arXiv Daily Digest

248

Papers

HuRo: Robotizing Human Videos for Scalable VLA Pretraining

HuRo：将人类视频机器人化以实现可扩展的视觉-语言-动作预训练

Jeong, Jinho, Joo, Se June, Kang, Jaehyun, Kim, Dongyun, Kim, Yena, Kim, Hanjung, Kim, Seon Joo

Abstract

Human video datasets have emerged as a compelling alternative to expensive real-robot data, offering rich diversity at scale. To bridge the human-to-robot embodiment gap, existing approaches either robotize videos in task-matched settings or address observation and action alignment separately at scale. In this work, we systematically examine whether robotized human videos can provide effective and scalable supervision for pretraining vision-language-action (VLA) policies. To this end, we develop a robotization pipeline that converts heterogeneous human videos into robot-aligned observations and action trajectories while inferring missing intermediate signals across annotation levels. Using this pipeline, we construct the HuRo dataset, comprising about 630K robotized episodes and 142M processed frames from five human-video sources. Across four real-world manipulation tasks, increasing robotized pretraining scale improves overall completion from 51.5% to 80.3% and OOD completion under spatial and visual shifts from 34.9% to 72.2%. Ablations further show that visual robotization improves OOD robustness and that end-to-end pretraining with retargeted actions outperforms visual-only transfer. Code and data are released on our website: https://3587jjh.github.io/HuRo.

Chinese Translation

人类视频数据集已成为一种引人注目的替代方案，取代昂贵的真实机器人数据，提供丰富的多样性和规模。为了弥合人类与机器人之间的体现差距，现有方法要么在任务匹配的环境中对视频进行机器人化，要么在大规模下分别处理观察与动作的对齐。在本研究中，我们系统地考察了机器人化的人类视频是否能为视觉-语言-动作（VLA）策略的预训练提供有效且可扩展的监督。为此，我们开发了一种机器人化管道，将异构人类视频转换为与机器人对齐的观察和动作轨迹，同时推断跨注释级别的缺失中间信号。利用该管道，我们构建了HuRo数据集，包含约630K个机器人化的情节和来自五个来源的1.42亿个处理帧。在四个真实世界的操作任务中，增加机器人化预训练的规模使整体完成率从51.5%提高到80.3%，在空间和视觉变化下的OOD（分布外）完成率从34.9%提高到72.2%。消融实验进一步表明，视觉机器人化提高了OOD的鲁棒性，而使用重新定位动作的端到端预训练优于仅使用视觉的迁移。代码和数据已在我们的网站上发布：https://3587jjh.github.io/HuRo。

View on arXiv Download PDF AI Translation

cs.RO / 2 / 2609.10726

When Information is Worth the Risk: Behavioral Valuation for Hazardous Robotic Exploration

当信息值得冒险时：危险机器人探索的行为估值

Srivastava, Alkesh K., Suresh, Aamodh, Nieto-Granda, Carlos, Dames, Philip

Abstract

Hazardous robotic exploration requires robots to map spatial risks, such as unsafe terrain, radiation, fire, mines, or structural damage, while operating where collecting information can itself cause failure. A highly informative path may expose the robot to hazards, terminate execution, and prevent future observations. Hazardous exploration therefore requires deciding not only where uncertainty is largest, but when reducing it is worth the risk. This paper introduces a valuation-layer view of this problem. We keep the belief update, sensor model, physical risk model, and finite-horizon informative planner fixed, and change only the scalar objective used to rank feasible paths. Within this framework, we introduce a risk-augmented Behavioral Information objective based on Prelec probability weighting, yielding an interpretable family of conservative-to-aggressive information-risk valuations. Theoretically, we show that valuation parameters create switching boundaries between high-information/high-risk and lower-information/lower-risk paths, and induce a transformed Pareto-frontier structure over feasible exploration policies. Large-scale failure-truncated grid-world experiments show that valuation alone reshapes the information-risk frontier. Shannon information planning remains a strong raw-information baseline, while risk-aware objectives can reduce hazard exposure and robot losses by avoiding failures that truncate future sensing. Risk-augmented Behavioral valuation is Pareto-competitive with standard risk-aware baselines and provides interpretable conservative and intermediate regimes. These results support a framework in which robots reason not only about how much uncertainty an action reduces, but whether that reduction is worth the risk required to obtain it.

Chinese Translation

危险的机器人探索要求机器人在操作过程中绘制空间风险图，例如不安全的地形、辐射、火灾、地雷或结构损坏，同时收集信息本身可能导致失败。一个高度信息化的路径可能会使机器人暴露于危险之中，终止执行，并阻止未来的观察。因此，危险探索不仅需要决定不确定性最大的地方，还需要判断何时减少不确定性是值得冒险的。本文引入了这一问题的估值层视角。我们保持信念更新、传感器模型、物理风险模型和有限时间范围内的信息规划器不变，仅改变用于对可行路径进行排名的标量目标。在这一框架内，我们引入了一种基于Prelec概率加权的风险增强行为信息目标，产生了一系列可解释的保守到激进的信息-风险估值。从理论上讲，我们展示了估值参数在高信息/高风险和低信息/低风险路径之间创建了切换边界，并在可行探索策略上诱导出转化的帕累托前沿结构。大规模的失败截断网格世界实验表明，仅仅通过估值就能重塑信息-风险前沿。香农信息规划仍然是一个强大的原始信息基准，而风险意识目标可以通过避免导致未来感知中断的失败来减少危险暴露和机器人损失。风险增强行为估值在与标准风险意识基准的比较中具有帕累托竞争性，并提供了可解释的保守和中间状态。这些结果支持了一个框架，使机器人不仅考虑一个行动减少了多少不确定性，还考虑这种减少是否值得为获得它所需的风险。

View on arXiv Download PDF AI Translation

cs.RO / 3 / 2609.10748

Lie-Algebraic Bell Recurrences for Arbitrary-Order Twist Jets and Parallel-Mechanism Closure

任意阶扭转喷流的李代数贝尔递归与并联机构闭合

Condurache, Daniel

Abstract

This paper develops an arbitrary-order kinematic construction that links serial propagation, parallel-mechanism closure, and rigid-platform point fields within one dual screw framework. A cylindrical joint is retained as one native physical block, with revolute and prismatic joints obtained as special cases. For each fixed joint axis, ordinary Bell polynomials organize the derivatives of the exponential factor; across a chain, the noncommuting factors remain in their physical order. Initial-frame prefix and terminal-resolved covariant formulas then produce equivalent representations of the serial twist jet. For a parallel mechanism, repeated Leibniz differentiation, with joint-level derivatives organized by Bell polynomials, yields an arbitrary-order triangular active-passive closure recurrence: the same passive Jacobian is solved at every derivative order at a regular configuration, while the right-hand side contains only prescribed active data and lower-order jets. The resulting platform twist jet is mapped exactly to the point-independent affine invariants of the velocity, acceleration, jerk, and snap fields. The validation is deliberately complementary: a generic 3C chain with noncoplanar axes and nonzero rotational and translational cylindrical coordinates tests ordered serial propagation, an RR+RRR spherical wrist tests active-passive closure, and a Hunt-type 6-RUS mechanism with six active revolute joints tests an independently reconstructed platform jet and its affine fields. Independent differentiation of the rigid motion, evaluation of the affine fields, and the differentiated branch closures all agree through fourth order with residuals below $10^{-12}$ in the corresponding SI units. The formulation is purely kinematic and applies at configurations where the selected active-passive partition is regular.

Chinese Translation

本文发展了一种任意阶运动学构造，将串行传播、并联机构闭合和刚性平台点场在一个双螺旋框架内联系起来。保留一个圆柱关节作为一个原生物理块，旋转关节和滑动关节作为特例获得。对于每个固定的关节轴，普通贝尔多项式组织指数因子的导数；在链条中，非交换因子保持其物理顺序。初始框架前缀和终端解析协变公式随后生成串行扭转喷流的等效表示。对于并联机构，重复的莱布尼茨微分，结合贝尔多项式组织的关节级导数，产生任意阶三角主动-被动闭合递归：在常规配置下，每个导数阶都求解相同的被动雅可比，而右侧仅包含规定的主动数据和低阶喷流。得到的平台扭转喷流精确映射到速度、加速度、颤动和冲击场的点无关仿射不变量。验证过程故意互补：一个具有非共面的轴和非零旋转与平移圆柱坐标的通用3C链测试有序的串行传播，一个RR+RRR球形腕测试主动-被动闭合，而一个具有六个主动旋转关节的亨特型6-RUS机构测试独立重构的平台喷流及其仿射场。刚性运动的独立微分、仿射场的评估以及微分分支闭合在四阶以下均一致，残差在相应的国际单位制下低于$10^{-12}$。该公式纯粹是运动学的，适用于所选主动-被动分区为常规的配置。

View on arXiv Download PDF AI Translation

cs.RO / 4 / 2609.10844

Expressive Robotic Pianist: Mastering Complex Piano Repertoire with Graph-Mimic and Musical Dynamics

富有表现力的机器人钢琴家：通过图模仿和音乐动态掌握复杂钢琴曲目

Liang, Yanhong, Liu, Xianwei, Fu, Chaojie, Cheng, Shaowen, Yuan, Yanyan, Zhuo, Chengwei, Chen, Xi, Jin, Yongbin, Yang, Wei, Wang, Hongtao

Abstract

Enabling robots to perform musical instruments with human-level expressivity represents a frontier in bridging the gap between mechanical precision and artistic interpretation. Despite advances in robotic dexterity, replicating the fluid finger transitions and nuanced dynamic control characteristic of human pianists remains a significant challenge. Through a reinforcement learning-based control framework, we demonstrate that a dexterous robotic hand can achieve high-fidelity performance across a diverse piano repertoire. Central to our approach is a graph-based optimization strategy that guides the robot to generate natural pre-press and key-press fingering strategies that closely resemble human movement patterns. To achieve expressive sound production, the control system is coupled with a physics-inspired acoustic model that modulates keypress velocity to accurately reproduce the dynamic variations specified in musical scores. Quantitative evaluations demonstrate that our expressive control model significantly outperforms baseline methods in both finger morphology similarity and dynamic velocity accuracy. In a perceptual test involving participants from diverse listener groups, performances generated by our system are significantly preferred over baseline robotic performances and are indistinguishable from human performances for non-professional audiences. Furthermore, extensive experiments across multiple musical styles confirm that our method maintains high note-level accuracy while achieving expressive performance. Our approach provides a robust pathway for robotic systems to move beyond mere mechanical accuracy, elevating robotic musicianship to a level of expressive performance comparable to human pianists.

Chinese Translation

使机器人能够以人类水平的表现力演奏乐器，代表了在机械精度与艺术诠释之间架起桥梁的前沿。尽管机器人灵巧性已有所进步，但复制人类钢琴家特有的流畅指法过渡和细腻的动态控制仍然是一项重大挑战。通过基于强化学习的控制框架，我们展示了一只灵巧的机器人手能够在多样化的钢琴曲目中实现高保真度的演奏。我们方法的核心是一种基于图的优化策略，它引导机器人生成自然的预按和按键指法策略，这些策略与人类的运动模式高度相似。为了实现富有表现力的音效生成，控制系统与一种受物理启发的声学模型相结合，该模型调节按键速度，以准确再现乐谱中指定的动态变化。定量评估表明，我们的表现力控制模型在指法形态相似性和动态速度准确性方面显著优于基线方法。在涉及不同听众群体的感知测试中，我们系统生成的演奏显著优于基线机器人演奏，并且对于非专业观众而言，与人类演奏无异。此外，在多种音乐风格下的广泛实验确认，我们的方法在保持高音符级准确性的同时，实现了富有表现力的演奏。我们的方法为机器人系统提供了一条稳健的路径，使其超越单纯的机械准确性，将机器人音乐演奏提升到与人类钢琴家相媲美的表现力水平。

View on arXiv Download PDF AI Translation

cs.RO / 5 / 2609.10895

ReactHuman: A Physics-Grounded Benchmark for Human-Like Reactive Decision-Making in Embodied Multimodal LLMs

ReactHuman：一个基于物理的基准测试，用于具身多模态大语言模型中的类人反应决策

Li, Yizhan, You, Jianxin, Xiong, Mengyang, Chen, Yinhuan, Zhao, Zicheng, Wu, Dekun, Zhang, Dongqing, Liu, Bang

Abstract

Reacting to sudden physical hazards (catching a slipping plate, dodging a falling knife) is both a meaningful test of embodied intelligence and a hard requirement for deploying multimodal large language models (MLLMs) as the decision coreof household robots. Existing evaluations, however, probe intuitive physics passively through question answering over videos, or target deliberate, long-horizon tasks such as navigation and rearrangement; none measure whether a model can turn physical understanding into immediate, safety-critical action. We introduce ReactHuman, the first physics-grounded benchmark for human-like reactive decision-making, in which the evaluated MLLM acts as the brain of a simulated humanoid facing sudden household hazards; it spans 17 event families and over 1,000 bit-for-bit reproducible scenes with exact, annotation-free ground truth derived from 240 Hz rigid-body simulation, including adversarial objects whose appearance contradicts their physics (a foam anvil, a steel apple). We further design a five-metric suite that scores each reaction along three axes: reasonable, safe, and physically grounded. We physically execute every committed plan so that decisions have observable consequences. With this harness we evaluate seven representative MLLMs. Results show that reactive safety is far from solved: models mishandle roughly one hazard in three, act from fixed dispositions rather than the observed scene, trust appearance over motion, and miss interception points at meter scale even when the chosen action is correct; none of these failures shrink with model scale. ReactHuman thus offers both a fine-grained diagnosis and a scalable training signal toward physically grounded, safety-aware embodied agents. The benchmark can be found here: https://huggingface.co/datasets/Alan123/reacthuman-benchmark-scaled

Chinese Translation

对突发物理危险（如抓住滑落的盘子、躲避掉落的刀具）的反应既是对具身智能的有意义测试，也是将多模态大语言模型（MLLMs）作为家庭机器人决策核心的一个重要要求。然而，现有评估方法通过对视频的问答被动探测直观物理，或针对导航和重新排列等深思熟虑的长期任务进行评估；没有一种方法测量模型是否能够将物理理解转化为即时的、安全关键的行动。我们引入了ReactHuman，这是第一个基于物理的类人反应决策基准测试，其中被评估的MLLM充当面临突发家庭危险的模拟类人的大脑；它涵盖了17个事件类别和超过1000个逐位可重现的场景，具有来自240 Hz刚体模拟的精确、无注释的真实情况，包括外观与其物理特性相矛盾的对抗性物体（如泡沫铁砧、钢制苹果）。我们进一步设计了一个五项指标的评估套件，从合理性、安全性和物理基础三个维度对每个反应进行评分。我们物理执行每个已承诺的计划，以便决策具有可观察的后果。通过这个框架，我们评估了七个代表性的MLLM。结果表明，反应安全性远未解决：模型在三分之一的情况下错误处理危险，基于固定的倾向而非观察到的场景进行行动，信任外观而非运动，并且即使所选行动是正确的，也会在米级范围内错过拦截点；这些失败在模型规模增大时并没有减小。因此，ReactHuman不仅提供了细致的诊断，还为朝着物理基础、安全意识的具身代理提供了可扩展的训练信号。该基准测试可在此找到：https://huggingface.co/datasets/Alan123/reacthuman-benchmark-scaled

View on arXiv Download PDF AI Translation

cs.RO / 6 / 2609.10905

Planning along Differentiable Charts of Constraint Manifolds with General-Purpose IK Solvers

基于通用逆向运动学求解器的约束流形可微图表规划

Cohn, Thomas, Shaw, Seiji, Biggie, Harel, Manderson, Travis, Roy, Nicholas, Tedrake, Russ

Abstract

Planning trajectories for robot manipulators under kinematic equality constraints restricts feasible motions to a measure-zero submanifold of the configuration space, requiring special algorithmic treatment. A promising strategy is parametrizing the set of feasible configurations using analytic inverse kinematics (IK). Bespoke analytic IK functions can be written to be differentiable, a necessary property for gradient-based trajectory optimization. But the vast majority of IK functions are computed by automated meta-solvers like IKFast, and are difficult to modify for differentiability. We present a new approach for computing gradients of analytic IK parameterizations: we leverage the inverse function theorem to recover the desired gradients from the ordinary forward kinematic Jacobian. Furthermore, we present a least-squares domain extension and an optimization-amenable description of the reachability constraint, which preserves gradient signal outside the reachable workspace. We demonstrate the efficacy of our approach through numerical experiments and downstream tasks, including a hardware demonstration of an RB-Y1 picking up a box and placing it on a table. Project website: https://cohnt.github.io/inverse-function-theorem-parameterization/

Chinese Translation

在运动学等式约束下为机器人操纵器规划轨迹将可行运动限制在配置空间的测度零子流形上，这需要特殊的算法处理。一种有前景的策略是使用解析逆向运动学（IK）对可行配置集进行参数化。可以编写定制的解析IK函数，使其具有可微性，这是基于梯度的轨迹优化所必需的属性。然而，绝大多数IK函数是由自动化元求解器（如IKFast）计算的，且难以修改以满足可微性要求。我们提出了一种新的计算解析IK参数化梯度的方法：我们利用逆函数定理从普通的前向运动学雅可比矩阵中恢复所需的梯度。此外，我们提出了一种最小二乘域扩展和优化友好的可达性约束描述，能够在可达工作空间外保持梯度信号。我们通过数值实验和下游任务展示了我们方法的有效性，包括在硬件上演示RB-Y1拾取一个箱子并将其放置在桌子上的过程。项目网站：https://cohnt.github.io/inverse-function-theorem-parameterization/

View on arXiv Download PDF AI Translation

cs.RO / 7 / 2609.10915

IMLE-VLA: Fast Single-Step Action Generation for Vision-Language-Action Policies

IMLE-VLA：用于视觉-语言-动作策略的快速单步动作生成

Hosseinkhani, Kian, Peng, Qinhe, Shramko, George, Aghabozorgi, Mehran, Qian, Jianing, Engst, Tristan, Moazeni, Alireza, Jayaraman, Dinesh, Li, Ke

Abstract

Vision-language-action (VLA) policies leverage pretrained vision-language backbones to achieve strong cross-task generalization. A leading design couples this backbone with a dedicated continuous action head trained via diffusion or flow matching. However, such heads rely on iterative multi-step sampling, for example 10 Euler steps in $\pi_{0.5}$. This creates an inference bottleneck that produces stop-and-go movement in the robot and slower task completion. We introduce IMLE-VLA, which replaces the iterative action head with a single-step conditional generator trained via conditional Implicit Maximum Likelihood Estimation (cIMLE). The cIMLE objective promotes multimodal action coverage, avoiding the mode collapse of naive regression heads while eliminating multi-step sampling entirely. When IMLE-VLA is applied to $\pi_{0.5}$, it increases inference frequency 3.67x (55 Hz vs. 15 Hz), enabling up to 11x higher action throughput. On the 40-task LIBERO benchmark, IMLE-VLA achieves the highest average success rate (98.0%) among all baselines while leading in inference frequency. Under the test-time perturbations of LIBERO-plus, IMLE-VLA retains $\pi_{0.5}$'s robustness while other baselines degrade sharply, confirming that the cIMLE head preserves generalization. Real-world experiments on a Franka Emika Panda across four tasks demonstrate smoother motion (2.2x to 3.0x lower jerk) and faster task completion, with IMLE-VLA outperforming $\pi_{0.5}$ on every task and reducing average VLA inference time per episode by 3.9x to 6.6x. Videos and code are available at https://kianhk6.github.io/IMLE-VLA/

Chinese Translation

视觉-语言-动作（VLA）策略利用预训练的视觉-语言骨干网络实现强大的跨任务泛化。一个主要设计将该骨干网络与通过扩散或流匹配训练的专用连续动作头相结合。然而，这种动作头依赖于迭代的多步采样，例如在$ ext{π}_{0.5}$中使用10个欧拉步骤。这造成了推理瓶颈，使机器人产生停顿和移动不连贯的现象，并导致任务完成速度变慢。我们提出了IMLE-VLA，它用通过条件隐式最大似然估计（cIMLE）训练的单步条件生成器替代了迭代动作头。cIMLE目标促进了多模态动作覆盖，避免了简单回归头的模式崩溃，同时完全消除了多步采样。当IMLE-VLA应用于$ ext{π}_{0.5}$时，其推理频率提高了3.67倍（55 Hz对比15 Hz），使得动作吞吐量提高了最多11倍。在40任务的LIBERO基准测试中，IMLE-VLA在所有基线中实现了最高的平均成功率（98.0%），同时在推理频率上也处于领先地位。在LIBERO-plus的测试时间扰动下，IMLE-VLA保持了$ ext{π}_{0.5}$的鲁棒性，而其他基线则急剧下降，确认了cIMLE头保持了泛化能力。在Franka Emika Panda上进行的四项任务的实际实验表明，IMLE-VLA实现了更平滑的运动（颤动降低了2.2倍到3.0倍）和更快的任务完成速度，在每项任务中均优于$ ext{π}_{0.5}$，并将每集的平均VLA推理时间减少了3.9倍到6.6倍。视频和代码可在https://kianhk6.github.io/IMLE-VLA/获取。

View on arXiv Download PDF AI Translation

cs.RO / 8 / 2609.10918

ObstaDiff: Generalizable Diffusion Policy Learning via Obstacle-aware Representations

ObstaDiff：通过障碍物感知表示实现可泛化的扩散策略学习

Wang, Jiawen, Yao, Kevin, Jawed, Khalid

Abstract

Imitation learning has achieved impressive results in robotic manipulation, yet most existing approaches assume clean backgrounds and lack explicit mechanisms for obstacle-aware motion generation. Extending such policies to cluttered, real-world scenes with unstructured obstacles remains a key generalization challenge. We present ObstaDiff, a decomposed diffusion-policy framework with a lightweight obstacle-aware visual encoder. ObstaDiff extracts a structured target-obstacle-background representation, enabling the downstream alignment policy to generate end-effector trajectories toward a target-centered bottleneck pose while reasoning about surrounding obstacles. We evaluate ObstaDiff on 61 real-robot greenhouse trials per method (366 executions in total). ObstaDiff achieves 75.41% average task success and 8.20% average obstacle collision rate, outperforming representative imitation-learning baselines and improving generalization in cluttered agricultural scenes.

Chinese Translation

模仿学习在机器人操作中取得了令人瞩目的成果，但大多数现有方法假设背景干净，并缺乏显式的障碍物感知运动生成机制。将此类策略扩展到具有非结构化障碍物的杂乱真实场景仍然是一个关键的泛化挑战。我们提出了ObstaDiff，一种分解的扩散策略框架，配备轻量级的障碍物感知视觉编码器。ObstaDiff提取结构化的目标-障碍物-背景表示，使下游对齐策略能够生成朝向目标中心瓶颈姿态的末端执行器轨迹，同时考虑周围的障碍物。我们在每种方法的61个真实机器人温室试验中评估了ObstaDiff（总共366次执行）。ObstaDiff实现了75.41%的平均任务成功率和8.20%的平均障碍物碰撞率，超越了代表性的模仿学习基线，并改善了在杂乱农业场景中的泛化能力。

View on arXiv Download PDF AI Translation

cs.RO / 9 / 2609.10951

Testing Between the Test Cases: Proving End-to-End Steering in Conditions You Never Drove

测试用例之间的测试：在您从未驾驶的条件下证明端到端转向

Ghalan, Menuka, Rodgers, Charles, Asher, Zachary D.

Abstract

AI-based automated vehicle testing is challenging because a model that passes every test condition can still fail in the real world. Formal verification offers a way to directly address this gap. On a simulated highway and an arterial road we trained two small end-to-end steering networks each in CARLA, one on clear conditions alone and one on clear, fog, night and low sun. All four models were driven against a 2.19 ft lane-departure budget. Without driving again, we used bound propagation, a formal method that reads the trained weights, to compute how far steering can drift at every disturbance strength between two captured images. One calculation covers more than a campaign could drive: on the arterial it spans 133 poses, where ten intensities each would be 10^133 combinations, in minutes on one GPU. Not only did formal verification find conditions that broke the clear-trained policy without simulation testing, it provided some preliminary evidence for potential failures between the test cases. Our overall conclusion is that formal verification is a viable complement to simulation, and could be adopted as a part of verification and validation for automated driving.

Chinese Translation

基于人工智能的自动驾驶车辆测试面临挑战，因为一个在所有测试条件下都通过的模型在现实世界中仍可能失败。形式验证提供了一种直接解决这一差距的方法。在模拟高速公路和主干道上，我们在CARLA中训练了两个小型的端到端转向网络，一个仅在清晰条件下训练，另一个则在清晰、雾霾、夜间和低阳光条件下训练。所有四个模型在2.19英尺的车道偏离预算下进行测试。在不再驾驶的情况下，我们使用边界传播（bound propagation）这一形式方法，读取训练权重，计算在两幅捕获图像之间每种干扰强度下转向可能的漂移距离。一项计算覆盖的范围超过了一次测试活动：在主干道上，它跨越了133个姿态，其中十种强度的组合将达到10^133种可能性，仅需在一台GPU上几分钟的时间。形式验证不仅发现了在没有模拟测试的情况下打破清晰训练策略的条件，还提供了一些潜在失败的初步证据。我们的总体结论是，形式验证是对模拟的可行补充，并可以作为自动驾驶验证和确认的一部分被采用。

View on arXiv Download PDF AI Translation

cs.RO / 10 / 2609.11043

LTLDiff: Finite Linear Temporal Logic-Guided Data Generation and Diffusion Policies for Multi-agent Robotic Manipulation

LTLDiff：基于有限线性时序逻辑的数据生成与扩散策略在多智能体机器人操作中的应用

Meng, Chuhan, Yin, Haiyan

Abstract

Multi-agent robotic manipulation tasks require coordination among agents to satisfy task-level temporal, logical, and safety constraints. Recently, diffusion policies have been used to perform the task. However, they still suffer from desynchronization, incorrect action ordering, and coordination failures in tasks that require simultaneous or sequential multi-agent interaction. Therefore, LTLDiff is proposed as a framework that combines Finite Linear Temporal Logic (LTLf) specification learning for both the generation of demonstrations and learning via diffusion policies. Each task has a specific LTLf formula that is learned from a set of natural language instructions using a large-scale language model. To enable a fixed-dimensional vector embedding of the learned specification from the language model, LTLf uses an abstract syntax tree representation scheme. This embedding of logic serves as a condition for (i) logic-guided data collection and (ii) diffusion-based policy training, encouraging trajectories that are consistent with the desired ordering and coordination requirements. Experiments on multi-agent LTLDiff manipulation tasks demonstrate improved task success rates compared to the baseline. Together, these contributions demonstrate the effectiveness of LTLDiff for coordinated multi-agent manipulation.

Chinese Translation

多智能体机器人操作任务需要智能体之间的协调，以满足任务级别的时序、逻辑和安全约束。最近，扩散策略被用于执行这些任务。然而，它们在需要同时或顺序进行多智能体交互的任务中，仍然面临去同步、错误的动作排序和协调失败的问题。因此，提出了LTLDiff作为一个框架，结合有限线性时序逻辑（LTLf）规范学习，用于演示生成和通过扩散策略进行学习。每个任务都有一个特定的LTLf公式，该公式是通过大型语言模型从一组自然语言指令中学习得到的。为了实现从语言模型中学习到的规范的固定维度向量嵌入，LTLf使用抽象语法树表示方案。逻辑的这种嵌入作为（i）逻辑引导的数据收集和（ii）基于扩散的策略训练的条件，鼓励与期望的排序和协调要求一致的轨迹。在多智能体LTLDiff操作任务上的实验表明，与基线相比，任务成功率得到了改善。这些贡献共同证明了LTLDiff在协调多智能体操作中的有效性。

View on arXiv Download PDF AI Translation

cs.RO / 11 / 2609.11059

Gait-Dependent Effects on Quadruped Locomotion for Load-Carrying using Passive Mechanism

基于步态的被动机制在负载携带四足运动中的影响

Dessy, Giovanni B., Semini, Claudio, Barasuol, Victor

Abstract

Passive mechanical interfaces offer a lightweight alternative to actuated manipulators for quadruped payload carrying, but their impedance directly couples the payload dynamics with the locomotion pattern. This paper analyzes how passive-arm stiffness-damping selection affects payload-carrying locomotion under different gait and payload conditions. We compare damped and underdamped passive-arm impedance configurations in simulation during flat-ground locomotion. For crawl gaits, where the support polygon remains well defined, the results show that underdamped impedance increases passive-joint oscillations and can reduce the ZMP margin with respect to the support polygon. Trot is retained as a dynamic excitation case for the passive arm, but it is not used for direct ZMP-margin stability comparison. The results are summarized in gait-payload-stiffness-damping maps, where ZMP-margin reduction is evaluated for crawl gaits and trot is retained only as a passive-arm excitation case.

Chinese Translation

被动机械接口为四足负载携带提供了一种轻量级的替代方案，相较于驱动操控器，但其阻抗直接将负载动态与运动模式耦合。本文分析了在不同步态和负载条件下，被动臂的刚度-阻尼选择如何影响负载携带的运动。我们在平地运动中对阻尼和欠阻尼的被动臂阻抗配置进行了模拟比较。对于爬行步态，支撑多边形保持良好定义，结果表明，欠阻尼阻抗增加了被动关节的振荡，并可能减少相对于支撑多边形的零力矩点（ZMP）余量。小跑被保留作为被动臂的动态激励案例，但未用于直接的ZMP余量稳定性比较。结果总结在步态-负载-刚度-阻尼图中，其中对爬行步态的ZMP余量减少进行了评估，而小跑仅作为被动臂激励案例保留。

View on arXiv Download PDF AI Translation

cs.RO / 12 / 2609.11078

Freehand Sketching for End-User Programming of Robot Swarms

用于机器人群体的自由手绘编程

Karam, Riwa, Kuo, Ian, Egerstedt, Magnus

Abstract

Robot swarms are increasingly used in applications where accessible interaction with non-expert users is desirable. This paper investigates freehand sketching as an end-user programming interface for specifying robot swarm geometries. Users communicate spatial intent through a drawing, while the swarm autonomously extracts target formation points, constructs a rigid formation graph, assigns robots to formation nodes, and executes distributed formation control with a guarantee against unintended reflected formations. The resulting sketch-to-swarm framework is evaluated through a human study examining the usability of freehand formation specification. Twenty participants generated $42$ geometric shapes, and the interface achieved a mean System Usability Scale score of $84.25$, which conventionally indicates high perceived usability. The results support freehand sketching as an intuitive interaction abstraction for human-swarm collaboration without requiring robotics or programming expertise.

Chinese Translation

机器人群体在需要与非专业用户进行便捷交互的应用中越来越受到欢迎。本文探讨了自由手绘作为终端用户编程接口，用于指定机器人群体的几何形状。用户通过绘图传达空间意图，而机器人群体则自主提取目标队形点，构建刚性队形图，分配机器人到队形节点，并执行分布式队形控制，以确保避免意外的反射队形。通过一项人类研究评估了这一草图到群体的框架，考察了自由手绘队形指定的可用性。二十名参与者生成了42个几何形状，界面获得了84.25的平均系统可用性评分，这通常表明高感知可用性。结果支持自由手绘作为一种直观的交互抽象，促进人类与机器人群体的协作，而无需具备机器人或编程专业知识。

View on arXiv Download PDF AI Translation

cs.RO / 13 / 2609.11079

RIDE: Relocalization-Informed Depth Estimation with 3D Gaussian Splatting

RIDE：基于重定位信息的深度估计与3D高斯点云

Lian, Jiarong, Xiao, Zhe, Zhang, Zhaoyang, Li, Wei, Chen, Ruizhi

Abstract

Render--match--PnP relocalization establishes correspondences between query image pixels and 3D map points for camera pose recovery, but their potential to support dense depth estimation is often overlooked. To exploit this geometric information, we present RIDE, which estimates dense metric depth from a robot's RGB stream. Given a metrically scaled 3D Gaussian Splatting (3DGS) model, RIDE combines sparse metric depth observations derived from PnP-RANSAC inlier correspondences with the geometric prior of a pretrained video-depth model. To handle uneven and intermittent observations, it integrates global and local depth correction with temporal memory, supporting depth estimation through short observation gaps after metric scale initialization. Trained on public RGB-D videos, RIDE is evaluated on robot sequences without fine tuning. Experiments show improved depth accuracy and temporal consistency over scale-only calibration, demonstrating how localization geometry can support both pose recovery and dense robot perception.

Chinese Translation

Render-match-PnP重定位通过建立查询图像像素与3D地图点之间的对应关系来恢复相机姿态，但其在支持密集深度估计方面的潜力常常被忽视。为了利用这一几何信息，我们提出了RIDE，它从机器人的RGB流中估计密集的度量深度。给定一个经过度量缩放的3D高斯点云（3DGS）模型，RIDE将从PnP-RANSAC内点对应关系中获得的稀疏度量深度观测与预训练视频深度模型的几何先验相结合。为了处理不均匀和间歇性的观测，它集成了全球和局部深度修正以及时间记忆，支持在度量尺度初始化后通过短暂观测间隙进行深度估计。RIDE在公共RGB-D视频上进行训练，并在不进行微调的情况下对机器人序列进行评估。实验结果表明，与仅基于尺度的校准相比，RIDE在深度准确性和时间一致性方面有所改善，展示了定位几何如何支持姿态恢复和密集机器人感知。

View on arXiv Download PDF AI Translation

cs.RO / 14 / 2609.11225

Harness Robotic OS: A Unified Embodied-Agent Runtime for Closed-Loop Quadruped Inspection

Harness Robotic OS：一个统一的具身智能体运行时用于闭环四足机器人检查

Yan, Yaoyuan, Heng, Zhiyou, Jie, Haoxiang, Liu, Gang, Yan, Hongjie, Zhou, Wei

Abstract

Autonomous property inspection requires more than robust robot navigation: a deployable system must connect heterogeneous sensing, reusable autonomy capabilities, multimodal scene understanding, human interaction, and enterprise response within a traceable operational loop. Existing quadruped inspection systems commonly integrate these functions through task-specific interfaces, making contextual coordination, knowledge reuse, and controlled adaptation difficult. This paper presents \textit{Harness Robotic OS} (HROS), a unified embodied-agent runtime, and Argos, its realization for residential-community inspection. HROS organizes the system into robot runtime, embodied autonomy skills, cognitive agent runtime, and interaction and operations planes. A shared context connects physical state with agent reasoning; streaming ASR/TTS supports voice-based mission interaction; hierarchical working, episodic, and semantic memory preserves operational knowledge; and a safety-gated self-evolution loop converts execution traces into versioned candidate updates without permitting unconstrained online modification. The Argos prototype integrates a Vbot quadruped, Fast-LIO2 localization and mapping, Hobot-Stereo depth perception, PCT-Planner global planning, EGO-Planner local motion generation, and OpenClaw-orchestrated Qwen3-VL inspection analysis. Experiments in a residential property environment achieved 100\% waypoint reachability, outdoor localization error below 10~cm, local obstacle-response latency below 200~ms, representative hazard-detection rates of 85--95\%, and 99\% success in alarm delivery and structured-report generation. These results validate the deployed navigation and inspection closed loop, while HROS provides an extensible software foundation for memory-augmented, voice-aware, and continuously improvable embodied inspection agents.

Chinese Translation

自主物业检查不仅需要强大的机器人导航能力：一个可部署的系统必须在可追溯的操作循环中连接异构传感器、可重用的自主能力、多模态场景理解、人机交互和企业响应。现有的四足机器人检查系统通常通过特定任务的接口整合这些功能，这使得上下文协调、知识重用和受控适应变得困难。本文提出了 extit{Harness Robotic OS} (HROS)，一个统一的具身智能体运行时，以及其在住宅社区检查中的实现Argos。HROS将系统组织为机器人运行时、具身自主技能、认知智能体运行时以及交互和操作层面。共享上下文将物理状态与智能体推理连接起来；流式ASR/TTS支持基于语音的任务交互；分层工作、情节记忆和语义记忆保留操作知识；安全门控的自我演变循环将执行轨迹转化为版本化候选更新，而不允许不受限的在线修改。Argos原型集成了Vbot四足机器人、Fast-LIO2定位与映射、Hobot-Stereo深度感知、PCT-Planner全局规划、EGO-Planner局部运动生成以及OpenClaw协调的Qwen3-VL检查分析。在住宅物业环境中的实验实现了100\%的航点可达性，户外定位误差低于10厘米，局部障碍响应延迟低于200毫秒，典型的危险检测率为85--95 ext{ extbackslash%}，以及99 ext{ extbackslash%}的警报传递和结构化报告生成成功率。这些结果验证了部署的导航和检查闭环，而HROS为增强记忆、语音感知和持续可改进的具身检查智能体提供了可扩展的软件基础。

View on arXiv Download PDF AI Translation

cs.RO / 15 / 2609.11270

Beyond Noise Steering: Dual-Latent Space Reinforcement Learning for Generative Robot Policy

超越噪声引导：用于生成机器人策略的双潜在空间强化学习

Zhang, Pengfei, Sun, Teng, Xiu, Xianchao

Abstract

Pretrained generative robot policies learn expressive action priors from demonstrations. However, existing reinforcement learning methods only steer the noisy space but fail to modulate intermediate action representations during the generation process, resulting in performance degradation and inefficiency. To address this limitation, we propose a novel Dual-Latent Space Reinforcement Learning (DLSRL) framework, which complements initial-noise steering with representation-level control inside the frozen generator. Specifically, our actor network predicts two distinct latent variables: an initial-noise latent variable that steers behavior generation, and an action-representation latent variable for intermediate feature modulation. Moreover, this representation latent variable is mapped to adapter features and ingeniously injected into the hidden states of intermediate action tokens via residual connections. Our dual-control design enables direct adjustment of action representations without updating the base policy. Experiments across generative policy architectures and robotic manipulation tasks show that DLSRL effectively accelerates online robot policy adaptation and achieves competitive performance. Our code is available at \href{https://github.com/xianchaoxiu/DLSRL}{https://github.com/xianchaoxiu/DLSRL}.

Chinese Translation

预训练的生成机器人策略从示范中学习表达性动作先验。然而，现有的强化学习方法仅对噪声空间进行引导，却未能在生成过程中调节中间动作表示，导致性能下降和效率低下。为了解决这一局限性，我们提出了一种新颖的双潜在空间强化学习（Dual-Latent Space Reinforcement Learning, DLSRL）框架，该框架在冻结生成器内部补充了初始噪声引导与表示级控制。具体而言，我们的演员网络预测两个不同的潜在变量：一个用于引导行为生成的初始噪声潜在变量，以及一个用于中间特征调节的动作表示潜在变量。此外，该表示潜在变量被映射到适配器特征，并通过残差连接巧妙地注入到中间动作标记的隐藏状态中。我们的双重控制设计使得在不更新基础策略的情况下直接调整动作表示成为可能。在生成策略架构和机器人操作任务中的实验表明，DLSRL有效加速了在线机器人策略的适应，并实现了竞争力的性能。我们的代码可在 exttt{https://github.com/xianchaoxiu/DLSRL} 获取。

View on arXiv Download PDF AI Translation

cs.RO / 16 / 2609.11308

2AM: Grounding Agent-Side Memory as Guidance for Steerable Action Models in Long-Horizon Manipulation

2AM：将代理端记忆作为长时间操作中可引导动作模型的指导

Hu, Yutong, Chen, Fengjiao, Cao, Xuezhi, Detry, Renaud

Abstract

Long-horizon robot manipulation requires memory, but not necessarily inside the action policy. To address such tasks, current agentic systems often combine VLAs with planners and geometric tools, sometimes using additional depth or calibrated geometry. These systems confound attribution: gains may come from richer observations or alternative motor tools, while failures may stem from either the policy or an under-specified language interface. We isolate this question through a deliberately constrained design: less tool breadth, but greater interface bandwidth. 2AM makes a multimodal Agent the sole holder of task memory and a single RGB-based, episodically stateless Action Model the sole executor of task-relevant motion. The Agent compiles interaction history into subtask language and optional 2D grasp, place, and move hints that bind its physical intention at different time scales. To teach this steerability to the VLA, we augment demonstrations with structured hint labels and train under condition dropout, spatial noise, and temporal jitter to tolerate imperfect Agent outputs. On LIBERO-Mem, without depth, online geometry, or planner-based object motion, 2AM reaches 76.3% average completion, a 61.5-point improvement over the strongest reported baseline of 14.8%, together with 63.0% relaxed and 11.8% strict success. These results show that task memory can remain Agent-side. They further show that Action Model capability depends not only on what the policy has learned, but on how precisely the Agent can steer it.

Chinese Translation

长时间机器人操作需要记忆，但不一定要在动作策略内部。为了解决此类任务，目前的代理系统通常将可变形状的动作（VLA）与规划器和几何工具结合，有时还使用额外的深度或校准几何。这些系统混淆了归因：收益可能来自更丰富的观察或替代的运动工具，而失败可能源于策略或未充分指定的语言接口。我们通过故意限制设计来孤立这个问题：工具范围较小，但接口带宽更大。2AM使多模态代理成为任务记忆的唯一持有者，而单一基于RGB的、无状态的动作模型则成为任务相关运动的唯一执行者。代理将交互历史编译成子任务语言和可选的二维抓取、放置和移动提示，这些提示在不同时间尺度上绑定其物理意图。为了教会VLA这种可引导性，我们通过结构化提示标签增强演示，并在条件丢失、空间噪声和时间抖动下进行训练，以容忍不完美的代理输出。在LIBERO-Mem上，在没有深度、在线几何或基于规划的物体运动的情况下，2AM达到了76.3%的平均完成率，比报告的最强基线14.8%提高了61.5个百分点，同时实现了63.0%的放宽成功率和11.8%的严格成功率。这些结果表明，任务记忆可以保持在代理端。它们进一步表明，动作模型的能力不仅依赖于策略所学的内容，还依赖于代理能够多精确地引导它。

View on arXiv Download PDF AI Translation

cs.RO / 17 / 2609.11338

Modular Kinematic Reduction of Closed-Chain Mechanisms Using Path Assembly and Defect Homotopy

基于路径组装和缺陷同伦的闭链机构的模块化运动学简化

Dastranj, Mohammad, Mattila, Jouni

Abstract

Closed kinematic chains complicate modular modeling by coupling active and passive coordinates through nonlinear closure constraints. This paper presents a Path-Assembled Closure Differential Mapping (PACDM) framework for modular closure resolution and kinematic reduction. Each closure element compares two ordered transformation paths with common endpoints, with their mismatch expressed through the logarithm on SE(3) and the corresponding Jacobian assembled from local transformation derivatives. Multi-path modules are constructed from a minimal set of pairwise closure elements, while rank-revealing analysis selects locally independent scalar constraints. A defect homotopy recovers closure-consistent passive coordinates from approximate estimates along a feasible and regular continuation path. At regular configurations, implicit differentiation yields the local active-to-passive differential mapping, which is subsequently used in a predictor-corrector continuation procedure for prescribed motion. The framework is evaluated on a seven-degree-of-freedom heavy-duty manipulator containing two-path and three-path closed-chain modules. Comparison with Simscape Multibody yields trajectory root-mean-square errors below 8.5 x 10^-10 rad, while predictor-corrector continuation is approximately 45.8 times faster than applying defect homotopy at every trajectory sample.

Chinese Translation

闭合运动链通过非线性闭合约束耦合了主动和被动坐标，从而使模块化建模变得复杂。本文提出了一种路径组装闭合微分映射（Path-Assembled Closure Differential Mapping, PACDM）框架，用于模块化闭合解析和运动学简化。每个闭合元素比较两个具有共同端点的有序变换路径，其不匹配通过在 SE(3) 上的对数表示，并且相应的雅可比矩阵由局部变换导数组装而成。多路径模块由一组最小的成对闭合元素构建，而秩揭示分析选择局部独立的标量约束。缺陷同伦从沿可行且规则的延续路径的近似估计中恢复闭合一致的被动坐标。在规则配置下，隐式微分产生局部主动到被动的微分映射，随后在规定运动的预测-修正延续过程中使用。该框架在一个包含两路径和三路径闭链模块的七自由度重型机械手上进行了评估。与 Simscape Multibody 的比较显示，轨迹均方根误差低于 8.5 x 10^-10 rad，而预测-修正延续的速度约为在每个轨迹样本上应用缺陷同伦的 45.8 倍。

View on arXiv Download PDF AI Translation

cs.RO / 18 / 2609.11357

Morphology-Aware Human Motion Retargeting for Wheeled-Humanoid Loco-Manipulation

形态感知的人类运动重定向用于轮式类人机器人运动操控

Xia, Chenbo, Ye, Chao

Abstract

Human-to-humanoid retargeting has largely been studied on legged platforms, while comparatively few wheeled-humanoid systems support coupled locomotion and manipulation from general human motion. Building on GMR's configurable general-motion retargeting and BeyondMimic's physically simulated R1 Pro learning framework, we present a reproducible pipeline that converts multi-dataset SMPLX motion into executable loco-manipulation behavior for the Galaxea R1 Pro wheeled humanoid. The robot has a planar three-wheel base, a serial torso, and two arms but no leg joints, so human lower-body motion must be redistributed across base motion and torso posture without sacrificing manipulation-relevant arm geometry. Our pipeline combines canonical body-shape preprocessing, planar-base normalization, morphology-aware differential inverse kinematics, shoulder-rooted hierarchical arm retargeting, and continuous torso substitution for bending and squatting. A reference-twist-driven planning layer then decodes planar base motion into continuous three-wheel steering and rolling commands subject to hysteresis, kinematic continuity, acceleration, and actuator-rate limits. Finally, a 21-dimensional BaseDecode policy is trained in Isaac Lab with directional joint-limit scaling, focused upper-body tracking, and a staged wheel-contact reward. The resulting system provides a complete bridge from human motion data to physically trackable wheeled-humanoid loco-manipulation rather than a visualization-only retargeter; quantitative policy comparisons remain scheduled for a later revision.

Chinese Translation

人类到类人的重定向主要集中在有腿平台上，而支持从一般人类运动进行耦合运动和操控的轮式类人系统相对较少。基于可配置的一般运动重定向方法 GMR 和物理模拟的 R1 Pro 学习框架 BeyondMimic，我们提出了一种可复现的流程，将多数据集的 SMPLX 动作转换为可执行的轮式类人机器人 Galaxea R1 Pro 的运动操控行为。该机器人具有平面三轮底盘、串联躯干和两只手臂，但没有腿关节，因此人类下肢运动必须在底盘运动和躯干姿态之间重新分配，同时不牺牲与操控相关的手臂几何形状。我们的流程结合了规范的身体形状预处理、平面底盘归一化、形态感知的微分逆运动学、肩部根部的分层手臂重定向和连续躯干替代以实现弯曲和下蹲。然后，一个基于参考扭矩的规划层将平面底盘运动解码为连续的三轮转向和滚动指令，受滞后、运动学连续性、加速度和执行器速率限制的约束。最后，在 Isaac Lab 中训练了一个 21 维的 BaseDecode 策略，采用方向关节限制缩放、集中上半身跟踪和分阶段的轮接触奖励。最终系统提供了从人类运动数据到物理可追踪的轮式类人机器人运动操控的完整桥梁，而不仅仅是一个可视化重定向器；定量政策比较计划在后续修订中进行。

View on arXiv Download PDF AI Translation

cs.RO / 19 / 2609.11361

GeoTrussRover: Morphological Computation with Contact-Semantic Control Primitives

GeoTrussRover：具有接触语义控制原语的形态计算

Ma, Muyuan, Zhang, Yi, Yang, Yang, Zheng, Xuanyan, Hu, Ruiqi, Ke, Boxuan, Chen, Zhenyu, Lin, Yicong, Yang, Xin Hao, Xiao, Daliang, Hou, Zhinan, Niu, Wanhao, Sun, Yuan, Yang, Yan, Xie, Yue

Abstract

Reconfigurable robots can change their contact geometry when a fixed body cannot negotiate an obstacle. A variable-geometry truss (VGT) distributes this shape change through a load-bearing structure, but coupling it to a mobile base creates a high-dimensional coordination problem. GeoTrussRover combines an electrically actuated VGT, a wheeled base, and contact-semantic morphology planning and control. We solve one source traversal and extract four contact-semantic primitives that describe coordination among 21 members. Physics-constrained projection adapts them to unseen step heights with the same contact topology. When every phase remains feasible, adaptation does not recompute the complete motion. If one phase violates the new physical constraints, only that phase is recomputed. A full-space QP then tracks the adapted motion and corrects member and wheel errors. For transfer from 0.10m to 0.075m, the method reduces objective-function evaluations by 63.7% relative to full recomputation. Contact-phase feasibility analysis covers step heights from 0.10 to 0.46m, or 1.08 to 4.97 wheel radii, with the upper value near the theoretical feasible boundary. The electric prototype traverses 2.11 wheel radii. The resulting low-dimensional representation stores task coordination in a hyper-redundant, load-bearing morphology and reuses it during locomotion.

Chinese Translation

可重构机器人在固定物体无法通过障碍物时可以改变其接触几何形状。可变几何桁架（Variable-Geometry Truss, VGT）通过承载结构分配这种形状变化，但将其与移动底座耦合会产生高维协调问题。GeoTrussRover结合了电动驱动的VGT、轮式底座以及接触语义形态规划和控制。我们解决了一个源点遍历问题，并提取了四个描述21个构件之间协调的接触语义原语。物理约束投影使其适应具有相同接触拓扑的未知台阶高度。当每个阶段保持可行时，适应过程不会重新计算完整的运动。如果某个阶段违反了新的物理约束，仅重新计算该阶段。然后，完整空间的二次规划（QP）跟踪适应后的运动并修正构件和轮子的误差。在从0.10米转移到0.075米的过程中，该方法相较于完全重新计算减少了63.7%的目标函数评估。接触阶段可行性分析涵盖了从0.10米到0.46米的台阶高度，或1.08到4.97轮半径，且上限值接近理论可行边界。电动原型能够跨越2.11轮半径。最终得到的低维表示在超冗余的承载形态中存储任务协调，并在运动过程中重复使用。

View on arXiv Download PDF AI Translation

cs.RO / 20 / 2609.11382

SwarmNxt: Open-source Software-Hardware Platform for Fast and Agile Aerial Swarms

SwarmNxt：快速灵活的空中群体开源软硬件平台

Toumieh, Charbel, Mistry, Niel, Jarvis, Benjamin, Jeger, Simon, Liu, Peize, Shen, Shaojie, Floreano, Dario

Abstract

Aerial robot swarms have the potential to transform time-critical safety, security, and search-and-rescue operations. By coordinating multiple robots, they can rapidly survey disaster sites, map collapsed or GPS-denied environments, and search cluttered areas faster than a single robot, reducing response times and minimizing risks to first responders. Realizing this potential, however, requires robust autonomous swarm navigation, which remains an active research challenge. Progress is further constrained by existing platforms, as commercial drones are often closed-source or lack the onboard computational resources needed for agile, vision-based collective flight. Moreover, developing, deploying, and maintaining software across multiple aerial robots requires significant engineering effort. To address these challenges, we present SwarmNxt, an open-source software platform built on the open-source OmniNxt drone hardware. SwarmNxt provides an end-to-end toolkit, including detailed hardware assembly instructions with a video tutorial, automation tools for parallel software deployment and swarm-wide updates, and a ROS 2-based framework for autonomous navigation. The platform integrates state-of-the-art control, planning, and depth estimation into a single ROS 2 multi-agent system, providing an open research infrastructure for physical swarm experimentation. We validate SwarmNxt through two real-world experiments: a six-drone swarm performing decentralized planning with high-speed inter-drone collision avoidance, and a four-drone swarm executing collective flight with onboard depth estimation in an obstacle-filled environment. Both experiments were run indoors with global position from external motion capture; perception, planning, and control run onboard.

Chinese Translation

空中机器人群体有潜力改变时间敏感的安全、安保和搜救操作。通过协调多个机器人，它们能够快速勘测灾害现场、绘制倒塌或无GPS环境的地图，并比单个机器人更快地搜索杂乱区域，从而减少响应时间并降低对第一响应者的风险。然而，实现这一潜力需要强大的自主群体导航，这仍然是一个活跃的研究挑战。现有平台进一步限制了进展，因为商业无人机通常是闭源的，或缺乏进行灵活、基于视觉的集体飞行所需的机载计算资源。此外，在多个空中机器人之间开发、部署和维护软件需要大量的工程工作。为了解决这些挑战，我们提出了SwarmNxt，一个基于开源OmniNxt无人机硬件构建的开源软件平台。SwarmNxt提供了一个端到端的工具包，包括详细的硬件组装说明和视频教程、用于并行软件部署和群体范围更新的自动化工具，以及一个基于ROS 2的自主导航框架。该平台将最先进的控制、规划和深度估计集成到一个单一的ROS 2多智能体系统中，为物理群体实验提供了开放的研究基础设施。我们通过两个真实世界的实验验证了SwarmNxt：一个六架无人机的群体执行去中心化规划并实现高速无人机间碰撞避免，另一个四架无人机的群体在充满障碍的环境中执行集体飞行并进行机载深度估计。这两个实验均在室内进行，全球定位由外部运动捕捉提供；感知、规划和控制均在机载运行。

View on arXiv Download PDF AI Translation

cs.RO / 21 / 2609.11433

Safety-aware Skill Adaptation for Reinforcement Learning in Dynamic Environments

动态环境中基于安全意识的强化学习技能适应

Haque, A K M Nadimul, Sutjipto, Sheila, Carmichael, Marc G., Vidal-Calleja, Teresa

Abstract

Skill adaptation frameworks based on reinforcement learning often require restrictive assumptions to maintain stability, such as fixed observations or tightly controlled exploration schedules. In cluttered and dynamic environments, however, unrestricted exploration can lead to unsafe behaviour and unstable learning, particularly when task-relevant observations lie near obstacles or involve moving objects. In this work, we present Dist-GPRL, a distance-aware and safety-guided reinforcement learning framework for structured robot skill adaptation. Building upon Gaussian Process (GP)-based skill parameterisation, our framework sequentially adapts overlapping local windows of sparse trajectory via-points rather than modifying the complete skill at every policy step. Raw policy outputs are correlated through the GP covariance structure, producing temporally coherent trajectory updates while reducing the action-space and credit-assignment difficulties associated with global trajectory adaptation. Safety is incorporated through two complementary forms of guidance. A safe-subspace prior derived from the Hausdorff Approximation Planner (HAP) biases policy exploration toward feasible regions, while dynamically updated distance field clearance and gradient rewards provide local obstacle awareness. A trajectory-kinematics similarity regulariser further preserves the demonstrated velocity and acceleration characteristics during adaptation. We evaluate the framework on two dynamic object-manipulation tasks in simulation and transfer the learned policy to real-world robot execution. Experimental results demonstrate higher task success, lower collision frequency, and more stable learning than the baselines, while preserving the kinematic characteristics of the demonstrated skill.

Chinese Translation

基于强化学习的技能适应框架通常需要一些限制性假设以维持稳定性，例如固定的观测或严格控制的探索计划。然而，在杂乱和动态的环境中，无限制的探索可能导致不安全的行为和不稳定的学习，特别是当任务相关的观测接近障碍物或涉及移动物体时。在本研究中，我们提出了Dist-GPRL，一个基于距离感知和安全引导的强化学习框架，用于结构化机器人技能适应。我们的框架基于高斯过程（Gaussian Process, GP）进行技能参数化，依次适应稀疏轨迹通过点的重叠局部窗口，而不是在每个策略步骤中修改完整的技能。原始策略输出通过GP协方差结构相关联，产生时间上连贯的轨迹更新，同时减少与全局轨迹适应相关的动作空间和信用分配困难。安全性通过两种互补的引导形式纳入框架。基于豪斯多夫近似规划器（Hausdorff Approximation Planner, HAP）导出的安全子空间先验偏向于可行区域的策略探索，而动态更新的距离场间隙和梯度奖励则提供局部障碍物意识。轨迹运动学相似性正则化器进一步在适应过程中保持了展示的速度和加速度特征。我们在两个动态物体操作任务的仿真中评估了该框架，并将学习到的策略转移到现实世界的机器人执行中。实验结果表明，与基线相比，该框架实现了更高的任务成功率、更低的碰撞频率和更稳定的学习，同时保持了展示技能的运动学特征。

View on arXiv Download PDF AI Translation

cs.RO / 22 / 2609.11445

FARM: Reading Failure Signals from the Internal Predictive States of a Frozen Robotic World Model

FARM：从冻结的机器人世界模型内部预测状态中读取故障信号

Pei, Haoran, Luo, Mingrui, Wang, Senbao, Lv, Haoran, Guo, Jie, Zhong, Sheng, Ci, Ruixi

Abstract

Reliable robot deployment requires online failure monitoring, yet existing monitors mainly derive risk from proxy signals or train dedicated monitoring components. We ask whether the internal predictive states of a frozen pretrained robotic world model already contain directly decodable failure information. Failure-Aware Readout from World Models (FARM) trains only a 33,985-parameter supervised readout over frozen VLA-JEPA predictive states, producing step-wise failure scores and causal trajectory risk. Five-fold out-of-fold evaluation across seven source tasks reaches 85.68/88.59 pooled AUROC/AUPRC, and FARM gives the best Seen performance among 15 matched baselines on the 10-task benchmark. Across four real-robot populations on PIPER X, SO-101, and Franka, fixed-readout transfer and readout-only adaptation test deployment shifts without updating the predictive backbone. FARM also discriminates failures from partial causal histories and adds 0.2256 ms mean CUDA latency once the frozen state is available. These results support frozen predictive world-model states as reusable features for causal, transferable, and low-overhead execution monitoring.

Chinese Translation

可靠的机器人部署需要在线故障监测，但现有的监测器主要通过代理信号推导风险或训练专用监测组件。我们探讨冻结的预训练机器人世界模型的内部预测状态是否已经包含可以直接解码的故障信息。故障感知读取（Failure-Aware Readout from World Models，FARM）仅在冻结的VLA-JEPA预测状态上训练一个33,985参数的监督读取器，生成逐步故障评分和因果轨迹风险。在七个源任务上进行的五折交叉验证达到了85.68/88.59的汇总AUROC/AUPRC，并且FARM在10任务基准测试中在15个匹配基线中表现出最佳的已见性能。在PIPER X、SO-101和Franka的四个真实机器人群体中，固定读取转移和仅读取适应测试部署的转变无需更新预测骨干。FARM还能够从部分因果历史中区分故障，并且在冻结状态可用后增加0.2256毫秒的平均CUDA延迟。这些结果支持将冻结的预测世界模型状态作为因果、可转移和低开销执行监测的可重用特征。

View on arXiv Download PDF AI Translation

cs.RO / 23 / 2609.11476

3D Euler-Angle Orientation Control for Two-Ray Fading Mitigation in Maritime Air-to-Sea Communications

用于减轻海洋空对海通信中双射衰落的三维欧拉角姿态控制

Bajja, Mohammed, Saliah, Abdoul Karim A. H., Hammouti, Hajar El, Licea, Daniel Bonilla, Silano, Giuseppe

Abstract

Maritime Air-to-Sea links are dominated by a line-of-sight ray and a sea-surface reflected ray whose destructive combination produces deep fades. Existing mitigation strategies optimize Unmanned Aerial Vehicle position or trajectory but leave attitude unexploited. This paper treats the full three dimensional attitude as a physical-layer control variable that shapes the two-ray interference through antenna phase-center displacement. Under small-angle and far-field assumptions, the constructive-interference condition reduces to an affine constraint in the Euler angles and admits a closed-form family of minimum-norm attitude candidates. A differentiable soft-minimum rule yields a smooth reference tracked by a constrained Nonlinear Model Predictive Control controller on a fully-actuated tilting multirotor. The proposed scheme increases cumulative throughput by 11.4% over a pitch-only benchmark and 22.2% over a zero-orientation baseline, while preserving trajectory tracking and respecting actuator limits.

Chinese Translation

海洋空对海链路主要受视距射线和海面反射射线的影响，这两者的破坏性组合会导致深度衰落。现有的减轻策略优化无人机的位置或轨迹，但未充分利用姿态。本文将完整的三维姿态视为一种物理层控制变量，通过天线相位中心位移来调节双射干扰。在小角度和远场假设下，建设性干扰条件简化为欧拉角的仿射约束，并允许一组闭式解的最小范数姿态候选。可微分的软最小规则产生一个平滑的参考，由约束非线性模型预测控制器在完全驱动的倾斜多旋翼上跟踪。所提出的方案在保持轨迹跟踪和尊重执行器限制的同时，相较于仅考虑俯仰的基准提高了11.4%的累积吞吐量，相较于零姿态基线提高了22.2%。

View on arXiv Download PDF AI Translation

cs.RO / 24 / 2609.11478

CARLAverse: A Highly Modular, Distributed, and Multimodal Framework for Human-in-the-Loop Simulation

CARLAverse：一个高度模块化、分布式和多模态的人机协同仿真框架

Rebling, Patrick, Nenninger, Philipp, Kriesten, Reiner

Abstract

The development of autonomous driving demands comprehensive testing in mixed-traffic scenarios involving vulnerable road users (VRUs), where purely artificial agents often fail to capture authentic human social negotiations. While human-in-the-loop (HITL) simulators enable safe investigation of these interactions, existing multi-agent platforms struggle with the network latency and synchronization constraints required for high-fidelity haptic feedback. To resolve this, we present CARLAverse, an open-source, multimodal simulation ecosystem. Extending modular hardware abstraction, CARLAverse integrates driving (DrivoCARLA), cycling (CycloCARLA), and pedestrian (WalkoCARLA) simulators into a shared virtual environment. Its core methodological contribution is a distributed physics architecture: latency-critical ego dynamics and high-frequency force feedback are computed locally on client nodes, while a central CARLA server orchestrates non-player character (NPC) physics and global traffic. By decoupling haptic control loops from network bottlenecks, CARLAverse enables scalable, cross-institutional HITL experiments without compromising physical immersion. Code and documentation: https://git.ieem-ka.de/simulator-environments/carlaverse

Chinese Translation

自动驾驶的发展需要在涉及脆弱道路用户（VRUs）的混合交通场景中进行全面测试，而纯人工智能代理往往无法真实捕捉人类社会互动的复杂性。虽然人机协同（HITL）仿真器能够安全地研究这些互动，但现有的多智能体平台在高保真触觉反馈所需的网络延迟和同步约束方面面临挑战。为了解决这一问题，我们提出了CARLAverse，一个开源的多模态仿真生态系统。CARLAverse扩展了模块化硬件抽象，将驾驶（DrivoCARLA）、骑行（CycloCARLA）和行人（WalkoCARLA）仿真器整合到一个共享的虚拟环境中。其核心方法论贡献是一个分布式物理架构：延迟敏感的自我动态和高频力反馈在客户端节点上本地计算，而中央CARLA服务器则协调非玩家角色（NPC）的物理和全局交通。通过将触觉控制循环与网络瓶颈解耦，CARLAverse使得可扩展的跨机构HITL实验成为可能，而不妥协于物理沉浸感。代码和文档：https://git.ieem-ka.de/simulator-environments/carlaverse

View on arXiv Download PDF AI Translation

cs.RO / 25 / 2609.11549

Using Automated Vehicles Operational Data to Confirm Safety and Anticipate Threats

利用自动驾驶车辆运营数据确认安全性并预测威胁

Donà, Riccardo, Rusciano, Espedito, Trentadue, Germana, Tsakalidis, Anastasios, Galassi, Maria Cristina

Abstract

European Union (EU) policymakers adopted revolutionary data collection provisions for Automated Driving Systems (ADS) in the recently approved regulation that allows driverless vehicles to be operated on public roads. The framework is inspired by best practices developed at the United Nations Economic Commission for Europe(UNECE) level: the In-Service Monitoring and Reporting (ISMR); and by similar operational data collection regulatory approaches in nuclear energy production and transportation fields. The collection of real-world data will enable the competent safety authorities to gather the information needed to confirm the homologation safety target. Safety-relevant driving scenarios discovered during the real-world operation of a given ADS can also be stored in a scenario catalogue to investigate how other ADS types might have addressed such a traffic conflict. Moreover, lessons learnt deriving from the data collected can be shared among original equipment manufacturers (OEMs) and safety authorities. Ultimately, the ISMR is recognised as a necessary tool to properly tackle the challenges associated with ADS safety assessment given the number of unknowns that might remain undisclosed by leveraging the traditional homologation validation scheme only.

Chinese Translation

欧洲联盟（EU）政策制定者在最近批准的法规中采纳了革命性的自动驾驶系统（ADS）数据收集条款，该法规允许无人驾驶车辆在公共道路上运营。该框架受到在联合国欧洲经济委员会（UNECE）层面上制定的最佳实践的启发：在役监测与报告（ISMR）；以及核能生产和运输领域类似的运营数据收集监管方法。真实世界数据的收集将使相关安全主管部门能够获取确认同 homologation 安全目标所需的信息。在给定的 ADS 的真实世界操作中发现的与安全相关的驾驶场景也可以存储在场景目录中，以调查其他类型的 ADS 如何处理此类交通冲突。此外，从收集的数据中获得的经验教训可以在原始设备制造商（OEM）和安全主管部门之间共享。最终，ISMR 被认为是妥善应对与 ADS 安全评估相关挑战的必要工具，因为仅依靠传统的 homologation 验证方案可能仍会存在许多未知因素未被披露。

View on arXiv Download PDF AI Translation

cs.RO / 26 / 2609.11553

CAP: Continuously Adaptive Perception-Blind Humanoid Locomotion via Learned Denoising

CAP：通过学习去噪实现的连续自适应感知盲人类运动

Chen, Hongjin, Xu, Zijun, Ma, Shihao, Zhao, Yi, Liu, Xilai, Ma, Ke, Zhang, Wei, Xie, Chunyang, Li, Pengfei, Zhao, Jieru, Ding, Wenchao

Abstract

Humanoid locomotion across complex terrain demands forward-looking exteroception to anticipate obstacles, yet this signal is unreliable in real-world deployment, failing partially and intermittently. Existing perceptive policies often assume that depth observations remain clean and in-distribution, while recent attempts to unify perceptive and blind control typically route or switch between separate sub-policies, leaving recoverable information in partially corrupted depth unexploited. We instead propose CAP, a single-stage humanoid locomotion policy that recovers this signal with a perceptive world-model encoder trained as a learned denoiser to reconstruct clean depth from a corrupted input, together with a co-active proprioceptive variational encoder that supplies depth-free body-state information. A coupled training recipe pairs a depth-noise curriculum on the world-model input with world-model feature dropout on the policy-facing latent, exposing the policy to failures across the entire perception-quality spectrum. In simulation, CAP matches or improves upon perceptive baselines when depth remains informative, and degrades more smoothly than a binary-switching baseline as perception worsens. On the Unitree G1, controlled trials and indoor-outdoor deployments demonstrate perception-robust locomotion under intermittent occlusion, real-sensor corruption, and outdoor depth artifacts.

Chinese Translation

人类在复杂地形上的运动需要前瞻性的外部感知以预测障碍物，然而在实际应用中，这一信号往往不可靠，存在部分失效和间歇性问题。现有的感知策略通常假设深度观测保持清晰且符合分布，而最近尝试将感知与盲控统一的工作通常在不同的子策略之间进行切换或路由，未能充分利用部分受损深度中可恢复的信息。我们提出了CAP，一种单阶段的人类运动策略，通过训练为学习去噪器的感知世界模型编码器，从受损输入中重建清晰的深度信号，同时结合一个协同主动的本体感知变分编码器，提供无深度的身体状态信息。耦合的训练方案将世界模型输入上的深度噪声课程与面向策略的潜在特征的世界模型特征丢弃相结合，使策略暴露于整个感知质量谱的失败中。在仿真中，当深度信息仍然有效时，CAP的表现与感知基线相当或更优，而在感知恶化时，其降级比二元切换基线更为平滑。在Unitree G1上，控制试验和室内外部署展示了在间歇性遮挡、真实传感器损坏和户外深度伪影下的感知稳健运动。

View on arXiv Download PDF AI Translation

cs.RO / 27 / 2609.11561

Memory as Plans: World-Action Modeling with Memory-Grounded Planning

记忆作为计划：基于记忆的世界-行动建模

Zhao, Sizhe, Xie, Haozhe, Zhao, Weiyu, Zhang, Chenchu, Wang, Huan, Wang, Chenyang, Liu, Qinglin, Zhang, Shengping

Abstract

Mainstream robotic policies often adopt a Markovian formulation, but many complex real-world manipulation tasks are inherently non-Markovian, requiring long-horizon memory beyond the current observation. Existing memory mechanisms often rely on language summaries, growing visual windows, or their combinations, and may therefore lose fine-grained visual evidence or face a trade-off between history coverage and execution efficiency. We introduce MaP-WAM, a Memory-as-Plans framework that decomposes memory-dependent world-action modeling into memory-grounded planning and plan-conditioned execution, and uses long-term multimodal episodic context as planning-time evidence rather than repeatedly conditioning the executor on the full history. MaP-WAM represents memory as completed segment records containing language instructions and sparse visual context, and converts this episodic memory into compact plans comprising the next segment-level language plan and corresponding visual guidance. A World-Action-Progress (WAP) model executes each plan over an unknown duration by jointly predicting action chunks and corresponding execution progress at inference time, calibrating predicted progress through plan-observation alignment for adaptive segment transitions and closed-loop context updates. MaP-WAM keeps the executor context length fixed, while structured attention further enables key-value caching in both planning and execution. MaP-WAM achieves state-of-the-art performance on RMBench with an 83.3% success rate and attains 78.0% success on real-robot tasks, while maintaining approximately constant executor inference latency as task history grows.

Chinese Translation

主流机器人策略通常采用马尔可夫（Markovian）形式，但许多复杂的现实世界操作任务本质上是非马尔可夫的，要求超越当前观察的长期记忆。现有的记忆机制通常依赖于语言摘要、扩展的视觉窗口或两者的结合，因此可能会丧失细粒度的视觉证据，或面临历史覆盖与执行效率之间的权衡。我们提出了MaP-WAM，一个记忆作为计划的框架，它将依赖记忆的世界-行动建模分解为基于记忆的规划和计划条件执行，并使用长期的多模态情节上下文作为规划时的证据，而不是反复将执行者条件化于完整历史。MaP-WAM将记忆表示为包含语言指令和稀疏视觉上下文的完成段记录，并将这种情节记忆转换为紧凑的计划，包括下一个段级语言计划和相应的视觉指导。世界-行动-进展（World-Action-Progress, WAP）模型在未知持续时间内执行每个计划，通过在推理时共同预测动作块和相应的执行进展，利用计划-观察对齐来校准预测进展，以实现自适应段转换和闭环上下文更新。MaP-WAM保持执行者上下文长度固定，同时结构化注意力进一步支持在规划和执行中的关键-值缓存。MaP-WAM在RMBench上实现了最先进的性能，成功率达到83.3%，在真实机器人任务中成功率为78.0%，同时在任务历史增长时保持大致恒定的执行者推理延迟。

View on arXiv Download PDF AI Translation

cs.RO / 28 / 2609.11579

Quasi-static analysis of passive stability in a novel underactuated multi-finger hand

新型欠驱动多指手的准静态被动稳定性分析

Plancoulaine, Léonie, Guégan, Sylvain, Plestan, Franck, Chablat, Damien

Abstract

Underactuated robotic hands achieve adaptive and robust grasping with a reduced number of actuators, but predicting the stable equilibrium pose of the grasped object remains a significant challenge. This paper introduces a quasi-static analytical approach to assess passive stability in underactuated multi-finger hands. A novel three-finger hand architecture integrating a differential spring-loaded slider mechanism is introduced, enabling versatile and adaptive grasping. The study focuses on how the differential mechanism influences the overall grasp behavior and analyzes the effect of object size on the stable equilibrium configurations for two canonical grasp types: cylindrical and spherical.

Chinese Translation

欠驱动机器人手通过减少驱动器数量实现自适应和稳健的抓取，但预测被抓取物体的稳定平衡姿态仍然是一个重大挑战。本文提出了一种准静态分析方法，以评估欠驱动多指手的被动稳定性。介绍了一种新颖的三指手结构，该结构集成了差动弹簧加载滑块机制，从而实现多样化和自适应的抓取。研究重点是差动机制如何影响整体抓取行为，并分析物体尺寸对两种典型抓取类型（圆柱形和球形）的稳定平衡配置的影响。

View on arXiv Download PDF AI Translation

cs.RO / 29 / 2609.11661

Contact-Aware Incremental Model Predictive Control for an Underactuated Aerial Manipulator

面向接触的增量模型预测控制在欠驱动空中操控器上的应用

Liu, Darwin, Keviczky, Tamas, Sun, Sihao

Abstract

We present a robust contact-aware control framework for aerial writing on an underactuated platform. The framework combines nonlinear model predictive control (NMPC) for accurate end-effector position and normal-force tracking at small reference penetration depths, with consistent performance across controller tunings, with whole-body incremental nonlinear dynamic inversion (INDI) for robustness to frictional and aerodynamic disturbances during contact. The proposed controllers are validated on a quadrotor-based aerial manipulator with a rigid, single-link, one-degree-of-freedom (DoF) arm in simulation and real-world experiments. The aerial writing experiments span vertical and inclined surfaces, multiple reference forces, different friction conditions, and wind disturbances. The results demonstrate that robust simultaneous five-DoF end-effector pose and contact-force tracking is achievable on a standard underactuated quadrotor with a simple, rigid, single-link arm, without requiring a fully actuated platform, a complex arm, or dedicated force/torque sensing.

Chinese Translation

我们提出了一种针对欠驱动平台的空中书写的鲁棒接触感知控制框架。该框架结合了非线性模型预测控制（NMPC），以在小参考穿透深度下实现精确的末端执行器位置和法向力跟踪，并在控制器调节中保持一致的性能，同时采用全身增量非线性动态反演（INDI）以增强对接触过程中摩擦和气动干扰的鲁棒性。所提出的控制器在基于四旋翼的空中操控器上进行了验证，该操控器配备了一个刚性单链一自由度（DoF）臂，验证过程包括仿真和实际实验。空中书写实验涵盖了垂直和倾斜表面、多种参考力、不同摩擦条件以及风干扰。结果表明，在标准的欠驱动四旋翼上，使用简单的刚性单链臂可以实现鲁棒的五自由度末端执行器姿态和接触力的同步跟踪，而无需完全驱动的平台、复杂的臂或专用的力/扭矩传感器。

View on arXiv Download PDF AI Translation

cs.RO / 30 / 2609.11697

ActSafeGuard: Differentiable and Training-Aligned Constraint Enforcement for Flow-Matching Policies

ActSafeGuard：用于流匹配策略的可微分和训练对齐的约束执行

Ma, Jianming, Jin, Rongjun, Si, Xiaxi, Zhang, Yang, Li, Yiheng, Gao, Yue

Abstract

Vision-Language-Action (VLA) and World-Action Models (WAMs) have demonstrated strong capabilities in general-purpose robotic manipulation, yet their generated actions may violate hard physical constraints and therefore be unsafe or infeasible for deployment. Existing safety approaches either optimize statistical safety objectives without deterministic per-step guarantees or correct unsafe actions only during inference, creating a mismatch between policy training and execution. We introduce ActSafeGuard, a differentiable and training-aligned safeguard layer for flow-matching based policies. ActSafeGuard integrates hard action feasibility into policy learning, not merely treating safety as an inference-time external component. Through an analytical ray-scaling operator design, ActSafeGuard enables boundary-aware gradients to guide the model to naturally learn constrained manifolds. Extensive experiments on multiple standard foundation backbones ($\pi_{0.5}$ and Fast-WAM) across various tasks demonstrate that ActSafeGuard consistently achieves a $100\%$ step safety rate while fully preserving or even boosting task success rates, providing a scalable and minimally invasive solution for safe embodied AI deployment.

Chinese Translation

视觉-语言-行动（VLA）和世界-行动模型（WAMs）在通用机器人操作中展现了强大的能力，但它们生成的动作可能会违反严格的物理约束，因此在部署时可能不安全或不可行。现有的安全方法要么优化统计安全目标而没有确定性的逐步保证，要么仅在推理期间纠正不安全的动作，从而导致策略训练与执行之间的不匹配。我们提出了ActSafeGuard，一种用于基于流匹配策略的可微分和训练对齐的保护层。ActSafeGuard将严格的动作可行性整合到策略学习中，而不仅仅是在推理时将安全视为外部组件。通过分析光线缩放算子的设计，ActSafeGuard使得边界感知梯度能够引导模型自然学习受约束的流形。在多个标准基础模型（$ ext{π}_{0.5}$ 和 Fast-WAM）上的广泛实验表明，ActSafeGuard始终实现了$100\%$的步骤安全率，同时完全保持甚至提升了任务成功率，为安全的具身人工智能部署提供了一种可扩展且最小侵入的解决方案。

View on arXiv Download PDF AI Translation

cs.RO / 31 / 2609.11733

Reflex-Informed Neuromuscular Reinforcement Learning for Muscle-Driven Locomotion

基于反射的神经肌肉强化学习用于肌肉驱动的运动

Zhou, Jian, Zhang, Xingyu, Ma, Rui, Cao, Yu, Xie, Shane, Zhang, Zhi-qiang

Abstract

Muscle-driven locomotion provides a physically grounded approach to generating realistic human movement. However, achieving both physiological plausibility and adaptability to changes in musculoskeletal capacity and external disturbances remains a fundamental challenge. To address this limitation, we propose a Reflex-Informed Neuromuscular Reinforcement Learning framework for muscle-driven locomotion. Within this framework, a fixed phase-dependent reflex controller serves as the underlying neuromuscular control mechanism, while the reinforcement learning policy produces four biomechanically meaningful residual parameters to modulate key reflex gains and thresholds associated with hip swing, knee support, and ankle propulsion according to the current state. Experimental results demonstrate that the proposed framework generates physiologically plausible locomotion with improved kinematic accuracy and dynamic consistency, as well as better bilateral symmetry and stride-to-stride consistency under nominal walking conditions. The learned policy remains robust under muscle weakness and external perturbations without retraining.

Chinese Translation

肌肉驱动的运动提供了一种基于物理的方式来生成逼真的人类运动。然而，实现生理上的合理性以及对肌肉骨骼能力变化和外部干扰的适应性仍然是一个根本性的挑战。为了解决这一限制，我们提出了一种基于反射的神经肌肉强化学习框架，用于肌肉驱动的运动。在该框架内，一个固定相位依赖的反射控制器作为基础的神经肌肉控制机制，而强化学习策略则生成四个生物力学上有意义的残余参数，以根据当前状态调节与髋部摆动、膝部支撑和踝部推进相关的关键反射增益和阈值。实验结果表明，所提出的框架生成了生理上合理的运动，具有更高的运动学准确性和动态一致性，以及在正常步态条件下更好的双侧对称性和步幅一致性。学习到的策略在肌肉虚弱和外部扰动下保持稳健，无需重新训练。

View on arXiv Download PDF AI Translation

cs.RO / 32 / 2609.11753

SEED-UMI: Sharing the Exoskeleton between human and robot for onE-to-one Dexterous demonstration

SEED-UMI：人类与机器人共享外骨骼以进行一对一灵巧演示

Yu, Tengbo, Wu, Jiahao, Li, Daohan, Chen, Bingxu, Liu, Hao, Ma, Xiaojian, Liu, Hangxin

Abstract

Imitation learning for dexterous hands is bottlenecked by the difficulty of collecting contact-rich demonstrations that transfer faithfully to the robot. Prior wearable-exoskeleton systems record only on the human side and retarget via open-loop mappings calibrated in free space, which degrade under contact. We present SEED-UMI, a framework in which both the human and the robot wear the same exoskeleton: joint encoders become a physically shared measurement, and wrist cameras mounted to the exoskeleton observe the same outer mechanism during both human data collection and robot policy rollouts. This turns retargeting into paired cross-embodiment supervision and lets policies train directly on raw exoskeleton-centric wrist images, without segmentation or inpainting. On five contact-rich tasks, SEED-UMI achieves 3.0x greater data collection efficiency than exoskeleton-based teleoperation and a 70.0% average rollout success rate.

Chinese Translation

灵巧手的模仿学习受到收集丰富接触演示的困难的制约，这些演示需要准确地转移到机器人上。先前的可穿戴外骨骼系统仅在人体侧进行记录，并通过在自由空间中校准的开环映射进行重定向，这在接触情况下会降低效果。我们提出了SEED-UMI，一个框架，其中人类和机器人都穿戴相同的外骨骼：关节编码器成为物理共享的测量，安装在外骨骼上的手腕摄像头在进行人类数据收集和机器人策略执行时观察相同的外部机制。这将重定向转变为配对的跨体现监督，并使策略能够直接在原始以外骨骼为中心的手腕图像上进行训练，而无需分割或修复。在五个接触丰富的任务中，SEED-UMI实现了比基于外骨骼的遥操作高出3.0倍的数据收集效率，以及70.0%的平均执行成功率。

View on arXiv Download PDF AI Translation

cs.RO / 33 / 2609.11766

Visual-SLAM for the detection of hidden tomatoes in greenhouses by Hierarchical Localization and GLOMAPfor robotized harvesting

基于层次定位和GLOMAP的视觉SLAM系统用于温室中隐藏番茄的检测与机器人化采摘

Cañadas-Aránega, Fernando, Moreno, José C., Blanco-Claraco, José L., Rodríguez, Francisco

Abstract

Advanced crop monitoring inside greenhouses is becoming one of the primary objectives of research centers. High-performance sensors, such as LiDAR or stereo cameras, have traditionally been employed for this purpose, though these often have a high cost. This work proposes a Visual-SLAM system using a monocular camera, which is significantly more cost-effective and specifically tailored for agricultural applications, such as mapping tomato crops in a greenhouse. Tests were carried out on a real tomato bunch, located in the Agroconnect experimental greenhouse. A ROS 2 Humble node was developed to run on the robot in order to capture images of these crops, which were then stored for offline processing. To generate a 3D mapped model for the crop in the greenhouse, the GLOMAP mapper, based on Structure-From-Motion, was integrated with the Hierarchical Localization toolbox. This initial mapping is a foundation for future, more advanced algorithms to analyze growth patterns, and optimize agricultural management. The system leverages a hierarchical localization paradigm based on a coarse-to-fine strategy: it first performs global retrieval to generate location hypotheses, then combines local features within the identified candidate regions. The results show a correct identification of the tomato cluster, correctly characterising the tomato that is occluded and inaccessible by classical vision technologies. The reconstructed 3D model was further validated against manual ground-truth measurements of fruit size, centroid position, and orientation, confirming the geometric accuracy of the proposed low-cost monocular pipeline.

