arXiv Daily Digest

180

Papers

Support-Constrained RL Enables Real-World Policy Improvement without Real-World Experience

支持约束强化学习实现无真实经验的现实世界策略改进

Yu, Raymond, Huey, William, Mukadam, Mustafa, Nagabandi, Anusha, Gupta, Abhishek

Abstract

Robots trained on real world data tend to be imprecise, slow, and brittle to perturbations. Improving these policies with reinforcement learning (RL) is an appealing alternative, but this process often requires expensive training in the real world. Performing policy improvement in simulation instead provides a far cheaper alternative, but unconstrained RL in simulation can exploit contact and dynamics mismatches, resulting in unsafe behaviors that do not transfer to hardware. Common forms of regularization can furthermore limit improvement by overconstraining to an imperfect behavior prior. In this work, we propose Support-Constrained Off-Domain REinforcement (SCORE), a real-to-sim-to-real framework that constrains RL in simulation to the support of a generative policy pretrained on real data. We instantiate this constraint through flow steering, restricting SCORE to actions the base policy can already produce, which ensures transferable behaviors while maximizing policy improvement. Improving a policy with SCORE requires minimal effort: it learns from sparse rewards, avoids distillation, and leaves the base policy untouched. Across eight real-world dexterous multi-fingered robotic manipulation tasks, SCORE improves average success rate from 37.8% to 89.9%, compared to 59.5% for the best baseline, and reaches success in 36.8% fewer steps than the base policy. Ultimately, through extensive experiments and ablations, we show that simulation can substantially improve real-world manipulation policies when policy optimization is appropriately constrained, introducing a new paradigm for real-to-sim-to-real policy improvement. Videos and code are available at https://weirdlabuw.github.io/score/.

Chinese Translation

在真实世界数据上训练的机器人往往表现不精确、反应缓慢，并且对扰动敏感。利用强化学习（RL）改进这些策略是一种吸引人的替代方案，但这一过程通常需要在真实世界中进行昂贵的训练。相反，在仿真中进行策略改进提供了一个更便宜的选择，但在仿真中不受约束的RL可能会利用接触和动态不匹配，从而导致不安全的行为，这些行为无法转移到硬件上。此外，常见的正则化形式可能通过过度约束不完美的行为先验而限制改进。在本研究中，我们提出了支持约束离域强化学习（SCORE），这是一个将RL在仿真中约束于在真实数据上预训练的生成策略支持的真实-仿真-真实框架。我们通过流引导实现这一约束，将SCORE限制为基础策略已经能够产生的动作，这确保了可转移的行为，同时最大化策略改进。使用SCORE改进策略所需的努力最小：它从稀疏奖励中学习，避免蒸馏，并且不改变基础策略。在八个真实世界的灵巧多指机器人操控任务中，SCORE将平均成功率从37.8%提高到89.9%，而最佳基线为59.5%，并且在比基础策略少36.8%的步骤中达成成功。最终，通过广泛的实验和消融研究，我们表明，当策略优化适当地受到约束时，仿真可以显著改善现实世界的操控策略，从而引入了一种新的真实-仿真-真实策略改进范式。视频和代码可在 https://weirdlabuw.github.io/score/ 获取。

View on arXiv Download PDF AI Translation

cs.RO / 2 / 2606.27495

AO-ARC: Almost-Surely Asymptotically Optimal Multi-Robot Motion Planning with ARC

AO-ARC：几乎肯定渐近最优的多机器人运动规划方法

Motes, James D., Morales, Marco, Amato, Nancy M.

Abstract

We present AO-ARC, an anytime multi-robot motion planning (MRMP) method that achieves initial solution times on par with state-of-the-art MRMP feasibility solvers while converging faster and more reliably than existing anytime MRMP methods as the number of robots increases. AO-ARC adapts the AO-x meta-algorithm for converting feasibility solvers into anytime algorithms by iteratively calling the original ARC method on bounded MRMP instances under a makespan cost metric. This exploits the adaptive (de)coupling of ARC while maintaining the consistent cost bound across robot (de)compositions needed for AO-x. We provide theoretical analysis proving the asymptotic optimality properties of AO- ARC and conduct empirical evaluation on a set of 2D scenarios with different levels of coordination complexity and a 3D manipulator scenario representative of real-world applications.

Chinese Translation

我们提出了AO-ARC，一种随时可用的多机器人运动规划（MRMP）方法，该方法在初始解决时间上与最先进的MRMP可行性求解器相当，同时在机器人数量增加时比现有的随时可用MRMP方法收敛得更快、更可靠。AO-ARC通过在有限的MRMP实例上迭代调用原始的ARC方法，适应了AO-x元算法，将可行性求解器转换为随时可用的算法，采用了以工期成本度量为基础的方式。这利用了ARC的自适应（解耦）特性，同时保持了AO-x所需的跨机器人（解耦）组合的一致成本界限。我们提供了理论分析，证明了AO-ARC的渐近最优性特征，并在一组具有不同协调复杂度的二维场景和一个代表真实世界应用的三维操控器场景上进行了实证评估。

View on arXiv Download PDF AI Translation

cs.RO / 3 / 2606.27566

Spacecraft Fiducial Marker for Autonomous Rendezvous, Proximity Operations, and Docking

用于自主会合、近距离操作和对接的航天器基准标记

Thakur, Ravi Kumar, Vrba, Matouš, Saska, Martin

Abstract

Robotic operations in space are challenging due to the harsh environment and the high cost of failure. Fiducial markers provide visual references that aid autonomous rendezvous, proximity operations, and docking for space robots. However, existing fiducial markers are mostly single-scale and largely designed for terrestrial robotics. Such markers leave the camera's field of view at close range, precisely during the proximity and docking phases where reliable tracking is most critical. This paper presents AstraTag, a fiducial marker designed for autonomous on-orbit robotic operations. The marker template is based on a square Spidron pattern whose recursive, self-similar structure enables detection across multiple spatial scales. Marker identification uses a 48-bit signature derived from triangular sub-regions of the template and encoded with a Generalised Reed-Solomon (GRS) code. The detection pipeline performs contour-based quadrilateral localisation, perspective normalisation, and signature matching against a pre-computed dictionary. To handle markers affixed to curved spacecraft surfaces, it incorporates a Thin-Plate Spline (TPS) re-warp fallback that exploits the marker's internal rectangular borders as additional geometric correspondences. We benchmark AstraTag against three-layer Fractal ArUco and AprilTag on spacecraft mockups with flat and curved surfaces. On curved surfaces, AstraTag achieves a higher detection rate than both baselines, offering a robust recursive-marker option for space robotics.

Chinese Translation

由于恶劣的环境和高昂的失败成本，太空中的机器人操作面临挑战。基准标记提供视觉参考，帮助航天机器人实现自主会合、近距离操作和对接。然而，现有的基准标记大多是单一尺度，并且主要设计用于地面机器人。这些标记在近距离时常常超出相机的视野，尤其是在近距离和对接阶段，此时可靠跟踪至关重要。本文提出了AstraTag，一种为自主在轨机器人操作设计的基准标记。该标记模板基于方形Spidron图案，其递归自相似结构使得在多个空间尺度上均能检测。标记识别使用从模板的三角子区域派生的48位签名，并采用广义里德-所罗门（Generalised Reed-Solomon, GRS）编码。检测流程执行基于轮廓的四边形定位、透视归一化以及与预先计算字典的签名匹配。为了处理附着在曲面航天器上的标记，它结合了薄板样条（Thin-Plate Spline, TPS）重映射回退，利用标记内部的矩形边界作为额外的几何对应关系。我们在具有平面和曲面航天器模型上对AstraTag与三层Fractal ArUco和AprilTag进行了基准测试。在曲面上，AstraTag的检测率高于两个基线，提供了一个稳健的递归标记选项，适用于太空机器人。

View on arXiv Download PDF AI Translation

cs.RO / 4 / 2606.27581

SceneBot: Contact-Prompted General Humanoid Whole Body Tracking with Scene-Interaction

SceneBot：基于接触提示的通用类人全身跟踪与场景交互

Chen, Sirui, Zhao, Shibo, Wu, Zhen, Li, Jiaman, Shi, Guanya, Liu, C. Karen

Abstract

Current humanoid reinforcement-learning policies excel at free-space motions but struggle with contact-rich tasks, as pure kinematic tracking cannot resolve the physical ambiguities of interacting with objects and uneven terrain. To address this, we introduce SceneBot, a unified motion-tracking framework capable of handling freespace locomotion, terrain traversal, and whole-body manipulation. SceneBot conditions a single policy on both reference motions and per-link contact labels, explicitly defining expected environmental interactions. To overcome the lack of annotated interaction data, we propose a hindsight scene reconstruction approach that infers scene-interaction graphs from retargeted human motion. Trained on 7.5 hours of this reconstructed, contact-rich data, SceneBot successfully generalizes to unseen motions and environments. Our results demonstrate that SceneBot is the first general framework to seamlessly unify free-space and contact-rich behaviors executing complex, long-horizon tasks like carrying a box upstairs and establishing contact conditioning as a powerful interface for humanoid control. All code and data will be open-sourced. More demos and information are available at: https://ericcsr.github.io/scenebot/

Chinese Translation

当前的类人强化学习策略在自由空间运动方面表现出色，但在接触丰富的任务中却面临挑战，因为纯运动学跟踪无法解决与物体和不平坦地形交互时的物理模糊性。为了解决这一问题，我们提出了SceneBot，一个统一的运动跟踪框架，能够处理自由空间运动、地形穿越和全身操控。SceneBot将单一策略条件化于参考运动和每个关节的接触标签，明确界定预期的环境交互。为了克服缺乏注释交互数据的问题，我们提出了一种事后场景重建方法，从重新定向的人类运动中推断场景交互图。SceneBot在7.5小时的重建接触丰富数据上进行训练，成功地推广到未见过的运动和环境。我们的结果表明，SceneBot是第一个能够无缝统一自由空间和接触丰富行为的通用框架，能够执行复杂的长时间任务，如将箱子搬上楼梯，并将接触条件建立为类人控制的强大接口。所有代码和数据将开源。更多演示和信息请访问：https://ericcsr.github.io/scenebot/

View on arXiv Download PDF AI Translation

cs.RO / 5 / 2606.27603

Learning to Throw: Agile and Accurate Cable-Suspended Payload Delivery with a Quadrotor

学习投掷：四旋翼无人机灵活且精确的电缆悬挂载荷投递

Zhai, Yifan, Raimondi, Elia, Ren, Yunfan, Geles, Ismail, Armati, Yannick, Xing, Jiaxu, Scaramuzza, Davide

Abstract

Quadrotors offer the agility needed to rapidly transport suspended payloads during time-critical applications, including search-and-rescue and medical delivery. While suspended-payload transport and traversal for these missions are well studied, the highly dynamic targeted release of the payload remains comparatively underexplored. State-of-the-art approaches typically rely on model-based trajectory optimization and tracking; however, these methods often yield sub-optimal performance due to conservative feasibility constraints, tracking errors, and the inherent difficulty of analytically modeling flexible rope dynamics. To overcome these limitations, we propose a hybrid simulation framework that couples a high-fidelity analytical quadrotor model with a physics solver for complex rope and payload interactions. By exchanging forces between the two domains at every step, we obtain a physically accurate simulation of the suspended-payload system. Leveraging this environment, we train a deep reinforcement learning (RL) policy that executes agile, accurate payload throws to designated targets. Deployed zero-shot on hardware, our RL policy pushes the boundary of the agility-accuracy trade-off, outperforming the model-based baseline by reducing the landing error by up to 50% and the throw duration by up to 30%. Ablation studies confirm that the coupled simulation is the key enabler of these gains. We further show that the same pipeline trains a policy driven by visual observations rather than an explicit state estimate, achieving accuracy comparable to that of the state-based policy. To accelerate future research in dynamic aerial manipulation, we open-source the simulator to the community upon acceptance.

