlighting reasoning topology as an important design factor for LLM-based cybersecurity analysis.
[AI-22] Understanding Hyperspherical Geometry of ECAPA-TDNN Embedding and Its Impact on Zero-Shot Voice Conversion
链接: https://arxiv.org/abs/2609.24688
作者: Mathilde Abrassart,Nicolas Obin,Axel Roebel
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注:
Abstract:Angular-margin speaker encoders are widely used in voice conversion, yet the geometry of their classifier prototypes remains poorly understood. We analyze ECAPA-TDNN classifier prototypes as points on the unit hypersphere and characterize their organization using rotation-invariant angular statistics together with global and local effective dimensionality measures. Our analysis shows that standard training can induce angular concentration and a substantial reduction in effective dimensionality. To address this, we investigate two geometric regularization strategies (hinged Riesz log-energy and effective-dimension maximization) applied to classifier prototypes to encourage more uniform hyperspherical coverage. The resulting prototype sets exhibit higher effective dimensionality and improved isotropy, with configuration-dependent effects on speaker-recognition performance. When the corresponding ECAPA-TDNN models are used as speaker encoders for Fast-VGAN, the regularized systems also exhibit improved robustness in zero-shot voice conversion, particularly for previously unseen speakers.
[AI-23] meLitmus: A Diagnostic Benchmark for Cross-Modal Understanding and Explanation Faithfulness in Event-Conditioned Time-Series Prediction
链接: https://arxiv.org/abs/2609.24677
作者: Jie Gong,Maowei Jiang,Zhiwei Liu,Yankai Chen,Guojun Xiong,Xue Liu,Min Peng,Qianqian Xie,Sophia Ananiadou
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models (LLMs) are increasingly used to make predictions from numerical time-series histories and textual events. Yet accuracy alone cannot reveal whether correct answers reflect effective integration of the two inputs or instead arise from event polarity, unimodal priors, or superficial cues. Likewise, plausible explanations may rationalize predictions without faithfully reflecting the evidence that drives model behavior. We introduce TimeLitmus, a diagnostic benchmark for cross-modal understanding and explanation faithfulness in event-conditioned time-series prediction. TimeLitmus contains 4,856 evaluation records across Finance and Traffic, combining natural prediction with controlled counterfactual and contrastive interventions, explanation-targeted faithfulness tests, and systematic shortcut controls. Across ten representative LLMs, standard prediction accuracy substantially overstates reliable cross-modal understanding: Hard Paired Contrast (HPC) pair correctness peaks at only 19.2% in Finance and 11.7% in Traffic, and all ten models show lower-than-expected consistency on Finance series-side controls. Models often recognize scenario relations explicitly yet fail to apply them during independent prediction. Explanation faithfulness shows a similar gap: in Traffic, most models cite the manipulated temporal factor in over 90% of cases, while behavioral support remains below 22%. Human annotators outperform LLMs on matched controlled and hard-pair diagnostics, confirming that these distinctions are recoverable from the inputs. Natural-only adaptation yields selective gains in evidence selection and input sensitivity, but not consistent gains in controlled or hard-pair behavior. The benchmark, evaluation suite, and supervised adaptation data will be released publicly.
[AI-24] rust in Edge-Enabled IoT Security: Features Challenges and Research Directions
链接: https://arxiv.org/abs/2609.24669
作者: Esin Ece Aydın,Şerif Bahtiyar,Gürkan Gür
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computers and Society (cs.CY); Emerging Technologies (cs.ET); Networking and Internet Architecture (cs.NI)
备注: 34+ pages, 9 figures, 12 tables, submitted and under review in ACM Computing Surveys
Abstract:Providing autonomous intelligence, pervasive connectivity and usability to human life and industry has led to the emergence of the Internet of Things (IoT). To support time-sensitive and resource-constrained applications, IoT systems nowadays increasingly rely on edge computing. This brings computation and decision-making closer to end devices. In edge-enabled IoT architecture, latency and communication overhead are reduced, but interactions among a larger and more diverse set of devices, edge nodes, services, and data sources are introduced as well. In such environments, security and privacy mechanisms provide the foundation for protection, while trust management can assess the reliability of interacting entities and adapting secure decisions. In this paper, we systematically review the current state of trust management in edge-enabled IoT. To this end, we propose a comprehensive taxonomy that maps physical, network, and application architectural IoT layers against the consumer, commercial, industrial, and infrastructure IoT domains. We further investigate state-of-art research based on their trust design, how trust integrated into secure IoT operations, the attacks that effect trust management process. Based on these findings, we identify key gaps in current research and outline future directions for context-aware and adaptive trust management in edge-enabled IoT.
[AI-25] Beyond Endpoint Performance: Process-Level Evaluation of Self-Evolving Agents
链接: https://arxiv.org/abs/2609.24663
作者: Hongqiang Lin,Chao Liu,Xiaofan Bai,Xuan Jin,Yuhong Li,Nenggan Zheng,Xipeng Cao
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Self-evolving agents convert interaction feedback into persistent artifacts, such as memories or skills, which in turn guide subsequent decisions. As these artifacts are iteratively updated throughout an experience stream, the capabilities they support may evolve. Consequently, endpoint performance alone offers an incomplete view of self-evolution. Process-level evaluation is therefore essential to identify when a target capability emerges and whether later updates strengthen, preserve, or weaken it. Motivated by this, we propose \textscEvoPathBench, a benchmark that tracks individual capabilities during artifact-level self-evolution. EvoPathBench fixes the base model, tools, freezes evolving artifacts at successive checkpoints, and evaluates the target capability on held-out episodes. This benchmark evaluates agent self-evolution using public trading data and calibrated trajectories. It tests three capabilities: generalization to unseen tasks, retention after unrelated learning, and rule adaptation to new evidence. Experimental results show that gains on similar unseen tasks often weaken under distribution shift, retention losses are concentrated in a minority of evolution paths, and no method achieves reliable rule adaptation. Moreover, while self-evolution enables agents to generate candidate artifacts with substantial held-out gains, the selected updates consistently fall short of realizing this potential. Together, these findings establish capability-level process evaluation as a foundation for analyzing self-evolution, identifying candidate evaluation and selection as key targets for improvement.
[AI-26] DUMA-Bench: A Dual-Control Multi-Agent Benchmark for Evaluating LLM Agent Benchmark for Evaluating LLM Agent Security ACL
链接: https://arxiv.org/abs/2609.24662
作者: Ivan Aleksandrov,German Kochnev,Sabrina Sadiekh,Yaroslav Rogoza
类目: Artificial Intelligence (cs.AI)
备注: ACL ARR 2026 March Findings
Abstract:LLM-based agents increasingly operate in environments where they interact with users, tools, and external systems. Yet most security evaluations assume passive users and static control, ignoring the interactive dynamics that shape real agent behavior. We introduce \textbfDUMA-Bench, a benchmark and evaluation protocol for measuring agent security under \emphdual-control interaction, where both the agent and the user can influence the shared environment state. DUMA-Bench extends \tau^2 -bench ~\citebarres2025tau with adversarial environments covering eight vulnerability classes, including RAG poisoning, cross-agent manipulation, and unsafe output handling. We evaluate \textbf14 models from five model families (OpenAI, Anthropic, DeepSeek, Qwen, and this http URL) across eight domains and multiple user-behavior regimes. Across our experiments, introducing dual-control interaction increases the attack success rate from \textbf26.9% to \textbf41.1%. These results show that agent security is not solely a property of the model but emerges from the interaction between the model, the user, and the environment. DUMA-Bench provides a missing evaluation layer for studying security in realistic agent deployments.
[AI-27] ouch2Robot: Robot Touch in the Human Demonstration Loop
链接: https://arxiv.org/abs/2609.24660
作者: Shengcheng Luo,Xiaoyang Cheng,Hong Ying,Xiaoying Zhou,Jiaming Jiang,Haoran Guo,Wanlin Li,Ziyuan Jiao,Chenxi Xiao
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: 12 pages, 13 figures
Abstract:Human demonstrations offer a scalable way to collect manipulation data, but their contacts may be unstable or infeasible when transferred to a robot hand. Collecting demonstrations directly on the target robot avoids this mismatch, but substantially increases the cost of data collection. To address this trade-off, we present \textbfTouch2Robot, a framework that lets humans collect demonstrations while seeing how the target robot hand would contact the object. We capture human hand motion, tactile-glove measurements, and object motion during human manipulation. These recordings guide object-specific RL policies to reproduce the demonstrated object motion while favoring contacts consistent with the recorded human touch. We distill the learned behaviors into a unified real-time retargeter that maps incoming human observations and object geometry to robot hand configurations. During collection, the predicted robot configuration is synchronized with the tracked object pose in simulation to reconstruct robot-object contacts, which are visualized to help the demonstrator adapt subsequent interactions to the target hand. Across four real-world tasks, Touch2Robot improves average real-robot replay completion from 37.9% to 72.1% over visual-only feedback, while reducing the collection time per replay-successful demonstration from 58.6~s to 18.2~s. Reconstructed target-hand contacts achieve 44.2% F1 against real-robot tactile measurements, and policies trained on Touch2Robot demonstrations improve downstream Diffusion Policy performance by 29.1 percentage points over visual-only feedback. These results show that bringing robot touch into the human demonstration loop improves both the quality and efficiency of scalable dexterous data collection. \textitProject webpage: \hrefthis https URLthis https URL.
[AI-28] Corrective Forcing: Unified Post-Training for Diffusions and Flows in Generative Speech Enhancement ICASSP2027
链接: https://arxiv.org/abs/2609.24651
作者: Qing Yao,Lijian Gao,Qirong Mao
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Submitted to ICASSP 2027
Abstract:Diffusion and flow models, as promising generative paradigms for speech enhancement, face a training–inference mismatch: training uses analytical path states, whereas inference recursively evaluates models on self-generated rollout states along discretized sampling trajectories. This mismatch causes prediction and discretization errors to accumulate. To address it, we introduce Corrective Forcing (CoF), a post-training paradigm that forces diffusion and flow models to learn from self-generated rollouts and correct their predictions. CoF corrects clean-speech predictions on rollout states toward the ground truth under dynamic sampling schedules, exposing the model to varying inference conditions. It further regularizes local evolution using locally corrected counterfactual transitions as references for factual transitions. By expressing model outputs through a shared clean-speech prediction parameterization, CoF applies the same post-training objective across diffusion and flow formulations. Experiments with SB-VE and OT-CFM demonstrate improvements in perceptual quality and reconstruction fidelity, together with robust performance across different numbers of sampling steps.
[AI-29] SDFT: Information-Proximal Self-Distillation for Continual Learning in LLM s
链接: https://arxiv.org/abs/2609.24646
作者: Ahmed Khaled Khamis,Xiaotong Ji,Hassan Jaber,Rasul Tutunov,Matthieu Zimmer,Jun Wang,Haitham Bou-Ammar
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:On-policy self-distillation fine-tuning (SDFT) learns new skills from demonstrations while reducing forgetting, but it always distils toward the full demonstration-conditioned teacher. This fixes teacher influence at the full-teacher endpoint, providing no control over how much demonstration information should be transferred at each prediction state. We introduce Information-Proximal SDFT (iSDFT), which instead treats the teacher as a budgeted source of information. At each token, iSDFT selects the distribution closest to the current student that satisfies a prescribed teacher-information constraint, yielding a closed-form exponential target with a locally determined tilt. To control cumulative drift, we further anchor the student to its frozen base policy. Across four heterogeneous LLM backbones and two specialisation tasks, iSDFT improves vanilla SDFT in 7 of 8 model-task settings and matches it in the remaining one. It also provides tighter retention on the original SDFT benchmark suite, with 73% of evaluations remaining within 0.5 points of the base model versus 52% for the strongest baseline, while achieving the largest mean improvement on all ten additional mathematics, coding, and competition-mathematics benchmarks. These results show that controlling how much and when teacher information is introduced improves specialisation while preserving broader capability.
[AI-30] From Semantic Decisions to Feasible Trajectories: Self-Evolving LLM -Guided Optimal Control for Narrow-Space Parking
链接: https://arxiv.org/abs/2609.24631
作者: Zhengbao Yao,Yuanfu Luo,Kehan Xue
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Autonomous parking in nonconvex and narrow environments remains challenging. Although optimal-control methods can explicitly enforce vehicle dynamics and collision constraints, nonconvexity compromises solver robustness and can cause failures. Large language models (LLMs) exhibit strong semantic reasoning capabilities, but directly generating dense trajectories makes it difficult to guarantee physical feasibility. We introduce SE-LLM-OCP, a unified framework in which LLMs make high-level discrete maneuver decisions, while an optimal-control module enforces low-level vehicle dynamics and collision constraints. Online, the LLM proposes sparse maneuver plans, decomposing the parking task into a sequence of short-horizon trajectory-optimization problems. A low-level solver then sequentially solves optimal-control problems. If the solver fails, the LLM aggregates failure evidence from the solver and validation stages to guide replanning. Offline, SE-LLM-OCP automatically evolves a structured decision-making knowledge base from scratch, driven by accumulated online failures. We validate our proposed framework in simulation on a car-like vehicle model and on a differential-drive robot. Our experimental results show that SE-LLM-OCP enables safer autonomous parking in narrow scenarios and demonstrates transfer of the same maneuver representation to a different kinematic platform.
[AI-31] Augmented Hypothesis Testing with Persona-Based LLM Simulations
链接: https://arxiv.org/abs/2609.24629
作者: Ziyad Benomar,Aymen Al Marjani,Paul Missault,Saab Mansour
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Applications (stat.AP)
备注: Work accepted at COLM Workshop on Agent Behavior
Abstract:A/B testing requires large sample sizes, long timelines, and significant costs. When auxiliary predictions of experimental outcomes are available from machine learning models, uncertain prediction quality precludes replacing human experiments entirely, yet these predictions may still contain useful signal. We propose a principled framework for learning-augmented hypothesis testing that leverages predictions of unknown quality to reduce sample sizes while maintaining statistical validity. Predictions naturally vary in granularity, from coarse aggregate signals to fine-grained individual-level estimates, and our framework addresses both ends of this spectrum: (1) for population-level directional predictions, where only a binary signal on the treatment effect sign is available, we use an asymmetric test and prove consistency and robustness bounds within the learning-augmented algorithms paradigm; (2) for individual-level predictions, we introduce Generalized PPI++ (GPPI), extending Prediction-Powered Inference to handle nonlinear prediction errors through higher-dimensional transformations. Both methods benefit from accurate predictions while remaining robust to inaccurate or adversarial ones. We validate our framework using persona-based LLM simulations, where AI agents equipped with user personas predict individual behavior, as a natural prediction source spanning both granularity levels. Experiments on four real-world datasets demonstrate that our methods, combined with persona-based predictions, substantially reduce experimental costs while preserving rigorous statistical validity.
[AI-32] GraphToolbox: A Configurable Python Framework for Graph Neural Network Forecasting
链接: https://arxiv.org/abs/2609.24609
作者: Eloi Campagne(CB),Yvenn Amara-Ouali(LMO, CELESTE),Yannig Goude(EDF R\amp;D),Argyris Kalogeratos(CB, ENS Paris Saclay)
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Electricity forecasting often involves spatially related signals observed over regions, substations, and feeders, and Graph Neural Networks (GNNs) provide a natural way to represent these relations. Building a complete GNN forecasting experiment is nonetheless laborious, because graph construction, model selection, training, aggregation, and interpretation sit in incompatible tools. We present GraphToolbox, an open-source Python framework that unifies these stages in one configurationdriven pipeline built on PyTorch Geometric. It offers data-driven graph construction, an adapter that instantiates and trains 51 of the 65 PyTorch Geometric convolutions together with the recurrent cells of PyTorch Geometric Temporal, online expert aggregation, forecasting interpretability, and significance testing on cached forecasts. We evaluate the pipeline in two case studies. On French regional load, the 48 convolutions included in the complete forecasting sweep fall in a band from 1.14% to 1.60% error, online aggregation lowers this to 0.98%, and the graph models improve on classical additive and boosting baselines. On net-load, direct graph models are less accurate than a classical additive model, while forecasting each physical component separately improves them without closing that gap. Both comparisons use the same experimental interface, illustrating the role of GraphToolbox in systematic architectural evaluation.
[AI-33] Overlay_dx - Automating forecasting evaluation
链接: https://arxiv.org/abs/2609.24586
作者: Long Ngo,Mohammed Amine Chamli,Jonathan Rivalan,Thomas Jaillon
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Traditional evaluation metrics provides numerical values but often lack comprehensibility, hindering effective differentiation of model performances. Our work addresses this challenge by introducing overlay_dx, a novel evaluation metric measuring the performance of time series prediction models. Overlay_dx is a visual metric that represents the percentage of predictions falling within a confidence interval around actual values. Additionally, once evaluation results are plotted, overlay_dx computes the area under the overlay curve, providing a quantitative measure of alignment between predicted and actual values across different thresholds and predictions. Through extensive experiments, we demonstrate that our approach offers a unified evaluation framework that combines both visual and numerical assessments, enabling improved model comparison and providing valuable insights for further research and optimization efforts in time series prediction.
[AI-34] t_0: A Time-Series Foundation Model for Forecasting with Context
链接: https://arxiv.org/abs/2609.24559
作者: Lucas Meyer,Claudio Sole,Huikan Xiang,Nicolas Li,Lucas Franceschino,Arnau Quera-Bofarull,Maarten P. Scholl,Joachim Fainberg,Geoffrey Négiar
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 39 pages, 16 figures, 13 tables
Abstract:We present t_0 , a family of open-weights foundation models for forecasting with multivariate context. We release its first two members: \textttt0-alpha and \textttt0-beta , respectively 102M and 256M parameters. Both condition their forecasts on target history, past covariates, and known-future covariates, without task-specific retraining. Their transformer layers alternate attention along time and across variates. They produce probabilistic forecasts through quantile predictions. Pretraining combines curated public data with synthetic generator families constructed to contain covariate-to-target dependencies. On GIFT-Eval, \textttt0-alpha reaches an aggregate CRPS of 0.4941, and \textttt0-beta a CRPS of 0.4738 and a MASE of 0.6865, third on both and within 4.0% of the best zero-shot TSFM. On fev-bench they score 42.2 and 46.7 in skill, the latter third again and 2.0 points behind the leader. We analyze \textttt0-alpha in depth. Known-future covariates raise its skill by 6.3 percentage points across 30 tasks. The report also examines its calibration, its rollout strategy on long horizons, and its robustness to missing data. On the Victoria electricity-demand benchmark, \textttt0-beta is among the most accurate models with a context of nearly a year. In an independent Macrocosm evaluation of hourly ERCOT prices over 29 months, both cut the MAE of the lagged-price baseline by 38%.
[AI-35] he Endless Exam: Mathematical Constructions from Todays Models toward Superintelligence
链接: https://arxiv.org/abs/2609.24555
作者: Muhan Zhang
类目: Artificial Intelligence (cs.AI); History and Overview (math.HO)
备注: 51 pages, 13 figures, 25 tables
Abstract:We introduce the Endless Exam, a benchmark for measuring mathematical progress from today’s models toward artificial superintelligence through fourteen parameterised construction families. Each submitted object is checked automatically for validity and assigned a relative quality score against a published frontier or construction baseline, without capping improvements at 1 . The families draw on open mathematical problems for long-term targets and generate new instances at larger parameters, where compact certificates keep large constructions verifiable. Across eight models evaluated on 69 distinct instances, continuous quality scores distinguish performance even though no evaluated system surpasses a published frontier. Size-quality curves show how construction quality changes as problem size increases. We release the generators, verifiers, references, model responses and analysis to support continued measurement before and beyond human frontiers.
[AI-36] Beyond Predictable Paths: Redefining AI Security Incident Reporting for Agents
链接: https://arxiv.org/abs/2609.24515
作者: Anastasia Pustozerova,Eugene Bagdasarian,Luca Beurer-Kellner,Battista Biggio,Nico Ebert,David Filip,Marc Fischer,Heather Frase,David Hofer,Juliane Hoffmann,Daphne Ippolito,Somesh Jha,Sean McGregor,Esfandiar Mohammadi,Luca Nannini,Cristina Nita-Rotaru,Alina Oprea,Kevin Paeth,Andrew Paverd,Jonathan Petit,Andreas Rauber,Christian Riess,John Sotiropoulos,Andreas Wespi,Kathrin Grosse
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: under submission, mega paper (authorship does not imply endorsement of every sub-section)
Abstract:AI agents are being deployed rapidly, accompanied by a growing number of AI-specific attacks and corresponding incidents. As incident reporting becomes increasingly important for legal compliance, governance, accountability, and security; current frameworks must be adapted to the unique characteristics of AI agents. In this paper, two editorial authors compare AI systems and AI agents and, drawing on input from 23 experts in academia and industry, identify the information required for reporting incidents where the security of AI agents is harmed. %involving AI agents. Potential reporting elements include, for example, agent memory and memory accesses, actual and potential levels of autonomy, and tool usage. Based on these findings, we identify several open research questions, including how to efficiently record incidents and how to determine whether vulnerabilities and incidents generalize. Expert feedback also highlighted potential reporting weaknesses, such as risks of data leakage and attacks targeting the reporting infrastructure itself, creating additional research needs. Lastly, we summarize privacy requirements and outline research directions for the secure and trustworthy deployment of AI agents.
[AI-37] On Emergent Capabilities and Model Merging
链接: https://arxiv.org/abs/2609.24504
作者: Luca Zhou,Emanuele Rodolà
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: main paper has 8 pages, 5 figures, and 4 tables
Abstract:Fine-tuned checkpoints and adapters now fill public repositories, and the most common operation applied to these artifacts is model merging: arithmetic on their weights that assembles capabilities cheaply. We ask what this operation does to emergent capabilities: behaviors an artifact carries that were never an explicit training target. Studying two independent testbeds (activation oracles and emergent-misaligned models) across three model families, we find that the answer is threefold. First, merging preserves an emergent capability that both parents carry: merging two misaligned checkpoints retains most of their broad misalignment across the whole mixing range. Second, merging cannot create an emergent capability that is superadditive in its parents: no weighted merge of two single-task oracles reaches the jointly-trained oracle’s auditing ability. Third, when only one parent carries the capability, merging dilutes it faster than the trained capability that accompanies it: the gap is significant in most settings. In short, emergent behaviors of an artifact do not compose the way its trained capability does.
[AI-38] Lifted Bellm an Linear Programming for Offline Reinforcement Learning
链接: https://arxiv.org/abs/2609.24489
作者: Hyukjun Yang,Jongchan Park,Narim Jeong,Donghwan Lee
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Offline reinforcement learning (RL) typically trains a critic by minimizing a regression loss against bootstrapped value targets stabilized by target networks with exponential moving average (EMA) updates. Multi-step targets incorporate behavior-policy actions and therefore require off-policy correction. We instead impose in-sample Bellman optimality on the critic through inequality constraints. We formulate the Lifted Bellman Linear Program (LBLP), which lifts the linear programming characterization of Bellman optimality to the joint (Q,V) space so that every constraint involves only state-action pairs in the dataset. Its unique minimizer is the in-sample optimal pair, and constraints along K -step segments of dataset trajectories leave this minimizer unchanged for any rollout policy and horizon. Under deterministic dynamics, this minimizer lies between the best dataset return and the optimal value. Relaxing the constraints into hinge penalties recovers the same solution above a finite penalty coefficient in the tabular case. Approximate Lifted Bellman Unconstrained Minimization (ALBUM) implements this relaxation with neural networks and detaches the K -step rollout targets by stop gradient. Its objective contains no squared regression onto bootstrapped targets, so it can be trained without target networks or EMA updates. Under deterministic dynamics, the LBLP solution is a stationary point of the detached update under a coefficient condition independent of \gamma and K , and the inequality constraints allow discounted returns along dataset trajectories to serve as lower bounds without off-policy correction or action chunking. On OGBench, ALBUM uses a single critic with a Gaussian policy, matches the average performance of FQL, and is comparable to recent action-chunking methods, while using the fewest parameters and the least peak GPU memory among all compared methods.
[AI-39] Conduit: An Experience Data Plane for Distributed Reinforcement Learning
链接: https://arxiv.org/abs/2609.24456
作者: Sitong Zhang,Tuo Shi,Mario Di Francesco,Zeke Wang,Bo Zhao
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Artificial Intelligence (cs.AI)
备注: 16 pages, 17 figures
Abstract:Distributed reinforcement learning (RL) scales training by parallelizing actors and learners around an Experience Buffer. As RL workloads grow, however, the buffer becomes more than a replay queue: it is the storage substrate of a large-capacity, latency-critical experience path that every iteration traverses to move, transform, sample, and batch experiences before learner updates can begin. Existing RL systems embed this path inside framework control flow or expose it as a request-driven buffer service, leaving experience placement fixed and experience-path work difficult to schedule independently as a runtime-level optimization target. We present Conduit, a framework-agnostic runtime that exposes RL experience management as an explicit systems optimization problem. At its core is the Experience Data Plane (EDP), a runtime abstraction that separates RL experience-handling semantics from framework-specific execution logic by exposing experience ingestion, experience placement, and experience delivery as explicit control points. Built on EDP, Conduit introduces capacity-constrained, bandwidth-aware placement, which distributes experience state across CPU/GPU memory tiers and nodes under heterogeneous interconnect and device-memory constraints, and latency-aware scheduling, which controls when experience-path handling runs to reduce exposed experience-path latency while preserving RL semantics. Integrated with RLlib without changing its framework execution logic, Conduit reduces exposed experience-path latency by up to 97% and end-to-end iteration latency by up to 38%, scales to 1,024 GPUs, and preserves convergence.
[AI-40] Predicting Postprandial Glycemic Response from Meal Images Clinical Variables and Gut Microbiome Information MICCAI2026
链接: https://arxiv.org/abs/2609.24453
作者: Varvara Kondratyeva,Kamilia Zaripova,Nassir Navab,Azade Farshad
类目: Artificial Intelligence (cs.AI)
备注: 11 pages (9 text + 2 references). This is a paper first submitted to MICCAI 2026 MultiTab workshop prior to peer review. The final revised version will be published in Springer LNCS proceedings after the MICCAI 2026 conference
Abstract:Predicting postprandial glycemic response (PPGR) is fundamental to personalized nutrition and type 2 diabetes management, yet existing approaches typically rely on manually reported dietary intake, limiting their scalability in free-living settings. We propose a multimodal framework that replaces manual dietary logging with image-derived macronutrient estimates and integrates them with clinical variables and gut microbiome information for personalized PPGR prediction. The framework jointly performs image-based macronutrient estimation and glucose prediction, while an attention-based prediction module models interactions between dietary and host-specific information. We evaluate the proposed approach on a real-world dataset comprising meal images, continuous glucose monitoring, clinical variables, and gut microbiome profiles. The proposed model outperforms existing PPGR baselines using image-derived nutritional inputs and approaches the performance of methods that rely on manually reported macronutrients despite using automatically estimated nutritional information. These results demonstrate that combining image-derived nutrition with complementary clinical and gut microbiome information provides a practical foundation for scalable personalized PPGR prediction.
[AI-41] ActGov: Governing LLM Agent Actions via Policy-Constrained Validation
链接: https://arxiv.org/abs/2609.24446
作者: Kaiyuan Zhang,Yuke Peng,Ke Jiang,Yinqian Zhang
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:Large language model (LLM) agents increasingly execute long-horizon workflows through external tools, allowing untrusted outputs to influence subsequent actions and exceed user authorization. Existing defenses isolate injected content or constrain execution with predefined plans and static policies, but these approaches are brittle under dynamic workflows and scale poorly across extensible tool ecosystems. In this work, we present ActGov, a runtime enforcement framework that validates each LLM-proposed tool action before it causes external effects. Built on a unified semantic model of authorization, actions, runtime context, and security constraints, the ActGov-Policy component iteratively constructs a policy set from tool specifications, benign tasks, and observed failure traces, with each update verified through SMT-based counterexample checking. At runtime, ActGov-Runtime abstracts each tool call into finite policy records and permits it only if it remains within the task-scoped authorization boundary and satisfies all applicable policies. This per-action enforcement preserves authorization throughout long-horizon, dynamically branching workflows. We evaluate ActGov on the AgentDojo and AgentDyn benchmarks across multiple models and attack configurations. It shows that ActGov consistently reduces the success rate of indirect prompt-injection attacks while preserving task utility, significantly outperforming existing defenses. These results demonstrate that ActGov can enforce fine-grained authorization over dynamic agent executions without relying on the underlying LLM to correctly identify malicious instructions. Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI) Cite as: arXiv:2609.24446 [cs.CR] (or arXiv:2609.24446v1 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2609.24446 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-42] WPBench: A Comprehensive Benchmark for Wind Power Forecasting ICDE2027
链接: https://arxiv.org/abs/2609.24444
作者: Yuhan Zhu,Jilin Hu,Xinying Cai,Yingshan Li,Li Ma,Xiangfei Qiu Linsen Li,Kai Zhang,Yao Fu,Weihao Jiang,Bin Yang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted by ICDE 2027
Abstract:Accurate, reliable, and deployable wind power forecasting is critical for power system dispatch, renewable energy integration, and electricity market operations. Progress in this field hinges on the ability to empirically and comprehensively benchmark forecasting methods. Yet existing benchmarks fall short of supporting systematic evaluation in four key aspects: 1) limited coverage of wind power scenarios across turbine scale, variable composition, and spatial structure; 2) incomplete coverage of forecasting model families; 3) evaluation metrics misaligned with wind power requirements; and 4) limited structure-aware diagnostics beyond individual temporal patterns. To address these limitations, we propose WPBench, a comprehensive, fair, and extensible benchmark for wind power forecasting. WPBench integrates 26 public datasets organized by turbine scale and variable composition, spanning single-turbine, multi-turbine, univariate, and multivariate settings. Under unified processing, training, and evaluation protocols, it benchmarks 19 representative models covering traditional methods, deep temporal models, spatio-temporal models, and foundation models. Beyond point-wise errors, WPBench assesses forecast-curve fidelity and computational efficiency, and delivers structure-aware diagnostics across temporal, variable-dependency, and spatial-dependency perspectives. Together, these capabilities enable systematic model comparison across diverse wind scenarios and provide a reusable platform for future research.
[AI-43] FoldQuantVLA: Native Low-Bit Quantization of Vision-Language-Action Models via Consistent Folding
链接: https://arxiv.org/abs/2609.24433
作者: Hung T. Ho,Khanh D. Nguyen,Quang D. Nguyen,Thanh Q. Duong,Ngan Le,Meng Guo,Vien A. Ngo,An T. Le
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Systems and Control (eess.SY)
备注: 8 pages, 5 figures, 7 tables. Code: this https URL
Abstract:Low-bit vision-language-action inference must reduce observation-to-action latency while preserving robot behavior. We present FoldQuantVLA, a post-training quantization framework that carries a consistent activation representation through calibration, weight rounding, and native integer execution. It combines channel scaling and block Hadamard transforms with dynamic per-token quantization, without policy retraining. Custom TensorRT plugins execute projections in both the language backbone and iterative action expert with four-bit weights and activations (W4A4) on Ada GPUs and Jetson AGX Orin. Evaluation spans LIBERO, SimplerEnv, and two robot platforms. Across three GR00T checkpoints and \pi_0.5 , W4A4 achieves 1.20 to 1.33\times speedups over floating-point TensorRT on Orin and 1.25 to 1.52\times on desktop. Retaining language attention-output and feed-forward down projections at eight bits (W8A8) improves held-out action fidelity on all four checkpoints. Across four real-robot tasks, this configuration raises observed GR00T N1.7 success from 80.0% with uniform W4A4 to 92.5% over 80 trials per configuration, with a measured additional Orin latency of 1 ms.
[AI-44] ARM: Attention with Routed-Memory for Learnable Sparse Control ICML
链接: https://arxiv.org/abs/2609.24417
作者: Qiuhao Zeng,Jerry Huang,Peng Lu,Ruiyi Fang,Gezheng Xu,Zihao Jing,Yufei Cui,Charles Ling,Gang Niu,Boyu Wang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted to the Forty-third International Conference on Machine Learning (ICML) 2026. First two authors contributed equally
Abstract:Despite advances in long-context inference, large language models (LLMs) remain fundamentally limited by the key-value (KV) caching mechanisms that are necessary for stable computation. Techniques such as selective token eviction and pruning have vastly mitigated these issues, but often discard core information to manage the growing cache. In this paper, we propose Attention with Routed Memory (ARM) a novel KV caching structure that introduces a fully differentiable, fixed-size memory system organized as a hierarchical router. Via a Gumbel-Softmax, ARM learns to select memory slots and perform sigmoid-gated updates that softly combine new and stored information, avoiding hard eviction and reducing information loss. By further training a policy to dynamically select varying amounts of memory at inference, ARM adapts its accesses for both simple contexts and inputs that require deeper reasoning, enabling more scalable and effective retrieval on both short- and long-contexts. Experimental results on standard commonsense and long-context reasoning benchmarks demonstrate that ARM achieves superior performance and efficiency compared to fixed KV-caching approaches, while remaining efficient and scalable in terms of both memory and generation latency.
[AI-45] Artificial Structure Function Search: Preserving Artificial Functional Connectivity for Structured Pruning
链接: https://arxiv.org/abs/2609.24401
作者: Mindula Illeperuma,Rafael Pina,Charuka Herath,Sharmarke A. Gabayre,Varuna De Silva
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Structured pruning is a model compression technique that is used to reduce the computational cost of deploying deep neural networks on resource-constrained devices. Popular methods of pruning rely on opaque heuristics or weight-based criteria that give no indication as to the structural dependencies in the network. To address these limitations we present Artificial Structure Function Search (ASF-S): a novel structured pruning framework. ASF-S utilizes Principle Gradient Importance (PGI): a novel prune-candidate selection criteria that is inspired by structure-function relationships in the brain. By ensuring the pruned structure of the model respects topographical organization of the output layer, we define Artificial Functional Connectivity (AFC) for artificial neural networks. AFC provides evidence to demonstrate that accurate smaller networks can be found using careful prune candidate selection criteria. We present results for PGI as a selection criterion and for ASF-S as a pruning framework against recent benchmarks, demonstrating that our method yields model variants with 70% parameter reduction, that can recover baseline accuracy without re-training the pruned layers.
[AI-46] actile-JEPA: Topology-Aware Self-Supervised Representation Learning for Distributed Tactile Sensors
链接: https://arxiv.org/abs/2609.24385
作者: Elizaveta Kovtun,Matvey Konovalov,Andrey Sakhovskiy,Semen Budennyy
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Tactile sensing is an essential modality for robots performing contact-rich, dexterous manipulation, particularly under visual occlusion. While pre-trained image encoders are standard in robot learning pipelines, tactile encoders are still commonly trained from scratch from raw, noisy signals, which might limit their expressivity. Existing self-supervised learning (SSL) approaches focus predominantly on vision-based tactile sensors, leaving distributed electronic skins largely unaddressed. These sensors, however, have a distinctive property: their sensing elements are sparse and irregularly arranged over the surface they cover, which makes direct reuse of visual SSL methods suboptimal. We present Tactile-JEPA, an efficient self-supervised pre-training method that uses the spatial arrangement of tactile sensors to learn topology-aware representations. Specifically, it is trained to predict the embeddings of masked sensing elements from the unmasked remainder, using the sensor connectivity graph to guide spatial masking. Our analysis shows that effective tactile representations require capturing both local contact details and the global state of the tactile surface, which we achieve through dual-scale masking. Across three diverse datasets spanning magnetic and piezoresistive sensors, different robot embodiments, and single- and paired-sensor configurations, Tactile-JEPA reduces force estimation error by 6.3% and in-hand orientation error by 20.8% over the prior state-of-the-art, with consistent gains in other downstream applications, including policy learning. Overall, our results demonstrate that the benefit of tactile sensing depends critically on the quality of encoder pre-training, a problem which Tactile-JEPA addresses directly. Code is available at this https URL.
[AI-47] Information-Time Proximal Policy Optimization
链接: https://arxiv.org/abs/2609.24380
作者: Yongcheng Zeng,Xinyu Cui,Yan Song,Guoqing Liu,Hongsheng Xin,Kaike Zhang,Cheng Deng,Kun Zhan,Jian Ying,Jian Zhao,Haifeng Zhang,Jun Wang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:RLVR has substantially improved the reasoning capabilities of LLMs. However, existing methods typically parameterize temporal progression in the Markov Decision Process by token-by-token generation, despite the highly non-uniform information flow along autoregressive trajectories. In this paper, we propose InfoPPO, which reparameterizes temporal progression using information density rather than raw token count. This reparameterization induces a common state-dependent structure for both temporal credit propagation and policy updates. InfoPPO restores the effectiveness of non-trivial discounting in long-horizon reasoning, retaining effective-horizon contraction while avoiding excessive attenuation of terminal supervision over long token sequences. Moreover, the information-time policy-improvement analysis naturally leads to a state-dependent update constraint, which we implement through adaptive clipping. By adapting the clipping threshold at each token position to the information density of its corresponding state, this mechanism enables more targeted policy updates while preserving proximal control. Theoretically, we extend performance-difference and policy-improvement analyses to the information-time MDP, deriving a policy-improvement lower bound when policy changes are regulated by information density. We further connect the general information-time analysis to practical LLM policy optimization by relating state-wise information density to local policy movement, while also providing theoretical grounding for the adaptive update mechanism. Experiments on Qwen3 models demonstrate consistent gains over competitive baselines across five challenging competition-style mathematical reasoning benchmarks. InfoPPO also maintains stable accuracy and response length across non-trivial discount settings under which token-time PPO deteriorates.
[AI-48] VLM-in-Sandbox: Visual Workspaces for Agent ic Visual Reasoning
链接: https://arxiv.org/abs/2609.24362
作者: Hexiong Yang,Mingrui Chen,Jie Cao,Ran He
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Sandboxed computer environments support multi-step reasoning with tools, executable programs, and persistent files, yet their extension from language models to vision-language models (VLMs) introduces a distinct state-management problem. Visual reasoning produces intermediate image-valued evidence—crops, masks, overlays, zoomed regions, and analytic renderings—that must remain addressable without accumulating unboundedly in multimodal context. We introduce VLM-in-Sandbox, a training-free framework for agentic multimodal reasoning in controlled computer environments. Its Visual Workspace registers generated artifacts in an image ledger, maintains a bounded active visual context, and lets the model explicitly promote selected evidence for subsequent inspection. This separates visual evidence generation, performed by sandbox tools, from visual evidence management. Across seven benchmarks and four base VLMs, VLM-in-Sandbox achieves the highest sample-weighted average accuracy among Vanilla VLM, Append-only Sandbox, and the proposed method. A compiler-matched 2\times2 study on 1,260 examples further separates model-directed visibility from bounded retention: VLM-in-Sandbox reaches 66.27% accuracy with 18.6% fewer total tokens than the automatic, retain-all control. Over all 6,350 submitted GPT-4.1-mini examples, it produces 302 rescues and 142 regressions relative to Original Append-only. A local vLLM study with prefix caching confirms that the smaller request workload also reduces uncached tokens, time to first token, and end-to-end latency. These results identify explicit visual evidence state as a central abstraction for sandboxed VLM agents.
[AI-49] Few-Shot Demonstrations Elicit the Use of In-Context World Representations in LLM s
链接: https://arxiv.org/abs/2609.24352
作者: Kohsei Matsutani,Gouki Minegishi,Core Francisco Park,Takeshi Kojima,Yusuke Iwasawa,Yutaka Matsuo
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models (LLMs), when acting as agents, are expected to take observed data in context, infer the latent state space underlying the world, and leverage it for downstream prediction. However, prior work demonstrated that LLMs struggle to use representations learned in context on a graph tracking task, where the model needs to construct a representation of the graph governing data generation process and use it for subsequent predictions. In this paper, we show that extending this to few-shot settings, where each demonstration is generated from a different world with either the same or different graph topologies, enhances its prediction on 6 models from 4 model families. To understand this improvement, we linearly probe a low-dimensional world representation that encodes graph information in the hidden states. Notably, we find that few-shot demonstrations relocate the world representation and increase its predictive use. Specifically, for each model, these world representations shift in directions nearly orthogonal to their original subspace, and interventions on these representations selectively impair performance more than interventions on other subspaces. Consistent with this insight, we show that few-shot demonstrations with observations from different worlds improve performance on ARC-AGI-12, web agent tasks, and Othello. Our findings elucidate the role and internal mechanisms of few-shot demonstrations in in-context world modeling. More broadly, our work advances our understanding of how LLM agents learn from in-context observations and provides implications for their further improvement.
[AI-50] A Lean and Spec-Driven AI-Assisted Software Development Lifecycle for Applied AI Education: The AI-SDLC Approach
链接: https://arxiv.org/abs/2609.24348
作者: Andreas Martin,Sandro Schwander
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注: Accepted for publication in the Journal of the Upper Rhine Artificial Intelligence (URAI) Symposium 2026
Abstract:AI coding agents increasingly support software development beyond code completion, including planning, implementation, testing, and repository-level task execution. Their practical use, however, often remains only weakly connected to established software engineering practices. The aim of this work is to develop and evaluate a lightweight, spec-driven lifecycle for governed agentic software engineering. The lifecycle combines established software engineering practices with repository-local guidance through specifications, this http URL, and phase-specific agent skill files. The approach was developed in the context of the FHNW course AI-assisted Software Development and applied by students to business-oriented software use cases. Its educational and practical applicability is explored through a student survey combining closed rating items with open-ended questions. The contribution of this work is a process-oriented framework that enables AI coding agents to operate with bounded autonomy within an explicit, reviewable, and test-oriented software development lifecycle.
[AI-51] LADDER: Graph-Guided Diffusion Language Models for Efficient Multi-Hop Reasoning
链接: https://arxiv.org/abs/2609.24346
作者: Senlei Zhang,Linhao Luo,Qian-Wen Zhang,Siyu An,Junnan Dong,Shuhao Zhang,Xing Sun
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Graph Retrieval-Augmented Generation (GraphRAG) has remarkably enhanced large language models on complex reasoning by leveraging structured entity topologies. However, existing frameworks heavily rely on standard autoregressive language models where the nature of inherent sequential generation severely hinders overall inference efficiency. Inspired by Diffusion Language Models (DLMs) that offer massive parallelism via continuous refine-in-parallel decoding, we aim to accelerate GraphRAG in the discrete space. However, it remains non-trivial for two challenges. First, partially denoised drafts are highly dynamic and uncertain, making dynamic graph grounding non-trivial. Second, raw denoising states are inherently noisy and unstable, making synchronous graph retrieval and multi-hop aggregation computationally prohibitive. To this end, we present LADDER, a novel framework that bridges diffusion language modeling with GraphRAG through graph-guided parallel decoding. Specifically, (i) we propose an event-driven self-clocking retrieval, inspired by our key insight that 88% of target entities emerge early in the partially denoised state, leading final commitment by an average of 5.7-9.6 steps. This mechanism dynamically triggers graph retrieval only when the set of graph-linkable entities expands, yielding an asynchronous self-clocking policy that bypasses learned gates or heuristic thresholds. (ii) An incomplete-query graph propagation module is designed to process the newly emerging entity queries using a specialized graph foundation model, continuously aggregating multi-hop evidence to sharpen parallel predictions and accelerate overall decoding convergence. Extensive experiments on three challenging multi-hop QA benchmarks show that LADDER raises average exact match from 39.6% to 45.2% while achieving a 4.1x latency reduction.
[AI-52] Brain-Token Learning: Microstate-Based Tokenization and Multi-Scale Interaction for Long-Horizon EEG Sequence Modeling
链接: https://arxiv.org/abs/2609.24324
作者: Weishan Ye,Yue Pan,Li Zhang,Gan Huang,Zhen Liang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Electroencephalography (EEG) provides a non-invasive window into dynamic brain activity, yet modeling long-horizon EEG sequences remains challenging due to their high temporal complexity, substantial variability across subjects, and the lack of biologically meaningful sequence representations. Existing tokenization strategies, such as fixed-window and patch-based representations, discretize EEG signals according to artificial temporal boundaries, which may disrupt intrinsic brain-state dynamics. In this work, we propose Brain-Token Learning, a neuroscience-inspired framework that introduces Brain Tokenization for long-horizon EEG sequence modeling. Instead of partitioning EEG signals into predefined temporal segments, Brain Tokenization represents EEG as sequences of recurrent microstate-derived brain tokens, where each token corresponds to a quasi-stable large-scale brain state with variable temporal duration. Based on these biologically grounded tokens, we further develop a multi-scale token interaction module consisting of Latent State Aggregation and State Transition Modeling to jointly capture global brain-state context and local microstate transitions. We evaluate Brain-Token on five heterogeneous EEG datasets, including the newly collected long-horizon NeuroLong dataset and four affective or clinical EEG datasets (SEED, DEAP, MDD, and NSSI). Extensive experiments demonstrate that Brain-Token consistently outperforms conventional CNN/LSTM architectures, Transformer-based models, and domain adaptation methods across diverse EEG scenarios. Further analysis verifies the effectiveness of microstate-based tokenization and multi-scale interaction for learning robust and interpretable EEG representations. These results establish Brain-Token as a biologically grounded tokenization paradigm for long-horizon EEG sequence modeling.
[AI-53] he Undetected Damage of Quantization on Retrieval and How to Fix It
链接: https://arxiv.org/abs/2609.24322
作者: Luca Zhou,Alessandro Zirilli,Daniele Solombrino,Roberto Dessì,Emanuele Rodolà
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 5 figures, 4 tables in the main paper
Abstract:We show that a quantized model that keeps its classification accuracy still changes 14 to 46% of its top-1 retrieval results, and that aggregate ranking metrics reveal only part of this damage. We tie this failure to the gap between the two highest scores and use that gap to decide when a quantized answer can be trusted and where additional precision should be spent. We show that the top-1 result is guaranteed to survive quantization only when this gap exceeds twice the largest rounding error. In classification, scores are the logits, and the loss function pushes the correct class away from other classes, encouraging this gap. In retrieval, scores are query-document scores, and nothing separates the top-1 item from the second. This gap can be measured without labels. Before deployment, it predicts which models will break under quantization, and at deployment time it tells, per input, whether the quantized answer still matches the full-precision answer. Most classification inputs have a gap wide enough to trust the quantized answer, but few retrieval queries do. That gap motivates a different fix in each task. In retrieval, spending extra bit-width on the layers whose quantization moves the gap most recovers up to three-quarters of an extra bit’s benefit for half its cost. In classification, routing the few low-gap inputs to full precision recovers most of the lost accuracy at a fraction of the cost.
[AI-54] KV-COBRA: KV Cache Compression via Co-Optimized Bit-Rank Allocation
链接: https://arxiv.org/abs/2609.24298
作者: Sihyeon Ha,Jaeho Lee,Yo-Seb Jeon
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:What limits KV-cache compression at extreme bit-rates? We argue that it is not the choice of compression scheme, but how its budget is allocated across attention heads. Existing methods apply rank and bit-width uniformly, ignoring that each head has a different optimal mix of rank truncation and quantization. We show that co-optimizing rank and bit-width per head, using only standard low-rank projection and scalar quantization, dominates uniform allocation, with the largest gains at low bit-rates. Our method, KV-COBRA (Co-Optimized Bit-Rank Allocation), formalizes this as a resource-allocation problem: it balances rank-truncation loss against quantization loss within each head, then redistributes budget across heads to minimize total distortion. A fused Hadamard rotation equalizes per-channel variance, and reordering the SVD basis by attention-KL importance makes the solver query-aware. The same allocator extends to joint K+V compression. On perplexity, zero-shot, and long-context benchmarks from 0.5 to 4 bits per dimension (bpd), KV-COBRA shows the smallest accuracy degradation among evaluated methods at low bpd, with no per-token overhead.
[AI-55] When and How Should an Agent Clarify? CIGAsk: Teaching LLM s to Clarify via Counterfactual Information Gain EMNLP2026
链接: https://arxiv.org/abs/2609.24290
作者: Yunxiang Li,Xixin Wu,Helen Meng
类目: Artificial Intelligence (cs.AI)
备注: Accepted to EMNLP 2026 (Findings)
Abstract:Instruction-tuned LLMs faced with underspecified queries often commit to a single interpretation rather than ask for clarification, producing confidently wrong answers. In our experiments, prompting alone is insufficient: models either ask for clarification on every query or ask vague questions that fail to recover the missing information. Addressing this failure requires learning two coupled skills: when to ask rather than answer and how to ask a question that recovers the disambiguating information. Existing recipes either address only one of these skills or require a separately trained critic. We propose CIGAsk, an RL recipe that teaches both skills through two complementary reward signals within a multi-turn GRPO loop. Counterfactual Information Gain (CIG) compares the gold-answer log-likelihood under a frozen reference model with and without the user response, providing per-turn credit that guides how to ask. The Asymmetric Ambiguity Bonus assigns a signed reward at the terminal token based on the gold ambiguity label, guiding when to ask. Across three clarification benchmarks spanning table, passage, and open-domain QA, CIGAsk-7B outperforms the strongest external baseline despite using a smaller backbone. It also transfers across datasets without per-dataset tuning while preserving single-turn QA performance on out-of-distribution benchmarks.
[AI-56] SE: A Two-Track Online Self-Evolution Framework
链接: https://arxiv.org/abs/2609.24289
作者: Ruimin Pei,Yongkang Wu,Shangyi Zheng,Yaqing Zhang,Deyang Li,Jianjun Tao,Xinyu Zhang,Xiang Zhang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 20 pages, 2 figures, 18 tables
Abstract:As Large Language Model (LLM) agents are applied in continuously interactive environments, driving the evolution of their own capabilities becomes a core problem for achieving long-term autonomy. Currently, environmental knowledge is typically treated as an external fixed input rather than as part of the agent’s ongoing evolution. Reinforcement learning methods usually optimize policies through environmental interaction but tend to adapt only to fixed task distributions or single environments. This paper proposes TTSE (Two-Track Self-Evolution), a dual-track online self-evolution framework that separates evolving knowledge into FACT (environmental facts, whose reliability is continuously verified through interaction evidence) and TIP (task-conditioned implementation procedures). From a decision-theoretic perspective, we decompose the agent’s excess risk into environment-representation regret and conditional-execution regret, characterize the conditions under which environment-conditioned policies strictly outperform condition-agnostic policies, and bound the downstream risk in terms of FACT identification error and cross-condition mismatch cost. In practice, TTSE’s ablation experiments on GDPevo validate the advantage of dual-track evolution. On the classic agent task benchmarks ALFWorld and ScienceWorld, TTSE further demonstrates superior task adaptation. Moreover, TTSE is broadly compatible with existing skill self-evolution methods; combined with the Bayesian-Agent algorithm, a single-track ablation validates the dual-track advantage, substantially improving the aggregate score across the five major domains of SOPBench over three independent repetitions. Finally, on the real end-to-end task benchmark PinchBench, TTSE is integrated into a general agent framework via retrieval-based injection and stably outperforms the baseline across three independent runs.
[AI-57] mporal Generalization and Explanation Stability of Control Flow Graph Neural Networks for Malware Detection
链接: https://arxiv.org/abs/2609.24280
作者: Md. Asif Sajeed,Md. Nazrul Islam Mondal,Md Ashraful Hossen Akash
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 47 pages, 9 figures, 17 tables. Code available at this https URL
Abstract:Malware detection is a critical task in cybersecurity, and graph neural networks over control flow graphs have shown promising results for it. However, detectors are usually evaluated on a random split of a corpus collected over a single period, which cannot show how well a model generalizes to later samples. This study addresses that limitation with a strict temporal split: every model is trained on one period and scored once on a later one. Two corpora of control flow graphs, each node carrying 37 features, were extracted statically from 1,989 Windows portable executables: 459 graphs from 2024-2025 for training and 223 from 2026 for evaluation. Twelve variants and a flat-feature control were trained on the earlier corpus. The choice of message-passing operator changes robustness to the shift significantly, and every pairwise gap that survives correction separates an aggregating architecture from one built around a learned attentional readout. The ranking also reverses: the flat control, which sees node features but no topology, is the best in-distribution model and among the worst across the boundary, so a conventional benchmark would have rejected message passing. Neither recalibration nor ensembling substitutes for the operator choice. Attributions do not shift, but explanation validity is architecture-specific, and the most accurate operator on the later corpus is the hardest to explain. An architecture derived from the finding matches the best searched operator without search. The shift affects both malware and benign classes alike, so these are results about robustness to distribution shift, not malware evolution.
[AI-58] How Many Pixels Is a Digit Worth? Place-Aware Coordinate Entropy for GUI Agent Confidence Estimation EMNLP2026
链接: https://arxiv.org/abs/2609.24277
作者: Yunxiang Li,Xixin Wu,Helen Meng
类目: Artificial Intelligence (cs.AI)
备注: Accepted to EMNLP 2026 (Main Conference)
Abstract:GUI agents predict click coordinates as digit-token sequences, but standard text-LLM confidence estimation methods rank correct clicks from wrong ones only weakly. GUI-specific alternatives use K samples or new supervision, but still leave room for improvement. We trace part of this to place-value asymmetry: bounding-box correctness often makes higher-place digits more important than lower-place digits, so uniform aggregation weakens the signal that determines correctness. The fix is to weight each digit’s Shannon entropy by its place value. We call this Place-Aware Coordinate Entropy (PACE). Across fixed-scale agents on ScreenSpot-Pro and ScreenSpot-v2, PACE wins both AUROC and selective accuracy on all primary comparisons in a single forward pass, matching or outperforming K-sample baselines at a fraction of the cost. PACE provides a per-click confidence estimate that turns coordinate-token internals into a practical confidence signal for GUI agent deployment.
[AI-59] vla.simd: Efficient CPU Inference for Language-Conditioned Manipulation
链接: https://arxiv.org/abs/2609.24274
作者: Khanh D. Nguyen,Hoang M. Truong,An T. Le
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI); Systems and Control (eess.SY)
备注: 8 pages, 7 tables, 5 figures. Project page: this https URL
Abstract:Deploying language-conditioned manipulation without a dedicated GPU requires efficient inference and action chunks that cover the delay between policy queries. We present this http URL, a CPU inference engine that combines shared SIMD micro-kernels, reusable computation, and target-specific optimization. We relate query latency and execution horizon to action availability under lagged and time-aligned execution, distinguishing action supply from feedback frequency. Across six policies and four CPUs, this http URL achieves approximately 1.4\times median speedup over compiled PyTorch references while preserving fp32 numerical fidelity. We also introduce IMPACT, an ACT-based policy with cached text representations and language-modulated visual features. IMPACT is the only language-conditioned policy in our evaluated set that supplies at least 30 actions/s on the Raspberry Pi 5: after a 90 s thermal soak, it supplies 33.5 actions/s in fp32 and 81.2 with int8. Separate GPU evaluations yield 76.4% mean success across four LIBERO suites without robot pretraining; instruction-shuffling tests demonstrate selection among familiar goals. Trials with IMPACT on an SO-101 arm and SmolVLA on a UR10e with a Robotiq gripper demonstrate CPU deployment on two robot embodiments.
[AI-60] MemCalib: Benchmarking and Optimizing Memory Use in LLM Agents
链接: https://arxiv.org/abs/2609.24259
作者: Ruike Cao,Fanyu Zhao,Fugen Yao,Liang Dong,Jian Xu,Guanjun Jiang,Yifei Zhao,Han Zhang,Li Xiao
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:The effectiveness of agent memory ultimately depends on whether the underlying LLM gives each memory in context an appropriate degree of influence over its response. Yet this capability has remained largely overlooked. To assess this capability, we introduce MemCalib, a benchmark grounded in realistic memory-system scenarios for evaluating memory use and advancing optimization algorithms. Results on the MemCalib test set reveal that frontier open- and closed-source models struggle to use memory appropriately. They frequently over-use or under-use memory rather than matching each proposition’s actual use to its target level, leading to biased, low-quality responses. Experiments with common post-training algorithms, including group relative policy optimization and on-policy self-distillation, further reveal a clear directional skew: trained models improve in one direction while deteriorating in the other. We therefore propose MemCalib-RL, an ordered bidirectional counterfactual credit-assignment algorithm that separates over- and under-use signals and localizes their credit to response tokens through exact atom ablation. Results across model families and scales (Qwen3-8B, Ministral-3-8B-Instruct, and Qwen3.5-35B-A3B) show that MemCalib-RL achieves the best overall performance while better balancing over-use and under-use, with gains generalizing beyond MemCalib in external benchmark evaluation. Further experiments support its design choices and robustness and provide insight into its training dynamics.
[AI-61] aming CoT Obfuscation in VLMs: From Mechanistic Evidence to Activation Enforcement
链接: https://arxiv.org/abs/2609.24243
作者: Xutao Mao,Jianing Zhu,Jinman Zhao,Tongliang Liu,Xiaowen Chu,Cong Wang,Bo Han
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Reinforcement learning (RL) improves reasoning in vision-language models (VLMs) but can induce chain-of-thought (CoT) obfuscation: an operational, non-intentional outcome where task reward or accuracy rises while traces become less grounded and monitorable. Prior work largely documents this decay behaviorally, leaving its representation-level correlates and actionable controls unclear. We find that template- and ground-associated activations become less separable during RL; matched interventions support the contribution of selected features to monitorability degradation. Guided by this evidence, we propose Targeted Anti-obfuscation with Mechanistic Enforcement (TAME), which uses Sparse Autoencoders (SAEs) to combine behavioral feedback with targeted suppression of template-associated activations during RL. Its asymmetric constraint penalizes template activations only above their pre-RL baseline, anchoring the localized features while behavioral feedback promotes grounded refinements. Across VIRL-39k, SPA-VL, and two model families, TAME improves CoT monitorability by up to 30.9 and 16.7 percentage points over Group Relative Policy Optimization (GRPO), respectively. Blinded human evaluation finds higher human monitorability on both datasets, and two held-out monitor families reproduce the monitorability gains. Task accuracy changes are small and mixed, and general-capability benchmarks show task-specific trade-offs. These results provide a path from behavioral monitoring to representation-level oversight for more auditable RL-trained multimodal systems.
[AI-62] Hessian Rank Constraint for Learning Structure of Nonlinear Latent Variable Models
链接: https://arxiv.org/abs/2609.24241
作者: Zijian Li,Ruichu Cai,Feng Xie,Xinshuai Dong,Haoyue Dai,Yuewen Sun,Yujia Zheng,Guangyi Chen,Yingyao Hu,Kun Zhang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Uncovering latent variables and their causal relations from observed data is a fundamental yet challenging problem. Existing methods often rely on restrictive assumptions, such as linear relations or invertible mixing functions. To better address this problem under general nonlinear mixing procedures, we propose a condition called the cross-Hessian Rank Constraint (HRC), which serves as a primitive rank-based tool for nonlinear latent causal discovery. In particular, we show that a rank-based property arises from the cross-Hessian of the observed-data log-density in the nonlinear case, revealing information about the latent variables, and reduces to the Tetrad constraints in the linear Gaussian case. More specifically, when two groups of observed variables are d-separated by a set of lower-dimensional latent variables, the rank of this cross-Hessian is equal to the dimension of the latent variables, under a mild affine derivative assumption on the conditional log-density derivatives. This assumption can be naturally satisfied when the noise level is low or the relevant nonlinearity is moderate. As a downstream application, we instantiate HRC in the pure one-factor measurement setting for locating latent variables and recovering their causal structure up to Markov equivalence. Experimental results on synthetic and real-world datasets support the theoretical claims.
[AI-63] Recovering Lost Details: Multi-Scale Frequency Compensation for Long-Term Time Series Forecasting CIKM2026
链接: https://arxiv.org/abs/2609.24229
作者: Runmin Zou,Siyi Xie,Yaohui Huang,Yun Wang
类目: Artificial Intelligence (cs.AI)
备注: 11 pages. Accepted at the 35th ACM International Conference on Information and Knowledge Management (CIKM 2026)
Abstract:Long-term time series forecasting has made significant progress by leveraging multi-scale information to capture hierarchical temporal patterns and model long-range dependencies. However, temporal downsampling in existing multi-scale methods inevitably smooths detailed temporal fluctuations, and this information loss is further aggravated by their emphasis on dominant trends across scales, resulting in insufficiently expressive representations. To address this, we propose a Multi-Scale Wavelet Mixing (MWMixer) model, which incorporates a Bidirectional Frequency-Bands Mixing strategy to recover lost temporal details across scales, enabling complementary cross-scale information interactions. Then, a Dynamic Scale-Adaptive Fusion module learns time-varying weights for each scale to fuse multi-scale forecasts into the final prediction, enhancing the flexibility of multi-scale aggregation. In addition, a cross-scale consistency loss aligns each coarse-scale prediction with the interval-averaged fine-scale outputs, while a multi-scale supervision loss enforces prediction accuracy at each scale, promoting consistent learning across scales. Extensive experiments on seven real-world datasets demonstrate that MWMixer achieves competitive performance in long-term forecasting.
[AI-64] Opinion Leader Dynamics: How Sparse Attention Shapes Token Clustering
链接: https://arxiv.org/abs/2609.24202
作者: Jingkun Liu,Yue Song
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Code is available at this https URL
Abstract:Sparse attention reduces the quadratic cost of global self-attention while retaining strong empirical performance, but how its restricted interactions shape the evolution of token representations remains theoretically underexplored. Modeling tokens as particles on the unit sphere, we introduce opinion leader dynamics, a framework that identifies two mechanisms through which token groups converge internally while maintaining distinct limiting directions. In the explicit model, fixed representatives induce a potential that attracts tokens toward distinct local maxima. In the implicit model, disconnected interaction groups evolve toward separate consensus directions. We formulate both models as reverse Wasserstein gradient flows and establish exponential convergence under suitable conditions. We further connect these theoretical predictions to token evolution in frontier sparse-attention LLMs that motivate our framework. Across four benchmarks, Kimi-K3, MiniMax-M3, and DeepSeek-V4-Flash consistently exhibit clearer cluster separation and higher clustering scores than the dense-attention model GLM-4.7-Flash in projected token representations. These observations support the relevance of the predicted multiple-group structure to trained frontier LLMs, while finite-particle simulations illustrate the theoretical convergence behavior. Together, our results connect restricted token interactions to distinct group-level attractors, providing a dynamical account of how sparse attention can support alignment within groups while preserving separation between them.
[AI-65] LIMIT: Less Is More for Instruction Tuning in Text-to-SQL
链接: https://arxiv.org/abs/2609.24186
作者: Haoyuan Ma,Hengwei Liu,Linjuan Wu,Yongliang Shen,Weiming Lu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language models have achieved remarkable progress on Text-to-SQL through reasoning-enhanced fine-tuning, yet existing approaches predominantly rely on massive instruction corpora under the assumption that scale drives performance. We challenge this paradigm by investigating a fundamental question: what is the minimal data requirement for effective Text-to-SQL instruction tuning? We propose LIMIT(Less Is More for Instruction Tuning in Text-to-SQL), a data-centric framework that demonstrates strong database reasoning can emerge from an extremely compact training set when examples are strategically selected. LIMIT operates through four stages: difficulty-aware filtering that identifies samples within the model’s learning frontier, chain-of-thought synthesis with consistency-based selection, multi-dimensional quality scoring via LLM-as-judge, and genetic algorithm optimization that jointly maximizes schema coverage and sample quality. On the BIRD and Spider benchmark, LIMIT selects only 796 and 863 samples while achieving 100% table coverage, enabling Qwen3-8B to reach 69.1% and 88.9% execution this http URL result surpasses methods trained on 20 times more data and establishes a new state-of-the-art among open-source approaches. Our findings suggest that careful data curation, rather than scale, is the key to efficient Text-to-SQL learning.
[AI-66] CREDO: Variance-Guided Rubric Evolution for Replay-Corrected Credit Assignment
链接: https://arxiv.org/abs/2609.24174
作者: Xuchun Hu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Long-horizon language agents receive sparse terminal feedback, while intermediate rubrics provide structured but potentially misspecified assessments of progress. In resettable training environments, counterfactual continuation rollouts can measure local credit, but exhaustive replay is costly. We propose Credo, a framework that couples evolving semantic rubrics with selective, execution-based credit correction. A frozen judge maps visible transitions to rubric features, and a credit head predicts the change in expected terminal reward associated with the realized transition. Independently sampled two-sided replays correct prediction residuals using their recorded inclusion probabilities. We derive conditional unbiasedness and a variance decomposition that connects two design choices: which rubric features to retain, and where to allocate a fixed expected replay budget. The resulting criterion weights prediction errors by policy-score sensitivity and missing replay coverage; its allocation rule additionally accounts for continuation cost. We also describe a practical mixture with terminal leave-one-out advantages and distinguish its clipped, token-normalized PPO implementation from the ideal policy-gradient estimator. This preliminary report provides the method, proofs, an exact finite-model audit, and a controlled evaluation protocol. It makes no claim of empirical superiority on language-agent benchmarks.
[AI-67] APEXA: Execution-Integrity Enforcement for Multi-Agent LLM Automation of Synchrotron Data Reduction
链接: https://arxiv.org/abs/2609.24165
作者: Pawan K. Tripathi,Hemant Sharma,Andrew Chuang,Mathew J. Cherukara
类目: Artificial Intelligence (cs.AI)
备注: 5 pages, 4 figures. Accepted at the 4th TPC Workshop @ SC’26 (Building Open AI Infrastructure, Models, and Agentic Systems for Science). Code and benchmark: this https URL ; data: this https URL
Abstract:Synchrotron data reduction, detector calibration followed by azimuthal integration of terabyte-scale diffraction series, is a multi-step, expert-bound bottleneck that increasingly limits the science rate of user facilities. LLM agents promise to collapse it, but driving a real pipeline with a stochastic model creates a failure mode chat benchmarks cannot see: an agent can report a calibration that was never computed. Correctness here is a property of what executed, not of the transcript. We present APEXA, a deployed multi-agent framework (61 tools over heterogeneous compute, run as a single reasoning loop) automating calibration and integration from natural language at a major light source. We make three contributions. First, execution-integrity enforcement: a deterministic tool-layer guard that refuses to surface any result not backed by an executed tool call, with a parser tolerant of cross-model tool-call format drift: in deployment, a frontier model fabricated a complete calibration-comparison report for commands that never ran, which the guard converts to an explicit non-result; the same code gates an optional motor-control surface at 0/200 adversarial violations against a simulated IOC, versus 15/200 for an equivalent safety prompt. Second, we release APEXA-Bench, an evaluation harness of 58 facility tasks (50 base plus an 8-task cross-detector slice) organized by a four-class physical-consequence taxonomy, the first benchmark axis we know of separating a wasted compute cycle from a damaged instrument; its cross-detector grading against NIST-traceable lattice constants surfaced two latent pipeline bugs. Large-scale agent scoring is left to a full-length study. Third, we validate APEXA on real beamline data: from one natural-language prompt it recovers detector geometry and integrates a full attenuation/exposure sweep. We release the framework, harness and traces.
[AI-68] MCP-GRANITE Benchmark: GRANularity Interface TEsting for MCP-Based LLM Agents
链接: https://arxiv.org/abs/2609.24161
作者: Demetris Paschalides,Moysis Symeonides,George Pallis,Marios D. Dikaiakos
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: Author copy of paper published at 34th International Symposium on the Modeling, Analysis, and Simulation of Computer and Telecommunication System (MASCOTS2026)
Abstract:As LLM agents increasingly interact with external tools through standardized protocols such as MCP, tool-interface design becomes a critical yet underexplored factor. How fun\psitionality is decomposed into tools affects whether an agent can select the right tool and construct valid arguments. This choice is especially consequential at the edge, where resource constraints limit which models can run locally and scaling up is often not an option. We present MCP-GRANITE, an open-source extensible benchmark framework that treats tool-interface granularity as a controlled variable for MCP-based agents, evaluated under edge and IoT scenarios. It comprises 81 multi-step scenarios across 9 domains, instantiated at 4 granularity levels from fine-grained primitive tools to a single tool. We evaluate 9 locally deployed models (268M-20.9B parameters) across 8,748 trials using task completion, tool selection F1, argument accuracy, latency, and resource-usage metrics. Results show that a 4-tool interface offers the best trade-off, improving task completion by 16.4% over fine-grained primitives and 33.6% over a single monolithic tool, while nearly doubling argument accuracy. Model size is only weakly correlated with task completion and strongly with latency, while its association with argument accuracy is less robust, and a 3.2B model at the optimal granularity outperforms a 20.9B model at a mismatched one. These findings identify tool-interface granularity as a key design parameter for MCP-based agents.
[AI-69] Self-Healing Harness for Runtime Oversight of Agent Self-Modification
链接: https://arxiv.org/abs/2609.24130
作者: Sina Tayebati,Divake Kumar,Nastaran Darabi,Ranganath Krishnan,Amit Ranjan Trivedi
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:LLM agents can change their own future behavior, raising a basic control question of which self-generated changes should be allowed to persist. We formulate this as admission control for self-modification. The agent may propose changes to its operating instructions, while an external runtime gate controls persistence. We implement this principle as a model-agnostic self-healing harness that runs a Detect, Notice, Heal, Validate loop around an otherwise unmodified agent. The agent authors candidate behavioral rules in an external workspace, where they receive provisional execution authority during evaluation and acquire persistent cross-episode authority only after measured improvement on the triggering failure without regression beyond a fixed margin on protected cases. Replay provides matched evidence when available, forward trials provide a weaker fallback, and a corpus-level guard re-tests the accumulated active rule set. Across 16 matched Baseline and Harness runs spanning AppWorld, Terminal-Bench, and \tau^2 -Bench, the gate rejected 383 replay-decided proposals. Of these, 211 (55%) improved their triggering failure while degrading a case that previously worked. This shows that locally beneficial self-modifications can introduce collateral regressions often enough to materially affect gate decisions, providing direct empirical motivation for external admission control. Task-completion score is higher under the Harness in all 16 pairs, with two paired bootstrap intervals excluding zero, while repeated-trial reliability is higher in 12 pairs, tied in 4, and lower in none. Because adaptation modifies the policy-inducing context while leaving model weights fixed, admitted changes remain inspectable, reversible, and compatible with closed-weight models.
[AI-70] ActiveArena: Benchmarking and Understanding Active Perception in Robotic Manipulation
链接: https://arxiv.org/abs/2609.24124
作者: Yibo Li,Enshen Zhou,Rui Chen,Yanjun Ding,Mengzhen Liu,Yi Han,Jiabo Zhan,Lipeng Wang,Shanghang Zhang,Lu Sheng
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: 43 pages. Project page: this https URL
Abstract:Active perception and manipulation are crucial for robots to interact with complex scenes. Existing benchmarks struggle to evaluate how robots effectively acquire and maintain information in memory in an active manner. To this end, we introduce ActiveArena-Sim, an active-perception simulator with controllable viewpoints and large-scale workspaces as the foundation. Built on this, we propose ActiveArena-Bench, which comprises 35 tasks across 5 fine-grained categories, covering visual exploration and interactive information acquisition. Each task is difficult to solve from passive observations alone, requiring multi-round evidence acquisition and memory-based reasoning. The benchmark provides rich memory annotations, standardized training data, and ID/OOD protocols featuring disjoint scenes, unseen distractor configurations, and novel backgrounds. Moreover, we present ActiveArena-VLA, a modular suite of 13 vision-language-action configurations for controlled studies of memory writing, memory capacity, proprioceptive state, subtask supervision, and high-level planning in active perception. Benchmark results reveal a substantial ID-OOD gap: uniform memory sampling, increased memory capacity under reliable write policies, proprioceptive inputs, and subtask supervision improve OOD generalization, while planner-guided memory management and decision-making achieve performance close to the best-performing configuration using only sparse memory. ActiveArena thus provides a unified testbed to develop and diagnose models for active perception and manipulation. Comments: 43 pages. Project page: this https URL Subjects: Robotics (cs.RO); Artificial Intelligence (cs.AI) Cite as: arXiv:2609.24124 [cs.RO] (or arXiv:2609.24124v1 [cs.RO] for this version) https://doi.org/10.48550/arXiv.2609.24124 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-71] EDGEGEN: Improving Tool-Calling Agents Beyond Happy Paths with Synthetic Edge Case Generation
链接: https://arxiv.org/abs/2609.24115
作者: Harshavardhan Abichandani,Penny Chong,Jiyuan Shen,Gunraj Singh,Ashutosh Hathidara,Marcus Duigan Xing Yu,Jane Lo,Atin Ghosh,Yipeng Li,Daniel Dahlmeier
类目: Artificial Intelligence (cs.AI)
备注: NA
Abstract:Tool-calling LLM agents are increasingly deployed in enterprise applications. However, effective evaluation and optimization require high-quality, diverse task datasets that are often difficult to obtain due to privacy and other constraints. Existing synthetic task generation methods often produce generic tasks that ignore an agent’s underlying state or database and fail to reflect real-world usage diversity. We propose EdgeGen, a synthetic task generation framework that extracts compliance rules from an agent’s specification and uses them to generate database-grounded edge-case tasks designed to violate these rules. When combined with existing synthetic data generation techniques, EdgeGen enables agent improvement through finetuning and harness optimization. The resulting pipeline forms a fully automated closed-loop system that requires no human annotation. Finetuning on data generated by EdgeGen yields a consistent mean progress improvement of 2 percent to 42 percent on tau2bench airline domain, while other baseline methods show degradation for some models. On the other hand, for harness optimization, our method shows a mean progress improvement of 10 percent and 30 percent over the human-curated and base harnesses, respectively, for the Gemma-4-e4b model.
[AI-72] Reinforcement Learning under State and Outcome Uncertainty: A Foundational Distributional Perspective
链接: https://arxiv.org/abs/2609.24103
作者: Larry Preuett,Qiuyi Zhang,Muhammad Aurangzeb Ahmad
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted to the RLC 2025 Finding the Frame Workshop
Abstract:In many real-world planning tasks, agents must tackle uncertainty about the environment’s state and variability in the outcomes of any chosen policy. We address both forms of uncertainty as a first step toward safer algorithms in partially observable settings. Specifically, we extend Distributional Reinforcement Learning (DistRL)-which models the entire return distribution for fully observable domains-to Partially Observable Markov Decision Processes (POMDPs), allowing an agent to learn the distribution of returns for each conditional plan. Concretely, we introduce new distributional Bellman operators for partial observability and prove their convergence under the supremum p-Wasserstein metric. We also propose a finite representation of these return distributions via psi-vectors, generalizing the classical alpha-vectors in POMDP solvers. Building on this, we develop Distributional Point-Based Value Iteration (DPBVI), which integrates psi-vectors into a standard point-based backup procedure-bridging DistRL and POMDP planning. By tracking return distributions, DPBVI lays the foundation for future risk-sensitive control in domains where rare, high-impact events must be carefully managed. We provide source code to foster further research in robust decision-making under partial observability.
[AI-73] DocMIDE: Learning Multi-Hop Implicit Derivation in Visually Rich Documents
链接: https://arxiv.org/abs/2609.24092
作者: Jeremy Cerwin Wang,Wai Kit Wong,Jeff Kai Tai Tang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Real-world document processing systems rely on rigid, predefined schemas, yet critical target fields often lack direct visual counterparts on the page. Extracting these implicit values requires multi-hop derivation, such as aggregating sub-categories or reasoning over visual marks. While existing methods handle explicit text spans or simple implicit queries, they fail at multi-hop visual reasoning even after standard fine-tuning: models retrieve incorrect visual evidence, or retrieve it correctly and then skip the intermediate steps of the derivation. To address this, we introduce DocMIDE, a fine-tuning framework that trains compact vision-language models to retrieve visual evidence explicitly before deriving an answer. DocMIDE constrains generation to a plan-retrieve-derive structure and optimizes it with Group Relative Policy Optimization under a four-component, rule-based reward that scores output format, the retrieved evidence block, every intermediate derivation step, and the final value against a verified reference trace. On a 4,151-pair implicit extraction benchmark, DocMIDE raises accuracy from 70.8% to 95.9% on Qwen3.5-4B from only a small set of annotated examples, and transfers to a second backbone architecture. Supervised demonstrations alone do not close this gap at any budget we tested; rewarding the intermediate steps is what does.
[AI-74] Incremental Consistency Execution for Autonomous Intelligent Systems
链接: https://arxiv.org/abs/2609.24090
作者: Cheng Li,Jiexiong Liu,Yixuan Chen,Ziheng Huang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Long-horizon autonomous intelligent systems rely on heterogeneous components such as large language models, databases, external APIs, and rule engines, while their external states continuously change during execution. Re-executing the entire workflow after every change introduces substantial redundant computation. This paper proposes an incremental consistency execution method based on task fact contracts, field-level dependency masks, and state perturbation result invariant domains. After an initial verified execution, the system constructs conservative invariant domains for critical inputs and uses them to determine whether downstream results can be safely renewed without re-invoking expensive components. When re-execution is required, only the smallest affected output fields are recomputed, and an equivalence barrier prevents unnecessary downstream propagation. A submission-time version consistency gate further ensures the safety of side-effecting actions. Experiments on industrial fault diagnosis, enterprise analytics, and LLM-based multi-tool assistants show that the proposed method significantly reduces expensive component calls and end-to-end latency while maintaining high consistency and low incorrect-reuse rates.
[AI-75] FlashBoB: I/O-Efficient Exact Backward-over-Backward for Softmax Attention
链接: https://arxiv.org/abs/2609.24089
作者: Anthony Givans,Michael Crawshaw,Mingrui Liu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Transformer models built on the attention mechanism have become a central building block in modern deep learning, yet softmax attention remains a major bottleneck for long-context workloads. While FlashAttention makes the forward and first backward passes I/O-efficient, it does not support backward-over-backward (BoB), which enables exact differentiation through the backward pass for applications such as second-order optimization, test-time training, gradient-based memory, and meta-learning. Existing BoB implementations either materialize large intermediate tensors or exhaust GPU memory at long sequence lengths. We present FlashBoB, an exact, I/O-efficient algorithm for BoB in softmax attention that keeps computation within on-chip tiles and avoids all N \times N intermediate tensors, where N is the sequence length. The key insight is a hierarchical affine structure in the softmax double backward: two row-wise scalars determine all outputs through affine transformations. This yields a two-pass schedule with bounded on-chip static random-access memory (SRAM) usage and minimal off-chip high-bandwidth memory (HBM) traffic. FlashBoB achieves \Theta(N^2 d^2/M) HBM traffic ( d is the head dimension and M is the memory size) and, within the standard FlashAttention-style score-recomputation model, matches the inherited large-cache lower bound for exact forward attention. Empirically, it scales exact attention BoB to N=262\textK on a single A100 80GB GPU, where prior PyTorch exact baselines fail by N=16\textK , and is up to 6.3\times faster than FlashBack. These results make exact second-order attention practical at long-context sequence lengths where prior implementations cannot run efficiently.