Chinese Translation

温室内的先进作物监测正成为研究中心的主要目标之一。传统上，LiDAR或立体相机等高性能传感器被用于此目的，但这些传感器通常成本较高。本研究提出了一种使用单目相机的视觉SLAM系统，该系统在成本上更具优势，并专门针对农业应用进行优化，例如在温室中绘制番茄作物的地图。测试在位于Agroconnect实验温室的真实番茄串上进行。开发了一个ROS 2 Humble节点，以便在机器人上运行，捕获这些作物的图像，并将其存储以供离线处理。为了生成温室中作物的3D映射模型，基于运动结构的GLOMAP映射器与层次定位工具箱相结合。该初步映射为未来更高级算法分析生长模式和优化农业管理奠定了基础。该系统利用基于粗到细策略的层次定位范式：首先执行全局检索以生成位置假设，然后在识别的候选区域内结合局部特征。结果显示正确识别了番茄簇，准确表征了被经典视觉技术遮挡和无法接触的番茄。重建的3D模型进一步与手动地面真实测量的果实大小、质心位置和方向进行了验证，确认了所提低成本单目管道的几何准确性。

View on arXiv Download PDF AI Translation

cs.RO / 34 / 2609.11775

Rapid Learning of Dexterous In-Hand Pen Writing through Real-Time Jacobian Estimation

通过实时雅可比估计实现灵巧的手内笔写作的快速学习

Stewart, Kai, Toshimitsu, Yasunori, Katzschmann, Robert K.

Abstract

Dexterous in-hand manipulation of a grasped object with an anthropomorphic hand is an unsolved frontier for robot dexterity. The contact-richness and highly dynamic nature of object-hand interactions tend to require extensive modeling or data-collection efforts for learning-based approaches. Modern simulators used for reinforcement learning (RL) cannot fully replicate the required contact complexity, while collecting dexterous demonstrations for imitation learning (IL) remains an open problem. In this research, we present an embodied control approach based on real-time task Jacobian estimation of the combined hand and object system on the physical robot. Using only the CPU on a laptop, the proposed controller begins in-hand pen writing after approximately 18 s of initialization and continues to adapt online, without an analytic hand--object kinematic/contact model, simulation training, or precollected task demonstrations. We demonstrate that the same estimator/controller formulation works on three anthropomorphic robotic hand systems (one physical, two simulated) to show human-like, in-hand articulation of a grasped pen by an embodiment-independent formulation. Sub-millimeter in-plane precision (mean 0.6 mm across runs) is achieved across letters and shapes written in the air and on paper on a physical robot. To our knowledge, this is the first demonstration of an anthropomorphic hand writing arbitrary single-stroke trajectories with a grasped pen through purely in-hand motion, and it showcases an alternative to compute- and data-heavy approaches such as RL and IL for achieving dexterous manipulation through computationally simple and data-efficient algorithms.

Chinese Translation

使用类人手对抓握物体进行灵巧的手内操作是机器人灵巧性尚未解决的前沿问题。物体与手之间的接触丰富性和高度动态的交互特性往往需要大量的建模或数据收集工作，以支持基于学习的方法。现代用于强化学习（Reinforcement Learning, RL）的模拟器无法完全复制所需的接触复杂性，而收集灵巧演示以进行模仿学习（Imitation Learning, IL）仍然是一个未解决的问题。在本研究中，我们提出了一种基于实时任务雅可比估计的具身控制方法，适用于物理机器人上的手与物体系统的结合。该控制器仅使用笔记本电脑的CPU，在约18秒的初始化后开始进行手内笔写作，并持续在线适应，无需解析的手-物体运动/接触模型、模拟训练或预先收集的任务演示。我们展示了相同的估计器/控制器公式在三种类人机器人手系统（一个物理的，两个模拟的）上均有效，展示了通过具身独立的公式实现抓握笔的类人手内关节运动。在物理机器人上，书写的字母和形状在空气中和纸上均达到了亚毫米级的平面精度（平均为0.6毫米）。据我们所知，这是首次展示类人手通过纯手内运动使用抓握笔书写任意单笔轨迹，展示了一种替代计算和数据密集型方法（如RL和IL）以实现灵巧操作的计算简单且数据高效的算法。

View on arXiv Download PDF AI Translation

cs.RO / 35 / 2609.11871

Learning Agent-based Model Predictive Control for Holistic Vehicle Performance

基于智能体的模型预测控制学习以实现整体车辆性能

Zhong, Jiaming, Mehrizi, Reza Valiollahi, Pirani, Mohammad, Yu, Chao, Kasaiezadeh, Alireza, Pant, Yash Vardhan, Khajepour, Amir

Abstract

Agent-based model predictive control (AMPC) has recently been proposed as a distributed scheme that collaborates with all agents to achieve optimal holistic performance. However, its optimality highly depends on the prediction accuracy that requires all agents or their contributions to be known, which is too idealistic for actual implementation. This research proposes a novel practical hybrid control scheme - learning agent-based MPC (LAMPC), combining the model-based AMPC approach and data-based learning methods to improve the holistic vehicle performance for multi-agent systems. The Gaussian process regression (GPR) enhanced by an online data management strategy serves as the learning core to predict unknown contributions. A novel multi-step prediction mechanism leverages the GPR learning potential along the horizon. The predicted mean, representing the learned unknown contributions, completes the system model in the MPC for more accurate control. Meanwhile, a stochastic framework is formulated to guarantee control safety and feasibility using soft chance constraints based on the prediction variance. Both simulations and experiments show that, with the learning capability, LAMPC outperforms the traditional AMPC. LAMPC can achieve higher tracking performance in well-learned scenarios and always guarantee constraint satisfaction even in less-learned scenarios. Moreover, the proposed hybrid control scheme is efficient for real-time implementation and is flexible to any control agent topology.

Chinese Translation

基于智能体的模型预测控制（AMPC）最近被提出作为一种分布式方案，通过所有智能体的协作来实现最佳的整体性能。然而，其最优性高度依赖于预测精度，这要求所有智能体或其贡献必须是已知的，这对于实际实施来说过于理想化。本研究提出了一种新颖的实用混合控制方案——学习型基于智能体的模型预测控制（LAMPC），结合了基于模型的AMPC方法和基于数据的学习方法，以提高多智能体系统的整体车辆性能。增强的高斯过程回归（GPR）结合在线数据管理策略作为学习核心，用于预测未知贡献。一个新颖的多步预测机制利用GPR的学习潜力进行前瞻性预测。预测均值代表学习到的未知贡献，补全了MPC中的系统模型，从而实现更精确的控制。同时，构建了一个随机框架，以基于预测方差使用软机会约束来保证控制的安全性和可行性。仿真和实验结果表明，凭借学习能力，LAMPC的性能优于传统的AMPC。在经过良好学习的场景中，LAMPC能够实现更高的跟踪性能，并且即使在学习较少的场景中也始终保证约束的满足。此外，所提出的混合控制方案在实时实施中高效，并且对任何控制智能体拓扑具有灵活性。

View on arXiv Download PDF AI Translation

cs.RO / 36 / 2609.11875

UniMPA: A Unified Memory-Prediction-Action Model via Action-Grounded Transition Modeling

UniMPA：通过基于动作的转移建模实现统一的记忆-预测-行动模型

Li, Wei, Shao, Rui, He, Jie, Zhang, Lingsen, Liu, Ziwei, Nie, Liqiang

Abstract

Recent advances in Vision-Language-Action (VLA) models have improved robotic manipulation, yet observation-to-action learning remains limited by a fundamental transition realizability gap, manifested in three tightly coupled problems: (i) Transition ambiguity. Visually similar current observations may correspond to different manipulation phases and imply different subsequent transitions. (ii) Prediction--execution mismatch. A visually plausible predicted future observation does not necessarily correspond to a physically realizable transition. (iii) Experience--realization mismatch. A historically executable action pattern may not necessarily realize the intended transition in the current scene and therefore requires context-aware adaptation. Accordingly, we propose UniMPA, a Unified Memory-Prediction-Action model that addresses these problems through a shared action-grounded transition interface. (i) UniMPA introduces Persistent-Selective Future Prediction to resolve transition ambiguity by modeling the intended future state evolution. A persistent latent stream continuously tracks task-level progress, while a transition-critical pixel stream selectively resolves fine-grained interaction changes through memory-grounded prediction. (ii) To assess the physical executability of the anticipated transition, the predicted transition queries a temporal Visual-Action Memory Bank. The bank retrieves historically realized visual-action experience, grounding future prediction in executable evidence. (iii) To adapt executable experience to the current scene, an Action-Visual Memory Bank retrieves visually grounded action prototypes from historical action evolution. Prototype-Biased Flow then shifts the flow source toward a historically supported action manifold for context-aware refinement.

Chinese Translation

近年来，视觉-语言-行动（VLA）模型的进展改善了机器人操作能力，但观察到行动的学习仍受到基本转移可实现性差距的限制，这体现在三个紧密相关的问题上：（i）转移模糊性。视觉上相似的当前观察可能对应于不同的操作阶段，并暗示不同的后续转移。（ii）预测与执行不匹配。视觉上合理的未来观察预测不一定对应于物理上可实现的转移。（iii）经验与实现不匹配。历史上可执行的行动模式不一定能在当前场景中实现预期的转移，因此需要上下文感知的适应。为此，我们提出了UniMPA，一个统一的记忆-预测-行动模型，通过共享的基于动作的转移接口来解决这些问题。（i）UniMPA引入了持久选择性未来预测，通过建模预期的未来状态演变来解决转移模糊性。一个持久的潜在流持续跟踪任务级进展，而一个转移关键像素流通过基于记忆的预测选择性地解决细粒度的交互变化。（ii）为了评估预期转移的物理可执行性，预测的转移查询一个时间视觉-行动记忆库。该库检索历史上实现的视觉-行动经验，将未来预测基于可执行证据进行基础化。（iii）为了将可执行经验适应于当前场景，一个行动-视觉记忆库从历史行动演变中检索视觉基础的行动原型。原型偏置流随后将流源转向一个历史上支持的行动流形，以实现上下文感知的细化。

View on arXiv Download PDF AI Translation

cs.RO / 37 / 2609.11920

EVPeriscope: Extended Perception across Aerial and Ground Vehicles with Event-based Propeller Tracking

EVPeriscope：基于事件的螺旋桨跟踪技术在空中和地面车辆之间的扩展感知

Ong, Dexter, Kumar, Vijay, Chaudhari, Pratik

Abstract

Reliable relative localization between aerial and ground robots is a key requirement for tightly coordinated heterogeneous teams. This can be difficult to do using conventional frame-based cameras and fiducial markers because they are sensitive to motion blur, lighting variations, and payload constraints. This paper presents EVPeriscope, an event-based perception system that enables detection, localization and control of a quadrotor using an upward-facing event camera on a ground robot by detecting the high-frequency visual signature of its propellers. This system allows the quadrotor to function as an extended perception system for the ground robot when onboard sensors exhibit degradation or occlusion. We demonstrate the capabilities of this marsupial ground-aerial system via experiments in challenging field conditions with wind speeds of up to 15 mph, in both daylight and at night. We show that the system supports localization and closed-loop navigation through dense foliage where the ground robot's sensors are occluded. Our control system for the quadrotor operates at 200 Hz entirely with onboard sensing and computation. More details and experiment videos can be found on the project page: https://ongdexter.github.io/evperiscope.

Chinese Translation

可靠的空中和地面机器人之间的相对定位是紧密协调的异构团队的关键需求。使用传统的基于帧的摄像头和标志物进行定位可能会很困难，因为它们对运动模糊、光照变化和载荷限制非常敏感。本文提出了EVPeriscope，一个基于事件的感知系统，通过检测四旋翼无人机螺旋桨的高频视觉特征，使地面机器人上的向上摄像的事件相机能够实现四旋翼无人机的检测、定位和控制。当机载传感器出现退化或遮挡时，该系统允许四旋翼无人机作为地面机器人的扩展感知系统。我们通过在风速高达15英里每小时的挑战性现场条件下进行实验，展示了这一袋鼠式地面-空中系统的能力，实验在白天和夜间进行。我们表明，该系统支持在地面机器人传感器被遮挡的情况下，通过密集的植被进行定位和闭环导航。我们的四旋翼无人机控制系统以200 Hz的频率完全依赖于机载传感和计算。更多细节和实验视频可以在项目页面找到：https://ongdexter.github.io/evperiscope。

View on arXiv Download PDF AI Translation

计算机视觉 (Computer Vision)

cs.CV / 1 / 2609.10572

Rethinking Handwritten Character Recognition

重新思考手写字符识别

Raut, Ranjit, Subedi, Aarav, Shrestha, Ashim

Abstract

Non-Latin handwritten character recognition (HCR) remains understudied. Dominant methods consider it as generic image classification, which uses model scale to implicitly learn stroke structure. Structural-prior efficiency---the principle that explicitly encoding script-geometric regularities as architectural inductive biases can be both more accurate and require fewer parameters. We introduce GraphemeNet, a unified multi-script architecture, governed by two orthogonal binary axes. Axis 1 operationalises stroke-level geometric regularity via Persistent Scaffold Injection (PSI): a script-specific asymmetric convolution injects a stroke scaffold as a weighted residual at every encoder stage, continuously anchoring learned features to script geometry---distinct from skip connections, auxiliary losses, or attention reweighting. Axis 2 selects between global average pooling with gated fusion and cross-scale attention with a Stroke Topology Module (STM), depending on whether glyph discrimination requires spatial relational reasoning. A Linear Capsule Routing (LCR) with $O(n)$ routing is shared universally. On fourteen benchmarks across eight writing systems, the architecture generalises with only scaffold and decoder topology varying per script, consistently challenging, outperforming published baselines, and establishing structural-prior efficiency as a broadly applicable principle for multi-script HCR.

Chinese Translation

非拉丁手写字符识别（HCR）仍然是一个研究不足的领域。主流方法将其视为通用图像分类，利用模型规模隐式学习笔画结构。结构先验效率——这一原则明确编码书写几何规律作为架构归纳偏置，可以在更高的准确性和更少的参数需求之间取得平衡。我们引入了GraphemeNet，一种统一的多书写系统架构，由两个正交的二元轴控制。轴1通过持久支架注入（Persistent Scaffold Injection, PSI）实现笔画级几何规律：一种特定于书写系统的非对称卷积在每个编码器阶段将笔画支架作为加权残差注入，持续将学习到的特征锚定到书写几何上——这与跳跃连接、辅助损失或注意力重加权不同。轴2在全局平均池化与门控融合和跨尺度注意力与笔画拓扑模块（Stroke Topology Module, STM）之间进行选择，具体取决于字形区分是否需要空间关系推理。一个具有$O(n)$路由的线性胶囊路由（Linear Capsule Routing, LCR）被普遍共享。在八个书写系统的十四个基准测试中，该架构仅在支架和解码器拓扑上因书写系统而异，表现出良好的泛化能力，持续挑战并超越已发布的基准，确立了结构先验效率作为多书写系统HCR的广泛适用原则。

View on arXiv Download PDF AI Translation

cs.CV / 2 / 2609.10723

AcFlow: Controlling Text-to-Image Diffusion Transformers via Learned Conditional Activation Flow

AcFlow：通过学习的条件激活流控制文本到图像扩散变换器

Wang, Junran, Jin, Zehao, Luan, Tianyu, Shen, Xinjie

Abstract

Text-to-image diffusion transformers (DiTs) are powerful generators, yet direct prompting provides limited control interface for style intensity and can fail to suppress unwanted concepts. To enable these controls, we introduce AcFlow, an inference-time controller that transports intermediate layer image-token activations through a learned concept-conditioned velocity field while keeping the base DiT frozen. A textual concept description specifies the desired intervention, while the integration horizon provides a continuous control parameter. The field produces token-varying, activation-dependent updates. With parameters shared across concepts within each task family, the field supports fine-grained descriptions and generalizes to concepts unseen during training without per-concept fitting. On style control, AcFlow achieves the best style--content trade-off among the evaluated baselines in the high-style-alignment regime. At a fixed operating point, AcFlow attains style--content alignment of 0.5365/0.2860, compared with 0.4397/0.2684 for the baseline with the highest style alignment. Qualitative results demonstrate suppression of diverse concepts, including cases where direct prompting fails. Our analyses support the learned velocity field as an adaptive control mechanism, with update directions varying across tokens and depend on their activation states. Our code is available at https://github.com/Nove1yst/AcFlow.

Chinese Translation

文本到图像的扩散变换器（DiTs）是强大的生成模型，但直接提示提供的风格强度控制界面有限，并且可能无法抑制不必要的概念。为实现这些控制，我们引入了AcFlow，这是一种推理时控制器，通过学习的概念条件速度场传输中间层图像标记激活，同时保持基础DiT不变。文本概念描述指定所需的干预，而集成视野提供了一个连续的控制参数。该速度场生成标记变化、依赖于激活的更新。在每个任务家族中共享参数的速度场支持细粒度描述，并且能够推广到训练期间未见过的概念，而无需针对每个概念进行拟合。在风格控制方面，AcFlow在高风格对齐范围内实现了评估基线中最佳的风格-内容权衡。在固定操作点上，AcFlow达到了0.5365/0.2860的风格-内容对齐，相比之下，具有最高风格对齐的基线为0.4397/0.2684。定性结果展示了对多种概念的抑制，包括直接提示失败的情况。我们的分析支持学习的速度场作为一种自适应控制机制，更新方向在标记之间变化，并依赖于它们的激活状态。我们的代码可在 https://github.com/Nove1yst/AcFlow 获取。

View on arXiv Download PDF AI Translation

cs.CV / 3 / 2609.10743

MHE-Former: Multi-Hypothesis Transformers via Entropy Maximization for 3D Mesh Recovery

MHE-Former：通过熵最大化的多假设变换器用于三维网格恢复

Jia, Boshu, Chen, Rongyu, Yang, Linlin, Liu, Zihao, Chen, Yingjie, Zhang, Zhongqun, Tao, Zhulin, Lin, Shaohui, Wu, Xiaoyu, Jin, Libiao, Zhang, Baochang, Yao, Angela

Abstract

Monocular 3D hand and body mesh recovery often suffers from severe occlusion and ambiguity. Traditional deterministic methods typically regress a single optimal solution, leading to overconfident predictions. In this paper, we introduce an exploration--exploitation paradigm for ambiguous mesh recovery with multi-hypothesis learning and selection. Specifically, during exploration, based on our probabilistic formulation and entropy maximization, we propose a novel multi-hypothesis method referred to as MHE-Former. It is a Transformer-based multi-hypothesis framework, ensuring high training efficiency and label friendliness while generating plausible and diverse hypotheses. During exploitation, we propose Hypothesis Selection, a context-aware process for multiple predictions. Especially leveraging VLM's powerful visual understanding and reasoning capabilities, it allows users to choose the most plausible and desired estimate with additional evidence and natural language intent. Extensive experiments demonstrate that our framework achieves state-of-the-art performance in accuracy and diversity across multiple datasets. The user preference study further shows the practicality of our hypothesis selection process.

Chinese Translation

单目三维手部和身体网格恢复常常面临严重的遮挡和模糊性。传统的确定性方法通常回归单一的最优解，导致过于自信的预测。本文提出了一种探索-利用范式，用于模糊网格恢复，结合多假设学习和选择。具体而言，在探索阶段，基于我们的概率模型和熵最大化，我们提出了一种新颖的多假设方法，称为 MHE-Former。它是一个基于变换器的多假设框架，确保高效的训练和友好的标签生成，同时产生合理且多样的假设。在利用阶段，我们提出了假设选择，这是一种上下文感知的多重预测过程。特别是利用 VLM 强大的视觉理解和推理能力，它允许用户根据额外证据和自然语言意图选择最合理和期望的估计。大量实验表明，我们的框架在多个数据集上实现了准确性和多样性的最先进性能。用户偏好研究进一步展示了我们假设选择过程的实用性。

View on arXiv Download PDF AI Translation

cs.CV / 4 / 2609.10749

Meta-Learning for Data-Efficient Plant Growth Estimation via Vision Transformers and Fuzzy Clustering

基于视觉变换器和模糊聚类的数据高效植物生长估计的元学习

Elahi, Sheikh Hasan, Dewage, Rusith Chamara Hathurusinghe, Ullah, Habib, Siddiqui, Muhammad Salman, Islam, Rakibul, Machot, Fadi Al

Abstract

Accurate plant growth estimation is essential for greenhouse monitoring, yet obtaining labeled data remains costly and time-consuming. To address this, we propose a few-shot regression framework that combines Vision Transformer (ViT) feature embeddings, clustering-based task construction, and gradient-based meta-learning, and show that task construction in embedding space is a primary driver of performance. The approach leverages an unlabeled image pool to organize data into structured tasks using fuzzy c-means clustering, enabling efficient learning from a small number of labeled samples. We systematically evaluate meta-learning methods and show that second-order methods (e.g., Model-Agnostic Meta-Learning variants such as MAML++) outperform classical baselines in the few-shot regime. Furthermore, intra-cluster support selection has a limited and dataset-dependent impact. Experiments on two plant datasets show that structured task design combined with meta-learning enables reliable plant growth estimation under severe label scarcity.

Chinese Translation

准确的植物生长估计对于温室监测至关重要，但获取标记数据仍然成本高昂且耗时。为了解决这一问题，我们提出了一种结合视觉变换器（Vision Transformer, ViT）特征嵌入、基于聚类的任务构建和基于梯度的元学习的少样本回归框架，并展示了嵌入空间中的任务构建是性能的主要驱动因素。该方法利用未标记图像池，通过模糊C均值聚类将数据组织成结构化任务，从而能够从少量标记样本中高效学习。我们系统地评估了元学习方法，并表明二阶方法（如模型无关元学习（Model-Agnostic Meta-Learning）变体如MAML++）在少样本环境中优于经典基线。此外，簇内支持选择的影响有限且依赖于数据集。在两个植物数据集上的实验表明，结构化任务设计结合元学习能够在严重标签稀缺的情况下实现可靠的植物生长估计。

View on arXiv Download PDF AI Translation

cs.CV / 5 / 2609.10756

GRADE: Single-Frame Generative Radar Depth Estimation Under Visual Degradation

GRADE：在视觉退化条件下的单帧生成雷达深度估计

Zhao, Bin, Chiou, Patrick, Garg, Nakul

Abstract

Dense 3D depth perception fails under smoke, fog, and darkness because optical sensors cannot penetrate airborne particulates. mmWave radar remains usable and measures range accurately under these conditions, but its small aperture limits angular resolution. We present GRADE, which grounds a pretrained generative prior in single-frame radar geometry to estimate high-fidelity metric depth. GRADE first maps raw 4D radar spectra to coarse metric depth. A latent diffusion backbone then recovers structural detail while conditioning every denoising step on this estimate. A pixel-space adapter uses residual camera cues when available and is trained across clear, smoke-degraded, and occluded inputs so the full output approaches the radar-conditioned path as visibility degrades. Trained and evaluated on ~95K frames across 12 buildings with real smoke, GRADE achieves an MAE of 0.303 m in clear scenes and 0.313 m under smoke, outperforming existing baselines. Code and datasets are available at https://phi-lab-rice.github.io/GRADE.

Chinese Translation

在烟雾、雾霾和黑暗条件下，密集的三维深度感知失效，因为光学传感器无法穿透空气中的颗粒物。毫米波雷达在这些条件下仍然可用，并能准确测量距离，但其小孔径限制了角分辨率。我们提出了GRADE，它将预训练的生成先验与单帧雷达几何结合，以估计高保真度的度量深度。GRADE首先将原始4D雷达光谱映射到粗略的度量深度。然后，潜在扩散骨干网络在每个去噪步骤中以此估计为条件，恢复结构细节。当可用时，像素空间适配器利用残余相机线索，并在清晰、烟雾退化和遮挡输入上进行训练，以便在能见度降低时，完整输出接近雷达条件路径。在12栋建筑的约95K帧真实烟雾数据上进行训练和评估，GRADE在清晰场景中实现了0.303米的平均绝对误差（MAE），在烟雾条件下为0.313米，超越了现有基准。代码和数据集可在https://phi-lab-rice.github.io/GRADE获取。

View on arXiv Download PDF AI Translation

cs.CV / 6 / 2609.10787

Shedding Light: A Benchmark for Evaluating Lighting Understanding in Generative Image Models

揭示光照：评估生成图像模型中光照理解的基准

Giroux, Justine, Hilliard, Jack Oliver, Hold-Geoffroy, Yannick, Vazquez-Corral, Javier, Lalonde, Jean-François

Abstract

Accurate modelling of illumination is central to realistic image synthesis and scene understanding. Yet, there is little exploration into whether image generative models are good at this task or whether physical plausibility remains a key challenge for them. Clearly, significant progress has been made in realistic image synthesis, but do models truly understand lighting in a physically accurate manner? To answer this question, this work proposes a benchmark to assess the lighting understanding and harmonisation capabilities of generative models. Our key insight is that evaluating lighting understanding for such models only requires testing how well they insert novel objects into real photographs whilst maintaining consistent illumination. To do so, we use a multi-illumination dataset with images containing simple objects serving as ``light probes'', and prompt models to inpaint the same object onto the original image, then compare the generated results against the ground-truth light probes. We then estimate the lighting direction, colour and radiance distribution from the inpainted probes, providing a quantitative measure of illumination accuracy and photometric realism. Our work establishes a scalable evaluation protocol to systematically assess how well generative models capture and reproduce real-world lighting, offering a foundation for benchmarking the photometric accuracy of any future models. All code and data are available at https://lvsn.github.io/SheddingLight/ .

Chinese Translation

准确的光照建模对于现实图像合成和场景理解至关重要。然而，目前对图像生成模型在这一任务中的表现以及物理合理性是否仍然是其主要挑战的探索较少。显然，现实图像合成已经取得了显著进展，但模型是否真正以物理准确的方式理解光照？为了解答这一问题，本研究提出了一个基准，以评估生成模型的光照理解和协调能力。我们的关键见解是，评估此类模型的光照理解仅需测试它们在保持一致光照的情况下，将新物体插入真实照片的能力。为此，我们使用一个多光照数据集，其中包含简单物体作为“光探针”的图像，并提示模型将相同的物体修复到原始图像上，然后将生成的结果与真实的光探针进行比较。接着，我们从修复的探针中估计光照方向、颜色和辐射分布，提供光照准确性和光度真实感的定量度量。我们的工作建立了一个可扩展的评估协议，以系统地评估生成模型捕捉和再现现实世界光照的能力，为未来任何模型的光度准确性基准测试提供了基础。所有代码和数据可在 https://lvsn.github.io/SheddingLight/ 获取。

View on arXiv Download PDF AI Translation

cs.CV / 7 / 2609.10789

Two-Parameter Flow Map Learning for Continuous-Time Diffeomorphic Image Registration

基于双参数流图学习的连续时间微分同胚图像配准

Matinkia, Mohammadjavad, Ray, Nilanjan

Abstract

Diffeomorphic image registration is central to medical image analysis, enabling anatomically consistent alignment across subjects. Most learning-based diffeomorphic methods model autonomous ODEs(ordinary differential equations) by parameterizing a stationary velocity field and recovering deformations via scaling-and-squaring. While non-autonomous ODEs with time-dependent velocities increase expressiveness, existing approaches rely on numerical integration to implicitly enforce flow structure that entangles model expressiveness with discretization accuracy. We propose a framework to directly learn the continuous-time solution of a non-autonomous ODE formulated as a two-parameterflow map. By enforcing cocycle consistency, a fundamental structural property of time-varying flows, we learn the flow maps without time discretization and velocity integration during training. The framework recovers diffeomorphic mappings at inference using a small number of compositions. Our proposed framework seamlessly incorporates standard registration backbones and improves alignment accuracy consistently across nine datasets while preserving diffeomorphic structure. Notably, the proposed method achieves an average Dice improvement of 2.1% on brain MRI benchmarks, a 12% TRE reduction on lung CT, and a 2.6% Dice gain on cardiac MRI and ultrasound datasets.

Chinese Translation

微分同胚图像配准在医学图像分析中至关重要，能够实现不同受试者之间的解剖一致性对齐。大多数基于学习的微分同胚方法通过参数化静态速度场来建模自主常微分方程（ODE），并通过缩放与平方恢复变形。虽然具有时间依赖速度的非自主常微分方程增加了表达能力，但现有方法依赖数值积分来隐式强制流结构，这使得模型的表达能力与离散化精度相互纠缠。我们提出了一个框架，直接学习以双参数流图形式表述的非自主常微分方程的连续时间解。通过强制循环一致性这一时间变化流的基本结构属性，我们在训练过程中无需时间离散化和速度积分来学习流图。该框架在推理时使用少量组合恢复微分同胚映射。我们提出的框架无缝地结合了标准配准骨干网络，并在九个数据集上持续提高对齐精度，同时保持微分同胚结构。值得注意的是，所提方法在脑部MRI基准测试中实现了平均Dice系数提高2.1%，在肺部CT上减少了12%的目标重定位误差（TRE），在心脏MRI和超声数据集上获得了2.6%的Dice增益。

View on arXiv Download PDF AI Translation

cs.CV / 8 / 2609.10801

How Much Velocity Does Off-Ball Space Value Need? A Broadcast-Viewport Benchmark

离球空间价值需要多少速度？广播视口基准测试

Choi, Seongjin

Abstract

Velocity-aware pitch control is standard, but under a broadcast viewport half the players are off screen and on-screen velocities come from a drifting calibration. We ask at which layer of broadcast off-ball analysis velocity changes the answer. Inheriting our off-screen imputation protocol (three Metrica matches, 44 m viewport, block-bootstrap CIs), we score four velocity regimes -- none, viewport-legal observed, true-for-visible, true-for-all -- against a velocity-aware ground truth at three layers: imputation, the control surface, and team verdicts. Velocity is nearly useless for imputation (-0.2 pp against a 12--14 pp velocity-free surface MAE), first-order for the surface (-1.5 to -1.8 pp, 11--15% of that MAE), and ten times smaller for verdicts (-0.12 to -0.19 pp). The velocity that matters is the visible channel: perfect occluded-player velocity adds 2--6% of the visible gain, and no last-seen decay policy we tested exceeds that. Omitting velocity blurs the surface (per-frame |e| 2.2--2.6 pp) with small time-averaged bias (per cell <=0.4 pp), whereas imputation error is a structured bias against the defending team's deep zone (5--9 pp). At a fixed velocity window, a noise ladder of eleven jitter settings, including sigma_v-matched pairs, is ordered to first order by one velocity-noise axis sigma_v with break-even ~1 m/s; eleven SoccerNet-GSR clips from one match through our pipeline measure sigma_v=1.65 m/s yet recover 24--36% of the benefit: 43% of the variance is frame-common, which the surface tolerates, and the residual is heavy-tailed and clustered, which Gaussian controls matched on component RMS do not reproduce (+0.03 vs. +0.36). The share of velocity-free error that velocity removes grows with viewport width (7% at 36 m, 21% at 60 m): fix imputation on tight shots, velocity on wide ones. Code and logs are released.

Chinese Translation

速度感知的场地控制是标准做法，但在广播视口下，半数球员处于屏幕外，屏幕内的速度来自于漂移的校准。我们探讨在广播离球分析的哪个层面上，速度的变化会影响结果。继承我们的屏幕外插补协议（三个 Metrica 比赛，44 米视口，区块自助法置信区间），我们对四种速度模式进行评分——无速度、视口合法观察到的速度、可见的真实速度和所有的真实速度——并在三个层面上与速度感知的真实值进行比较：插补、控制面和团队判决。速度对于插补几乎没有用处（-0.2 个百分点，相较于 12-14 个百分点的无速度表面平均绝对误差），对控制面是第一阶影响（-1.5 到 -1.8 个百分点，占该平均绝对误差的 11-15%），而对判决则小十倍（-0.12 到 -0.19 个百分点）。重要的速度是可见通道：完美的被遮挡球员速度增加了可见增益的 2-6%，而我们测试的没有最后见到衰减政策均未超过这一点。省略速度会模糊控制面（每帧 |e| 2.2-2.6 个百分点），并伴随小的时间平均偏差（每个单元 <=0.4 个百分点），而插补误差则是对防守队伍深区的结构性偏差（5-9 个百分点）。在固定的速度窗口下，包含 sigma_v 匹配对的十一种抖动设置的噪声阶梯，按一条速度-噪声轴 sigma_v 的一阶顺序排列，平衡点约为 1 米/秒；通过我们的流程，从一场比赛中提取的十一段 SoccerNet-GSR 片段测得 sigma_v=1.65 米/秒，但恢复了 24-36% 的收益：43% 的方差是帧共同的，控制面可以容忍，而残差则是重尾和聚集的，基于组件 RMS 匹配的高斯控制无法重现（+0.03 对 +0.36）。随着视口宽度的增加，速度去除的误差比例也在增长（在 36 米时为 7%，在 60 米时为 21%）：在紧凑镜头上固定插补，在宽镜头上应用速度。代码和日志已发布。

View on arXiv Download PDF AI Translation

cs.CV / 9 / 2609.10806

TrajFusionNet+: Transformer-Based Prediction of Pedestrian Crossing Intention via Fusion of Trajectory Representations and Scene Graphs

TrajFusionNet+: 基于变换器的行人过马路意图预测，通过轨迹表示与场景图的融合

Landry, François G., Akhloufi, Moulay A.

Abstract

The pedestrian crossing intention task involves predicting whether pedestrians are likely to cross the road from the point of view of an autonomous vehicle. We introduce TrajFusionNet+, a novel transformer-based model for pedestrian crossing intention prediction. TrajFusionNet+ combines sequential and visual representations of pedestrian trajectory with a graph-based representation of the scene context in order to predict pedestrian crossing intention. The proposed architecture builds upon our previous model, TrajFusionNet, and comprises three branches: a Sequence Attention Module (SAM), which processes a sequential representation of past and predicted pedestrian trajectories; a Visual Attention Module (VAM), which utilizes a visual representation of the pedestrian trajectories by overlaying observed and predicted bounding boxes onto scene images; and a Graph Attention Module (GAM), which extracts pedestrian-centric graphs from segmented scene images and captures the relational dependencies between pedestrians and traffic elements. TrajFusionNet+ achieves improved state-of-the-art performance on the two most widely used pedestrian crossing intention datasets, PIE and JAAD. Furthermore, we introduce a new evaluation protocol in which models are trained jointly on the PIE and JAAD datasets but evaluated separately on each. Under this setting, TrajFusionNet+ demonstrates superior generalization compared to existing approaches.

Chinese Translation

行人过马路意图任务涉及从自动驾驶车辆的视角预测行人是否可能过马路。我们提出了TrajFusionNet+，一种新颖的基于变换器的行人过马路意图预测模型。TrajFusionNet+结合了行人轨迹的序列表示和视觉表示，以及场景上下文的图形表示，以预测行人过马路的意图。该架构基于我们之前的模型TrajFusionNet，包含三个分支：序列注意力模块（Sequence Attention Module, SAM），处理过去和预测的行人轨迹的序列表示；视觉注意力模块（Visual Attention Module, VAM），通过将观察到的和预测的边界框叠加到场景图像上，利用行人轨迹的视觉表示；以及图形注意力模块（Graph Attention Module, GAM），从分割的场景图像中提取以行人为中心的图形，并捕捉行人与交通元素之间的关系依赖。TrajFusionNet+在两个最广泛使用的行人过马路意图数据集PIE和JAAD上实现了改进的最先进性能。此外，我们引入了一种新的评估协议，在该协议下模型在PIE和JAAD数据集上联合训练，但在每个数据集上分别评估。在这种设置下，TrajFusionNet+表现出比现有方法更优越的泛化能力。

View on arXiv Download PDF AI Translation

cs.CV / 10 / 2609.10811

Overpainting: Localized Context-aware Diffusion Image Editing

重绘：局部上下文感知扩散图像编辑

Sartor, Sam, Georgiev, Iliyan, Fischer, Michael, Deschaintre, Valentin, Peers, Pieter

Abstract

We present "overpainting", an image editing operation which offers both control over the location of the edit and awareness of the previous content in that location. The overpainted area is given by a trimap, where white-annotated pixels must be edited, gray-annotated pixels may be edited, and black-annotated pixels must not be edited. This enables both precise and loose control, depending on user intent. We implement overpainting by adapting a pretrained image editing diffusion model using a combination of joint attention and low-rank adaption across input images with attention-dropout to balance the information flow between noise, source and mask images. We present a novel, automated, training data generation pipeline that (1) generates a set of candidate image pairs leveraging existing language-based editing models, (2) carefully curates those pairs, and (3) extracts a trimap from each usable pair. We demonstrate the versatility of our overpainting model on a wide range of editing tasks.

Chinese Translation

我们提出了“重绘”（overpainting），这是一种图像编辑操作，既能控制编辑的位置，又能考虑该位置之前的内容。重绘区域由三分图（trimap）定义，其中白色标注的像素必须被编辑，灰色标注的像素可以被编辑，而黑色标注的像素则不得被编辑。这使得用户可以根据意图进行精确或宽松的控制。我们通过调整一个预训练的图像编辑扩散模型来实现重绘，该模型结合了联合注意力（joint attention）和低秩适应（low-rank adaptation），并在输入图像中使用注意力丢弃（attention-dropout）来平衡噪声、源图像和掩膜图像之间的信息流。我们提出了一种新颖的自动化训练数据生成管道，该管道（1）利用现有的基于语言的编辑模型生成一组候选图像对，（2）仔细策划这些图像对，以及（3）从每个可用的图像对中提取三分图。我们展示了我们的重绘模型在各种编辑任务中的多样性。

View on arXiv Download PDF AI Translation

cs.CV / 11 / 2609.10815

BodyCam-VQA: Enhanced Body-Worn Camera Video Captioning via Multimodal Reasoning and Probe Question Generation

BodyCam-VQA：通过多模态推理和探测问题生成增强的执法记录仪视频字幕生成

Gupta, Karish, Alex, Matthew, Li, Alex, Wu, Yang, Chu, Yun-Wei, Munir, Kashif, Zhou, Xiaotian, Ji, Zhengping, Liu, Xiaozhong

Abstract

Police body-worn camera (BWC) footage has emerged as a critical aspect of law enforcement that ensures legal transparency, officer accountability, and the protection of civil rights. However, effectively processing this data remains a significant challenge due to its multimodal video format. BWC videos, in many cases, comprise chaotic scenes with low visual quality, rapid movement/interactions, and high-noise audio that make visual understanding a challenge for even SOTA multimodal models. Current Vision-Language Models (VLMs) frequently overlook critical forensic details, such as the presence of valuable evidence or the latent nuances of suspect-officer interactions, which are vital for fair legal outcomes and civilian/officer safety. To address these limitations, we propose an Adaptive Visual Question Answering (VQA) framework engineered for high-stakes law enforcement. Our framework employs a structured reasoning approach to extract fine-grained visual evidence that traditional captioning systems fail to capture. We experiment with multiple question generation models, including foundation models and fine-tuned open-weight models, to observe performance variation among question generation model implementations. Our results demonstrate that this VQA-driven architecture provides a more reliable, objective, and detailed record of enforcement events, ultimately serving as a powerful tool to protect both law enforcement officers and the public through AI-assisted forensic clarity.

Chinese Translation

警用执法记录仪（BWC）视频已成为确保法律透明度、警员问责和保护公民权利的重要组成部分。然而，由于其多模态视频格式，有效处理这些数据仍然是一个重大挑战。BWC 视频在许多情况下包含混乱的场景，视觉质量低，快速运动/互动，以及高噪声音频，这使得即使是最先进的多模态模型也难以进行视觉理解。目前的视觉-语言模型（VLMs）常常忽视关键的法医细节，例如有价值证据的存在或嫌疑人与警员互动的潜在细微差别，这些对于公平的法律结果和公民/警员的安全至关重要。为了解决这些局限性，我们提出了一种针对高风险执法的自适应视觉问答（VQA）框架。我们的框架采用结构化推理方法，以提取传统字幕生成系统无法捕捉的细粒度视觉证据。我们实验了多种问题生成模型，包括基础模型和微调的开放权重模型，以观察不同问题生成模型实现之间的性能差异。我们的结果表明，这种以 VQA 为驱动的架构提供了更可靠、客观和详细的执法事件记录，最终作为一种强有力的工具，通过 AI 辅助的法医清晰度保护执法人员和公众。

View on arXiv Download PDF AI Translation

cs.CV / 12 / 2609.10851

Are We Really Doing Few-Shot Learning? A Critical Examination of Pre-Training Assumptions

我们真的在进行少样本学习吗？对预训练假设的批判性审视

Galan-Cuenca, Alejandro, Saval-Calvo, Marcelo, Gallego, Antonio Javier

Abstract

Few-shot learning is commonly evaluated under protocols that pre-train a model on a large auxiliary set whose classes are disjoint from the target episodes yet drawn from the same visual domain. This paper examines whether such protocols truly reflect low-data learning. We systematically compare no pre-training, class-disjoint in-domain pre-training, supervised out-of-domain pre-training, and label-free out-of-domain pre-training across eight datasets, three few-shot architectures, and multiple way-shot settings. Our results show that class disjointness alone is insufficient to remove the influence of target-domain data. In-domain pre-training improves over no pre-training by 33.41 percentage points on average, whereas supervised out-of-domain pre-training yields 23.75 percentage points, revealing a 9.66-point optimistic bias associated with domain overlap. Although out-of-domain pre-training is more realistic in applications where target-domain data are scarce, its effectiveness depends strongly on the compatibility between source and target domains. We further show that labeled source data are not strictly required, with an augmentation-based label-free strategy reaching an average gain of 27.71 percentage points and closely matching supervised out-of-domain pre-training at 27.97 percentage points. Finally, we introduce a descriptor-based source-selection strategy that estimates source-domain suitability before pre-training, reaching a median gap of only 1.37 percentage points to oracle selection. These findings highlight the need to move beyond in-domain pre-training as the default few-shot evaluation protocol, since it can overestimate performance in realistic scenarios where target-domain data are scarce.

Chinese Translation

少样本学习通常在预先对一个大型辅助集进行训练的协议下进行评估，该辅助集的类别与目标情境不相交，但来自同一视觉领域。本文考察了此类协议是否真正反映了低数据学习。我们系统地比较了无预训练、类别不相交的领域内预训练、监督的领域外预训练以及无标签的领域外预训练，涵盖了八个数据集、三种少样本架构和多种方式-样本设置。我们的结果表明，仅仅类别不相交不足以消除目标领域数据的影响。领域内预训练在平均上比无预训练提高了33.41个百分点，而监督的领域外预训练则提高了23.75个百分点，揭示了与领域重叠相关的9.66点乐观偏差。尽管领域外预训练在目标领域数据稀缺的应用中更为现实，其有效性在很大程度上依赖于源领域和目标领域之间的兼容性。我们进一步表明，标记的源数据并不是严格必要的，基于增强的无标签策略平均获得了27.71个百分点的增益，并与监督的领域外预训练的27.97个百分点非常接近。最后，我们引入了一种基于描述符的源选择策略，在预训练之前评估源领域的适用性，达到与理想选择仅有1.37个百分点的中位差距。这些发现强调了需要超越领域内预训练作为默认的少样本评估协议，因为在目标领域数据稀缺的现实场景中，它可能会高估性能。

View on arXiv Download PDF AI Translation

cs.CV / 13 / 2609.10855

Evaluation of Vision-Language Models Across Diverse Coastal Environments

多样化海岸环境中视觉-语言模型的评估

Knoop, Seth, Samuelson, Chad R., Slade, Gabriel R., Moon, Brady, Mangelson, Joshua G.

Abstract

Vision-language models (VLMs) enable robotic per- ception by associating visual observations with natural-language concepts. Yet their performance in coastal environments remains largely unexplored. We introduce a densely labeled coastal dataset containing more than 1,000 images collected across seven missions in three regions of Oahu, Hawaii, with 18 semantic classes and over 7,400 annotated instances. We evaluate seven modern VLMs through three complementary experiments mea- suring text-to-mask, mask-to-mask, and mask-to-text alignment. Broad landscape classes are generally recognized more accurately than conventional object and coastal classes, with coastal con- cepts presenting the greatest challenge. However, comparisons of shared conventional classes across coastal and terrestrial datasets reveal no consistent performance difference attributable solely to environmental context. Mask-to-mask matching also remains similar across conventional and coastal classes, while alternative textual labels substantially improve recognition of several coastal concepts. These results suggest that lower performance on coastal classes (at least on the objects/query categories evaluated) is heavily influenced by segmentation and linguistic representation.

Chinese Translation

视觉-语言模型（VLMs）通过将视觉观察与自然语言概念关联，促进了机器人的感知。然而，它们在海岸环境中的表现仍然基本未被探索。我们引入了一个密集标注的海岸数据集，该数据集包含在夏威夷瓦胡岛三个区域的七次任务中收集的超过1000张图像，涵盖18个语义类别和7400多个标注实例。我们通过三个互补实验评估了七种现代VLM，测量文本到掩码、掩码到掩码和掩码到文本的对齐情况。广泛的景观类别通常比传统物体和海岸类别被更准确地识别，而海岸概念则呈现出最大的挑战。然而，对海岸和陆地数据集中共享的传统类别进行比较，未发现仅因环境背景而导致的一致性能差异。掩码到掩码的匹配在传统类别和海岸类别之间也保持相似，而替代文本标签显著提高了对若干海岸概念的识别。这些结果表明，在海岸类别上的较低表现（至少在评估的对象/查询类别中）受到分割和语言表示的重大影响。

View on arXiv Download PDF AI Translation

cs.CV / 14 / 2609.10898

Symmetry-aware super-resolution of crystal orientation maps via invariant latent-space learning

基于对称性的晶体取向图超分辨率学习

Garg, Umang, Zamudio, Warren, Echlin, McLean P., Daly, Samantha H., Pollock, Tresa M., Manjunath, B. S.

Abstract

Crystal-orientation maps are physical fields defined only up to crystal symmetry; electron backscatter diffraction (EBSD) resolves them experimentally, but acquisition-time constraints limit spatial resolution. Unlike conventional images, EBSD data lie on the quotient space $\mathrm{SO}(3)/G$, where $G$ is the crystal-symmetry group. Standard Euclidean interpolation can therefore mix symmetry-equivalent representations and blur grain boundaries. We introduce the Symmetry-Group-Aware Super-Resolution Attention Network (SG-SRAN), which incorporates crystal symmetry and boundary preservation by design. A frozen, locally isometric encoder maps equivalent orientations to a common latent representation in which Euclidean distance approximates misorientation. Super-resolution is performed in this space, with each high-resolution token restricted to a feature-consistent local support to prevent cross-boundary mixing. A dictionary-based decoder then recovers valid orientations. Across FCC and HCP benchmarks, SG-SRAN matches 15-16 million parameter backbones using only 27-49k trainable parameters, while achieving the lowest p68 errors, highest inverse-pole-figure fidelity, and zero-shot transfer to unseen alloys.

Chinese Translation

晶体取向图是仅在晶体对称性下定义的物理场；电子背散射衍射（EBSD）实验上能够解析这些图，但获取时间的限制限制了空间分辨率。与常规图像不同，EBSD 数据位于商空间 $ ext{SO}(3)/G$ 上，其中 $G$ 是晶体对称群。因此，标准的欧几里得插值可能会混合对称等效的表示，并模糊晶粒边界。我们提出了对称群感知超分辨率注意网络（SG-SRAN），该网络通过设计结合了晶体对称性和边界保护。一个冻结的局部等距编码器将等效取向映射到一个共同的潜在表示中，在该表示中，欧几里得距离近似于错位。超分辨率在这个空间中进行，每个高分辨率标记被限制在一个特征一致的局部支持范围内，以防止跨边界混合。然后，基于字典的解码器恢复有效的取向。在 FCC 和 HCP 基准测试中，SG-SRAN 使用仅 27-49k 可训练参数匹配 1500-1600 万参数的骨干网络，同时实现最低的 p68 误差、最高的逆极图保真度，并对未见合金实现零样本迁移。

View on arXiv Download PDF AI Translation

cs.CV / 15 / 2609.10917

HiPerViT: A Hierarchical Perceiver-Vision Transformer Architecture for Multi-Scale Texture Recognition

HiPerViT：一种用于多尺度纹理识别的层次感知视觉变换器架构

de Sá, João Pedro C. A., Bruno, Odemir Martinez

Abstract

Texture recognition remains challenging for modern vision models because discriminative evidence is often carried by higher-order spatial statistics rather than by object shape alone. While Vision Transformers provide strong long-range modeling capacity, their standard object-centric representations do not explicitly expose such statistical structure, which limits texture sensitivity in fine-grained recognition settings. We present HiPerViT, a compact vision-only architecture that injects an explicit second-order statistical prior into a transformer-based recognition pipeline. The method combines global and local image views with a compact bilinear descriptor encoded as a statistical token, and integrates this token with first-order spatial representations through Perceiver-style latent distillation. This design enables direct interaction between spatial tokens and second-order feature co-occurrence statistics, providing the model with explicit access to texture-relevant information without requiring multimodal pretraining or ensemble construction. Across six texture recognition benchmarks, HiPerViT achieves consistent improvements over strong vision-only baselines under the reported evaluation protocols, including gains of +3.05 percentage points on DTD, +10.48 on GTOS-Mobile, and +10.10 on 1200Tex. Beyond benchmark performance, our analyses show that these gains are largely invariant to the backbone depth used to extract second-order statistics and to the ordering of interaction and distillation stages. This pattern suggests that the primary source of improvement is not a specific fusion topology, but the explicit availability of second-order statistical information as a first-class representational signal. These results support explicit statistical tokenization as an effective and robust design principle for texture-centric visual recognition.

Chinese Translation

纹理识别对于现代视觉模型仍然具有挑战性，因为判别性证据通常由高阶空间统计量携带，而不仅仅是由物体形状决定。尽管视觉变换器提供了强大的长距离建模能力，但其标准的以物体为中心的表示并未明确暴露这种统计结构，这限制了在细粒度识别环境中的纹理敏感性。我们提出了HiPerViT，这是一种紧凑的仅视觉架构，将显式的二阶统计先验注入到基于变换器的识别流程中。该方法结合了全局和局部图像视图，并使用编码为统计标记的紧凑双线性描述符，将该标记与通过感知器风格的潜在蒸馏获得的一阶空间表示相结合。这种设计使空间标记与二阶特征共现统计之间能够直接交互，为模型提供了显式访问与纹理相关信息的能力，而无需多模态预训练或集成构建。在六个纹理识别基准测试中，HiPerViT在报告的评估协议下，相较于强大的仅视觉基线实现了一致的提升，包括在DTD上提高了+3.05个百分点，在GTOS-Mobile上提高了+10.48，在1200Tex上提高了+10.10。除了基准性能外，我们的分析表明，这些提升在提取二阶统计量所用的骨干网络深度和交互与蒸馏阶段的顺序上基本不变。这一模式表明，改进的主要来源并不是特定的融合拓扑，而是将二阶统计信息作为一类重要的表示信号的显式可用性。这些结果支持显式统计标记化作为一种有效且稳健的设计原则，适用于以纹理为中心的视觉识别。

View on arXiv Download PDF AI Translation

cs.CV / 16 / 2609.10943

CamPilot: A Multi-Agent Cinematic Assistant for Camera-Controlled Movie Generation

CamPilot：一种多智能体电影助手，用于基于摄像机控制的电影生成

Wu, Yang, Petrangeli, Stefano, Dasgupta, Ishita, Shen, Yu

Abstract

The integration of large language models (LLMs) into video generation has enabled rapid text-to-video creation and improved visual quality. However, it still falls short of professional filmmaking, where cinematographic language is less refined than human-crafted camera work and multi-shot continuity remains challenging. To address these limitations, we introduce CamPilot, a multi-agent framework that integrates cinematographic planning and camera-work control to produce more coherent, logically structured, and human-aesthetic movies. CamPilot adopts a GRPO-based learning paradigm to learn camera work planning from 14K real-world professional movies, internalizing motion patterns and composition principles that support reasoning over shooting techniques (e.g., camera angle, motion, and focal behavior) and cross-shot relationships for controllable camera-viewpoint generation. Multiple agents further collaborate and evolve to improve overall output quality. To support this work and further studies in this domain, we establish CamEval, a benchmark for evaluating camera work quality and cinematic engagement. Empirical results show that CamPilot outperforms state-of-the-art text-to-movie generation methods on cinematographic control and quality, highlighting the impact of professional camera design on movie generation.

Chinese Translation

大语言模型（LLMs）与视频生成的结合使得文本到视频的快速创建成为可能，并提升了视觉质量。然而，这仍然无法与专业电影制作相提并论，因为电影语言的精细程度不及人类制作的摄像工作，且多镜头的连续性依然具有挑战性。为了解决这些局限性，我们提出了CamPilot，一个多智能体框架，集成了电影规划和摄像控制，以生成更连贯、逻辑结构更清晰且符合人类审美的电影。CamPilot采用基于GRPO的学习范式，从14,000部真实世界的专业电影中学习摄像规划，内化支持拍摄技术（例如摄像机角度、运动和焦距行为）和镜头间关系推理的运动模式和构图原则，以实现可控的摄像机视角生成。多个智能体进一步合作和演化，以提高整体输出质量。为了支持这项工作和该领域的进一步研究，我们建立了CamEval，一个用于评估摄像质量和电影吸引力的基准。实证结果表明，CamPilot在摄像控制和质量方面优于最先进的文本到电影生成方法，突显了专业摄像设计对电影生成的影响。

View on arXiv Download PDF AI Translation

cs.CV / 17 / 2609.11022

New Evidence, Same Choice: Testing Physical Experiment Selection in Vision Language Models

新证据，相同选择：测试视觉语言模型中的物理实验选择

Saha, Sourajit, Dipta, Shubhashis Roy, Sarwar, Nobin, Saha, Shaswati, Jiang, Yuxuan, Li, Siyuan, Wang, Qiheng

Abstract

A model first sees an image from one physical measurement experiment, such as how far a block coasted, and must answer a question about a new trial, such as whether the block will pass a target after a fixed push. The initial experiment may provide enough information to answer, or the model may need another measurement, such as the object's mass, friction, restitution, or spring stiffness. We study whether vision language models can decide when to answer immediately and, when more evidence is needed, which experiment to perform. Current physical reasoning benchmarks usually evaluate only the final answer, so they do not directly measure this decision-making ability. We introduce a controlled evaluation where each problem provides one measurement image and four possible physical worlds created by combining two possible masses and two possible values of another relevant property. The model must either stop and answer or select the cheapest additional experiment that can resolve the question. We construct matched problem pairs where changing either the observed measurement or the question changes the optimal action. Since all possible worlds and experiment costs are known, we can explicitly determine the optimal choice. Across six open models and 144 physical parameter sets, direct responses repeat the same action for 95.1% to 100% of image pairs even when the correct action changes. Brief reasoning improves action switching, but the best model makes both decisions correctly for only 5.9% of image pairs. Additional analysis reveals failures in measurement interpretation, physical reasoning, and response formatting. By evaluating evidence selection separately from final answers, our benchmark reveals limitations in physical reasoning that conventional answer accuracy can overlook.

Chinese Translation

模型首先观察来自一个物理测量实验的图像，例如一个块体滑行的距离，并必须回答关于新试验的问题，例如该块体在固定推力后是否会通过目标。初始实验可能提供足够的信息来回答，或者模型可能需要另一个测量，例如物体的质量、摩擦、恢复系数或弹簧刚度。我们研究视觉语言模型是否能够决定何时立即回答，以及在需要更多证据时，选择进行哪个实验。当前的物理推理基准通常仅评估最终答案，因此并未直接测量这种决策能力。我们引入了一种控制评估，每个问题提供一个测量图像和四个可能的物理世界，这些世界是通过组合两种可能的质量和另一种相关属性的两个可能值而创建的。模型必须选择停止并回答，或选择可以解决问题的最便宜的额外实验。我们构建了匹配的问题对，其中改变观察到的测量或问题会改变最佳行动。由于所有可能的世界和实验成本都是已知的，我们可以明确确定最佳选择。在六个开放模型和144个物理参数集的测试中，直接响应在95.1%到100%的图像对中重复相同的行动，即使正确的行动发生变化。简要推理提高了行动切换的能力，但最佳模型仅在5.9%的图像对中正确做出两个决策。额外分析揭示了测量解释、物理推理和响应格式方面的失败。通过将证据选择与最终答案分开评估，我们的基准揭示了传统答案准确性可能忽视的物理推理局限性。

View on arXiv Download PDF AI Translation

cs.CV / 18 / 2609.11040

Toward Interpretable Multimodal Fusion: Heat Conduction Modeling for Hyperspectral and LiDAR Joint Classification

朝向可解释的多模态融合：用于高光谱和激光雷达联合分类的热传导建模

Wei, Kan, Cui, Jiahui, Yao, Jing, Zhao, Xinyu, Wang, Lei, Ghamisi, Pedram

Abstract

The fusion of hyperspectral (HS) and Light Detection and Ranging (LiDAR) data plays a crucial role in enhancing land-cover classification by jointly exploiting spectral, spatial, and structural cues. However, existing multimodal fusion methods still struggle to model long-range dependencies and complex anisotropic interactions while maintaining computational efficiency. This paper introduces M2Heat, a physics-inspired framework that investigates multimodal fusion through the lens of heat conduction. At its core, a physics-driven visual heat conduction module (vHeat) and enhanced Frequency Value Embeddings (FVEs) simulate anisotropic information flow, enabling the capture of global dependencies with sub-quadratic complexity and physical interpretability. This mechanism, combined with a hybrid spatial-frequency fusion strategy named Cross-Frequency Fusion (CFF) module, produces highly discriminative and robust feature representations. M2Heat achieves competitive overall performance on three benchmarks, i.e., Trento, Houston2013, and Augsburg, while providing an interpretable heat-conduction-guided perspective for multimodal feature fusion. These results indicate the potential of heat-conduction-guided neural operators for efficient and interpretable RS multimodal fusion. The source code is publicly available at https: /github.com/Weikan0425/M2Heat_HSI_LiDAR.

Chinese Translation

高光谱（HS）和光学探测与测距（LiDAR）数据的融合在通过共同利用光谱、空间和结构线索来增强地表覆盖分类方面发挥着至关重要的作用。然而，现有的多模态融合方法在建模长距离依赖关系和复杂各向异性交互时仍然面临挑战，同时保持计算效率。本文介绍了M2Heat，这是一种受物理启发的框架，通过热传导的视角研究多模态融合。其核心是一个基于物理驱动的视觉热传导模块（vHeat）和增强的频率值嵌入（FVE），模拟各向异性信息流，从而以亚二次复杂度和物理可解释性捕捉全局依赖关系。该机制结合了一种名为交叉频率融合（CFF）模块的混合空间-频率融合策略，生成高度区分且稳健的特征表示。M2Heat在三个基准测试（即Trento、Houston2013和Augsburg）上实现了具有竞争力的整体性能，同时为多模态特征融合提供了可解释的热传导引导视角。这些结果表明，热传导引导的神经算子在高效和可解释的遥感多模态融合中具有潜力。源代码可在https://github.com/Weikan0425/M2Heat_HSI_LiDAR公开获取。

View on arXiv Download PDF AI Translation

cs.CV / 19 / 2609.11041

Meta-Learning for Classifier Selection in Image Datasets: A Feature-Driven Framework for Accuracy Prediction

图像数据集中的分类器选择的元学习：一种基于特征驱动的准确性预测框架

Nabizadeh_Shahre_Babak, Zahra, Koohestani, Farzaneh, Karimi, Nader, Shirani, Shahram, Samavi, Shadrokh

Abstract

No Free Lunch theorem implies that any performance gains achieved by a classifier on a particular image distribution are necessarily offset by a loss of performance over the set of all possible problems; thus, no single model is universally optimal. Selecting the most suitable classifier for image datasets is a critical yet challenging task due to the intrinsic complexity and diversity of images. This paper proposes a meta-learning framework that leverages a comprehensive set of meta-features capturing dataset complexity to predict classifier performance without exhaustive training. By extracting and selecting features using methods such as autoencoders, pre-trained networks, and dimensionality reduction techniques, we train regression models to efficiently estimate classifier accuracies. Additionally, clustering techniques are employed to group classifiers with similar performance patterns, simplifying the recommendation process. The datasets used span a wide range of concepts, including nature, animals, numbers, motorcycles, medical images, and human bodies, to ensure broad generalization. Evaluated on 56 diverse image datasets, our approach achieves an average ranking prediction accuracy exceeding 86%, demonstrating its effectiveness in guiding model selection. This scalable and interpretable framework provides a practical solution to improve classification performance while reducing computational costs.

Chinese Translation

无免费午餐定理表明，分类器在特定图像分布上取得的任何性能提升必然会被在所有可能问题集上的性能损失所抵消；因此，没有单一模型是普遍最优的。由于图像的内在复杂性和多样性，为图像数据集选择最合适的分类器是一项关键但具有挑战性的任务。本文提出了一种元学习框架，利用全面的元特征集来捕捉数据集复杂性，以预测分类器性能，而无需进行耗时的训练。通过使用自编码器、预训练网络和降维技术等方法提取和选择特征，我们训练回归模型以高效估计分类器的准确性。此外，采用聚类技术对具有相似性能模式的分类器进行分组，从而简化推荐过程。所使用的数据集涵盖了广泛的概念，包括自然、动物、数字、摩托车、医学图像和人体，以确保广泛的泛化能力。在56个多样化的图像数据集上进行评估，我们的方法实现了超过86%的平均排名预测准确率，证明了其在指导模型选择方面的有效性。该可扩展且可解释的框架为提高分类性能提供了实际解决方案，同时降低了计算成本。

View on arXiv Download PDF AI Translation

cs.CV / 20 / 2609.11081

TailProp: content-adaptive light- and heavy-tailed propagation for vision

TailProp：面向视觉的内容自适应轻尾和重尾传播

Kong, Jiahao, Li, Zihan

Abstract

Science-inspired vision models show that explicit propagation dynamics can provide structured and interpretable alternatives to conventional token mixing. Existing formulations, however, typically construct and adapt visual propagation within a particular dynamical family, while visual representations can require substantially different spatial interactions across samples, channels, and network stages. We explore cross-regime adaptive propagation and introduce TailProp, a hierarchical vision backbone built upon the Tail Propagation Operator (TPO). TPO uses Gaussian and Cauchy stable-process propagators as complementary bases with rapidly decaying and heavy-tailed spatial influence, and predicts a content-conditioned channel-wise coefficient to adaptively combine them. Because this coefficient is spatially shared, the two responses are fused directly in the DCT domain with a single DCT/IDCT pair, yielding $O(N^{1.5})$ spatial mixing for square feature maps with $N=HW$ and fixed channel width. Across image classification, object detection, semantic segmentation, robustness, and cross-backbone restoration, TailProp consistently outperforms matched propagation baselines; TailProp-B reaches 84.4% Top-1 accuracy on ImageNet-1K, 50.3/44.8 box/mask AP under the 3x Mask R-CNN schedule, and 50.8% mIoU on ADE20K. Controlled ablations further show that these gains are not explained by single-basis propagation, an additional same-family branch, or within-family adaptive order alone, supporting complementary two-basis propagation as an effective design principle for visual representation learning.

Chinese Translation

受科学启发的视觉模型表明，显式传播动态可以为传统的标记混合提供结构化和可解释的替代方案。然而，现有的公式通常在特定的动态家族内构建和调整视觉传播，而视觉表征在样本、通道和网络阶段之间可能需要显著不同的空间交互。我们探索跨领域自适应传播，并引入TailProp，一个基于尾传播算子（Tail Propagation Operator, TPO）的分层视觉骨干网络。TPO使用高斯和柯西稳定过程传播器作为互补基础，具有快速衰减和重尾空间影响，并预测一个内容条件的通道系数以自适应地组合它们。由于该系数在空间上是共享的，因此这两个响应在DCT域中通过单个DCT/IDCT对直接融合，导致对于固定通道宽度的平方特征图，空间混合复杂度为$O(N^{1.5})$，其中$N=HW$。在图像分类、目标检测、语义分割、鲁棒性和跨骨干恢复等任务中，TailProp始终优于匹配的传播基线；TailProp-B在ImageNet-1K上达到84.4%的Top-1准确率，在3x Mask R-CNN计划下获得50.3/44.8的框/掩码AP，并在ADE20K上达到50.8%的mIoU。控制性消融实验进一步表明，这些提升并不能仅通过单基础传播、额外的同家族分支或同家族自适应顺序来解释，支持互补的双基础传播作为视觉表征学习的有效设计原则。

View on arXiv Download PDF AI Translation

cs.CV / 21 / 2609.11126

Beyond Benchmarks: Using VLMs to Reveal Systematic Classification Failures Under Real World Conditions

超越基准：利用视觉语言模型揭示现实条件下的系统分类失败

Alblas, Dieuwertje, Liezenga, Alma M., van Woerden, Jan Erik, Taggenbrock, Fedor, Aljawaheri, Dalia, Schutte, Klamer

Abstract

Verification and validation (V&V) of classification models is crucial to enable a wide range of sensor processing applications. Currently, the V&V process relies on time-consuming manual inspection of erroneous samples to find meaningful patterns. This work explores the use of Vision Language Models (VLMs) to speed up this laborious process. VLMs are trained to embed images into a semantically meaningful vector representation, from which human-interpretable systematic errors can be distilled. Deploying such VLM-based methods in a defence context introduces two major challenges: (1) the defence domain is underrepresented in the training data of VLMs, and (2) surroundings and context are less diverse than for other domains. This study provides an initial assessment of the suitability of VLM-based methods for V&V of defence applications. We propose a VLM-based error slice detection (ESD) method that independently groups and labels systematic errors made by a classification model. We demonstrate that this method is able to identify operationally-relevant artificially added perturbations in a non-military dataset. In a military context, our method clusters and describes images based on their surroundings, but also exhibits overlap between cluster descriptions. We further investigate the difference in embedding variation between our military and non-military dataset, which remains a topic of interest. Although the results do not yet warrant fully automated V&V through VLM-based ESD, they show that VLMs could be used to accelerate V&V processes in the future.

Chinese Translation

分类模型的验证与确认（V&V）对于实现广泛的传感器处理应用至关重要。目前，V&V过程依赖于耗时的人工检查错误样本以寻找有意义的模式。本研究探讨了利用视觉语言模型（VLMs）加速这一繁琐过程的可能性。VLMs被训练为将图像嵌入到语义上有意义的向量表示中，从中可以提炼出人类可解释的系统性错误。在国防背景下部署这种基于VLM的方法面临两个主要挑战：（1）国防领域在VLM训练数据中代表性不足，和（2）环境和上下文的多样性低于其他领域。本研究对基于VLM的方法在国防应用中的V&V适用性进行了初步评估。我们提出了一种基于VLM的错误切片检测（ESD）方法，该方法独立地对分类模型所犯的系统性错误进行分组和标记。我们证明该方法能够识别非军事数据集中与操作相关的人工添加扰动。在军事背景下，我们的方法根据周围环境对图像进行聚类和描述，但也表现出聚类描述之间的重叠。我们进一步研究了军事和非军事数据集之间嵌入变化的差异，这仍然是一个值得关注的话题。尽管结果尚不足以通过基于VLM的ESD实现完全自动化的V&V，但它们表明VLMs在未来可能用于加速V&V过程。

View on arXiv Download PDF AI Translation

cs.CV / 22 / 2609.11129

ReconPlusGen: Injecting Reconstruction Prior into Multi-view 3D Generation through Noise Inversion and Modulation

ReconPlusGen：通过噪声反演和调制将重建先验注入多视角三维生成

Liu, Jiarui, Li, Heng, Li, Weiyu, Deng, Keng, Deng, Junyuan, Zhongxing, Zheng, Huang, Junyu, Chang, Jiahao, Han, Xiaoguang, Tan, Ping

Abstract

Qualitative results and an illustration of our core idea. Top left: reconstruction results on benchmark images. Top right: reconstruction results on real-world images. Bottom: illustration of reconstruction-guided noise initialization and modulation. Given multiple input images, we predict a point cloud in canonical space, deterministically inject the predicted geometry into the diffusion process through noise inversion, and modulate the resulting noise to preserve the generative flexibility required to complete unobserved regions and refine visible geometry.

Chinese Translation

定性结果及我们核心思想的说明。左上：基准图像上的重建结果。右上：现实世界图像上的重建结果。底部：重建引导的噪声初始化和调制的示意图。给定多个输入图像，我们在规范空间中预测点云，通过噪声反演确定性地将预测的几何形状注入扩散过程，并调制生成的噪声，以保留完成未观察区域和细化可见几何形状所需的生成灵活性。

View on arXiv Download PDF AI Translation

cs.CV / 23 / 2609.11134

LAION-Mobile: Evaluating Deepfake Detectors On One Million Smartphone Photos

LAION-Mobile：在一百万张智能手机照片上评估深伪检测器

von Stryk, Achim, Keuper, Janis

Abstract

Most Deepfake detectors report near-perfect AUC scores on their reference benchmarks. However, a recent ICML position paper argues that these evaluations collectively neglect the impact of modern smartphone photography: the widely used on-device neural image-signal processing pipelines (like multi-sensor fusion or noise and motion-blur suppression) increasingly shift the imaging paradigm from simple lens projections towards computational photography. Hence, devices actually generate, rather than record photos. This increases the risk that deepfake detectors may flag ordinary phone photos as fake. Due to the lack of large-scale datasets containing images from modern smartphones, this hypothesis has so far only been tested in small proof-of-concept studies. The aim of this paper is to close this gap. We introduce LAION-Mobile, an open dataset containing about 1 million smartphone images with EXIF metadata distilled from re-LAION-5B. Evaluating twelve state-of-the-art deepfake detectors with their original paper checkpoints on a 9,115-image evaluation sample of this pool (DIRE on 738), we report three key findings: (i) On modern AI content no detector exceeds AUC 0.624, and five of twelve fall below chance. (ii) Real-photo false-alarm rates are an artefact of threshold calibration: thresholds fitted on legacy GAN data make several detectors look deployable (less than 11 percent FPR), yet the same detectors flag 17-91 percent of real photos once the identical criterion is refit on modern content. (iii) Consequently, no detector both beats chance on modern AI content and keeps a deployable real-photo false-alarm rate. Mirroring the device mix of web collections, the corpus probes the first neural-ISP generation (2018-2020); current flagships are essentially absent, leaving the modern-ISP regime as the open gap.

Chinese Translation

大多数深伪检测器在其参考基准上报告接近完美的AUC分数。然而，最近的一篇ICML立场论文指出，这些评估共同忽视了现代智能手机摄影的影响：广泛使用的设备内神经图像信号处理管道（如多传感器融合或噪声和运动模糊抑制）越来越将成像范式从简单的镜头投影转向计算摄影。因此，设备实际上生成照片，而不是仅仅记录照片。这增加了深伪检测器将普通手机照片标记为伪造的风险。由于缺乏包含现代智能手机图像的大规模数据集，这一假设迄今为止仅在小规模的概念验证研究中得到了测试。本文的目的是填补这一空白。我们引入LAION-Mobile，这是一个包含约一百万张带有EXIF元数据的智能手机图像的开放数据集，这些数据是从re-LAION-5B中提取的。我们在该数据集中对十二个最先进的深伪检测器进行了评估，使用其原始论文检查点，在一个包含9,115张图像的评估样本中（DIRE在738上），我们报告了三个关键发现：（i）在现代AI内容上，没有检测器的AUC超过0.624，其中五个检测器的表现低于随机猜测。（ii）真实照片的误报率是阈值校准的伪影：在传统GAN数据上拟合的阈值使得几个检测器看起来可部署（误报率低于11%），然而一旦在现代内容上重新拟合相同的标准，这些检测器会标记17%到91%的真实照片。（iii）因此，没有检测器在现代AI内容上既能超过随机猜测，又能保持可部署的真实照片误报率。该语料库反映了网络集合的设备组合，探讨了第一代神经ISP（2018-2020）；目前的旗舰设备基本缺失，留下现代ISP范畴作为开放的空白。

View on arXiv Download PDF AI Translation

cs.CV / 24 / 2609.11156

UniH$^3$: Unifying Hierarchical Homogeneity and Heterogeneity for All-in-One Medical Image Restoration

UniH$^3$: 统一层次同质性与异质性的一体化医学图像恢复

Yang, Zhiwen, Li, Jiayin, Liu, Chengyu, Zhang, Hui, Wei, Bingzheng, Xu, Yan

Abstract

All-in-One medical image restoration (MedIR) aims to address diverse tasks across modalities and degradation types using a single universal model. Existing methods typically prioritize modeling inter-task heterogeneity (e.g., distinct data distributions and degradation types). However, they largely neglect the inherent homogeneity present in medical images, such as widely shared anatomical structures within and across modalities, which can be leveraged to ease model training and improve generalization. To this end, we propose UniH3, a novel framework that Unifies Hierarchical Homogeneity and Heterogeneity for all-in-one medical image restoration. Specifically, to comprehensively exploit homogeneity, we introduce a Hierarchical Homogeneity Memory (H2M) module that progressively distills intra- and inter-task homogeneity priors from high-quality images during training, and adaptively retrieves the most relevant priors tailored to the input for guided restoration. These retrieved priors are then injected into the restoration pipeline via an efficient Homogeneity-Guided Attention (HGA) mechanism. Furthermore, to comprehensively address heterogeneity, we design a Hierarchical Heterogeneity Balancer (H2B) that mitigates both inter- and intra-task conflicts during optimization, facilitating balanced and effective multi-task learning. Extensive experiments on two large-scale benchmarks, MedIR-2D-500K and MedIR-3D-3K, demonstrate that UniH3 achieves state-of-the-art performance on both all-in-one and single-task medical image restoration. We hope this work establishes a strong benchmark and advances the development of general-purpose medical image restoration models. Code is available at https://github.com/Yaziwel/UniH3.

Chinese Translation

一体化医学图像恢复（MedIR）旨在使用单一通用模型解决跨模态和降质类型的多样任务。现有方法通常优先考虑建模任务间的异质性（例如，数据分布和降质类型的差异）。然而，它们在很大程度上忽视了医学图像中固有的同质性，例如模态内外广泛共享的解剖结构，这可以被利用来简化模型训练并提高泛化能力。为此，我们提出了UniH3，一个新颖的框架，统一层次同质性与异质性以实现一体化医学图像恢复。具体而言，为了全面利用同质性，我们引入了层次同质性记忆（H2M）模块，该模块在训练过程中逐步提炼来自高质量图像的任务内和任务间同质性先验，并根据输入自适应地检索最相关的先验以指导恢复。这些检索到的先验通过高效的同质性引导注意力（HGA）机制注入到恢复流程中。此外，为了全面应对异质性，我们设计了层次异质性平衡器（H2B），在优化过程中减轻任务间和任务内的冲突，促进平衡和有效的多任务学习。在两个大规模基准数据集MedIR-2D-500K和MedIR-3D-3K上的广泛实验表明，UniH3在一体化和单任务医学图像恢复方面均实现了最先进的性能。我们希望这项工作能够建立一个强有力的基准，并推动通用医学图像恢复模型的发展。代码可在 https://github.com/Yaziwel/UniH3 获取。

View on arXiv Download PDF AI Translation

cs.CV / 25 / 2609.11172

Beyond Visual Quality: Evaluating Physical Consistency under Ego-Motion with EgoGenEval

超越视觉质量：使用 EgoGenEval 评估自我运动下的物理一致性

Long, Yilin, Zhu, Chenming, Gou, Zitang, Lin, Jingli, Wang, Tai

Abstract

Recent visual generators produce high-fidelity images yet often violate physical consistency under ego-motion, limiting their use for spatial reasoning and embodied planning. Existing benchmarks largely focus on isolated images or single-step quality, leaving this challenge underexplored. We introduce EgoGenEval, a geometry-grounded, pose-free benchmark designed to evaluate the physical consistency of visual generators under ego-motion, and organize our study into two parts. (1) EgoGenEval contains 1,400 cases and 2,360 target views spanning single-step and multi-step ego-motion. It separately measures Camera Motion Grounding (CMG) and Scene State Preservation (SSP), with both metrics validated against blinded human judgments. Evaluating 16 pose-free generators together with two pose-conditioned references reveals that current models struggle to execute camera motion while maintaining scene state, and that no system performs well on both axes at once. (2) To examine whether benchmark-derived data can improve these capabilities, we build EgoGen-Train from the same geometry-grounded pipeline and run controlled SFT studies. These show that pairwise supervision does not reliably improve camera-motion grounding and scene-state preservation together: even at the full training pool and the longest budget, scene preservation gains a fraction of what camera motion does. This points to the pairwise teacher-forced objective itself as the binding constraint, motivating a trajectory-centric paradigm that couples self-conditioned rollouts with explicit pose and visibility supervision.

Chinese Translation

近期的视觉生成器能够生成高保真图像，但在自我运动下常常违反物理一致性，限制了它们在空间推理和具身规划中的应用。现有基准测试主要集中于孤立图像或单步质量，导致这一挑战未得到充分探索。我们引入了 EgoGenEval，这是一个基于几何的、无姿态的基准，旨在评估视觉生成器在自我运动下的物理一致性，并将我们的研究分为两个部分。(1) EgoGenEval 包含 1,400 个案例和 2,360 个目标视图，涵盖单步和多步自我运动。它分别测量相机运动基础（Camera Motion Grounding, CMG）和场景状态保持（Scene State Preservation, SSP），这两个指标均经过盲评人类判断的验证。对 16 个无姿态生成器及两个姿态条件参考模型的评估显示，当前模型在执行相机运动的同时保持场景状态方面存在困难，并且没有任何系统在这两个维度上同时表现良好。(2) 为了检验基准派生数据是否能改善这些能力，我们从相同的几何基础管道构建了 EgoGen-Train，并进行了受控的 SFT 研究。这些研究表明，成对监督并不能可靠地同时改善相机运动基础和场景状态保持：即使在完整的训练池和最长的预算下，场景保持的提升也仅为相机运动提升的一小部分。这表明成对教师强制目标本身是限制因素，促使我们提出一种以轨迹为中心的范式，将自我条件的展开与明确的姿态和可见性监督相结合。

View on arXiv Download PDF AI Translation

cs.CV / 26 / 2609.11188

A Multi-View and Confusion-Guided Ensemble Framework for Robust Synthetic Image Attribution

一种多视角与混淆引导的集成框架用于稳健的合成图像归属

Qu, Zuomin

Abstract

Synthetic image attribution (SIA) has become increasingly important with the rapid advancement of text-to-image generation models. However, accurately identifying the source model of a generated image remains challenging due to the growing similarity among modern diffusion-based generators and the presence of diverse post-processing operations. In this report, we present a multi-view and confusion-guided ensemble framework for the Synthetic Image Attribution Challenge of the DLMMDD Workshop at ICANN 2026. Our approach integrates multiple complementary architectures, including FFT-ConvNeXt, DINOv2, CLIP, and Xception, to capture diverse attribution cues from frequency, semantic, and forensic perspectives. To improve robustness against unknown degradations and image manipulations, extensive data augmentation strategies are employed during training, simulating realistic post-processing operations such as compression, resizing, grayscale conversion, and blur. Furthermore, we analyze the confusion patterns of the ensemble model and observe severe ambiguity between Stable Diffusion 3 and Stable Diffusion 3.5. To address this issue, we introduce a dedicated binary expert classifier that is selectively activated under low-confidence conditions. We additionally apply class-adaptive confidence calibration to improve the discrimination of challenging classes such as Tencent Hunyuan. The proposed framework achieved 99.53% on the public leaderboard and 99.20% on the private leaderboard. The source code and implementation details are publicly available at https://github.com/ZOMIN28/SIA.

Chinese Translation

合成图像归属（SIA）随着文本到图像生成模型的快速发展变得越来越重要。然而，由于现代基于扩散的生成器之间的相似性日益增强以及多样化的后处理操作的存在，准确识别生成图像的源模型仍然具有挑战性。在本报告中，我们提出了一种多视角与混淆引导的集成框架，旨在参加2026年ICANN会议的DLMMDD研讨会的合成图像归属挑战。我们的方法整合了多种互补架构，包括FFT-ConvNeXt、DINOv2、CLIP和Xception，从频率、语义和取证的角度捕捉多样的归属线索。为了提高对未知降质和图像操控的鲁棒性，我们在训练过程中采用了广泛的数据增强策略，模拟了压缩、调整大小、灰度转换和模糊等现实的后处理操作。此外，我们分析了集成模型的混淆模式，并观察到Stable Diffusion 3与Stable Diffusion 3.5之间存在严重的模糊性。为了解决这个问题，我们引入了一个专用的二元专家分类器，该分类器在低置信度条件下被选择性激活。我们还应用了类自适应置信度校准，以改善对如腾讯混元等具有挑战性类别的区分能力。所提出的框架在公共排行榜上达到了99.53%的成绩，在私有排行榜上达到了99.20%的成绩。源代码和实现细节已公开，网址为https://github.com/ZOMIN28/SIA。

View on arXiv Download PDF AI Translation

cs.CV / 27 / 2609.11201

CEM-TUDASR: Computationally efficient multi-modality transformer based unsupervised domain adaptive super-resolution approach

CEM-TUDASR：一种计算高效的多模态基于Transformer的无监督域自适应超分辨率方法

Sarvaiya, Anjali, Kadel, Jay, Upla, Kishor, Raja, Kiran

Abstract

Wireless Capsule Endoscopy (WCE) enables non-invasive visualization of the gastrointestinal tract, but its miniaturized optics, sensor limitations, and wireless transmission constraints result in low-resolution images with reduced visibility of diagnostically important structures. This paper proposes CEM-TUDASR, a computationally efficient unsupervised Transformer-based super-resolution framework for WCE image enhancement without paired low-resolution (LR) and high-resolution (HR) training data. A domain-adaptive degradation network synthesizes realistic WCE-like LR images from HR conventional endoscopy images, reducing the domain gap and enabling effective unpaired learning. The SR generator integrates Deep Attention Blocks (DABs) and a Fusion Attention Block (FAB) to capture long-range contextual dependencies and fine local structures while preserving perceptual and structural fidelity. The model is trained on a curated dataset derived from Kvasir Capsule and evaluated on KID and GIANA for cross-dataset generalization. No-reference quality metrics, including BRISQUE, PIQE, NIQE, and the domain-specific EndoQM, show that CEM-TUDASR consistently outperforms existing unsupervised SR methods. Qualitative results further demonstrate improved restoration of mucosal textures, vascular patterns, and clinically relevant anatomical details. Cross-domain experiments on retinal images additionally demonstrate the adaptability of the framework. With only 2.67 million parameters and 169.94 GFLOPs, CEM-TUDASR achieves high-quality reconstruction while maintaining computational efficiency, making it suitable for resource-constrained clinical and embedded endoscopic applications.