Chinese Translation

四旋翼无人机提供了在时间关键应用中快速运输悬挂载荷所需的灵活性，包括搜索与救援和医疗投递。虽然这些任务中的悬挂载荷运输和穿越已得到充分研究，但高动态目标释放载荷的过程仍然相对欠缺探索。现有的最先进方法通常依赖于基于模型的轨迹优化和跟踪；然而，由于保守的可行性约束、跟踪误差以及灵活绳索动力学的解析建模固有的困难，这些方法往往导致次优的性能。为克服这些局限性，我们提出了一种混合仿真框架，将高保真度的分析四旋翼模型与复杂绳索和载荷交互的物理求解器相结合。通过在每一步之间交换两个领域的力，我们获得了悬挂载荷系统的物理准确仿真。利用这一环境，我们训练了一种深度强化学习（RL）策略，以执行灵活且精确的载荷投掷，目标指向指定目标。在硬件上零样本部署，我们的RL策略推动了灵活性与精确性之间的权衡，表现优于基于模型的基线，将着陆误差减少了多达50%，投掷持续时间减少了多达30%。消融研究确认，耦合仿真是实现这些增益的关键因素。我们进一步展示了相同的流程可以训练一种基于视觉观察而非显式状态估计的策略，达到与基于状态的策略相当的精度。为了加速未来在动态空中操作方面的研究，我们在论文接受后将仿真器开源给社区。

View on arXiv Download PDF AI Translation

cs.RO / 6 / 2606.27624

Physics-Guided Robotic Radiation Source Localization along Arbitrary Measurement Paths in Unstructured Environments

基于物理引导的机器人辐射源定位方法：沿任意测量路径在非结构化环境中的应用

Son, Hojoon, Tan, Kai, Zhang, Fan

Abstract

Using robots to estimate the location of the radiation source is an effective way to improve efficiency and safety. Existing methods focus on planning the robot's path to achieve precise estimation, typically approaching the source. However, approaching the source increases the risk of radiation damage to a robot. In addition, a path-planning algorithm designed solely for radiation source localization (RSL) limits the flexibility of missions that deploy robots into radioactive environments. This study presents an automation framework for robotic RSL that leverages a physics-informed machine learning (PIML) model to precisely estimate the source location, regardless of measurement paths, in unknown environments. Physics-inspired model tensors have been designed for PIML to handle attenuated gamma-ray flux signals from unknown obstacles, and multiple models are computed in parallel to improve the robustness and precision of the RSL. The proposed method is evaluated in high-fidelity simulation environments using Monte Carlo particle transport across diverse randomized domains, including spatial scales, radiation source types, obstacle materials and geometries, and robot trajectories. The method is also validated through physical experiments on configurations that are not included in the simulation-based evaluation. The continuous learning technique is applied in real-robot deployment to enhance the practical applicability of the online robotic RSL system. The proposed method advances robot radiation perception from pointwise flux detection to spatial intelligence.

Chinese Translation

利用机器人估计辐射源的位置是一种提高效率和安全性的有效方法。现有方法主要集中在规划机器人的路径以实现精确估计，通常是接近辐射源。然而，接近辐射源会增加机器人遭受辐射损伤的风险。此外，专门为辐射源定位（RSL）设计的路径规划算法限制了将机器人部署到放射性环境中的任务灵活性。本研究提出了一种机器人RSL的自动化框架，该框架利用物理信息机器学习（PIML）模型在未知环境中无论测量路径如何都能精确估计源位置。为PIML设计了基于物理的模型张量，以处理来自未知障碍物的衰减伽马射线通量信号，并且多个模型并行计算以提高RSL的鲁棒性和精确性。所提出的方法在高保真模拟环境中进行了评估，采用蒙特卡洛粒子传输技术，涵盖了多种随机化领域，包括空间尺度、辐射源类型、障碍物材料和几何形状以及机器人轨迹。该方法还通过物理实验进行了验证，实验配置不包括在基于模拟的评估中。连续学习技术被应用于真实机器人部署，以增强在线机器人RSL系统的实际适用性。所提出的方法将机器人辐射感知从点状通量检测提升到空间智能。

View on arXiv Download PDF AI Translation

cs.RO / 7 / 2606.27625

P-ARC: Exploiting Subproblem Independence for Parallel Multi-Robot Motion Planning

P-ARC：利用子问题独立性进行并行多机器人运动规划

Motes, James D., Morales, Marco, Amato, Nancy M.

Abstract

This paper presents Parallel ARC (P-ARC), a parallel variant of the Adaptive Robot Coordination (ARC) approach to multi-robot motion planning (MRMP). P-ARC proposes a parallel variant for each of the three main stages in ARC: initial individual solutions, conflict detection, and conflict resolution, exploiting the independence created by ARC's decomposition of the MRMP problem. Additionally, we employ an OR-parallel multi-start strategy to both ARC and P-ARC, creating a hybrid parallel strategy OR-P-ARC. We evaluate the impact of the different parallel strategies for ARC using a set of scaling 2D mobile and planar manipulator scenarios with up to 128 robots to control for conflicts and work distribution across the stages of ARC. Additionally, we demonstrate planning time speedups approaching 4X over the sequential version for large Panda multi-manipulator teams in real-world inspired scenarios when deploying 16 CPU cores.

Chinese Translation

本文提出了并行自适应机器人协调（P-ARC），这是自适应机器人协调（ARC）方法在多机器人运动规划（MRMP）中的并行变体。P-ARC为ARC的三个主要阶段提出了并行变体：初始个体解、冲突检测和冲突解决，利用ARC对MRMP问题的分解所创造的独立性。此外，我们对ARC和P-ARC采用了OR-并行多启动策略，形成了一种混合并行策略OR-P-ARC。我们使用一组规模化的二维移动和平面操控器场景对ARC的不同并行策略的影响进行了评估，场景中最多可控制128个机器人，以管理冲突和在ARC各阶段之间的工作分配。此外，我们展示了在真实世界启发场景中，当部署16个CPU核心时，针对大型Panda多操控器团队的规划时间加速接近4倍，相较于顺序版本。

View on arXiv Download PDF AI Translation

cs.RO / 8 / 2606.27663

Direct Action-Head Injection of A Grounded 3D Point Unlocks Spatial and Task Generalization

直接作用-头部注入一个基础的3D点以解锁空间和任务泛化

Tsai, Shiang-Feng, Jhang, Jin-Cheng, Tai, Yen-Ling, Lai, Jia-Hong, Wong, Shih-Yun, KangTung-Hsu, Chen, Yi-Ting

Abstract

Vision-Language-Action (VLA) models leverage large-scale vision-language pretraining for flexible robot manipulation, yet at test time they remain brittle along two axes: spatial generalization, when object positions differ from those seen during training, and task generalization, when a familiar scene is paired with a different language instruction than the one seen in training. A growing family of methods addresses this brittleness by endowing a policy with the spatial and task-aware information such as 2D pixel-coordinate for object localization and placement. However, we find that existing representation through language prompting or visual prompting does not address the limitations; in contrast, exploiting a 3D point-based representation and feeding it directly to the action head leads to substantial improvements-revealing that how the grounding signal is represented and injected into the VLA is the true game changer. Thus, we propose a lightweight, model-agnostic module that represents the grounding signal in 3D, computes its relative displacement to the gripper, and injects the resulting spatial embedding directly into the action head through adaptive layer normalization. The entire module is a two-layer MLP that requires no changes to the VLA backbone or pretraining pipeline. On LIBERO-PRO, our method improves the average success rate of GR00T-N1.6 from 31.2 to 77.5 points under task perturbation and from 28.1 to 60.2 points under position perturbation (gains of 46.3 and 32.1 points). Comparable gains are achieved for $\pi_{0.5}$ as well, demonstrating that the mechanism is backbone-agnostic. Together, these results support our central finding: given adequate grounding lifted into 3D, injecting it directly into the action head is what unlocks both spatial and task generalization in VLAs-achievable with nothing more than a lightweight module on top of a pretrained backbone.

Chinese Translation

视觉-语言-动作（VLA）模型利用大规模的视觉-语言预训练实现灵活的机器人操作，但在测试时仍然在两个方面表现脆弱：空间泛化，即当物体位置与训练期间所见不同；以及任务泛化，即当熟悉的场景与不同于训练期间所见的语言指令配对时。越来越多的方法通过赋予策略空间和任务感知信息（例如用于物体定位和放置的2D像素坐标）来解决这种脆弱性。然而，我们发现现有的通过语言提示或视觉提示的表示并未解决这些局限性；相反，利用基于3D点的表示并直接将其输入到动作头部会带来显著的改善——揭示了如何表示和注入基础信号到VLA中才是真正的游戏改变者。因此，我们提出了一个轻量级的、模型无关的模块，该模块在3D中表示基础信号，计算其相对于夹持器的相对位移，并通过自适应层归一化将生成的空间嵌入直接注入到动作头部。整个模块是一个两层的多层感知器（MLP），无需对VLA主干或预训练管道进行任何更改。在LIBERO-PRO上，我们的方法在任务扰动下将GR00T-N1.6的平均成功率从31.2提高到77.5，在位置扰动下从28.1提高到60.2（分别增加了46.3和32.1点）。对于$ ext{π}_{0.5}$也取得了类似的增益，证明该机制与主干无关。综合这些结果支持我们的核心发现：在3D中提供足够的基础信号，直接将其注入到动作头部是解锁VLA的空间和任务泛化的关键——这一切仅需在预训练主干之上添加一个轻量级模块即可实现。

View on arXiv Download PDF AI Translation

cs.RO / 9 / 2606.27676

CWI: Composite Humanoid Whole-Body Imitation System for Loco-manipulation

CWI：复合类人全身模仿系统用于运动操控

Ge, Wenqi, Guo, Junde, Fu, Zhen, Yang, Shunpeng, Chen, Jiayu, Chen, Hua

Abstract

Achieving everyday tasks with humanoid robots requires coordinating stable locomotion with versatile manipulation. However, existing whole-body controllers still face significant challenges. Methods trained solely via command sampling, without motion-capture (MoCap) data, often struggle with sparse rewards and require carefully tuned curricula to converge. This is especially problematic for upper-body control, where the resulting motions deviate from human-like statistics and degrade whole-body coordination. Conversely, approaches that imitate full-body MoCap data suffer from dataset imbalance, as many locomotion trajectories are overly aggressive for stable-locomotion scenarios, necessitating extensive data filtering and augmentation. To address this, we present Composite Whole-Body Imitation (CWI), a framework that decouples the use of MoCap data for upper-body manipulation and lower-body locomotion. This decoupling allows us to exploit the full MoCap dataset of diverse manipulation references, while stable, command-conditioned lower-body locomotion is guided by dual discriminators trained on curated expert-quality walking and squatting clips via an Adversarial Motion Prior (AMP). A multi-critic architecture reduces conflicts among locomotion, manipulation, and motion-style objectives, and a teacher--student distillation stage yields a whole-body policy conditioned only on bimanual hand poses and velocity/height commands. We evaluate CWI through simulation experiments and real-world deployment on a full-size LimX Oli humanoid. The results show competitive loco-manipulation performance, robust whole-body coordination, and practical teleoperation without full-body motion-capture equipment. A project page with supplementary material can be found at https://cwi-ral.github.io/CWI-RAL-Webpage.