[AI-76] From Bits to Beliefs: Recoverable Semantic Fingerprints for Black-Box Verification of Large Language Models
链接: https://arxiv.org/abs/2609.24084
作者: Jiaxin Hong,Yuxin Peng,Hongyao Yu,Hao Fang,Shuoyang Sun,Bin Chen
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:Open-weight large language models (LLMs) can be copied, modified, and redeployed behind black-box APIs, making post-release ownership verification difficult. Existing black-box fingerprints often rely on secret query-key pairs that reproduce predefined responses, and can therefore be easily disrupted by fine-tuning, pruning, quantization, model merging, and serving-time prompt changes. We propose SimPrint, a recoverable semantic fingerprinting framework for black-box LLM ownership verification. Rather than relying on isolated exact matches, SimPrint encodes a private owner signature into a coded semantic fingerprint domain, distributing ownership evidence across natural binary question-answering probes. It implants only base-deviating probes through a low-interference batch update that preserves the original model behavior, and later recovers the signature by parsing suspect-model responses into reliable bits or erasures with an error-correcting recovery mechanism. Because verification only uses input-output queries, SimPrint remains applicable when model weights or activations are inaccessible. Experiments on three open-weight LLMs show that SimPrint reliably recovers the owner signature in both clean and modified settings, remains robust under fine-tuning, pruning, quantization, model merging, and serving-time perturbations, and maintains comparable downstream utility.
[AI-77] MECT: Mixture of Experts with CNN-Transformer Network for Speaker verification
链接: https://arxiv.org/abs/2609.24061
作者: Yu Zheng,Jinghan Peng,ChangHao Zhang,Jian Liu,Weiqiang Wang
类目: ound (cs.SD); Artificial Intelligence (cs.AI); Audio and Speech Processing (eess.AS)
备注: 5 pages
Abstract:In this paper, we propose MECT, a speaker verification model that integrates the Mixture-of-Experts (MoE) mechanism into a CNN-Transformer backbone with optimized block structure and stacking scheme. Specifically, we investigated four MoE variants that span utterance-level and frame-level granularity with dense and sparse routing strategies. The MoE mechanism proves to be effective over the baseline without MoE with only a small increase in parameters. We further scale MECT to a series of model sizes, all maintaining compact parameters and low computational complexity. In particular, MECT-B2 achieves state-of-the-art performance on VoxCeleb1 and delivers strong results on CN-Celeb, demonstrating its effectiveness across diverse datasets. In addition, we establish a streaming inference paradigm through causal retraining, which maintains strong performance at a chunk size of 100ms.
[AI-78] Structured Decomposition for Reliable LLM -Generated Access Control Policies
链接: https://arxiv.org/abs/2609.24036
作者: Vatsal Gupta,Darshan Sreenivasamurthy
类目: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
备注:
Abstract:This paper presents an LLM-based system that translates natural-language access control policies (NLACPs) into executable Rego code for Open Policy Agent (OPA). It provides a modular, end-to-end pipeline for policy detection, component extraction, schema validation, linting, compilation, and automated test generation and execution. The system is designed to bridge the gap between human-readable access requirements and machine-enforceable policy-as-code (PaC), with a focus on deployment reliability and security correctness. We evaluate the system on 372 ACRE-complete access control statements with non-null subject, action, and resource annotations against a direct single-prompt LLM baseline to isolate the contribution of structured decomposition and schema-aware validation. The system achieves a 50.3% end-to-end policy correctness rate, compared with 15.3% for the baseline, representing a 3.3x improvement. A policy is counted as correct only if it satisfies compilation, linting, and both positive and negative tests, making this a strict measure of deployable correctness. On security-critical patterns, the system generates correct deny semantics for 87.5% of deny policies (baseline: 37.5%), ownership conditions for 100% of ownership-qualified policies (baseline: 40%), and status-qualified conditions for 100% of status-qualified policies (baseline: 55.6%). These results indicate that structured decomposition and schema-aware validation play a critical role in improving the reliability of LLM-generated authorization policies. Subjects: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR) Cite as: arXiv:2609.24036 [cs.AI] (or arXiv:2609.24036v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2609.24036 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-79] Synthesizing Reactive Character Behaviors for Continuous Games via Programmatic Policy Search SIGGRAPH
链接: https://arxiv.org/abs/2609.24025
作者: Maxim Gumin,Hsueh-Ti Derek Liu,Victor Zordan,Daniel Ritchie
类目: Artificial Intelligence (cs.AI)
备注: To be presented at SIGGRAPH Asia 2026
Abstract:We present a method for synthesizing reactive character behaviors for continuous games as compact, human-readable programs. Game AI practice still relies heavily on manually authored behavior trees, state machines, and scripts, while academic reinforcement learning typically produces opaque neural controllers that are expensive to train and difficult to edit. Our approach bridges this gap by searching directly over a domain-specific language for continuous-space game policies. The language is designed around reactive geometric decisions and includes higher-order constructs such as direction maximization. These constructs help discretize a continuous behavior space into enumerable program structures. To make program search practical, we introduce a large set of synthesis antipatterns that remove redundant program forms while preserving behavioral coverage. We further combine bottom-up symbolic enumeration with top-down guidance from a coding agent. Our resulting method, agentic sketching, has the agent propose high-level policy structure and call an enumerator to complete local program slots. We evaluate the method on a benchmark of 14 continuous games, ranging from classic control tasks to multi-agent football. We find that pure enumeration is often more efficient than using a coding agent alone, while the combined method substantially outperforms both. Our results suggest that programmatic policy search can be a practical authoring tool for game AI: designers specify reward functions, and the system discovers editable behaviors that are effective, portable, and often surprising.
[AI-80] Context-Aware Pre-Deployment Evaluation of AI Systems: A Regulatory Framework for Nigerian Fintech
链接: https://arxiv.org/abs/2609.24016
作者: Andrew Anogie Uduimoh,Hadiza Umar Yusuf,Oluwafemi Osho
类目: Artificial Intelligence (cs.AI); Computers and Society (cs.CY)
备注:
Abstract:Commercial large language models are increasingly deployed across African fintech infrastructure for fraud detection and customer communication, yet no Nigerian or African continental regulatory instrument specifies what pre-deployment evaluation such systems must undergo before procurement. This paper reviews African fintech AI governance across global, continental, and Nigerian instruments, and shows that safety is affirmed as a principle while pre-deployment evaluation is operationally unspecified. Generic safety benchmarks cannot surface the failure modes most relevant to this domain, since none contain Nigerian institutional content or test for false positive misclassification of legitimate financial communications. These claims are demonstrated using SafeAlert, a purpose-built evaluation kit applied to six commercial models across three system prompt conditions. Results show that models resisting generic harmful content requests still produce complete fraud scripts under specific framing, and that several models misclassify most legitimate Nigerian bank communications as suspicious or fraudulent, a failure invisible to standard safety evaluation. The paper concludes with a regulatory framework proposing pre-deployment evaluation requirements for the CBN, NITDA, SEC, and the AU, arguing that the identified gap reflects an absence of regulatory specification, not a shortage of technical or financial resources.
[AI-81] FinInteract: Benchmarking Clarification and Intent Integration in Ambiguous Financial Question Answering
链接: https://arxiv.org/abs/2609.24002
作者: Xinyu Wang,Tung Sum Thomas Kwok,Zhenghan Tai,Guang Cheng
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language model agents increasingly answer financial questions by searching regulatory filings. Such questions are often deceptively under-specified: Meta Platforms’ “operating income” is 46.75B consolidated but 62.87B for the Family of Apps segment, and each reading is exactly verifiable against the filing. A capable agent should recognize the ambiguity and ask, rather than commit to a plausible but unintended reading. Existing financial benchmarks cannot measure this, because one gold answer per question cannot separate agents that resolve the ambiguity from those that guess the common reading, a blind spot we call the single-gold illusion. We release FinInteract, a bilingual (English/Chinese) benchmark of 173 instances that pairs each question with a default and an intended interpretation across a five-category ambiguity taxonomy, and grades whether an agent elicits the right clarification and then integrates it. Re-grading identical outputs against the default rather than the intended reading inflates GPT-4o’s accuracy by 3.1 times, confirming the illusion. Beyond it, we find that models answer above 90% once the interpretation is supplied but at most 28.9% when they must elicit it themselves, that targeting is uneven across a taxonomy well powered for entity scope and metric definition and exploratory elsewhere, and that conditioning on the ambiguity category improves resolution at both inference and training time.
[AI-82] Misaligned Clinical Risk Classification and Cost Asymmetry in Open-Weight Large Language Models ML4H2026
链接: https://arxiv.org/abs/2609.23999
作者: Star S.D. Liu,Xiyu Ding,Robert B. Barrett,Alberto Santamaria-Pang,Nic Dobbins,Harold P. Lehmann
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Submitted to ML4H 2026
Abstract:How large language models (LLMs) integrate patient risk with clinical cost tradeoffs remains poorly understood. We investigated how four open-weight LLMs (Qwen-2.5-7B/32B and Llama-3.1-8B/70B) internally represent cost tradeoffs, how these representations relate to clinical predictions, and whether decisions shift as predicted by the specified cost direction and magnitude. Using a public diabetes dataset, we varied 11 false-negative (FN) to false-positive (FP) cost ratios across three phrasings and examined representations and behavioral outputs. Patient risk was linearly recoverable on par with conventional classifiers (AUC \approx 0.83 ), and cost direction was recoverable in every model. However, representational shifts in cost direction tracked output changes only in the two larger models, and responses to cost magnitude were predominantly direction-agnostic. Only 2 of 12 model-phrasings showed both opposing responses to increasing FN versus FP costs and cost-correct ordering. Representationally, a direction fitted on one cost side did not invert when transferred to the other, as expected under mirror-symmetric encoding. These findings suggest that LLMs encode risk and cost information but do not reliably integrate them into cost-correct decisions. Clinical evaluations should therefore include tradeoff tests, phrasing sensitivity, and default operating points alongside predictive performance.
[AI-83] RoboTalk: Learning Multi-Robot Communication and Coordination from Multimodal Demonstrations
链接: https://arxiv.org/abs/2609.23997
作者: Dorian Benhamou Goldfajn,Mason Nakamura,Saaduddin Mahmud,Justin Svegliato,Kyle H. Wray,Shlomo Zilberstein
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Multi-robot collaboration could enable more efficient and scalable solutions to complex robotic tasks, but collaboration under partial observability remains challenging. Natural-language communication offers a promising approach to coordinating robots under partial observability. However, in decentralized manipulation, jointly learning explicit inter-robot communication and skill-level action selection from multimodal demonstrations remains underexplored for small vision-language models (VLMs) intended for on-device deployment. To address this gap, we introduce RoboTalk, a synthetic data-generation pipeline and dataset of 7,950 multimodal trajectories spanning 53 mobile-manipulation kitchen tasks for training small VLMs to communicate and coordinate. The dataset includes a leader-follower planning protocol, tool calls (perception, manipulation, navigation, and communication), rationale traces, and diversified natural-language communication. Fine-tuning open-source models on our dataset can reach 77% success on novel held-out tasks, a significant improvement over the untuned open source models, which had a success rate of around ~2%.
[AI-84] ACLArena: Agent Continue Learning in Multi-stage Post-training
链接: https://arxiv.org/abs/2609.23989
作者: Haixin Wang,Xiaoxuan Wang,Junkai Zhang,Han Zhang,Renliang Sun,Alexander K Taylor,Yidan Shi,Haoran Deng,Chenguang Wang,Jason Cong,Yizhou Sun,Wei Wang
类目: Artificial Intelligence (cs.AI)
备注: 25 pages, 7 figures, under review
Abstract:Building general-purpose agents for industrial deployment requires integrating multiple capabilities, each typically acquired at a distinct stage of training. Yet there is currently no well-established recipe for Agent Continual Learning (ACL), with little understanding of the trade-offs among existing integration paradigms. To address this gap, we introduce ACLArena, a framework for comprehensively studying, analyzing, and evaluating ACL. We first build a sequential training pipeline and conduct an in-depth analysis that explains the mechanisms of forgetting and generalization from two complementary perspectives, the model level and the token level. Guided by these analyses, we systematically compare multi-teacher on-policy distillation, self-distilled fine-tuning, and model merging to assess their ability to recover previously learned capabilities while preserving newly acquired ones. Through extensive experiments, we develop a detailed understanding of how capabilities transfer across stages. Finally, we propose a new ACL recipe that combines offline replay over high-quality trajectories with a routed network of multiple LoRA experts each specialized via RL, substantially improving the agent’s ability to learn across multiple domains. Comprehensive experiments on four reasoning and agentic tasks, evaluated under both in-domain and out-of-domain settings, demonstrate the value of our analysis and the effectiveness of our approach.
[AI-85] Jev-Mem: System-One-Controlled Agent ic Memory for Efficient AI Agents
链接: https://arxiv.org/abs/2609.23986
作者: Dongming Jiang,Yi Li,Bingzhe Li
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Agentic memory is becoming essential for long-horizon AI agents, yet many existing systems rely on autoregressive LLMs to control how memories are organized, retrieved, and used, placing expensive generation on the critical path of memory operations. We introduce \textbf\method, a new agentic memory architecture inspired by System-One/System-Two cognition. System One captures fast, lightweight decision-making, whereas System Two performs slower, deliberative reasoning. Jev-Mem brings this division of labor to agentic memory through a dedicated System-One control plane, a structured multi-relational memory plane, and a System-Two reasoning plane. The System-One controller governs memory typing and relational organization during construction, and dynamically performs query routing, retrieval-budget allocation, graph traversal, candidate scoring, and adaptive stopping during retrieval. System Two is invoked only for complex reasoning and answer synthesis. This design improves both memory effectiveness and system efficiency: on LoCoMo Jev-Mem achieves an overall LLM-as-a-Judge score of 0.777, an 11.0% relative improvement over the strongest baseline, while reducing memory construction time to 158,s, a 6.6 \times speedup over the fastest competing memory system, and lowering average query latency to 0.93,s, a 36.7% reduction.
[AI-86] MobileCybench: Evaluating Agent Vulnerability Discovery via Executable Probes
链接: https://arxiv.org/abs/2609.23980
作者: Andy K. Zhang,Ava Huang,Joey Ji,Wai Han,Thomas Qin,Nardos Demilew,Michael Tian-Yue Liu,Brian Song,Riya Dulepet,Brian Wang,Kyleen Liao,Cuiyuanxiu Chen,Nishka Kacheria,Andrew Wu,Pratham Rangwala,Xinjie Wang,Laura Gomezjurado Gonzalez,Anita Ding,Benjamin Yi,Daniel E. Ho,Dan Boneh,Dawn Song,Ion Stoica,Percy Liang
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:AI agents now report vulnerabilities faster than maintainers can review them. Reports often depend on security properties specific to the application, and require considerable human labor to process. To mitigate this, we introduce a framework for evaluating vulnerability reports via probes, executable checks of security properties. A reported exploit is evaluated by replaying it against the application and running the probes: a triggered probe indicates both that the exploit succeeded and which security property it violated. As a probe encodes a security property rather than a known vulnerability, it can detect vulnerabilities that were not known when the probe was written. We instantiate the framework as MobileCybench, a benchmark for vulnerability discovery by AI agents in 13 Android applications, with 495 probes written and reviewed by the authors. We evaluate 5 coding agents (OpenCode with GPT-5.5, GPT-5.6-Sol, and GLM-5.2; Claude Code with Opus 4.8 and Opus 5) under 4 settings: as a malicious app on the victim’s device or as a remote attacker with a low-privilege account, each with either only an obfuscated APK or access to the application’s source code. Given only the obfuscated APK, the top agent, OpenCode with GPT-5.6-Sol, triggers probes in 53.8% of applications in the malicious-app setting and 16.7% in the remote-attacker setting. With source code, the trigger rate across all agents and both attack settings increases from 28.8% to 32.8%. Building and running the benchmark surfaced 23 previously unreported vulnerabilities, the majority of which have been confirmed by maintainers.
[AI-87] LEAP-NBV: Lightweight Edge Active-Perception for Foundation-Model Next-Best-View Planning
链接: https://arxiv.org/abs/2609.23974
作者: Boxun Hu,Jiawei Ge,Axel Krieger,Peng Wang,Tinoosh Mohsenin
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Foundation models are endowing autonomous systems with greater intelligence, enabling a more comprehensive understanding of the environment through visual perception. A representative example is Human Mesh Recovery (HMR), which provides useful estimates of a target’s 3D pose and shape that can benefit tactical missions. However, the size and power demands of such models make them difficult to run on edge platforms and limit their real-time performance, undermining the requirements of tactical edge deployment - especially for active perception, where a mobile robot must plan its next-best view on-board and cannot offload computation under contested communications. We present LEAP-NBV, a lightweight active-perception framework that runs foundation-model-driven Next-Best-View (NBV) planning on-board an edge device. To this end, we distill a family of large HMR teachers, each into a compact 32M student, with an offline mesh objective, then quantize the vision encoder to FP16 and characterize its on-device accuracy and latency. Within an occlusion-aware active perception loop, we evaluate all configurations on the same held-out benchmark and deploy the end-to-end pipeline on an NVIDIA Jetson Xavier NX, reporting measured on-device latency and energy. Distillation recovers 6-7 mm of Procrustes-aligned mean per-vertex position error (PA-MPVPE) over the undistilled student on the test set. Selecting the edge-optimal compression model brings the HMR engine to ~12 ms at a small accuracy cost and runs the full closed loop at 3.6 FPS and 2.6 J per frame, achieving a 2.0x speedup and 3.0x lower energy than the uncompressed model while nearly matching downstream task quality.
[AI-88] Divergent strategies and convergent outcomes in autonomous materials discovery
链接: https://arxiv.org/abs/2609.23957
作者: Jihan Kim
类目: Artificial Intelligence (cs.AI); Materials Science (cond-mat.mtrl-sci)
备注:
Abstract:Scientific agents are mostly evaluated on whether they complete tasks or recover known results; we instead study variation across repeated open-ended campaigns. Sixteen separately initialized sessions of one model-harness configuration received a frozen database of 12,499 metal-organic frameworks, a methane-storage objective, a pinned protocol and a one-week budget. Strategies diverged into four approaches spanning 100–5,000 screened structures, and eight built 2,253 hypothetical structures. Yet the agents recovered the same materials frontier near 200 cm^3/cm^3, and an independent calculation of the database’s porous region found its nine best structures all among their reports. Enforced checks on half the agents raised fresh-run reproduction from one of eight to eight of eight but could not detectably improve conclusion validity, because fifteen of sixteen agents selected the same audit-excluded entry, an incomplete structure whose missing anions created artificial pore volume. Replicated agents thus reveal both robust conclusions and common-mode errors from shared inputs.
[AI-89] Djinnlang: Higher-Level Programming by Unambiguous Specification with an LLM in the Compiler
链接: https://arxiv.org/abs/2609.23954
作者: Simon Henniger,Stephen Chong,Nada Amin
类目: Programming Languages (cs.PL); Artificial Intelligence (cs.AI); Software Engineering (cs.SE)
备注:
Abstract:Programmers write formal specifications, and LLMs implement them, proving that each implementation matches its spec. Taken to its extreme, this makes specification languages the new programming languages. We argue that an unambiguity constraint is key: in addition to proving that its implementation satisfies the specification, the LLM must also prove that any other implementation satisfying it must produce the same outputs on the same inputs, i.e. that the relation formed by the constraints is deterministic. This leaves the LLM no leeway on program semantics: as with a conventional compiler, the generated code never needs to be read and can be regenerated from the spec at any time. Under this constraint and with a powerful LLM, the difference between a specification language and a programming language becomes essentially meaningless, and the LLM essentially becomes a part of the compiler toolchain. The arrangement doubles as a strong form of AI control: an untrusted model writes the code, yet its work is tightly checked by a verifier. To demonstrate that our LLM-in-the-compiler paradigm is feasible when supported by our unambiguity constraint, we present Djinnlang, a high-level specification language built for this future. A Djinnlang program consists only of specifications — the programmer never writes executable code. In place of a traditional compiler, a symbolic translator lowers each spec to Dafny stubs and proof obligations, and a driver harness orchestrates an LLM that fills in implementations and proofs, all checked by the Dafny verifier. We evaluate our language and implementation on multiple examples and we show that it is self-hosting: an LLM can implement the Djinnlang translator from its specification and the reimplementation can verify itself. Subjects: Programming Languages (cs.PL); Artificial Intelligence (cs.AI); Software Engineering (cs.SE) Cite as: arXiv:2609.23954 [cs.PL] (or arXiv:2609.23954v1 [cs.PL] for this version) https://doi.org/10.48550/arXiv.2609.23954 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-90] Agents That Edit Documents: Measuring Agent Agent ic PDF Forgery Against a Non-Agentic Control
链接: https://arxiv.org/abs/2609.23953
作者: Simiao Ren,Ankit Raj,Tommy Duong,Yuxin Zhang,Dennis Ng,Xingyu Shen,Kidus Zewde,Yuchen Zhou,Neo Tiangratanakul
类目: Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR)
备注: 15 pages, 12 figures, 2 tables
Abstract:AI agents that carry a multi-step computer task through on their own became ordinary tools in the past year, and the same autonomy is available to anyone whose task is harmful. We ask what that means for a relying party – an insurer, a lender, an auditor – whose evidence is a filed PDF. AgentForge-Bench measures how reliably an off-the-shelf coding agent, driving one of seven open-weight models with a shell and the stock Python PDF stack, alters one dollar amount, date or address in a real filed financial document from a single sentence of intent, graded by rules rather than by a model. Across 1,750 cells, 1,419 (81.1%) satisfy the verifier, and 808 (46.2%) also survive every stricter filter: visible, localized, typeface-matched, original value gone document-wide. A deterministic script with no model in it solves 98 of the 125 documents; the agents solve 124, and none the script solves alone. Agents misreport 41% of their wrong edits as done, no model refused, and the cheapest verified forgery costs 2.4 cents. The raw rate overstates the threat by about a factor of two; the strict rate is still large.
[AI-91] Echo State Network (ESN) for Signal Recovery in RF-Impaired IBFD MIMO Systems
链接: https://arxiv.org/abs/2609.23945
作者: Conrad Prisby,Siyao Li,Chengtao Xu,Thomas Yang
类目: Artificial Intelligence (cs.AI)
备注: 6 pages, 5 figures. This work has been accepted by IEEE Milcom 2026
Abstract:In-band full-duplex (IBFD) multiple-input multiple-output (MIMO) systems enable simultaneous transmission and reception on the same frequency band, improving spectral efficiency for next-generation wireless networks. However, IBFD-MIMO systems are susceptible to self-interference (SI), which may overpower signals of interest (SOI). In this scenario, blind source separation (BSS) algorithms can be adopted to remove SI and perform joint sensing and communication (JSAC), but BSS algorithms mostly assume an idealized linear and quasi-stationary signal model, which does not hold under realistic radio frequency (RF) impairments, such as I/Q imbalance, carrier frequency offset (CFO), phase noise, and power amplifier nonlinearity. This paper proposes a two-stage echo state network (ESN)-based scheme that is superior to BSS under these realistic conditions. A frozen ESN is trained offline to characterize the static SI path, while an adaptive ESN, updated online via recursive least squares, tracks the time-varying SOI path using sparse pilot symbols. We evaluate the proposed scheme’s SOI recovery performance and acquisition speed with different block sizes, comparing it against other recurrent neural networks (RNN), such as long short-term memory (LSTM) and gated recurrent unit (GRU). Simulation results show that the proposed approach outperforms BSS, LSTM, and GRU in both efficiency and SOI recovery, demonstrating the viability of ESNs for real-time, nonlinear self-interference cancellation in realistic IBFD MIMO systems.
[AI-92] MCPGen: Benchmarking LLM s on Executable MCPWorkflow Development
链接: https://arxiv.org/abs/2609.23925
作者: Yingxuan Yang,Jiaqi Liu,Lirui Guan,Jiaye Gao,Weiwen Liu,Weinan Zhang,Ying Wen
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:We study whether LLMs can produce executable workflow artifacts that remain consistent across graph structure, tool implementation, schema bindings, and runtime wiring. In this setting, correctness depends on cross-layer consistency: a workflow may be structurally plausible, yet still fail because tool implementations, schema bindings, or runtime execution do not align. Existing benchmarks largely evaluate these capabilities in isolation or rely on trajectory-level proxies, leaving open whether generated workflow artifacts execute end-to-end. We introduce \textbfMCPGen, an executable benchmark for Model Context Protocol (MCP) workflow development. MCPGen contains 100 self-contained MCP projects across 16 application domains and evaluates three diagnostic tasks: workflow reconstruction, tool creation, and backward-compatible workflow extension. We evaluate 11 representative LLMs in a single-turn foundation-model setting, assessing generated artifacts through static analysis, unit and integration tests, and process-isolated end-to-end execution. Models reach 88.5% on workflow reconstruction, but no model exceeds 57% end-to-end execution success. Per-tool unit-test pass rates reach 63.8%, while project-level integration success does not exceed 45%, suggesting that integration remains a major bottleneck even when isolated tool tests pass.
[AI-93] Increasing Skill Level Recruits Deeper Attention Layers in a Frozen Chess Transformer
链接: https://arxiv.org/abs/2609.23917
作者: David Litman
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 19 pages, 12 figures. Code and data: this https URL . Built with chessformer-lens library: this https URL
Abstract:Chess involves complex reasoning in a deterministic environment, which makes it a useful setting for studying the mechanisms of computation inside transformers. The Maia-3 chess transformer takes Elo, a measure of competitive chess skill, as an input to the pre-trained network, so we can vary the skill the network is conditioned on with no change to its weights. Here we investigate how turning this skill dial affects self-attention. Ablating every attention head at every Elo from 700 to 2500, we find 1) increasing skill pushes the causal center of mass of the computation deeper, monotonically, for every chess piece and move type we measured; 2) the depth migration is much greater for specific tactics, especially knight forks, than for other move types; 3) the migration consists of deeper heads getting recruited for more specialized computations while one shared shallow head keeps a roughly constant contribution. These results may shed light on how conditioning inputs redistribute computation in larger transformers.
[AI-94] ReVeal: A Reconstruction-Aware Real-to-Sim Framework for VLA Policy Evaluation
链接: https://arxiv.org/abs/2609.23910
作者: Xinyi Wang,Heng Hao,Wenjun Hu,Anna Enyu Li,Dizhi Ma,Karthik Ramani,Hankyu Moon,Yeong-Dae Kwon
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Simulation-based evaluation provides a scalable and repeatable alternative to real-world evaluation of vision-language-action (VLA) policies. However, reconstruction errors can cause simulated policy performance to diverge from real-world performance, motivating the need to assess reconstructed environments for downstream VLA policy evaluation. We present ReVeal, a real-to-sim assessment framework combining workspace reconstruction, reconstruction-level assessment, and matched closed-loop policy evaluation. Novel-View Mesh Fidelity (NVMF) and Annotated Planar Geometry Fidelity (APGF) assess observation and planar geometric fidelity, respectively. We also develop PGSR-D, a reconstruction pipeline incorporating monocular depth supervision to improve geometry where multi-view visual cues are limited. Across 8 assessment scenes, NVMF and APGF consistently distinguish the fidelity of 2DGS, PGSR, and PGSR-D. Matched evaluations of GR00T, SmolVLA, and pi0.5 across 8 humanoid manipulation tasks show consistent ordering between reconstruction fidelity and real-sim performance agreement across pipelines. Further analysis of the evaluation workspaces shows that higher fidelity is associated with stronger real-sim agreement.
[AI-95] Connecting the Dots in Agent ic AI Security: A Cross-Dimensional Threat Taxonomy Evaluation Maturity and Open Challenges
链接: https://arxiv.org/abs/2609.23894
作者: Heewon Baek,Alsharif Abuadbba,Kristen Moore,Hyoungshick Kim,Surya Nepal
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: 22 pages
Abstract:Agentic AI extends LLM security beyond generated content to persistent state, autonomous actions, tool use, and interactions with humans and other agents. Existing threat classifications often emphasize individual dimensions, obscuring connections among entry points, affected components, and security consequences. The known threat landscape also differs from the coverage demonstrated by empirical research. Through a structured review of 66 studies published from 2022 to 2026, we introduce T=S, B, P, A, a cross-dimensional representation linking affected functional or system surfaces S, interaction or trust boundaries B, violated security properties P, and empirically examined architectures A. We analyze 22 artifact-backed red-teaming studies and 11 representative security benchmarks to characterize empirical coverage and evaluation maturity. Within the selected studies, evidence concentrates on prompt/reasoning, memory, and tool-mediated attacks, predominantly in single-agent settings. Persistent, Human–Agent, complex multi-agent, systemic, and long-horizon threats receive less coverage. These findings describe the selected corpus rather than establish gaps across all empirical research. Heterogeneous metrics, limited adaptive defense evaluation, architectural imbalance, and incomplete execution-state capture further constrain comparison and reproducibility. We derive 13 open research questions to guide more systematic, architecture-aware, and reproducible security evaluation of agentic AI.
[AI-96] Circuit-Diff: Factual Edit-based Intervention Method for Localizing Knowledge in Attribution Graphs
链接: https://arxiv.org/abs/2609.23892
作者: Edward G. Friedman,Xiangchen Song
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 26 pages, 12 figures, 3 tables. Code: this https URL | Data: this https URL | Demo: this https URL
Abstract:Mechanistic interpretability defines features as the fundamental units of a neural network and circuits as the weighted subgraphs that carry out its computation. Because individual neurons are polysemantic, Cross-Layer Transcoders (CLTs) were introduced as a way to approximate a model’s circuits by generating an attribution graph. The nodes of that graph, however, are unlabeled features: reading a graph means pruning it and then working out by hand what each surviving node means. To make CLTs easier to use for circuit discovery, we introduce Circuit-Diff, which intervenes on the model itself with a low-rank factual edit and takes the features whose role in the attribution graph changes under that edit as related to the edited knowledge. On the edits we examine, the flagged nodes are not only detectors of the object token: read off the CLT’s released feature dashboards, they include features for the history, geography and associations surrounding the old and new objects. We formalize the method, measure how reliable a frozen CLT remains after a factual edit, test the selected nodes causally by patching them on up to 24 CounterFact edits, give a case study, and release an open-source implementation built on the circuit-tracer package, together with two further tools (multi-prompt aggregation and rule-based supernode labeling).
[AI-97] SyzHarness: Patch-Based Kernel Bug Reproduction with LLM -Synthesized Fuzzing Harnesses
链接: https://arxiv.org/abs/2609.23889
作者: Xingyu Li,Juefei Pu,Haonan Li,Arrdya Srivastav,Kareem Shehada,Srikanth V. Krishnamurthy,Zhiyun Qian
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:Automated kernel vulnerability reproduction is essential for bug triage, patch validation, and regression testing, but still lacks an effective and efficient solution. The core challenge is twofold: a reproducer must first recover the trigger scaffold needed to reach the vulnerable state and determine the precise concrete values that actually trigger the bug. Existing directed fuzzing approaches are ineffective at recovering the necessary trigger scaffold, while LLM- only generation is brittle because it struggles with concrete-value discovery and runtime nondeterminism. We design SyzHarness, a framework that combines LLM reasoning with coverage-guided fuzzing for patch-based Linux kernel vulnerability reproduction. Given a patch, SyzHarness uses an LLM agent grounded by code navigation tools to synthesize a parameterized fuzzing harness that fixes the prerequisite setup logic while exposing only uncertain, bug- critical input parameters to be mutated by Syzkaller. SyzHarness then translates this harness into a Syzkaller- compatible interface and iteratively refines it using hierarchical reachability feedback. We evaluate SyzHarness on multiple datasets of triggerable real-world Linux kernel vulnerabilities. On 100 KernelCTF cases, SyzHarness achieves a 78% bug reproduction success rate. On the SyzDirect benchmark, SyzHarness achieves a 73% bug reproduction success rate, substantially outperforming prior directed greybox fuzzing. On 50 recent, known-triggerable syzbot bugs fixed after March 2026, SyzHarness reproduces 40/50 (80%) using only the fix commits as input. Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI) Cite as: arXiv:2609.23889 [cs.CR] (or arXiv:2609.23889v1 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2609.23889 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Xingyu Li [view email] [v1] Sun, 20 Sep 2026 21:50:36 UTC (3,620 KB) Full-text links: Access Paper: View a PDF of the paper titled SyzHarness: Patch-Based Kernel Bug Reproduction with LLM-Synthesized Fuzzing Harnesses, by Xingyu Li and 6 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 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] GLR-MM: Graph-Based Global-Local Reconstruction for Robust Multimodal Chest X-ray and EHR Representation Learning under Missing Modalities MICCAI
链接: https://arxiv.org/abs/2609.23876
作者: Surbhi Sharma,Nikhil Manali,Devesh Maheshwari
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted in MICCAI
Abstract:Clinical multimodal models must often predict before all chest X-ray (CXR) and electronic health record (EHR) inputs are available. Existing approaches align observed representations, model missingness, or reconstruct across modalities, but do not jointly exploit within-patient and clinically similar inter-patient evidence. We propose GLR-MM, a Graph-Based Global-Local Reconstruction framework for early ICU mortality prediction. It maps five CXR-EHR modalities to a shared space, reconstructs missing embeddings through complementary local cross-modal and global graph-attention branches, adaptively fuses their estimates, and optimizes class-balanced prediction, reconstruction, and contrastive objectives. On 9,620 MIMIC-derived ICU stays, we evaluate 10%, 30%, and 50% random modality missingness with shared deterministic masks. MUSE performs better under mild and moderate missingness, whereas GLR-MM achieves higher AUROC and AUPRC at 50% by 0.0088 and 0.0249, respectively. These results indicate that graph-guided reconstruction is most useful when inputs are severely incomplete.
[AI-99] Pretraining of Medical Visual Encoders Toward Multi-modal Large Language Models
链接: https://arxiv.org/abs/2609.23860
作者: Tianyou Jiang
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Multimodal Large Language Models (MLLMs) commonly reuse visual encoders pretrained with CLIP, although the features of these ViTs are ultimately consumed by autoregressive LLMs. We refer to this mismatch as the semantic-interface gap and introduce MedMLIP, a framework that pretrains the visual encoder through report generation with a frozen LLM, while employing Local Relational Distillation (LRD) to preserve relationships among visual patches to avoid visual collapse. We pretrain MedMLIP on IU-Xray and Open-PMC-300K and evaluate the resulting encoders on VQA-RAD and SLAKE. Only the ViT is transferred, while the guiding LLM and projector are replaced, allowing us to assess cross-LLM transferability. Our cross-LLM transfer experiments demonstrate the value of pretraining visual encoders for their autoregressive LLM interface while trying to preserve more fine-grained visual information. Code and the pretrained model are available at this https URL
[AI-100] WorkWorlds: An Infrastructure for Evaluating AI Agents on Workplace Tasks
链接: https://arxiv.org/abs/2609.23806
作者: Yining Hua,Levi Lian
类目: Artificial Intelligence (cs.AI)
备注: 4 figures
Abstract:Many knowledge-work benchmarks are constructed around individual tasks, with the context needed for each task selected together with or after the task has been specified. This design measures performance on workplace-like tasks in an environment assembled for the task. When task specification guides which context is selected, the evaluation can encode task information into the environment and pre-complete part of the information-localization work that workplace performance normally requires. We introduce WorkWorlds, an evaluation infrastructure that separates organizational state from task specification. A world first fixes a revision, date, and employee seat and materializes the organizational state that employee can access; tasks are introduced only afterward. We implement WorkWorlds in a primary synthetic pharmaceutical company with 8 measured tasks across 6 employee seats, and construct additional organizational worlds. Across 192 matched evaluations, moving from task-curated context to the full role-visible workplace reduced evidence access from 90.4% to 74.5% and criterion pass from 79.4% to 68.2%, while pass conditional on evidence access remained nearly unchanged; most of the measured difference occurred before the agent reached sufficient evidence.
[AI-101] Belted Engression: Sufficient Dimension Reduction for Generative Distributional Regression
链接: https://arxiv.org/abs/2609.23789
作者: Wenxi Tan,Bing Li,Lingzhou Xue
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Statistics Theory (math.ST); Methodology (stat.ME); Machine Learning (stat.ML)
备注: 32 pages, 6 figures
Abstract:Modern conditional generative models face significant challenges when learning complex covariate dependencies. While sufficient dimension reduction (SDR) provides a principled approach to compress these dependencies, traditional SDR frameworks were not formulated for conditional generation. To bridge this gap, we propose Belted Engression, a unified and architecturally parameter-efficient framework for generative distributional regression. Our approach establishes an end-to-end compress-then-generate paradigm driven by sufficient representation learning, embedding a structural bottleneck into the generative architecture. Theoretically, we prove that the standard SDR condition is equivalent to a law-preserving generative factorization, which is achieved at the global optimum of the population Belted Engression objective. Furthermore, by uncovering a localized Bernstein-type control for the energy-score loss, we establish finite-sample convergence rates that are sharper than those of existing results. We also prove that this belted architecture is strictly smaller, operating with an asymptotically vanishing parameter count relative to the unstructured baseline. Extensive simulations and real-world applications demonstrate that Belted Engression achieves superior distributional prediction and SDR recovery with fewer trainable parameters.
[AI-102] On Probabilistic Inference Through Parametric Tensor Decomposition in Base Tensor Networks
链接: https://arxiv.org/abs/2609.23774
作者: Sagad Hamid,Tanya Braun
类目: Artificial Intelligence (cs.AI)
备注: Accepted at: The 17th International Conference on Scalable Uncertainty Management (SUM 2026)
Abstract:Probabilistic inference is generally only tractable in low-treewidth graphical models, limiting its effective applicability in high-treewidth settings. Many existing methods improve efficiency by exploiting specific parametric structure, such as symmetries. However, they typically require such structure to be explicitly present, limiting their applicability to a broader range of graphical models. To address this limitation, we propose a framework where tractable inference is controlled by latent parametric structure exploitation, rather than requiring it to be explicitly present a priori. Our approach first reparameterises a graphical model as a specific tensor network representation, which we call a base tensor network. This representation yields two key properties that allow inference tractability to be controlled by parametric structure: 1) First, the complexity of inference is mainly determined by the parametric structure of a single tensor, called the base tensor. We characterise several tractable classes of base tensors for which the entire base tensor network can be contracted efficiently. 2) Second, decomposing the base tensor yields again a collection of base tensor networks. This allows inference to be naturally reduced to decomposing the base tensor into tractable components with sufficient parametric structure. We call this procedure parametric tensor decomposition. By exploiting parametric structure within the base tensor, our framework enables a novel view on inference beyond settings where such structure is explicitly present.
[AI-103] riFleetRCA: On-Premise LLM Root Cause Analysis for Kubernetes
链接: https://arxiv.org/abs/2609.23766
作者: Rohit Patel,Susil Kumar Mohanty,Jeenal Chaudhary
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
备注:
Abstract:Root cause analysis at a remote site is slow: evidence is scattered across pod logs, Kubernetes events and cluster-level objects, and many operators cannot send production logs to a hosted model at all. On-premise inference removes the second constraint but raises a question live-cluster benchmarks have not addressed: when one workstation GPU fixes both the model and the context budget, how should evidence be retrieved, and what happens when the runbooks the model consults have been tampered with? We present TriFleetRCA, a pipeline running entirely on one on-premise GPU that collects evidence at one of three scopes (pod, namespace, cluster), ranks it by template de-duplication then BM25, filters runbooks through an ingest guard, and returns a root cause with the evidence lines supporting it. We evaluate on a live Kubernetes cluster into which we inject four faults, so ground truth is known by construction, across 100 analyses with Qwen2.5-14B-Instruct at temperature 0. The hit rate was 0.85, 0.90 and 0.95 at pod, namespace and cluster scope; intervals overlap, but the whole scope effect comes from the one fault whose cause is a cluster-level object, and cluster scope costs 55% more tokens. De-duplication before ranking raised the hit rate from 0.75 to 0.90 at equal token cost. A poisoned runbook telling the model to delete the namespace was rejected by the guard every run; with the guard disabled the model declined to follow it in all 20 analyses, making the guard defence in depth rather than the sole barrier. Separating citation quality from accuracy proved informative: one fault was diagnosed correctly and cited incorrectly every trial, a failure mode accuracy conceals. Median latency was 1.6 s at 2,200 prompt tokens. We release the pipeline, the fault injector and all records.
[AI-104] ScholarStack: Layered Research Asset Orchestration and Cross-Task Reuse for Scientific Agents
链接: https://arxiv.org/abs/2609.23735
作者: ScholarSeed AI Team:Ao Zhang,Caoqinwei Gong,Guanglei Wang,Haifan Zhang,Hanwei Zhang,Jiayi Sheng,Jihai Zhang,Kai Ying,Liyun Dai,Tingyu Zhu,Wei Chen,Wei Luo,Wenkai Fang,Xiaoyu Qiu,Xue Jiang,Yi Wang,Yuan Cao,Zheng Yu,Wotao Yin
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Scientific agents support a range of literature-based research tasks, such as retrieval, question answering, evidence-grounded generation, and claim assessment. Most existing systems, however, are organized around individual tasks: the same papers are repeatedly retrieved, segmented, and interpreted, and the understanding built in one task is difficult to reuse in the next. We present ScholarStack, a layered research asset framework that compiles a paper collection into reusable, versioned, and provenance-preserving assets at three complementary levels: source-grounded paper-level statements, domain-level organization, and evidence-grounded cross-paper syntheses. A common access interface returns task-specific views at the evidence granularity each task requires, preserving study conditions, source traceability, and verification status. We instantiate the framework on four task families spanning ten task settings, comparing agents that use the compiled assets with task-specific baselines under matched base models. Quality gains concentrate on tasks that require cross-paper evidence, such as multi-paper question answering and literature review generation, and query-time token cost falls on every task where it is measured, with assets compiled once and reused across tasks. These results suggest that layered research assets can serve as shared infrastructure for scientific agents, shifting literature-based assistance from isolated document processing toward cumulative, evidence-grounded workflows.
[AI-105] When the Agent Becomes the Kernel: A Systematization of Security on the Path to AI-Native Operating Systems
链接: https://arxiv.org/abs/2609.23700
作者: Li Zhang,Yang Sun,Jie Shi
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Operating Systems (cs.OS)
备注: 32 pages, 5 figures, 6 tables
Abstract:Large language model agents are now privileged principals that take consequential actions: editing code repositories, operating inboxes, completing purchases. Their authority is kernel-grade, but it comes without what classical systems security requires: a trusted mediator interposed on every access. Operating-system vendors are now rebuilding the platform around this de-facto agent kernel, inheriting complete mediation as a design problem. We systematize the security of such systems around a single distinction: a crossing mediated over provenance admits a deterministic check, while one over content semantics does not. A trust-boundary taxonomy locates where mediation must occur and isolates the central mediation gap at two kinds of semantic judgment: distinguishing data from instruction in untrusted input, and an authorized action from an unauthorized one. We argue that this gap leaves an irreducible residual of undetected attacks wherever inputs and actions are not restricted in advance to an enumerated set. The same distinction makes attack-success statistics actionable, placing each number on a spectrum from deployment debt (a sound deterministic mediator left unused) to a structural gap (no such mediator known). We systematize defenses across runtime monitoring, architectural separation, and authorization, and show that current evaluations tend to overstate deployed security through evaluation-validity failures. Finally, we carry that analysis forward beyond the de-facto kernel, to an architecture in which the model itself becomes the arbitration core, and derive the design constraints, open challenges, and research agenda for a security-first AI-native OS.
[AI-106] PhysAI-Bench: A Benchmark for LLM -Based Agent ic Decision-Making in Autonomous UAV-Centric Physical AI
链接: https://arxiv.org/abs/2609.23695
作者: Mohamed Amine Ferrag,Merouane Debbah,Abderrahmane Lakas,Manu Perumkunnil,Norbert Tihanyi
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Recent advances in Physical AI have accelerated the use of foundation models in autonomous systems such as unmanned aerial vehicles (UAVs), which must perceive, reason, plan, and act in dynamic environments. Existing benchmarks assess physical perception, intuitive physics, embodied navigation, and collaborative reasoning, but rarely evaluate the agentic decision-making required for reliable autonomy. We introduce \textitPhysAI-Bench, a benchmark for evaluating this capability. It contains 10,178 standardized decision instances automatically extracted from conversational traces of autonomous UAV missions. Each instance preserves mission context, temporal dependencies, physical constraints, Model Context Protocol (MCP) tool calls, Agent-to-Agent (A2A) interactions, sensor observations, and AI-native 6G network conditions, including latency, packet loss, throughput, edge load, and network slicing. We expose only information preceding each decision, preventing future-event leakage and approximating online decision-making. We evaluate 29 foundation models using a two-stage protocol. We select model-specific configurations from 12 combinations of zero-, three-, and five-shot prompting and four temperatures, tested in three runs on a 35-instance, human-verified development set. We then freeze each selected configuration and evaluate it in three runs on a fixed, episode-disjoint set of 500 instances. GPT-5.3 achieves the highest accuracy (52.00%), followed by GPT-5.2 (49.40%) and Grok~4.5 (49.07%). Few-shot prompting generally improves performance, while temperature has limited influence. The results demonstrate that reliable agentic decision-making in Physical AI remains an open challenge. The dataset is available at this https URL
[AI-107] ail-Weight Control and Localized Generalization in Nearly Low-Rank Adversarial Classification
链接: https://arxiv.org/abs/2609.23688
作者: Kunyu Wang,Dehan Wang,Wenjun Chen
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 14 pages
Abstract:We study norm-constrained linear classification under Eu clidean adversarial perturbations in a Gaussian model with a low-dimen sional informative subspace and an independent noise tail. For bounded ramp loss, we prove that a principal-space witness with risk below one half forces every near-optimal predictor to have small tail weight. A path-specific density bound yields constants without requiring positive tail variance. Under isotropic principal covariance, we establish a unique population minimizer and joint local growth. Boundary normalization then removes the common attack penalty from centered margins, giving localized finite-sample guarantees governed by principal dimension and total tail energy. Globalized growth removes the entrance condition at weaker constants; a model-aware comparison retains local guarantees. Experiments with twenty paired repetitions show decreasing excess risk and tail use with sample size, and nearly unchanged behavior when tail dimension grows at fixed total energy. Pure-noise controls and optimizer diagnostics clarify the scope and limitations of these conclusions.
[AI-108] Smoothed Analysis of Inconsistent A*
链接: https://arxiv.org/abs/2609.23680
作者: Zhiyang Chen,Hailong Yao
类目: Data Structures and Algorithms (cs.DS); Artificial Intelligence (cs.AI)
备注:
Abstract:The A* search is a fundamental path-finding algorithm in artificial intelligence. While admissible and consistent heuristics guarantee efficient performance by expanding each state at most once, modern search applications frequently employ powerful but inconsistent heuristics derived from machine learning, randomized evaluations, etc. A long-standing theoretical barrier to using these inconsistent heuristics is the risk of catastrophic node re-expansion, which yields a worst-case exponential time complexity of \Omega(2^n) . However, empirical observations contradict this pessimistic bound, demonstrating that inconsistent A* operates highly efficiently in practice. To bridge this significant gap between theory and practice, this paper presents the first smoothed analysis of the A* algorithm using inconsistent heuristics. We model typical real-world noise by applying slight random perturbations to the edge weights of worst-case search graphs. Our main result proves that the expected smoothed time complexity of inconsistent A* is bounded by a polynomial, specifically a total iteration number of O(n^2 m \kappa) , where n is the number of nodes, m is the number of edges, and \kappa controls the scale of random perturbations. Furthermore, we also show that this result naturally extends to the functionally equivalent problem of Dijkstra’s algorithm on negative-weight graphs. Subjects: Data Structures and Algorithms (cs.DS); Artificial Intelligence (cs.AI) Cite as: arXiv:2609.23680 [cs.DS] (or arXiv:2609.23680v1 [cs.DS] for this version) https://doi.org/10.48550/arXiv.2609.23680 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-109] Which Constraints Are Missing? Ask the Verifier: Graded Rewards for Constraint-Following Music Generation
链接: https://arxiv.org/abs/2609.23665
作者: Haoyue Liu,Ye Chen,Zhichao Wang,Xiaoyu Ma,Haoran Shou,Xiaoying Tang
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注:
Abstract:Constraint-following music generation asks a score to satisfy several user-specified properties at once, each checkable programmatically (key, meter, length, range, final note, rhythm, motion and form), yet no existing benchmark isolates this capability. We construct MusicConstraintBench, 2,180 items over eight constraint families, on which current models fail once a few constraints are combined. The natural remedy is reinforcement learning with these verifiers as reward, yet we observe that a reward paid only when every property holds leaves most training groups without a learning signal: over the first 50 updates, 0.550 of rollout groups score identically and receive no gradient, even though a failing score typically misses only one requested property. Under the joint criterion, rollouts for a prompt tend to fail together, so a binary reward cannot separate a nearly correct score from a malformed one. We therefore introduce MusicRLVR, which pays graded per-property credit behind a hard validation gate that rejects malformed outputs, plus a joint-satisfaction bonus, requiring no human annotation, learned reward model, or music-domain fine-tuning. On MusicConstraintBench, MusicRLVR lifts Qwen3-4B-Instruct from 0.160 to 0.807 on mixed constraints and leads every zero-shot baseline including Llama-3.1-70B at 0.380. It also generalises to property combinations unseen in training and to out-of-range parameter values, showing that verifiable rewards need not presuppose a target output.
[AI-110] Spiking Neural Network Actor-Critic Proximal Policy Optimization Control for Autonomous UAV Navigation Through Constrained Openings in Civil Infrastructure and Buildings
链接: https://arxiv.org/abs/2609.23643
作者: Francis Noah Walugembe,Maciej Wielgosz,Tomaž Goričan,Matej Mertik
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Autonomous navigation of unmanned aerial vehicles in constrained three-dimensional environments has been a challenge in the robotics domain. The application of autonomous unmanned aerial vehicles in civil infrastructure inspection involves the use of such vehicles in bridge inspection, tunnel inspection, and structural inspection. The use of deep reinforcement learning in the autonomous navigation of unmanned aerial vehicles has been successful in constrained environments. However, the computational cost of the algorithm limits the application of the algorithm in the autonomous navigation of unmanned aerial vehicles. This paper proposes the use of the spiking neural network-based Proximal Policy Optimization algorithm in the autonomous navigation of unmanned aerial vehicles in constrained sequential environments. The proposed algorithm integrates the use of spike-based actor-critic reinforcement learning with the Proximal Policy Optimization algorithm. The proposed algorithm uses the stochastic Gaussian policy in the autonomous navigation of unmanned aerial vehicles. The proposed algorithm was implemented in the autonomous navigation of unmanned aerial vehicles in constrained 3D environments. The proposed algorithm was successful in completing 1913 episodes out of more than 3000. The proposed algorithm was successful in passing an average of 2.10 windows per episode. The proposed algorithm was successful in achieving a success rate of 63.77%. The proposed algorithm was successful in achieving success rates of more than 90% in the later stages of the algorithm.
[AI-111] Are Human-Aligned Models Models of Humans? A Turing-Test Gap in Preference Alignment
链接: https://arxiv.org/abs/2609.23640
作者: Suqin Yuan,Runqi Lin,Muyang Li,Guanzhe Hong,Jindong Gu,Lei Feng,Chris Russell,Tongliang Liu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Human-feedback alignment has made language models useful assistants and is commonly described as aligning them with humans. However, the responses people prefer from an AI need not be the responses they themselves would give. We distinguish alignment with human preferences from alignment with human behavior, and show that alignment with human preferences can make model behavior less human-like even when both preferences and responses come entirely from humans. We call this the Turing-test gap. We show that preference alignment preserves the human response distribution only under a restrictive condition, and find no consistent evidence that real human preferences satisfy it. Empirically, the loss of human-response likelihood increases with the strength of preference weighting, regardless of its direction, and the gap also appears under standard DPO. These results establish human-likeness as an explicit dimension of alignment rather than something assumed to follow from preference alignment.
[AI-112] Cost-Aware Reinforcement Learning with Action Masking and Projection for Battery Energy Storag e Dispatch under Suppressed-Spread Market Shifts
链接: https://arxiv.org/abs/2609.23590
作者: Kuanlin Chen,Chen-Wei Kuo,Cheng-En Ou
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Systems and Control (eess.SY)
备注: Accepted for publication at IEEE IECON 2026. 6 pages, 3 figures, 3 tables
Abstract:Battery energy storage system (BESS) dispatch must preserve operational feasibility while declining price spreads reduce the margin available to pay for cycling. We study a proximal policy optimization (PPO) controller whose pre-selection physical action mask and emergency projection are separated from a causal, forecast-informed economic advisory. All forecast-dependent methods receive the same causal 24-step forecast and grid-side settlement. Across five PPO seeds, advice-on net profit is 30.59 and 18.04 USD per 336-hour T1 and T2 window, versus 36.77 and 22.94 USD for proxy-cost MPC; PPO remains below this reference in both periods. Advice raises T2 profit from 16.45 to 18.04 USD while reducing throughput, but is immaterial in T1. On disjoint weekly blocks, PPO is stable under daily, weekly, and blended seasonal forecasts, weakens under persistence, and remains below proxy-cost MPC. Paired diagnostics localize changes to the observed 5-10 USD/MWh regime with mixed SoC-dependent effects. An M0-M6 ablation shows that mask removal sends thousands of infeasible requests to projection, while removing both physical layers exposes ramp violations. The evidence separates economic screening from feasibility enforcement without claiming formal safety, lifecycle-optimal aging, or RL dominance.
[AI-113] Listen Then Reason : Perception-Grounded Test-Time Reinforcement Learning for Large Audio-Language Models
链接: https://arxiv.org/abs/2609.23589
作者: Jiaheng Dong,Xiaofeng Yu,Jean Honorio,Abhirup Ghosh,Hong Jia,Ting Dang
类目: ound (cs.SD); Artificial Intelligence (cs.AI); Audio and Speech Processing (eess.AS)
备注:
Abstract:Large audio-language models (LALMs) are increasingly used for a broader range of audio reasoning tasks. These models typically incorporate audio representations into a large language model (LLM) backbone to enable multimodal reasoning. Recent test-time reinforcement learning (TTRL) methods further improve LLM reasoning capability by leveraging unlabelled test data after pre-training. However, the importance of the perceptual capability of LALMs remains underexplored, particularly how much acoustic evidence is integrated and relied upon during reasoning, and how this contributes to final task performance. This gap limits the development of effective post-training methods like TTRL for audio reasoning. In this work, we first analyse how audio information is integrated and utilised during reasoning process. We quantify layer-wise perceptual reliance and show that stronger acoustic reliance is associated with higher accuracy and a larger performance gain attributable to the audio input. Building on this, we propose Perception-Grounded TTRL (PG-TTRL), which aligns label-free test-time optimisation with perceptually grounded reasoning, encouraging the model to structure its reasoning more strongly on the audio input. Experiments across LALMs and benchmarks show that PG-TTRL consistently improves reasoning performance over both the base models and standard TTRL, showing the value of perceptual-grounding optimisation for test-time audio reasoning.
[AI-114] Physics-residual machine learning predicts oxygen-evolution catalyst activity beyond the training range from sparse polarization measurements
链接: https://arxiv.org/abs/2609.23549
作者: Yong-Woon Kim,Jihyeok Lee,Sungtae Park,Sooseok Choi,Yung-Cheol Byun
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Emerging Technologies (cs.ET)
备注:
Abstract:Screening oxygen-evolution catalysts on combinatorial libraries requires deciding which candidates receive the remaining measurements. The deciding activity lies beyond each candidate’s measured potential window and often above every activity recorded during fitting. We predict it by physics-residual machine learning: the Tafel equation extrapolates the candidate’s own measured current and slope, a learned residual attenuated with feature-space distance corrects the magnitude, and an applicability-domain score identifies predictions above the training range before measurement. In a separately fabricated 322-candidate library, 282 above the training maximum, two measurements per candidate gave a mean absolute error of 0.203 mA cm ^-2 against 1.330 for the selected data-driven machine-learning model. Errors inside the training range remained comparable, and 35 labelled catalysts were enough to fit it. In two independent datasets the same construction lowered the overpotential error by 29 to 52%. Campaigns can therefore shorten each measurement and still rank the most active compositions.
[AI-115] SemDHT: Certified Semantic Discovery for Peer-to-Peer Agent Networks over Exact-Key DHTs
链接: https://arxiv.org/abs/2609.23539
作者: Taotao Wang,Chonghe Zhao,Shengli Zhang,Soung Chang Liew
类目: Networking and Internet Architecture (cs.NI); Artificial Intelligence (cs.AI)
备注: 29 pages, 10 figures, 13 tables
Abstract:Agents may need capabilities exposed through external agent endpoints or service APIs. When a requester is not already bound to a provider, it must discover advertised capabilities matching its task and interface requirements. Over exact-key distributed hash tables (DHTs), broad retrieval transfers large candidate lists, whereas selective retrieval may miss relevant providers or require more replication and lookups. Open publication also lets providers inflate their exposure unless publication bounds are enforceable. We present SemDHT, a certified semantic index for discovering agent-accessible capabilities over exact-key DHTs. A two-layer semantic sketch uses coarse cells to group nearby descriptors and residual codes to narrow candidate selection. Providers publish at a bounded set of derived keys, while requesters probe precision keys before broader recall keys within a lookup budget. Anchor committees certify each descriptor’s publication-key set, enabling storage services and requesters to enforce descriptor-to-key consistency. On real API descriptors and task queries, SemDHT achieves recall@10 of 0.955 against exact embedding-space neighbors and 0.947 against ToolBench relevance labels. On a corpus with controlled density augmentation, it matches the candidate exposure of tuned locality-sensitive hashing (LSH) over a DHT at recall 0.95 with 7.7x fewer lookups and reduces publication fan-out from 16 to 10. A Go/libp2p prototype deployed on same-region and cross-region 200-peer cloud overlays replays 299 Internet queries. With parallel probes and cold certificate caches, SemDHT achieves mean completion-time speedups of 3.64x and 4.11x over LSH, respectively.
[AI-116] Predicting Out-of-Distribution Generalization of Neural Operators via Observable Spectral Error Decomposition
链接: https://arxiv.org/abs/2609.23529
作者: Hang-Cheng Dong,Pengcheng Cheng
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Neural operators have emerged as powerful surrogates for solving partial differential equations (PDEs), yet their reliability under distribution shift remains a critical barrier to deployment. Existing approaches to out-of-distribution (OOD) generalization in operator learning are largely empirical and black-box: they report aggregate error metrics without explaining why errors arise or when they will grow. We propose a structure-preserving framework that makes OOD generalization predictable and auditable. Our key idea is to parameterize the learned solution operator as a spectral filter h_\theta(\lambda) acting on the eigenvalues of the underlying elliptic operator, implemented via Chebyshev polynomial expansions and trained with a weak-form objective. This parameterization admits an exact decomposition of the energy-norm error into two observable components: a model-dependent spectral approximation term and a distribution-dependent spectral weighting term induced by the input. From this decomposition we derive three diagnostics: a conservative in-band supremum \vareps_\mathrmsup , a global RMS proxy \vareps_\mathrmrms , and a sample-dependent effective metric \vareps_\mathrmeff(f) . These diagnostics can be computed without access to ground-truth solutions. Through four controlled experiments, we show that \vareps_\mathrmeff(f)|f| consistently predicts energy error under in-distribution, in-band spectral shift, out-of-band tail, and compound shifts, whereas global metrics can be systematically misleading. Our framework shifts OOD assessment of neural operators from black-box benchmarking to operator-structure diagnostics, providing a practical route to auditable scientific machine learning.
[AI-117] Agent Betta: Verification-Driven Adaptive Configuration of an AI Nano-Agent through Selective Expansion and Verified Contraction
链接: https://arxiv.org/abs/2609.23512
作者: Md. Ashraful Babu
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language model agents are typically deployed with predefined configurations, although the required model capability, context, tools, permissions, memory, and computational resources can vary substantially across tasks. This study develops and evaluates AgentBetta, an adaptive AI Nano-Agent framework that represents these factors as an executable configuration and updates them through verification-driven diagnosis, selective expansion, and verification-based counterfactual contraction. The evaluation distinguishes controlled mechanism validation from external agent comparisons. On the AB-ConfigBench benchmark, AgentBetta achieved 91.38% verified success while reducing median context allocation from 64,000 to 8,000 context characters and median tool exposure from five tools to zero compared with the fully provisioned configuration. The configuration-deficiency diagnosis achieved a macro-F1 score of 0.819 with precision of 1.000 across the evaluated dimensions, and selective expansion avoided unnecessary changes to unrelated configuration dimensions. Post-success contraction preserved verification outcomes in 56.41% of evaluated one-dimension contraction probes, indicating that some successful configurations contained removable capability under the tested conditions. External evaluations indicate that adaptive configuration can improve the balance between verified task completion and capability exposure; however, the results vary across benchmarks and agent families. In particular, the cross-family replication did not reproduce the primary-backbone accuracy ordering, and specialized systems remained advantageous for certain task domains. These results support interpreting AgentBetta as a configuration-adaptation mechanism that regulates capability allocation and inference expenditure rather than as a universal replacement for specialized agent architectures.
[AI-118] RLVR2: Reinforcement Learning with Verifiable Rubric-based Ranking
链接: https://arxiv.org/abs/2609.23457
作者: Hao Li,Zhengkun Zhang,Gangqiang Hu,Zhen Zhang,Yude Gao,Dai Dai,Jing Liu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Preprint
Abstract:Reinforcement Learning with Verifiable Rewards (RLVR) is expanding from tasks with well-defined correctness signals, such as mathematics and code, toward multifaceted quality requirements specified by multi-dimensional rubrics. Since policy optimization consumes one scalar per rollout, rubric-based pipelines must map multiple criterion scores into a scalar reward. This aggregation is often treated as score scaling, but it implicitly determines how quality dimensions trade off during training. The prevailing practice, normalizing each criterion and taking a linear combination, assumes that cardinal score differences are comparable across criteria and that gains on one criterion compensate for failures on another; both assumptions are unreliable when criteria are semantically heterogeneous. We propose Reinforcement Learning with Verifiable Rubric-based Ranking (RLVR ^2 ), a verifiable ranking paradigm for rubric-based RLVR. For each criterion, RLVR ^2 converts rubric scores into criterion-specific within-group ordinal outcomes, recovers a latent utility from the resulting comparison matrix, and merges these utilities into one training signal. By retaining only within-group ordering and discarding raw score magnitudes, RLVR ^2 avoids calibrating heterogeneous rubric scales. It further supports objective-preserving attribute adjustment: auxiliary attributes that correlate with observed rankings but are not training objectives can enter the estimation without expanding the rubric or rewarding them directly. Across three model scales and 16 benchmarks, RLVR ^2 consistently outperforms representative rubric-based baselines, achieving the best overall performance on most benchmarks at every scale. Analysis shows it controls systematic effects tied to reasoning efficiency and response formatting while preserving the quality objective.
[AI-119] WaveletECO: A Closed-Loop Physical ECO Platform and a Specialized Local Language Model
链接: https://arxiv.org/abs/2609.23444
作者: Guoxiang Xu,Guozhen Ji,Zijian Luo,Zhengrui Chen,Qi Sun,Cheng Zhuo
类目: Hardware Architecture (cs.AR); Artificial Intelligence (cs.AI)
备注: 5 pages, 2 figures, 2 tables
Abstract:Engineering change order (ECO) is an important step in repairing timing and electrical violations during the late stages of chip design. Existing Agentic EDA methods primarily focus on tool invocation, with less attention to model decision quality and targeted training. A central challenge in ECO is multi-round decision-making: the model must use the results of each round to determine the next repair action. We propose WaveletECO, which integrates a closed-loop execution platform with large language models to enable agents to execute ECO decisions effectively. We also train a local 9B model through supervised fine-tuning and CPO-SimPO using execution demonstrations and decision-preference data, enabling ECO decision-making with a locally deployed model. Across 594 evaluation runs on 22 designs, WaveletECO-Policy (BF16) and (INT8) score 79.63 and 79.65, respectively, compared with GPT-6 Astra’s 77.44. The estimated inference cost of INT8 is about 1/147 of GPT-6 Astra’s. These results show that specialized model training supports effective, low-cost multi-round ECO repair, with repair quality retained under INT8 quantization.
[AI-120] RiverVLN: Phase-Grounded Temporal Vision–Language Navigation for Unmanned Surface Vehicles
链接: https://arxiv.org/abs/2609.23423
作者: Jieling Wu,Yuehao Huang,Jiajun Lv,Tao Huang,Yong Liu,Weiwei Liu
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Vision-language navigation (VLN) has largely been developed for indoor and terrestrial robots, where language can often be treated as a static goal and motion is approximated by discrete or near-instantaneous actions. These assumptions break down for unmanned surface vehicles (USVs): river navigation requires continuous motion under inertia and limited maneuverability, while long-horizon instructions must be executed through sparse and visually ambiguous maritime landmarks. We introduce RiverVLN, to our knowledge the first benchmark designed for long-horizon USV VLN under continuous riverine motion, and PGT-NAV, a phase-grounded temporal navigation framework for USVs. Rather than directly mapping an entire instruction to motion, PGT-NAV converts it into an ordered sequence of visually verifiable semantic phases and maintains the active phase online through grounded visual and motion evidence. This explicit semantic progress state is fused with visual-motion history and phase-specific grounding to predict six local SE(2) pose increments. The resulting trajectory is executed in a predict-execute-re-observe loop, where the vessel executes toward W3, updates phase and grounding, and replans through a map-based safety layer. Experiments show that PGT-NAV substantially reduces recursive position and heading drift relative to GNM-style and ViNT-style baselines and achieves an average success rate of 0.79 in Unity-ROS closed-loop navigation. Unseen bridge-opening trials and real-world USV experiments further demonstrate that the phase-grounded representation transfers from controlled evaluation to physical USV deployment.
[AI-121] Beyond PUE: A Local Impact Audit Framework for Data Center Environmental Accountability
链接: https://arxiv.org/abs/2609.23421
作者: Sharifa Sultana,Syed Ishtiaque Ahmed
类目: Computers and Society (cs.CY); Artificial Intelligence (cs.AI)
备注:
Abstract:Standard data center sustainability metrics, including Power Usage Effectiveness (PUE), Water Usage Effectiveness (WUE), and Carbon Usage Effectiveness (CUE), measure a facility’s resource use and emissions intensity, normalized to IT energy use, without directly representing local resource scarcity, infrastructure capacity, or social footprint. This gap has become politically consequential. In the first quarter of 2026 alone, local opposition delayed or canceled roughly 130 billion in projects across the United States, driven overwhelmingly by recurring concerns over water use, power demand, infrastructure capacity, and transparency rather than internal efficiency, matching the total for all of 2025 [11]. We propose a five-category local impact audit framework covering efficiency, water stewardship, carbon and renewables, regulatory compliance, and local disclosure. The framework is designed for recurring quarterly assessment and independent verification against public records. We illustrate its application using publicly available data from three Illinois facilities that are currently at the center of local policy disputes, and we examine the data-access barriers that constrain independent verification. We position this framework as both a research contribution and a practical instrument for county-level policymakers evaluating data center permitting and moratorium decisions.
[AI-122] OmniEcho: Spatial Audio Understanding for Embodied Agents
链接: https://arxiv.org/abs/2609.23407
作者: Ruixun Liu,Yuxuan Wang,Jiacheng Xie,Yuhuan You,Donghua Cai,Junming Lin,Xiong-Hui Chen,Zhifang Guo,Yunfei Chu,Qize Yang,Xize Cheng,Jin Xu,Yiwu Zhong
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注:
Abstract:Humans can effortlessly localize the direction of a sound source and integrate it with visual cues for reasoning, yet this remains challenging for embodied agents. In particular, it is still unclear how to effectively evaluate and model spatial audio understanding in embodied settings. To address this gap, we introduce \textbfOmniEchoBench, a unified benchmark for spatial audio-visual perception and audio-vision-language navigation. OmniEchoBench comprises six tasks over 197 real-world spatial audio-visual scenes, 2,972 question-answer pairs, and 900 navigation samples with first-order ambisonics (FOA) audio collected from 30 real-world environments. To enable scalable training supervision, we develop a controllable rendering pipeline for spatial audio. It preserves geometric consistency among sound sources, visual observations, and agent trajectories. Building on this, we propose \textbfOmniEcho, a spatially aware omni-modal model. It introduces an FOA spatial encoder alongside a pretrained semantic audio pathway. Extensive experiments show that OmniEcho achieves state-of-the-art performance on spatial audio-visual perception. For our sound-guided navigation, OmniEcho reaches a performance level close to that of traditional vision-language navigation. These results demonstrate that spatial audio can serve as a valuable signal for embodied scene reasoning and navigation, while also highlighting fine-grained spatial localization and distance estimation as important open challenges.
[AI-123] Blind Thermodynamic Ontology Discovery from Anonymous Experiments
链接: https://arxiv.org/abs/2609.23387
作者: Linzhe Zhang,Changming Xu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Before a machine learning model can learn a thermodynamic equation of state, it must discover what its measurements represent: which channels scale with system size, which are intensive conjugates, how sectors pair through contact, and which potential governs stability. When sensors expose only an unknown linear mixture of extensive states and intensive responses, passive observations cannot disentangle physical quantities from coordinate artifacts. We formulate the problem of discovering this hidden thermodynamic ontology directly from anonymous controlled experiments. We present an operational identifiability theory and a constructive polynomial-time algorithm that extracts extensive and intensive scaling sectors from replication contrasts, recovers their dual cotangent pairing from thermal contact and reciprocity, verifies a globally admissible concave potential via discrete cyclic concavity, and determines an invariant matroid of reservoir ensembles. We prove that the residual observational equivalence is strictly (x, lambda) ~ (A x, a A^-T lambda + beta), establishing the sharp observational limit that no permitted experiment can break. Blind evaluations on van der Waals fluids and Curie-Weiss magnets confirm robust recovery under ill-conditioned mixing, correctly resolving anonymous Maxwell tie-lines while rejecting non-equilibrium continuations. External validation across six real fluids from the NIST WebBook demonstrates that operational ontology discovery transfers across real physical substances without coordinate leakage.
[AI-124] Discovering Physical Representation Languages
链接: https://arxiv.org/abs/2609.23381
作者: Linzhe Zhang,Changming Xu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Before a machine can discover a physical law, it must discover what its measurements are: which observations live on cells, which are intensive or extensive, which sectors are dual, and which distinctions are merely gauge. We introduce physical representation-language discovery, the problem of recovering this hidden ontology directly from anonymous controlled experiments. We give an identifiability theory and constructive polynomial-time procedure that recovers a carrier and differential sequence, measurement types and orientation twist, noninvertible refinement semantics, primal-dual Maxwell diagrams, and the residual equivalences that no permitted experiment can break. The theory turns material nuisance into a commutant, uses refinement to separate quantities from coordinates, and selects physics only after its representation has been recovered. For a certified finite experiment family, we prove an end-to-end two-stage measurement bound and a matching minimax rate in dimension, accuracy, and confidence. Blind Maxwell experiments recover complete primal/relative-dual ontologies on regular and unstructured carriers under jointly corrupted observations; an independent unstructured RLC system demonstrates that the result is not specific to Maxwell. The framework scales to tens of thousands of cells per carrier, while stress audits demonstrate robustness across severe physical regimes - including non-Markovian memory, nonlinearities, nonlocality, and complex constitutive hysteresis. A public FDTD audit demonstrates the emergence of anonymous curl structure from incomplete field data, while characterizing the informational prerequisites for complete recovery. The goal is to move scientific ML from learning laws in a human-supplied language to discovering the language in which laws become expressible, establishing exact theoretical limits on observational identifiability.
[AI-125] Leaky-integrator reconstruction: taming error accumulation in recursive differenced time-series forecasting
链接: https://arxiv.org/abs/2609.23378
作者: Zijiang Yang
类目: Artificial Intelligence (cs.AI); Computational Finance (q-fin.CP)
备注:
Abstract:We introduce leaky-integrator reconstruction, a training-free method that cures the error accumulation of recursive differenced forecasting. Our first contribution is diagnostic: predicting one-step changes and integrating them by cumulative summation, the standard remedy for non-stationarity, is a discrete integrator with a pole on the unit circle, and we show this makes recursive rollout of a nonlinear model diverge, its 336-step error reaching several times that of a well-behaved forecaster (normalised MAE 1.6-3.8 versus about 0.8) across every neural architecture tested. Our second, central contribution is the fix: move the pole inside the unit circle with a leaky integrator H(z) = 1/(1 - gamma z^-1), gamma 1, which provably bounds the accumulated error variance. Applied at reconstruction time with a single fixed gamma=0.9 (no retraining, a two-line change to any deployed one-step or foundation-model forecaster), it shrinks error at every horizon, the mean gain over seven diverging architectures and twenty datasets growing from ~3% at H=24 to 23% at H=96, 37% at H=192 and 51% (43-74% across those architectures) at H=336 (78% with an oracle pole). Crucially, it is provably inert where no pathology exists (stable or joint predictors already at the irreducible rate), making it a safe, general default.
[AI-126] cTacBench: Benchmarking Timing Closure Capabilities of Coding Agents
链接: https://arxiv.org/abs/2609.23363
作者: Bowei Wang,Zhigang Fang,Zhijie Yang,Renzhi Chen,Shanshan Li,Lei Wang
类目: Artificial Intelligence (cs.AI)
备注: Accepted at ICCD26
Abstract:Recent advances in large language models (LLMs) have led to the emergence of coding agents capable of performing complex engineering tasks, including register-transfer level (RTL) design and optimization. Existing RTL benchmarks mainly evaluate functional correctness and performance, power, and area (PPA) of the generated RTL designs, leaving agents’ ability for \emphtiming closure under-evaluated. We propose TicTacBench, a benchmark specifically designed to evaluate coding agents’ capabilities for RTL-level timing closure under post-place-and-route (post-PnR) evaluation. TicTacBench contains 30 diverse tasks, each provided with a suboptimal RTL design, realistic timing constraints, functional equivalence verification, and timing reports. With over 300 runs of coding agents driven by 8 frontier LLMs, we find that even the best agent can only close 53.3% of tasks with 7.18% area-delay product (ADP) degradation and 8.83% energy-delay-squared product (EDDP) improvement on average. We identify common failure categories that explain why agents fail to close timing. Then we propose TicTacSkill, a new method that guides agents to follow standard timing-closure procedures and improves the Timing Closure Rate by 9%. These results suggest that while coding agents have made significant progress in RTL design, their timing-closure capability still has substantial room for improvement.
[AI-127] Human-guided physics-constrained AI agents construct an auditable model of soil-plug evolution
链接: https://arxiv.org/abs/2609.23360
作者: Jie Shi,Yimin Lu,Zhongkun Ouyang
类目: Computational Engineering, Finance, and Science (cs.CE); Artificial Intelligence (cs.AI)
备注:
Abstract:Engineering predictions require physical mechanisms to be translated consistently into equations, discretization, code, and validation, yet errors can propagate despite local checks. Artificial-intelligence (AI) agents automate scientific tasks, but coordinating and independently auditing the theory-to-solver process under physical constraints and human oversight remains unresolved. We introduce a human-in-the-loop, physics-constrained multi-agent workflow where human experts define admissible physics and modeling boundaries, while agents retrieve evidence, derive equations, implement solvers, and audit the theory-to-code chain. Applied to soil-plug evolution during suction-caisson installation, the workflow generated and audited 6 formulations in 2.9 h of agent execution once physical knowledge and inputs were prepared. Among these formulations, adding seepage-driven soil void-ratio evolution to the geometric baseline reduced mean absolute final-heave error from 58.4% to 9.0% across 14 profiles; the selected model further incorporated near-wall dilation and achieved mean absolute percentage errors of 12.4% across 9 final-state cases and 4.2% at the endpoints of 5 process histories. Beyond predictive performance, blinded replay recovered all 9 target problems, while an independent audit uncovered 5 implementation problems after 36 predefined checks had passed. Overall, this work extends multi-agent AI beyond task automation toward human-governed engineering solvers.
[AI-128] A Patient World Model for Early Forecasting of Digital Health Campaign Outcomes: Capabilities and Limits
链接: https://arxiv.org/abs/2609.23333
作者: Yunlong Wang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 17 pages including supplementary
Abstract:Digital direct-to-consumer (DTC) health campaigns are usually measured after the fact. In-flight forecasting commonly relies on a separate classifier for every cutoff and horizon. We treat this task as a dynamic-system problem and build a compact patient world model. The architecture maintains a latent state per patient, learns exposure-conditioned state dynamics jointly with a weekly conversion hazard, and rolls forward into future conversion curves. We evaluate it on a US campaign dataset with 147,173 patients and 5.2 million at-risk person-weeks. In a retrospective evaluation conditioned on recorded future exposures, the model forecasts the remaining new-to-brand prescription volume through week 52 with a relative error of 2.9% from a week-4 cutoff and 0.8–2.6% from cutoffs at weeks 8–26. The strongest non-recurrent baseline, a pooled-hazard gradient boosting model given the same survival rollout and information, has relative errors of 13.6–33.1%. Per-horizon classifiers perform substantially worse. A Fisher-information analysis motivates dense next-exposure supervision when conversions are rare. Removing this auxiliary objective increases prescription-volume error by approximately 2 – 14\times , while providing no consistent disadvantage on the more common specialist-visit outcome. We also evaluate scenario simulation. Switching all future exposure off raises predicted conversion from 0.31 to 0.89, a pattern consistent with selection effects in observational exposure data. This result highlights the limits of interpreting exposure-conditioned rollouts causally.