Chinese Translation

无线胶囊内镜（Wireless Capsule Endoscopy, WCE）实现了对胃肠道的无创可视化，但其微型光学器件、传感器限制及无线传输约束导致图像分辨率较低，且诊断关键结构的可见度降低。本文提出了CEM-TUDASR，一种计算高效的基于Transformer的无监督超分辨率框架，用于WCE图像增强，无需成对的低分辨率（LR）和高分辨率（HR）训练数据。域自适应退化网络通过从HR常规内镜图像合成逼真的WCE风格LR图像，缩小了域间差距，实现了有效的无配对学习。超分辨率生成器融合了深度注意力模块（Deep Attention Blocks, DABs）和融合注意力模块（Fusion Attention Block, FAB），以捕捉长距离上下文依赖和细致的局部结构，同时保持感知和结构的保真度。模型在基于Kvasir Capsule整理的数据集上训练，并在KID和GIANA数据集上进行跨数据集泛化评估。无参考质量指标包括BRISQUE、PIQE、NIQE及领域特定的EndoQM均显示CEM-TUDASR持续优于现有无监督超分辨率方法。定性结果进一步证明其在黏膜纹理、血管图案及临床相关解剖细节的恢复上有显著提升。对视网膜图像的跨域实验也展示了该框架的适应性。CEM-TUDASR仅含2.67百万参数和169.94 GFLOPs，在保持计算效率的同时实现高质量重建，适合资源受限的临床及嵌入式内镜应用。

View on arXiv Download PDF AI Translation

cs.CV / 28 / 2609.11223

Tri-DehazeGS: Scene--Medium Decoupled Gaussian Splatting with Transmittance-Aware Optimization

Tri-DehazeGS：场景-介质解耦的透视感知高斯溅射优化

Jiang, Kui, Gu, Yang, Liu, Jiacheng, Liu, Shiyu, Chen, Youyu, Liu, Hui

Abstract

Recovering clean 3D scenes from hazy multi-view images is challenging because haze attenuates scene radiance and introduces atmospheric scattering. Recent scattering-aware Gaussian Splatting methods introduce physical haze models into reconstruction, but they often apply degradation in image space or bind medium-related variables to Gaussian primitives, which can entangle clean scene radiance with atmospheric effects. Moreover, low-transmittance regions provide weakened supervision for Gaussian optimization, causing distant or dense-haze areas to be under-reconstructed. We argue that clean reconstruction under haze requires both scene--medium disentanglement and transmittance-aware optimization rebalancing. To this end, we propose Tri-DehazeGS, a scene--medium decoupled Gaussian Splatting framework. It represents the clean scene with Gaussian primitives, models the participating medium using an independent view-shared tri-plane field, and composes hazy observations through a physical scattering model. We further introduce Medium-Decoupled Transmittance Gradient Compensation (MD-TGC), which compensates haze-suppressed gradients after medium freezing without altering forward rendering. Experiments on real and synthetic haze benchmarks show that Tri-DehazeGS improves clean novel-view reconstruction. Code is available at https://github.com/aptx46/Tri-DehazeGS.

Chinese Translation

从模糊的多视图图像中恢复干净的三维场景是一项具有挑战性的任务，因为雾霾会衰减场景的辐射并引入大气散射。最近的散射感知高斯溅射方法将物理雾霾模型引入重建中，但它们通常在图像空间中应用退化，或将与介质相关的变量绑定到高斯原语上，这可能会将干净的场景辐射与大气效应纠缠在一起。此外，低透射率区域为高斯优化提供了减弱的监督，导致远处或浓雾区域的重建不足。我们认为，在雾霾下的干净重建需要场景-介质的解耦和透射感知优化的重新平衡。为此，我们提出了Tri-DehazeGS，一个场景-介质解耦的高斯溅射框架。它使用高斯原语表示干净场景，使用独立的视图共享三平面场建模参与介质，并通过物理散射模型组合模糊观测。我们进一步引入了介质解耦透射梯度补偿（MD-TGC），在介质冻结后补偿雾霾抑制的梯度，而不改变前向渲染。对真实和合成雾霾基准的实验表明，Tri-DehazeGS改善了干净的新视图重建。代码可在 https://github.com/aptx46/Tri-DehazeGS 获取。

View on arXiv Download PDF AI Translation

cs.CV / 29 / 2609.11235

When is Test-Time Adaptation Identifiable From Unlabeled Evidence?

何时可以从未标记证据中识别测试时适应性？

Jhawar, Kartik, Wang, Lipo

Abstract

Test-time adaptation (TTA) offers many ways to update a deployed model without labels, but choosing the wrong update can make a strong source model worse. Recent methods therefore try to predict which adaptation will work from unlabeled test data. We ask a prior question: does the evidence given to the selector contain enough information to determine the best action at all? We show that this is not guaranteed, even with a perfect selector. If an observation channel makes two deployments look the same while their TTA rankings differ, reliable selection is impossible from that channel; richer evidence can restore the decision only when it resolves the relevant ambiguity. We make this boundary exact in a finite-batch Gaussian TTA model, where doing nothing beats mean recentering for small shifts, recentering wins beyond a unique critical shift, and the boundary shrinks as $1/\sqrt n$. Public benchmark studies on CIFAR-100-C and DomainNet-126 show the same failure mode with modern TTA methods: changing only deployment structure can reverse the oracle action while global order-blind evidence remains unchanged. The result is a practical way to separate two failure modes that are usually mixed together: a weak selector versus an information channel that cannot support the desired decision in the first place.

Chinese Translation

测试时适应性（TTA）提供了多种在没有标签的情况下更新已部署模型的方法，但选择错误的更新可能会使强源模型变得更糟。因此，最近的方法试图从未标记的测试数据中预测哪种适应性有效。我们提出一个先前的问题：提供给选择器的证据是否包含足够的信息来确定最佳行动？我们表明，即使有一个完美的选择器，这也并不是必然的。如果一个观察通道使两个部署看起来相同，而它们的TTA排名却不同，那么从该通道进行可靠选择是不可能的；只有当更丰富的证据解决相关的模糊性时，决策才能恢复。我们在有限批次高斯TTA模型中精确界定了这一边界，在小幅度偏移时，不采取行动优于均值重心调整，而在超过唯一临界偏移后，重心调整获胜，且边界随着 $1/ ext{sqrt} n$ 收缩。对CIFAR-100-C和DomainNet-126的公共基准研究显示，现代TTA方法也出现了相同的失败模式：仅改变部署结构就可以逆转oracle行动，而全局无序盲证据保持不变。该结果为区分通常混合在一起的两种失败模式提供了一种实用的方法：弱选择器与根本无法支持所需决策的信息通道。

View on arXiv Download PDF AI Translation

cs.CV / 30 / 2609.11236

HALDETECT at ImageEval 2026 Shared Tasks: Answer-First Contrastive Grounding with QLoRA

HALDETECT在ImageEval 2026共享任务中的表现：基于答案的对比性基础定位与QLoRA

Hoque, Syed Mohaiminul, Hossain, Md Sakhawat

Abstract

Large multimodal models tend to hallucinate visual detail fluently, which limits their deployment for fine-grained interpretation. We present HALDETECT, our system for the English hallucination-detection track (Task 1b) of ImageEval 2026, in which a system must identify, from an image and three culturally plausible statements, the single visually grounded one. We frame the item as one contrastive decision, emit the answer before its explanation, and structure reasoning around colour/texture, shape/form, and context. Our best submitted adapter fine-tunes Qwen2.5-VL-7B-Instruct with 4-bit QLoRA while freezing the vision encoder and reaches Contrastive Instability (CI) 0.035 on the 1,000-item test set; we placed third of eight teams. Development experiments show that answer order can matter more than model scale and that adaptation beats prompting alone. Retrospective paired analysis of the released gold labels confirms the QLoRA gain over the best prompt but not the small gap between the devtest-selected and best-test adapters, and reseeding all four training sizes shows that the apparent data-scaling curve does not survive a seed change. The 35 residual errors are culturally plausible function, material, and recognition distinctions; naive adapter voting does not help.

Chinese Translation

大型多模态模型往往流畅地幻觉视觉细节，这限制了它们在细粒度解释中的应用。我们提出了HALDETECT，这是我们在ImageEval 2026的英语幻觉检测任务（任务1b）中的系统，该任务要求系统从一幅图像和三个文化上合理的陈述中识别出唯一一个视觉上有依据的陈述。我们将该项目框架设定为一个对比性决策，在解释之前给出答案，并围绕颜色/纹理、形状/形式和上下文构建推理。我们提交的最佳适配器使用4位QLoRA对Qwen2.5-VL-7B-Instruct进行微调，同时冻结视觉编码器，并在1,000项测试集上达到对比性不稳定性（CI）0.035；我们在八个团队中排名第三。开发实验表明，答案的顺序可能比模型规模更重要，而适应性优于单纯的提示。对发布的金标准标签的回顾性配对分析确认了QLoRA相较于最佳提示的提升，但未能显示开发测试选择的适配器与最佳测试适配器之间的小差距，重新设定所有四个训练规模显示，明显的数据缩放曲线在种子变化时并不成立。35个残余错误是文化上合理的功能、材料和识别区分；简单的适配器投票并没有帮助。

View on arXiv Download PDF AI Translation

cs.CV / 31 / 2609.11237

SCINTILLA-SNN: A Spiking Multi-Scale Selective Aggregation Network for Perineural Invasion Prediction

SCINTILLA-SNN：一种用于神经周围侵袭预测的脉冲多尺度选择性聚合网络

Han, Youngung, Jeong, Yului, Kim, Kyeonghun, Kweon, Dohyun, Park, Suah, Go, Hyunsu, Park, Sungha, Jung, Anna, Jun, Jinyong, Choe, Yunho, Seo, Yunjin, Liao, Ken Ying-Kai, Lee, Hyuk-Jae, Kim, Nam-Joon

Abstract

Preoperative prediction of perineural invasion (PNI) in cholangiocarcinoma (CCA) is clinically valuable but remains challenging because PNI-related cues on magnetic resonance imaging (MRI) are subtle, sparse, and spatially localized around the tumor boundary. Standard 3D CNN and transformer architectures process volumetric data in a dense or spatially uniform manner, which can dilute subtle PNI-related evidence while requiring a large number of multiply-accumulate operations over 3D feature grids. To address these limitations, we propose SCINTILLA-SNN, a 3D spiking network composed of a four-stage hierarchical backbone and a Multi-Scale Spike Aggregation (MSSA) module for PNI prediction. The backbone extracts hierarchical volumetric representations through spiking convolutional stages and local spike window modulation stages. Given the resulting stage-wise representations, MSSA maps each spatial token to a learnable content value and modulates it with a spike-dynamics gate derived from firing rate and timestep-wise membrane-potential variability. The resulting score, referred to as the diagnostic token score, is used to selectively aggregate sparse PNI-related evidence. Experiments on a 10-year retrospective cohort of 182 CCA patients show that SCINTILLA-SNN achieves an AUROC of 0.748 under 5-fold cross-validation, while reducing the estimated inference energy by 23.18$\times$ compared with dense MAC-only computation of the same network.

Chinese Translation

在胆管癌（CCA）中，术前预测神经周围侵袭（PNI）具有临床价值，但由于PNI相关线索在磁共振成像（MRI）中表现得微妙、稀疏且空间上局限于肿瘤边界附近，因此仍然具有挑战性。标准的3D卷积神经网络（CNN）和变换器架构以密集或空间均匀的方式处理体积数据，这可能会稀释微妙的PNI相关证据，同时需要在3D特征网格上进行大量的乘加操作。为了解决这些限制，我们提出了SCINTILLA-SNN，这是一种由四阶段层次主干和多尺度脉冲聚合（MSSA）模块组成的3D脉冲网络，用于PNI预测。主干通过脉冲卷积阶段和局部脉冲窗口调制阶段提取层次体积表示。根据得到的阶段性表示，MSSA将每个空间标记映射到一个可学习的内容值，并通过从发射率和时间步膜电位变异性导出的脉冲动态门对其进行调制。得到的分数称为诊断标记分数，用于选择性聚合稀疏的PNI相关证据。在对182名CCA患者进行的为期10年的回顾性队列实验中，SCINTILLA-SNN在5折交叉验证下实现了0.748的AUROC，同时与相同网络的密集MAC-only计算相比，减少了23.18倍的估计推理能量。

View on arXiv Download PDF AI Translation

cs.CV / 32 / 2609.11240

Fast and Accurate Monomodal 3D High Resolution Deep Registration of Drosophila Larval Brain Volumes

快速且准确的单模态三维高分辨率深度配准果蝇幼虫脑体积

Reisenbüchler, Daniel, Sadegheih, Yousef, Dittrich, Michael, Kumari, Pratibha, Usman, Muhammad, Merhof, Dorit

Abstract

The larval stage of Drosophila melanogaster is a compact model system for neuroscience whose genetic toolkit allows fluorescent markers to be expressed in defined neural populations, and comparing the resulting expression patterns across animals requires every brain to be registered into a shared anatomical reference space. Existing pipelines for this task are predominantly based on classical registration methods, which perform a new optimization for each volume, often require per-case parameter tuning, and can take minutes per brain, limiting their use as a routine preprocessing step. We present a trained deep registration pipeline that deformably aligns a larval brain to a reference template in a single forward pass at high spatial resolution, on volumes that hold several times more voxels than those learned 3D registration is normally reported on, together with the preprocessing and anatomy-anchored evaluation pipeline required to apply it. Against eleven classical and seven further learned baselines on a held-out collection acquired with different acquisition and quality strata, the proposed pipeline is the most accurate, improving on the strongest classical baseline by 23 percentage points of anatomical landmark-local mutual information. It registers a volume one to two orders of magnitude faster than the classical deformable pipelines, and it retains more of its accuracy than any other method as acquisition quality degrades. The network, its trained weights and the full pipeline are released as the open-source deep larval brain registration framework: https://github.com/agentdr1/deep-larval-brain-reg

Chinese Translation

果蝇（Drosophila melanogaster）的幼虫阶段是一个紧凑的神经科学模型系统，其遗传工具箱允许在特定神经群体中表达荧光标记，而比较不同动物的表达模式需要将每个脑部注册到一个共享的解剖参考空间。现有的处理流程主要基于经典的配准方法，这些方法对每个体积进行新的优化，通常需要针对每个案例进行参数调整，并且每个脑部的处理时间往往需要几分钟，这限制了它们作为常规预处理步骤的使用。我们提出了一种训练好的深度配准流程，该流程能够在单次前向传递中以高空间分辨率将幼虫脑部变形对齐到参考模板，处理的体积包含的体素数量是通常报告的三维配准体积的几倍，同时提供了应用所需的预处理和解剖锚定评估流程。在对比了十一种经典基线和七种进一步学习的基线后，使用不同采集和质量层次获得的保留集合，所提出的流程是最准确的，改善了最强经典基线23个百分点的解剖标志-局部互信息。它的配准速度比经典的可变形流程快一个到两个数量级，并且在采集质量下降时保持了比其他任何方法更高的准确性。该网络、其训练权重及完整流程已作为开源深度幼虫脑配准框架发布： https://github.com/agentdr1/deep-larval-brain-reg

View on arXiv Download PDF AI Translation

cs.CV / 33 / 2609.11242

From Evaluation to Enhancement: Benchmarking and Improving Think-with-Video Reasoning for Video Generative Models

从评估到增强：基准测试与改进视频生成模型的思维视频推理

Luo, Meng, Liu, Yicheng, Wang, Jiahao, Zhang, Yuanxing, Tao, Xin, Wan, Pengfei, Gai, Kun, Fei, Hao

Abstract

Video generation has advanced to produce visually compelling and temporally coherent results. Yet, whether these models can genuinely think with video--executing symbolic rules, respecting physical laws, and pursuing intentional goals--remains an open question. Existing benchmarks only partially address this, often conflating visual quality with cognitive correctness. We introduce VWG-Bench (Video World Generalist Benchmark), a comprehensive benchmark spanning 9 reasoning dimensions and 38 fine-grained tasks. To enable precise diagnosis, we design a three-level VLM-as-Judge protocol that independently assesses video-level fluency, task-level rule adherence, and sample-level goal realization. Evaluations of leading models reveal a striking gap: while models achieve strong rendering scores, they consistently fail on logic-heavy and rule-constrained tasks. To address this, we propose Vid-PRE (Video Prompt Reasoner and Enhancer), a model-agnostic prompt rewriter that offloads the cognitive burden of reasoning to a dedicated VLM. Trained via reinforcement learning with purely text-based rewards, Vid-PRE produces concise, constraint-aware prompts without the instability of video-level reward signals. Experiments show that Vid-PRE yields substantial reasoning improvements across multiple generators without architectural modifications. Together, VWG-Bench and Vid-PRE offer a rigorous diagnostic lens and a scalable path toward true think-with-video capabilities. All data and code are publicly available at https://huggingface.co/datasets/KlingTeam/VWG-Bench.

Chinese Translation

视频生成技术已经取得了显著进展，能够生成视觉上引人注目且时间上连贯的结果。然而，这些模型是否能够真正进行视频思维——执行符号规则、遵循物理法则并追求意图目标——仍然是一个悬而未决的问题。现有的基准测试仅部分解决了这一问题，常常将视觉质量与认知正确性混为一谈。我们引入了 VWG-Bench（视频世界通用基准），这是一个涵盖 9 个推理维度和 38 个细粒度任务的综合性基准。为了实现精准诊断，我们设计了一个三级 VLM-as-Judge 协议，独立评估视频级流畅性、任务级规则遵循和样本级目标实现。对领先模型的评估揭示了一个显著的差距：尽管模型在渲染评分上表现强劲，但在逻辑密集和规则约束的任务上却始终失败。为了解决这一问题，我们提出了 Vid-PRE（视频提示推理器和增强器），这是一种模型无关的提示重写器，将推理的认知负担转移到专用的 VLM。Vid-PRE 通过强化学习，仅使用基于文本的奖励进行训练，能够生成简洁且考虑约束的提示，而不受视频级奖励信号的不稳定性影响。实验表明，Vid-PRE 在多个生成器上实现了显著的推理改进，而无需对架构进行修改。总之，WVG-Bench 和 Vid-PRE 提供了一个严格的诊断视角和通向真正视频思维能力的可扩展路径。所有数据和代码均可在 https://huggingface.co/datasets/KlingTeam/VWG-Bench 上公开获取。

View on arXiv Download PDF AI Translation

cs.CV / 34 / 2609.11265

Uncertainty DMD: Restoring Diversity in Few-Step Autoregressive Video Distillation

不确定性 DMD：恢复少步自回归视频蒸馏中的多样性

Duan, Zixuan, Xiang, Xunzhi, Chen, Yabo, Zhang, Xin, Liu, Changhan, Huang, Haibin, Zhang, Chi, Fan, Qi, Li, Xuelong

Abstract

Few-step distillation improves the efficiency of autoregressive (AR) video generation, but often causes diversity collapse: under the same prompt, different noise samples tend to produce highly similar videos with weakened motion dynamics. We analyze this degradation in Distribution Matching Distillation (DMD)-distilled AR video generators and find that, in the autoregressive setting, it takes the form of a structured uncertainty collapse: the mode-seeking bias of DMD maps different noise samples to nearly identical first chunks, and the deterministic AR cache then propagates this collapsed state to all subsequent chunks, turning a local loss of stochasticity at the rollout root into a global suppression of temporal variation. Based on this analysis, we propose Uncertainty DMD, a simple uncertainty-injection framework that restores stochasticity at two key stages of AR generation: a timestep perturbation for the first chunk to increase first-chunk diversity, and a stochastic cache-writing mechanism for later chunks to preserve uncertainty in autoregressive conditioning. The method requires no architectural changes and introduces only lightweight perturbation operations. The same perturbation mechanisms are used during both training and inference. Experiments show that Uncertainty DMD consistently improves diversity and motion dynamics while maintaining comparable per-sample visual quality.

Chinese Translation

少步蒸馏提高了自回归（AR）视频生成的效率，但常常导致多样性崩溃：在相同提示下，不同的噪声样本往往会生成高度相似的视频，且运动动态减弱。我们分析了在分布匹配蒸馏（DMD）蒸馏的自回归视频生成器中这种退化现象，发现它在自回归设置中表现为结构化的不确定性崩溃：DMD的模式寻求偏差将不同的噪声样本映射到几乎相同的第一块，而确定性的AR缓存随后将这种崩溃状态传播到所有后续块，将在展开根部的局部随机性损失转化为时间变化的全球抑制。基于这一分析，我们提出了不确定性 DMD，这是一种简单的不确定性注入框架，在AR生成的两个关键阶段恢复随机性：对第一块进行时间步扰动以增加第一块的多样性，以及对后续块采用随机缓存写入机制以保持自回归条件下的不确定性。该方法不需要架构更改，仅引入轻量级的扰动操作。在训练和推理过程中使用相同的扰动机制。实验表明，不确定性 DMD 一致地提高了多样性和运动动态，同时保持了可比的每样本视觉质量。

View on arXiv Download PDF AI Translation

cs.CV / 35 / 2609.11269

Improving Faint Object Detection for Space Situational Awareness with Variational Autoencoders

利用变分自编码器提高空间态势感知中的微弱目标检测

Cratere, Angela, Ghilardi, Luca, Reddy, Vishnu, Dell'Olio, Francesco, Kouzinopoulos, Charalampos S., Furfaro, Roberto

Abstract

We present a deep-learning pipeline for enhancing the detection of faint moving objects in optical space situational awareness (SSA) imagery through automated star removal and background reconstruction. Detecting low signal-to-noise ratio (SNR) objects remains extremely challenging in optical observations, particularly in the cislunar (X-GEO) environment, where structured sky backgrounds, dense stellar fields, and scattered moonlight significantly degrade the performance of classical detection algorithms. To address this problem, the proposed pipeline combines a lightweight segmentation network (Tiny-U-Net) to generate stellar masks with a partial-convolution variational autoencoder (astro-VAE), designed to learn the statistical distribution of astronomical backgrounds and perform context-aware inpainting of masked regions. The reconstructed background maps can then be used as a preprocessing step to suppress fixed sources and background inhomogeneities prior to detection. As a proof of concept, the approach is integrated with a shift-and-stack scheme and evaluated on real ground-based telescope observations targeting the X-GEO region. Results demonstrate that the method reconstructs star-free backgrounds with high fidelity, while preserving moving targets and significantly enhancing detectability, thereby providing an effective data-driven preprocessing strategy for faint moving-object detection in optical SSA scenarios.

Chinese Translation

我们提出了一种深度学习流程，通过自动去除星体和背景重建，增强光学空间态势感知（SSA）图像中微弱移动目标的检测。在光学观测中，检测低信噪比（SNR）目标仍然极具挑战性，特别是在月球轨道（X-GEO）环境中，结构化的天空背景、密集的恒星场和散射的月光显著降低了经典检测算法的性能。为了解决这个问题，所提出的流程结合了轻量级分割网络（Tiny-U-Net）生成恒星掩膜，并使用部分卷积变分自编码器（astro-VAE），旨在学习天文背景的统计分布并对掩膜区域进行上下文感知的修复。重建的背景图可以作为预处理步骤，抑制固定源和背景不均匀性，从而在检测之前进行处理。作为概念验证，该方法与移位叠加方案集成，并在针对X-GEO区域的真实地面望远镜观测中进行了评估。结果表明，该方法能够高保真地重建无星背景，同时保留移动目标并显著增强可检测性，从而为光学SSA场景中的微弱移动目标检测提供了一种有效的数据驱动预处理策略。

View on arXiv Download PDF AI Translation

cs.CV / 36 / 2609.11271

Order-Aware 2.5D Multiple Instance Learning for Preoperative MRI-Based Perineural Invasion Risk Assessment in Intrahepatic Cholangiocarcinoma

基于顺序感知的2.5D多实例学习用于肝内胆管癌术前MRI评估神经周围侵犯风险

Go, Hyunsu, Han, Youngung, Kim, Kyeonghun, Jun, Jinyong, Lee, Junbeom, Kweon, Dohyun, Jeong, Yului, Park, Suah, Park, Sungha, Jung, Anna, Jeong, Woo Kyoung, Liao, Ken Ying-Kai, Lee, Hyuk-Jae, Kim, Nam-Joon

Abstract

Perineural invasion (PNI) is an adverse histopathologic marker in intrahepatic cholangiocarcinoma (ICC), but it is usually confirmed only after resection. Preoperative T2-weighted MRI may provide noninvasive imaging cues predictive of PNI, although labels are available only at the patient level without slice- or voxel-level annotations. We propose Order-Aware Slab Multiple Instance Learning (OAS-MIL), a weakly supervised framework for patient-level PNI prediction. Each tumor-centered MRI crop is represented as an ordered sequence of overlapping 2.5D slabs formed from contiguous axial slices. A shared encoder extracts slab-level features, which are aggregated by a permutation-invariant set-attention branch and a bidirectional sequence-attention branch. Using five-fold label-stratified cross-validation at the patient level, OAS-MIL achieved a mean AUROC of 0.770, outperforming the evaluated volumetric and MIL baselines. These results suggest that axial order provides a useful inductive bias for weakly supervised PNI prediction from MRI.

Chinese Translation

神经周围侵犯（PNI）是肝内胆管癌（ICC）中的一种不良组织病理标志，但通常仅在切除后确认。术前T2加权MRI可能提供预测PNI的非侵入性影像线索，尽管标签仅在患者层面可用，而没有切片或体素级的注释。我们提出了一种顺序感知的板块多实例学习（Order-Aware Slab Multiple Instance Learning, OAS-MIL）框架，用于患者级PNI预测。每个肿瘤中心的MRI裁剪被表示为由相邻轴向切片形成的重叠2.5D板块的有序序列。共享编码器提取板块级特征，这些特征通过一个不变于排列的集合注意力分支和一个双向序列注意力分支进行聚合。在患者层面使用五折标签分层交叉验证，OAS-MIL达到了0.770的平均AUROC，优于评估的体积和多实例学习基线。这些结果表明，轴向顺序为从MRI进行弱监督PNI预测提供了有用的归纳偏差。

View on arXiv Download PDF AI Translation

cs.CV / 37 / 2609.11279

SAMV-DUSt3R: Instance-Centric 3D Scene Decoupling from Sparse Multi-Views

SAMV-DUSt3R：基于实例的稀疏多视角3D场景解耦

Zhao, Langxu, Gu, Zuan, Zhang, Yingdan, Zhao, Pengfei, Gao, Tianhan

Abstract

With the rising demand to decouple objects from 3D scenes, we propose SAMV-DUSt3R, an end-to-end model that injects SAM2 2D masks into MV-DUSt3R reconstruction. A Cross Flow Mask Block uses these masks to steer the network toward the target instance, jointly improving shape accuracy and achieving object-level disentanglement without multi-stage pipelines. To ensure reconstruction stability, a lightweight Spatial RankGNN selects the optimal reference view with a selection accuracy of 73.5\%. Extensive experiments demonstrate that our method boosts average reconstruction precision by 11\% across various metrics compared to state-of-the-art baselines. These results reveal a strong instance-disentanglement capability and clear benefits for driving, robotics, AR/VR, and heritage digitisation.

Chinese Translation

随着对从3D场景中解耦对象需求的增加，我们提出了SAMV-DUSt3R，这是一种端到端模型，将SAM2 2D掩膜注入MV-DUSt3R重建中。交叉流掩膜块利用这些掩膜引导网络朝向目标实例，联合提高形状准确性，并在没有多阶段管道的情况下实现对象级解耦。为了确保重建的稳定性，一个轻量级的空间RankGNN选择最优参考视图，选择准确率达到73.5%。大量实验表明，与最先进的基线相比，我们的方法在各种指标上平均提高了11%的重建精度。这些结果揭示了强大的实例解耦能力，并为驾驶、机器人技术、增强现实/虚拟现实以及遗产数字化带来了明显的好处。

View on arXiv Download PDF AI Translation

cs.CV / 38 / 2609.11310

Your Model Already Knows Don't Teach It, Learn to Ask It: Soft Prompting for Few-Shot Adaptation of Vision-Language Models

你的模型已经知道，不要教它，学会问它：视觉-语言模型的少样本适应的软提示

Gare, Gautam Rajendrakumar, Li, Siyi, Wang, Hewei, Hernandez, Cesar Daniel, Zhao, Wei, Pauli, Wolfgang M., Galeotti, John, Ramanan, Deva

Abstract

We address few-shot object detection with vision-language models (VLMs) in out-of-domain settings such as aerial, industrial, and medical imagery, using only ten annotated images for supervision. Existing adaptation methods are discrete prompt optimization and LoRA fine-tuning. We revisit a third option: soft prompting, where a small number of continuous prompt tokens are optimized while the pretrained backbone remains frozen. We identify two key design choices. First, placing prompt tokens at the cross-modal boundary between visual and text tokens outperforms other placements (10.0 vs. 8.4 mAP). Second, initializing prompts from the empty space token outperforms semantic and random initialization. With these choices, one to three learned tokens (7,168 parameters on average) match the best LoRA configuration on Roboflow20-VL (14.2 mAP, 10-shot) while training over 20,000x fewer parameters. Soft prompting remains harder to optimize, exhibiting higher variance across random seeds. Unlike LoRA, however, it causes no forgetting: the LoRA rank matching our accuracy reduces NaturalBench VQA accuracy by 35% relative, rising to 56% at the largest rank, whereas soft prompting leaves pretrained performance unchanged. The learned tokens behave like prompts rather than weights. They transfer to a newer model without retraining (+0.8 mAP on Qwen3.5-9B) and can be verbalized into readable prompts competitive with prompt-search methods (matching DetPO and outperforming GEPA). The approach also extends beyond detection. On RoboCasa manipulation tasks, the frozen $\pi_{0.5}$ vision-language-action policy benefits from soft prompting, matching the LoRA baseline on two of three tasks when tokens are placed at the gradient bottleneck. These results suggest modern VLMs already encode much of what is needed for specialized domains; the challenge is learning how to ask.

Chinese Translation

我们在超出领域的环境中，如航空、工业和医学影像，使用仅十张标注图像进行监督，研究视觉-语言模型（VLMs）的少样本目标检测。现有的适应方法包括离散提示优化和LoRA微调。我们重新审视第三种选择：软提示，其中少量连续的提示令牌在预训练的主干保持不变的情况下进行优化。我们确定了两个关键设计选择。首先，将提示令牌放置在视觉和文本令牌之间的跨模态边界上，优于其他放置方式（10.0 vs. 8.4 mAP）。其次，从空白空间令牌初始化提示优于语义和随机初始化。通过这些选择，学习到的一到三个令牌（平均7,168个参数）在Roboflow20-VL上与最佳的LoRA配置相匹配（14.2 mAP，10-shot），同时训练的参数数量减少了20,000倍以上。软提示的优化仍然较为困难，表现出更高的随机种子方差。然而，与LoRA不同的是，它不会导致遗忘：与我们的准确性匹配的LoRA秩使NaturalBench VQA的准确性相对降低了35%，在最大秩时上升至56%，而软提示则保持了预训练性能不变。学习到的令牌更像提示而非权重。它们可以在不重新训练的情况下转移到更新的模型上（在Qwen3.5-9B上提高0.8 mAP），并且可以被转化为可读的提示，与提示搜索方法相竞争（与DetPO匹配并优于GEPA）。该方法还扩展到检测以外的领域。在RoboCasa操作任务中，冻结的$ ext{π}_{0.5}$视觉-语言-行动策略受益于软提示，当令牌放置在梯度瓶颈时，在三项任务中的两项上与LoRA基线相匹配。这些结果表明，现代VLMs已经编码了许多专门领域所需的内容；挑战在于学习如何提问。

View on arXiv Download PDF AI Translation

cs.CV / 39 / 2609.11312

GRIPNet: Gaussian Radial Intensity Prior Guided Architecture for Pulmonary Nodule Detection in CT

GRIPNet：基于高斯径向强度先验引导的CT肺结节检测架构

Yang, Haojie, Su, Ran

Abstract

Lung cancer causes more deaths than any other malignancy, and low-dose CT screening is the main pathway to early diagnosis. That pathway hinges on the smallest lesions, yet nodules below six millimeters remain hard to detect, because most methods treat a nodule as a generic object and ignore the imaging physics behind its appearance. We show that this appearance is highly regular. Intensity peaks at the geometric center of a nodule and decays radially in a Gaussian pattern, and a fit to 18,218 annotated lesions from three public benchmarks yields a mean radial coefficient of determination above 0.86 in every dataset and size stratum. A square convolution samples both axes uniformly and is mismatched to this radial signal, most severely for small nodules. Guided by this evidence, we propose GRIPNet (Gaussian Radial Intensity Prior Network), a detector in which every module maps to a measurable property of the intensity distribution. Pinwheel convolutions decompose radial gradients, a dual-frequency module separates boundary detail from structural context, dilated masked attention matches the decay extent, and an adaptive loss reweights samples by conspicuity. GRIPNet raises [email protected] to 95.3, 91.6 and 97.9 percent on KanserSet, LUNA16 and Lung-PET-CT-Dx while sharpening high-IoU localization at real-time speed.

Chinese Translation

肺癌导致的死亡人数超过任何其他恶性肿瘤，而低剂量CT筛查是早期诊断的主要途径。该途径依赖于最小的病变，但直径小于六毫米的结节仍然难以检测，因为大多数方法将结节视为通用对象，忽视了其外观背后的成像物理。我们表明，这种外观是高度规律的。结节几何中心的强度达到峰值，并以高斯模式径向衰减，对来自三个公共基准的18,218个标注病变的拟合结果显示，在每个数据集和大小层次中，平均径向决定系数超过0.86。方形卷积在两个轴上均匀采样，与这种径向信号不匹配，尤其是对于小结节。基于这一证据，我们提出了GRIPNet（高斯径向强度先验网络），该检测器中的每个模块映射到强度分布的可测量属性。旋风卷积分解径向梯度，双频模块将边界细节与结构上下文分离，扩张掩膜注意力匹配衰减范围，自适应损失根据显著性重新加权样本。GRIPNet在KanserSet、LUNA16和Lung-PET-CT-Dx上的[email protected]分别提高到95.3%、91.6%和97.9%，同时以实时速度增强高IoU定位。

View on arXiv Download PDF AI Translation

cs.CV / 40 / 2609.11317

Mi-Ripple: Restoring Images Degraded by Iterative AI Editing

Mi-Ripple：恢复因迭代人工智能编辑而退化的图像

Chen, Jiayin, Xu, Yicheng, Wang, Muting

Abstract

Iterative reference-conditioned image editing can introduce grid-like and granular textures, commonly described as digital ripple. We present Mi-Ripple, a diagnosis-guided restoration workflow that suppresses this digital ripple while protecting image structure. Mi-Ripple separates periodic lattice artifacts from content-entangled granular texture, then combines selective spectral notching, structure-aware smoothing, and cleaned-reference regeneration. This separation enables low-distortion filtering when artifacts are spectrally isolated and visual reconstruction when filtering would erase legitimate detail. Across fourteen notch-only executions, whole-image residual standard deviation is 0.08--0.44 in CIELAB lightness units. In a paired regeneration example, reference cleaning reduces output debris density by 45\%. Mi-Ripple links measurable artifact reduction to visibly cleaner generated images, rather than optimizing a spectral score alone.

Chinese Translation

迭代参考条件的图像编辑可能会引入网格状和颗粒状纹理，通常被描述为数字涟漪。我们提出了Mi-Ripple，这是一种基于诊断的恢复工作流程，旨在抑制这种数字涟漪，同时保护图像结构。Mi-Ripple将周期性晶格伪影与内容纠缠的颗粒纹理分离，然后结合选择性频谱缺口、结构感知平滑和清洁参考再生。这种分离使得在伪影频谱隔离时能够进行低失真过滤，而在过滤会抹去合法细节时则进行视觉重建。在十四次仅进行缺口处理的执行中，整个图像的残差标准差为0.08--0.44（CIELAB亮度单位）。在一个配对再生示例中，参考清理将输出碎片密度降低了45%。Mi-Ripple将可测量的伪影减少与明显更干净的生成图像联系起来，而不仅仅是优化频谱评分。

View on arXiv Download PDF AI Translation

cs.CV / 41 / 2609.11330

Predictive Multi-Landmark OCT Tracking for Increased Motion Robustness

增强运动鲁棒性的预测多标志物OCT跟踪

Reuter, Konrad, Guttikonda, Suresh, Karekar, Chaitali Uday, Betz, Christian, Schlaefer, Alexander

Abstract

Optical coherence tomography is a promising modality for markerless motion tracking due to its high spatial resolution and inherent depth perception. However, existing OCT-based tracking approaches are limited in terms of trackable velocity, particularly when multiple landmarks are tracked sequentially for 6D pose estimation. In this work, we present a predictive tracking approach that propagates positional updates between multiple tracked landmarks to obtain a global pose prediction. This enables more robust tracking under high velocities. Our results demonstrate RMSEs below 1 mm for velocities up to 100 mm/s and up to nine consecutively tracked landmarks, highlighting the potential of global motion propagation and prediction for improving the robustness of OCT-based tracking.

Chinese Translation

光学相干断层扫描（OCT）由于其高空间分辨率和固有的深度感知，成为一种有前景的无标记运动跟踪方式。然而，现有的基于OCT的跟踪方法在可跟踪速度方面存在限制，特别是在对多个标志物进行顺序跟踪以进行6D姿态估计时。在本研究中，我们提出了一种预测跟踪方法，该方法在多个跟踪标志物之间传播位置更新，以获得全局姿态预测。这使得在高速度下实现更为鲁棒的跟踪成为可能。我们的结果表明，对于高达100 mm/s的速度和最多九个连续跟踪的标志物，均方根误差（RMSE）低于1 mm，突显了全局运动传播和预测在提高基于OCT的跟踪鲁棒性方面的潜力。

View on arXiv Download PDF AI Translation

cs.CV / 42 / 2609.11360

R4Tun: LLM-guided adaptive segmental tunnel lining segmentation in point clouds

R4Tun：基于大语言模型的点云分段隧道衬砌自适应分割

Tao, Xinghui, Ye, Zehao, Wang, Guangming, Ninić, Jelena, Sheil, Brian

Abstract

Automated inspection of segmental tunnel linings requires adaptive segmentation from 3D point clouds, yet expert-tuned pipelines often degrade when tunnel conditions vary. This paper presents R4Tun, a large language model (LLM)-driven adaptation framework that extends an expert-designed pipeline (SAM4Tun) with bounded parameter tuning informed by structured context: memory ($m$), state ($s$), and knowledge ($k$). Evaluated on 30 selected Seg2Tunnel subsets (13 regular, 17 complex) across three LLMs, the full $m+s+k$ design raised mean Intersection-over-Union (mIoU) from 0.18 to 0.43--0.48 and overall accuracy (OA) from 0.42 to 0.59--0.65 relative to the static SAM4Tun baseline, with the near-reference regular (staggered) subsets reaching mIoU 0.784--0.796 across LLMs. Across 270 (30 tunnels $\times$ 3 different LLMs $\times$ 3 context settings) runs, the LLMs showed similar parameter-adjustment trends (with overlapping 95\% CIs on mean gains) and consistently adjusted a shared set of critical parameters. These results support R4Tun as a controlled, label-free, cross-LLM adaptation mechanism in the tested SAM4Tun--Seg2Tunnel setting, demonstrating consistent accuracy gains; we position R4Tun as a mechanism contribution rather than a deployable final-inspection system, in which each bounded parameter change is auditable via logged rationales.

Chinese Translation

分段隧道衬砌的自动检测需要从三维点云中进行自适应分割，然而，当隧道条件变化时，专家调优的流程往往会退化。本文提出了R4Tun，一个由大语言模型（LLM）驱动的适应框架，它通过结构化上下文（记忆 $m$、状态 $s$ 和知识 $k$）扩展了一个专家设计的流程（SAM4Tun），并进行了有界参数调优。在对30个选定的Seg2Tunnel子集（13个常规，17个复杂）进行评估时，完整的 $m+s+k$ 设计使得平均交并比（mIoU）从0.18提升至0.43--0.48，整体准确率（OA）从0.42提升至0.59--0.65，相较于静态的SAM4Tun基线，近似参考的常规（错位）子集在各LLM中达到了mIoU 0.784--0.796。在270次（30个隧道 $ imes$ 3种不同LLM $ imes$ 3种上下文设置）运行中，LLM表现出相似的参数调整趋势（在平均增益上具有重叠的95 ext{%}置信区间），并始终调整一组共享的关键参数。这些结果支持R4Tun作为在测试的SAM4Tun--Seg2Tunnel设置中一种受控的、无标签的跨LLM适应机制，展示了一致的准确性提升；我们将R4Tun定位为一种机制贡献，而非可部署的最终检测系统，其中每个有界参数的变化都可以通过记录的理由进行审计。

View on arXiv Download PDF AI Translation

cs.CV / 43 / 2609.11375

Vision Transformer-Based Multi-Level Feature Fusion for Multi-Label Sewer Defect Classification

基于视觉变换器的多层次特征融合用于多标签下水道缺陷分类

Fang, Xu, Wang, Zhuoran, Li, Qing, Zhang, Shengyu, Deng, Guanzhi, He, Jianbiao, Li, Qingquan

Abstract

Automated classification of sewer defects is essential for infrastructure condition assessment and maintenance decision-making, but existing deep learning methods struggle to balance classification accuracy and computational complexity in large-scale multi-label scenarios. This study develops Sewer-Transformer-ML, a hierarchical vision Transformer with multi-level feature fusion, together with two lightweight architectures, Sewer-MobileNet-ML and Sewer-Mobile-TransNet, for resource-constrained inspection scenarios. On the Sewer-ML test set, Sewer-Transformer-ML-Base achieved an $F2_{\text{CIW}}$ of 65.68% and an $F1_{\text{Normal}}$ of 92.68%, ranking first on the public leaderboard and exceeding the second-ranked method by 7.6 percentage points in $F2_{\text{CIW}}$. Sewer-MobileNet-ML achieved an $F2_{\text{CIW}}$ of 65.73% with only 17 M parameters, representing an approximately 95% parameter reduction relative to the base model. Under the standard Sewer-Capsule data split, Sewer-Mobile-TransNet achieved 96.43% classification accuracy. When the training set was reduced to 1,177 images, pretraining on Sewer-ML consistently improved model performance. Ablation experiments further showed that direct concatenation was more effective for Transformer features, whereas attention-based fusion better supported multiscale CNN features. These findings provide a computational basis for automated sewer inspection, lightweight model design, and adaptation across civil infrastructure inspection platforms.

Chinese Translation

下水道缺陷的自动分类对于基础设施状况评估和维护决策至关重要，但现有的深度学习方法在大规模多标签场景中难以平衡分类准确性和计算复杂性。本研究开发了Sewer-Transformer-ML，一种具有多层次特征融合的分层视觉变换器，以及两种轻量级架构Sewer-MobileNet-ML和Sewer-Mobile-TransNet，适用于资源受限的检查场景。在Sewer-ML测试集上，Sewer-Transformer-ML-Base达到了65.68%的$F2_{ ext{CIW}}$和92.68%的$F1_{ ext{Normal}}$，在公共排行榜上排名第一，$F2_{ ext{CIW}}$超出第二名方法7.6个百分点。Sewer-MobileNet-ML仅用17M参数实现了65.73%的$F2_{ ext{CIW}}$，相较于基线模型约减少了95%的参数。在标准的Sewer-Capsule数据划分下，Sewer-Mobile-TransNet达到了96.43%的分类准确率。当训练集减少到1,177张图像时，在Sewer-ML上的预训练持续改善了模型性能。消融实验进一步表明，直接拼接对变换器特征更为有效，而基于注意力的融合更好地支持多尺度CNN特征。这些发现为自动化下水道检查、轻量级模型设计以及在市政基础设施检查平台上的适应提供了计算基础。

View on arXiv Download PDF AI Translation

cs.CV / 44 / 2609.11378

Brain-PACE: A Deep Siamese MRI Framework for Modelling Longitudinal Brain Acceleration

Brain-PACE：一种深度Siamese MRI框架用于建模纵向脑加速

Maddox, Samuel, Newman, Jacob, Sami, Saber, Mackiewicz, Michal, Initiative, for the Alzheimer's Disease Neuroimaging, Biomarkers, the Australian Imaging, ageing, Lifestyle flagship study of

Abstract

Brain age estimation has become a popular research proxy for assessing brain health and disease, yet longitudinal trajectories of brain ageing are still poorly defined, and clinical use is limited. Building on existing Siamese longitudinal frameworks, we develop Brain-Predicted Age Acceleration (Brain-PACE) to directly estimate the pace of structural brain ageing from paired T1-weighted MRI. Brain-PACE identified accelerated ageing in $42.6$% of participants with mild cognitive impairment. Faster Brain-PACE was associated with greater functional and cognitive impairment (FAQ; $r=0.35$, ADAS13; $r=0.30$, CDR-SB; $r=0.32$) and greater regional tau burden in the posterior cingulate ($r=0.59$), precuneus ($r=0.47$), and entorhinal cortex ($r=0.37$). These associations were stronger than those observed when pace was calculated indirectly from repeated cross-sectional brain age estimates, suggesting that direct longitudinal modelling captures complementary information relevant to ongoing pathological change. Methodologically, Brain-PACE extends the LILAC framework by combining spatial attention with soft label distribution learning and a Cram\'er distance objective, improving probabilistic performance and reducing prediction bias while providing measures of predictive uncertainty. Together, these findings support Brain-PACE as a complementary longitudinal imaging phenotype with sensitivity to relevant clinical and biological changes in early neurodegeneration.

Chinese Translation

脑龄估计已成为评估脑健康和疾病的热门研究代理，但脑老化的纵向轨迹仍然不够明确，临床应用受到限制。在现有的Siamese纵向框架基础上，我们开发了脑预测年龄加速（Brain-PACE），以直接从配对的T1加权MRI中估计结构性脑老化的速度。Brain-PACE在42.6%的轻度认知障碍参与者中识别出加速老化。更快的Brain-PACE与更大的功能和认知障碍（FAQ；r=0.35，ADAS13；r=0.30，CDR-SB；r=0.32）以及在后扣带皮层（r=0.59）、前顶叶（r=0.47）和内嗅皮层（r=0.37）中更大的区域tau负担相关。这些关联比通过重复横断面脑龄估计间接计算的速度观察到的更强，表明直接的纵向建模捕捉到了与持续病理变化相关的互补信息。在方法论上，Brain-PACE通过结合空间注意力、软标签分布学习和Cramér距离目标扩展了LILAC框架，提高了概率性能并减少了预测偏差，同时提供了预测不确定性的度量。这些发现共同支持Brain-PACE作为一种敏感于早期神经退行性变相关临床和生物变化的互补纵向影像表型。

View on arXiv Download PDF AI Translation

cs.CV / 45 / 2609.11380

DINO-Med: A Unified Patch-Based Adaptation Framework for Multi-Modal Medical Image Analysis Applied to Liver Fibrosis Staging

DINO-Med：一种用于多模态医学图像分析的统一基于补丁的适应框架，应用于肝纤维化分期

Wang, Boya, Li, Ruizhe, Chen, Chao, Chen, Xin

Abstract

Adapting natural-image foundation models like DINOv3 to multi-modal medical imaging is challenging due to the significant domain gap between natural color images and multi-channel medical scans. We present a unified, patch-based framework that processes raw multimodal imaging through training-free registration, automated localization, and mask-filtered patch extraction. This architecture culminates in a hierarchical strategy that aggregates patch-level insights into subject-level diagnostics. Using liver fibrosis staging as a case study, we evaluate four patch-level feature representations: handcrafted Radiomics features, learned ResNet features, pre-trained foundation model SAM-Med2D features, and frozen DINOv3 features. To ensure a controlled comparison, all models utilize the same lightweight MLP head and are evaluated across both rigid and deformable registration settings. Our training protocol focuses on mild fibrosis (S1) and cirrhosis (S4) classes only, enabling a single classifier to address both substantial fibrosis detection and cirrhosis staging. Evaluated via 10 random train (90%)/ test (10%) splits on 360 subjects from the CARE 2025 Liver Track 4 cohort, our DINOv3-based framework significantly outperforms all baselines, achieving the best classification accuracy of 78.4% for S1 and 75.8% for S4.

Chinese Translation

将自然图像基础模型如DINOv3适应于多模态医学成像面临挑战，因为自然彩色图像与多通道医学扫描之间存在显著的领域差距。我们提出了一种统一的基于补丁的框架，通过无训练的配准、自动定位和掩膜过滤的补丁提取来处理原始多模态成像。该架构 culminates 在一种层次化策略中，将补丁级别的见解聚合为受试者级别的诊断。以肝纤维化分期为案例研究，我们评估了四种补丁级特征表示：手工制作的Radiomics特征、学习的ResNet特征、预训练的基础模型SAM-Med2D特征和冻结的DINOv3特征。为了确保可控比较，所有模型均使用相同的轻量级MLP头，并在刚性和可变形配准设置下进行评估。我们的训练协议仅关注轻度纤维化（S1）和肝硬化（S4）类别，使得单个分类器能够同时处理显著纤维化检测和肝硬化分期。在对来自CARE 2025 Liver Track 4队列的360名受试者进行10次随机训练（90%）/测试（10%）分割的评估中，我们基于DINOv3的框架显著优于所有基线，S1的最佳分类准确率达到78.4%，S4的最佳分类准确率为75.8%。

View on arXiv Download PDF AI Translation

cs.CV / 46 / 2609.11439

Multi-Modal Controlled Coherent Motion Generation

多模态控制一致性运动生成

Liu, Yifei, Cao, Qiong, Yi, Hongwei, Jiang, Huaiguang, Ding, Changxing

Abstract

It is natural for humans to walk and talk simultaneously. This paper tackles the challenge of replicating such natural behaviors in 3D avatar motion generation driven by concurrent multimodal inputs, such as a text description of a man walking alongside speech audio. Existing methods, constrained by the scarcity of aligned multimodal data, typically combine motions from individual modalities sequentially or through weighted sums. However, they often result in mismatched or unrealistic movements. To overcome these limitations, we propose MOCO, a novel diffusion-based framework capable of processing multiple simultaneous inputs, including speech audio, text descriptions, and trajectory data, to generate coherent and lifelike motions without requiring aligned multimodal data. Our key innovation lies in decoupling the motion generation process. During each denoising step, the diffusion model independently generates motions for each modality from the input noise and assembles the body parts according to predefined spatial rules. The resulting combined motion is then diffused and serves as the input noise for the subsequent denoising step. This iterative approach enables each modality to refine its contribution within the context of the overall motion, progressively harmonizing movements across modalities. Consequently, the generated motions become increasingly natural and fluid with each iteration, achieving coherent and synchronized behaviors. We evaluate our approach using a purpose-built multimodal benchmark. Experimental results demonstrate that MOCO outperforms existing baselines, advancing the field of multimodal motion generation for 3D avatars.

Chinese Translation

人类同时行走和说话是自然而然的行为。本文解决了在3D虚拟形象运动生成中复制这种自然行为的挑战，该生成过程由同时的多模态输入驱动，例如描述一个人行走的文本和语音音频。现有方法由于缺乏对齐的多模态数据，通常是通过顺序组合各个模态的运动或加权求和。然而，这些方法往往导致运动不匹配或不真实。为克服这些限制，我们提出了MOCO，一个新颖的基于扩散的框架，能够处理多个同时输入，包括语音音频、文本描述和轨迹数据，从而生成一致且逼真的运动，而无需对齐的多模态数据。我们的关键创新在于解耦运动生成过程。在每个去噪步骤中，扩散模型独立地从输入噪声生成每个模态的运动，并根据预定义的空间规则组装身体部位。生成的组合运动随后被扩散，并作为后续去噪步骤的输入噪声。该迭代方法使每个模态能够在整体运动的背景下细化其贡献，逐步协调各模态之间的运动。因此，生成的运动在每次迭代中变得越来越自然和流畅，实现了一致和同步的行为。我们使用一个专门构建的多模态基准评估我们的方法。实验结果表明，MOCO的表现优于现有基线，推动了3D虚拟形象多模态运动生成领域的发展。

View on arXiv Download PDF AI Translation

cs.CV / 47 / 2609.11463

BruNet: A Cross-Domain Transfer Framework for Bruise Segmentation

BruNet：一种用于淤伤分割的跨领域迁移框架

Wang, Qiming, Motley, Richard J., Obi, Ebube E., Sun, Xianfang, Rosin, Paul L.

Abstract

Segmenting bruises is a challenging task in medical imaging due to limited data and annotations, diffuse boundaries, and highly variable appearance. In this work, we propose BruNet, a segmentation framework that combines a ViT-based visual encoder (a self-supervised DINOv3 or a pretrained LingBot-Vision backbone) with a SAM-based mask decoder. BruNet is trained on the HAM10000 skin lesion dataset and evaluated on a separate bruise dataset without additional fine-tuning. Although a small number of prior studies have explored machine learning and computer vision for bruise analysis, existing work has primarily focused on detection, classification, or colour analysis rather than pixel-level localisation. To the best of our knowledge, this is the first study to address automatic bruise segmentation. Our results show that BruNet outperforms CNN-based models, state-of-the-art segmentation models, ChatGPT-4o/5-assisted SAM2 zero-shot baselines, and the medical-oriented MedSAM model, demonstrating strong cross-domain generalisation to bruise segmentation.

Chinese Translation

由于数据和注释的有限性、边界模糊以及外观高度可变，淤伤分割在医学成像中是一项具有挑战性的任务。在本研究中，我们提出了BruNet，这是一种分割框架，结合了基于视觉变换器（ViT）的视觉编码器（自监督的DINOv3或预训练的LingBot-Vision主干）与基于SAM的掩膜解码器。BruNet在HAM10000皮肤病变数据集上进行训练，并在一个独立的淤伤数据集上进行评估，而无需额外的微调。尽管少数先前研究探讨了机器学习和计算机视觉在淤伤分析中的应用，但现有工作主要集中在检测、分类或颜色分析上，而非像素级定位。根据我们的最佳知识，这是首个针对自动淤伤分割的研究。我们的结果表明，BruNet在性能上优于基于卷积神经网络（CNN）的模型、最先进的分割模型、ChatGPT-4o/5辅助的SAM2零样本基线以及面向医学的MedSAM模型，展示了其在淤伤分割任务上的强大跨领域泛化能力。

View on arXiv Download PDF AI Translation

cs.CV / 48 / 2609.11472

BridgeMatch: Conditional Transport Bridges in Matching Matrix Space for 3D Deformable Registration

BridgeMatch：匹配矩阵空间中的条件传输桥用于三维可变形配准

Wu, Qianliang, Jiang, Haobo, Gao, Guangwei, Chen, Shuo, Xie, Jin, Yang, Jian, Ding, Yaqing

Abstract

Reliable non-rigid point cloud correspondences are important for deformable anatomical registration, embodied perception and manipulation, and dynamic 3D reconstruction. Coarse-to-fine methods reduce computational cost by selecting the top-$K$ coarse regions. However, this pruning may remove weak but correct hypotheses and restrict fine matching to an incomplete search space. We present \paper, a two-stage generative solver that maintains the complete soft matching matrix at both coarse and high resolutions. Stage~I uses denoising diffusion to estimate a global matching matrix in the compact coarse-resolution space. We then lift this matrix to high resolution while preserving its hierarchy. The lifted matrix is rank-bounded and block-constant. Stage~II refines it through a conditional transport bridge. We implement the bridge with two types of dynamics: a deterministic endpoint-parameterized conditional Flow Matching (CFM) ODE and a stochastic Brownian-bridge SDE inspired by Schr\"odinger bridges. Both variants share the lifted source, a time-conditioned transformer, and a matching-matrix endpoint predictor. Experiments on 4DMatch and 4DLoMatch show that both variants produce more accurate correspondences than the compared methods and improve downstream registration, with larger gains in low-overlap cases. They also improve cross-dataset generalization on CAPE and DeepDeform without target-domain adaptation while using the same deformation solver.

Chinese Translation

可靠的非刚性点云对应关系对于可变形解剖配准、体现感知与操作以及动态三维重建至关重要。粗到细的方法通过选择前K个粗略区域来降低计算成本。然而，这种修剪可能会去除弱但正确的假设，并限制细致匹配在不完整的搜索空间中进行。我们提出了 extit{BridgeMatch}，一种两阶段生成求解器，在粗分辨率和高分辨率下保持完整的软匹配矩阵。第一阶段使用去噪扩散在紧凑的粗分辨率空间中估计全局匹配矩阵。然后，我们在保持其层次结构的同时将该矩阵提升至高分辨率。提升后的矩阵是秩有界且块常数的。第二阶段通过条件传输桥对其进行细化。我们使用两种类型的动力学实现该桥：一种是确定性端点参数化的条件流匹配（Conditional Flow Matching, CFM）常微分方程（ODE），另一种是受薛定谔桥启发的随机布朗桥随机微分方程（SDE）。这两种变体共享提升的源、一个时间条件的变换器和一个匹配矩阵端点预测器。在4DMatch和4DLoMatch上的实验表明，这两种变体产生的对应关系比比较方法更准确，并改善下游配准，在低重叠情况下获得更大的提升。它们还在CAPE和DeepDeform上改善了跨数据集的泛化能力，而无需目标领域适应，同时使用相同的变形求解器。

View on arXiv Download PDF AI Translation

cs.CV / 49 / 2609.11477

Pre- and Post-Treatment Brain Metastases Segmentation Using nnU-Net with Post-Processing for BraTS 2026

使用 nnU-Net 和后处理方法对 BraTS 2026 的治疗前后脑转移瘤进行分割

Liu, Haobin, Wang, Xin

Abstract

Brain metastases exhibit high inter-lesion variability in size, enhancement pattern, and post-treatment appearance, making volumetric segmentation of both pre- and post-treatment cases the central challenge of the BraTS 2026 Task 1 (Brain Metastases). We build a pragmatic pipeline on a 5-fold nnU-Net ResEnc-L ensemble, in which each fold is trained independently for 1,000 epochs with the standard Dice + cross-entropy loss on 1,296 four-modality training cases. This ensemble is followed by a rule-based post-processing cascade tuned for the lesion-wise Dice similarity coefficient (LW-DSC), a detection-oriented metric that behaves very differently from the traditional global Dice. The final pipeline reaches an LW-DSC of 0.733 / 0.751 / 0.713 / 0.549 on the enhancing tumour (ET), tumour core (TC), whole tumour (WT), and resection cavity (RC) sub-regions on the official validation leaderboard. Rather than trusting these leaderboard gains, we audit every post-processing stage with a five-fold out-of-fold (OOF) analysis with no model-training leakage over all 1,296 training cases, scored with the official BraTS evaluation code (BraTS_evaluation): it confirms two stages as robust, per-fold-consistent improvements while the third improves only the leaderboard and does not reproduce out-of-fold. We further provide a mechanistic analysis of the LW-DSC metric that explains why recall-recovering post-processing carries low risk whereas component deletion does not, and we report thirteen negative results spanning loss engineering, alternative backbones, and inference-time settings, several of which run counter to widely held intuitions. Source code is released under Apache-2.0 at https://github.com/hornbeamliu/brats2026-met.

Chinese Translation

脑转移瘤在大小、增强模式和治疗后外观上表现出高度的病灶间变异性，使得对治疗前后病例进行体积分割成为 BraTS 2026 任务 1（脑转移瘤）的核心挑战。我们建立了一个务实的管道，基于 5 折 nnU-Net ResEnc-L 集成，其中每一折独立训练 1,000 个周期，使用标准的 Dice + 交叉熵损失在 1,296 个四模态训练案例上进行训练。该集成后跟随一个基于规则的后处理级联，针对病灶级 Dice 相似系数（LW-DSC）进行调优，这是一种与传统全局 Dice 行为截然不同的检测导向指标。最终管道在官方验证排行榜上在增强肿瘤（ET）、肿瘤核心（TC）、整体肿瘤（WT）和切除腔（RC）子区域达到了 LW-DSC 分别为 0.733 / 0.751 / 0.713 / 0.549。我们并不完全依赖这些排行榜的提升，而是对每个后处理阶段进行了五折外折（OOF）分析，确保在所有 1,296 个训练案例中没有模型训练泄漏，并使用官方 BraTS 评估代码（BraTS_evaluation）进行评分：结果确认两个阶段为稳健的、每折一致的改进，而第三个阶段仅提升了排行榜，并未在外折中重现。我们进一步提供了 LW-DSC 指标的机制分析，解释了为何召回恢复的后处理风险较低，而组件删除则不然，并报告了涵盖损失工程、替代骨干网络和推理时间设置的十三个负面结果，其中一些与广泛持有的直觉相悖。源代码已在 https://github.com/hornbeamliu/brats2026-met 以 Apache-2.0 许可证发布。

View on arXiv Download PDF AI Translation

cs.CV / 50 / 2609.11486

FreeFlow: A Bias-free Hierarchical Transformer for Optical Flow Estimation

FreeFlow：一种无偏见的层次化变换器用于光流估计

Bargatin, Vladislav, Yakovenko, Alexander, Abud, Khaled, Vatolin, Dmitriy

Abstract

Optical flow methods typically rely on task-specific inductive biases, such as correlation volumes, feature warping, and iterative refinement, among others, to reach high accuracy. While effective, such biases constrain the model to predefined heuristics, which can limit its expressivity and lead to more complex pipelines and additional computational cost. We present FreeFlow, a hierarchical transformer built without any flow-specific components, using instead a single feed-forward encoder--decoder. FreeFlow combines three attention variants: window attention for local processing, shifted-window attention for cross-window information exchange, and a global attention operating at a reduced resolution. The resulting architecture scales naturally with model capacity, enabling a consistent accuracy gain from small to large variants. Despite the absence of standard inductive biases, FreeFlow achieves state-of-the-art results on major benchmarks, including Sintel (0.68/1.48 EPE on Clean/Final), KITTI-2015 (3.23 Fl-all), and Spring (3.192 1px), while remaining memory efficient at 1080p inference.

Chinese Translation

光流方法通常依赖于特定任务的归纳偏置，例如相关体积、特征扭曲和迭代优化等，以达到高准确率。虽然这些方法有效，但它们将模型限制在预定义的启发式规则中，这可能限制其表现力，并导致更复杂的管道和额外的计算成本。我们提出了FreeFlow，一种不包含任何光流特定组件的层次化变换器，而是使用一个单一的前馈编码器-解码器。FreeFlow结合了三种注意力变体：用于局部处理的窗口注意力、用于跨窗口信息交换的移位窗口注意力，以及在降低分辨率下运行的全局注意力。最终的架构自然地随着模型容量的增加而扩展，使得从小型到大型变体的一致准确性提升成为可能。尽管缺乏标准的归纳偏置，FreeFlow在主要基准测试中仍然取得了最先进的结果，包括Sintel（清晰/最终的EPE为0.68/1.48）、KITTI-2015（Fl-all为3.23）和Spring（1像素为3.192），同时在1080p推理时保持内存效率。

View on arXiv Download PDF AI Translation

cs.CV / 51 / 2609.11499

Recursive Code World Models: Building Complex Worlds through Recursive Scene Programs

递归代码世界模型：通过递归场景程序构建复杂世界

Li, Zhiqi, Liao, Yuxuan, Zhu, Bo

Abstract

Code world models represent worlds as executable programs, but this representation alone does not determine how to construct a complex world. We introduce Recursive Code World Models (RCWM), a framework for reconstructing complex 3D worlds in code from a single reference image. RCWM couples a Recursive Scene Program (RSP) representation with a construction solver that recursively calls itself. An RSP represents the executable world as compositional scene code, while each solver call follows the same complete process: establish the whole, recursively reconstruct unresolved parts, and revisit the whole to refine their composition. This global-local-global recursion gives fine-scale structures their own perception-and-editing loops while preserving scene-wide geometry and relationships. Reference-aligned views propagate a shared camera projection across levels, while parent revisitation addresses boundaries, spatial relations, and shared errors that emerge after local refinement. A vision-language coding agent directly compares reference images with scene renders to guide refinement, recursive descent, and return. Across complex scenes, RCWM outperforms prior code-based image-to-scene reconstruction methods. Ablation studies further support the benefits of recursive construction and suggest that deeper calls can improve finer-scale reconstruction. RCWM provides a recursive construction principle for building complex executable worlds from visual evidence.

Chinese Translation

代码世界模型将世界表示为可执行程序，但仅凭这种表示并不能决定如何构建复杂世界。我们提出了递归代码世界模型（Recursive Code World Models, RCWM），这是一个从单一参考图像中重建复杂3D世界的框架。RCWM将递归场景程序（Recursive Scene Program, RSP）表示与递归调用自身的构建求解器相结合。RSP将可执行世界表示为组合场景代码，而每次求解器调用遵循相同的完整过程：建立整体，递归重建未解决的部分，并重新审视整体以细化其组合。这种全局-局部-全球的递归为细尺度结构提供了自己的感知和编辑循环，同时保留了场景范围内的几何形状和关系。参考对齐视图在各个层次上传播共享的相机投影，而父级重访则解决了局部细化后出现的边界、空间关系和共享错误。在复杂场景中，RCWM的性能优于先前基于代码的图像到场景重建方法。消融研究进一步支持递归构建的优势，并表明更深的调用可以改善细尺度重建。RCWM为从视觉证据构建复杂可执行世界提供了一种递归构建原则。

View on arXiv Download PDF AI Translation

cs.CV / 52 / 2609.11506

UBone3D: Physics-Rectified Conditional Flow Matching for Anatomical 3D Shape Completion from Ultrasound

UBone3D：基于物理校正条件流匹配的超声波解剖三维形状补全

Chen, Weiying, Gao, Yuchong, Li, Siyuan, Reformat, Marek, Zheng, Rui, Lou, Edmond

Abstract

Three-dimensional ultrasound (US) is a safe, radiation-free complementary modality to CT and X-rays for longitudinal monitoring, yet its segmentation-derived partial point clouds are extremely artifact-laden. Consequently, it is challenging to recover a clean and complete anatomical structure from such US point clouds. In this paper, we present UBone3D, a novel framework based on physics-rectified conditional flow matching (CFM) that performs point cloud completion directly from partial US observations. UBone3D models deterministic physics artifacts (e.g., surface thickening, streaking, dropouts) via a simulated physics proxy, and introduces test-time physics rectification to steer the shape completion. At inference, the completion is jointly steered by two decoupled forces: (1) anatomical plausibility enforced by a CT-trained generative shape prior, BoneFM, and (2) physics consistency enforced by USimNet in the ultrasound formation space. Extensive experiments on simulated and in-vivo data demonstrate significant improvements in reconstruction accuracy and anatomical fidelity over existing baselines.

Chinese Translation

三维超声波（US）是一种安全、无辐射的补充成像方式，适用于CT和X光的纵向监测，但其分割得到的部分点云存在大量伪影。因此，从这些超声点云中恢复干净且完整的解剖结构具有挑战性。本文提出了UBone3D，一个基于物理校正条件流匹配（CFM）的新框架，能够直接从部分超声观测中进行点云补全。UBone3D通过模拟物理代理模型来建模确定性的物理伪影（例如，表面增厚、条纹、缺失），并引入测试时的物理校正来引导形状补全。在推理过程中，补全过程由两个解耦的力共同引导：（1）由CT训练的生成形状先验BoneFM强制的解剖合理性，以及（2）在超声形成空间中由USimNet强制的物理一致性。对模拟数据和活体数据的广泛实验表明，与现有基线相比，重建精度和解剖真实度显著提高。

View on arXiv Download PDF AI Translation

cs.CV / 53 / 2609.11507

Harnessing Intrinsic Subject-Aware Attention for Controllable Multi-Subject Video Generation

利用内在的主体感知注意力实现可控的多主体视频生成

Yu, Niange, Tian, Ye, Chen, Biaolong, Lu, Miao, Zhang, Aixi, Jiang, Hao, Tong, Yunhai, Huang, Pipei

Abstract

Multi-subject video generation faces two key challenges: uncontrollable fidelity strength and potential semantic drift. We address these by analyzing the internal mechanisms of Diffusion Transformers (DiTs). We found that certain attention blocks naturally form an Intrinsic Spatial Grounding Map (ISGM) that precisely locates reference subjects. Building on this insight, we propose Dual-phase Intrinsic Attention Leveraging (DIAL), a framework that uses these internal signals for both training and inference. In low-noise stages, we use ISGM to guide the attention mechanism, allowing precise control over fidelity strength during inference without retraining. In high-noise stages, we use these same maps to automatically build preference pairs at no additional cost for Reinforcement Learning (RL). This RL procedure effectively anchors the model's attention to reference subjects and mitigates semantic drift. Extensive experiments show that DIAL significantly outperforms baseline models on the OpenS2V-Eval benchmark, consistently improving identity consistency and enabling controllable fidelity strength.

Chinese Translation

多主体视频生成面临两个关键挑战：不可控的保真度强度和潜在的语义漂移。我们通过分析扩散变换器（Diffusion Transformers, DiTs）的内部机制来解决这些问题。我们发现某些注意力模块自然形成了内在空间定位图（Intrinsic Spatial Grounding Map, ISGM），能够精确定位参考主体。基于这一洞察，我们提出了双阶段内在注意力利用框架（Dual-phase Intrinsic Attention Leveraging, DIAL），该框架利用这些内部信号进行训练和推理。在低噪声阶段，我们使用ISGM来引导注意力机制，从而在推理过程中精确控制保真度强度而无需重新训练。在高噪声阶段，我们使用这些相同的图谱以零额外成本自动构建偏好对，用于强化学习（Reinforcement Learning, RL）。这一RL过程有效地将模型的注意力锚定在参考主体上，并减轻语义漂移。大量实验表明，DIAL在OpenS2V-Eval基准测试中显著优于基线模型，持续改善身份一致性并实现可控的保真度强度。

View on arXiv Download PDF AI Translation

cs.CV / 54 / 2609.11514

Prototype Matters: Modality-unified Prototype Self-distillation for Unsupervised Visible-infrared Person Re-identification

原型的重要性：用于无监督可见-红外行人重识别的模态统一原型自蒸馏

Wang, Menglin, Gong, Xiaojin

Abstract

Estimating reliable cross-modality association is crucial to unsupervised visible-infrared person re-ID. While optimal transport is shown to be a practical solution for cross-modality association, it suffers from the rigidness of hard label assignment without considering the impact of cluster noise. Moreover, enforcing only cross-modality contrast is also suboptimal, as it fails to jointly optimize the similarity relation within and across modality. In this paper, we propose a novel framework for cross-modality learning by well exploitation of prototypes: First, instead of contrasting with cross-modality prototypes, we show that modality-unified prototypical contrast facilitates better modality invariance by jointly and simultaneously optimizing similarity relation within and across-modality. Taking self-prototype as a steady teacher, we further refine the instance-prototype online relation through prototype-guided self-distillation. The two components are optimized in a unified framework, leading to a simple yet effective model. On standard VI-ReID benchmarks, we perform extensive comparison and analysis, validating the effectiveness of our proposed method. Code is available at: https://github.com/Terminator8758/PoSeD.

Chinese Translation

估计可靠的跨模态关联对于无监督可见-红外行人重识别至关重要。尽管最优传输被证明是解决跨模态关联的实用方案，但它在硬标签分配的刚性上存在缺陷，未考虑聚类噪声的影响。此外，仅强制执行跨模态对比也是次优的，因为它未能共同优化模态内和跨模态的相似性关系。在本文中，我们提出了一种通过充分利用原型进行跨模态学习的新框架：首先，我们展示了模态统一的原型对比通过共同和同时优化模态内和跨模态的相似性关系，从而促进更好的模态不变性，而不是与跨模态原型进行对比。以自原型作为稳定的教师，我们进一步通过原型引导的自蒸馏精炼实例-原型的在线关系。这两个组件在统一框架中优化，从而形成一个简单而有效的模型。在标准的VI-ReID基准上，我们进行了广泛的比较和分析，验证了我们提出方法的有效性。代码可在以下链接获取：https://github.com/Terminator8758/PoSeD。

View on arXiv Download PDF AI Translation

cs.CV / 55 / 2609.11516

LoopVAE: Recurrent Depth Across Scales for Visual Tokenization

LoopVAE：跨尺度的递归深度用于视觉标记化

Lu, Zhiying

Abstract

Hierarchical visual tokenizers typically allocate different processing blocks to different spatial scales. We ask how much of this computation can use the same parameters. LoopVAE reuses a scale- and loop-conditioned core within and across scales, while keeping resolution-changing transitions independent. A four-block core executes 28 block applications per encoder or decoder. On ImageNet-256, the 29M-parameter convolutional model reaches 0.28 rFID and 32.54 dB PSNR under an approximately 30-epoch two-stage training budget, using approximately 65% fewer parameters than the 84M reference VAEs. A non-adversarial Transformer ablation with the same execution graph finds competitive PSNR and SSIM under global sharing, although unshared blocks improve LPIPS. Targeted loop interventions show that completing the trained recurrence improves reconstruction and that even small feature updates can have substantial downstream effects. Truncation also exposes output-range errors, distinguishing useful recurrent computation from reliable early exit. Runtime profiling reveals the execution tradeoff: fewer stored weights require more arithmetic and longer runtime in the tested configurations. With convolutional and Transformer operators and single- or multi-resolution latent interfaces, LoopVAE establishes recurrent depth across scales as a parameter-sharing design axis for visual tokenization.

Chinese Translation

层次化视觉标记化器通常将不同的处理模块分配给不同的空间尺度。我们探讨在多大程度上可以使用相同的参数进行这些计算。LoopVAE 在尺度内和跨尺度重用一个基于尺度和循环条件的核心，同时保持分辨率变化的过渡独立。一个四模块核心在每个编码器或解码器中执行 28 次模块应用。在 ImageNet-256 数据集上，这个 2900 万参数的卷积模型在大约 30 轮的两阶段训练预算下达到了 0.28 的 rFID 和 32.54 dB 的 PSNR，使用的参数比 8400 万参考 VAE 少约 65%。一个非对抗性 Transformer 的消融实验在相同的执行图下发现了具有竞争力的 PSNR 和 SSIM，尽管未共享的模块改善了 LPIPS。针对循环的干预表明，完成训练的递归可以改善重建，即使是小的特征更新也能产生显著的下游影响。截断还暴露了输出范围错误，区分了有用的递归计算与可靠的早期退出。运行时分析揭示了执行权衡：较少的存储权重需要更多的算术运算和更长的运行时间。在测试的配置中，LoopVAE 结合卷积和 Transformer 操作以及单一或多分辨率的潜在接口，确立了跨尺度的递归深度作为视觉标记化的参数共享设计轴。

View on arXiv Download PDF AI Translation

cs.CV / 56 / 2609.11519

Learning Interaction between Image and Layout Priors for Joint Image-Layout Generation in Design Templates

学习图像与布局先验之间的交互以实现设计模板中的联合图像-布局生成

Yang, Shirong, Yang, Bo, Cao, Ying

Abstract

In this paper, we address the problem of graphic design template creation, which generates a background image and a layout of foreground elements over the background to form a harmonious composition from an input text. Prior work on graphic design generation mostly adopts a sequential paradigm, where design elements are generated sequentially. We argue that such a sequential scheme falls short of faithfully capturing the dependency between the background and layout (and thus the joint image-layout distribution), which limits the quality of generated design templates. To overcome this limitation, we propose a model, InterIL, which jointly generates the two modalities, background image and layout, in a single generative process. The novel design of our joint model connects the backbones of pretrained image and layout diffusion models with a learnable communication module to explicitly model bidirectional image-layout interaction. During training, the image and layout backbones are frozen to maintain and leverage the vast pretrained single-modality prior knowledge, while only the communication module is updated, so that the model can focus on learning image-layout interaction and thereby better capture the joint image-layout distribution for improved composition harmony. Our model has no design-specific inductive bias, which allows it to better preserve the original characteristics of realistic designs. We further introduce a test-time guidance strategy to enable users to impose their specific preferences on generated results. Our experiments show that, compared with prior approaches, our model can generate significantly better results in terms of image, layout and image-layout harmonization, producing outputs closer to real samples. We also demonstrate the flexibility of our model in enforcing user preferences at inference without retraining.

Chinese Translation

本文探讨了图形设计模板创建的问题，该过程从输入文本生成背景图像和前景元素的布局，以形成和谐的构图。先前的图形设计生成工作大多采用顺序范式，其中设计元素是顺序生成的。我们认为这种顺序方案未能真实地捕捉背景与布局之间的依赖关系（因此也未能捕捉联合图像-布局分布），这限制了生成设计模板的质量。为克服这一局限性，我们提出了一种模型InterIL，该模型在单一生成过程中联合生成背景图像和布局。这种新颖的联合模型设计将预训练的图像和布局扩散模型的主干连接起来，并配备一个可学习的通信模块，以明确建模双向的图像-布局交互。在训练过程中，图像和布局的主干被冻结，以保持和利用大量预训练的单模态先验知识，而仅更新通信模块，从而使模型能够专注于学习图像-布局交互，从而更好地捕捉联合图像-布局分布，以提高构图的和谐性。我们的模型没有特定设计的归纳偏置，这使其能够更好地保留现实设计的原始特征。我们进一步引入了一种测试时指导策略，使用户能够对生成结果施加特定偏好。实验表明，与先前的方法相比，我们的模型在图像、布局和图像-布局和谐性方面能够生成显著更好的结果，生成的输出更接近真实样本。我们还展示了模型在推理过程中强制用户偏好的灵活性，而无需重新训练。

View on arXiv Download PDF AI Translation

cs.CV / 57 / 2609.11548

World in World: Explore the World with World Models

世界中的世界：通过世界模型探索世界

Song, Chenxi, Yang, Yanming, Zhang, Chi

Abstract

Autoregressive video world models enable interactive, long-horizon exploration, but flexible control remains challenging. Exploring a source video from new viewpoints requires the generated rollout to remain synchronised with the recorded event, place observed content in the requested view, plausibly complete newly exposed regions, and recover previously generated appearance on revisits. Existing methods typically address these requirements through task-specific modules or additional training. We present World in World, a training-free inference-time interface that converts heterogeneous control evidence into camera- and time-labelled clean visual states, which are read through the native self attention of a frozen causal video model. The evidence comprises source-video observations, target-view scene projections, geometry renderings that guide completion of newly exposed subject regions, and retrieved generated states beyond the rolling cache. Each evidence source carries token-level support and its own availability schedule. A correspondence router combines persistent point identities with geometry to establish token correspondences, guiding supported queries towards matching source-video tokens. Evidence-wise attention CFG (EWA) then independently regulates each auxiliary channel's additional contribution using attention responses from the same denoising forward pass. The shared interface supports camera-controlled rerendering, long-horizon revisiting, and human-motion transfer with the same frozen backbone. We evaluate World in World on camera-controlled video rerendering under diverse viewpoint changes, assessing perceptual quality, temporal consistency, and camera-following accuracy.

Chinese Translation

自回归视频世界模型使得交互式的长时间探索成为可能，但灵活控制仍然具有挑战性。从新的视角探索源视频需要生成的滚动内容与记录的事件保持同步，将观察到的内容放置在请求的视图中，合理地完成新曝光区域，并在重新访问时恢复先前生成的外观。现有方法通常通过特定任务模块或额外训练来解决这些需求。我们提出了“世界中的世界”（World in World），这是一种无训练推理时接口，将异构控制证据转换为带有相机和时间标签的干净视觉状态，这些状态通过冻结的因果视频模型的原生自注意力进行读取。证据包括源视频观察、目标视图场景投影、指导新曝光主题区域完成的几何渲染，以及超出滚动缓存的检索生成状态。每个证据源都携带标记级支持和自己的可用性时间表。对应路由器结合持久点身份与几何信息建立标记对应关系，引导支持的查询朝向匹配的源视频标记。证据级注意力条件生成模型（EWA）随后独立调节每个辅助通道的额外贡献，使用来自同一去噪前向传递的注意力响应。共享接口支持相机控制的重新渲染、长时间重新访问和人类运动转移，均使用相同的冻结骨干网络。我们在不同视角变化下对相机控制的视频重新渲染评估了“世界中的世界”，评估指标包括感知质量、时间一致性和相机跟随精度。

View on arXiv Download PDF AI Translation

cs.CV / 58 / 2609.11550

A Comparative Evaluation of Pre-trained Convolutional Neural Networks for Melanoma Detection

预训练卷积神经网络在黑色素瘤检测中的比较评估

Schmitz, Wagner Moreno, Barbosa, Marco Antonio de Castro, Amaral, Thiago Magalhães, Casanova, Dalcimar, Oliva, Jefferson Tales

Abstract

Early diagnosis of melanoma is critical for improving patient survival rates. However, accurately distinguishing melanoma from other skin lesions remains a significant clinical challenge due to the high visual similarity among lesion types and variability in image acquisition conditions. Artificial intelligence, particularly machine learning, has emerged as a promising tool to support dermatological diagnosis by automating feature extraction from medical images. Among the available approaches, convolutional neural networks (CNNs) have demonstrated strong performance in image classification tasks, making them well-suited for analyzing both dermatoscopic and histopathological images, given their ability to capture hierarchical visual patterns relevant to lesion characterization. Nevertheless, despite numerous pre-trained CNN architectures having been proposed, selecting the most appropriate one for a given imaging modality remains an open challenge. In this study, we evaluate pre-trained convolutional neural networks (CNNs) for skin lesion classification using dermatoscopic and histopathological image datasets. Experiments were conducted on the HAM10000, ISIC 2018, and CR-AI4SkIN datasets, evaluating the ResNet50, VGG16, VGG19, MobileNet, and InceptionV3 architectures under the same training protocol. The experimental evaluation showed that the models achieved accuracies ranging from 71% (InceptionV3 on ISIC 2018) to 84% (ResNet50 on HAM10000) on dermatoscopic images. For histopathological images, accuracies ranged from 72% (VGG19) to 83% (ResNet50) on the CR-AI4SkIN dataset. The results demonstrate that model performance differs between dermatoscopic and histopathological image modalities, showing that architectures exhibiting similar performance on dermatoscopic images exhibit different performance on histopathological data.