Chinese Translation

实现类人机器人完成日常任务需要协调稳定的运动与多样的操控。然而，现有的全身控制器仍面临重大挑战。仅通过命令采样训练的方法，在没有运动捕捉（MoCap）数据的情况下，往往难以应对稀疏奖励，并需要精心调整的课程才能收敛。这对于上半身控制尤其成问题，导致生成的动作偏离类人统计特征，降低全身协调性。相反，模仿全身MoCap数据的方法则受到数据集不平衡的影响，因为许多运动轨迹对于稳定运动场景而言过于激进，需进行大量数据过滤和增强。为了解决这一问题，我们提出了复合全身模仿（CWI）框架，该框架将上半身操控与下半身运动的MoCap数据使用解耦。这种解耦使我们能够利用多样化操控参考的完整MoCap数据集，同时稳定的、基于命令的下半身运动由通过对策运动先验（AMP）训练的双鉴别器引导，这些鉴别器基于策划的专家级行走和下蹲视频。多批评家架构减少了运动、操控和动作风格目标之间的冲突，而教师-学生蒸馏阶段则生成仅基于双手姿态和速度/高度命令的全身策略。我们通过仿真实验和在全尺寸LimX Oli类人机器人上的实际部署评估了CWI。结果显示出竞争力的运动操控性能、稳健的全身协调性，以及在没有全身运动捕捉设备的情况下的实用远程操作。项目页面及补充材料可在 https://cwi-ral.github.io/CWI-RAL-Webpage 找到。

View on arXiv Download PDF AI Translation

cs.RO / 10 / 2606.27677

DIM-WAM: World-Action Modeling with Diverse Historical Event Memory

DIM-WAM：具有多样化历史事件记忆的世界行动建模

Wang, Kai, Gu, Zhaopeng, Chen, Yixiang, Xu, Yuan, Ma, Qisen, Su, Peng, Li, Zhaowen, Huang, Yan, Wang, Liang

Abstract

World-action models have shown promising robot-manipulation performance by jointly predicting future visual states and actions. However, existing methods mainly rely on short-term history and short-horizon future prediction, which is insufficient for long-horizon tasks whose correct execution depends on earlier observations and task progress. Such temporally dependent tasks require effective use of complementary temporal information, including recent local context, cross-stage historical events, immediate future dynamics, and global task progress. To address long-term forgetting and poor awareness of the global task state, we introduce DiM-WAM, a memory-augmented world-action model that integrates multi-scale historical context, local future dynamics, and global task progress. The memory extracts compact visual event information from real observations, updates multiple memory banks through independent similarity-based merging, and then reads the bank-identity- and time-embedded long-term context to condition video and action denoising. A progress-supervision objective further encourages memory tokens to encode not only completed historical events but also the current task stage and its implications for the remaining task. On RMBench, DiM-WAM raises average success from 28.4% with LingBot-VA to 69.8%, exceeding the explicit-memory Mem-0 baseline at 42.0%. On four real-world Franka tasks, it improves average stage success from 70.7% to 91.5% and full-task success from 52.5% to 80.0%. Project page: https://wangkai-casia.github.io/dim-wam/{\texttt{https://wangkai-casia.github.io/dim-wam/}}.

Chinese Translation

世界行动模型通过联合预测未来视觉状态和动作，展现了令人期待的机器人操控性能。然而，现有方法主要依赖于短期历史和短期未来预测，这对于长时间跨度的任务而言是不够的，因为这些任务的正确执行依赖于早期观察和任务进展。这类时间依赖的任务需要有效利用互补的时间信息，包括近期的局部上下文、跨阶段的历史事件、即时的未来动态以及全球任务进展。为了解决长期遗忘和对全球任务状态的认知不足的问题，我们提出了DiM-WAM，这是一种增强记忆的世界行动模型，整合了多尺度的历史上下文、局部未来动态和全球任务进展。该记忆模块从真实观察中提取紧凑的视觉事件信息，通过独立的基于相似度的合并更新多个记忆库，然后读取嵌入银行身份和时间的长期上下文，以调节视频和动作去噪。进度监督目标进一步鼓励记忆标记不仅编码已完成的历史事件，还编码当前任务阶段及其对剩余任务的影响。在RMBench上，DiM-WAM将LingBot-VA的平均成功率从28.4%提升至69.8%，超过了显式记忆Mem-0基线的42.0%。在四个真实世界的Franka任务中，它将平均阶段成功率从70.7%提升至91.5%，将完整任务成功率从52.5%提升至80.0%。项目页面：https://wangkai-casia.github.io/dim-wam/

View on arXiv Download PDF AI Translation

cs.RO / 11 / 2606.27755

Drop-Then-Recovery: How Redundant Are Vision-Language-Action Models?

丢弃后恢复：视觉-语言-动作模型的冗余性如何？

Sun, Guoheng, Feng, Kaixi, He, Shwai, Gong, Xiaochuan, He, Yexiao, Wang, Ziyao, Shen, Zheyu, Ye, Wanghao, Kompella, Ramana Rao, Liu, Gaowen, Li, Ang

Abstract

Vision-Language-Action (VLA) models enable instruction-driven robotic manipulation, but they inherit oversized language backbones from pretrained VLMs whose capacity far exceeds what is needed for short robotic instructions. This raises a basic question: how much of a VLA model is actually necessary for closed-loop control? In this work, we study architectural redundancy in VLA models by using transformer block removal as a controlled intervention. We introduce \textbf{Drop-Then-Recovery (DTR)}, an analysis protocol that removes selected blocks from a pretrained VLA model and then fine-tunes the resulting model to measure whether the removed capacity was necessary for downstream control. To make this intervention reliable, we propose \textbf{GateProbe}, a one-shot virtual-gate sensitivity metric that ranks blocks by their contribution to the downstream action loss. Across multiple VLA architectures, manipulation benchmarks and even real-robot industrial scenarios, we find a strong asymmetry in post-removal recoverability: \ul{\textit{language backbones are highly redundant for standard robotic manipulation tasks, whereas vision and action pathways are substantially less tolerant to removal}}. On LIBERO, removing half of the LLM blocks even improves OpenVLA-OFT from 95.0% to 98.3% under the same downstream fine-tuning budget, and retaining only two language blocks still recovers baseline-level performance. These results suggest that current VLA benchmarks may exert limited pressure on deep language grounding and compositional instruction understanding, and that future VLA architectures should allocate capacity more deliberately across language, vision, and action components. The code is available at https://github.com/s1ghhh/VLADrop.

Chinese Translation

视觉-语言-动作（VLA）模型使得基于指令的机器人操作成为可能，但它们继承了来自预训练视觉-语言模型（VLM）的过大语言骨干，其容量远超短期机器人指令所需。这引发了一个基本问题：VLA 模型中实际需要多少部分用于闭环控制？在本研究中，我们通过使用变换器块移除作为一种受控干预，研究 VLA 模型中的架构冗余。我们引入了 extbf{丢弃后恢复（Drop-Then-Recovery, DTR）}，一种分析协议，该协议从预训练的 VLA 模型中移除选定的块，然后微调结果模型，以测量被移除的容量是否对下游控制是必要的。为了使这一干预可靠，我们提出了 extbf{门探测（GateProbe）}，一种一次性虚拟门敏感度指标，通过其对下游动作损失的贡献对块进行排序。在多种 VLA 架构、操作基准以及实际机器人工业场景中，我们发现移除后的恢复性存在显著不对称性： extit{语言骨干对于标准机器人操作任务高度冗余，而视觉和动作路径对移除的容忍度则显著较低}。在 LIBERO 上，移除一半的 LLM 块甚至在相同的下游微调预算下将 OpenVLA-OFT 的性能从 95.0% 提高到 98.3%，而仅保留两个语言块仍能恢复到基线水平的性能。这些结果表明，当前的 VLA 基准可能对深层语言基础和组合指令理解施加的压力有限，未来的 VLA 架构应更有意识地在语言、视觉和动作组件之间分配容量。代码可在 https://github.com/s1ghhh/VLADrop 获取。

View on arXiv Download PDF AI Translation

cs.RO / 12 / 2606.27807

SpikeVLA: Vision-Language-Action Models with Spiking Neural Networks

SpikeVLA：基于脉冲神经网络的视觉-语言-动作模型

Song, Ruiqi, Nie, Dujun, Teng, Siyu, Ding, Baiyong, Zhang, Xiaotong, Li, Dong, Zhang, Chenming, Li, Yuchen, Wu, Hangbin, Chen, Long

Abstract

Vision-Language-Action (VLA) models have become a dominant paradigm for embodied intelligence. However, most existing approaches are built on large-scale transformers, resulting in substantial inference latency and energy consumption that limit their practical deployment in low-power, real-time scenarios. We propose SpikeVLA, a spiking VLA architecture for embodied navigation with energy-efficient inference, consisting of three key components. (i) A spiking vision encoder, Spike-V, that replaces dense continuous layers with event-driven spiking layers to reduce the energy consumption of visual representation learning. (ii) A multi-modal spiking large language model, Spike-L, that reformulates cross-modal reasoning with spiking dynamics and token-level event-driven sparsity to further lower computational cost. (iii) A spiking action policy network, Spike-A employs Laplacian-kernel population coding with a multi-layer fully connected SNN, and decodes spiking activities into stable and robust continuous control with energy-efficient inference under low-power constraints. Experiments on navigation and robotic control tasks show that SpikeVLA significantly reduces energy consumption and computational cost while maintaining competitive performance, highlighting its potential for low-power, real-time embodied intelligence.

Chinese Translation

视觉-语言-动作（VLA）模型已成为具身智能的主导范式。然而，现有的大多数方法都是基于大规模变换器（transformers），这导致了显著的推理延迟和能耗，限制了它们在低功耗实时场景中的实际应用。我们提出了SpikeVLA，一种用于具身导航的脉冲VLA架构，具有能效推理，包含三个关键组件。(i) 脉冲视觉编码器Spike-V，用事件驱动的脉冲层替代密集连续层，以减少视觉表征学习的能耗。(ii) 多模态脉冲大语言模型Spike-L，通过脉冲动态和基于令牌的事件驱动稀疏性重新构建跨模态推理，以进一步降低计算成本。(iii) 脉冲动作策略网络Spike-A，采用拉普拉斯核（Laplacian-kernel）群体编码与多层全连接脉冲神经网络（SNN），并在低功耗约束下将脉冲活动解码为稳定且鲁棒的连续控制，具有能效推理。导航和机器人控制任务的实验表明，SpikeVLA显著降低了能耗和计算成本，同时保持了竞争性能，突显了其在低功耗实时具身智能中的潜力。

View on arXiv Download PDF AI Translation

cs.RO / 13 / 2606.27811

LXD-SLAM: LiDAR+X Dense SLAM with $\sum_{i=0}^{5}C_5^i$ Configurable Sensor Combinations

LXD-SLAM：具有 $ extstyle extsum_{i=0}^{5}C_5^i$ 可配置传感器组合的 LiDAR+X 稠密 SLAM

Wang, Zhong, Zhang, Lin, Li, Linfei, Shen, Ying, Zhang, Shaoming, Shi, Pengcheng, Zhao, Shengjie

Abstract

Simultaneous Localization and Mapping (SLAM) is essential for autonomous systems, yet achieving reliable, globally consistent pose estimation and dense mapping in complex environments remains challenging due to geometric degeneracy and sensor drift. While multi-sensor fusion addresses these issues, existing systems often lack the modularity to adapt to diverse platforms and rely on mathematically inconsistent fusion or suboptimal map representations. To address these limitations, we propose LXD-SLAM (LiDAR+X Dense SLAM), a highly versatile and unified multi-sensor fusion framework. Centered around 3D LiDAR, our system allows for the plug-and-play integration of LiDAR, Camera, IMU, Wheel Encoder, and GNSS, supporting up to 32 distinct sensor combinations. We employ a mathematically unified Iterative Error-Sate Kalman Filter with an adaptive hierarchical prediction strategy and an update step that minimizes point-to-mesh distances and visual reprojection errors. To support this, the environment is modeled using continuous multi-layered Gaussian Process (GP) sub-meshes, which enables efficient ray-to-mesh depth recovery for visual features. For global consistency, we introduce an Extended Scan Context (ESC) descriptor derived from the GP sub-meshes alongside a Bidirectional PnP optimization for robust multi-modal loop closure within a hybrid pose graph. Extensive evaluations on public datasets and real-world experiments demonstrate that LXD-SLAM matches or exceeds state-of-the-art specialized odometry solutions across various configurations while generating high-fidelity, globally consistent dense meshes in real-time. The relevant codes and data will be made available at https://github.com/peterWon/LXD-SLAM upon publication.