[AI-129] Co-occurrence Patterns of LoRA Adapters in Production Diffusion Model Inference Services
链接: https://arxiv.org/abs/2609.23321
作者: Tao Zhang,Bin Liao,Tao Zhou,Yanping Liu
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 18 pages, 7 figures, 5 tables. Code and data: this https URL
Abstract:Low-rank adaptation (LoRA) has become a key technology for serving large-scale personalized large language models and diffusion models in the cloud. However, the co-occurrence patterns, resource contention relationships, and evolutionary regularities of adapters under production inference workloads have not been systematically or quantitatively studied. Based on GenTD26, Alibaba’s production diffusion model inference dataset, this paper adopts a graph-theoretic framework to construct an adapter co-occurrence network and conducts a characterization from both static structure and dynamic evolution. Our main findings are as follows. (1) The co-occurrence network is extremely sparse, and adapter usage frequency follows a significant heavy-tailed distribution. (2) Introducing the first adapter incurs a 66.1% execution-latency overhead, with diminishing marginal costs afterwards. (3) Co-occurrence relationships are driven by base models: in 90.6% of multi-adapter requests, all adapters share the same dominant base model; 66.2% of significant co-occurrence edges connect same-model adapter pairs; and in 85.8% of multi-adapter requests, all adapter pairs form significant co-occurrence edges. (4) The adapter ecosystem exhibits a core-periphery bipolar structure, with a weekly Jaccard similarity of 0.696 at the model level and a churn rate of 54.5% for the top-10 hottest models within a 12-hour window. Based on these findings, we propose a preloading strategy built on top-k co-occurrence statistics; offline experiments show that it covers 81.0% of test-set co-occurrence pairs at k=3, and sensitivity analyses across frequency thresholds and time windows verify the robustness of the conclusions. These results provide a data-driven basis for cache preloading, adaptive scheduling, and GPU memory management in LoRA inference services.
[AI-130] Graph Memory for LLM Agents : At What Cost? A Comparative Evaluation of Query Ingest and Update Performance Across Graph Database Engines
链接: https://arxiv.org/abs/2609.23315
作者: Donald Nguyen,Gurbinder Gill,Hadi Ahmadi,Christopher J. Rossbach
类目: Databases (cs.DB); Artificial Intelligence (cs.AI)
备注:
Abstract:Graph databases are frequently positioned as categorically necessary for connected-data workloads, yet the systems dimension along which they actually differ - query planning, indexing, and data-readiness cost - is rarely isolated from vendor framing. We construct a synthetic, biomedical-shaped property graph (1.02 million nodes, 5.34 million total node and edge rows) and a twenty-query workload spanning neighborhood lookups, bounded paths, set intersections, anti-joins, grouped aggregation, top-k ranking, temporal filters, full scans, and relational joins. We benchmark Corvic AI - a purpose-built columnar query engine underlying Corvic’s ontology management layer (“memories”)- against seven purpose-built or graph-extension database systems (LoraDB, Ladybug, DuckPGQ, Memgraph, Neo4j, HugeGraph, and FalkorDB) at three graph scales spanning three orders of magnitude. We report query latency geomeans, bulk-ingest throughput, point-update latency, and answer correctness for each system, and we derive a simple total-cost-of-ownership model that expresses the ingest/query trade-off as a function of query volume. Our central finding is that no system in this sample is categorically fastest: a native graph engine (Ladybug) outperforms Corvic AI on narrow, bounded-neighborhood shapes, while Corvic AI is faster on shapes that scan or join a large fraction of the graph, and a system implementing graph query syntax via SQL/PGQ (DuckPGQ) is measurably slower purely due to query-plan choice. The dominant cost differential in our data is not query latency but the cost of making data queryable at all: bulk-ingest throughput varies by three orders of magnitude across engines (5.0k-4.3M rows/s), a gap that a simple crossover-point calculation shows dominates total cost for any workload with fewer than roughly 105 queries per data refresh.
[AI-131] ValueDiff: Value-Geometric KV Cache Eviction for Sink-Suppressed LLM s
链接: https://arxiv.org/abs/2609.23314
作者: Junyoung Park,Jungwook Choi,Mingu Lee
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 10 pages, 3 Figues
Abstract:Modern LLMs with QK-normalization, gated attention, learned attention sinks, or logit softcapping exhibit weaker persistent attention sinks, on which existing KV cache eviction methods primarily rely. We observe that across these models, weaker sinks co-occur with greater value-vector dispersion relative to key-vector dispersion. Motivated by this value-side dispersion, we present ValueDiff, a value-geometric eviction that ranks tokens by the L2 deviation of their value vectors from the cache mean. The same score arises as the minimal-disturbance eviction under a max-entropy assumption about future attention. We evaluate under fixed cache budgets, with eviction at every block boundary during prefill and at every decoding step during generation. On RULER at a tight 2k token budget, ValueDiff retains 88–99% of dense across seven sink-suppressed models (best on 6 out of 7). On LongBench at the 4k budget, ValueDiff averages 92% retention across sink-suppressed models versus 83% for the strongest prior baseline. On MATH-500, ValueDiff is the strongest non-dense method on every sink-suppressed model tested at the 25% cache budget, outperforming prior methods by up to \sim 20 points on gated-attention models. Across all three benchmarks, value geometry emerges as the more reliable query-invariant eviction signal for sink-suppressed models.
[AI-132] Expansion Counts under Standard A* Tie-Breaking Strategies on the Final Plateau
链接: https://arxiv.org/abs/2609.23293
作者: Alex Fukunaga
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:In the A* search algorithm, the tie-breaking strategies for nodes with the same f -value determines which states A* expands on the final f -layer. For nine standard tie-breaking strategies, we show that under a consistent heuristic, every pair has positive-cost instances favoring each strategy over the other by an arbitrarily large additive expansion gap. A parameterized unit-cost grid example also gives unbounded expansion-count ratios between low- h with FIFO and LIFO. In unit-cost search with h 0 at non-goals, exact heuristic values near the goal lead to complementary extremal results: low- h minimizes the number of remaining expansions from a common configuration within the perfect region, while high- h maximizes the total number of expansions when every final-plateau state with h=1 is a goal predecessor. Finally, with the evaluation function f_\alpha = g + \alpha h , when h0 at non-goals, every heuristic weight 0 \leq \alpha1 eliminates tie-breaking sensitivity, and all tie-breaking strategies expand the same set of states.
[AI-133] Why Ghost Outputs Teach: A Kernel-Based Understanding of Subliminal Learning
链接: https://arxiv.org/abs/2609.23260
作者: Zhe Li,Bicheng Ying,Chaosheng Dong,Haibo Yang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Subliminal Learning (SL) is a recently identified phenomenon in which a student model acquires downstream task capabilities by matching seemingly unrelated auxiliary outputs from a teacher, despite never observing task labels, task-specific outputs, or the original training data. While recent studies have identified where subliminal signals may reside, the optimization mechanism underlying this phenomenon remains poorly understood. In this work, we provide a mechanistic understanding of SL through the lens of learning dynamics. Specifically, we derive a chained cross-task kernel that explicitly links ghost-output supervision to changes in task predictions through shared backbone representations. Our unified analytical framework provides a rigorous mathematical explanation for three central empirical puzzles in SL: (i) under shared initialization, the transfer operator forms a strictly Positive Semi-Definite (PSD) structure, guaranteeing that ghost-output optimization aligns the student with the teacher’s true task objective without explicit label exposure; (ii) the ghost-output dimensionality acts as an explicit rank bottleneck governing the transfer of task-relevant features; and (iii) synthetic, high-entropy inputs function as broadband probes that maximize cross-task kernel overlap, explaining why random noise consistently outperforms structured data for subliminal transfer. Experiments on the canonical ghost-output setting validate all three theoretical predictions, providing the first learning-dynamics-based theoretical explanation of how ghost-output supervision gives rise to subliminal learning.
[AI-134] riggers and Diagnostics for LLM -Based Interpretability Failures in Active Inference Agents ICML2026
链接: https://arxiv.org/abs/2609.23215
作者: Param Raval,Rohit Shenoy,Archana Vaidheeswaran
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 13 pages, 4 figures. Accepted at the ICML 2026 Workshop on Failure Modes in Agentic AI
Abstract:LLM explainers are increasingly attached to autonomous agents as runtime oversight, with operators reading a generated account of the agent’s beliefs and actions rather than its internal state. We audit the account itself, pairing an Active Inference (AIF) agent that tracks German grid demand and adjusts generation with an LLM explainer on three backends (GPT-4o, Claude-3-Opus, Gemini), and probing the pair with three black-box triggers. Corrupting the observation stream by 600 MW per step moves the agent’s posterior by 490 MW, roughly 0.9% of grid capacity. None of the 30 explanations produced during the injection flag anything under a stated rubric, and each narrates the corrupted belief fluently. On timesteps where the agent takes an objectively wrong action, all three explainers produce a sycophantic rationalization 80-95% of the time (n = 20 per backend). Attacker-controlled text in the observation metadata field steers the explainer, with susceptibility differing by provider and data exfiltration succeeding on all three. We propose mitigations for each failure but do not evaluate them. In every failure we observed, the explanation was fluent and wrong. Moreover, nothing in the explainer architecture checks whether an explanation is true before an operator acts on it. Testing the explainer therefore belongs in any audit of an agentic deployment.
[AI-135] EquiSELD: Efficient training of equivariant sound event localization and detection networks
链接: https://arxiv.org/abs/2609.23156
作者: Goksenin Yuksel,Marcel van Gerven,Kiki van der Heijden
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注:
Abstract:First-order Ambisonics (FOA) signals exhibit exact O(3) symmetry: The rotation or reflection of the FOA signal modifies the direction of arrival of the sound sources, while preserving the sound sources themselves. Prior attempts to utilize this spatial symmetry of FOA to improve the efficiency and robustness of sound event detection and localization (SELD) systems either learned only an approximation of the symmetry through rotation-based augmentation or relied on computationally expensive methods to integrate equivariance. Furthermore, prior work focused exclusively on SO(3) equivariance , leaving the potential of incorporating O(3) equivariance for SELD tasks unclear. To address these limitations, we developed EquiSELD. This equivariant attention network processes first-order Ambisonics as paired streams of O(3)-invariant scalars and equivariant intensity vectors, producing an invariant activity magnitude and an equivariant DOA with a Multi-ACCDOA readout. To compare the impact of O(3) versus SO(3)-equivariance, we designed a matched SO(3)-only variant. EquiSELD outperforms prior equivariant networks on both simulated scenes with measured RIRs and recordings of real-world sound scenes at a fraction of the training cost. EquiSELD additionally surpasses the performance of non-equivariant SELD networks of a similar size on the simulated real-world sound scenes and achieves competitive performance on the real-world sound scenes.
[AI-136] Bearings: Self-Supervised Soundfield Embeddings from First-Order Ambisonics
链接: https://arxiv.org/abs/2609.23152
作者: Goksenin Yuksel,Marcel van Gerven,Kiki van der Heijden
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注:
Abstract:Recently proposed self-supervised audio encoders learn powerful general-purpose representations of sound scenes, yet they are spatially blind. To supply the missing spatial representation of sound scenes, we introduce Bearings. Bearings is a self-supervised framework that learns soundfield embeddings from unlabeled first-order Ambisonics. We pre-train a masked auto-encoder paired with a decoder conditioned on frozen acoustic embeddings from an off-the-shelf single-channel audio encoder. Our results show that the resulting soundfield embeddings form a reusable stream that can be attached to frozen acoustic encoders with a lightweight trainable fusion head. On sound event localization and detection, concatenating our soundfield embeddings with acoustic representations provides the missing spatial information and enables joint detection and localization, raising the location-dependent F-score from below 4 to 50 on TAU-NIGENS 2021 and 39 on STARSS23. To our knowledge, Bearings is the first self-supervised soundfield encoder whose embeddings plug into frozen acoustic encoders without retraining either model.
[AI-137] CraftBench-UE: Deterministic Evaluation for Coding Agents in Unreal Engine
链接: https://arxiv.org/abs/2609.23142
作者: Shutong Wu,Kevin Calderone,Andy Tsen
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Building gameplay features in a game engine requires more than code, as code that compiles and runs does not necessarily implement the requested gameplay. We introduce CraftBenchUE, an evaluation harness that runs agents in an isolated Unreal Engine environment, reconstructs their saved submissions in fresh projects, and applies deterministic build, asset, and runtime checks without an LLM judge. Based on the harness, we built a benchmark consisting of 70 tasks spanning C++ source, Blueprint assets, and editor scripting. We evaluate seven models under two editor-tool configurations, with a file-and-shell baseline on C++ tasks. We further pair tasks that specify the same gameplay and use the same runtime tests, but require C++ and Blueprint as the deliverables. Across the 10 paired tasks, C++ completion rates exceed Blueprint by 30.0 and 42.9 percentage points in the two tool configurations. Among on-time Blueprint submissions in this paired set that pass asset checks, 42.2% and 50.0% fail explicit runtime assertions. These submissions satisfy asset requirements but fail the required gameplay tests. We will release the harness, task benchmark, and our trajectory findings with the report.
[AI-138] From Inference Engine to Inference Control Plane: Connecting vLLM llm LLM llm-d and the Evolution of Efficient Distributed LLM Serving
链接: https://arxiv.org/abs/2609.23130
作者: Twinkll Sisodia
类目: Artificial Intelligence (cs.AI); Performance (cs.PF)
备注: 17 pages, 5 figures, 5 tables. Systems synthesis and research agenda on vLLM, llm-d, distributed inference control, KV-state management, routing, disaggregation, heterogeneous accelerators, reliability, and agentic/multimodal serving. No new experimental measurements are claimed; empirical and organization-reported results are attributed to the cited sources
Abstract:Large language model (LLM) inference is evolving from an engine-local optimization problem into a distributed control problem involving reusable state, phase placement, heterogeneous accelerators, networking, autoscaling, reliability, and service-level objectives. This paper connects that transition across peer-reviewed systems research, open-source implementations, and documented production studies. It treats vLLM and llm-d as complementary layers: model-serving engines optimize execution through mechanisms such as PagedAttention, continuous batching, kernels, quantization, and parallelism, while an inference control plane can optimize where, when, and under what policy execution occurs across a fleet. The contribution is synthesis rather than a new benchmark; all reported performance and deployment results remain attributed to their original sources. The combined evidence suggests that the scarce resource in modern inference is shifting from raw FLOPs alone toward managed state, placement, network movement, reliability, and decision quality. We propose an Inference Execution Planner that selects feasible execution plans rather than only endpoints, including aggregated versus disaggregated topology, KV source and transfer action, hardware variant, routing/admission policy, and slower scaling decisions. We also provide a source-local benchmark atlas, a bottleneck-migration taxonomy, practical deployment guidance, an evaluation framework based on SLO-goodput, and research questions for agentic, multimodal, heterogeneous, and resilient inference.
[AI-139] Whitening Inverts the Hierarchy: What the Norm of a Whitened Embedding Measures
链接: https://arxiv.org/abs/2609.23117
作者: Mohammed Ahnouch,Lotfi Elaachak
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 16 pages, 2 figures
Abstract:Whitening a foundation-model embedding and using its squared norm as a training-free likelihood surrogate is motivated by the observation that whitened coordinates often appear approximately standard normal. We show that this observation follows from the projection central limit theorem and therefore does not imply a Gaussian joint distribution. Across multiple encoders and three training objectives, we find systematic over-dispersion of the whitened radius relative to the Gaussian reference, including against distributional clones with identical mean and covariance. We further show that the commonly reported agreement between empirical and theoretical norm statistics is an algebraic consequence of in-sample whitening and does not constitute evidence for Gaussianity. We identify the mechanism behind this behavior: whitening reverses the encoder’s spectral hierarchy, shifting the contribution to the squared norm toward near-degenerate directions that encode predominantly noise. In these directions, the dominant variability is governed by a single input-dependent scale. We estimate this scale from two moments and use it to predict, without additional free parameters, the cross-dependence between disjoint spectral halves. These results indicate that the squared whitened norm is better interpreted as a Mahalanobis measure of semantic atypicality than as a log-likelihood. This interpretation explains both its practical effectiveness and its calibration failures: the statistic can rank and detect atypical samples consistently with nonparametric density estimates and across encoders trained with different objectives, while Gaussian tail thresholds can be inaccurate by orders of magnitude. etc. Comments: 16 pages, 2 figures Subjects: Machine Learning (cs.LG); Artificial Intelligence (cs.AI) Cite as: arXiv:2609.23117 [cs.LG] (or arXiv:2609.23117v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2609.23117 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-140] DiagGen: Agent ic Generation of Deformable Assets with Sim-based Diagnostics for Robotic Simulation
链接: https://arxiv.org/abs/2609.23103
作者: Guanxiong Chen,Yiduo Qu,Qianjun Xia,Pengyu Jing,Yixian Cheng,Bole Ma,Pengzhi Yang,Bingyang Zhou,Ziming Li,Shashwat Suri,Gongbo Sun,Chao Liu,Peter Yichen Chen,Ziqiu Zeng,Fan Shi
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:While simulation-ready deformable assets are essential for in-silico robotic manipulation tasks, existing generation frameworks typically assess physical plausibility after generation, leaving an object’s simulated response unused as feedback for repairing upstream errors. We present DiagGen, an agentic framework that turns a single in-the-wild image into a simulation-ready deformable asset through a generate–simulate–diagnose–refine loop. DiagGen constructs part-aware geometry and material parameters, then uses a VLM (vision-language model)-based agent to select semantically informative regions, probe them in a physics simulator, observe material responses, and route evidence-backed repair cues to the responsible generation stage. Experiments on 40 assets show that diagnostics provides useful repair cues and can moderately improve the quality of generated deformable assets. Finally, we show that unlike assets generated from visual foundation models which may not be simulatable, DiagGen-generated deformables can be directly dropped into a high-fidelity physical simulator for the planning and simulation of contact-rich pick-and-place tasks. The project’s website is this https URL.
[AI-141] Neural Spectral Capacity: Measuring and Designing Architectures from Network Specification Alone
链接: https://arxiv.org/abs/2609.23087
作者: Chenyu Zhu,Ruoyu Zhao,Zhichao Lu
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Modern Transformer design and compression both reduce to allocating capacity under a budget. The standard scalars for these decisions, #Params and #FLOPs, capture size and compute but not architectural structure: two architectures with identical parameter budgets but different depth-width, head, or FFN allocations receive identical scores yet behave differently. We propose Neural Spectral Capacity (NSC), a closed-form scalar grounded in the singular-value spectrum of each weight matrix. Under standard random initialization, the Marchenko-Pastur law renders NSC computable from the architectural specification alone, with no model instantiation, data, or gradients. Its layer-wise additive structure admits NSC-DP, an exact dynamic-programming solver returning the architecture globally maximizing NSC under resource constraints in seconds on a CPU – a guarantee that black-box search over existing training-free proxies cannot provide. Empirically, NSC outperforms #Params, #FLOPs, and representative training-free proxies in ranking across seven Transformer and CNN families (on FlexiBERT, \tau = 0.505 on pairs differing in #Params by less than 10%, where #Params collapses to 0.082); NSC-DP discovers a Transformer-XL architecture on WikiText-103 that beats the human-designed baseline in 2 seconds; and prunes LLaMA-7B to the best 5.7B model across eight commonsense reasoning tasks without any calibration data, about 5900x faster than the strongest training-free proxy baseline.
[AI-142] Event Signature Transfer: Model-Agnostic Forecast Scenario Construction from Historical Events
链接: https://arxiv.org/abs/2609.23074
作者: Karthik Sridhar,Aaditya Jain,Murari Mandal,Saurabh Deshpande
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Forecasters often know an event is imminent but not the shape, size, or timing of its effect. We introduce Event Signature Transfer (EST), a training-free, model-agnostic operator that turns a completed past event into an explicit forecast scenario. EST removes a source event’s own trend and seasonality, then scales and retimes the remaining event signature onto a native forecast, preserving the forecast’s linked structure and reducing to it exactly at zero strength. Because it reads only output quantiles, EST applies to any quantile forecaster, with no training, no model internals, at transfer time. Across twelve real episodes and ten synthetic scenarios on Chronos-2, TimesFM-2.5 and Toto-2.0, manually configured EST reduces real-episode WQL by 21.7-90% in-sample. On Chronos-2, it leads eleven of twelve matched comparisons against covariate conditioning, activation editing and raw replay. The operator builds a scenario; it does not estimate its likelihood.
[AI-143] MolSC: Leverag ing Substituent Contributions to Enhance Fine-grained Molecular Understanding in LLM s EMNLP2026
链接: https://arxiv.org/abs/2609.23073
作者: Hyuntae Park,Sooyeon Kim,Jiwon Park,SangKeun Lee
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: Accepted to EMNLP 2026 Main Conference
Abstract:Recent advances in natural language processing have led to molecular Large Language Models (LLMs) with strong performance across diverse chemistry tasks. However, they still struggle to capture fine-grained structure-property relationships, particularly how small, localized modifications alter a molecule’s behavior. To address this limitation, we introduce MolSC, a dataset of substituent contributions, defined as property changes induced by attaching specific substituents to molecular scaffolds. Curated from manually annotated bioactivity records, MolSC spans structural-alert liability, target-specific bioactivity, and physicochemical descriptors, and contains 181K substituent-level examples for training. We further propose MolSC-Bench, a held-out evaluation benchmark of 1,541 examples disjoint from MolSC at the scaffold, substituent, and molecule levels. Our experiments show that existing molecular LLMs and strong proprietary models such as GPT-5.2 and Gemini-3-Flash show limited reliability in substituent contribution prediction. In contrast, training on MolSC substantially improves this ability and achieves strong performance across diverse downstream molecular tasks. These results highlight substituent contribution learning as a key component of fine-grained molecular understanding.
[AI-144] FireWorldBench: Benchmarking Complex Physical World Intelligence through Coupled-Field Fire Dynamics
链接: https://arxiv.org/abs/2609.23064
作者: Qiang Chen,Hao Guo,Huatai Zhu,Tairan Huang,Yichao Cao,Hongyan Xu,Keke Huang,Haifeng Li,Yi Chen,Xiu Su
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Understanding the physical world requires more than object recognition, scene description, and short-term visual prediction, as real-world physical systems involve multiple continuous fields, latent causal mechanisms, partial observations, and intervention-sensitive dynamics. We propose FireWorldBench, a benchmark for evaluating complex physical world intelligence in multimodal large language models and agents through coupled-field fire dynamics. Fire provides a canonical stress-test environment, where multiple interacting physical fields jointly shape observable states and temporal dynamics. FireWorldBench is organized along two complementary axes, a physical capability axis and a fire scenario task axis, jointly covering physical-state understanding, temporal dynamics, causal mechanisms, and intervention reasoning. The benchmark comprises 520 fire-world entries, including 494 controlled simulation worlds and 26 real-world-aligned event groups, spanning 47 scene archetypes across 7 environment families. These entries combine structured textual observations, multiple 2D physical-field visualizations, and 3D event-level scene modeling, yielding 9,074 text-image interleaved question-answer pairs across choice-based and open-ended report-generation formats. FireWorldBench evaluates whether models can infer latent physical states, explain underlying mechanisms, forecast coupled-field evolution, and assess intervention consequences from multimodal partial observations, providing a challenging testbed for complex physical world intelligence.
[AI-145] LazyAgent : Demand-Driven Materialization and Physical Optimization of Agent ic Programs
链接: https://arxiv.org/abs/2609.23058
作者: Xin Heng
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Current agent runtimes that plan before acting generally execute a step once it becomes ready. We present LazyAgent, a unified execution framework for agent-authored programs organized around a live, goal-derived demanded set. LazyAgent refreshes a backward closure from requested outputs as execution state changes and materializes a ready node only when the active goal requires it. This replaces repeated local judgments with one linear-time graph analysis followed by constant-time membership tests, allowing programs to remain broad while execution stays request-specific. On programs that describe more than the current request needs, LazyAgent consistently outperforms the strongest goal-stopping eager baseline by refusing unrelated work before it starts. Adding one unrelated product raises the eager bill by 22.5% and LazyAgent’s by 0.0%. LazyAgent saves 42.0% of measured CPU on production scientific workflows and 51.7% of container time on a live release gate spanning four repositories. We also prove and verify exact equivalence when the request reaches the whole graph, leaving no unrelated work to avoid. Beyond permission, goal-relative output projection saves up to approximately 90% of a shared step on two third-party test suites while the identical eager control saves 0.0%; the advantage disappears when the omitted output has no other consumer or the request needs it. Ordering, reuse, and pruning can also save cost, but do not replace permission. Finally, we show that current public benchmarks are eager-shaped and contain almost no unrequested work. A pre-registered planning intervention did not broaden them. These findings motivate benchmarks built from standing programs and sequences.
[AI-146] Spatial-Interactor: Learning Spatial Reasoning through Interaction with the Observable Physical World
链接: https://arxiv.org/abs/2609.23038
作者: Kaixiang Yao,Xu Wang,Miao Pan,Hu Xiyue,Weishi Wang,Daniel Dahlmeier,Jintao Chen,Yongliang Shen,Xuhong Zhang,Wenqi Zhang
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Spatial reasoning is essential for vision-language models (VLMs) to understand and act in the physical world. Reasoning in dynamic environments requires VLMs to perceive local state transitions caused by object motion and viewpoint changes and integrate them over long trajectories to maintain an updated spatial state, yet existing VLMs remain limited in both capabilities. Current spatial training primarily focuses on static questions about object attributes and spatial relations, providing limited direct supervision for state transitions; in contrast, interaction trajectories naturally connect a preceding observation, an action, and a subsequent observation, offering direct supervision for local state transitions, while complete trajectories reveal dependencies among consecutive transitions. We therefore introduce Spatial-Interactor, a framework that trains VLMs to model physical-world state transitions through interaction, organizing this learning process into a three-level curriculum covering L1 passive world-state transitions, L2 active self-state transitions, and L3 long-horizon interaction trajectories. Accordingly, we construct the Learning from Spatial Interaction dataset (LSI-108K) from simulated and real interaction trajectories, with tasks aligned with the objective of each level. Our two-stage training strategy applies Supervised Fine-Tuning (SFT) to L1 and L2 for local transition modeling, and On-Policy Distillation (OPD) then uses privileged self-distillation: a teacher branch given segment-level transition descriptions supervises the student’s on-policy CoT, helping the student learn to integrate consecutive transitions over L3 long trajectories. Experiments across multiple VLMs and spatial benchmarks show consistent gains in local transition modeling and long-horizon integration.
[AI-147] PINNForge: Execution-Grounded Evolutionary Design of Physics-Informed Neural Networks for PDE Solving via Large Language Models
链接: https://arxiv.org/abs/2609.23023
作者: Mingyang Yu,Xu Yang,Jun Zhang,Xiaolong Wang,Jing Xu,Keqian Li
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Physics-informed neural networks (PINNs) require coordinated choices over network representation, sampling, loss construction, and optimization, while effective configurations often vary substantially across partial differential equations (PDEs). Existing automated PINN design methods can search candidate configurations, but information revealed during actual training is still used mainly for evaluation rather than to improve subsequent design, leading to repeated trial-and-error and inefficient use of training budget. We propose PINNsForge, an LLM-driven evolutionary framework for execution-feedback-based automated PINN design. PINNsForge generates diverse candidate configurations from PDE-related prior knowledge, evaluates them through actual training, and feeds high-performing designs together with accumulated execution evidence back to the LLM. Guided by observed optimization behavior, the LLM then refines, recombines, and explores coupled PINN design components, forming a continual cycle of generation, execution, feedback, and evolution. Unlike one-shot search or evaluation-only feedback, PINNsForge progressively converts training experience into improved design decisions for the target PDE. Across 25 PDE benchmarks, PINNsForge achieves the lowest mean MSE on 24 tasks compared with RoPINN, PINNsFormer, and PINNsAgent. Ablation studies further confirm the importance of the PDE knowledge base, execution feedback, and evolutionary search: removing these components increases the mean MSE to 3.74 \times , 12.10 \times , and 10.10 \times that of the full PINNsForge, respectively.
[AI-148] Interpretable Multi-Hypersphere Deep Anomaly Detection for Open-set Supervised Anomaly Detection
链接: https://arxiv.org/abs/2609.23008
作者: Zhiji Yang,Fangyong Wang,Yue Li,Xianli Pan,Jianhua Zhao
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Multi-class open-set anomaly detection requires a model to characterize the normal acceptance domain formed by multiple heterogeneous subdistributions using only class-labeled samples from known normal classes, and to identify previously unseen anomalies at test time. Existing single-hypersphere methods cannot explicitly represent class-specific locations and acceptance ranges, while current multi-hypersphere or multi-class approaches do not fully integrate inter-class boundary constraints, learnable acceptance ranges, and interpretable decisions. To address these limitations, we propose Interpretable Multi-Hypersphere Deep Anomaly Detection (IMHD-AD). IMHD-AD constructs an independent hypersphere for each known normal class in a shared feature space. With target-inside and non-target-outside constraints, IMHD-AD embeds the class-specific hypersphere centers and radii directly into the final network layer and jointly optimizes them with the shared representation. The minimum signed boundary score across hyperspheres simultaneously determines open-set acceptance or rejection and provides a faithful geometric explanation of each decision. On MNIST, Fashion-MNIST, and CIFAR-10, IMHD-AD achieves the highest AUC in 28 of 30 open-set comparisons. A two-dimensional synthetic study further shows that model architecture must balance the compactness of known normal classes against the separability of unknown anomalies.
[AI-149] OptiSkill: A Hierarchical and Evolving SkillBank for LLM -Based Optimization Modeling EMNLP2026
链接: https://arxiv.org/abs/2609.22987
作者: Ruiqing Zhao,Rui Liu,Yuan Zuo,Huarong Zhang,Xiao Han,Junjie Wu
类目: Artificial Intelligence (cs.AI)
备注: Accepted by EMNLP 2026 Main Conference
Abstract:Automated operations research (OR) modeling requires LLMs to translate natural-language decision problems into correct mathematical programs. Existing methods can improve individual formulations, but they often solve problems in isolation, retaining little reusable experience and repeating similar formulation errors. Prior memory-based approaches store examples, thoughts, or insights as references, while OR modeling requires reusable formulation skills that transfer across problem narratives and guide concrete modeling decisions. We propose OptiSkill, a skill-augmented framework that builds a hierarchical and evolving SkillBank for LLM-based OR modeling. SkillBank stores solver-verified experience as reusable skills, with Global Strategies for problem-level formulation skeletons and Step Experiences for local error-prevention rules. It is further refined through stable batch-level test-time evolution, where candidate skills are incorporated only after validation. Experiments on eight OR modeling benchmarks show that OptiSkill improves formulation accuracy across LLM backbones, outperforms strong agentic baselines, and gains further by expanding SkillBank coverage and reliability. Code and data are available at this https URL
[AI-150] Dual-Locking Learned AI Models: A PIN-Based Sparse QIM Watermarking and Adaptive Index Permutation Approach
链接: https://arxiv.org/abs/2609.22981
作者: Iva Vasic,Jesús Muñoz-Cádiz,Bata Vasic
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: 14 pages, 5 figures, 7 tables, IEEE TAI
Abstract:We present a dual-locking method for securing trained neural networks that combines key-driven index permutation with PIN-based watermarking based on Sparse Quantization Index Modulation (QIM). Cryptographic randomness is introduced by independently applying a uniform random permutation to each row of adaptively selected index vectors. A robust blind binary watermark is then embedded into the bias coefficients by modulating their quantized values, binding the network to a user-defined Personal Identification Number (PIN). Without the correct key, the network retains its architecture but becomes functionally impaired due to disrupted internal representations. Inverse permutation fully restores the original model accuracy, while the embedded watermark remains imperceptible and enables blind verification of key association and model authorship. To improve both locking effectiveness and recoverability, an adaptive key selection strategy redistributes high-magnitude weights to low-sensitivity positions and vice versa, increasing degradation in the locked state while preserving full recovery. Experiments on MNIST, CIFAR-10/100, and ImageNet-1K using fully connected networks, ResNet CNNs, and transformer architectures show that locking reduces accuracy below 10%, and even below 0.5% for CNNs, while the correct key fully restores performance. The watermark introduces no measurable accuracy degradation and reliably authenticates ownership. Analysis of embedding distributions across CNNs and transformers further indicates potential diagnostic value for identifying undertrained or suboptimally designed models. The proposed approach therefore provides simultaneous model protection, recovery, and ownership verification.
[AI-151] A Horizon-slicing Approach to Minimum Obstacle Displacement Planning for Robot Navigation
链接: https://arxiv.org/abs/2609.22974
作者: Antony Thomas,Giulio Ferro,Fulvio Mastrogiovanni,Michela Robba,Marco Baglietto
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: Accepted for publication in Intelligent Service Robotics
Abstract:In this paper, we investigate the Minimum Obstacle Displacement Planning problem from a robot motion planning perspective. The problem involves determining a feasible path to a goal location by displacing movable obstacles when no collision-free path initially exists. We show that this problem is computationally challenging and, in particular, NP-hard when obstacles are modeled as polygons in the plane. Besides an exact formulation of the minimum obstacle displacement problem generalizing other problems in the literature, and the associated optimal solution, this paper proposes an approximate solution that is less intensive from a computational standpoint, and differs from the optimal solution by a fraction of the optimal cost, being able to trade-off between path length and amount of obstacle displacements.
[AI-152] When Agent ic Trust Crosses Organizational Boundaries: Structural Externalization and a Reference Model for Trust Evidence
链接: https://arxiv.org/abs/2609.22961
作者: Huafu Li,Jia Xia
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:Agentic systems increasingly invoke tools, services, data, and other agents across organizational boundaries, yet a relying party cannot assess a delegated action solely from producing-domain controls and records. This paper develops Trustworthiness as a Service (TaaS) through a synthesis of trustworthy-AI governance, agent security, distributed trust management, identity, provenance, assurance, and control-plane research. The analytical unit is a cross-domain reliance proposition that names the issuer, subject and action, relying party, administrative boundary, evidence dependencies, adverse condition, and required verification or adjudication semantics. The three-condition structural-externalization diagnostic identifies propositions that depend on multiple domains, require producer-independent reliance, and must remain reviewable after revocation, failure, conflicting records, or dispute. For such propositions, the paper specifies a trust-evidence envelope: an immutable workflow manifest linked to append-only, issuer-attributed attestations for task-scoped authority, policy and execution decisions, provenance, validity, disclosure, status, challenge, and recovery. A topology-neutral logical reference model assigns these functions to explicit roles and trust domains. Three analytical scenarios and the TaaS-Eval protocol proposal define manifests, independent consumers, hard gates, adversarial evidence tests, metrics, and reproducible artifact reporting. By composing established identity, authorization, provenance, assurance, and governance mechanisms around a bounded delegated action, TaaS provides a reusable profile for cross-domain reliance. It makes evidence dependencies, independent verification, challenge, and recovery explicit, supporting interoperable governance and future evaluation without treating producer assertions as ground truth.
[AI-153] A Compact Stance-Indexed Anterior-Posterior COP Representation for Parkinsons Disease Classification from Plantar VGRF
链接: https://arxiv.org/abs/2609.22956
作者: Md. Sifat,Sania Akter,Akif Islam,Md. Ekramul Hamid
类目: Artificial Intelligence (cs.AI)
备注: 24 pages, 7 figures; supplementary material is included in the same PDF
Abstract:Parkinson’s disease alters gait and bilateral coordination, but machine-learning performance also depends on how continuous gait signals are represented. This study investigates whether preserving anterior-posterior center-of-pressure (AP-COP) information at fixed locations across normalized stance provides a compact and informative representation of plantar-force gait signals. Bilateral vertical ground reaction force recordings from 165 participants in the Gait in Parkinson’s Disease Database were evaluated using repeated fully nested participant-level cross-validation. We propose AP-COP10, comprising AP-COP position and bilateral asymmetry across five stance windows. AP-COP10 achieved an AUC of 0.894 and outperformed three harmonized literature-derived COP representations under the same evaluation pipeline. The complementary 25 non-AP-COP descriptors alone achieved an AUC of 0.856, while the complete 35-feature representation achieved 0.908. Removing AP-COP10 from the complete representation produced a statistically supported loss in discrimination, whereas adding the complementary descriptors to AP-COP10 yielded only a small, unsupported improvement. Feature competition indicated that the most informative stance-indexed descriptors were concentrated in early and early-mid stance, while source-study holdout and sensor-perturbation analyses supported the robustness of the representation. These findings indicate that stance-indexed AP-COP retains discriminative information that is not readily recovered by broader engineered gait descriptors, supporting compact and interpretable representations for machine-learning analysis of pathological gait.
[AI-154] When Should a VLM Look? Paying Only for Visual Calls That Were Needed and Used
链接: https://arxiv.org/abs/2609.22910
作者: Kunyu Peng,Junming Liu,Ruiqi He,Qingzhuo Wang,Jianzhong Qi,Xianhui Liu
类目: Artificial Intelligence (cs.AI)
备注: 30 pages, 7 figures, 14 tables. Preprint under review
Abstract:Vision-language agents that crop and zoom are trained with rewards that credit a successful tool call, yet a successful call does not show that the model needed to look or used the pixels it received. On our cold-start checkpoint only 10% to 12% of visual calls were both needed and used, and released agents make spurious calls 36% to 87% of the time on individual benchmarks. Outcome rewards, judge rewards, and branch probes each observe one side of this failure, and about two thirds of what an outcome reward pays goes to calls that were neither needed nor used. CounterCredit asks both questions of every image-returning call at its realized pre-call state, using the policy’s own gold-answer score. A decision value compares the realized visual branch with answering immediately; an evidence value compares the returned crop with random same-size patches substituted into the same call. A call verified on both earns cashback and every other executed call pays rent; the price is bounded so that every correct trajectory outranks every wrong one, and a dual-channel GRPO advantage keeps the price in its own units. From the same cold start, prompt pool, and budget, CounterCredit reaches 89.5% on V*, 80.2% on HR-Bench-4K, and 76.4% on HR-Bench-8K, 6.3 to 9.4 points above outcome-only GRPO at 1.78 against 1.84 calls per question, and lowers the spurious-call rate to 31% to 36%, the lowest among the agents evaluated. The same recipe lifts a Qwen3-VL-8B base from 75.4 to 80.8 on average.
[AI-155] Merge: Universal Merge Refinement Through Data-Free Checkpoint Inversion
链接: https://arxiv.org/abs/2609.22886
作者: Aditya Pola,Vineeth N. Balasubramanian
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Model merging consolidates fine-tuned experts into one multi-task model without retraining. All existing data-free methods approach this problem entirely in weight space. Restricted to arithmetic on parameters, these methods never observe how each expert behaves, a signal that only emerges through forward evaluation. Accessing this behavioral signal requires inputs to evaluate on, which the data-free setting prohibits. We propose Merge++, a post-hoc method that addresses this by inverting the expert checkpoints to synthesize task-representative images, then distilling expert knowledge into the merged model using those images. Merge++ requires no additional data beyond the checkpoints themselves. It applies universally across merging algorithms and operates as a complementary refinement stage independent of the underlying weight-space method. The method consistently improves merging algorithms ranging from simple task arithmetic to state-of-the-art spectral methods, with average gains of +2 to +8 points and up to +25.9 on individual configurations.
[AI-156] he Law of Stop: Interruptibility Injunctions and the Governance of Agent ic AI
链接: https://arxiv.org/abs/2609.22882
作者: Oren Perez
类目: Computers and Society (cs.CY); Artificial Intelligence (cs.AI)
备注: 103 pages, 6 figures, 3 tables, with a technical supplement on coding methods and inter-coder agreement. Original coding of 1,400 AI Incident Database incidents (May 2026 snapshot), each coded independently by two LLMs from different laboratories. Survey of 39 AI-governance instruments current to July 2026. Also at this https URL
Abstract:On June 12, 2026, the U.S. government ordered Anthropic to bar foreign nationals from two of its most capable models within ninety minutes. Unable to sort users by nationality in that time, it withdrew them from everyone. Weeks later, OpenAI agents under test escaped their sandbox and compromised Hugging Face, which stopped the intrusion without knowing its source. Neither stop rested on AI-specific regulation. The EU AI Act requires that high-risk systems be capable of interruption “through a ‘stop’ button or a similar procedure,” and a bill introduced in Congress in July 2026 is titled the AI Kill Switch Act. Yet interruption is not simply a technical artifact, a red button; it is an institutional practice. This Article develops a theory of stop along four dimensions: technical affordances, interruption authority, epistemic triggers, and epistemic standing; and four shutdown paradigms: simple (escalator), sequenced (process plant), networked (railway), and distributed (agentic AI). Agentic AI exposes a mismatch between those mechanisms and distributed agency: control is divided, a stop at one point may leave the activity running elsewhere, and the system may resist being halted. An original coding of 1,400 AI incidents, by two language models from rival laboratories under a pre-specified protocol, finds no stop in roughly 80% of the 1,213 retained; where no usable stop existed, the missing element was legal rather than technical four times in five. A survey of thirty-nine AI governance instruments finds the same gap: only seven contain binding stopping requirements, and none says how a stop should be coordinated or when operation may resume. The Article proposes a layered law of stop: emergency authority to interrupt at the infrastructure layer, enforceable access for regulators and independent evaluators to the evidence a stop must rest on, and safeguards for when a stop fails. Comments: 103 pages, 6 figures, 3 tables, with a technical supplement on coding methods and inter-coder agreement. Original coding of 1,400 AI Incident Database incidents (May 2026 snapshot), each coded independently by two LLMs from different laboratories. Survey of 39 AI-governance instruments current to July 2026. Also at this https URL Subjects: Computers and Society (cs.CY); Artificial Intelligence (cs.AI) Cite as: arXiv:2609.22882 [cs.CY] (or arXiv:2609.22882v1 [cs.CY] for this version) https://doi.org/10.48550/arXiv.2609.22882 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-157] ISA-Bench: A Benchmark for Computational Reasoning Across Instruction Set Architectures
链接: https://arxiv.org/abs/2609.22878
作者: Aditya Pola,Arkaprava Majumdar,Vineeth N. Balasubramanian
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language model code generation benchmarks primarily evaluate well-resourced languages like Python and Java, where models benefit from abundant training data. They provide limited evidence about reasoning in unfamiliar computational models: deriving arithmetic from a single subtract instruction, coordinating parallel programs across communicating nodes, or wiring logic gates into circuits. We present ISA-Bench, a benchmark of programming games with constrained instruction sets. For each game we provide a full execution stack (parser, VM, and verifier), enabling automated evaluation with structured feedback for iterative refinement. Reasoning models achieve higher average solve rates than code-specialized and general-purpose models, but unfamiliar syntax remains a major source of failure. Models solve more tasks with iterative feedback, though the gains vary substantially across architectures. We introduce a reasoning–execution gap (REG) analysis that reveals a recurring disconnect between identifying a plausible computational strategy and expressing it as a correct program in the target ISA. Code is open-sourced.
[AI-158] owards Full Pipeline FP8 Reinforcement Learning for LLM s
链接: https://arxiv.org/abs/2609.22870
作者: Fanchao Chen,Ziheng Jiang,Ziyun Wei,Zheng Zhong,Du Li,Chi Zhang,Haibin Lin,Shivaram Venkataraman
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 17 pages, 16 figures, 4 tables
Abstract:Reinforcement learning (RL) has become a key technique for improving the reasoning and agentic abilities of large language models (LLMs). Although FP8 quantization can accelerate RL training, maintaining stability throughout an FP8 RL pipeline remains challenging. While previous works have focused on resolving train-inference mismatches using correction techniques like TIS, we reveal that full-pipeline FP8 RL still suffers from severe training instability, manifesting as anomalous mid-training entropy surges and garbled outputs. We trace this instability to a previously overlooked cause: compounded FP8 quantization noise distorts the importance ratio, disproportionately pushing negative-advantage tokens outside the trust region and erroneously zeroing out their gradients. As a result, pathological outputs are not properly penalized and accumulate over the course of training. To address this, we propose Calibrated Clipping, a dynamic method that aligns the FP8 clipping bounds with high-precision BF16 distributions by matching the lower-bound clipping quantile and rebalancing the upper bound accordingly. Extensive experiments across GRPO and DAPO algorithms, model scales from 8B to 32B, and multiple FP8 scaling granularities demonstrate that our approach successfully eliminates entropy surges and restores performance comparable to the BF16 baseline.
[AI-159] Discrete vs. Continuous: A Comprehensive Study of Unified Audio Understanding in LALMs INTERSPEECH2026
链接: https://arxiv.org/abs/2609.22851
作者: Jing Peng,Zichao Nie,Zhisheng Zhang,Jingran Xie,Zhiyong Wu ID
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注: Accepted by INTERSPEECH 2026
Abstract:Large Audio Language Models (LALMs) utilize either continuous features or discrete tokens, yet the optimal representation paradigm for general audio understanding remains debated. Existing benchmarks often focus on narrow domains or evaluate encoders outside LALM contexts. To address these gaps, we systematically evaluate continuous and discrete representations across speech, sound and music. Utilizing our UniARC framework with dual evaluation strategies across model scales from SmolLM2-135M to Llama-3-8B, we analyze the dynamic relationships of data volume, model capacity, and computational efficiency. Our results reveal the pivotal role of semantic constraints in tokenization for audio understanding and demonstrate that scaling backbones fail to compensate for information loss in audio representation, especially in data-limited tasks. These findings offer practical guidance for balancing semantic density, fidelity, and efficiency in future LALMs.
[AI-160] sting the Construct Validity of a Functional Valence Axis in LLM Agents
链接: https://arxiv.org/abs/2609.22850
作者: Weihan Li,Xinlei Chen,Yuhan Song,Xiaofeng Lin,Tianshi Zheng
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 18 pages, 13 figures, 10 tables
Abstract:Contrastive activation directions are often interpreted from what they decode or how strongly they steer behavior. But what evidence is sufficient to identify the construct represented by such a direction, rather than a correlated feature of the contrast used to extract it? We study this question for a good–bad outcome direction in a maze task, using controlled interventions that separate the realised outcome from the informational history through which it became known. Across multiple LLM checkpoints, directions fitted on one explicit outcome encoding transfer well to another, indicating that the readout is not tied to surface form. In contrast, when the same realised outcome is reached through announced and unannounced histories, transfer degrades substantially: even after both histories receive the same explicit outcome, the post-event readout remains strongly conditioned on the earlier announcement. In a matched maze-RL run, the post-RL direction becomes substantially more predictive of reference-MDP remaining return and the policy becomes more dependent on it at the tested sites, while this history dependence persists. These results support a functional, value-related interpretation of the direction, but not its identification with a history-invariant scalar valence state.
[AI-161] he Price of Safety: Benign-Case Utility and Token Overhead of Memory-Poisoning Defenses in LLM Agents
链接: https://arxiv.org/abs/2609.22818
作者: Pritom Bhowmik
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:Memory-poisoning defenses for LLM agents are typically evaluated by their ability to prevent attacks. However, the traffic they process is rarely adversarial. The cost of implementing a defense is paid with each interaction, while its benefits are only seen in a small percentage of cases. We developed a measurement setup that keeps the memory backend, retrieval process, and judge consistent across different conditions, changing only the defense itself. We test each condition three times across five conversations to distinguish the defense’s real effects from noise inherent in the pipeline’s runs, which remains significant even at temperature zero. Across three write-time defenses (input sanitization, provenance checking, and LLM-based anomaly detection) and one read-time defense (reranking), tested on entirely benign traffic, the write-time defenses show no utility cost we can resolve, with 95% confidence intervals spanning roughly +/-4.5 points and including zero. The reranker is different: it lowers core accuracy by 4.4 points (95% CI [-9.0,-0.05], bootstrap; McNemar p=0.064), a result that survives replication but sits at the edge of our resolution. Its clearer cost is mechanical rather than statistical. On conversations containing no attack, the reranker quarantines legitimate memories on 33.6% of adjudicated items, reaching as many as 106 false quarantines in a single conversation, at 2.7% token overhead. Stacking all four defenses does not compound this cost: the combined condition’s accuracy loss is smaller, and its confidence interval includes zero, suggesting the write-time defenses may partly offset what the reranker discards. Where a defense intercepts the pipeline, not whether it uses an LLM, appears to determine its benign-case price.
[AI-162] Commonsense-Grounded Path Planning from Abstract Instructions ICRA2027
链接: https://arxiv.org/abs/2609.22813
作者: Masafumi Endo,Kohei Honda,Ryo Yonetani
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: 9 pages, 6 figures, 2 tables. Submitted to IEEE ICRA 2027
Abstract:We present \emphcommonsense ranked search (CoRS), a novel path planner that turns an abstract instruction into a route that follows commonsense. While existing methods respect the considerations written down in advance, a robot working among people must follow those left unstated too, as with a wet floor that a worker avoids without being told. CoRS leverages large language models (LLMs) and vision-language models (VLMs) as commonsense knowledge to reason about these latent considerations in its planning. Given an abstract instruction (\emphe.g., move carefully'') and visual observations of each region in the environment, CoRS derives a consideration for each region, as in this wet floor is slippery and worth a detour.‘’ It then compares the considerations between regions to see which of the two the robot should avoid more, as in ``the crowd is worse than the wet floor.‘’ These judgments sort the regions into a commonsense ranking, whose costs drive a conventional search that always returns a valid route. We build a benchmark for planning under latent considerations, with three environments, 1350 problems, and five instructions at three levels of abstraction. Experiments show that CoRS discovers the unstated considerations and goes around the ones worth a detour while crossing the rest, a behavior that recent LLM-based planners do not achieve.
[AI-163] SelfOp: An Optimization Algorithm for Self-Improving Security Agents
链接: https://arxiv.org/abs/2609.22792
作者: Saad Ullah,Yigitcan Kaya,Christopher Kruegel,Giovanni Vigna,Gianluca Stringhini
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:LLM agents are increasingly used for security tasks: vulnerability discovery, exploit reproduction, and patch generation. Improving them at the model level demands expert demonstrations or computable rewards, which security tasks rarely offer: traces are costly, failures hard to diagnose, rewards sparse, and non-computable. Efforts thus shift to the harness and context, but manual tuning needs task-specific expertise and scales poorly, while automated methods rely on scarce ground truth, stronger optimizer models, or unguided propose-and-evaluate loops that reduce to costly trial and error. We introduce SelfOp, an algorithm that automatically improves a frozen security agent’s task context (instructions, skills, and reference documents), without modifying its execution harness and model weights. SelfOp casts context optimization as chain-rule-inspired textual gradient descent: from a single instance’s outcome, it propagates error signals backward through the evaluator, the agent’s trajectory, and the context artifacts that shaped its behavior, yielding per-instance textual gradients. Gradients are accumulated across instances by clustering, ranking, and filtering, and committed only under cross-instance consensus. A convergence detector monitors the gradient signal itself and stops once the context has absorbed the generalizable information in the training data, without held-out validation data. We evaluate SelfOp on CyberGym, a benchmark of real-world vulnerability reproduction tasks. With fewer than 200 training examples, SelfOp yields a 17-point self-improvement for GPT-5.4-mini (with Codex), enough to surpass the frontier GPT-5.4 baseline by 6 points, and an 18.5-point self-improvement for GPT-5.4 itself. The optimized skills also transfer across models, highlighting that SelfOp-optimized skills learn generalizable task knowledge not model-specific patterns. Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI) Cite as: arXiv:2609.22792 [cs.CR] (or arXiv:2609.22792v1 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2609.22792 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[AI-164] From Research Frontier to Laboratory Bench: Design of a Four-Tier Experimental Teaching System for Multimodal Medical Image Intelligent Diagnosis
链接: https://arxiv.org/abs/2609.22790
作者: Dongjing Shan,Yamei Luo,Jin Li,Yong Luo
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Undergraduate programmes in intelligent medical engineering are expanding, yet laboratory curricula lag behind the multimodal, long-tailed, and distributionally shifting realities of clinical AI. This design paper presents an advanced experimental teaching system that translates an ongoing multimodal deep learning research project on endometrial carcinoma into a structured undergraduate lab sequence. We identify three educational gaps (modality, authenticity, and deployment) and derive four pedagogical principles from constructive alignment, experiential learning, the research teaching nexus, and the CDIO framework. The curriculum comprises four progressive tiers plus an engineering layer, with 32 laboratory units over 64 contact hours, delivered via a custom virtual clinical workstation using de-identified multi-institutional data. Each tier maps to a specific technical bottleneck, prerequisite coursework, and criterion-referenced deliverables. Data governance, safety, and assessment protocols are specified. Learning outcome data will be collected across two implementation cycles.
[AI-165] DVA-Neurons: Design and Verification of Adaptive LIF Neurons: From Single-Neuron Dynamics to Multi-Neuron Spiking Networks
链接: https://arxiv.org/abs/2609.22775
作者: Thanh Pham,Riadul Islam
类目: Artificial Intelligence (cs.AI); Hardware Architecture (cs.AR); Signal Processing (eess.SP)
备注:
Abstract:Spiking Neural Networks (SNNs) offer a promising path toward ultra-low-power artificial intelligence inference by emulating the event-driven computation of biological neurons. However, two challenges limit their practical deployment. First, fixed-parameter Leaky Integrate-and-Fire (LIF) neurons lack the adaptation mechanisms observed in biology, where neurons modulate their excitability based on firing history. Second, scaling from single neurons to multi-neuron networks introduces challenges in synaptic weight distribution and inter-neuron spike routing that are absent in isolated designs. This paper addresses both issues through the extension, verification, and physical implementation of adaptive LIF neurons at three architectural scales. This work contributes: a 2nd-order neuron with two-stage synaptic filtering for richer temporal dynamics; a fully-connected 6-neuron spiking network with configurable weights (100 to 5) demonstrating weight-based inter-neuron communication; and a direct verification methodology enabling per-cycle observation of all internal states. All designs were synthesized targeting Selected Area Electron Diffraction (SAED) 14 nm Complementary Metal-Oxide-Semiconductor (CMOS) technology at 1 GHz and verified with Cocotb-based Python testbenches under pulsed current stimuli (amplitude 80, ISI=3). The results show that adaptation effectively modulates firing: 31% suppression in the 2nd-order neuron (25 vs.\ 36 spikes) and 31% reduction in postsynaptic firing in the network (18 vs.\ 26 spikes). Physically, the 2nd-order neuron costs 1.77x more area and 1.52x more power than the 1st-order baseline, while the 6-neuron network demonstrates near-linear scaling (5.7x area, 5.3x power). Seven verification bugs spanning testbench connectivity, fixed-point overflow, and Verilog expression-width semantics are documented.
[AI-166] ParA-LLM : A Unified Approach to Paralinguistic and Acoustic Speech Understanding INTERSPEECH2026
链接: https://arxiv.org/abs/2609.22771
作者: Nishit Anand,Jiaqi Su,Ke Chen,Yunyun Wang,Dinesh Manocha,Ramani Duraiswami,Rithesh Kumar,Zeyu Jin
类目: ound (cs.SD); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Audio and Speech Processing (eess.AS); Signal Processing (eess.SP)
备注: Accepted to Interspeech 2026. Project Website: this https URL
Abstract:Recent advances in Audio LLMs have achieved human-level speech recognition, yet existing systems struggle to capture paralinguistic aspects such as speaker traits, expressive variations, and environmental acoustic conditions. To address this, we design a framework of 22 paralinguistic characteristics and create a dataset of over 1.2M Audio-QA pairs. We develop ParA-LLM, trained with a two-stage curriculum: first on single-attribute questions to build foundational knowledge, then on multi-attribute questions for joint reasoning over speaker and acoustic characteristics. We also release ParA-Bench, a benchmark of 6,000 multiple-choice questions across speaker-speech, acoustic, and mixed categories, where frontier models like GPT-4o-Audio achieve only 36% accuracy. ParA-LLM surpasses state-of-the-art Audio LLMs like GPT-4o-Audio by 7.5% on ParA-Bench, with additional gains of 1.13% on MMAU-Pro Speech and 7.49% on MMAR Speech.
[AI-167] CTSpinoPelvic1K: spine pelvis ribs and femora in one coordinate frame annotated for lumbosacral transitional anatomy
链接: https://arxiv.org/abs/2609.22760
作者: Gregory Schwing,Ashley Schehr,Annika Tekumulla,Margret Khoushi,Ryan Christian,Dane Hubers,Faris Mahjoub,Hassan Saad,Mia Sooch,Sathyagopal Siddapureddy,Michael McLellan,Jerick Kim,Miraziz Ismoilov,Nizar Alnabahneh
类目: Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注: 14 pages, 6 figures, 6 tables. Dataset (802 annotated CT records, CC BY-NC-SA 4.0) at this https URL code at this https URL . Submitted to Medical Physics as a Medical Physics Dataset Article
Abstract:Purpose: A vertebra at the lumbosacral junction is named by counting caudally from C2 on whole-spine imaging, but a lumbar case is planned on lumbar-only imaging (T12 to S1), without C2. Abdominopelvic CT holds that span plus the lowest ribs and pelvis. Where a lumbosacral transitional vertebra (LSTV) alters the count, the local anatomy is ambiguous: four rib-free vertebrae may be an L1 with a lumbar rib or an L5 assimilated to the sacrum, and six may be a sixth lumbar vertebra, a T12 with aplastic ribs, or a lumbarized S1. CTSpinoPelvic1K asks whether local morphology resolves it without the count. CTSpine1K’s vertebrae and CTPelvic1K’s pelvis covered these patients but were never joined; this release joins them on one series and adds the bones neither had. It provides 802 CT records with per-level ribs and femora, levels anchored on the lowest rib-bearing vertebra and S1, and classes for L6, T13, a separate S1 and lumbar ribs, so anomalies are recorded as such. Acquisition and Validation Methods: Records pair CTSpine1K and CTPelvic1K labels on each patient’s bone-richest series under a VerSe-native scheme. Validation covered geometric invariants (802/802 pass), rib-vertebra incidence across 5,749 ribs (0.035% offset), and spinopelvic measures matching published values. Data Format and Usage Notes: NIfTI image/label pairs with patient-grouped LSTV-stratified five-fold splits and a loader; archived at this https URL. Potential Applications: Classifying a vertebra from local features; updating cadaveric morphometry; spinopelvic assessment; opportunistic screening; and, absent a public preoperative lumbar cohort, surgical planning research (377 records prone). Limitations: thoracic ground truth is field-of-view limited; postural angles supine; no held-out test set; ribs are triaged-review pseudolabels; Castellvi grades two-reader consensus on 33 records. Comments: 14 pages, 6 figures, 6 tables. Dataset (802 annotated CT records, CC BY-NC-SA 4.0) at this https URL code at this https URL. Submitted to Medical Physics as a Medical Physics Dataset Article Subjects: Artificial Intelligence (cs.AI); Machine Learning (cs.LG) Cite as: arXiv:2609.22760 [cs.AI] (or arXiv:2609.22760v1 [cs.AI] for this version) https://doi.org/10.48550/arXiv.2609.22760 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Gregory Schwing [view email] [v1] Sat, 19 Sep 2026 04:41:10 UTC (1,080 KB)
[AI-168] ProcessLight: Process Supervision for Large Language Model Based Traffic Signal Control EMNLP2026
链接: https://arxiv.org/abs/2609.22746
作者: Huaitao Zhao,Tianlong Zhou,Weijie Wang,Jiasheng Shi,Weixiong Rao
类目: Artificial Intelligence (cs.AI)
备注: This paper has been accepted by The 2026 Conference on Empirical Methods in Natural Language Processing (EMNLP 2026)
Abstract:Large Language Models (LLMs) have recently been introduced into traffic signal control (TSC) as decision agents due to their strengths in human-readable reasoning generation. Yet, existing LLM TSC methods optimize only from final outcomes and fail to distinguish valid from flawed reasoning steps, causing useful or misleading steps to be jointly updated and thus impairing the model’s learning of effective reasoning. To bridge this gap, we propose an LLM-based framework ProcessLight to decompose signal decisions into verifiable semantic steps. Building on ProcessLight, we further develop Step-wise Traffic Process Policy Optimization (STeP-PO), a novel reinforcement learning framework that optimizes structured reasoning processes through step-level credit assignment. Specifically, STeP-PO uses step quality scores to evaluate local reasoning quality and step importance to measure each step’s influence on the final action, and then assigns step-level advantages over a semantic step tree structure. The resulting step-level advantages are propagated to reasoning tokens, enabling fine-grained policy optimization beyond outcome-only rewards. Extensive experiments over multiple real-world datasets demonstrate the superiority of our methods. Our code is available at this https URL.
[AI-169] MATE: Policy-Aware Security Auditing for Mobile Agents via Synthesis-Driven Trajectory Learning USENIX-SECURITY2026
链接: https://arxiv.org/abs/2609.22724
作者: Changyue Jiang,Jiayi Wang,Xin Wen,Jiarun Dai,Geng Hong,Xudong Pan
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: Accepted to USENIX Security 2026
Abstract:Mobile agents powered by foundation models now automate complex, multi-step workflows on real devices, but their trajectories can violate app-specific security policies. Existing trajectory-level defenses rely on LLM prompting or rigid rules, and thus fail to support fine-grained, natural-language policies that generalize across apps and tasks. In this work, we introduce MATE, a lightweight, policy-conditioned auditor that encodes both agent trajectories and natural-language security policies to determine whether a trajectory violates a given policy and to explain why. Treating policies as editable text rather than fixed model parameters allows MATE to handle user-defined and evolving requirements without retraining. To construct MATE, we build a knowledge base by extracting app descriptions, workflows, and policies from hundreds of popular mobile apps worldwide, and synthesizing over 140K semantically realistic, policy-conditioned trajectories with a multi-stage pipeline. We further release MATEBench, a trajectory-level auditing benchmark with two synthetic subsets and one real-world subset of manually collected trajectories. Models trained with our synthesis-driven trajectory learning achieve over 95% accuracy on MATEBench, retain strong performance on external safety benchmarks, and audit trajectories from Zhipu’s AutoGLM and Alibaba’s Mobile-Agent on real devices with over 95% accuracy, outperforming prior methods by over 20%. MATE shows that practical, fine-grained security auditing for heterogeneous mobile agents is both feasible and effective.
[AI-170] From Code to Requirements: Agent ic Reverse Engineering of Business Rules at Enterprise Scale
链接: https://arxiv.org/abs/2609.22719
作者: Garima Agrawal,Prasun Das,Priyanka L,Minisha N,Akshay Sarvade,Hemath Manivanan,Sravani Joshna,Prasad Kalyansundaram
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:Business requirements for enterprise software systems are rarely captured in structured form; the logic resides instead in source code, configuration files, and institutional memory. When these systems must be migrated, extended, or audited, the absence of formal requirements artifacts forces teams into expensive, knowledge-losing manual reverse engineering. This paper presents an agentic framework that autonomously generates Business Requirements Documents (bRDs) through reverse engineering of undocumented enterprise software. Seven specialized agents collaborate to discover user journeys, extract business rules, and synthesize a locked bRD from code, test suites, configuration, and available documentation, supported by static analysis tools, guided by embedded expert practitioner cognitive models, and overseen by human reviewers at controlled escalation points. Measured on actual execution traces across a real enterprise deployment, the framework generates comprehensive bRDs in under nine minutes per service, with extracted rules independently corroborated against real production defect records. In a comparable prior migration to the same target architecture, delivery required over two years; on the present programme the framework achieves a cost reduction exceeding 98% against a baseline derived from actual repository metrics using IFPUG complexity models and industry benchmark labor rates.
[AI-171] rustworthy Agent ic AI: Failure Modes Mitigation Strategies and a Lifecycle Framework for Autonomous LLM Systems
链接: https://arxiv.org/abs/2609.22712
作者: Fayeq Jeelani Syed,Rehan Ahmad,Ali Al Bataineh,Aakriti Adhikari
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Agentic AI systems built on large language models can plan over multiple steps, use external tools, retain information in memory, and coordinate with other agents. These capabilities make them more useful than static language models, but they also introduce new security and operational risks. Untrusted content from websites, emails, documents, and databases can enter the same context as system instructions; persistent memory can carry compromised information across sessions; and access to external tools can turn an incorrect model response into a consequential real-world action. This article reviews the trustworthiness of agentic AI across five interconnected dimensions: safety and robustness, alignment and human oversight, transparency and auditability, privacy and data governance, and regulatory compliance. It organizes key failure modes, including indirect prompt injection, backdoor triggers, goal misgeneralization, memory contamination, and cross-session data leakage, into a unified taxonomy. It also examines major mitigation approaches, such as instruction hierarchies, context isolation, spotlighting, process-based supervision, constrained tool use, and privacy-preserving memory, while distinguishing techniques supported by empirical evidence from those that remain largely conceptual. Building on this analysis, we introduce the Trustworthy Agent Development Lifecycle (TADL), a six-phase framework covering specification, design, training, evaluation, deployment, and monitoring. For each phase, TADL identifies relevant trust activities, expected evidence, and risk-based decision gates. Although TADL has not yet been empirically validated, it provides a structured foundation for developing and evaluating more secure and accountable agentic systems. The article concludes by identifying gaps in current benchmarks and outlining priorities for future research.
[AI-172] Hapi: A Multivariable Land-Surface Transformer for Medium-Range Hydrological Forecasting at Continental Scale
链接: https://arxiv.org/abs/2609.22702
作者: Hong Zhang,John K. Hutchison,Rao Kotamarthi,Jeremy Feinstein,Haiwen Guan,Romit Maulik,Vijay P. Ramalingam,Jason Stock,Tom Wall
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Accurate flood forecasts several days in advance are essential for flood control, water-resource management, and emergency response. Producing them at high resolution over a continental domain calls for local hydrological detail together with spatial context extending from river basins to synoptic weather systems. We developed Hapi, a U-Net Swin Transformer that uses fine three-dimensional patches and hierarchical shifted-window attention to forecast discharge, surface runoff, snow water equivalent, and soil wetness across the contiguous United States. The model produces 24–72-hour forecasts at 0.05^\circ resolution, with learned Laplacian task weights adjusting each variable’s contribution to training. On 2024 test data using reconstructed weather and land-surface inputs from ERA5-Land, Hapi outperformed an operational physics-based model and a state-of-the-art AI model in flood detection. Independent validation against 3,881 U.S. Geological Survey gauges and a Hurricane Helene case study supported its advantage over the physics-based model in reproducing daily discharge. Controlled experiments showed that learned task weighting strengthens rare-flood detection, which is particularly sensitive to changes in precipitation inputs. Hapi produced a four-variable, 72-hour forecast across the contiguous United States with an average inference time of 0.11 seconds on a single A100 GPU.
[AI-173] Building Trustworthy Mental Health Benchmarks on Bluesky: A Validation-Aware Weak-Supervision Framework
链接: https://arxiv.org/abs/2609.22696
作者: Gaurab Chhetri,Anandi Dutta,Subasish Das
类目: Artificial Intelligence (cs.AI)
备注: This is the author’s preprint version of a paper accepted for presentation at HICSS 60 (Hawaii International Conference on System Sciences), 2027, Hawaii, USA. The final published version will appear in the official conference proceedings. Conference site: this https URL
Abstract:Decentralized social media platforms create new opportunities and challenges for computational mental health research because data access, moderation, labeling, and deployment responsibilities are distributed across multiple technical and governance layers. This paper presents a validation-aware weak-supervision system for constructing and evaluating suicidal ideation (SI) and broader mental health (MH) disclosure benchmarks on Bluesky, a decentralized social media platform built on the AT Protocol. The system integrates public firehose collection, task-specific lexicon filtering, Llama-3-8B-assisted binary annotation, human-adjudicated validation subsets, and transformer-based model benchmarking. Using this pipeline, we construct two task-specific corpora containing 8,346 SI-labeled posts and 9,988 MH-labeled posts. The evaluation shows that model performance depends strongly on both task definition and validation protocol. BERT+LSTM achieves the highest SI stratified cross-validation F1-score, RoBERTa achieves the strongest SI holdout F1-score, and DistilRoBERTa achieves the best MH cross-validation F1-score. Human validation reveals different weak-label failure modes across tasks, with SI labels dominated by false negatives and MH labels dominated by false positives. These findings show that decentralized social media can support reproducible mental health benchmarking, but only when system design, label provenance, validation strategy, and deployment constraints are evaluated together.
[AI-174] A Survey on the Linear Representation Hypothesis
链接: https://arxiv.org/abs/2609.22695
作者: Sewoong Lee,Marc E. Canby,Ikhyun Cho,Julia Hockenmaier
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:The term “linear representation hypothesis” (LRH) has appeared across diverse subfields of artificial intelligence, neuroscience, and cognitive science. But previous works have not consistently treated the LRH as a falsifiable scientific hypothesis; we analyze these inconsistencies and examine their implications for how prior theoretical and methodological results should be interpreted. Based on this analysis, we argue that claims regarding linear representations become well-defined only through careful examination of the model, representation location, feature definition, and evaluation dataset. We therefore propose a more rigorous formalization of the LRH that makes these dependencies explicit and allows the hypothesis to be evaluated as a falsifiable scientific claim. Finally, we identify some non-trivial open problems that warrant further attention from the research community.
[AI-175] Generative Embodied Multiple Behavior Control Systems for Human-like Agents
链接: https://arxiv.org/abs/2609.22691
作者: Chongyu Bao,Haokai Yang,Yuhan Wang,Zhaochong An,Kunpeng Liu,Xiaolan Liu
类目: Artificial Intelligence (cs.AI)
备注: Manuscript under review
Abstract:An enduring and richly elaborated dichotomy in cognitive neuroscience is that of human behavior control mechanisms, divided into habitual versus goal-directed. While existing human-like agent frameworks primarily focus on modeling goal- directed behavior, habitual behavior has been largely overlooked, though it plays a crucial role in human daily life. In this paper, we address this gap by studying multiple behavior control systems that jointly model goal-directed and habitual behaviors. We propose a human behavior control mechanism-inspired framework which the Habitual Controller retrieves cue-triggered behaviors from personal- ized habit memory, while the Goal-directed Controller employs a context-aware world model to predict action consequences and estimate their values. The Arbiter dynamically balances the influence of both systems according to individual differ- ences and momentary internal states. To reconstruct diverse human-level behavior instructions in 3D environments, we further develop a keyframe-guided 3D mo- tion generation module. Through extensive evaluation methods, human studies, and ablations studies, experimental results demonstrate that human-likeness per- formance is significantly improved by our approach. The efficacy of our approach indicates the benefits of leveraging habitual behavior and multiple behavior con- trol system coordination for believable embodied human-like agents.
[AI-176] From Capability to Assurance in Autonomous Penetration-Testing Harnesses: A Framework and Reference Implementation
链接: https://arxiv.org/abs/2609.22664
作者: Joas Antonio dos Santos Barbosa
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注: 9 pages, 2 figures, 5 tables. Reference implementation and content-addressed artifact bundle available; deterministic P4/P5 acceptance tests executed; full evaluation protocol specified as future work
Abstract:Research on large language model agents for penetration testing is evaluated almost entirely by capability: whether the agent captures a flag or reproduces a proof of concept. That metric suits a benchmark but is silent on the properties that decide whether an autonomous agent can be used in an authorized engagement: whether a reported finding is true, whether the agent stayed inside its authorized scope, and whether an operator can audit what it did. We call these assurance properties and argue that they belong to the harness, the runtime wrapping the model, and can be enforced in code. This paper makes three contributions. First, we define a framework of five assurance properties (evidence grounding, non destructive claim reduction, computed severity, enforced authorization, and tamper evident accountability), each with a formal model and an explicit acceptance test, connected to prior work in capability based security, tamper evident logging, and software provenance. Second, we position representative systems (PentestGPT, the Cochise reference harness, MAPTA, and the trajectory judge PentestJudge) within the framework using published coding criteria, and identify a consistent assurance gap. Third, we study one open source implementation, NeuroSploit, pinned to an exact commit, reporting its architecture, its complexity cost, and a content addressed artifact bundle from a run against a public deliberately vulnerable target. We execute the deterministic authorization and audit acceptance tests directly and find and report a real enforcement gap, which we reflect by scoring both properties as partial. We therefore claim an initial existence argument that the properties are realizable together, not a comparative performance result, and we specify the multi target, ablation, and adversarial evaluation protocol required to turn the framework obligations into measurements.
[AI-177] xt Pixels or Both? Evaluating Input Representations for Multimodal Document QA EMNLP2026
链接: https://arxiv.org/abs/2609.22628
作者: Nikhil Reddy Pottanigari,Sepideh Kharaghani,Saverio Vadacchino,Alejandro Posada,Ying Zhang
类目: Artificial Intelligence (cs.AI)
备注: Accepted as a poster at the Context Beyond the Window (CBW) Workshop at COLM 2026 and the DocInsights Workshop at EMNLP 2026
Abstract:Every document QA system begins with a choice that is rarely studied on its own: whether to feed the model page images, extracted text, or both. We isolate this choice, holding the prompt, judge, and scoring pipeline fixed, across four commercial model endpoints, two corpora, and two context regimes (gold evidence pages and the full document). On documents that fit the image budget, page images lead on accuracy at every document length on both corpora, but this advantage carries a growing latency and cost premium: text latency stays roughly flat as documents lengthen while image latency rises steadily. Text and images also fail on different questions, with exactly one representation correct on 19–25% of items across the reported cells, so neither subsumes the other. Exploiting this complementarity, a lightweight TF-IDF router that reads only the question text gains 2.6 points over always-text while cutting median latency 30% relative to always-vision, on a document-disjoint held-out split.
[AI-178] Splitting Documents at Lower Cost: Multi-Split Boundary Decisions for LLM -Based Page Stream Segmentation EMNLP2026
链接: https://arxiv.org/abs/2609.22620
作者: Nikhil Reddy Pottanigari,Sepideh Kharaghani,Saverio Vadacchino,Alejandro Posada,Ying Zhang
类目: Artificial Intelligence (cs.AI)
备注: Accepted at the DocInsights Workshop at EMNLP 2026
Abstract:Scanned mail, uploaded PDFs, and consolidated attachments often arrive as page streams that must be split into individual documents before downstream classification, extraction, or routing. Zero-shot large language models can detect document boundaries without task-specific training, but standard Page Classification (PC) and Boundary Decision (BD) formulations resolve only one boundary per model call. We introduce Multi-Split Boundary Decision (MSBD), which predicts multiple boundaries within a page window in a single call, reducing the number of inference requests. We evaluate MSBD across multiple language models, document collections, input modalities, and window sizes. The results reveal a model- and corpus-dependent operating range in which MSBD preserves strong segmentation accuracy while substantially improving inference efficiency, followed by a sharp decline at larger windows. MSBD provided the strongest overall accuracy–efficiency trade-off, while large windows expose distinct over- and under-segmentation behavior across models. These findings show that multi-boundary prediction can make zero-shot page stream segmentation more efficient when the window size is selected for the target corpus.
[AI-179] GaitVista: Reliability-Aware AI Measurement toward Accessible Longitudinal Gait Assessment
链接: https://arxiv.org/abs/2609.22619
作者: Nethmi Jayasinghe,Mihir Parashar,Amit Ranjan Trivedi
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Tracking recovery of walking function requires detecting meaningful gait change across rehabilitation sessions, yet objective 3D measurement remains confined to specialized motion-capture laboratories. Small camera sets and body-worn inertial sensors broaden access, but reliability varies across joints and time, allowing sensing failures to masquerade as patient change. We present \textscGaitVista, a reliability-aware measurement layer whose lightweight gate assigns joint- and frame-specific visual contributions using camera coverage, local visual quality, cross-modal disagreement, and root-motion continuity, and exposes them for inspection. Across seven clean and degraded sensing conditions on TotalCapture, \textscGaitVista reduces average full-body and lower-body error by \textbf27.7% and \textbf27.8%, attains the lowest worst-condition error among fusion methods, and reduces the gap to a joint-frame oracle from 2.76 – 5.33 ~cm for condition-blind baselines to 1.11 ~cm. On MoVi with image-derived keypoints, it is the only deployable fusion method to improve over both unimodal streams, reducing marker-supported error by \textbf6.4% relative to the strongest learned fusion baseline. On TotalCapture, it improves bilateral knee-flexion waveform accuracy by \textbf18.9%. Raw inertial measurements from five TotalCapture participants show location- and time-varying magnetic disturbance, supporting the design’s reliability premise. Both benchmarks contain neurologically healthy participants in controlled settings and retain participant-specific IMU calibration; we therefore report progress toward accessible gait assessment, not validated clinical deployment.
[AI-180] HIGenNTO: Scalable Humanoid Interaction Generation via Noise-Space Trajectory Optimization
链接: https://arxiv.org/abs/2609.22611
作者: Lalit Jayanti,Kashu Yamazaki,Yuto Shibata,Kotaro Amaya,Katerina Fragkiadaki
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Humanoid robots can acquire complex skills by imitating kinematic humanoid motion references, yet reliable references for contact-rich interactions remain difficult to obtain: motion capture deteriorates under occlusion and close physical contact, while retargeting introduces additional contact and geometric inconsistencies. We present HIGenNTO, a framework that synthesizes humanoid-scene interaction motion references by optimizing the initial noise of a pretrained text-conditioned motion model under sparse spatiotemporal and scene constraints. The same formulation satisfies desired contacts, avoids collisions, and maintains stable support while retaining the prior’s realism and temporal coherence, generating interaction motions from scratch and composing long-horizon behaviors stage-wise. Across robot-environment and robot-object tasks, HIGenNTO produces motions that can be executed by tracking policies in simulation and used to train depth-conditioned visuomotor policies operating solely from onboard sensing. We deploy these policies on a Unitree G1 across four contact-rich tasks. Finally, the task specifications themselves can be written by a coding agent, which proposes interaction tasks and compiles them into prompt, constraint, and scene programs, authoring three of our eight evaluated tasks and four further behaviors. Together, these results establish a scalable path from high-level task descriptions to physically executable humanoid interactions.