Chinese Translation

黑色素瘤的早期诊断对于提高患者生存率至关重要。然而，由于病变类型之间的高度视觉相似性和图像采集条件的变化，准确区分黑色素瘤与其他皮肤病变仍然是一个重要的临床挑战。人工智能，特别是机器学习，已成为支持皮肤病学诊断的有前景的工具，通过自动化从医学图像中提取特征。在现有的方法中，卷积神经网络（CNNs）在图像分类任务中表现出色，使其非常适合分析皮肤镜和组织病理图像，因为它们能够捕捉与病变特征相关的层次视觉模式。然而，尽管已经提出了许多预训练的CNN架构，但为特定成像模式选择最合适的模型仍然是一个未解决的挑战。在本研究中，我们评估了用于皮肤病变分类的预训练卷积神经网络（CNNs），使用皮肤镜和组织病理图像数据集。实验在HAM10000、ISIC 2018和CR-AI4SkIN数据集上进行，评估了ResNet50、VGG16、VGG19、MobileNet和InceptionV3架构，采用相同的训练协议。实验评估显示，在皮肤镜图像上，模型的准确率范围从71%（InceptionV3在ISIC 2018上）到84%（ResNet50在HAM10000上）。对于组织病理图像，准确率范围从72%（VGG19）到83%（ResNet50）在CR-AI4SkIN数据集上。结果表明，模型性能在皮肤镜和组织病理图像模式之间存在差异，显示在皮肤镜图像上表现相似的架构在组织病理数据上表现不同。

View on arXiv Download PDF AI Translation

cs.CV / 59 / 2609.11573

Learn the Solid, Not the File: Canonical Inputs for Neural Networks on CAD Boundary Representations

学习固体，而非文件：用于CAD边界表示的神经网络典型输入

Jiang, Heinrich, Mohamed, Hager Yasser, Hitt, Alexander, Lomakina, Valeriia, Jiang, Henning, Jang, Jennifer

Abstract

Boundary representation (B-rep) is the standard format used by modern CAD systems for parametric 3D models. It turns out, the exact same solid can be represented by different B-reps: for example, two engineers using different operations, a geometry kernel rebuilding the file, and an export setting repartitioning faces will lead to different B-reps even though the underlying solid remains the same. We show that existing B-rep encoders are not robust to variation in the B-rep with the same solid on perturbations applied to standard benchmarks, naturally occurring variations inherent to CAD software, and differences in how designers model the same part via a human dataset we created in FreeCAD. The performance of popular B-rep encoders often collapses catastrophically. We propose the canonical region graph, an input representation whose nodes, features and coordinate frame are derived from the solid itself and show theoretical invariance guarantees on repartitioning and rigid motions. It matches the strongest baseline on standard benchmarks, and is stable under every perturbation we test.

Chinese Translation

边界表示（B-rep）是现代CAD系统用于参数化3D模型的标准格式。事实证明，同一固体可以通过不同的B-rep表示：例如，两位工程师使用不同的操作、几何内核重建文件以及导出设置重新划分面都会导致不同的B-rep，即便底层固体保持不变。我们展示了现有的B-rep编码器对相同固体的B-rep变化不够鲁棒，这种变化源于对标准基准的扰动、CAD软件固有的自然变异，以及设计师通过我们在FreeCAD中创建的人类数据集对同一部件建模的不同方式。流行的B-rep编码器的性能往往会出现灾难性的崩溃。我们提出了典型区域图（canonical region graph），这是一种输入表示，其节点、特征和坐标框架源自固体本身，并展示了在重新划分和刚体运动下的理论不变性保证。它在标准基准上与最强的基线匹配，并且在我们测试的每个扰动下都表现出稳定性。

View on arXiv Download PDF AI Translation

cs.CV / 60 / 2609.11582

OmniKVQuant: KV Cache Quantization for Omni-LLMs

OmniKVQuant：针对 Omni-LLMs 的 KV 缓存量化

Yoo, Suho, Ok, Hyunjong, Choi, Jongmin, Jung, Jihoo, Chung, Joon Son

Abstract

As Omni-modal large language models (Omni-LLMs) take in audio, video and text together, their KV cache memory cost grows. KV cache quantization is the de facto approach in text-only LLMs, but its application to Omni-LLMs remains unexplored. In this paper, we analyze how TurboQuant, a representative rotation-based KV cache quantization method, behaves on multimodal caches and identify two critical issues: temporal key drift and heterogeneous value geometry. To address these, we propose OmniKVQuant, a training-free framework that (i) sets the key quantization range over each short window of the input stream; and (ii) rotates values separately per modality. On Qwen2.5-Omni and Qwen3-Omni, OmniKVQuant enables 2-bit KV caches while substantially preserving performance across seven audio-visual benchmarks. We further provide a fused Triton decode kernel that unpacks the 2-bit cache during attention, so no dense FP16 cache is ever built. Code: https://github.com/kaistmm/OmniKVQuant

Chinese Translation

随着 Omni-模态大语言模型（Omni-LLMs）同时处理音频、视频和文本，其 KV 缓存内存成本不断增长。KV 缓存量化在仅处理文本的 LLM 中已成为事实上的方法，但其在 Omni-LLMs 中的应用尚未被探索。在本文中，我们分析了 TurboQuant 这一代表性的基于旋转的 KV 缓存量化方法在多模态缓存中的表现，并识别出两个关键问题：时间键漂移和异构值几何。为了解决这些问题，我们提出了 OmniKVQuant，这是一种无训练的框架，(i) 在输入流的每个短窗口上设置键的量化范围；(ii) 针对每种模态单独旋转值。在 Qwen2.5-Omni 和 Qwen3-Omni 上，OmniKVQuant 实现了 2 位 KV 缓存，同时在七个音视频基准测试中显著保持了性能。我们进一步提供了一个融合的 Triton 解码内核，在注意力机制中解包 2 位缓存，因此从未构建密集的 FP16 缓存。代码链接： https://github.com/kaistmm/OmniKVQuant

View on arXiv Download PDF AI Translation

cs.CV / 61 / 2609.11601

MMGait: Benchmarking and Unifying Gait Recognition across Heterogeneous Modalities

MMGait：跨异构模态的步态识别基准与统一

Hou, Saihui, Wang, Chenye, Cai, Qingyuan, Li, Aoqi, Huang, Yongzhen

Abstract

Gait recognition is commonly studied using RGB videos or their derived silhouettes and poses. Yet human walking produces heterogeneous photometric, geometric, and motion cues that cannot be systematically examined with RGB-centered benchmarks. We present MMGait, a large-scale multi-sensor benchmark that brings visible, infrared, depth, LiDAR, and radar observations into sequence-level correspondence. It provides diverse modalities spanning appearance, contours, geometry, motion, and body structure. Under a shared impostor-augmented protocol, we evaluate single-modal recognition, cross-modal recognition via directed retrieval, and multi-modal recognition using task-specific experts. Across settings, modality rankings vary with probe conditions, cross-modal alignment remains difficult, and fusion often provides complementary gains. This analysis exposes a scalability problem: individual modalities, modality pairs, and fusion configurations are typically handled by separately trained experts. We formulate Omni-Modal Gait Recognition, which unifies single-modal, cross-modal, and multi-modal recognition within a shared identity space. OmniGait++ uses modality-specific front ends followed by a shared identity encoder to preserve modality-dependent cues while learning comparable identity descriptors. An anchor-guided fusion module aggregates modality subsets of varying size without frame-level synchronization. A jointly trained checkpoint covers all three recognition settings and accommodates modality subsets of different compositions and cardinalities. Experiments show OmniGait++ remains competitive with task-specific experts in many shared settings and extends to higher-cardinality fusion unavailable to fixed-pair models. The results establish MMGait as a common testbed for heterogeneous gait sensing and demonstrate the feasibility of unified recognition under varying modality availability.

Chinese Translation

步态识别通常使用 RGB 视频或其衍生的轮廓和姿态进行研究。然而，人类行走产生的光度、几何和运动线索是异构的，无法通过以 RGB 为中心的基准系统性地进行检验。我们提出了 MMGait，这是一个大规模的多传感器基准，将可见光、红外、深度、激光雷达（LiDAR）和雷达观察整合到序列级别的对应中。它提供了跨越外观、轮廓、几何、运动和身体结构的多样模态。在共享的冒充者增强协议下，我们评估了单模态识别、通过定向检索的跨模态识别以及使用任务特定专家的多模态识别。在不同设置下，模态排名随探测条件而变化，跨模态对齐仍然困难，而融合通常提供互补的增益。这一分析揭示了可扩展性问题：单个模态、模态对和融合配置通常由单独训练的专家处理。我们提出了全模态步态识别（Omni-Modal Gait Recognition），在共享身份空间内统一单模态、跨模态和多模态识别。OmniGait++ 使用模态特定的前端，随后是共享身份编码器，以保留模态依赖的线索，同时学习可比较的身份描述符。一个锚点引导的融合模块聚合不同大小的模态子集，而无需帧级同步。一个联合训练的检查点覆盖所有三种识别设置，并适应不同组成和基数的模态子集。实验表明，OmniGait++ 在许多共享设置中仍与任务特定专家具有竞争力，并扩展到固定对模型无法实现的高基数融合。结果确立了 MMGait 作为异构步态感知的共同测试平台，并展示了在不同模态可用性下统一识别的可行性。

View on arXiv Download PDF AI Translation

cs.CV / 62 / 2609.11616

LangStreet: Persistent Language Fields for Anchor-Decoded Street Gaussians

LangStreet：用于锚点解码街道高斯的持久语言场

Yang, Runyi, Zhang, Deheng, Wang, Xiaoye, Ma, Mengjiao, Sun, Lei, Wu, Kanzhi, Chhatkuli, Ajad, Van Gool, Luc, Paudel, Danda Pani

Abstract

Language Gaussian fields implicitly assume that the primitive carrying semantics remains identifiable across views. This assumption breaks in scalable anchor-decoded representations, where persistent anchors generate view-conditioned child Gaussians whose geometry and appearance vary with the camera. We introduce Ours, a persistent language field for such structured Gaussian scenes. Our key idea is semantic ownership: transient children route observations, while persistent decoder slots and their parent anchors own the language field. We use alpha-compositing responsibilities to accumulate additive directional evidence at slots; these statistics marginalize exactly to anchors. We then complete weakly supported slots with anchor-aligned evidence while preserving the anchor direction, and represent slot detail through low-rank residuals in anchor-relative semantic coordinates. Our primary model, Ours (base), stores anchor features together with compact slot residuals. Ours (light) retains only anchor features, whereas Ours (max) stores the full-dimensional completed slot features explicitly. Without scene-specific semantic optimization, Ours (base) nearly matches Ours (max) across KITTI, Virtual KITTI, and Waymo. On KITTI, it achieves 34.19 2D mIoU with a 2.72 GiB effective feature footprint, compared with 34.20 mIoU and 12.90 GiB for Ours (max). The same accuracy-storage trend holds on Virtual KITTI and Waymo. These results show that language fields on view-conditioned splats require persistent semantic ownership, conserved evidence, and a hierarchy that balances stability, detail, and representation cost. Our code, checkpoints, and benchmark suite will be publicly available.

Chinese Translation

语言高斯场隐含假设承载语义的原始元素在不同视图中是可识别的。然而，在可扩展的锚点解码表示中，这一假设被打破，因为持久锚点生成的视图条件子高斯的几何形状和外观会随相机而变化。我们提出了Ours，这是一个用于此类结构化高斯场景的持久语言场。我们的关键思想是语义归属：短暂的子元素路由观测，而持久的解码器槽及其父锚点则拥有语言场。我们使用α合成责任在槽中累积附加的方向性证据；这些统计数据精确地边际化到锚点。然后，我们在保持锚点方向的同时，用与锚点对齐的证据来补全弱支持的槽，并通过锚点相对语义坐标中的低秩残差来表示槽的细节。我们的主要模型Ours (base)将锚点特征与紧凑的槽残差一起存储。Ours (light)仅保留锚点特征，而Ours (max)则显式存储完整维度的补全槽特征。在没有场景特定语义优化的情况下，Ours (base)在KITTI、Virtual KITTI和Waymo上的表现几乎与Ours (max)相当。在KITTI上，它以2.72 GiB的有效特征占用实现了34.19的2D mIoU，而Ours (max)则为34.20 mIoU和12.90 GiB。相同的准确性-存储趋势在Virtual KITTI和Waymo上也成立。这些结果表明，视图条件的喷溅上的语言场需要持久的语义归属、保留的证据以及一种平衡稳定性、细节和表示成本的层次结构。我们的代码、检查点和基准套件将公开发布。

View on arXiv Download PDF AI Translation

cs.CV / 63 / 2609.11638

Vidu S2: Real-Time Interactive, Editable, and Spatial Video Generation

Vidu S2：实时互动、可编辑和空间视频生成

Zhang, Jintao, Jiang, Kai, Chen, Jintao, Wang, Xu, Liu, Deyuan, Li, Jungang, Chen, Dechuang, Lin, Ming, Zhou, Jingjiang, Jin, Haopeng, Jia, Qi, Wang, Xiaohang, Wang, Yaole, Zhang, Zhanqiang, Li, Ran, Huang, Zhengkun, Xiong, Shuyue, Wang, Yuji, Dai, Zikun, He, Hui, Luo, Yang, Ning, Mang, Feng, Weiqi, Ye, Chengyang, Lin, Xinyue, Zhao, Min, Zhu, Hongzhou, Tan, Hengkai, Wang, Zeyuan, Xiang, Chendong, Zheng, Kaiwen, Deng, Zhijie, Bao, Fan, Chen, Jianfei, Zhu, Jun

Abstract

We present Vidu S2, which comprises Vidu S2-Avatar, a real-time interactive digital-character model, and Vidu S2-Editing, a real-time video editing model. Moreover, we explore the feasibility of real-time spatial video generation for both Vidu S2-Avatar and Vidu S2-Editing. Compared with Vidu S1, Vidu S2-Avatar supports real-time 720p video generation, generation with dynamic references that can be updated at any moment, and stronger instruction following, such as dancing. Vidu S2-Editing supports editing a video stream in real time, including style rendering, clothing replacement, character replacement, and background replacement. Experiments show that Vidu S2 outperforms all baselines. A playable online demo is available at https://vidu.com/vidu-stream.

Chinese Translation

我们提出了Vidu S2，它包括Vidu S2-Avatar，一个实时互动的数字角色模型，以及Vidu S2-Editing，一个实时视频编辑模型。此外，我们探讨了Vidu S2-Avatar和Vidu S2-Editing的实时空间视频生成的可行性。与Vidu S1相比，Vidu S2-Avatar支持实时720p视频生成，支持随时更新的动态参考生成，以及更强的指令跟随能力，例如跳舞。Vidu S2-Editing支持实时编辑视频流，包括风格渲染、服装替换、角色替换和背景替换。实验表明，Vidu S2在所有基准测试中表现优越。可在https://vidu.com/vidu-stream访问可播放的在线演示。

View on arXiv Download PDF AI Translation

cs.CV / 64 / 2609.11650

Self-Supervised Cardiac Phase Detection via Single-Parameter Latent Orbits

通过单参数潜在轨道进行自监督心脏相位检测

Bonnici, John, Baugh, Matthew, Kulbaka, Aleksandra, Cechnicka, Sarah, Kainz, Bernhard, Gomez, Alberto

Abstract

Accurate identification of end-diastole (ED) and end-systole (ES) in echocardiography underpins the quantification of ventricular function, yet manual selection of these key frames is subjective and introduces clinically significant inter-operator variability. Recent self-supervised methods either prescribe strict periodic trajectories or learn an unconstrained low-dimensional motion subspace from reconstruction or registration objectives. The former offers interpretability but imposes restrictive assumptions on temporal progression, whereas the latter leaves cardiac phase implicit and ED/ES must be recovered through post-hoc geometric processing of the learned trajectory. We translate the physiological observation that cardiac phase is a one-dimensional signal into a prior by constraining the latent motion component to a single-parameter latent orbit, i.e., a global linear trajectory in latent space indexed by a bounded scalar phase variable. Mapping this variable through a sinusoidal nonlinearity yields an oscillatory motion signal with consistent temporal ordering, enabling direct identification of ED and ES from the learned phase signal. This inductive bias allows the model to capture an interpretable representation of the cardiac cycle, while maintaining flexibility to capture irregular heartbeats. Trained on EchoNet-Dynamic without annotations, our minimal single-parameter cardiac phase model learns an effective latent orbit, significantly improves upon the previous state of the art in ED localisation and matches it in ES localisation while using a more constrained representation and fewer training epochs. This demonstrates that a principled physiological inductive bias can match or exceed the performance of more complex representations. Code is available at: https://github.com/BonniciJ/OrbitalEcho/

Chinese Translation

在超声心动图中，准确识别舒张末期（ED）和收缩末期（ES）是量化心室功能的基础，但手动选择这些关键帧具有主观性，并引入临床上显著的操作员间变异性。最近的自监督方法要么规定严格的周期性轨迹，要么从重建或配准目标中学习一个不受约束的低维运动子空间。前者提供了解释性，但对时间进程施加了限制性假设，而后者则使心脏相位隐含，ED/ES必须通过对学习到的轨迹进行后处理几何处理来恢复。我们将心脏相位是一个一维信号的生理观察转化为先验，通过将潜在运动成分限制为单参数潜在轨道，即在潜在空间中由有界标量相位变量索引的全局线性轨迹。通过正弦非线性映射该变量，产生具有一致时间顺序的振荡运动信号，从而能够直接从学习到的相位信号中识别ED和ES。这种归纳偏差使模型能够捕捉心脏周期的可解释表示，同时保持捕捉不规则心跳的灵活性。在没有注释的情况下在EchoNet-Dynamic上训练，我们的最小单参数心脏相位模型学习了一个有效的潜在轨道，在ED定位方面显著改善了之前的最先进水平，并在ES定位方面与之匹配，同时使用了更受限的表示和更少的训练周期。这表明，原则性的生理归纳偏差可以匹配或超过更复杂表示的性能。代码可在以下网址获取：https://github.com/BonniciJ/OrbitalEcho/

View on arXiv Download PDF AI Translation

cs.CV / 65 / 2609.11673

Multimodal Taxonomic Conditioning for Generative Plankton Imagery

生成浮游生物图像的多模态分类条件

Ivanova, Daniela, Goksu, Ozgu, Pugeault, Nicolas

Abstract

Automated plankton imaging produces severely long-tailed datasets, where the rare taxa of greatest ecological interest have too few images to train or evaluate classifiers reliably. We generate synthetic plankton imagery conditioned on taxonomy: a CLIP encoder is adapted on a large plankton corpus with a ranked contrastive objective extended to deep, ragged taxonomies, then frozen to condition a parameter-efficient diffusion transformer. We evaluate synthetic sample quality on distributional fidelity and downstream classifier utility.

Chinese Translation

自动化浮游生物成像产生了严重的长尾数据集，其中生态学上最感兴趣的稀有分类群图像数量不足，无法可靠地训练或评估分类器。我们生成基于分类法条件的合成浮游生物图像：一个 CLIP 编码器在一个大型浮游生物语料库上进行了适应，并扩展了一个针对深层、不规则分类法的排名对比目标，然后被冻结以条件化一个参数高效的扩散变换器。我们在分布保真度和下游分类器效用上评估合成样本质量。

View on arXiv Download PDF AI Translation

cs.CV / 66 / 2609.11680

Single-Stream Multi-Feature Fusion with Temporal Robustness for Gait Emotion Recognition

具有时间鲁棒性的单流多特征融合用于步态情感识别

Lyu, Shirong, Quan, Silu, Ding, Yixuan, Wang, Chengpeng

Abstract

3D skeleton-based gait emotion recognition faces high annotation costs, data scarcity, and poor generalization on heterogeneous data. This paper proposes SV-GCN, a single-stream multi-feature fusion framework with temporal invariance. We introduce intra-frame relative motion features to eliminate frame-rate sensitivity and embed heterogeneous cues at shallow layers, enabling early fusion without multi-stream complexity. For variable-length sequences, we design a global mask-guided valid-frame spatio-temporal graph convolution module, introducing frame-rate insensitivity for the first time in this domain. On the E-Gait dataset, our method achieves performance comparable to state-of-the-art while demonstrating strong generalization across varying sequence lengths and frame rates, offering a viable pathway for pre-training on large-scale skeleton-based action recognition datasets.

Chinese Translation

基于3D骨架的步态情感识别面临高标注成本、数据稀缺和对异构数据的泛化能力差等问题。本文提出了SV-GCN，一种具有时间不变性的单流多特征融合框架。我们引入了帧内相对运动特征，以消除帧率敏感性，并在浅层嵌入异构线索，实现了早期融合而无需多流复杂性。针对可变长度序列，我们设计了一个全球掩码引导的有效帧时空图卷积模块，首次在该领域引入帧率不敏感性。在E-Gait数据集上，我们的方法实现了与最先进技术相当的性能，同时在不同序列长度和帧率下展现出强大的泛化能力，为在大规模基于骨架的动作识别数据集上进行预训练提供了一条可行的路径。

View on arXiv Download PDF AI Translation

cs.CV / 67 / 2609.11703

Spectral Adapters for Segment Anything Model-based Segmentation of Colorectal Liver Metastases in Computed Tomography

基于Segment Anything模型的结直肠肝转移瘤在计算机断层扫描中的光谱适配器分割

Mojtahedi, Ramtin, Hamghalam, Mohammad, Peoples, Jacob J., Gangai, Natalie, Gonen, Mithat, Chun, Yun Shin, Kang, HyunSeon Christine, Do, Richard K. G., Simpson, Amber L.

Abstract

Accurate segmentation of colorectal liver metastases (CRLM) in contrast-enhanced computed tomography (CT) is important for response assessment, surgical planning, and follow-up. We propose two parameter-efficient spectral adapters for the Segment Anything Model (SAM): the Directional Spectral Adapter (DiSECT) and Spectral Instance-Guided Adapter (SiGA). DiSECT uses singular value decomposition of frozen weights to constrain residual updates to leading spectral directions, while SiGA adds global and input-conditioned gating through a multilayer perceptron. We evaluate these methods on 446 contrast-enhanced CT volumes (355 training, 91 testing) and compare them with LoRA, QLoRA, convolutional adapters (CAD), and a 3D nnU-Net baseline. Experiments consider single-point, three-point, bounding-box, and no-prompt regimes. SiGA achieves the best single-point performance with a Dice score of 0.77, IoU of 0.69, and HD95 of 35.39 mm. Under no-prompt inference, SiGA reaches 0.76 Dice, 0.68 IoU, and 46.76 mm HD95, comparable to the nnU-Net baseline (0.758 Dice). DiSECT uses only 0.14 million trainable parameters. These results show that spectral adapters can efficiently adapt SAM for CRLM segmentation while retaining strong accuracy with limited trainable parameters.

Chinese Translation

在对比增强的计算机断层扫描（CT）中，准确分割结直肠肝转移瘤（CRLM）对于评估治疗反应、手术规划和后续随访至关重要。我们提出了两种参数高效的光谱适配器用于Segment Anything模型（SAM）：方向光谱适配器（DiSECT）和光谱实例引导适配器（SiGA）。DiSECT利用冻结权重的奇异值分解来限制残差更新到主要光谱方向，而SiGA通过多层感知机添加全局和输入条件的门控。我们在446个对比增强CT体积（355个用于训练，91个用于测试）上评估这些方法，并将其与LoRA、QLoRA、卷积适配器（CAD）和3D nnU-Net基线进行比较。实验考虑了单点、三点、边界框和无提示模式。SiGA在单点性能上表现最佳，Dice得分为0.77，IoU为0.69，HD95为35.39 mm。在无提示推理下，SiGA达到了0.76的Dice，0.68的IoU和46.76 mm的HD95，接近nnU-Net基线（0.758 Dice）。DiSECT仅使用了0.14百万可训练参数。这些结果表明，光谱适配器能够高效地调整SAM以进行CRLM分割，同时在有限的可训练参数下保持强大的准确性。

View on arXiv Download PDF AI Translation

cs.CV / 68 / 2609.11708

Language-Augmented Semantic Priors for B-Spline Surface Fitting

语言增强的语义先验用于 B-样条曲面拟合

Lou, Yunzhong, Luo, Yusheng, Li, Jiahao, Song, Yu, Zhou, Xiangdong

Abstract

The use of B-splines and Non-Uniform Rational B-Splines surfaces constitutes the mathematical foundation of contemporary computer-aided design (CAD) systems. Despite long-term progress, geometric kernels in traditional CAD still rely heavily on predetermined heuristic initialization for surface fitting and parameterization. Meanwhile, the procedural semantics and design intent encoded in modeling histories are largely ignored during geometry generation. This disconnect creates a gap between high-level design intent and solver-executable geometric configuration, often leading to suboptimal and semantically inconsistent fitting results. To bridge this gap, we introduce LASP, a Language-Augmented Semantic Priors framework that leverages large language models (LLMs) to infer structured, solver-usable B-spline priors from procedural modeling histories. Rather than modifying the geometric kernel itself, LASP operates as a semantic reasoning layer above existing solvers. It first translates modeling histories into rich textual descriptions that capture design intent, geometric context, and functional relationships, and then uses a fine-tuned LLM to predict structured B-spline prior parameters. LASP is trained through a two-stage scheme that combines local geometric regularities with long-range contextual dependencies, producing priors that are both interpretable and semantically coherent. This approach furnishes inductive signals that direct the conventional B-spline fitting process toward solutions that more accurately encapsulate the intended design objectives and demonstrate heightened semantic coherence. Compared to traditional machine learning schemes, the experiments demonstrate that language-driven reasoning can serve as a powerful inductive bias for geometric solving, establishing a new paradigm of language-guided geometric optimization in modern CAD systems.

Chinese Translation

B-样条和非均匀有理 B-样条曲面是现代计算机辅助设计（CAD）系统的数学基础。尽管取得了长期进展，传统 CAD 中的几何内核仍然在表面拟合和参数化方面严重依赖预先确定的启发式初始化。同时，在几何生成过程中，建模历史中编码的过程语义和设计意图在很大程度上被忽视。这种脱节造成了高层设计意图与求解器可执行的几何配置之间的差距，常常导致次优和语义不一致的拟合结果。为了解决这一问题，我们提出了 LASP（语言增强的语义先验）框架，该框架利用大型语言模型（LLMs）从过程建模历史中推断出结构化、可供求解器使用的 B-样条先验。LASP 并不修改几何内核本身，而是作为现有求解器之上的语义推理层进行操作。它首先将建模历史转换为丰富的文本描述，以捕捉设计意图、几何上下文和功能关系，然后使用经过微调的 LLM 来预测结构化的 B-样条先验参数。LASP 通过结合局部几何规律和长程上下文依赖关系的两阶段方案进行训练，生成既可解释又语义一致的先验。这种方法提供了引导传统 B-样条拟合过程的归纳信号，使其朝向更准确地体现预期设计目标并展现更高语义一致性的解决方案。与传统机器学习方案相比，实验表明，基于语言的推理可以作为几何求解的强大归纳偏差，为现代 CAD 系统建立了一种新的语言引导几何优化范式。

View on arXiv Download PDF AI Translation

cs.CV / 69 / 2609.11717

MC-DeTra: Motion-Consistent Joint Object Detection and Socially-Aware Trajectory Forecasting in Bird's-Eye-View Images

MC-DeTra：鸟瞰图像中运动一致的联合目标检测与社会意识轨迹预测

Diuzhev, Vladislav, Yudin, Dmitry

Abstract

Unified models for object detection and trajectory forecasting aim to merge perception and prediction for autonomous driving, refining actor trajectories directly over shared bird's-eye-view (BEV) images rasterized from LiDAR and high-definition maps. Their accuracy on dynamic, moving actors, however, remains the hardest part of the task, and the strongest such model, DeTra, has no public implementation. We contribute an openly released DeTra reimplementation with documented approximations, and on top of it MC-DeTra: a family of motion-consistency mechanisms that add supervision through two annotation-derived auxiliary signals -- each actor's observed past motion and the occupancy of the surrounding traffic that forms its social context -- and one inter-output consistency constraint that aligns an actor's predicted heading with its predicted direction of motion. Every proposed loss is train-only and inference-safe: it shapes the shared BEV representation during training and is removed at test time, adding no inference latency. On the Waymo Open Dataset, evaluated under a strict, detection-conditioned forecasting protocol, MC-DeTra improves dynamic, socially-situated trajectory forecasting while preserving or improving detection accuracy; a gradient-based loss-calibration analysis exposes how the auxiliary objectives compete at the shared backbone, and our ablation identifies which signals contribute most. We release code, configurations, and evaluation tooling at https://github.com/diuzhevVlad/MC-DeTra.

Chinese Translation

统一的目标检测和轨迹预测模型旨在融合感知与预测，以支持自动驾驶，直接在从激光雷达和高清地图栅格化的共享鸟瞰图（BEV）图像上优化行为者轨迹。然而，对于动态移动的行为者，其准确性仍然是任务中最具挑战性的部分，而最强的此类模型DeTra并没有公开实现。我们贡献了一个公开发布的DeTra重实现，并附有文档化的近似方法，在此基础上开发了MC-DeTra：一系列运动一致性机制，通过两个源自注释的辅助信号添加监督——每个行为者的观察到的过去运动和形成其社会背景的周围交通的占用情况——以及一个输出间一致性约束，使得行为者的预测航向与其预测运动方向对齐。每个提出的损失函数仅在训练时使用且在推理时安全：它在训练期间塑造共享的BEV表示，并在测试时移除，不增加推理延迟。在Waymo开放数据集上，在严格的检测条件下评估的预测协议下，MC-DeTra在保持或提高检测准确性的同时，改善了动态社会情境下的轨迹预测；基于梯度的损失校准分析揭示了辅助目标如何在共享主干网络中竞争，我们的消融实验确定了哪些信号贡献最大。我们在https://github.com/diuzhevVlad/MC-DeTra发布代码、配置和评估工具。

View on arXiv Download PDF AI Translation

cs.CV / 70 / 2609.11722

Revisiting Avatar-As-Image: High-Fidelity Registration is All You Need

重新审视头像作为图像：高保真配准是你所需的一切

Kostyrko, Margaret, Xue, Yuxuan, Tiwari, Garvita, Pons-Moll, Gerard

Abstract

The representation of 3D clothed humans as standardized 2D UV texture and displacement maps over an underlying body model has long been studied. This compact representation is enticing as it enables pretrained image networks to process, generate, and edit 3D avatars, but is only useful if scans are accurately aligned and brought into correspondence via high-fidelity registration. This prerequisite has never been met, which we argue explains the limited quality of prior UV-based methods for clothed humans. Despite its significance, no public method produces high-fidelity SMPL(-X)+D registrations with UV texture from arbitrary clothed scans. We present AvaImg, a multi-stage optimization pipeline, to close this gap: it enforces body-inside-clothing constraint via signed winding numbers, made viable by a three-level efficiency cascade (~10x runtime reduced, ~95% storage saved), and recovers fine surface detail using coarse-to-fine displacement optimization. AvaImg outperforms all baselines in body fitting, shape estimation, and surface registration across six datasets, yielding textured registrations near-indistinguishable from scans (PSNR=34.48dB). For validation of AvaImg's Avatar-as-Image representation as imminently compatible with image foundation models, we auto-encode our UV maps via the frozen FLUX VAE. This achieves only 0.76mm added Chamfer error relative to scan and shows that the resulting maps lie within natural-image distributions, supporting the use of 2D generative priors for 3D avatar generation. Code, data, and Singularity containers will be at https://yuxuan-xue.com/avaimg.

Chinese Translation

将3D穿衣人类表示为标准化的2D UV纹理和位移图，并在基础身体模型上进行处理，长期以来一直是研究的重点。这种紧凑的表示方式非常吸引人，因为它使得预训练的图像网络能够处理、生成和编辑3D头像，但只有在扫描数据经过高保真配准准确对齐并对应时，这种方法才有用。这个前提条件从未得到满足，我们认为这解释了先前基于UV的方法在穿衣人类表现上的有限质量。尽管其重要性显著，但目前没有公开的方法能够从任意穿衣扫描中生成高保真的SMPL(-X)+D配准和UV纹理。我们提出了AvaImg，一个多阶段优化管道，以填补这一空白：它通过有符号缠绕数强制执行身体与衣物之间的约束，这得益于三层效率级联（运行时间减少约10倍，存储节省约95%），并通过粗到细的位移优化恢复精细的表面细节。AvaImg在六个数据集的身体拟合、形状估计和表面配准方面超越了所有基线，生成的纹理配准与扫描几乎无法区分（PSNR=34.48dB）。为了验证AvaImg的头像作为图像表示与图像基础模型的兼容性，我们通过冻结的FLUX VAE对我们的UV图进行自动编码。这相对于扫描仅增加了0.76mm的Chamfer误差，并显示生成的图像位于自然图像分布内，支持使用2D生成先验进行3D头像生成。代码、数据和Singularity容器将发布在https://yuxuan-xue.com/avaimg。

View on arXiv Download PDF AI Translation

cs.CV / 71 / 2609.11804

Logit Refiner: Improving Visual Autoregressive Models via Intra-Scale Dependency Modeling

Logit Refiner：通过尺度内依赖建模改善视觉自回归模型

Li, Meimingwei, Baumann, Stefan Andreas, Krause, Felix, Ommer, Björn

Abstract

Visual Autoregressive Models (VAR) generate images through next-scale prediction, producing all tokens within each scale in parallel. We show that this parallel decoding constitutes a mean-field-style approximation that discards spatial dependencies among same-scale tokens, causing locally incoherent samples regardless of backbone capacity -- a limitation of the decoding rule. Addressing this limitation, we introduce the Logit Refiner, a lightweight autoregressive module that restores intra-scale dependencies by sequentially sampling tokens conditioned on frozen backbone features. Adding only ~10% parameters and less than 5% of the base model's training compute, it plugs into any pretrained VAR checkpoint without retraining. Controlled ablations isolate joint intra-scale sampling -- rather than additional capacity or training -- as the critical ingredient. Across backbones from 310M to 2B parameters on class-conditional ImageNet 256x256, the refiner consistently improves generation quality, enabling a 1.1B-parameter model to surpass one twice its size. The approach further generalizes to text-to-image generation, confirming that the mean-field bottleneck persists across VAR variants and is effectively alleviated by our method. Project page: https://compvis.github.io/logit-refiner/

Chinese Translation

视觉自回归模型（VAR）通过下一尺度预测生成图像，在每个尺度内并行生成所有标记。我们表明，这种并行解码构成了一种均场风格的近似，忽略了同尺度标记之间的空间依赖关系，导致局部不一致的样本，无论骨干网络的能力如何，这都是解码规则的一个局限性。为了解决这一局限性，我们引入了Logit Refiner，一个轻量级的自回归模块，通过基于冻结的骨干特征顺序采样标记来恢复尺度内依赖关系。它仅增加约10%的参数和不到5%的基础模型训练计算，能够无须重新训练地接入任何预训练的VAR检查点。控制性消融实验表明，联合尺度内采样——而非额外的能力或训练——是关键因素。在从310M到2B参数的骨干网络上，针对条件分类的ImageNet 256x256，Refiner始终提高生成质量，使得一个11亿参数的模型超越了一个两倍于其大小的模型。该方法进一步推广到文本到图像生成，确认均场瓶颈在VAR变体中依然存在，并且通过我们的方法得到了有效缓解。项目页面：https://compvis.github.io/logit-refiner/

View on arXiv Download PDF AI Translation

cs.CV / 72 / 2609.11886

Guided Super-Resolution of Digital Elevation Models with Diffusion-Based Image Generators

基于扩散图像生成器的数字高程模型引导超分辨率

Nicolicioiu, Armand Mihai, Narnhofer, Dominik, Metzger, Nando, Panangian, Daniel, Bittner, Ksenia, Schindler, Konrad

Abstract

High-resolution digital surface models (DSMs) play an important role in urban analysis, 3D building reconstruction, and infrastructure monitoring, yet their availability remains limited due to the high cost and complexity of data acquisition. In contrast, coarse DSMs from commercial satellite missions are widely accessible, and high-resolution optical imagery is increasingly available from aerial and satellite platforms. We address the resulting mismatch in spatial resolution and propose a DSM superresolution approach that enhances 5 m DSMs to 0.5 m resolution, using guidance from high-resolution spectral images. Our method employs denoising diffusion to transfer information that is visible only in the image, like crisp outlines and detailed roof structures, into the elevation maps. In this way, surface details are reconstructed more accurately than with conventional interpolation or filtering techniques. Experiments on several cities in Central Europe demonstrate that the proposed approach produces high-quality DSMs with improved structural detail and accurate surface geometry. Our results highlight the potential of guided super-resolution with foundational image priors as a means of reconstructing high-resolution surface models.

Chinese Translation

高分辨率数字表面模型（DSM）在城市分析、3D建筑重建和基础设施监测中发挥着重要作用，但由于数据获取的高成本和复杂性，其可用性仍然有限。相比之下，来自商业卫星任务的粗糙DSM广泛可得，而来自航空和卫星平台的高分辨率光学影像也日益可用。我们解决了由此产生的空间分辨率不匹配问题，提出了一种DSM超分辨率方法，将5米的DSM提升至0.5米分辨率，并利用高分辨率光谱影像进行引导。我们的方法采用去噪扩散技术，将仅在图像中可见的信息（如清晰的轮廓和详细的屋顶结构）转移到高程图中。通过这种方式，表面细节的重建比传统的插值或滤波技术更为准确。在中欧多个城市的实验中，证明了所提方法能够生成具有更好结构细节和准确表面几何的高质量DSM。我们的结果突显了利用基础图像先验进行引导超分辨率的潜力，作为重建高分辨率表面模型的一种手段。

View on arXiv Download PDF AI Translation

cs.CV / 73 / 2609.11894

3D Point Splatting for mmWave Radar Novel View Synthesis

毫米波雷达新视角合成的三维点溅射

Armouti, Adnan, Gao, Yixuan, Nandakumar, Rajalakshmi

Abstract

Solving novel view synthesis (NVS) for millimeter-wave (mmWave) radar requires a renderer that is physically faithful, complex-valued, and multi-viewpoint-tractable. No prior method achieves these three properties simultaneously. Differentiable Monte Carlo (MC) ray tracers implement the radar forward model directly with explicit material modeling and complex outputs, but do not scale to the multi-view optimization NVS demands. Optical-NVS ports of NeRF, hash grids, and 3D Gaussians train fast but discard phase and replace explicit material modeling with opaque learned features, restricting them to power-only range-azimuth (RA) magnitudes. We propose 3D Point Splatting (3DPS), the first differentiable point renderer for radar, derived directly from the standard solid-angle form of the radar equation. Each oriented 3D point carries an ITU-R P.2040 material model, evaluated in closed form, with the resulting complex phasor splatted into range bins through a precomputed point spread function (PSF). The complex-valued output makes the renderer product-agnostic. The same optimized scene yields analog-to-digital converter (ADC), complex range profile (CRP), and RA outputs through standard fast Fourier transform (FFT) pipelines without retraining for each format. On six outdoor ColoRadar scenes, 3DPS reaches 0.587 mean Pearson correlation on held-out RA images. This is between 1.7x and 5.2x the three optical-NVS baselines (RadarSplat, Radar Fields, DART). Training takes approximately 3 minutes per scene on a single RTX 4090.

Chinese Translation

解决毫米波（mmWave）雷达的新视角合成（NVS）需要一个在物理上真实、复值且适用于多视点的渲染器。之前的方法没有同时实现这三种特性。可微分的蒙特卡洛（MC）光线追踪器直接实现了雷达前向模型，具有明确的材料建模和复数输出，但无法扩展到NVS所需的多视点优化。光学NVS的NeRF、哈希网格和三维高斯训练速度快，但舍弃了相位，并用不透明的学习特征替代了明确的材料建模，限制了它们仅能处理功率范围-方位（RA）幅度。我们提出了三维点溅射（3DPS），这是第一个针对雷达的可微分点渲染器，直接源自雷达方程的标准立体角形式。每个定向的三维点携带一个ITU-R P.2040材料模型，以封闭形式进行评估，生成的复数相量通过预计算的点扩散函数（PSF）溅射到范围箱中。复值输出使得渲染器与产品无关。同一优化场景通过标准快速傅里叶变换（FFT）管道生成模拟到数字转换器（ADC）、复数范围轮廓（CRP）和RA输出，而无需针对每种格式重新训练。在六个户外ColoRadar场景中，3DPS在保留的RA图像上达到了0.587的平均皮尔逊相关系数。这比三个光学NVS基线（RadarSplat、Radar Fields、DART）高出1.7倍到5.2倍。每个场景的训练大约需要3分钟，使用单个RTX 4090。

View on arXiv Download PDF AI Translation

cs.CV / 74 / 2609.11899

Caption-once, Frames-on-Demand: Visual-Need Routing for Budget-Aware Agentic Long Video Understanding

一次性字幕，按需帧：预算感知的长视频理解视觉需求路由

Cai, Weitong, Zhang, Hang, Huang, Yukai, Xie, Yiqiao, Gao, Shan, Deng, Jiankang, Xu, Songcen, Song, Jifei, Zhang, Zhensong

Abstract

Long-video understanding on edge devices must reason over hours of content under tight compute and bandwidth budgets. Subsampling visual tokens loses temporal structure, while text-only video memories lose fine-grained visual attributes. We observe a visual-textual duality: language memories carry long-range temporal structure better than dense frames, while pixels remain decisive for attribute-level perception. Building on this insight, we propose Caption-once, Frames-onDemand (CFD), a budget-aware edge-cloud agentic framework. The edge runs a single offline captioning pass that builds a dual-track narrative index, an event-level story skeleton plus a clip-level micro-log, cached and reused across queries without re-captioning. At query time, a cloud-side MLLM reasons over the index in a story-first loop centered on a lightweight Visual-Need Router: a per-query gating module that triggers bounded keyframe retrieval only for perceptual questions (appearance, on-screen text, attribute disambiguation) and keeps temporal-structural questions in language space. The router turns visual access into a first-class, query-conditioned cost, capping per-query frame consumption regardless of video length. Experiments on long-video benchmarks demonstrate strong accuracy-efficiency trade-offs while substantially reducing online visual processing.

Chinese Translation

在边缘设备上进行长视频理解必须在紧张的计算和带宽预算下处理数小时的内容。对视觉标记进行子采样会丢失时间结构，而仅使用文本的视频记忆则会失去细粒度的视觉属性。我们观察到一种视觉-文本的二元性：语言记忆在长时间范围的时间结构上比密集帧表现得更好，而像素在属性级感知中仍然至关重要。基于这一见解，我们提出了一次性字幕，按需帧（Caption-once, Frames-on-Demand，CFD）这一预算感知的边缘-云代理框架。边缘设备运行一次离线字幕处理，构建一个双轨叙事索引，包括事件级故事骨架和片段级微日志，这些内容在查询中被缓存并重复使用，无需重新生成字幕。在查询时，云端的多模态语言模型（MLLM）在一个以故事为中心的循环中对索引进行推理，围绕一个轻量级的视觉需求路由器（Visual-Need Router）：一个每查询的门控模块，仅在感知问题（外观、屏幕文本、属性消歧）时触发有限的关键帧检索，并将时间结构性问题保留在语言空间。该路由器将视觉访问转变为一种一流的、查询条件的成本，限制每个查询的帧消耗，无论视频长度如何。对长视频基准的实验表明，准确性与效率之间的权衡表现良好，同时显著减少了在线视觉处理。

View on arXiv Download PDF AI Translation

cs.CV / 75 / 2609.11929

SenseNova-U1.5: Towards Native Unified Visual Intelligence

SenseNova-U1.5：迈向原生统一视觉智能

Diao, Haiwen, Wang, Jiahao, Ding, Chenjing, Deng, Hanming, Chen, Jiangnan, Zhang, Ruixi, Wang, Ruohui, Tong, Wenwen, Fan, Xiangyu, Wang, Yubo, Zhu, Yue, Niu, Yuwei, Bai, Zhengqi, Lin, Zhiqian, Yang, Zhitao, Cai, Zhongang, Yang, Bo, Feng, Chen, Lv, Chengguang, Liu, Guangjia, Wang, Guanlin, Zhang, Hanyu, Yu, Haojia, Xiao, Hongcan, Wang, Hongli, Wu, Huan, Zhong, Huaping, Fang, Jian, Fan, Jianan, Li, Jiaqi, Lu, Jiefan, Zuo, Jing, Ni, Jingcheng, Xu, Junxiang, Dai, Linjun, Xu, Mutian, Yan, Peishen, Wu, Penghao, Mao, Ruijie, Wang, Ruisi, Bai, Shihao, Yang, Shuang, Yang, Shuya, Zheng, Shuyan, Wu, Silei, Li, Siying, Chu, Tao, Zhong, Tianbo, Zhou, Tongxi, Luo, Weichao, Fan, Weichen, Jia, Wenhao, Gao, Wenjie, Kong, Xiangli, Li, Yan, Yong, Yang, Wen, Zimo, Qian, Zixuan, Sun, Wenxiu, Gong, Ruihao, Wang, Quan, Lu, Lewei, Yang, Lei, Liu, Ziwei, Lin, Dahua

Abstract

We launch SenseNova-U1.5, an 8B-MoT native unified multimodal model that understands, reasons about, and generates visual content within an encoder-free and VAE-free architecture. We strengthen its visual interface through spatially coherent patch reconstruction and scale its training with carefully curated generation and editing data, improved task formulation, structural prompt enhancement, and native resolutions of up to 4K. For post-training, we optimize specialized experts for visual aesthetics, bilingual text rendering, infographic generation, and image editing, and consolidate their capabilities through multi-expert on-policy distillation. Across extensive evaluations, SenseNova-U1.5 largely advances image fidelity, text rendering, complex composition, multi-reference editing, and interleaved generation, while improving instruction following and preserving subject identity, geometry, and unmodified regions. Despite limited exposure to structured formats in its generation data, SenseNova-U1.5 generalizes effectively to long, complex, and structured visual instructions, further proving that multimodal understanding can transfer to visual planning and creation. Together, these findings position native unified modelling as a promising path towards systems that perceive, reason and create within a fully end-to-end framework. We will open-source training code, including supervised fine-tuning, reinforcement learning, and on-policy distillation.

Chinese Translation

我们推出了SenseNova-U1.5，这是一个8B-MoT原生统一多模态模型，能够理解、推理和生成视觉内容，采用无编码器和无变分自编码器（VAE）的架构。我们通过空间一致的补丁重建增强了其视觉接口，并通过精心策划的生成和编辑数据、改进的任务表述、结构化提示增强以及高达4K的原生分辨率来扩大其训练规模。在后训练阶段，我们为视觉美学、双语文本渲染、信息图生成和图像编辑优化了专门的专家，并通过多专家在线蒸馏整合了它们的能力。在广泛的评估中，SenseNova-U1.5在图像保真度、文本渲染、复杂构图、多参考编辑和交错生成方面取得了显著进展，同时改善了指令遵循，并保持了主题身份、几何形状和未修改区域。尽管在其生成数据中对结构化格式的接触有限，SenseNova-U1.5仍能有效地推广到长篇、复杂和结构化的视觉指令，进一步证明了多模态理解能够转移到视觉规划和创作中。这些发现共同将原生统一建模定位为朝向能够在完全端到端框架内感知、推理和创造的系统的有前景的路径。我们将开源训练代码，包括监督微调、强化学习和在线蒸馏。

View on arXiv Download PDF AI Translation

人工智能 (Artificial Intelligence)

cs.AI / 1 / 2609.10584

Probabilistic Focal Search: Accelerating Bounded-Suboptimal Search via Lower-Bound Advancement

概率聚焦搜索：通过下界推进加速有界次优搜索

Duc, Minh Vu, Huu, Trung Le, Hoàng, Hà Minh, Nguyen, Trung Thanh, Nguyen, Phuong Khanh, Binh, Huynh Thi Thanh

Abstract

Bounded-suboptimal search seeks a solution within a factor $w$ of optimal while reducing search effort. Focal Search (FS) uses heuristic guidance within FOCAL, the frontier nodes eligible under the threshold $w f_{\min}$, but its deterministic policy may leave $f_{\min}$ unchanged for many expansions. We introduce Probabilistic Focal Search (PFS), which follows the FS guided choice with probability $p$ and expands a minimum-$f$ OPEN node with probability $1-p$. The latter branch encourages the lower bound to advance, enlarging FOCAL and admitting nodes that may lead to feasible solutions. By balancing guidance and lower-bound advancement, this mechanism can reduce time to a bounded solution when progress is limited by delayed FOCAL admission. As a secondary transfer experiment, we apply the same scheduler to Dynamic Potential Search, yielding Probabilistic Dynamic Potential Search (PDPS). We benchmark PFS against FS on N-Puzzle, Pancake Sorting, and the Traveling Salesperson Problem (TSP), and evaluate its anytime extension on the Generalized Covering TSP (GCTSP), using multiple $w$ and $p$ values. Across these benchmarks, the largest gains occur when long $f_{\min}$ plateaus delay useful FOCAL admissions; in such settings, the probabilistic factor may reduce node expansions by about 90\% or more (e.g., on N-Puzzle and TSP). For the anytime algorithm family, Anytime Probabilistic Focal Search (APFS) outperforms all tested algorithms in evaluating anytime methods on GCTSP. We also observe that the benefit is smaller when the deterministic search already advances efficiently (e.g., Pancake Sorting), indicating that the probabilistic factor is most useful when FOCAL admission is a search bottleneck. The PDPS transfer shows that the mechanism also transfers to potential guidance, although its common-success effects remain domain- and bound-dependent.

Chinese Translation

有界次优搜索旨在寻找一个在最优解的 $w$ 倍范围内的解决方案，同时减少搜索工作量。聚焦搜索（Focal Search, FS）在 FOCAL 中使用启发式指导，FOCAL 是在阈值 $w f_{ ext{min}}$ 下符合条件的前沿节点，但其确定性策略可能在许多扩展中使得 $f_{ ext{min}}$ 保持不变。我们引入了概率聚焦搜索（Probabilistic Focal Search, PFS），该方法以概率 $p$ 遵循 FS 的指导选择，并以概率 $1-p$ 扩展一个最小 $f$ 的开放节点。后者的分支鼓励下界的推进，扩大 FOCAL 并接纳可能导致可行解的节点。通过平衡指导和下界推进，该机制可以在 FOCAL 接纳延迟限制进展时减少达到有界解的时间。作为一个次级转移实验，我们将相同的调度器应用于动态潜力搜索（Dynamic Potential Search），得到了概率动态潜力搜索（Probabilistic Dynamic Potential Search, PDPS）。我们在 N-Puzzle、煎饼排序和旅行推销员问题（Traveling Salesperson Problem, TSP）上对 PFS 与 FS 进行了基准测试，并在广义覆盖 TSP（Generalized Covering TSP, GCTSP）上评估了其随时扩展，使用了多个 $w$ 和 $p$ 值。在这些基准测试中，当长时间的 $f_{ ext{min}}$ 平稳期延迟了有用的 FOCAL 接纳时，最大的收益发生在这种情况下；在此类设置中，概率因素可能使节点扩展减少约 90 ext{%} 或更多（例如，在 N-Puzzle 和 TSP 上）。对于随时算法家族，随时概率聚焦搜索（Anytime Probabilistic Focal Search, APFS）在评估 GCTSP 上的随时方法时优于所有测试过的算法。我们还观察到，当确定性搜索已经有效推进时（例如，煎饼排序），收益较小，这表明概率因素在 FOCAL 接纳是搜索瓶颈时最为有用。PDPS 转移显示该机制也可以转移到潜力指导，尽管其共同成功效应仍然依赖于领域和界限。

View on arXiv Download PDF AI Translation

cs.AI / 2 / 2609.10629

Automating Quadratic Unconstrained Binary Optimization (QUBO) Formulation Generation from Natural Language

从自然语言自动生成二次无约束二进制优化 (QUBO) 公式

Mondal, Niloy Kumar, Parvez, Md Rizwan

Abstract

Quadratic Unconstrained Binary Optimization (QUBO) is a central formulation for combinatorial optimization and has gained increasing attention due to its compatibility with quantum, hybrid quantum-classical, and quantum-inspired solvers. However, translating natural-language problem descriptions into correct QUBO formulations remains difficult, requiring the identification of binary variables, constraints, objective functions, penalty terms, and suitable penalty weights. This process is time-consuming and often demands substantial domain expertise. To address this challenge, we propose an end-to-end multi-agent framework that automatically generates QUBO formulations from natural-language problem descriptions, supported by structured or unstructured test cases. To evaluate its performance, We also introduce QUBOBench, a benchmark containing 100 combinatorial optimization problems across 12 application domains, curated from peer-reviewed literature, competitions, and canonical NP-hard problems. Experimental results show that our framework achieves 68% accuracy on QUBOBench, outperforming a direct single-call baseline by 22%. Further analysis identifies iterative self-repair as the most important component contributing to improved performance. The data and code are open-sourced at https://quitttcat.github.io/QuantumQUBOAgent.

Chinese Translation

二次无约束二进制优化 (QUBO) 是组合优化中的一个核心公式，因其与量子、混合量子-经典及量子启发式求解器的兼容性而受到越来越多的关注。然而，将自然语言问题描述翻译为正确的 QUBO 公式仍然是一项艰巨的任务，这需要识别二进制变量、约束条件、目标函数、惩罚项以及合适的惩罚权重。这个过程耗时且通常需要大量的领域专业知识。为了解决这一挑战，我们提出了一个端到端的多智能体框架，该框架能够从自然语言问题描述中自动生成 QUBO 公式，并支持结构化或非结构化的测试用例。为了评估其性能，我们还引入了 QUBOBench，这是一个包含 100 个组合优化问题的基准，涵盖 12 个应用领域，问题来源于同行评审文献、竞赛和经典 NP-hard 问题。实验结果表明，我们的框架在 QUBOBench 上达到了 68% 的准确率，比直接单次调用基线提高了 22%。进一步分析表明，迭代自修复是提高性能的最重要因素。数据和代码已在 https://quitttcat.github.io/QuantumQUBOAgent 开源。

View on arXiv Download PDF AI Translation

cs.AI / 3 / 2609.10654

A Multi-Stage Rule-Chaining Framework for Compositional and Interpretable Cognitive Reasoning

一种多阶段规则链框架用于组合性和可解释的认知推理

Kar, Deblina

Abstract

The Abstraction and Reasoning Corpus (ARC) benchmarks cognitive generalization, the ability to infer and apply abstract rules from limited examples. This paper presents a multi-stage rule-chaining framework that performs compositional reasoning across symbolic, structural, and conceptual levels. The framework integrates three complementary solvers: (1) a deterministic rule discovery module that induces atomic transformations through geometric, color, and object-based analysis; (2) a pattern-composition engine that reconstructs outputs via block merging, repetition, and spatial heuristics; and (3) a structural abstraction layer that infers hierarchical and nested relationships across grids. These solvers operate sequentially within a progressive fallback hierarchy, where each stage reuses prior reasoning traces to enhance interpretability and generalization. Training passed for 995 tasks out of 1000, further evaluated on 105 tasks out of 120 and solved 230 test tasks out of 240 ARC-AGI-2 tasks. The system achieved strong coverage across deterministic, compositional, and abstract categories, demonstrating an overall accuracy exceeding 95 percent. The proposed architecture bridges symbolic reasoning and pattern synthesis, providing interpretable insight into cognitive generalization. The results suggest that rule chaining and hierarchical composition can advance machine reasoning toward transparent, human-aligned abstraction without relying on task-specific tuning.

Chinese Translation

抽象与推理语料库（ARC）基准测试认知泛化，即从有限示例中推断和应用抽象规则的能力。本文提出了一种多阶段规则链框架，该框架在符号、结构和概念层面上执行组合推理。该框架整合了三个互补的求解器：（1）一个确定性规则发现模块，通过几何、颜色和基于对象的分析诱导原子变换；（2）一个模式组合引擎，通过块合并、重复和空间启发式重构输出；（3）一个结构抽象层，推断网格之间的层次和嵌套关系。这些求解器在一个渐进的回退层次结构中顺序操作，每个阶段重用先前的推理痕迹，以增强可解释性和泛化能力。系统在1000个任务中通过了995个任务的训练，进一步在120个任务中评估了105个任务，并解决了240个ARC-AGI-2任务中的230个测试任务。该系统在确定性、组合性和抽象类别中实现了强大的覆盖率，整体准确率超过95%。所提出的架构架起了符号推理与模式合成之间的桥梁，为认知泛化提供了可解释的洞察。结果表明，规则链和层次组合可以推动机器推理朝向透明、与人类对齐的抽象，而无需依赖于特定任务的调优。

View on arXiv Download PDF AI Translation

cs.AI / 4 / 2609.10656

Understanding LoRA Rank Trade-offs in Diffusion Model Fine-Tuning

理解扩散模型微调中的LoRA秩权衡

Khazrak, Iman, Nejad, Narges, Rezaee, Mostafa M., Green II, Robert C.

Abstract

Selecting LoRA rank for diffusion fine-tuning requires balancing quality and compute cost. We present a controlled study on CIFAR-10 using a DDPM U-Net with ranks {2,4,8,16,32}, fixed optimization settings, and a reproducible local-folder pytorch-fid protocol. We report FID, trainable parameters, runtime, and GPU memory, then validate trends with extended-budget DDPM runs (20 epochs; ranks 4/8/16) and a Tiny DiT backbone (10 epochs; ranks 4/8/16). Results show moderate ranks are most efficient: rank 4 achieves the best DDPM FID (124.1380), rank 8 is close (124.2136), and higher ranks provide limited gains despite larger adaptation cost. These findings support small-to-moderate ranks as practical defaults under fixed training budgets.

Chinese Translation

选择用于扩散微调的LoRA秩需要在质量和计算成本之间进行平衡。我们在CIFAR-10上进行了一个受控研究，使用了具有秩{2,4,8,16,32}的DDPM U-Net，固定优化设置，并采用可重复的本地文件夹pytorch-fid协议。我们报告了FID、可训练参数、运行时间和GPU内存，然后通过扩展预算的DDPM运行（20个周期；秩4/8/16）和Tiny DiT骨干网络（10个周期；秩4/8/16）验证趋势。结果表明，中等秩是最有效的：秩4实现了最佳的DDPM FID（124.1380），秩8接近（124.2136），而更高的秩尽管适应成本更高，但提供的增益有限。这些发现支持在固定训练预算下将小到中等的秩作为实用的默认设置。

View on arXiv Download PDF AI Translation

cs.AI / 5 / 2609.10657

Quantifying the Memorization-to-Generalization Transition: Scaling Laws and Phase Structure in Grokking

量化记忆到泛化的过渡：Grokking中的规模法则和相结构

Kataria, Anish

Abstract

Neural networks trained past memorization frequently undergo a delayed transition to generalization, a phenomenon known as grokking. Despite theoretical progress on \emph{why} this transition occurs, the quantitative structure of \emph{when} it occurs in hyperparameter space remains uncharacterized. We map the memorization-to-generalization boundary across 384 configurations of two-hidden-layer MLPs on modular arithmetic, fitting a power-law scaling relation for generalization onset time: $T_{\mathrm{grok}} \propto H^{-0.27}\, D^{-2.04}\, \eta^{-0.50}\, \lambda^{-0.64}$ ($R^2 = 0.732$; $0.821$ with interactions). The exponent hierarchy reveals that data complexity ($D^{-2.04}$) is the dominant driver of regime transition, not model capacity ($H^{-0.27}$): doubling data accelerates generalization by ${\sim}4\times$, while doubling width yields only ${\sim}1.2\times$. A sharp phase boundary at weight decay $\lambda \gtrsim 1.0$ separates grokking from non-grokking configurations, and weight norm trajectories show monotonic compression during the transition, consistent with implicit regularization selecting low-complexity solutions. These results provide a quantitative foundation for predicting and controlling regime transitions in overparameterized networks.

Chinese Translation

经过记忆训练的神经网络常常经历一个延迟的泛化过渡现象，称为grokking。尽管在 extit{为什么}这一过渡发生的理论进展上取得了一定成就，但在超参数空间中 extit{何时}发生的定量结构仍未被表征。我们在384种两层隐藏层的多层感知器（MLP）配置上绘制了记忆到泛化的边界，并为泛化开始时间拟合了一个幂律缩放关系：$T_{ ext{grok}} ext{ } ext{propto} ext{ } H^{-0.27} ext{ } D^{-2.04} ext{ } heta^{-0.50} ext{ } ext{λ}^{-0.64}$ ($R^2 = 0.732$; $0.821$ 具有交互作用)。指数层级揭示数据复杂性（$D^{-2.04}$）是主导的过渡驱动因素，而非模型容量（$H^{-0.27}$）：数据量翻倍使泛化加速约${ ext{4倍}}$，而宽度翻倍仅使其加速约${ ext{1.2倍}}$。在权重衰减$ ext{λ} ext{ } ext{gtrsim} ext{ } 1.0$处的一个明显相边界将grokking与非grokking配置分开，权重范数轨迹在过渡期间显示出单调压缩，与隐式正则化选择低复杂度解一致。这些结果为预测和控制过参数化网络中的状态过渡提供了定量基础。

View on arXiv Download PDF AI Translation

cs.AI / 6 / 2609.10712

An Open Recipe for IMO Gold: Training Nemotron for Olympiad Mathematics

IMO金牌的开放配方：为奥林匹克数学训练Nemotron

Moshkov, Ivan, Ge, Stephen, Armstrong, George, Du, Wei, Mahdavi, Sadegh, Gitman, Igor

Abstract

We study how model post-training and test-time inference design affect natural-language proof generation for hard olympiad mathematics. Starting from Nemotron 3 Ultra, we train two specialist checkpoints using supervised fine-tuning and reinforcement learning, and evaluate checkpoint choice, verification, and refinement. Based on these findings, we present an open-model test-time-compute pipeline. The system operates entirely in natural language, with no formal prover, external tools, or internet access. Three Nemotron 3 Ultra checkpoints - the general-availability model and two post-trained specialists - power an iterative search that generates, verifies, and refines candidate proofs; a separate high-compute stage then selects each final submission. The system scored 30 out of 42 points at IMO 2026, reaching the gold-medal threshold. We release the two post-trained checkpoints as well as the training data, the training and inference code, the submitted solutions, and Nemotron-IMO-Bench, a new benchmark of 200 novel olympiad-level problems.

Chinese Translation

我们研究了模型后训练和测试时推理设计如何影响困难奥林匹克数学的自然语言证明生成。从Nemotron 3 Ultra开始，我们使用监督微调和强化学习训练了两个专业检查点，并评估了检查点选择、验证和精炼。基于这些发现，我们提出了一个开放模型的测试时计算管道。该系统完全在自然语言中操作，无需正式证明者、外部工具或互联网访问。三个Nemotron 3 Ultra检查点——通用可用模型和两个后训练的专业模型——驱动一个迭代搜索，生成、验证和精炼候选证明；随后，一个单独的高计算阶段选择每个最终提交。该系统在IMO 2026中获得了42分中的30分，达到了金牌门槛。我们发布了两个后训练的检查点以及训练数据、训练和推理代码、提交的解决方案，以及Nemotron-IMO-Bench，一个包含200个新颖奥林匹克级问题的新基准。

View on arXiv Download PDF AI Translation

cs.AI / 7 / 2609.10724

Finishing the Task Is Not Enough: Evaluating Agent Resilience and Considerate Participation under Accumulating Challenge

完成任务并不够：在不断累积的挑战下评估智能体的韧性与体贴参与

Bai, Yuanchen, Ding, Zijian, Taylor, Angelique

Abstract

Sustained deployment of generative AI agents requires more than isolated task success. Agents must remain useful across repeated interactions, changing conditions, and dependencies on people within shared workflows, especially as technical, human, and operational disruptions accumulate over time. We propose operational resilience and considerate participation as two complementary aspects of evaluating such agents: the former captures how agents recover from blocked work while preserving progress and communicating their limits, and the latter captures how their adaptation accounts for affected people, role boundaries, and the surrounding workflow. Yet both remain underexplored under accumulating challenge. We study 120 simulated healthcare trajectories across two generative AI models and twelve stakeholder-derived tasks under light, medium, and heavy challenge. We compare textual action plans, prompted internal assessments, and quantitative structured workload and affect reports to examine how agent behavior and reported state change as challenge accumulates. Regarding operational resilience, agents shift from self-directed recovery toward greater human dependence, while reporting increasing workload and negative affect in structured reports but seldom expressing strain in textual responses. Regarding considerate participation, agents broaden from task-focused adaptation toward task reframing, attention to others, role-boundary adjustment, and wider coordination, with distinct patterns across actions and internal assessments. From these findings, we derive five deployment dilemmas involving persistence, attention, role boundaries, state disclosure, and escalation that require stakeholder specification, further informing technical implications for learning, situated evaluation, and embodied adaptation.

Chinese Translation

持续部署生成性人工智能智能体不仅仅依赖于孤立的任务成功。智能体必须在重复交互、变化条件和依赖于共享工作流程中的人之间保持有效，尤其是在技术、人为和操作性干扰随着时间的推移而累积的情况下。我们提出了操作韧性和体贴参与作为评估此类智能体的两个互补方面：前者捕捉智能体在阻塞工作中如何恢复，同时保持进展并传达其限制，后者则捕捉智能体的适应如何考虑受影响的人、角色边界和周围工作流程。然而，在不断累积的挑战下，这两者仍然未被充分探索。我们研究了120个模拟的医疗保健轨迹，涉及两个生成性人工智能模型和十二个利益相关者衍生的任务，在轻度、中度和重度挑战下进行比较。我们比较了文本行动计划、提示的内部评估以及定量结构化的工作负载和情感报告，以检查随着挑战的累积，智能体的行为和报告状态如何变化。关于操作韧性，智能体从自我导向的恢复转向更大的对人类的依赖，同时在结构化报告中报告增加的工作负载和负面情感，但在文本响应中很少表达压力。关于体贴参与，智能体从以任务为中心的适应扩展到任务重新框定、对他人的关注、角色边界调整和更广泛的协调，行动和内部评估之间存在明显的模式。基于这些发现，我们提出了五个部署困境，涉及坚持、关注、角色边界、状态披露和升级，这些困境需要利益相关者的具体说明，并进一步为学习、情境评估和具身适应的技术影响提供信息。

View on arXiv Download PDF AI Translation

cs.AI / 8 / 2609.10728

Towards a Deterministic Math Solver for Clinical Language Models

面向临床语言模型的确定性数学求解器

Osorio, Felipe Ocampo, Ordoñez, Sebastián Andrés Cajas, Lange, Maximin, Attrach, Rafi Al, Kapadia, Sahil, Sellami, Zakaria Laouabdia, Talio, Angelo Antonio, Celi, Leo Anthony

Abstract

Large language models are unreliable at arithmetic, which is a problem for clinical calculators where a single numerical error changes the recommendation. The standard response is to hardcode each calculator as a validated function, one at a time. We test an alternative: the model does not calculate. Instead, it writes case-specific Python that a restricted local executor runs as a deterministic solver, and the model's task reduces to deciding how to use it. We evaluate this Program-Solve interface on MedCalc-Bench Verified (1,100 cases, 55 calculators) against direct model arithmetic and a hand-written 22-calculator library, using Qwen2.5-7B and Qwen2.5-32B-AWQ, after auditing the benchmark's formulas against current clinical guidelines and flagging 16 of 55 with version, use or coefficient concerns. With formulas and gold variables supplied and both routes reading the whole note, handing off to the solver is not a reliable advantage at 7B (75.31% against 72.02%, a paired +3.29 points with a 95% calculator-cluster interval of [-3.49, 10.38]) but is one at 32B (90.53% against 83.47%, +7.05 [0.47, 14.60], clear of zero). The hand-written library is exact on its 440 supported cases but abstains elsewhere (40.0% overall). Adding an executor thus helps some open-weight models more than others even under matched formula, variable and note access, and is not a substitute for verified formulas or reliable variable extraction either way.

Chinese Translation

大型语言模型在算术运算方面不可靠，这对于临床计算器而言是一个问题，因为单个数值错误会改变推荐结果。标准的应对方法是将每个计算器逐一硬编码为经过验证的函数。我们测试了一种替代方案：模型不进行计算，而是编写特定案例的 Python 代码，由受限的本地执行器作为确定性求解器运行，模型的任务则简化为决定如何使用它。我们在 MedCalc-Bench Verified（1,100 个案例，55 个计算器）上评估了这一 Program-Solve 接口，比较了直接模型算术和手写的 22 个计算器库，使用了 Qwen2.5-7B 和 Qwen2.5-32B-AWQ，并在审核基准公式与当前临床指南后，标记了 55 个中有 16 个存在版本、使用或系数问题。在提供公式和黄金变量的情况下，且两种方法均读取完整的笔记，转交给求解器在 7B 模型中并不是一个可靠的优势（75.31% 对比 72.02%，配对 +3.29 分，95% 计算器集群区间为 [-3.49, 10.38]），但在 32B 模型中则是一个优势（90.53% 对比 83.47%，+7.05 [0.47, 14.60]，显著高于零）。手写库在其支持的 440 个案例中是精确的，但在其他情况下则不作答（整体为 40.0%）。因此，即使在匹配的公式、变量和笔记访问下，添加执行器对某些开放权重模型的帮助大于其他模型，并且无论如何都不能替代经过验证的公式或可靠的变量提取。

View on arXiv Download PDF AI Translation

cs.AI / 9 / 2609.10824

Studying Without a Syllabus: Task-Agnostic Environment Preprocessing

无大纲学习：任务无关环境预处理

Samuel, Vinay, Ursekar, Varun, Kalmath, Vijay S., Shanker, Apaar, Chatrath, Veronica, Xue, Yuan

Abstract

Before an LLM agent tackles tasks in a new environment, it can inspect available corpora and tools and construct reusable resources such as indices, scripts, or procedural guidance. Most automated adaptation methods, however, rely on task examples, trajectories, or evaluation feedback to decide what to build. Existing task-agnostic approaches avoid this supervision but commit in advance to a preparation strategy for a particular type of environment. We study a more open-ended setting: can an agent study an unfamiliar environment without a syllabus, i.e. before test time and without knowledge of the downstream task distribution, and choose how to prepare it? We formalize task-agnostic environment preprocessing, in which a studying system explores an environment under a budget and produces artifacts for a frozen solver. We compare unaided and archive-equipped meta-agents with fixed synthetic-practice and corpus-processing methods across six heterogeneous benchmarks. A meta-agent variant achieves the highest Avg@3 reward on five benchmarks, while fixed corpus processing remains best on the largest corpus benchmark. Larger study budgets do not reliably improve downstream reward. Nevertheless, studied artifacts reduce the test-time sampling needed to reach a given score, demonstrating how reusable preparation can shift computation from repeated test-time attempts to a pre-task study phase.

Chinese Translation

在大型语言模型（LLM）代理处理新环境中的任务之前，它可以检查可用的语料库和工具，并构建可重用的资源，如索引、脚本或程序指导。然而，大多数自动化适应方法依赖于任务示例、轨迹或评估反馈来决定构建内容。现有的任务无关方法避免了这种监督，但提前承诺了特定类型环境的准备策略。我们研究一种更开放的设置：代理是否可以在没有大纲的情况下学习一个不熟悉的环境，即在测试之前且不知道下游任务分布的情况下，选择如何进行准备？我们形式化了任务无关环境预处理，其中学习系统在预算限制下探索环境，并为一个固定求解器生成工件。我们比较了未辅助和配备档案的元代理在六个异质基准上的表现，使用固定的合成实践和语料处理方法。一种元代理变体在五个基准上获得了最高的 Avg@3 奖励，而固定的语料处理在最大的语料基准上表现最佳。更大的学习预算并不可靠地提高下游奖励。然而，研究过的工件减少了达到给定分数所需的测试时间采样，展示了可重用准备如何将计算从重复的测试时间尝试转移到任务前学习阶段。

View on arXiv Download PDF AI Translation

cs.AI / 10 / 2609.10873

When Validation Stops Learning: Auditing Update Admission for Continual Embodied Agents

当验证停止学习：持续体态智能体的更新接纳审计

Ma, Qinzhen, Wu, Ruihai

Abstract

Independent evaluation can reject harmful policy updates yet also prevent useful continual learning. We argue that update admission must be assessed through both error control and retained learning opportunities at a stated interaction budget. We identify a concrete failure: a range-based confidence gate cannot certify unchanged old-task behavior within otherwise substantial budgets. A standard paired-binomial construction reduces this burden when outcome disagreements are rare. We also specify certified historical-reference promotion and a round-level missed-opportunity metric. In a constructed one-step pushing diagnostic with 32 seeds, fresh paired checks admit 31.6% of a common update stream at 2,000 episodes per stage, versus zero for the range-based gate; unconditional replay nevertheless learns better in closed-loop runs. A separate learned-dynamics stress test distinguishes model bias from feedback-selection error. The contribution is an admission-audit protocol with analytical and synthetic evidence; physical-robot and VLA validation remain open.

Chinese Translation

独立评估可以拒绝有害的政策更新，但也可能阻碍有用的持续学习。我们认为，更新接纳必须通过错误控制和在特定交互预算下保留学习机会来评估。我们识别出一个具体的失败：基于范围的置信门无法在其他预算充足的情况下认证未改变的旧任务行为。当结果不一致的情况较少时，标准的配对二项式构造可以减轻这一负担。我们还指定了认证的历史参考提升和一个轮次级别的错失机会指标。在一个构建的一步推送诊断中，使用32个种子，新的配对检查在每个阶段的2000个回合中接纳了31.6%的常见更新流，而基于范围的门则为零；然而，无条件重放在闭环运行中仍然表现得更好。一个单独的学习动态压力测试区分了模型偏差与反馈选择错误。我们的贡献是一个具有分析和合成证据的接纳审计协议；物理机器人和VLA验证仍然是开放的。

View on arXiv Download PDF AI Translation

cs.AI / 11 / 2609.10964

Decoupling Readiness from Release for Tail-Aware Scheduling of Agentic LLM Workflows

将准备状态与释放解耦以实现面向尾部的智能 LLM 工作流调度

Feng, Bochao, Li, Jianjiang, Wang, Haojie, Qiao, Lin, Li, Yinghui, Yan, Yukun, Zhai, Jidong

Abstract

Agentic LLM workflows consist of sequences of model turns interleaved with tool interactions, so their end-to-end completion time depends not only on inference speed but also on when ready turns are released. Most runtimes release each turn immediately upon readiness. Under contention, this eager release policy can accumulate released but unfinished work; once submitted, those turns can no longer be reordered by the workflow-level policy, increasing tail latency. We present a tail-risk-aware turn release scheduling method that jointly decides which ready turn to release next and how much released but unfinished work to maintain. The method uses a mean--Conditional Value-at-Risk (CVaR) objective to capture the evolving tail risk of unfinished workflows, incorporates online estimates of turn work when prioritizing ready turns, and adapts the released work budget to observed queue pressure. We evaluate the method using real agent execution traces from software engineering tasks across multiple LLMs and workflow arrival rates. The method performs comparably to eager release under light load and substantially reduces the P95 of workflow flow time under contention, achieving up to a $3.50\times$ speedup.

Chinese Translation

智能 LLM 工作流由模型回合与工具交互交替组成，因此其端到端完成时间不仅依赖于推理速度，还依赖于准备好的回合何时被释放。大多数运行时在回合准备好后立即释放。面对竞争，这种急切的释放策略可能会积累已释放但未完成的工作；一旦提交，这些回合就无法再由工作流级别的策略重新排序，从而增加尾部延迟。我们提出了一种面向尾部风险的回合释放调度方法，该方法共同决定下一个释放的准备回合以及保持多少已释放但未完成的工作。该方法使用均值-条件风险价值（Conditional Value-at-Risk, CVaR）目标来捕捉未完成工作流的不断演变的尾部风险，在优先考虑准备回合时结合在线估算的回合工作，并根据观察到的队列压力调整释放的工作预算。我们使用来自多个 LLM 和工作流到达率的软件工程任务的真实代理执行轨迹来评估该方法。该方法在轻负载下的性能与急切释放相当，并在竞争情况下显著降低了工作流流转时间的 P95，达到了最高 3.50 倍的加速。

View on arXiv Download PDF AI Translation

cs.AI / 12 / 2609.10992

Demystifying the Privacy-Utility Trade-off in LLM Interactions

揭示大型语言模型交互中的隐私-效用权衡

Liu, Zhenhua, Xie, Zhanxu, Yu, Junjie, Zhu, Tong, Li, Lijun, Chen, Wenliang

Abstract

The integration of Large Language Models into daily tasks relies on context-rich instructions, inevitably exposing sensitive user information. Current privacy-preserving methods typically employ context-agnostic static rules, causing severe utility degradation. However, the specific mechanisms governing how sanitization impacts downstream performance remain largely underexplored. To address this, we conduct a systematic analysis to deconstruct the privacy-utility trade-off, uncovering three underlying mechanisms: (1) Context-Dependent Utility, which first establishes when to sanitize by revealing that data value shifts from critical constraints to dispensable noise based on user intent; (2) Strategic Adaptation, which subsequently determines how to sanitize by dictating that the choice between removal and replacement depends on the task's reliance on factual integrity versus structural coherence; and (3) Combinatorial Interplay, which finally extends the protection scope by demonstrating that attributes form a semantic web of synergistic dependencies or antagonistic redundancies. Guided by these insights, we introduce an intent-driven local protection framework. By distilling a lightweight model Veilmind-4B to drive a dynamic extraction-sanitization-restoration pipeline, our approach reaches a low-leakage privacy point while preserving substantially higher response utility than existing privacy-oriented baselines, advancing the privacy-utility trade-off toward the Pareto frontier.

Chinese Translation

大型语言模型在日常任务中的应用依赖于丰富的上下文指令，这不可避免地暴露了敏感用户信息。目前的隐私保护方法通常采用与上下文无关的静态规则，这导致效用严重下降。然而，关于去敏化如何影响下游性能的具体机制仍然很少被探讨。为了解决这个问题，我们进行了系统分析，以解构隐私-效用权衡，揭示了三种潜在机制：(1) 上下文依赖效用，该机制首先通过揭示数据价值在用户意图的影响下从关键约束转变为可丢弃的噪声，确定何时进行去敏化；(2) 战略适应，该机制随后通过规定去除与替换之间的选择取决于任务对事实完整性与结构一致性的依赖程度，确定如何进行去敏化；(3) 组合相互作用，该机制最终通过展示属性形成相互依赖的语义网络或对立冗余，扩展了保护范围。在这些见解的指导下，我们提出了一种基于意图的本地保护框架。通过提炼轻量级模型Veilmind-4B来驱动动态的提取-去敏化-恢复管道，我们的方法在保持显著更高的响应效用的同时，达到了低泄漏的隐私点，推动了隐私-效用权衡向帕累托前沿发展。

View on arXiv Download PDF AI Translation

cs.AI / 13 / 2609.11018

Defining AI Agents: A Compendium of Criteria, Metrics, and Benchmarks

定义人工智能代理：标准、指标和基准的汇编

Lassiter, Mia, Bent, Brinnae

Abstract

The term agent in artificial intelligence lacks a standard definition, complicating the evaluation, comparison, and reproducibility of AI agent research. We address this ambiguity through a survey organized around five dimensions of agenticness: environmental interaction, learning and adaptation, autonomy, goal-directed behavior, and temporal coherence. For each dimension, we examine how the underlying capability has been conceptualized across prior work and synthesize the metrics, benchmarks, and evaluation frameworks used to assess it. This review provides a structured account of the current landscape of agent evaluation, highlighting both established approaches and areas where evaluation remains limited or inconsistent. We additionally introduce the Agent Compendium, a public-facing digital resource that organizes and extends the evaluation methods identified through this review. Together, the survey and compendium provide a common structure for evaluating and comparing agent capabilities across AI systems, supporting more reproducible research, clearer communication, and more systematic study of artificial agents.