Chinese Translation

同步定位与地图构建（SLAM）对于自主系统至关重要，但在复杂环境中实现可靠的、全局一致的姿态估计和稠密地图构建仍然面临挑战，这主要是由于几何退化和传感器漂移。虽然多传感器融合可以解决这些问题，但现有系统往往缺乏适应多样化平台的模块化能力，并依赖于数学上不一致的融合或次优的地图表示。为了解决这些局限性，我们提出了 LXD-SLAM（LiDAR+X 稠密 SLAM），这是一个高度灵活且统一的多传感器融合框架。我们的系统以 3D LiDAR 为核心，允许 LiDAR、相机、IMU、轮编码器和 GNSS 的即插即用集成，支持多达 32 种不同的传感器组合。我们采用了一个数学上统一的迭代误差状态卡尔曼滤波器，结合自适应分层预测策略和一个更新步骤，以最小化点到网格的距离和视觉重投影误差。为支持这一点，环境使用连续多层高斯过程（GP）子网格建模，这使得视觉特征的高效光线到网格深度恢复成为可能。为了实现全局一致性，我们引入了一个扩展扫描上下文（ESC）描述符，该描述符源自 GP 子网格，并结合双向 PnP 优化，以实现混合姿态图中的稳健多模态回环闭合。在公共数据集和实际实验上的广泛评估表明，LXD-SLAM 在各种配置下与最先进的专业里程计解决方案相匹配或超越，同时实时生成高保真、全局一致的稠密网格。相关代码和数据将在出版后提供于 https://github.com/peterWon/LXD-SLAM。

View on arXiv Download PDF AI Translation

cs.RO / 14 / 2606.27813

Booster Lab: A Data-Centric Pipeline for Learning Deployable Humanoid Locomotion Policies

助推器实验室：一个以数据为中心的可部署类人运动策略学习管道

Chen, Penghui, Zheng, Tinglong, Zhang, Yufeng, Zhao, Mingguo

Abstract

Humanoid robot motion learning requires not only task-oriented control policies but also physically feasible and natural behaviors that can be transferred to real robots. However, robot-feasible motion data are often scarce: raw human demonstrations may be incompatible with the robot morphology, open-source clips vary in quality, and simulation-collected robot trajectories still require feasibility checking. To address these challenges, we propose a data-centric training and deployment pipeline that integrates motion data curation, real-to-sim model adaptation, AMP-based reinforcement learning, and sim-to-real deployment. We validate the framework on the Booster T1 robot and further provide preliminary cross-platform validation on Booster K1.

Chinese Translation

类人机器人运动学习不仅需要面向任务的控制策略，还需要可物理实现和自然的行为，这些行为能够转移到真实机器人上。然而，适合机器人使用的运动数据往往稀缺：原始的人类示范可能与机器人形态不兼容，开源视频片段的质量参差不齐，而通过仿真收集的机器人轨迹仍需进行可行性检查。为了解决这些挑战，我们提出了一种以数据为中心的训练和部署管道，该管道集成了运动数据整理、真实到仿真模型适配、基于AMP（自适应多策略）的强化学习以及仿真到真实的部署。我们在助推器T1机器人上验证了该框架，并进一步提供了助推器K1的初步跨平台验证。

View on arXiv Download PDF AI Translation

cs.RO / 15 / 2606.27861

PPO-EAL: Exact Augmented Lagrangian Proximal Policy Optimization for Safe Robotic Control

PPO-EAL：用于安全机器人控制的精确增强拉格朗日近端策略优化

Ding, Jiatao, Gao, Songqun, Del Prete, Andrea, Saveriano, Matteo

Abstract

Reinforcement learning (RL) has emerged as a promising solution to accomplish complex robotic control tasks; however, most of the current work ignores the safety requirements. Safe RL seeks to maximize task performance while satisfying explicit physical constraints, but current algorithms struggle to learn the policy efficiently with precise constraint satisfaction. This work proposes PPO-EAL, a novel first-order constrained policy optimization framework that integrates exact augmented Lagrangian optimization into proximal policy optimization for safe robotic control. By combining clipped policy updates with exact quadratic penalty terms, PPO-EAL achieves theoretically grounded constraint enforcement without requiring impractically large penalty factors. A momentum-regulated multiplier update further improves dual-variable stability, reducing constraint oscillation and unsafe behavior while preserving task performance. We provide exactness and convergence analysis under standard stochastic approximation assumptions. Extensive validation across diverse GPU-accelerated robotic benchmarks-including cart-pole balancing, cart-double-pendulum stabilization, 7-DoF Franka end-effector reaching, and quadrupedal locomotion-demonstrates superior safety precision and reward performance compared with state-of-the-art first-order safe RL baselines. Finally, we demonstrate zero-shot sim-to-real deployment in a contact-rich gear assembly task, where PPO-EAL substantially improves task success, reduces peak contact force, and enhances operational robustness. These results establish PPO-EAL as a general and practically deployable safe RL framework for diverse safety-critical robotic systems.

Chinese Translation

强化学习（RL）已成为实现复杂机器人控制任务的有前景的解决方案；然而，目前大多数研究忽视了安全要求。安全强化学习旨在最大化任务性能，同时满足明确的物理约束，但当前算法在高效学习策略和精确满足约束方面面临挑战。本研究提出了PPO-EAL，一种新颖的一阶约束策略优化框架，将精确增强拉格朗日优化集成到近端策略优化中，以实现安全的机器人控制。通过结合裁剪的策略更新与精确的二次惩罚项，PPO-EAL在不需要不切实际的大惩罚因子的情况下，实现了理论上有依据的约束执行。动量调节的乘子更新进一步提高了对偶变量的稳定性，减少了约束振荡和不安全行为，同时保持了任务性能。我们在标准随机逼近假设下提供了精确性和收敛性分析。在多种GPU加速的机器人基准测试中进行的广泛验证——包括小车-杆平衡、小车-双摆稳定、7自由度Franka末端执行器到达和四足运动——显示出与最先进的一阶安全强化学习基准相比，PPO-EAL在安全精度和奖励性能上具有优越性。最后，我们在一个接触丰富的齿轮组装任务中展示了零-shot模拟到真实的部署，其中PPO-EAL显著提高了任务成功率，减少了峰值接触力，并增强了操作的鲁棒性。这些结果确立了PPO-EAL作为一种通用且可实际部署的安全强化学习框架，适用于各种安全关键的机器人系统。

View on arXiv Download PDF AI Translation

cs.RO / 16 / 2606.27871

LocalNav: Distilling Frontier VLMs and Embodied RL for On-Device Object Goal Navigation

LocalNav：将前沿视觉语言模型与具身强化学习提炼用于设备端目标导航

Baumann, Nicolas, Boyle, Liam, Deng, Pu, Ghignone, Edoardo, Sun, Boyang, Pollefeys, Marc, Benini, Luca, Magno, Michele

Abstract

Vision Language Models (VLMs) have emerged in the robotic domain as a powerful tool that enables environmental perception with language context, serving as a catalyst for open-vocabulary tasks like ObjectNav. Yet, their computational footprint typically confines them to cloud execution, hindering low-latency inference with local deployment on resource-constrained robots. To address this challenge, we present a distillation strategy that transfers complex spatial-semantic reasoning from large frontier models into a lightweight, 4B-parameter local VLM for edge execution on embedded GPU devices (e.g., Jetson Orin). We first establish a State of the Art (SotA), Scene Graph (SG)-based pipeline using Claude Sonnet 4.6, achieving a 39.7% Success Rate (SR) on the HM3D OVON benchmark. We then demonstrate that fine-tuning Qwen3.5-4B on just 500 frontier reasoning traces effectively enables navigation capabilities, yielding a SR of 34.5%, narrowing the gap to the performance of large cloud models. Finally, we introduce E-RLVR with Token Generation (TG) regularization to compress output sequence lengths for physical deployment while grounding the agent in its task. This downstream optimization reduces TG overhead by 72.1% and latency by 71.8%. Combined with quantization, this joint strategy yields a cumulative 82.8% reduction in overall inference latency without significantly sacrificing performance, presenting a viable paradigm for local, low-latency VLM execution on mobile robots.

Chinese Translation

视觉语言模型（VLMs）在机器人领域中作为一种强大的工具出现，能够结合语言上下文进行环境感知，为开放词汇任务（如目标导航）提供了催化剂。然而，它们的计算开销通常限制了它们只能在云端执行，从而妨碍了在资源受限的机器人上进行低延迟推理。为了解决这一挑战，我们提出了一种提炼策略，将大型前沿模型中的复杂空间-语义推理转移到一个轻量级、具有40亿参数的本地VLM，以便在嵌入式GPU设备（如Jetson Orin）上进行边缘执行。我们首先建立了一个基于状态图（Scene Graph, SG）的最先进（State of the Art, SotA）管道，使用Claude Sonnet 4.6，在HM3D OVON基准测试中实现了39.7%的成功率（Success Rate, SR）。然后，我们展示了在仅使用500个前沿推理轨迹对Qwen3.5-4B进行微调，能够有效地启用导航能力，获得34.5%的成功率，缩小了与大型云模型性能之间的差距。最后，我们引入了带有令牌生成（Token Generation, TG）正则化的E-RLVR，以压缩物理部署时的输出序列长度，同时将代理与其任务相结合。这一下游优化将TG开销减少了72.1%，延迟减少了71.8%。结合量化，这一联合策略在不显著牺牲性能的情况下，实现了整体推理延迟的累计82.8%的减少，展示了一种在移动机器人上进行本地、低延迟VLM执行的可行范式。

View on arXiv Download PDF AI Translation

cs.RO / 17 / 2606.27872

S$^2$-VLA: State-Space Guided Vision-Language-Action Models for Long-Horizon Manipulation

S$^2$-VLA：用于长时间操作的状态空间引导视觉-语言-动作模型

Xie, Zhipeng, Han, Zongyi, Wei, Xiangyi, Sun, Shiliang, Li, Yang, Zhao, Jing

Abstract

Vision-Language-Action (VLA) models have demonstrated strong capabilities in robotic manipulation, but their performance degrades significantly in long-horizon tasks due to cumulative error propagation. This limitation largely arises from static feature fusion mechanisms that rely on fixed weights to combine visual, language, and action representations, preventing the model from adapting to different phases of task execution. To address this limitation, we propose S$^2$-VLA, a framework that introduces a State-Space Guided Adaptive Attention (SSGAA) mechanism. SSGAA maintains a belief state that tracks task progression and generates dynamic gating weights to adaptively fuse information from three complementary sources visual features for spatial perception, task intents for high-level task planning, and temporal action sequences for execution consistency. This adaptive fusion allows the model to shift its focus throughout task execution, aligning with the evolving requirements of different task stages. Despite its compact 2B parameter size, S$^2$-VLA consistently outperforms larger 7B-scale models and achieves state-of-the-art performance on long-horizon manipulation benchmarks, including LIBERO and SimplerEnv. highlighting the importance of adaptive feature fusion for long-horizon robotic manipulation.