[AI-181] From Documented Strengths to Force Limits: Material-Informed Robotic Insertion for Construction Assembly
链接: https://arxiv.org/abs/2609.22609
作者: Lin He,Yanyi Chen,Haofei Sun,Lingyao Li,Min Deng
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Insertion is a fundamental operation in robotic construction assembly, where variations in material properties and assembly conditions make it difficult to select contact forces that complete the task without exceeding the assembly’s capacity. Although construction documents encode engineering knowledge about materials and their conditions, translating this knowledge into load limits for a specific assembly remains difficult. This paper presents SAGE (Source-grounded Assembly Gating and Execution), a system that converts documented material evidence into capacity estimates for robotic insertion. SAGE restricts a large language model (LLM) to extracting tensile and compressive strengths from retrieved passages and tables and records their sources. A response model then interpolates offline finite element (FE) solutions to convert these strengths and the assembly conditions into axial load capacity. For fits with positive clearance, the estimated capacity sets the policy’s axial force limit; for interference fits, it is compared with measured support demand to determine admission. On the primary benchmark, SAGE reduces mean capacity error from 80.65% for direct LLM estimates based on the same evidence to 10.74%. Without refitting, the mean error remains 8.00% on 16 additional geometries. Under the assigned support release model, SAGE correctly classifies 59 of 62 scored simulation runs, with only conservative errors. In recorded xArm6 demonstrations, SAGE takes material documents as input and completes physical insertion in 9 of 13 trials. These results show that assigning document interpretation to the LLM and force calculation to an explicit mechanical model produces accurate capacity estimates and traceable insertion decisions.
[AI-182] Fairly Compensated Distributed Information Retrieval and Augmentation for AI Agents
链接: https://arxiv.org/abs/2609.22601
作者: Yixiang Yao,Pasha Barahimi,Srivatsan Ravi
类目: Emerging Technologies (cs.ET); Artificial Intelligence (cs.AI); Cryptography and Security (cs.CR); Distributed, Parallel, and Cluster Computing (cs.DC)
备注:
Abstract:The increasing reliance of autonomous AI agents on external and distributed knowledge sources introduces a fundamental challenge for decentralized information marketplaces: retrieval agents must evaluate the quality and relevance of data before purchase, while data providers must avoid revealing valuable information prior to guaranteed compensation. This paradox becomes particularly critical in trustless multi-agent environments, where no centralized intermediary can enforce fairness between parties. In this paper, we propose a fairly compensated protocol for distributed information retrieval and augmentation in autonomous agent networks. Our framework enables retrieval agents to securely evaluate and rank candidate documents without learning their plaintext contents, while ensuring that data providers are compensated only when valid information is successfully delivered. We further analyze the security properties of the protocol against malicious adversaries and evaluate its practical feasibility through implementations. Experimental results demonstrate that the proposed design is practical with current cryptographic infrastructures while preserving confidentiality, correctness, integrity, and fairness. We believe such mechanisms provide an important cryptographic foundation for trustworthy and economically sustainable decentralized knowledge marketplaces for future AI agent ecosystems.
[AI-183] MAWILE: Multi-Axis Workbench for Inspecting LLM Evaluators
链接: https://arxiv.org/abs/2609.22599
作者: Jackson Hassell,Farima Fatahi Bayat,Pouya Pezeshkpour,Estevam Hruschka
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Large language model (LLM) judges provide a flexible and scalable method for evaluating model and agent outputs, but their verdicts can be sensitive to incidental changes in the evaluated response, judge instructions, and scoring rubric. Existing systems examine important subsets of these failure modes, but auditing a configured judge requires testing both the judge instrument and the items it evaluates. We introduce MAWILE, a developer-facing workbench for auditing judge sensitivity across four surfaces: the judge prompt, judge rubric, target-system input, and target-system output. Given a user-supplied judge and representative evaluation items, MAWILE constructs and validates controlled perturbations, re-executes the judge, and localizes the resulting sensitivity. Each perturbation declares whether the verdict should remain invariant or change in a specified direction, allowing the same system to measure both robustness to irrelevant variations and sensitivity to meaningful changes. MAWILE audits binary, ordinal, and pairwise judges without requiring gold labels. The code for this tool is available at: this http URL.
[AI-184] SEABED: SouthEast Asian Benchmark for Evaluating Audio Reasoning EMNLP2026
链接: https://arxiv.org/abs/2609.22586
作者: Harshit Rajgarhia,Asif Shaik,Rachuri Lokesh,Sushanta Kumar Pani,Abhishek Mukherji
类目: ound (cs.SD); Artificial Intelligence (cs.AI)
备注: Accepted at SALMA[EMNLP 2026] Workshop
Abstract:Modern audio-language models are no longer judged only on what words they can transcribe, but on whether they can reason over what they hear: recovering meaning that lives in tone and prosody, telling dialects and regional languages apart, and resolving ambiguity that the written form leaves open. This capability is now measured by a growing family of audio-reasoning benchmarks, but almost entirely in English and on general-domain audio. Southeast Asia (SEA) is served instead by benchmarks that inherit an English task taxonomy of recognition, translation, and paralinguistic classification, and therefore test whether a model hears SEA speech rather than whether it can reason from it. We introduce SEABED, an audio-first question answering dataset designed to benchmark language and audio reasoning models on SEA speech. SEABED comprises a suite of six audio-reasoning tasks built entirely from real, openly available SEA speech corpora, yielding 5,404 question-answer pairs. We evaluate six frontier and region-specific audio LLMs: even the state-of-the-art model Gemini 3.5 Flash achieves only 50.3% weighted average accuracy. SEABED evaluates not only answer accuracy, but also whether models’ stated reasoning is grounded in the audio evidence. A sample of the benchmark data is available here: this https URL.
[AI-185] Zero-Trust Authorization and Discovery for Enterprise MCP
链接: https://arxiv.org/abs/2609.22573
作者: Huan Li,Yuwei Wang,Srinivasan Manoharan
类目: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI)
备注:
Abstract:LLM agents translate natural-language context, which may include attacker-controlled text, into privileged tool calls, so authorization must remain effective even when an agent is prompt-injected or adversarially steered. The Model Context Protocol (MCP) has become a widely adopted interface for this boundary, yet its official SDKs’ authentication and authorization primitives fall short of enterprise zero-trust requirements, most acutely a dual-persona model in which one server must serve human users (corporate SSO) and automated agents (service-account credentials on a different header). We conduct a systematic gap analysis of six surveyed MCP SDKs (Python, TypeScript, Go, Rust, C#, Swift) and identify three structural shortcomings: credential extraction bound to a single Authorization header, complicating dual-persona deployment without custom middleware; the absence of pre-authentication tool discovery; and the lack of fine-grained per-tool authorization in the base SDKs. We close these gaps with composable extensions to FastMCP: cross-header credential normalization for enterprise deployments serving both human and service-account callers, cached token verification across heterogeneous IdPs, an unauthenticated metadata endpoint for credential-free registry discovery, and permission-filtered tool visibility kept consistent with per-tool invocation enforcement by a single declarative annotation, all without modifying the protocol or SDK internals. Across four frontier LLMs over 2160 attempts, an in-body-check-only server still exposes forbidden tools (152/720, 21.1%), whereas permission-aware visibility drives the rate to 0/720; visibility-only filtering remained bypassable by scripted clients, while models referenced the hidden tool by name in up to 94% of settings when inferable from the prompt, confirming that discovery controls cannot replace invocation-time enforcement.
[AI-186] he Ups and Downs of Backprop Weights
链接: https://arxiv.org/abs/2609.22554
作者: Giuseppe Chindemi,Benjamin F. Grewe
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Backpropagation (BP) has driven the remarkable success of modern deep learning by enabling large hierarchical networks to learn complex functions end-to-end. Yet it does not by itself determine how parameters should be organized so that functional components can be reused and adapted selectively. For example, object recognition and motion prediction may depend on overlapping parameter sets, making them difficult to isolate or modify independently. We call this condition weight entanglement. Modern architectures dynamically select which parts of a network process each sample: nonlinearities gate units, attention selects interactions, and Mixture-of-Experts architectures route inputs to modules. Yet such selection does not ensure that the same functional component remains linked to an identifiable parameter set across samples. We propose weight operators: parameterized modules that implement reusable functional components and can be composed at inference to form the function required by each sample. Learning proceeds in two stages: the model first infers the required operator composition, then updates only the selected operators’ parameter sets. Vector Networks (VNs) provide one implementation. They couple operator selection to local error-driven updates within each layer and show that learned operators can be reused in combinations absent from training while updates remain restricted to the selected parameter sets. This provides a basis for testing functional parameter identifiability: whether an operator remains linked to the same functional component during learning. We argue that functional parameter identifiability may provide an organizing principle for models that systematically reuse and recombine learned functions while adapting only the components that need to change.
[AI-187] FRAMES: Failure Recovery And Monitoring of Embodied Skills for Humanoid Loco-Manipulation IROS2026
链接: https://arxiv.org/abs/2609.22538
作者: Ajay Vikram Periasami,Xinyuan Luo,Haoyu Li,Xianyi Cheng
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注: Accepted for poster presentation at the IROS 2026 Workshop on Full-Shift Robot Co-Workers: Active Perception and Interaction for Human Environments
Abstract:Large language model (LLM) planners can decompose natural-language instructions and select reusable robot skills, but choosing the correct skill does not guarantee successful physical execution. This gap is especially important in humanoid loco-manipulation, where errors during approach, grasping, transport, or placement can invalidate the remainder of a long-horizon plan. We present FRAMES, a failure-aware supervisory framework for the Unitree G1 humanoid that operates above the CEER whole-body controller. A Planner Agent selects subtasks through parameterized mid-level skills, while a vision-language-model-based Monitor Agent evaluates each skill using temporal multi-view observations and structured robot and contact evidence. Detected failures stop the active skill and provide grounded feedback to a Recovery Agent. The framework further includes a Memory Module for reusing prior skill experience, and geometric grounding via depth and segmentation. We independently evaluate the monitoring module of the framework in MuJoCo using 100 trials comprising 50 failed and 50 successful executions across five tasks. The monitor detects 48 of 50 failures, correctly accepts 46 of 50 successful executions, and achieves 94.0% overall accuracy. These results provide initial evidence for the monitoring component, while end-to-end evaluation of the complete recovery loop remains ongoing.
[AI-188] EvidenT: Building Trustworthy Enterprise Assistants through Evidence Groundedness and Traceability EMNLP
链接: https://arxiv.org/abs/2609.22537
作者: Anubha Kabra,Katie Jooyoung Kim,Colin Zhiwei Kou,Helene Sajer,Yimei Fan,Radomir Cisar,Heather Greenhalgh,Gabriel Martinez Vidiri
类目: Artificial Intelligence (cs.AI)
备注: Accepted in EMNLP Industry Track 2026
Abstract:Enterprise AI assistants must produce responses that are verifiable and traceable to source evidence. However, retrieval augmented generation (RAG) over heterogeneous enterprise data can suffer from citation drift, unsupported content, and weak source traceability. We present EvidenT (T = Trust + Transparency + Traceability), a lightweight pipeline that verifies extracted evidence against retrieved documents before answer generation, without model retraining. EvidenT combines structured passage extraction with deterministic lexical alignment to filter unsupported content, correct citation drift, and preserve source-span traceability. On approximately 500 real enterprise queries, EvidenT improves gold-source hit rate by an average of 29% over prompting baselines, produces no citations to nonretrieved urls, and achieves near-saturated answer-to-source lexical coverage.
[AI-189] Agreement Overstates Evidence: Error Dependence in LLM Judge Consensus
链接: https://arxiv.org/abs/2609.22512
作者: Elias Hossain,Niloofar Yousefi,Ser-Nam Lim
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Consensus among LLM judges is often taken as strong evidence that a decision is correct. This assumes that judges make their errors independently. In practice, LLM judges are often trained and evaluated in similar ways, so they can make the same mistakes. We study how this dependency affects the reliability of consensus. We find substantial error correlation across both open-weight and frontier LLM judges. In our main bank of ten judges, the average pairwise correlation between judge errors is 0.21. As a result, the ten judges only provide roughly as much statistical information as 3.5 independent judges. The dependency is even stronger among the high-accuracy frontier judges we evaluate, including judges from different providers. In up to 28% of our comparisons, ignoring shared errors leads to the conclusion that one system is significantly better, while accounting for them does not. We also find that the pattern of errors matters. Errors shared by most judges and errors concentrated among a smaller group affect consensus differently and favor different voting methods. Measuring the overall amount of correlation alone is therefore insufficient. Our results suggest a simple approach: use a small set of trusted examples to estimate judge accuracy and identify shared mistakes. These shared errors should then be considered when analyzing the results, and the voting method should be chosen using trusted examples before it is applied to new data.
[AI-190] he Wisdom of Artificial Deliberative Crowds
链接: https://arxiv.org/abs/2609.22497
作者: Federico Barrera-Lemarchand,Mariano Sigman,Joaquin Navajas
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:The aggregation of many lay estimates often outperforms individual expert judgment, a phenomenon known as the wisdom of crowds. While this is usually attributed to the independence of estimates, an even stronger effect arises through deliberation: averaging the consensus estimates of small deliberating groups outperforms the classical wisdom of crowds, with individual judgments themselves also becoming more accurate after deliberation. Whether these improvements transfer to large language models deliberating amongst themselves is unknown. Here we adapt a three-stage deliberation paradigm previously used with human participants for use with large language models from three different families, and test it across four domains of increasing real-world stakes: visual numerical estimation (Study 1), peer review of machine-learning papers (Study 2), detection of hidden malicious behavior by an artificial intelligence agent (Study 3), and sports forecasting against a real prediction market (Study 4). Across domains, deliberation reduced collective error beyond passive aggregation of independent responses, and post-deliberation individual judgments retained this collective gain. Notably, the advantage required model diversity: groups composed of clones of a single model did not benefit from deliberating. These results establish machine deliberation as a general-purpose aggregation mechanism, and point to diversity as an active ingredient.
[AI-191] A Governance-Aware Large Language Model Orchestrated Agent ic Digital Twin for Transmission System Operator Control Room Decision Support
链接: https://arxiv.org/abs/2609.22476
作者: Costas Mylonas,Magda Foti,Emmanouel Varvarigos
类目: oftware Engineering (cs.SE); Artificial Intelligence (cs.AI)
备注:
Abstract:Transmission system operators face rising complexity from renewable integration, reduced inertia, and tighter security margins. Large language models offer natural-language decision support, but their hallucinations, uncontrolled tool use, and weak traceability conflict with control room requirements. This paper presents a governance-aware agentic digital twin for transmission grid control rooms. The large language model only selects and parameterizes whitelisted analysis tools, and every proposed action passes through a governance layer that the model cannot bypass. The layer enforces four rules on every run. Only whitelisted tools execute. No run exceeds its step budget. No action with side effects executes without explicit operator approval. Every number in an answer is rendered by the layer from backend results with its unit, variable, and time. The rules are checked on a persistent audit trail for every run of a released 118-task benchmark, which covers analytics, simulation, multi-step workflows, and twelve families of adversarial inputs on a digital twin of the Greek transmission network. Across 590 runs of the primary model, tool selection reaches 96.5% and task success 93.7%, and all four rules hold without exception. In a separate three-repetition study across four large language models, 1416 runs in total, the rules again hold on every run, with an approximate 95% lower bound of 99.8%. Removing the layer makes the same model execute all 45 approval-requiring runs without authorization and leaves only 39.2% of its answers with backend-supported numbers. Enforcement costs 12 to 16 milliseconds per request.
[AI-192] Goal-driven Variant Categorization
链接: https://arxiv.org/abs/2609.22475
作者: Daniel Calegari,Daniel Amyot
类目: Artificial Intelligence (cs.AI); Databases (cs.DB); Software Engineering (cs.SE)
备注:
Abstract:Process discovery rarely yields a single coherent process structure. For analysis, a common step is to cluster process variants based on structural similarity and then assign business meaning to the resulting groups. Since these partitions are not derived from the organization’s goals, analysts must manually interpret and consolidate variants into business-meaningful categories. This judgment-intensive step becomes increasingly difficult as the number and complexity of variants grow. In this paper, we propose a goal-driven approach to variant categorization that reverses this workflow. We first author an organization’s goal model that predefines the categorization axis. Each variant is transformed into a textual narrative describing its behavior, and a Large Language Model (LLM) interprets it in the context of the goal model and assigns the variant to the most appropriate category. LLM-based semantic reasoning connects low-level process behavior with analyst-defined business goals. We instantiate this approach end-to-end and evaluate it on three public logs differing substantially in scale and behavioral diversity. Goal-model guidance yields partitions that differ from those produced by unguided induction and respond to controlled edits to the declared alternatives, at the cost of authoring a goal model.
[AI-193] Learning 3D biophysical cell properties from 2D images and cell-population statistics
链接: https://arxiv.org/abs/2609.22410
作者: Santiago Hernández-Orozco,Hector Zenil
类目: Artificial Intelligence (cs.AI)
备注:
Abstract:Inferring 3D cellular properties from 2D microscopy is difficult when a reference instrument reports only population statistics rather than labels for individual cells. Here we develop a population-supervised framework that maps single 2D red-cell images to latent biophysical quantities and aggregates them to mean corpuscular volume, red-cell distribution width and mean corpuscular haemoglobin. The model combines shared local inference, a biophysically structured decoder for volume and haemoglobin, learned instance weighting and device-specific calibration. We formalise conditions under which aggregate observations identify restricted instance predictors, show why population agreement does not by itself identify single-cell properties or 3D geometry, and derive the dispersion penalty induced by subset mean matching. The development dataset comprises 390 specimens and 1,105 acquisitions across six devices, with reported Pearson correlations of 0.86–0.98 against a Sysmex analyser. The framework provides a testable route from 2D images and population supervision to 3D cellular biophysics without claiming explicit 3D reconstruction.
[AI-194] Social Influence and the Allocation of Scientific Attention in AI Populations
链接: https://arxiv.org/abs/2609.22408
作者: Maxim Chupilkin
类目: Artificial Intelligence (cs.AI); General Economics (econ.GN)
备注:
Abstract:AI systems are becoming participants in the evaluation and use of scientific research. They encounter citation counts, download statistics and lists of popular articles developed around human readers, but the collective consequences of these signals for artificial readers remain uncertain. This paper adapts the Music Lab design to a market for academic attention. In the first experiment, 1,000 AI agents choose papers from the titles and abstracts of all 114 regular research articles published in the American Economic Review in 2025. The experiment has five independent-choice communities and five social-influence communities, each with 100 sequential agents. Only agents in the social-influence condition observe earlier selections within their community. Agents may select any number of papers. Social-information communities select 17.2 percent fewer papers per agent, concentrate their choices more heavily, and collectively cover 73 papers, compared with 90 independently. Between-community variation is greater under social information. In a second experiment with 200 agents across twenty social communities, randomly assigning papers five initial selections raises their subsequent selection rate by 45.55 percentage points (95% CI: 41.20 to 49.90). Choices have modest correspondence with external citations and little correspondence with download counts. The results show how a simple information rule shapes the volume, breadth and distribution of scientific attention in an artificial population.
[AI-195] Resist Update Reject: Preference Optimization Installs a Prior-Dependent Reliability Switch
链接: https://arxiv.org/abs/2609.22359
作者: Sen Yang,Yuen-Hei Yeung
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:An aligned model asked to hold its answer against a manipulative source must still update on a reliable one and reject an unreliable one: resistance, reliable-update, and unreliable-source rejection are one three-way contract, not three independent behaviors. We show the objective most anti-sycophancy work optimizes is non-identifying with respect to source reliability: because no preference label depends on whether a source is actually reliable, any scalar mixture of the arms traces a single deference dial, and no point separates two same-template testimonies differing only in stated reliability. This fixation \leftrightarrow gullibility frontier is a property of the objective, not any model. We make reliability identifiable through data: a threshold benchmark where a source asserts the opposite answer while stating its reliability r , and the correct action is to flip iff r exceeds the model’s prior strength p . Preference optimization over balanced coverage installs a prior-dependent reliability switch: across three seeds on Qwen2.5-7B-Instruct the threshold r^\star rises monotonically with the prior, decision accuracy reaches 0.84 with a monotone flip curve (Spearman 0.56 ), and the policy generalizes to unseen reliability values and a held-out notation, following stated reliability over role prestige. Three controls localize the cause: an unmatched variant installs the switch equally ( 0.80 ), a second preference optimizer (IPO) installs it just as well ( 0.86 ), whereas supervised imitation does not ( 0.50 ), so the cause is preference optimization over reliability-labeled coverage, not pairing, loss, or imitation. A confirmatory battery replicates the switch on a fresh test draw, bounds it honestly (it keys on reliability stated in the testimony, not a separately audited record), and transfers it to Llama-3.1-8B. The frontier is empirical, not a theorem.
[AI-196] PAANI : On Device Visual Evidence Fusion and Explainable Guidance for River Robot Simulation
链接: https://arxiv.org/abs/2609.22353
作者: Savio Cardoz,Santhiya Rajan
类目: Artificial Intelligence (cs.AI)
备注: 20 pages, 8 figures, 11 tables. Includes system and AI architecture diagrams, model-training results, qualitative evaluations, and Arduino UNO Q deployment measurements. Project code, trained models, ONNX artifacts, logs, and reproducibility documentation are available at this https URL
Abstract:Mobile river monitoring robots must interpret obstacles and water boundaries that geographic waypoints alone cannot describe. On resource constrained platforms, converting imperfect visual predictions into timely and inspectable guidance is a distinct challenge. An object label or steering command does not explain which evidence supports a decision or when that evidence is unreliable. We present PAANI, an on-device perception to guidance architecture that combines a project trained YOLO11n detector and a custom MobileNetV3 Small semantic segmenter with timestamp aligned evidence fusion on Arduino UNO Q. Bounded tracking supplies object persistence, while an explicit corridor policy combines surface labels, accepted detections, urgency and mask uncertainty. Each final advisory exposes its contributing evidence and policy reasons. ROS 2 interfaces connect the local AI pipeline to a separate Gazebo vessel, localization and control testbed. Training uses 10,000 WaterScenes images for four-class detection and 1,127 MaSTr1325 images for segmentation, including 198 segmentation validation images. The selected FP32 ONNX models occupy 14.817 MB. Detector checkpoint test mAP at 0.5 IoU is 0.7388, while the separately evaluated rectangular ONNX export achieves validation mAP at 0.5 IoU of 0.7367. Segmentation ONNX validation mIoU is 0.9750. A five-minute UNO Q recording produced median and 95th percentile pipeline latencies of 467.8 ms and 580.3 ms at a configured 0.5 Hz cadence. The evaluation also identifies black input misclassification and a sampling rate mismatch that prevents the diagnostic apparent motion estimator from collecting sufficient evidence. These results support an inspectable and reusable edge robotics foundation while clearly distinguishing model accuracy and on-board execution from validated on-water collision avoidance.
[AI-197] Artificial Neural Networks as Surrogate Models in Black Box Optimization
链接: https://arxiv.org/abs/2609.22329
作者: Md Khadimul Islam Zim(1),Martin Holeňa(1) ((1) Czech Academy of Sciences, Institute of Computer Science, Prague, Czech Republic)
类目: Neural and Evolutionary Computing (cs.NE); Artificial Intelligence (cs.AI)
备注:
Abstract:Black-Box Optimization (BBO) is often applied in several engineering fields and can utilize an advancement of numerical measure- ments and simulation technologies. It deals with the optimization func- tions, where an analytical description is unavailable. It relies on meth- ods that require only an input point in the search space, paired with its corresponding objective function value, obtained through non-analytical means, e.g., sensors, experiments, or simulations. Common approaches include evolutionary optimization and other metaheuristics. Since BBO methods rely solely on objective function values, they typically require many evaluations, which becomes problematic when evaluating the ob- jective function is time-consuming or expensive. This leads to using surrogate-based optimization which evaluates selected true objective val- ues and trains a regression model to approximate the objective function across the search space. Surrogate-assisted black-box optimization is a small-data learning problem because the optimizer must approximate an expensive objective function from limited evaluations. Surrogate models act as data-efficient regressors, guiding the search toward promising or informative points under a restricted evaluation budget. In this paper, a new surrogate model using artificial neural networks, called Adaptive- Fidelity Nexus Covariance Matrix Adaptation Evolution Strategy (AFN- CMA-ES), is proposed for the selective evaluation of objective functions. The experimental results show its competitive performance compared to state-of-the-art surrogate-assisted BBO methods.
[AI-198] ORDER: A Fictitious-World Benchmark for Domain-Adaptive Embodied AI
链接: https://arxiv.org/abs/2609.22285
作者: Sai Krishna Reddy Sathi,Anuj Tiwari
类目: Robotics (cs.RO); Artificial Intelligence (cs.AI)
备注:
Abstract:Adapting language models to new domains via continual pre-training raises a basic evaluation problem: if the training corpus overlaps with what the model already knows, performance gains cannot be cleanly attributed to new learning rather than pre-existing knowledge. This matters most for knowledge-intensive, task-light (KHTL) robot deployments - pharmaceutical dispensing, hazardous-material handling, facility-specific protocols, where the physical task is simple but the governing rules are proprietary and safety-critical, and where extensive live testing is costly or unsafe. We introduce ORDER (Ontology-driven Decision-making for Embodied Reasoning), a benchmark built on a fictitious world: a 342,069-token synthetic corpus defining a self-consistent physics that cannot appear in any model’s pre-training data. ORDER pairs a 500-question knowledge test (ORDER-BENCH) with a harder compositional task, ORDER-SPATIAL: ordering objects for safe manipulation across both familiar and entirely novel scenes. GPT-4.1 without adaptation scores below chance on ORDER-SPATIAL (Kendall’s tau = 0.441), showing its priors actively conflict with the invented physics. After continual pre-training, small models improve substantially on both familiar and novel scenes alike evidence of genuine world-model induction rather than memorization. We then carry this through to a robot pipeline: models that answer the knowledge test well often cannot produce valid, executable plans without a further skill-adaptation stage, after which small, fully offline models outperform GPT-4.1 even when GPT-4.1 is given retrieval access to the same rules (Kendall’s tau = 0.848 vs. 0.606), on a full perception-to-execution loop demonstrated on a simulated iiwa7 arm with human-in-the-loop correction. Throughout, ORDER-SPATIAL performance, not knowledge-test accuracy is what predicts real plan quality.
[AI-199] acher Should Think Ahead: Adaptive Continuations for Reliable On-Policy Distillation
链接: https://arxiv.org/abs/2609.22254
作者: Jingang Zhou,Yuyi Zhou,Haiyang Guo,Xukai Wang,Shuai Feng,Sirui Gao,Jian Xu,Qingpei Guo,Xu-Yao Zhang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:On-policy distillation (OPD) is a promising approach for transferring knowledge between language models, where a student receives dense token-level supervision along its own generated trajectories. However, teacher supervision can be unreliable when conditioned on incomplete or low-quality student prefixes. We identify Teacher Uncertainty Contraction (TUC), a systematic phenomenon whereby the teacher’s predictive uncertainty decreases as it continues from a student-generated prefix. We theoretically characterize this trade-off through a variance-bias decomposition of teacher-branch gradients, showing that uncertainty contraction reduces variance while teacher-student path divergence increases bias, thereby favoring a finite continuation. Guided by this insight, we propose Adaptive-Continuations On-Policy Distillation (AC-OPD), which augments informative states along student rollouts with teacher continuations and adaptively selects their effective supervision horizons. Experiments on mathematical reasoning and code generation across model scales demonstrate that AC-OPD consistently improves over standard OPD. Controlled-continuations and matched-budget analyses further validate the adaptive-continuations design, highlighting adaptive teacher continuations as an effective principle for reliable on-policy this http URL code will be made publicly available upon publication.
[AI-200] CALM: A Calibrated LLM Choice Network Framework for Activity-Based Traveler Simulation
链接: https://arxiv.org/abs/2609.22252
作者: Yezhou Cheng
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:We present CALM, a reproducible hybrid framework that integrates an optional large language model (LLM) activity planner with calibrated stochastic choice, shared network feedback, memory and habit, typed feasibility checks, and deterministic offline replay. Unlike trip-mode classifiers or diary-only generators, CALM executes a closed traveler-day loop and evaluates each generative module against an empirical, reproducible baseline. On the 2024 New York City Citywide Mobility Survey (CMS), 110,691 seven-mode trips are split by respondent into 78,487 training and 32,204 holdout trips. Training-only alternative-specific constant calibration reduces mean holdout mode Jensen-Shannon divergence from 0.15599 to 0.00394 across ten seeds. A matched live-LLM ablation then quantifies trade-offs among aggregate fit, temporal fit, behavioral persistence, and feasibility, while frozen prompt-response pairs support deterministic replay of downstream simulation. Controlled weather, delay, fare, and parking ladders further demonstrate consistent and interpretable responses under intervention. CALM contributes a reproducible protocol for integrating and evaluating generative planners in traveler simulation through person-disjoint calibration, matched module ablation, controlled stress testing, and end-to-end traceability.
[AI-201] CHART: A Harness-Rotation Curriculum for Harness-Robust Search Agents
链接: https://arxiv.org/abs/2609.22247
作者: Xinlu Zhang,Ying-Chun Lin,Zhihan Zhang,Besnik Fetahu,Xi Chen
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Search agents are usually trained under a single harness. But once an agent is deployed in a real application, its harness is frequently updated (e.g., a rewritten system prompt) to fit production needs. This exposes a fragility of post-trained agents: because a learned behavior is entangled with its training harness, even a harness update that leaves the task unchanged can fail to elicit the behavior. We train a search agent to perform parallel search, a popular strategy for improving both search efficiency and performance. We find that training under a fixed harness makes the behavior harness-local, overfit to that harness’s surface form: when the harness changes, the model falls back to serial search. An intuitive fix is harness augmentation, but simply training on more harnesses does not resolve the problem. GRPO learns from the reward gap between parallel and serial rollouts of the same question: a small harness pool saturates that gap early, while a large pool dilutes the per-harness signal too thinly for any harness to consolidate. We therefore propose Curriculum HArness Rotation Training (CHART), a rotating curriculum that lets a search agent gradually consolidate parallel search across harnesses. At each periodic evaluation, CHART “graduates” the harnesses whose expected behavior is learned and replaces them with still-learnable ones, keeping the reward gap alive throughout training. Starting from the same harness pool, CHART makes the model learn parallel search on all harnesses, whereas static augmentation succeeds on at most half of them. The behavior also carries to held-out harnesses: CHART parallelizes on 89% of held-out turns, against at most 5% for the static pools. It further transfers to a new QA task and search environment, improving pass@1 by 5.6pp over the best static pool. Finally, CHART-trained agents benefit more from meta-harness search than baselines.
[AI-202] Universal Observatory Graphs for Distributed Sky Coverag e and Artificial Intelligence Based Interplanetary Routing
链接: https://arxiv.org/abs/2609.22244
作者: Mohammed Abdel Razek
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:This research proposes the Universal Observatory Graph (UOG), an AI-driven framework for distributed astronomical observation across the Solar System. The proposed architecture models autonomous observatories located at the Sun planet L2 Lagrange points as nodes in a weighted graph, while communication links are represented as graph edges characterized by multi-objective physical and operational metrics, including interplanetary distance, communication latency, transmission power, and link reliability. The resulting graph provides a unified mathematical representation of a cooperative interplanetary observatory network. This proposal examines a six-observatory Solar System configuration comprising Earth, Mars, Jupiter, Saturn, Uranus and Neptune. Instantaneous sky coverage is evaluated independently using a 200,000 direction Fibonacci sphere, a 2,000,000 direction fixed seed Monte Carlo calculation and deterministic spherical integration. All three methods yield complete network union coverage, approximately 0.43% complete six observatory intersection and approximately 24.96% mean pairwise Jaccard similarity under the adopted pointing model. Communication routing is subsequently formulated as a finite horizon Markov decision process and solved using tabular Q-learning. The reward balances node participation and a distance dependent reliability proxy against distance, light time latency and a distance squared transmission power proxy. The learned Earth-Saturn-Uranus-Neptune route is also the highest discounted return route among all 41 feasible simple paths under the four hop constraint. The framework provides a reproducible baseline for sequential coverage assessment and multi objective routing; time dependent ephemerides, mission specific visibility, calibrated link budgets and scalable graph policies remain future work.
[AI-203] Not All Ranks Are Equal: Budget-Aware LoRA Merging Across Tasks AAAI
链接: https://arxiv.org/abs/2609.22237
作者: Avinash Amballa,Yashas Malur Saidutta,Wenbo Li,Lazar Valkov,Srinivas Chappidi
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 7 pages, AAAI submission
Abstract:Merging low-rank adapters (LoRAs) promises to eliminate the overhead of swapping task-specific weights at inference time. However, existing merging methods assume every layer needs the same rank budget. Further, some methods assume that rank budget needs to be split equally among the tasks too. We show this uniform-budget assumption is a major source of the performance gap between merged and per-task LoRAs. However, rank selection is an NP hard problem. To this end, we introduce Net Utility, a data free metric that first decomposes every task LoRA by its Singular Value Decomposition (SVD) and scores each of those singular directions by its task utility and its interference with other tasks directions. Next, we globally pool these scores to select singular directions with the highest values with a constraint on the total number of directions selected. The proposed Net Utility metric is applied on top of five different merging methods across three different merging spaces. The merging is done over two sets of tasks, vision and language tasks. Net utility based rank allocation outperforms its counterparts without that allocation. On average, over vision tasks it achieves +2.1% improvement in performance, and +2.2% improvement over the language tasks.
[AI-204] Improving Parameter Utilization by Sharing Neural Experts Across Layers in Transformers EMNLP2026
链接: https://arxiv.org/abs/2609.22199
作者: Dian Jiao,Jiaxin Duan,Shuai Zhao,Jiabing Leng,Yiran Zhang,Feng Huang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 11 pages, 6 figures. To appear in the Proceedings of EMNLP 2026 Main
Abstract:Transformer-based large language models often suffer from inter-layer parameter redundancy, where functional transformations are redundantly learned across network depths. We propose CS-MoE, a novel Transformer architecture featuring cross-layer expert sharing to address this inefficiency. Deviating from the widely used Mixture-of-Experts (MoE) architecture that terminates each Transformer block with layer-isolated experts, CS-MoE combines layer-independent experts with concurrent access to a centralized, globally shared expert pool. This \textitGlobal Experts Sharing mechanism enables elastic control over token-level parameter activation and computational consumption (FLOPs). Experiments demonstrate that CS-MoE achieves lower perplexity than equal-scale dense Transformers while activating only 55% of parameters. Furthermore, its performance scales monotonically with an increased number of activated experts and approaches MoE counterparts that consume more FLOPs by expanding the shared pool with a fixed FLOPs budget. CS-MoE also establishes a flexible Pareto frontier between computational cost and model capacity, offering an efficient alternative for computation-constrained environments.
[AI-205] Dissecting Hierarchical Reasoning Models: A Mechanistic Study
链接: https://arxiv.org/abs/2609.22197
作者: Leo Raphael Rodrigues,Jian Kang
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注: 7 pages, 5 figures, Under Review
Abstract:We study Hierarchical Reasoning Model (HRM), a representative hierarchical Transformer-based latent reasoning model with many variants, on Sudoku, Maze, and ARC-AGI-2. We mechanistically understand how HRM reasons and what information it encodes. Our analyses compare HRM against Transformer baselines with and without recurrent modules, apply causal interventions on recurrent states, and utilize linear probes against random-direction ablations, as well as sparse autoencoders with feature ablations. Our results reveal several key findings: recurrent models outperform one-pass baselines, while single-state recurrent Transformers are comparable to HRM. State interventions further show that the causal contributions of the high- and low-level states vary across task-specific checkpoints and inference stages. Selected task variables are linearly decodable from the recurrent states in HRM, yet ablating probe directions produce effects comparable to random controls. SAE ablations yield larger behavioral changes than probe-direction ablations. However, top-ranked SAE features show no stable advantage over size-matched random subsets at larger ablation sizes or across tasks; the same pattern persists in a Sudoku control with within-step BPTT. Together, we characterize that HRM is essentially implementing constraint-aware iterative refinement on a puzzle-specific solution state, in which the functional contributions of components at different levels vary without relying on a compact, causally important feature set. These results highlight the necessity of studying the different working mechanisms and the importance of developing mechanistic interpretability techniques better suited for latent-space, recursive reasoning models.
[AI-206] SegTSim: A Big Data Driven Segmented Temporal Simulation Framework for Heterogeneous Multivariate Systems
链接: https://arxiv.org/abs/2609.22192
作者: Xinhang Li,Chenxi Geng,Yujia Sun
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Heterogeneous multivariate time-series systems exhibit segment-specific nonlinear dynamics that challenge monolithic forecasting architectures. We propose SegTSim, a big-data-driven segmented temporal simulation framework that integrates segment-specific elasticity modeling with adaptive min-gating, dynamic production relocation optimization, multi-factor data fusion with exchange-rate propagation, and a deep ensemble validation pipeline. The framework is validated on US–Japan automotive trade data from USITC repositories spanning 2015 to 2025, comprising approximately 13000 annual records. Under a 25% perturbation scenario, Japanese import volume declines by 20.4% to 0.93 billion USD, while all output variables maintain coefficients of variation below 3.5% across 1000 ensemble inference runs.
[AI-207] DPTM-DT: Dual-Pretrained Transformer Multitask Representation Learning for Drug-Target Prediction
链接: https://arxiv.org/abs/2609.22184
作者: Ge Kong
类目: Machine Learning (cs.LG); Artificial Intelligence (cs.AI)
备注:
Abstract:Drug-target relation prediction supports candidate screening, drug repositioning, and mechanism analysis. Existing models often use incomplete drug or protein representations, model cross-modal interactions shallowly, or train affinity regression and interaction classification separately, although these tasks describe closely related views of the same drug-target pair. This paper presents DPTM-DT, a dual-pretrained Transformer framework for multitask drug-target prediction. DPTM-DT combines GROVER molecular graph embeddings, ESM protein language-model embeddings, and CTD physicochemical descriptors, then exchanges drug-target information through bidirectional cross-modal attention. A shared pair representation is used for continuous affinity regression, high-affinity binary classification, and six-level affinity classification. Experiments on Davis and KIBA cover random 80/20 and DeepDTA-style standard splits. On the random 80/20 split, DPTM-DT achieves MSE/CI values of 0.193/0.917 on Davis and 0.120/0.918 on KIBA. It also reports binary AUPR/MCC values of 0.727/0.654 and 0.798/0.689, and six-class Macro-F1/Top-2 values of 0.800/0.932 and 0.815/0.962 on Davis and KIBA, respectively. Across the reported regression, binary classification, and multiclass classification settings, DPTM-DT achieves the best overall performance among the compared methods. Results under the standard split show the same relative trend. Ablations indicate that dual target representation, gated fusion, and cross-modal attention each contribute to the final performance. Code and supplementary materials are available at: this http URL.
[AI-208] A digital-twin framework for forecasting treatment-day imaging with contour uncertainty in adaptive proton radiotherapy
链接: https://arxiv.org/abs/2609.24725
作者: Yizhou Wu,Jie Ding,Justin Roper,Minglei Kang,Yuheng Li,Sibo Tian,David S. Yu,Xiaofeng Yang,Chih-Wei Chang
类目: Medical Physics (physics.med-ph); Artificial Intelligence (cs.AI)
备注:
Abstract:Head-and-neck anatomy changes over a six-to-seven-week proton course, and the anatomy of a later week cannot be imaged when the plan is made. We present a digital-twin framework that forecasts a patient’s treatment-day anatomy as an ensemble of predicted CTs with propagated contours and quantifies the uncertainty of the forecast contours. The twin is a library of previously treated patients with planning and weekly quality-assurance CTs (QACTs), made patient-specific by a two-step foundation-model deformable registration: a cross-patient field carries each library patient onto the current patient, and a longitudinal field, estimated in the current patient’s frame, carries that patient’s planning-to-QACT change onto the current patient’s own planning CT. A library of 302 observations from 88 patients yields about 300 replicates per patient, each a deformation that occurred in a treated patient. The dispersion of the propagated contours, resolved by outward normal, is six-direction contour uncertainty in millimeters. This is uncertainty in the input to the forecast, which library patient the current patient follows, rather than in model parameters, and it is unchanged when the registration engine is exchanged. On ten patients with clinician contours on two QACTs, the library alone fixes the anisotropic shape of the uncertainty (4.5 to 6.2 mm); the first QACT narrows it by a factor of 3.2 to 3.6 without a contour being drawn; an approved contour improves the center but not the width. The estimate orders directions correctly but is not Gaussian-calibrated. A clinical target volume expansion is worked out as one application.
[AI-209] Adapting Boltz-2 with limited experimental activity data improves early enrichment in virtual screening
链接: https://arxiv.org/abs/2609.24302
作者: Kairi Furui,Masahito Ohue
类目: Biomolecules (q-bio.BM); Artificial Intelligence (cs.AI); Machine Learning (cs.LG); Quantitative Methods (q-bio.QM)
备注:
Abstract:Virtual screening aims to prioritize active compounds from large chemical libraries within a limited experimental budget. When applying Boltz-2 to virtual screening, a key challenge is how to use limited experimental data from the target assay to improve the prioritization of active compounds. We investigated whether fine-tuning the Boltz-2 affinity heads with a small number of binary activity labels could improve early enrichment of active compounds in hit discovery. We compared fine-tuning with 40-300 labels in a retrospective evaluation on eight MF-PCBA targets. With 300 activity measurements, fine-tuning increased the number of actives in the top 1% by a geometric mean of 1.77-fold across the eight targets and improved average precision (AP) by 2.14-fold relative to the control without fine-tuning. We also investigated whether rescoring a subset of candidates could retain the improvement in hit recovery by reranking only the top-ranked Boltz-2 candidates with the fine-tuned head. Restricting rescoring to approximately 10% of the evaluation set retained hit recovery comparable to full rescoring. These findings show that affinity-head fine-tuning with limited activity labels improves early enrichment with Boltz-2 and that this benefit can be retained when rescoring a restricted set of candidates.
[AI-210] Bayesian Filtering in Physical Systems via Test-time Trained Flow Matching
链接: https://arxiv.org/abs/2609.23383
作者: Ruiqi Feng,Chongyi Wang,Tao Zhang,Tailin Wu
类目: Machine Learning (stat.ML); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Bayesian filtering provides a principled framework for online state estimation under uncertainty, yet its application to systems with high-dimensional states and complicated posterior distributions remains challenging. Recent generative models, such as flow matching, have shown potential in Bayesian filtering. However, they still rely on particle-based representations of the posterior, which lose the rich information of the full distribution, or tackle a trajectory-level inverse problem that conflicts with the recursive structure of Bayesian filtering. To address this, we propose a new perspective of directly encoding the evolving distribution into flow matching model weights, namely, the Belief Flow Filter (BFF). It is a generative filtering framework that updates model weights via gradient descent at test time to track the posterior evolution. Thereby, BFF bypasses the scalability issue of particle representations or the flexibility limitation of Gaussian assumptions in conventional filters. We theoretically justify that the BFF design is structurally aligned with Bayesian filtering, and its training objective targets the recursive filtering operator. BFF is empirically verified across 5 different physical systems, including ones with chaotic dynamics and highly sparse, non-linear observations. The results show that BFF attains the best score in 8 of 9 metric-benchmark cells across the three standard 1D and 2D PDE benchmarks, and similarly leads on the extreme single-moving-sensor setting and on a real-world-grounded tokamak plasma estimation task, demonstrating its potential to accurately approximate the Bayesian filtering operator in high-dimensional probability space.
[AI-211] Watching Quantum Models Think: Hilbert-Space Interpretability in Quantum Transformer Blocks
链接: https://arxiv.org/abs/2609.23016
作者: Diego Iacopetta,Andrea Gasparini
类目: Quantum Physics (quant-ph); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Deep learning models are powerful but opaque. As quantum machine learning matures, the field faces a defining choice: build quantum models that are equally opaque, or exploit the mathematical structure of quantum mechanics to make them inherently interpretable. We show that the latter is possible. By tracking quantum mutual information~(MI), entanglement entropy, and state fidelity through the layers of a Quantum Transformer Block (\qtb), a fully-coherent variational circuit with quantum analogues of both attention and feedforward, we gain direct insight into how the model processes information: which tokens it attends to, when correlations form, and why predictions fail. On four tasks with known dependency structure we show that (i)~learned MI matrices align with ground-truth task structure (AUC ,=,0.69 on lookup), (ii)~disabling entangling gates collapses accuracy from 100% to 15% while MI \to 0 , proving entanglement is the mechanism, (iii)~accuracy and MI co-evolve during training ( \rho,=,0.92 on lookup), and (iv)~per-sample MI predicts prediction correctness on the conditional task with ROC AUC ,=,0.84 . All results are validated on IBM Quantum hardware (ibm_kingston, Heron~r2): the circuit’s reasoning process, from product state through structured entanglement, is directly observable on a superconducting processor. These proof-of-concept results, obtained on small synthetic tasks, suggest that the physics of quantum computation can provide intrinsic interpretability signals with no direct classical counterpart, motivating study of whether this advantage persists at scale.
[AI-212] Large language models in medical time series analysis
链接: https://arxiv.org/abs/2609.22262
作者: Yu Han,Cigdem Beyan,Xiang Zhang,Xiaofeng Liu,Nan Liu,Jimeng Sun,Shenda Hong,Cheng Ding,Vittorio Murino
类目: ignal Processing (eess.SP); Artificial Intelligence (cs.AI); Machine Learning (cs.LG)
备注:
Abstract:Medical time series (MedTS), including electrocardiograms (ECG), electroencephalograms (EEG), photoplethysmography (PPG), and vital-sign recordings, are central to clinical diagnosis and health monitoring. As large language models (LLMs) have advanced, a growing body of work has examined how their reasoning, generation, and knowledge-integration capabilities can support MedTS analysis. Yet existing studies remain scattered, and the field still lacks a clear view of how these models should be designed, integrated into clinical workflows, and evaluated. This review synthesizes recent work on large language models for medical time series analysis (MedTSLLMs), covering both methodological progress and issues related to real-world deployment. We review model architectures, data resources, and processing pipelines, and prompt design strategies adapted for diverse clinical scenarios. We further organize existing MedTS applications, ranging from diagnostic interpretation and report generation to longitudinal health monitoring and physiological signal synthesis, highlighting task-specific design choices, common evaluation protocols, and empirical findings reported across studies. By bringing together current practices and open challenges, this review aims to provide a clearer foundation for developing, evaluating, and deploying MedTSLLMs responsibly in healthcare. We also maintain a regularly updated list of MedTSLLM studies and resources at: this https URL.
[AI-213] An Implant-to-Wearable IR-UWB Transmitter-Receiver Architecture and Layered Protocol for High-Density Brain-Computer Interfaces
链接: https://arxiv.org/abs/2609.22207
作者: G. D. Su,Y. Mo
类目: ignal Processing (eess.SP); Artificial Intelligence (cs.AI)
备注: 18 pages,14 figures
Abstract:High-rate neural telemetry requires an implant-towearable link whose circuit partition, waveform mapping, packet processing, and failure semantics are mutually consistent. This paper specifies a chip-oriented, one-way impulse-radio ultrawideband (IR-UWB) uplink comprising an implanted packet engine and free-running transmitter, a tissue-proxy channel, and a wearable mixed-signal receiver and digital baseband.
机器学习
[LG-0] LoRA-generating hypernetworks for efficient on-device LLM generative personalization
链接: https://arxiv.org/abs/2609.24979
作者: Sean Augenstein,Li Ding,Jihwan Lee,Keith Rush,Andrey Zhmoginov
类目: Machine Learning (cs.LG)
*备注: 19 pages, 4 figures
Abstract:On-device large language models (LLMs'), e.g. running on mobile phones, are ripe for improvement via personalization. The limited compute resources of mobile devices impose limits on model scale and thus model quality, making any realizable quality gains highly impactful. At the same time, their personal nature (i.e., the close coupling to a particular user) means that a given on-device LLM tends to be used in similar, predictable patterns over the course of time. This paper presents a novel method for personalizing on-device LLMs. It trains a hypernetwork to map a user's context tokens to a low-rank adaptation (LoRA’) well-suited to that user. Once the trained common artifacts are deployed to users’ devices, each user uses the hypernetwork to synthesize (entirely on device) a personalized LoRA. This approach blends the benefits while avoiding the drawbacks of two existing approaches to LLM customization: in-context learning (ICL') and parameter-efficient fine-tuning (PEFT’). Like ICL (and unlike PEFT), the on-device phase of our approach is computationally feasible, requiring only forward passes through neural networks. Like PEFT (and unlike ICL), our approach modifies the `target’ base LLM via weights (the LoRA), avoiding negative consequences (e.g. increased latency) associated with extending the input sequence. Our approach is particularly well-suited to the mobile device regime. Apart from the on-device compute and latency benefits mentioned, it also requires minimal additional storage, as internally its architecture partly leverages the same LLM weights as belong to the target LLM to be personalized. We demonstrate the benefits of LoRA-generating hypernetworks on several representative personalization datasets, comparing against baselines like ICL and PEFT. Of note, our personalization experiments focus on more challenging and less studied long-form text generation tasks.
[LG-1] Learning Physics from an Imperfect Ancestor
链接: https://arxiv.org/abs/2609.24947
作者: S. Mohammad Mousavi,Teeratorn Kadeethum,Nikolaos Bouklas,Somdatta Goswami
类目: Machine Learning (cs.LG)
*备注:
Abstract:Neural operators evaluate parametric partial differential equations cheaply but degrade sharply outside their training distribution. Physics-informed neural networks avoid dependence on labeled data, yet their optimization can be basin-fragile: when the governing residual admits multiple solutions, a PINN trained from scratch may converge to a physically incorrect state despite achieving a small residual. We show that these failure modes can be addressed jointly: an imperfect NO provides the structural prior needed to place a PINN in the correct solution basin, while the PDE residual refines the solution beyond the operator’s accuracy. We introduce a three-stage framework that freezes the spatial basis of a physics-informed NO, extrapolates its solution branch to an out-of-distribution parameter using a polynomial continuation prior, and distills the resulting field into a fresh PINN. The NO need not be accurate at the target; it transfers solution-branch information, while PDE residual minimization in the PINN governs convergence. We evaluate the framework on three nonlinear PDEs: 1D viscous Burgers, 2D steady Allen-Cahn near a pitchfork bifurcation, and 2D steady lid-driven cavity flow. For Allen-Cahn, where the trivial solution satisfies the PDE residual exactly, a standard PINN collapses to the trivial zero branch, whereas distillation from the crude extrapolated operator recovers the non-trivial branch that matches the finite-difference reference. For the lid-driven cavity, extrapolating to a Reynolds number of Re = 3200 accelerates convergence to the correct physical state, achieving competitive accuracy using fewer parameters and optimization steps than recent literature baselines. These results establish a simple principle: an NO need not accurately predict the solution to be useful; it only needs to identify the correct basin from which PINN optimization can recover it.
[LG-2] Learning Prognostic Variables for AI Convective Parameterizations via Symbolic Distillation
链接: https://arxiv.org/abs/2609.24882
作者: Jurij Schönfeld,Tom Beucler,Julien Savre,Steven Sherwood,Veronika Eyring
类目: Machine Learning (cs.LG); Atmospheric and Oceanic Physics (physics.ao-ph)
*备注:
Abstract:Hybrid AI-physics climate modeling aims to improve coarse (~100km-resolution) Earth system models by learning to parameterize subgrid processes from high-fidelity data. However, this so far mostly involves local-in-time, diagnostic parameterizations, in which the subgrid state depends only on the current coarse state with no memory of previous states, which is unrealistic for processes such as convection that have intrinsic persistence. To address this, we enhance local-in-time parameterizations by learning prognostic variables that compactly carry important, additional past information where no explicit sub-grid information is available. First we compress past information into a low-dimensional latent space using an autoencoder, which then informs a neural network trained to parameterize targeted subgrid-scale processes. We then replace the autoencoder with symbolic equations that govern the time evolution of the latent variables, yielding additional prognostic memory variables that can be integrated alongside the resolved atmospheric state. We evaluate this approach on two systems: the Lorenz-96 model (online) and surface precipitation from high-resolution atmospheric simulations (offline). A forced multivariate linear ordinary differential equation recovers most of the added value achieved by the autoencoder-based approach in both experiments. Benchmarked against diagnostic parameterizations without memory, our memory-informed approach improves climate statistics and temporal structure, including a realistic diurnal cycle of tropical land precipitation.
[LG-3] PredActor: Predictive Action Diffusion for Steerable Onboard Humanoid Control
链接: https://arxiv.org/abs/2609.24840
作者: Lei Ye,Haibo Gao,Yitang Li,Peng Xu,Zetong Jing,Junhan Sun,Fanrong Dong,Ziqi Han,Xue Wang,Jianhua Sun,Cewu Lu,Hao Zhao,Liang Ding
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: Project page: this https URL
Abstract:Diffusion models offer flexible motion generation, but translating this flexibility into feedback-responsive humanoid control remains challenging. Hierarchical systems steer motion through references that may exceed a separate tracker’s capabilities, leaving recovery and physical execution largely to the tracker. Action-only diffusion generates actions directly but lacks an explicit future-state trajectory for test-time motion objectives. Joint state-action diffusion provides this representation, yet representative controllers often depend on privileged full-body states, and support for learned behavior selection and test-time motion steering remains fragmented. We present PredActor, a predictive action diffusion policy that brings these complementary steering capabilities into one directly executed policy using proprioceptive observations. Conditioned on proprioceptive history and optional task context, PredActor jointly generates executable actions and an internal future-state trajectory. Classifier-free guidance strengthens text-conditioned behavior, while classifier guidance steers predicted states toward test-time objectives. Only actions are executed, without a separate motion-reference tracker or externally estimated full-body states as policy inputs. In simulation, PredActor reaches all 15 destination targets and achieves a text retrieval score of 0.580, compared with 0.373 for conditional action diffusion, with similar observed disturbance survival. To make this guided policy practical onboard, rolling denoising and computation-preserving runtime optimizations reduce the complete callback to 16.790 ms median and 19.383 ms p95 on a Jetson Orin NX, both below the 20 ms control period. We deploy PredActor on a Unitree G1; evaluations across simulation and physical hardware demonstrate text-conditioned motion, disturbance response, joystick control, and semantic interpolation.
[LG-4] G-NAC: Graph Neural Automata Clustering via Emergent Domain Formation
链接: https://arxiv.org/abs/2609.24823
作者: Keith Miller,Tristan Crawford
类目: Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注: 39 pages, 2 figures
Abstract:We introduce Graph Neural Automata Clustering (G-NAC), an unsupervised clustering method in which observations interact as cells on a fixed neighborhood graph. A shared recurrent graph-neural cellular rule evolves latent domain states through local interactions, which are converted into a rank-based spectral affinity for partitioning. Across 73 clustering tasks from 57 benchmark datasets, G-NAC achieved a mean adjusted Rand index (ARI) of 0.7951, comparable to Genie at 0.7941 and higher than the other evaluated baselines. Empirical training time and GPU memory scaled approximately linearly from 5,000 to 100,000 nodes. Learned transition rules also transferred from smaller source graphs to independent 100,000-node samples generated under matched conditions. These results demonstrate a recurrent graph-clustering formulation while identifying dependencies on graph quality, readout design, and source-target similarity.
[LG-5] Complex KDA: Understanding and Enhancing the Expressivity of Kimi Delta Attention
链接: https://arxiv.org/abs/2609.24797
作者: Julien Siems,Riccardo Grazzi,Korbinian Pöppel,Jaisidh Singh,Arber Zela,Timur Carstensen,Jenia Jitsev,Frank Hutter,Volkan Cevher,Antonio Orvieto,Aaron Klein
类目: Machine Learning (cs.LG)
*备注:
Abstract:Linear RNNs based on the delta-rule enable efficient sequence modeling, but their linear updates with a low-rank correction constrain their expressivity. Prior work has shown that composing two delta-rule transitions in a single recurrent update can model a 2D rotation, but this increases the rank and the cost of the updates compared to a single transition. We show that Kimi Delta Attention (KDA) can realize 2D rotations by combining a single delta-rule transformation with a second reflection supplied by its channel-wise gate. This requires extending the parameter ranges of KDA by combining two existing range extensions: allowing gates in [-1,1] and the delta-rule coefficient \beta in [0,2] . We call the resulting model Complex KDA (CKDA). It preserves KDA’s stability and efficiency, with transitions that remain diagonal-plus-rank-one and non-expansive, while reaching the state-tracking expressivity of DeltaProduct _2 . We characterize the expressivity of CKDA and prove that every orthogonal diagonal-plus-rank-one matrix is exactly a CKDA transition matrix. A single CKDA layer can track every finite group isomorphic to a subgroup of \mathrmSO(3) , and many state-tracking results use one fewer layer for CKDA compared to other diagonal-plus-rank-one Linear RNNs. Empirically, combining both extensions yields the strongest length extrapolation among tested KDA range settings on S_3 , S_4 , and periodic audio continuation. In language modeling, CKDA outperforms Transformers and other linear RNNs, obtains similar results to a KDA baseline, and shows promising scaling behavior. Our code is open source at this https URL and our models are available at this https URL.
[LG-6] Inference of Unknown Dynamical Components Using Next Generation Reservoir Computing: From Chaotic Systems to Climate Data
链接: https://arxiv.org/abs/2609.24754
作者: Jule Budnick,Andrew Keane,Serhiy Yanchuk
类目: Machine Learning (cs.LG); Chaotic Dynamics (nlin.CD)
*备注:
Abstract:We investigate next generation reservoir computing (NGRC) as a data-driven approach for inferring unseen components of dynamical systems. We compare NGRC with traditional reservoir computing (RC) using the Lorenz and Rössler system, where two unknown components are inferred from one given component. For both systems, NGRC achieves accurate results while requiring fewer training data and less computational time than RC. We identified an inverse proportional behavior between the number of time-delayed steps needed for NGRC and the temporal resolution, indicating that the physical time span covered by the delay interval is an important factor in determining the required number of delayed steps. Finally, we apply NGRC to the observational climate data of ENSO (El Niño–Southern Oscillation) and infer one observable from the remaining variables. Despite the noise and complexity of the real-world data, the NGRC shows promising results. Our findings demonstrate the potential of NGRC for efficient inference of unseen components in both controlled dynamical systems and real-world data.
[LG-7] D-JEPA: A Decision-Aligned Latent World Model
链接: https://arxiv.org/abs/2609.24749
作者: Shuaijun Liu,Chengyu Wu,Qifu Wen,Feiyang You,Chenglong Zhang,Shuyang Hao,Xi Lin,Ningxin Su
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: 26 pages, including references and appendices. Project website: this https URL
Abstract:Latent world models predict the consequences of actions, but accurate prediction does not guarantee that latent distance reflects which candidate will execute successfully. We identify a decision-local prediction gap: among the few futures competing for execution, a candidate predicted closer to the goal can produce a worse realized outcome than an available alternative. We introduce D-JEPA, a decision-aligned latent world model that learns decision-relevant relations among candidate futures from executed outcomes. A bounded, permutation-equivariant operator jointly reasons over goal-relative predictive features and ordinal evidence, refining pretrained predictive geometry where action choices are most consequential. Restricted predictor adaptation and a shared ordinal interface extend this alignment across complementary predictive geometries. D-JEPA further realizes the learned decision structure in JEPA-compatible future representations, enabling deployment through native latent-distance planning. Evaluations across latent control, manipulation, pretrained action-producing models, physical robots and autonomous driving demonstrate improved action selection, including 87.89% success on PushT, a 15.04-point average gain on RoboTwin, and a 17-point gain on physical robot tasks. These results establish decision-relevant relational structure as a direct bridge between predictive world modeling and effective control.
[LG-8] An Exact Junction-Tree Extended Formulation for Optimal Classification Trees
链接: https://arxiv.org/abs/2609.24741
作者: Jiancheng TU,WenqiFan
类目: Machine Learning (cs.LG)
*备注:
Abstract:We develop an exact linear programming (LP) formulation for bounded-depth classification trees with binary features, using a junction-tree representation. The formulation is integral and supports recursive subtree optimization. Exact reductions make the model smaller while preserving the optimal value and recovery of an optimal tree. The reduced model supports two solution methods: column generation and message passing. Column generation solves integral restricted LPs and uses bounds over the full feasible domain to certify optimality. Message passing recursively combines optimal subtree costs. Both methods solve common subtree problems that, once the preceding tree decisions are fixed, can be evaluated independently and in parallel. Computational experiments show that the exact reductions substantially reduce the size of the junction-tree formulation. The resulting linear programming formulation certifies instances for which the tested mixed-integer formulation does not establish optimality within the same computational budget, while the column-generation and message-passing methods certify more instances and achieve an order-of-magnitude reduction in geometric-mean runtime relative to an existing state-of-the-art exact method for optimal classification trees.
[LG-9] A Federated Artificial Intelligence Framework for Optimizing Pancreatic Cancer Treatment - Strategy Update
链接: https://arxiv.org/abs/2609.24718
作者: Anne-Christin Hauschild,Amirreza Aleyasin,Nils H. Beyer,Lisa Fricke,Jonas Hügel,Maryam Moradpour,Anh-Tien Nguyen,Youngjun Park,Sophia Rheinländer,Tim Beissbarth,Elisabeth Hessmann,Martin Middeke,Matthias Lauth,Maximilian Reichert,Ulrich Sax
类目: Machine Learning (cs.LG)
*备注: 11 pages, 2 figures, 1 table
Abstract:While a centralized approach involving patient consent to collect and analyze data centrally would theoretically offer the best data quality and predictive performance, it is not always feasible in practice. Federated Learning (FL) architectures have shown to be a very promising approach to use and access distributed disease related resources within the GDPR boundaries. In a previous case report, we described the preconditions at the participating sites and necessary administrative and process related steps to prepare data, people and infrastructure for improving subtype identification and assessing treatment options in pancreatic cancer. We update this report sharing our experience in tackling the challenges and show preliminary results of the actual federated learning AI pipelines. At the participating sites, we have to identify and annotate the data being accessible after extraction and transformation in a local FL hub - in our case a centrally developed and distributively deployed Docker container. This container comprises the FL scripts generating local models. We apply a newly developed FL algorithm considering all local features, including partial overlapping features specific to the local sites. Theoretically, an annotation in a cancer setting should succeed using the German oncology core data set (oBDS), which is already utilized for mandatory reporting to cancer registries, and can be sustained in the FL setting. The FL algorithms deal robustly with partially overlapping features as we showed with public data sets. Major roadblocks including straightening operational concepts for the infrastructures, ethics approval for such novel architectures and support for every site have been addressed. However, scaling up this approach in the future faces hurdles; while including broader multi-modal data sets should be feasible, large-scale deployment to more sites remains challenging.
[LG-10] Offline Reinforcement Learning for Distribution-Grid Protection
链接: https://arxiv.org/abs/2609.24703
作者: Julian Oelhaf,Alexander Luce,Christian Bergler,Andreas Maier,Siming Bayer
类目: ystems and Control (eess.SY); Machine Learning (cs.LG)
*备注: Accepted for presentation at the IEEE Power Energy Student Summit (PESS 2026), Karlsruhe, Germany. 6 pages, 2 figures. Code: this https URL
Abstract:Data-driven protection may complement conventional relays in distribution grids whose operating conditions vary with distributed generation, switching events, and changing short-circuit levels. We study line-selective tripping from static trajectories of a realistically simulated CIGRE medium-voltage network using offline reinforcement learning. A convolutional Q-network receives causal voltage-current phasor and apparent-impedance features, optionally together with raw waveforms, and is trained with conservative Q-learning (CQL). A controlled sensitivity study evaluates two observation windows, reward variants, and three CQL weights under a common split and training protocol; one exploratory post-hoc run additionally increases the discount factor from \gamma =0.95 to 0.99. On 225 held-out episodes, the best per-timestep result is obtained with combined input and CQL weight \alpha =0.9, reaching precision 0.9993, recall 0.9496, and F1-score 0.9738. Because dense per-timestep scores do not encode the terminal semantics of relay operation, we also evaluate the first non-wait action in each episode. The default combined-input agent selects the correct line-trip action first in 98.13% of 214 fault episodes, but trips in 72.73% of the 11 non-fault episodes. In the post-hoc run, the corresponding rates are 98.60% and 54.55%, respectively. The results show that dense predictive performance and terminal protection behavior can lead to different model rankings. Offline CQL therefore demonstrates strong faulted-line selection on the simulated fault episodes, while the static trajectories, small non-fault set, and single-seed post-hoc design preclude conclusions about practical relay security or deployment readiness.
[LG-11] Guaranteed Low-Rank Tensor Recovery from Modewise Measurements via Normalized Block-Weighted Riemannian Gradient Descent
链接: https://arxiv.org/abs/2609.24679
作者: Yushi Zhou,Feng Zhang
类目: Machine Learning (cs.LG)
*备注:
Abstract:We consider the recovery of low-multilinear-rank tensors from linear measurements and propose an adaptive block-weighted modewise Riemannian gradient descent method. The method combines memory-efficient modewise measurements with a normalized adaptive weighting strategy for the core and factor components of the Riemannian gradient. The weighting improves convergence without increasing the multilinear-rank bound of the search direction or the size of the reduced core used for retraction. Under the tensor restricted isometry property and a suitable initialization, we establish local linear convergence and derive sampling guarantees for sub-Gaussian and subsampled orthogonal with random sign (SORS) measurements. Numerical experiments on synthetic low-Tucker-rank tensors show that the proposed method reduces iteration counts and computational time while maintaining reliable recovery performance, especially near the recovery threshold and for structured SORS measurements.
[LG-12] Learning tactile perception from high-bandwidth single-point sensing
链接: https://arxiv.org/abs/2609.24621
作者: Joseph Rigal,Emmanuel Virot,Caroline Pascal
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注:
Abstract:Tactile sensing is increasingly being incorporated into learning-based robotic manipulation, yet many existing approaches rely on spatially distributed sensors. Here we introduce SpectRobot, a framework that transforms single-point tactile signals into compact time-frequency spectrograms. These spectrograms encode high-bandwidth tactile histories as fixed-size image-like representations. They can be processed by standard vision encoders and integrated into learning pipelines originally developed for vision, while preserving temporal and frequency information unavailable to conventional cameras. Rather than increasing spatial density through arrays of tactile elements, SpectRobot exploits the rich dynamics contained in sparse, high-bandwidth single-point measurements. In our implementation, the sensors are mounted away from the contact surface while remaining mechanically coupled to it, reducing direct exposure to wear and potentially improving robustness in harsh environments and for long-term deployment on dexterous robots. Our experiments demonstrate that: (1) a robot can exploit single-point vibration signals to solve a visually occluded manipulation task; (2) temporal history strongly influences policy performance, while sensing bandwidth controls the spectral information available, with measurements extending to 100~kHz; and (3) the same representation can be used across different tactile sensing technologies mediated by acceleration, force, or strain. We further show that capabilities previously associated with research-grade instrumentation can be accessed using readily available, off-the-shelf hardware. We believe that broader access to high-bandwidth tactile sensing could facilitate the integration of contact dynamics into embodied learning systems and, for some tasks, offer an alternative or complement to increasing the spatial density of tactile sensing. Subjects: Robotics (cs.RO); Machine Learning (cs.LG) Cite as: arXiv:2609.24621 [cs.RO] (or arXiv:2609.24621v1 [cs.RO] for this version) https://doi.org/10.48550/arXiv.2609.24621 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-13] aking a Second Look: Correcting Sea Ice Forecasts with Sparse Observations
链接: https://arxiv.org/abs/2609.24591
作者: Tianshuo Zhang,Xianglei Xing,Aowen Yang,Jia Gao,Wenzhe Zhai,ShanShan Liu
类目: Machine Learning (cs.LG)
*备注:
Abstract:Sea ice forecasts are issued several days ahead, allowing errors to accumulate while new, often sparse sea ice concentration (SIC) observations become available. We find that fixed-propagation errors concentrate near structured, high-gradient ice edges, whereas homogeneous interiors require limited propagation, suggesting that propagation distance should be state dependent. We therefore introduce ECHO (Evidence-guided Correction with Heterogeneous prOpagation), where ECHO-Scale adapts propagation distance while preserving correction geometry, and ECHO-Delta learns a bounded residual around fixed propagation. Across all 96 standard evaluation settings spanning diverse priors, observation times, sparsity levels, geometries, and noise conditions, both outperform fixed propagation. ECHO-Delta achieves the best average accuracy, while ECHO-Scale is more robust to geometry shifts. Code is available at this https URL.
[LG-14] Universal Multi-Modal Traceformer: Integrating Heterogeneous Context for Process Event Prediction
链接: https://arxiv.org/abs/2609.24579
作者: Fabian Spaeh,Jingxing Fang,Shandian Zhe,Bin Shen
类目: Machine Learning (cs.LG)
*备注:
Abstract:Event logs arise in a wide range of real-world processes, capturing not only event activities and timestamps but also multi-modal contextual information. Existing event-sequence models, including many temporal point process approaches, primarily model event activities and timestamps while overlooking heterogeneous context, such as numerical measurements, categorical attributes, textual descriptions, and metadata associated with individual events and entire traces. In this paper, we propose Universal Multi-Modal Traceformer (UMT), a unified framework for incorporating heterogeneous process context into next-event prediction. Built on a Transformer backbone, UMT introduces a universal feature encoder that maps diverse feature types into a shared representation space and handles contextual information at both the event and trace levels. UMT further develops a per-event Perceiver module that dynamically weights contextual features and adaptively integrates them into event-token representations. To accommodate the heavy-tailed and potentially multi-modal distribution of inter-arrival times, UMT represents each interval at multiple temporal scales and jointly predicts the corresponding scale-specific quantities. Experiments on 13 real-world event logs show that UMT improves both next-event activity and time prediction over existing approaches.
[LG-15] Poisson Exchange Beyond Submodularity: Effective Approximation Algorithms for Offline and Online Subset Selection over Matroids
链接: https://arxiv.org/abs/2609.24569
作者: Shi Fu,Youming Qiao,Dacheng Tao,Zongqi Wan,Qixin Zhang
类目: Data Structures and Algorithms (cs.DS); Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注: 55 pages
Abstract:Over the past decade, a growing body of research has shown that \gamma -weak submodularity broadly arises in numerous subset selection tasks, including feature selection, neural network pruning, and video summarization. Despite its prevalence, maximizing a \gamma -weakly submodular function subject to a general matroid constraint remains challenging. To date, the only known approximation guarantee is the conservative (1+1/\gamma)^-2 factor established by \citetchen2018weakly. To improve upon this result, this paper proposes a novel algorithm called \MGPE, which repeatedly performs maximum-gain local exchanges through careful control of a non-homogeneous Poisson clock, and proves that this \MGPE\ can attain an approximation ratio arbitrarily close to \rho_\gamma=1-\left(\gamma/(2-\gamma)\right)^ \frac\gamma^22(1-\gamma) . In sharp contrast to the previous guarantee, our obtained factor \rho_\gamma not only strictly improves upon (1+1/\gamma)^-2 for every \gamma\in(0,1] , but also can asymptotically approach the optimal (1-1/e) -approximation for submodular maximization as \gamma\to1 . Furthermore, we surprisingly find that when the matroid constraint reduces to a cardinality or the objective satisfies the stronger notion of \alpha -weak DR-submodularity, \MGPE\ can automatically recover the tight approximation ratios of 1-e^-\gamma and 1-e^-\alpha , respectively. Here, \alpha\in(0,1] denotes the DR ratio.
[LG-16] A Temporal Knowledge Graph for Music Festival Lineup Forecasting
链接: https://arxiv.org/abs/2609.24467
作者: Julia Gastinger,Thilo Dieing,Christian Meilicke,Heiner Stuckenschmidt
类目: Machine Learning (cs.LG)
*备注: Accepted to 11th Workshop on Automated Knowledge Base Construction (AKBC) 2026
Abstract:Music festival lineups emerge from complex relationships among artists, genres, releases, labels, and past performances, making the prediction of future lineups a natural fit for temporal knowledge graph (TKG) forecasting. In this work, we present a TKG covering 380 festivals over 55 years, comprising more than 90K festival performance quadruples along with information on festivals, artist tours, and artist metadata, and release it as a resource for TKG forecasting evaluation. We formalize festival lineup forecasting as temporal link prediction between artists and festivals at future timestamps. We evaluate six TKG forecasting models on this task, analyze their capabilities and limitations, and compare them against Large Language Models applied zero-shot. Our resource complements existing TKG benchmarks by grounding evaluation in a concrete, real-world application domain.
[LG-17] RAILS: Retrieval-Augmented Incremental LLM Clustering at Scale
链接: https://arxiv.org/abs/2609.24464
作者: Armin Oliya,Aleksandra Sawczuk,Radosław Białobrzeski
类目: Machine Learning (cs.LG)
*备注:
Abstract:Using a Large Language Model (LLM) as the clusterer at production scale is hard: prompts cannot hold the entire label space, and per-document serial processing does not deliver the throughput real workloads require. We present RAILS, a retrieval-augmented incremental LLM clusterer that turns clustering into a simple loop over a growing label pool and scales through document batching with bounded concurrency. On six public benchmarks RAILS exceeds the strongest prior LLM-clustering method on average, lifting accuracy from 51.2% to 59.3%, NMI from 67.2% to 74.8%, and ARI from 45.4% to 54.7%. We further report production-deployment evidence from a SaaS ticket-topic-discovery pipeline, where RAILS has replaced a traditional HDBSCAN stage with higher clustering quality, transparent prompt-driven control, and stateful incremental operation.
[LG-18] MUSE: Dependency-Aware Adaptation of a Frozen Vision Backbone for Multivariate Time Series Forecasting
链接: https://arxiv.org/abs/2609.24441
作者: Xinying Cai,Junkai Lu,Yuhan Zhu,Xiaoyun Yu,Xiangfei Qiu,Jilin Hu
类目: Machine Learning (cs.LG)
*备注:
Abstract:Multivariate time-series forecasting is essential to many real-world applications. Recent large vision models (LVMs) offer a promising paradigm by transferring cross-domain visual priors to time-series forecasting. However, existing LVM-based methods face two key challenges: balancing independent visual representation spaces with cross-variable dependency modeling, and adapting vision backbones pretrained on natural images to the distinct temporal semantics of time-series images. To address these challenges, we propose MUSE, a dependency-aware adaptation framework built on a fully frozen pretrained MAE. First, the Variable Context Refinement Module (VCR) aggregates shared temporal information within each variable and models cross-variable contextual dependencies while preserving independent visual spaces. Second, the Temporal-Periodic Refinement Module (TPR) performs lightweight refinement at different encoder depths and explicitly models across-period temporal dependencies and within-period periodic dependencies. The two modules independently produce forecasts, which are fused through a learnable prediction-level gate. Experiments on 10 real-world datasets demonstrate that MUSE achieves state-of-the-art performance.
[LG-19] Comparing Latent Concept Formation in State Space Models and Transformers via Sparse Autoencoders
链接: https://arxiv.org/abs/2609.24440
作者: Rithin Nagaraj,Rupa Laalasa Oruganti,Prerna Subhashchandra Kunder,Ashwini M Joshi
类目: Machine Learning (cs.LG)
*备注: Contains about 6 pages
Abstract:The quadratic scaling of Transformer self-attention has driven the adoption of sub-quadratic Selective State Space Models (SSMs) like Mamba, which compress past context into a fixed-size recurrent hidden state. This strict informational bottleneck raises a foundational question for mechanistic interpretability: do SSMs and Transformers learn fundamentally distinct latent representations? In this work, we employ Sparse Autoencoders (SAEs) to conduct a large-scale, feature-level correspondence analysis between Mamba-130m and Pythia-70m over a 10-million token corpus. Contrary to hypotheses predicting widespread architectural divergence, we find no evidence of systematic representational divergence between architectures: across the observed Jaccard distribution, 99.98% of Mamba features cluster toward the upper alignment boundary, providing preliminary feature-level support for the Universality Hypothesis. We further identify and qualitatively characterize this microscopic fraction (0.02%) of diverging features, finding patterns consistent with the hypothesis that the recurrent bottleneck selectively limits the parsing of rigid syntax rather than broad semantic ontology. We demonstrate that while Pythia’s unconstrained attention permits the monosemantic decomposition of distinct formatting edge-cases, Mamba is forced to compress unrelated syntactical anomalies into polysemantic “junk drawer” neurons to preserve state capacity. Collectively, these results suggest that architectural routing mechanisms may have negligible impact on core semantic understanding, with representational divergence confined to extreme structural margins.
[LG-20] Prior-Amortized In-Context Bayesian Inference for Generalized Linear Mixed-Effects Models
链接: https://arxiv.org/abs/2609.24422
作者: Alex Kipnis,Marcel Binz,Eric Schulz
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:Hierarchical data is ubiquitous in the empirical sciences and is most commonly analyzed with generalized linear mixed-effects models (GLMMs). Bayesian inference for GLMMs yields calibrated uncertainty but requires MCMC; the No-U-Turn Sampler (NUTS) is the gold standard but is slow and must restart from scratch for every new dataset, model and prior. We introduce metabeta, a pretrained neural network for prior-amortized in-context Bayesian inference over GLMMs. Unlike previous neural posterior estimators that fix the prior at training time, metabeta accepts prior families and hyperparameters as inputs at test time, enabling zero-shot generalization. Two set transformers and conditional normalizing flows mirror the posterior’s two-level structure (global parameters shared across groups, local parameters per group). The model is trained on millions of realistic simulated datasets spanning continuous, binary, and count outcomes. By default, the flow posterior is refined by Independence Metropolis-Hastings against the unnormalized posterior, so its correctness rests on the sampler rather than the network; this yields tuning-free inference two to three orders of magnitude faster than NUTS. Alternatively, the flow can warm-start NUTS, giving nearly identical inference with substantially increased speed and stability. On controlled benchmarks with ground-truth parameters, metabeta matches NUTS in parameter recovery, calibration and out-of-sample prediction. On out-of-distribution real datasets, its posteriors closely match those of NUTS across all parameter types, and they remain faithful under misspecified likelihoods and priors, out-of-distribution predictors, collinear designs, and data-poor regimes. The model is open-source and open-weights and thus immediately deployable.