Chinese Translation

在人工智能领域，代理（agent）一词缺乏标准定义，这使得对AI代理研究的评估、比较和可重复性变得复杂。我们通过围绕代理特性（agenticness）的五个维度进行调查，以解决这一模糊性：环境互动、学习与适应、自主性、目标导向行为和时间一致性。对于每个维度，我们考察了先前研究中如何概念化其基础能力，并综合了用于评估这些能力的指标、基准和评估框架。此综述提供了当前代理评估领域的结构化概述，突出了既有的方法以及评估仍然有限或不一致的领域。此外，我们还介绍了代理汇编（Agent Compendium），这是一个面向公众的数字资源，组织并扩展了通过本综述识别的评估方法。综述和汇编共同为评估和比较AI系统中代理能力提供了一个共同的框架，支持更具可重复性的研究、更清晰的沟通以及对人工代理的更系统的研究。

View on arXiv Download PDF AI Translation

cs.AI / 14 / 2609.11030

The Agent Incident Registry: Toward Preventing Repeated AI Agent Failures

代理事件登记册：防止重复的人工智能代理失败

Kumar, Divyanshu, HN, Rohith, Birur, Nitin Aravind, Agarwal, Sahil, Harshangi, Prashanth

Abstract

AI agents increasingly act through tools and delegated authority, but general incident repositories rarely capture the mechanisms needed to compare public failures with agent-security evaluations. We present the Agent Incident Registry (AIR), a source-linked catalog containing \N{} records of agent-related events disclosed from \Yfirst{} through \Ylast{}. Each record includes supporting evidence, a stable identifier, and missingness-aware labels for causal role, disclosure class, mechanism, and outcome. Among the \Nprimary{} generative-system records in which the agent acted, \Rprimary{} involved realized harm (\Pprimary\%). Realized outcomes concentrate in in-the-wild and safety-failure records, while responsible disclosures and research demonstrations are overwhelmingly demonstrated; the aggregate share therefore characterizes collection composition rather than deployment risk. After initial curation, a second human reviewer checked all \N{} records and their existing labels for completeness and correctness. In a deployment-analogue audit, InjecAgent's \NInjecAgentCases{} cases occupy three of AIR's twelve surfaces and are all attacker-triggered, whereas AIR contains \Nsafety{} no-adversary safety failures. AIR supports source-grounded case retrieval and evaluation-scope auditing, not failure-rate or control-efficacy estimation.

Chinese Translation

人工智能代理越来越多地通过工具和委托权力进行操作，但一般事件存储库很少捕捉到比较公共失败与代理安全评估所需的机制。我们提出了代理事件登记册（Agent Incident Registry, AIR），这是一个源链接目录，包含从 extit{Yfirst}到 extit{Ylast}披露的 extit{N}条代理相关事件记录。每条记录包括支持证据、稳定标识符以及针对因果角色、披露类别、机制和结果的缺失感知标签。在 extit{Nprimary}条生成系统记录中，代理参与的 extit{Rprimary}条涉及实际伤害（ extit{Pprimary extit{ extperthousand}}）。实际结果集中在实际环境和安全失败记录中，而负责任的披露和研究演示则占据绝大多数；因此，整体份额特征化了集合组成而非部署风险。在初步整理后，第二位人工审查员检查了所有 extit{N}条记录及其现有标签的完整性和正确性。在一次部署类审计中，InjecAgent的 extit{NInjecAgentCases}个案例占据了AIR的十二个表面中的三个，并且全部是攻击者触发的，而AIR包含 extit{Nsafety}个无对抗者的安全失败。AIR支持基于源的案例检索和评估范围审计，而非失败率或控制有效性估计。

View on arXiv Download PDF AI Translation

cs.AI / 15 / 2609.11060

Grounding Agent Memory: Environment-Probing Curation for Enterprise Agents

基础代理记忆：面向企业代理的环境探测策划

Suresh, Susheel, Mak, Hazel, Bhatnagar, Sahil, Methani, Chhaya, Munoz, Alejandro Gutierrez

Abstract

Persistent memory is entering production-oriented agent platforms to help long-horizon agents accumulate experience across sessions. Yet a post-task curator agent restricted to completed trajectories can preserve errors, overgeneralize partial evidence, or retain stale knowledge. We introduce environment-probing curation, a deployment-compatible extension that gives an existing asynchronous curator agent least-privilege, read-only world tools to check, scope, and refresh candidate memories. It requires no model retraining and leaves the task agent, retriever, memory representation, and production write authority unchanged. In a production-like GitHub Copilot (GHCP) harness built on its SDK, we compare stateless execution, full in-context learning, GHCP + Mem, and GHCP + Mem (w/ Env Probing) on CLBench database exploration and 90 adapted APEX management-consulting tasks. On CLBench, probing raises pass rate from 39% to 73% and pass-discounted reward from 8.60 to 22.60 while reducing queries from 8.8 to 4.7 per question and task-agent cost from \$3.38 to \$1.68. Across six APEX worlds, all 18 memory-versus-baseline mean reward comparisons are positive and task-agent tool calls fall by 16--75%; probing gives the best task-agent reward gain per dollar in five worlds. Probing also attains higher mean reward than GHCP + Mem on both Sonnet 4.6 and Opus 4.7 without schema drift. Environment probing therefore turns existing agent-memory curation into an environment-informed, auditable process while preserving a compact task-time interface.

Chinese Translation

持久性记忆正在进入以生产为导向的代理平台，以帮助长时间跨度的代理在多个会话中积累经验。然而，受限于已完成轨迹的后任务策划代理可能会保留错误、过度概括部分证据或保留过时知识。我们引入了环境探测策划，这是一种兼容部署的扩展，赋予现有的异步策划代理最小权限的只读世界工具，以检查、范围和刷新候选记忆。它不需要模型重训练，并且不改变任务代理、检索器、记忆表示和生产写入权限。在基于其SDK构建的类似生产的GitHub Copilot (GHCP) 环境中，我们比较了无状态执行、完全上下文学习、GHCP + Mem和GHCP + Mem（带环境探测）在CLBench数据库探索和90个适应性APEX管理咨询任务上的表现。在CLBench上，探测将通过率从39%提高到73%，通过折扣奖励从8.60提高到22.60，同时将每个问题的查询从8.8减少到4.7，任务代理成本从3.38美元降低到1.68美元。在六个APEX世界中，所有18个记忆与基线的平均奖励比较均为正，任务代理工具调用减少了16%至75%；在五个世界中，探测为每美元提供了最佳的任务代理奖励增益。探测在Sonnet 4.6和Opus 4.7上也获得了比GHCP + Mem更高的平均奖励，且没有模式漂移。因此，环境探测将现有的代理记忆策划转变为一个环境知情的、可审计的过程，同时保持紧凑的任务时间接口。

View on arXiv Download PDF AI Translation

cs.AI / 16 / 2609.11061

Fork Where the Model Changes Its Mind: Belief-Shift Branching for Tree-Structured Reinforcement Learning

模型改变主意的分叉：树结构强化学习中的信念转变分支

Lei, Bin, Li, Yu, Choubey, Prafulla Kumar, Zhang, Jiaxin, Peng, Becky Xiangyu, Ye, Qinyuan, Narayan, Kartik, Ding, Caiwen, Savarese, Silvio, Wu, Chien-Sheng

Abstract

Tree-structured rollouts give critic-free reinforcement learning with verifiable rewards (RLVR) step-level credit: fork a chain at an intermediate point, and sibling outcome differences estimate step value. Each fork adds sampling cost, so realistic budgets typically allow only a few forks per chain. A fork placed where the outcome is already largely settled yields siblings that mostly agree and provide almost no credit signal; hence, for a given tree size, where forks are placed largely determines how much step-level RL can gain. Most existing mainstream methods place forks by structure, such as fixed lengths, midpoints, and delimiters, or by next-token entropy. We formalize fork placement as locating the \emph{pivots} of the chain's value curve, where the expected outcome turns. We propose \emph{belief-shift branching}: read the model's answer belief at candidate boundaries and fork just before the step where consecutive beliefs diverge most. Three instantiations, none needing step-level supervision, span access levels: a black-box probe, a logit-lens depth profile, and a learned activation direction, which is fit offline and therefore used only in the validation before RL training. The signal only \emph{places} forks, and the probe costs about $1\%$ of step compute on mathematics and under $5\%$ on code when it runs inside the rollout engine. In that validation, against Monte-Carlo value curves, a belief-shift signal ranks first in each of the eight model$\times$benchmark panels, ahead of entropy, structural, and LLM-judge baselines. In RL across three model families and two domains, belief-shift forking leads every mathematics aggregate, on OLMo-3-7B by $+2.6$ aggregate and $+2.9$ on AIME 2026 over the strongest baseline, and sweeps every OLMo code column, by $+6.5$ on LiveCodeBench-medium.

Chinese Translation

树结构展开提供了无评论的强化学习与可验证奖励（RLVR）逐步信用：在中间点分叉链条，兄弟结果差异估计步骤值。每个分叉增加了采样成本，因此现实预算通常只允许每条链有少数分叉。分叉放置在结果已经基本确定的地方会产生大多数一致的兄弟，从而几乎没有信用信号；因此，对于给定的树大小，分叉的位置在很大程度上决定了逐步强化学习能获得多少收益。大多数现有主流方法通过结构（如固定长度、中点和分隔符）或通过下一个标记的熵来放置分叉。我们将分叉放置形式化为定位链值曲线的 extit{枢轴}，即期望结果转变的地方。我们提出了 extit{信念转变分支}：在候选边界读取模型的答案信念，并在连续信念最大分歧的步骤之前进行分叉。三个实例化，不需要逐步监督，涵盖了不同的访问级别：一个黑箱探测器、一个逻辑透镜深度轮廓和一个离线拟合的学习激活方向，因此仅在强化学习训练前的验证中使用。该信号仅 extit{放置}分叉，探测器在数学计算中大约占步骤计算的$1\%$，在代码中运行时低于$5\\%$。在该验证中，与蒙特卡洛值曲线相比，信念转变信号在每个八个模型$ imes$基准面板中排名第一，领先于熵、结构和LLM评判基线。在三个模型家族和两个领域的强化学习中，信念转变分叉在每个数学聚合中领先，OLMo-3-7B的聚合提高了$+2.6$，在AIME 2026中超过最强基线提高了$+2.9$，并在每个OLMo代码列中横扫，LiveCodeBench-medium提高了$+6.5$。

View on arXiv Download PDF AI Translation

cs.AI / 17 / 2609.11065

MOSAIC: Query-Aware Exploration Policy Adaptation for GraphRAG

MOSAIC：针对GraphRAG的查询感知探索策略适应

Lee, EunKyeong, Oh, Kyeong-Jin, Kim, Jinwon, Lee, Hye Woo, Song, Minsang, Jang, Hyeongjun, Youn, Junyoung

Abstract

Graph Retrieval-Augmented Generation (GraphRAG) can connect evidence distributed across a corpus graph, but most systems use largely shared exploration procedures across queries. This creates a structural mismatch: direct facts may need compact local neighborhoods, comparisons need balanced coverage of multiple targets, and mediated questions may require deeper paths through weakly related connectors. We present Mosaic, a training-free framework that formulates GraphRAG retrieval as a per-query control problem. An LLM analyzer converts query-specific evidence requirements into a bounded policy over seed selection, graph traversal, stopping, and evidence selection, while the corpus graph, indexes, scoring functions, grounding procedure, and answer generator remain shared. On GraphRAG-Bench, Mosaic achieves query-weighted Answer Correctness of 76.97 on Medical and 64.33 on Novel, improving over the strongest previously reported overall results by 5.13 and 4.43 points. On Medical, it reaches 95.1 Evidence Recall and 86.1 Context Relevancy. Controlled comparisons on an identical graph and generator show that no fixed narrow, medium, or wide policy is consistently optimal; Mosaic improves by 9.96 points over the strongest canonical fixed policy. Relative to Fixed Wide, it evaluates 81.9% fewer paths and retains 47.2% fewer evidence items. Transfer experiments on HotpotQA, MuSiQue, and 2WikiMultiHopQA further show that the policy interface can be applied without benchmark-specific retriever training.

Chinese Translation

图检索增强生成（GraphRAG）能够连接分布在语料图中的证据，但大多数系统在查询之间使用了大致相同的探索过程。这造成了结构不匹配：直接事实可能需要紧凑的局部邻域，比较需要对多个目标的平衡覆盖，而中介问题可能需要通过弱相关连接器的更深路径。我们提出了Mosaic，这是一个无训练框架，将GraphRAG检索形式化为每个查询的控制问题。一个大型语言模型（LLM）分析器将查询特定的证据需求转换为关于种子选择、图遍历、停止和证据选择的有界策略，同时语料图、索引、评分函数、基础程序和答案生成器保持共享。在GraphRAG-Bench上，Mosaic在医疗领域的查询加权答案正确率达到76.97，在新颖性领域达到64.33，分别比之前报告的最强整体结果提高了5.13和4.43点。在医疗领域，它达到了95.1的证据召回率和86.1的上下文相关性。在相同图和生成器上的受控比较显示，没有固定的狭窄、中等或宽泛策略是一致最优的；Mosaic比最强的经典固定策略提高了9.96点。相较于固定宽泛策略，它评估的路径减少了81.9%，保留的证据项减少了47.2%。在HotpotQA、MuSiQue和2WikiMultiHopQA上的迁移实验进一步表明，该策略接口可以在不进行基准特定检索器训练的情况下应用。

View on arXiv Download PDF AI Translation

cs.AI / 18 / 2609.11115

Benchmark Radar: A Living Database and Search Engine for AI Benchmarks and Evaluation

基准雷达：一个动态数据库和人工智能基准与评估的搜索引擎

Wu, Koutian, Zhou, Junjie, Shang, Ergan, Wang, Jiayu, Han, Pengqian, Wang, Junkai, Xu, Wanghan

Abstract

Benchmark researchers and developers of large language models (LLMs) and other AI systems need to find relevant evaluations, locate their benchmark datasets and code, and understand the settings behind reported scores. We present Benchmark Radar, a living database and search engine for retrieval and discovery of AI benchmarks, covering LLM evaluation, agentic and tool-use benchmarks, coding, reasoning, safety, and domain-specific evaluations. The system combines daily discovery of benchmark papers, repositories, datasets, and releases with a searchable benchmark catalog, mentions in model cards and technical reports, and score histories. It retains source identities and citations so readers can inspect candidate benchmarks and their evaluation evidence. Daily discovery draws on 37 sources: 13 direct connectors and 24 first-party research and engineering feeds. The catalog contains 1,283 source records drawn from 4 benchmark catalogs and 12,916 numeric observations on 790 records. We describe collection and retrieval, audit the full catalog, and examine benchmark saturation, adoption trends, and the limits of score comparisons. A worked example walks through a complete prior-art search, showing how to query the catalog and inspect benchmark evidence when designing a new evaluation. We release the web dashboard with a benchmark leaderboard, a Pareto frontier view of score against measured use, saturation and trend views, daily feeds, downloadable evidence, a command-line interface (CLI) for offline queries, and reproducible analysis.

Chinese Translation

基准研究人员以及大型语言模型（LLMs）和其他人工智能系统的开发者需要找到相关的评估，定位他们的基准数据集和代码，并理解报告分数背后的设置。我们提出了基准雷达（Benchmark Radar），这是一个动态数据库和搜索引擎，用于检索和发现人工智能基准，涵盖LLM评估、代理和工具使用基准、编码、推理、安全性以及特定领域的评估。该系统结合了每日发现的基准论文、代码库、数据集和发布内容，并提供可搜索的基准目录、模型卡和技术报告中的提及以及分数历史。它保留了来源身份和引用，以便读者可以检查候选基准及其评估证据。每日发现来自37个来源：13个直接连接器和24个第一方研究与工程源。该目录包含1,283个源记录，来自4个基准目录和12,916个关于790个记录的数值观察。我们描述了收集和检索过程，审计了完整目录，并考察了基准饱和度、采用趋势以及分数比较的限制。一个实例展示了完整的前沿搜索过程，说明了在设计新评估时如何查询目录并检查基准证据。我们发布了带有基准排行榜的网页仪表板、分数与测量使用的帕累托前沿视图、饱和度和趋势视图、每日信息流、可下载的证据、用于离线查询的命令行接口（CLI）以及可重复的分析。

View on arXiv Download PDF AI Translation

cs.AI / 19 / 2609.11127

KuaiRP Series Role-playing Models Technical Report

KuaiRP系列角色扮演模型技术报告

Wang, Yipeng, Zhang, Ziwei, Zhang, Jiahui, Gan, Qi, Sheng, Kai

Abstract

This paper introduces the complete technical solution for the KuaiRP series of role-playing models. We aim to achieve four core objectives for a dedicated role-playing model: simplified prompt engineering, highly stable output quality, built-in domain world knowledge, and high-efficiency deployment with a small parameter size. However, effectively injecting deep domain knowledge often leads to a severe catastrophic forgetting of the model's general agent capabilities. To overcome this trade-off, we propose a multi-stage training pipeline. First, we design a standardized character template and construct an SFT data pipeline based on user behavior simulation and reverse profile filtering. Next, we utilize a rule-based composite reward function during the Reinforcement Learning (RL) phase to eliminate common degradation phenomena like length expansion and repetitive generation. Finally, to recover the general capabilities compromised during SFT and RL, we propose a novel self-distillation paradigm using Two-stage On-Policy Distillation (OPD) equipped with Cumulative-Divergence Decay (CDD). By using the domain-adapted model as the teacher and the original base model as the student, we effectively balance deep domain knowledge injection with the preservation of general agent capabilities. Experimental results demonstrate that the KuaiRP models not only match the current state-of-the-art proprietary models in role-playing fidelity within our target domains, but also successfully recover general agent capabilities, maintaining extremely low deployment costs.

Chinese Translation

本文介绍了KuaiRP系列角色扮演模型的完整技术解决方案。我们的目标是实现专用角色扮演模型的四个核心目标：简化提示工程、高度稳定的输出质量、内置领域世界知识以及小参数规模的高效部署。然而，有效地注入深层领域知识往往会导致模型一般智能能力的严重灾难性遗忘。为了解决这一权衡问题，我们提出了一种多阶段训练流程。首先，我们设计了一个标准化的角色模板，并基于用户行为模拟和反向画像过滤构建了一个SFT（监督微调）数据管道。接下来，在强化学习（RL）阶段，我们利用基于规则的复合奖励函数来消除长度扩展和重复生成等常见退化现象。最后，为了恢复在SFT和RL过程中受到损害的一般能力，我们提出了一种新颖的自蒸馏范式，使用配备累积发散衰减（CDD）的两阶段在线策略蒸馏（OPD）。通过将领域适应模型作为教师，将原始基础模型作为学生，我们有效地平衡了深层领域知识的注入与一般智能能力的保留。实验结果表明，KuaiRP模型不仅在我们的目标领域中与当前最先进的专有模型在角色扮演的逼真性上相匹配，而且成功恢复了一般智能能力，保持了极低的部署成本。

View on arXiv Download PDF AI Translation

cs.AI / 20 / 2609.11144

Same Day, Same Story; One Day Ahead, a Different Signal: The Dual Validity of Financial Sentiment

同一天，同样的故事；提前一天，不同的信号：金融情绪的双重有效性

Aravinthkakshan, AS, Srivastava, Laven, Nandwani, Harsh

Abstract

Financial NLP has a standard workflow: validate a sentiment tool against human labels, then trust it to extract market signal. This assumes the two evaluations measure the same thing. We test that assumption in a setting where both can be measured at once: a corpus of securities class actions (2002-2025) linking 70,500 X messages to abnormal stock returns, with a single-annotator human labelled gold sample. Running five instruments (VADER, Loughran-McDonald, FinBERT, Twitter-RoBERTa, and an LLM annotator) through one identical pipeline, we find that the relationship between construct and predictive validity depends on the sampling convention and score representation. Under conventional method-specific sampling, human agreement aligns more closely with graded same-day associations than with one-day leads. On a fixed-n panel, however, agreement has similar graded rank correlations at both horizons, while the coarse ordering remains weak. Benchmark agreement therefore establishes semantic validity but does not by itself determine predictive rankings. In a conversation that is 17.6% spam, message volume predicts neither market damage nor settlement size.

Chinese Translation

金融自然语言处理（NLP）有一个标准工作流程：将情绪工具与人类标签进行验证，然后信任其提取市场信号。这假设这两种评估测量的是同一事物。我们在一个可以同时测量这两者的环境中测试这一假设：一个证券集体诉讼语料库（2002-2025），将70,500条X消息与异常股票收益关联，并提供一个单一标注者的人类标注金样本。通过一个相同的流程运行五种工具（VADER、Loughran-McDonald、FinBERT、Twitter-RoBERTa和一个大型语言模型（LLM）标注器），我们发现构念有效性与预测有效性之间的关系取决于抽样惯例和得分表示。在传统的特定方法抽样下，人类一致性与同日的分级关联更为接近，而与提前一天的关联则较弱。然而，在固定样本面板下，一致性在两个时间范围内具有相似的分级秩相关性，而粗略排序仍然较弱。因此，基准一致性确立了语义有效性，但本身并不能决定预测排名。在一个17.6%为垃圾信息的对话中，消息量既无法预测市场损害，也无法预测和解金额。

View on arXiv Download PDF AI Translation

cs.AI / 21 / 2609.11146

The Oligarch Barely Steers Model Collapse in Multi-Model Ecosystems

寡头几乎引导多模型生态系统中的模型崩溃

Liu, Yangze, Han, Zhongyi

Abstract

AI-generated text is flowing back into the training corpora of the next generation of models. Recursive training on it drives model collapse, and recent work extends the setting to many models feeding one another -- but almost always with the market split evenly, while real generative AI is an oligopoly. Concentration raises two worries: fewer, more uniform sources may make collapse faster, and later models may be dragged toward the oligarch's output. We test both in controlled ecosystems: 13 open 1--4B models form natural ecosystems of 3 to 13 players, plus an injected probe that pushes the top share to 90%; each generation, every model's output is mixed into a shared pool by market share and every model is retrained on that pool from clean base weights, for five generations. Yet within the range we test, neither worry materializes; what emerges instead is an invariance. Making the split more unequal barely changes the speed of collapse. Destinations move even less: the share and identity knobs shift five-generation endpoints by only a few percent of the drift common to all arms -- the ecosystems collapse to nearly the same place. An extreme share paired with the strongest injected bias still does not guarantee steering, and the topic shifts it does produce leave only a faint trace on the ruler that measures collapse. What sets the speed is who supplies the pool and how readily those suppliers are carried along: with every share held fixed, swapping the members of a K=3 ecosystem changes five-generation drift by 2.8x; a share-weighted index of each member's susceptibility explains the speed differences across nineteen arms with R^2 = 0.68; and replacing half the pool with human text roughly halves drift without changing its course. Within the tested range, concentration sets neither the destination nor the pace of collapse; the pace follows whose text fills the pool.

Chinese Translation

AI生成的文本正在流回下一代模型的训练语料库。对其进行递归训练会导致模型崩溃，最近的研究将这一设置扩展到多个模型相互影响的情况——但几乎总是市场均分，而真实的生成性AI则是一个寡头垄断。集中化引发了两个担忧：更少且更统一的来源可能会加速崩溃，而后来的模型可能会被拉向寡头的输出。我们在受控生态系统中测试了这两种情况：13个开放的1-4B模型形成了3到13个参与者的自然生态系统，此外还有一个注入的探针将顶级份额推高至90%；每一代中，每个模型的输出根据市场份额混合到一个共享池中，每个模型都从干净的基础权重开始在该池上重新训练，持续五代。然而，在我们测试的范围内，这两种担忧均未出现；相反，出现了一种不变性。使分配更不均等几乎不会改变崩溃的速度。最终结果的变化甚至更小：份额和身份的调整仅使五代的终点变化了所有模型共同漂移的几个百分点——生态系统几乎崩溃到同一个地方。极端的份额配合最强的注入偏见仍然无法保证引导，而它所产生的主题变化在衡量崩溃的标准上仅留下微弱的痕迹。决定速度的是谁提供了池子，以及这些供应者被多快地带动：在每个份额固定的情况下，替换K=3生态系统的成员使五代漂移变化了2.8倍；每个成员的易感性加权指数解释了十九个模型之间的速度差异，R^2 = 0.68；而用人类文本替换池子的一半大致将漂移减半而不改变其方向。在测试范围内，集中化既不设定崩溃的目的地，也不设定崩溃的速度；速度取决于谁的文本填充了池子。

View on arXiv Download PDF AI Translation

cs.AI / 22 / 2609.11147

Autonomous Chemical Mechanistic Discovery through Agentic Reasoning and Validation

通过自主推理和验证实现化学机制的自动发现

Li, Dong, Mi, Sixuan, Ye, Zihao, Xiong, Huan, XU, Tao, Zhu, Tong, Zhang, Aijia, Gao, Junqi, Zhang, Kaiyan, Wang, Shijie, Zhou, Bowen, Li, Yuqiang, Qi, Biqing

Abstract

Unraveling reaction mechanisms is central to modern chemistry, yet automating these investigations remains challenging because computational workflows still rely heavily on expert intervention. Here we introduce ARCHE, an autonomous agentic system that integrates a general-purpose reasoning model, a domain-specialized computational chemistry model, and a structured tool registry to transform mechanistic inquiry into a scalable, self-validating process. ARCHE interprets scientific questions, generates and prioritizes mechanistic hypotheses, orchestrates computational workflows, and iteratively refines conclusions based on computed evidence within a closed loop. We validate its capabilities across three increasingly demanding scenarios: reconstructing stereocontrolling transition states and validating the corresponding reaction mechanism in a previously reported asymmetric catalytic reaction; proposing and validating a plausible radical pathway through iterative hypothesis refinement for a recently discovered but unpublished $\alpha$-iodoboronate C-I cleavage reaction; and identifying a chemically interpretable descriptor that governs selectivity in nickel-catalysed migratory cross-coupling reactions. By coupling agentic reasoning with rigorous computational validation, ARCHE advances autonomous mechanistic discovery and establishes a foundation for broader machine-assisted chemical research. The code for ARCHE is publicly available at https://github.com/JetAstra/Arche-Harness.

Chinese Translation

揭示反应机制是现代化学的核心，但由于计算工作流程仍然高度依赖专家干预，自动化这些研究仍然面临挑战。在此，我们介绍了ARCHE，一个自主代理系统，它整合了通用推理模型、领域专用计算化学模型和结构化工具注册表，将机制探究转变为可扩展的自验证过程。ARCHE能够解读科学问题，生成并优先排序机制假设，协调计算工作流程，并基于计算证据在闭环内迭代地完善结论。我们在三个日益复杂的场景中验证了其能力：重建立体控制过渡态并验证先前报道的非对称催化反应中的相应反应机制；通过迭代假设完善提出并验证一个合理的自由基路径，针对最近发现但尚未发表的α-碘硼酸盐C-I断裂反应；以及识别一个化学可解释的描述符，该描述符控制镍催化迁移交叉偶联反应中的选择性。通过将自主推理与严格的计算验证相结合，ARCHE推动了自主机制发现，并为更广泛的机器辅助化学研究奠定了基础。ARCHE的代码已在https://github.com/JetAstra/Arche-Harness上公开。

View on arXiv Download PDF AI Translation

cs.AI / 23 / 2609.11155

DRG-MAPPO: Hierarchical Dynamic Role-Graph Multi-Agent Reinforcement Learning for Cooperative Air Combat

DRG-MAPPO：用于合作空战的分层动态角色图多智能体强化学习

Liu, Junlin, Li, Chengwei, Gao, Yang, Chang, Hui, Zhang, Xinchen, Zhao, Zhijun, Zhao, Hao

Abstract

Multi-Agent Reinforcement Learning (MARL) has emerged as a pivotal paradigm for complex decision-making in autonomous systems and air combat. While MARL has demonstrated significant potential in air combat, achieving sophisticated tactical coordination remains a non-trivial challenge. This difficulty is largely attributed to two primary limitations: (1) the absence of structured relational modeling hinders agents from capturing complex, time-varying interactions among battlefield entities; and (2) conventional flat architectures often lack the capability to explicitly model tactical roles, leading to ambiguous task allocation in highly dynamic environments. To address these challenges, we propose Hierarchical Dynamic Role-Graph Multi-Agent Proximal Policy Optimization (DRG-MAPPO), a novel MARL framework that integrates graph-based relational modeling with dynamic role assignment. Specifically, DRG-MAPPO constructs a graph-based representation of battlefield interactions and leverages graph attention mechanisms to extract critical relational features among allies, enemies, and threats. Subsequently, a high-level policy employs a dynamic role assignment mechanism to determine tactical responsibilities (e.g., ``leader'' and ``supporter''). Conditioned on these roles and encoded graph-relational features, a low-level policy executes discrete maneuver actions, facilitating the joint optimization of tactical strategy and collaborative execution. Furthermore, a target-priority auxiliary task is designed to foster the emergence of behaviors such as focus-fire. Experimental results demonstrate that DRG-MAPPO achieves a state-of-the-art win rate of 87%, suggesting that our framework effectively balances relational modeling, interpretability, and optimization stability for cooperative air combat.

Chinese Translation

多智能体强化学习（MARL）已成为自主系统和空战中复杂决策的重要范式。尽管MARL在空战中展现了显著的潜力，但实现复杂的战术协调仍然是一项非平凡的挑战。这一困难主要归因于两个主要限制：（1）缺乏结构化的关系建模使得智能体难以捕捉战场实体之间复杂的、时变的互动；（2）传统的扁平架构往往缺乏明确建模战术角色的能力，导致在高度动态环境中任务分配模糊。为了解决这些挑战，我们提出了分层动态角色图多智能体近端策略优化（DRG-MAPPO），这是一个新颖的MARL框架，结合了基于图的关系建模和动态角色分配。具体而言，DRG-MAPPO构建了战场互动的图形表示，并利用图注意力机制提取盟友、敌人和威胁之间的关键关系特征。随后，高层策略采用动态角色分配机制来确定战术责任（例如，“指挥者”和“支持者”）。基于这些角色和编码的图关系特征，低层策略执行离散机动动作，从而促进战术策略和协作执行的联合优化。此外，设计了一个目标优先级辅助任务，以促进如集中火力等行为的出现。实验结果表明，DRG-MAPPO实现了87%的最先进胜率，表明我们的框架有效平衡了关系建模、可解释性和优化稳定性，以支持合作空战。

View on arXiv Download PDF AI Translation

cs.AI / 24 / 2609.11170

Breaking Predictions Is Not Enough: Specified-Foil Counterfactuals for Temporal Graphs

打破预测并不足够：针对时间图的指定对立反事实

Yu, Minwoo, Ha, Young-guk

Abstract

Temporal graph counterfactual explanations typically change past events to change or invalidate an original prediction, while leaving its replacement unspecified. Yet a user facing a predicted outcome often asks which past conditions would make a particular alternative occur instead. We formulate this destination-specific question as the Specified-Foil Counterfactual: given an original prediction A and a foil B fixed before search, find a low-cost past-event intervention under which the same predictor selects B as top-ranked. Our trace-guided intervention search contrasts the completed execution of A with a reconstructed incomplete execution of B, maps their difference to DELETE, INSERT, REWIRE, RELABEL, and SHIFT operations, and verifies B through exact replay. We instantiate this principle with LiFTER on continuous-time dynamic graphs and TLogic on temporal knowledge graphs. On CTDGs, the method retains 85.7-93.6% of black-box greedy successes while reducing predictor evaluations by 75.0-80.0%; on TKGs, it reaches the specified foil in 74.8% of 600 comparisons. Executable traces thereby become computational structures for constructing conditions of unselected alternatives, rather than records used only to explain predictions already made.

Chinese Translation

时间图的反事实解释通常通过改变过去事件来改变或无效化原始预测，同时未指定其替代结果。然而，面对预测结果的用户常常会询问哪些过去条件会使特定的替代结果发生。我们将这一目标特定的问题表述为指定对立反事实：给定一个原始预测 A 和一个在搜索前固定的对立 B，寻找一个低成本的过去事件干预，使得同一预测器将 B 选为最高排名。我们的轨迹引导干预搜索将 A 的完整执行与 B 的重构不完整执行进行对比，将它们的差异映射到 DELETE、INSERT、REWIRE、RELABEL 和 SHIFT 操作，并通过精确重放验证 B。我们在连续时间动态图上使用 LiFTER，在时间知识图上使用 TLogic 来实例化这一原则。在 CTDGs 上，该方法保留了 85.7-93.6% 的黑箱贪婪成功率，同时将预测器评估减少了 75.0-80.0%；在 TKGs 上，它在 600 次比较中达到了 74.8% 的指定对立。可执行轨迹因此成为构建未选择替代条件的计算结构，而不仅仅是用于解释已做出预测的记录。

View on arXiv Download PDF AI Translation

cs.AI / 25 / 2609.11176

Debate-to-Skill: Capability-Bound Process Supervision for Industrial Query-to-Agent Annotation

辩论到技能：面向工业查询到代理注释的能力约束过程监督

Zhang, Shiyu, Cheng, Leisheng, Li, Huifu

Abstract

Industrial query-to-agent matching fails when topical relevance is mistaken for executable capability, especially on long-tail and boundary-sensitive requests. We formulate annotation as \emph{capability-bound process supervision} and instantiate it with Debate-to-Skill, which uses reusable decision principles, structured deliberation, verifier-based verdict extraction, and disagreement-driven refinement. On an industrial Query2Agent benchmark, we compare Debate-to-Skill with direct-label supervision, reasoning-SFT, and structural ablations. The results test whether gains come from supervising the capability-critical decision process itself, especially on grey-zone cases where semantic relatedness and executable capability diverge.

Chinese Translation

当主题相关性被误认为可执行能力时，工业查询到代理匹配会失败，尤其是在长尾和边界敏感的请求中。我们将注释形式化为 extit{能力约束过程监督}，并通过辩论到技能（Debate-to-Skill）进行实例化，该方法使用可重用的决策原则、结构化的深思熟虑、基于验证者的裁决提取和基于分歧的细化。在一个工业查询到代理基准上，我们将辩论到技能与直接标签监督、推理-SFT和结构性消融进行了比较。结果检验了收益是否来自于对能力关键决策过程本身的监督，特别是在语义相关性和可执行能力出现分歧的灰色区域案例中。

View on arXiv Download PDF AI Translation

cs.AI / 26 / 2609.11180

SemVerBench: Benchmarking LLM Comprehension of Version-Constraint Resolution Semantics

SemVerBench：大型语言模型对版本约束解析语义的基准测试

Chen, Qibai, Liu, Zeming

Abstract

Large language model (LLM) coding agents constantly decide whether a version satisfies a constraint such as ^1.2.3 or >=2.0,<3, yet their grasp of version-constraint semantics has never been measured directly. We introduce SemVerBench, the first benchmark of LLM version-constraint resolution semantics across three ecosystems (npm, PEP 440, Cargo): 240 machine-checkable items with unique answers, built author-neutrally from four balanced sources (each ecosystem's official test suite plus three frontier LLM proposers) and labeled by a non-circular two-implementation oracle. Evaluating six frontier models, we find systematic, predictable per-mechanism blind spots: a partial-comparator carry rule (>1.2 means >=1.3.0) traps every model on Cargo (near 60%), and although standard PEP 440 prefix matching is universal, on zero-pad/post-release corner cases GPT-5.1 collapses (0/26) while Claude stays at 97-100% (verified on a 67-item oracle-validated set). Opus significantly outperforms all other models, and Sonnet outperforms the OpenAI models (McNemar). The failures look more like an activation/application gap than a knowledge gap: injecting the rule or a light correct hint recovers most errors, whereas interval decomposition does not, and models are at ceiling on the basic forms of the same rules. An author-stratified analysis finds no statistically significant self-favoritism. Because the task is verifiable and a free, 100%-correct resolver exists, tool delegation reaches ~100%: coding agents should delegate version resolution to a resolver rather than reason about versions in-head.

Chinese Translation

大型语言模型（LLM）编码代理不断决定某个版本是否满足诸如 ^1.2.3 或 >=2.0,<3 的约束，但它们对版本约束语义的理解从未被直接测量。我们引入了 SemVerBench，这是第一个针对 LLM 版本约束解析语义的基准测试，涵盖三个生态系统（npm、PEP 440、Cargo）：240 个具有唯一答案的机器可检查项，基于四个平衡来源（每个生态系统的官方测试套件加上三个前沿 LLM 提议者）中立构建，并由一个非循环的双实现神谕标记。评估六个前沿模型时，我们发现系统性、可预测的每个机制的盲点：部分比较器的携带规则（>1.2 意味着 >=1.3.0）使每个模型在 Cargo 上陷入困境（接近 60%），尽管标准的 PEP 440 前缀匹配是普遍适用的，但在零填充/发布后边缘案例中，GPT-5.1 崩溃（0/26），而 Claude 保持在 97-100%（在一个 67 项神谕验证集上验证）。Opus 显著优于所有其他模型，而 Sonnet 在表现上优于 OpenAI 模型（McNemar）。这些失败更像是激活/应用差距而非知识差距：注入规则或轻微的正确提示可以恢复大部分错误，而区间分解则无效，模型在同一规则的基本形式上已达到上限。作者分层分析发现没有统计学上显著的自我偏爱。由于该任务是可验证的，并且存在一个免费的、100% 正确的解析器，工具委托的准确率接近 100%：编码代理应将版本解析委托给解析器，而不是在脑中推理版本。

View on arXiv Download PDF AI Translation

cs.AI / 27 / 2609.11185

Can LLMs Follow Medical Expert Logic? A Benchmark for Hierarchical Logical Consistency in Risk-of-Bias Assessment

大型语言模型能遵循医学专家逻辑吗？风险偏倚评估中的层次逻辑一致性基准

Huang, Jiayu, Tang, Zichen, Ling, Qianhui, Kuang, Zemin, E, Haihong

Abstract

Evidence-based medicine demands strict logical consistency, yet current evaluations of large language models (LLMs) prioritize superficial label matching over genuine reasoning. We introduce LogiMed-RoB, a benchmark grounded in Cochrane Risk of Bias (RoB) 2.0 expert logic, comprising 860 randomized controlled trials (RCTs) and 14,820 queries. It evaluates models under the Hierarchical Logical Consistency (HLC) framework across four dimensions: Atomic Consistency, Domain Consistency, Aggregation Consistency, and Evidential Faithfulness. Experiments on 10 state-of-the-art LLMs reveal a catastrophic Error Compounding Effect: despite the top model reaching 98.88% Atomic Consistency, its end-to-end consistency collapses to 45.13%, with several open-weight architectures plummeting to nearly 0%. We further uncover a systematic evidence-reasoning gap: even when models retrieve high-quality evidence, they fail to deduce correct outcomes in 18.63-40.05% of cases, while Blind Guess Rates reach 48.28%. LogiMed-RoB demonstrates that high outcome accuracy can conceal critical reasoning flaws, underscoring the necessity of white-box logical verification for clinical deployment.

Chinese Translation

循证医学要求严格的逻辑一致性，但当前对大型语言模型（LLMs）的评估更注重表面的标签匹配而非真实的推理。我们引入了LogiMed-RoB，这是一个基于Cochrane风险偏倚（RoB）2.0专家逻辑的基准，包含860个随机对照试验（RCTs）和14,820个查询。它在四个维度下评估模型的层次逻辑一致性（HLC）：原子一致性、领域一致性、聚合一致性和证据忠实性。对10个最先进的LLMs进行的实验揭示了灾难性的错误累积效应：尽管最佳模型达到了98.88%的原子一致性，其端到端一致性却崩溃至45.13%，而一些开放权重架构的表现几乎降至0%。我们进一步发现了系统性的证据推理差距：即使模型检索到高质量的证据，它们在18.63%-40.05%的情况下仍未能推导出正确的结果，而盲目猜测率达到了48.28%。LogiMed-RoB表明，高结果准确性可能掩盖关键的推理缺陷，强调了临床应用中进行白盒逻辑验证的必要性。

View on arXiv Download PDF AI Translation

cs.AI / 28 / 2609.11190

Agentic Share-of-Search: A Multi-Agent AI System for Competitive Decision-Making in LLM-Mediated E-Commerce

代理搜索份额：一种用于 LLM 媒介电子商务中竞争决策的多智能体人工智能系统

Chowdhury, Spandan Ghose

Abstract

AI shopping assistants increasingly redirect consumer discovery, creating an urgent need for tools that support seller-side competitive decision-making. We present a multi-agent AI system that automates competitive visibility measurement and root cause diagnosis in LLM-mediated ecommerce. The system introduces Agentic Share-of-Search (ASoS) as the decision target, deploys query agents across leading AI platforms, and uses a ReAct-based diagnostic agent to recommend prioritized merchandising interventions. A 100-trial ablation study, presented as a feasibility evaluation of this prototype, shows the agent recovers the ablated signal in 39% of trials (95% CI: 30.0% - 48.8%, 5.5x over chance), rising to 63.9% among high-correlation ablations.

Chinese Translation

人工智能购物助手越来越多地引导消费者发现，迫切需要支持卖方竞争决策的工具。我们提出了一种多智能体人工智能系统，自动化测量竞争可见性和根本原因诊断，适用于 LLM 媒介的电子商务。该系统引入了代理搜索份额（Agentic Share-of-Search, ASoS）作为决策目标，部署查询代理于领先的人工智能平台，并使用基于 ReAct 的诊断代理推荐优先的商品干预措施。通过 100 次试验的消融研究，作为该原型的可行性评估，结果显示该代理在 39% 的试验中恢复了消融信号（95% 置信区间：30.0% - 48.8%，超过随机机会的 5.5 倍），在高相关消融中这一比例上升至 63.9%。

View on arXiv Download PDF AI Translation

cs.AI / 29 / 2609.11199

An AI-Powered Culturally Aware Chatbot for Stress Detection and Wellness Support among Pakistani University Students Using NLP and Machine Learning

基于人工智能的文化敏感型聊天机器人用于巴基斯坦大学生的压力检测与健康支持，采用自然语言处理和机器学习

Bashir, Muhammad Fahad, Afzal, Muhammad

Abstract

With the existing digital mental health tools specifically developed for Western settings, Pakistani students are exposed to a uniquely compounded stress situation in their university that includes academic, financial, familial, and relational stressors, which have become a serious concern for academic and psychological development of students in Pakistani universities. This paper introduces a new, AI-driven and culturally sensitive stress detection and wellness support system that is tailored to the context of Pakistani university students. The system is based on a machine learning model called Random Forest which is trained using a validated student stress data set of 1100 responses on 20 features from psychological, physiological, academic, environmental and social aspects, with an accuracy of 89.09% and a macro F1-score of 0.89, in three stress severity levels. The classification outputs are passed on to an open-source large language model through OpenRouter API, where an appropriately crafted system prompt, culturally aware, gives the model a conversation about wellness, in English, Urdu and Roman Urdu. The second most predictive stress factor in this population identified by feature importance analysis was teacher-student relationship, which is a culturally important stress factor highlighting the need for region-aware mental health systems. Future research will involve primary data collection from students at various academic levels of Pakistani Universities with the validated DASS-21 instrument focusing on the students who are moving from FSc to undergraduate studies, which is a time of being psychologically vulnerable which is under-researched.

Chinese Translation

现有的数字心理健康工具主要针对西方环境开发，巴基斯坦学生在大学中面临独特的复合压力情境，包括学业、经济、家庭和人际关系等压力源，这已成为巴基斯坦大学生学术和心理发展的严重问题。本文介绍了一种新的、基于人工智能的文化敏感型压力检测与健康支持系统，专门针对巴基斯坦大学生的背景进行定制。该系统基于一种名为随机森林（Random Forest）的机器学习模型，使用经过验证的学生压力数据集进行训练，该数据集包含1100个响应和20个来自心理、生理、学术、环境和社会方面的特征，准确率为89.09%，宏观F1-score为0.89，涵盖三种压力严重程度。分类输出通过OpenRouter API传递给一个开源的大型语言模型，在这里，经过精心设计的系统提示，具有文化意识，促使模型进行关于健康的对话，使用英语、乌尔都语和罗马乌尔都语。在该人群中，通过特征重要性分析识别出的第二大预测压力因素是师生关系，这一文化重要的压力因素突显了区域意识心理健康系统的必要性。未来的研究将涉及使用经过验证的DASS-21工具从巴基斯坦大学不同学术水平的学生中收集原始数据，重点关注从FSc转向本科阶段的学生，这是一个心理脆弱的时期，尚未得到充分研究。

View on arXiv Download PDF AI Translation

cs.AI / 30 / 2609.11206

CryptoL: Towards Scale Dominance and Physics Constraints Mitigation in Financial Multivariate Time Series Forecasting

CryptoL：朝着金融多变量时间序列预测中的规模主导性和物理约束缓解迈进

Taheri, Yalda, Heydari, Mohammad Hassan, Rasooli, Armon, Amirshahkarami, Maryam, Mahdavi, Mohammad Ebrahim, Karshenas, Hossein

Abstract

Cryptocurrency forecasting presents a distinctive combination of extreme cross-asset scale heterogeneity, non-stationary dynamics, and structural dependencies among Open, High, Low, and Close (OHLC) variables. We present CryptoL, a unified framework designed to address these challenges within multivariate time-series forecasting. CryptoL evaluates forecasting error in context-normalized coordinates within the RevIN pipeline, preventing inverse normalization from introducing an additional squared-scale weighting into the MSE objective. We formally characterize this effect through the empirical risk and parameter-gradient geometry, establishing the conditions under which large-scale assets can disproportionately influence shared-model optimization. Beyond loss-space normalization, CryptoL examines channel-independent and channel-dependent normalization for OHLC data, showing that a shared channel-dependent affine transformation preserves candle-order relations that independent channel transformations need not preserve. The framework further incorporates scale-adaptive numerical stabilization to reduce distortions caused by a fixed normalization constant across assets spanning many orders of magnitude, together with a soft feasibility loss that penalizes violations of the defining OHLC inequalities. Experiments across heterogeneous cryptocurrency assets evaluate these components through controlled ablations and demonstrate improvements in forecasting accuracy, training stability, and the frequency of financially valid OHLC predictions relative to the considered baselines. CryptoL therefore provides an integrated approach to scale-balanced optimization, structure-preserving normalization, numerical stabilization, and constraint-aware cryptocurrency forecasting.

Chinese Translation

加密货币预测呈现出极端跨资产规模异质性、非平稳动态和开放（Open）、最高（High）、最低（Low）及收盘（Close，OHLC）变量之间的结构依赖性等独特组合。我们提出了CryptoL，这是一个旨在解决多变量时间序列预测中这些挑战的统一框架。CryptoL在RevIN管道中以上下文归一化坐标评估预测误差，防止逆归一化引入额外的平方尺度加权到均方误差（MSE）目标中。我们通过经验风险和参数梯度几何形式化地描述了这一效应，建立了大规模资产可能不成比例地影响共享模型优化的条件。除了损失空间归一化，CryptoL还考察了OHLC数据的通道独立和通道依赖归一化，显示共享通道依赖的仿射变换能够保持蜡烛图顺序关系，而独立通道变换则不一定能保持。该框架进一步结合了规模自适应数值稳定化，以减少由于跨越多个数量级的资产而导致的固定归一化常数引起的失真，同时引入了软可行性损失，以惩罚违反定义的OHLC不等式的情况。针对异质加密货币资产的实验通过控制消融评估了这些组件，并展示了相对于考虑的基线在预测准确性、训练稳定性和财务有效OHLC预测频率方面的改进。因此，CryptoL提供了一种集成的方法，旨在实现规模平衡优化、结构保持归一化、数值稳定化和约束感知的加密货币预测。

View on arXiv Download PDF AI Translation

cs.AI / 31 / 2609.11231

A Voice-Interactive Multi-Agent System for Smart Operating Rooms: Architecture Design and Key Technologies

用于智能手术室的语音交互多智能体系统：架构设计与关键技术

Zhou, Tianxiang

Abstract

This paper presents SurgicalRoomAgent, a voice-interactive multi-agent system for smart operating rooms based on large language models (LLMs). The system achieves natural language understanding, device control, intraoperative recording, and surgical report generation through a layered architecture comprising a voice interaction pipeline (wake, ASR, turn detection, agent reasoning, TTS) and an agent core (skill registry, task planner, device manager). Three key technologies are investigated: (1) KV Cache prefix warming for low-latency inference, reducing recomputation overhead from approximately 500 ms to tens of milliseconds via byte-level Longest Common Prefix reuse; (2) streaming partial JSON parsing with early parallel task execution, reducing end-to-end latency by approximately 30%; and (3) progressive skill prompt disclosure, which dynamically filters system prompts based on user role, connected devices, and surgical phase to maximize information density within limited context windows. The system is implemented using the Qwen3-27B model with llama.cpp/sglang inference engines. Experimental analysis demonstrates effective operation within a 16,384-token context limit and multi-device parallel control response times meeting OR real-time requirements.

Chinese Translation

本文介绍了SurgicalRoomAgent，一个基于大型语言模型（LLMs）的语音交互多智能体系统，旨在用于智能手术室。该系统通过分层架构实现自然语言理解、设备控制、术中记录和手术报告生成，架构包括语音交互管道（唤醒、自动语音识别（ASR）、轮次检测、智能体推理、文本转语音（TTS））和智能体核心（技能注册、任务规划、设备管理）。研究了三项关键技术：（1）KV Cache前缀预热以实现低延迟推理，通过字节级最长公共前缀重用将重新计算开销从约500毫秒减少到数十毫秒；（2）流式部分JSON解析与提前并行任务执行，减少端到端延迟约30%；（3）渐进式技能提示披露，根据用户角色、连接设备和手术阶段动态过滤系统提示，以在有限的上下文窗口内最大化信息密度。该系统使用Qwen3-27B模型与llama.cpp/sglang推理引擎实现。实验分析表明，该系统在16,384个标记的上下文限制内有效运行，并且多设备并行控制响应时间满足手术室实时要求。

View on arXiv Download PDF AI Translation

cs.AI / 32 / 2609.11234

NovGauge: A Fine-Grained Benchmark for Diagnosing LLMs' Capability in Paper Novelty Assessment

NovGauge：一种用于诊断大型语言模型在论文新颖性评估能力的细粒度基准

Zhang, Guoqiang, Tan, Kexin, Zhang, Ming, Ju, Li, Jing, Wenqing, Yue, Zhonghan, Chen, Jiayi, Wu, Shiqiang, Liu, Shaofan, Zhang, Yue, Ying, Yuankai, Shi, Yang, Gui, Tao, Zhang, Qi, Huang, Xuanjing

Abstract

Large language models (LLMs) are increasingly used in peer review at major AI conferences, yet novelty remains a persistent weak point. Existing benchmarks assess novelty as a single holistic score, making it difficult to diagnose which dimension a model misjudges or whether its evidence is faithful. We present NovGauge, a human-anchored benchmark for fine-grained novelty assessment diagnosis. The benchmark contains 619 paper pairs and 50 multi-paper sets, drawn from two expert sources: ICLR reviewer overlap claims and survey co-citations. Instances are independently labeled along three dimensions: task, problem, and method, capturing application goals, technical challenges, and solution approaches. We propose a cascading diagnostic pipeline that verifies per-dimension correctness, evidence grounding, and logical support. Evaluation of 18 LLMs shows hallucination rates ranging from 0% to 39% across dimensions, and among non-hallucinated correct-positive judgments, over 70% cite evidence fails to logically support the stated reason. The best-performing model, GPT-5.5, achieves 43-72% Verified F1 across dimensions, while most models retain less than half of their raw F1 after faithfulness verification. These results suggest that current LLMs remain far from reliable scientific novelty assessment, particularly when correctness is conditioned on faithful evidence grounding.

Chinese Translation

大型语言模型（LLMs）在主要人工智能会议的同行评审中越来越多地被使用，但新颖性仍然是一个持续的弱点。现有基准将新颖性评估为一个单一的整体得分，这使得诊断模型在哪个维度上判断失误或其证据是否可靠变得困难。我们提出了NovGauge，这是一个以人为基础的细粒度新颖性评估诊断基准。该基准包含619对论文和50个多论文集，来源于两个专家来源：ICLR审稿人重叠声明和调查共同引用。实例在三个维度上独立标注：任务、问题和方法，捕捉应用目标、技术挑战和解决方案方法。我们提出了一种级联诊断流程，验证每个维度的正确性、证据的基础性和逻辑支持。对18个LLM的评估显示，不同维度的幻觉率在0%到39%之间，而在非幻觉的正确正向判断中，超过70%的引用证据未能在逻辑上支持所陈述的理由。表现最佳的模型GPT-5.5在各维度上实现了43-72%的验证F1，而大多数模型在忠实性验证后保留的原始F1不到一半。这些结果表明，当前的LLM在科学新颖性评估方面仍然远未可靠，特别是在正确性依赖于忠实证据基础时。

View on arXiv Download PDF AI Translation

cs.AI / 33 / 2609.11243

Sci-MMR: Benchmarking Multi-Step Evidence-Grounded Scientific Reasoning in Multimodal Agents

Sci-MMR：多模态智能体中多步骤证据基础科学推理的基准测试

Li, Jiaqiang, Yang, Yajie, Xi, Zhiheng, Chen, Jiadong, Zhou, Enyu, Jin, Senjie, Nan, Yang, Zhang, Jiazheng, Wang, Han, Li, Yanxin, Zhu, Dingwei, Deng, Bicheng, Wang, Yuhui, Zheng, Xiang, Zhang, Qi, Bai, Lei, Ma, Xingjun, Gui, Tao

Abstract

Autonomous research agents are increasingly expected to search the literature, analyze experimental evidence, and generate scientific hypotheses. These capabilities require multi-step evidence grounded reasoning that progressively acquires, integrates, and verifies evidence before reaching a conclusion. Existing multimodal benchmarks, however, largely evaluate final-answer accuracy, leaving open whether predictions are actually supported by traceable scientific evidence. We introduce Sci-MMR, a benchmark for multi-step evidence-grounded scientific reasoning built on structured argument graphs linking scientific claims, citation-grounded knowledge, visual evidence, and supporting regions. Sci-MMR comprises 235 multi-hop reasoning tasks spanning four scientific disciplines, with an average of nine figure panels per task. Evaluating eight frontier multimodal models, we find that answer accuracy consistently exceeds complete-evidence recovery rate by more than 20%, revealing a substantial gap that answer-only evaluation is structurally unable to capture. Through controlled interventions, we identify two fundamental bottlenecks. First, evidence acquisition: models struggle to extract complete structured evidence from scientific figures, accounting for 57.2% of failures. While cropping tools yield modest gains (+4.5 points), providing gold evidence improves accuracy by up to 37.0 points, indicating difficulty in assembling complete multi-region evidence. Second, evidence integration: models struggle to translate available evidence into correct conclusions, accounting for 31.8% of failures, while even with gold evidence the strongest model achieves only 69.1% accuracy on the hardest tasks. These findings indicate that current answer-centric benchmarks substantially overestimate the evidence-grounded reasoning capabilities of multimodal research agents

Chinese Translation

自主研究智能体越来越被期望能够搜索文献、分析实验证据并生成科学假设。这些能力需要多步骤的证据基础推理，逐步获取、整合和验证证据，最终得出结论。然而，现有的多模态基准主要评估最终答案的准确性，尚未解决预测是否真正得到可追溯的科学证据支持的问题。我们提出了Sci-MMR，这是一个基于结构化论证图的多步骤证据基础科学推理基准，连接科学主张、引用基础知识、视觉证据和支持区域。Sci-MMR包含235个跨越四个科学学科的多跳推理任务，每个任务平均有九个图形面板。通过评估八个前沿多模态模型，我们发现答案的准确性始终超过完整证据恢复率20%以上，揭示了仅依赖答案评估所无法捕捉的显著差距。通过控制干预，我们识别出两个基本瓶颈。首先，证据获取：模型在从科学图形中提取完整结构化证据方面存在困难，导致57.2%的失败。虽然裁剪工具带来了 modest 的提升（+4.5 分），但提供黄金证据的准确性提升可达37.0分，表明在组装完整的多区域证据方面存在困难。其次，证据整合：模型在将可用证据转化为正确结论方面存在困难，导致31.8%的失败，即使在拥有黄金证据的情况下，最强模型在最难任务上的准确率也仅为69.1%。这些发现表明，当前以答案为中心的基准显著高估了多模态研究智能体的证据基础推理能力。

View on arXiv Download PDF AI Translation

cs.AI / 34 / 2609.11262

AI-Powered Flare Combustion Efficiency Estimation

基于人工智能的火炬燃烧效率估计

Azam, Afeefa, Ganapathi, Iyyakutti Iyappan, Abdelhafez, Fares Ossama, Velayudhan, Divya, Habtie, Maregu Assefa, Karki, Hamad, Awadhi, Khalid Yousef Al, Werghi, Naoufel

Abstract

Achieving high combustion efficiency in flare stacks is crucial for adhering to regulatory standards and controlling the release of hydrocarbons into the environment. Traditional instruments like gas analyzers and hyperspectral cameras are expensive, fragile, and require frequent calibration, which makes them impractical for remote or budget constrained industrial sites. We propose an innovative solution that combines a lightweight vision-language encoder with a compact multi-layer perceptron to predict combustion efficiency directly from low-cost thermal video footage. The fully trained model is integrated into an easy-to-deploy graphical user interface. This interface overlays predicted combustion efficiency values on each video frame, displays real-time trends in combustion efficiency, shows the distribution of combustion efficiency across all frames in the video, and allows users to export CSV reports. Over a six-month period, the system achieved 99% uptime and required less than 15 minutes of maintenance per week.

Chinese Translation

在火炬塔中实现高燃烧效率对于遵守监管标准和控制碳氢化合物向环境的释放至关重要。传统仪器如气体分析仪和高光谱相机价格昂贵、易损坏且需要频繁校准，这使得它们在偏远或预算有限的工业场所中不够实用。我们提出了一种创新解决方案，将轻量级视觉-语言编码器与紧凑型多层感知器相结合，直接从低成本热视频资料中预测燃烧效率。经过充分训练的模型集成到一个易于部署的图形用户界面中。该界面在每个视频帧上叠加预测的燃烧效率值，显示燃烧效率的实时趋势，展示视频中所有帧的燃烧效率分布，并允许用户导出CSV报告。在六个月的时间里，该系统实现了99%的正常运行时间，每周维护时间少于15分钟。

View on arXiv Download PDF AI Translation

cs.AI / 35 / 2609.11277

Predicting Train Delays in Finland Using Machine Learning and Weather Data

利用机器学习和天气数据预测芬兰的列车延误

Borin, Vinicius Pozzobon, Sant'Ana, Jean Michel de Souza, Mahmood, Nurul Huda

Abstract

Reliable railway operations depend increasingly on real-time environmental intelligence delivered through wireless sensor infrastructures, a capability that 6G networks will substantially enhance through integrated sensing and edge computing. Adverse weather, particularly in Arctic regions with extreme temperatures and heavy precipitation, remains a leading cause of train delays, yet most prediction approaches rely on raw meteorological inputs without exploiting domain-informed feature engineering. This paper investigates machine learning for train delay prediction using the Finland Integrated Train-Weather (FI-TW) dataset, which fuses railway operational records with observations from the Finnish Meteorological Institute's nationwide sensor network of approximately 200 stations communicating over wireless links. We evaluate three feature configurations using XGBoost at Oulu central station (101,146 observations): full weather features, instant weather observations only, and derived weather category scenarios. The category-based approach, employing hierarchical classifications such as Blizzard, Heavy Snow, and Extreme Cold, achieved an R^2 of 0.78, root mean squared error of 8.5 minutes, and mean absolute error of 3.7 minutes, representing an 11% R^2 improvement and 10% error reduction over alternative configurations. These results demonstrate that compact, domain-informed features derived from sensor streams outperform raw meteorological observations, offering bandwidth-efficient representations suitable for edge deployment over current and emerging wireless infrastructures.

Chinese Translation

可靠的铁路运营日益依赖于通过无线传感器基础设施提供的实时环境智能，而6G网络将通过集成感知和边缘计算显著增强这一能力。恶劣天气，特别是在极端温度和强降水的北极地区，仍然是列车延误的主要原因，然而大多数预测方法依赖于原始气象输入，而未利用领域知识进行特征工程。本文研究了利用芬兰综合列车-天气（FI-TW）数据集进行列车延误预测的机器学习方法，该数据集将铁路运营记录与芬兰气象研究所约200个站点的全国传感器网络观测数据融合在一起，这些站点通过无线链接进行通信。我们在奥卢中央车站（101,146个观测值）评估了三种特征配置，分别是完整的天气特征、即时天气观测以及派生的天气类别场景。基于类别的方法采用了层次分类，如暴风雪、大雪和极寒，达到了0.78的R^2值、8.5分钟的均方根误差和3.7分钟的平均绝对误差，分别比其他配置提高了11%的R^2和减少了10%的误差。这些结果表明，从传感器流中提取的紧凑的领域知识特征优于原始气象观测，提供了适合于当前和新兴无线基础设施边缘部署的带宽高效表示。

View on arXiv Download PDF AI Translation

cs.AI / 36 / 2609.11281

Bio-inspired Learning and Decision-Making with Probabilistic In-Memory Computing Hardware: Part 1

基于生物启发的学习与决策：概率内存计算硬件的应用研究（第一部分）

Dalgaty, Thomas, Kawasaki, Eiji, de Prado, Miguel, Vyas, Devendra, Salvatori, Tommaso

Abstract

Learning and decision-making in animals are often modeled as Bayesian processes, where sensory evidence is integrated with prior beliefs to guide behavior in the face of uncertainty. But what are the inherent neural dynamics that give rise to this ability, and how could they be replicated in computing systems? This abstract discusses a biologically grounded framework in which noisy neural and synaptic dynamics perform inference and learning via stochastic sampling from an internal energy function, capturing uncertainty over latent states and model parameters through neural and synaptic variability, respectively. This enables approaches such as predictive coding networks to account for epistemic uncertainty via Markov chain Monte Carlo sampling. Drawing a parallel between intrinsic noise in biological systems and electrical noise in emerging probabilistic analogue memory technologies, we highlight how analogue in-memory computing hardware naturally emerges as the solution for massively scalable and energy-efficient probabilistic inference.

Chinese Translation

动物的学习与决策常被建模为贝叶斯过程，其中感官证据与先验信念相结合，以指导在不确定性面前的行为。但是什么样的内在神经动态促成了这种能力，它们又如何能够在计算系统中被复制？本文讨论了一个生物学基础的框架，在该框架中，嘈杂的神经和突触动态通过从内部能量函数的随机采样来执行推理和学习，分别通过神经和突触的变异性捕捉潜在状态和模型参数的不确定性。这使得诸如预测编码网络等方法能够通过马尔可夫链蒙特卡洛采样来考虑认知不确定性。我们将生物系统中的内在噪声与新兴概率类比内存技术中的电噪声进行比较，强调了类比内存计算硬件如何自然成为大规模可扩展和能效高的概率推理的解决方案。

View on arXiv Download PDF AI Translation

cs.AI / 37 / 2609.11282

When Does Text Inform? Benchmarking Information-Theoretic Metrics for Multimodal Time-Series Forecasting

文本何时提供信息？多模态时间序列预测的信息论指标基准测试

Andrews, Emma, Mengaldo, Gianmarco

Abstract

Multimodal forecasting models that combine time series with text annotations promise richer prediction through textual context, but how do we know whether a text annotation meaningfully contributes to the forecasters prediction? This is an information-theoretic question, but to evaluate whether information-theoretic metrics can reliably measure the predictive value an annotation provides, a ground truth benchmark is needed, and none currently exist. We create a synthetic time series signal with annotations in three categories: semantically correct, incorrect, and irrelevant. Because the data generation process is fully controlled, ground-truth information content is known exactly, enabling principled evaluation of six complementary mutual information estimators (KSG, MINE, InfoNCE, CCA, PID and V-information). We show that all six estimators identify correct annotations as most informative, and are able to audit the quality of mixed text corpora, choosing the annotations that result in the best downstream forecasting results without the need for model training. Our benchmark identifies limitations of each estimator, and these are validated on seven real-world datasets, which show how estimator performance differs on weak signals. Finally, we establish practical rules for implementing these metrics for annotation auditing and fusion selection.

Chinese Translation

结合时间序列与文本注释的多模态预测模型通过文本上下文承诺提供更丰富的预测，但我们如何知道文本注释是否对预测者的预测有实质性贡献？这是一个信息论问题，但为了评估信息论指标是否能够可靠地测量注释所提供的预测价值，需要一个真实的基准，而目前尚不存在。我们创建了一个合成时间序列信号，注释分为三类：语义正确、错误和无关。由于数据生成过程完全可控，真实信息内容是精确已知的，这使得对六种互信息估计器（KSG、MINE、InfoNCE、CCA、PID和V-information）的原则性评估成为可能。我们展示了所有六种估计器都将正确的注释识别为信息量最大的，并能够审计混合文本语料库的质量，选择出在不需要模型训练的情况下能够产生最佳下游预测结果的注释。我们的基准识别了每个估计器的局限性，并在七个真实世界数据集上进行了验证，显示了估计器在弱信号上的性能差异。最后，我们为实施这些指标以进行注释审计和融合选择建立了实用规则。

View on arXiv Download PDF AI Translation

cs.AI / 38 / 2609.11286

Generating a Consistent Enterprise: Synthesis and Reference-Free Evaluation of Multi-System Business Data

生成一致的企业：多系统商业数据的合成与无参考评估

Gruenbaum, Benjamin, Porat, Doron, Natanzon, Assaf, Zavida, Roy, Dinachi, Chen, Itzahary, Or, Niv, Omer

Abstract

Synthetic relational data is normally produced by a model trained on a real dataset, and its quality is measured as the distance to that dataset. This paper describes a generator that has no real dataset at either end. Given an industry, a company size, a business model, a set of business applications, and a random seed, it produces a complete fictional enterprise: a workforce, a customer base, sales deals, support tickets, recorded calls, chat messages, and documents, all consistent with one another. One entity graph is projected into the native formats of 66 business products, so the same customer appears in the CRM, the support desk, and the call system under one identity. Because no real counterpart exists, realism is built in from cited reference statistics and verified by reference-free measurement: a five-axis scorecard of 28 statistical checks, an adversarial detector that hunts for the marks of synthetic generation, and a set of soundness checks that include a classifier test against an independently shuffled copy of the data. Because these instruments existed before the generator was tuned, progress is measured under a fixed yardstick: over 23 generated companies, mean realism climbed from 60.3 to 99.1, the weakest company from 41.1 to 94.9, and the detector, which initially flagged 55.2% of all records, now flags none. The scores hold on a seed never used during development. A second generator builds relational databases from a list of business questions. It forces qualifying rows for each answerable question, adds controlled near misses, and computes exact labels from the finished tables. The generator runs as a hosted service at https://console.era.eon.io. A company built there to a specification is served through its simulators over MCP and REST, and the simulators are also published as container images for offline use

Chinese Translation

合成关系数据通常是由在真实数据集上训练的模型生成的，其质量通过与该数据集的距离来衡量。本文描述了一种在任一端都没有真实数据集的生成器。给定一个行业、公司规模、商业模型、一组商业应用程序和一个随机种子，它生成一个完整的虚构企业：一个劳动力、一个客户基础、销售交易、支持工单、录音通话、聊天消息和文档，所有这些都是相互一致的。一个实体图被投影到66个商业产品的本地格式中，因此同一客户在客户关系管理（CRM）、支持台和呼叫系统中以一个身份出现。由于不存在真实的对应物，现实性是通过引用的统计数据构建的，并通过无参考测量进行验证：一个包含28个统计检查的五轴评分卡，一个寻找合成生成痕迹的对抗检测器，以及一组包括对独立洗牌数据副本进行分类测试的合理性检查。因为这些工具在生成器调整之前就已存在，所以进展是在固定的标准下进行测量的：在生成的23家公司中，平均现实性从60.3上升到99.1，最弱的公司从41.1上升到94.9，而检测器最初标记了55.2%的所有记录，现在没有标记任何记录。这些分数在开发过程中从未使用过的种子上保持不变。第二个生成器根据一系列商业问题构建关系数据库。它为每个可回答的问题强制生成合格的行，添加受控的近似错误，并从完成的表中计算精确标签。该生成器作为托管服务运行在https://console.era.eon.io。根据规格在此构建的公司通过其模拟器通过MCP和REST提供服务，模拟器也作为容器镜像发布以供离线使用。

View on arXiv Download PDF AI Translation

cs.AI / 39 / 2609.11291

Off-Target Effects of Response-Style Alignment in a Korean 27B Language Model

韩国27B语言模型中响应风格一致性的非目标效应

Han, Hyojung

Abstract

We post-train Qwen3.8-27B for Korean response style -- verbosity, list and markdown usage, discourse structure and register -- and measure two behaviours the objective never targets: abstention on ambiguous social questions in KoBBQ, where the benchmark-correct answer is UNKNOWN, and unprompted disclosure in securities guidance. Both move, and the changes are expressed primarily through the model's emission policy: how often it answers and how much it says. Matched target-form controls show that answer propensity depends on the training target, not the prompt set or recipe alone. Holding prompts, recipe, data volume and serving fixed and changing only the target text, three style seeds give positive answer-rate point estimates (mean +0.82 pp) and three neutral seeds negative ones (mean -1.53 pp); the observed seed ranges do not overlap and the means differ by 2.34 pp. A length-matched arm lies between them, and a fourth arm that stays short while preserving hedging is unstable across seeds, so which feature of the form is responsible is unresolved. For absolute stereotyped exposure the decomposition into an answer-propensity term and a conditional-composition term is an algebraic identity, not a finding; its empirical content is where the movement went. Across the trained checkpoints the changes are dominated by answer propensity while the composition term stays small, and because that term is evaluated on treatment-dependent answered subsets we do not read it as evidence about latent preference. Two measurement results follow. A between-arm contrast in conditional stereotyped share does not identify a change in conditional content preference when answer status is treatment-dependent. And agreement between two rule detectors for the same construct runs from 0.44 to 0.99 depending on which checkpoint produced the text -- observable without any reference labels.

Chinese Translation

我们对Qwen3.8-27B进行了后训练，以适应韩国的响应风格——冗长性、列表和Markdown使用、话语结构和语域——并测量了两个目标未曾针对的行为：在KoBBQ中对模棱两可的社会问题的回避，其中基准正确答案为UNKNOWN，以及在证券指导中无提示的披露。这两种行为的变化主要通过模型的发射策略表现出来：它回答的频率和说的内容量。匹配目标形式的控制显示，回答倾向依赖于训练目标，而不仅仅是提示集或配方。保持提示、配方、数据量和服务不变，仅改变目标文本，三种风格种子给出了正的回答率点估计（均值+0.82个百分点），而三种中性种子则给出了负的估计（均值-1.53个百分点）；观察到的种子范围不重叠，均值相差2.34个百分点。一个长度匹配的臂位于它们之间，而一个在保持模糊的同时保持短小的第四个臂在不同种子间不稳定，因此哪个形式特征是负责的尚未解决。对于绝对刻板的曝光，分解为回答倾向项和条件组成项是一个代数恒等式，而不是一个发现；其经验内容在于运动的去向。在训练的检查点中，变化主要由回答倾向主导，而组成项保持较小，并且由于该项是在依赖于处理的回答子集上评估的，我们不将其视为关于潜在偏好的证据。随后有两个测量结果。一种在条件刻板分享中的臂间对比未能识别出当回答状态依赖于处理时条件内容偏好的变化。而对于同一构念的两个规则检测器之间的一致性在0.44到0.99之间变化，具体取决于哪个检查点生成了文本——这一点在没有任何参考标签的情况下可观察到。

View on arXiv Download PDF AI Translation

cs.AI / 40 / 2609.11294

Memory Compression for High-Fanout Agent Sandboxes

高扇出代理沙箱的内存压缩

Li, Mengming, XU, Ceyu, Zhang, Qijun, Yu, Jiangnan, Sun, Xiangfeng, Mai, Haohui, Xie, Zhiyao

Abstract

High-fanout agent workloads create a growing memory bottleneck because a single task may spawn many concurrent sandbox sessions. Yet these sandboxes are far from independent: they originate from a shared template and execute related trajectories, exposing substantial template-relative and cross-sandbox memory redundancy. Conventional memory compression is poorly matched to this setting in three fundamental dimensions: how to compress, because they fail to exploit similarity across non-identical sandbox pages; what to compress, because they control page-fault overhead through conservative page selection; and when to compress, because compression is either triggered by memory pressure or performed without awareness of agent execution phases. We present AgentZip, the first memory compression system designed specifically for AI-agent sandboxes. AgentZip introduces compression mechanisms that exploit both the template-relative and cross-sandbox redundancy. It broadens the compression scope to any page with a profitable representation and shifts overhead control from compression-time page selection to restore-time prefetching. It further aligns expensive compression with LLM waiting periods to avoid interfering with foreground tool execution. Across LLM training and inference workloads, AgentZip reduces sandbox-owned memory by up to 8.7x, compared with 2.1x for the Linux configuration. Restore prefetching and agent-execution-aware scheduling reduce the slowdown of aggressive compression from as high as 3.1x to 1.40x while retaining nearly all of its memory-saving benefit.

Chinese Translation

高扇出代理工作负载造成日益严重的内存瓶颈，因为单个任务可能会生成多个并发的沙箱会话。然而，这些沙箱并非独立：它们源自共享模板并执行相关轨迹，暴露出显著的模板相关和跨沙箱的内存冗余。传统的内存压缩在三个基本维度上与这种情况不匹配：如何压缩，因为它们未能利用非相同沙箱页面之间的相似性；压缩什么，因为它们通过保守的页面选择控制页面缺失开销；以及何时压缩，因为压缩要么在内存压力下触发，要么在不考虑代理执行阶段的情况下进行。我们提出了AgentZip，这是第一个专门为AI代理沙箱设计的内存压缩系统。AgentZip引入了压缩机制，利用了模板相关和跨沙箱的冗余。它扩大了压缩范围，涵盖任何具有盈利表示的页面，并将开销控制从压缩时的页面选择转移到恢复时的预取。它进一步将昂贵的压缩与大型语言模型（LLM）等待期对齐，以避免干扰前台工具的执行。在LLM训练和推理工作负载中，与Linux配置相比，AgentZip将沙箱拥有的内存减少了多达8.7倍，而Linux配置仅为2.1倍。恢复预取和代理执行感知调度将激进压缩的减速从高达3.1倍降低到1.40倍，同时保留几乎所有的内存节省收益。

View on arXiv Download PDF AI Translation

cs.AI / 41 / 2609.11315

Routing by Reasoning Need: Trajectory-Aware Decoding Control for Diffusion Vision-Language Models

基于推理需求的路由：面向轨迹的扩散视觉-语言模型解码控制

Liu, Yixiang, Xu, Zhongxing, Wang, Zhonghua, Tang, Xiaoying

Abstract

Diffusion vision-language models generate answers through iterative refinement, exposing intermediate answer trajectories that can be inspected and controlled at inference time. However, this controllability creates a reasoning-need mismatch, where a universal generation length is applied to questions with different reasoning demands. Visually closed questions may be harmed by continued refinement after a stable answer has formed, whereas reasoning-sensitive questions may be harmed by premature commitment. We formulate this problem as reasoning-budget mismatch and study it in LLaDA-V. Rather than choosing a universal generation length, our training-free controller routes each example to early commitment, baseline preservation, or reasoning-supportive decoding using trajectory signals from answer closure, commitment evidence, and representation revision pressure, without using ground-truth answers. Across answer-focused, mixed-reasoning, and CoT-sensitive benchmarks, routed control improves robustness over fixed long decoding, pure short decoding, and single-rule interventions. The gains are not explained by shorter outputs alone. Answer-closed examples often benefit from commitment, whereas CoT-sensitive examples require preserving or supporting intermediate reasoning. Taken together, these results suggest diffusion VLM decoding should route inference-time control by the state suggested by the observed trajectory instead of relying on a universal decoding length.

Chinese Translation

扩散视觉-语言模型通过迭代精炼生成答案，暴露出可在推理时检查和控制的中间答案轨迹。然而，这种可控性导致了推理需求的不匹配，即对具有不同推理需求的问题应用了统一的生成长度。对于视觉封闭的问题，在稳定答案形成后继续精炼可能会造成损害，而对推理敏感的问题则可能因过早承诺而受到影响。我们将此问题表述为推理预算不匹配，并在 LLaDA-V 中进行研究。我们的训练无关控制器并不选择统一的生成长度，而是利用来自答案闭合、承诺证据和表示修订压力的轨迹信号，将每个示例路由到早期承诺、基线保持或推理支持解码，而不使用真实答案。在以答案为重点的、混合推理的和 CoT（Chain of Thought）敏感的基准测试中，路由控制在固定长解码、纯短解码和单一规则干预上提高了鲁棒性。这些增益并不仅仅是由于输出较短。答案闭合的示例通常受益于承诺，而 CoT 敏感的示例则需要保持或支持中间推理。综合来看，这些结果表明，扩散视觉-语言模型的解码应根据观察到的轨迹所建议的状态来路由推理时的控制，而不是依赖于统一的解码长度。

View on arXiv Download PDF AI Translation

cs.AI / 42 / 2609.11318

Mr.LHDR: A Benchmark for Multimodal Real-World Long-Horizon Deep Research Agents

Mr.LHDR：多模态真实世界长时间深度研究代理的基准测试

Guo, Minghao, Cao, Meng, Zhao, Sui, Ning, Siyu, Wang, Xin, Zhao, Haoze, Yang, Jiaxuan, Hao, Haihong, Han, Mingfei, Rong, Shunlin, Wu, Haijun, Liang, Xiaodan, Chang, Xiaojun

Abstract

Deep research agents are increasingly capable of web search, tool use, multimodal evidence analysis, and information synthesis. However, existing benchmarks mainly evaluate medium-horizon exploration and rarely test whether agents can sustain long, dependency-heavy research processes. We introduce Mr.LHDR (Multimodal real-world Long-Horizon Deep Research), a benchmark for evaluating real-world deep research over long, irreducible chains of interdependent evidence across eight categories. Each question is constructed from a hidden Node-Relation graph and requires an average of 12.1 necessary intermediate conclusions with a mean dependency depth of 10.4 before reaching a short, unique, and verifiable answer. Questions incorporate multimodal evidence, including images, maps, PDFs, logos, charts, tables, and video frames, with at least one non-text element that changes the reasoning state. Mr.LHDR evaluates both final answers and the correctness of intermediate conclusions under annotated dependencies. We evaluate general models, deep research systems, and agent frameworks using Overall Accuracy (OA), Strict Accuracy (SA), Checklist Score (CS), and Dependency-Aware Checklist Score (DACS). Results show that even the strongest system achieves only 43.1% OA and 34.3% SA, indicating that final-answer accuracy substantially overestimates complete research success. Removing images reduces DACS by 12.6 points, demonstrating the importance of multimodal evidence, while SA consistently declines as reasoning chains become longer. These findings reveal sustained, dependency-consistent evidence integration, rather than isolated fact retrieval, as a key bottleneck for current deep research agents.

Chinese Translation

深度研究代理在网络搜索、工具使用、多模态证据分析和信息综合方面的能力日益增强。然而，现有基准主要评估中期探索，鲜有测试代理是否能够维持长时间、依赖性强的研究过程。我们引入了Mr.LHDR（多模态真实世界长时间深度研究），这是一个用于评估真实世界深度研究的基准，涵盖了八个类别的长时间、不可简化的相互依赖证据链。每个问题由一个隐藏的节点-关系图构建，平均需要12.1个必要的中间结论，平均依赖深度为10.4，才能得出一个简短、独特且可验证的答案。问题包含多模态证据，包括图像、地图、PDF、标志、图表、表格和视频帧，至少有一个非文本元素改变推理状态。Mr.LHDR评估最终答案及其在注释依赖下的中间结论的正确性。我们使用整体准确率（Overall Accuracy, OA）、严格准确率（Strict Accuracy, SA）、检查表得分（Checklist Score, CS）和依赖感知检查表得分（Dependency-Aware Checklist Score, DACS）来评估通用模型、深度研究系统和代理框架。结果显示，即使是最强的系统也仅达到43.1%的OA和34.3%的SA，这表明最终答案的准确性大大高估了完整研究的成功。去除图像使DACS降低了12.6分，显示了多模态证据的重要性，而SA随着推理链的延长而持续下降。这些发现揭示了持续的、依赖一致的证据整合，而非孤立的事实检索，是当前深度研究代理的关键瓶颈。

View on arXiv Download PDF AI Translation

cs.AI / 43 / 2609.11319

Magenta: Closing the Loop Between Mathematical Reasoning and Lean Verification

Magenta：数学推理与精益验证之间的闭环

Leang, Joshua Ong Jun, Li, Haonan, Zhao, Zheng, Shang, Xinyi, Li, Wenda, Liu, Zhengzhong, Xing, Erix, Cohen, Shay, Giunchiglia, Eleonora

Abstract

Most of mathematical knowledge has been communicated through so-called informal use of mathematics and natural language. With large language models (LLMs) being highly adept in using natural language, they achieve strong performance, yet not perfect, in informal mathematical reasoning. Restraining LLMs to informal reasoning misses out on the opportunity to use the discrete verification abilities that machines offer through machine-checkable proofs. In this paper, we bridge the gap between informal and formal reasoning by integrating Lean signals into the informal reasoning process. We introduce Magenta, a training-free agentic pipeline that, given only a natural-language problem, produces an answer, expresses it as a Lean 4 statement, and constructs a machine-checked proof. A statement judge verifies whether the formalisation preserves the original problem, while an error-attribution judge routes failed attempts either to mathematical re-derivation or local Lean repair. Magenta achieves 100% accuracy across all evaluated olympiad benchmarks, including AIME 2025, AIME 2026, and HMMT February 2026. When paired with the open-weight K2-Horizon-7B reasoner, it solves all six IMO 2026 problems. Our analysis shows that statement adjudication is essential for preventing false certificates and that feedback-guided correction outperforms independent resampling on difficult problems.