Chinese Translation

视觉-语言-动作（VLA）模型在机器人操作中展现了强大的能力，但在长时间任务中，由于累积误差传播，其性能显著下降。这一限制主要源于依赖固定权重的静态特征融合机制，该机制将视觉、语言和动作表示结合在一起，阻碍了模型适应任务执行不同阶段的能力。为了解决这一限制，我们提出了S$^2$-VLA框架，该框架引入了一种状态空间引导自适应注意力（SSGAA）机制。SSGAA维护一个信念状态，用于跟踪任务进展，并生成动态门控权重，以自适应地融合来自三个互补来源的信息：用于空间感知的视觉特征、高层任务规划的任务意图，以及用于执行一致性的时间动作序列。这种自适应融合使模型能够在任务执行过程中调整其关注点，以适应不同任务阶段不断变化的需求。尽管其参数规模仅为2B，S$^2$-VLA仍然持续超越更大规模的7B模型，并在长时间操作基准测试中取得了最先进的性能，包括LIBERO和SimplerEnv，突显了自适应特征融合在长时间机器人操作中的重要性。

View on arXiv Download PDF AI Translation

cs.RO / 18 / 2606.27883

Swarm sign language: motion-based communication between drones

群体手语：无人机之间的基于运动的通信

Rey, Thomas, Moras, Julien, Eudes, Alexandre, Manzanera, Antoine

Abstract

In stealth-constrained swarm robotics, visual communication provides a critical alternative to active radio transmissions, which might be jammed. This research investigates motion-based communication for non-active information exchange, utilizing modular, dynamically feasible planar trajectories as visual cues. On the receiver drone end, a pose estimator tracks the transmitting drone's pose, feeding it into our custom 3DTrajDecoder. The decoder is designed to classify and segment the spatiotemporal sequence while simultaneously regressing its size and normal vector. To robustly train the decoder on both communicative and non-communicative trajectories, we developed a configurable online procedural generation pipeline. We validate our system through real-world testing and simulation to define its operating domain, supported by an extensive ablation study detailing our architectural choices and system limitations.

Chinese Translation

在隐身受限的群体机器人中，视觉通信提供了一种关键的替代方案，以应对可能被干扰的主动无线电传输。本研究探讨了基于运动的通信，用于非主动信息交换，利用模块化、动态可行的平面轨迹作为视觉线索。在接收无人机端，姿态估计器跟踪发送无人机的姿态，并将其输入到我们定制的3DTrajDecoder中。该解码器旨在对时空序列进行分类和分段，同时回归其大小和法向量。为了在通信和非通信轨迹上稳健地训练解码器，我们开发了一个可配置的在线过程生成管道。我们通过实际测试和仿真验证了我们的系统，以定义其操作域，并通过广泛的消融研究支持我们的架构选择和系统限制。

View on arXiv Download PDF AI Translation

cs.RO / 19 / 2606.27914

Drifting in the Future: Stabilizing Path Following Drifting on High-Latency Vehicle Systems

未来的漂移：在高延迟车辆系统上稳定路径跟随漂移

Werner, Frederik, Heintzenberg, Till, Lienkamp, Markus, Betz, Johannes

Abstract

Autonomously controlling and handling a vehicle at and beyond its stability limit is a mathematically and computationally demanding task. Prior demonstrations of automated drifting have been limited to research platforms with instantaneous torque delivery and independently actuated wheels, leaving their applicability to production vehicles with actuator latencies and mechanically coupled axles uncertain. To overcome these issues, we design a predictor to compensate for powertrain delays, develop a revised control formulation to accommodate higher actuation latencies as well as a differential coupling on the driven axle, and introduce brake-based velocity stabilization. This paper presents the controller framework, the model extensions, and real-world experimental results. We observe that our controller enables a production sports car with a combustion engine to robustly sustain circular and figure-eight drifts, limiting lateral error to 1.1 m and sideslip overshoot to 0.06 rad despite actuator delays exceeding 250 ms, while mitigating oscillations and maintaining stable path and sideslip tracking. In conclusion, our results establish that autonomous drifting is feasible on production-ready vehicles, opening pathways to advanced safety systems capable of stabilizing cars in scenarios where traditional control fails.

Chinese Translation

在稳定性极限及其以上自主控制和处理车辆是一项在数学和计算上都要求极高的任务。以往的自动漂移演示仅限于具有瞬时扭矩输出和独立驱动轮的研究平台，这使得其在具有执行器延迟和机械耦合轴的量产车辆上的适用性不确定。为了解决这些问题，我们设计了一种预测器来补偿动力系统延迟，开发了一种修订的控制公式以适应更高的执行延迟以及驱动轴上的差动耦合，并引入了基于制动的速度稳定化。本文介绍了控制器框架、模型扩展以及实际实验结果。我们观察到，我们的控制器使一辆配备内燃机的量产跑车能够稳健地维持圆形和八字形漂移，将横向误差限制在1.1米，侧滑过冲限制在0.06弧度，尽管执行器延迟超过250毫秒，同时减轻了振荡并保持了稳定的路径和侧滑跟踪。总之，我们的结果表明，量产车辆上实现自主漂移是可行的，为能够在传统控制失效的情况下稳定车辆的先进安全系统开辟了新的途径。

View on arXiv Download PDF AI Translation

cs.RO / 20 / 2606.27929

When Multi-Robot Systems Meet Agentic AI:Towards Embodied Collective Intelligence

当多机器人系统遇上自主智能：迈向具身集体智能

Yan, Yuxuan, Jia, Yuanyuan, Yang, Qianqian

Abstract

Embodied AI is increasingly becoming agentic, shifting robots from perception--control pipelines towards closed-loop systems that can retrieve context, deliberate during execution, monitor feedback, and refine future behavior. In parallel, robotics research has also moved from single-robot autonomy towards multi-robot systems, driven by the need for wider sensing, distributed action, heterogeneous capabilities, and fault tolerance. As AI agents move from single-agent use towards multi-agent collaboration, robotics faces a parallel challenge: robot teams must move beyond sharing maps, task assignments, and datasets towards sharing the state produced by embodied agent loops. This article explores Embodied Collective Intelligence (ECI), a future multi-robot paradigm in which a robot team accumulates and uses world context, task progress, and skill experience as shared resources. Specifically, we first review how embodied AI is becoming agentic and how multi-robot cooperation has evolved. We then present Embodied Collective Intelligence through Co-Perception, Co-Action, and Co-Evolution. Finally, we use an illustrative navigation study to examine one concrete component of the concept: shared world-memory inheritance. The study shows that a newly added robot can benefit from merged team memory, but it is not intended as a full evaluation of the ECI framework. Taken together, the review and conceptual framework motivate Embodied Collective Intelligence as a direction for embodied multi-agent intelligence, while the case study grounds one measurable part of the concept.

Chinese Translation

具身人工智能正日益变得自主，将机器人从感知-控制管道转向能够检索上下文、在执行过程中进行推理、监测反馈并优化未来行为的闭环系统。与此同时，机器人研究也从单机器人自主性向多机器人系统发展，这一转变源于对更广泛感知、分布式行动、异构能力和容错性的需求。随着人工智能代理从单一代理使用转向多代理协作，机器人面临着一个相应的挑战：机器人团队必须超越共享地图、任务分配和数据集，转向共享由具身代理循环产生的状态。本文探讨了具身集体智能（Embodied Collective Intelligence, ECI），这一未来多机器人范式中，机器人团队将世界上下文、任务进展和技能经验作为共享资源进行积累和利用。具体而言，我们首先回顾了具身人工智能如何变得自主以及多机器人合作的演变。然后，我们通过共同感知（Co-Perception）、共同行动（Co-Action）和共同进化（Co-Evolution）来呈现具身集体智能。最后，我们通过一个示例导航研究来检验这一概念的一个具体组成部分：共享世界记忆继承。研究表明，新加入的机器人可以从合并的团队记忆中受益，但这并不旨在对ECI框架进行全面评估。综上所述，回顾和概念框架为具身多代理智能的方向提供了动机，而案例研究则为这一概念的一个可测量部分奠定了基础。

View on arXiv Download PDF AI Translation

cs.RO / 21 / 2606.27962

Building a Scalable, Reproducible, Evaluatable, and Closed-Loop Simulation Environment Foundation for Embodied Intelligence Cloud-Native Simulation Infrastructure for Embodied Intelligence Training, Evaluation, and Data Collection

构建可扩展、可重复、可评估的闭环仿真环境基础设施：面向具身智能的云原生仿真基础设施用于具身智能的训练、评估和数据收集

Xiong, Junwu, Guo, Yongjian, Luo, Mingxi, Qiao, Ning, Kang, Lei, Wang, Song, Gao, Yince

Abstract

This paper presents a cloud-native simulation infrastructure framework for embodied intelligence that supports large-scale training, standardized evaluation, and simulation-based data collection. The framework unifies simulation environment generation, task execution, trajectory collection, model evaluation, data management, and cloud services into a scalable and reproducible platform. To address the high cost, limited scalability, and poor reproducibility of real-world robotic data collection, the framework adopts cloud-native technologies including elastic resource scheduling, containerized simulation, unified data management, and service-oriented system design, enabling efficient large-scale simulation for multi-model and multi-task workloads. Built on a four-layer architecture, the framework provides standardized environment assets, automated task generation, trajectory collection, benchmark evaluation, and closed-loop data optimization. It further integrates representative systems including D-VLA, RL-VLA3, Sword, and Pre-VLA to support scalable simulation, dynamic scheduling, visual augmentation, and real-time data filtering. We argue that cloud-native simulation infrastructure provides a unified foundation for data generation, model training, standardized evaluation, and real-world deployment, and will play a key role in the future development of embodied intelligence.