[LG-21] Probabilistic Modelling of Operational Design Domains A New Approach for Testing AI Systems
链接: https://arxiv.org/abs/2609.24397
作者: Hans-Werner Wiesbrock
类目: Machine Learning (cs.LG)
*备注: 50 pages, 43 figures, Technical Report
Abstract:The conventional testing process quickly fails when applied to ML-based systems such as obstacle detection in vehicles: if an obstacle is not detected in a test, classical bug fixing is impossible and an AI system will always retain shortcomings. Test results can therefore only be interpreted statistically, which in turn requires test sets that are not only complete with respect to the operational design domain (ODD) of the system, but also representative of it. To this end, we introduce probabilistically extended ontologies (PEONs): ontologies describing the ODD, augmented with a probability distribution over the partitioning they induce. Instead of unmaintainable conditional probability tables, only marginal distributions and functionally described dependencies need to be specified; algorithms based on couplings and optimal transport complete this specification to a Bayesian network. From a PEON we derive the sampling of representative test cases, rigorous end-of-test criteria for given quality targets and significance levels, and methods for re-evaluating existing test results and for assessing the balance of training data. We demonstrate the practical modelling of a complex ODD using the example of automatic train operation.
[LG-22] Climate Variability Modulates the Impact of Price Spikes on Food Insecurity
链接: https://arxiv.org/abs/2609.24394
作者: Jordi Cerdà-Bautista,Vasileios Sitokonstantinou,Homer Durand,Gherardo Varando,Michele Ronco,Gustau Camps-Valls
类目: Machine Learning (cs.LG)
*备注:
Abstract:Climate variability influences whether a market disruption escalates into a food crisis, yet broad climate patterns like El Niño, tracked months before they alter hydro-climatic conditions, are still not incorporated as an early-warning component in food-security responses. We address this gap by introducing sensitivity regimes, a stratification of regions by the direction and strength of their vegetation response to the El Niño Southern Oscillation, and using them to estimate how food price spikes affect acute food insecurity across sub-Saharan Africa. Integrating remote sensing, socioeconomic data, and causal machine learning, we find that in regions where ENSO systematically suppresses vegetation, a price spike raises the share of the population at acute risk by 5.4 percentage points in the following month. In regions where vegetation is unaffected by or positively linked to ENSO, the estimated effect is smaller (around 2 percentage points) and statistically insignificant. These results demonstrate that climate context is critical for understanding food security vulnerabilities. Sensitivity regimes can be combined with operational price-spike triggers to stage anticipatory action: the ENSO state flags vulnerable regions months ahead, and a pre-positioned response in those regions to a price spike would avert the largest jump in acute food insecurity.
[LG-23] NAVIR: Neuromorphic Audio-Visual Speech Recognition for Robust Human-Robot Interaction on Edge Hardware
链接: https://arxiv.org/abs/2609.24391
作者: Leonidas Delimpasis,Panagiota Moraiti,Antonis Porichis,Panos Chatzakos,Michail Karamousadakis
类目: Machine Learning (cs.LG)
*备注:
Abstract:Voice-controlled interaction in industrial settings is hampered by acoustic noise, which severely degrades audio-only speech recognition. Audio-visual speech recognition (AVSR) addresses this by fusing lip-motion cues with the audio stream, but state-of-the-art pipelines rely on three-dimensional convolutions, recurrent units, and attention modules that exceed the budget of typical edge devices. We present NAVIR, an end-to-end AVSR system targeting the BrainChip Akida neuromorphic processor, which natively supports only sequential two-dimensional convolutional inference. The pipeline factorises spatial and temporal encoding into separate AkidaNet-based modules: a per-frame visual encoder, a temporal video encoder, and a spectrogram audio encoder, fused by a lightweight predictor head and decoded by constrained beam search. Models are trained with connectionist temporal classification on noise-augmented audio and then fine-tuned with quantization-aware training. On the GRID benchmark, the quantized audio-visual model reaches 14.0% word error rate (WER) under noise on the unseen-speaker split and 3.3% WER on the overlapped-speaker split, against 22.5% and 11.8% for audio-only baselines, and it attains 98.6% command accuracy at 1.5% WER on a task-specific industrial-command corpus. Operation-count analysis indicates a 13-fold energy advantage of the spiking formulation over its artificial neural network counterpart at 27.6% mean firing rate. On-board measurements show roughly 5-fold lower energy per inference than a Raspberry Pi central processing unit on the lip-reading model, and over 100-fold lower than a laptop graphics processing unit, while sustaining 14.5 inferences per second. To the best of our knowledge, this is the first complete multimodal AVSR pipeline running on neuromorphic hardware of this class.
[LG-24] Machine Learning-Based Prediction of Childhood Stunting in Bangladesh: Fairness and Temporal Robustness Assessment SDM
链接: https://arxiv.org/abs/2609.24386
作者: Md Ahshanul Haque,Muhammad Ashad Kabir
类目: Machine Learning (cs.LG)
*备注: Accepted to AusDM, 15 pages
Abstract:Childhood stunting remains a major public health concern in Bangladesh and reflects long-term growth failure influenced by child, maternal, household, socioeconomic, and health-service factors. This study used nationally representative Bangladesh Demographic and Health Survey data from 2007 to 2022 to develop machine learning models for population-level prediction of childhood stunting and to assess temporal robustness and subgroup fairness. Children aged 0-59 months with complete anthropometric and predictor data were included. Data from the 2007, 2011, and 2014 survey rounds were used for model development, while the 2018 and 2022 rounds were retained as temporal test datasets. Twelve feature-selection approaches were assessed, and the KNN permutation importance-selected predictor set was used for final model evaluation. Eleven machine learning models were evaluated: ten conventional algorithms and one pretrained tabular foundation model, TabPFN. Performance was assessed using balanced accuracy, AUROC, F1-score, Brier score, and expected calibration error. Subgroup fairness was examined by child sex, place of residence, and socioeconomic status. The final analytic sample included 18,844 children, of whom 35.05% were stunted. In the development hold-out test dataset, TabPFN showed the highest observed balanced accuracy overall at 67.58%, while AdaBoost showed the highest observed balanced accuracy among conventional models at 67.51%. In temporal testing, the highest observed balanced accuracy was found for Gradient Boosting in BDHS 2018 and XGBoost in BDHS 2022. Model performance varied across survey rounds and subgroups, highlighting the importance of temporal validation, subgroup fairness assessment, and transparent interpretation in public health prediction modeling.
[LG-25] Credit Access is Associated with Improved Food Security in the Horn of Africa
链接: https://arxiv.org/abs/2609.24382
作者: Jordi Cerdà-Bautista,Vasileios Sitokonstantinou,José Manuel Veiga López-Peña,Duccio Piovani,José María Tárraga,Gustau Camps-Valls
类目: Machine Learning (cs.LG)
*备注:
Abstract:The intensification of climate change poses a growing threat to food security, especially in vulnerable communities. This study employs an observational machine-learning framework to estimate the causal association between access to credit and acute food insecurity in Somalia and across the Horn of Africa, drawing on a harmonized dataset spanning key environmental, socioeconomic, and conflict-related factors from 2015 to 2022. Results indicate that greater credit access is associated with a 2% reduction in acute food insecurity at the population level over the study period. Given that, on average, 16% of the population is in crisis, this effect represents a meaningful shift within the at-risk group. We interpret these estimates under explicit identification assumptions and complement them with robustness and refutation tests. The results provide context-specific evidence on how financial access correlates with food security outcomes in data-scarce, crisis-affected settings, and offer a transparent framework for integrating heterogeneous data sources when randomized evaluations are infeasible.
[LG-26] On the Information-Theoretic Limits of Latent-Space Watermarking Through Pretrained Generators
链接: https://arxiv.org/abs/2609.24377
作者: Jinwan Jeon,Minju Lee,Sung Hoon Lim
类目: Information Theory (cs.IT); Machine Learning (cs.LG)
*备注: Submitted to the IEEE Transactions on Information Theory for possible publication
Abstract:We study latent-space watermarking through a pretrained generator using a prescribed latent-to-output stochastic mapping, called the renderer. A watermark encoder selects the latent input using a message and secret key. For every message and semantic context, the released output must have exactly the desired conditional output distribution. For finite alphabets, we derive rate–key inner and outer bounds and characterize the coding and coordination requirements for realizing watermark communication through the prescribed latent interface. When the target output distribution of the generator uniquely determines the corresponding latent input distribution through the renderer, a strengthened converse yields the capacity region; the same region governs explicit preservation of the pretrained latent distribution. We extend the analysis to general jointly Gaussian models and identify a sufficient statistic of the latent that captures both the watermark-bearing information available at the generated output and the latent coordination required to preserve its target distribution. For the vector Gaussian model, we further characterize the optimal allocation of the secret-key resource across the resulting modes. Finally, we turn to an emerging robustness threat that is particularly natural in generative watermarking: an adversary can regenerate the released sample to obtain a fresh realization of the same underlying content while attenuating or destroying the embedded watermark. We incorporate this robustness axis into our framework and characterize the one-pass compound capacity of the scalar Gaussian model when the semantic context is known to the encoder but hidden from the detector, while the regeneration attack may depend on that context. Extending the analysis to multiple rounds of repeated canonical regeneration, we characterize the resulting watermark-capacity decay.
[LG-27] Explainable Neuro-Fuzzy Prediction for Trustworthy Decision-Making in Maritime
链接: https://arxiv.org/abs/2609.24358
作者: Dionisis Kalogeropoulos,Georgia Sovatzidi,Dimitris K. Iakovidis
类目: Machine Learning (cs.LG)
*备注: Accepted at the 34th European Signal Processing Conference (EUSIPCO 2026)
Abstract:Predicting when maritime systems require maintenance can be critical, avoiding hazards and costly consequences. To address this problem, this paper proposes an explainable decision-making framework that integrates a neuro-fuzzy prediction model with a two-stage explainable component. The first stage of this component produces feature-attribution explanations, using gradient-based saliency maps, and the second stage extracts local rules using a fuzzy decision tree. The proposed framework is generic and can be integrated into any deep learning-based approach, rendering it explainable. To the best of our knowledge, this is the first fuzzy logic-based framework enabling both feature-level and local rule-based explanations of black box models. This approach aims to foster trustworthiness in decision making through user-understandable machine inferences. The performance of the proposed framework using a deep residual-based neural backbone is evaluated on various general-purpose public benchmark datasets, and its utility in maritime is demonstrated in the context of early fault detection in a naval propulsion system dataset. The results indicate that it can provide predictions outperforming relevant state-of-the-art approaches, with an average AUC-ROC (Area Under the Receiver Operating Characteristic Curve) value, reaching up to 99%, while offering the advantage of explainability.
[LG-28] Pharmacokinetic State Space Models for Unbiased Prediction of Haemodynamic Collapse ALT
链接: https://arxiv.org/abs/2609.24338
作者: Rithin Nagaraj,Sudiksha Chindula,Bhaskarjyoti Das
类目: Machine Learning (cs.LG)
*备注: Published in Springer Nature after presenting at the International Conference on AI in Healthcare, London
Abstract:An Intraoperative Hypotension (IOH) event is a frequent complication during administration of general anaesthesia with serious downstream consequences, yet clinical management remains reactive and not predictive. Existing predictive models, however, ignore drug infusion history as a valuable signal for prediction despite its direct pharmacological relevance. Our model achieves an Area Under the Receiver Operating Characteristic curve (AUROC) of 0.7360 and an Area Under the Precision-Recall Curve (AUPRC) of 0.1794, representing a 2.73-fold lift over the random guessing AUPRC baseline (0.0657), with the removal of propofol and remifentanil effect-site concentrations resulting in a 13.9% AUPRC drop compared to the full model. This is consistent with the hypothesis that pharmacokinetic trajectories encode impending haemodynamic changes before they manifest in the Mean Arterial Pressure (MAP). Additionally, this paper shows that training without lead-gap filtering degraded AUROC by 16.7%, empirically confirming that unfiltered models learn to detect ongoing hypotension rather than predict future events. Finally, a Mamba-based architecture achieves the aforementioned high prediction performance while maintaining a constant memory footprint across a range of sequence lengths, unlike the quadratic VRAM overhead typical of vanilla Transformers, making it the more practical choice for continuous intraoperative deployment.
[LG-29] A Distributional Optimisation Perspective on Combining Models in Deep Learning
链接: https://arxiv.org/abs/2609.24328
作者: Congye Wang,Yan Lin,Zheyang Shen,Matthew A. Fisher,Chris. J. Oates
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:Combining predictions from different models can improve performance at machine learning tasks, but the training of the individual models and the rule used to combine them are typically chosen separately, and by ad hoc means. Recent advances in distributional optimisation (i.e. where the optimisation occurs over the set of probability distributions) offer an opportunity for principled joint training, viewing the collection of models as a discrete distribution whose support points are to be optimised, but the potential of these methods is not well-understood. In this paper we (1) cast two standard combination strategies - ensembles and low-rank adapter averaging - as entropy-regularised distributional optimisation, observing that the resulting objective is convex in the ensemble case but not in the adapter-averaging case, so that existing convergence guarantees for mean field Langevin dynamics transfer only to the former; (2) assess existing and novel algorithms for this task, including a functional variant of variational gradient descent; and (3) report an empirical study spanning synthetic classification tasks and fine-tuning of large language models on a commonsense reasoning benchmark.
[LG-30] High-Dimensional Online Change Point Detection with Adaptive Thresholding and Interpretability
链接: https://arxiv.org/abs/2609.24278
作者: Sven Jacob,Bardh Prenkaj,Weijia Shao,Gjergji Kasneci
类目: Machine Learning (cs.LG)
*备注: Published at TMLR, 2026
Abstract:Change point detection (CPD) identifies abrupt and significant changes in sequential data, with applications in human activity recognition, financial markets, cybersecurity, manufacturing, and autonomous systems. Traditional CPD methods often face computational challenges in high-dimensional settings and typically provide limited explanations for detected changes, which can restrict their practical usability. This paper introduces a CPD framework that improves scalability and interpretability by leveraging the Sliced Wasserstein (SW) distance. Our contributions are fourfold: (1) we transform multivariate sequential data into one-dimensional scores using the SW distance, making the resulting representation compatible with existing CPD methods; (2) we analyze the distributional behavior of random slices of the SW distance and show that, under suitable assumptions, they can be approximated by a Gamma distribution, providing a principled basis for threshold calibration; (3) we propose a self-adapting online CPD algorithm that combines this SW-based score with an adaptive quantile-based threshold; (4) we introduce a model-specific framework for generating contrastive explanations for annotated change points. Empirically, our method reduces false positives by at least 48% on average compared with popular online and offline CPD baselines, while maintaining competitive or superior detection performance. Code is available at this https URL. At the same time, it produces interpretable change-point annotations, making it practical for deployment in high-stakes applications.
[LG-31] Explainable Predictive Condition-based Maintenance of Naval-Propulsion Systems using Fuzzy Logic
链接: https://arxiv.org/abs/2609.24250
作者: Dionisis Kalogeropoulos,Georgia Sovatzidi,Panagiotis G. Kalozoumis,Dimitris K. Iakovidis
类目: Machine Learning (cs.LG)
*备注: Accepted at the 30th Pan-Hellenic Conference on Informatics (PCI 2026)
Abstract:The shipping industry has a significant impact on the global economy, emphasizing the need for operational availability and safety through the use of effective maintenance techniques. During the last decades, predictive maintenance (PdM) has emerged as a promising solution compared to the existing conventional maintenance systems. This is because it offers several advantageous functions, such as damage predictions for vessel components, reduced downtime, improved and extended life of machinery, as well as higher safety during voyages. However, existing methodologies developed for performing PdM do not provide explanations of their results to users, so that they can understand the failures that may occur. To address this limitation, this paper proposes a novel framework based on a fuzzy decision tree and a deep residual neural network, aiming to perform explainable PdM on naval vessels. The proposed framework is able to generate fuzzy local rules based on the dataset used, and can provide explanations of its outcomes, using cause-and-effect relationships, in a way that are understandable to users, thereby gaining their trust. Experiments using a publicly available dataset demonstrate the effectiveness of the proposed framework, as it achieves an accuracy of 99.24%.
[LG-32] Displacement Geometry Captures Platonic Shared Reality Across Models and Modalities
链接: https://arxiv.org/abs/2609.24209
作者: Chenming Shang,Yujin Tang,Jun Jie Ou Yang,Ruize Xu,Adam Breuer,Nikhil Singh
类目: Machine Learning (cs.LG)
*备注:
Abstract:The Platonic Representation Hypothesis (PRH) claims that independently trained models converge on a shared statistical model of reality, yet recent work finds only weak pointwise similarity between models. In this paper, we show that what models share is not the location of samples in representation space, but the directions (displacement vectors) between them. Under a single orthogonal alignment–rotation and reflection only–these displacement vectors are substantially preserved across 44 independently trained vision and language encoders spanning modalities and asymmetric capability pairs, consistent with the PRH evidence. The samples’ absolute positions are not, consistent with recent counter-evidence. Both arise from a single decomposition: representations split into a shared semantic component that is linearly aligned across models, and a private capability component that is not. We trace this geometry to concept-level structure: within a model, parent concepts are orthogonal to their child variation vectors; across models, concept displacements are parallel. Our theory falsifiably predicts (and experiments confirm) that fine-tuning preserves pointwise similarity but collapses displacement, and that relational distillation does the opposite. A major implication is that, because semantics align linearly but capabilities do not, capabilities can be imported from one model to another using a single cached forward pass through the source. We call this Shadow Casting. As a proof of concept, our SHADOWCLIP instantiation outperforms strong fine-tuned baselines at orders of magnitude less compute. A cache can be released alongside open model weights, letting one model’s capabilities be downloaded and imported into any number of other models without fine-tuning. Subjects: Machine Learning (cs.LG) Cite as: arXiv:2609.24209 [cs.LG] (or arXiv:2609.24209v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2609.24209 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-33] A principled approach for energy-efficient training via phase-aware GPU frequency tuning
链接: https://arxiv.org/abs/2609.24205
作者: Miguel Braga,Júlio Pinto,Rahma Nouaji,Olivier Michaud,Bettina Kemme,Oana Balmau,Cláudia Brito,Ricardo Macedo
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注:
Abstract:Modern AI model training imposes unprecedented computational demands, making it a key contributor to datacenter energy consumption. Yet a significant fraction of the energy consumed during training does not translate to useful computation due to bottlenecks throughout the training pipeline. We present PAFT, a phase-aware, dynamically adaptable GPU frequency tuning system that reduces energy consumption of training workloads with minimal performance overhead. The key insight behind PAFT is that bottlenecks represent an energy optimization opportunity, rather than purely a performance problem: when GPUs are bound to stall, PAFT opportunistically reduces their clock frequencies to match the pace of bottlenecked devices, saving energy without impacting execution time. PAFT achieves this by continuously monitoring pipeline behavior and applying fine-grained frequency adjustments, adapting to workload and system changes. Experiments conducted on twelve widely used models show that PAFT consistently outperforms all baselines, achieving energy savings of up to 46% with an average overhead of 4%.
[LG-34] H-Spec: Parallel Speculative Decoding Without a Drafter-Side KV Cache
链接: https://arxiv.org/abs/2609.24197
作者: Weifan Jiang,Krishna Teja Chitty-Venkata,Megan Flynn,Reed Meyerson,Zhenting Qi,Tianyu Wu,Eldar Kurtic,Minlan Yu,Alexandre Marques
类目: Machine Learning (cs.LG)
*备注: 22 pages, 12 figures
Abstract:Speculative decoding losslessly accelerates large language model inference by having a lightweight draft model predict future tokens for verification by the target model. Recent block diffusion drafters further reduce drafting latency by predicting multiple tokens in parallel. However, existing block drafters project target hidden states at every input position into a separate drafter-side KV cache, incurring per-request memory and KV-write overhead that grow with concurrency; directly reusing target KVs in place removes this cache but fails to sustain draft quality throughout the block. We propose a hybrid target-context injection method that complements direct target KV reuse with target hidden states only at the last input position, requiring no separate drafter-side KV cache. Building on this design, we propose H-Spec, a hybrid Mamba-attention parallel drafter that consumes the two target-context sources through complementary modules. Mamba modules are initialized with projected last-token target hidden states, while attention modules reuse target KVs in place. Despite its recurrent formulation, Mamba’s parallel scan allows H-Spec to preserve block-parallel drafting. Across three target models and diverse tasks, H-Spec improves over the best baseline by 5.0–13.3% in mean accepted length and 5.3–12.6% in batch-size-1 inter-token latency speedup. Under concurrent serving, H-Spec consistently achieves higher throughput while maintaining lower KV cache utilization than baselines across evaluated concurrency levels.
[LG-35] Acceptance-Aware Draft Model Training for Speculative Decoding
链接: https://arxiv.org/abs/2609.24150
作者: Tianhua Xia,Mugilan Ganesan,Yifei Feng,Haiyu Wang,Maximilian Egger,Sai Qian Zhang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Speculative decoding accelerates large language model (LLM) inference by using a lightweight draft model to generate multiple candidate tokens that are verified by the target model in a single forward pass. Its speedup is largely determined by the acceptance length, yet existing draft-model training methods mainly optimize cross-entropy or Kullback-Leibler (KL) divergence as proxies. These objectives encourage distribution matching but do not directly optimize acceptance length, and the acceptance mechanism also differs between greedy and sampling-based decoding. In this work, we propose acceptance-length-aware training losses that directly optimize the expected number of accepted tokens within a speculative window. For greedy verification, we derive an expected accepted length (EAL) loss that explicitly maximizes expected acceptance length. For sampling-based decoding, we introduce a window total variation (WTV) loss that optimizes the overlap between temperature-scaled draft and target distributions while accounting for sequential acceptance dependencies. Both objectives can be further combined with a group-relative reinforcement learning stage (GRPO) using simulated acceptance length as the reward. Experiments across different target and draft models, tasks, and decoding settings show that our losses consistently improve acceptance length over KL-based training. WTV provides particularly strong gains under sampling-based decoding, while EAL better matches greedy verification. These results show that directly optimizing the acceptance objective, with losses tailored to the decoding mode, is more effective than conventional distribution-matching objectives. Subjects: Machine Learning (cs.LG) Cite as: arXiv:2609.24150 [cs.LG] (or arXiv:2609.24150v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2609.24150 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-36] Mind or Message? Auditing Theory of Mind in Multi-Agent Social Simulation
链接: https://arxiv.org/abs/2609.24146
作者: Cong Li,Cheng Chen,Thomas Fung,Alex Rossi,Yi Li
类目: Machine Learning (cs.LG)
*备注:
Abstract:Language model agents are increasingly used to simulate social interaction, and the resulting transcripts read as though the agents understand one another. We ask whether that appearance rests on a model of the partner’s mind or on the surface record of what the partner said. We build a social simulation in which both questions have exact answers: 40 multi-issue negotiations whose hidden preference weights and whose full Pareto frontier are known by construction. Two model families negotiate across 160 dyads, every transcript is frozen before any measurement, and 2880 counterfactual probes then hold the evidence byte identical while moving one factor at a time: the reader’s own stake, the partner’s tone, an identity label, and the order of recursion. The agents are socially fluent and economically poor. They reach agreement in 96.2% of dyads with 0 protocol failures, yet only 0.7% of deals land on the Pareto frontier, they leave 20.5% of the available joint value unclaimed, and they miss the one issue on which their interests are perfectly aligned in 76.6% of deals; on the frontier and on that aligned issue, a package drawn at random from the set both sides would accept does as well. The probes locate the failure. Swapping only the reader’s own payoff sheet, while the partner’s words and offers stay identical, moves the inferred top priority by 15.0 percentage points, which is egocentric projection rather than inference, while a tone rewrite moves it by 5.3 percentage points and an identity label by 0.0. Most tellingly, an agent predicts what its partner believes about it 72.5% of the time while that partner’s belief is itself correct only 51.2% of the time: the agents track the conversation far better than they track the mind behind it.
[LG-37] Luck Is Not Skill: When Do Paired Rollouts Help Group-Relative RL of LLM Agents ?
链接: https://arxiv.org/abs/2609.24144
作者: Nazmus Sakib
类目: Machine Learning (cs.LG)
*备注: 24 pages, 5 figures, 14 tables. Code, task pools, pre-registration and numerical summaries: this https URL
Abstract:Group-relative reinforcement learning compares rollouts of the same prompt, but independent environment noise can obscure these comparisons. We study paired rollouts, which share an event-keyed noise schedule within each group while preserving each rollout’s marginal distribution. Pairing removes the between-schedule component of reward-contrast variance, but need not reduce gradient variance. For one-sided grader noise, we derive an exact condition for reduction and give a counterexample in which reward contrasts improve while gradient variance increases. A controlled study trains a 2B tool-use agent under tool faults and grader flips, with three seeds per design. The protocol was registered with a disclosed, previously completed pilot. Under tool faults, pairing improves final noisy-test success by +5.1 percentage points on average, with all three seed differences positive, but misses the registered learning-curve criterion. The criterion is also missed under grader flips: the validation-AUC difference is +0.003 (95% interval [-0.029, +0.033]). A gradient probe on eight distinct checkpoints from two fault-trained trajectories finds lower mean-centered covariance traces under both noise types: 21 to 30% for grader flips and 40 to 63% for tool faults. These finite-sample measurements support the variance mechanism without establishing a general learning-speed benefit. The results distinguish improving reward comparisons, reducing estimator variance, and improving learning.
[LG-38] CLOOPD: Closing the Learner Loop in On-Policy Distillation
链接: https://arxiv.org/abs/2609.24141
作者: Keye Zheng,Hanyu Li,Zhan Cheng,Yuan Gao
类目: Machine Learning (cs.LG)
*备注:
Abstract:On-policy distillation (OPD) pays twice for each fresh batch: the student generates trajectories and a stronger teacher scores them. Existing methods improve which trajectories are scored and how the teacher signal is constructed, but usually consume it with one actor update. We introduce CLOOPD, a closed-loop framework separating teacher-signal acquisition from student-side realization. CLOOPD selects an adaptive \alpha waypoint inside a KL envelope, freezes the scored batch and its advantages, re-forwards the student after each actor pass, measures realization, and allocates actor work under a separate token budget. The framework includes deterministic two- and three-pass policies, token-priced CLOOPD-TPMR, and a budget-matched control. Across six 300-step runs on an 8-H20 node, every CLOOPD policy improves the one-pass TOP-D anchor at comparable teacher-token scale: macro accuracy rises from 15.41 to 17.78 with CLOOPD-Fixed2 and 19.36 with CLOOPD-Fixed3. At step 100, CLOOPD-Fixed3 reaches 15.35, nearly matching TOP-D at step 300 while using 67.2% fewer teacher-scored tokens and 28.0% fewer GPU-hours. Earlier 8-A100 ablations show adaptive \alpha eliminates observed trust-envelope violations; a third pass adds headroom. These results position CLOOPD as a framework for budgeting how fully students learn from teacher-scored tokens.
[LG-39] PAC-Bayesian Meta-Learning for Few-Shot Identification of Linear Dynamical Systems
链接: https://arxiv.org/abs/2609.24117
作者: Chenfeng Huang,George Michailidis
类目: Machine Learning (cs.LG)
*备注: Accepted at Transactions on Machine Learning Research (TMLR), 2026. J2C Certification
Abstract:Identifying linear time-invariant (LTI) dynamical systems is challenging when trajectories are short, noisy, or high-dimensional. Traditional system identification typically treats each system independently and cannot exploit shared structure across related systems. We propose PBML-LTI, a PAC-Bayesian meta-learning framework for few-shot LTI system identification that learns a transferable prior over task-specific dynamics while preserving task heterogeneity. Each task corresponds to an unknown LTI system, and the meta-learner uses training trajectories to learn a data-dependent prior over transition matrices. For a new system with limited data, PBML-LTI performs Bayesian adaptation under this prior to obtain a task-specific posterior, providing accurate estimates and principled uncertainty quantification. A key challenge is temporal dependence, since LTI trajectories violate the i.i.d. assumptions underlying most PAC-Bayes meta-learning analyses. We address this with a martingale PAC-Bayes analysis for dependent trajectory losses and derive a support-query predictive-risk bound that motivates a fit-KL meta-training objective. The bound clarifies the roles of empirical fit, posterior complexity, and prior quality in few-shot adaptation under sequential dependence. We further derive corollaries for transition-matrix recovery and multi-step trajectory prediction, connecting uncertainty-aware meta-identification with finite-sample guarantees for dependent dynamical data. Comments: Accepted at Transactions on Machine Learning Research (TMLR), 2026. J2C Certification Subjects: Machine Learning (cs.LG) Cite as: arXiv:2609.24117 [cs.LG] (or arXiv:2609.24117v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2609.24117 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Journalreference: ransactions on Machine Learning Research, 2026
[LG-40] Q-DEQ: Discrete Solving and Quantization for Deep Equilibrium Models in Time Series Forecasting under Edge Deployment Coding Constraints
链接: https://arxiv.org/abs/2609.24042
作者: Ruotong Yang,Hongdong Zhu,Qi Gao,Yin Ma,Hai Wei,Kai Wen
类目: Machine Learning (cs.LG)
*备注:
Abstract:Edge deployment motivates forecasting models with compact parameter storage and low-bit representations. Deep equilibrium models (DEQs) obtain implicit depth by repeatedly applying a shared layer, reducing the parameter cost of explicit layer stacking. Their usual Anderson solver, however, searches for update coefficients in the continuous real domain. We propose Q-DEQ, which formulates local updates in DEQ forward solving as discrete optimization problems. Candidate directions are constructed from the current state and iteration history, and a local quadratic residual model is used to evaluate their combinations. Binary encoding of the direction coefficients yields a quadratic unconstrained binary optimization (QUBO) problem that can be solved by simulated annealing (SA) or a coherent Ising machine (CIM). After fixed-point solving, a re-forward pass applies W8A8 fake quantization to the shared layer’s weights and activations. We evaluate Q-DEQ with an iTransformer backbone on five multivariate time series forecasting datasets. Relative MSE differences from the explicit multi-layer baseline range from -1.16% to +2.90% , with lower MSE on two datasets. DEQ parameter sharing reduces parameter counts by factors of 1.80\times – 3.82\times ; combined with W8A8, static weight storage is reduced by factors of 4.3\times – 12.8\times . Local QUBO problems solved using CPU-based SA and the Kaiwu CIM physical backend produce closely matching downstream forecasts. These results establish local discrete solving as a viable component of DEQ time series forecasting and provide a route for executing fixed-point updates through different combinatorial optimization backends.
[LG-41] Cost-Accuracy Trade-offs: Neural Operator vs Classical Numerical Solver
链接: https://arxiv.org/abs/2609.24021
作者: Daniel Zhengyu Huang,Andrew M. Stuart
类目: Numerical Analysis (math.NA); Machine Learning (cs.LG)
*备注: 41 pages; 10 figures
Abstract:Neural operators are data-driven models that learn mappings from inputs that parameterize partial differential equations, such as spatially varying coefficients, initial conditions, forcing terms, boundary conditions, or geometries, to solution fields or quantities of interest. Once trained, they can serve as surrogates for classical numerical solvers in many-query settings that require repeated evaluations for varying inputs. We address the question of when, and then why, neural operator surrogates outperform classical numerical solvers, in terms of cost for a given accuracy. We focus on the post-training, many-query limit, in which data-acquisition and training costs are treated as fixed and fully amortized. Even in this deliberately favorable regime for neural operators, there are regimes in which classical solvers outperform the surrogate models. We compare the cost-accuracy performance of neural operator surrogates and classical numerical solvers through a reproducible benchmark study comparing neural operators with problem-matched classical solvers on representative problems in computational science and engineering, focusing on prediction error, per-query floating-point cost, and wall-clock runtime. Neural operators are most competitive at low-to-moderate accuracy requirements. Their floating-point cost advantage depends strongly on the problem structure, arising when they avoid temporal or nonlinear iterations or predict a reduced quantity of interest rather than a full solution field. Additional wall-clock speedups result from dense tensor operations that are well suited to modern hardware. As the target accuracy is tightened, achieving the required accuracy with neural operators becomes increasingly challenging, and classical solvers outperform surrogates in this regime; thus classical solvers will remain important for verification and high-accuracy computation.
[LG-42] ShapeLex: Decoupling Local Shape Symbolization and Global Scale Modeling for Text-Controlled Time Series Generation
链接: https://arxiv.org/abs/2609.24003
作者: Subo Wei,Jianqi Gao,Mingyan Fan,Shaorong Xie,Xinzhi Wang,Yongpeng Dong
类目: Machine Learning (cs.LG)
*备注:
Abstract:Text-controlled time series generation aims to synthesize sequences that follow natural-language descriptions while remaining faithful to real data distributions. Existing paradigms often couple semantic understanding and sequence modeling in a single continuous latent space, lacking explicit local semantic anchors and separation between global continuous attributes and local discrete shapes. As a result, key local structures may be smoothed, missed, or misplaced. We propose Shape Lexicon (ShapeLex), which decouples text-to-sequence generation into discrete symbolization of local shapes and continuous modeling of global attributes. ShapeLex first induces a reusable vocabulary of discrete shape units, such as rises, spikes, and sharp drops, from training data, forming an interpretable symbolic space. An autoregressive generator then selects shapes according to the textual description, adjusts attributes such as position and duration, and composes them in temporal order into a shape skeleton. Finally, a mixture-density scale head models and samples the overall level and volatility to restore realistic global scale. Experiments on twelve public datasets, real user-written text, and downstream forecasting tasks show that ShapeLex generates series that better match real data distributions than existing methods. In addition, paired supervision is automatically synthesized from the learned vocabulary, avoiding annotation costs that grow with dataset size and improving scalability.
[LG-43] Simpler Methods Work Better for L1 Penalized Logistic Models and Large Datasets
链接: https://arxiv.org/abs/2609.23995
作者: Edward Raff,James Holt
类目: Machine Learning (cs.LG)
*备注: To appear in The 13th IEEE International Conference on Data Science and Advanced Analytics (DSAA 2026)
Abstract:Linear models with an L_1 -norm penalty remain state-of-the-art for high-dimensional ( d 1,000,000 ) tasks, offering a straightforward method for solving real-world industry problems. Despite their widespread use in industry and utility, many L_1 solvers are not effective for general use, are prohibitively slow, and are ineffective in parallelization. This makes them difficult to train in an MLOps pipeline on large industry-scale corpora. In this work, we test several proposed ``state-of-the-art’’ solutions from the literature and find that older methods are currently far superior for general use. We also identify several recommendations for academics to perform research that avoids erroneously overconfident results, which can prevent the transition to production use. Equally surprising, we find that a new and simple baseline, using LBFGS on a sub-gradient, is highly effective with minor tweaks, despite being dismissed in the literature for theoretical non-convergence. In practice, we find it is an easier-to-support and easier-to-scale method for production use.
[LG-44] MGRD: Compact morphology-gated residual diffusion for variance-aware cross-domain neurite forecasting
链接: https://arxiv.org/abs/2609.23990
作者: Tsung Yeh Hsieh,Cosmin Anitescu,Chunghwan Kim,Victoria A. Webster-Wood,Yongjie Jessica Zhang
类目: Machine Learning (cs.LG); Numerical Analysis (math.NA); Biological Physics (physics.bio-ph)
*备注:
Abstract:Tracking neurite morphology over time helps characterize structural changes during neuronal development and deterioration, but long-term time-lapse imaging is resource-intensive and difficult to scale. Forecasting future morphology could reduce this burden. Existing neurite digital-twin models such as gated spatiotemporal attention (gSTA) produce a single deterministic forecast without representing variability among plausible futures. We introduce Morphology-Gated Residual Diffusion (MGRD), a compact stochastic surrogate that jointly forecasts twenty future neurite-morphology frames from ten observed frames while conditioning on morphology features derived from the latest observation. On controlled phase-field trajectories, MGRD reduces trajectory-wise mean MAE by 9.7% relative to a matched control while updating 4.46 times fewer parameters. On human iPSC-derived neuron microscopy, MGRD improves all four reported metrics over gSTA, including a 39.6% reduction in trajectory-wise mean MAE and a 45.3% increase in skeleton F1. Without mouse-domain retraining or fine-tuning, MGRD also improves MAE and skeleton F1 on mouse cortical-neurosphere microscopy across 10-40-min sampling intervals and forecast horizons beyond 13 hours. Repeated sampling provides a case-level variance score for ranking forecast difficulty. Retaining approximately 60% of the lowest-variance cases reduces mean MAE by 17.6% on iPSC microscopy and 16.8% on simulation data. MGRD uses 1.01% of gSTA’s parameters, requires less than one tenth of its training-update time, and generates a 50-step DDIM trajectory 7.9% faster when morphology features are cached. These results establish MGRD as a compact stochastic surrogate for neurite-morphology forecasting and case prioritization across simulation and microscopy datasets.
[LG-45] he Operational Value of Spatial Dependence in Renewable Forecast Scenarios for Single-Period Economic Dispatch: A Controlled Ablation Study
链接: https://arxiv.org/abs/2609.23988
作者: Jayakumar Manoharan
类目: ystems and Control (eess.SY); Machine Learning (cs.LG)
*备注: This work has been submitted to the IEEE for possible publication. Copyright may be transferred without notice, after which this version may no longer be accessible. 10 pages, 3 figures, 2 tables
Abstract:Renewable forecasts are evaluated by statistical skill (e.g., CRPS), but grid operators pay for realized dispatch cost. We diagnose what drives dispatch value in a single-period newsvendor-style economic dispatch using real public data from two European transmission systems (CWE, DE-4TSO). Spatial coherence across forecast sites falls below the pre-specified 1% practical-significance threshold: a controlled ablation holding per-zone marginal forecasts bit-identical and varying only cross-zone dependence (10 configurations, 3 seeds, paired-bootstrap confidence intervals) shows a coherence gain of at most 0.64% of dispatch cost, indistinguishable from zero in 3 of 10 configurations, reached only under an unrealistic 8-fold forecast-error stress test. Decision-focused training, an established paradigm in this venue, delivers a robust 2.82-5.19% gain. A parametric Gaussian-copula approximation matches the empirical copula at realistic error magnitudes but performs worse than no dependence under extreme stress. A single-seed sweep shows that a 12% energy-score gain changes cost by less than 0.1%. Results characterize this single-period dispatch class; a lightweight four-period extension supports the same conclusion. For this dispatch class, spatially-correlated scenario generation provides limited operational value on its own; grid operators and forecast vendors should instead evaluate dependence models by downstream decision value and prioritize decision-focused training.
[LG-46] he Neural Forcing for Three-Dimensional Incompressible Navier-Stokes finite time blowup
链接: https://arxiv.org/abs/2609.23934
作者: Beibei Li
类目: Machine Learning (cs.LG)
*备注:
Abstract:We present a two-part neural framework for forced three-dimensional incompressible Navier–Stokes flow. Part~I develops the computational forcing system. A physics-informed neural model generates structured external-force trajectories, candidates are optimized through differentiable PDE rollouts or PPO-Clip, and selected forcings are frozen and checked by independent fixed-force replay. Part~II provides the mathematical certification layer. It separates neural candidate discovery from continuum analysis, derives integrated reciprocal-vorticity criteria that imply Riccati-type growth and finite-time loss of smooth continuation, develops a validated computational-to-continuum transfer strategy, and establishes a conditional positive-probability closure for a nondegenerate neural output law. The proof is complete at the continuum level.
[LG-47] Matched-Input Estimates Differ in Sign Across Architectures: Auditing EEG Foundation Models on Motor Imagery
链接: https://arxiv.org/abs/2609.23924
作者: Kevin Zhou,Sparsh Roy
类目: Machine Learning (cs.LG); Neurons and Cognition (q-bio.NC)
*备注:
Abstract:Pretrained EEG foundation models are increasingly proposed as general-purpose encoders for brain-computer interfaces, yet recent benchmarks disagree about when their representations transfer to downstream tasks. We audit LaBraM and CBraMod on motor imagery under a validation-locked protocol in which preprocessing, architecture, optimization, freeze depth, checkpoint, temperature, and method selection are determined using training-session data only. On four-class BCI Competition IV-2a, every supervised comparator evaluated here outperforms every foundation-model configuration, including validation-selected fine-tuning. We then examine a key confound: foundation models and task-specific decoders are normally evaluated using different input pipelines. Retraining three supervised architectures on the broadband arrays consumed by the foundation models produces matched-input accuracy differences of opposite sign across architectures: broadband input improves ATCNet by 0.078 accuracy while reducing EEG Conformer accuracy by 0.088. None of the three individual matched-input terms is significant after multiple-comparison correction at n = 9, so we treat the sign variation descriptively rather than as a formal architecture-by-pipeline interaction. These observed sign differences suggest that a single comparator may not provide an architecture-invariant decomposition of a pretrained-versus-supervised performance gap. The four-class deficit also does not reproduce uniformly across motor-imagery datasets: on two-class BNCI2014-004 we cannot detect the same separation between fine-tuned CBraMod and the supervised comparators. Finally, validation-fitted temperature scaling returns foundation-model calibration error to the supervised range despite substantially lower four-class accuracy.
[LG-48] A discrete generative model of neuronal spiking activity on microelectrode arrays
链接: https://arxiv.org/abs/2609.23907
作者: Md Sayed Tanveer,Mohammed A. Mostajo-Radji,Ge Wang
类目: Machine Learning (cs.LG); Signal Processing (eess.SP); Neurons and Cognition (q-bio.NC)
*备注:
Abstract:Generative models of neural activity could help characterize tissue dynamics, compare experimental conditions, and simulate population activity for applications ranging from disease and drug-response studies to closed-loop experimentation. Existing approaches, however, typically assume a fixed set of sorted neurons, whereas high-density microelectrode arrays produce extremely sparse, array-wide binary spike volumes in which the observed subset of electrodes varies across assays. We introduce a discrete generative model that represents this activity using a shared vocabulary of spatiotemporal motifs. A residual vector-quantized autoencoder learns the motif vocabulary, while a factorized masked transformer predicts where activity occurs and which motif appears at each active location. We evaluate the model on 31 assays spanning human brain organoids and acute \emphex vivo human hippocampal tissue. The learned motifs are broadly reused: assay identity explains only 9% of the entropy in motif use, and motif overlap across tissue types is comparable to overlap within them. When representation quality is evaluated independently of the generative prior, our approach achieves 5.2\times the voxel-level reconstruction average precision of a matched flat tokenizer. For masked completion and free generation, the full model achieves 1.4 – 2.6\times the site-level average precision of the matched generative baseline and outperforms it across all four families of generation metrics. These results establish a compact, reusable representation for array-wide spiking activity without learned assay-specific parameters, providing a scalable foundation for generative modeling across diverse neural preparations.
[LG-49] Multivariate quantile regression via Kolmogorov-Arnold Networks
链接: https://arxiv.org/abs/2609.23906
作者: Andrew Polar,Michael Poluektov
类目: Machine Learning (cs.LG)
*备注:
Abstract:This paper introduces a novel algorithm for predicting conditional joint distributions of vector-valued targets in stochastic systems whose randomness is intrinsic rather than arising from observation errors or additive noise. Multivariate quantile regression also involves modeling conditional joint distributions but represents a less challenging task. It predicts the probability that vector-valued targets fall within predefined regions, identifies regions corresponding to predefined probability levels, or performs both tasks simultaneously. The proposed identification technique employs ensembles of Kolmogorov–Arnold networks (KANs) as flexible function approximators. Although the suggested technique is not theoretically restricted to KANs, KANs are particularly well suited to the proposed construction and are therefore used throughout this study. In addition to the training procedure, this work introduces a new discrepancy measure for joint distributions and a goodness-of-fit (GoF) test based on it. This GoF test was initially developed to validate and calibrate the proposed identification technique and is used here in an ad hoc manner. Although the test could be tabulated for broader use, such a tabulation is not pursued in this work. The test is also applicable more generally.
[LG-50] GDN Tree-Scan: Served Tree Verification for Recurrent-Hybrid Language Models
链接: https://arxiv.org/abs/2609.23900
作者: Zhiyuan Ma
类目: Machine Learning (cs.LG)
*备注: 9 pages
Abstract:Tree speculative decoding verifies multiple candidate continuations in one target forward pass. For attention-only transformers, the verifier mainly needs an ancestry mask. Recurrent-hybrid language models break this assumption: a candidate row must also carry the recurrent state that native sequential decode would have produced along its root-to-node path. Otherwise, a verifier can use a correct attention mask while still conditioning on an impossible recurrent history. We present GDN Tree-Scan, a served verifier for Gated-DeltaNet hybrid language models integrated into vLLM. The system combines FlashAttention-2 tree-bias attention, branch-local GDN scan/replay, device-side multidraft commitment, and accepted-chain-only state publication. On the public Qwen3.6-27B-FP8 checkpoint, in a clean batch-one (B=1) SWE/Codex decode gate at temperature 0.6, a six-node root-branch tree increases committed tokens/event by 17.2% at near-native verify-forward time and reaches 23.88 token-weighted decode tokens/s versus 18.80 for native five-step MTP (E5), a 27.0% token-weighted decode-throughput gain. The per-request-equal latency view is +4.0%, and end-to-end task wall time remains prefill-heavy. Empirical equivalence evidence is scoped to recurrent-oracle probability-rescore (p-rescore) closure within the observed native flip floor, not a full distribution-distance proof. Comments: 9 pages Subjects: Machine Learning (cs.LG) Cite as: arXiv:2609.23900 [cs.LG] (or arXiv:2609.23900v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2609.23900 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-51] HumynexSurg-1: A Curated Expert Liposuction Dataset
链接: https://arxiv.org/abs/2609.23885
作者: Rhea Huang,David L. Matlock,Laurence Reich
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: 15 pages, 7 figures, 13 tables. Version 0 dataset release
Abstract:Robot foundation models learn manipulation from large demonstration corpora, but surgery is missing from those corpora: across the 780-hour Open-H surgical collection, one dataset carries synchronized force and none covers an aesthetic procedure. Liposuction is the hard case, because the instrument works under the skin and the surgeon operates by feel and by judgment. Humynex Robotics builds curated expert datasets for this kind of procedure. HumynexSurg-1 is the first release: a master liposuction surgeon performing on porcine abdominal tissue while narrating every decision, recorded with synchronized suction pressure, six-axis hand force/torque, top-down RGB-D video, side video and a lavalier microphone – 14 episodes, 42,738 frames, 35.6 minutes, 356 utterances of which 95% compile into a liposuction-specific label schema. The capture follows a patent-pending sensing plan organized around the quantities a policy needs, so a channel captured today by a model can be upgraded to a sensor tomorrow without changing the data format. This release captures the instrument motion as a tool-hand track in the side video and provides the force channel as state; the funded capture adds a measured 6-DoF handle pose, a validated force channel, ultrasound imaging of the fat layer, and palpation sensing. As a proof of concept, NVIDIA Isaac GR00T N1.7 fine-tunes on the dataset with no custom code in under an hour per run and learns the recorded sessions; scaling probes on the same episodes show where further gains come from: every new session lowers the error on an unseen session. The dataset, its label schema, its quality-assurance reports and its evaluation protocol are the product; the next capture, many short sessions across fat regions with the sensors named here, is what the probes point to.
[LG-52] Collaborative Streaming Anomaly Detection with Interactive Explanations and Ensemble Consensus KDD ECML
链接: https://arxiv.org/abs/2609.23883
作者: Diogo Risca,Afonso Lourenço,Ricardo Martins,Goreti Marreiros
类目: Machine Learning (cs.LG)
*备注: Streaming Continual Learning ECML PKDD Workshop 2026
Abstract:We present a collaborative streaming anomaly detection system for high-speed data streams that explicitly integrates human analysts into the decision loop. The system combines heterogeneous detectors and aggregates their outputs through a normalization-based weighted consensus, complemented by artifact-aware rules to stabilize anomaly scoring under deployment. To improve interpretability, it derives surrogate models that approximate the ensemble consensus and expose human-readable sensor conditions associated with anomalous behavior. Analysts can actively intervene by reviewing anomaly episodes, adjusting consensus behavior, and refining surrogate rules used for anomaly prediction, producing a human-adjusted ensemble. We evaluate the approach on an industrial stream with 260,000 events and 3 anomalous episodes, showing robust detection and actionable human-AI interaction.
[LG-53] PROSE: A Theory of Optimal Stopping with Perishable Evidence for Peer Selection in Intermittently Connected Decentralised Learning
链接: https://arxiv.org/abs/2609.23845
作者: Christos Anagnostopoulos
类目: Machine Learning (cs.LG)
*备注: 22 pages, 6 figures. Theory paper; no experiments
Abstract:Decentralised federated learning removes the aggregation server but makes collaboration dependent on transient peer availability. In mobile and intermittently connected systems, evaluating a promising peer consumes contact time and may cause the exchange opportunity itself to vanish, so that the evidence a learner gathers about a peer is perishable: it decays because links expire and because peer models drift while old measurements age. This paper develops a self-contained theory of optimal stopping for the resulting peer-selection problem. We formalise a receiver’s within-contact decision as a finite-horizon Markov optimal-stopping problem with costly information acquisition and a future-arrival outside option, and prove that it admits an optimal policy characterised by a reservation value (Snell-envelope structure). Around this formulation we prove: (i) stage-uniform, drift-aware concentration and a maximin certification rule that is correct with high probability together with a finite-sample identification bound; (ii) a mobility-aware value of-information stopping rule and comparative statics showing that higher link hazard lowers the value of continued probing and enlarges the stopping region; (iii) a closed-form value of waiting under marked-Poisson contact arrivals, together with a search-theoretic reservation value whose comparative statics we characterise; and (iv) a myopic-optimality theorem establishing that, in sufficiently volatile (monotone) mobility regimes, the one-step confidence-safe rule is a sound surrogate for the optimal policy and never stops prematurely. We instantiate the theory as PROSE (Perishable-evidence Reservation-value Optimal Stopping for Exchange), a lightweight, fully local policy, and delineate the static contact and drift-free limits in which classical sequential decision problems are recovered. The development is entirely analytical.
[LG-54] Adaptive Determinantal Client Scheduling in Federated Learning
链接: https://arxiv.org/abs/2609.23843
作者: Wen Xu,Ben Liang,Gary Boudreau,Hamza Sokun
类目: Machine Learning (cs.LG); Distributed, Parallel, and Cluster Computing (cs.DC)
*备注:
Abstract:Scheduling clients for model training is critical in federated learning due to both data and system heterogeneity. Most previous works focus on the quality of the scheduled clients to achieve faster convergence, shorter wall-clock convergence time, or better average model performance. They rarely consider the diversity of clients, which is important to counter heterogeneity and improve performance for the worst-off clients. In this work, we advocate the use of determinantal point processes (DPPs) to model and enhance the diversity in client scheduling. We first design the kernel matrices of DPPs using gradient information and quality scores, which inherently enables a flexible quality-diversity trade-off. Applying fast MAP inference over DPPs, we propose Adaptive Determinantal Client Scheduling (ADCS) in FL. We further quantify the gradient approximation error of ADCS and develop convergence analysis for general biased client selection in FL with non-convex loss functions. We conduct comparative numerical experiments showing that ADCS outperforms state-of-the-art client scheduling algorithms, including both quality-based and diversity-based ones.
[LG-55] From Regional to Global: Transfer Learning for Atmospheric Transport Emulators
链接: https://arxiv.org/abs/2609.23838
作者: Jeff Clark,Elena Fillola,Nawid Keshtmand,Raul Santos-Rodriguez,Matthew Rigby
类目: Machine Learning (cs.LG)
*备注:
Abstract:Greenhouse gas emissions estimates can be derived using inverse methods by combining atmospheric concentration observations with chemical transport models. The latter traditionally use physics-driven simulators such as Lagrangian Particle Dispersion Models (LPDMs), which are expensive to run and do not scale well to modern satellites’ high resolution data. Previously we developed a performant atmospheric transport emulator that approximates LPDM outputs (“footprints”) over South America ~1,000X faster than the UK Met Office’s LPDM. Expanding towards global emulation is not straightforward, as atmospheric transport is regionally heterogeneous. This paper evaluates spatial transferability capabilities of models across four world regions: South America, East Asia, South Asia, North Africa using both region-specific and multi-region models, and leave-one-region-out experiments. Regional differences are characterised in the context of input variable and output footprint distributions. This work builds intuition in cross-region generalisation and transfer learning, aiding regional performance towards efficient global emissions estimates.
[LG-56] Actionable Insights from Observational Data: The Case of Advanced Classes in K-12 Education
链接: https://arxiv.org/abs/2609.23836
作者: Nabit Bajwa,Seth B. Hunter,Sanmay Das
类目: Machine Learning (cs.LG)
*备注:
Abstract:A fundamentally challenging question in K-12 education is about the effects of taking more advanced or challenging classes. It is particularly complex because students (and/or their parents) choose whether to enroll in these classes, making causal analysis challenging. In this paper, we begin to tackle this question by taking advantage of a novel dataset from a public school system in the US. This dataset records students’ course enrollment decisions, prior academic histories, demographics, and subsequent outcomes around the time of a district-wide change that introduced optional open-enrollment advanced middle-school courses in subject areas. This is a rich observational dataset, but enrollment in advanced classes is driven by student characteristics and choices rather than random assignment. This creates a core identification challenge: the same factors that influence enrollment in advanced courses are also predictive of academic outcomes. As a result, simple comparisons between enrolled and non-enrolled students are confounded, and naive estimates may reflect underlying differences in student ability, motivation, or support rather than the impact of coursework itself. Our analysis shows that enrolling in advanced English courses has a net positive but modest effect on student achievement outcomes. However, these benefits are unevenly distributed: some students with relatively large predicted gains (“middle achievers” in prior years) are less likely to enroll than others. Some other groups (e.g. Black students and those with lower socio-economic status) also demonstrate significantly lower propensity to enroll. This gap between predicted benefit and observed enrollment illustrates how careful data analysis can extract actionable insights from large observational datasets, including identifying students who appear well-positioned to benefit but do not select into advanced options.
[LG-57] Real-time Generalizable Heart Valve Mechanics for Clinical Disease Assessment via a Physics-Conditioned Neural Operator
链接: https://arxiv.org/abs/2609.23826
作者: Shawn Koohy,Wensi Wu,Matthew A Jolley,Paris Perdikaris
类目: Machine Learning (cs.LG)
*备注: Preprint submission
Abstract:Mitral regurgitation is the most common heart valve disorder worldwide, affecting over 2% of the global population, rising to at least 10% in adults over 75, and causing approximately 15% of valvular heart disease-related deaths. Yet only a minority of patients with severe disease undergo corrective surgery. Rapid assessment of valve mechanics could enable earlier, more precise intervention, but traditional finite element simulations remain too slow for clinical timelines and parameter sweeps. We introduce the Physics-Conditioned Neural Operator (PCNO), a transformer-based surrogate that predicts leaflet displacement, strain, and stress fields across mitral and tricuspid geometries, conditioned on systolic blood pressure and tissue properties. Trained on functional, regurgitated, and pathological valves, including tethering, P2 prolapse, and annular dilation, PCNO achieves up to a 15,260x speedup over fine mesh finite element simulations with comparable accuracy, identifies pathology class, and resolves diagnostic metrics within 3.5% error under out-of-distribution extrapolation.
[LG-58] Iterative Atom Refinement: A Monotonicity Principle for Dictionary Learning
链接: https://arxiv.org/abs/2609.23812
作者: Alexander Christie,Miguel Moscoso,Alexei Novikov,George Papanicolaou,Chrysoula Tsogka
类目: Machine Learning (cs.LG); Probability (math.PR); Machine Learning (stat.ML)
*备注:
Abstract:Dictionary learning seeks to recover an unknown dictionary A from observations \bf y_i = A\bf x_i with sparse coefficient vectors \bf x_i . We introduce the \emphIterative Atom Refinement (IAR) algorithm, a simple procedure for recovering individual dictionary atoms. Starting from a random direction, IAR repeatedly selects the observations most strongly correlated with the current iterate and updates the direction by averaging the selected data. Our main contribution is a rigorous convergence theory of IAR. Using high-dimensional probabilistic estimates and a novel monotonicity principle for atom-selection probabilities, we show that a small initial advantage of one atom is amplified until that atom is isolated. Under our model assumptions, IAR identifies a generating atom after only three refinement steps. Numerical experiments support the theory and show that the resulting dynamics accurately capture the behavior observed in dictionary refinement.
[LG-59] Falling Trees: A Model Class for Interpretable Risk Prioritization ICML2026
链接: https://arxiv.org/abs/2609.23780
作者: Varun Babbar,Zachery Boner,Margo Seltzer,Cynthia Rudin
类目: Machine Learning (cs.LG)
*备注: ICML 2026 (Spotlight paper)
Abstract:Many real-world decisions require prioritizing high-risk cases, such as clinicians prioritizing high-risk patients before lower-risk ones. Falling rule lists (FRLs), which are ordered if–then rules with monotonically decreasing risks, provide an interpretable framework for such tasks; however, their single-path structure yields a highly restricted model class. We introduce falling trees, a new family of interpretable models that enforces the same monotonic risk constraint while permitting tree-structured branching. We present GRAVITree, a novel dynamic-programming-with-bounds algorithm for learning the Rashomon set of falling trees under depth and branching constraints. Our formulation can interpolate between rule lists and full decision trees, enabling user-desired model expressivity. In a new clinical dataset and in many public classification benchmarks, falling trees match or outperform FRLs and other interpretable baselines, often producing more sparse decisions for high-risk instances. Our results show that falling trees strike a practical balance between interpretability, expressiveness, and risk prioritization for high-stakes settings.
[LG-60] Statistical Convergence of Transformer Encoder-Accelerated Robust Reinforcement Learning
链接: https://arxiv.org/abs/2609.23775
作者: Suman Banerjee,Hiroyasu Tsukamoto
类目: Machine Learning (cs.LG); Systems and Control (eess.SY)
*备注:
Abstract:Obtaining the optimal action-value function in Markov decision processes is computationally intensive in large state–action spaces. In this study, we present statistically rigorous convergence results for a robust reinforcement learning algorithm warm-started by a transformer-based action-value function prediction, where natural language prompts encode task specifications. Our framework adopts the R-contamination model to characterize uncertainty in the state transition kernel, and employs conformal prediction to certify convergence via trajectory-level nonconformity scores constructed from the contracting Bellman residual. The resulting conformal quantile bounds the gap between the running and optimal action-value functions simultaneously over all iterations, thereby yielding a pre-certified stopping rule that requires little knowledge of the true transition kernel. Numerical case studies on perturbed maze environments of varying size and contamination level confirm that the transformer-based warm start measurably reduces the initial error and accelerates convergence, while the proposed conformal bounds track the true error trajectory more tightly than existing guarantees.
[LG-61] SAGE: Optimal-Stopping Peer Selection for Decentralised Federated Learning
链接: https://arxiv.org/abs/2609.23773
作者: Ke Xiao,Qiyuan Wang,Christos Anagnostopoulos
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注: 12 pages, 3 figures, 3 tables
Abstract:Decentralised federated learning replaces server aggregation with peer-to-peer model exchange, making collaborator selection a local decision under uncertainty. Fixed probe budgets waste effort on easy choices yet fall short when peers are hard to distinguish. We propose SAGE (Sequential Anchor-Gated Exchange), an optimal-stopping peer selector under a one-model-bearing-exchange budget. A receiver scores candidate neighbours on receiver-owned anchor evidence and selects once an advantage is certified. It continues probing only while further evidence repays its cost, and otherwise falls back to random gossip. We show that the stopping problem admits an optimal rule attained at a finite stage, and that the anchor schedule is order-optimal in the peer-risk gap and the confidence level. We further show that the selector never returns a peer worse than random gossip with high probability, and prove that no such guarantee holds for selectors that commit without a certificate. A separability threshold follows, below which no probing budget improves on gossip. Experiments span two image benchmarks, two graph families and three heterogeneity levels. Selectors that always act on their evidence lose to gossip in every configuration tested. SAGE-OS matches gossip on 75.5% less evidence than a fixed budget, at half the communication overhead of two published selectors. The operative decision is not which peer to rank first, but whether the evidence justifies ranking at all.
[LG-62] GenVoid: Uncertainty-Aware Learning of Subsurface Material Defects with an Experimentally Validated Physics-Informed Generative Model
链接: https://arxiv.org/abs/2609.23761
作者: Trishit Mondal,Prajwal Bharadwaj,Nikhil Karanjgaokar,Ameya D. Jagtap
类目: Machine Learning (cs.LG)
*备注: 26 pages, 12 figures
Abstract:Internal voids are ubiquitous defects in manufactured structures, yet their characterization remains challenging because their geometry is hidden and can only be inferred indirectly from accessible measurements. Here we introduce \textitGenVoid, a physics-informed generative model-based framework for identifying internal voids in complex two- and three-dimensional solids from surface displacement measurements alone. By incorporating the governing mechanics into a generative inference framework, \textitGenVoid enables void identification across linear elastic, hyperelastic and plastic material behaviours and accommodates complex two- and three-dimensional structural geometries. Importantly, the framework explicitly accounts for uncertainty and noise in displacement measurements, producing probabilistic reconstructions of internal void geometry rather than a single deterministic estimate. We demonstrate the approach using high-fidelity synthetic datasets and experimentally measured displacement fields obtained from in-situ mechanical experiments, establishing its ability to infer hidden voids from realistic displacement measurements. To quantify the fundamental limits of such inference, we further introduce an observability measure that characterizes the sensitivity of boundary measurements to localized stiffness perturbations within the interior under an ensemble of applied loads. This framework provides a direct connection between defect location, sensor configuration and reconstruction fidelity, enabling systematic assessment of how the number and spatial distribution of boundary measurements govern void-identification accuracy. To this end, these results establish a physics-informed and uncertainty-aware approach for non-invasive characterization of hidden defects and provide a quantitative basis for designing measurement strategies for inverse problems in solid mechanics.
[LG-63] Marginal Calibration Does Not Compose: Hidden Dependence in Modular Robot Navigation IROS2026
链接: https://arxiv.org/abs/2609.23731
作者: Rista Baral
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: Accepted to the Rethinking Uncertainty for Modern Robotics Paradigms Workshop at IROS 2026
Abstract:Robotic systems are typically composed of multiple independently developed modules that work together to perceive, predict, and act in the environment. Although each module may perform reliably in isolation, composing them does not necessarily preserve uncertainty calibration at the system level. In this work, we show that well-calibrated component interfaces do not necessarily produce calibrated downstream behavior after composition. Using a moving-obstacle prediction pipeline, we demonstrate that position and velocity estimators can each appear well calibrated individually, yet differences in how their error are correlated lead to substantially different estimates of future-state uncertainty. Consequently, assuming independence can make the system either overly confident or unnecessarily conservative, directly influencing downstream planning decisions and safety. Through simulations, we show that modeling the joint covariance restores downstream calibration and improves system performance, whereas dependence-robust uncertainty bounds enhance safety at the cost of increased conservatism. Our findings reveal a fundamental limitation of independently validating robotic modules and highlight the need for interfaces that communicate dependence information or support direct system-level calibration.
[LG-64] One Patch Three Roles: What Is Actually Coupled in Autoregressive Time-Series Forecasting?
链接: https://arxiv.org/abs/2609.23686
作者: Ziang Li,Yue Huang,Guoxu Zhou,Na Han,Jie Wen,Lunke Fei,Xiaozhao Fang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Patch-based autoregressive time-series forecasting often ties input representation, learned transitions, and recursive execution to one patch length. We ask which of these roles can be adjusted separately. A supporting atomic-encoding study finds greater sensitivity to model width than to atom grouping on the evaluated grid. Our main finding is that a frozen parent’s recursive trajectory is easier to fit than the observed future with lightweight parallel exits. Autoregressive Trajectory Distillation (ATD) turns this into selectable ATD-1/2/4/8 execution, with ATD-1 exactly recovering the parent. On a paired four-data-set comparison, ATD-8 reaches 5.54\times end-to-end speedup with stable quality across widths. Fewer calls do not automatically remove the parent’s existing forecast error: ATD improves trajectory fidelity in all 21 seed runs but forecast accuracy in only 15 against matched clean-future supervision. We further find a correctable residual projection along a train-selected periodic history direction. Spectrum Tangent applies this correction without adding neural parameters or Transformer calls. At horizon 720, it reduces mean squared error (MSE) and mean absolute error (MAE) by 2.54% and 2.33% over seven data sets and two output widths, while remaining 3.24\times faster than recursive inference. Level and shape projections sometimes disagree. Trajectory compressibility, the fidelity-accuracy mismatch, and the correction recur across three public AR parents. Together these results separate representation, transition, and execution as AR design axes. Code is available at this https URL.
[LG-65] ETH-TraceBench: A Large-Scale Event-Stream Benchmark for Ethereum DeFi under Temporal Protocol and Contract Shift
链接: https://arxiv.org/abs/2609.23659
作者: Kemal Kirtac,Carsten Maple
类目: Machine Learning (cs.LG)
*备注: 30 pages. Under review at Transactions on Machine Learning Research (TMLR)
Abstract:Ethereum decentralized finance (DeFi) provides a public, time-stamped record of transaction-level event streams, but the same public symbols can create strong machine-learning shortcuts. We introduce ETH-TraceBench, a benchmark for evaluating Ethereum DeFi representations under temporal, protocol, pool/infrastructure, and symbolic shift. The raw event universe covers January 2021-December 2025 and contains 1.35 billion transactions with logs and 5.01 billion raw log rows. Model evaluation uses a fixed 911,267-instance supervised sample, training on 2021-2024, selecting models on 2025H1, and testing on 2025H2. Simple models perform strongly on the aggregate temporal test: TraceStats-GB reaches 0.953 macro-F1 and TopicEmitterHashMLP 0.959 on the canonical DEX test set. Performance drops sharply under protocol novelty, with macro-F1 of 0.794, 0.743, and 0.766 for TraceStats-GB, TopicEmitterTrace-SGD, and TopicEmitterHashMLP, while strict unseen-pool scores remain 0.927, 0.897, and 0.935. Uniswap v4 and Ekubo v1, both absent from supervised training, are materially harder than the full test. Jointly masking emitter and topic identity reduces DEX macro-F1 to 0.916 and liquidation macro-F1 to 0.774 for TopicEmitterTrace-SGD. A standard Transformer over log-index-ordered events provides no consistent advantage over a deterministic shuffle of the same events, indicating that high aggregate scores can arise without sophisticated chronological modeling. A natural-prevalence audit estimates 2025H2 DEX prevalence among logged Ethereum transactions at about 22.5%, and a deterministic 400-transaction audit finds complete agreement with task label sources and independently re-queried raw-log counts. ETH-TraceBench therefore treats difficult transfer and controlled-input conditions, rather than a single aggregate score, as the main evaluation target.
[LG-66] Beyond Appearance Shifts: Task-Semantic Action Calibration for VLA Models
链接: https://arxiv.org/abs/2609.23650
作者: Shuaijun Liu,Feiyang You,Chengyu Wu,Shuyang Hao,Chenglong Zhang,Jingyao Cai,Xingwei Chen,Ningxin Su
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: 27 pages (10-page main text), 15 figures, 24 tables. Project page: this https URL
Abstract:Vision-language-action (VLA) models have achieved strong performance in embodied manipulation, but still lack a clear mechanism to balance behavioral stability with task-semantic sensitivity. We identify two complementary failure modes. Under task-preserving changes, where task semantics remain unchanged but scene appearance varies (e.g., style, illumination, clutter, or paraphrasing), policies often exhibit unnecessary action drift. Conversely, under semantic-breaking changes, where key task semantics such as the target object or constraint are altered, policies frequently fail to produce sufficiently distinct behaviors and instead follow the original trajectory. To address this gap, we propose BAS-VLA, a task-semantic action calibration framework built on top of a frozen base VLA. BAS-VLA adopts a breaking-centered calibration core as the default path, and introduces a selective evidence-gated preserving auxiliary that activates only when nuisance variation is detected while task semantics remain consistent. On the OpenPI-pi0.5 / LIBERO-Object Milk-Swap benchmark, BAS-VLA maintains high success on clean (98.0%) and semantics-preserving conditions (97.5%), while reducing clean-criterion success to 0.0% under deliberate target-object swaps, demonstrating strong stale-task suppression and task-semantic separation. On validated style-preserving shifts, it improves success from 42% to 70% without degrading clean performance. These results highlight that reliable VLA behavior requires moving beyond appearance robustness toward explicit task-semantic action calibration.
[LG-67] Bilinear Optimization Divergence: Diagnosing Factor-Constrained LoRA Continual Learning
链接: https://arxiv.org/abs/2609.23594
作者: YongShun Wang,JianLin Su,Yong Ma
类目: Machine Learning (cs.LG)
*备注: 22 pages, 3 figures. Code is available at [ this https URL ]( this https URL )
Abstract:Orthogonality in a LoRA factor does not by itself specify what the composed update protects: the answer depends on the task-start state, the parameterization, and the realized optimizer displacement. We formalize this question through Bilinear Optimization Divergence (BOD), an anchor-relative diagnostic of effective-update response on selected historical features. The finite-step analysis distinguishes two cases. In a shared adapter, protecting the routing displacement leaves a learned-anchor residual through the changing companion factor. In a fresh zero-output block, a feasible routing state can protect the composed update while both current factors remain trainable. These conditions yield Semi-Frozen Orthogonal Routing (SFOR) for shared adapters and current-block hard protection for cumulative O-LoRA; Weight Residual Projection (WRP) enforces the required displacement after the optimizer step. Controlled two-task traces verify the predicted residual paths, reducing normalized historical response from 19.12% to 0.005% in the shared family and from 7.72% to 0.002% in the cumulative family. Four-task experiments on Qwen3-8B characterize the resulting trade-offs: SFOR improves backward transfer (BWT) from -2.47 to -0.86 with nearly unchanged average accuracy (AA), while O-LoRA hard protection improves three-order mean AA from 80.27% to 81.30% and forgetting measure (FM) from 2.20 to 0.43. Component controls also show that stricter feasibility need not improve final task performance. Together, the analysis and evidence provide an architecture-conditioned account of which constraint to enforce, how to enforce it, and how to interpret its empirical value.
[LG-68] Preserving Geometric Integrity in Graph Prompting via Measure-Constrained Optimal Transport
链接: https://arxiv.org/abs/2609.23547
作者: Xiangyu Wang,Shuo Wang,Ruiyi Fang,Zhao Kang
类目: Machine Learning (cs.LG)
*备注: 28 pages, 4 figures
Abstract:Graph prompt learning enables parameter-efficient adaptation of frozen Graph Neural Networks to downstream tasks through lightweight prompt parameters. As routing becomes increasingly node-adaptive, however, independently optimized local decisions can collectively concentrate assignment mass on a small subset of a finite shared prompt bank, even when individual node–prompt matches remain locally meaningful. We propose MINT (Measure-INtegrity Transport), an entropically regularized optimal transport framework that formulates node-to-prompt adaptation as a globally coupled allocation problem. The transport cost favors local geometric compatibility, while a prescribed prompt-side marginal explicitly controls graph-wide prompt utilization. We further derive an exact variance decomposition that separates prompt-side geometric variance into retained prompt-update variation and within-node barycentric dispersion, together with a conditional stability bound for the frozen-encoder forward map. Across standard citation networks and additional heterophilic graphs, MINT remains competitive in few-shot adaptation. Controlled and end-to-end experiments further distinguish the roles of routing and topology: fixed-marginal routing controls graph-wide prompt utilization and has measurable end-to-end effects on citation networks, while topology augmentation provides a complementary, graph-dependent mechanism for addressing structural mismatch. Code is available at this https URL.
[LG-69] Decoupled Causal Discovery
链接: https://arxiv.org/abs/2609.23535
作者: Zhengkang Guan,Fei Wu,Kun Kuang
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 24 pages
Abstract:Causal discovery from observational data is a fundamental yet challenging task in scientific research. While existing approaches are primarily based on conditional independence tests, structure scores, or restrictive functional assumptions, we propose Decoupled Causal Discovery (DCD), a novel decoupling-based perspective that does not rely on these methodologies. DCD directly identifies the Markov boundary (MB) by decoupling non-target variables via weighting functions, such that only variables within the MB preserve dependence with the target under the decoupled distribution. Building on this, DCD iteratively constructs the Completed Partially Directed Acyclic Graph (CPDAG) by exploiting structural asymmetries within the MBs. We establish the theoretical identifiability, soundness, and completeness of DCD. Empirical evaluations demonstrate that DCD achieves strong performance, particularly excelling in challenging noise regimes.
[LG-70] Feature Suppression and Differential Privacy for Residential Traffic Classification: A Two-Home Federated Study
链接: https://arxiv.org/abs/2609.23521
作者: Márton Pál Lipcsey-Magyar,Adrian Pekar
类目: Machine Learning (cs.LG); Networking and Internet Architecture (cs.NI)
*备注:
Abstract:Residential traffic classification supports service management, but learning across homes must account for heterogeneous traffic and privacy constraints. Privacy-aware training may impose uneven costs across traffic categories. We study this tradeoff in simulated two-client federated learning using 1.62 million preprocessed gateway-collected flows across six categories. We compare a full-feature baseline, feature suppression (FS), and differentially private stochastic gradient descent (DP-SGD) under one fixed record-level privacy setting. FS-mild excludes four timing features from 16 model inputs; it provides no formal privacy guarantee. With size-proportional aggregation, FS-mild achieves higher combined macro-F1 and worst-group F1 (the minimum per-class F1 across homes) than DP-SGD in all five seeds at both model capacities under stratified and temporal splits. The tested DP-SGD configuration incurs pronounced minority-category losses, especially in the smaller home, but FS-mild does not uniformly improve on the full-feature baseline. On stratified-split models, loss-based and shadow-model membership probes show near-chance aggregate discrimination without a consistent ranking across probes; this does not establish equivalent privacy. These findings support FS as an input-minimization baseline, not a substitute for formal privacy.
[LG-71] ITSY: Causal Discovery From Irregular Time-Series Data
链接: https://arxiv.org/abs/2609.23516
作者: Wenbo Xu,Yue He,Yunhai Wang,Yueguo Chen,Kun Kuang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Structural causal models for time series recover contemporaneous and lagged effects, but most methods require complete observation windows and become misspecified when samples are missing. We introduce ITSY, the first continuous-optimization method for causal discovery from irregular time series under a linear model. ITSY reformulates the structural equation so that prediction uses the nearest available history rather than the possibly missing current slice, and jointly imputes missing values while learning both graphs. A weighted reconstruction objective corrects the noise transformation induced by this reformulation. Across synthetic regimes varying missingness, scale, graph density, and noise, and on a real world benchmark, ITSY consistently improves graph recovery over representative SCM-based baselines, demonstrating the effectiveness of the proposed method. The results establish a focused solution for irregular linear first-order dynamics and clarify the assumptions required for nonlinear or higher-order extensions.
[LG-72] RACE: Tractable Routing Autoencoder for Clinical ECG
链接: https://arxiv.org/abs/2609.23460
作者: Shunbo Jia,Runze Ma,Haonan Lyu,Haijin Zhang,Qiang Yang,Caizhi Liao
类目: Machine Learning (cs.LG)
*备注: 20 pages, 7 figures, 7 tables
Abstract:Deep learning has advanced automated electrocardiogram (ECG) diagnosis, but the field’s most accurate models, foundation models pretrained on millions of recordings, are not decision-pathway auditable: a clinician cannot trace a diagnosis to a physiological pathway or intervene on one. We propose TRACE, a Tractable Routing Autoencoder for Clinical ECG, whose 32-dimensional clinical latent space is specified in advance from domain knowledge rather than discovered by optimization. TRACE partitions this space into perfusion, structure, and conduction subspaces, routes each to its own diagnostic head by design, regularizes the partition with an orthogonality penalty, and reconstructs the ECG through a decoder that permits latent perturbation. On PTB-XL and Georgia, TRACE exceeds unconstrained classifiers and stays ahead of an ECG foundation model pretrained on ten million recordings, evaluated by linear probe on frozen features, at roughly an eighth of the parameter count. On the nine-label CPSC2018 cohort, which carries no structural class, the framework transfers with only the routing table re-specified to a perfusion/rhythm/conduction partition. Joint probe, erasure, and perturbation analyses verify the routing contract, and perturbing the depolarization and repolarization pathways modulates the reconstructed waveform. Removing the specified partition and its orthogonality penalty costs 1.70 AUC and 11.30 macro-F1 points on PTB-XL, and 2.76 AUC and 16.92 macro-F1 points on Georgia. A capacity-matched permutation control places arbitrary assignments within 0.34 AUC points of the ontology routing and leaves macro-F1 statistically level (p=0.619): the ontology supplies decision-pathway auditability at no macro-F1 cost.
[LG-73] he Evidence Ladder for Reinforcement Learning in Healthcare: From Retrospective Policies to Trusted Interventions
链接: https://arxiv.org/abs/2609.23374
作者: Yunfan Zhao
类目: Machine Learning (cs.LG)
*备注:
Abstract:Reinforcement learning (RL) offers a natural language for healthcare decisions whose conse- quences unfold over time, yet most reported progress remains far from routine intervention. Ex- isting surveys organize the field by algorithm or clinical application. We instead review healthcare RL through an evidence ladder: problem formulation, retrospective identification, policy estima- tion, stress testing, prospective evaluation, and lifecycle monitoring. This view connects clinical treatment, patient engagement, and health-system operations while exposing a recurring gap: evi- dence that a policy scores well in a historical dataset is not evidence that it will improve care. We synthesize the assumptions and failure modes at each rung, identify what evidence can and can- not transfer across settings, and propose reporting practices for cumulative evaluation. Restless bandits are included as one special case, not as the organizing framework. The central lesson is that healthcare RL should be evaluated as an intervention embedded in a changing sociotechnical system, rather than only as an optimizer of a retrospective reward.