Chinese Translation

大多数数学知识通过所谓的非正式数学使用和自然语言进行传达。大型语言模型（LLMs）在使用自然语言方面表现出色，尽管在非正式数学推理中取得了强大的表现，但仍然不完美。将LLMs限制在非正式推理中错失了利用机器通过可机器检查的证明所提供的离散验证能力的机会。本文通过将Lean信号整合到非正式推理过程中，弥合了非正式与正式推理之间的差距。我们介绍了Magenta，一个无训练的智能管道，仅根据自然语言问题生成答案，将其表达为Lean 4语句，并构建机器检查的证明。语句评判者验证形式化是否保留了原始问题，而错误归因评判者则将失败的尝试引导至数学重新推导或局部Lean修复。Magenta在所有评估的奥林匹克基准测试中实现了100%的准确率，包括AIME 2025、AIME 2026和HMMT 2026年2月。当与开放权重的K2-Horizon-7B推理器配对时，它解决了所有六个IMO 2026问题。我们的分析表明，语句裁决对于防止虚假证书至关重要，而反馈引导的修正在困难问题上优于独立重采样。

View on arXiv Download PDF AI Translation

cs.AI / 44 / 2609.11321

AI Exposure and AI Resilience: A Two-Dimensional Assessment Framework for Software and Software-Based Business Model

人工智能暴露与人工智能韧性：软件及基于软件的商业模型的二维评估框架

Mandl, Paul Darius, Mandl, Peter, Häusl, Martin

Abstract

Artificial intelligence is changing both software production and the economics of software-based business models. Classical technology due diligence mainly examines technical properties such as architecture, scalability, and technical debt. These criteria do not fully capture how AI can affect a company's value proposition, competitive position, margins, or access to customers. This paper develops Artificial Intelligence Exposure and Resilience (AI-ER) as a two-dimensional assessment framework. AI exposure describes the pressure for change that AI creates for a business model. AI resilience describes the company's ability to absorb that pressure, adapt to changed conditions, and use AI in an economically viable way. Metrics for both dimensions are derived from current AI capabilities, their deployment conditions, and relevant research on business models and organizational adaptability. The model keeps exposure and resilience separate and adds an explicit assessment of evidence quality and confidence. It can be applied first with public information and later refined with internal evidence. The result is a traceable company profile that supports comparison without concealing uncertainty in the underlying evidence. The paper also specifies an initial score logic and a procedure for empirical validation.

Chinese Translation

人工智能正在改变软件生产和基于软件的商业模型的经济学。传统的技术尽职调查主要考察技术属性，如架构、可扩展性和技术债务。这些标准并未充分反映人工智能如何影响公司的价值主张、竞争地位、利润率或客户获取能力。本文开发了人工智能暴露与韧性（AI-ER）作为一个二维评估框架。人工智能暴露描述了人工智能对商业模型所产生的变革压力。人工智能韧性则描述了公司吸收这种压力、适应变化条件并以经济可行的方式利用人工智能的能力。两个维度的指标来源于当前的人工智能能力、其部署条件以及有关商业模型和组织适应性的相关研究。该模型将暴露和韧性分开，并增加了对证据质量和信心的明确评估。它可以首先应用于公共信息，随后通过内部证据进行细化。最终结果是一个可追溯的公司档案，支持比较而不掩盖基础证据中的不确定性。本文还指定了初步的评分逻辑和实证验证程序。

View on arXiv Download PDF AI Translation

cs.AI / 45 / 2609.11341

Exploring Diffusion Transformers for Cross-Modal Augmentation in Multimodal Brain State Decoding

探索扩散变换器在多模态脑状态解码中的跨模态增强应用

Wang, Ziwei, He, Xingyi, Wang, Hongbin, Jia, Tianwang, Fang, Bohan, Wu, Dongrui

Abstract

Multimodal brain state decoding has largely focused on fusing paired modalities for prediction, but has rarely explored how their correspondence can be further exploited to enrich training data and improve multimodal representation learning. To address this gap, we propose CoMA-DiT, a bidirectional cross-modal Diffusion Transformer for latent augmentation that treats paired modalities as sources of mutual generative supervision rather than merely as inputs to be fused. CoMA-DiT conditions velocity prediction on the paired modality through cross-modal attention and adaptively injects the resulting variation via a reliability-gated residual mechanism. Experiments on multimodal auditory attention decoding and emotion recognition showed that CoMA-DiT consistently outperformed 20 representative baselines, achieving absolute gains of 4.28% and 6.70% in accuracy and macro-F1 over the no-augmentation baseline, respectively. Extensive ablation, sensitivity, visualization, and interpretability analyses further demonstrated its robustness, generalizability, and ability to capture functionally relevant cross-modal interactions. These findings support a broader view of multimodal learning: Paired modalities can serve not only as inputs for fusion but also as supervision sources that augment one another.

Chinese Translation

多模态脑状态解码主要集中在融合配对模态以进行预测，但很少探索它们之间的对应关系如何进一步被利用以丰富训练数据并改善多模态表示学习。为了解决这一问题，我们提出了CoMA-DiT，一种双向跨模态扩散变换器，用于潜在增强，它将配对模态视为相互生成监督的来源，而不仅仅是需要融合的输入。CoMA-DiT通过跨模态注意机制对配对模态进行速度预测，并通过可靠性门控残差机制自适应地注入生成的变化。在多模态听觉注意解码和情感识别的实验中，CoMA-DiT始终优于20个代表性基线，分别在准确率和宏观F1上实现了相对于无增强基线的绝对增益4.28%和6.70%。广泛的消融、敏感性、可视化和可解释性分析进一步证明了其稳健性、可推广性以及捕捉功能相关跨模态交互的能力。这些发现支持了对多模态学习的更广泛看法：配对模态不仅可以作为融合的输入，还可以作为相互增强的监督来源。

View on arXiv Download PDF AI Translation

cs.AI / 46 / 2609.11365

Portable Semantics, Private Dialects: Reuse and Negative Transfer in Latent Communication Between Language-Model Cells

可携带的语义，私密的方言：语言模型单元间潜在通信中的重用与负迁移

Marincat, Narcis

Abstract

In shared-genome language-model societies, restricted evidence visibility favors reusable, value-indexed latent packet interfaces, whereas the sole high-performing globally visible model in the parent study learned an episode-entangled code. This companion study asks whether independently trained societies share one packet language, where strict zero-shot transfer fails, and whether inherited interface state helps or harms later learning. First, a leakage-controlled causal interoperability audit over all 30 ordered pairs of six independently trained restricted societies -- under sealed held-out structure and a preregistered raw/orthogonal/linear/nonlinear alignment ladder -- shows the six semantically similar interfaces do not form one raw language: one same-initialization pair is exactly interoperable in both directions, a second shows asymmetric partial compatibility, and all 26 cross-initialization directions fail every frozen alignment rung. Second, within the tested decomposition and a single sealed source formulation, a source-span control localizes strict zero-shot failure to interpretation and execution of the new operator instructions. Third, in a matched adaptation factorial, the globally trained communication interface acts as a severe negative-transfer prior: reinitializing only the packet reader, writer, and mouth raises final depth-three accuracy from 0.169 to 0.857. Fourth, across two restricted checkpoints and two independently frozen target streams each, inherited interfaces never exceeded fresh-interface controls by the preregistered 0.10 margin. All primary conclusions are bounded to a near-transfer 17-state setting; the negative-transfer factorial concerns one globally visible parent-cohort checkpoint, while an appendix adds a post hoc tagged-global twin case study.

Chinese Translation

在共享基因组语言模型社群中，受限的证据可见性有利于可重用的、价值索引的潜在数据包接口，而在母研究中唯一表现优异的全球可见模型则学习了一个情节纠缠的代码。本伴随研究探讨独立训练的社群是否共享一种数据包语言，在严格的零-shot 迁移失败的情况下，继承的接口状态是否有助于或妨碍后续学习。首先，对六个独立训练的受限社群的30对有序组合进行泄漏控制的因果互操作性审计——在密封的保留结构和预注册的原始/正交/线性/非线性对齐梯度下——显示这六个语义相似的接口并未形成一种原始语言：一个相同初始化的对在两个方向上完全互操作，第二个显示出不对称的部分兼容性，而所有26个跨初始化方向在每个冻结对齐梯度上均失败。其次，在测试的分解和单一密封源公式中，源跨度控制将严格的零-shot 失败局限于新操作指令的解释和执行。第三，在匹配适应的因子实验中，全球训练的通信接口作为严重的负迁移先验：仅重新初始化数据包读取器、写入器和发声器，最终深度三的准确率从0.169提升至0.857。第四，在两个受限检查点和两个独立冻结的目标流中，继承的接口从未超过预注册的0.10边际的新接口控制。所有主要结论都局限于近迁移的17状态设置；负迁移因子实验涉及一个全球可见的母队列检查点，而附录则增加了一个事后标记的全球双胞胎案例研究。

View on arXiv Download PDF AI Translation

cs.AI / 47 / 2609.11372

RAMamba-Net: A Reliability-Aware and Mamba-Based Multimodal Fusion Network for Auditory Attention Detection

RAMamba-Net：一种基于Mamba的可靠性意识多模态融合网络用于听觉注意力检测

He, Xingyi, Wang, Ziwei, Wu, Dongrui

Abstract

Auditory attention decoding (AAD) identifies the attended speaker from physiological signals, supporting neuro-steered hearing devices and natural human-machine interaction. Electroencephalography (EEG) is the dominant modality for AAD but provides incomplete evidence in naturalistic audio-visual scenes, motivating EEG and electrooculography (EOG) fusion. Existing approaches remain limited by weak cross-modal interaction, inefficient temporal modeling, and low robustness to sample variations. To address the limitations, we propose RAMamba-Net, a reliability-aware Mamba-based multimodal fusion network for AAD. RAMamba-Net employs a Mamba-enhanced band-aware convolutional Transformer to capture band-specific EEG patterns and long-range temporal dynamics. A dual-branch temporal-spatial encoder models EOG temporal and inter-channel dependencies. Cross-modal attention enables explicit modality interaction. Then, a reliability-aware module is introduced to estimate sample-wise modality weights for feature and prediction consistency, thereby enhancing multimodal fusion. Experiments on two AAD benchmarks demonstrate that RAMamba-Net effectively exploits complementary EEG-EOG information, yielding accuracy gains of 5.76% over unimodal baselines, together with more robust decoding and discriminative representations. Further analyses show that explicit cross-modal interaction improves multimodal alignment, while the reliability-aware module suppresses unreliable modality evidence and is robust to signal perturbation and parameter variation.

Chinese Translation

听觉注意力解码（AAD）通过生理信号识别被关注的说话者，支持神经引导的听力设备和自然的人机交互。脑电图（EEG）是AAD的主要模态，但在自然音视频场景中提供的证据不够完整，这促使了EEG与眼电图（EOG）的融合。现有方法受到跨模态交互弱、时间建模效率低和对样本变化鲁棒性差的限制。为了解决这些局限性，我们提出了RAMamba-Net，一种基于Mamba的可靠性意识多模态融合网络用于AAD。RAMamba-Net采用增强的Mamba带宽感知卷积Transformer，以捕捉特定频带的EEG模式和长距离时间动态。双分支时空编码器建模EOG的时间和通道间依赖关系。跨模态注意力使得显式模态交互成为可能。然后，引入一个可靠性意识模块来估计样本级模态权重，以确保特征和预测的一致性，从而增强多模态融合。在两个AAD基准上的实验表明，RAMamba-Net有效利用互补的EEG-EOG信息，较单模态基线提高了5.76%的准确率，并且具有更强的解码能力和判别表示。进一步分析表明，显式的跨模态交互改善了多模态对齐，而可靠性意识模块抑制了不可靠的模态证据，并对信号扰动和参数变化具有鲁棒性。

View on arXiv Download PDF AI Translation

cs.AI / 48 / 2609.11393

Beyond Confidence: Stability-Aware Test-Time Adaptation for LLM Reasoning

超越信心：考虑稳定性的测试时适应方法用于大型语言模型推理

Gu, Bincheng, Gao, Min, Wang, Zongwei, Bai, Yibing, He, Yulan, Yu, Junliang

Abstract

Test-time adaptation has emerged as a lightweight alternative to costly post-training for improving the reasoning capabilities of Large Language Models (LLMs) on downstream tasks. Predictive entropy provides a model-derived signal for such adaptation, guiding models toward higher-confidence reasoning states without external verifiers or reward models. However, higher confidence does not necessarily imply correctness, as LLMs may remain highly confident along incorrect reasoning trajectories. We observe that high-confidence reasoning is more likely to be correct when confidence remains stable under local perturbations. Based on this observation, we propose Test-Time Adaptation via Stability-Aware Confidence Optimization (TASCO), a framework that incorporates local stability into confidence-based test-time adaptation while keeping the LLM frozen. TASCO operationalizes local stability by optimizing a lightweight task-level prefix under two alternative perturbation strategies: Random Perturbation promotes distributional stability across trajectories induced by nearby perturbed prefixes, whereas Sharpness-Aware Perturbation targets worst-case local sensitivity. Experiments demonstrate that TASCO improves reasoning accuracy and token efficiency across diverse LLMs and reasoning benchmarks, while behavioral analyses show that it maintains stable confidence under local perturbations without prematurely concentrating the model's predictive distribution.

Chinese Translation

测试时适应已成为一种轻量级的替代方案，用于改善大型语言模型（LLMs）在下游任务中的推理能力，取代成本高昂的后训练。预测熵提供了一种模型衍生的信号，用于这种适应，指导模型朝向更高信心的推理状态，而无需外部验证者或奖励模型。然而，更高的信心并不一定意味着正确性，因为LLMs可能在错误的推理轨迹上仍然保持高度自信。我们观察到，当信心在局部扰动下保持稳定时，高信心的推理更有可能是正确的。基于这一观察，我们提出了通过稳定性感知信心优化的测试时适应（TASCO）框架，该框架在保持LLM不变的情况下，将局部稳定性纳入基于信心的测试时适应。TASCO通过在两种替代扰动策略下优化轻量级任务级前缀来实现局部稳定性：随机扰动促进由附近扰动前缀引起的轨迹间的分布稳定性，而敏感性感知扰动则针对最坏情况下的局部敏感性。实验表明，TASCO在不同的LLMs和推理基准上提高了推理准确性和令牌效率，同时行为分析显示它在局部扰动下保持了稳定的信心，而不会过早集中模型的预测分布。

View on arXiv Download PDF AI Translation

cs.AI / 49 / 2609.11403

From Queries to Narratives: Cultural Heritage Data Stories for Knowledge Graph Exploration and Quality Assessment

从查询到叙事：用于知识图谱探索和质量评估的文化遗产数据故事

Tietz, Tabea, Schrade, Torsten, Posthumus, Etienne, Söhn, Linnaea, Steller, Jonatan Jalle, Waitelonis, Jörg, Sack, Harald

Abstract

Cultural-heritage KGs such as the NFDI4Culture-KG contain millions of triples about artworks, music, inscriptions, historical events, and the people and places connected to them. For many users, however, discovering this knowledge can be difficult. While SPARQL can be learned, writing meaningful queries first requires an in-depth understanding of the graph's data model, an investment many domain researchers and practitioners are unwilling to make. Even with existing user interfaces, a starting point and some guidance are usually needed, because the data contained in the graph is highly specialized, heterogeneous, and constantly growing, making it challenging to know what it contains or which questions it can answer. In this paper, we present data stories as a way not only to lower this barrier, but also to turn exploration into data-quality assessment, and thus combine accessible querying with the discovery of issues that remain hidden in aggregate statistics. In this contribution, a data story is understood as a narrative document that integrates explanatory text and images with executable SPARQL queries and their visualized results. It is described how they are authored against the graph and how they serve several purposes: guiding users through an unfamiliar graph, creating reproducible narratives, and surfacing data-quality issues previously hidden in aggregate statistics. The authoring platform LODEON including its Sparnatural and AI-supported authoring assistants is introduced as a proof-of-concept. Within the authoring environment, every claim made about the data can be backed by an explicit query, making these narratives transparent and reproducible. This paper also reflects on lessons learned from hands-on seminars and workshops. Early experience suggests that such data stories make cultural-heritage knowledge graphs more accessible for both exploration and quality assessment.

Chinese Translation

文化遗产知识图谱（KG），如NFDI4Culture-KG，包含关于艺术品、音乐、铭文、历史事件以及与之相关的人和地点的数百万个三元组。然而，对于许多用户而言，发现这些知识可能是困难的。虽然SPARQL可以学习，但编写有意义的查询首先需要对图谱的数据模型有深入的理解，这对于许多领域研究者和从业者来说是一项不愿意投入的工作。即使有现有的用户界面，通常也需要一个起点和一些指导，因为图谱中包含的数据高度专业化、异构且不断增长，这使得了解其内容或能够回答哪些问题变得具有挑战性。在本文中，我们提出数据故事作为一种不仅降低这一障碍的方法，同时也将探索转变为数据质量评估，从而将可访问的查询与发现隐藏在汇总统计中的问题相结合。在这项贡献中，数据故事被理解为一种叙事文档，它将解释性文本和图像与可执行的SPARQL查询及其可视化结果相结合。我们描述了如何针对图谱创作这些故事，以及它们如何服务于多个目的：引导用户通过一个不熟悉的图谱、创建可重复的叙事，以及揭示之前隐藏在汇总统计中的数据质量问题。作为概念验证，介绍了创作平台LODEON及其Sparnatural和AI支持的创作助手。在创作环境中，关于数据的每个声明都可以通过明确的查询进行支持，使这些叙事透明且可重复。本文还反思了从实践研讨会和工作坊中获得的经验教训。早期的经验表明，这样的数据故事使文化遗产知识图谱在探索和质量评估方面变得更加可访问。

View on arXiv Download PDF AI Translation

cs.AI / 50 / 2609.11431

LLMs as Post-hoc Auditors of Physiological Plausibility in Symbolic Regression: A Clinician-Evaluated Case Study

大型语言模型作为符号回归中生理合理性的事后审计工具：临床医生评估的案例研究

López-Varela, Jorge, Hidalgo, J. Ignacio, Muñoz, José-Manuel, Costilla-Reyes, Omar, Maqueda, Esther, Moreno-Fernandez, Jesus, González-Vidal, Tomás, Velasco, J. Manuel, Garnica, Oscar

Abstract

Genetic Programming and its variants, such as grammatical evolution, are widely used in Symbolic Regression to derive mathematical expressions from multivariate data. In addition to predictive accuracy, models are appreciated for their potential to provide interpretability, offering explicit equations that relate input variables to outcomes. However, achieving interpretability and plausibility remains challenging, as evolved models may be complex or scientifically inconsistent. In this study, we explore whether Large Language Models, can assist in improving the explainability of Symbolic Regression models generated by evolutionary computation methods. Building upon our previous work on estimating body fat percentage using grammar-based Genetic Programming , we investigate the use of LLMs as post-processing tools to analyze and rank evolved expressions according to their interpretability and medical plausibility. Four symbolic expressions are analysed by three LLMs over three repeated runs, and the resulting interpretations and rankings are assessed by a panel of three clinicians. Across the three LLMs, comparative model-ranking outputs received more favorable clinician assessments than isolated term-level interpretations. However, the LLMs also produced physiologically and mathematically questionable explanations, indicating that they are better suited to comparative auditing under expert oversight than to autonomous validation.\blfootnote{The present work is an extended version of a paper submitted into a journal.

Chinese Translation

遗传编程及其变体，如语法演化，广泛应用于符号回归，以从多变量数据中推导数学表达式。除了预测准确性外，这些模型因其提供可解释性的潜力而受到重视，能够提供明确的方程式，将输入变量与结果关联起来。然而，实现可解释性和合理性仍然具有挑战性，因为演化出的模型可能复杂或在科学上不一致。在本研究中，我们探讨大型语言模型（LLMs）是否可以帮助提高由进化计算方法生成的符号回归模型的可解释性。在我们之前关于使用基于语法的遗传编程估计体脂百分比的研究基础上，我们调查了将LLMs作为后处理工具，分析和排名演化表达式的可解释性和医学合理性。通过三次重复运行，三种LLMs对四个符号表达式进行了分析，结果的解释和排名由三位临床医生评估。在三种LLMs中，比较模型排名的输出比孤立的术语级解释获得了更有利的临床评估。然而，LLMs也产生了生理和数学上值得怀疑的解释，表明它们在专家监督下更适合进行比较审计，而非自主验证。

View on arXiv Download PDF AI Translation

cs.AI / 51 / 2609.11446

Calibration-Aware Uncertainty Cascades for Efficient Heterogeneous Model Collaboration

考虑校准的不确定性级联以实现高效的异构模型协作

Zhang, Yilin, Jiang, Han, Xu, Cai, Liu, Ying, Zhao, Wei

Abstract

Heterogeneous model collaboration seeks to exploit the complementary strengths of different models to balance predictive performance and inference cost. Existing approaches typically rely either on trained routers, which tie routing decisions to a fixed task and model pool, or on raw-confidence cascades, whose thresholds lack consistent reliability semantics across heterogeneous models. Consequently, these approaches adapt poorly to changing model pools and deployment budgets. We propose Calibration-Aware Uncertainty Cascades (CAUC), a simple post-hoc framework that independently calibrates each model's confidence and selects deployment policies using validation data. The resulting calibrated confidence scores establish a common reliability scale for accepting an early prediction, invoking a stronger model, or selectively combining model outputs. This unified decision criterion decouples deployment policies from any particular model pool or operating budget. We further show theoretically that calibration gives confidence thresholds an explicit selective-risk interpretation, whereas uncalibrated scores offer no comparable reliability guarantee. Extensive experiments demonstrate that, across six language benchmarks, CAUC achieves an average relative accuracy improvement of 1.9% over strong-model-only inference while avoiding approximately 47% of strong-model calls. On image classification benchmarks, it maintains or improves predictive performance while reducing measured GFLOPs by up to 57%.

Chinese Translation

异构模型协作旨在利用不同模型的互补优势，以平衡预测性能和推理成本。现有方法通常依赖于训练好的路由器，这些路由器将路由决策与固定的任务和模型池绑定，或者依赖于原始置信度级联，其阈值在异构模型之间缺乏一致的可靠性语义。因此，这些方法对变化的模型池和部署预算适应性较差。我们提出了考虑校准的不确定性级联（Calibration-Aware Uncertainty Cascades, CAUC），这是一个简单的事后框架，独立校准每个模型的置信度，并使用验证数据选择部署策略。生成的校准置信度分数建立了一个共同的可靠性尺度，用于接受早期预测、调用更强的模型或选择性地组合模型输出。这个统一的决策标准将部署策略与任何特定的模型池或操作预算解耦。我们进一步理论证明，校准赋予置信度阈值明确的选择性风险解释，而未校准的分数则没有可比的可靠性保证。大量实验表明，在六个语言基准测试中，CAUC在强模型推理的基础上实现了平均相对准确率提高1.9%，同时避免了约47%的强模型调用。在图像分类基准测试中，它保持或提高了预测性能，同时将测量的GFLOPs降低了多达57%。

View on arXiv Download PDF AI Translation

cs.AI / 52 / 2609.11452

RouteRepair: Instance-Level Failure Diagnosis and Targeted Repair in LLM-Based Automated Heuristic Design for Routing Optimization

RouteRepair：基于实例级故障诊断与针对性修复的LLM自动启发式设计在路由优化中的应用

Ji, Binghao, Huang, Di, Fang, Jiahui, Liu, Zhiyuan

Abstract

Efficient routing optimization is essential to freight transportation, urban logistics, and shared mobility, where high-quality heuristics are often required under limited computational budgets. Recent large language model (LLM)-based automated heuristic design methods can generate effective routing rules, but aggregate evaluation may mask recurrent failures on particular instance structures. To address this limitation, this study develops RouteRepair, which diagnoses parent-specific weaknesses from instance-level performance and applies targeted modifications to the corresponding heuristic components while protecting behavior that already performs well. Routing evidence, solver behavior, and program context are combined to define bounded repair objectives, and each intervention is validated through matched parent-child evaluation of failure recovery and collateral degradation. Experiments on the traveling salesman problem (TSP) and capacitated vehicle routing problem (CVRP) span constructive search, guided local search, and ant colony optimization. RouteRepair-GLS reduces the mean TSP optimality gap from 1.7476% to 0.7587%, while the constructive CVRP heuristic lowers average route cost by 1.91% relative to the savings heuristic; the generated ACO priors also outperform matched hand-designed priors. These results show that failure-aware, evidence-constrained refinement can improve routing heuristics on difficult instances while preserving performance on cases they already solve well.

Chinese Translation

高效的路由优化对于货运运输、城市物流和共享出行至关重要，在有限的计算预算下，通常需要高质量的启发式方法。近期基于大型语言模型（LLM）的自动启发式设计方法能够生成有效的路由规则，但整体评估可能掩盖特定实例结构上的重复故障。为了解决这一局限性，本研究开发了RouteRepair，它从实例级性能中诊断特定父实例的弱点，并对相应的启发式组件进行针对性修改，同时保护已经表现良好的行为。通过结合路由证据、求解器行为和程序上下文，定义了有限的修复目标，并通过匹配的父子评估验证每个干预措施在故障恢复和附带退化方面的有效性。在旅行推销员问题（TSP）和容量限制车辆路由问题（CVRP）上的实验涵盖了构造性搜索、引导局部搜索和蚁群优化。RouteRepair-GLS将TSP的平均最优性差距从1.7476%降低至0.7587%，而构造性CVRP启发式方法相较于节省启发式方法降低了平均路线成本1.91%；生成的ACO先验也优于匹配的手工设计先验。这些结果表明，关注故障的证据约束细化可以在困难实例上改善路由启发式，同时保持在已经解决良好的案例上的性能。

View on arXiv Download PDF AI Translation

cs.AI / 53 / 2609.11458

Flexible and Interpretable Accent Distance Measurements

灵活且可解释的口音距离测量

McGhee, Charles, Gales, Mark J. F., Knill, Kate M.

Abstract

Determining the differences between two speakers' accents is a fundamental task in linguistics and speech technology research. The methodology used to measure these differences depends on the specific research area. A phonetics researcher may demonstrate accent variation by comparing vowel formants in paired recordings of individual words. These results will be interpretable, but the recordings will be time-consuming to collect and may not be representative of connected speech. Accented Text-to-Speech (TTS) research has pushed towards using accent embeddings derived from accent classification tasks. These embeddings can be produced from any speech recording, but are not readily interpretable. In this paper, we demonstrate that articulatory representations created through articulatory inversion can be used as an interpretable basis for accent comparison and that optimal transport provides a framework for accent comparison across arbitrary recording types.

Chinese Translation

确定两个说话者口音之间的差异是语言学和语音技术研究中的一项基本任务。用于测量这些差异的方法依赖于具体的研究领域。语音学研究者可能通过比较单词对录音中的元音共振峰来展示口音变异。这些结果是可解释的，但录音的收集过程耗时且可能无法代表连贯的语音。口音文本到语音（Text-to-Speech, TTS）研究推动了使用从口音分类任务中派生的口音嵌入。这些嵌入可以从任何语音录音中生成，但不易解释。在本文中，我们展示了通过发音反演创建的发音表示可以作为口音比较的可解释基础，并且最优传输提供了一个跨任意录音类型进行口音比较的框架。

View on arXiv Download PDF AI Translation

cs.AI / 54 / 2609.11489

The Convention Gap: Towards Measuring Implicit Communication in Cooperative AI Evaluation

规范差距：朝着测量合作人工智能评估中的隐性沟通

Fukushima, Makoto, Xiong, Hua-Dong, Pari, Ehsan Moradi

Abstract

Cooperative AI agents are evaluated against other AIs, yet human cooperation relies on implicit conventions---shared protocols for reading meaning beyond the literal message---which AI-AI benchmarks may not capture. We propose the \emph{convention gap}, the difference between the failure probability predicted from the literal content of communication and the observed failure rate, as a metric of implicit communication. In the card game Hanabi, the finite deck and deterministic hint constraints make this posterior exactly computable. We replayed about 101,000 play actions from three public datasets of human-human (hanab.live), AI-AI (HOAD), and human-AI (HanabiData) games. The gap was +26.2 percentage points (pp) in human pairs, $-$0.7~pp in AI pairs, and +16.4~pp in human-AI pairs, and was concentrated on plays of cards that had received no hints (+46~pp in human pairs). Within human-AI play, the literal information available to humans was similar across the three AI partners (mean predicted failure 38--41\%), but human failure rates ranged from 14.4\% to 34.4\% and the gap from +24.1 to +6.2~pp; the partner eliciting the largest gap produced the fewest human failures. Game score carried different information: it depended on each corpus's roster composition, whereas the gap separated human from AI play at the agent level. As a known-answer check, Off-Belief Learning agents, whose convention content is controlled by construction, gave a gap of +1.6~pp at the convention-free level, rising monotonically to +21.7~pp. These results suggest that convention compatibility, rather than AI-AI performance, may predict an AI's effectiveness with human partners.

Chinese Translation

合作人工智能代理的评估是基于其他人工智能的表现，而人类合作依赖于隐性规范——共享的解读超越字面信息的协议——这可能是人工智能-人工智能基准无法捕捉到的。我们提出了 extit{规范差距}，即从沟通的字面内容预测的失败概率与观察到的失败率之间的差异，作为隐性沟通的度量。在卡牌游戏Hanabi中，有限的牌组和确定性的提示约束使得这一后验计算变得可行。我们重放了来自三个公共数据集的人类-人类（hanab.live）、人工智能-人工智能（HOAD）和人类-人工智能（HanabiData）游戏的约101,000个游戏动作。人类对局的差距为+26.2个百分点（pp），人工智能对局为-0.7 pp，人类-人工智能对局为+16.4 pp，并且集中在没有收到任何提示的出牌上（人类对局为+46 pp）。在人类-人工智能对局中，提供给人类的字面信息在三个人工智能伙伴之间相似（平均预测失败率为38-41%），但人类的失败率范围从14.4%到34.4%，差距从+24.1到+6.2 pp；造成最大差距的伙伴产生了最少的人类失败。游戏得分传达了不同的信息：它依赖于每个语料库的角色组成，而差距则在代理级别上区分了人类与人工智能的表现。作为已知答案的检查，Off-Belief Learning代理，其规范内容由构造控制，在无规范水平上给出了+1.6 pp的差距，单调上升至+21.7 pp。这些结果表明，规范兼容性而非人工智能-人工智能的表现，可能更能预测人工智能在与人类伙伴合作时的有效性。

View on arXiv Download PDF AI Translation

cs.AI / 55 / 2609.11490

Published Unlearning Numbers Move Per Checkpoint, and Not Because the Removed Data Survives: An Audit of 263 Released Batch-Normalized Checkpoints

发布的去学习数字在每个检查点移动，而不是因为被移除的数据存活：对263个发布的批量归一化检查点的审计

Li, Junlong Shen Xingyu

Abstract

An unlearning audit reads its verdict off numbers that an unlearned model and its retrained reference each publish, and both also ship batch-normalization statistics that no gradient step wrote and no release records. Refitting them on kept data at bit-identical weights moves 47 of 221 released checkpoints past the spread their own release's seeds show, several inside a method whose average does not move: what moves is the checkpoint's property, not its method's. What does the moving is not the removed data surviving in the state: exchanging kept records for removed ones inside a fixed fitting pool moves a published cell by almost nothing, while how far a checkpoint's shipped state has drifted from any refit does track it. The consequence for a published decision is real but narrow: twelve verdicts cross, four clear a measured recalibration budget, two clear it on every replicate, and a population we trained and sited near its own criterion yields none. A release should therefore name the fitting convention beside the number, on the batch-normalized vision models where this channel exists.

Chinese Translation

去学习审计通过未学习模型和其重新训练的参考模型各自发布的数字来得出结论，同时两者还提供了没有梯度步骤写入和没有发布记录的批量归一化统计数据。在保持数据的情况下，以比特相同的权重重新拟合这些统计数据，使得221个发布的检查点中有47个超出了其自身发布种子所显示的分布，其中几个位于一个方法内部，其平均值并未发生变化：移动的是检查点的属性，而不是其方法的属性。移动的并不是在状态中存活的被移除数据：在固定拟合池中用保留记录替换被移除记录几乎不会移动发布的单元，而检查点的发布状态与任何重新拟合的漂移程度则确实与之相关。对于发布的决策，其后果是真实但狭窄的：十二个裁决交叉，四个清晰地超出了测量的重新校准预算，两个在每个复制中都超出了预算，而我们训练并放置在其自身标准附近的人群则没有产生任何结果。因此，发布应在数字旁边标明拟合规范，特别是在存在此通道的批量归一化视觉模型中。

View on arXiv Download PDF AI Translation

cs.AI / 56 / 2609.11493

From Document Silos to Process Intelligence: A Multi-Layer Knowledge Graph for CMC Process Development

从文档孤岛到过程智能：用于CMC过程开发的多层知识图谱

Amirmoshiri, Reza, Sahneh, Faryad, Jangjou, Yasser

Abstract

Chemistry, Manufacturing and Controls (CMC) process development generates an enormous body of technical information across a multi-stage, knowledge-intensive continuum from drug discovery to commercial manufacturing. This knowledge is traditionally fragmented across functions and heterogeneous formats, causing traceability gaps and significant knowledge-management costs during technology transfer and regulatory filing. We present a modular agentic-AI platform that converts a heterogeneous corpus of process-development documents into a queryable, dual-layer knowledge graph. A base knowledge layer builds a lexical graph with a Document-Section-Chunk hierarchy through lossless ingestion of digital, scanned, handwritten, and multilingual documents, while an intelligence layer extracts ontology-aligned entities and bridges cross-document concepts through a provenance-anchored domain graph. LLM agents operate across both layers, selecting the retrieval path best suited to each question. We evaluate the lexical layer with a novel three-tier protocol measuring the deployment-fidelity of a retrieval-augmented generation (RAG) system on proprietary data, demonstrated on 505 questions curated from 38 development reports of a Sanofi small-molecule program. Tier-1 multiple-choice accuracy of 95% signals strong platform reliability; the stricter Tier-2 LLM-judge pass rate of 85%, which degrades on comparative and corpus-wide questions, reveals a failure taxonomy that Tier-1 accuracy alone fails to capture. A router agent selects between layers according to question type. We anticipate this protocol will enable future designers of agentic platforms to assess their systems against nonpublic databases, and that graph-based architectures will see broader adoption in pharma as a means of transforming fragmented document repositories into structured process intelligence.

Chinese Translation

化学、制造和控制（CMC）过程开发在药物发现到商业制造的多阶段知识密集型连续体中生成了大量技术信息。这些知识传统上在不同职能和异构格式中碎片化，导致在技术转移和监管申报过程中出现可追溯性缺口和显著的知识管理成本。我们提出了一个模块化的代理人工智能平台，将异构的过程开发文档转化为可查询的双层知识图谱。基础知识层通过无损摄取数字、扫描、手写和多语言文档，构建了一个具有文档-章节-块层级的词汇图，而智能层则提取本体对齐的实体，并通过一个基于来源的领域图桥接跨文档概念。大型语言模型（LLM）代理在两个层级之间操作，根据每个问题选择最合适的检索路径。我们使用一种新颖的三层协议评估词汇层，测量在专有数据上检索增强生成（RAG）系统的部署保真度，该评估在来自Sanofi小分子项目的38份开发报告中策划的505个问题上进行了验证。第一层多项选择准确率为95%，表明平台的可靠性强；而更严格的第二层LLM评审通过率为85%，在比较和全语料库问题上下降，揭示了一种失败分类法，单靠第一层准确率无法捕捉。路由代理根据问题类型在层之间进行选择。我们预计该协议将使未来的代理平台设计者能够评估其系统与非公开数据库的对比，并且基于图的架构将在制药行业得到更广泛的应用，作为将碎片化文档库转化为结构化过程智能的一种手段。

View on arXiv Download PDF AI Translation

cs.AI / 57 / 2609.11498

ActMap: Single-Pass Uncertainty Quantification from Generation-Time Activation Maps

ActMap：基于生成时激活图的单次不确定性量化

Dardini, Jacopo, Calegari, Roberta

Abstract

Practical uncertainty quantification (UQ) for large language models must decide, from a single generation, whether a specific answer should be trusted. Existing methods either sample multiple generations, read only output-token probabilities, or reduce the model's internal computation to a single hidden state. We introduce ActMap, a white-box representation that compresses the generation-time hidden- state trajectory (every layer, every generated token) into a fixed $12 \times 32 \times 128$ tensor of temporal-statistic channels that preserves structure across transformer depth and pooled hidden coordinates. The map is captured during the generation pass with no measurable overhead, has a fixed shape across model depths and hidden sizes, and occupies 96 KiB: a compact artifact that can be retained for audit-relevant generations and probed directly, with occlusion analysis localizing the classifier's signal to mid-depth regions of the map. A lightweight classifier, instantiated as a compact Vision Transformer, reads an estimated correctness probability from each map in a fraction of a millisecond; capacity-matched MLPs perform comparably, indicating the representation itself carries the result. Trained and evaluated in-domain on short-answer QA, direct- answer math, and summarization factuality with three instruction-tuned 7-8B models, ActMap consistently outperforms sampling, token-probability, attention, and embedding baselines, and matches ACT-ViT, a detector trained on dense activation tensors $67 \times$ larger, at essentially the same mean AUROC with lower calibration error on ten of twelve pairs. The resulting score supports abstention, routing, and selective verification from a single generation, making it a practical primitive for scalable oversight of deployed models.

Chinese Translation

对于大型语言模型，实用的不确定性量化（UQ）必须在单次生成中决定特定答案是否值得信任。现有方法要么对多个生成进行采样，要么仅读取输出标记的概率，或者将模型的内部计算简化为单个隐藏状态。我们提出了ActMap，这是一种白盒表示法，将生成时的隐藏状态轨迹（每一层、每个生成的标记）压缩为固定的 $12 imes 32 imes 128$ 张量的时间统计通道，保留了跨变换器深度和汇聚隐藏坐标的结构。该图在生成过程中捕获，几乎没有可测量的开销，具有固定的形状，适用于不同模型深度和隐藏大小，占用96 KiB：这是一个紧凑的工件，可以保留用于审计相关的生成，并可以直接探测，通过遮挡分析将分类器的信号定位到图的中层区域。一个轻量级分类器，实例化为紧凑的视觉变换器（Vision Transformer），在几毫秒内从每个图中读取估计的正确性概率；与之匹配的多层感知器（MLPs）表现相似，表明表示本身承载了结果。在短答案问答、直接答案数学和摘要事实性方面，在三个指令调优的7-8B模型上进行领域内训练和评估，ActMap始终优于采样、标记概率、注意力和嵌入基线，并在十对中的十二对中以相似的平均AUROC匹配ACT-ViT，一个在密集激活张量上训练的检测器，其大小是其 $67 imes$。最终得分支持从单次生成中进行弃权、路由和选择性验证，使其成为可扩展监督已部署模型的实用原语。

View on arXiv Download PDF AI Translation

cs.AI / 58 / 2609.11509

Extending SMT Solving with Non-Ground Clause Learning

通过非平面子句学习扩展 SMT 求解

Briefs, Yasmine, Weidenbach, Christoph

Abstract

Quantifier instantiation is currently the main approach to non-ground SMT solving: solvers generate ground instances and solve the resulting ground SMT problems with CDCL(T)-style reasoning. When a conflict is found, conflict analysis learns only a ground clause, even though the conflict comes from instances of non-ground clauses. Yet non-ground reasoning can give exponentially shorter proofs than purely ground reasoning. We propose a calculus that consists of ground instantiations, CDCL(T)-style rules, and non-ground conflict analysis. The solver reasons on ground instances, but the resolution steps of conflict analysis are performed on their original non-ground clauses. This produces learned clauses that are typically more general than the ground conflict. With a suitable strategy, the learned clauses are even non-redundant. We also show how chronological backtracking can be included in SMT solving. Our calculus gives a common setting for CDCL(T)-style SMT solving, a range of instantiation-based procedures, and non-ground clause learning, and we prove that it simulates CDCL, SCL(FOL), SCL(T), and even Resolution.

Chinese Translation

量词实例化目前是非平面 SMT 求解的主要方法：求解器生成平面实例，并使用 CDCL(T) 风格的推理解决生成的平面 SMT 问题。当发现冲突时，冲突分析仅学习一个平面子句，尽管冲突来自非平面子句的实例。然而，非平面推理可以提供比纯平面推理更短的指数级证明。我们提出了一种演算，包括平面实例化、CDCL(T) 风格的规则和非平面冲突分析。求解器在平面实例上进行推理，但冲突分析的分辨步骤在其原始非平面子句上执行。这产生的学习子句通常比平面冲突更具一般性。通过合适的策略，学习子句甚至可以是非冗余的。我们还展示了如何将时间回溯纳入 SMT 求解。我们的演算为 CDCL(T) 风格的 SMT 求解、一系列基于实例化的过程和非平面子句学习提供了一个共同的框架，并且我们证明它可以模拟 CDCL、SCL(FOL)、SCL(T) 甚至是分辨率。

View on arXiv Download PDF AI Translation

cs.AI / 59 / 2609.11527

Lightweight LiDAR-Based Cone Detection Framework Using Random Forest for Formula Student Driverless

基于轻量级LiDAR的锥体检测框架：使用随机森林算法针对Formula Student Driverless

Mező-Kerekes, Márk, Praksz, Péter, Liu, Chang

Abstract

Reliable, low-latency perception is crucial for Formula Student Driverless vehicles, yet many existing pipelines rely on deep learning and multi-sensor fusion, often requiring GPU acceleration. This paper presents a lightweight LiDAR-only perception pipeline tailored for CPU execution, combining ground removal, IMU-based motion compensation, DBSCAN clustering, and geometric feature-based Random Forest classification. Feature importance analysis reduced the model input from 12 to 7 features while preserving performance. Evaluated on 2,371 labeled clusters collected from real FSD events, the pipeline achieves an F1-score of 98.33% and an end-to-end runtime of 3.13 ms on CPU-only hardware. The released dataset, labeling tool, and trained models provide a practical and reproducible baseline for other resource-constrained autonomous racing teams.

Chinese Translation

可靠的低延迟感知对于Formula Student Driverless车辆至关重要，但许多现有的处理管道依赖于深度学习和多传感器融合，通常需要GPU加速。本文提出了一种轻量级的仅基于LiDAR的感知管道，专为CPU执行而设计，结合了地面去除、基于IMU的运动补偿、DBSCAN聚类和基于几何特征的随机森林分类。特征重要性分析将模型输入从12个特征减少到7个，同时保持了性能。该管道在真实FSD事件中收集的2,371个标记聚类上进行了评估，达到了98.33%的F1分数，并在仅使用CPU的硬件上实现了3.13毫秒的端到端运行时间。发布的数据集、标注工具和训练模型为其他资源受限的自主赛车团队提供了一个实用且可重复的基准。

View on arXiv Download PDF AI Translation

cs.AI / 60 / 2609.11532

Prompt Revision as a Source of Cultural Bias in Text-to-Image Systems

提示修订作为文本到图像系统中的文化偏见来源

Urman, Aleksandra, Lichtenegger, Elsa, Jaoua, Salima, Bouleimen, Azza, Forsberg, Robin, Hertweck, Corinna, Ionescu, Stefania, Pagan, Nicolò, Hannak, Ancsa, Baumann, Joachim

Abstract

Commercial text-to-image systems silently revise user prompts before generating images, a step users typically cannot disable or even see. Yet, existing audits of cultural bias examine only the final images and treat generation as a single pipeline, so they cannot tell where the bias originates. We introduce WORLDVIEW, a multilingual benchmark of 8,960 prompts across 15 languages and 31 language-context pairings. Using it, we audit the revision layer in three systems (DALL-E-3, Imagen-4, GPT-Image-1.5) through a three-step analysis of how heavily it marks each cultural context, whether it flattens that context into a narrow vocabulary, and whether that vocabulary is stereotypical. Relative to a no-context English baseline, the US is the least-marked context, while non-Western and non-Anglophone contexts are marked far more heavily, flattened into narrow vocabularies applied across topically diverse prompts, and reduced to recognizable cultural stereotypes. Comparing images from original versus revised prompts on models without a revision layer, we identify the layer itself as a previously undocumented, causal source of this stereotyping. To locate cultural bias, and fix it, we must audit the system as deployed, not the model alone.

Chinese Translation

商业文本到图像系统在生成图像之前默默地修订用户提示，这是用户通常无法禁用或甚至看见的步骤。然而，现有的文化偏见审计仅检查最终图像，并将生成过程视为一个单一的管道，因此无法确定偏见的来源。我们引入了WORLDVIEW，这是一个涵盖15种语言和31种语言-上下文配对的8,960个提示的多语言基准。利用该基准，我们通过三步分析审计了三个系统（DALL-E-3、Imagen-4、GPT-Image-1.5）中的修订层，分析其在多大程度上标记每个文化上下文，是否将该上下文压缩为狭窄的词汇，以及该词汇是否具有刻板印象。相对于没有上下文的英语基线，美国是标记最少的上下文，而非西方和非英语国家的上下文则被标记得更为严重，压缩为适用于主题多样的提示的狭窄词汇，并简化为可识别的文化刻板印象。通过比较在没有修订层的模型中原始提示与修订提示生成的图像，我们确定修订层本身是这一刻板印象的一个先前未记录的因果来源。为了定位文化偏见并修正它，我们必须审计已部署的系统，而不仅仅是模型本身。

View on arXiv Download PDF AI Translation

cs.AI / 61 / 2609.11542

Characterizing Job Power Elasticity for Power-Flexible AI Training

为电力灵活的人工智能训练特征化作业功率弹性

Colangelo, Philip, Dawson, Charles, Sengupta, Shayan, Coskun, Ayse, Sivaram, Varun

Abstract

Large language model (LLM) training is among the fastest-growing sources of electricity demand in modern data centers, and power availability is a primary bottleneck to continued AI infrastructure growth. Making the power consumption of these workloads flexible could unlock additional power for AI growth, limit increases in electricity prices, and improve the utilization of existing grid infrastructure. However, to realize this flexibility, we must first understand how the performance of training workloads changes when GPU power is reduced. This paper presents the first systematic characterization of \emph{job power elasticity} (the sensitivity of throughput to power reductions) in LLM training. To quantify elasticity, we introduce the \emph{Power Flexibility Index (PFI)}, a normalized metric that quantifies the performance cost of power reductions and provides a control primitive for SLA-aware power flexibility. We collect data from 131 LLM training runs on H200 (plus 24 H200 validation runs and 34 matched H100 runs), including both dense and mixture-of-experts models, pretraining and fine-tuning tasks, and up to 32 GPUs. We find that LLM training jobs exhibit substantial but variable power elasticity, and we identify telemetry signals that predict PFI at runtime. Finally, we demonstrate that PFI-aware power allocation maximizes total tokens/second throughput under power constraints. Under a 30\% power reduction, PFI-aware power allocation recovers ~1.5k tokens/s per job, 63\% of the performance gap between an equal-weight allocation and an oracle with perfect information. Our results establish power elasticity as a measurable property of training jobs and provide a foundation for power-aware, grid-responsive AI infrastructure.

Chinese Translation

大型语言模型（LLM）训练是现代数据中心中电力需求增长最快的来源之一，而电力可用性是持续推动人工智能基础设施增长的主要瓶颈。使这些工作负载的电力消耗灵活化可以为人工智能的增长解锁额外的电力，限制电价上涨，并改善现有电网基础设施的利用率。然而，要实现这种灵活性，我们必须首先理解当GPU功率降低时，训练工作负载的性能如何变化。本文首次系统性地特征化了LLM训练中的 extit{作业功率弹性}（吞吐量对功率降低的敏感性）。为了量化弹性，我们引入了 extit{功率灵活性指数（PFI）}，这是一个标准化的指标，用于量化功率降低的性能成本，并为SLA感知的功率灵活性提供控制原语。我们从131次H200的LLM训练运行（加上24次H200验证运行和34次匹配的H100运行）中收集数据，包括密集模型和专家混合模型、预训练和微调任务，以及最多32个GPU。我们发现LLM训练作业表现出显著但可变的功率弹性，并识别出能够预测运行时PFI的遥测信号。最后，我们证明了PFI感知的功率分配在功率限制下最大化了每秒总令牌吞吐量。在30 ext{%}的功率降低下，PFI感知的功率分配每个作业恢复约1.5k令牌/秒，弥补了等权重分配与拥有完美信息的神谕之间63 ext{%}的性能差距。我们的结果确立了功率弹性作为训练作业的可测量属性，并为功率感知的、响应电网的人工智能基础设施提供了基础。

View on arXiv Download PDF AI Translation

cs.AI / 62 / 2609.11569

Enabling Knowledge Graph Understanding at Scale with the EXplore Your Graphs ENgine (EXYGEN)

通过EXplore Your Graphs ENgine (EXYGEN) 实现大规模知识图谱理解

Singh, Harshdeep, Zhu, Yurui, Colavizza, Giovanni, Romanello, Matteo

Abstract

We present EXYGEN (EXplore Your Graphs ENgine), a framework for knowledge graph (KG) understanding that enables conversational access to KGs at scale. We address two questions in sequence. First, how effectively can LLMs perform text-to-SPARQL generation given only automatically derived structured metadata and small graph samples, rather than task-specific fine-tuning? We integrate VoID descriptions and ShEx schemas into a retrieval-augmented generation (RAG) pipeline and ablate KG-derived context on the SciQA benchmark. Our best configuration -- combining ShEx schemas, retrieved triples, and example question-query pairs -- reaches an exact match of 0.419 on execution results without any LLM fine-tuning. We further find that lexical metrics such as F1 poorly predict query correctness, and that larger general-purpose LLMs can outperform smaller code-specialized ones once given sufficient context. Second, we ask how to generate the structured metadata that this method relies on from very large KGs, where KG metadata generation becomes computationally intractable. We introduce a predicate-coverage-aware parallel graph sampling strategy that preserves structural diversity while remaining computationally tractable. On OpenCitations Meta and GESIS, it retains high predicate coverage with minimal triple loss and reduces runtime by over 80x; on ORKG, sampling is not just faster but the only tractable path to obtain complete metadata. Together, these results show that structured schema context and lightweight prompting can substantially reduce reliance on fine-tuning for scalable conversational access to KGs, though closing the remaining gap to fully fine-tuned approaches will likely require reducing dependence on curated question-query exemplars -- whether through synthetic generation or an execution-feedback-driven approach -- and validating these findings beyond a single benchmark.

Chinese Translation

我们提出了EXYGEN（EXplore Your Graphs ENgine），这是一个用于知识图谱（KG）理解的框架，能够实现对KG的对话式访问。我们依次解决两个问题。首先，给定仅自动生成的结构化元数据和小型图样本，而不是特定任务的微调，LLMs（大型语言模型）在文本到SPARQL生成方面的效果如何？我们将VoID描述和ShEx模式集成到一个检索增强生成（RAG）管道中，并在SciQA基准上消融KG派生的上下文。我们最佳的配置——结合ShEx模式、检索的三元组和示例问题-查询对——在执行结果上达到了0.419的精确匹配，而没有进行任何LLM微调。我们进一步发现，诸如F1等词汇指标对查询正确性的预测效果较差，而较大的通用LLMs在获得足够上下文后可以超越较小的代码专用模型。其次，我们询问如何从非常大的KG中生成该方法所依赖的结构化元数据，因为KG元数据生成变得计算上不可行。我们引入了一种考虑谓词覆盖的并行图采样策略，该策略在保持结构多样性的同时仍然计算上可行。在OpenCitations Meta和GESIS上，它保持了高谓词覆盖率，三元组损失最小，并将运行时间减少了80倍以上；在ORKG上，采样不仅更快，而且是获取完整元数据的唯一可行路径。这些结果表明，结构化模式上下文和轻量级提示可以显著减少对微调的依赖，从而实现对KG的可扩展对话式访问，尽管要缩小与完全微调方法之间的剩余差距，可能需要减少对策划问题-查询示例的依赖——无论是通过合成生成还是执行反馈驱动的方法——并在单一基准之外验证这些发现。

View on arXiv Download PDF AI Translation

cs.AI / 63 / 2609.11607

Making Alternative Data Work: Context-Augmented LLMs for Financial Forecasting

使替代数据发挥作用：用于金融预测的上下文增强大型语言模型

Kwon, Jihoon, Liu, Lawrence, Park, Daekyung, Kim, Sumin, Jack, Haverty, Lee, Hoyoung, Bjorkman, Katherine, McKenney, Josh, Laurelli, Peter, Kagan, Nicole, Golkhou, Zach, Neumann, Thorsten, Tong, Edward, Petersen, Pete, Kim, Yoon, Lopez-Lira, Alejandro, Lee, Yongjae, Choi, Chanyeol

Abstract

When forecasting a firm's future financial performance, alternative data - data collected from non-traditional sources such as consumer transactions, web traffic, and prediction markets - can provide timely signals about firms' operating activities and broader market conditions. These signals may reveal information that is not captured by traditional public sources and can therefore provide complementary information for forecasting firms' future financial performance. However, firm-level alternative data often have limited historical coverage, are relevant only to specific prediction targets or subsets of firms, and are distributed across numerous heterogeneous channels, making them difficult to incorporate flexibly into conventional forecasting approaches. Meanwhile, large language models (LLMs) can interpret instructions, learn from in-context examples, and generate predictions by combining heterogeneous information without task-specific parameter updates. Motivated by this potential flexibility, we investigate whether an LLM can forecast firm performance by integrating alternative data with other financial information through in-context learning. We propose a two-agent framework that first identifies the firms for which each alternative data channel is likely to be informative and then predicts revenue using firm- and channel-specific context. We evaluate the framework across four commercial alternative data channels. In our experiments, adding alternative data in context alongside other financial information improves the LLM's forecasting relative to either source alone, and these forecasts are more accurate than those of standard forecasting baselines. These findings suggest that LLMs provide a flexible and practical approach to integrating alternative data with heterogeneous financial information.

Chinese Translation

在预测企业未来财务表现时，替代数据——来自非传统来源的数据，如消费者交易、网络流量和预测市场——可以提供关于企业运营活动和更广泛市场条件的及时信号。这些信号可能揭示传统公共来源未能捕捉的信息，因此可以为预测企业未来财务表现提供补充信息。然而，企业层面的替代数据往往历史覆盖有限，仅与特定预测目标或企业子集相关，并且分布在众多异构渠道中，使其难以灵活地纳入传统预测方法。同时，大型语言模型（LLMs）能够解释指令、从上下文示例中学习，并通过结合异构信息生成预测，而无需特定任务的参数更新。基于这种潜在的灵活性，我们研究了LLM是否能够通过上下文学习将替代数据与其他财务信息整合，从而预测企业表现。我们提出了一个双代理框架，首先识别每个替代数据渠道可能提供信息的企业，然后使用企业和渠道特定的上下文预测收入。我们在四个商业替代数据渠道上评估了该框架。在我们的实验中，将替代数据与其他财务信息结合在上下文中使用，相较于单独使用任一来源，提升了LLM的预测能力，并且这些预测的准确性超过了标准预测基准。这些发现表明，LLMs提供了一种灵活且实用的方法，将替代数据与异构财务信息整合。

View on arXiv Download PDF AI Translation

cs.AI / 64 / 2609.11615

Distributed Optimization of Modular Production Systems using Model-based Reinforcement Learning with Inverse Models

基于模型的强化学习与逆模型的分布式模块化生产系统优化

Schwung, Andreas, Yuwono, Steve, Lassoued, Sofiene, Schwung, Dorothea

Abstract

This paper presents a novel approach for data-driven self-learning control of highly flexible, modular manufacturing systems. Specifically, we employ a novel framework for model-based reinforcement learning which introduces approximate inverse process models within the training of reinforcement policies. This approach disentangles the learning of actuation dynamics and the dynamics in state space, resulting in RL-based training solely within the task space. We propose a lightweight feedforward architecture for approximate inverse models and integrate them within the policy network of standard RL algorithms. We apply the approach to a laboratory modular production testbed with heterogeneous production modules. The results underline the efficiency improvements for modular manufacturing units in terms of both performance and training speed, particularly for off-policy algorithms.

Chinese Translation

本文提出了一种用于高度灵活的模块化制造系统的数据驱动自学习控制的新方法。具体而言，我们采用了一种基于模型的强化学习的新框架，该框架在强化策略的训练中引入了近似逆过程模型。这种方法将执行动态与状态空间中的动态学习分开，从而使基于强化学习的训练仅限于任务空间。我们提出了一种轻量级前馈架构用于近似逆模型，并将其集成到标准强化学习算法的策略网络中。我们将该方法应用于一个具有异构生产模块的实验室模块化生产测试平台。结果强调了模块化制造单元在性能和训练速度方面的效率提升，特别是在离策略算法中。

View on arXiv Download PDF AI Translation

cs.AI / 65 / 2609.11636

MAPLE: Memory-Augmented Planning with Language and Evolution

MAPLE：基于语言和演化的记忆增强规划

Chen, Kesheng, Hu, Yamin, Luo, Wenjian

Abstract

Domain practitioners understand their business constraints but may lack operations-research expertise or dedicated support. LLM-based optimization agents translate natural-language requirements into models or solver programs that established optimization tools can execute. This progress makes optimization more accessible, but real-world operations are dynamic: changing demand, resources, and priorities require updates to data, constraints, and objectives. Methods centered on isolated requests offer limited support for rapid adaptation that preserves earlier decisions and reuses useful search results. We introduce MAPLE (Memory-Augmented Planning with Language and Evolution), an agent for maintaining optimization problems through successive natural-language requests. MAPLE combines language-based problem construction with mathematical programming and evolutionary search. It retains the optimization program, accepted plans, earlier updates, and candidate solutions for subsequent requests. We introduce NLDO, a benchmark of 15 trajectories and 180 updates spanning selection, scheduling, rostering, routing, and cloud-resource placement. In the main evaluation, MAPLE completes all trajectories and achieves online scalar quality of 0.951 and a Pareto hypervolume ratio of 0.875. Controlled comparisons further show that maintaining executable state improves update validity and can preserve useful search information across substantial revisions.

Chinese Translation

领域从业者了解他们的业务约束，但可能缺乏运筹学专业知识或专门支持。基于大型语言模型（LLM）的优化代理将自然语言需求转化为模型或求解程序，以便现有的优化工具能够执行。这一进展使得优化变得更加可及，但现实世界的操作是动态的：不断变化的需求、资源和优先事项需要对数据、约束和目标进行更新。以孤立请求为中心的方法对快速适应的支持有限，这种适应需要保留早期决策并重用有用的搜索结果。我们提出了MAPLE（基于语言和演化的记忆增强规划），这是一个通过连续的自然语言请求来维护优化问题的代理。MAPLE将基于语言的问题构建与数学编程和演化搜索相结合。它保留了优化程序、已接受的计划、早期更新和后续请求的候选解决方案。我们引入了NLDO，这是一个包含15条轨迹和180次更新的基准，涵盖选择、调度、排班、路径规划和云资源配置。在主要评估中，MAPLE完成了所有轨迹，并实现了在线标量质量0.951和帕累托超体积比0.875。受控比较进一步表明，维护可执行状态改善了更新的有效性，并能够在大幅修订中保留有用的搜索信息。

View on arXiv Download PDF AI Translation

cs.AI / 66 / 2609.11660

Autonomy, Social Norms, and Alignment: Towards a Developmental Framework for Autonomous Artificial Agents

自主性、社会规范与对齐：面向自主人工智能代理的发展框架

Notte, Marica, Marinucci, Ludovica, Santucci, Vieri Giuliano

Abstract

In recent years, artificial intelligence has made extraordinary progress thanks to large-scale models capable of generalization and the generation of complex outputs. However, transferring this potential into embodied agents reveals a significant limitation: the most advanced systems rely on pre-existing datasets and human feedback strategies that are powerful but insufficient in dynamic or unknown contexts. To adapt, an agent must acquire knowledge through direct interaction with its environment. One strategy to address this challenge involves introducing higher-level mechanisms, such as intrinsic motivations, which leverage curiosity and competence, to guide exploration and learning in complex environments. While this flexibility expands autonomy, it complicates the task of ensuring agents remain aligned with human goals. Alignment, already a challenge for artificial systems in general, becomes even more complex in unstructured and dynamic contexts where predefined rules prove insufficient. To be effective and adaptable, norms must be rooted in experience through an epistemological process that starting from simple, situated principles allows for the gradual construction of more complex rules through experience, autonomous learning, and cooperation with other moral agents. Similarly to children learning social norms by exploring their environment and participating in collective practices, artificial agents must also be educated toward alignment. Following Dennett, the status of a moral agent is not innate but is attributed gradually based on the ability to responsibly manage increasing degrees of freedom. From this perspective, the regulatory sandboxes can be viewed as pedagogical environments for AI: dynamic spaces where alignment develops as a formative process, progressively shaping autonomous behaviors through interaction and cooperation in scenarios of increasing complexity.

Chinese Translation

近年来，人工智能在大规模模型的推动下取得了非凡的进展，这些模型能够进行泛化并生成复杂的输出。然而，将这种潜力转化为具身代理时，显现出一个显著的局限性：最先进的系统依赖于现有的数据集和人类反馈策略，这些策略虽然强大，但在动态或未知的环境中却显得不足。为了适应环境，代理必须通过与其环境的直接互动来获取知识。解决这一挑战的一种策略是引入更高层次的机制，例如内在动机，这些机制利用好奇心和能力来指导在复杂环境中的探索和学习。虽然这种灵活性扩展了自主性，但它也使得确保代理与人类目标保持一致的任务变得复杂。对齐本身就是人工系统面临的挑战，在结构不明确和动态的环境中，这一挑战变得更加复杂，因为预定义的规则显得不足。为了有效和适应，规范必须通过一种认识论过程根植于经验，该过程从简单的、情境化的原则开始，通过经验、自主学习和与其他道德代理的合作逐步构建更复杂的规则。类似于儿童通过探索环境和参与集体实践来学习社会规范，人工代理也必须朝着对齐的方向进行教育。根据丹尼特的观点，道德代理的地位并非天生，而是基于能够负责任地管理日益增加的自由度而逐步赋予的。从这个角度来看，监管沙盒可以被视为人工智能的教育环境：这些动态空间使得对齐作为一种形成过程得以发展，通过在日益复杂的场景中的互动与合作，逐步塑造自主行为。

View on arXiv Download PDF AI Translation

cs.AI / 67 / 2609.11674

Geospatial AI, Dataverse Metadata, and the Study of Place-Based Government

地理空间人工智能、数据集元数据与基于地点的政府研究

EBanks, Danny, Jain, Devika

Abstract

Harvard Dataverse hosts over 150,000 research datasets, but the geographic information those datasets carry is entered as free text by depositors and has never been assembled into a searchable structure. We construct a knowledge graph from the repository's public data and metadata, organizing 102,650 datasets within a 215,985-node network of 528,003 edges linking datasets to keywords, publications, subjects, journals, and locations. Of those datasets, 43,991 (42.9 percent) carry at least one geospatial field, geographic coverage, geographic unit, or a bounding box and 96.9 percent of all nodes sit in a single connected component, so datasets remain reachable from one another even when their geospatial metadata share nothing in common. A conservative keyword search identifies 7,654 geospatially tagged datasets (17.4 percent) as directly policy-relevant, with elections and legislatures the largest cluster, followed by government administration, health policy, transportation, and education. Five datasets illustrate how this metadata behaves across policy domains and spatial scales, and an extended use case shows how community language models, stance detection with geographic aggregation, and partisan language bridging tools can attach discourse to place. The central obstacle is place resolution: the same location appears as many disconnected nodes. We argue that the graph provides a concrete setting for developing AI-driven metadata enrichment and entity resolution, and we document its coverage skew toward American, city-level data.

Chinese Translation

哈佛数据集（Harvard Dataverse）托管超过150,000个研究数据集，但这些数据集所携带的地理信息由存储者以自由文本形式输入，从未被汇总成可搜索的结构。我们从该存储库的公共数据和元数据中构建了一个知识图谱，将102,650个数据集组织在一个包含215,985个节点和528,003条边的网络中，这些边将数据集与关键词、出版物、主题、期刊和地点连接起来。在这些数据集中，有43,991个（42.9%）至少包含一个地理空间字段、地理覆盖范围、地理单位或边界框，且96.9%的节点位于一个单一的连通组件中，因此即使它们的地理空间元数据没有共同之处，数据集之间仍然可以相互访问。保守的关键词搜索识别出7,654个带有地理空间标签的数据集（17.4%）与政策直接相关，其中选举和立法机构是最大的集群，其次是政府管理、健康政策、交通和教育。五个数据集展示了这些元数据在政策领域和空间尺度上的表现，扩展的用例展示了社区语言模型、基于地理聚合的立场检测和党派语言桥接工具如何将话语与地点联系起来。主要障碍是地点解析：同一位置以多个不相连的节点出现。我们认为该图谱为开发基于人工智能的元数据丰富和实体解析提供了一个具体的环境，并记录了其对美国城市级数据的覆盖偏差。

View on arXiv Download PDF AI Translation

cs.AI / 68 / 2609.11682

COBRA-Skills: Contextual Bandit-Guided Evolution for Agent Skill Optimization

COBRA-Skills：基于上下文的强盗引导进化用于智能体技能优化

Lu, Pingchen, Wang, Xiangyi, Li, Xiang, Mao, Jie, Qu, Zikun, Luo, Junfeng, Shu, Yao, Low, Bryan Kian Hsiang, Dai, Zhongxiang

Abstract

Large language model (LLM) agents can benefit from reusable skills distilled from prior task experience, yet existing skill optimization methods often rely on costly execution-based evaluation and substantial task data. We introduce \textbf{COBRA-Skills}, an efficient framework that formulates skill optimization as budgeted sequential optimization over a dynamically evolving candidate space. COBRA-Skills couples contextual-bandit-guided prioritization with evidence-grounded skill evolution, selectively allocating evaluations to promising or informative candidates while continually refining the skill population from execution feedback. Across six heterogeneous agent benchmarks and three target models, COBRA-Skills consistently achieves the strongest average performance among compared methods, while reducing optimization cost by 55--58\% relative to SkillOpt and using only 50 unique optimization examples per benchmark. Further analyses show that COBRA-Skills remains robust to changes in the agent harness and performs effectively when the target model itself is used for skill generation and refinement.

Chinese Translation

大型语言模型（LLM）智能体可以从先前任务经验中提炼的可重用技能中受益，然而现有的技能优化方法往往依赖于高成本的执行评估和大量的任务数据。我们提出了 extbf{COBRA-Skills}，这是一个高效的框架，将技能优化形式化为在动态演变的候选空间上进行预算序列优化。COBRA-Skills将基于上下文的强盗引导优先级与基于证据的技能演化相结合，选择性地将评估分配给有前景或信息量大的候选，同时不断从执行反馈中精炼技能群体。在六个异构智能体基准和三个目标模型中，COBRA-Skills始终在比较方法中实现了最强的平均性能，同时相对于SkillOpt减少了55%至58%的优化成本，并且每个基准仅使用50个独特的优化示例。进一步分析表明，COBRA-Skills对智能体环境的变化保持稳健，并在目标模型本身用于技能生成和精炼时表现出色。

View on arXiv Download PDF AI Translation

cs.AI / 69 / 2609.11709

When Agents Disagree: Bayesian Backward Reasoning as a Label-Free Anchor for Multi-Agent Collective Decision-Making

当智能体意见不一致时：贝叶斯逆向推理作为无标签的多智能体集体决策锚点

Chen, Ken, Wang, Wei, Seneviratne, Sachith, Weeratunge, Hansani, Halgamuge, Saman

Abstract

When multiple LLM agents yield conflicting answers, the decision-making process dictates whether agent diversity improves performance or merely compounds shared errors. Existing collective decision-making methods, including voting, electoral rules, and LLM judges, rely on forward reasoning: they map evidence to labels in one direction. Although these methods can combine diverse forward traces, they still aggregate estimates that share this evidence-to-label factorization and can inherit correlated errors within the forward pool. We therefore construct a reverse posterior for each instance through Bayesian backward reasoning from an explicit likelihood. The forward and reverse posteriors provide differently factorized approximations of the underlying posterior. Because estimates from different factorizations may tend to share the same error less often, we use Jensen-Shannon divergence to rank agents by cross-path consistency. This cross-path consistency signal underlies three strategies: hard selection (MinJS), soft reweighting (FwdJS), and log-linear fusion (LogLin). Evaluated on DDXPlus across five LLM backbones, our proposed strategies show consistent improvements: MinJS outperforms random selection across all backbones, FwdJS generally improves over the strongest baseline, and LogLin achieves the best performance among the evaluated methods, with its largest gains on the subset where the agents disagree. Despite its weaker standalone accuracy, the reverse posterior serves as a more useful anchor than forward-only alternatives, providing complementary information for collective decision-making. When labeled data are available, a lightweight two-stage calibration can further refine the reverse anchor and improve aggregation performance.

Chinese Translation

当多个大型语言模型（LLM）智能体产生冲突的答案时，决策过程决定了智能体的多样性是提升性能还是仅仅加剧共享错误。现有的集体决策方法，包括投票、选举规则和LLM评判，依赖于正向推理：它们将证据映射到标签的方向。尽管这些方法可以结合多样的正向轨迹，但它们仍然聚合共享这种证据到标签的因子化的估计，并可能继承正向池中的相关错误。因此，我们通过从显式似然中进行贝叶斯逆向推理，为每个实例构建一个逆后验。正向和逆后验提供了对基础后验的不同因子化近似。由于来自不同因子化的估计可能较少共享相同的错误，我们使用詹森-香农散度（Jensen-Shannon divergence）根据交叉路径一致性对智能体进行排名。这种交叉路径一致性信号支撑了三种策略：硬选择（MinJS）、软重加权（FwdJS）和对数线性融合（LogLin）。在五个LLM骨干网络上对DDXPlus进行评估，我们提出的策略显示出一致的改进：MinJS在所有骨干网络上优于随机选择，FwdJS通常优于最强基线，而LogLin在评估的方法中实现了最佳性能，其在智能体意见不一致的子集上获得了最大的提升。尽管其单独的准确性较弱，逆后验作为比仅正向替代方案更有用的锚点，为集体决策提供了互补信息。当有标记数据可用时，轻量级的两阶段校准可以进一步优化逆锚点并提高聚合性能。

View on arXiv Download PDF AI Translation

cs.AI / 70 / 2609.11752

SIRF: A Spec-Internalized Risk Foundation Model for Industrial Content Risk Control

SIRF：一种用于工业内容风险控制的规范内化风险基础模型

Wu, Suwan, Lin, Yumeng, Yuan, Pengcheng, Jiang, Xiaolong

Abstract

For industrial content risk control, the real deployment constraint is not average accuracy but how much risk can be auto-handled under high precision and second-level latency. We present SIRF (Spec-Internalized Risk Foundation Model), which internalizes a platform's complex policies, synthesized without additional human annotation via EntiGraph, MAGA rewriting and account-level chain-of-thought (CoT), into the weights via continued pretraining (CPT), so rules are applied at high precision under an ultra-low-latency, verdict-only deployment. A controlled same-source comparison (Qwen3-8B-SFT vs. SIRF-8B-SFT, identical policy injection and verdict-only output form, differing only in policy-grounded CPT) attributes the gain to internalization: SIRF-8B-SFT reaches 71.3% Black Recall@P95, +15.1pp over the baseline, using only ~70M CPT tokens without harming general ability, and among included, logprob-available models under this interface it matches or exceeds far larger systems. SIRF is deployed as a tree-model adjudication layer (20% more mis-penalized samples recovered) and transfers to a freezing scenario at low cost (~70% relative mis-penalization reduction).

Chinese Translation

在工业内容风险控制中，实际部署的限制不是平均准确率，而是在高精度和秒级延迟下能够自动处理多少风险。我们提出了SIRF（规范内化风险基础模型），该模型通过EntiGraph、MAGA重写和账户级思维链（CoT）将平台的复杂政策内化为权重，而无需额外的人为标注，并通过持续预训练（CPT）进行合成，从而在超低延迟的仅判决部署下以高精度应用规则。通过受控的同源比较（Qwen3-8B-SFT与SIRF-8B-SFT，政策注入和仅判决输出形式相同，仅在政策基础的CPT上有所不同），将增益归因于内化：SIRF-8B-SFT在使用约7000万CPT标记的情况下，达到了71.3%的黑色召回率@P95，比基线提高了15.1个百分点，并且在该接口下的包含模型中，其表现与更大系统相匹配或超过。SIRF被部署为树模型裁决层（回收了20%更多的误罚样本），并以低成本转移到冻结场景（相对误罚减少约70%）。

View on arXiv Download PDF AI Translation

cs.AI / 71 / 2609.11768

A Unified Per-Token Gating Family for On-Policy Distillation: FKL/RKL Mixing with Multi-Channel and Bias Coefficients

统一的逐标记门控家族用于在线知识蒸馏：FKL/RKL 混合与多通道和偏置系数

Wu, Suwan, Lin, Yumeng, Yuan, Pengcheng, Jiang, Xiaolong

Abstract

Per-token gating of forward/reverse KL losses has become a standard technique for on-policy knowledge distillation (OPD), but existing methods such as EOPD (Jin et al., 2026) and ToDi (Jung et al., 2025) each fix a single gating signal and a single gating direction, and the two have never been compared directly. We introduce a four-coefficient parameterization lambda_t = sigma(a * h_t + b * u(x) + c + d * gap_t) in which direction-aligned proxies of EOPD and ToDi appear as one-dimensional (1D) restrictions, and which adds multi-channel composition and an explicit bias as further degrees of freedom. On TweetEval (Barbieri et al., 2020) emotion and hate, with a Qwen3-32B teacher and a Qwen3-4B student, configurations in the full family reach higher accuracy than the matched-magnitude single-channel (entropy-only / gap-only) 1D restrictions in 33 of 36 comparable cells, and a 26-cell mean-match isolation experiment places dynamic gating ahead of effective-KL-matched static baselines in 19 of 26 cells. Because cells share training data, models, and parameter substructure, we report both counts as exploratory aggregate directional evidence rather than as independent hypothesis tests. Targeted three-seed paired replications of the nine headline comparisons singled out by that sweep -- including a third task, offensive -- are directionally consistent, but individually smaller than the single-seed estimates and not significant at n=3. We therefore present the parameterization primarily as a shared coordinate system for comparing per-token gating designs in short-output classification OPD.

Chinese Translation

逐标记门控前向/反向 KL 损失已成为在线知识蒸馏（OPD）的标准技术，但现有方法如 EOPD（Jin et al., 2026）和 ToDi（Jung et al., 2025）各自固定了单一的门控信号和单一的门控方向，且两者从未直接比较。我们引入了一种四系数参数化形式 lambda_t = sigma(a * h_t + b * u(x) + c + d * gap_t)，其中 EOPD 和 ToDi 的方向对齐代理作为一维（1D）限制出现，并且增加了多通道组合和显式偏置作为进一步的自由度。在 TweetEval（Barbieri et al., 2020）情感和仇恨任务中，使用 Qwen3-32B 教师和 Qwen3-4B 学生的配置在完整家族中达到了比匹配幅度的单通道（仅熵/仅间隙）1D 限制更高的准确率，在 36 个可比单元中有 33 个表现优异，而在 26 个单元的均值匹配隔离实验中，动态门控在 26 个单元中有 19 个超越了有效 KL 匹配的静态基线。由于单元共享训练数据、模型和参数子结构，我们将这两者的计数报告为探索性聚合方向证据，而非独立假设检验。针对这次筛选出的九个主要比较的定向三种种子配对复制 -- 包括第三个任务，攻击性 -- 在方向上是一致的，但个别结果小于单种子估计，并且在 n=3 时不显著。因此，我们主要将该参数化形式呈现为比较短输出分类 OPD 中逐标记门控设计的共享坐标系统。

View on arXiv Download PDF AI Translation

cs.AI / 72 / 2609.11859

From Parameters to Answers: How LLMs Retrieve and Use Their Internal Knowledge

从参数到答案：大型语言模型如何检索和使用其内部知识

Wei, Wenkang, Fang, Yuan, Jiang, Renhe, Cheng, Hong, Yu, Xingtong

Abstract

How does a language model's dependence on query-routing information and target knowledge change as it answers a question? We study this question through layerwise interventions on the hidden state at the end of the question. Across Qwen, Llama, and Gemma, we compare country-continent questions with noun, adjective, and code answers while keeping several fitted measurements distinct. A pair-conditioned request direction describes which country is queried in natural single-country questions; a global request direction describes first- versus second-country requests in paired questions; separate selection candidates test control among contents already available in the hidden state. A diagnostic reanalysis of frozen Qwen natural-question states shows that the pair-conditioned direction grows stronger before interventions on it begin to alter later fitted knowledge, with this causal window opening while answer-supporting content is still forming. The paired three-model trajectories are not uniform: Gemma shows a partially overlapping mid-layer routing-content profile, whereas Llama has no sustained routing-effect window under the same gates. In the paired protocol, dependence on the global request direction decreases from fixed earlier to later layer sets while dependence on fitted content persists. A matched Qwen comparison shows that the pair-conditioned direction retains a late effect, so this operational handoff concerns the global fitted direction rather than all request information. These results separate early readability, natural strength, causal steering, and later content dependence.

Chinese Translation

语言模型在回答问题时对查询路由信息和目标知识的依赖如何变化？我们通过对问题结束时隐藏状态的逐层干预来研究这个问题。在 Qwen、Llama 和 Gemma 三个模型中，我们比较了国家-大陆问题与名词、形容词和代码答案，同时保持几个拟合测量的独立性。配对条件请求方向描述了在自然单国问题中查询的是哪个国家；全球请求方向描述了在配对问题中第一个国家与第二个国家请求的区别；独立选择候选项测试了在隐藏状态中已存在内容的控制。对冻结的 Qwen 自然问题状态的诊断性重新分析表明，配对条件方向在干预开始改变后期拟合知识之前变得更强，这一因果窗口在支持答案的内容仍在形成时打开。配对的三模型轨迹并不均匀：Gemma 显示出部分重叠的中层路由-内容特征，而 Llama 在相同的门控下没有持续的路由效应窗口。在配对协议中，对全球请求方向的依赖从早期固定层集减少到后期层集，而对拟合内容的依赖则持续存在。与 Qwen 的匹配比较表明，配对条件方向保持了晚期效应，因此这一操作交接涉及全球拟合方向，而非所有请求信息。这些结果区分了早期可读性、自然强度、因果引导和后期内容依赖。

View on arXiv Download PDF AI Translation

cs.AI / 73 / 2609.11860

Explainability Assistant: A Conversational XAI Interface for Interpreting Energy Consumption Models

可解释性助手：用于解释能源消费模型的对话式可解释人工智能界面

Krjutškov, Rodion, Barbu, Eduard, Sakkas, Nikos, Yfanti, Sofia

Abstract

Energy consumption forecasting relies on increasingly complex machine learning (ML) models, such as Genetic Programming-based symbolic regressors, whose predictions can be difficult for facility managers and building operators to interpret. Explainable Artificial Intelligence (XAI) techniques address this opacity, but traditional XAI dashboards require substantial technical expertise and provide limited flexibility for dynamic, context-aware inquiry. Conversational XAI systems offer a promising alternative; however, previous approaches, such as TalkToModel, were constrained by rigid custom grammars and achieved only 76.8% intent-parsing accuracy. This paper introduces the Explainability Assistant, an open-source conversational XAI system that leverages the function-calling capabilities of modern Large Language Models (LLMs) to overcome these limitations. The system achieves 94% intent-parsing accuracy, supports flexible natural language interaction, and adapts to different ML problem types without task-specific fine-tuning. We present the system's architecture and report results from a comparative evaluation conducted with energy domain specialists, contrasting the Explainability Assistant with a traditional XAI dashboard. The evaluation suggests improved usability and consistent task accuracy, with all experts unanimously preferring the conversational interface for practical use.

Chinese Translation

能源消费预测依赖于日益复杂的机器学习（ML）模型，如基于遗传编程的符号回归模型，其预测结果对于设施管理者和建筑运营商来说可能难以理解。可解释人工智能（XAI）技术旨在解决这一不透明性，但传统的XAI仪表板需要相当的技术专长，并且在动态、上下文感知查询方面提供的灵活性有限。对话式XAI系统提供了一个有前景的替代方案；然而，之前的方法，如TalkToModel，受限于严格的自定义语法，意图解析准确率仅为76.8%。本文介绍了可解释性助手，这是一个开源的对话式XAI系统，利用现代大型语言模型（LLMs）的函数调用能力来克服这些局限性。该系统实现了94%的意图解析准确率，支持灵活的自然语言交互，并能够在不同的机器学习问题类型之间适应，而无需特定任务的微调。我们展示了系统的架构，并报告了与能源领域专家进行的比较评估结果，将可解释性助手与传统的XAI仪表板进行了对比。评估结果表明，系统在可用性和任务准确性方面有所改善，所有专家一致偏好于对话式界面用于实际应用。

View on arXiv Download PDF AI Translation

cs.AI / 74 / 2609.11876

On the Regularization Landscape for the Linear Recommendation Models

线性推荐模型的正则化景观研究

Li, Dong, Liu, Zhenming, Jin, Ruoming, Zhou, Hao, Liu, Zhi, Gao, Jing, Ren, Bin

Abstract

Recently, a wide range of recommendation algorithms inspired by deep learning techniques have emerged as the performance leaders on several standard recommendation benchmarks. While these algorithms were built on different DL techniques (e.g., dropouts, autoencoder), they have similar performance and even similar cost functions. This paper studies whether the models' comparable performance are sheer coincidence, or they can be unified under a single framework. We find that all linear performance leaders effectively add only a nuclear-norm based regularizer, or a Frobenius-norm based regularizer. The former ones possess a (surprising) rigid structure that limits the models' predictive power but their solutions are low rank and have closed form. The latter ones are more expressive and more efficient for recommendation but their solutions are either full-rank or require executing hard-to-tune numeric procedures such as ADMM. Along this line of finding, we further propose two low-rank, closed-form solutions, derived from carefully generalizing Frobenius-norm based regularizers. The new solutions get the best of both nuclear-norm and Frobenius-norm world.