Chinese Translation

本文提出了一种面向具身智能的云原生仿真基础设施框架，支持大规模训练、标准化评估和基于仿真的数据收集。该框架将仿真环境生成、任务执行、轨迹收集、模型评估、数据管理和云服务统一为一个可扩展和可重复的平台。为了应对现实世界机器人数据收集的高成本、有限的可扩展性和较差的可重复性，该框架采用了包括弹性资源调度、容器化仿真、统一数据管理和面向服务的系统设计等云原生技术，从而实现了多模型和多任务工作负载的高效大规模仿真。该框架基于四层架构，提供标准化的环境资产、自动化任务生成、轨迹收集、基准评估和闭环数据优化。它进一步整合了代表性系统，包括 D-VLA、RL-VLA3、Sword 和 Pre-VLA，以支持可扩展仿真、动态调度、视觉增强和实时数据过滤。我们认为，云原生仿真基础设施为数据生成、模型训练、标准化评估和现实世界部署提供了统一的基础，并将在具身智能的未来发展中发挥关键作用。

View on arXiv Download PDF AI Translation

cs.RO / 22 / 2606.28133

Translation as a Bridging Action: Transferring Manipulation Skills from Humans to Robots

作为桥接行动的翻译：将操作技能从人类转移到机器人

Chen, Sijin, Jiang, Kaixuan, Shi, Haixin, Wang, Yanhui, Zhong, Weiheng, Li, Haosheng, Jiang, Bo, Liu, Yuxiao, Liu, Xihui

Abstract

We study whether we can learn novel manipulation skills from human actions to a bi-manual robot with parallel grippers. Human action data is cheap, abundant, and diverse, making it one of the most promising resources for scaling up robot learning. Yet transferring skills from humans to robots remains hard: most prior work treats humans as just another bi-manual 6DoF embodiment, where hand-pose estimates are noisy and the contact patterns of human fingers differ fundamentally from those of a parallel gripper. We argue that learning rotation-inclusive action signals from human data is therefore sub-optimal, and instead propose a bridging action representation: the relative wrist translation within the initial head-camera frame, an action space shared by humans and robots. To handle the potential absence of certain action components in different embodiments, we build a $\pi_0$-like vision-language-action model with interleaved action tokens and attention masking. On a suite of novel bi-manual manipulation tasks, our bridging action transfers human manipulation knowledge to robots far more effectively than noisy 6DoF human actions and scales with the amount of human data.

Chinese Translation

我们研究了是否可以从人类动作中学习新颖的操作技能，并将其应用于具有平行夹爪的双手机器人。人类动作数据便宜、丰富且多样，使其成为扩展机器人学习的最有前景的资源之一。然而，将技能从人类转移到机器人仍然困难：大多数先前的研究将人类视为另一种双手6自由度（6DoF）体现，其中手部姿态估计噪声较大，而人类手指的接触模式与平行夹爪的接触模式在根本上存在差异。因此，我们认为从人类数据中学习包含旋转的动作信号是次优的，进而提出了一种桥接行动表示：在初始头部相机框架内的相对手腕位移，这是人类和机器人共享的一个动作空间。为了处理不同体现中某些动作组件可能缺失的情况，我们构建了一个类似于$ ext{π}_0$的视觉-语言-动作模型，采用交错的动作标记和注意力掩蔽。在一系列新颖的双手操作任务中，我们的桥接行动比噪声较大的6DoF人类动作更有效地将人类操作知识转移到机器人，并且随着人类数据量的增加而扩展。

View on arXiv Download PDF AI Translation

cs.RO / 23 / 2606.28192

PA-BiCoop: A Primary-Auxiliary Cooperative Framework for General Bimanual Manipulation

PA-BiCoop：一种用于一般双手操作的主辅协作框架

Qicheng, Bai, Ziru, Wang, Teli, Ma, Guang, Dai, Jingdong, Wang, Mengmeng, Wang

Abstract

Bimanual manipulation is essential for advanced robotic systems because it offers higher efficiency and flexibility compared to single-arm configurations. However, existing approaches either lack inter-arm interaction or ignore the need for a dynamic division of labor, treating the arms as functionally equivalent. To address these limitations, this paper draws inspiration from human bimanual manipulation where one arm handles core operations and the other provides auxiliary support, and proposes PA-BiCoop, a new single-model bimanual cooperation framework with dynamic primary-auxiliary arm differentiation. PA-BiCoop categorizes robotic arms into primary and auxiliary arms with adaptively adjustable roles across task stages, employs two specialized decoders that share a global feature encoder: the primary decoder generates the primary arm's base-coordinate pose and core-task affordance heatmaps, and the auxiliary decoder outputs the auxiliary arm's relative pose in the primary arm's coordinate system. Moreover, we design a dynamic role assignment module to automatically map roles to left/right arms without manual pre-definition. This design facilitates inter-arm knowledge sharing and coordinated manipulation. Extensive experiments demonstrate that our PA-BiCoop achieves superior performance: it outperforms state-of-the-art baselines by 48% on average in RLBench2 simulation tasks and by over 50% on average in real world tasks, thereby verifying its effectiveness and advancement in bimanual manipulation.

Chinese Translation

双手操作对于先进的机器人系统至关重要，因为它相比单臂配置提供了更高的效率和灵活性。然而，现有的方法要么缺乏臂间互动，要么忽视了动态分工的需求，将双臂视为功能上等同。为了解决这些局限性，本文借鉴了人类的双手操作，其中一只手负责核心操作，另一只手提供辅助支持，并提出了PA-BiCoop，一种具有动态主辅臂区分的新型单模型双手协作框架。PA-BiCoop将机器人臂分为主臂和辅臂，并在任务阶段之间自适应调整角色，采用两个共享全局特征编码器的专用解码器：主解码器生成主臂的基坐标姿态和核心任务可用性热图，而辅解码器则输出辅臂在主臂坐标系中的相对姿态。此外，我们设计了一个动态角色分配模块，能够自动将角色映射到左右臂，而无需手动预定义。该设计促进了臂间知识共享和协调操作。大量实验表明，我们的PA-BiCoop在性能上表现优越：在RLBench2模拟任务中，平均超越最先进的基线48%，在现实世界任务中平均超越50%以上，从而验证了其在双手操作中的有效性和先进性。

View on arXiv Download PDF AI Translation

cs.RO / 24 / 2606.28196

Learning Stable In-Grasp Manipulation in a Non-Dropping Action Space

在非掉落动作空间中学习稳定的抓取操控

Doan, Ha Thang Long, Arita, Hikaru, Nakashima, Kazuto, Tahara, Kenji

Abstract

Traditionally, dexterous manipulation controllers are designed using analytic models constrained by strong assumptions about the hand and the objects being manipulated. Reinforcement learning (RL) has become another common approach in which skills are explored openly in an end-to-end manner but is inefficient because of unnoticeable instability and conflicts in learning objectives. This paper attempts to efficiently explore stable and accurate manipulation skills by decomposing dexterous skills into multiple simpler/analyzable components. Each skill component is subsequently learned with constraints and guidance from classical physics and control theory. Our work shows that for stable grasp, in-grasp reposition/reorientation with different objects, sensor/motor noise, latency, and frictional conditions, skill learning becomes efficient and stable with prior knowledge from theory.

Chinese Translation

传统上，灵巧操控控制器是基于对手部和被操控物体的强假设设计的解析模型。强化学习（Reinforcement Learning, RL）成为另一种常见的方法，通过端到端的方式开放探索技能，但由于学习目标的不稳定性和冲突，效率较低。本文试图通过将灵巧技能分解为多个更简单/可分析的组件，来高效探索稳定且准确的操控技能。每个技能组件随后在经典物理和控制理论的约束和指导下进行学习。我们的研究表明，对于稳定的抓取，在不同物体、传感器/电机噪声、延迟和摩擦条件下的抓取重新定位/重新定向，技能学习在理论的先验知识支持下变得高效且稳定。

View on arXiv Download PDF AI Translation

cs.RO / 25 / 2606.28237

Unleashing Infinite Motion: Scaling Expressive Quadrupedal Motion via Generative Video Priors

释放无限运动：通过生成视频先验扩展表现力丰富的四足运动

Liu, Youzhi, Gao, Li, Qian, Yifei, Liu, Liu, Cai, Yang, Li, Ziqiao

Abstract

Quadruped robots have achieved remarkable locomotion, yet their behavioral repertoire remains confined to a few gaits--far from the expressive, companion-like presence long envisioned for them. Attempts to import the humanoid recipe of large-scale motion data have inherited one tacit assumption: that robot motion must first pass through an animal body, making data collection dependent on cooperative animals, reconstruction fragile across species, and retargeting ill-posed across incompatible morphologies. We propose Uni-Mo, a fully automated pipeline that removes the animal from the loop by reframing data scarcity as a generation problem: an LLM proposes motion prompts, a video diffusion model synthesizes the corresponding robot behaviors, and the generated videos are lifted into 3D reference trajectories used to train tracking policies deployed on a real Unitree Go2. To make naively-drifting generations reliably extractable, we introduce an Identity Consistency Loss that enforces appearance coherence across frames. We release Quad-Imaginarium at https://github.com/GaoLii/Quad-Imaginarium.git, the resulting open-source dataset of 7,488 language-annotated quadruped motions (18.5 hours) spanning acrobatic and performative behaviors. We validate 392 randomly sampled motions on a real Unitree Go2 with a 96.7% deployment success rate, complemented by a 97.6% success rate across the full dataset in simulation.

Chinese Translation

四足机器人在运动能力上取得了显著的进展，但其行为范围仍然局限于少数几种步态，远未达到人们长期设想的表现力丰富、如同伴般的存在。尝试引入类人机器人大规模运动数据的做法，隐含着一个默默的假设：机器人运动必须首先通过动物身体，这使得数据收集依赖于合作动物，跨物种的重建变得脆弱，而在不兼容的形态上进行重定向则显得不适当。我们提出了Uni-Mo，一个完全自动化的流程，通过将数据稀缺重新框定为生成问题，消除了动物的参与：一个大型语言模型（LLM）提出运动提示，一个视频扩散模型合成相应的机器人行为，生成的视频被提升为用于训练在真实Unitree Go2上部署的跟踪策略的3D参考轨迹。为了使简单漂移的生成结果能够可靠提取，我们引入了一种身份一致性损失（Identity Consistency Loss），以强制实现帧间外观的一致性。我们在 https://github.com/GaoLii/Quad-Imaginarium.git 发布了Quad-Imaginarium，这是一个包含7488个语言注释的四足运动（18.5小时）的开源数据集，涵盖了杂技和表演行为。我们在真实的Unitree Go2上验证了392个随机采样的运动，部署成功率为96.7%，在整个数据集的模拟中成功率为97.6%。

View on arXiv Download PDF AI Translation

cs.RO / 26 / 2606.28276

SimFoundry: Modular and Automated Scene Generation for Policy Learning and Evaluation

SimFoundry：用于政策学习和评估的模块化自动场景生成

Ranawaka, Nadun, Wong, Josiah, Pai, Wei-Lin, Chu, Wei-Teng, Dai, Tianyuan, Moghani, Masoud, Yin, Hang, Jiang, Yunfan, Durbano, Wesley, Huynh, Brandon, Fang, Yu, Fan, Linxi, Xu, Danfei, Zhang, Ruohan, Fei-Fei, Li, Wen, Bowen, Mandlekar, Ajay, Zhu, Yuke

Abstract

Training and evaluating robot policies in the real world is costly and difficult to scale. We introduce SimFoundry, a modular and automated system for zero-shot real-to-sim scene construction from a video. SimFoundry generates sim-ready digital twins and supports object, scene, and task editing, enabling the automated generation of diverse digital cousins: affordance-preserving variations of reconstructed real-world scenes. Policies trained on SimFoundry data transfer zero-shot to challenging real tasks involving multi-step manipulation, articulated object interaction, and bimanual interaction, and its digital cousins (variations of the original scene, objects, and tasks) facilitate generalization to new real-world conditions. Across 7 manipulation tasks and 5 policy architectures, SimFoundry simulation evaluations strongly predict real-world performance, with mean Pearson correlation 0.911 and mean maximum ranking violation 0.018. When evaluating sim-trained policies zero-shot in the real world, policies trained with object, scene, and task cousins in simulation show average task success rate improvements of 17%, 21%, and 40%, respectively. Additional details at https://research.nvidia.com/labs/gear/simfoundry/ .