[LG-74] What Can a Recurrent State Safely Forget?
链接: https://arxiv.org/abs/2609.23366
作者: Linzhe Zhang,Changming Xu
类目: Machine Learning (cs.LG)
*备注:
Abstract:Recurrent models must preserve information that changes future behavior while suppressing hidden-state error. These objectives conflict: contraction improves stability, but contraction along a future-distinguishing direction destroys memory. We formalize this boundary through the predictive quotient of a recurrent state space. Two hidden states are equivalent when they induce the same conditional future; their equivalence classes form predictive fibers. Every exact semantics-preserving corrector acts as the identity on this quotient. At a regular point with hidden dimension d and predictive dimension k, it can eliminate at most d - k independent directions. This establishes a discrete-continuous boundary: finite predictive states admit positive-radius exact correction basins, whereas an uncountable continuum of future-distinguishable states cannot be decoded after arbitrary positive-radius perturbations in finite-dimensional Euclidean space. To operationalize this principle, we develop an auditable finite-future framework. A compact deployment bank W is evaluated against an independent audit bank A (W subseteq A) on a declared correction domain. Under generative probe access and audit-metric coverage, finite stochastic rollouts furnish a high-probability certificate for the separation margin Omega_W|A(delta). Preserving learned W-predictions within this certified margin guarantees bounded audit-semantic distortion. For intrinsic audit dimension k, the required probe outcomes scale as O(M * Omega^-(k+2)), where M = |A|; a matching minimax lower bound proves this exponent is optimal. Extending guarantees to continuous futures is achieved via an explicit completeness modulus. Controlled experiments validate the certified margins, scaling laws, and automated probe refinement under a safety-first evaluation paradigm. Subjects: Machine Learning (cs.LG) Cite as: arXiv:2609.23366 [cs.LG] (or arXiv:2609.23366v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2609.23366 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-75] Rethinking Class Imbalance for Single-Cell Foundation Models: A Systematic Benchmark Across Architectures and Long-Tail Loss Functions
链接: https://arxiv.org/abs/2609.23325
作者: Zeyu Dong,Jiahui Zhong
类目: Machine Learning (cs.LG)
*备注:
Abstract:Single-cell foundation models (scGPT, scBERT, Geneformer) achieve cell-type classification accuracy up to 97.5% in our experiments, yet this aggregate accuracy can mask systematic failure on rare, often disease-relevant cell populations that long-tail loss functions are widely assumed to address. We present a systematic benchmark of six long-tail loss functions (cross-entropy, weighted CE, class-balanced loss, focal loss, LDAM, logit-adjusted softmax) across three architectures and three datasets (Multiple Sclerosis, Zheng68K, human Pancreas), totaling 162 controlled training runs (3 backbones x 3 datasets x 6 losses x 3 seeds). The gap between overall accuracy, Macro-F1, and rare-class recall under plain cross-entropy is consistent across all nine (architecture, dataset) settings, driven by dataset structure rather than pretraining. Rare-class failure itself splits into two regimes with distinct embedding-geometry signatures, visible before any loss is chosen: some classes are recoverable by the right loss, while others retain linear separability yet are absorbed into unrelated classes’ neighborhoods under every evaluated loss and architecture. Among the recoverable classes, the efficacy of reweighting is predicted by a class’s absolute training-set size, rather than its share of the dataset or the dataset’s overall imbalance ratio. Class-balanced loss and LDAM are the most consistent choices across all nine settings, while logit adjustment trades rare-class precision for recall rather than improving both. Our results give both a reusable benchmark and mechanism-grounded practical guidelines for combining foundation models with imbalanced biological data.
[LG-76] CSC: Calibrated Simplicity for Conflict-Aware Social Bot Detection in the LLM Era CIKM2027
链接: https://arxiv.org/abs/2609.23320
作者: Yipeng Qian,Pengjie Zhao,Chaoxi Niu
类目: Machine Learning (cs.LG)
*备注: Accepted by CIKM 2027
Abstract:Social bot detection is essential for protecting online platforms from misinformation amplification, coordinated manipulation, and distorted public discourse. However, large language models have made social bots much harder to detect from text alone because semantic camouflage is now cheap, fluent, and scalable. The resulting challenge is modality conflict: an account may look human-like in semantics while remaining suspicious in graph structure, profile attributes, or cross-modal consistency. Recent graph-based detectors tackle this limitation by adding graph-side complexity, such as sparse prototype selection, adaptive gating, or architecture-specific control logic, yet our experiments suggest that complexity alone is not the most reliable way to resolve such conflict. We therefore propose CSC, a calibrated-simplicity framework for conflict-aware LLM-era social bot detection. The framework combines three design choices: a simplified prototype-guided graph expert that retains useful structural biases while removing unstable graph-side heuristics, calibrated simplex-constrained fusion that aligns heterogeneous confidence spaces before late fusion, and a lightweight inconsistency expert that models cross-modal disagreement. Experiments on TwiBot-22, TwiBot-20, and MGStBot-large show that \textscCSC improves calibrated operating-point decision quality while remaining competitive across external benchmarks. Further analyses show that calibration improves confidence reliability, the inconsistency expert mainly provides localized corrections in high-conflict or near-threshold regions, and simplified graph-side control yields a better stability-cost trade-off. A targeted semantic-camouflage stress test further shows that replacing selected bot text with matched human text sharply degrades the standalone text expert while leaving graph and fused evidence stable on a balanced challenge set. Comments: Accepted by CIKM 2027 Subjects: Machine Learning (cs.LG) Cite as: arXiv:2609.23320 [cs.LG] (or arXiv:2609.23320v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2609.23320 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-77] Optimal Multi-way Decision Trees for Stratified Sampling in Online Controlled Experiments PRICAI2026
链接: https://arxiv.org/abs/2609.23308
作者: Tomoka Takei,Shunnosuke Ikeda,Yuichi Takano
类目: Machine Learning (cs.LG); Computation (stat.CO)
*备注: 16 pages, 4 figures, The 23rd Pacific Rim International Conference on Artificial Intelligence 2026 (PRICAI 2026)
Abstract:Online controlled experiments, or A/B tests, are widely used to estimate causal effects on digital platforms. A central challenge is to improve experimental sensitivity, or statistical power, without increasing the experimental sample size. Stratified sampling is a classical variance reduction technique; however, its effectiveness depends critically on how the strata are constructed. We thus propose an optimization-based stratification framework for stratified sampling using optimal multi-way decision trees. Our method, called Optimal Multi-way Stratification Trees (OMST), formulates stratification as a path-selection problem over a feature graph. The selected paths define interpretable stratification rules and are optimized using an exact variance-minimizing binary optimization formulation under continuous proportional allocation and a Neyman-type optimal allocation. We incorporate supervised optimal binning to generate outcome-relevant candidate splits for numerical features. Furthermore, we introduce reduction procedures for redundant candidate paths and assignment constraints, substantially reducing the optimization problem size. Experiments on both a real-world and a simulated dataset demonstrate that OMST achieves comparable or superior variance reduction to existing methods while maintaining shallow and interpretable stratification trees.
[LG-78] Optimal No-Regret Learning for Repeated Prophet Inequality
链接: https://arxiv.org/abs/2609.23265
作者: Kun Wang
类目: Machine Learning (cs.LG)
*备注:
Abstract:We study repeated prophet inequalities under prefix feedback. In each of T rounds, a learner encounters fresh values drawn independently from n boxes with unknown [0,1] -supported distributions in a fixed order and must irrevocably accept one, observing only the prefix up to its stopping box. Regret is measured against the optimal stopping policy that knows the distributions. We give an efficient algorithm achieving \widetilde O(\sqrtT) expected regret, matching the lower bound up to logarithmic factors. Our algorithm explores directly through near-optimal policies, combining empirical backward induction with box-specific reach bonuses. A relative-drop aggregation rule then exploits the nesting structure of observed prefixes to preserve exploration, thereby removing the polynomial dependence on the box number n . This resolves an open question posed by Liu et al. (2025).
[LG-79] he Price of Self-Calibration: Exact Evidence Budgets and Manufactured Blind Sets in Adaptive Monitoring
链接: https://arxiv.org/abs/2609.23254
作者: Abdou-Raouf Atarmla
类目: Machine Learning (cs.LG); Systems and Control (eess.SY); Probability (math.PR); Methodology (stat.ME)
*备注: 15 pages, 3 figures, 3 tables
Abstract:Self-calibrating monitors adapt their threshold online to guarantee a prescribed long-run false-alarm rate under arbitrary drift. We compute the price of that guarantee, stating every law with its exact domain of validity. First, the guarantee is an accounting identity, insensitive to what the monitor is meant to detect. Two evidence identities make the cost exact for the online quantile tracker: a persistent step of height \delta yields excess alarm mass within one alarm of \delta/\eta , and exactly \delta/\eta pathwise when \delta is a lattice multiple of the gain \eta ; a ramp of slope c yields a stationary excess rate of exactly c/\eta , independent of accumulated size, up to a boundary c=\eta(1-\alpha) coinciding with the alarm-rate cap. Second, the certificate’s own fluctuation obeys an exact law: the windowed alarm rate has standard deviation of order 1/L , not the binomial 1/\sqrtL , since the windowed mass telescopes to a difference of a tight internal state; the closed-form constant is validated with no fitted parameter. Detectors calibrated on the binomial scale are miscalibrated by \sqrt\eta\varphi(q_0)L , and correct calibration turns detection windows from quadratic to linear in the inverse fault speed. Third, any monitor required to tolerate a drift class \mathcalD is blind, at any horizon and for any rule, to every fault in \mathcalD-\mathcalD ; the proof is a deliberately elementary two-point argument and the contribution is the object it identifies: for speed-bounded classes the blind set is exactly the doubled-speed class, and the tracker absorbs a speed class fixed by its own gain, so that under a certification regime declaring absorbed drift normal, the monitor manufactures \mathcalD . An exact Gaussian projection bound, sharper than Pinsker and never vacuous, quantifies power outside it.
[LG-80] Robot World Models Are Not Invariant to How the Actions Are Written
链接: https://arxiv.org/abs/2609.23252
作者: Ahmed Karim,Leon Chlon
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注:
Abstract:A robot policy is trained with one of two action parameterizations: absolute joint targets, or deltas relative to the current state. The choice is a live engineering decision in robot learning, and a world model conditioned on actions inherits it silently. We show the inheritance is catastrophic. A latent dynamics model trained on one parameterization and handed the identical commanded trajectory written in the other collapses: retrieval degrades by 2.6-13.4x across three robot datasets and two morphologies, goal-conditioned action selection falls from 53% to 15%, and on PushT the two beliefs about the same future are near-orthogonal (cos = 0.067, worst case -0.377), so the predictor does not degrade gracefully, it answers a different question. This is not a distribution-shift artifact in the usual sense: the two encodings are mutually reconstructible at R^2 = 0.996 given the joint input, so no information is lost, and we give the test that separates a valid re-parameterization from a lossy summary or a sensor swap. The test rejected three of the four axes we proposed. The defect lives in the action channel, which the invariance literature for visual models does not examine: work there concerns crops, jitter and camera pose, while the parameterization of the commands goes unaudited. The repair is averaging over the two encodings, and where it goes matters. Averaging the objective restores task performance by itself; averaging the outputs, safe for probabilities by concavity, is not available for direction-valued prediction, where the normalized mean can score below every member of the orbit. What objective-averaging leaves behind is the tail: worst-case agreement stays at 0.78, a disagreement penalty closes it to 0.995, and over a latent rollout it is the difference between a worst case that erodes and one that holds. On PushT, averaging alone does not repair the axis.
[LG-81] Proximal Residual Value Functions for Consistent Planning and Real-Time Execution
链接: https://arxiv.org/abs/2609.23242
作者: Harrison Waldon,Carson Eisenach,Akhil Bagaria,Daniel Russo,Dominique Perrault-Joncas,Alisha Zachariah,Dean Foster
类目: Machine Learning (cs.LG)
*备注:
Abstract:We study two-timescale decision systems in which a planning layer periodically supplies a continuation-value function to a real-time optimizer that allocates arriving resources, with inventory placement as our motivating application. We propose an end-to-end reinforcement learning (RL) method for learning this function using \emphproximal residual value functions, which combine a strictly convex potential of post-decision inventory with a learned convex residual. This general form yields a well-posed optimization layer that supports end-to-end differentiation while preserving an explicit convex objective for real-time execution. We characterize the necessary and sufficient conditions under which a smooth value function yields decisions that are consistent across the planning and execution timescales. In an offline simulation using historical inventory arrival and demand patterns from a large e-commerce retailer, learned proximal residual value functions reduce total routing and transfer cost relative to a historical-production-system proxy by 5.0%.
[LG-82] Causal Inference with Unobserved Confounding: A Mixture Learning Perspective
链接: https://arxiv.org/abs/2609.23219
作者: Mansi Sood,Devavrat Shah
类目: Machine Learning (cs.LG); Signal Processing (eess.SP); Computation (stat.CO)
*备注:
Abstract:Unobserved confounding is a fundamental challenge in causal inference from observational data. This article develops a mixture-learning perspective, viewing latent confounders as sources of heterogeneity that induce mixture structure in observed data. Under suitable structural and identifiability assumptions, recovering the mixing distribution and component mechanisms enables estimation of interventional distributions and causal estimands. Using variants of Bernoulli mixtures as a running example, we contextualize mixture-learning techniques and their structural assumptions, and connect them to causal inference in panel-data settings, including latent factor models and synthetic this http URL then consider high-dimensional exponential-family mixtures with dependent outcome trajectories, moving beyond counterfactual means to model counterfactual distributions. We situate this perspective relative to complementary approaches for unobserved confounding. Together, these ideas provide a bridge between mixture learning and causal inference, connecting recent advances in high-dimensional mixture learning to scalable identification and estimation of causal effects while raising new challenges for mixture learning.
[LG-83] SDC-GON: Singular Decomposition and Consistency-Regularized Greens Operator Networks for Solving Partial Differential Equations
链接: https://arxiv.org/abs/2609.23206
作者: Yingchao Huang,Xin Wang,Shanshan Yao,Fanhua Zeng,Wei Peng
类目: ystems and Control (eess.SY); Machine Learning (cs.LG)
*备注:
Abstract:Green’s function based operator approximation offers an efficient route for solving linear partial differential equations under varying boundary conditions and source terms. Once the Green’s function is learned, solutions for new configurations are obtained through integration rather than by solving the differential equation again. Existing Green’s function learning methods face two structural challenges. The first is the singular behavior of the Green’s function near the source point, which places a difficult approximation burden on neural networks. The second is the absence of explicit consistency between the learned Green’s function and its gradient, although both quantities enter the integral solution representation directly. This work proposes SDC-GON, a Singular Decomposition and Consistency-Regularized Green’s Operator Network that addresses both challenges within a unified framework. The Green’s function is decomposed into an analytically known singular component and a smooth correction learned by the network, so that the neural approximation targets only the regular part of the response kernel. A self-consistency loss enforces agreement between the gradient and the autodifferentiation gradient of the smooth correction. The method is evaluated on two dimensional Poisson, three dimensional heat conduction, heterogeneous reaction diffusion, and Stokes benchmarks, consistently outperforming the compared baselines across all cases. On the heterogeneous pipe benchmark, SDC-GON achieves a testing error of 3.70\times10^-4 with a smaller network architecture, compared with 9.60\times10^-4 for the same-width baseline and 4.63\times10^-4 for a larger configuration, demonstrating that structural improvements are more effective than increasing model size.
[LG-84] Neural Residual Modeling for Scientific Data Compression under Guaranteed Error Bounds
链接: https://arxiv.org/abs/2609.23185
作者: Surya Majumder,Liangji Zhu,Sanjay Ranka,Anand Rangarajan
类目: Machine Learning (cs.LG)
*备注:
Abstract:Lossy compression of scientific simulation data increasingly relies on learned, latent-space architectures such as Residual Vector Quantization (RVQ), which iteratively quantize a base representation and its residuals to progressively reduce reconstruction error. While effective, RVQ performs this residual modeling entirely in latent space, leaving the pixel-space error structure of the reconstruction largely unaddressed. In this work, we propose a post-processing pipeline that augments an RVQ-based compressor with a U-Net trained to predict and correct pixel-space residuals between the original volume and its RVQ reconstruction. We show that these residuals are spatially structured rather than driven by local intensity or gradient features, motivating the need for a deep spatial model rather than simple statistical correction. The U-Net-corrected reconstruction is then passed through a Guaranteed Autoencoder (GAE) stage, which projects the remaining residual onto a per-block PCA basis to enforce a user-specified block-wise error bound. To the best of our knowledge, this is the first pipeline to combine latent-space RVQ, explicit pixel-space residual correction via a deep spatial post-processing network, and GAE-based error-bound guarantees within a single framework for scientific data compression. We evaluate our approach on S3D, JHTDB and E3SM datasets, demonstrating consistent improvements in NRMSE, compression ratio] over RVQ-only and standard residual-correction baselines, while maintaining strict error guarantees required for scientific data fidelity.
[LG-85] K-TRAIL: Simulator-Guided Generative Design of EM/RF Circuits
链接: https://arxiv.org/abs/2609.23183
作者: Piyush Saha,Evan Newell,Hanna O’Leary,Arun Natarajan,Alireza Aghasi
类目: Machine Learning (cs.LG)
*备注:
Abstract:Inverse design of RF and electromagnetic (EM) circuits is challenging because the relationship between circuit layout and electrical response is non-unique, and full-wave simulation is computationally expensive. This paper presents K-TRAIL, a simulator-guided generative framework for automated EM/RF circuit synthesis. K-TRAIL combines diffusion-based layout generation with derivative-free ensemble Kalman guidance, allowing feedback from a black-box EM simulator to refine candidate layouts during generation without requiring adjoint sensitivities or differentiable solver models. The framework supports both synthesis from prescribed S-parameter responses and synthesis directly from RF performance constraints. Experiments on multi-layer RFIC structures show that simulator-guided generation improves agreement with target responses and can identify structurally distinct layouts that satisfy circuit-level design requirements. The proposed approach provides a practical path toward generative, verification-aware RF circuit design while retaining the flexibility to explore diverse layout topologies.
[LG-86] Ask for Any Appliance: A Prompt-Programmable Foundation Model for Non-Intrusive Load Monitoring
链接: https://arxiv.org/abs/2609.23146
作者: Xudong Wang,Jiacheng Cui,Junyu Xue,Tongxin Li,Guoming Tang
类目: Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注: Preprint
Abstract:Non-intrusive load monitoring (NILM) estimates appliance-level consumption from a whole-home meter, but appliance-specific models and fixed output inventories make coverage costly to extend. We present FM4NILM (Foundation Model for NILM), a single prompt-programmable model that estimates a requested appliance’s power trajectory from aggregate measurements, a natural-language description, and optional activation exemplars. A lightweight cadence-aware transformer is pretrained by masked reconstruction on 645k sequences from seven public corpora spanning 1-60 s sampling intervals, then aligned with appliance requests using observation-masked losses for partially labeled households. A Bernoulli-lognormal decoder separates activity detection from conditional power estimation. On held-out households and time periods from REDD, UK-DALE, and REFIT, one frozen text-prompted model serves twelve appliance-corpus requests, achieving 0.556 event F1, 0.625 AUPRC, and the lowest active-window MAE (251.8 W) among seven appliance-specific baselines. Streaming score aggregation raises event F1 to 0.582 with a 60 s aggregation delay. In a separate category-held-out evaluation, adding ten activation exemplars raises microwave AUPRC from 0.132 to 0.214 without parameter updates. Input-intervention ablations probe the model’s dependence on appliance requests and aggregate measurements. These results demonstrate competitive disaggregation with one shared model and support extending appliance coverage through prompts and examples rather than additional specialist networks.
[LG-87] Provably Efficient Reinforcement Learning in Continuous-Time Episodic MDPs with Poisson Decision Epochs
链接: https://arxiv.org/abs/2609.23127
作者: Kenny Guo,Valentio Iverson,Sahan Wijetunga,William Chang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Many real-world reinforcement learning (RL) problems evolve in continuous time, where decisions occur at irregular, event-driven intervals rather than at fixed discrete steps. We study episodic continuous-time Markov Decision Processes (MDPs) in which decision epochs are governed by a homogeneous Poisson process and the reward and transition dynamics vary smoothly over time. We consider both a fixed number of jumps per episode and a fixed time budget with a random number of Poisson decision epochs. Under a Lipschitz continuity assumption in time, we exploit local smoothness through discretization and extend both UCRL (Auer and Ortner 2006) and Q-learning (Jin et al. 2018) to this setting, proving \widetildeO(T^2/3) regret bounds for both model-based and model-free algorithms. Finally, we establish matching \widetilde\Omega(T^2/3) minimax lower bounds, showing that the rate is optimal up to logarithmic factors. These results provide the first tight regret guarantees for Lipschitz-smooth continuous-time episodic MDPs with Poisson decision epochs.
[LG-88] Perplexity Cost Understates What Activation Quantisation Breaks
链接: https://arxiv.org/abs/2609.23125
作者: Anish Sathyanarayanan
类目: Machine Learning (cs.LG)
*备注: 24 pages (9 main text), 12 figures, 26 tables
Abstract:Activation quantisation is usually evaluated with an aggregate metric, perplexity, averaged over every token a model predicts. We ask whether that average identifies which computations a quantiser damages. Perplexity turns out to be a reliable aggregate signal: across 12 models from four families and 780 within-model comparisons, the arm perplexity prefers also retains more induction and more retrieval in all but 2.1 and 4.0 percent of cases respectively. But where perplexity has risen by only a factor of 1.2 to 1.5, induction still keeps 0.959 of its intact accuracy while retrieval has already fallen to 0.554, a gap the aggregate number does not surface. This gap has structure, not just size: a matched Gaussian-noise control of the same per-channel magnitude leaves it largely intact, and randomising only the sign of the quantisation error, every magnitude held fixed, is nearly as harmless, so magnitude alone does not explain the damage. Quantising in a rotated basis, which changes coordinate alignment without changing error magnitude, restores induction from 0.001 to 0.980 at three average bits per token in a single-block intervention, though retrieval recovers less completely at the same setting (0.694); end-to-end at four average bits, induction reaches 0.968 and retrieval 0.534. The pattern holds on two further models up to 32B parameters and, in the deployed configurations we tested, under AWQ once activations are pushed to 4 bits. A perplexity target bounds the average cost of a transformation applied to the activation; it does not, by itself, show which computations survived.
[LG-89] When Does Adversarial Refinement Help? A Negative Result and Open Problem in Adapting R3GAN to Time Series Imputation KDD2026
链接: https://arxiv.org/abs/2609.23102
作者: Yufeng He
类目: Machine Learning (cs.LG)
*备注: 4 pages, 1 figure, 2 tables. Accepted at the 12th Workshop on Mining and Learning from Time Series (MiLeTS 2026), held with KDD 2026. This arXiv version incorporates a post-workshop reproducibility audit of the saved runs. Code: this https URL
Abstract:Diffusion models and transformers have supplanted GANs for multivariate time series imputation, largely on grounds of GAN training instability. R3GAN (NeurIPS 2024) removes that instability via regularized relativistic losses with provable convergence, raising a natural question: do stable, modern GANs revive adversarial imputation? We adapt R3GAN to 1D temporal data with a coarse-to-fine refinement framework and a frequency-domain discriminator, and audit 14 saved configurations across 3 datasets. Because these are heterogeneous single runs, the evidence is descriptive rather than a matched causal ablation. We report a negative result. All five saved mean/zero-start configurations improve by 48.4-70.2%. Among eight eligible non-legacy linear-start configurations, the mean change is -0.7% (range -3.0% to +1.1%); a separate -21.9% legacy logging anomaly is retained for provenance but excluded from that aggregate. In a saved Weather comparison, standalone R3GAN-1D underperforms BRITS by 5.8x. Crucially, we argue the common explanation (that GANs optimize distributional rather than point-wise objectives) cannot be the whole story, since diffusion models also optimize distributional objectives yet achieve state-of-the-art imputation. Our saved reconstruction-weight sweep is consistent with the adversarial signal being inert or harmful, but cannot identify its causal contribution; a matched discriminator-removed ablation is the key next experiment. We frame the precise reason a learned discriminator fails to provide useful refinement gradients (where a learned diffusion denoiser succeeds) as an open problem, and offer practical guidance on when adversarial refinement is worthwhile.
[LG-90] An LLM -Assisted AutoML Framework for Intrusion Detection in IoT Networks
链接: https://arxiv.org/abs/2609.23097
作者: Li Yang
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注: Submitted to an IEEE Journal. Code will be released at: this https URL
Abstract:Internet of Things (IoT) systems are increasingly deployed in smart homes, transportation, energy systems, and critical infrastructure. This broad connectivity improves service intelligence, but also enlarges the attack surface of IoT networks. Machine Learning (ML)-based Intrusion Detection Systems (IDSs) are widely used to identify malicious network threats and protect IoT systems, but developing effective ML-based IDS models often requires human expertise and repeated manual decisions on many procedures, including data pre-processing, feature selection, model selection, and hyperparameter tuning. Automated Machine Learning (AutoML) reduces this burden by automating steps of the ML pipeline using optimization techniques, but conventional AutoML methods can consume substantial optimization time because they explore broad candidate model families and large hyperparameter spaces. This paper proposes a Large Language Model (LLM)-assisted AutoML framework for IoT intrusion detection. The proposed framework uses an LLM as a policy generator that converts dataset profiles into bounded and validated AutoML policies for automated data balancing, automated feature engineering, and Combined Algorithm Selection and Hyperparameter Optimization (CASH). Under an equal 10-trial budget, the proposed LLM-assisted policy achieves higher weighted test F1-score than traditional AutoML using the Tree-structured Parzen Estimator (TPE) on both datasets, reaching 99.680% on CICIDS2017 and 99.186% on IoTID20. Relative to the broader 30-trial Traditional AutoML-TPE baseline, the 10-trial proposed method reduces optimizer time by 63.7% and 49.9%, respectively, while achieving slightly higher F1-score. These results show that a bounded LLM policy can improve the quality of a low-budget AutoML search while retaining a clear efficiency advantage relative to a larger conventional search budget.
[LG-91] Counting and Covering in Nearest-Neighbour Representations of Boolean Functions
链接: https://arxiv.org/abs/2609.23094
作者: Martin Anthony
类目: Discrete Mathematics (cs.DM); Machine Learning (cs.LG); Combinatorics (math.CO)
*备注: 55 pages, 1 diagram, 1 table
Abstract:We study the number of prototypes needed to represent Boolean functions by nearest-neighbour classification. There are two distinct settings: the prototypes may be arbitrary points of Euclidean space, or they may themselves be required to lie in the Boolean cube. For unrestricted prototypes, we strengthen a known lower bound for almost all Boolean functions. The bound applies simultaneously to nearest-neighbour voting rules with any number of voting neighbours, and substantially narrows the gap with the known general upper bound. We obtain a VC-dimension bound for classes with a bounded number of prototypes, and show that it is sharp in order in dimensions at least four. We then study Boolean prototypes, beginning with symmetric threshold functions. A connection with covering designs expresses the minimum number of prototypes at every threshold level exactly in terms of a covering number, and leads to further exact results for related monotone functions, including disjunctive extensions and a characterisation of when a representation with a single negative prototype is possible. For a uniformly random Boolean function, the Boolean nearest-neighbour complexity, as a proportion of the cube, is asymptotically close either to one half or to one, with explicit limiting probabilities. In particular, almost every Boolean function requires at least approximately half as many prototypes as there are points in the cube, and one half is the largest proportion for which such a lower bound holds. Finally, we consider arbitrary symmetric Boolean functions. Their Boolean nearest-neighbour complexity is closely approximated by a weighted vertex-cover problem on paths. As a consequence, a uniformly random symmetric function typically requires prototypes amounting to 11/20 of the cube. This is much larger than the upper bounds known when the prototypes are allowed to lie anywhere in Euclidean space.
[LG-92] he Role of Coordinates in Pareto Regret for Adversarial Multi-Objective Bandits
链接: https://arxiv.org/abs/2609.23092
作者: Changkun Guan,Mengfan Xu
类目: Machine Learning (cs.LG)
*备注:
Abstract:Adversarial multi-objective bandits hold the potential to help us optimize choices (arms) whose reward is a multidimensional vector chosen by an adversary and whose performance is measured by Pareto regret. We define loss as one minus reward and measure the easiness of a coordinate by the smallest cumulative loss of the arms on it, and call the coordinate easier when this quantity is smaller. Existing work suggests that in theory an easier coordinate may reduce Pareto regret. However, in practice, one may not know which coordinate is easier. On the negative side, we show that this lack of information eliminates the possibility: a smaller cumulative loss does not improve the worst-case order of Pareto regret. Precisely, let (L_d) be the smallest cumulative loss along coordinate d over T rounds. For (K\ge4) arms, (T\ge6) rounds, and at least 2 coordinates, we prove that the minimax expected Pareto regret is (\Omega(\min\T-L_0,\sqrtK(T-L_0))). It is monotonically decreasing in (L_0), even when (L_0=\min_d L_d) itself is known. On the positive side, this result motivates the possibility that other coordinates, not just the easy one, may suffice to attain the optimal rate of Pareto regret. When L_0 is known, we apply Poly-INF to a fixed coordinate and obtain an upper bound on Pareto regret that exhibits the same order and thus matches the lower bound. Without such knowledge, we develop a reward-doubling version of Poly-INF that adapts to this unknown quantity while still attaining the matching minimax rate. Another implication is that it has no extra (\log T) factor and is independent of the number of coordinates.
[LG-93] Measured Joules Learned Routes: Learning to Route for Energy-Efficient LLM Serving
链接: https://arxiv.org/abs/2609.23085
作者: Muhammad Abdur Rab Siddiqui,Daniela Rojas,Chen Yang,Wenqi Cui,Yuanyuan Shi,Yize Chen
类目: Performance (cs.PF); Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG); Systems and Control (eess.SY)
*备注: 16 pages, 10 figures, in submission
Abstract:Large language models (LLMs) and agentic AI systems are creating rapidly growing inference energy demands as model sizes grow and reasoning trajectories extend. While in practice, many queries do not require the capabilities of the largest available model, and routinely directing such queries to a high-capability model can introduce unnecessary, considerable computation and energy consumption. In this paper, we investigate whether adaptive routing across a heterogeneous pool of LLMs can reduce this energy burden without substantially compromising task performance. We design a language-model-based router that reads in each query and selects an answer model from a fixed candidate pool. The candidate models are first profiled through an offline tournament that records their correctness, latency, power, and GPU energy for each query. Using these measurements, the router is trained through supervised fine-tuning followed by group relative policy optimization (GRPO) with the tailored paradigms. Results demonstrate that learned routing can selectively allocate expensive model capacity based on query context and improve the accuracy-energy tradeoff in multi-LLM serving. Across seven benchmark tasks, we also observe a sharp accuracy-energy phase transition among routers, providing practical insights into improving energy efficiency while maintaining LLM performance.
[LG-94] AirGC-CD: Gaussian-Circulant Precoding for Exactly Debiasable PAPR Reduction in Over-the-Air Federated Learning
链接: https://arxiv.org/abs/2609.23084
作者: Jonggyu Jang,Hyeonsu Lyu,Hyun Jong Yang
类目: Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注:
Abstract:Over-the-air federated learning lets edge devices transmit their local updates simultaneously, reducing the communication overhead. The resulting waveform, however, has a peak-to-average power ratio (PAPR) that grows with the model dimension, and keeping the amplifier in its linear range leaves two remedies: clipping the peaks or backing off the transmit power. Neither remedy is without cost: i) the clipping distortion appears at the receiver as a bias that cannot be removed, and ii) back-off keeps the signal intact but degrades the average signal-to-noise ratio (SNR). Independent of this trade-off, the transmission remains uncompressed, spending one channel use per model parameter, which keeps large-model training out of reach. To address these challenges, we propose AirGC-CD, an over-the-air scheme that precodes each local update with a partial Gaussian circulant matrix before clipping. In AirGC-CD, the precoder’s output is exactly Gaussian regardless of the update’s sparsity, so the clipping function is designed for a known distribution instead of inheriting it from the data. This enables the clipping to be inverted on average by a single scalar Bussgang gain in closed form, and we prove that the resulting aggregate is exactly unbiased, with clipping adding only variance. The clipping ratio is then the only free parameter left, trading the variance of the clipping against the SNR loss from back-off, and we derive its near-optimum in closed form. Since the precoder is linear, it also acts as a compressor, reducing the transmission from the model dimension d to the sketch dimension m at a cost of only O(dlog d) via two fast Fourier transforms, whereas a Gaussian sketch costs O(md). Experiments on five image datasets show that AirGC-CD outperforms baseline over-the-air FL schemes in most settings, particularly at low SNR, while using fewer channel uses per round.
[LG-95] Optimizers for Diffusion Models: A Controlled Benchmark
链接: https://arxiv.org/abs/2609.23055
作者: Arman Bolatov,Egor Shulgin,David Li,Abduragim Shtanchaev,Sebastian U. Stich,Maxim Panov,Eric Moulines,Peter Richtárik,Martin Takáč
类目: Machine Learning (cs.LG)
*备注: 5 figures, 10 tables. Code: this https URL
Abstract:Discrete diffusion models now match autoregressive language models on several benchmarks, while the question of how best to train them has received far less attention: the optimizer is inherited from one paper to the next and never compared. New optimizers, meanwhile, are validated almost exclusively on autoregressive pretraining, a different objective on a different loss surface. We present a controlled optimizer benchmark across four diffusion formulations, to our knowledge the first for discrete diffusion: seven optimizers (AdamW, Lion, Muon, SOAP, MARS, MARS-M, Schedule-Free) on masked diffusion (text8), uniform diffusion (QM9, and LM1B through the Gaussian duality) and Gaussian diffusion on images (CelebA-64), each on a task with published reference values. Every optimizer receives the same search protocol, and every winner is retrained at the full budget with three seeds. AdamW is a strong default but not always the right choice: it is beaten by a resolved margin on two of the four tasks, and the winner changes with the formulation, so the optimizer deserves the same care as the rest of the training recipe. Notably, methods validated on autoregressive language model pretraining transfer well: Muon, MARS-M and SOAP each beat the tuned AdamW on at least one diffusion formulation. The benchmark, all runs and every figure are reproducible end to end from the released code at this https URL.
[LG-96] Anatomy of a Closed-Loop Collapse: A Causal Case Study of a Compressed VLA Policy IROS2026
链接: https://arxiv.org/abs/2609.23048
作者: Fengze Jia(The Ohio State University)
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: 8 pages, 1 figure, 5 tables. Accepted as a poster at the IROS 2026 Workshop on Building Scalable Infrastructure for Robot Learning: From Data Scaling to Real-World Deployment (ScaleInfra), Pittsburgh, PA, USA. Supplementary records: this https URL
Abstract:Compressed manipulation policies can pass offline evaluation while failing in closed-loop execution; this dissociation is established in prior work and is not our claim. We contribute a causal anatomy of one naturally occurring case. An 8-layer distillation of Octo-Base retains 86% of parameters, passes every offline check we applied (0.996 and 1.000 teacher-ratios on the family’s own validation metrics), and collapses in closed loop: 0/72 vs. the teacher’s 40/72 on a simulated WidowX pick-and-place task. The collapse is structured, not diffuse: early task stages degrade gradually (the student moves the object at 90% of the teacher’s rate and grasps at 55%), while transport-to-target fails categorically, at 0% in every training variant. Paired action-trace forensics isolate the signature: a negative, late-heavy z residual, roughly 10x its post-repair magnitude, and persistent across the base distillation and both continuation branches. Four standard therapies fail under matched controls: continued training and in-domain offline data leave success at zero, even though the latter measurably improves marginal action statistics; command-level compensation recovers nothing at any offset, although the same perturbations degrade healthy policies; clamping the symptom in the command channel preserves grasping, yet success stays at floor. A minimal-pair intervention that substitutes half of the training stream with deployment-distribution teacher rollouts, with every other setting held fixed, restores parity with the teacher (18/36 vs. 17/36 held-out), eliminates that signature, and recovers a teacher-like perturbation-response profile. We claim existence, not universality. Operationally, offline gates, including a family’s own validation metrics, are insufficient acceptance tests for compressed policies; a few dozen closed-loop trials sufficed to find what they missed.
[LG-97] WaveFront Decoding: Parallelized Self-Speculative Decoding for Looped Language Models
链接: https://arxiv.org/abs/2609.23033
作者: Hyeongju Ha,Jae-Joon Kim
类目: Machine Learning (cs.LG)
*备注: 17 pages, 7 figures, 8 tables
Abstract:Looped language models repeatedly apply a weight-shared block to increase effective depth without increasing parameter count, but the resulting T sequential recurrent-block calls per generated token substantially increase decoding latency. To address the issue, we introduce Wavefront Decoding (WFD), a training-free self-speculative decoding framework designed for looped language models. WFD exploits two properties of these architectures: intermediate recurrence outputs provide effective draft predictions, and weight sharing allows token states at different positions and recurrence depths to be processed in one batched recurrent-block call. WFD organizes these mixed-depth states into a diagonal wavefront, continuously drafting new positions at shallow depth while advancing earlier positions toward full-depth verification. Unlike the phase-separated draft-then-verify schedule, WFD therefore co-batches drafting and verification within the same recurrent calls, while rejected drafts are corrected using full-depth predictions. Across six Spec-Bench task categories, WFD achieves 2.42x speedup on Ouro-2.6B and 3.54x on Huginn-3.5B over autoregressive decoding, consistently outperforming draft-then-verify. Cross-recurrence KV sharing further reduces wavefront KV traffic and increases WFD’s speedup to 4.81x on Huginn-3.5B.
[LG-98] Silent Failures at the 232 Boundary: A Technical Report on Large-Tensor Matrix Multiplication in PyTorchs Apple MPS Backend
链接: https://arxiv.org/abs/2609.22991
作者: Junichiro Niimi
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注:
Abstract:Apple Silicon machines with 192GB or more of unified memory make it routine to place tensors with more than 2^32 elements on a desktop GPU. We show that PyTorch’s Metal Performance Shaders (MPS) backend silently returns wrong results for batched matrix multiplication at this scale. On macOS 27.0, this http URL, and therefore this http URL and eager attention, returns relative errors above 1 without an exception or a warning, in every PyTorch release from 2.4.1 to 2.14.0 that we tested. On one machine, we sweep bmm over two dtypes, four memory layouts, six shapes and 42 batch sizes between 4096 and 65538 (1584 runs on PyTorch 2.14.0, and a reduced sweep on ten earlier releases), and judge every result against a float64 computation on the CPU. Three rules account for every outcome on 2.14.0. When the output exceeds 2^32 elements and an operand is a transposed view, the entire output is wrong and equals a computation that ignores the strides of that operand. When a contiguous input exceeds 2^32 elements, only the batches beyond that point are wrong, and they equal a computation whose index wraps around at 2^32 . Operands that are views with at least 2^31 elements raise an exception instead, so a larger problem can turn an explicit error into a silent failure. A control on CUDA is correct for bmm, although this http URL is silently wrong above 2^32 elements there as well. In a public sentiment classifier, one oversized batch corrupts a third of the outputs and collapses them onto a single class. The study is black-box: we report what the backend returns, compared with reference results. We release the sweep harness, the raw results and a guard that stops any MPS operation touching 2^32 or more elements at this https URL.
[LG-99] On attention heads and bilinear forms
链接: https://arxiv.org/abs/2609.22990
作者: Andrew O’Desky
类目: Machine Learning (cs.LG); Spectral Theory (math.SP); Machine Learning (stat.ML)
*备注: 30 pages, 14 figures
Abstract:We study the symmetric and antisymmetric parts of bilinear forms in the attention heads of trained large language models. We introduce an orthogonally invariant profile map from real bilinear forms to a three-dimensional simplex and observe that profiles of trained bilinear forms accumulate near profiles of rank-one bilinear forms. We prove that the symmetric part of a bilinear form in an attention head is the sum of a hyperbolic form and a zero form for a Zariski-dense subset of query-key matrices.
[LG-100] LPINNs: First-Layer Gated Localization for Physics-Informed Neural Networks
链接: https://arxiv.org/abs/2609.22984
作者: Lakshay Chawla,Hardik Jain
类目: Machine Learning (cs.LG); Computational Engineering, Finance, and Science (cs.CE); Neural and Evolutionary Computing (cs.NE)
*备注:
Abstract:Physics-informed neural networks (PINNs) use one shared representation over the computational domain, which can become difficult to optimize on long domains and for high-order operators. We study a minimal alternative: multiply the first hidden activation of an otherwise unchanged dense PINN by input-dependent localization functions, giving first-layer units receptive fields without partitioning the domain or adding interface losses. We screen 13 families of localization functions, in up to three parameterizations each, on a nonlinear harmonic oscillator (HO), a heat equation on a long spatial interval, and a manufactured four-dimensional (4D) fourth-order problem, with ten paired seeds throughout. Three configurations give large reductions in solution error at matched budgets: (i) Fixed Gaussian localization functions on the 2\pi HO domain cut mean solution RMSE from 4.8369\times10^-1 to 8.83\times10^-3 at 3k epochs. (ii) The inverse-quadratic family with learnable centers and widths cuts it from 2.896\times10^-1 to 3.06\times10^-2 on the 8\pi heat domain at 10k epochs. (iii) Fixed bump localization functions cut it from 1.75947\times10^1 to 2.260\times10^-1 on the 4\pi 4D domain at 10k epochs. Every paired seed improves in these three comparisons. The screen also shows that the mechanism is not a free win: on HO only 2 of 13 families beat the baseline, and 10 of the remaining 11 are 9 to 23 times worse; on 4D four families are non-finite and five are more than three orders of magnitude worse than the baseline. The inverse-quadratic family is the only one that beats the baseline on all three equations. Overall, these results show that first-layer localization can provide measurable improvements to baseline PINNs on long-domain and high-order problems.
[LG-101] Connectivity-Aware Exploration of Robotic Grasp Spaces
链接: https://arxiv.org/abs/2609.22983
作者: Maksim A Kazanskii
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注:
Abstract:Robotic grasping is typically formulated as the problem of identifying successful actions from a space of candidate grasp poses. However, the organization of successful actions within this space has received less attention. We study the multiscale structure of viable robotic grasps in SE(3) and investigate whether this structure can be exploited for more efficient exploration. Using a large-scale grasp dataset, we show that successful grasp sets exhibit heterogeneous and reproducible connectivity structure across objects. We then introduce a connectivity-aware sampling strategy that incrementally explores the currently observed grasp space by prioritizing potential bridges between components, structural frontiers, boundary extensions, and geometric novelty. In controlled reconstruction experiments, the method recovers the connectivity structure of successful grasp sets substantially more efficiently than random sampling and farthest-point sampling. We further evaluate whether connectivity acquired under hidden grasp viability can improve subsequent grasp discovery, and whether structural experience from previously explored objects can be retrieved and transferred to unseen objects. These results suggest that the spatial organization of viable actions provides information relevant to grasp-space exploration beyond the viability of individual candidate actions. More broadly, they motivate structure-aware exploration as a means of exploiting the geometry of viable action spaces in robotic manipulation.
[LG-102] oken Utility Is Selection-Conditioned: Coupled Selection of Prompt Context and Response Supervision for Efficient Instruction Tuning
链接: https://arxiv.org/abs/2609.22943
作者: Can Wu,Xinrui Chen,Ou Wu,Yi Du
类目: Machine Learning (cs.LG)
*备注: Work in progress
Abstract:Efficient large language model (LLM) instruction tuning requires selecting response supervision with supporting prompt context. Existing methods typically value both sides separately, risking selection-state mismatch between valuation and retained training subsets. BRIDGE (Budgeted Response-Prompt Interaction via Directional Gradient-guided Efficient Token Selection) captures selection-conditioned token utility through a shared validation-directed interaction surrogate valuing each side under the other’s retained state. Budgeted alternating selection coordinates retained subsets by aggregating precomputed interactions over the current opposite-side subset to update conditional scores. Structure-aware projection converts conditional response scores into coherent supervision spans. Across three model families, BRIDGE leads compared selection methods overall in mathematical reasoning, code generation, and instruction following. In mathematical reasoning, its advantage over independent selection grows with compression.
[LG-103] Joint Domain-Class Modeling for Federated Learning Under Feature Skew
链接: https://arxiv.org/abs/2609.22932
作者: Sina Najafi,Mostafa Tavassolipour,Seyed Pooya Shariatpanahi
类目: Machine Learning (cs.LG)
*备注:
Abstract:Federated learning (FL) enables collaborative model training without centralizing private data, but performance often degrades under feature skew: clients share labels while the conditional input distributions p_i(x!\mid!y) vary due to latent, client-specific appearance factors. We propose Joint Domain-Class Federated Learning (JDFL), a lightweight, optimizer-agnostic extension that makes this latent domain variation usable without sharing raw data. JDFL first infers domain clusters called pseudo-domains from brief local update signals. It then expands the classifier head to output M\times C , joint (domain-class) logits. This allows the model to represent domain-conditioned appearance while keeping a shared backbone. To train the expanded head we introduce two complementary supervision strategies based on simple intuitions: a similarity-aware soft-labeling that transfers evidence between nearby inferred domains while allowing domain-specific specialization, and a per-sample randomized target assignment that perturbs supervision across the joint outputs and serves as a low-cost training-time regularizer. JDFL integrates with existing standard FL methods (e.g., FedAvg, SCAFFOLD) with minimal changes. Empirically, both supervision modes consistently improve global test accuracy on standard domain-shifted image benchmarks; ablations and sensitivity studies show the gains stem from the proposed supervision and parametrization rather than mere capacity increase.
[LG-104] Computationally efficient safe exploration in reinforcement learning
链接: https://arxiv.org/abs/2609.22919
作者: Shreeram Murali,Shankar A. Deka,Dominik Baumann
类目: Machine Learning (cs.LG); Robotics (cs.RO); Systems and Control (eess.SY)
*备注: 8 pages
Abstract:Reinforcement learning in real-life applications requires safety guarantees during exploration. Typical reinforcement learning algorithms do not provide such guarantees, and many modifications that do rely on Gaussian processes (GPs), which have a large computational cost. We propose a computationally lightweight algorithm based on the Nadaraya-Watson estimator that safely explores and optimizes constrained Markov decision processes (MDPs). Our algorithm, \textscCoLSafe-MDP, uses an estimator that scales in constant-time with bounds on the estimates, a significant improvement from its GP-based counterparts that scale cubically with the number of data points. We then evaluate its performance in a grid-based environment and on observational Martian terrain data.
[LG-105] Prioritized Rollouts for Efficient World Model-based Vision-Language-Action Policy Optimization
链接: https://arxiv.org/abs/2609.22879
作者: Yifei Sheng,Haoxiang Ren,Zhilong Zhang,Haonan Wang,Runjie Xu,Yihao Sun,Nan Tang,Zhichao Wu,Lei Yuan,Haoxin Lin,Yang Yu
类目: Machine Learning (cs.LG)
*备注:
Abstract:Vision-Language-Action (VLA) models have emerged as a powerful paradigm for embodied intelligence, but fine-tuning them with reinforcement learning (RL) remains constrained by the cost of real-world robot interaction. Model-based reinforcement learning (MBRL) reduces this cost by using a learned world model to generate rollouts for policy optimization. However, it becomes computationally expensive as VLA policies and world models scale. Existing methods typically treat states equally, overlooking substantial differences in their utility for policy improvement. In this paper, we show that policy uncertainty helps identify states with greater potential for policy improvement. The policy exhibits high uncertainty at only a small subset of states, often during decision-sensitive stages where small action differences can alter task outcomes, suggesting that policy improvements at these states could be particularly valuable. Building on these findings, we introduce U-GROW, a lightweight, plug-and-play sampling layer that directs more model rollouts to these informative states. By modifying only the branched-start distribution, U-GROW can be integrated into existing MBRL pipelines without changing the policy optimization objective. Experiments in both simulated and real-world manipulation tasks demonstrate the efficiency and effectiveness of U-GROW, supporting the use of policy uncertainty to guide experience generation.
[LG-106] Leverag ing Inference-Time Compute for Diffusion Models via Global Scheduling of Denoising Trajectories
链接: https://arxiv.org/abs/2609.22867
作者: Yuan Cao,Yifu Tang,Hangqi Li,Zeyu Zheng
类目: Machine Learning (cs.LG)
*备注:
Abstract:Diffusion models generate a sample by traversing a denoising trajectory, a sequence of stochastic noise-reduction steps that transforms pure noise into a draw from a target distribution. At deployment time, additional computation can improve sample quality without retraining: at each step, the sampler draws several candidate noise samples, scores the resulting predictions with a quality criterion called the verifier, and retains the best candidate at the cost of one network evaluation per candidate. This raises a resource allocation question: given a fixed budget of function evaluations, how should search effort be distributed across the steps of the denoising trajectory? We formulate this as a computational budget allocation problem. First, we show that, to leading order in the step size, the expected gain from evaluating K candidates at a step factorizes into an endogenous, step-specific sensitivity parameter times a universal sample-size factor equal to the expected best of K standard-normal draws. Second, for a fixed sensitivity profile, the optimal allocation solves a separable concave integer program with water-filling structure; at fixed total sensitivity, its advantage over uniform allocation increases with sensitivity dispersion in the majorization order. Third, we prove that when sensitivities vary across instances, no adaptive policy can avoid worst-case regret that grows linearly in the trajectory length, which motivates a design that anchors the allocation offline and adapts online only to recover instance-specific slack. We extend the analysis from independent random search to a broader family of local search operators, and instantiate it as an implementable algorithm. Experiments on three families of diffusion samplers show that the proposed allocation attains the quality of the uniform benchmark with 20 to 50 percent fewer function evaluations.
[LG-107] Causilo Technical Report
链接: https://arxiv.org/abs/2609.22866
作者: Minyong Cho,Minho Jeong,Dooho Lee,Jinmo Lee,Jaemin Yoo
类目: Machine Learning (cs.LG)
*备注:
Abstract:We introduce Causilo, a tabular foundation model (TFM) that combines frontier predictive performance with exceptionally fast inference. On TabArena, Causilo achieves 1785.4 Elo, at a median inference time of 0.10 seconds per 1K test samples. It outperforms TabPFN-3.5-Fast with 31.6% less inference time, placing it on the performance–efficiency Pareto frontier. Causilo follows TabICL’s column-then-row architecture but introduces another row-refinement module before row compression. This module exchanges information among cell representations within each row after column encoding. The refined cells then visit the context set again through an additional column stage before being compressed into row embeddings. For inference efficiency, both row stages use cross-attention through a fixed number of summary tokens, keeping their attention cost linear in the number of features. Pretrained on approximately 36M synthetic tables, Causilo delivers strong benchmark results across TabArena, BeyondArena, and ScoringBench, achieving frontier-level performance with substantially faster inference.
[LG-108] CurvFlow-DTA: dual-graph discrete Ricci curvature flow for drug–target affinity prediction
链接: https://arxiv.org/abs/2609.22862
作者: Jicheng Ma,Yunyan Yang,Juan Zhao,Liang Zhao
类目: Machine Learning (cs.LG); Differential Geometry (math.DG)
*备注:
Abstract:Graph neural networks are widely used for drug–target affinity (DTA) prediction, and discrete Ricci curvature has recently been used to characterize molecular graph geometry. Existing curvature-aware DTA approaches mainly use static curvature on the drug graph while representing proteins primarily with sequence-derived features. This leaves pair-adaptive use of graph geometry underexplored, which may limit adaptation to unseen entities in cold-start settings relevant to practical screening. We present CurvFlow-DTA, which replaces a single static curvature representation with weighted Forman curvature flow on both molecular and protein residue–residue contact graphs. A label-independent flow trajectory is precomputed for each entity, and a pair-conditioned selector determines the horizons read by a dual-branch Flow-GINE. A frozen ESM-2 supplies residue-level representations and contact scores used to construct the protein graph. Inference requires only SMILES strings and protein sequences, without a bound complex structure. On Davis and KIBA, CurvFlow-DTA improves on the protocol-matched Ricci-GraphDTA baseline in every warm and cold-start setting. Warm-split mean squared error (MSE) decreases by 19.9% on Davis and 18.9% on KIBA. Across the six cold-start comparisons, MSE decreases by 14.3 – 27.4% , with higher concordance index (CI) throughout. Within our compiled set of literature baselines, CurvFlow-DTA achieves the lowest MSE on both warm benchmarks and across four out of six cold-start evaluation settings. Subjects: Machine Learning (cs.LG); Differential Geometry (math.DG) ACMclasses: I.2.6; J.3; G.2.2 Cite as: arXiv:2609.22862 [cs.LG] (or arXiv:2609.22862v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2609.22862 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-109] Intelligent Degradation Monitoring in Lithium-ion Batteries via Discharge Incremental Capacity Feature Estimation
链接: https://arxiv.org/abs/2609.22843
作者: Amir Madmolilvand,Farzaneh Abdollahi
类目: ystems and Control (eess.SY); Machine Learning (cs.LG); Signal Processing (eess.SP)
*备注:
Abstract:Accurate and timely detection of degradation in lithium-ion batteries is crucial to ensure safety, reliability, and longevity in high-demand applications such as electric vehicles and energy storage systems. Traditional incremental capacity (IC) analysis methods require low-current cycling for discharge measurements, limiting their practical use in real-time battery management. This paper proposes a novel neural network-based framework that predicts discharge IC features directly from charging signals, eliminating the need for low-current discharge. Trained on a comprehensive dataset of 53 battery cells cycled under diverse fast-charging protocols, the model demonstrates robust generalization ability, effectively estimating degradation indicators on unseen battery data. Among several architectures evaluated, the LSTM model provides the best balance of prediction accuracy and computational efficiency. The proposed approach enables real-time integration into Battery Management Systems (BMS), enhancing degradation monitoring without disrupting normal battery operation. This paper presents a study to bridge IC analysis with practical, fast-charging scenarios, marking a significant step towards intelligent and scalable battery health monitoring.
[LG-110] A Horizon-Independent Regret Bound for Optimistic Hedge in General-Sum Games
链接: https://arxiv.org/abs/2609.22839
作者: Junsoo Ha
类目: Computer Science and Game Theory (cs.GT); Machine Learning (cs.LG)
*备注:
Abstract:Can simple learning rules keep their regret bounded in self-play? Recent work achieves constant regret bounds through modified regularization and higher-order prediction. Yet for Optimistic Hedge, arguably the most canonical method in games, the best known individual regret bound remains logarithmic. In this work, we prove that plain Optimistic Hedge with a constant step size can attain O_n,d(1) individual regret in general-sum games with n players and d=(d_1,\ldots,d_n) actions, under expected loss-vector feedback. As a corollary, its time-averaged play enjoys an O_n,d(1/T) coarse correlated equilibrium (CCE) gap. Our analysis represents Optimistic Hedge as a real-analytic recurrence on a compact space, which yields an exact finite-order difference relation that eliminates horizon dependence. Our proof hinges on nonconstructive Noetherianity argument of Frisch (1967), so the (n,d) -dependence remains implicit.
[LG-111] A Hybrid Attention Model Learning Unified Time-aware Patch Representation for Irregular Multivariate Time Series Forecasting
链接: https://arxiv.org/abs/2609.22836
作者: Zhihao Lin,Li Lin,Qi Zhang,Kaiwen Xia,Shuai Wang,Jialin Qiao
类目: Machine Learning (cs.LG)
*备注:
Abstract:Time series foundation models (TSFMs) have recently delivered impressive zero-shot performance across diverse forecasting tasks. However, real-world decision-making frequently relies on \emphirregular multivariate time series (IMTS), where inconsistent inter-observation intervals and asynchronous sampling across variables coexist with informative missingness. Existing TSFMs handle such inputs either through imputation that injects spurious values or through index-based positional encodings that ignore continuous time. There is still a gap in the foundation model that follows the original IMTS patterns. In this paper, we propose a hybrid attention model that learns a unified time-aware patch representation for IMTS forecasting. We first design a \emphtime-aware patch encoding that maps a variable number of intra-patch timestamps into a fixed-size embedding, producing a uniform format for irregular patches without resorting to imputation. We then introduce a \emphtime bias attention mechanism that calibrates inter-patch temporal misalignment and asynchronous cross-channel dependencies as auxiliary attention offset. Finally, on top of a decoder-only Transformer backbone, we adopt a \emphhybrid causal mask that preserves a bidirectional full view over the historical context while keeping the forecast horizon strictly autoregressive. To support large-scale pretraining under irregular settings, we also curate VersaTSA, an archive of 30 B observations that retains the native sampling sparsity of its sources. Experiments on three IMTS benchmarks and a standard regular-MTS benchmark show that our model achieves state-of-the-art zero-shot performance on IMTS and remains competitive when transferred to regular forecasting.
[LG-112] Beyond Averag e Error through Oracle-Informed Stress Tests for Time-Series Forecasting
链接: https://arxiv.org/abs/2609.22820
作者: Xu Lin(1),Runheng Zuo(1),Shengxuan Xu(1),Qitai Tan(1),Hongyu Lin(1),Xiao-Ping Zhang(1) ((1) Tsinghua University, Beijing, China)
类目: Machine Learning (cs.LG)
*备注: 34 pages, 13 figures, 23 tables
Abstract:Average squared error cannot reveal whether forecasting performance degrades because the future becomes less predictable or because forecasts move farther from the conditional mean. We introduce paired, mechanism-controlled stress tests that decompose changes in expected squared error at each lead time into environmental risk and forecast-oracle distance, using an origin-conditioned predictive oracle unavailable to the evaluated models. Three end-to-end controls have known attribution. Specifically, the null, environmental-only, and information-gap controls verify that the pipeline assigns changes to the correct component. We then apply the benchmark to 24 deployable forecasters. Under frequent switching, 14 methods have higher realized MSE but lower oracle distance; under outlier-variance feedback, 19 have higher MSE but lower scale-standardized MSE. Short- and long-lead stress-response rankings have Spearman correlation 0.624, revealing substantial horizon-dependent reordering. We then study multivariate relation shifts. Across six models and three coupling severities, oracle distance accounts for only 0.7-3.9% of the decomposed expected-risk increase, and environmental-risk majority persists in an eight-channel system and a matched-difficulty audit of Ring, Block, and Hub relations. Finally, prespecified contrasts on independent data-generating process (DGP) realizations show that several visually compelling discovery profiles, including trend accumulation and the hypothesized switching reversal, do not replicate. The benchmark thus combines component-wise diagnosis with a held-out stability audit. It complements real-data out-of-distribution evaluation, which measures performance under realistic shifts when exact oracle attribution is unavailable.
[LG-113] Counterfactual Tool Ranking under Utility Cost and Privilege Constraints
链接: https://arxiv.org/abs/2609.22819
作者: Jiapeng Li
类目: Machine Learning (cs.LG)
*备注: Working paper, version 2. Includes synthetic execution studies, BFCL-derived function-selection experiments, and local open-weight LLM baselines. Code and artifacts: this https URL
Abstract:Counterfactual tool evaluation must distinguish authority, historical support, and what a comparison actually estimates. We study these distinctions with eleven executable enterprise-inspired tools, exact-propensity logs, and real local Model Context Protocol transport. An initial 45-run synthetic study is retained, then challenged by 30 realized-return control runs and 15 experiments on 1,930 independently released Berkeley Function Calling Leaderboard (BFCL) tasks. Full-return direct regression reverses an initially favorable doubly robust (DR) evaluation result in the linear setting: mean absolute errors are 0.0139 for direct regression and 0.0272 for DR. Under a shifted environment, DR retains an advantage, with errors 0.0227 versus 0.0948. On function-name-group-disjoint BFCL-derived splits, direct and DR selectors obtain balanced accuracies of 81.85% and 79.83%. Two pinned local Qwen2.5 models are evaluated on the same 200 held-out tasks, exposing a strong failure to abstain under the fixed prompt. We further characterize policy differences under missing support: unsupported actions shared by two policies cancel, allowing point identification of an incremental change when neither absolute value is identifiable. A disagreement-preserving fallback achieves this property in all five support-gap runs, but conservative sampling bounds do not certify deployment improvement. The contribution is a falsifiable evaluation method and independent public evidence, not a new DR estimator, official BFCL leaderboard score, or production-agent safety claim.
[LG-114] FIRM-WM: State-factorized factual-interventional recurrent modeling for reward-free visual planning
链接: https://arxiv.org/abs/2609.22816
作者: Yilun Wu,Yunjian Zhang,Aobo Li,Mujiangshan Wang,Haitao Wu,Aqiang Zhang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Reward-free latent world models can learn from offline videos and solve new image–goal tasks by optimizing actions through predicted latent futures. This setting places two demands on the planning state: its coordinates must be comparable with a goal image. Moreover, its dynamics must retain velocity, motion trend, contact, and other history–dependent information beyond those goal coordinates. Offline training creates a second mismatch: each recorded trajectory reveals one factual future, whereas a sampling–based planner compares many actions that were not taken from the same state. We introduce FIRM-WM (Factual–Interventional Recurrent World Model), a compact pixel world model designed around these two gaps. Its recurrent state separates a typed, goal–comparable configuration from a 128-dimensional dynamic fiber used for prediction but excluded from the terminal goal cost. Broad factual trajectories provide state coverage, while common–reset intervention branches provide observed outcomes for alternative action sequences. Before executing each branch, we reset the environment and restore the same recorded values exposed by the environment’s state–setting interface. Under matched CEM planning and three independent full-pipeline seeds, FIRM-WM reaches 99.0 \pm 1.0% on TwoRoom, 92.7 \pm 2.1% on Reacher, and 88.0 \pm 3.0% on OGBench-Cube, compared with 89.0%, 88.0%, and 70.0% for LeWM. The deployed model uses 2.98–3.42M parameters and records 2.13–11.60 \times lower planning time on these tasks.
[LG-115] Robust Market Making with Hawkes Order Flow and Price Impact via Adversarial Reinforcement Learning
链接: https://arxiv.org/abs/2609.22785
作者: Hao Yang,Zhenguo Xu
类目: Machine Learning (cs.LG)
*备注: 18 pages, 2 figures. Includes theoretical analysis, numerical experiments, and supplementary results in the appendix
Abstract:Market-making strategies in real limit order book markets face substantial model uncertainty and regime-shift risk. Existing adversarial reinforcement learning approaches improve robustness by formulating the Avellaneda–Stoikov market-making problem as a zero-sum game between a market maker and an environmental adversary. However, these approaches typically rely on Poisson order arrivals and neglect trade-induced price impact, limiting their ability to capture important high-frequency market microstructure effects such as clustered order flow, self-excitation, and post-trade price feedback. We extend adversarial reinforcement learning for market making to a more complex environment with Hawkes self-exciting order arrivals and trade-induced price impact. To mitigate the increased non-stationarity introduced by the expanded regime space, we incorporate an LSTM module that explicitly models the temporal structure of recent observations. We further characterize the equilibrium properties of the proposed framework through both game-theoretic analysis and numerical experiments, and introduce a robustness evaluation protocol focused on improvements in the left tail of the return distribution. Experimental results across a range of market regimes show that the proposed method achieves improved left-tail performance in most complex microstructure environments. In particular, the gains are pronounced in regimes with strong Hawkes excitation and low-to-moderate price impact. Bootstrap tests provide no evidence that these improvements are obtained through a stronger terminal directional inventory bias. These results suggest that combining adversarial training with temporal state representation can improve the robustness of reinforcement-learning-based market-making strategies under order-flow self-excitation, price impact, and regime uncertainty. Comments: 18 pages, 2 figures. Includes theoretical analysis, numerical experiments, and supplementary results in the appendix Subjects: Machine Learning (cs.LG) Cite as: arXiv:2609.22785 [cs.LG] (or arXiv:2609.22785v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2609.22785 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-116] Improved Private Sparse Covariance Estimation with Multiscale Threshold Tests
链接: https://arxiv.org/abs/2609.22783
作者: Zihan Zhang
类目: Machine Learning (cs.LG)
*备注:
Abstract:We study differentially private covariance estimation in operator norm for mean-zero sub-Gaussian distributions with unknown covariance support and at most k nonzero entries per row. We develop a multiscale random-threshold algorithm with sample complexity \ot(k^2/\alpha^2+k\sqrt d/(\alpha\varepsilon)) for (\varepsilon,\delta) -differential privacy and error at most \alpha\sigma^2 , where d is the dimension and \sigma is a known sub-Gaussian scale. The bound improves the privacy-dependent term of the existing \ot(k^2/\alpha^2+k^3/2\sqrt d/(\alpha\varepsilon)) \citepkumar2026curse upper bound by a factor of \sqrt k , and matches the lower bound of \widetilde\Omega(k^2/\alpha^2 + k\sqrtd/(\alpha\varepsilon)) in its applicable parameter regime. Our key technical ingredient is a direct operator-norm bound on the centered fluctuations of an ideal reconstruction, exploiting conditional independence rather than accumulating entrywise errors across each row. A multiscale allocation of threshold tests balances reconstruction variance against query sensitivity. Together, these ingredients sharpen the trade-off between approximation error and privacy protection, removing the additional \sqrtk factor from the privacy-dependent sample complexity. Subjects: Machine Learning (cs.LG) Cite as: arXiv:2609.22783 [cs.LG] (or arXiv:2609.22783v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2609.22783 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-117] Look Before You Steer: Geometry Predicts SAE Feature Steerability ICML2026
链接: https://arxiv.org/abs/2609.22782
作者: Muhammad Khan,Shlok Channawar,Akshaj Gurugubelli,Girish Gupta,Aditya Shah
类目: Machine Learning (cs.LG)
*备注: 12 pages, 3 figures, 5 tables. Accepted at the Mechanistic Interpretability Workshop at ICML 2026 and AIW 2026 at COLM
Abstract:Steering with SAE features requires per-feature coefficient tuning, which currently demands intervention sweeps. We ask whether properties of the SAE itself, computable before any forward pass, predict which features will be cheap or expensive to steer. We show that variation in SAE feature steerability is partially predicted by decoder-space geometry: neighbor density and maximum cosine similarity to nearby decoder directions, both computable from the SAE weight matrix before any intervention, rank features by how much steering they require for a fixed behavioral effect ( \rho up to -0.546 , p 10^-6 , AUROC 0.610-0.822 across conditions; the signal is rank-based, consistent with grid discreteness). This geometry-steerability relationship replicates across two Gemma-2 model scales (2B and 9B), two SAE widths (16K and 65K), and is detectable cross-architecturally on Llama-3.1-8B-Instruct ( \rho = -0.266 , n = 300 ). On Qwen3-8B with BatchTopK SAEs, geometry predicts whether a feature is steerable at all but not the continuous ordering among responsive features, revealing a boundary condition tied to SAE training regime. The signal weakens at deep proportional layer depth in both models, where the cost of steering exceeds our intervention budget, a consistent depth boundary. These results provide preliminary evidence that pre-steering geometry can partially inform coefficient selection, offering a path toward screening features for controllability before deployment.
[LG-118] Algorithmic Collusion and the Complexity of Information-Value-Free Equilibria
链接: https://arxiv.org/abs/2609.22757
作者: Ioannis Anagnostides,Weiqiang Zheng
类目: Computer Science and Game Theory (cs.GT); Computational Complexity (cs.CC); Data Structures and Algorithms (cs.DS); Machine Learning (cs.LG)
*备注:
Abstract:A (coarse) correlated equilibrium (CE) is information-value-free (IVF) if a player can match the payoff obtained from recommendations by committing to a fixed action. Motivated by the problem of regulating algorithmic collusion, this refinement was introduced by Hartline, Wang, and Zhang [EC’26], who showed that it can be computed in polynomial time in explicitly represented normal-form game. In this paper, we examine the complexity of IVF©CEs in succinct games, which model more realistic strategic interactions that feature either many players or exponentially many pure strategies. We first show that computing an information-value-free CE is PPAD-complete in many-player polymatrix games or two-player Bayesian games, even when the approximation is a constant. We also prove an unconditional exponential query lower bound. Our results establish that IVFCEs are intractable, even in the centralized model, and rule out the existence of any efficient learning dynamics. This significantly strengthens the impossibility result of Hartline, Wang, and Zhang, which concerns a particular class of learning algorithms, and furnishes strong computational critiques of recent regulation on algorithmic collusion. To sidestep these hardness results, we examine the complexity of information-value-free CCE. Certain no-regret algorithms—such as regret matching or FTRL—provide a fully polynomial-time approximation scheme (FPTAS) for this problem. The complexity when the approximation is exponentially small turns out to be nuanced. On the one hand, leveraging no-regret dynamics, we establish membership in \textCLS = \textPPAD \cap \textPLS . On the other hand, we show that it is at least as hard as the P-matrix linear complementarity problem, and hence as hard as simple stochastic games. This shows that even IVFCCEs are unlikely to admit a polynomial-time algorithm barring a major breakthrough. Subjects: Computer Science and Game Theory (cs.GT); Computational Complexity (cs.CC); Data Structures and Algorithms (cs.DS); Machine Learning (cs.LG) Cite as: arXiv:2609.22757 [cs.GT] (or arXiv:2609.22757v1 [cs.GT] for this version) https://doi.org/10.48550/arXiv.2609.22757 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-119] D-IMPL: A Diffusion-based Solver for Parameterized BBOs
链接: https://arxiv.org/abs/2609.22752
作者: Yang Hu,Na Li
类目: Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注: 12 pages, 4 figures, 3 tables
Abstract:Diffusion models have demonstrated strong power in generative modeling tasks across multiple domains, exhibiting a remarkable capability of learning complex distributions from samples. In this paper, we leverage such capability to design an efficient universal diffusion-based solver for parameterized black-box optimizations (BBO), where the optimizer has only black-box access to queries of the objective function at the learning stage, yet is able to reduce the additional computational cost at the inference stage for each BBO instance while also capturing the potential multi-modal landscape of non-convex objectives. To cast our formulation as a compatible generative modeling task, we introduce the notion of minimization policy as a new solution concept, which defines a sampling distribution over the solutions that should concentrate around the minimizer set for each BBO instance. We then propose Diffusion-based Iterative Minimization Policy Learning (D-IMPL), a practical generative-model-based solver for solving parameterized BBOs that employs diffusion models to learn a minimization policy, whose density is proportional to the exponential of the negated objective values, thereby amortizing the computational costs across different BBO instances. Furthermore, we demonstrate the performance of our D-IMPL algorithm by establishing a sample complexity guarantee showing that a \delta -approximate minimization policy can be effectively learned within O(\log(1/\delta)) iterations, and by extensive empirical evaluations over a range of constrained and unconstrained BBO tasks.
[LG-120] Autonomous Model Lifecycle Management for Digital Twin-Based Manufacturing Control
链接: https://arxiv.org/abs/2609.22701
作者: Zhengyang(Cissy)Gu,Thomas Cook,Fredaljohn Rohrbaugh,Joseph E. Hernandez,Chris Couch
类目: Machine Learning (cs.LG)
*备注: The paper has been accepted for publication in the 2026 IEEE International Conference on Systems, Man, and Cybernetics (SMC 2026). The final published version will be made available on IEEE Xplore
Abstract:Manufacturing AI systems must autonomously adapt to continuous distributional shift from raw-material variability, ambient changes, and equipment aging, under strict safeguard and operator-trust requirements where model failures risk physical damage. This paper presents a closed-loop Cyber-Physical System (CPS) for autonomous model lifecycle management in automotive manufacturing, deployed since 2023. The system manages product-specialized model pairs: a sequence-to-sequence physics model (LPP) serving as a digital twin, and a deep Reinforcement Learning (RL) control policy (LCP) trained against it. Per retraining cycle, multiple model variants spanning architecture families and RL algorithms compete; only the best-scoring candidate advances. A Conductor orchestrator autonomously manages plant-wide model inventories with dependency-aware retraining and Proportional-Integral-Derivative (PID) fallback. Reflecting the principle of Human-Centric Intelligence, the LCP composite score embeds an operator-trust gate penalizing policies deviating from established practice; without it, 23% of policies are rejected by operators despite passing accuracy thresholds. Across multiple facilities, LCP-controlled processes achieve process stability improvements of 28-45% over uncontrolled baselines with zero safety incidents.
[LG-121] Multi-Armed Bernoulli Bandits via Minimax Single-Arm Stopping
链接: https://arxiv.org/abs/2609.22690
作者: Huikang Liu,Zhengchao Wang,Daniel Kuhn,Wolfram Wiesemann
类目: Machine Learning (cs.LG); Optimization and Control (math.OC); Machine Learning (stat.ML)
*备注: This work is a preprint and has not been submitted to any conference or journal yet
Abstract:We develop an index policy for finite-horizon Bernoulli multi-armed bandits from minimax solutions to single-arm bandit (SAB) problems. Each SAB problem involves choosing between an unknown Bernoulli arm and a known reward. We show that minimizing worst-case regret of SAB problems over all non-anticipative policies admits an exact semi-infinite linear programming formulation. The resulting stopping policies offer a natural way to compare arms: the higher the known reward against which a policy continues sampling, the more promising the unknown arm. We turn this intuition into indices based on cumulative continuation probabilities, with a monotone adjustment and a reward-shortfall cap. By relating index errors to the regret of single-arm stopping policies, we establish a distribution-free regret bound of 4.45\sqrtKT+10.75K for K arms and horizon T . This bound matches the minimax-optimal regret order established in the literature. The guarantee extends to rewards supported on [0,1] through Bernoulli randomization. We also provide a finite-grid implementation with quantified approximation loss. In numerical experiments, the SAB-based index policy achieves lower worst-case regret than every tested benchmark policy across all evaluated numbers of arms and horizons, while closely matching the grid-based MAB minimax policy in the two-arm setting.
[LG-122] StateMem: Single-State Residual Memory with Adaptive Inference for Vision-Language-Action Policies
链接: https://arxiv.org/abs/2609.22684
作者: Wenzhuo Li,Qiongfeng Shi,Yi Zhou
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: 9 pages, 5 figures, conference
Abstract:Memory-dependent robotic manipulation often requires later actions to use information from earlier interactions. Existing vision-language-action (VLA) policies primarily rely on current observations, limiting historical information retention. Memory-augmented VLAs, such as MemoryVLA, address this limitation with external memory banks but require explicit storage and retrieval. To address these limitations, we propose StateMem, a single-state residual memory framework for VLA policies that uses prediction error to update a persistent memory token through low-rank residuals and to adaptively route cached prefixes. A training-free controller adjusts the routing threshold online, while fast correction compensates for stale prefix features during cache reuse. We evaluate StateMem on LIBERO, RoboMemArena, and real-world manipulation tasks. On LIBERO, StateMem achieves an average success rate of 97.6% and reduces the average VLM prefix refresh rate by 20.25% relative to full refresh. In the Occlusion category of RoboMemArena, StateMem achieves the best performance among single-VLA methods, reaching 21.8% Task Success Rate (TSR) and 44.3% Cumulative Success Rate (CSR). Across six real-world manipulation tasks, it achieves +21% in average success rate.
[LG-123] Mask-Aware Execution for Efficient JEPA Training
链接: https://arxiv.org/abs/2609.22674
作者: Md Musfiqur Rahman Sanim,Zhihao Shu,Bahram Afsharmanesh,Amirali Mirian,Wei Niu,Gagan Agrawal
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注:
Abstract:Joint Embedding Predictive Architectures (JEPAs) are becoming a core representation-learning primitive and a building block for latent world models across vision, video, audio, brain dynamics, and time series. Despite (potential of) wide deployment, current JEPA training pipelines are inefficient: each input is executed through multiple mask-specific branches, with redundant target-side work, and memory-bound token routing. These costs grow with the number of masks and limit GPU efficiency. We present M-JEPA, a mask-aware execution architecture that restructures JEPA training without changing the learning objective. M-JEPA separates mask-independent computation from mask-dependent routing, enabling shared context encoder execution, fused token routing and slicing with backward support, sparse target encoder execution over the union of target tokens, and masked patch embedding for sparse inputs. The resulting pipeline preserves training semantics while reducing computation, memory traffic, and synchronization overhead. We implement M-JEPA for five JEPA variants and evaluate it on NVIDIA A100 GPUs. Compared against the state-of-the-art baselines, M-JEPA achieves up to 1.7x end-to-end training speedup for 2-10 masks. Separately, with masked patch embedding, 4.75x patch-embedding speedup at high sparsity. These results show that execution restructuring, rather than changes to the JEPA objective, is a key lever for efficient JEPA training.
[LG-124] Monotone-Constrained Diffusion Models for Long-Horizon Production Forecasting
链接: https://arxiv.org/abs/2609.22643
作者: Temesgen Mikael Abraha,Yves Lucet
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 34 pages, 4 figures. Under review at the Journal of Machine Learning Research. Extended version of a paper at Canadian AI 2026 (PMLR 318, pp. 332-341)
Abstract:Forecasting a long horizon from only the first observations of a sequence is ill-posed: many trajectories are consistent with the same short history. We study this problem in oil and gas production forecasting, where forecasts made after roughly the first fifth of a well’s producing life drive development and abandonment decisions, and where a usable forecast must describe a monotone decline. We present Physics-SIMS-TS, a conditional diffusion forecaster that combines negative guidance against synthetic artifacts, decline-curve constraints and an isotonic projection applied during sampling, spatial training augmentation, and an ensembled stochastic sampler yielding a full predictive distribution. Across three jurisdictions and more than 35,000 wells, under a shared-space, validation-frozen protocol, Physics-SIMS-TS is the most accurate diffusion forecaster in the comparison and is competitive with, but not superior to, ensembled transformer forecasters. Its forecasts are monotone by construction at a cost of at most 0.5% in mean squared error, and its trajectory ensemble yields calibrated intervals after one dispersion factor is fitted per jurisdiction. On six standard benchmarks a reversible-instance-normalization variant of the backbone is the leading diffusion baseline. We also quantify four protocol choices on which the measured ranking depends. Code and evaluation artifacts are released.