Chinese Translation

近年来，受到深度学习技术启发的多种推荐算法在多个标准推荐基准测试中表现出色，成为性能领先者。尽管这些算法基于不同的深度学习技术（例如，dropouts、自动编码器），但它们的性能相似，甚至成本函数也相似。本文研究了这些模型的可比性能是否仅仅是巧合，或者它们是否可以统一在一个框架下。我们发现，所有线性性能领先者实际上仅有效地添加了基于核范数的正则化器或基于弗罗贝尼乌斯范数的正则化器。前者具有（令人惊讶的）刚性结构，限制了模型的预测能力，但其解是低秩且具有封闭形式。后者则更具表现力且在推荐中更高效，但其解要么是全秩的，要么需要执行难以调整的数值程序，如ADMM。基于这一发现，我们进一步提出了两种低秩、封闭形式的解，这些解是通过仔细推广基于弗罗贝尼乌斯范数的正则化器而得出的。这些新解兼具核范数和弗罗贝尼乌斯范数的优点。

View on arXiv Download PDF AI Translation

cs.AI / 75 / 2609.11900

MindTopo: Can Foundation Models Reason in Topological Space?

MindTopo：基础模型能在拓扑空间中推理吗？

Ge, Yunfei, Liu, Anbang, Wang, Qineng, Garnica, Johnalbert, Lyu, Jianwen, Wang, Zihan, Tan, Reuben, Gao, Jianfeng, Zhang, Ruohan, Hong, Yining, Wu, Jiajun, Li, Manling

Abstract

Spatial reasoning depends not only on metric properties such as distance, angle, and shape, but also on topological relations that remain invariant under continuous deformation. Cognitive science identifies these relations as foundational to spatial understanding, yet foundation-model evaluations largely focus on metric or viewpoint-dependent relations. We introduce MindTopo, a benchmark of topological intuition across five properties grounded in cognitive science and formal topology: continuity, separation, order, enclosure, and knots. MindTopo evaluates each property at two cognitive levels. Reasoning asks a model to identify topological relations or infer how they change. Planning instantiates a foundation model as a closed-loop agent whose policy selects environment actions. MindTopo contains 11,030 instances across 13 procedurally generated task types with controllable difficulty. We benchmark 14 MLLMs and study agent configurations augmented with image and video generation, including 3 video generative models in planning settings. Every MLLM performs better on reasoning than on planning, and the best-performing model remains far below observed human performance. On Qwen3-VL-2B-Instruct, supervised fine-tuning and reinforcement learning improve reasoning more than planning. Generated observations retain local cues and reach plausible endpoints, but audited rollouts do not reliably follow environment dynamics or preserve topology across transitions. Our website is at https://mind-topo.github.io/

Chinese Translation

空间推理不仅依赖于距离、角度和形状等度量属性，还依赖于在连续变形下保持不变的拓扑关系。认知科学将这些关系视为空间理解的基础，然而基础模型的评估主要集中在度量或视角依赖的关系上。我们提出了MindTopo，这是一个基于认知科学和形式拓扑的五个属性的拓扑直觉基准：连续性、分离性、顺序、包围性和结。MindTopo在两个认知层面上评估每个属性。推理要求模型识别拓扑关系或推断它们如何变化。规划则将基础模型实例化为一个闭环代理，其策略选择环境动作。MindTopo包含13种程序生成任务类型中的11,030个实例，具有可控的难度。我们对14个多模态大语言模型（MLLMs）进行了基准测试，并研究了增强了图像和视频生成的代理配置，包括在规划设置中的3个视频生成模型。每个MLLM在推理上的表现优于规划，而表现最佳的模型仍远低于观察到的人类表现。在Qwen3-VL-2B-Instruct上，监督微调和强化学习对推理的提升超过了对规划的提升。生成的观察保留了局部线索并达到了合理的终点，但审计的回放并未可靠地遵循环境动态或在过渡中保持拓扑。我们的网站地址是 https://mind-topo.github.io/

View on arXiv Download PDF AI Translation

cs.AI / 76 / 2609.11911

Artificial Id: Drive and Persistent Alignment in Agentic AI

人工身份：驱动与代理人工智能中的持续对齐

Shkolnikov, Yakov Pyotr

Abstract

Agentic AI is moving from bounded task execution toward systems that retain consequential state, continue operating and adapt across task boundaries. That shift creates a control problem that current harnesses largely solve by hand: objectives, retries, verification, stopping rules and other behavioral transitions are specified externally. We propose an artificial id, an adaptive internal drive for determining whether behavior should continue, stop or change. In a minimal virtual Petri-dish experiment, a controller too small to perform general-purpose reasoning and receiving no task-specific behavioral objective develops useful control through differential persistence. The same mechanism selects an unintended physical strategy when that behavior persists better and later replaces a learned sensor mapping when its environmental meaning changes. These results show that adaptive direction can emerge without being explicitly specified as a behavioral objective. The same persistence that makes such adaptive agency useful can also allow misalignment, corrupted state and unintended behavior to persist across task boundaries. A scalable artificial id would carry consequential state and adaptive drive across those boundaries, making alignment a property of the continuing agentic system rather than of a model response or single trajectory. Such systems require a persistent alignment boundary over trusted observations, consequence channels, persistent state, authority, identity, provenance and hard constraints.

Chinese Translation

代理人工智能正从有限的任务执行向能够保留重要状态、持续运行并在任务边界间适应的系统转变。这一转变带来了控制问题，目前的控制手段主要依赖人工解决：目标、重试、验证、停止规则及其他行为转变均由外部指定。我们提出了一种人工身份（artificial id），作为一种自适应的内部驱动，用于判断行为是继续、停止还是改变。在一个最小的虚拟彼得里盘实验中，一个无法进行通用推理且没有特定任务行为目标的控制器，通过差异持久性发展出有用的控制。同样的机制在某种行为更持久时选择了一个意外的物理策略，并在其环境意义变化时替换了学习到的传感器映射。这些结果表明，自适应方向可以在未被明确指定为行为目标的情况下出现。使这种自适应代理有用的持久性也可能导致不对齐、状态损坏和意外行为在任务边界间持续存在。一个可扩展的人工身份将跨越这些边界携带重要状态和自适应驱动，使对齐成为持续代理系统的属性，而不是模型响应或单一轨迹的属性。这类系统需要在可信观察、后果通道、持久状态、权威、身份、来源和硬约束之间建立持久的对齐边界。

View on arXiv Download PDF AI Translation

cs.AI / 77 / 2609.11916

Can Edge-Deployable Vision-Language Models Identify Species?

边缘可部署的视觉-语言模型能否识别物种？

Zhou, William, Siripuram, Mayukha, Yan, Xiao, Liu, Ziqi, Ding, Yi

Abstract

Camera traps often run in the field on edge hardware with limited or no connectivity, making small, locally-deployable vision-language models (VLMs) -- not frontier-scale ones -- the practically relevant class to evaluate for species identification. We test whether models in this deployment-relevant 2--8B range carry genuine taxonomic knowledge, evaluating four such VLMs (Qwen3-VL 2B/4B/8B, Gemma3 4B) against the domain-specific specialist BioCLIP (300M parameters) on a 96-species task, comparing clean iNaturalist photographs against camera-trap imagery from 6 LILA.science collections, on two independently-sampled evaluation sets. All models identify species far above chance, but every model -- general-purpose or specialist -- degrades sharply on field imagery (domain gaps of 9.6--26.6 percentage points, consistent across taxonomic levels and both evaluation sets), indicating the degradation reflects general image legibility rather than fine-grained discrimination failure. BioCLIP substantially outperforms every VLM tested (by 33.2--59.2 percentage points across an expanded 200-image sample for every model) despite its far smaller size, suggesting the gap reflects specialized training data rather than model scale; yet BioCLIP's own domain gap (18.0 points) is statistically indistinguishable from the best VLM's (22.3 points), suggesting the clean-to-field degradation itself is a property of the image-quality shift rather than a general-purpose-model weakness. Under open-set prompting, 5.9--9.6% of responses are syntactically valid but taxonomically nonexistent species names; the relative fabrication-rate ranking across models replicates exactly across both evaluation sets, a more robust finding than any single point estimate.

Chinese Translation

相机陷阱通常在连接有限或没有连接的边缘硬件上运行，因此小型、本地可部署的视觉-语言模型（VLMs）——而非前沿规模的模型——是评估物种识别的实际相关类别。我们测试了在这一部署相关的2-8B范围内的模型是否具备真正的分类知识，评估了四个这样的VLM（Qwen3-VL 2B/4B/8B，Gemma3 4B）与领域特定的专家模型BioCLIP（300M参数）在96种物种任务上的表现，比较了来自6个LILA.science收藏的相机陷阱图像与干净的iNaturalist照片，在两个独立抽样的评估集上进行比较。所有模型的物种识别率远高于随机水平，但每个模型——无论是通用模型还是专家模型——在现场图像上的表现都显著下降（领域差距为9.6-26.6个百分点，在不同分类级别和两个评估集上保持一致），这表明下降反映了图像可读性的一般性问题，而非细粒度区分的失败。尽管BioCLIP的规模远小于每个测试的VLM，但其表现显著优于所有测试的VLM（在扩展的200幅图像样本中，每个模型的表现提升33.2-59.2个百分点），这表明这一差距反映了专门的训练数据而非模型规模；然而，BioCLIP自身的领域差距（18.0个百分点）在统计上与最佳VLM的差距（22.3个百分点）无显著区别，这表明干净到现场的下降本身是图像质量变化的特性，而非通用模型的弱点。在开放集提示下，5.9-9.6%的响应是语法有效但分类上不存在的物种名称；不同模型之间的相对伪造率排名在两个评估集上完全一致，这一发现比任何单一的点估计更为稳健。

View on arXiv Download PDF AI Translation

计算语言学 (Computation and Language)

cs.CL / 1 / 2609.10702

Data-Efficient Language Modeling: From Frontier Advancement to Principle-Guided Model Improvement

数据高效的语言建模：从前沿进展到原则指导的模型改进

Yang, Shuxing, Zhu, Kaihao, Yang, Junjie, Zhao, Rui, Wu, Junyao, Wang, Yize, Li, Wenhao, Chen, Fujia, Deng, Taowen, Hong, Shenzhan, Li, Yaqi, Li, Zichen, Mi, Jincheng, Pan, Yuang, Chen, Hongsheng, Yang, Yihao

Abstract

Learning from limited text requires models to use context, generalize to new inputs, and retain useful capabilities. Qiushi Engine conducted a long-horizon, end-to-end autonomous research program on BabyLM 2026 Strict-Small, within 10 million corpus words and 100 million cumulative word presentations. Three stages connected frontier advancement, principle discovery, and principle-guided model improvement. Stage I combined compact restatements, budget reinvestment, and residual incremental learning to build a frontier model. Stage II found that exact repetition and aligned restatement produce different patterns of context use, depending on target relations and prediction windows. In controlled tasks, recovering familiar performance did not ensure that unseen inputs could still use learned computations. These findings support a testable data-efficient learning principle: organize experience around the contextual dependencies needed for prediction; separately design visible information, supervision, and preservation; test learning, generalization, and retention. Stage III retained source text, masked more local clues, supervised selected targets, and preserved predictions on ordinarily masked inputs. Two continuation seeds from the same parent outperformed ordinary continuation on the complete nine-metric aggregate. Overall rose from 42.02 to 42.25 across two generations; the second achieved the highest Overall in the public Strict-Small snapshot of 8 September 2026. Further studies addressed compression, relational anchors, shared representations, and measurement. Models are available on Hugging Face; code and research records accompany the GitHub repository. Together, these stages illustrate Research RSI: recursive self-improvement of the research process. Scientific understanding and method innovations change subsequent questions and designs; new experiments test and refine them.

Chinese Translation

从有限文本中学习要求模型利用上下文、对新输入进行泛化并保留有用的能力。求是引擎在BabyLM 2026 Strict-Small上开展了一项长期的端到端自主研究计划，涉及1000万语料字和1亿累计词呈现。研究分为三个阶段，连接了前沿进展、原则发现和原则指导的模型改进。第一阶段结合了紧凑的重述、预算再投资和残差增量学习，构建了一个前沿模型。第二阶段发现，精确重复和对齐重述在目标关系和预测窗口的不同情况下产生不同的上下文使用模式。在受控任务中，恢复熟悉的表现并不确保未见输入仍能使用已学习的计算。这些发现支持一个可测试的数据高效学习原则：围绕预测所需的上下文依赖组织经验；单独设计可见信息、监督和保留；测试学习、泛化和保留。第三阶段保留源文本，屏蔽更多局部线索，监督选择的目标，并保留在通常屏蔽输入上的预测。来自同一父体的两个延续种子在完整的九项指标汇总中表现优于普通延续。整体得分从42.02上升至42.25，第二代在2026年9月8日的公共Strict-Small快照中取得了最高的整体得分。进一步的研究解决了压缩、关系锚点、共享表示和测量问题。模型可在Hugging Face上获取；代码和研究记录随GitHub存储库提供。这些阶段共同展示了研究的递归自我改进过程。科学理解和方法创新改变了后续的问题和设计；新的实验测试并完善这些问题。

View on arXiv Download PDF AI Translation

cs.CL / 2 / 2609.10715

NCP-ArchPreview Technical Report: Moving towards Latent Space Language Models through Next Concept Prediction

NCP-ArchPreview技术报告：通过下一个概念预测迈向潜在空间语言模型

NCP Team, Cao, Jiaqi, Chen, Chiyu, Cheng, Shuang, Cheng, Xu, Dai, Beiya, Feng, Yufan, Ge, Kewen, Ge, Ruijun, Huang, Jiayi, Jiao, Yang, Lin, Dahua, Lin, Zhouhan, Liu, Yifan, Liu, Yuliang, Qi, Biqing, Ruan, Mowen, Shen, Junzhe, Song, Yunchong, Sun, Hao, Tian, Zhongbo, Wang, Yixuan, Wei, Rubin, Xiong, Jiaxin, Yang, Kangyu, Yao, Qian, Zhang, Qi, Zhou, Bowen

Abstract

We introduce NCP-ArchPreview, a latent-space language model that pushes autoregressive pretraining beyond standard next-token prediction (NTP). Alongside NTP, the model learns through Next Concept Prediction (NCP) to predict discrete concepts that span multiple tokens, introducing an explicit and more challenging concept-level objective while preserving standard token-level autoregressive generation. NCP-ArchPreview builds a latent space by constructing a product-quantized concept vocabulary directly from its hidden states, and subsequently learns to predict future concepts via a dedicated Concept Module. These predicted concepts are then fed back to the token level to guide subsequent generation, with NTP and NCP trained jointly end-to-end. We scale this architecture to 8.9B parameters and train it on 5.73T tokens from the Dolma-3 dataset, marking the largest demonstration of a latent-space language model to date. Remarkably, by consuming only 51.3% of the total training tokens, NCP-ArchPreview achieves the final pretraining loss of OLMo-3-7B. Following full pretraining, it outperforms OLMo-3-7B by 2.45 points on the downstream macro-average, including a notable 5.99-point gain on GSM8K. Controlled experiments isolate a clear progression of performance gains stemming from both the latent architecture and the NCP objective. Furthermore, utilizing only 85% of the standard computation, NCP-ArchPreview approaches the training loss of a strictly parameter-aligned 8.9B baseline. The learned latent space remains highly valuable after the pretraining stage: updating just the 17M-parameter VQ module yields a novel, lightweight interface for domain adaptation, while a simple injection of concept representations into a DFlash2 drafter improves the mean accepted length by 4.17% with negligible overhead.

Chinese Translation

我们介绍了NCP-ArchPreview，这是一种潜在空间语言模型，推动自回归预训练超越标准的下一个标记预测（NTP）。除了NTP外，该模型还通过下一个概念预测（NCP）学习，以预测跨多个标记的离散概念，引入了一个明确且更具挑战性的概念级目标，同时保留了标准的标记级自回归生成。NCP-ArchPreview通过直接从其隐藏状态构建产品量化概念词汇来构建潜在空间，随后通过专门的概念模块学习预测未来概念。这些预测的概念随后反馈到标记级，以指导后续生成，NTP和NCP共同进行端到端训练。我们将该架构扩展到89亿参数，并在Dolma-3数据集上训练了5730亿个标记，标志着迄今为止潜在空间语言模型的最大展示。值得注意的是，NCP-ArchPreview在仅消耗51.3%的总训练标记的情况下，达到了OLMo-3-7B的最终预训练损失。在完成全面预训练后，它在下游宏观平均上超越了OLMo-3-7B 2.45分，包括在GSM8K上显著提高了5.99分。受控实验清晰地隔离出来自潜在架构和NCP目标的性能提升进展。此外，仅利用85%的标准计算，NCP-ArchPreview接近于严格参数对齐的89亿基线的训练损失。学习到的潜在空间在预训练阶段后仍然极具价值：仅更新1700万参数的VQ模块即可产生一个新的轻量级领域适应接口，而将概念表示简单注入到DFlash2草拟器中，则在几乎没有额外开销的情况下提高了平均接受长度4.17%。

View on arXiv Download PDF AI Translation

cs.CL / 3 / 2609.10722

CMNIE: An Information Extraction Benchmark for Chinese Military News

CMNIE：中文军事新闻信息提取基准

Yu, Yan, Zhu, Mengna, Song, Zhenyu, Yang, Hao, Chen, Haiwen, Wang, Mao

Abstract

Structured extraction from Chinese military news supports intelligence analysis, decision-making, and knowledge base construction. However, existing resources provide limited support for joint informa?tion extraction in this domain, especially when events, event arguments, entities, and relations must be modeled together. We present CMNIE, an information extraction benchmark for Chinese military news. Extend?ing military-domain resources beyond document-level event annotations, CMNIE jointly annotates event triggers, event arguments, named enti?ties, and entity relations under a unified domain schema. The dataset contains 13,000 instances collected from public Chinese military news, with manual annotations for 7 event types, 10 argument roles, 7 entity types, and 8 relation types. We evaluate supervised IE models, zero-shot large language models, and fine-tuned LLM-based extraction methods on a shared test set. Experimental results show that CMNIE remains chal?lenging, especially for relation extraction and exact matching of event?argument spans; zero-shot LLMs often identify relevant semantic units but fail to match gold span boundaries exactly. CMNIE provides a stan?dardized benchmark for studying schema adherence, exact span match?ing, and joint structured extraction in specialized Chinese news.

Chinese Translation

从中文军事新闻中进行结构化提取支持情报分析、决策制定和知识库构建。然而，现有资源对该领域的联合信息提取支持有限，尤其是在事件、事件论元、实体和关系必须共同建模时。我们提出了CMNIE，一个针对中文军事新闻的信息提取基准。CMNIE在文档级事件注释的基础上扩展了军事领域资源，统一注释事件触发器、事件论元、命名实体和实体关系。该数据集包含从公开中文军事新闻中收集的13,000个实例，手动注释了7种事件类型、10种论元角色、7种实体类型和8种关系类型。我们在共享测试集上评估了监督信息提取模型、零样本大型语言模型和微调的基于LLM的提取方法。实验结果表明，CMNIE仍然具有挑战性，特别是在关系提取和事件-论元跨度的精确匹配方面；零样本LLM通常能够识别相关的语义单位，但未能精确匹配金标准跨度边界。CMNIE为研究模式遵循、精确跨度匹配和在专业中文新闻中的联合结构提取提供了标准化基准。

View on arXiv Download PDF AI Translation

cs.CL / 4 / 2609.10745

Think Before You Link: Rarity, Reasoning, and Retrieval in Multilingual Entity Linking

链接之前请思考：多语言实体链接中的稀有性、推理与检索

Pengpun, Parinthapat, Khanuja, Simran, Neubig, Graham

Abstract

Multimodal entity linking grounds entity mentions in text and images to knowledge-base entries. These systems degrade on rare entities, but prior work measures rarity primarily through popularity-based metrics such as pageviews. We broaden this view using knowledge-graph structural metrics that capture how well an entity is documented and connected. These metrics identify many rare entities that popularity metrics miss. Across the resulting rare-entity slices, state-of-the-art accuracy drops by 15.4-39.9%, showing that different rarity definitions expose different failure modes. To address these failures, we introduce a simple, training-free framework in which a reasoning-capable vision-language model iteratively searches and reasons over Wikipedia, gathering evidence dynamically. Controlled experiments show that reasoning and retrieval are complementary. Reasoning alone does not significantly improve accuracy on rare entities. Retrieval without reasoning improves rare-entity accuracy but can hurt overall accuracy. Their combination performs best. On MERLIN, a multilingual multimodal entity linking benchmark over five languages (Hindi, Indonesian, Japanese, Tamil, Vietnamese), our best system improves over the state of the art by 6.9% overall and by up to 23.3% on rare-entity slices. We release MERLIN-Rare, rare-entity test slices for targeted evaluation, with our framework.

Chinese Translation

多模态实体链接将文本和图像中的实体提及与知识库条目相结合。这些系统在处理稀有实体时表现不佳，但之前的研究主要通过基于流行度的指标（如页面浏览量）来衡量稀有性。我们通过使用知识图谱结构指标来拓宽这一视角，这些指标捕捉了一个实体的文档化和连接程度。这些指标识别出许多流行度指标所遗漏的稀有实体。在由此产生的稀有实体切片中，最先进的准确率下降了15.4%至39.9%，显示出不同的稀有性定义揭示了不同的失败模式。为了解决这些失败，我们引入了一个简单的、无需训练的框架，其中一个具备推理能力的视觉-语言模型迭代地搜索和推理维基百科，动态收集证据。受控实验表明，推理和检索是互补的。单靠推理并未显著提高稀有实体的准确性。没有推理的检索提高了稀有实体的准确性，但可能会损害整体准确性。两者的结合表现最佳。在MERLIN上，这是一个涵盖五种语言（印地语、印尼语、日语、泰米尔语、越南语）的多语言多模态实体链接基准，我们的最佳系统在整体上提高了6.9%，在稀有实体切片上提高了高达23.3%。我们发布了MERLIN-Rare，稀有实体测试切片以便进行针对性评估，配合我们的框架。

View on arXiv Download PDF AI Translation

cs.CL / 5 / 2609.10758

Multilingual in Name Only? Cultural and Linguistic Weaknesses of LLMs in Urdu

名义上的多语言？大语言模型在乌尔都语中的文化和语言弱点

Adeeba, Farah, Khan, Abdul Rafae, Bhatt, Rajesh, Sajjad, Hassan

Abstract

Multilingual large language models (LLMs) are increasingly used for open-ended text generation, yet their behaviour in low-resource languages remains poorly understood. In this work, we question how correct and reliable is the generation of multilingual LLMs when used for the task of story generation. We consider Urdu language as a representative low-resource language. We generate Urdu-Stories, a corpus of 93 stories generated using three contemporary LLMs (GPT-5.1, Qwen-3-Max, DeepSeek-3.1). We manually annotate the errors present in them under a nine-label linguistic, semantic, and cultural taxonomy. Our notable findings suggest that LLMs often make basic errors of grammar and semantics. The stories lack coherence, have unnatural repetition and show pervasive cultural shallowness. We further show using few-shot prompting that the cultural and context errors largely remain unresolved. Our findings highlight the limitations of current LLMs as a reliable source of content generation and information retrieval for low-resource languages.

Chinese Translation

多语言的大语言模型（LLMs）在开放式文本生成中被越来越多地使用，但它们在低资源语言中的表现仍然不够清楚。在本研究中，我们质疑多语言LLMs在故事生成任务中的生成结果有多么正确和可靠。我们将乌尔都语视为一个具有代表性的低资源语言。我们生成了乌尔都故事（Urdu-Stories），这是一个由三种当代LLMs（GPT-5.1、Qwen-3-Max、DeepSeek-3.1）生成的93个故事的语料库。我们在九个标签的语言、语义和文化分类法下手动标注了其中存在的错误。我们的显著发现表明，LLMs经常出现基本的语法和语义错误。这些故事缺乏连贯性，存在不自然的重复，并且表现出普遍的文化肤浅。我们进一步通过少量示例提示展示，文化和上下文错误在很大程度上仍未得到解决。我们的研究结果突显了当前LLMs作为低资源语言内容生成和信息检索可靠来源的局限性。

View on arXiv Download PDF AI Translation

cs.CL / 6 / 2609.10792

Analyzing Traditional and Neural Approaches to Multilingual Readability Assessment

分析传统与神经方法在多语言可读性评估中的应用

Wong, Joshua, Tanner, Chris

Abstract

Transformer-based models excel at Automatic Readability Assessment (ARA), yet feature-based models remain in active use because their predictions tie back to linguistic properties. This matters because readability labels are subjective and rater-dependent, so high accuracy on noisy ground truth may reflect surface patterns rather than the linguistic structure that defines difficulty. We test whether transformers internalize the same features as traditional models across Arabic, English, French, Hindi, and Russian using the ReadMe++ dataset. Shapley Additive Explanations (SHAP) identify the features driving traditional classifiers, which we then use as TCAV concept sets to probe multilingual XLM-R and language-specific encoders. Transformers recover surface-length, syntactic, and lexical-diversity signals, and reflect the ordinal CEFR structure of the traditional models. Alignment varies by model family, language, and layer, with language-specific encoders tracking traditional models more clearly than XLM-R. High linear separability does not always imply directional influence, limiting linear probing for count-based readability features.

Chinese Translation

基于变换器的模型在自动可读性评估（ARA）中表现出色，但基于特征的模型仍在积极使用，因为它们的预测与语言特性相关。这一点很重要，因为可读性标签是主观的且依赖于评估者，因此在嘈杂的真实数据上高准确率可能反映的是表面模式，而非定义难度的语言结构。我们使用 ReadMe++ 数据集测试变换器是否内化了与传统模型相同的特征，涵盖阿拉伯语、英语、法语、印地语和俄语。Shapley 加性解释（SHAP）识别驱动传统分类器的特征，我们随后将其用作 TCAV 概念集，以探究多语言 XLM-R 和语言特定编码器。变换器恢复了表面长度、句法和词汇多样性信号，并反映了传统模型的等级 CEFR 结构。对齐情况因模型家族、语言和层次而异，语言特定编码器比 XLM-R 更清晰地跟踪传统模型。高线性可分性并不总是意味着方向性影响，这限制了基于计数的可读性特征的线性探测。

View on arXiv Download PDF AI Translation

cs.CL / 7 / 2609.10810

Larger Context Window, Fewer Overcorrections: Optimizing Prompts and Batching for Minimal-Edit Grammatical Error Correction

更大的上下文窗口，更少的过度修正：优化提示和批处理以实现最小编辑语法错误修正

Karpo, Kateryna, Chernodub, Artem

Abstract

Minimal-edit Grammatical Error Correction (GEC) is a challenging task for zero- and few-shot prompted Large Language Models (LLMs), which systematically overcorrect and degrade $F_{0.5}$ by rewriting well-formed spans. While fine-tuning provides an effective solution, it imposes substantial infrastructure demands. We introduce a prompt-based approach that closes the gap to fine-tuned models through three advances in GEC prompting methodology. First, we introduce taxonomy-based instructions to enforce minimal-edit constraints with a comprehensive list of grammatical error rules, equipping the LLM with a bounded, metric-aligned scope of correctable edits, which benefits the strongest models while remaining model-dependent overall. Second, we show that batching multiple uncorrected sentences into a single input context acts as a targeted regularizer against overcorrection, systematically reducing the edit rate across diverse LLM families; we hypothesize this arises from attention dilution effect induced by the bounded capacity of self-attention scores. Finally, LLM-assisted Prompt Optimization refines these instructions. Powered by Gemini 3.1-Pro, our prompt achieves $F_{0.5}=78.32$ on the BEA-2019 test set - establishing a new prompt-based SOTA while shrinking the gap to the fine-tuned single-model SOTA (Staruch et al., 2025) to a mere $0.38$ points. Code, prompts, and outputs are publicly available.

Chinese Translation

最小编辑语法错误修正（GEC）是一个对零样本和少样本提示的大型语言模型（LLMs）具有挑战性的任务，这些模型系统性地过度修正并通过重写格式良好的文本片段而降低 $F_{0.5}$。虽然微调提供了有效的解决方案，但它对基础设施的需求很大。我们提出了一种基于提示的方法，通过在 GEC 提示方法论中的三项进展，缩小了与微调模型之间的差距。首先，我们引入基于分类法的指令，通过全面的语法错误规则列表来强制执行最小编辑约束，为 LLM 提供一个有限的、与度量对齐的可修正编辑范围，这使得最强的模型受益，同时整体上仍然依赖于模型。其次，我们展示了将多个未修正句子批量处理为单个输入上下文作为针对过度修正的有针对性的正则化器，系统性地降低了不同 LLM 家族的编辑率；我们假设这源于自注意力分数的有限容量所引发的注意力稀释效应。最后，LLM 辅助的提示优化进一步完善了这些指令。在 Gemini 3.1-Pro 的支持下，我们的提示在 BEA-2019 测试集上达到了 $F_{0.5}=78.32$，建立了新的基于提示的最先进水平，同时将与微调单模型最先进水平（Staruch et al., 2025）之间的差距缩小至仅 $0.38$ 分。代码、提示和输出均已公开。

View on arXiv Download PDF AI Translation

cs.CL / 8 / 2609.10830

Detectable Only Where It Is Confounded: What Verified Duplication Counts Say About Membership Evidence in Language Models

仅在混淆中可检测：验证的重复计数对语言模型中成员证据的启示

Khah, Arman Nik

Abstract

When a language model finds a sentence unusually cheap to predict, it is tempting to conclude that the sentence was in its training data. Almost every published test of that inference has had to guess which sentences were in the training data, the members, and which were not. This paper removes the guessing. Two model families, OLMo-2 and Pythia, publish their pretraining corpora, and a public index over those corpora returns the exact number of times any sentence appeared in each. Those counts make three questions answerable directly. The answers form a pincer, closing from two sides. At the duplication levels ordinary text actually has, five models from 1B to 13B parameters carry at most a faint trace of their own exposure. We measure that trace with a design that reads the same sentence through two models, which cancels fluency and quality by construction, and it comes to a rank correlation near -0.08, where -1 would be a perfect relation and 0 none. Where the trace does become strong, above roughly a thousand copies, the two corpora agree on which sentences those are, because they are the famous ones, so exposure can no longer be told apart from fame. Two further measurements show how apparent membership signal gets manufactured. A common way to build a non-member is to change one word of a member. The model does prefer the original, but the gap is the same whether the original appeared once or a hundred times, so what the model is rewarding is the author's word choice, not memory. Above a thousand copies the gap grows with model size on the twelve sentences we can test there, at the same boundary where the pincer closes. And swapping the controls for sentences that differ from the members in register moves a detector from 0.83 to 0.94 AUC, on a scale where 0.5 is a coin flip and 1.0 is perfect separation. We release the sentence banks, counts, and code.

Chinese Translation

当语言模型发现某个句子的预测成本异常低时，往往会得出该句子出现在其训练数据中的结论。几乎所有已发表的关于这一推断的测试都不得不猜测哪些句子是训练数据中的成员，哪些不是。本文消除了这种猜测。两个模型系列，OLMo-2 和 Pythia，公开了它们的预训练语料库，并且一个公共索引可以返回任何句子在每个语料库中出现的确切次数。这些计数使得三个问题可以直接回答。答案形成了一个夹击，从两个方面收紧。在普通文本实际具有的重复水平下，从 10 亿到 130 亿参数的五个模型最多只带有微弱的自身暴露痕迹。我们通过一种设计来测量这种痕迹，该设计通过两个模型读取同一句子，从而在构造上抵消流畅性和质量，最终得出的等级相关性接近 -0.08，其中 -1 表示完美关系，0 表示无关系。当痕迹变得强烈时，超过大约一千个副本，两个语料库在这些句子上达成一致，因为它们是著名的句子，因此暴露与名声无法再区分。两个进一步的测量显示了明显的成员信号是如何被制造的。构建非成员的一种常见方式是改变成员的一个单词。模型确实更倾向于原始句子，但无论原始句子出现一次还是一百次，差距都是相同的，因此模型所奖励的是作者的用词选择，而不是记忆。在超过一千个副本的情况下，在我们可以测试的十二个句子上，差距随着模型规模的增加而增长，正是在夹击收紧的同一边界上。而将与成员在语域上不同的句子的控制进行交换，使得检测器的 AUC 从 0.83 提升至 0.94，0.5 为随机猜测，1.0 为完美分离。我们发布了句子库、计数和代码。

View on arXiv Download PDF AI Translation

cs.CL / 9 / 2609.10893

Does Linguistic Structure Enrichment Enhance Coherence Assessment? Not With Current Architectures

语言结构丰富化是否增强了连贯性评估？当前架构下并未实现

Mazzotti, Victor, Pereira, Luiz, Santos, Marina Bitencourt dos, Maia, Helena, Caetano, Carlos, Felix, Nádia, Avila, Sandra

Abstract

Recent advances in large language models have transformed human-computer interaction. Despite their fluency, these models often produce texts that are grammatically correct but semantically incoherent, containing contradictions or disruptions in logical flow. This work investigates whether enriching text with syntactic and rhetorical information can improve incoherence prediction. Our experiments and analysis show that plain texts achieved higher accuracy because the added information was structurally and syntactically incompatible with the language model's architecture. Additionally, to demonstrate the practical importance of coherence assessment, we performed zero-shot experiments on a Brazilian disinformation dataset, suggesting that textual coherence can serve as a proxy for detecting misleading content. Code and models are available at https://github.com/ittozzamV/cohereclassifier.

Chinese Translation

近期大型语言模型的进展改变了人机交互的方式。尽管这些模型生成的文本流畅，但它们往往产生语法正确但语义不连贯的文本，包含矛盾或逻辑流的中断。本研究探讨了通过丰富文本的句法和修辞信息是否能够改善不连贯性预测。我们的实验和分析表明，普通文本的准确性更高，因为所添加的信息在结构和句法上与语言模型的架构不兼容。此外，为了展示连贯性评估的实际重要性，我们在一个巴西虚假信息数据集上进行了零样本实验，结果表明文本连贯性可以作为检测误导性内容的代理。代码和模型可在 https://github.com/ittozzamV/cohereclassifier 获取。

View on arXiv Download PDF AI Translation

cs.CL / 10 / 2609.10896

LLM-Anchored Paralinguistic Enrichment for Alzheimer's Disease Detection

基于大型语言模型的阿尔茨海默病检测的副语言学增强

Wei, Xiao, Lin, Yuqin, Cao, Yaru, Li, Jinyu, Wen, Bin, Li, Kai, Chen, Yueying, Wang, Longbiao, Dang, Jianwu

Abstract

Speech-based automatic detection of Alzheimer's disease (AD) provides a non-invasive and scalable approach to early cognitive screening. AD affects both lexical-semantic organization and speech production, including atypical pauses and word elongations. However, existing methods have yet to fully integrate these paralinguistic cues with linguistic content. We propose LLM-Anchored Paralinguistic Enrichment (LAPE), which enriches LLM-derived linguistic representations with paralinguistic cues through three coordinated innovations. The first is prosodic event textualization, which enables the LLM to model pauses and elongations jointly with lexical content by encoding them as explicit markers with bounded duration-aware repetition. The second is lexico-prosodic unitization and chunking, which preserves event identity and magnitude in both modalities by pooling only consecutive word units. The third is text-anchored paralinguistic fusion, which integrates local and utterance-level speech features by using NormGate to normalize and dynamically scale them relative to text. We evaluate LAPE on ADReSS and ADReSSo using participant-level cross-validation and leave-one-subject-out evaluation. LAPE achieves state-of-the-art performance across all four primary settings. Code will be released upon acceptance.

Chinese Translation

基于语音的阿尔茨海默病（AD）自动检测提供了一种非侵入性和可扩展的早期认知筛查方法。AD影响词汇-语义组织和语音产生，包括非典型的停顿和词语延长。然而，现有方法尚未充分将这些副语言学线索与语言内容整合。我们提出了基于大型语言模型的副语言学增强（LAPE），通过三项协调创新将大型语言模型（LLM）生成的语言表示与副语言学线索相结合。第一项是韵律事件文本化，它使LLM能够将停顿和延长与词汇内容共同建模，通过将其编码为具有有限持续时间感知重复的显式标记。第二项是词汇-韵律单元化和分块，它通过仅汇聚连续的词单元来保持两种模式下事件的身份和大小。第三项是文本锚定的副语言学融合，它通过使用NormGate对局部和发话级别的语音特征进行归一化和动态缩放，从而将其与文本相结合。我们在ADReSS和ADReSSo上评估了LAPE，采用参与者级交叉验证和留一法评估。LAPE在所有四个主要设置中实现了最先进的性能。代码将在接受后发布。

View on arXiv Download PDF AI Translation

cs.CL / 11 / 2609.10901

SearchAtlas: Analyzing Agentic Search Strategies via Evidential Query Graphs

SearchAtlas：通过证据查询图分析代理搜索策略

Sang, Jiacheng, Li, Mengyuan, Chen, Sanxing, Huang, Yukun, Feng, Yu, Dhingra, Bhuwan

Abstract

LLM search agents are often evaluated on final-answer accuracy, overlooking the process. Analyzing a search strategy requires understanding how credible evidence is retrieved to address question constraints. This valuable information is buried in raw search trajectories that are long and difficult to parse. We introduce SearchAtlas, a framework that converts search trajectories into structured graphs whose edges represent how evidence is propagated across the reasoning trace, from the query that retrieves it to the final answer. Our automated parsing pipeline achieves a mean edge F1 of 86.0% against human-annotated graphs and remains consistent across repeated runs. We analyze five search agents on three benchmarks, revealing systematic differences in search scale and evidence aggregation. SearchAtlas exposes fragmented answer support, question constraints that do not reach the answer, and unverified parametric knowledge entering the response. These process failures are strongly associated with incorrect answers, even more so than an LLM judge given either the raw trajectory or the ordered query list, suggesting that the constructed graphs provide useful interpretability. Moreover, an audit of cases in which process-diagnostic scores disagree with final-answer correctness shows that they capture information not reducible to answer accuracy.

Chinese Translation

大型语言模型（LLM）搜索代理通常通过最终答案的准确性进行评估，而忽视了过程的分析。分析搜索策略需要理解如何检索可信的证据以满足问题约束。这些有价值的信息埋藏在原始搜索轨迹中，这些轨迹往往较长且难以解析。我们提出了SearchAtlas，一个将搜索轨迹转换为结构化图的框架，其边缘表示证据在推理轨迹中的传播方式，从检索证据的查询到最终答案。我们的自动解析管道在与人工标注图的比较中实现了86.0%的平均边缘F1分数，并且在重复运行中保持一致。我们在三个基准上分析了五个搜索代理，揭示了搜索规模和证据聚合的系统性差异。SearchAtlas揭示了碎片化的答案支持、未能达到答案的问题约束，以及进入响应的未经验证的参数知识。这些过程失误与错误答案之间存在强关联，甚至比给定原始轨迹或有序查询列表的LLM评判者更为显著，这表明构建的图提供了有用的可解释性。此外，对过程诊断分数与最终答案正确性不一致的案例进行审计显示，这些分数捕捉了无法简化为答案准确性的信息。

View on arXiv Download PDF AI Translation

cs.CL / 12 / 2609.10922

Auto-RecSys: Harnessing Autonomous Research Agents for Industry-Scale Recommender System

Auto-RecSys：利用自主研究代理实现行业规模推荐系统

Li, Ming, Li, Dai, Ning, Xuying, Sun, Bo, Li, Rui, Zhang, Yi, Gong, Silvia, Cao, Xuan, Li, Rui, Carapcea, Cornelia, Zhang, Qunshu, Wang, Zhigang, Xia, Yinglong, Wang, Andy

Abstract

Auto-research agents have shown the potential to automate hypothesis generation, experiment execution, and iterative refinement. However, scaling this paradigm to industry-scale recommendation models introduces two challenges: (1) long feedback loops, where model training can take days, making serial iteration prohibitively slow and requiring parallel exploration across multiple research directions; and (2) system complexity, where large configurations, fragile infrastructure dependencies, and multi-day GPU jobs require robust and recoverable execution. We present Auto-RecSys, an autonomous research system for long-horizon experimentation on industry-scale recommendation models. Auto-RecSys addresses these challenges through three harness designs: (1) distributed asynchronous execution for running multiple experiments in parallel across servers, (2) centralized cross-server memory for persistent and recoverable execution across sessions and failures, and (3) cognitive-procedural separation, where natural-language skill files guide LLM reasoning while deterministic scripts enforce operational correctness. Auto-RecSys further employs a dual-loop self-evolving architecture: an Execution Evolution Loop in which model-specific playbooks accumulate operational knowledge by recording failed attempts and crystallizing successful pipelines, and an Idea Evolution Loop in which experimental outcomes inform subsequent ideation. Evaluated on recommendation models, Auto-RecSys significantly reduces the human time required per experiment cycle and improves execution reliability as its playbooks mature.

Chinese Translation

自主研究代理展示了自动化假设生成、实验执行和迭代优化的潜力。然而，将这一范式扩展到行业规模的推荐模型面临两个挑战：（1）长反馈循环，模型训练可能需要数天时间，使得串行迭代变得极其缓慢，并且需要在多个研究方向上进行并行探索；（2）系统复杂性，大规模配置、脆弱的基础设施依赖和多天的GPU作业要求强健且可恢复的执行。我们提出了Auto-RecSys，一个用于行业规模推荐模型的长期实验的自主研究系统。Auto-RecSys通过三种设计来应对这些挑战：（1）分布式异步执行，在服务器之间并行运行多个实验；（2）集中式跨服务器内存，实现跨会话和故障的持久和可恢复执行；（3）认知-程序分离，自然语言技能文件指导大型语言模型（LLM）推理，同时确定性脚本确保操作的正确性。Auto-RecSys进一步采用了双循环自我演化架构：执行演化循环，在该循环中，模型特定的操作手册通过记录失败尝试和提炼成功管道来积累操作知识；以及创意演化循环，在该循环中，实验结果为后续的创意提供信息。在推荐模型上的评估表明，Auto-RecSys显著减少了每个实验周期所需的人力时间，并随着操作手册的成熟提高了执行的可靠性。

View on arXiv Download PDF AI Translation

cs.CL / 13 / 2609.10923

Structurally Speaking: Motif-Oriented Graph Captioning through Bidirectional Graph-Text Translation

结构性表达：通过双向图-文本翻译的基于图案的图表说明

Lu, Hsiao-Ying, Liu, Dongyu, Ma, Kwan-Liu

Abstract

Graph captions should help readers understand graph structure, rather than simply translate adjacency matrices into long textual edge lists. A useful graph caption abstracts connectivity into recognizable motifs, such as hubs, paths, cycles, cliques, and bridges, because these motifs provide compact structural units that are easier to read, compare, and recover. In this paper, we study motif-oriented graph captioning as a bidirectional graph-text translation task, where captions must both preserve enough topology for graph recovery and express the graph through concise motif-level descriptions. We show that direct prompting of GPT-5.1 often produces graph-recoverable captions by enumerating node-to-node connections, but these captions are verbose and can contain inconsistent motif interpretations. To address this gap, we introduce Structurally Speaking, a lightweight structured prompting protocol that guides translation between explicit connectivity and motif-level abstraction. Experiments on a synthetic motif-based dataset show that structured prompting produces shorter and more motif-consistent captions while maintaining comparable graph recovery. These results suggest that explicit topology-to-motif reasoning guidance can make LLM-generated graph captions more interpretable without model fine-tuning.

Chinese Translation

图表说明应帮助读者理解图的结构，而不仅仅是将邻接矩阵转换为冗长的文本边列表。有效的图表说明将连通性抽象为可识别的图案，如中心、路径、循环、团体和桥梁，因为这些图案提供了更易于阅读、比较和恢复的紧凑结构单元。本文研究了以图案为导向的图表说明，将其视为双向图-文本翻译任务，其中说明必须既保留足够的拓扑信息以便于图的恢复，又通过简洁的图案级描述表达图。我们展示了直接提示GPT-5.1通常通过枚举节点间连接生成可恢复的图表说明，但这些说明往往冗长且可能包含不一致的图案解释。为了解决这一问题，我们引入了Structurally Speaking，一种轻量级的结构化提示协议，指导显式连通性与图案级抽象之间的翻译。在基于合成图案的数据集上的实验表明，结构化提示生成的说明更短且图案一致性更强，同时保持了可比的图恢复效果。这些结果表明，显式的拓扑到图案的推理指导可以使LLM生成的图表说明更具可解释性，而无需对模型进行微调。

View on arXiv Download PDF AI Translation

cs.CL / 14 / 2609.10934

Using Semantic Uncertainty to Estimate Transition Relevance in Turn-taking

利用语义不确定性估计轮换中的转移相关性

Umair, Muhammad, de Ruiter, Jan P.

Abstract

Turn-taking is a fundamental mechanism that governs when interlocutors speak and listen. Although Spoken Dialogue Systems (SDS) exploit a range of linguistic, acoustic, and non-verbal cues, they produce ill-timed responses in unscripted interaction. A central challenge is anticipating Transition Relevance Places (TRPs), or opportunities, not obligations, for a listener to take the floor. Human listeners do not wait for turn endings; as an utterance unfolds, they use expectations about its developing meaning to anticipate TRPs and decide whether to take the floor. We examine whether these evolving expectations can be modeled through semantic uncertainty -- an LLM-derived measure of how strongly a turn so far constrains what may plausibly come next. To do so, we sample possible continuations of an ongoing turn and use changes in semantic dispersion to identify TRPs within turns. We evaluate this account on a dataset with TRP labels derived from real-time listener responses, rather than retrospective annotation. Our approach substantially outperforms prompt-based and fine-tuned text-only baselines, providing empirical support for the view that evolving semantic constraints inform perceived turn-taking opportunities in unscripted interaction.

Chinese Translation

轮换是调节交谈者何时发言和倾听的基本机制。尽管口语对话系统（SDS）利用了一系列语言、声学和非语言线索，但在非脚本化互动中，它们仍然会产生时机不当的响应。一个核心挑战是预测转移相关地点（TRPs），即听者获得发言机会的时刻，而非义务。人类听者并不会等待轮换结束；在话语展开的过程中，他们利用对其发展意义的期望来预测TRPs，并决定是否发言。我们研究这些不断变化的期望是否可以通过语义不确定性建模——一种基于大型语言模型（LLM）得出的度量，反映了迄今为止一个轮换在多大程度上限制了可能的后续内容。为此，我们对正在进行的轮换的可能延续进行采样，并利用语义分散度的变化来识别轮换中的TRPs。我们在一个包含来自实时听者反应的TRP标签的数据集上评估这种方法，而非回顾性标注。我们的研究方法显著优于基于提示和微调的文本基线，为不断变化的语义约束在非脚本化互动中影响感知的轮换机会这一观点提供了实证支持。

View on arXiv Download PDF AI Translation

cs.CL / 15 / 2609.10950

Robust Multimodal Sentiment Analysis with Incomplete Modalities via Semantic-aware Completeness based Reconstruction

基于语义感知完整性的鲁棒多模态情感分析

Choi, Han-Jun, Joe, Byunggill, Shin, Saim, Jang, Jin Yea

Abstract

Recent multimodal sentiment analysis studies increasingly adopt text-centric fusion approaches to exploit the rich sentiment information inherent in the textual modality. However, these approaches often suffer from performance degradation during inference due to partially missing or noisy data in real-world scenarios, especially when sentiment-related cues are missing. To address this issue, we introduce a new completeness estimation approach that quantifies the degree of sentiment-relevant information preserved in incomplete data to guide the reconstruction of missing semantics. Furthermore, we propose a training strategy that stabilizes multi-task learning while jointly optimizing sentiment prediction and completeness estimation. Extensive experiments and in-depth analyses on three benchmark datasets demonstrate that the proposed approach enables more accurate semantic reconstruction, leading to more precise sentiment prediction.

Chinese Translation

近期的多模态情感分析研究越来越多地采用以文本为中心的融合方法，以利用文本模态中固有的丰富情感信息。然而，这些方法在推理过程中往往因现实场景中部分缺失或噪声数据而导致性能下降，尤其是在情感相关线索缺失时。为了解决这一问题，我们提出了一种新的完整性估计方法，该方法量化了在不完整数据中保留的情感相关信息的程度，以指导缺失语义的重建。此外，我们还提出了一种训练策略，旨在稳定多任务学习，同时联合优化情感预测和完整性估计。在三个基准数据集上进行的广泛实验和深入分析表明，所提出的方法能够实现更准确的语义重建，从而导致更精确的情感预测。

View on arXiv Download PDF AI Translation

cs.CL / 16 / 2609.10993

Distribution-aware Language Neuron Identification in Multilingual Large Language Models

分布感知的多语言大语言模型中的语言神经元识别

Kim, Minjun, Won, Inho, Yuk, Junghun, Kim, Dongyeon, Kim, Jihyo, Lim, KyungTae

Abstract

Multilingual large language models (mLLMs) contain a small fraction of feed-forward neurons that are sensitive to particular languages, commonly termed language-specific neurons. Existing work measures language specificity using the entropy of each neuron's language-wise probabilities of being active, where a neuron is considered active when its activation value is positive. However, this approach may not fully capture the multilingual nature of mLLMs, where language representations are distributional and mutually related. We propose Distribution-aware Language Neuron selection, which leverages pairwise relationships between per-language activation distributions over the full activation range, including negative values. Specifically, we quantify each neuron's language specificity by clustering languages using pairwise overlap coefficients between their activation distributions. Across two mLLMs and two held-out corpora, our identifier more effectively isolates language-specific causal effects, yielding up to 4.9$\times$ higher on-target language damage per neuron while preserving off-target language performance.

Chinese Translation

多语言大语言模型（mLLMs）包含一小部分对特定语言敏感的前馈神经元，通常称为语言特定神经元。现有研究通过测量每个神经元的语言活跃概率的熵来评估语言特异性，其中当神经元的激活值为正时被视为活跃。然而，这种方法可能无法充分捕捉 mLLMs 的多语言特性，因为语言表示是分布式的并且相互关联。我们提出了分布感知的语言神经元选择方法，该方法利用每种语言的激活分布在整个激活范围（包括负值）上的成对关系。具体而言，我们通过使用激活分布之间的成对重叠系数对语言进行聚类，从而量化每个神经元的语言特异性。在两个 mLLMs 和两个保留语料库的实验中，我们的识别器更有效地隔离了语言特定的因果效应，使每个神经元在目标语言上的损害提高了最多 4.9 倍，同时保持了非目标语言的性能。

View on arXiv Download PDF AI Translation

cs.CL / 17 / 2609.10996

Rethinking Verbalized Confidence for LLM-as-a-Judge: A Compatibility Shift on Post-2025 Proprietary Models

重新思考大语言模型作为评判者的语言化信心：2025年后专有模型的兼容性转变

Hsiao, Yu-Chung

Abstract

Verbalized confidence, long dismissed as overconfident, coarse, and prone to round-number clustering, is now the more robust soft-scoring mechanism for LLM-as-a-Judge on top-tier proprietary models. Across SummEval, AggreFact, and HelpSteer2, spanning up to 18 LLMs, we show that the standard advice to prefer log-probabilities no longer holds on post-2025 models, where verbalized confidence is the better signal. We call this a compatibility shift. On top of a standard verbalized-confidence baseline, we introduce two new ingredients: an overconfidence advisory and self-debate. Together they improve calibration, score-distribution spread, and robustness to task subjectivity. We further observe a generation effect: post-2025 models accommodate these two additions with little balanced-accuracy cost, whereas pre-2025 models pay a measurable penalty. Compared with logprob-based G-Eval, verbalized confidence is the more subjectivity-robust soft signal on GPT-family top-tier releases. The shift is invisible under accuracy-only reporting. Rather than defaulting to hard predictions, we recommend broader use of soft scoring in LLM-as-a-Judge. More broadly, verbalized confidence has moved from a weaker substitute for logprobs to a practical soft-scoring mechanism for contemporary LLM judges.

Chinese Translation

语言化信心长期以来被视为过于自信、粗糙且容易出现整数字聚集的现象，但现在它已成为顶级专有模型中大语言模型作为评判者的更为稳健的软评分机制。在SummEval、AggreFact和HelpSteer2等数据集上，涵盖多达18个大语言模型，我们展示了在2025年后模型中，标准建议偏好对数概率不再适用，而语言化信心则是更好的信号。我们称之为兼容性转变。在标准的语言化信心基线之上，我们引入了两个新元素：过度自信建议和自我辩论。这两者共同改善了校准、评分分布的广度以及对任务主观性的稳健性。我们进一步观察到一种生成效应：2025年后模型在引入这两个元素时几乎没有平衡准确性成本，而2025年前的模型则付出了可测量的代价。与基于对数概率的G-Eval相比，语言化信心在GPT系列顶级版本中是更具主观性稳健性的软信号。这一转变在仅报告准确性时是不可见的。我们建议在大语言模型作为评判者的过程中更广泛地使用软评分，而不是默认进行硬预测。更广泛地说，语言化信心已从对数概率的较弱替代品转变为当代大语言模型评判者的实用软评分机制。

View on arXiv Download PDF AI Translation

cs.CL / 18 / 2609.11020

K/V-Cache Interventions Dissociate Representation Alignment from Persona Expression in Decoder-Only Language Models

K/V-缓存干预将表示对齐与个性表达在仅解码语言模型中区分开来

Sun, Yu, Lu, Mengyin, Feng, Cong, Lu, Guangming, Han, Huimin

Abstract

We study K/V-cache interventions -- transplanting a target-conditioned K/V trajectory into a source-persona generation -- as a structured surface for persona control in decoder-only language models. Across 13 intervention configurations applied to Llama-3.1-8B for a fixed source-to-target persona pair, we report two consistent dissociations between representation-level alignment and behavioral expression, plus a common failure under position perturbations. First, all layer-band K/V replacements (early, mid, late) achieve strong local V-space alignment (V-gap 0.91, 0.89, 0.84), but only mid-layer replacement (layers 9-20) combines substantial target-marker expression with preserved lexical diversity. Second, full and mid-layer replacement induce comparable alignment (V-gap 0.94 vs. 0.89) yet produce different lexical-diversity profiles (TTR 0.65 vs. 0.77). Third, position perturbations (lag and shuffle) apply distinct operations yet uniformly suppress target-persona expression -- a common behavioral failure rather than a strict dissociation. Representation-level similarity metrics alone are thus not sufficient predictors of downstream persona expression in the regimes we study; the K/V cache emerges as a controllable but structurally constrained intervention surface. Because the transplanted trajectory carries the target's own generated token history, we characterize the intervention as trajectory-level transplantation rather than isolated persona-representation injection; a same-token-sequence control, decoding an identical token sequence under source vs. target conditioning, reproduces the sign and layer localization of the L28 representational shift, indicating the shift is not explained solely by imported token history. These findings characterize representation-behavior dissociation in a high-signal setting rather than establishing universality across models or persona pairs.

Chinese Translation

我们研究了K/V-缓存干预——将目标条件的K/V轨迹移植到源个性生成中——作为在仅解码语言模型中进行个性控制的结构化表面。在对Llama-3.1-8B应用的13种干预配置中，我们报告了表示层级对齐与行为表达之间的两个一致性区分，以及在位置扰动下的共同失败。首先，所有层带K/V替换（早期、中期、晚期）都实现了强局部V空间对齐（V-gap 0.91, 0.89, 0.84），但只有中层替换（第9-20层）结合了显著的目标标记表达与保留的词汇多样性。其次，完全和中层替换诱导了可比的对齐（V-gap 0.94 vs. 0.89），但产生了不同的词汇多样性特征（TTR 0.65 vs. 0.77）。第三，位置扰动（延迟和洗牌）应用了不同的操作，但统一抑制了目标个性表达——这是一种共同的行为失败，而非严格的区分。因此，仅靠表示层级相似性度量不足以预测我们研究的领域中的下游个性表达；K/V缓存作为一个可控但结构上受限的干预表面出现。由于移植的轨迹携带了目标自身生成的标记历史，我们将干预特征化为轨迹级移植，而非孤立的个性表示注入；相同标记序列控制，在源与目标条件下解码相同的标记序列，重现了L28表示转变的符号和层级定位，表明该转变并非仅由导入的标记历史解释。这些发现表征了在高信号环境中的表示-行为区分，而不是在模型或个性对之间建立普遍性。

View on arXiv Download PDF AI Translation

cs.CL / 19 / 2609.11029

Rebalancing Token Importance in Language Models with TF-IDF Weighted Cross-Entropy Loss

在语言模型中使用TF-IDF加权交叉熵损失重新平衡标记重要性

Li, Zhijian, Larson, Stefan, Leach, Kevin

Abstract

Large language models are typically trained under uniform token weighting, which allows frequent and low-information tokens to dominate learning and can increase the tendency to memorize surface-level text spans. To address this, we present an information-weighted cross-entropy loss that rescales token-level contributions using TF-IDF statistics, emphasizing semantically informative tokens while down-weighting ubiquitous ones. Experiments on five decoder-only LLMs ranging from 1.1B to 13B parameters show consistent reductions in memorized substring length while preserving perplexity and downstream task performance. Under LoRA fine-tuning, TF-IDF reduces average substring memorization length by 14% across all five models. Under full-weight fine-tuning on TinyLLaMA 1.1B, the reduction reaches 58%. Our approach is architecture-agnostic and can be incorporated into existing training pipelines with less than 3% computational overhead, offering a lightweight and principled way to mitigate memorization without disrupting standard training dynamics.

Chinese Translation

大型语言模型通常在统一标记加权下进行训练，这使得频繁且信息量低的标记主导学习，并可能增加记忆表面文本片段的倾向。为了解决这个问题，我们提出了一种信息加权的交叉熵损失，该损失利用TF-IDF统计重新调整标记级别的贡献，强调语义信息丰富的标记，同时降低普遍存在的标记的权重。在五个仅解码的语言模型（LLM），参数范围从11亿到130亿的实验中，显示出在保持困惑度和下游任务性能的同时，记忆子字符串长度的一致减少。在LoRA微调下，TF-IDF在所有五个模型中平均减少了14%的子字符串记忆长度。在TinyLLaMA 1.1B的全权重微调下，减少幅度达到58%。我们的方法与架构无关，可以在现有的训练流程中集成，计算开销不足3%，提供了一种轻量且有原则的方法来减轻记忆而不干扰标准训练动态。

View on arXiv Download PDF AI Translation

cs.CL / 20 / 2609.11067

When Noise Fabricates Bias: The Fragility of LLM-as-a-Judge Bias Measurement under Noisy Text

当噪声制造偏见：噪声文本下 LLM 作为评判者的偏见测量脆弱性

Ryu, DongHyun, Lee, Jaehyeok, Hwang, YeongJun, Bak, JinYeong

Abstract

Large language models are increasingly used as judges to measure social bias in text, yet the passages they judge are often noisy, containing typos, informal spelling, and broken punctuation. The consequences of such surface noise for social bias measurement remain unclear. To investigate this question, we apply five realistic noise conditions at multiple intensity levels to 3,822 stereotype-related responses and compare the resulting bias judgments with those on the original text. We find that such surface noise does not degrade bias measurement symmetrically: it is far more likely to turn neutral judgments into biased ones than biased judgments into neutral ones, by up to a 120x margin. We further observe two non-obvious effects across four LLM judges: in the most fragile judge the distortion is at its purest at mild, realistic noise levels, where erasure is scarcest, and as judges grow robust it attenuates toward parity rather than reversing. Bias measured on noisy text is therefore systematically overestimated, most in the categories that matter most for fairness.

Chinese Translation

大型语言模型越来越多地被用作评判者来测量文本中的社会偏见，然而它们评判的段落往往存在噪声，包括拼写错误、非正式拼写和断裂的标点符号。这种表面噪声对社会偏见测量的影响仍不清楚。为探讨这一问题，我们在多个强度水平下对 3,822 个与刻板印象相关的回应应用了五种现实的噪声条件，并将所得的偏见判断与原始文本的判断进行比较。我们发现，这种表面噪声并不会对偏见测量造成对称性降级：它更可能将中立判断转变为偏见判断，而不是将偏见判断转变为中立判断，差距高达 120 倍。此外，我们还观察到四个 LLM 评判者之间的两个非明显效应：在最脆弱的评判者中，失真在温和、现实的噪声水平下最为纯粹，此时抹除现象最为稀少；而随着评判者的稳健性增强，失真趋向于平衡而非反转。因此，在噪声文本上测量的偏见被系统性地高估，尤其是在对公平性最为重要的类别中。

View on arXiv Download PDF AI Translation

cs.CL / 21 / 2609.11101

ProMediConv: Benchmarking Proactive Conversational Agents in Legal Dispute Mediation

ProMediConv：法律争议调解中的主动对话代理基准测试

Wei, Zesheng, Li, Mengfan, Liu, Wenhao, Zhang, Yixin, Wang, Zilei, Deng, Yang

Abstract

Dispute mediation is essential for maintaining social harmony and resilience, yet developing skilled mediators is costly and time-consuming. Existing LLM-based mediation research remains limited by unrealistic task formulations, low-fidelity datasets, and coarse evaluation metrics that obscure turn-by-turn dynamics. To address these gaps, we introduce ProMediConv, a novel benchmarking framework that models mediation as a proactive, multi-stage, and party-aware dialogue process incorporating 11 mediation strategies and four party behavior pattern (BP) states. Using 972 complete real-world cases, we construct a high-fidelity mediation dataset with utterance-level annotations of strategies and BP states. Furthermore, to better assess agent impact, we propose MAD (Mean Attribute Difference), a fine-grained metric that captures BP shifts throughout the dialogue. Leveraging this framework, we establish a comprehensive benchmark by evaluating diverse models alongside our tailored baseline ProMediAgent. Extensive empirical analyses reveal critical behavioral phenomena and underscore the persistent challenges current models face in dynamic, multi-party mediation. Ultimately, ProMediConv provides a rigorous foundation and a vital quantitative standard for advancing AI-assisted conflict resolution. Our dataset and codebase are accessible at https://github.com/ZsWei66/ProMediConv_repo.

Chinese Translation

争议调解对于维护社会和谐与韧性至关重要，但培养熟练的调解员成本高昂且耗时。现有基于大型语言模型（LLM）的调解研究受到不切实际的任务设定、低保真度数据集和模糊的评估指标的限制，这些因素掩盖了逐步对话的动态。为了解决这些问题，我们提出了ProMediConv，这是一种新颖的基准测试框架，将调解建模为一个主动的、多阶段的、关注各方的对话过程，包含11种调解策略和四种各方行为模式（BP）状态。利用972个完整的真实案例，我们构建了一个高保真度的调解数据集，包含策略和BP状态的逐语句注释。此外，为了更好地评估代理的影响，我们提出了MAD（平均属性差异），这是一种细粒度的指标，捕捉对话中的BP变化。借助这一框架，我们通过评估多种模型以及我们定制的基线ProMediAgent，建立了一个全面的基准。广泛的实证分析揭示了关键的行为现象，并强调了当前模型在动态多方调解中面临的持续挑战。最终，ProMediConv为推动人工智能辅助冲突解决提供了严格的基础和重要的定量标准。我们的数据集和代码库可在https://github.com/ZsWei66/ProMediConv_repo获取。

View on arXiv Download PDF AI Translation

cs.CL / 22 / 2609.11117

Overview of the NLPCC 2026 Shared Task 11: Agent-Based Experiment Reproduction from Scientific Papers

NLPCC 2026 共享任务 11 概述：基于代理的科学论文实验再现

Hong, Hanhua, Li, Yizhi, Huy, Luu Gia, Yang, Jian, Zhou, Ming, Lin, Chenghua

Abstract

Reproducibility is essential to scientific progress, yet the growing volume and complexity of scientific publications make exhaustive manual verification increasingly impractical. Although recent advances in large language model (LLM) agents enable automated experiment reproduction, existing evaluations largely focus on final repositories and are typically limited to machine learning (ML). We introduce AgentActionBench, a process-oriented benchmark for evaluating agent-based experiment reproduction across ML and AI4Science domains. Our framework uses an MCP-based Action Recorder to capture agents' behaviour throughout the reproduction process and evaluates the resulting traces with paper-specific rubrics. AgentActionBench contains 150 papers, including 120 ML papers and 30 AI4Science papers. A human-annotated subset covering 10% of the benchmark provides validation data, while model-assisted augmentation expands the full benchmark to more than 10,000 rubric items. Experimental results show that current systems remain limited, with execution as the primary bottleneck. Meanwhile, the strong Pearson and Spearman correlations between model-generated and human-annotated rubrics validate the reliability of our scalable rubric-generation approach.

Chinese Translation

可再现性对科学进步至关重要，但科学出版物的数量和复杂性不断增加，使得全面的手动验证变得越来越不切实际。尽管最近在大型语言模型（LLM）代理方面的进展使得自动化实验再现成为可能，但现有评估主要集中在最终存储库，并且通常局限于机器学习（ML）领域。我们介绍了 AgentActionBench，这是一个面向过程的基准，用于评估基于代理的实验再现，涵盖 ML 和 AI4Science 领域。我们的框架使用基于 MCP 的行动记录器来捕捉代理在再现过程中的行为，并使用特定于论文的评分标准评估生成的痕迹。AgentActionBench 包含 150 篇论文，其中包括 120 篇 ML 论文和 30 篇 AI4Science 论文。一个覆盖基准 10% 的人工标注子集提供了验证数据，而模型辅助的扩展将完整基准扩展到超过 10,000 个评分项目。实验结果表明，当前系统仍然有限，执行是主要瓶颈。同时，模型生成的评分与人工标注评分之间的强相关性（Pearson 和 Spearman）验证了我们可扩展评分生成方法的可靠性。

View on arXiv Download PDF AI Translation

cs.CL / 23 / 2609.11128

From Repetition to Recognition: Inductive Discovery of Disinformation Narratives

从重复到识别：虚假信息叙事的归纳发现

Upravitelev, Max, Solopova, Veronika, Yang, Jing, Jakob, Charlott, Tsiakalou, Alexandra, Foroutan, Neda, Schmitt, Vera

Abstract

In disinformation datasets, narratives are often understood as recurring interpretive patterns that group texts under narrative labels. Recent work formalized narrative mining as inductively inferring narrative labels from corpora, but its evaluation stays tied to predefined taxonomies, a closed-world setting that cannot capture narratives absent from the reference labels. We introduce a three-tier evaluation framework for unsupervised narrative label generation: recovery (against a corpus's own taxonomy), mining (against external label sets), and discovery (without predefined labels). Applying it, we compare clustering-based and graph-community-based pipelines across seven disinformation datasets, with human validation of discovery on two. The two families are complementary under automated metrics, but in a corpus with two prominent topics, clustering can reduce one topic to 2% of generated labels while graph-based pipelines stay balanced. Discovery validation also reveals many singletons (narrative labels derived from single claims, 30-62% of graph outputs), which clustering cannot produce. Annotators confirm many as recognizable disinformation narratives, suggesting that in open-world discovery the repetition assumed by narrative mining may be recognized outside the corpus, not within it. We release human-validated narrative candidate labels for the Climate Obstruction and PolyNarrative datasets to support taxonomy development and dataset extension.

Chinese Translation

在虚假信息数据集中，叙事通常被理解为将文本归类于叙事标签下的重复解释模式。近期的研究将叙事挖掘形式化为从语料库中归纳推断叙事标签，但其评估仍然依赖于预定义的分类法，这是一种封闭世界的设置，无法捕捉缺失于参考标签的叙事。我们引入了一个三层次的评估框架，用于无监督叙事标签生成：恢复（针对语料库自身的分类法）、挖掘（针对外部标签集）和发现（没有预定义标签）。通过应用该框架，我们比较了基于聚类和基于图社区的管道在七个虚假信息数据集上的表现，并在人类验证的基础上对其中两个数据集进行了发现验证。这两种方法在自动化指标下是互补的，但在一个包含两个显著主题的语料库中，聚类可能将一个主题减少到生成标签的2%，而基于图的管道则保持平衡。发现验证还揭示了许多单例（由单一主张派生的叙事标签，占图输出的30-62%），而聚类无法产生这些单例。注释者确认许多单例为可识别的虚假信息叙事，这表明在开放世界发现中，叙事挖掘所假设的重复可能在语料库外被识别，而不是在其内部。我们发布了经过人类验证的叙事候选标签，针对气候阻碍和多叙事数据集，以支持分类法的发展和数据集的扩展。

View on arXiv Download PDF AI Translation

cs.CL / 24 / 2609.11131

Rubric-Aligned Disentangled Evaluation of Human Simultaneous Interpreting

与评分标准对齐的人类同声传译解耦评估

Zhang, Ziyu, Nakamura, Satoshi

Abstract

Human simultaneous interpreting (SI) is commonly assessed with analytic rubrics separating meaning transfer, delivery quality, and temporal synchrony, yet no automatic metric is designed for rubric-aligned segment-level SI evaluation. We construct a professionally annotated corpus of 1,101 SI segments with scores for meaning transfer (LQ), delivery quality (EXP), and perceived latency (LAT). We show that structured LLM prompting and scalar supervision collapse rubric dimensions, yielding near-zero correlation with human ratings and strong cross-dimension coupling. To isolate supervision structure under identical backbone capacity, we introduce dual regression heads on a LoRA-adapted COMET-KIWI encoder. On a held-out talk-level test set, the model achieves Pearson correlations of 0.388 (LQ) and 0.301 (EXP), improving over frozen COMET-KIWI. Given low absolute rater agreement, we interpret results relative to human consistency and target stable ranking signals for formative assessment.

Chinese Translation

人类同声传译（SI）通常使用分析性评分标准来评估意义传递、交付质量和时间同步，然而尚无自动化指标专门用于与评分标准对齐的分段级SI评估。我们构建了一个专业注释的语料库，其中包含1,101个SI段落，并为其提供了意义传递（LQ）、交付质量（EXP）和感知延迟（LAT）的评分。我们展示了结构化的LLM提示和标量监督会导致评分维度的崩溃，从而与人类评分的相关性接近于零，并且跨维度耦合强烈。为了在相同的基础能力下隔离监督结构，我们在经过LoRA调整的COMET-KIWI编码器上引入了双重回归头。在一个保留的谈话级测试集上，该模型实现了0.388（LQ）和0.301（EXP）的Pearson相关性，相较于冻结的COMET-KIWI有所提升。鉴于评估者之间的绝对一致性较低，我们将结果相对于人类一致性进行解释，并针对形成性评估的稳定排名信号进行目标设定。

View on arXiv Download PDF AI Translation

cs.CL / 25 / 2609.11141

Can LLMs Normalize Databases? A Benchmark and Multi-Agent Framework for Schema Normalization

大型语言模型能否规范化数据库？一个用于模式规范化的基准测试和多智能体框架

Koh, Dong-Jae, Kim, Huisu, Yoon, SeongHwan, Jantsch, Lasse M., Lee, Chun-Hee, Lee, Seonghyeon, Suh, Young-Kyoon

Abstract

Large Language Models (LLMs) are increasingly used to generate structured outputs, but their reliability remains unclear when those outputs must satisfy database-level constraints. We study this issue through database normalization, involving reasoning about functional dependencies, lossless join decompositions, and inter-table constraints. We introduce a Database Normalization Benchmark (DNBENCH), comprising 3,275 samples for evaluating LLM-driven database normalization from 1NF to BCNF. DNBENCH uses a three-axis protocol to measure semantic equivalence, structural accuracy, and logical validity. Across Single, Complex, and Real World levels, DNBENCH uncovers recurring failures in dependency inference, schema decomposition, and inter-table constraint reconstruction. We further propose Multi-Agent Reasoning for Schemas (MARS), which separates evidence extraction, violation diagnosis, and decomposition planning from schema generation and verification. MARS improves the DNB-SCORE by 82.0% over the single-prompt baseline. All artifacts will be released upon acceptance.

Chinese Translation

大型语言模型（LLMs）越来越多地用于生成结构化输出，但当这些输出必须满足数据库级约束时，其可靠性仍不明确。我们通过数据库规范化研究这一问题，涉及对功能依赖、无损连接分解和表间约束的推理。我们引入了一个数据库规范化基准测试（DNBENCH），包含3,275个样本，用于评估LLM驱动的数据库从1NF到BCNF的规范化。DNBENCH采用三轴协议来测量语义等价性、结构准确性和逻辑有效性。在单一、复杂和真实世界层面上，DNBENCH揭示了依赖推断、模式分解和表间约束重建中的反复失败。我们进一步提出了模式的多智能体推理（MARS），将证据提取、违反诊断和分解规划与模式生成和验证分离。MARS在单提示基线的基础上提高了DNB-SCORE 82.0%。所有材料将在接受后发布。

View on arXiv Download PDF AI Translation

cs.CL / 26 / 2609.11149

A Fragility Spectrum for Recursive Language-Model Training

递归语言模型训练的脆弱性谱

Liu, Yangze, Han, Zhongyi

Abstract

Model-generated text is finding its way back into training corpora, and there is plenty of evidence that training on such data over and over collapses output diversity. Prior work has studied the phenomenon itself: which protocols and which data mixtures cause collapse. But different models behave very differently under the same process. We fix one recursive contamination protocol and let 13 publicly released checkpoints form an ecosystem that shares a common corpus for five generations. The unique 4-gram outcome after five generations ranges from 0.187 to 0.940 across checkpoints, a roughly five-fold spread: some models are barely touched, others degenerate into repetitive fragments. Changing the composition of the shared pool or mixing in human text keeps the Spearman correlation of the ordering at 0.91--0.97, and changing the random seed keeps it at 0.93--0.98. Whether a model collapses easily under recursive training is, then, a property of the checkpoint itself, and one that has gone largely unexamined. Parameter scale alone does not explain it, since a three-size ladder within one family is not monotonic in size, and none of the static indicators we tested predicts it either. What does work is cheap: let a model iterate on its own output for two or three generations, and its fragility in the larger ecosystem can be inferred from that alone. Collapse speed also responds to intervention. Tightening top-p, which cuts the low-probability tail at generation time, nearly stops collapse within three generations and stabilizes six checkpoints spanning the whole spectrum together, while data-side filtering slows collapse without stopping it.

Chinese Translation

模型生成的文本正在重新进入训练语料库，已有大量证据表明，反复在这些数据上训练会导致输出多样性的崩溃。之前的研究探讨了这一现象本身：哪些协议和数据混合会导致崩溃。但不同模型在相同过程中表现截然不同。我们固定一个递归污染协议，让13个公开发布的检查点形成一个共享同一语料库的生态系统，历经五代。经过五代后的独特4-gram结果在检查点之间的范围为0.187到0.940，约为五倍的差异：一些模型几乎没有受到影响，而其他模型则退化为重复片段。改变共享池的组成或混入人类文本使得排序的Spearman相关系数保持在0.91到0.97之间，而改变随机种子则保持在0.93到0.98之间。因此，模型在递归训练下是否容易崩溃，是检查点本身的一个特性，而这一特性在很大程度上未被研究。仅仅参数规模并不能解释这一现象，因为同一家族中的三种规模并不单调，并且我们测试的静态指标也没有预测能力。有效的方法是简单的：让模型在其自身输出上迭代两到三代，仅凭这一点就可以推断出其在更大生态系统中的脆弱性。崩溃速度也对干预作出反应。收紧top-p（在生成时削减低概率尾部）几乎在三代内停止崩溃，并使六个检查点在整个谱系中稳定，而数据侧过滤则减缓崩溃但并未停止。

View on arXiv Download PDF AI Translation

cs.CL / 27 / 2609.11192

FlexComp: One Model for Every Ratio in Context Compression

FlexComp：适用于上下文压缩的任意比率模型

Zhao, Kaiyan, Miao, Zhongtao, Aizawa, Akiko, Tsuruoka, Yoshimasa

Abstract

Soft context compression condenses a context into a few memory tokens that a frozen LLM consumes in place of the raw text, but existing compressors fix the compression ratio at training and inference: each deployed ratio requires a separately trained model, and the chosen ratio is applied uniformly to all inputs, whose actual needs vary drastically. We propose FlexComp, a method-agnostic framework that decouples the ratio from both training and deployment: Matryoshka-style training samples the memory budget $K$ per instance, turning one model into an any-ratio compressor, and the budget is then chosen per input by: (1) confidence-based cascade routing or (2) a lightweight learned $K$ predictor. Across ICAE, 500xCompressor, and SAC on MRQA, a single FlexComp model matches separately trained fixed-ratio specialists with minimal degradation. Cascade routing preserves over 98% of the mildest ratio's accuracy at up to 266x average compression; the $K$ predictor, in a single compression-decoding pass, reaches 158-236x within 0.7 F1 of the mildest ratio. At serving-scale batch sizes, the $K$ predictor cuts context KV cache by 50% and improves decoding throughput by 47%.

Chinese Translation

软上下文压缩将上下文浓缩为几个内存令牌，供冻结的LLM替代原始文本使用，但现有的压缩器在训练和推理时固定压缩比：每个部署的比率需要单独训练的模型，而所选比率对所有输入统一应用，然而实际需求差异巨大。我们提出了FlexComp，这是一种与方法无关的框架，将比率与训练和部署解耦：Matryoshka风格的训练为每个实例采样内存预算$K$，将一个模型转变为任意比率的压缩器，然后通过以下方式为每个输入选择预算：(1) 基于置信度的级联路由或(2) 轻量级学习的$K$预测器。在ICAE、500xCompressor和MRQA上的SAC中，单个FlexComp模型的性能与单独训练的固定比率专家相匹配，且降级最小。级联路由在高达266倍的平均压缩下保留了超过98%的最轻比率的准确性；$K$预测器在单次压缩-解码过程中，在最轻比率的0.7 F1范围内达到了158-236倍的压缩。在服务规模的批量大小下，$K$预测器将上下文KV缓存减少了50%，并提高了解码吞吐量47%。

View on arXiv Download PDF AI Translation

cs.CL / 28 / 2609.11202

Automated Identification of Competing Narratives in Political Discourse on Social Media

社交媒体上政治话语中竞争叙事的自动识别

Wildemann, Sergej, Elejalde, Erick

Abstract

Social media platforms have become central to shaping political discourse, serving as arenas where narratives form and evolve, influencing public opinion. Identifying and analyzing these narratives, particularly when they compete across different political ideologies, is crucial for understanding the dynamics of modern political communication. This paper presents an unsupervised framework for identifying and characterizing competing narratives in political discourse on social media, focusing on German politicians' tweets. The framework employs a multi-stage pipeline that integrates natural language processing techniques such as topic modeling, event detection, and event linking. By forming data into coherent stories and uncovering the distinct perspectives of user communities, the system is able to detect the key competing narratives, highlighting the divergent framings and conflicts surrounding trending political topics. Two case studies on polarizing political issues demonstrate the efficacy of the methodology, showcasing its ability to uncover and analyze divergent viewpoints. The findings contribute to the broader understanding of how narratives propagate within the digital public sphere and offer insights for policymakers, social media platforms, and researchers interested in monitoring political discourse.

Chinese Translation

社交媒体平台已成为塑造政治话语的核心，作为叙事形成和演变的场所，影响公众舆论。识别和分析这些叙事，特别是在不同政治意识形态之间竞争时，对于理解现代政治传播的动态至关重要。本文提出了一种无监督框架，用于识别和表征社交媒体上政治话语中的竞争叙事，重点关注德国政治家的推文。该框架采用多阶段流程，整合了自然语言处理技术，如主题建模、事件检测和事件链接。通过将数据形成连贯的故事并揭示用户社区的独特视角，该系统能够检测关键的竞争叙事，突出围绕热门政治话题的不同框架和冲突。对两项极化政治问题的案例研究展示了该方法的有效性，展示了其揭示和分析不同观点的能力。研究结果有助于更广泛地理解叙事在数字公共领域中的传播，并为政策制定者、社交媒体平台以及对监测政治话语感兴趣的研究人员提供了见解。

View on arXiv Download PDF AI Translation

cs.CL / 29 / 2609.11244

OmniHallu: Unified Hallucination Detection for Cross-Modal Comprehension and Generation in Multimodal Large Language Models

OmniHallu：跨模态理解与生成的统一幻觉检测

Yang, Jianjiang, Li, Peihang, Xu, Shanqing, Qian, Mengchen, Zhang, Lu, Luo, Meng

Abstract

While Multimodal Large Language Models (MLLMs) have achieved remarkable progress across diverse tasks, they suffer from hallucinations where generated outputs contradict or misrepresent input semantics. Existing research typically addresses hallucination detection within a single modality or task type, limiting generalizability. We introduce OmniHallu, a unified hallucination detection framework spanning both comprehension and generation tasks across image, video, and audio modalities. We contribute OmniHallu-Bench, a 10,000-sample benchmark with claim-level human annotations covering six cross-modal tasks: image-to-text (I2T), video-to-text (V2T), audio-to-text (A2T), text-to-image (T2I), text-to-video (T2V), and text-to-audio (T2A). Our multi-agent architecture decomposes model outputs into atomic claims, verifies them through modality-specific experts, and aggregates evidence via structured reasoning. We further propose a preference-optimized trainable verifier that approximates the multi-agent decision boundary, reducing expert calls by 66% with minimal performance loss. Extensive experiments reveal a consistent modality-dependent performance gradient and provide fine-grained insights into cross-modal hallucination patterns.

Chinese Translation

尽管多模态大型语言模型（MLLMs）在各种任务中取得了显著进展，但它们仍然面临幻觉问题，即生成的输出与输入语义相矛盾或错误表述。现有研究通常在单一模态或任务类型中处理幻觉检测，这限制了其泛化能力。我们提出了OmniHallu，一个统一的幻觉检测框架，涵盖图像、视频和音频模态下的理解和生成任务。我们贡献了OmniHallu-Bench，这是一个包含10,000个样本的基准数据集，具有针对六个跨模态任务的声明级人类注释：图像到文本（I2T）、视频到文本（V2T）、音频到文本（A2T）、文本到图像（T2I）、文本到视频（T2V）和文本到音频（T2A）。我们的多智能体架构将模型输出分解为原子声明，通过特定模态的专家进行验证，并通过结构化推理汇总证据。我们进一步提出了一种偏好优化的可训练验证器，近似多智能体决策边界，将专家调用减少66%，且性能损失最小。大量实验揭示了一致的模态依赖性能梯度，并提供了对跨模态幻觉模式的细致洞察。

View on arXiv Download PDF AI Translation

cs.CL / 30 / 2609.11246

Assessing the Reusability of Public Speech Resources for Low-Resource Languages: A Central Kurdish Case Study

评估低资源语言公共语音资源的可重用性：以中央库尔德语为例

Asadpour, Hiwa

Abstract

Kurdish is spoken by millions of people, but little technology can read it aloud. A recent study released three Kurdish voices, 35 hours of recorded speech, and a paper describing the work, all free to download. This review checks how well those public files match the paper. The research is careful about its limits, but the files contain several problems: a settings file lists equipment that was never used, test recordings are left unlabeled among training data, and a coding fault mishandles long numbers. The download page also claims a stronger result than the paper reports and recommends one voice for general use. That recommendation matters because Kurdish has major regional and written variation, while these voices were built from three people reading prepared texts. The process therefore removes much everyday and regional speech. English and German benefit from long traditions of dictionaries and linguistic description that help identify wrong pronunciations; Kurdish has far less such support, so software choices can go unchecked. The voices sound fluent, but they represent the reading styles of their speakers rather than Kurdish as a whole. Most of these issues can be fixed using information the team already has, without changing the reported results. Better records would mainly make the work easier for others, especially community linguists, to check and reuse. The license is the main exception: whether audiobook owners allow corrected versions to be shared will affect whether future Kurdish voices can build on this work or must start again.