Chinese Translation

在现实世界中训练和评估机器人策略既昂贵又难以扩展。我们介绍了SimFoundry，一个用于从视频中进行零-shot真实到仿真场景构建的模块化自动化系统。SimFoundry生成适合仿真的数字双胞胎，并支持对象、场景和任务的编辑，从而实现多样化数字副本的自动生成：保留功能的重建现实世界场景的变体。在SimFoundry数据上训练的策略能够零-shot转移到涉及多步骤操作、关节对象交互和双手交互的具有挑战性的真实任务中，其数字副本（原始场景、对象和任务的变体）有助于在新的现实世界条件下的泛化。在7个操作任务和5种策略架构中，SimFoundry的仿真评估强烈预测现实世界的表现，平均皮尔逊相关系数为0.911，平均最大排名违反为0.018。在现实世界中零-shot评估经过仿真训练的策略时，使用对象、场景和任务副本进行训练的策略显示出平均任务成功率分别提高了17%、21%和40%。更多详细信息请访问 https://research.nvidia.com/labs/gear/simfoundry/ 。

View on arXiv Download PDF AI Translation

cs.RO / 27 / 2606.28300

CacheMPC: Certified Cached Model Predictive Control for Quadruped Locomotion

CacheMPC：四足运动的认证缓存模型预测控制

Khandelwal, Nimesh, Anand, Mehul, Gupta, Shakti S., Kothari, Mangal

Abstract

Model Predictive Control (MPC) is the standard predictive layer in hierarchical quadruped controllers, but the per-cycle QP solve limits the update rate achievable on embedded processors. Because legged gaits revisit a bounded region of state space, MPC solutions admit caching and reuse. This paper proposes \emph{Certified CacheMPC}: a Locality-Sensitive-Hashed cache of horizon contact-force trajectories, partitioned by contact mode, retrieved at query time and accepted only when an a-posteriori per-query certificate confirms primal feasibility and a Lagrangian dual-gap upper bound on cost suboptimality. A bounded-budget controller schedule combines top-$K$ certified retrieval, a deadline-bounded QP solve, and a shifted last-certified fallback. The framework is evaluated on a Unitree Go2 across $2{,}038$ usable cold-controller MuJoCo trials, including a $600$-trial $n\!=\!50$ campaign at three failure-boundary cells, and a first-deploy session on the on-robot NVIDIA Orin NX. The un-gated cache delivers a $25\times$ median solve-time speedup in simulation and an $18.7\times$ median speedup on hardware. At $n\!=\!50$ no statistically significant difference in closed-loop stable rate is detected between the cache variants and the no-cache baseline at any tested cell. The certificate's contribution to closed-loop safety is not resolvable at the present sample size.

Chinese Translation

模型预测控制（MPC）是分层四足控制器中的标准预测层，但每个周期的二次规划（QP）求解限制了嵌入式处理器可实现的更新速率。由于腿部步态会重新访问有限的状态空间区域，MPC 解决方案允许缓存和重用。本文提出了 extit{认证缓存模型预测控制（Certified CacheMPC）}：一种基于局部敏感哈希（Locality-Sensitive Hashing）的接触力轨迹缓存，按接触模式进行分区，在查询时检索，仅在后验每查询证书确认原始可行性和拉格朗日对偶成本次优性的上界时接受。一个有预算限制的控制器调度结合了前$K$个认证检索、一个截止时间限制的QP求解和一个偏移的最后认证回退。该框架在Unitree Go2上进行了评估，共进行了$2{,}038$次可用的冷控制器MuJoCo试验，包括在三个失败边界单元进行的$600$次试验的$n eq50$活动，以及在机器人上的NVIDIA Orin NX的首次部署会话。未加门控的缓存在仿真中提供了$25 imes$的中位数求解时间加速，在硬件上提供了$18.7 imes$的中位数加速。在$n=50$时，在任何测试单元中，缓存变体与无缓存基线之间的闭环稳定率没有检测到统计显著差异。证书对闭环安全性的贡献在当前样本大小下无法解决。

View on arXiv Download PDF AI Translation

cs.RO / 28 / 2606.28320

WARP-RM: A Warp-Augmented Relative Progress Reward Model for Data Curation

WARP-RM：一种用于数据整理的扭曲增强相对进展奖励模型

Yu, Justin, Goldberg, Andrew, Kondap, Kavish, El-Refai, Karim, Ransing, Ethan, Chen, Qianzhong, Schwager, Mac, Shentu, Fred, Wu, Philipp, Goldberg, Ken

Abstract

Scaling imitation learning requires large datasets, yet human teleoperation inevitably produces mixed-quality demonstrations containing hesitations and recoveries. Prior frame-level progress reward models supervise on absolute temporal progress proxies that suffer from label noise, or require costly human annotations to define subtask boundaries. We present WARP (Warp-Augmented Relative Progress), a novel fully self-supervised algorithm for learning dense, signed relative progress magnitudes directly from successful demonstrations. WARP generates per-frame progress targets via time-warp augmentations of demonstrations (variable playback speeds and reversals) and we train WARP-RM to predict the normalized elapsed time between input frames. Aggregating these predictions across overlapping windows yields a dense frame-level progress signal. We then introduce WARP-BC, which leverages these scalar reward estimates to upweight high-advantage action chunks during behavior cloning, where chunk-level advantage is obtained by aggregating per-frame rewards. We evaluate our approach on a physical bimanual robot system performing a long-horizon deformable object manipulation task: folding T-shirts from a random crumpled start. To evaluate policy robustness against suboptimal data, we construct training datasets of varying quality using episode length as a proxy for teleoperation sub-optimality. As the dataset is widened to admit more inefficiencies, WARP-BC maintains a 19/20 success rate compared to vanilla BC's collapse to 2/20, improving throughput by up to 18x.

Chinese Translation

扩展模仿学习需要大量数据集，但人类遥控操作不可避免地产生混合质量的演示，其中包含犹豫和恢复。之前的帧级进展奖励模型依赖于绝对时间进展代理，这些代理受到标签噪声的影响，或者需要昂贵的人类注释来定义子任务边界。我们提出了WARP（扭曲增强相对进展），这是一种新颖的完全自我监督算法，旨在直接从成功演示中学习密集的、有符号的相对进展幅度。WARP通过对演示进行时间扭曲增强（可变播放速度和反转）生成每帧进展目标，并且我们训练WARP-RM以预测输入帧之间的标准化经过时间。通过在重叠窗口中聚合这些预测，产生密集的帧级进展信号。然后，我们引入WARP-BC，它利用这些标量奖励估计在行为克隆过程中对高优势动作块进行加权，其中块级优势是通过聚合每帧奖励获得的。我们在一个物理双手机器人系统上评估我们的方法，该系统执行一个长时间范围的可变形物体操作任务：从随机皱褶的起始状态折叠T恤。为了评估策略对次优数据的鲁棒性，我们构建了不同质量的训练数据集，使用剧集长度作为遥控次优性的代理。随着数据集的扩大以容纳更多低效，WARP-BC保持了19/20的成功率，而普通BC则崩溃至2/20，吞吐量提高了多达18倍。

View on arXiv Download PDF AI Translation

cs.RO / 29 / 2606.28323

DexCompose: Reusing Dexterous Policies for Multi-Task Manipulation with a Single Hand

DexCompose：利用灵巧策略在单手多任务操作中的重用

Huang, Dihong, Wei, Zhenyu, Xu, Zhuxiu, Yao, Yunchao, Li, Sikai, Ding, Mingyu

Abstract

Dexterous manipulation policies can solve individual skills, but composing them to perform multiple tasks with a single hand remains challenging. Adding a new task on top of an existing manipulation skill often imposes conflicting demands on overlapping fingers and contact modes, causing destructive interference between preserving an existing manipulation outcome and executing a new one. We propose DexCompose, a role-aware residual composition framework that reuses pretrained dexterous policies for multi-task manipulation through explicit finger-level action ownership. Given two pretrained full-hand policies, DexCompose first collects successful post-task states from the first skill and performs release tests over candidate finger masks to identify which fingers are necessary for maintaining the established skill state. It then trains two asymmetric residual modules: a bounded residual stabilizer for task preservation, and a context-aware residual that adapts the frozen downstream policy only within the action subspace assigned to the new task. We evaluate the framework on 16 composite dexterous manipulation tasks spanning four object-retention skills and four downstream interactions. DexCompose achieves a 77.4% average composite success rate, demonstrating that structural action ownership with dual residuals offers a promising direction for composing dexterous skills beyond conventional policy chaining.

Chinese Translation

灵巧操作策略能够解决单个技能，但将它们组合以使用单手执行多个任务仍然具有挑战性。在现有操作技能的基础上添加新任务往往会对重叠的手指和接触模式施加相互冲突的要求，从而导致在保持现有操作结果和执行新任务之间的破坏性干扰。我们提出了DexCompose，一种角色感知的残差组合框架，通过显式的手指级动作归属重用预训练的灵巧策略以实现多任务操作。给定两个预训练的全手策略，DexCompose首先从第一个技能中收集成功的后任务状态，并对候选手指掩码进行释放测试，以识别哪些手指对于维持已建立的技能状态是必要的。然后，它训练两个不对称的残差模块：一个用于任务保持的有界残差稳定器，以及一个上下文感知的残差，该残差仅在分配给新任务的动作子空间内调整冻结的下游策略。我们在涵盖四个物体保持技能和四个下游交互的16个复合灵巧操作任务上评估了该框架。DexCompose实现了77.4%的平均复合成功率，证明了具有双重残差的结构性动作归属为超越传统策略链的灵巧技能组合提供了一个有前景的方向。

View on arXiv Download PDF AI Translation

计算机视觉 (Computer Vision)

cs.CV / 1 / 2606.27412

Not All Relations Rotate Alike: Transformation-Aware Decoupling for Viewpoint-Robust 3D Scene Graph Generation

并非所有关系都以相同方式旋转：面向变换的解耦以实现视角鲁棒的三维场景图生成

Sun, Jingjun, Wang, Chaowei, Liu, Zhirui, Tian, Jiaxu, Yang, Ming, Wang, Yaoxing, Gao, Shan

Abstract

3D Scene Graph Generation (3DSGG) represents 3D scenes as structured object-relation-object graphs, providing a compact relational abstraction for spatial understanding. In embodied intelligence settings, the same 3D scene may be observed by agents from viewpoints that differ by yaw rotations. However, current 3DSGG models often fail to produce relation predictions that follow the expected transformation behavior under such viewpoint shifts. This behavior reveals an empirical mismatch related to predicate-level transformation heterogeneity: directional predicates such as left, front, right, and behind should transform with the observation frame, whereas most contact, support, and semantic predicates such as standing on and attached to should remain stable. To reduce this mismatch, we propose Transformation-Aware Decoupling (TAD), a viewpoint-robust 3DSGG framework that decouples relation reasoning according to predicate transformation behavior and is supported by viewpoint-stable object representations. TAD decomposes relation reasoning into two parts: one learns cues that should stay stable across viewpoints, while the other learns directional cues that should change with the observation frame. The two parts are merged for standard multi-label predicate prediction. Transformation-specific descriptors and group-aware auxiliary supervision encourage the two branches to capture complementary relation cues. Extensive experiments on 3DSSG show that TAD achieves state-of-the-art robustness under yaw viewpoint changes without training-time rotation augmentation, while maintaining competitive performance under the standard benchmark. The project page is available at https://tad-predicate.github.io/.