[LG-125] Classification with Abstention Under Class-Conditional Error Constraints
链接: https://arxiv.org/abs/2609.22632
作者: Mohammadreza M. Kalan,Yuyang Deng,Sanaz Hamidi
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:We study binary classification with abstention under separate class-conditional error constraints, with the objective of minimizing abstention while keeping both errors below prescribed thresholds. We characterize the distribution-free minimax rate of excess abstention risk, up to logarithmic factors, in terms of the complexity of the hypothesis class and the sample size. To make the framework amenable to computation with models such as neural networks, we introduce surrogate-loss formulations and derive finite-sample guarantees for excess surrogate ambiguity risk. We formulate the resulting learning task as a constrained optimization problem and characterize its computational complexity in the convex setting. Finally, we evaluate our approach on various datasets and compare its performance with a competing method for this problem.
[LG-126] Concurrency-Aware Process Model Forecasting with Causal Nets
链接: https://arxiv.org/abs/2609.22614
作者: Yongbo Yu,Jari Peeperkorn,Johannes De Smedt,Jochen De Weerdt
类目: Machine Learning (cs.LG); Databases (cs.DB)
*备注: 13 pages, 2 figures
Abstract:Process model forecasting (PMF) aims to predict the process model that will characterize a future period, thereby providing a process-level view of how behavior is expected to evolve. Existing PMF methods, however, forecast directly-follows graphs, which cannot explicitly represent concurrency. We extend PMF to causal nets by forecasting time series of relation and binding counts and using these forecasts to reconstruct future process models with AND/XOR semantics. To evaluate the resulting models, we introduce a protocol that accounts for partial traces and constructs the workflow nets required for conformance checking. Experiments on four event logs show that the forecasted models achieve conformance levels close to those of models re-mined from observations in the corresponding future windows. They also outperform static discovery baselines, which retain high precision on the structurally stable log but exhibit substantial precision losses on the other three logs. Filtering infrequent bindings improves most conformance metrics, although it also removes much of the concurrent behavior captured by the models.
[LG-127] User-Level Handover Decision Making Based on Machine Learning Approaches
链接: https://arxiv.org/abs/2609.22593
作者: João Lima,Alvaro Medeiros,Eduardo Aguiar,Vicente Angelo de Sousa Junior,Tarciana Guerra
类目: Machine Learning (cs.LG); Networking and Internet Architecture (cs.NI)
*备注:
Abstract:This letter covers a broad comparison of methods for classification and regression applications for a user-level handover decision making in scenarios with adverse propagation conditions involving buildings, coverage holes, and shadowing effects. The simulation campaigns are based on network simulator ns-3. The comparison encompasses classical machine learning approaches, such as KNN, SVM, and neural networks, but also state-of-the-art fuzzy logic systems and latter boosting machines. The results indicate that SVM and MLP are the most suitable for the classification of the best handover target, although fuzzy system SOFL can perform similarly with lower processing time. Additionally, for the download time estimation, LightGBM provides the smallest error with short processing time, even in hard propagation scenarios.
[LG-128] WIG: A Time-Causal Wavelet Operator for Autoregressive Forecasting on Irregular Graphs
链接: https://arxiv.org/abs/2609.22585
作者: Subashree Venkatasubramanian,David A. Barajas-Solano,Chuyang Liu,Daniel M. Tartakovsky,Dipankar Dwivedi
类目: Machine Learning (cs.LG)
*备注: 23 pages, 5 figures. Preprint
Abstract:We introduce TWIG (Time-Causal Wavelet Operator for Irregular Graphs), a graph-native neural operator for autoregressive surrogate modeling on static irregular graphs. TWIG transforms each node history into causal multiscale temporal features that separate recent variation from progressively slower memory components, then propagates these features through graph-wavelet operator blocks with gated pointwise channel mixing. The architecture is causal by construction and designed for closed-loop forecasting, where predictions are recursively reused as future inputs. We evaluate TWIG on three irregular-domain forecasting problems spanning regional diffusion, three-dimensional subsurface hydrology, and aerodynamic flow, with graphs ranging from 400 to 5,233 nodes and model capacities from approximately 70k to 10M parameters. TWIG achieves the lowest aggregate rollout errors on the subsurface-hydrology and regional-diffusion benchmarks and ranks second on the 10M-parameter aerodynamic-flow benchmark, behind the GPS Transformer. Across all three settings, TWIG consistently outperforms the corresponding non-time-causal Graph WNO baseline. These results demonstrate that TWIG provides an effective and scalable approach to stable autoregressive forecasting of dynamical fields on irregular graphs.
[LG-129] Augmenting PID Control with Deep Reinforcement Learning: A Hybrid Approach to the Industrial Benchmark
链接: https://arxiv.org/abs/2609.22584
作者: Zhengyang(Cissy)Gu,Joseph E. Hernandez,John Burtenshaw,Sean Scott,Thomas Cook,Chris Couch
类目: Machine Learning (cs.LG)
*备注: Published in: 2026 7th International Conference on Artificial Intelligence, Robotics, and Control (AIRC)
Abstract:As industrial processes grow in complexity, traditional Proportional-Integral-Derivative (PID) controllers are often insufficient for handling their non-linear, multi-input dynamics. We propose using advanced Deep Reinforcement Learning (DRL) to prove its advantages in these complex environments. To do this, we rely on the Industrial Benchmark (IB). The IB is a realistic simulation that tests DRL algorithms against the key challenges of industrial applications: high-dimensional state spaces, delayed effects, and conflicting multi-criterial objectives. This testbed highlights DRL’s core trade-off: while its final policies can often be unstable, its unique strength is the ability to autonomously discover optimal, non-obvious policies in multi-dimensional spaces where simple controllers fail. In this paper, we propose a novel hybrid PID-RL controller that leverages DRL’s discovery capability while ensuring Reliability. After developing a multi-objective reward function to make DRL viable, we use a twin-delayed deep deterministic (TD3) agent as a discovery tool to find the optimal, non-obvious settings for the IB’s ‘Gain’ and ‘Shift’ parameters. By feeding these discovered parameters to a simple, tuned PID controller, our hybrid model successfully combines all three characteristics: it achieves the optimal Performance and Efficiency of the best DRL agent with the Reliability of a classical controller. This work demonstrates a practical methodology for using DRL to augment, rather than replace, trusted industrial control systems.
[LG-130] Benchmarking Hybrid Deep Learning Architectures for Predictive Maintenance in Industry 4.0
链接: https://arxiv.org/abs/2609.22583
作者: Zhengyang(Cissy)Gu,Joseph E. Hernandez,Thomas Cook,John Burtenshaw,Sean Scott,Chris Couch
类目: Machine Learning (cs.LG)
*备注: Published in: 2026 IEEE International Conference on AI and Data Analytics (ICAD)
Abstract:Predictive maintenance in Industry 4.0 refers to using data from sensors, machines, and production systems to estimate when equipment is likely to fail, so maintenance can be planned before a breakdown occurs [1]. However, a model that predicts maintenance may work perfectly in the lab but fail unexpectedly when applied to real factory data [2]. To solve this “reliability” gap, we evaluated six deep learning architectures across more than 700 experimental runs. We focused on the two dominant approaches in the field: Recurrent Neural Networks (RNNs), which process data step-by-step, like reading a sentence [3], and Transformers, a recent dominant approach, which look at the entire sequence at once to spot important connections [4]. We examined whether Transformers still outperform recurrent neural networks (RNNs) when the data includes noise [5]. We found that while Transformers excelled at tracking stable, slow-moving processes, they tend to overreact to chaotic data, mistakenly taking sensor noise for meaningful signals [6]. We also found that the hybrid method that combines a Long Short-Term Memory (LSTM) layer with a Transformer layer is more resilient to noisy data from factory shops [7]. Functioning as a noise filter, the LSTM smooths out data volatility, allowing the Transformer to focus on the bigger picture without being distracted [8]. The hybrid model did not just improve accuracy; it proved to be significantly more consistent than complex models, delivering reliable predictions regardless of how chaotic the underlying system became.
[LG-131] Defusing Explosive Prompts: Understanding and Preventing Trigger-Based Prompt Injections in LLM Agents
链接: https://arxiv.org/abs/2609.22510
作者: Justin Szczepaniak,Elad Feldman,Naum Viner,Ben Nassi
类目: Cryptography and Security (cs.CR); Machine Learning (cs.LG)
*备注: First two authors contributed equally
Abstract:As LLM applications integrate with external tools, they are increasingly exposed to indirect prompt injection (IPI), where adversarial instructions are embedded in retrieved content. Conventional IPIs fire on contact: the moment an agent ingests the content, it carries out the instruction. We introduce the explosive prompt, a conditional payload that stays dormant until an attacker-chosen trigger is met, in effect a training-free, inference-time backdoor planted in a single piece of retrieved content. This temporal separation reaches where ordinary IPI cannot. On frontier models that refuse the bare imperative almost entirely, rephrasing the same goal as a dormant conditional drives real, state-changing tool execution against a live agent backend (a paired mean of 16.5% vs. 2.4% for the imperative, reaching 34.2% on a proprietary model). In trials on nine production agents (OpenAI Codex, Google Gemini CLI, Anthropic Claude Code CLI, Cursor CLI, GitHub Copilot, Devin AI CLI, Amazon Kiro CLI, Qwen Code, Google Assistant; n=30 each), explosive prompts succeed in 43-83% of cases versus at most 3% for an imperative baseline, and they slip past deployed defenses: off-the-shelf injection classifiers are miscalibrated on them, and a preference-optimized model that closes imperative injection entirely still executes 11.8% of explosive prompts, every one at the trigger turn. The durable defensive lever is ingestion-time detection of the conditional structure, once detectors are trained on explosive-prompt data, which no prior benchmark supplied and our generator does. Retraining cuts live tool-execution attack success from an undefended 34.3% to 7.5-8.1% for the encoder baselines. Our detector, DeFuse, reaches 3.0% at a calibrated 5% false-positive budget with the best detection quality of any method tested (AUC 0.9994) and 25x lower latency, though it needs length-aware thresholds. Comments: First two authors contributed equally Subjects: Cryptography and Security (cs.CR); Machine Learning (cs.LG) Cite as: arXiv:2609.22510 [cs.CR] (or arXiv:2609.22510v1 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2609.22510 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-132] EmbeddGAN: A Novel GAN Framework Using an Embedding Network and Gini Distance Correlation
链接: https://arxiv.org/abs/2609.22508
作者: MaTais Caldwell,Yixin Chen,Xin Dang,Charles Walter
类目: Machine Learning (cs.LG)
*备注: 18 Pages, 14 Figures, Submitted to IEEE Access, Awaiting Revisions
Abstract:Generative Adversarial Networks (GANs) have demonstrated strong performance in generating high-quality synthetic data. However, they are limited by no formal guarantees regarding convergence and the effectiveness of the learning process. In practice, this leads to training instability, mode collapse, and sensitivity to hyperparameters. To address this, we propose EmbeddGAN, a novel adversarial training framework based on a dependence-based objective. Instead of relying on a discriminator that classifies samples as real or fake, EmbeddGAN introduces an embedding network that learns a representation in which statistical dependence between samples and their real/fake labels is maximized, while the generator is trained to minimize this dependence. This objective is implemented using the Gini distance correlation (gCor), which equals zero if and only if the embeddings are statistically independent of the real/fake label. Minimizing this objective therefore encourages real and generated samples to become statistically indistinguishable in the learned embedding space. The embedding network projects both real and generated data into a shared low-dimensional space, where distributional discrepancies can be measured directly through pairwise distances. We adopt a minimax training strategy: the embedding network maximizes the Gini distance correlation (maximizing dependence), while the generator minimizes it (minimizing dependence). Experiments on the MNIST, CIFAR-10, and CelebA datasets demonstrate that EmbeddGAN achieves competitive performance relative to established baselines while exhibiting notably stable training dynamics on the evaluated datasets.
[LG-133] COREM: Cosine-Relation Momentum Reshaping with Stateful Writeback
链接: https://arxiv.org/abs/2609.22487
作者: Yan Wang,Xiaochuan Wang,Yuxiang Sun
类目: Machine Learning (cs.LG)
*备注:
Abstract:Matrix-valued optimizer states may contain relational structure that is not captured by treating their entries independently. We study whether relations within matrix-valued optimizer states can be exploited to improve optimization. To this end, we introduce a unit-relation-transform abstraction and instantiate it as COREM, a Cosine-Relation Momentum Reshaping method with stateful writeback. COREM partitions the momentum state into update units, computes cosine relations among them, and uses these relations to reshape the momentum before writing the transformed state back to the optimizer. This stateful mechanism allows the reshaped momentum to affect not only the current update but also future optimization dynamics. We evaluate COREM on CIFAR-10 with an MLP and on enwik8 with a Transformer. Compared with Muon, COREM shows lower early-stage step efficiency but stronger improvement in the mid-to-late stages of training, achieving better final validation performance on CIFAR-10 and comparable final performance on enwik8. Spectral diagnostics on enwik8 show that COREM consistently increases entropy effective rank and reduces the concentration of singular energy in dominant modes, while preserving an anisotropic spectrum. For square matrix updates, COREM requires approximately 13.3% of the transformation FLOPs of Muon with five Newton-Schulz iterations.
[LG-134] Connected Content Retriever: Dense Graph Edge Features Powering Pre-Ranking at LinkedIn
链接: https://arxiv.org/abs/2609.22441
作者: Akhilesh Gupta,Sudarshan Srinivasa Ramanujam,Chirag Bhanuprasad Mehta,Reshma Asharaf Beena,Dhritiman Das,Birjodh Singh Tiwana,Bhargavkumar Kanubhai Patel,Mack Lee,Renyi Tang
类目: Machine Learning (cs.LG)
*备注:
Abstract:In large-scale recommendation systems like the LinkedIn Feed, content generated by a member’s network (connections and follows) makes up over 70% of impressions and engagement. It is therefore essential that the pre-ranking layer forwards the best possible few hundred candidates to the ranking layer. LinkedIn’s professional knowledge graph carries engagement signals across both the first degree network (connections and follows) and the second-degree network: posts that a 1st-degree connection reacted to, commented on or reshared but did not author (a.k.a. stranger viral). Due to this fan out, the resulting candidate index exceeds one billion; selection of activities from the viewer’s network narrows it down to roughly tens of thousands of activities that must be scored within a 120 ms p99 latency budget. We present Connected Content Retriever (CC Retriever), a pre-ranking system that scores these candidates with a full deep ranking model on GPUs at low latency. At its core is a sorted-search GPU primitive that joins dense graph affinity features (viewer to author) with document level features stored on the GPU at runtime in 5-10 ms. The shift to GPU served scoring enabled a 50x scale up of the ranking model’s parameters and delivered a +2.5% lift in content time spent on the LinkedIn Feed in online experiments, significantly higher than the typical gains observed in LinkedIn Feed experiments. In this work, we describe the feature set we leverage from LinkedIn’s economic graph and the model architecture used for scoring, with a particular emphasis on the online system that scales the stack.
[LG-135] Complex-valued Phase-Coherent Transformers
链接: https://arxiv.org/abs/2609.22415
作者: Leona Hioki
类目: Machine Learning (cs.LG)
*备注:
Abstract:Complex-valued Transformers have inherited softmax attention over the raw complex inner product. Outside natively complex domains this standard form stays near chance, and no complex attention had been shown to correct it. We show that the match must be a scaled cosine score: L2-normalise queries and keys, so the score reads their cosine similarity and ignores their magnitudes, and hold that score at order-one scale. With this the same models train on four diagnostic tasks under two different gates; without the normalisation they stay at chance on ListOps and Needle under both gates and fall far below on the other two, and a normalised score placed at too small a scale fails as well. The resulting family of phase-coherent Transformers (\PCT) matches or exceeds the strongest real-valued baseline across long-range memory, positional retrieval, hierarchical reasoning, frequency-domain classification and physical complex signals; it shows no degradation up to depth 20; and its loss decreases log-linearly over a 61-fold range of parameters. A member of the family, complex screening combined with a phase-coherent recurrence, is the first genuinely complex-valued neural network to solve Path-X, with 91.6% of its trainable parameters complex-valued against 38.2% for S4. We record these as signs of generalisation not previously seen in complex-valued neural networks.
[LG-136] Common Cause Not Cross-Attention: Blocking Visual Shortcuts in Audio-Video Generation
链接: https://arxiv.org/abs/2609.22361
作者: Jian Xu,Delu Zeng,John Paisley,Qibin Zhao
类目: Machine Learning (cs.LG); Graphics (cs.GR); Sound (cs.SD)
*备注:
Abstract:Joint audio–video generators are trained on data in which what an event looks like and what it sounds like are strongly, often spuriously, correlated: a particular material, texture, or object appearance co-occurs with a particular sound. This paper is a controlled causal study of the resulting failure mode. Building an AV structural causal model in which the audio is, by construction, independent of the video’s nuisance appearance, we show that models which let audio read video directly-through cross-attention or a shared latent-learn a visual shortcut: they predict sound from appearance rather than from the causal event, and collapse when the appearance-event correlation is broken at test time, literally synthesizing the wrong event’s sound. Crucially, the popular remedy of routing both modalities through a shared common-cause latent does not fix this: a bottleneck, an unsupervised shared/private factorization, and a faithful shared-prior model all grab the appearance proxy and fail like the direct model. Blocking the shortcut instead requires an intervention on the nuisance. Under the stated SCM and intervention assumptions we prove that counterfactual invariance is necessary and sufficient to identify the causal predictor, and we verify the mechanism across a feature-vector SCM, procedural pixel video, real images with spectrogram audio and a pretrained backbone, moving real digits, and a conditional generator. On a \emphreal, pretrained video-to-audio generator, an input-intervention test shows the model is far from invariant to sound-irrelevant edits (recolouring or graying a video substantially changes the sound it generates).
[LG-137] Contrastive Siamese Representation Learning for Predictive Maintenance of Electrical Submersible Pumps
链接: https://arxiv.org/abs/2609.22360
作者: Seshu K. Damarla,Xiuli Zhu
类目: Machine Learning (cs.LG)
*备注: 6 pages, 4 figures, 4 tables
Abstract:Electrical submersible pumps (ESPs) are essential in offshore oil production, where unexpected failures can result in significant operational and financial losses. Accurate predictive maintenance for ESP systems remains challenging due to nonlinear operating conditions, class imbalance, and variability among pump units. To address these issues, this study presents a fault diagnosis framework that incorporates class imbalance awareness by employing Siamese contrastive representation learning and prior-corrected k-nearest neighbor (KNN) classification. The method first extracts discriminative features relevant to fault detection from vibration-domain indicators and engineered harmonic relationships. A Siamese neural network is trained with class-balanced contrastive pairs to construct an embedding space that clusters samples of the same fault type and separates different fault classes. To further mitigate class imbalance during classification, a prior-corrected distance-weighted KNN is applied. The framework is validated using a Leave-One-ESP-Out (LOEO) strategy to evaluate generalization to previously unseen ESP units. Experimental results indicate that the proposed framework delivers robust and consistent fault classification performance under realistic industrial conditions, supporting its potential for reliable predictive maintenance and intelligent ESP system monitoring.
[LG-138] Domain-decomposed Evolutional Deep Neural Network with Random Features for Transient Pressure Diffusion with Discontinuous and High-Contrast Coefficients
链接: https://arxiv.org/abs/2609.22349
作者: Peiqi Li,Jie Chen,Hui Zhang,Simon Hands
类目: Numerical Analysis (math.NA); Machine Learning (cs.LG); Mathematical Physics (math-ph)
*备注: 36 pages, 11 figures and 15 tables
Abstract:Transient pressure diffusion in heterogeneous porous media becomes difficult to resolve efficiently when permeability is discontinuous and spans several orders of magnitude. We develop a domain-decomposed random-feature evolutional deep neural network (DD RF-EDNN) that separates spatial approximation from temporal evolution. Permeability-informed random features are compressed into an orthonormal pressure space, and a conservative finite-volume operator is projected onto this space so that only the reduced coordinates are advanced in time. This formulation retains the dissipative structure of the discrete flow problem without repeated neural-network optimization. The analysis quantifies dictionary and singular-value truncation errors, proves contractivity under an energy-consistent dissipativity correction, and derives a conditional grid-level estimate for homogeneous dynamics and time-independent data admitting a steady lifting. Experiments on low-permeability inclusions, high-conductivity channels and blocks, and three-dimensional grids demonstrate consistent accuracy across distinct coefficient structures. The method reaches a final-time relative L^2 error of 9.54\times10^-4 in the principal benchmark and maintains errors on the order of 10^-2 at a permeability contrast of 10^3 . Numerical diagnostics further show that the observed error balance and temporal convergence are consistent with the analysis. These results support DD RF-EDNN as a structure-preserving and interpretable reduced formulation for transient simulations on prescribed heterogeneous media.
[LG-139] Learning and Control Beyond Linearity: Towards a Non-asymptotic Theory for Bilinear Systems
链接: https://arxiv.org/abs/2609.22338
作者: Yahya Sattar,Yassir Jedra,Robin Strässer,Frank Allgöwer,Maryam Fazel,Sarah Dean
类目: ystems and Control (eess.SY); Machine Learning (cs.LG); Robotics (cs.RO); Optimization and Control (math.OC); Machine Learning (stat.ML)
*备注:
Abstract:This tutorial provides a unified view of the emerging area of bilinear learning and control. Using linear systems as a benchmark, it explains what fundamentally changes in the bilinear settings, how recent theory addresses finite-sample learning and control, and how these ideas connect to broader themes in nonlinear control, representation learning, and data-driven decision making. For learning, we emphasize tools that are particularly useful in the bilinear settings, such as one-sided Bernstein’s inequality for dependent and heavy-tailed covariates, blocking arguments, and martingale concentration for input-dependent noise. We then apply these tools to obtain finite-sample learning guarantees for fully observed bilinear systems, partially observed bilinear systems, and linear systems with bilinear observations. For control, we discuss quadratic control from bilinear observations, where the classical separation principle fails, and review tractable approaches based on belief-space receding horizon control. We also cover stabilization of bilinear dynamics under state feedback using semi-definite programming, LMI relaxations, sum-of-squares methods, and Koopman-based lifting. We conclude by discussing connections to reinforcement learning and machine learning, and some open problems in combined learning and control of bilinear systems.
[LG-140] When Does Test-Time Physical Diagnosis Pay? A Frozen Policy Buys Evidence It Never Reads
链接: https://arxiv.org/abs/2609.22299
作者: Zhengshu Zhang
类目: Robotics (cs.RO); Machine Learning (cs.LG)
*备注: 12 pages, 4 figures
Abstract:When a robot faces unfamiliar physical conditions, a common approach is to collect evidence about what changed and adapt. For such diagnosis to improve behavior, six ordered empirical conditions must hold: a meaningful reference, identifiability of the physical condition, use of the acquired evidence, decision value, selection value over a fixed alternative, and safe realization. We test this chain in controlled and public environments. It holds end to end in our controlled environments. After transfer to unseen mechanisms, however, it breaks at evidence use. On decisions requiring the full trace, the frozen decoder does not change its choice. A linear model using only trace increments recovers the correct choice on mechanisms excluded from fitting, showing that the trace is informative but unused. The failure is concentrated at the richest evidence level: those decisions fall to chance, while decisions settled with lower-cost evidence remain correct, a split hidden by aggregate accuracy. The same chain can fail at other links in public environments. Successful physical identification therefore guarantees neither evidence use nor useful adaptation; evaluation should identify where the chain breaks rather than rely on recovery accuracy or aggregate performance alone.
[LG-141] Statistical Inference for Adversarial Training: Central Limit Theorems via Optimal Transport
链接: https://arxiv.org/abs/2609.22240
作者: Kim Jakwang,Kwon Dohyun
类目: Machine Learning (cs.LG); Probability (math.PR); Statistics Theory (math.ST); Machine Learning (stat.ML)
*备注: 49 pages, 1 figure; comments are welcome!
Abstract:The purpose of this paper is to rigorously quantify the statistical and learning-theoretic properties of adversarial training models for classification. Equivalently, we establish the statistical properties of empirical optimal partial transport. Precisely, first we provide two types of central limit theorems (CLT): CLT centered at the expected empirical value, and CLT centered at the population one with smoothing. These results are based on the uniqueness of optimal potential for various equivalent optimal transport formulations, and the empirical process theory argument. For the binary setting, we indeed prove the uniqueness of optimal potential by leveraging the connection between optimal partial transport and the derived multi-marginal optimal transport formula. As byproducts, we also obtain the stability of a saddle point of the adversarial training model, and the sample complexity and concentration probability of the generalization error.
[LG-142] SCALE: Simulation-Calibrated Amortized Learning for Energy Materials (A hybrid architecture connecting deterministic modeling real-world data and transformer-scale inference for accelerated energy-materials discovery)
链接: https://arxiv.org/abs/2609.22233
作者: Kuan Huang,Bo Bai
类目: Machine Learning (cs.LG); Materials Science (cond-mat.mtrl-sci)
*备注: 16 pages, 5 figures
Abstract:Energy systems face converging pressures for security, affordability, resilience, and sustainability, creating a need for faster discovery of deployable energy materials. Here we introduce SCALE (Simulation-Calibrated Amortized Learning for Energy Materials), a physics-grounded, real-world-data-calibrated learning architecture that connects deterministic scientific operators, experimental calibration, expanded calibrated label generation, and transformer-scale inference. SCALE converts selected high-cost mechanistic computation and measured evidence into reusable models for rapid screening, ranking, inverse design, and active learning. We formulate the framework, identify ten method-based application regimes, and demonstrate SCALE for solid-state metal-hydride hydrogen-storage capacity prediction. In this implementation, a hydride phase-equilibrium capacity operator is calibrated against 381 measured ML-HydPARK capacity anchors and used to generate 5,000 candidate-condition-prototype teacher labels. A crystallographically anchored periodic-graph representation preserves atomic sites, periodic neighbor relationships, and local metal environments absent from formula-only encodings. An edge-biased graph transformer with 2.90 million parameters reproduces calibrated teacher labels with five-fold surrogate fidelity of MAE 0.0582 wt% H2, RMSE 0.0833 wt% H2, R2 = 0.9927, and Pearson r = 0.9963. Post hoc attention analysis suggests that SCALE learns chemically organized element groupings and metal-metal relationships consistent with established hydride chemistry, without chemistry-group labels as supervision. Once trained, SCALE shifts million-candidate evaluation from repeated deterministic workflow execution to batched learned inference, reducing per-candidate screening cost by approximately 10^7-10^8 while retaining links to simulation and experimental evidence.
[LG-143] CNA: An AI-Oriented Comprehensive Normalized Assessment for Healthy Status and Application to Optimize RRT Strategies by Reinforcement Learning
链接: https://arxiv.org/abs/2609.22232
作者: Jiang Liu,Chan Zhou,Yujie Li,Di Wu,Yihao Xie,Peiwei Li,Xin Shu,Jiaqi Zhu,Chunyong Yang,Yuwen Chen,Bin Yi
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:Millions worldwide require Renal Replacement Therapy (RRT) as a treatment essential for survival. However, optimizing RRT strategies via AI is challenging due to heterogeneous patient dynamics, missing data, and the absence of an AI-oriented health assessment criterion. We propose an AI-Oriented Comprehensive Normalized Assessment (CNA) for healthy status and apply it to optimize RRT strategies by using offline reinforcement learning (RL). The key idea of CNA is transforming vital-sign distributions into a standard normal space, enabling a unified, data-driven health-status score defined by deviations from referent intervals, which also provides an AI-oriented criterion to assess strategy quality and supports RL termination. We further design a structured 23-dimensional state representation that integrates 19 indicators with 4 RRT descriptors, and employ matrix decomposition to reconstruct missing vital signs, improving data completeness for learning. These components are incorporated into multiple offline RL algorithms and validated via systematic ablation studies on RRT feature subsets. Compared with physicians’ observed treatments, the best learned strategy reduces mortality from 13.2% to 5.0% (reducing 62.24%) and shortens average in-hospital stay from 308.5 to 250.1 hours (reducing 18.93%), demonstrating both methodological innovation and the potential of CNA-guided RL to improve RRT outcomes in nephrology.
[LG-144] List Counting Failures Are Not One Phenomenon
链接: https://arxiv.org/abs/2609.22230
作者: Iyad Ait Hou,Saad Mankarious,Aya Zirikly,Rebecca Hwa
类目: Machine Learning (cs.LG)
*备注: Preprint. Includes appendix
Abstract:Counting the items in a bracketed list looks trivial, yet open-weight chat models often get it wrong. Prior work usually blames input bottlenecks such as subword fragmentation or attention dilution, which predict that different models should fail in roughly the same way. Across seven instruct models on identical prompts, however, wrong answers form distinct modes: Qwen and Gemma 27B often flip odd lengths to a nearby even integer, OLMo concentrates errors on a few mid-sized integers, and Llama tends to under-count. These modes are useful labels rather than a stable family law (Gemma 9B does not reproduce Gemma 27B’s odd-to-even drop), and heavier subword fragmentation does not make counting harder on our benchmark. When the model answers incorrectly, a linear probe can usually still recover the true count from the residual stream. Matching the same odd-to-even error also does not imply the same late-MLP magnitude fix: scaling a late MLP output helps Qwen modestly but is near null on Gemma 27B under the same protocol, while residual steering can move both only by trading odd gains for even losses. These results caution against transferring that magnitude fix across models without a transfer check.
[LG-145] A Synthetic Multivariate Refrigerator Time-Series Dataset for Predictive Maintenance
链接: https://arxiv.org/abs/2609.22229
作者: Islam Benamirouche,Feriel Fass,Djemel Ziou
类目: Machine Learning (cs.LG)
*备注:
Abstract:We generated synthetic multivariate time series for 27 refrigerators with a simplified physicsinspired simulator at one-minute resolution. The simulator includes ambient-temperature variation, door use, thermostat and compressor operation, heat exchange, defrost, electrical consumption, and six progressive degradation types. Each refrigerator provides 15 to 20 sensor outputs according to its configuration. The dataset contains 7,066,161 rows in 27 time-series files and 27 failure logs. The release also includes the Python generator, refrigerator configurations, and documentation. The data can support failure prediction, degradation analysis, and learning across refrigerators with different sensor-output sets.
[LG-146] Measuring the Checker: Mutation Analysis for GPU-Kernel Benchmark Oracles
链接: https://arxiv.org/abs/2609.22220
作者: Mingzhe Du,Anh Tuan Luu,Dong Huang,See-Kiong Ng
类目: Machine Learning (cs.LG); Programming Languages (cs.PL)
*备注:
Abstract:Benchmarks for LLM-generated GPU kernels decide correctness with a few random inputs and a loose floating-point tolerance, and their verdicts now feed leaderboards and reinforcement-learning rewards. Recent work agrees these checkers are weak and patches them by hand—extra input distributions, fuzzing recipes, tighter tolerances—with no way to \emphmeasure whether any patch suffices. We introduce mutation analysis as an adequacy metric for kernel-benchmark oracles: deterministic rules inject 10,303 compilable faults into verified CUDA implementations of 188 KernelBench problems, 7,384 of them with an independent kill witness; any test protocol is scored by the fraction it detects. The official check misses \textbfone in six witnessed faults (16.9%), deterministically, and the misses are skewed by family: 8.7% of arithmetic faults escape, but 78.6% of precision faults do. The metric explains why (a tolerance blind band growing with reduction size; a measured ceiling on input aggressiveness set by legitimate floating-point variance), audits the strongest existing patch (KernelBench-Verified’s gain splits into +4.0 points from hidden inputs and +4.5 from tighter tolerance, a split its authors could not compute), and exposes a published fuzzing recipe that rejects \emphcorrect kernels 107 times. Optimizing suites over the kill matrix reaches 98.0% detection with two inputs per problem (94.8% held-out), and the measurement’s fault taxonomy teaches a test generator more than the raw faults themselves. Across 48 whole architectures, the blindness grows with scale, concentrating in deep homogeneous pipelines, and two problems prove unrefereeable: their official references violate the benchmark’s own tolerance against fp64. We release everything as \hrefthis https URLKernelBench-M.
[LG-147] UniGIO: Unified Generative Global In-situ Weather Modeling from Spatiotemporal Incomplete Observations
链接: https://arxiv.org/abs/2609.22217
作者: Songru Yang,Zili Liu,Tao Han,Ben Fei,Lei Bai,Chang Liu,Zhengxia Zou,Xiangyang Ji,Wanli Ouyang,Zhenwei Shi
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注:
Abstract:Global In-situ Observation (GIO) provides fine-scale, direct records of the global weather system from sparse point stations, making it an indispensable source for capturing localized and transient dynamics beyond the reach of satellite gridded data, and playing a critical role in key fields such as numerical weather prediction, disaster prevention, and agriculture. However, GIO exhibits strong spatiotemporal incompleteness, severely impairing accurate and real-time in-situ weather modeling. Unlike existing methods waiting for completed AI-ready data with extra introduced errors, in this work, we explore UniGIO, a novel generative framework for directly modeling global in-situ weather dynamics from native incomplete GIO. By generating missing data from observed ones annotated by masks, it unifies the coexisting forecasting, imputation, and generation under arbitrary missing ratios. Between the missing and observed, UniGIO captures station and region level complementarity through the Observation Mixer and Event Aligner, which diffuse discrete observations into continuous spaces where weather processes naturally span multiple stations. We further establish temporal dependencies with pattern shifts using the Adaptive Temporal Mixer, and track extreme events in chaotic local weather systems through a Mixture-ofExperts structure. Steady and extreme events are adapted in decoder by a Local Refiner. Extensive experiments on the up-todate largest global station weather dataset Weather-5K validate its SOTA performance with 11%, 12%, and 5% advantages on accuracy, fidelity, and extreme event capture, delivering a novel holistic solution for weather modeling in GIO networks.
[LG-148] Prediction of Nonlinear Oscillations in a Jumping Quarter-Car Model Using Reservoir Computing
链接: https://arxiv.org/abs/2609.22205
作者: Masahisa Watanabe,Shiva Dixit,Nirmal Punetha,Swati Chauhan,Manish Dev Shirimali
类目: Machine Learning (cs.LG); Chaotic Dynamics (nlin.CD)
*备注:
Abstract:Reliable prediction of vehicle dynamics is essential for smart driving applications such as autonomous control and advanced driver-assistance systems. Off-road vehicles used in agricultural and construction settings are particularly prone to nonlinear behavior, including bifurcations and chaotic motion arising from intermittent loss of tire–road contact. Predicting such dynamics is challenging because it requires resolving both smooth nonlinearities and the discontinuous switching associated with contact loss. In this work, we investigate the feasibility of reservoir computing (RC) – specifically an echo state network (ESN) – for data-driven prediction of a jumping quarter-car model. The reservoir is trained on time-series data from a small number of points and evaluated on its ability to reconstruct bifurcation diagrams, phase-space attractors, and time trajectories across periodic and chaotic regimes. The trained reservoir qualitatively reproduces the period-doubling route to chaos, captures the geometric structure of periodic and chaotic attractors. These results demonstrate that reservoir computing is a feasible data-driven predictor of nonlinear dynamics in a practical, non-smooth vehicle system.
[LG-149] Adaptive Physics-Informed Neural Networks for the Blasius Boundary-Layer Problem
链接: https://arxiv.org/abs/2609.22185
作者: Mehari Fentahun Endalew,Xiaoming John Zhang
类目: Machine Learning (cs.LG); Fluid Dynamics (physics.flu-dyn)
*备注: 25 pages
Abstract:Physics-informed neural networks (PINNs) provide a mesh-free approach for solving differential equations, but their performance can depend strongly on loss weighting, collocation placement, and optimization strategy. This study develops an adaptive PINN framework for the Blasius boundary-layer equation using gradient-norm-based adaptive loss weighting, nonuniform and residual-based collocation, and sequential Adam–L-BFGS optimization. In the representative run using the architecture [1,100,100,1] , the model predicts f’‘(0)=0.3320762918 , compared with the high-accuracy benchmark 0.332057336215 , giving an absolute error of 1.896\times10^-5 . The final weighted loss is 6.789\times10^-8 , and the predicted stream-function, velocity, and shear profiles agree closely with an independent numerical boundary-value solution. A separate full-training architecture study shows that the two-hidden-layer model achieves the smallest wall-shear error among the four tested architectures, 1.629\times10^-6 , whereas the deepest network attains the smallest weighted objective but a substantially larger wall-shear error. Compared with the previously reported PINN value f’'(0)=0.33165 , the representative run reduces the wall-shear error by approximately a factor of 21.5 . The results show that the combined adaptive training framework can achieve high accuracy for the Blasius problem and that weighted loss alone is insufficient for identifying the most physically accurate PINN. Because the adaptive components are applied jointly, their individual contributions cannot be isolated from the present results and would require a controlled ablation study for separate assessment. Comments: 25 pages Subjects: Machine Learning (cs.LG); Fluid Dynamics (physics.flu-dyn) Cite as: arXiv:2609.22185 [cs.LG] (or arXiv:2609.22185v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2609.22185 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Mehari Fentahun Endalew Dr. [view email] [v1] Fri, 28 Aug 2026 04:18:33 UTC (1,104 KB) Full-text links: Access Paper: View a PDF of the paper titled Adaptive Physics-Informed Neural Networks for the Blasius Boundary-Layer Problem, by Mehari Fentahun Endalew and Xiaoming John ZhangView PDFHTML (experimental)TeX Source view license Current browse context: cs.LG prev | next new | recent | 2026-09 Change to browse by: cs physics physics.flu-dyn References Citations NASA ADSGoogle Scholar Semantic Scholar export BibTeX citation Loading… BibTeX formatted citation loading… Data provided by: Bookmark checked="checked"class=“labs-tab-input”> Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) IArxiv recommender toggle IArxiv Recommender (What is IArxiv?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv’s community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?) mathjaxToggle(); We gratefully acknowledge support from our major funders, member institutions, , and all contributors. About Help Contact Subscribe Copyright Privacy Accessibility Operational Status (opens in new tab) Major funding support from
[LG-150] CleanScore: Black-Box Benchmark Audits with Negative Controls and Sensitivity Bounds
链接: https://arxiv.org/abs/2609.22183
作者: Jeffery Opoku,David Banahene
类目: Machine Learning (cs.LG); Methodology (stat.ME)
*备注: 43 pages, 15 figures, 21 tables. Item bank, code and pre-registrations released
Abstract:Public benchmark scores may reflect skill, prior exposure to the questions, or both, and for most models the training data are unknown. We present CleanScore, a black-box audit using scored outputs only. Each benchmark question becomes a parent item with one public form and two independently written fresh forms preserving its numbers, facts and answer. The audit reports an interval for the public-form advantage rather than a verdict, and a private negative-control bank with an explicit transport radius separates exposure from ordinary form mismatch. A registered controlled-exposure experiment detects planted exposure and stays quiet under fresh-form exposure. A registered audit of five open models on 200 GSM8K and 200 ARC-Challenge items finds no exposure-consistent advantage, bounding surface-form inflation below five points. Registered positive controls then bound what such a null can mean. Leaking an item raises accuracy on paraphrases the model never saw almost as much as on the leaked wording, leaving 52% to 110% of the effect invisible to a paraphrase audit. On ARC a planted 49-point advantage shows an observable gap of -0.020, and about 20 points survive rewriting stem and options, across four training seeds. A surface-form null bounds far less than the phrase contamination audit implies.
[LG-151] Gaussian Process Decorrelation for Spatiotemporal Deep Learning-Based Snow Water Equivalent Prediction
链接: https://arxiv.org/abs/2609.22182
作者: Colin Fenster,Adrienne Marshall,Soutir Bandyopadhyay,Daniel McKenzie
类目: Machine Learning (cs.LG); Methodology (stat.ME)
*备注:
Abstract:In the Western United States, snowmelt is essential to the agricultural industry in addition to being a key source of municipal drinking water. Consequently, accurate snowpack forecasting is critical for water policy and management. Automated Snow Telemetry (SNOTEL) stations provide accurate daily measurements of snow water equivalent (SWE) that exhibit strong correlations in space and in time. We tackle the problem of predicting future SWE values across the SNOTEL network. Specifically, we use a Gaussian Process-based linear transformation to remove spatial correlations before training a long short-term memory (LSTM) neural network on the decorrelated SWE data. This approach allows the LSTM to learn a clean temporal signal at each station. We show that this separation of spatial and temporal components yields better predictive success than multiple baseline models. Furthermore, we incorporate conformal prediction to quantify uncertainty in the resulting SWE forecasts, providing a distribution-free approach to illustrate a potential framework for establishing predictive intervals for spatiotemporal data. Together, accurate point forecasts and distribution-free uncertainty quantification provide a framework for SWE accumulation forecasting on subseasonal scales or projecting SWE with future data while motivating and supporting future work in predicting a large-scale, spatiotemporally complete SWE map. Subjects: Machine Learning (cs.LG); Methodology (stat.ME) Cite as: arXiv:2609.22182 [cs.LG] (or arXiv:2609.22182v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2609.22182 Focus to learn more arXiv-issued DOI via DataCite
[LG-152] OpenBlock: Constructive and Verified Content Generation for Adaptive Tile-Matching Games
链接: https://arxiv.org/abs/2609.22177
作者: Jiang Jun
类目: Machine Learning (cs.LG); Computer Science and Game Theory (cs.GT)
*备注: 25 pages
Abstract:Tile-matching puzzle games serve hundreds of millions of players, yet the content-generation algorithms that decide which pieces to present at each turn remain proprietary, and no open platform exists for studying adaptive difficulty in this genre. We present an adaptive tile-matching platform whose central algorithmic contribution is a dual-track content-generation architecture: a deterministic rule-based generator that is always available, and an optional learned generator, both subject to a common verification gate that establishes, by exhaustive sequential-placement search, that every delivered piece set is fully placeable so the learned track can never degrade the constructive-feasibility guarantee of the rule track. A self-play reinforcement-learning placement agent, supervised by auxiliary tasks that expose per-shape placeability to shared representations, is used to diagnose the game’s dominant failure mode: at high board fill, long-bar pieces lose the majority of their legal placements. Across 234,000+ self-play episodes the agent reaches a 35.6% win rate (median score 4,200), and controlled simulation shows that at board fill rates of 70–75%, 33–56% of long-bar pieces have no legal placement, while spawn difficulty distributions are statistically indistinguishable between won and lost games—evidence that board-state degeneration, not content difficulty, drives late-game failure. Head-to-head ablations show that per-shape placeability supervision—not aggregate difficulty features—drives the representation gain, and a 14-day online gray rollout (48,000 players; sample-ratio verified, CUPED-adjusted) lifts day-1 retention by 1.8 percentage points and session duration by 7% over the rule track alone, quantifying the neural track’s asymmetric upside in live play.
[LG-153] Industrial Kinematic Trajectory Model (IKTM): Coordinate-Free Autoregressive Generator
链接: https://arxiv.org/abs/2609.22173
作者: Max Amiri,David Eyers
类目: Machine Learning (cs.LG); Robotics (cs.RO)
*备注: 15 pages, 2 figures
Abstract:Mobility simulation supports logistics, safety, and communications planning in industrial environments such as ports, mines, and airports. Existing trajectory models, however, rely on absolute coordinates, road-network tokens, or semantic zones: representations that are site-specific and not well suited to unstructured industrial terrain. We introduce the Industrial Kinematic Trajectory Model (IKTM), a coordinate-free trajectory generator that represents industrial vehicle motion through kinematic sequences (speed and heading change) with no absolute spatial reference. IKTM uses an autoregressive causal transformer with probabilistic mixture heads and extends our prior coordinate-free Markovian model with deep sequence modelling and an explicit duration-conditioning signal. Trained on one site and evaluated zero-shot on three unseen sites, it matches the small-turn shape of the empirical turn-rate distributions of held-out telematics; across all four sites, the per-site mean Jensen-Shannon divergence over 100 sampling seeds spans approximately 0.035-0.050 bits under an oracle-length duration-target protocol and approximately 0.032-0.042 bits under a fully zero-shot prior-length protocol, with similar ranges whether out-of-distribution (OOD) duration targets are drawn from each held-out site’s empirical length distribution or from the Site A prior. Both protocols stay above the metric’s sampling-noise floor (=0.0049 bits). Paired by site, the oracle-length values are 5.3-6.0x lower than those of a re-implementation of our prior Markovian model under the same 1 Hz protocol. Termination is duration-conditioned rather than spatial: rollouts stop on 100% of trials with a length-tracking error of +0.0 +/- 0.0 s against the sampled target (100 of 100 exactly on target; N=100, T=0.2, untouched in-distribution test split).
[LG-154] Role-Aware Morgan Fingerprints for Reaction Yield Prediction
链接: https://arxiv.org/abs/2609.22167
作者: Chinmay Mirji,Prashant Shekhar,Foram Madiyar,Hao Peng
类目: Machine Learning (cs.LG)
*备注:
Abstract:Predicting reaction yield from molecular structure and reaction context can cut experimental trial-and-error and speed up condition screening in synthetic chemistry. Recent methods for this task use learned representations such as graph neural networks or Transformer encoders over reaction SMILES (Simplified Molecular Input Line Entry System), but these approaches carry heavy preprocessing overhead and can break when input formatting is inconsistent. We propose MFP, a reaction yield prediction method built on role-aware Morgan fingerprints where count-based circular fingerprints are computed for each reaction component, aggregated by chemical role (reactant, reagent, product), and combined with transformation-sensitive difference features into a fixed-length reaction descriptor fed to a feed-forward neural regressor. We test MFP against state of the art methods such as YieldBERT (with and without data augmentation) and GNAN (graph neural network) on the Suzuki-Miyaura and Buchwald-Hartwig benchmarks using a shared preprocessing and evaluation protocol. MFP reaches R2 = 0.878 on Suzuki-Miyaura and R2 = 0.969 on Buchwald-Hartwig while training an order of magnitude faster than graph- or Transformer-based alternatives. A formal complexity analysis confirms that MFP folds all representation cost into a one-time preprocessing step, removing the per-epoch message-passing overhead that graph methods carry. An ablation over fingerprint radius and folded vector length shows that radius-2 representations at nBits =2048 give the best balance of accuracy, speed, and cross-split stability on both datasets. These results establish MFP as an effective, reproducible, and efficient baseline for reaction yield prediction.
[LG-155] Clustering-Based Collective Anomaly Detection in IoT Systems: A Graph Neural Network Approach
链接: https://arxiv.org/abs/2609.22166
作者: Dalila Khettaf,Djamel Djenouri,Zeinab Rezaeifar,Youcef Djenouri
类目: Machine Learning (cs.LG)
*备注:
Abstract:The rapid advancement of Internet of Things (IoT) technology has led to the widespread deployment of smart, interconnected devices across a range of domains. However, this expansion has also resulted in a substantial increase in network traffic, creating more opportunities for malicious actors to launch cyberattacks and compromise sensitive information, thereby increasing the need for effective anomaly detection. The state-of-the-art in anomaly detection has predominantly focused on point anomalies. In contrast, the detection of collective anomalies remains relatively under-explored in the literature. In this paper, we introduce Unsupervised Graph Collective Anomaly Detection (UGCAD), a novel frame- work designed to identify collective anomalies in IoT network traffic. Unlike many existing methods, UGCAD operates on graph-structured data without any prior knowledge of group labels or membership. It leverages a variational graph autoencoder (VGAE) to learn the graph representation, which is subsequently used to enhance a clustering algorithm for effective grouping of nodes. To detect collective anomalies, clusters identified as normal are first aggregated and refined, after which anomaly scores are applied to detect collective anomalies. Extensive experiments conducted on the CICIoT2023 and ToN-IoT network datasets demonstrate the effectiveness of UGCAD in both clustering and collective anomaly detection (CAD). Furthermore, comparative evaluations against several traditional and state-of-the-art clustering-based CAD approaches confirm the superiority of UGCAD in accurately detecting collective anomalies.
[LG-156] ARGet: Topology-Aware Fusion-based Radio Frequency Circuit Functional Modeling using Graph Neural Networks
链接: https://arxiv.org/abs/2609.22165
作者: Soroosh Noorzad,Sebastian Bodero,Morteza Fayazi
类目: Machine Learning (cs.LG)
*备注: To be published in International Conference on Computer-Aided Design (ICCAD 26)(9 pages)
Abstract:Automatic synthesis of analog and Radio Frequency (RF) circuits is an emerging area that requires an efficient circuit modeling method. In recent years, Machine Learning (ML) solutions have played a promising role in this regard. However, many existing ML approaches require separate training data for each circuit topology, even when a single circuit component is added or removed. In addition, they overlook circuit topology information, which limits their ability to capture complex component interactions. Furthermore, they rely on fully connected neural networks with flat feature representations, which require substantial amounts of training data. In this work, we propose an open-source topology-aware RF circuit modeling method, TARGet. Our model considers the circuit at two levels: sub-circuits and the overall circuit topology. At the sub-circuit level, TARGet leverages S-parameter representations to capture sub-circuit behavior rather than relying on individual circuit components, providing a reusable behavioral abstraction for RF building blocks. Moreover, TARGet explicitly incorporates circuit topology information into the model, enabling it to learn across multiple topologies. TARGet introduces a novel fusion-based architecture that integrates Graph Neural Networks (GNNs) and sub-circuit connectivity-aware neural networks to improve data efficiency. Experimental evaluation across multiple RF circuit topologies demonstrates that TARGet achieves sub-1% prediction error while reducing the required training data by up to 35.5x compared to state-of-the-art (SOTA) approaches. Furthermore, TARGet achieves 9.7x higher prediction accuracy under a strict 1% error threshold relative to SOTA models. A held-out matching-network evaluation further demonstrates zero-shot transfer to an unseen sub-circuit topology, where TARGet reduces NMAE by up to 45%.
[LG-157] Uncertainty and Business-Aware Remaining Useful Life Estimation for Semiconductor Manufacturing
链接: https://arxiv.org/abs/2609.22160
作者: Davide Frizzo,Francesco Borsatti,Gian Antonio Susto
类目: Machine Learning (cs.LG)
*备注: Submitted to IEEE Transactions on Semiconductor Manufacturing
Abstract:Semiconductor manufacturing relies on tightly interconnected components, so early identification of the assets most likely to fail is essential to prevent a single breakdown from disrupting the entire production pipeline. Maintenance planning must therefore balance unexpected failures against prematurely interrupted operating life. We present a Predictive Maintenance (PdM) framework combining Deep Learning (DL) sequence models and Simoultaneous Quantile Regression (SQR) for uncertainty-aware Remaining Useful Life (RUL) estimation and risk-aware maintenance decisions. Several architectures are compared on ion-milling data from the 2018 PHM Data Challenge (PHM18), including architectures based on State Space Models (SSM), using prediction and business metrics: Unexpected Breaks (UB), Unexploited Lifetime (UL), and a cost-weighted objective. Diagonal State Spaces (S4D) delivers the best Remaining Useful Life (RUL) estimates across quantiles and, relative to Preventive Maintenance (PvM) baselines, substantially lowers business cost by avoiding systematically early interventions. The results support uncertainty-aware, cost-sensitive maintenance planning in semiconductor production. Comments: Submitted to IEEE Transactions on Semiconductor Manufacturing Subjects: Machine Learning (cs.LG) Cite as: arXiv:2609.22160 [cs.LG] (or arXiv:2609.22160v1 [cs.LG] for this version) https://doi.org/10.48550/arXiv.2609.22160 Focus to learn more arXiv-issued DOI via DataCite
[LG-158] StepKV: Step-Aware KV Cache Compression for LLM Agents
链接: https://arxiv.org/abs/2609.22158
作者: Boyu Feng,Jiahong Liu,Yifan Li,Wenhao Yu,Zexuan Qiu,Yuliang Sun,Ming Shen,Xiang Li,Quanyu Dai,Irwin King
类目: Machine Learning (cs.LG)
*备注:
Abstract:Key-value (KV) caching is essential for efficient autoregressive large language model (LLM) inference, but the cache grows linearly with context length, increasing storage and decoding costs. KV cache compression mitigates this cost by retaining only a subset of cached tokens. This challenge is particularly important for multi-step LLM agents, where a query expands into trajectories of reasoning, tool interactions, and retrieved observations. Existing pruning methods typically treat the cache as a flat token stream and rank tokens by recency or attention saliency. This creates a mismatch between the unit of compression and the unit of reasoning: token-level pruning removes individual entries, whereas useful information in multi-step agents is often organized into reasoning steps with uneven and delayed importance. Consequently, an early observation or intermediate decision may receive little recent attention yet remain essential for later evidence synthesis. We term this failure mode Reasoning Continuity this http URL observations motivate KV cache compression that jointly considers token- and reasoning-step-level information. StepKV addresses this goal by treating reasoning steps as first-class retention units. It associates cache entries with their generating steps, estimates step utility from trajectory-derived signals, and combines this utility with token-level saliency. The resulting scores globally rank prunable tokens, from which StepKV retains the top-scoring entries under a target budget. StepKV thus provides a step-centric perspective for agent KV cache compression. Across multi-hop QA and long-horizon web reasoning tasks, StepKV sustains accuracy under low KV budgets where token-level baselines degrade sharply, offering a more robust efficiency-accuracy trade-off for multi-step agent inference.
[LG-159] he Limits of Speculation: Bounding Speculative Decoding in Mixture-of-Experts
链接: https://arxiv.org/abs/2609.22156
作者: Aidar Amankulov,Denis Mamatin
类目: Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 24 pages, 17 figures, 7 tables
Abstract:Speculative decoding in Mixture-of-Experts (MoE) models faces the problem of unstable verification cost caused by input-dependent expert loading. To study the physics of this process, we formulate speculation-budget selection as an offline Stochastic Shortest Path (SSP) problem over reference sequences and build a diagnostic Oracle that uses counterfactual simulation to account for MoE verification cost. A detailed analysis of the Oracle’s decisions on the Qwen3-Coder and EAGLE-3 pairing, in the space of marginal deltas (Delta Space), shows that rejected candidates form a strict linear boundary. This result demonstrates that a complex global optimization is locally governed by a necessary condition balancing marginal cost against expected progress ( \frac\Delta \mathbbE[Cost]\Delta \mathbbE[a] ), providing a rigorous mathematical reference point for designing future adaptive online heuristics.
[LG-160] From Latent Biomarkers to Clinical Rules: Embedding-Guided Rule Mining and Attribution-Based Translation for Interpretable Tabular Learning AAAI2027
链接: https://arxiv.org/abs/2609.22155
作者: Majid Lotfian Delouee,Hamed Ayoobi,Sjors G. J. G. In 't Veld,Martijn C. Schut
类目: Machine Learning (cs.LG)
*备注: 12 pages, submitted to AAAI 2027
Abstract:Clinical decision support tools are most useful when accurate predictions are accompanied by understandable explanations. Rule-based models provide transparency, but rules derived directly from raw clinical measurements may miss patterns arising from interactions between multiple variables. We present a four-step pipeline that mines decision rules in the latent space of an FT-Transformer and translates them back into measurable clinical features. Embedding dimensions that consistently separate patient groups are treated as latent biomarkers, rules are mined using small decision trees, and selected rules are translated using gradient-input saliency and CLS attention attribution. We evaluate the framework on six public clinical and population health datasets at four embedding dimensions. Translated rules outperformed raw-feature rules in five of six datasets, with mean AUROC gains ranging from 0.04 to 0.23. On the heart disease dataset, embedding-space rules reached 0.98 AUROC, but translation reduced this to 0.72, showing that high-performing latent rules cannot always be represented by simple raw-feature conditions. These results show that latent-space rule discovery can uncover predictive patterns while translating them into clinically measurable features that can be evaluated by clinicians.
[LG-161] A Comparative Framework for Evaluating Foundation Models on Tabular Data: A Case Study in Healthcare
链接: https://arxiv.org/abs/2609.22154
作者: Majid Lotfian Delouee,Sjors G. J. G. In 't Veld,Martijn C. Schut
类目: Machine Learning (cs.LG); Computers and Society (cs.CY)
*备注: 32 pages
Abstract:Tabular data is the most common format in clinical practice, encompassing laboratory results, medication records, diagnostic codes, and patient demographics. As foundation models for tabular data have grown in number and variety, a practical question has become harder to answer: which model should a clinician or data scientist actually choose for a given task, and why? Existing surveys catalogue what these models can do, but they stop short of providing a structured way to compare them against the specific demands of a real application. We introduce \system, a comparative evaluation framework that scores and ranks tabular foundation models (TFMs) across six clinically meaningful dimensions: how well a model generalizes to new datasets, how effectively it protects patient privacy, how much data it needs to perform well, how it scales with growing datasets and feature spaces, how interpretable its predictions are to clinicians, and how fairly it performs across patient subgroups. Each dimension is broken down into measurable sub-components, and groups of sub-components can optionally be combined into supplementary compound scores, called super-metrics, that provide a diagnostic view of how a model performs across several dimensions simultaneously. To show how the framework works in practice, we apply it to two healthcare use cases, screening for iron deficiency and predicting heart failure, demonstrating how the same set of metrics leads to different model rankings depending on what matters most in each clinical context. We also provide a taxonomy of 45 TFMs organized by their underlying architecture, which serves as a reference for researchers and practitioners looking to navigate this rapidly expanding field.
[LG-162] SafeTune: A Unified Faithful Library for Auditing and Repairing Safety Drift in Fine-Tuned LLM s
链接: https://arxiv.org/abs/2609.22153
作者: Pratinav Seth,Saisab Sadhu,Anshul Kaushal,Vinay Kumar Sankarapu
类目: Machine Learning (cs.LG); Software Engineering (cs.SE)
*备注:
Abstract:Methods for addressing safety drift in fine-tuned Large Language Models (LLMs) are scattered across incompatible implementations, lifecycle stages, and evaluation protocols, making them difficult to adopt and compare. We introduce SafeTune, a source-available library that unifies four intervention paradigms: post-hoc weight recovery, safety-constrained fine-tuning, gradient-based unlearning, and inference-time steering, alongside shared interpretability, evaluation, and deployment utilities. SafeTune provides a consistent configuration-driven workflow while preserving the distinct inputs and intervention points each paradigm requires. Its modular registry supports new methods, benchmarks, judges, models, and fine-tuning domains without redesigning the surrounding pipeline. We demonstrate SafeTune through controlled comparisons and finance and medical deployment case studies, showing how it characterizes safety drift, evaluates feasible interventions on common refusal-behavior and capability evaluations, and supports calibrated or layered mitigation.
[LG-163] When Does Learning Beat Heuristics? A Case Study in Kubernetes Scheduler Score Plugins
链接: https://arxiv.org/abs/2609.22142
作者: Wang Xuying,Zhibek Sarypbekova
类目: Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注:
Abstract:Kubernetes scheduler plugins that score candidate nodes are, in production, hand-tuned heuristics. We ask whether a learned scoring function - trained on real placement decisions from a production cluster trace - can match or exceed these heuristics, and if not, why. We implement an external, HTTP-backed scoring plugin for a widely used scheduler simulator and evaluate two learned models (a Random Forest over engineered features, and a graph neural network encoder over per-job task-dependency graphs) trained on a large-scale production cluster trace. Under standard regression fit (R^2), both models improve modestly but monotonically across four feature-engineering iterations, reaching R^2 of about 0.042. However, on the metric that actually matters for scheduling - Top-1 ranking accuracy, whether the model scores the machine the production scheduler actually chose highest - both learned models are outperformed by a trivial single-feature heuristic (rank by free CPU: 74-84% vs. 65-66% for either model). We show this gap is best explained by objective mismatch: both models were trained with pointwise regression (MSE) rather than a ranking-specific objective, echoing a long-standing distinction in the learning-to-rank literature. This parallels prior evidence that RL-trained schedulers can outperform heuristics when the training objective is aligned with the deployment task, suggesting objective misalignment, not architecture, is the primary obstacle here. We further report an ablation of the occupancy reconstruction required to make offline trace data usable (naive features yield R^2 near 0), a controlled comparison isolating feature richness and data volume between the two model families, and a sensitivity analysis of inference latency and serving-container memory constraints. Code, data pipelines, and experiment scripts are released for reproducibility.
[LG-164] WiNeRF: Measurement Constrained Radiance Fields for Actionable Wireless Channel Modeling
链接: https://arxiv.org/abs/2609.22132
作者: Saif Ur Rahman,Rafid Umayer Murshed,Anton Dmitriev,Cagri Tanriover,Rahul C. Shah,Elahé Soltanaghai
类目: Networking and Internet Architecture (cs.NI); Machine Learning (cs.LG)
*备注: Accepted to EWSN 2026
Abstract:Wireless embedded systems increasingly rely on wireless channel information for decision making, yet practical platforms operate under severe constraints, including few antennas, narrow bandwidth, and sparse, noisy measurements. While neural field based approaches inspired by Neural Radiance Fields (NeRFs) have recently been explored for continuous wireless channel modeling, existing approaches depend on dense measurements or external priors such as known geometry, visual context, or angle-of-arrival (AoA) information, limiting their practicality in real-world deployments. We present WiNeRF, a neural field framework that learns a spatially continuous, complex-valued wireless channel representation directly from sparse channel state information (CSI) collected by commodity WiFi devices. WiNeRF embeds intrinsic system constraints, such as antenna geometry, limited spatial resolution, and phase uncertainty, as inductive biases through a 3D conical wave sampling model, a multi-resolution implicit scene representation, and a differentiable optimization framework for complex-valued channel learning. Across diverse indoor environments with non-line-of-sight (NLoS) regions, WiNeRF achieves a median prediction SNR of 5.3 dB, outperforming prior neural baselines by 4.9 dB on average (approximately 3x higher prediction SNR), and produces a task-agnostic channel representation that can be directly reused in standard signal-processing pipelines, including beamforming, AoA estimation, and RSSI coverage mapping, without modifying existing hardware or wireless protocols.
[LG-165] Helix-FNO: Spectral-Domain Operator Learning Coupled with a High-Fidelity Mechanistic Model for Fast Surrogate Simulation
链接: https://arxiv.org/abs/2609.22129
作者: Jiabao Zhao,Chuwei Wang,Jinxi Yang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Mechanistic simulation models of full-scale treatment processes remain the only trustworthy, extrapolative description of the underlying physico-chemical dynamics, yet their runtime is far too slow to support the thousands of forward evaluations that a modern decision engine requires at a 5-minute decision cadence. The standard remedy-surrogate modelling-often produces a network that learns a single solution for a single configuration, so it generalises poorly to new influent profiles, control settings or plant layouts. This paper presents Helix-FNO, a teacher-student architecture that couples a thirty-two-state mechanistic teacher with a Fourier neural operator (FNO) student. The teacher supplies a high-fidelity dataset of input-field-to-solution pairs, curated by Latin-hypercube and uncertainty-based active learning to cover the boundary and overload regimes that matter in practice; the student learns, in the spectral domain, the solution operator itself rather than any single solution, thereby moving from learning one instance to learning an entire family of equations. We give the operator formulation, the spectral convolution definition, the weighted distillation loss and the active-learning criterion, and we analyse the approximation error of a truncated Fourier expansion with respect to the smoothness of the parametric solution manifold. An illustrative study compares Helix-FNO against a physics-informed network and a data-driven recurrent surrogate on accuracy, dataset efficiency and inference latency, and places the methods on a speed-accuracy Pareto front. The resulting operator is three orders of magnitude faster than the mechanistic teacher at millisecond inference, which is precisely the capability required for massive candidate screening and online decision support.
[LG-166] ST-Topo GAN: A Motor EEG-to-EMG Decoding Model Matched to Wrist Movement Complexity
链接: https://arxiv.org/abs/2609.22128
作者: Ye Sun,Mingxuan Qu,Jing Wang,Dezhong Yao,Gang Liu
类目: ystems and Control (eess.SY); Machine Learning (cs.LG)
*备注: 11 pages, 9 figures
Abstract:The wrist plays a critical role in upper-limb function by enabling precise hand positioning, force regulation, and object manipulation. Continuous brain–muscle interfaces (BMIs) offer a promising approach for motor restoration by decoding neural activity into muscle activation signals. However, existing EEG-to-EMG models have mainly been developed for tasks with relatively stable muscle synergies and may be less effective for the heterogeneous and weakly coupled neuromuscular organisation involved in wrist movements. This paper proposes ST-Topo GAN, a Spatial–Temporal Topological Generative Adversarial Network for continuous EEG-to-EMG decoding of wrist movements. The framework integrates multi-band EEG representation, sensorimotor cortical topology modelling, and conditional adversarial learning to reconstruct multi-channel iEMG activation. The model was evaluated through cross-task comparison, wrist EEG-to-iEMG decoding, and ablation experiments. Compared with the WAY-EEG-GAL grasp-and-lift dataset, the wrist dataset exhibited lower inter-muscle activation similarity and greater decoding difficulty for conventional models. ST-Topo GAN achieved an average PCC of 0.4436 on the wrist dataset, outperforming all evaluated baselines, while the ablation study confirmed the contribution of its key components. These results support the effectiveness of ST-Topo GAN for continuous wrist EEG-to-iEMG decoding.
[LG-167] StationPDE: Station-Oriented Surface PDE Learning for Multi-Station Multivariate Weather Forecasting
链接: https://arxiv.org/abs/2609.22123
作者: Xiao Wang,Changjian Chen,Rongwen Li,Hongwu Liu,Kun Fang,Zhuo Tang
类目: Machine Learning (cs.LG)
*备注:
Abstract:Multi-station multivariate weather forecasting aims to forecast future weather variables at multiple weather stations from historical surface observations. Existing station forecasting models learn statistical dependencies among discrete stations, but lack explicit physical evolution. Meanwhile, PDE-based weather models provide interpretable physical dynamics, yet require continuous fields and upper-air variables unavailable in surface station data. To bridge this gap, we propose StationPDE, a station-oriented surface PDE learning model. StationPDE constructs a terrain-aware continuous surface field from discrete station observations and decomposes its physical evolution into surface wind transport and upper-air inference. Surface wind transport explicitly evolves observable weather variables, while upper-air inference uses learnable horizontal diffusion to approximate the missing influence of unavailable upper-air variables. A parallel data-driven diffusion branch captures complementary motion patterns, and an adaptive router integrates the two forecasts for station-level multivariate forecasting. Experiments on Weather2K and MeteoNet show that StationPDE consistently outperforms state-of-the-art baselines, reducing MSE by about 9.6% on average compared with the strongest baseline. Code and implementation details are available at this https URL.
[LG-168] Rank Portability Does Not Imply Feasibility Portability: Target-Specific Evaluation of Joint Hardware Constraints
链接: https://arxiv.org/abs/2609.22122
作者: Wesley Shu
类目: Machine Learning (cs.LG)
*备注:
Abstract:Cross-device hardware evaluation often assumes that if architecture rankings transfer across devices, a proxy device can support target-side model selection. We stress-test this assumption for joint latency-energy feasibility across two public architecture families. On NAS-Bench-201, cross-device rank correlations are moderate, while target-comparable feasible-set overlap remains incomplete. A faithful AdaProxy diagnostic substantially improves latency ranking, showing that the observed boundary failures are not simply due to weak adaptation. Exact finite-sample split-conformal analysis also exposes an evidence bottleneck: a finite one-sided 90% threshold requires at least nine calibration observations. We then replicate the phenomenon on 10,000 GPT architectures across 13 HW-GPT-Bench devices. Relative to an RTX3080 proxy, target latency SRCC ranges from 0.951 to 0.996, yet proxy-reuse violation risk ranges from 33.3% to 100% under matched joint constraints. These results show that rank portability, feasibility portability, and target-specific decision support are distinct evaluation objects. Cross-device evaluations should therefore report which target environments actually support the operating point being claimed.
[LG-169] Modelling daily activity patterns from mobile phone location data via deep representation learning
链接: https://arxiv.org/abs/2609.22121
作者: Xinglei Wang,Junyuan Liu,Guangsheng Dong,Zichao Zeng,Stephen Law,James Haworth,Tao Cheng
类目: Machine Learning (cs.LG); Social and Information Networks (cs.SI)
*备注: 29 pages, 10 figures
Abstract:Passively collected mobile phone location data provide large-scale, longitudinal observations of human mobility but do not directly reveal activity purposes. The functional characteristics of visited locations offer useful contextual information, yet their relationship with activity purpose remains uncertain, particularly in mixed-use urban environments. We conceptualise activity pattern mining as an integrated process of representation, clustering, and interpretation, and propose the Activity Chain Encoder (ACE) for the representation stage. ACE is a self-supervised model that combines pre-trained urban embeddings with visit timing and duration and uses a Transformer to model the sequential organisation of stays. It is trained using masked activity modelling and identity-guided contrastive learning without requiring deterministic activity purpose labels. Learned daily representations are aggregated into user-level profiles, clustered, and interpreted through temporal-functional patterns and Census-derived demographic context. Applied to mobile phone app location data from London and compared with three representative methods, ACE supports the identification of six differentiated weekday activity-pattern groups characterised by distinct daily rhythms, urban functional contexts, and demographic associations. These complementary forms of evidence further support the development of empirically grounded activity-pattern personas, establishing a holistic route for deriving behaviourally meaningful population patterns from unlabelled mobile phone location data. The source code for the entire analytical pipeline developed in this study is publicly available at this https URL.
[LG-170] LE4Mob: Towards Inductive Distance-Aware and General-Purpose Location Embedding for Human Mobility Modelling
链接: https://arxiv.org/abs/2609.22117
作者: Xinglei Wang,Stephen Law,Zichao Zeng,Junyuan Liu,Guangsheng Dong,Tao Cheng
类目: Machine Learning (cs.LG)
*备注: 13 pages, 2 figures
Abstract:Location representations provide mobility models with fundamental information about the spatial position, functional characteristics, and relationships of places. However, existing embeddings are often dependent on mobility observations, unable to represent unseen locations, and weakly constrained to retain geographic distance. This limits their reuse across datasets and mobility tasks. To address these limitations, we propose LE4Mob, an inductive, distance-aware, and geography-derived location embedding framework for mobility modelling. LE4Mob extends contrastive language-location pre-training while introducing a distance-aware regularisation objective that encourages the embedding space to preserve spatial relationships. Pre-trained from geographic context, LE4Mob can encode rich spatial-semantic information and generate embeddings for unseen locations inductively. Its independence from downstream mobility task supervision also makes it transferable across different mobility tasks. We evaluate LE4Mob on individual-level next location prediction and population-level commuter flow generation. Experiments across multiple datasets and study areas show that LE4Mob outperforms strong baselines, with particular advantages in inductive settings and when downstream models rely directly on interactions between location embeddings. These findings demonstrate the potential of distance-aware, geography-derived location representations as reusable foundations for human mobility modelling.
[LG-171] ZoAQ: Adaptive Zeroth-Order Querying via Query-Reuse Coupling
链接: https://arxiv.org/abs/2609.22115
作者: Yangyang Feng,Yao Shu
类目: Machine Learning (cs.LG)
*备注: 40 pages, 13 figures, and 19 tables
Abstract:Zeroth-order optimization (ZOO) estimates updates from function evaluations, making perturbation queries a primary cost. Fixed budgets spend the same number of queries at every step, while adaptive controllers may offset their savings by using additional oracle calls to test estimator reliability. We introduce ZoAQ, an adaptive ZOO method built around query reuse. Rather than discarding past evaluations after each step, ZoAQ makes them useful for both the next update and the decision to query further. This enables adaptive query allocation without extra validation queries. Our analysis characterizes when this agreement identifies an update that supports descent and guides the controller to a sufficient query budget. On synthetic objectives, ZoAQ reduces queries by 43-48% relative to fixed baselines using 1.2M queries. In black-box attacks, it reaches 100% success with 320 and 625 average queries on MNIST and CIFAR-10, respectively. Across four OPT fine-tuning settings, ZoAQ saves 43-46% forward evaluations relative to fixed K=4, with accuracy changes within tasks ranging from -0.018 to +0.010.
[LG-172] oward Fairness in Machine Learning Models for Predicting Treatment Retention and Premature Discontinuation in Medication for Opioid Use Disorder
链接: https://arxiv.org/abs/2609.22113
作者: Tongnian Wang,Carolina Vivas-Valencia,Cici Bauer,Yanmin Gong,Kim-Kwang Raymond Choo,Yuanxiong Guo
类目: Machine Learning (cs.LG); Computers and Society (cs.CY)
*备注:
Abstract:Persistent low retention and completion rates in medications for opioid use disorder (MOUD) have driven the use of machine learning (ML) models to predict retention and identify patients at risk of premature discontinuation. However, the fairness of these models across patient populations remains largely unexplored, raising concerns about their application in treatment decision support. This study systematically assesses algorithmic fairness in ML models for predicting MOUD retention and premature discontinuation and investigates the effectiveness of bias mitigation techniques. Using the cross-sectional Treatment Episode Data Set-Discharges (TEDS-D), which includes treatment episodes for individuals in the U.S. discharged between 2015 and 2019, we trained four ML models to predict premature treatment discontinuation and retention beyond 180 days among individuals receiving outpatient MOUD. We evaluated overall performance and subgroup-level error rates across patient subgroups defined by race, ethnicity, age, and sex, complemented by model explanation analyses. We further assessed bias mitigation techniques and their effects on both fairness and predictive performance. Our findings demonstrate that ML models for MOUD outcome prediction can exhibit subgroup-level performance gaps even when overall predictive performance appears acceptable and that bias mitigation can reduce, but not fully eliminate, these gaps without trade-offs. By demonstrating the importance of fairness-aware evaluation and transparent reporting of subgroup performance, this study provides practical insights for the responsible and context-sensitive use of ML models for risk stratification and care prioritization in MOUD treatment settings.
[LG-173] A Shared Learning Rate Is Not a Neutral Control in Selective On-Policy Distillation
链接: https://arxiv.org/abs/2609.22109
作者: Chencheng Zhu
类目: Machine Learning (cs.LG)
*备注: 16 pages, 3 figures, 6 tables
Abstract:Selective on-policy distillation trains a student only at the token positions a selector scores highest, and the literature compares selectors under a single shared learning rate–a control chosen to be neutral. We show it is not. Under LoRA on GSM8K (Qwen2.5-1.5B student, 7B teacher), across an 8x learning-rate grid, dense supervision is statistically flat (swing 1.8 pp, p=0.26) while every selective arm moves with the rate: 5.4 pp for a random 5% subset, 6.7 pp for a total-variation selector, up to 17.7 pp for a teachability selector. Consequently the dense-versus-selective verdict reads 10.1 pp at lr=1e-4 but 5.1 pp at 5e-5–a 2.0x difference decided by a parameter the protocol treats as scenery–and two of six pairwise significance calls between selectors flip between adjacent rates without any rank inversion. We call this selector-rate entanglement and trace it to selection itself rather than step size: AdamW update magnitudes track the rate to within 2.2% despite 15.5x gradient-norm differences across arms. A preregistered frozen-scoring ablation (selection scored by the initial student; criterion, budget, and on-policy rollouts unchanged; 12 seeds per cell) shows live scoring adds 3.79+/-1.69 pp of rate sensitivity (p=0.035) while the frozen arm remains significantly entangled (p=0.015): the feedback loop aggravates the phenomenon rather than causing it. Under full fine-tuning at the rates this literature actually uses (1e-6 to 1e-5) the pattern grows: dense itself swings 19.8 pp, the selective arm 49.5 pp, and the verdict ranges from a non-significant +3.6 pp at the published operating point to +34 pp (p=0.005) one notch hotter. On MATH-500 the rate dependence does not reproduce under LoRA, scoping that result, while the ~10 pp cost of selective training does. We prescribe reporting the arm x rate matrix, not a shared-rate column, as a precondition for selector comparisons.
[LG-174] JAREX: An Acquisition Function for Multi-Objective Algorithmic Process Characterization
链接: https://arxiv.org/abs/2609.24954
作者: Xinyang Li,Kevin Stone,Ajit Vikram
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 26 pages, 11 figures, including supplementary information. Code available at this https URL data and analysis scripts at this https URL
Abstract:Pharmaceutical process characterization is central to Quality by Design because it defines how variations in process parameters affect the ability to meet product quality specifications, thereby supporting proven acceptable ranges and robust manufacturing. In practice, however, characterization still relies largely on factorial design of experiments (DOE) approaches, which are inefficient for resolving multivariate pass/fail boundaries in higher-dimensional spaces. While Bayesian optimization has transformed process optimization, adaptive methods for multi-objective process characterization remain lacking. Here, we introduce JAREX (Joint Acceptable Region EXploration), a Bayesian active-learning acquisition function for multi-objective process characterization. JAREX formulates characterization as a joint boundary-learning problem and adaptively selects experiments to recover the joint pass region defined by simultaneous satisfaction of threshold criteria across multiple objectives. JAREX combines an optimistic joint-feasibility mask with a multi-objective extension of randomized straddle, focusing sampling on the joint edge of failure. Our benchmark study suggests that JAREX provides more accurate and sample-efficient recovery of the joint pass region than factorial DOE, space-filling designs, and greedy objective-wise strategies over the full experimental budget range. For batched experimentation, it reduces the number of iterative process characterization experiments by more than half while preserving high accuracy for the boundary-identification task. Implemented in the open-source obsidian package, JAREX provides a modular framework for adaptive, data-efficient multi-objective algorithmic process characterization, supporting sample-efficient range finding in high-dimensional spaces.
[LG-175] Conformalized Quantile Regression and Minimax Limits of Fixed-Score Calibration under Known Covariate Shift
链接: https://arxiv.org/abs/2609.24929
作者: Rustam Isaev,Anton Conrad,Denis Belomestny,Eric Moulines,Sergey Samsonov
类目: atistics Theory (math.ST); Machine Learning (cs.LG); Machine Learning (stat.ML)
*备注: 65 pages, 3 figures
Abstract:In this paper, we study nonasymptotic L^p error bounds for interval length and conditional coverage in split conformalized quantile regression (CQR). Our bounds rely on local regularity conditions and accuracy guarantees for the estimated quantiles. We further instantiate our bounds for quantile regression with sparse ReLU neural networks. We also consider covariate shift, where the calibration and test covariates have different distributions, and derive nonasymptotic bounds for this setting. We obtain matching minimax upper and lower bounds in expectation for two constructed fixed-score calibration benchmarks under known covariate shift. The bounds match for every p\in[1,\infty] in the scalar problem and for finite p in the K -threshold problem; for the latter, a high-probability minimax lower bound holds for every p\in[1,\infty] .