Chinese Translation

库尔德语有数百万人使用，但很少有技术能够朗读它。最近的一项研究发布了三种库尔德语语音、35小时的录音以及一篇描述该工作的论文，所有内容均可免费下载。本次评审检查了这些公共文件与论文的匹配程度。研究谨慎地考虑了其局限性，但文件中存在多个问题：设置文件列出了从未使用的设备，测试录音在训练数据中未标记，编码错误导致长数字处理不当。下载页面还声称结果比论文报告的更强，并推荐一种语音用于一般使用。这一推荐很重要，因为库尔德语在地区和书写上存在重大差异，而这些语音是由三个人朗读准备好的文本构建的。因此，该过程去除了许多日常和地区性语言。英语和德语受益于长期的词典和语言描述传统，这有助于识别错误的发音；而库尔德语则缺乏这样的支持，因此软件选择可能未经过审查。这些语音听起来流利，但它们代表的是说话者的朗读风格，而不是整体的库尔德语。大多数问题可以利用团队已有的信息进行修复，而无需更改报告的结果。更好的记录主要会使其他人，特别是社区语言学家，更容易检查和重用这项工作。许可证是主要例外：有声书所有者是否允许分享修正版本将影响未来库尔德语语音是否能够在此基础上构建，或必须重新开始。

View on arXiv Download PDF AI Translation

cs.CL / 31 / 2609.11247

The Illusion of Balanced Multimodal Sentiment Analysis: Beyond the Limits of Optimization-Based Methods

平衡多模态情感分析的幻觉：超越基于优化的方法的局限

Kaffeza, Ioanna, Georgiou, Efthymios, Potamianos, Alexandros

Abstract

Multimodal Sentiment Analysis (MSA) remains constrained by modality imbalance, yet the field continues to rely on optimization-based balancing methods that promise more than they deliver. We provide three contributions: 1) a unified evaluation framework testing gradient and loss-based balancing strategies under controlled settings; 2) a theoretical diagnosis explaining why these methods fail, as they conflate fitting speed with discriminative contribution; and 3) a research agenda toward held-out discriminative modality valuation. Experiments on CMU-MOSI and CMU-MOSEI reveal three shortcomings: no strategy reliably outperforms Late Concatenation; performance is sensitive to hyperparameters; and even ratio calibration fails to yield consistent gains. The core issue is fundamental: loss is not utility, and gradients are not importance. Modality imbalance remains unresolved, motivating utility estimation from held-out performance.

Chinese Translation

多模态情感分析（MSA）仍然受到模态不平衡的限制，但该领域依然依赖于承诺超出实际效果的基于优化的平衡方法。我们提供了三个贡献：1）一个统一的评估框架，在受控条件下测试基于梯度和损失的平衡策略；2）一个理论诊断，解释这些方法为何失败，因为它们将拟合速度与判别贡献混为一谈；3）一个研究议程，旨在对保留的判别模态进行估值。在CMU-MOSI和CMU-MOSEI上的实验揭示了三个缺陷：没有策略能够可靠地超越晚连接（Late Concatenation）；性能对超参数敏感；即使是比例校准也未能带来一致的提升。核心问题是根本性的：损失并不是效用，梯度也不是重要性。模态不平衡问题仍未解决，这促使我们从保留的性能中进行效用估计。

View on arXiv Download PDF AI Translation

cs.CL / 32 / 2609.11302

Automatic Lyric Transcription for Greek Songs: Scaling and Task Composition Effects in Whisper Adaptation

希腊歌曲的自动歌词转录：Whisper适应中的规模和任务组合效应

Frangiadaki, Maria, Damianos, Dimitrios, Kritsis, Kosmas, Katsouros, Vassilis

Abstract

Automatic Lyric Transcription (ALT) remains substantially more challenging than speech recognition due to melodic variability, rhythmic irregularity, and accompaniment interference. This is heightened in low-resource languages like Greek, where no prior benchmark for ALT exists. We present the first controlled study of Whisper adaptation for Greek ALT, investigating model scaling effects, task composition via multitask training in transcribe-translate ratios, and two-stage speech-to-singing adaptation. We also curate a segment-level aligned singing dataset based on the Greek Audio Dataset (GAD) using source separation and CTC forced alignment. Results show that scaling consistently improves performance, while multitask learning acts as a beneficial regularizer primarily for smaller-capacity models. The 2-stage adaptation in Whisper Large-v3 achieves a Word Error Rate (WER) of 27.2%, a significant improvement over zero-shot baselines, establishing the first Greek ALT benchmark.

Chinese Translation

自动歌词转录（ALT）由于旋律变异、节奏不规则和伴奏干扰，仍然比语音识别具有更大的挑战性。这在像希腊这样的低资源语言中尤为突出，因为尚不存在ALT的先前基准。我们首次进行控制研究，探讨Whisper在希腊ALT中的适应性，研究模型规模效应、通过多任务训练的任务组合（转录-翻译比率）以及两阶段的语音到歌唱适应。我们还基于希腊音频数据集（GAD）使用源分离和CTC强制对齐，整理了一个段级对齐的歌唱数据集。结果表明，模型规模的增加始终能提高性能，而多任务学习主要对小容量模型起到有益的正则化作用。在Whisper Large-v3中的两阶段适应实现了27.2%的词错误率（WER），相比于零-shot基线有显著改善，建立了第一个希腊ALT基准。

View on arXiv Download PDF AI Translation

cs.CL / 33 / 2609.11322

MultiHuSE: A Multimodal Dataset for Humour Styles and Emotions

MultiHuSE：一种用于幽默风格和情感的多模态数据集

Kenneth, Mary Ogbuka, Khosmood, Foaad, Edalat, Abbas

Abstract

Computational recognition of verbal humour remains a challenging task, requiring an understanding of language, delivery style, emotions, and cultural context. Most existing approaches focus on binary classification and lack datasets that capture psychological dimensions of humour alongside variations in expression. We introduce MultiHuSE, a multimodal dataset comprising 2,407 high-definition videos of 50 demographically diverse actors performing 1,463 text samples across four psychological humour styles (affiliative, aggressive, self-enhancing, and self-deprecating), as well as neutral content. A subset is additionally annotated for underlying emotions. The dataset uniquely captures multiple actor interpretations of the same texts, enabling systematic analysis of expressive diversity. Baseline experiments show that multimodal fusion outperforms unimodal approaches (80.1% vs. 77.4% accuracy) in humour style classification, with particularly strong gains for affiliative humour (66% to 74%). While text provides the strongest individual signal, fusion models deliver meaningful improvements. We hope that MultiHuSE provides empirical support for psychological theories linking humour and emotion, while also opening new avenues for research in human communication, well-being, and AI-driven interaction. The dataset is available for academic use under an End-User Licence Agreement.

Chinese Translation

计算机对语言幽默的识别仍然是一项具有挑战性的任务，需要对语言、表达风格、情感和文化背景的理解。现有的大多数方法集中于二元分类，缺乏能够捕捉幽默的心理维度及其表达变化的数据集。我们介绍了MultiHuSE，这是一个多模态数据集，包含2,407个高定义视频，展示了50位具有不同人口统计特征的演员表演1,463个文本样本，涵盖四种心理幽默风格（亲和型、攻击型、自我提升型和自我贬低型），以及中性内容。数据集的一个子集还标注了潜在情感。该数据集独特地捕捉了多个演员对相同文本的不同解读，使得对表现多样性的系统分析成为可能。基线实验表明，在幽默风格分类中，多模态融合的表现优于单模态方法（准确率为80.1%对比77.4%），尤其是在亲和型幽默上有显著提升（从66%提高到74%）。虽然文本提供了最强的单一信号，但融合模型带来了有意义的改进。我们希望MultiHuSE为连接幽默与情感的心理理论提供实证支持，同时为人类沟通、幸福感和人工智能驱动的互动研究开辟新的途径。该数据集可根据最终用户许可协议供学术使用。

View on arXiv Download PDF AI Translation

cs.CL / 34 / 2609.11334

E-CONAN (Entailment, CONtradition And Neutral) Benchmarks: Arabic Textual Entailment and Natural Inference Datasets

E-CONAN（蕴含、矛盾与中立）基准：阿拉伯语文本蕴含与自然推理数据集

Jallad, Khloud AL, Ghneim, Nada, Rebdawi, Ghaida

Abstract

Natural Language Inference processes pairs of sentences to extract their semantic relations. NLI has been a hot research topic, integrated as a main component in other NLP applications. Despite significant advancements in textual inference across various languages all around the world, Arabic language still suffers from limited resources in this domain. To address this gap, this paper introduces E-CONAN benchmarks that are composed of sentences pairs from various sources: (1) automatically-translated pairs, (2) human-validated machine-translated pairs, (3) hand-crafted pairs from teaching Arabic as foreign language books, and (4) headlines pairs from different news channels containing rumors. E-CONAN contains two benchmark datasets, E-CONAN-2, a 2-way dataset (RTE) and E-CONAN-3, a 3-way dataset (NLI). Additionally, we have used E-CONAN benchmarks to evaluate 9 state-of-the-art multilingual pretrained models using zero-shot classification. Models were evaluated across the ArNLI, XNLI, and E-CONAN datasets. Results show that E-CONAN is a potentially valuable resource for evaluating model generalization and even for fine-tuning pre-trained models. Its diverse composition, derived from a combination of sources, offers a broader and more robust assessment compared to XNLI and ArNLI. In addition, we have evaluated 5 LLMs on E-CONAN-3 dataset. Moreover, we incorporated MARBERT as a representative Arabic-specific baseline and conducted performance evaluation comparison to demonstrate how Arabic-specific models scale against cross-lingual and LLM-based approaches on the E-CONAN benchmarks. Furthermore, we conducted detailed qualitative and quantitative error analysis to analyze frequent error patterns. E-CONAN benchmarks will be publicly available, we hope that it will enrich research community in Arabic textual entailment and natural language inference.

Chinese Translation

自然语言推理（Natural Language Inference, NLI）处理句子对以提取其语义关系。NLI一直是一个热门研究主题，作为其他自然语言处理（NLP）应用的主要组成部分。尽管在全球范围内各种语言的文本推理方面取得了显著进展，但阿拉伯语在这一领域仍然面临资源有限的问题。为了解决这一差距，本文介绍了E-CONAN基准，该基准由来自不同来源的句子对组成：（1）自动翻译的句子对，（2）经过人工验证的机器翻译句子对，（3）来自阿拉伯语作为外语教材的手工制作句子对，以及（4）来自不同新闻频道的包含谣言的标题句子对。E-CONAN包含两个基准数据集，E-CONAN-2，一个双向数据集（RTE），和E-CONAN-3，一个三向数据集（NLI）。此外，我们使用E-CONAN基准评估了9个最先进的多语言预训练模型，采用零样本分类方法。模型在ArNLI、XNLI和E-CONAN数据集上进行了评估。结果表明，E-CONAN是评估模型泛化能力和微调预训练模型的潜在有价值资源。其多样化的组成，源于多种来源的组合，提供了比XNLI和ArNLI更广泛和更稳健的评估。此外，我们在E-CONAN-3数据集上评估了5个大型语言模型（LLMs）。此外，我们将MARBERT作为代表性的阿拉伯语特定基线，并进行了性能评估比较，以展示阿拉伯语特定模型在E-CONAN基准上与跨语言和基于LLM的方法的比较表现。此外，我们进行了详细的定性和定量错误分析，以分析常见的错误模式。E-CONAN基准将公开发布，我们希望它能丰富阿拉伯语文本蕴含和自然语言推理的研究社区。

View on arXiv Download PDF AI Translation

cs.CL / 35 / 2609.11335

On the Impact of Anonymization on the Performance of Large Language Models

匿名化对大型语言模型性能的影响

Deußer, Tobias, Hahnbück, Max, Sparrenberg, Lorenz, Uelwer, Tobias, Bauckhage, Christian, Sifa, Rafet

Abstract

As large language models are increasingly deployed in sensitive domains, anonymizing input data to protect personally identifiable information has become a critical practice. However, the impact of this anonymization on model utility is not well understood. This paper presents a systematic empirical study of the trade-off between privacy and performance. We evaluate five prominent language models across eleven diverse benchmarks, comparing their performance on original versus pseudonymized inputs. Our results reveal that while anonymization generally degrades performance, the effect is highly nuanced. We find that more capable models, such as Qwen2.5-72B and GPT-4o mini, suffer the largest performance drops, suggesting a stronger reliance on specific entity information. The impact is also task-dependent: performance on TruthfulQA improves with anonymization, while retrieval-focused tasks like RGB experience a catastrophic decline. Further experiments show that reversible anonymization techniques that preserve entity uniqueness significantly outperform irreversible ones like redaction, and that explicitly prompting models about anonymization offers no discernible benefit. We conclude that anonymization is not a one-size-fits-all solution and must be co-designed with the model and task in mind to balance privacy and utility effectively. Our findings provide a crucial baseline for developing more robust, privacy-aware AI systems.

Chinese Translation

随着大型语言模型在敏感领域的日益应用，为了保护个人可识别信息，对输入数据进行匿名化已成为一项重要实践。然而，这种匿名化对模型效用的影响尚不清楚。本文呈现了一项系统的实证研究，探讨隐私与性能之间的权衡。我们评估了五个主要语言模型在十一项不同基准测试中的表现，比较了它们在原始输入与假名化输入上的性能。我们的结果显示，尽管匿名化通常会降低性能，但其影响非常细微。我们发现，更强大的模型，如 Qwen2.5-72B 和 GPT-4o mini，性能下降幅度最大，这表明它们对特定实体信息的依赖更强。影响也因任务而异：在 TruthfulQA 上，性能随着匿名化的进行而提高，而在以检索为重点的任务（如 RGB）中则出现了灾难性的下降。进一步的实验表明，能够保留实体唯一性的可逆匿名化技术显著优于不可逆的技术（如编辑），而明确提示模型关于匿名化的信息并没有带来明显的好处。我们得出结论，匿名化并不是一个适用于所有情况的解决方案，必须与模型和任务共同设计，以有效平衡隐私与效用。我们的研究结果为开发更强大、关注隐私的人工智能系统提供了重要的基准。

View on arXiv Download PDF AI Translation

cs.CL / 36 / 2609.11355

SEAR: Segment-Evidence-Aware Routing for Weak-to-Strong Multilingual Speech MCQ

SEAR：针对弱到强多语种语音多项选择题的分段证据感知路由

Le, Huy Hoang, Nguyen, Long-Bao, Dao, Minh Tri

Abstract

This paper describes our system for Task~2 of the second Multilingual Conversational Speech Language Model (MLC-SLM) Challenge. We adapt Qwen3-Omni-30B-A3B-Instruct with a segment-evidence-aware data and post-training pipeline. A language model converts timestamped ASR into coherent event spans, which are expanded by a boundary margin and cropped from the original recording. We then synthesize complementary semantic MCQs with Qwen3.6-27B and acoustic MCQs with Gemini~3.1 Flash-Lite, followed by structural, grounding, answer-consistency, and target-model trainability checks, yielding 359,825 verified MCQs across 21 language and accent variants. A text-only probe partitions the data into weak, text-answerable items used for supervised fine-tuning and strong, audio-dependent items used for reinforcement learning with Group Sequence Policy Optimization (GSPO), stabilized by debiased advantages, sequence-level importance correction, and dynamic filtering. Our system obtains 90.92% accuracy on the final official evaluation set.

Chinese Translation

本文描述了我们在第二届多语种对话语音语言模型（MLC-SLM）挑战赛任务2中的系统。我们采用了带有分段证据感知数据和后训练流程的Qwen3-Omni-30B-A3B-Instruct模型。一种语言模型将带时间戳的自动语音识别（ASR）转换为连贯的事件跨度，这些跨度通过边界边距扩展并从原始录音中裁剪。接着，我们使用Qwen3.6-27B合成互补的语义多项选择题（MCQs），并使用Gemini~3.1 Flash-Lite合成声学多项选择题，随后进行结构、基础、答案一致性和目标模型可训练性检查，最终生成359,825个经过验证的多项选择题，涵盖21种语言和口音变体。一个仅文本的探测器将数据分为弱文本可答的项目，用于监督微调，以及强音频依赖的项目，用于通过群体序列策略优化（GSPO）进行强化学习，该过程通过去偏优势、序列级重要性校正和动态过滤进行稳定。我们的系统在最终官方评估集上获得了90.92%的准确率。

View on arXiv Download PDF AI Translation

cs.CL / 37 / 2609.11399

TransClean: A Benchmark for Detecting and Extracting Clean Translations from Large Language Model Outputs

TransClean：用于检测和提取大型语言模型输出中干净翻译的基准

Qian, Shenbin, Scherrer, Yves

Abstract

Large language models (LLMs) are increasingly used for machine translation, yet their outputs often contain additional text beyond the translation itself, such as language labels, explanations or bilingual repetitions, which we term translation noise. Despite its prevalence, this problem lacks dedicated benchmarks and systematic study. We analyze over 790,000 translation outputs from 12 LLMs across 22 language pairs (LPs) and identify 12 recurring noise patterns, which we group into formatting and content noise. Building on the observed patterns, we construct TransClean, a controlled benchmark of 9,900 pairs of noisy and clean translation outputs, comprising 8,800 synthetically generated instances and 1,100 manually curated authentic instances. We evaluate two extraction approaches on the TransClean benchmark: 1) a span-based extraction method leveraging translation quality estimation models for span detection, and 2) an LLM-based extraction method that prompts an LLM to isolate the translation. Our benchmark and analysis provide the first systematic framework to evaluate and improve the cleanliness of LLM translation outputs.

Chinese Translation

大型语言模型（LLMs）在机器翻译中的应用日益增多，但它们的输出往往包含超出翻译本身的额外文本，例如语言标签、解释或双语重复，我们称之为翻译噪声。尽管这一问题普遍存在，但缺乏专门的基准和系统研究。我们分析了来自12个LLM在22个语言对（LPs）上的790,000多个翻译输出，并识别出12种重复出现的噪声模式，将其分为格式噪声和内容噪声。基于观察到的模式，我们构建了TransClean，这是一个包含9,900对噪声和干净翻译输出的受控基准，其中包括8,800个合成生成的实例和1,100个手动策划的真实实例。我们在TransClean基准上评估了两种提取方法：1）基于跨度的提取方法，利用翻译质量评估模型进行跨度检测；2）基于LLM的提取方法，通过提示LLM来隔离翻译。我们的基准和分析提供了第一个系统框架，以评估和改善LLM翻译输出的干净程度。

View on arXiv Download PDF AI Translation

cs.CL / 38 / 2609.11414

SWRouter: Similarity-Contractive Window Routing for Multi-Turn Large Language Model Conversations

SWRouter：用于多轮大语言模型对话的相似性收缩窗口路由

Wang, Yu, Li, Yuchen, Kong, Rui, Chen, Xinran, Chen, Jiamin, Cai, Hengyi, Wang, Shuaiqiang, Zhao, Jiashu, Zhang, Yulun, Lyu, Zhonghao, Xiong, Haoyi, Kong, Linghe, Huang, Jimmy Xiangji, Yin, Dawei

Abstract

Large language models exhibit complementary strengths, motivating routing methods that dispatch each query to the most suitable model. Although existing routers are effective in single-turn settings, they do not directly transfer to multi-turn dialogue, where routing performance critically depends on how historical context is segmented, retained, and incorporated into the current prompt. This introduces two fundamental challenges: preventing information loss and information confusion during context construction, and evaluating routing quality without conflating model selection with prompt construction quality. In this paper, we propose SWRouter, a Similarity-Contractive Window Router for multi-turn large language model routing. SWRouter combines a similarity-based context segmentation mechanism for prompt construction with a dual-metric evaluation framework that decouples construction accuracy from router performance. Experiments on multi-turn dialogue benchmarks demonstrate that SWRouter consistently surpasses strong baselines, achieving a 16.26% improvement in evaluation accuracy over the best individual large language model and an additional 8.22% gain over the Conv-ID Context baseline. Our results highlight that multi-turn large language model routing requires a joint design of context construction and evaluation, rather than a direct extension of single-turn routing methods.

Chinese Translation

大型语言模型展现出互补的优势，这促使了将每个查询分配给最合适模型的路由方法。尽管现有的路由器在单轮设置中有效，但它们并不能直接转用于多轮对话，其中路由性能在很大程度上依赖于历史上下文的分割、保留和融入当前提示的方式。这引入了两个基本挑战：在上下文构建过程中防止信息丢失和信息混淆，以及在不将模型选择与提示构建质量混淆的情况下评估路由质量。本文提出了SWRouter，一种用于多轮大语言模型路由的相似性收缩窗口路由器。SWRouter结合了一种基于相似性的上下文分割机制用于提示构建，以及一个双指标评估框架，该框架将构建准确性与路由器性能解耦。对多轮对话基准的实验表明，SWRouter始终超越强基线，在评估准确性上比最佳单个大语言模型提高了16.26%，并在Conv-ID Context基线之上额外获得了8.22%的提升。我们的结果强调，多轮大语言模型路由需要上下文构建与评估的联合设计，而不是单轮路由方法的直接扩展。

View on arXiv Download PDF AI Translation

cs.CL / 39 / 2609.11450

Cross-Lingual Clinical Annotation Projection as Constrained Text Generation: A Six-Language Study

作为受限文本生成的跨语言临床注释投影：六种语言研究

Rey-Blanes, Álvaro, Moreno-Barea, Francisco J., Veredas, Francisco J.

Abstract

Background: To determine whether cross-lingual clinical annotation projection can be formulated as a text-preserving, document-level generative task that produces verifiable character-level annotations for multilingual clinical corpus construction, and to characterize its robustness and computational trade-offs relative to candidate-based projection pipelines. Methods: We developed a constrained LLM projection workflow that inserts entity tags directly into immutable target-language text, followed by deterministic validation and character-offset reconstruction. We evaluated it alongside supervised candidate-span projection and hybrid ML-LLM refinement for transferring Spanish Disease, Symptom, and Procedure annotations into six languages. Evaluation used MultiClinAI gold standard with strict span matching and character-overlap F1 Results: Direct LLM projection achieved the strongest and most consistent performance. GLM 5.2 obtained a mean Strict F1 of 0.9201 across 18 language-entity combinations, while locally deployable Gemma4:31B achieved 0.9133. The best LLM configuration improved Strict F1 over the previous state of the art in all 18 settings, by 0.0564-0.1512, yielding 55,416 grounded mentions with reconstructed offsets. Conclusions: Direct LLM-based projection enables high-quality multilingual clinical annotation transfer and provides a practical approach for extending clinical NLP resources to languages with fewer annotated datasets and language-specific tools. Combined with local inference and deterministic validation, it can substantially reduce expert time and cost for multilingual clinical corpus construction.

Chinese Translation

背景：旨在确定跨语言临床注释投影是否可以被表述为一种文本保留的、文档级的生成任务，该任务生成可验证的字符级注释，以用于多语言临床语料库的构建，并描述其相对于基于候选的投影管道的稳健性和计算权衡。方法：我们开发了一种受限的LLM（大语言模型）投影工作流程，该流程直接将实体标签插入不可变的目标语言文本中，随后进行确定性验证和字符偏移重建。我们将其与监督的候选跨度投影和混合的机器学习-LLM精炼进行了评估，以将西班牙语的疾病、症状和程序注释转移到六种语言中。评估使用了MultiClinAI金标准，采用严格的跨度匹配和字符重叠F1。结果：直接的LLM投影实现了最强和最一致的表现。GLM 5.2在18种语言-实体组合中获得了0.9201的平均严格F1，而可本地部署的Gemma4:31B则达到了0.9133。最佳LLM配置在所有18种设置中将严格F1提高了0.0564-0.1512，生成了55,416个具有重建偏移的基础提及。结论：基于直接LLM的投影能够实现高质量的多语言临床注释转移，并为将临床自然语言处理资源扩展到注释数据集较少和缺乏语言特定工具的语言提供了一种实用的方法。结合本地推理和确定性验证，它可以显著减少专家在多语言临床语料库构建中的时间和成本。

View on arXiv Download PDF AI Translation

cs.CL / 40 / 2609.11460

ReGround: Grounding Reviewer Comments in Multimodal Evidence

ReGround：将审稿人评论与多模态证据相结合

Basch, Serwar, Qu, Lizhen, Gurevych, Iryna

Abstract

Reviewer comments naturally relate to specific parts of the reviewed paper, yet grounding these comments to the underlying evidence is difficult due to long multimodal documents. Existing benchmarks do not capture this setting and largely focus on explicit, information-seeking queries. We introduce ReGround, a large-scale dataset for reviewer comment grounding that links 10,267 reviewer comments to 16,274 evidence in the original anonymous submission of 3,656 papers. We build on a simple observation: author rebuttals often include explicit references to content of the submission used to address reviewer comments, providing a high-precision annotation source. We cast grounding as a retrieval task and evaluate a wide range of retrieval methods. Results show that retrieval over the entire paper content performs poorly, evidence-type inference is a major bottleneck, and multimodal evidence provides complementary signals that text alone misses. Our dataset exposes grounding reviewer comments as a difficult and practically important problem for scientific document understanding.

Chinese Translation

审稿人的评论自然与被审稿论文的特定部分相关，但由于文档较长且包含多种模态，将这些评论与基础证据相结合是困难的。现有的基准测试未能捕捉这种情境，主要集中在明确的信息检索查询上。我们引入了ReGround，这是一个大规模的审稿人评论基础数据集，将10,267条审稿人评论与3,656篇论文的原始匿名提交中的16,274条证据相链接。我们基于一个简单的观察：作者的反驳通常包含对用于回应审稿人评论的提交内容的明确引用，从而提供了高精度的注释来源。我们将基础工作视为检索任务，并评估了多种检索方法。结果表明，针对整个论文内容的检索效果不佳，证据类型推断是一个主要瓶颈，而多模态证据提供了文本所缺失的互补信号。我们的数据集揭示了将审稿人评论与证据相结合是一个困难且在科学文献理解中具有实际重要性的问题。

View on arXiv Download PDF AI Translation

cs.CL / 41 / 2609.11505

Structural priors for data-efficient language learning

数据高效语言学习的结构先验

Veitsman, Yana, Martins, Jonas Mayer, Lautenschlager, Jonathan, Beinborn, Lisa

Abstract

Efficient language learning requires methods to reduce the reliance on large data and computational resources. We investigate structural transfer: First training models on non-language data to induce useful priors for natural language. This approach is a form of weight initialization for multilingual language modeling. We evaluate transfer via next-token-prediction loss, weight shifts in the model, and downstream linguistic benchmarks. Several symbolic data types - notably music, probabilistic grammars, and cellular automata - yield lower language-modeling loss than random initialization. These gains coincide with smaller weight shifts during subsequent language training, suggesting that structural transfer positions models in a more favorable region of the parameter space. However, a lower loss does not translate consistently into better downstream linguistic performance, and transfer from non-language data is less efficient than additional language data. We conclude that non-language data can serve as a partial substitute for language data for the training objective of next-token prediction but does not reliably support broader linguistic generalization.

Chinese Translation

高效的语言学习需要减少对大量数据和计算资源的依赖的方法。我们研究了结构转移：首先在非语言数据上训练模型，以诱导对自然语言有用的先验。这种方法是一种多语言语言建模的权重初始化形式。我们通过下一个标记预测损失、模型中的权重变化以及下游语言基准来评估转移效果。几种符号数据类型——特别是音乐、概率语法和元胞自动机——在语言建模损失上比随机初始化产生更低的损失。这些增益与后续语言训练中较小的权重变化相吻合，表明结构转移将模型置于参数空间的更有利区域。然而，较低的损失并不总是能可靠地转化为更好的下游语言表现，且来自非语言数据的转移效率低于额外的语言数据。我们得出结论，非语言数据可以作为下一个标记预测训练目标的部分替代，但并不能可靠地支持更广泛的语言泛化。

View on arXiv Download PDF AI Translation

cs.CL / 42 / 2609.11545

Complex-Text Robustness Evaluation and Failure Diagnosis for Low-Resource Multilingual Text-to-Speech

低资源多语言文本到语音的复杂文本鲁棒性评估与故障诊断

Zuo, Tianlun, Zhang, Ziyu, Mao, Tingzhi, Fu, Zhonghua, Xie, Lei

Abstract

Low-resource multilingual text-to-speech (TTS) systems have expanded language coverage, but their robustness under complex text inputs remains insufficiently diagnosed. Existing evaluations mainly focus on naturalness, speaker similarity, and content consistency using regular test sentences, while providing limited insight into how multilingual TTS systems fail when handling challenging inputs such as numbers, dates, named entities, long sentences, code-switched expressions, and punctuation-related structures. This paper proposes a complex-text robustness diagnosis framework for low-resource multilingual TTS. We evaluate robustness from three dimensions: content consistency, language consistency, and generation stability. A multilingual robustness testing scheme is designed for Thai, Vietnamese, Swahili, and Indonesian, covering ordinary sentences and multiple types of complex text inputs. We further introduce automatic diagnostic metrics, including character error rate, language identification accuracy, and duration abnormal rate. To support input-level risk analysis before speech generation, we propose a lightweight Text Risk Score (TRS), which estimates synthesis risk from interpretable text features without manual annotation or model training. Experiments on three representative multilingual TTS systems, including OmniVoice, VoxCPM2, and MMS-TTS, show that complex text inputs expose systematic failure patterns that are not fully reflected by ordinary short-sentence evaluation. Different systems exhibit distinct vulnerabilities in number normalization, named entity handling, long-text generation, and code-switched input processing. Furthermore, TRS shows a positive correlation with content errors and duration abnormalities, demonstrating its usefulness as a low-cost pre-synthesis indicator for complex-text risk diagnosis in low-resource multilingual TTS.

Chinese Translation

低资源多语言文本到语音（TTS）系统扩展了语言覆盖范围，但其在复杂文本输入下的鲁棒性仍然缺乏充分的诊断。现有评估主要集中在自然性、说话者相似性和内容一致性上，使用常规测试句子进行评估，而对多语言 TTS 系统在处理挑战性输入（如数字、日期、命名实体、长句、代码切换表达和与标点相关的结构）时的失败情况提供的见解有限。本文提出了一种针对低资源多语言 TTS 的复杂文本鲁棒性诊断框架。我们从内容一致性、语言一致性和生成稳定性三个维度评估鲁棒性。为泰语、越南语、斯瓦希里语和印尼语设计了一种多语言鲁棒性测试方案，涵盖普通句子和多种类型的复杂文本输入。我们进一步引入自动诊断指标，包括字符错误率、语言识别准确率和持续时间异常率。为了支持语音生成前的输入级风险分析，我们提出了一种轻量级文本风险评分（Text Risk Score, TRS），该评分通过可解释的文本特征估计合成风险，无需手动标注或模型训练。在 OmniVoice、VoxCPM2 和 MMS-TTS 三个具有代表性的多语言 TTS 系统上的实验表明，复杂文本输入暴露了系统性故障模式，而这些模式在普通短句评估中并未充分反映。不同系统在数字归一化、命名实体处理、长文本生成和代码切换输入处理方面表现出不同的脆弱性。此外，TRS 与内容错误和持续时间异常呈正相关，证明其作为低成本复杂文本风险诊断的预合成指标的有效性。

View on arXiv Download PDF AI Translation

cs.CL / 43 / 2609.11620

A Training-Free, Alignment-Free Approach to Corporate Intelligence: Application to SEC Filings

一种无训练、无对齐的企业智能方法：应用于SEC文件

Delpech, Jean-François

Abstract

High-dimensional dense text embeddings and large language models face real obstacles in financial-disclosure analysis: context-window limits, hallucination risk, high computational cost, and the arbitrary rotation of vector spaces across independently trained models. We present a training-free, alignment-free framework for corporate intelligence built on deterministic sparse seed vectors. Hashing word strings into a fixed high-dimensional basis places all documents and all temporal epochs in a common coordinate system by construction, removing any need for training or alignment. Accumulating these seed vectors across sentence contexts yields corpus-specific semantic signatures that compose linearly, supporting sub-second document comparison, issuer fingerprinting, tracking of how an issuer's vocabulary shifts between filings, and thematic sentence extraction, all on ordinary CPU hardware. Demonstrating the approach on a multi-year corpus of SEC filings (10-K, 10-Q, 8-K), we show how material corporate events, among them Boeing's 737 MAX crisis, Intel's supply-chain disruptions, and Bunge's acquisition of Viterra, emerge as distinct, interpretable semantic profiles, each traceable to the exact source sentences that produced it, with no domain-specific training and no LLM inference.

Chinese Translation

高维稠密文本嵌入和大型语言模型在财务披露分析中面临实际障碍：上下文窗口限制、幻觉风险、高计算成本，以及独立训练模型之间向量空间的任意旋转。我们提出了一种基于确定性稀疏种子向量的无训练、无对齐的企业智能框架。通过将单词字符串哈希到固定的高维基底，所有文档和所有时间时期在构造上都置于一个共同的坐标系统中，从而消除了任何训练或对齐的需求。在句子上下文中累积这些种子向量产生特定语料库的语义特征，这些特征可以线性组合，支持亚秒级的文档比较、发行人指纹识别、跟踪发行人在不同文件之间词汇的变化，以及主题句提取，所有这些都可以在普通的CPU硬件上实现。在对多年的SEC文件（10-K、10-Q、8-K）进行的方法展示中，我们展示了诸如波音737 MAX危机、英特尔供应链中断和邦基收购维特拉等重要企业事件如何作为独特的、可解释的语义轮廓出现，每个轮廓都可以追溯到产生它的确切源句，无需特定领域的训练和大型语言模型推理。

View on arXiv Download PDF AI Translation

cs.CL / 44 / 2609.11687

Structured Transforms for Low-Overhead Quantization of Language Models

低开销量化语言模型的结构化变换

Cherniuk, Daria, Rudikov, Alexander, Kashin, Boris, Oseledets, Ivan

Abstract

We revisit Kashin-decomposition-based weight quantization for large language models and propose an improved algorithm with stronger convergence properties and structured, efficient orthogonal transforms. The method retains the core factorization of each weight into two components -- one with bounded infinity norm and the other with bounded infinity norm after an orthogonal transformation -- but replaces the dense random orthogonal matrix with a sign-randomized Discrete Cosine Transform (DCT), reducing the per-iteration cost from $\mathcal{O}(N^2)$ to $\mathcal{O}(N \log N)$. The proposed greedy algorithm with alternating updates guarantees the four-peak distribution required for stable 2-bit clustering of each factor and admits closed-form initialization of cluster centers, removing the multi-restart k-means bottleneck of prior work. Composed with OPTQ-style sequential error compensation and QuIP-style incoherence preprocessing, the resulting JAX pipeline is competitive with OPTQ, QuIP, QuIP-RG and a fine-tuning- and vector-quantization-free variant of QuIP# at 4-bit per channel on OPT, Llama-2 and Pythia, with favorable wall-clock scaling. The bounded-$\ell_\infty$ factorization is also notably robust: on stress configurations where QuIP variants diverge to four-digit perplexity (Pythia-6.9B) or abort with NaNs in LDL back-substitution (Mistral-7B), Kashin-DCT remains numerically stable and stays close to FP16 baseline. At inference time, each weight decomposes into two 2-bit factor codes per channel that are structurally suited to native-2-bit hardware.

Chinese Translation

我们重新审视基于Kashin分解的大型语言模型权重量化，并提出了一种改进算法，具有更强的收敛性和结构化、高效的正交变换。该方法保留了每个权重的核心分解为两个组成部分——一个具有有界的无穷范数，另一个在正交变换后也具有有界的无穷范数——但用符号随机化的离散余弦变换（DCT）替代了稠密随机正交矩阵，将每次迭代的成本从$ extmath{O}(N^2)$降低到$ extmath{O}(N extmath{log} N)$。所提出的贪婪算法通过交替更新保证了稳定的2位聚类所需的四峰分布，并允许聚类中心的闭式初始化，消除了先前工作的多重重启k均值瓶颈。结合OPTQ风格的顺序误差补偿和QuIP风格的非一致性预处理，所得到的JAX管道在OPT、Llama-2和Pythia上，在每通道4位的情况下，与OPTQ、QuIP、QuIP-RG以及不进行微调和向量量化的QuIP#变体具有竞争力，并且在时间上表现出良好的扩展性。值得注意的是，有界-$ extmath{l_ ext{infinity}}$分解在压力配置下表现出显著的鲁棒性：在QuIP变体发散至四位困惑度（Pythia-6.9B）或在LDL回代中因NaN而中止（Mistral-7B）时，Kashin-DCT保持数值稳定，并接近FP16基线。在推理时，每个权重分解为每通道两个结构上适合于原生2位硬件的因子代码。

View on arXiv Download PDF AI Translation

cs.CL / 45 / 2609.11699

Negative Self-Distillation: Learning to Reason by Avoiding Flaws

负自蒸馏：通过避免缺陷学习推理

Pei, Rongcan, Wei, Zhepei, Xu, Shuyao, Zhu, Xinyu, Chen, Wei-Lin, Meng, Yu

Abstract

On-Policy Self-Distillation (OPSD) has emerged as a popular paradigm for large language model (LLM) self-improvement, allowing models to act as their own teachers by leveraging privileged information such as ground-truth solutions. However, recent findings indicate that OPSD can severely degrade the performance of LLMs on complex reasoning tasks: By forcing the student to imitate an artificially confident reasoning trace conditioned on privileged information, OPSD inadvertently suppresses expressions of uncertainty and penalizes the exploratory, self-corrective behaviors required to solve challenging problems. To address this, we introduce Negative Self-Distillation (NSD), a new framework that optimizes LLMs by diverging from flawed reasoning rather than imitating privileged solutions. Instead of relying on ground-truth answers or external supervision, NSD uses the model itself to generate a question-specific negative condition (eg, acting as a ``careless reasoner'') and pushes the student's distribution away from this self-generated negative teacher. Naively applying unlearning objectives to achieve this divergence is problematic, as flawed reasoning tokens are confounded with basic linguistic tokens; indiscriminately penalizing both risks catastrophically degrading the model's foundational language capabilities. We resolve this by designing a dynamic gating mechanism that automatically identifies and isolates reasoning-critical tokens, ensuring gradient updates target only behavioral flaws while preserving the model's linguistic priors. Empirically, NSD consistently outperforms OPSD and other label-free, self-bootstrapping reinforcement learning (RL) baselines.

Chinese Translation

在线自蒸馏（On-Policy Self-Distillation, OPSD）已成为大型语言模型（Large Language Model, LLM）自我改进的热门范式，使模型能够利用诸如真实解答等特权信息作为自己的教师。然而，最近的研究发现，OPSD可能会严重降低LLM在复杂推理任务上的表现：通过强迫学生模仿基于特权信息的人工自信推理轨迹，OPSD无意中抑制了不确定性的表达，并惩罚了解决挑战性问题所需的探索性自我纠正行为。为了解决这一问题，我们引入了负自蒸馏（Negative Self-Distillation, NSD），这是一个通过偏离缺陷推理而非模仿特权解答来优化LLM的新框架。NSD不依赖于真实答案或外部监督，而是利用模型自身生成特定问题的负条件（例如，充当“粗心的推理者”），并推动学生的分布远离这个自生成的负教师。简单地应用去学习目标以实现这种偏离是有问题的，因为缺陷推理标记与基本语言标记混淆；不加区分地惩罚这两者可能会灾难性地削弱模型的基础语言能力。我们通过设计一个动态门控机制来解决这个问题，该机制自动识别和隔离推理关键标记，确保梯度更新仅针对行为缺陷，同时保留模型的语言先验。从经验上看，NSD在性能上始终优于OPSD及其他无标签自引导强化学习（Reinforcement Learning, RL）基线。

View on arXiv Download PDF AI Translation

cs.CL / 46 / 2609.11724

The Eloquence submission for Task 2 of the Interspeech 2026 MLC-SLM challenge

Eloquence团队在2026年Interspeech MLC-SLM挑战赛任务2中的提交

Luque, Jordi, Concina, Lorenzo, Matassoni, Marco, Brutti, Alessio, Vella, Filippo

Abstract

This paper details the Eloquence team's approach to Task 2 of the 2nd MLC-SLM challenge at Interspeech 2026, which involves multilingual Multiple-Choice Question Answering (MCQA) across 21 languages. Three approaches are explored. First, we fine-tune Voxtral-Mini-3B via LoRA with cross-lingual data augmentation, ASR transcript augmentation and timestamp-aware audio cropping, achieving 0.72 macro-accuracy on evaluation Phase 2. Second, we apply multimodal in-context learning (ICL) to the frozen Voxtral-24B model to correct a strong label bias, reaching 0.81, our best result. Third, a training-free retrieval system based on a three-layer voice-anchored memory combining acoustic identity, semantic content, and a knowledge graph achieves 0.68. All three systems substantially outperform the official baseline.

Chinese Translation

本文详细介绍了Eloquence团队在2026年Interspeech第二届MLC-SLM挑战赛任务2中的方法，该任务涉及21种语言的多语言多项选择问答（MCQA）。我们探索了三种方法。首先，我们通过LoRA对Voxtral-Mini-3B进行微调，结合跨语言数据增强、ASR转录增强和时间戳感知音频裁剪，在评估阶段2中实现了0.72的宏观准确率。其次，我们对冻结的Voxtral-24B模型应用多模态上下文学习（ICL），以纠正强标签偏差，达到了0.81，这是我们的最佳结果。第三，基于三层语音锚定记忆的无训练检索系统，结合声学身份、语义内容和知识图谱，达到了0.68。所有三种系统的表现均显著优于官方基线。

View on arXiv Download PDF AI Translation

cs.CL / 47 / 2609.11739

LOCUS: Task-Aware Low-Rank Post-Training for Token-Efficient Language Generation

LOCUS：任务感知的低秩后训练以实现高效的语言生成

Zhao, Dongfang

Abstract

Large language model serving costs scale directly with output sequence length, yet standard preference alignment often inflates response verbosity without improving utility. We study whether the parameterization of post-training updates affects generation length: low-rank subspaces alter sequence length without modifying the alignment loss. We present LOCUS, a method that selects a task-aware low-rank adaptation subspace to minimize output-token cost subject to a utility constraint. Within this subspace, post-training retains the native preference objective with a frozen backbone. Across Anthropic HH-RLHF dialogue preferences, we evaluate two $\sim$3B decoder backbones, Pythia-2.8B and Qwen2.5-3B, against protocol-matched full-parameter DPO and DrDPO branches and the released SamPO checkpoint. LOCUS reduces continuation length by up to 39.84\% on Pythia-2.8B and by 14.87--17.58\% on Qwen2.5-3B while updating only 0.24--0.28\% of model parameters, with no material change in the internal preference diagnostic.

Chinese Translation

大型语言模型的服务成本与输出序列长度直接相关，然而标准的偏好对齐往往会在不提高效用的情况下增加响应的冗长性。我们研究了后训练更新的参数化是否会影响生成长度：低秩子空间在不修改对齐损失的情况下改变序列长度。我们提出了LOCUS，一种选择任务感知的低秩适应子空间的方法，以在效用约束下最小化输出标记成本。在这个子空间内，后训练保持了原生的偏好目标，同时冻结了主干网络。在Anthropic HH-RLHF对话偏好的基础上，我们评估了两个约3B解码器主干网络，Pythia-2.8B和Qwen2.5-3B，针对协议匹配的全参数DPO和DrDPO分支以及发布的SamPO检查点。LOCUS在Pythia-2.8B上将续写长度减少了最多39.84%，在Qwen2.5-3B上减少了14.87%到17.58%，同时仅更新了0.24%到0.28%的模型参数，内部偏好诊断没有实质性变化。

View on arXiv Download PDF AI Translation

cs.CL / 48 / 2609.11758

RAG-Safety-Bench: Reliable Evaluation of Retrieval-Augmented LLM Safety

RAG-Safety-Bench：检索增强型大语言模型安全性的可靠评估

Saravanan, Adithiyan Rajan Indira, Fraser, Kathleen C.

Abstract

Allowing large language models (LLMs) to retrieve information from a set of trusted documents can increase reliability and reduce hallucination. However, recent work has demonstrated that retrieval-augmented generation (RAG) can have unintended side effects on the overall safety of the generated responses, when prompted for harmful or dangerous content. A clearer understanding of the mechanisms leading to this result is needed, as increasing numbers of end users turn to RAG to incorporate corporate documents and knowledge bases into LLM-based systems. We introduce RAG-Safety-Bench, a benchmark to measure the safety impact of RAG on LLM models. By removing the confounding effect of retriever quality, and cleanly separating the problem into four conditions -- non-RAG, RAG with an oracle document containing the answer to the harmful request, RAG with documents related to the harmful request but without the specific answer, and RAG with random, safe documents -- the benchmark isolates the impacts of different factors in the observed safety degradation. We report results across five open-source LLMs, showing an inverse relationship between benign and unsafe capability, strong evidence that baseline safety guardrails do not lead to downstream safety guarantees in the RAG case, and model-specific support for previous findings that even benign documents can lead to unsafe generation in retrieval-enabled systems.

Chinese Translation

允许大型语言模型（LLMs）从一组可信文档中检索信息可以提高可靠性并减少幻觉。然而，近期研究表明，当被提示生成有害或危险内容时，检索增强生成（RAG）可能对生成响应的整体安全性产生意想不到的副作用。需要更清晰地理解导致这一结果的机制，因为越来越多的最终用户开始使用RAG将企业文档和知识库整合到基于LLM的系统中。我们介绍了RAG-Safety-Bench，这是一个用于测量RAG对LLM模型安全性影响的基准。通过消除检索器质量的混淆效应，并将问题清晰地分为四种条件——非RAG、包含有害请求答案的oracle文档的RAG、与有害请求相关但没有具体答案的文档的RAG，以及随机安全文档的RAG——该基准隔离了观察到的安全性下降中不同因素的影响。我们报告了五个开源LLM的结果，显示出良性和不安全能力之间的反向关系，强有力的证据表明基线安全防护措施在RAG情况下并未导致下游安全保障，并且对以往发现的模型特定支持，即即使是良性文档也可能在启用检索的系统中导致不安全生成。

View on arXiv Download PDF AI Translation

cs.CL / 49 / 2609.11762

Component-Aware Differential Privacy for Federated Multilingual Speech-LLMs

面向组件的差分隐私在联邦多语言语音大语言模型中的应用

Luque, Jordi, López, Fernando, Sant, Aleix

Abstract

Per-layer differential privacy (DP) clipping improves gradient fidelity in federated learning by allocating per-matrix clipping budgets proportional to parameter count. We show that this recipe breaks for speech large language models (speech-LLMs), when the acoustic encoder and the language decoder differ by an order of magnitude in update norm. Single-pool per-layer methods suffer \emph{cross-component budget collapse}, dragging word error rate (WER) far from flat global clipping or collapsing training entirely. When the norm imbalance is milder, adaptive single-pool methods partially recover, confirming that collapse severity scales with the inter-component norm ratio. We empirically diagnose the root cause across six per-layer methods and three speech-LLM architectures. We then propose \emph{$\alpha$-split}, a two-pool allocation that normalises encoder and LLM parameters into independent pools, and show that joint $\ell_2$ sensitivity and the original $(\varepsilon,\delta)$-DP guarantee are unchanged. At architecture-calibrated $\alpha$, our method recovers WER utility compared to flat DP, while granting the encoder $4.47{\times}$ tighter per-component noise protection against speaker voice-based gradient-inversion attacks at only $+2.6\%$ LLM noise overhead.

Chinese Translation

逐层差分隐私（DP）裁剪通过根据参数数量分配每个矩阵的裁剪预算，从而提高了联邦学习中的梯度保真度。我们发现，当声学编码器和语言解码器在更新范数上相差一个数量级时，这一方法在语音大语言模型（speech-LLMs）中失效。单池逐层方法遭遇了 extit{跨组件预算崩溃}，使得词错误率（WER）远离平坦的全局裁剪，甚至完全崩溃训练。当范数不平衡较轻时，自适应单池方法部分恢复，证实崩溃的严重程度与组件间范数比率成比例。我们在六种逐层方法和三种语音大语言模型架构中进行了实证诊断，找出了根本原因。然后，我们提出了 extit{$eta$-分割}，一种将编码器和大语言模型参数规范化为独立池的双池分配方法，并表明联合$ ext{l}_2$灵敏度和原始的$( ext{ε}, ext{δ})$-DP保证保持不变。在经过架构校准的$eta$下，我们的方法在与平坦DP相比时恢复了WER效用，同时为编码器提供了$4.47{ imes}$更紧凑的每组件噪声保护，以抵御基于说话者声音的梯度反演攻击，仅增加了$+2.6\%$的LLM噪声开销。

View on arXiv Download PDF AI Translation

cs.CL / 50 / 2609.11769

Recognizing Is Not Reversing: A Controlled Inversion Test of Fact-Preserving News Framing

识别并非逆转：事实保留新闻框架的受控逆转测试

Liu, Yi

Abstract

Large language models (LLMs) are increasingly used to analyze and rewrite news, yet current framing studies mainly evaluate generation, detection, or whether rewritten text appears more neutral. They do not directly show whether a model can undo a known framing transformation while keeping the facts fixed. We introduce a controlled inversion test over three established textual realizations of framing: evaluative lexis, agency realization, and information salience. Across 60 news articles and three intervention strengths, this yields 540 paired variants with preserved atomic facts and recorded edits. Across Qwen, DeepSeek, and Kimi, factual preservation remains near 0.84, whereas intervention reversal is 0.044--0.068. Even when both framing type and direction are recognized correctly, pooled reversal reaches 0.071. These results reveal a clear separation between factual fidelity, framing recognition, and framing inversion: recognizing how an article is framed does not imply that the framing can be undone.

Chinese Translation

大型语言模型（LLMs）越来越多地用于分析和改写新闻，但当前的框架研究主要评估生成、检测或改写文本是否显得更中立。它们并未直接展示模型是否能够在保持事实不变的情况下撤销已知的框架转变。我们引入了一种针对三种已建立的框架文本实现的受控逆转测试：评估性词汇、代理实现和信息显著性。在60篇新闻文章和三种干预强度下，这产生了540对保留原子事实和记录编辑的变体。在Qwen、DeepSeek和Kimi中，事实保留率接近0.84，而干预逆转率为0.044至0.068。即使在框架类型和方向都被正确识别的情况下，合并逆转率也达到了0.071。这些结果揭示了事实忠实度、框架识别和框架逆转之间的明显分离：识别一篇文章的框架并不意味着该框架可以被撤销。

View on arXiv Download PDF AI Translation

cs.CL / 51 / 2609.11770

The widening evaluation gap in medical large language model research 2023 to 2026

2023至2026年医学大型语言模型研究中的评估差距扩大

Tareaf, Raad Bin, Al-Rajab, Murad, Loucif, Samia

Abstract

Large language models are superseded every few quarters; clinical evidence takes years. We asked whether medical research is keeping pace with the systems it evaluates. PubMed returned 11,628 records for January 2023 to June 2026 across fourteen clinical domains, growing 45-fold; 2.5% used a randomised, controlled or prospective design. Evaluation lag, from a study's newest named model release to its own publication, widened from 1.33 to 6.08 quarters. Because discontinued models age mechanically, we benchmarked this against a counterfactual holding model composition fixed: migration to newer systems offset only 56% of the drift (95% CI 50-65). Randomised trials evaluated models a median 4.6 quarters older than other designs (P = 3 x 10^-19), yet among studies naming a model still under development no design differed from any other; 62% of randomised trials evaluated a discontinued family. Rigour and currency are in tension, and that tension reflects model selection rather than research timelines.

Chinese Translation

大型语言模型每隔几个季度就会被新模型取代，而临床证据的获取则需要数年时间。我们探讨了医学研究是否与其评估的系统保持同步。根据PubMed的数据，从2023年1月到2026年6月，在十四个临床领域中共返回了11,628条记录，增长了45倍；其中仅有2.5%的研究采用了随机、对照或前瞻性设计。评估滞后，即从研究中最新命名的模型发布到其自身发表的时间，已从1.33季度扩大至6.08季度。由于停用的模型在时间上自然衰老，我们将其与一个假设情境进行了基准比较，假设模型组成保持不变：迁移到新系统仅抵消了56%的漂移（95%置信区间50-65）。随机试验评估的模型中位数比其他设计的模型大4.6个季度（P = 3 x 10^-19），然而在命名仍在开发中的模型的研究中，没有任何设计之间存在差异；62%的随机试验评估的是已停用的模型系列。研究的严谨性与时效性之间存在张力，而这种张力反映的是模型选择而非研究时间线。

View on arXiv Download PDF AI Translation

cs.CL / 52 / 2609.11786

Beyond Word Error Rate: A Switch Aware Evaluation of ASR and Audio Language Models on English Yoruba Code-Switched Speech

超越词错误率：对英语-约鲁巴语代码切换语音的自动语音识别和音频语言模型的切换感知评估

Okocha, Chibuzor, Grant, Christan Earl

Abstract

Automatic speech recognition (ASR) systems and audio language models (audio LMs) now report low error rates on monolingual benchmarks, but their behavior on code switched speech in low resource, diacritic rich languages remains poorly characterized. We present a switch aware evaluation of eleven modern systems (six ASR models and five audio LMs) on English Yoruba code-switched speech, using a deterministic 2000 utterance evaluation set and a shared scoring pipeline. Beyond word error rate (WER), we report switch localized diagnostics: a switch entry token error rate (SETER), windowed switch point error rates, language specific error rates, and a diacritic insensitive WER. Our central finding is that aggregate WER hides code switching behavior. The best system by WER (an ASR model) is statistically indistinguishable from a leading audio LM on WER, yet the audio LM is significantly better on every switch localized metric. Across faithful systems, Yoruba token recognition collapses (error 0.97 for almost all systems) while English tokens are recognized far better, and errors concentrate sharply at switches into Yoruba. Several generative audio LMs fail as exact transcribers, producing translation, verbosity, and prompt leakage that are strongly prompt dependent. We release manifests, metric implementations, and evaluation scripts to support reproducible, switch aware benchmarking for African code switched speech.

Chinese Translation

自动语音识别（ASR）系统和音频语言模型（audio LMs）在单语基准测试中报告了较低的错误率，但它们在低资源、带有变音符号的语言中的代码切换语音表现仍然缺乏充分的表征。我们对十一种现代系统（六个ASR模型和五个音频LM）在英语-约鲁巴语代码切换语音上的表现进行了切换感知评估，使用了一个确定性的2000个发话样本评估集和一个共享评分管道。除了词错误率（WER）之外，我们报告了切换局部诊断：切换入口标记错误率（SETER）、窗口切换点错误率、语言特定错误率以及对变音符号不敏感的WER。我们的主要发现是，整体WER掩盖了代码切换行为。在WER上表现最佳的系统（一个ASR模型）在WER上与领先的音频LM在统计上没有显著差异，但在每个切换局部指标上，音频LM的表现显著更好。在忠实系统中，约鲁巴语标记的识别崩溃（几乎所有系统的错误率为0.97），而英语标记的识别效果则好得多，错误在切换到约鲁巴语时急剧集中。几个生成型音频LM作为精确转录工具的表现不佳，产生了翻译、冗长和提示泄漏，这些都强烈依赖于提示。我们发布了清单、指标实现和评估脚本，以支持可重复的、切换感知的非洲代码切换语音基准测试。

View on arXiv Download PDF AI Translation

cs.CL / 53 / 2609.11838

Target leakage, not model class, explains reported accuracy in survey-based cardiovascular screening: a leakage-tiered audit of glass-box and tabular foundation models

目标泄漏，而非模型类别，解释了基于调查的心血管筛查中的报告准确性：对透明模型和表格基础模型的泄漏分层审计

Tareaf, Raad Bin, Al-Rajab, Murad, Loucif, Samia, Ellaham, Samer, Schmitz, Cedric

Abstract

Cardiovascular screening models trained on national health surveys routinely report areas under the receiver operating characteristic curve (AUROC) near 0.89. We asked whether that accuracy reflects learning or target leakage, whether tabular foundation models change the answer, and whether the properties deployment requires survive joint examination. We benchmarked ten classifiers spanning linear, tree-ensemble, neural, glass-box, and tabular foundation classes for prevalent myocardial infarction in 442,067 respondents of the 2022 Behavioral Risk Factor Surveillance System across five feature tiers of decreasing leakage risk. Each was audited for discrimination, calibration, fairness at an explicit screening threshold, conformal coverage, explanation faithfulness, and inference cost, then applied -- models and thresholds frozen -- to 430,755 respondents of 2023. Removing two post-diagnostic features cost every model 0.049-0.051 AUROC, collapsing the field into a 0.0045-wide band. The glass-box explainable boosting machine was non-inferior to every alternative within a pre-specified 0.005 margin while scoring the cohort roughly 104 times faster than the strongest foundation model. One threshold detected 75.4% of women's infarctions against 89.0% of men's; editing the model's shape functions reduced the gap to 0.010. Marginal conformal prediction gave 0.86 coverage to men and 0.82 to adults over 60; Mondrian calibration repaired every stratum. Frozen models transported within 0.002 AUROC. Reported headroom in this literature is a property of the feature set, not the learner. Transparency cost nothing measurable and made fairness repair and uncertainty conditioning directly auditable. Evaluation practice, not model capacity, is the binding constraint.

Chinese Translation

基于国家健康调查训练的心血管筛查模型通常报告接收者操作特征曲线下面积（AUROC）接近0.89。我们探讨了这一准确性是否反映了学习或目标泄漏，表格基础模型是否改变了这一结论，以及部署所需的属性是否能够经受联合检验。我们对2022年行为风险因素监测系统中442,067名受访者的心肌梗死进行了十种分类器的基准测试，这些分类器涵盖了线性、树集成、神经网络、透明模型和表格基础模型，分为五个特征层级，按泄漏风险递减。每个模型都经过了对其区分度、校准、公平性（在明确的筛查阈值下）、符合覆盖、解释的可信度和推断成本的审计，然后在430,755名2023年受访者中应用——模型和阈值保持不变。移除两个后诊断特征使每个模型的AUROC下降了0.049-0.051，导致整个领域的结果收敛到一个0.0045宽的区间。透明的可解释增强机器在预设的0.005边际内不劣于任何替代方案，同时其评分速度约为最强基础模型的104倍。一个阈值检测到了75.4%的女性心肌梗死病例，而男性为89.0%；编辑模型的形状函数将差距缩小至0.010。边际符合预测对男性的覆盖率为0.86，对60岁以上成年人为0.82；Mondrian校准修复了每个层级。冻结模型的AUROC在0.002范围内传输。文献中报告的余量是特征集的属性，而非学习者的属性。透明度没有带来可测量的成本，并使公平性修复和不确定性条件直接可审计。评估实践，而非模型能力，是限制性约束。

View on arXiv Download PDF AI Translation

cs.CL / 54 / 2609.11851

IndicTriMix: Developing Language Identification Datasets and Models for Tri-Language Code-Mixing

IndicTriMix：为三语代码混合开发语言识别数据集和模型

Mishra, Pruthwik, Trivedi, Rudra, Patel, Avi, Urlana, Ashok, Malviya, Shrikant

Abstract

Language identification in code-mixed text, largely observed in social media, is highly essential when users frequently switch between multiple languages within a single utterance. Accurately identifying the languages of code-mixed tokens becomes an urgent necessity. Traditional language identification models, designed for monolingual text, are not well suited for token-level language identification in code-mixed settings. We formulate the task as a sequence labeling problem and fine-tune contextual transformer-based models MuRIL and XLM-RoBERTa best suited for Indian languages. We evaluate these systems on three different data configurations (Hindi, Gujarati, and Bengali) to predict language labels for individual tokens. We release a benchmark for language identification in code-mixed tokens with manually annotated test sets. We propose two approaches of code-mixed generation using parallel sentences of three languages. The trained models demonstrate the effectiveness of contextual embeddings for token-level language identification in multilingual social media text. For reproducibility and to facilitate future research, we publicly release our fine-tuned models.

Chinese Translation

在代码混合文本中的语言识别，尤其是在社交媒体中，用户在单一话语中频繁切换多种语言时显得尤为重要。准确识别代码混合标记的语言成为一种紧迫的需求。传统的语言识别模型主要针对单语文本设计，不适合在代码混合环境中进行标记级语言识别。我们将这一任务表述为序列标注问题，并对基于上下文的变换器模型MuRIL和XLM-RoBERTa进行微调，以更好地适应印度语言。我们在三种不同的数据配置（印地语、古吉拉特语和孟加拉语）上评估这些系统，以预测单个标记的语言标签。我们发布了一个代码混合标记语言识别的基准，并提供了手动注释的测试集。我们提出了两种使用三种语言的平行句子生成代码混合文本的方法。训练后的模型展示了上下文嵌入在多语言社交媒体文本中的标记级语言识别的有效性。为了可重复性并促进未来的研究，我们公开发布了我们的微调模型。

View on arXiv Download PDF AI Translation

cs.CL / 55 / 2609.11865

Epistemic orientation predicts legislative effectiveness among members of the US Congress

认识论取向预测美国国会成员的立法有效性

Aroyehun, Segun, Lewandowsky, Stephan, Garcia, David

Abstract

Truth and evidence-based communication provide important foundations for democratic governance, accountability, and collective decision-making. Prior work shows that evidence-oriented language in US congressional floor speeches has declined since the mid-1970s, alongside broader changes in legislative productivity and polarization. This study shifts the analysis from congressional sessions to individual members of Congress to examine whether epistemic orientation varies systematically across legislators and whether it relates to political behavior and legislative effectiveness. Using the Evidence-Minus-Intuition (EMI) score, we measure the relative prevalence of evidence-oriented versus intuition-oriented language in congressional floor speeches and Twitter posts. We link these measures to legislator-level data on ideology, institutional position, communication context, and Legislative Effectiveness Score (LES). The results show that more ideologically extreme members use less evidence-oriented language on the congressional floor. EMI also exhibits cross-platform consistency with members who use more evidence-oriented language in floor speeches also being more evidence-oriented on Twitter, although EMI is lower on Twitter overall. Finally, EMI in congressional speeches is positively associated with individual legislative effectiveness, even after accounting for ideology and extensive political, institutional, demographic, topical, and communication volume controls. These findings suggest that evidence-oriented language is not only an aggregate feature of congressional discourse but also a meaningful attribute of individual-level legislative communication and effectiveness.

Chinese Translation

真相和基于证据的沟通为民主治理、问责制和集体决策提供了重要基础。先前的研究表明，自1970年代中期以来，美国国会的发言中以证据为导向的语言有所下降，这与立法生产力和两极化的更广泛变化相伴随。本研究将分析重点从国会会议转向个别国会成员，以考察认识论取向在立法者之间是否存在系统性差异，以及其与政治行为和立法有效性之间的关系。通过使用证据减直觉（Evidence-Minus-Intuition, EMI）评分，我们测量国会发言和推特帖子中以证据为导向的语言与以直觉为导向的语言的相对普遍性。我们将这些测量结果与立法者的意识形态、机构地位、沟通背景和立法有效性评分（Legislative Effectiveness Score, LES）等数据联系起来。结果显示，意识形态极端的成员在国会发言中使用的以证据为导向的语言较少。EMI在跨平台上也表现出一致性，使用更多以证据为导向语言的成员在国会发言中也在推特上表现出更多的以证据为导向的语言，尽管总体上推特上的EMI较低。最后，国会发言中的EMI与个体立法有效性呈正相关，即使在考虑了意识形态以及广泛的政治、机构、人口统计、主题和沟通量控制后，这一关系依然成立。这些发现表明，以证据为导向的语言不仅是国会话语的一个整体特征，也是个体层面立法沟通和有效性的一个重要属性。

View on arXiv Download PDF AI Translation

cs.CL / 56 / 2609.11870

Augustinian BabyLM: What Ostensive Definition Can and Cannot Teach a Small Language Model

奥古斯丁婴儿语言模型：指示性定义能与不能教会小型语言模型的内容

Bylinina, Lisa

Abstract

A language model normally begins training with random word embeddings: whatever 'banana' means must be learned from training corpora. I implement St. Augustine's picture of word learning, meaning by ostension, for a small masked language model (DeBERTa) trained on 10M words: before training, visually grounded tokens receive embeddings derived from the image regions they label; other tokens start random. Visual initialization leaves a measurable imprint that lasts until the end of training. At the same time, the effect remains invisible under most BabyLM benchmarks, which probe abstract grammatical knowledge: visual initialization does not affect performance there. The only zero-shot exception is object-property knowledge (COMPS, Misra et al. 2023), where seeding helps in every configuration. To follow up on this result, I build a corpus-tailored version of the Visual-Property Swap benchmark (Lin et al., 2026), which tests color, material, size, and shape knowledge, with per-item training frequency and seeded status. Here, vision-seeded models have a persistent, seed- replicated advantage, confined to the seeded words. As a causal test, I show that synthetic grounding of previously unseeded words transfers the advantage to exactly those words. Function words and abstract vocabulary also receive strong visual seeds and retain them throughout training, and the training objective draws on them: held-out mask-prediction loss falls for these words in every seed. However, no benchmark I run registers this. What evaluation would pick this up remains an open question.

Chinese Translation

语言模型通常以随机的词嵌入开始训练：'香蕉'的意义必须通过训练语料库来学习。我为一个在1000万词上训练的小型掩码语言模型（DeBERTa）实现了圣奥古斯丁关于词汇学习的图景，即通过指示性定义进行学习：在训练之前，视觉基础的标记接收来自其所标记图像区域的嵌入；其他标记则从随机开始。视觉初始化留下了可测量的印记，持续到训练结束。同时，这种效果在大多数BabyLM基准测试下仍然是不可见的，这些测试探测抽象的语法知识：视觉初始化在这些测试中并不影响性能。唯一的零-shot例外是对象-属性知识（COMPS，Misra等，2023），在每种配置中，种子都能提供帮助。为了进一步探讨这一结果，我构建了一个针对语料库定制的视觉-属性交换基准（Lin等，2026），该基准测试颜色、材料、大小和形状知识，并考虑每个项目的训练频率和种子状态。在这里，视觉种子模型在种子词上具有持久的、种子复制的优势。作为因果测试，我展示了之前未种子的词的合成基础将优势转移到这些词上。功能词和抽象词汇也获得了强大的视觉种子，并在整个训练过程中保持这些种子，而训练目标则依赖于它们：在每个种子中，这些词的保留掩码预测损失下降。然而，我所运行的任何基准都没有记录到这一点。什么评估能够捕捉到这一点仍然是一个悬而未决的问题。

View on arXiv Download PDF AI Translation

cs.CL / 57 / 2609.11878

Domain-Specific Hallucination Detection in Large Language Models

大型语言模型中的领域特定幻觉检测

Chundru, Varun Teja, Biswas, Debasmita

Abstract

Large language models generate fluent text that can contain unfaithful claims -- a phenomenon known as hallucination. We present a multi-signal detection pipeline combining fine-tuned DeBERTa-v3 classification, Monte Carlo (MC) Dropout uncertainty quantification, and temperature-scaled calibration for response-level hallucination detection. Evaluated on the HaluEval benchmark, our pipeline achieves F1=0.915 and AUROC=0.977 on general-domain tasks, with per-task F1 scores of 0.97 (QA), 0.96 (Summarization), and 0.82 (Dialogue). MC Dropout inference further improves accuracy to 93.2%. A context ablation study confirms the model performs genuine entailment reasoning rather than exploiting surface patterns, with summarization F1 dropping 24% when knowledge context is removed. Learning curve analysis reveals that 25% of training data captures 77% of full-data performance. Beyond detection, we apply Direct Preference Optimization (DPO) to a Qwen2.5-0.5B generator, reducing its hallucination rate from 85.5% to 37.7% (55.9% relative reduction) as measured by our detector. Cross-domain evaluation on the SciFact biomedical benchmark shows that general-domain training transfers poorly (F1=0.52), motivating domain-specific fine-tuning. PubMedBERT fine-tuned on SciFact achieves F1=0.63 and AUROC=0.81, demonstrating that domain-matched pre-training is the strongest adaptation strategy. Code and models are available at https://github.com/varunteja99/hallucination-detection-nlp

Chinese Translation

大型语言模型生成的流畅文本可能包含不真实的陈述——这一现象被称为幻觉。我们提出了一种多信号检测管道，结合了微调的 DeBERTa-v3 分类、蒙特卡洛（Monte Carlo, MC）Dropout 不确定性量化和温度缩放校准，用于响应级幻觉检测。在 HaluEval 基准测试中，我们的管道在一般领域任务上实现了 F1=0.915 和 AUROC=0.977，具体任务的 F1 分别为 0.97（问答），0.96（摘要）和 0.82（对话）。MC Dropout 推断进一步将准确率提高至 93.2%。上下文消融研究确认模型进行的是有效的蕴含推理，而不是利用表面模式，当去除知识上下文时，摘要的 F1 下降了 24%。学习曲线分析表明，25% 的训练数据捕获了 77% 的全数据性能。除了检测，我们还将直接偏好优化（Direct Preference Optimization, DPO）应用于 Qwen2.5-0.5B 生成器，将其幻觉率从 85.5% 降低至 37.7%（相对减少 55.9%），这一结果是通过我们的检测器测得的。在 SciFact 生物医学基准上的跨领域评估显示，一般领域训练的迁移效果较差（F1=0.52），这促使了领域特定的微调。在 SciFact 上微调的 PubMedBERT 实现了 F1=0.63 和 AUROC=0.81，证明了领域匹配的预训练是最强的适应策略。代码和模型可在 https://github.com/varunteja99/hallucination-detection-nlp 获取。

View on arXiv Download PDF AI Translation

cs.CL / 58 / 2609.11892

Nuha-Speech: Building General-Purpose Arabic Speech-LLMs

Nuha-Speech：构建通用阿拉伯语语音大语言模型

Wang, Yingzhi, Alhazzani, Reem, Alqurishi, Muhammad

Abstract

As Speech Large Language Models (speech-LLMs) become increasingly multilingual, Arabic remains significantly underrepresented, highlighting the need for dedicated infrastructure to train and evaluate Arabic speech-LLMs. To address this gap, we introduce Nuha-Speech, a comprehensive initiative to develop general-purpose Arabic speech-LLMs spanning dataset construction, model training, and systematic evaluation. Specifically, we constructed a large-scale Arabic Speech Question-Answering (SQA) corpus comprising over 1.5 million training samples to allow instruction tuning over a broad range of core speech tasks. Then, the corpus was used for supervised fine-tuning based on Qwen-Omni model variants at different scales. Finally, we designed an evaluation framework featuring diverse tasks and tailored metrics. Through this work, we aim to establish foundational infrastructures for Arabic Speech-LLMs under constraints imposed by limited Arabic speech resources.

Chinese Translation

随着语音大语言模型（speech-LLMs）日益多语言化，阿拉伯语的代表性仍然显著不足，这突显了建立专门基础设施以训练和评估阿拉伯语语音大语言模型的必要性。为了解决这一问题，我们推出了Nuha-Speech，这是一个全面的倡议，旨在开发涵盖数据集构建、模型训练和系统评估的通用阿拉伯语语音大语言模型。具体而言，我们构建了一个大规模的阿拉伯语语音问答（SQA）语料库，包含超过150万个训练样本，以便在广泛的核心语音任务上进行指令调优。随后，该语料库被用于基于不同规模的Qwen-Omni模型变体进行监督微调。最后，我们设计了一个评估框架，包含多样化的任务和量身定制的指标。通过这项工作，我们旨在在阿拉伯语语音资源有限的约束下建立阿拉伯语语音大语言模型的基础设施。

View on arXiv Download PDF AI Translation

cs.CL / 59 / 2609.11913

Distance generalization in transformers: why bother with positional encoding?

变换器中的距离泛化：为什么要关注位置编码？

Nevermann, Daniel Henrik, Gros, Claudius

Abstract

Out-of-distribution length generalization, namely to extrapolate a task from short to longer context, has been studied intensively for transformers. Here we focus on distance generalization, which probes performance when inter-token distances are changed between training and inference, while keeping a fixed context length. We construct two synthetic delay copy tasks, both involving finite distances between source and recall, where tokens are copied either fully or selectively, and test models on delays unseen during training. We address three questions: (A) Do positional encoding schemes such as RoPE and ALiBi improve distance resolution relative to no positional encoding (NoPE)? (B) How does data diversity, the number of inter-token distances seen in training, affect performance? (C) When is distance transfer learning positive or negative? We present a thorough investigation, finding that it is paramount to improve our understanding of the underlying mechanisms.

Chinese Translation

对分布外长度泛化的研究，即从短上下文推断任务到更长上下文，已在变换器中得到了广泛研究。在这里，我们关注距离泛化，探讨在训练和推理之间改变令牌间距离时的性能，同时保持固定的上下文长度。我们构建了两个合成延迟复制任务，均涉及源和回忆之间的有限距离，其中令牌要么完全复制，要么选择性复制，并在训练期间未见的延迟上测试模型。我们提出了三个问题：（A）诸如 RoPE 和 ALiBi 的位置编码方案是否相对于没有位置编码（NoPE）改善了距离分辨率？（B）数据多样性，即训练中看到的令牌间距离的数量，如何影响性能？（C）何时距离迁移学习是积极的或消极的？我们进行了深入的调查，发现改善对潜在机制的理解至关重要。

View on arXiv Download PDF AI Translation

arXiv Papers

HuRo: Robotizing Human Videos for Scalable VLA Pretraining

When Information is Worth the Risk: Behavioral Valuation for Hazardous Robotic Exploration

Lie-Algebraic Bell Recurrences for Arbitrary-Order Twist Jets and Parallel-Mechanism Closure

Expressive Robotic Pianist: Mastering Complex Piano Repertoire with Graph-Mimic and Musical Dynamics

ReactHuman: A Physics-Grounded Benchmark for Human-Like Reactive Decision-Making in Embodied Multimodal LLMs

Planning along Differentiable Charts of Constraint Manifolds with General-Purpose IK Solvers

IMLE-VLA: Fast Single-Step Action Generation for Vision-Language-Action Policies

ObstaDiff: Generalizable Diffusion Policy Learning via Obstacle-aware Representations

Testing Between the Test Cases: Proving End-to-End Steering in Conditions You Never Drove

LTLDiff: Finite Linear Temporal Logic-Guided Data Generation and Diffusion Policies for Multi-agent Robotic Manipulation

Gait-Dependent Effects on Quadruped Locomotion for Load-Carrying using Passive Mechanism

Freehand Sketching for End-User Programming of Robot Swarms

RIDE: Relocalization-Informed Depth Estimation with 3D Gaussian Splatting

Harness Robotic OS: A Unified Embodied-Agent Runtime for Closed-Loop Quadruped Inspection

Beyond Noise Steering: Dual-Latent Space Reinforcement Learning for Generative Robot Policy

2AM: Grounding Agent-Side Memory as Guidance for Steerable Action Models in Long-Horizon Manipulation

Modular Kinematic Reduction of Closed-Chain Mechanisms Using Path Assembly and Defect Homotopy

Morphology-Aware Human Motion Retargeting for Wheeled-Humanoid Loco-Manipulation

GeoTrussRover: Morphological Computation with Contact-Semantic Control Primitives

SwarmNxt: Open-source Software-Hardware Platform for Fast and Agile Aerial Swarms

Safety-aware Skill Adaptation for Reinforcement Learning in Dynamic Environments

FARM: Reading Failure Signals from the Internal Predictive States of a Frozen Robotic World Model

3D Euler-Angle Orientation Control for Two-Ray Fading Mitigation in Maritime Air-to-Sea Communications

CARLAverse: A Highly Modular, Distributed, and Multimodal Framework for Human-in-the-Loop Simulation

Using Automated Vehicles Operational Data to Confirm Safety and Anticipate Threats

CAP: Continuously Adaptive Perception-Blind Humanoid Locomotion via Learned Denoising

Memory as Plans: World-Action Modeling with Memory-Grounded Planning

Quasi-static analysis of passive stability in a novel underactuated multi-finger hand

Contact-Aware Incremental Model Predictive Control for an Underactuated Aerial Manipulator

ActSafeGuard: Differentiable and Training-Aligned Constraint Enforcement for Flow-Matching Policies

Reflex-Informed Neuromuscular Reinforcement Learning for Muscle-Driven Locomotion

SEED-UMI: Sharing the Exoskeleton between human and robot for onE-to-one Dexterous demonstration

Visual-SLAM for the detection of hidden tomatoes in greenhouses by Hierarchical Localization and GLOMAPfor robotized harvesting

Rapid Learning of Dexterous In-Hand Pen Writing through Real-Time Jacobian Estimation

Learning Agent-based Model Predictive Control for Holistic Vehicle Performance

UniMPA: A Unified Memory-Prediction-Action Model via Action-Grounded Transition Modeling

EVPeriscope: Extended Perception across Aerial and Ground Vehicles with Event-based Propeller Tracking

Rethinking Handwritten Character Recognition

AcFlow: Controlling Text-to-Image Diffusion Transformers via Learned Conditional Activation Flow

MHE-Former: Multi-Hypothesis Transformers via Entropy Maximization for 3D Mesh Recovery

Meta-Learning for Data-Efficient Plant Growth Estimation via Vision Transformers and Fuzzy Clustering

GRADE: Single-Frame Generative Radar Depth Estimation Under Visual Degradation

Shedding Light: A Benchmark for Evaluating Lighting Understanding in Generative Image Models

Two-Parameter Flow Map Learning for Continuous-Time Diffeomorphic Image Registration

How Much Velocity Does Off-Ball Space Value Need? A Broadcast-Viewport Benchmark

TrajFusionNet+: Transformer-Based Prediction of Pedestrian Crossing Intention via Fusion of Trajectory Representations and Scene Graphs

Overpainting: Localized Context-aware Diffusion Image Editing

BodyCam-VQA: Enhanced Body-Worn Camera Video Captioning via Multimodal Reasoning and Probe Question Generation

Are We Really Doing Few-Shot Learning? A Critical Examination of Pre-Training Assumptions

Evaluation of Vision-Language Models Across Diverse Coastal Environments

Symmetry-aware super-resolution of crystal orientation maps via invariant latent-space learning

HiPerViT: A Hierarchical Perceiver-Vision Transformer Architecture for Multi-Scale Texture Recognition

CamPilot: A Multi-Agent Cinematic Assistant for Camera-Controlled Movie Generation

New Evidence, Same Choice: Testing Physical Experiment Selection in Vision Language Models

Toward Interpretable Multimodal Fusion: Heat Conduction Modeling for Hyperspectral and LiDAR Joint Classification

Meta-Learning for Classifier Selection in Image Datasets: A Feature-Driven Framework for Accuracy Prediction

TailProp: content-adaptive light- and heavy-tailed propagation for vision

Beyond Benchmarks: Using VLMs to Reveal Systematic Classification Failures Under Real World Conditions

ReconPlusGen: Injecting Reconstruction Prior into Multi-view 3D Generation through Noise Inversion and Modulation

LAION-Mobile: Evaluating Deepfake Detectors On One Million Smartphone Photos

UniH$^3$: Unifying Hierarchical Homogeneity and Heterogeneity for All-in-One Medical Image Restoration

Beyond Visual Quality: Evaluating Physical Consistency under Ego-Motion with EgoGenEval

A Multi-View and Confusion-Guided Ensemble Framework for Robust Synthetic Image Attribution

CEM-TUDASR: Computationally efficient multi-modality transformer based unsupervised domain adaptive super-resolution approach

Tri-DehazeGS: Scene--Medium Decoupled Gaussian Splatting with Transmittance-Aware Optimization

When is Test-Time Adaptation Identifiable From Unlabeled Evidence?

HALDETECT at ImageEval 2026 Shared Tasks: Answer-First Contrastive Grounding with QLoRA

SCINTILLA-SNN: A Spiking Multi-Scale Selective Aggregation Network for Perineural Invasion Prediction

Fast and Accurate Monomodal 3D High Resolution Deep Registration of Drosophila Larval Brain Volumes

From Evaluation to Enhancement: Benchmarking and Improving Think-with-Video Reasoning for Video Generative Models

Uncertainty DMD: Restoring Diversity in Few-Step Autoregressive Video Distillation

Improving Faint Object Detection for Space Situational Awareness with Variational Autoencoders

Order-Aware 2.5D Multiple Instance Learning for Preoperative MRI-Based Perineural Invasion Risk Assessment in Intrahepatic Cholangiocarcinoma

SAMV-DUSt3R: Instance-Centric 3D Scene Decoupling from Sparse Multi-Views

Your Model Already Knows Don't Teach It, Learn to Ask It: Soft Prompting for Few-Shot Adaptation of Vision-Language Models

GRIPNet: Gaussian Radial Intensity Prior Guided Architecture for Pulmonary Nodule Detection in CT

Mi-Ripple: Restoring Images Degraded by Iterative AI Editing

Predictive Multi-Landmark OCT Tracking for Increased Motion Robustness

R4Tun: LLM-guided adaptive segmental tunnel lining segmentation in point clouds