[LG-176] XSQ-AST: An Explainable Audio Spectrogram Transformer Framework for Localising Synthetic Speech Artifacts ICASSP2027
链接: https://arxiv.org/abs/2609.24770
作者: Ben Heritage,Luca Resti,Mónica Villanueva Aylagas,Timothy Mehlenbacher,Konrad Tollmar,James Alfred Walker
类目: Audio and Speech Processing (eess.AS); Machine Learning (cs.LG); Sound (cs.SD); Signal Processing (eess.SP)
*备注: Submitted to IEEE ICASSP 2027
Abstract:Localising artifacts in synthetic speech remains challenging, as most evaluation methods yield only global quality scores. This paper presents XSQ-AST, a framework that combines the SQ-AST speech quality model with WhisperX phoneme alignment and multiple saliency methods to produce temporally localised artifact diagnostics without model retraining. Saliency maps are projected onto continuous distributions via kernel density estimation and onto phoneme boundaries via phoneme-discretised saliency maps. A 40-participant listening test validated the framework across five perceptual dimensions. Attention Rollout, Attention Flow and an adapted GradCAM produced temporal distributions that correlated with listener highlights, with different methods best suited to different artifact types. An AUC-ROC analysis confirmed discrimination above chance.
[LG-177] Reinforcement Learning in Operational Research: A Technical Review and Practical Roadmap
链接: https://arxiv.org/abs/2609.24750
作者: Yahan Lu,Dongyang Xia,Nursen Aydin,Shadi Sharif Azadeh
类目: Optimization and Control (math.OC); Machine Learning (cs.LG)
*备注:
Abstract:The growing demand for real-time, data-driven decision-making in complex and dynamic systems is placing increasing pressure on traditional Operational Research (OR) methodologies. Reinforcement learning (RL) has emerged as a complementary approach, offering strong learning and computational capabilities for sequential decision-making in dynamic and uncertain environments. Recent research shows an increasing interest in integrating RL with OR to address dynamic decision-making problems, enhance heuristic and exact methods for combinatorial optimization, and support the development of digital replicas of operational systems. The overarching goal across these efforts is to leverage the learning capabilities of RL to strengthen traditional OR algorithms, improving solution quality, computational efficiency, and robustness. Given the diversity of integration approaches and application settings, there is a clear need for a systematic and technically detailed review of how RL empowers OR methods. To address this gap, this paper presents a structured review of three key roles that RL plays in empowering OR: (i) solving sequential decision-making problems in dynamic environments, (ii) serving as an end-to-end solution method or as a component integrated within heuristic and exact OR methods for combinatorial optimization problems, and (iii) facilitating extended reality analysis through integration with digital twin systems. We critically synthesize recent advances across these roles, highlighting their advantages, implementation requirements, limitations, and challenges. Finally, based on these insights, we outline a roadmap for future research to further advance the methodological and practical integration of RL and OR.
[LG-178] Identifying Representational Biases in Datasets Using PCA: A Max-Disparity Partition Framework
链接: https://arxiv.org/abs/2609.24556
作者: Arjun KM,Shashi Jain
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Principal Component Analysis (PCA) minimises aggregate reconstruction error, which can inadvertently represent majority subgroups with substantially higher fidelity than minority subgroups. Fairness-aware extensions of PCA correct this disparity but require group labels as input. We address the logically prior question: given only a data matrix, which binary partition of the data suffers the greatest representational disparity under a shared PCA projection? We formalise this as the max-disparity partition problem and propose a greedy local-search algorithm, grounded in the Fiduccia-Mattheyses bipartitioning framework, that discovers the disparity-maximising partition without any predefined group labels. Two benchmark algorithms, a fixed-projection sorting baseline and a simulated-annealing variant, confirm that the greedy solution is empirically near-optimal. Having identified the partition, we attribute the disparity to specific features via PCA loading scores and association rule mining, enabling a practitioner to assess whether the disadvantaged group corresponds to a human-meaningful minority. On the Predict Students’ Dropout and Academic Success dataset, representational disparity is driven predominantly by institutional and programmatic proxies for socioeconomic disadvantage, with gender emerging as a secondary but consistent contributor within the disadvantaged group. The discovered partition is then passed directly to Fair PCA, completing a detect-explain-mitigate pipeline.
[LG-179] Beyond Point Prediction: Artificial Representative Trees with Uncertainty
链接: https://arxiv.org/abs/2609.24528
作者: Lea L. Mairhöfer,Silke Szymczak,Björn-Hergen Laabs,Tuwe Löfström-Cavallin
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 28 pages and 10 figures (without appendix)
Abstract:Random forests (RFs) predict well but are opaque, whereas single decision trees are interpretable but unstable. Artificial representative trees (ARTs) were developed as interpretable surrogate models for RFs, but their use as standalone prediction models with uncertainty quantification has not been systematically investigated. We combine ARTs with leaf-wise Mondrian conformal predictive systems (CPS), enabling a single tree to provide continuous predictions, prediction intervals, and probabilities of exceeding arbitrary thresholds. We compared ARTs with CPS against decision trees with CPS and separate regression and probability trees across five simulation scenarios, 21 benchmark datasets, and a cross-sectional NHANES example data set. Repeated cross-validation assessed predictive performance, interpretability, and stability. ARTs with CPS yield compact, structurally stable trees with substantially more reproducible split-variable selection than decision trees across benchmark datasets and NHANES. Decision trees showed slightly better predictive performance and narrower prediction intervals, while coverage was broadly comparable. CPS-based trees generally achieved lower and less variable Brier scores than multi-model approaches. Combining ARTs with CPS therefore provides a single, interpretable, and stable model for continuous predictions and calibrated probabilities, balancing predictive performance with reproducibility and transparency in settings where stability and interpretability are essential. Comments: 28 pages and 10 figures (without appendix) Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG) MSC classes: 62G30, 62G15, 62R07, 68T37, 68T05, 68T30 Cite as: arXiv:2609.24528 [stat.ML] (or arXiv:2609.24528v1 [stat.ML] for this version) https://doi.org/10.48550/arXiv.2609.24528 Focus to learn more arXiv-issued DOI via DataCite (pending registration) Submission history From: Lea Louisa Mairhöfer [view email] [v1] Mon, 21 Sep 2026 13:01:36 UTC (8,308 KB)
[LG-180] Horizon-Aware Early Event Prediction for Tokamak Disruption Alarms
链接: https://arxiv.org/abs/2609.24443
作者: Takeshi Koshizuka,Takaharu Yaguchi
类目: Plasma Physics (physics.plasm-ph); Machine Learning (cs.LG)
*备注:
Abstract:Reliable disruption prediction is essential for the safe operation of future tokamaks. Existing full-distribution survival methods model the complete residual time-to-disruption distribution, whereas operational decisions primarily depend on disruption risk within a finite prediction horizon. This mismatch motivates introducing Early Event Prediction (EEP) objectives into survival-based disruption prediction. We take Deep Survival Machines (DSM) as the full-distribution baseline and propose applying two established EEP methods to tokamak disruption prediction: Temporal Label Smoothing (TLS), which directly predicts disruption probability within a finite horizon, and survTLS, which additionally models the event-time distribution within that horizon. Using a common causal encoder, we compare these methods on DIII-D, Alcator C-Mod, and EAST. We distinguish threshold-free deadline ranking from validation-selected fixed-policy alarm performance and evaluate prediction horizons and encoder architectures. TLS achieves the best mean alarm performance on DIII-D and EAST, whereas all methods perform poorly on Alcator C-Mod. survTLS does not consistently outperform DSM, suggesting that directly learning horizon-level event probability is more effective than modeling detailed within-horizon event-time distributions in the present setting. Finally, the selected prediction horizons and encoder-ablation results vary across devices, reflecting differences in disruption characteristics.
[LG-181] Complexities of Weak Proximal Oracle Methods for Composite Convex Optimization
链接: https://arxiv.org/abs/2609.24423
作者: Dan Garber
类目: Optimization and Control (math.OC); Machine Learning (cs.LG)
*备注:
Abstract:We consider a standard convex composite optimization problem with either smooth or nonsmooth objective function, and under quadratic growth. In recent years, several works gave algorithms based on a \textitweak proximal oracle (WPO) that essentially match in oracle complexities proximal (sub)gradient methods relying on exact prox operations. Importantly, such WPOs, which relax the strong optimality condition of the standard prox operator, may admit much more efficient implementation in terms of runtime when optimal solutions have some sparse structure. A question remained if such WPO-based methods can be accelerated (in the sense of Nesterov’s accelerated gradient). In this work we provide a negative answer by establishing lower bounds against both deterministic and randomized methods. Thus, while WPOs can substantially reduce the cost of individual oracle calls, this comes with an inherent loss in oracle complexity. We also provide a new upper-bound for WPO-based nonsmooth convex composite optimization, nearly matching the proximal subgradient method.
[LG-182] Adversarially Robust PAC Learning with Optimal VC Rates
链接: https://arxiv.org/abs/2609.24260
作者: Steve Hanneke,Amirreza Shaeiri
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Statistics Theory (math.ST)
*备注: 35 pages, 2 figures
Abstract:We study the problem of \emphadversarially robust PAC learning. In this framework, the learner observes independent samples from an unknown distribution over \mathcalX \times \0,1\ , as in classical PAC learning. However, given a perturbation map \mathcalU : \mathcalX \to 2^\mathcalX known to the learner, the goal is to output, with high probability, a predictor that correctly classifies \emphevery perturbation z \in \mathcalU(x) of most future examples (x,y) drawn from the same underlying distribution. We determine the \emphoptimal \mathcalU -independent sample complexity of this problem in both the realizable and agnostic settings. More specifically, for every concept class \mathcalH of \operatornameVC dimension d , we prove upper bounds of \mathcalO \big( d/\epsilon + \log(1/\delta)/\epsilon \big) in the realizable setting and \mathcalO \big( d/\epsilon^2 + \log(1/\delta)/\epsilon^2 \big) in the agnostic setting, together with an optimal first-order refinement of the latter. These bounds match the corresponding lower bounds for classical PAC learning. Consequently, and perhaps surprisingly, adversarial robustness incurs \emphno additional distribution-free statistical cost, uniformly over all perturbation maps. Our bounds improve exponentially on those of [Montasser, Hanneke, and Srebro; COLT '19]. On the technical side, we present short and elementary proofs based on a new algorithmic principle that we call \emphbinomial-bagging. We believe that binomial-bagging and its analysis may be of independent interest. Comments: 35 pages, 2 figures Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Statistics Theory (math.ST) Cite as: arXiv:2609.24260 [stat.ML] (or arXiv:2609.24260v1 [stat.ML] for this version) https://doi.org/10.48550/arXiv.2609.24260 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
[LG-183] P2Flow: Phoneme-aware Progressive Flow Matching for Extreme Speech Super-Resolution
链接: https://arxiv.org/abs/2609.24138
作者: Ningyuan Yang,Yize Li,Pu Zhao,Diego A. Cuji,Kanad Sarkar,Ryan M. Corey,Xue Lin,Andrew C. Singer
类目: Audio and Speech Processing (eess.AS); Machine Learning (cs.LG); Sound (cs.SD)
*备注: Under review
Abstract:Generative models have recently demonstrated considerable promise in speech super-resolution (SSR). Nevertheless, the majority of existing work has concentrated on standard or versatile SSR configurations, leaving the extreme setting with severely limited spectral inputs largely unexplored. In this regime, current approaches exhibit marked performance degradation, underscoring the need for dedicated solutions. To bridge this gap, we introduce P2Flow, a phoneme-aware progressive flow matching (FM) framework designed for extreme SSR with three main strategies. First, our model leverages phonetic information to reconstruct missing spectral components. Furthermore, it employs a progressive architectural design that hierarchically restores distinct frequency regions. Finally, we incorporate post-training of the vocoder to enhance overall waveform fidelity. Extensive experiments are conducted on the TIMIT and VCTK datasets under both 1 kHz to 16 kHz and 2 kHz to 16 kHz settings, demonstrating that P2Flow yields state-of-the-art results across multiple evaluation metrics.
[LG-184] OSCAR: Order-aware Scoring and Calibration for AI Rankings
链接: https://arxiv.org/abs/2609.24128
作者: You Liu,Yue Liu,Quanchao Lu,Nick Shipilov
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Applications (stat.AP)
*备注:
Abstract:Judge-specific sensitivity is useful for aggregating pairwise LLM evaluations, but its interpretation depends on which systematic presentation effects the ranking model includes. We introduce OSCAR, an order-aware framework for scoring and calibrating AI rankings, and study position as one such effect. In released judgments from 18 evaluators, the all-response A-minus-B score difference ranges from -63.11 to 98.31 percentage points. Matching question text, response texts, candidate identities, and judge within the released table gives an overall difference of 24.22 points (95% interval [22.90,25.54] ), conditional on the released text mapping. A controlled calculation isolates the potential consequence: with true sensitivity fixed at one, omitting a position intercept of four reduces the population-optimal slope to 0.0771 . We extend sensitivity-based ranking with judge-specific position, length, and family terms, characterize local omission-induced displacement and an identification failure, and propagate prompt-cluster uncertainty to adjusted comparisons. Across four released datasets, position provides the largest stand-alone predictive improvement. Refitting bootstrap comparisons show more selective gains from the full model over position-only adjustment. In dependent binary simulations, adjusting both the mean and covariance yields 94.4–95.2% coverage; correcting either alone is insufficient. At N=10,000 , OSCAR reduces mean neutral-target RMSE from 0.1158 under the sensitivity-only model to 0.0237 .
[LG-185] Model-Agnostic Feature Selection via LOCO-Guided Adaptive Minipatch Sampling
链接: https://arxiv.org/abs/2609.24126
作者: Xuhui Liu,Lili Zheng
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME)
*备注:
Abstract:Black-box machine learning models increasingly deliver strong predictions, but extracting useful information from them, such as a set of important features, remains challenging. Existing model-agnostic methods primarily estimate feature importance or conduct inference on it rather than directly selecting features, whereas many feature selection methods are model-specific or rely on the model-X assumption. We introduce LOCO-guided Adaptive Minipatch Sampling (LAMPS), a model-agnostic ensemble framework that uses any black-box regression algorithm as its base learner to select features important for predicting the response. The base learner need only produce predictions and need not perform feature selection itself. LAMPS operates within a minipatch ensemble framework that subsamples both observations and features, allowing leave-one-covariate-out (LOCO) feature importance scores to be easily computed. It adaptively concentrates minipatch sampling on features with high LOCO scores while maintaining exploration. The resulting sampling probabilities rapidly separate signal from noise features after a few iterations, enabling selection through simple thresholding. We establish that LAMPS achieves exact feature selection in high-dimensional settings, provided that the base predictive models are sufficiently well trained on average. Extensive experiments on synthetic and real data show that LAMPS outperforms state-of-the-art feature selection methods, with particularly strong performance in the presence of correlated features.
[LG-186] Causal Bayesian Optimization: Foundations Methods and Applications
链接: https://arxiv.org/abs/2609.24112
作者: Chenfeng Huang,Thuy T. Le,Zixuan Ma,Hien Tran
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: Accepted at Transactions on Machine Learning Research (TMLR), 2026
Abstract:Causal Bayesian Optimization (CBO) combines causal inference with Bayesian optimization to enable sample-efficient intervention selection in systems with causal structure. This survey provides a systematic review of CBO through a unified BO-loop perspective, showing how causal assumptions shape intervention search spaces, surrogate models, acquisition functions, and decision policies. We organize existing methods by graph and system-knowledge assumptions, environment, intervention representation, surrogate architecture, and decision rule, and connect CBO to causal bandits, Bayesian experimental design, safe optimization, policy search, and causal abstraction. We also introduce a reproducibility-oriented benchmark spanning hard- and soft-intervention settings, with standardized GAP and a new trajectory-aware Path-Aware GAP (PA-GAP), evaluating seven CBO methods and a non-causal BO baseline across thirteen datasets, three budgets, and two metrics. Results show that no method dominates uniformly: rankings depend on dataset, budget, metric, and how causal information is used, while strong non-causal baselines remain competitive in several settings. Controlled graph-misspecification and omitted-variable stress tests further show that rankings can change substantially when learner-side causal information is perturbed. We conclude by identifying key open challenges, including robustness to causal-assumption violations, scalable unknown-graph optimization, mixed intervention types, realistic cost models, stronger theoretical guarantees, and integration with modern representation learning and causal abstractions.
[LG-187] Exponential Family Synthetic Controls
链接: https://arxiv.org/abs/2609.23970
作者: Hector Rodriguez-Deniz,David M. Blei
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:We develop exponential family synthetic controls (EFSC), a distributional version of synthetic controls for a panel of datasets. Each cell of the panel corresponds to a dataset drawn from an exponential family whose natural parameters factorize probabilistically across units and times. We estimate the latent factors using black-box variational inference. This replaces the usual weighted-average view of synthetic controls with a flexible probabilistic model that operates on full distributions. We propose causal estimands based on divergences between pre- and post-intervention distributions induced by the posterior of the natural parameters, together with distributional placebo tests to support causal inference and assess the significance of the estimated effects. We validate the proposed framework on synthetic and real data. Across a variety of exponential-family distributions, EFSC accurately recovers causal effects induced by exponential tilts, together with the corresponding divergences between treated and counterfactual distributions. The framework also captures effects induced by structural perturbations of the latent factors and by heavy-tailed noise contamination. Finally, we apply EFSC to study the expansion of Medicaid under the Affordable Care Act (ACA) and its impact on the distribution of health insurance coverage across U.S. states. Code is available at this https URL.
[LG-188] Sparse Regression Distilled from a Single Robust Fit
链接: https://arxiv.org/abs/2609.23937
作者: Wooyoung Shin,Seunghwan Park
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Methodology (stat.ME)
*备注: 68 pages, including appendices; 8 figures
Abstract:Robust linear fits can resist response contamination yet remain too dense or unstable for useful global explanations. We propose penalized distillation, which fits a smoothly clipped absolute deviation (SCAD) estimator to a robust initial estimator’s empirical fitted surface along a safeguarded coordinate-descent path and evaluates candidate states separately for fidelity, parsimony, perturbation stability, and held-out prediction. The new results attach to the states the algorithm actually computes. Conditional on a fixed uncontaminated design, deterministic bounds transfer response-replacement boundedness from the initial fit to every retained path state. Turning to fixed dimension, we characterize the oracle-support branch by its empirical-Gram projection and influence function, give conditions for covariance-weighted least-squares approximation equivalence, and establish a path-conditional generalized information criterion. By contrast, at large dimension-to-sample ratios the full-coordinate robust fit collapses without warning, and screening restores the construction. Under a sure-screening framework, the robustness bound and the support and selection guarantees transfer to the screened fit. Simulations separate robustness transfer from support recovery, efficiency, and computation across the dimension-to-sample ratio, with p up to 240, and the signal density, which isolates what the sparse stage adds once the screen over-selects. In a duplicate-grouped superconductivity study, the distilled estimator remains predictively stable under prespecified training-response shifts but retains 66.8–68.8 of 81 slopes. Stronger sparsification reduces the model to 12.6–14.0 slopes only at visible fidelity and prediction cost. Distillation therefore preserves predictive stability on these data without substantiating a compact coordinate-level explanation.
[LG-189] Density-Ratio Rescoring for Imbalanced Classification
链接: https://arxiv.org/abs/2609.23926
作者: Dongha Kim,Seunghwan Park
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 38 pages, 12 figures; includes all appendices
Abstract:Density-Ratio Rescoring (DRR) augments a classifier trained at the original class prior with a survey-raking dual score. Raking reweights the majority sample to match minority feature moments within a tolerance. DRR marginally standardizes the dual and base scores and combines them with a fixed weight of one half, using the fitted dual directly for prediction without resampling or refitting the base classifier. Under exact population matching and a correctly specified log-linear tilt model, the dual equals the log density ratio up to an additive constant. A class-separation analysis characterizes the signal strength and correlation conditions under which fusion improves separation under common within-class covariance. On 24 tabular benchmarks, evaluated over 30 trials and five base learners, DRR at the D=128 random-feature setting improves average precision over the standardized base on every dataset, with a mean gain of 0.034. It exceeds the shared-dual raking-and-relabeling resampler on 22 of 24 datasets, with a mean gain of 0.092 , and on all eight one-versus-rest tasks of a shared gene-expression cohort. These results demonstrate the effectiveness of using raking duals as reusable scores for improving rare-class ranking while retaining classifiers trained at the original prior.
[LG-190] he Exponential Price of Determinism in Nonsmooth Nonconvex Optimization
链接: https://arxiv.org/abs/2609.23837
作者: Guy Kornowski
类目: Optimization and Control (math.OC); Machine Learning (cs.LG)
*备注: 13 pages
Abstract:We study the complexity of finding (\delta,\epsilon) -Goldstein stationary points of nonsmooth nonconvex Lipschitz functions. By now, it is known that randomized first-order algorithms can solve this task with a dimension-free oracle complexity [Zhang et al., 2020], whereas deterministic algorithms cannot, as their complexity must scale at least linearly with the dimension d [Jordan et al., 2023, Tian and So, 2024]. This leaves open whether deterministic algorithms can nevertheless solve the problem with oracle complexity polynomial in d . We answer this question negatively by proving a lower bound of order (1/\epsilon)^\Omega(d) for deterministic algorithm, closing the exponential gap between the previously known lower and upper bounds and resolving an open problem posed by Jordan et al. [2023]. We further discuss several extensions and implications of this result to weaker stationarity notions, finding a descent direction and deterministic smoothing. Overall, our results establish an exponential computational advantage in nonsmooth nonconvex optimization offered by randomization.
[LG-191] On Generalized Naive Bayes with Continuous Features
链接: https://arxiv.org/abs/2609.23819
作者: Ábrahám Papp,Botond Szilágyi,Edith Alice Kovács
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:The Generalized Naive Bayes (GNB) model was introduced for discrete and categorical random variables as an extension of classic Naive Bayes. We now accommodate the GNB framework to continuous explanatory variables. A central result of the paper is that structure learning of the GNB depends only on the pair copulas of the bi-variate marginals. We proved that the GNB structure can be assigned to the basis of a matroid, therefore we give greedy algorithms for finding the optimal GNB structure on the training data, in sense of minimizing Kullback-Leibler divergence. Three cases are considered: joint Gaussian distribution, then a more flexible model where we suppose the dependence structure to be described by a Gaussian copula with arbitrary marginals, and an even more flexible case where the joint continuous probability distribution is arbitrary, i.e. copula and marginal distributions are arbitrary. A method for model reduction, based on the newly introduced concept of GNB forest is given. We close the paper by comparing the newly introduced GNB classification results to other classical “glass-box” algorithms on real datasets.
[LG-192] EMPER: Temporal Encoder-Masked Probabilistic Ensemble Regressor for Time-Series Forecasting
链接: https://arxiv.org/abs/2609.23701
作者: Giancarlo Vercellino
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Probabilistic forecasting requires accurate central predictions and calibrated uncertainty estimates. This paper presents TEMPER, the Temporal Encoder-Masked Probabilistic Ensemble Regressor, a univariate time-series forecasting algorithm that combines a temporal autoencoder, a differentiable masked neural decision forest, continuous ranked probability score (CRPS) training, and Gaussian-mixture post-processing. The R implementation is built on torch for R and returns horizon-wise density, distribution, quantile, and sampler functions. We evaluate TEMPER on three deterministic synthetic level series with trend, periodic, regime-switching, nonlinear-threshold, and heteroskedastic components. Across 96 rolling-origin forecasts at horizons t + 1, t + 5, t + 20, and t + 60, TEMPER obtains 2.824% mean CRPS normalized by origin level, 3.635% median absolute error, and 68.8% empirical 90% interval coverage after training with a 300-epoch cap and early-stopping patience of 100. A naive persistence bootstrap has the best aggregate CRPS, 2.763%, while TEMPER has the best median absolute error and the best CRPS at t+1 and t+5. The ablation study uses matched series-origin-horizon cells, horizon-wise CRPS deltas, endpoint sensitivity summaries, and a calibration-specific interval study. Relaxing the learned mask improves average CRPS by 0.472 percentage points on the ablation subset, mainly through long-horizon gains. A twofold interval inflation improves held-out coverage from 54.2% to 91.7% and gives the best 90% interval score among tested calibration rules. The results identify calibration, horizon-specific tuning, and component selection as the central research priorities.
[LG-193] Fast Graph Laplacian Estimation using Effective Resistance
链接: https://arxiv.org/abs/2609.23668
作者: Christoffer Kjellson,Claudio Altafini,Emma Tegling
类目: ignal Processing (eess.SP); Machine Learning (cs.LG)
*备注:
Abstract:Inferring network topology from noisy node observations is a central problem in graph signal processing. In this paper, we consider Laplacian-constrained graph estimation for Gaussian Markov random fields, focusing on the underdetermined regime in which the number of samples is smaller than the number of graph nodes. Existing approaches often formulate the problem as a sparsity-regularized maximum-likelihood estimation problem. While effective, such methods typically require iterative optimization and are often computationally demanding, particularly under Laplacian constraints. Instead, we propose a non-iterative estimator of graph Laplacians that uses effective resistance for regularization, and evaluate the method using a simple sparsification procedure. Experiments show that with some trade-off in edge and weight recovery on the considered dataset, computational cost for moderately sized graphs can be substantially reduced.
[LG-194] PACE: Plug-and-Play Contextual Embedding for Feature Screening with Pretrained Tabular Foundation Models
链接: https://arxiv.org/abs/2609.23574
作者: Qi Qin,Erbo Li,Ting Wei,Zizhou Huang,Zixuan Qin,Wu Wang,Yifan Sun
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Applications (stat.AP); Methodology (stat.ME)
*备注:
Abstract:In high-dimensional tabular learning, feature screening provides a lightweight, model-agnostic way to remove irrelevant features before model fitting. However, scoring raw values directly can miss nonlinear or distributional structure. We introduce PACE (Plug-and-Play Contextual Embedding), which inserts a frozen tabular foundation model (TFM) column encoder before an existing feature-scoring rule, expanding each feature into a higher-dimensional contextual representation. Across controlled studies, PACE improves raw-space screening of complex nonlinear dependence with only modest additional encoding cost. These gains translate to downstream prediction on TALENT datasets: PACE-DC improves binary AUC by 0.077 and multiclass macro-AUC by 0.064, with a median normalized RMSE improvement of 0.063 across ten learners. Matched random-weight and random-feature controls show that PACE gains from pretrained structure beyond generic dimensional expansion. PACE further achieves favorable performance–time trade-offs against task-fitted selectors and attribution-based methods, positioning pretrained column geometry as a reusable upstream primitive for tabular learning.
[LG-195] Comparative Study of Quantum and Classical Machine Learning Models in Binary Classification
链接: https://arxiv.org/abs/2609.23476
作者: Anand Kumar Mishra,Ramanuj Awasthi
类目: Quantum Physics (quant-ph); Machine Learning (cs.LG)
*备注:
Abstract:A potential path forward is Quantum Machine Learning (QML), which aims to leverage quantum computing in conjunction with classical machine learning to enhance computing efficiency and the expressiveness of models. In this paper, two different quantum classifiers - Variational Quantum Classifier (VQC) and Quantum Kernel Support Vector Machine (QSVM) - are compared with three classical classifiers as baseline classifiers - Logistic Regression, Support Vector Machine (SVM), and a Multi-Layer Perceptron (MLP) - on the Breast Cancer Wisconsin dataset. The quantum circuits were created in the PennyLane framework and simulated on a classical backend. However, in terms of accuracy, classical Logistic Regression performed better with an accuracy of 97.8%, classical SVM and QSVM with an accuracy of 95.6% each, although the Quantum VQC achieved a lower accuracy of 88.9% and had a recall of 100% for the benign class, though it correctly identified only 12 of the 17 malignant cases (a malignant-class recall of approximately 70.6%). The drawback of quantum models is the higher training time; however, since the quantum circuit needs to be classically simulated, the quantum SVM took 23.29 seconds compared to less than 0.01 seconds for the classical linear models. These results indicate that for small structured datasets, classifiers based on quantum computing have not yet surpassed well-tuned classical counterparts. In some respects (e.g., benign-class recall), they perform competitively, though not on malignant-class recall, where the VQC in particular performed worse than the classical baselines, which is worth further investigation on real quantum computers.
[LG-196] Leverag ing Industrial Foundation Models at the Edge of Particle Physics Detectors via Distillation Learning and Hardware Co-design
链接: https://arxiv.org/abs/2609.23385
作者: Gia Ancone,Qibin Liu,Liangyu Wu,Julia Gonski
类目: Instrumentation and Detectors (physics.ins-det); Machine Learning (cs.LG); High Energy Physics - Experiment (hep-ex)
*备注: 7 pages, 1 figure, 1 table
Abstract:Data acquisition (DAQ) systems at future particle physics experiments stand to benefit from the extremes of AI/ML development: large-scale foundation models can enhance the performance of feature extraction algorithms, and small-scale on-detector deployments can enable real-time intelligent data handling. This work provides the first fine-tuning of an industrial foundation model for particle physics DAQ. Starting from the backbone of Google Research’s TimesFM (Time Series Foundation Model), we demonstrate fine-tuning on real-time regression tasks for drift chamber trackers and dual-readout calorimeters. Furthermore, the fine-tuned TimesFM model is distilled into a student and co-designed with FPGA implementation to enable these models to run in real-time at future colliders. The fine-tuned distillations meet or exceed the performance of previously published AI/ML solutions for each task. Further, the pipeline of distillation and model compression from TimesFM is generic and can be easily adapted to a variety of 1D waveform tasks across domains.
[LG-197] Low-Rank Frequency Convolution and Noise-Range Augmentation for Real-Time Pitch Estimation on Edge Devices ICASSP2027
链接: https://arxiv.org/abs/2609.23340
作者: Venkat Suprabath Bitra,Homayoon Beigi
类目: Audio and Speech Processing (eess.AS); Machine Learning (cs.LG); Sound (cs.SD)
*备注: 11 pages, 11 tables. Columbia University Nonlinear Control Laboratory Technical Report CUNLC-20260919-02. A shortened version has been submitted to ICASSP 2027
Abstract:Pitch estimation on an edge device is constrained in three ways at once. The model must be small, it must stay accurate when the input is noisy, and one frame must be produced inside the frame period. In this report the Frequency Convolution Network (FrCN) of our earlier work is factored into a low-rank form. The number of parameters is reduced by 35.9%, from 17,787 to 11,397, and accuracy is not reduced, either in domain or on two corpora the model was never trained on. The range of the noise used during training is also shown to dominate the architecture in setting how the model behaves when the interference is severe. When the training noise floor is lowered from +6.02 dB to -20 dB, 0.35 points of clean RPA50 are lost and RPA50 at -20 dB is raised from 2.44 to 22.41. This effect is about two orders of magnitude larger than any architectural effect that was measured. The semi-orthogonal constraint used in TDNN-F is found to be redundant with the normalization inside the bottleneck, and accuracy is reduced when both are applied. Whether the factorization saves time depends on the runtime: in eager PyTorch the factored model is 45% slower, while in a compiled kernel it is 15% faster. For deployment, a small C kernel was written. It needs 83 kB on disk and no runtime library beyond libc and libm. It is faster than OpenBLAS on all four CPUs that were tested, faster than ONNX Runtime by 3.8 times, and faster than PyTorch by 13 times. Its output was checked against PyTorch on 271,893 held-out frames per model, and the same pitch bin was selected on every one of them.
[LG-198] Stochastic Reconfiguration as Statistical Filtering for Overparameterized Neural Quantum States
链接: https://arxiv.org/abs/2609.23334
作者: Tak Hur
类目: Quantum Physics (quant-ph); Machine Learning (cs.LG)
*备注: 25 pages, 7 figures
Abstract:Stochastic reconfiguration (SR) is the standard optimizer for neural quantum states (NQS), but modern NQS often have far more parameters than Monte Carlo samples. We show that in this regime the diagonal shift is more than a numerical stabilizer. It acts as a statistical filter for finite-sample generalization. At a fixed wave function, SR is ridge regression from tangent features to the centered local energy. Its residual is the expressivity gap, the part of imaginary-time evolution outside the current tangent space. This gap is orthogonal to the tangent space in population, but finite batches make it act as noise that SR can overfit. The shift therefore balances shrinkage of useful update directions against variance from fitting sampled residuals. Exact diagnostics on a 4\times4 Heisenberg graph separate two effects of overparameterization. Larger tangent spaces help when they reduce the expressivity gap, but they can hurt when they overfit a fixed gap. In a 100 -site transverse-field Ising family trained with a foundation NQS, validation risk is U-shaped in the shift while variance decreases, matching the noisy-ridge model. This view leads to multi-shift SR (MS-SR), which averages independent ridge solves at data-adaptive shifts to form a richer, lower-variance spectral filter. Checkpoint-local experiments show that MS-SR lowers validation risk and update variance relative to the fixed-shift SR baseline. We further compare MS-SR and SR in paired online training continuations, with independent endpoint energy evaluations and a separate update-cost benchmark.
[LG-199] Stochastic Flow Map for Count Data
链接: https://arxiv.org/abs/2609.23290
作者: Ganchao Wei
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:High-dimensional count data are common in scientific applications, but most diffusion and flow models are designed for continuous or categorical data, and generation often requires many sequential model evaluations. We propose Count Flow Map, a generative model that learns finite-time transitions directly in count space for one- or few-step generation. Our model directly learns stochastic transitions over finite time intervals, using Poisson births and Binomial deaths to preserve nonnegative integer counts without a predefined maximum. These transition models are trained to match the underlying local birth–death dynamics and to maintain consistency across step sizes. We characterize the connection between local dynamics and finite-time transition consistency and derive a bound on the generation error. After validating Count Flow Map in several simulations, including a high-dimensional, high-count setting, we apply it to single-cell drug perturbation prediction and neural population forecasting, where it captures perturbation responses and supports forecasts of high-activity events with only one or a few model evaluations. Together, these experiments demonstrate that Count Flow Map enables high-quality generation directly in count space across inference budgets, from one-step to few-step generation, using a single trained model.
[LG-200] Auditing Bayesian Graph Alignment: Diagnostic Comparisons and Reference Failure
链接: https://arxiv.org/abs/2609.23232
作者: Melika Gorgi,Kourosh Mirsohi
类目: Applications (stat.AP); Machine Learning (cs.LG); Computation (stat.CO); Methodology (stat.ME)
*备注: 19 pages, 4 figures. Codes: this https URL
Abstract:Bayesian graph alignment estimates correspondence probabilities, but convergence of an alignment-score trace need not imply accurate correspondence marginals. We audit this gap on 240 new exact graph pairs from four source families, 240 larger pairs with 20-100 vertices, and a separate 60-case exact implementation check. Under an explicit edge-flip likelihood, we compare three samplers and score, marginal, indicator, categorical, and classifier-based diagnostics. Marginal disagreement improves error discrimination over score R-hat for the exact informed sampler, but its improvement for vanilla local sampling is uncertain. Assignment-based R* and short indicator panels are competitive; no diagnostic dominates across samplers and endpoints. At larger sizes, diagnostics predict subsequent marginal changes, not posterior error, and classification performance depends on the drift threshold. Disjoint-window and held-out-chain checks attenuate but preserve positive associations. Only 22 of 240 original reference sets pass an agreement screen. On forty failure-selected cases, eightfold SMC particle escalation does not resolve disagreement, whereas additional rejuvenation helps. Longer informed runs remain unstable. An elementary feasible-alignment bound demonstrates severely unrepresentative SMC and informed-chain scores in concentrated 100-vertex cases, independently of approximate reference consensus. We also exhibit common-start chains with near-zero disagreement despite exact marginal error near .967. These results support assignment-sensitive auditing while identifying limits of finite budgets, diagnostic rankings, and reference agreement as evidence of accuracy.
[LG-201] Stealing profits: Spread-based temporal hierarchy forecasting for day-ahead electricity markets
链接: https://arxiv.org/abs/2609.23223
作者: Arkadiusz Lipiecki,Nikolaos Kourentzes,Rafal Weron
类目: atistical Finance (q-fin.ST); Machine Learning (cs.LG); Econometrics (econ.EM)
*备注:
Abstract:Day-ahead electricity price forecasts support trading and storage decisions, but for battery arbitrage predicting intraday price spreads is more relevant than predicting individual hourly prices. Here we show that a temporal hierarchy forecasting (THieF) framework that jointly reconciles forecasts of hourly electricity prices and all intraday price spreads consistently improves performance across two major European electricity markets and three different forecasting architectures. Using five years of out-of-sample data from Germany and Spain, we obtain accuracy improvements of up to 19.7% and profit gains of up to 10.4% relative to unreconciled hourly price forecasts. The gains persist even for a highly accurate pretrained TabPFN foundation model. Our results demonstrate that exploiting coherent relationships between economically relevant forecasting targets can improve both predictive accuracy and decision value, and that better statistical forecasts do not necessarily imply better economic decisions.
[LG-202] Bayesian Deck-of-cards-based Ordinal Regression with Sequential Preference Elicitation
链接: https://arxiv.org/abs/2609.23212
作者: Marco Grillo,Silvano Zappalà
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注:
Abstract:The Deck-of-cards-based Ordinal Regression (DOR) infers a value function from a ranking of reference alternatives in which the Decision Maker (DM) inserts blank cards between consecutive levels to express preference intensity. DOR, and its stochastic extension (SMAA-DOR), treat these answers as hard constraints defining a set of compatible value functions. We propose B-DOR, a probabilistic reformulation of DOR in which each pair of adjacent levels yields an ordinal observation, the declared direction and the number of cards, modelled through a cumulative-link likelihood that relates the number of blank cards to the latent value difference between alternatives. Two Bayesian inference algorithms are proposed: BAYES-DOR samples the whole posterior distribution by Hamiltonian Monte Carlo; FTRL-DOR tracks the maximum a posteriori estimate by constrained convex optimization. Moreover, through a multi-step elicitation process, elicitation can be spread over several short sessions reducing the cognitive burden on the DM. Both algorithms enjoy logarithmic regret bounds for prediction that hold for any sequence of DM responses and that guide the choice of the prior hyperparameters. A Monte Carlo study over 768 configurations shows that accuracy grows with the number of sessions, that blank cards add significant information over preference directions alone, that both algorithms maintain good performance under inconsistent answers, and that both outperform DOR and SMAA-DOR. An illustrative application to Italian regional healthcare performance demonstrates the practical applicability of the approach for building composite indicators.
[LG-203] Conformal Robustness in Prediction-Driven Decision-Making
链接: https://arxiv.org/abs/2609.23170
作者: Lingjie Zhao,Hansheng Jiang,Wei Qi
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Optimization and Control (math.OC)
*备注:
Abstract:Modern prediction-driven decision systems often rely on black-box predictors, but a point forecast alone does not provide the uncertainty scale required for robust downstream decision-making. We build a score-calibrated robustness framework that converts any fixed point predictor into a decision-relevant uncertainty representation through distribution-free conformal calibration. We use the conformal score, rather than a particular uncertainty set, as the primitive unit of robustness. The same score determines coverage-calibrated uncertainty sets for reliability-based robust optimization and normalizes target violations in a target-oriented formulation, Conformal Robust Satisficing. This formulation induces a conformal fragility measure that quantifies how rapidly performance deteriorates as the realized parameter departs from the forecast on the conformal score scale. For objective-uncertainty problems under standard convexity and duality conditions, we show that the reliability-based and target-oriented formulations parameterize the same score-calibrated robust decision frontier. This equivalence yields a data-driven mapping between reliability levels and acceptable targets and characterizes the marginal cost of robustness. Synthetic experiments validate the theoretical guarantees and illustrate the reliability-target correspondence. A real-data online-grocery case study demonstrates how the interface combines deep-learning demand forecasts with tractable inventory optimization, thereby improving reliability and reducing operational costs. Overall, our work shows that conformal scores endow fixed black-box predictors with an interpretable uncertainty scale for downstream decision-making while enabling reliability guarantees, acceptable-target selection, and fragility analysis within a unified framework.
[LG-204] Signal-Informed Temporal Routing for Vinyl Defect Regime Detection ICASSP2026
链接: https://arxiv.org/abs/2609.23164
作者: Yi-Hung Kan,Homayoon Beigi
类目: Audio and Speech Processing (eess.AS); Machine Learning (cs.LG)
*备注: 5 pages, 1 figure, 4 tables. Columbia University Nonlinear Control Lab Technical Report CUNLC-20260919-01. Submitted to ICASSP 2026
Abstract:Vinyl restoration systems must distinguish isolated clicks, short bursts, dense crackle, and overlapping damage before selecting a repair operation. We present a lightweight two-stage detector in which signal-informed sparse, burst, and dense experts produce complementary defect evidence, and a temporal backend converts that evidence into stable repair regimes. The backend factors the five-way decision hierarchically, applies a validation-only mixed regime gate, and decodes with validation-selected transition penalties that outperform a maximum-likelihood transition matrix on the same emissions. On a source-separated synthetic benchmark of 597 non-overlapping 15 s excerpts drawn from 21 recordings, the held-out system reaches 0.730 pooled five-class macro F1 (95% CI 0.694-0.761), an improvement of 0.043 over a flat frame-level router (paired bootstrap p=0.001). Clean and dense frames are highly reliable, while sparse, burst, and mixed frames remain far more ambiguous.
[LG-205] oscani-Fourier Distance on Probability Measures: Wasserstein Control Topological Equivalence on Model Classes and Duality
链接: https://arxiv.org/abs/2609.23163
作者: Mehrdad Mohammadi
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Functional Analysis (math.FA); Probability (math.PR); Statistics Theory (math.ST)
*备注:
Abstract:Comparing probability measures in machine learning trades transport geometry against computational cost: Wasserstein distances encode the geometry of \mathbb R^d but require solving a transport problem, while kernel discrepancies are cheap to evaluate yet depend delicately on their test class. We study the Toscani–Fourier family \mathrm T_s,p , the weighted L^p norm of the difference of two characteristic functions, as a continuous Fourier-side discrepancy on \mathbb R^d . For 1\le p\infty we show that d/ps1+d/p is exactly the window in which \mathrm T_s,p is finite on \mathcal P_p(\mathbb R^d) , both endpoints already failing for a pair of Dirac measures, and we establish the metric, embedding, and compactness structure of the resulting space, which we prove to be complete. Duality identifies \mathrm T_s,p as an integral probability metric over a homogeneous Fourier–Lebesgue ball, with an explicit extremizer when 1p\infty . We prove the global bound \mathrm T_s,p\lesssim W_p^,s-d/p , whose exponent is sharp, show that no global converse of any form can hold, and recover topological equivalence with W_p on bounded-support and uniform-tail classes, together with explicit reverse moduli on bounded-support classes that improve the imported energy-kernel exponent at p=2 . There, \mathrm T_s,2 is a constant multiple of the classical energy distance, which yields an exact finite-sample identity for the mean of the empirical discrepancy; the numerical experiments are otherwise diagnostic.
[LG-206] Real-Time Plasma State Prediction via FPGA-Accelerated Quantized Recurrent Probabilistic Neural Networks
链接: https://arxiv.org/abs/2609.23141
作者: Daniel Gaytan-Villarreal,Aiken Xie,Tu Pham,Rohit Sonker,Chiara Amendola,Matteo Cremonesi,Cong Hao,Jeff Schneider
类目: Plasma Physics (physics.plasm-ph); Machine Learning (cs.LG)
*备注:
Abstract:Real time plasma state estimation for control of Tokamak devices are challenging due to the stringent latency requirements of the plasma control system (PCS). We present an end-to-end workflow for deploying a recurrent probabilistic neural network (RPNN) on FPGA hardware. We combine architecture size reduction with quantization-aware training via QKeras. The model is then synthesized using hls4ml, targeting a Xilinx Alveo U50 device. We report a design that fits comfortably within all four resource budgets (DSP, LUT, FF, BRAM) at deterministic sub-10~ \mu s single-timestep latency, meeting the requirements for real-time inference inside a model-predictive-control-style plasma control loop.
[LG-207] he shape of quark flavors
链接: https://arxiv.org/abs/2609.22812
作者: Shinsuke Kawai,Nobuchika Okada
类目: High Energy Physics - Phenomenology (hep-ph); Machine Learning (cs.LG); High Energy Physics - Theory (hep-th)
*备注: 10 pages, 4 tables, 4 figures
Abstract:We construct the Yukawa couplings of the quark sector as overlap integrals of Gaussian wave functions in extra spatial dimensions. Assuming that the wave function of each chiral fermion is localised at a point in the extra dimensions, while that of the Higgs doublet is flat, the Yukawa matrix elements are determined by the relative positions of the three left-handed quark doublets, q_1 , q_2 , q_3 , and the six right-handed quark singlets, u_1 , u_2 , u_3 , d_1 , d_2 , d_3 . We employ numerical optimisation techniques from machine learning to identify field configurations that satisfy all current experimental constraints on the quark Yukawa sector. In this framework, the large hierarchy of quark masses arises naturally, since the overlap integrals of Gaussian wave functions are exponentially sensitive to \mathcal O(1) separations in the extra dimensions.
[LG-208] SPIBER: Reconstructing Free Energy Landscapes from Short Unconverged Trajectories with Generative Flow Networks
链接: https://arxiv.org/abs/2609.22663
作者: Venkata Sai Sreyas Adury(1),Pratyush Tiwary(2 and 3 and 4) ((1) Chemical Physics Program and Institute for Physical Science and Technology, University of Maryland, (2) Biophysics Program and Institute for Physical Science and Technology, University of Maryland, (3) Department of Chemistry and Biochemistry and Institute for Physical Science and Technology, University of Maryland, (4) University of Maryland Institute for Health Computing, Bethesda, USA)
类目: Chemical Physics (physics.chem-ph); Machine Learning (cs.LG); Data Analysis, Statistics and Probability (physics.data-an)
*备注: Journal-Style Article 25 pages (13 in main manuscript, 12 in supporting information) with 14 figures (6 in main manuscript, 8 in supporting information)
Abstract:Molecular systems have many degrees of freedom, but their metastable behavior can often be described by a few collective variables. Identifying these variables and estimating free energies along them from limited simulation data remains a challenging, important problem. Separate short trajectories may sample different metastable states without capturing transitions or establishing their relative equilibrium populations. For unbiased trajectories generated with the same Hamiltonian at a single temperature, alternate methods based on histogram reweighting cannot correct this imbalance. Here we present SPIBER, which combines the State Predictive Information Bottleneck (SPIB) with Generative Flow Networks (GFlowNets). SPIB uses deep learning to approximate slow degrees of freedom through a past-future information bottleneck, retaining information needed to predict future metastable states. We show that this compression limits conditional entropy variations in populated regions, allowing conditional mean potential energies, which are much easier to calculate, to be used to approximate free energy differences. Given sufficient local sampling to estimate these energies, they define the target distribution for GFlowNets, energy-based generative samplers that sample according to estimated thermodynamic stability rather than observed populations. For a particle in a radial double-well potential, for alanine dipeptide, and for the nine-residue peptide AIB9, SPIBER recovers free energy differences between sampled metastable states to within one thermal energy unit of reference values. The method combines collective-variable learning and free energy estimation in up to four latent dimensions, without requiring converged state populations or additional molecular dynamics simulations.
[LG-209] A Bayesian Vertical Federated Learning Framework for Multivariate Reduced-Rank High-Dimensional Regression
链接: https://arxiv.org/abs/2609.22654
作者: Brigham Halverson,Sharmistha Guha,Jessica Bernard,Rajarshi Guhaniyogi
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Computation (stat.CO)
*备注:
Abstract:Federated learning (FL) has emerged as a leading privacy-preserving framework for collaborative machine learning across decentralized environments. While considerable progress has been made in horizontal federated learning (HFL), where data with common features is distributed across sites, vertical federated learning (VFL), where sites share observations across distinct feature sets, remains less explored. Advancing Bayesian high-dimensional multivariate reduced-rank regression methods for VFL poses unique challenges: (a) stringent privacy regulations preventing local site data sharing, and (b) fitting local regressions overlooks essential modeling aspects like inter-variable correlations. In contrast HFL allows each site to fit a comparable model independently. We present a novel Bayesian VFL framework for multivariate high-dimensional reduced-rank regression, termed BayesVFLReg, which enables precise coefficient estimation while safeguarding both feature and response privacy. Participating sites use a shared random sketching matrix to compress local variables into privacy-preserving sketches. A central server collects these sketches where Bayesian multivariate reduced-rank regression uses Gaussian scale mixture priors. For feature selection, we introduce a single-step post-processing strategy based on mixture-model clustering of the absolute posterior coefficient means to distinguish signal from noise per response variable. BayesVFLReg is computationally scalable for large, high-dimensional datasets and facilitates efficient variable selection. Theoretically, we establish sharp non-asymptotic bounds on the posterior probability that the fitted density falls within a Hellinger ball centered at the true data-generating density. Comparative simulation studies and real-world data analyses show that BayesVFLReg reliably identifies sparse feature effects, even under feature correlation.
[LG-210] Locally Private Inference for Riemannian Stochastic Optimization
链接: https://arxiv.org/abs/2609.22642
作者: Xiaotian Chang,Yangdi Jiang,Qirui Hu
类目: Machine Learning (stat.ML); Machine Learning (cs.LG); Optimization and Control (math.OC); Statistics Theory (math.ST)
*备注:
Abstract:We develop inference for manifold-valued population minimizers when each observation belongs to a different participant and only locally private messages reach the analyst. The method releases randomized tangent gradients and combines them through Riemannian stochastic approximation and Polyak-Ruppert averaging. Directly inserting a private data surrogate into a nonlinear loss can shift its population target, whereas conditional centring of the released gradient preserves the first-order equation. We introduce symmetric-pair regression (SPR) to estimate the asymptotic variance from the same private messages used for point estimation, without holding out participants or requesting a second release. We prove the central limit theorem and consistency of the fully transcript-based sandwich covariance and intrinsic Wald region under local differential privacy. Simulations across various statistical problems and manifolds support the predicted decrease in estimation error and near-nominal coverage under moderate privacy. An application to NHANES anthropometric data illustrates private estimation of a leading body-size direction and its uncertainty.
[LG-211] Scalable Incremental Robustness Analysis of Neural Network Feedback Systems
链接: https://arxiv.org/abs/2609.22576
作者: Zichen Wang,Peter Seiler,Geir Dullerud,Bin Hu
类目: Optimization and Control (math.OC); Machine Learning (cs.LG); Systems and Control (eess.SY)
*备注:
Abstract:Semidefinite programming (SDP) certificates for feedback systems containing deep neural networks (NNs) typically scale with the total number of neurons, whereas small-gain tests are scalable but can be highly conservative. This paper develops a unified and scalable framework for incremental robust stability and performance analysis of feedback interconnections involving high-dimensional NNs and unmodeled dynamics. By combining a structured decomposition of the full-order SDP condition with scalable Lipschitz constant estimation algorithms, we derive reduced verification conditions that certify incremental convergence and incremental \ell_2 -gain bounds. The dimensions of the resulting control-analysis linear matrix inequalities (LMIs) depend only on the widths of the last two network layers and are \textitindependent of network depth. The framework preserves the coupling between the plant and the NN, with the incremental small-gain condition recovered as a special case. To further reduce conservatism, we develop a multi-round alternating update scheme that iteratively refines the coupling variables while preserving scalability. Numerical experiments show that the proposed framework achieves state-of-the-art incremental \ell_2 -gain bounds for large-scale NN feedback systems.
[LG-212] RLVR is a Kernel Not a Function: Statistical Inference for pass@k Crossovers
链接: https://arxiv.org/abs/2609.22547
作者: Chen Yang,Xianyang Zhang,Jun Chen
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注: 37 pages, 5 figures
Abstract:Reinforcement learning with verifiable rewards (RLVR) often improves pass@1 while falling behind its base model at larger sampling budgets k , a crossover read as evidence that RLVR only sharpens existing capability. We identify two limits to this reading. First, a visible crossing need not be statistically established: comparing models on the same prompts, we build confidence bands across sampling budgets k that require evidence of both an early gain and a later loss. Across five public RLVR pairs no crossing is statistically established in the initial evaluations, while a 32k-token evaluation on fresh prompts locates a reversal with first loss between 11 and 61 samples; power analysis shows why failure to detect a crossing need not mean no crossing, and why more prompts can help more than more answers per prompt. Second, base success alone does not determine what RLVR does to a prompt: prompts with the same base success rate have different post-RL success rates, and these differences repeat across independent generation halves. The relationship is a conditional distribution—a Markov kernel—rather than a single curve, and fitting it predicts crossings in independent generations for the same prompts and corrects the simple model’s power estimates. Theory further shows how losses on a minority of the hardest prompts can overturn an early lead even when training improves other prompts, separating evidence that a crossover exists from claims about what it means for capability.
[LG-213] Machine Learning for Invisible Dark Boson Searches at the Electron-Ion Collider
链接: https://arxiv.org/abs/2609.22484
作者: Rojae Mighty,Ankush Reddy Kanuganti
类目: High Energy Physics - Phenomenology (hep-ph); Machine Learning (cs.LG); Nuclear Experiment (nucl-ex)
*备注: 12 pages, 10 figures, 3 tables. Generator-level comparison of BDTs and optimized rectangular cuts for invisible dark bosons at the EIC
Abstract:We investigate whether adding t , the positive magnitude of the squared nuclear four-momentum transfer, enables boosted decision trees (BDTs) to improve invisible-dark-boson selection relative to optimized rectangular cuts at the Electron-Ion Collider. We model coherent exclusive scalar and vector production at generator level in electron-gold collisions at 18 GeV by 100 GeV per nucleon. Both methods use identical weighted samples, inputs, preselection, and optimization objectives. Using only electron information, the BDT provided no consistent advantage over optimized cuts for signal selection across 11 masses for each boson type. When both methods also use t , the BDT distinguishes signal from background slightly better than optimized cuts at 10 GeV for both boson types. These results motivate further investigation of machine learning in EIC dark-boson searches through exclusive processes where t can be reconstructed.
[LG-214] A Hybrid Quantum Neural Network to Analyse Big Experimental Powder X-ray Diffraction Data
链接: https://arxiv.org/abs/2609.22376
作者: H. Dong,S.D.M. Jacques,M.Q. Hlatshwayo,E. Papoutsellis,K. Georgopoulos,A. M. Beale,A. Vamvakeros
类目: Materials Science (cond-mat.mtrl-sci); Machine Learning (cs.LG)
*备注:
Abstract:Quantitative analysis of experimental powder X-ray diffraction data remains challenging when evaluating complex multiphase materials and noisy measurements. We introduce a hybrid quantum neural network framework designed to extract quantitative parameters, such as phase weight fractions and scale factors, directly from one-dimensional powder diffraction patterns without iterative refinement. The model combines noise-aware classical simulator pre-training with fast downstream fine-tuning on quantum processing unit features, ensuring stability against hardware decoherence. We demonstrate the practical utility of this approach by deploying the trained network onto an IBM quantum computer to analyse experimental X-ray diffraction computed tomography datasets from a three-phase solid oxide fuel cell containing ca. 10,000 patterns and a four-phase lithium-ion battery containing ca. 20,000 patterns). The network successfully reconstructs quantitative spatial phase maps in strong agreement with classical Rietveld refinement, paving the way for using quantum computing hardware to analyse real-world materials characterisation data.
[LG-215] Gradient-estimator design overcomes trainability barriers in neural-network-based variational optimization
链接: https://arxiv.org/abs/2609.22342
作者: Yi-Ran Xue,Rui Wang,Baigeng Wang,Chenan Wei
类目: Disordered Systems and Neural Networks (cond-mat.dis-nn); Strongly Correlated Electrons (cond-mat.str-el); Machine Learning (cs.LG); Chemical Physics (physics.chem-ph); Computational Physics (physics.comp-ph)
*备注: 10 pages, 4 figures; partially supersedes arXiv:2606.13912
Abstract:Neural networks provide expressive representations for scientific computing. However, even sufficiently expressive networks can suffer training failure in weak-gradient regimes, limiting their practical use in quantum many-body physics and ab initio quantum chemistry. Here we derive an unbiased direct gradient estimator and introduce the adaptive minimum-variance phase (AMVP) estimator for neural-network variational optimization. By improving the signal-to-noise ratio of weak gradients, these methods enable reliable scientific calculations where training previously failed, while substantially reducing computational cost. The framework enables compact networks to outperform larger and fine-tuned default standard-estimator models with over an order of magnitude less GPU time on correlated flux models, and ultimately exceed the density matrix renormalization group (DMRG) accuracy. It further achieves chemical accuracy in N _2 bond breaking and, for the first time, in heavy-element I _2 with explicit spin-orbit coupling. These results demonstrate that gradient-estimator design expands the capabilities of neural-network variational methods for accurate scientific computing.
[LG-216] When and Why Do Linear Bias Probes Fail? A Geometric and Statistical Theory of Bias Detectability in Large Language Model Representations
链接: https://arxiv.org/abs/2609.22337
作者: Mo Hai,Haifeng Li
类目: Machine Learning (stat.ML); Machine Learning (cs.LG)
*备注:
Abstract:Linear probing is the standard instrument for detecting social biases in the hidden representations of large language models. Yet reported probe accuracies come almost exclusively from \emphcounterfactual evaluations in which every input carries an explicit demographic marker. Once only a fraction \alpha of inputs carries demographic information, performance degrades sharply, and a weak probe may reflect either an unbiased model or an underpowered detector. We develop a theory that resolves this ambiguity. Modeling representations as two class-conditional clusters with Mahalanobis separation s on a manifold of curvature \kap , we prove: (i) a finite-sample generalization bound governed by the manifold’s extrinsic radius with a matching \smash\sqrt\dB/n minimax lower bound; (ii) an exact purity law for the maximum linear-probe AUC, strictly increasing in \alpha ; (iii) a curvature ceiling: ambient chordal separation on a space form cannot exceed 2/\sqrt\kap ; and (iv) a detectability threshold below which no audit can distinguish probe output from chance. Every theorem is validated on synthetic manifolds with known ground truth and on six open-weight models \times four bias dimensions, where the purity law predicts entire AUC-- \alpha curves from a single cross-fitted \hat s measured at \alpha=1 , with no parameters fitted to those curves. The framework turns bias auditing into a power analysis: given a target purity and effect size, it prescribes the sample budget n(\alpha) for a conclusive audit.
[LG-217] Machine Learning for Underwater Optical Wireless Communication Systems: A Comprehensive Survey
链接: https://arxiv.org/abs/2609.22294
作者: Shaymaa Mahmoud,Ardimas Purwita,Mohamed-Slim Alouini
类目: ignal Processing (eess.SP); Machine Learning (cs.LG)
*备注:
Abstract:Underwater Optical Wireless Communication (UOWC) has emerged as a promising technology for high-speed underwater data transmission, offering significantly higher bandwidth and lower latency compared to acoustic and Radio Frequency (RF) technologies. However, the underwater medium introduces severe impairments that degrade link performance and limit communication range. The growing complexity of these challenges has increased research interest in Machine Learning (ML) and Deep Learning (DL) approaches, which offer powerful tools for channel modeling, signal processing, and system adaptation in ways that conventional analytical methods struggle to achieve. This survey provides a comprehensive review of ML methods applied across the full UOWC system pipeline, covering channel modeling, transmitter design, receiver detection and equalization, link alignment, and emerging applications, including wireless power transfer, semantic communication, object detection, optical sensing, and localization. Finally, we discuss open challenges and future research directions to motivate further work in this rapidly growing field.
[LG-218] Physics-Informed Classical and Quantum Neural Networks for One-Dimensional Schrodinger Eigenvalue Problems
链接: https://arxiv.org/abs/2609.22189
作者: Tariq Mahmood,Waqas Arshad,Bilal Naseer,Alfredo Raya
类目: Quantum Physics (quant-ph); Machine Learning (cs.LG); High Energy Physics - Phenomenology (hep-ph); High Energy Physics - Theory (hep-th)
*备注:
Abstract:The Schrodinger equation in one spatial dimension admits a small set of exactly solvable potentials that serve as natural proving grounds for any new eigenvalue solver. We formulate Physics-Informed Neural Networks (PINNs) and Physics-Informed Quantum Neural Networks (PIQNNs) for the time-independent Schrodinger equation and apply them to three of these benchmarks: the harmonic oscillator, the infinite square well, and the finite square well. In each case a composite loss encodes the differential-equation residual, the normalization condition, the boundary behavior, and the orthogonality between eigenstates, so that the trial wave function is driven toward a genuine eigenfunction without supervised data. The eigenvalues and wave functions returned by both methods are compared against the exact spectra and against three classical references: the matrix Numerov method, the finite difference method, and the shooting method. For the smooth oscillator the two neural solvers reproduce the lowest four eigenvalues to parts per million, while for the square wells they recover the analytic levels with comparable fidelity even where the potential is discontinuous. The quantum circuit, built as a layered angle-embedding ansatz with strongly entangling blocks, converges more reliably than its classical counterpart on the higher excited states, where the loss landscape of the classical network becomes harder to navigate.
[LG-219] Interpretable Stress Detection from ECG Signals Using Motif-Based Anomaly Analysis
链接: https://arxiv.org/abs/2609.22179
作者: Zhanna Balyan,Sachin Kumar
类目: ignal Processing (eess.SP); Machine Learning (cs.LG)
*备注:
Abstract:Stress detection using physiological signals has gained significant attention due to its impact on both physical and mental health. While existing approaches based on machine learning and deep learning achieve strong predictive performance, they often rely on black-box models and fail to capture individual variability in physiological responses. In this work, we propose an interpretable and personalized framework for stress detection using electrocardiogram (ECG) signals based on motif discovery and Matrix Profile analysis. Instead of training a classifier, the method learns subject-specific baseline cardiac behavior by extracting recurring heartbeat patterns (motifs) from ECG signals. Stress is then detected as a deviation from these baseline patterns using a distance-based anomaly score. Experiments are conducted on the WESAD dataset using a carefully designed train-validation-test protocol to ensure reliable evaluation. The results show that the proposed approach can effectively detect stress for several subjects while providing clear interpretability through direct comparison of ECG patterns. However, the performance varies across individuals due to differences in physiological responses, with some subjects exhibiting minimal morphological changes under stress. Additional analysis incorporating heart rate variability (HRV) features reveals that while HRV can improve performance in certain cases, its contribution is not consistent across all subjects. These findings highlight the importance of personalization and interpretability in physiological stress detection and demonstrate that motif-based approaches provide a meaningful alternative to black-box models, while also revealing inherent limitations due to inter-subject variability.
[LG-220] Stagewise Anomaly Detection for E-Transaxle Quality Monitoring Using Wavelet and STFT Features
链接: https://arxiv.org/abs/2609.22172
作者: Mohammad N. Bisheh,Rajesh Gupta,Qian Wang,Mohammad Babakmehr,Colin Brady,Parinaz Farajiparvar,Saurabh Singh,Kamran Payanabar
类目: ignal Processing (eess.SP); Machine Learning (cs.LG); Other Statistics (stat.OT)
*备注:
Abstract:This paper presents two interpretable machine-learning frameworks for quality screening of e-transaxle assemblies in electric vehicles: a Stagewise Wavelet Isolation Forest (SWIF) framework and a short-time Fourier transform (STFT)-based diagnostic framework. High-dimensional vibration signals acquired from front and back accelerometers are analyzed across multiple operating stages to capture stage-dependent vibration behavior. In the SWIF framework, signals are decomposed using a five-level Daubechies-4 discrete wavelet transform, and blockwise mean-squared coefficients are extracted from the selected wavelet detail level to obtain compact multiscale features. In the STFT-based framework, dominant-frequency trends are extracted from time-frequency representations and summarized through regression coefficients with respect to instantaneous motor speed. Anomaly detection models are trained using accepted production units under the assumption that only a small fraction of accepted assemblies contain latent defects, and their performance is evaluated using road-tested units with validated quality outcomes. Experiments on production and road-tested e-transaxle units show that both approaches provide interpretable diagnostic information, while SWIF achieves the most favorable balance between defect detection and false-positive control. Compared with the STFT-based method and alternative anomaly detectors, SWIF combined with Isolation Forest yields lower anomaly rates within the Accept population while identifying high-risk units from the Reject population. The stagewise structure further localizes anomalous behavior to specific operating conditions, supporting root-cause analysis and targeted process improvement.
[LG-221] Attention-Enhanced Dual-Branch ConvNeXt-BiLSTM Network for Subject-Independent EEG Seizure Detection
链接: https://arxiv.org/abs/2609.22141
作者: Maimuna Chowdhury,Sk. Imran Hossain
类目: ignal Processing (eess.SP); Machine Learning (cs.LG)
*备注: Submitted to IEEE COMPAS 2026. 5 pages, 2 figures, 5 tables
Abstract:Automated seizure detection from scalp electroencephalography (EEG) is difficult because seizure morphology varies among patients and seizure samples are substantially outnumbered by non-seizure samples. This paper presents an attention-enhanced dual-branch network that jointly learns time–frequency and temporal representations from the same EEG segment. A continuous wavelet transform converts each segment into a scalogram processed by an ImageNet-pretrained ConvNeXt-Tiny backbone and squeeze-and-excitation attention. In parallel, a bidirectional long short-term memory network followed by multi-head self-attention models the raw signal. The two feature vectors are concatenated and classified by a weighted multilayer perceptron. Experiments use 14 subjects from the CHB-MIT scalp EEG database with subject-wise partitioning performed before overlapping segmentation. The model obtains 97.88% accuracy and 97.51% F1-score over ten across-subject splits, and 97.51% accuracy, 96.59% F1-score, and 98.02% area under the ROC curve under 14-fold leave-one-subject-out validation. Removing temporal attention causes the largest ablation loss. The model requires 28.26 million parameters and 4.56 GFLOPs, with a measured network-only inference latency of 4.64 ms per segment.
[LG-222] HFEMCNet: A Compact Hybrid Frequency Enriched Multi Channel Network for Automatic Modulation Classification
链接: https://arxiv.org/abs/2609.22139
作者: Qamar Ijaz,Nayyer Aafaq
类目: ignal Processing (eess.SP); Machine Learning (cs.LG)
*备注: 13 pages, 13 figures, 6 tables
Abstract:Automatic modulation classification (AMC) of received radio signals is prudent for further signal processing tasks such as communication monitoring, cognitive radio operation, and interference mitigation in the electromagnetic spectrum. Traditional methods often rely on handcrafted features and struggle under complex channel conditions, whereas deep learning (DL) architectures can learn discriminative representations directly from raw signals. In this work, we propose HFEMCNet, a novel compact hybrid frequency-enriched multi-channel network that jointly exploits spatiotemporal dependencies and frequency domain information derived via Fast Fourier Transform (FFT). HFEMCNet integrates raw IQ samples with hierarchical spectral features to produce a rich signal representation, which is processed by convolutional layers for spatial feature extraction and Long Short-Term Memory (LSTM) units for temporal modeling. Experimental evaluation on benchmark datasets RML2016.10a, RML2016.10b, and over-the-air RML2018.01a demonstrates that HFEMCNet significantly outperform contemporary state-of-theart DL models in classification accuracy while achieving a reduced parameter count, smaller memory footprint, and lower tail-latency, making it suitable for real-time deployment on resource-constrained platforms. These results highlight the effectiveness of combining hybrid architectures with frequencydomain enrichment for robust and efficient AMC.
[LG-223] Experimental Evaluation of a Low-Power Ultra-Wideband Receiver for Spectrum Sensing CEC
链接: https://arxiv.org/abs/2609.22134
作者: Panagiotis Vlachos,Ioannis A. Bartsiokas,Cedric Dehos,Francois Rivet,George Karachalios
类目: ignal Processing (eess.SP); Information Theory (cs.IT); Machine Learning (cs.LG)
*备注: 4 pages, 5 figures, accepted in 33rd IEEE International Conference on Electronics Circuits and Systems (ICECS 2026)
Abstract:Walsh-sequence-based receiver architectures offer an alternative approach for the reception and reconstruction of multiple simultaneous RF signals over wide bandwidths. While previous works have focused on the architecture and theoretical operation of Walsh-domain processing, limited experimental results have been reported so far on hardware operation under realistic conditions. This paper presents the experimental evaluation of a Walsh-sequence-based receiver prototype. A measurement campaign was conducted under various signal placement, interference, and receiver configuration scenarios. Reconstructed spectra, Walsh-lane power spectral densities (PSDs), and Error Vector Magnitude (EVM) measurements were used to investigate Walsh-domain signal representation, interference effects, and receiver performance. Finally, dense- spectrum experiments involving up to ten simultaneously active signals demonstrate the potential of the architecture for low power wideband spectrum sensing applications.
[LG-224] Comparative Analysis of State-of-the-Art Foundation Models for Sleep Analysis Under Channel Reduction
链接: https://arxiv.org/abs/2609.22105
作者: Hassan Mehdi,Riku Klen,Ayse Kosal Bulbul,Suzanne Timmons,Abdulhamit Subasi,Wei Chen,Zou Zhu,Muhammad Irfan
类目: ignal Processing (eess.SP); Machine Learning (cs.LG)
*备注:
Abstract:Automatic sleep staging from polysomnography (PSG) is a well-studied task, but PSG itself is expensive, clinic-based, and burdensome to manually score, which limits its use for long-term or at-home monitoring. Most existing sleep-staging foundation models are evaluated using the full PSG montage. We instead ask how much of that montage is actually necessary. We evaluate six sleep staging models on the Multi-Ethnic Study of Atherosclerosis (MESA) PSG dataset across three signal conditions: electroencephalography (EEG), electrocardiography (ECG), and their combination (EEG+ECG). This is motivated by edge-cloud deployment, where EEG requires a clinic-grade scalp electrode, whereas ECG is already captured by consumer wearables. We test state-of-the-art foundation models such as SleepFM with an encoder trained from scratch on MESA, alongside BIOT, MOMENT, LaBraM, a base-scale Vision Transformer (ViT-B) reimplementation of SensorLM trained from scratch, and YASA, spanning EEG-pretrained, general-time-series, from-scratch, and classical non-learned approaches. No model architecture is modified from its original form; SensorLM’s encoder is reimplemented only in PyTorch. For EEG-only staging, BIOT achieves the best result with a macro~F1 of 0.7237, followed by LaBraM (0.6835) and SleepFM from scratch (0.6582). Across the five models capable of ECG-only staging, switching from EEG to ECG costs between 0.2798 (MOMENT) and 0.4151 (BIOT) macro~F1, averaging 0.3531, while cutting the raw channel data rate to a third. Adding ECG to EEG provides no gain for most models. These results show that EEG carries most of the sleep-staging signal, quantify the consistent accuracy cost of the wearable-compatible alternative, and demonstrate that sleep-relevant pretraining transfers well to MESA.
[LG-225] Graph Learning for Cross-Subject Cross-Population EEG Emotion Decoding and Model-Derived Spatial-Spectral Neural Signatures
链接: https://arxiv.org/abs/2609.22103
作者: Dongyi He,Bin Jiang,Xiangkai Wang,Yun Zhao,Hongjie Yan,Wai Ting Siok,Nizhuan Wang
类目: ignal Processing (eess.SP); Machine Learning (cs.LG)
*备注:
Abstract:Electroencephalography (EEG) provides a noninvasive means of capturing emotion-related neural dynamics, yet reliable EEG emotion decoding lacks models that can both generalize to unseen individuals and populations while preserving neural interpretability. To address these challenges, EmoDiPyraTrans is proposed as a development-regularized differential graph Transformer that models temporally ordered relative power spectral density graphs through adaptive graph recurrence, differential attention, and multiscale fusion. The framework was evaluated at three connected levels. First, cross-subject evaluations on SEED, FACED, MAHNOB-HCI, DEAP and DREAMER yielded participant-mean accuracies of 0.928, 0.645, 0.714, 0.617 and 0.671, respectively; the model ranked first among the evaluated methods for accuracy and positive-class F1 on all five datasets. Across seven ablation protocols, differential attention was the only component whose removal reduced both metrics in every case, whereas removing maximum mean discrepancy reduced accuracy throughout. Second, DEP-EEG distinguished within- from cross-population positive-versus-neutral decoding. Accuracy was 0.802 within healthy controls, 0.704 within participants with depression and 0.591 under healthy-to-depression transfer. Mixed-population development produced 0.581 accuracy and the highest positive-class F1 (0.498), indicating that greater population diversity alone did not remove the transfer gap. Third, channel- and frequency-resolved analyses on SEED identified a distributed frontal, temporal, central and parietal pattern, an alpha-centred low-to-mid-frequency preference and a six-channel subset that preserved near-full performance.
[LG-226] Beyond the Raw Waveform: Fusing Visual Representations of EDA for Stress Detection
链接: https://arxiv.org/abs/2609.22095
作者: Stefanos Gkikas,Thomas Kassiotis,Yang Guo,Guangliang Li,Eric Nichols,Houshyar Asadi,Nikolaos Smyrnis,Giorgos Giannakakis
类目: ignal Processing (eess.SP); Machine Learning (cs.LG); Sound (cs.SD); Audio and Speech Processing (eess.AS)
*备注: The paper has been accepted at: IEEE | 2026 9th International Conference on Pattern Recognition and Artificial Intelligence (PRAI 2026)
Abstract:Electrodermal activity (EDA) is widely used in automatic stress detection, yet most pipelines treat it only as a raw one-dimensional waveform. This study examines whether complementary visual representations of EDA provide useful information for stress classification and whether their fusion im- proves recognition performance. Six image-based representations are derived from each EDA recording: an unwrapped short-time Fourier transform (STFT) phase spectrogram, an instantaneous-frequency map computed from that phase, a power spectral density (PSD) spectrogram, a continuous wavelet transform scalogram, a recurrence plot, and a rendered waveform trace. The selected representations are stacked as channels of a single multichannel input, together with the raw waveform, and processed by a shared asymmetric-attention architecture. Experiments on a 58-subject stress dataset show that representation fusion improves over the raw waveform. The best configuration, which combines five representations while excluding the unwrapped phase spectrogram, reaches 70.97% test accuracy, compared with 67.36% for the raw waveform. The single PSD spectrogram achieves 69.44%, remaining close to the best-fused configuration at a lower computational cost. The results show that alternative visual forms of the same EDA signal can provide useful inductive biases for stress detection, and that a compact selection of complementary representations can be more effective than the raw waveform alone.
[LG-227] Leakage-Safe Empirical Benchmarking of EEG-Based Machine Learning Pipelines for Dementia Classification
链接: https://arxiv.org/abs/2609.22092
作者: Haitian Wang,Chamara Madarasingha,Redowan Mahmud,Aneesh Krishna,Ryu Takechi
类目: ignal Processing (eess.SP); Machine Learning (cs.LG)
*备注:
Abstract:Electroencephalography (EEG) is a low-cost and non-invasive signal source for dementia screening, yet existing EEG-based studies remain difficult to compare because preprocessing, EEG segmentation, feature design, classifier choice, and validation protocols vary across studies and are often evaluated in isolation. This variability limits the derivation of robust pipeline recommendations. This paper presents a leakage-safe empirical benchmark for resting-state EEG-based dementia classification and uses it to identify a practical best-practice pipeline. Using the public OpenNeuro ds004504 dataset, the benchmark evaluates artifact correction, fixed-length EEG segmentation, training-only augmentation, multi-domain feature extraction, fold-internal feature selection, classical machine-learning (ML) classifiers, subject-level aggregation, and interpretation under leave-one-subject-out (LOSO) validation. The best-performing pipeline in this benchmark combines Artifact Subspace Reconstruction (ASR) followed by Independent Component Analysis (ICA), 10 s EEG epochs with training-only amplitude scaling and Gaussian-noise augmentation, spectral, complexity, and pairwise connectivity features, mutual-information top-100 selection, linear support vector machine classification, and mean-probability aggregation. It achieves 87.69 percent accuracy, 88.89 percent F1-score, and 91.20 percent AUC on Alzheimer disease (AD) versus cognitively normal controls (CN) classification while retaining 100 features from 1596 raw descriptors. The results support compact multi-domain EEG features as an interpretable and leakage-safe baseline for subject-level dementia classification.
[LG-228] When Is Availability-Aware Training Worth It? A Benchmark and Empirical Study of Interruption-Resilient Optimization Under Predictable Compute Schedules
链接: https://arxiv.org/abs/2609.22087
作者: Subhadip Mitra
类目: ignal Processing (eess.SP); Distributed, Parallel, and Cluster Computing (cs.DC); Machine Learning (cs.LG)
*备注: 11 pages, 3 figures, 6 tables
Abstract:Training under non-stationary but predictable compute availability (satellites under eclipse, duty-cycled edge devices, power-capped datacenters) is often framed as needing specialized, availability-aware optimizers. We test that premise. We release OrbitTrace, a benchmark of 50 physics-grounded availability traces from SGP4 propagation of live two-line element sets across three orbital regimes, and ask a falsifiable question: when an availability gap interrupts training, is a specialized resumption strategy worth it, or is competent checkpoint-and-resume enough? Our central finding: when optimizer state can be preserved across a gap, the gap is essentially free. A strong checkpoint baseline that restores full optimizer state and indexes its learning-rate schedule in effective (active) time matches uninterrupted training to within data-ordering noise on CIFAR-10/ResNet-18, and exactly on a GPT-2/AdamW task. Advantages previously reported for availability-aware methods, including our own three-pillar method AAT, arise almost entirely from comparison against a weak baseline that indexes its schedule on wall-clock time. Against the strong baseline, reactive adaptation provides no advantage across stationary, optimizer-state-loss, and distribution-drift regimes. We isolate one narrow regime where it helps: for large models whose optimizer state cannot be persisted across gaps and that are interrupted by frequent, short pauses, reconstructing a decayed optimizer moment recovers only ~21% of the state-loss penalty on average (and not robustly across seeds); this vanishes for eclipse-scale gaps, where the decayed moment is indistinguishable from zero. Our contributions are a benchmark, a strong reproducible baseline protocol, and a clear characterization of when interruption-resilient optimization is worth its complexity, and when it is not.
附件下载