Chinese Translation

三维场景图生成（3DSGG）将三维场景表示为结构化的对象-关系-对象图，为空间理解提供了紧凑的关系抽象。在具身智能环境中，同一三维场景可能被代理从不同的视角观察，这些视角之间存在偏航旋转的差异。然而，目前的3DSGG模型往往无法在这种视角变化下产生符合预期变换行为的关系预测。这种行为揭示了与谓词级变换异质性相关的经验性不匹配：方向性谓词如左、前、右和后应随着观察框架而变换，而大多数接触、支持和语义谓词如站立在和附着于则应保持稳定。为了减少这种不匹配，我们提出了面向变换的解耦（Transformation-Aware Decoupling, TAD），这是一个视角鲁棒的3DSGG框架，根据谓词变换行为解耦关系推理，并由视角稳定的对象表示支持。TAD将关系推理分解为两个部分：一个学习在不同视角下应保持稳定的线索，另一个学习应随观察框架变化的方向性线索。这两个部分合并用于标准的多标签谓词预测。特定于变换的描述符和群体感知的辅助监督鼓励这两个分支捕捉互补的关系线索。在3DSSG上的大量实验表明，TAD在偏航视角变化下实现了最先进的鲁棒性，而无需在训练时进行旋转增强，同时在标准基准测试中保持了竞争力的性能。项目页面可访问 https://tad-predicate.github.io/。

View on arXiv Download PDF AI Translation

cs.CV / 2 / 2606.27444

SemCityLoc: Aerial 6DoF Localization Using Semantic 3D City Models

SemCityLoc：基于语义3D城市模型的空中6自由度定位

Mao, Jingfeng, Chen, Xuyang, Zhang, Qilin, Dhaouadi, Oussema, Wang, Guangming, Sheil, Brian, Cremers, Daniel, Xia, Yan, Wysocki, Olaf

Abstract

Aerial 6DoF localization typically relies on precise GNSS signals or radiometrically rich 3D reconstructions, limiting scalability and on-board deployment. We propose SemCityLoc, a semantic-geometric alignment system that reframes aerial pose estimation as structured surface registration between foundation-model-derived visual priors and standardized LoD-compliant 3D city models. Instead of matching sparse contours or dense texture, our method aligns semantic surfaces and monocular depth with lightweight semantic 3D building models, increasing pose discriminability in repetitive and occluded urban environments. To enable accurate evaluation, we introduce SemCityLockeD, the first real-world benchmark combining centimeter-accurate UAV poses with standardized LoD1--LoD3 semantic city models and challenging low-altitude imagery. Experiments demonstrate substantial improvements over existing map-based approaches, improving recall by up to 36% and reducing mean positional error from 9.89m to 2.62m in challenging urban canyons. Our results indicate that semantically structured geometry provides sufficient and scalable constraints for high-precision aerial localization without radiometric scene reconstructions. The code and data are available at https://albertchen98.github.io/SemCityLoc.

Chinese Translation

空中6自由度定位通常依赖于精确的全球导航卫星系统（GNSS）信号或具有丰富辐射特性的3D重建，这限制了其可扩展性和机载部署。我们提出了SemCityLoc，一种语义几何对齐系统，将空中姿态估计重新定义为基础模型派生的视觉先验与标准化的符合层次细节（LoD）3D城市模型之间的结构化表面配准。我们的方法不是匹配稀疏轮廓或密集纹理，而是对齐语义表面和单目深度与轻量级语义3D建筑模型，从而提高在重复和遮挡的城市环境中的姿态可辨识性。为了实现准确评估，我们引入了SemCityLockeD，这是第一个将厘米级精度的无人机（UAV）姿态与标准化的LoD1至LoD3语义城市模型以及具有挑战性的低空影像相结合的真实世界基准。实验表明，与现有的基于地图的方法相比，显著提高了召回率，提升幅度高达36%，并将在具有挑战性的城市峡谷中的平均位置误差从9.89米降低至2.62米。我们的结果表明，语义结构化几何为高精度空中定位提供了足够且可扩展的约束，而无需辐射场景重建。代码和数据可在 https://albertchen98.github.io/SemCityLoc 获取。

View on arXiv Download PDF AI Translation

cs.CV / 3 / 2606.27484

Fine-tuning a multimodal large language model for clinician-grade autism behavioral scoring from short home videos

针对临床级自闭症行为评分的多模态大型语言模型的微调：基于短视频的研究

Honarmand, Mohammadmahdi, Azizian, Parnian, Kline, Aaron, Nurge, Kae, Tumpa, Zerin Nasrin, Surabhi, Saimourya, Dunlap, Kaitlyn, Qian, Yang, Kargarandehkordi, Ali, Neupane, Sameer, Washington, Peter, Wall, Dennis P.

Abstract

Autism spectrum disorder (ASD) affects 1 in 31 US children, yet median age at diagnosis exceeds four years. Artificial intelligence pipelines that provide quantified diagnosis using easy to access observational data (e.g., home videos) could help with earlier diagnosis, and timely delivery of early treatments. We fine-tuned Gemini 2.5 Pro on 400 clinician-rated home videos with low-rank adaptation, training only on 30 behavioral features previously validated to produce reliable predictions when passed to various ML models. On 99 held-out children (49 ASD, 50 neurotypical), inter-rater reliability with clinicians (per-feature weighted Cohen's kappa) improved by 40% (p<0.001), with 27 of 28 evaluable features improving. As an emergent zero-shot capability, direct ASD diagnosis F1 improved by 53% (p<0.001), matching or exceeding clinician outcomes. Classifier-assisted pipelines using fine-tuned LLM-derived behavioral features matched clinician-scored inputs across all tested pathways and achieved 77% accuracy (95% CI: 68-85%) and an AUC of 86% (95% CI: 78-92%). Fine-tuned multimodal LLMs can serve as scalable behavioral feature extractors for use in autism assessment and diagnosis.

Chinese Translation

自闭症谱系障碍（ASD）影响着美国每31名儿童中的1名，但诊断的中位年龄超过四岁。利用易于获取的观察数据（例如家庭视频）提供量化诊断的人工智能管道可以帮助实现更早的诊断和及时的早期治疗。我们在400个由临床医生评分的家庭视频上对Gemini 2.5 Pro进行了微调，采用低秩适应，仅在30个先前验证的行为特征上进行训练，这些特征在传递给各种机器学习模型时能够产生可靠的预测。在99名被排除的儿童（49名ASD，50名神经典型）中，与临床医生的评分一致性（每个特征加权的Cohen's kappa）提高了40%（p<0.001），28个可评估特征中有27个得到了改善。作为一种新兴的零样本能力，直接的ASD诊断F1值提高了53%（p<0.001），与临床医生的结果相匹配或超过。使用微调的LLM衍生行为特征的分类器辅助管道在所有测试路径中与临床评分输入相匹配，达到了77%的准确率（95% CI: 68-85%）和86%的AUC（95% CI: 78-92%）。微调的多模态大型语言模型可以作为可扩展的行为特征提取器，用于自闭症评估和诊断。

View on arXiv Download PDF AI Translation

cs.CV / 4 / 2606.27491

SelectAnyTree: A Promptable Instance Segmentation Model for 3D Forest LiDAR Point Clouds

SelectAnyTree：一种可提示的3D森林LiDAR点云实例分割模型

Nguyen, Trung Thanh, Lusk, Daniel, Gerberding, Kilian, Vajna-Jehle, Janusch, Vu, Tuan-Anh, Le, Duc Viet, Vo, Tu, Nguyen, Phi Le, Kawanishi, Yasutomo, Komamizu, Takahiro, Ide, Ichiro, Frey, Julian, Kattenborn, Teja

Abstract

Automated instance segmentation of forest LiDAR point clouds is increasingly critical as forest monitoring moves toward scalable, detailed, 3D measurement. Yet, progress is constrained by label scarcity for tree instances; a single hectare can hold millions of points and hundreds of overlapping, complex crowns, making manual annotation from scratch with raw data laborious and error-prone. Annotations are often corrected from automatic pre-segmentations, but remain costly as these provide no interactive or AI-assisted refinement. Inspired by the promptable paradigm of foundation segmentation models, we propose SelectAnyTree, a promptable instance segmentation model that delineates any individual tree in a 3D forest point cloud from a few clicks. It introduces two key components: Click-to-query prompt encoder and Canopy Height Model (CHM)-guided first prompt. The former turns each click into a single content query, encoding its 3D position and positive/negative polarity together with a pooled local backbone feature. The latter provides treetops as a geometry- and ecologically guided first prompt without any user input. The resulting prompt query is then decoded into one tree mask by a state-space query decoder to efficiently capture long-range context in large-scale forest scenes with linear-time complexity. We evaluate SelectAnyTree in interactive and instance-level settings across seven diverse forest regions and an independent held-out test dataset, demonstrating strong generalization beyond the training domains. It segments a target tree to 78.2 Intersection over Union (IoU) from a single click, 24.8 points above the strongest promptable baseline, and reaches every accuracy target with the fewest clicks, while using far fewer parameters and less inference time than prior promptable models. The source code is available at https://github.com/thanhhff/SelectAnyTree.

Chinese Translation

随着森林监测向可扩展、详细的3D测量发展，森林LiDAR点云的自动实例分割变得愈发重要。然而，树木实例的标签稀缺限制了这一进展；单公顷土地上可能包含数百万个点和数百个重叠的复杂树冠，使得从原始数据手动标注变得繁琐且容易出错。注释通常是从自动预分割中纠正的，但由于这些方法没有提供交互式或AI辅助的细化，成本依然高昂。受到基础分割模型的可提示范式的启发，我们提出了SelectAnyTree，这是一种可提示的实例分割模型，可以通过少量点击在3D森林点云中勾勒出任何单独的树木。该模型引入了两个关键组件：点击查询提示编码器和冠层高度模型（Canopy Height Model, CHM）引导的首个提示。前者将每次点击转换为一个单一的内容查询，编码其3D位置和正/负极性，以及一个汇聚的局部主干特征。后者在没有任何用户输入的情况下提供树顶作为几何和生态引导的首个提示。生成的提示查询随后通过状态空间查询解码器解码为一个树木掩膜，以高效捕捉大规模森林场景中的长距离上下文，且具有线性时间复杂度。我们在七个不同的森林区域和一个独立的保留测试数据集上评估了SelectAnyTree在交互式和实例级设置中的表现，展示了其在训练领域之外的强泛化能力。它在单次点击下将目标树的交并比（Intersection over Union, IoU）分割到78.2，超出最强可提示基线24.8个点，并以最少的点击次数达到每个准确性目标，同时使用的参数更少，推理时间也比之前的可提示模型更短。源代码可在https://github.com/thanhhff/SelectAnyTree获取。

View on arXiv Download PDF AI Translation

cs.CV / 5 / 2606.27499

DMV-Bench: Diagnosing Long-Horizon Multimodal Agents' Visual Memory with Incidental Cue Injection

DMV-Bench：通过偶然线索注入诊断长时间跨度的多模态智能体视觉记忆

Tang, Yujin, Shang, Chenming, Xu, Ruize, Singh, Nikhil

Abstract

Research on agent memory has matured rapidly, but almost entirely on the text side: few existing benchmarks ask, in an interactive environment, when an agent genuinely needs to remember what it saw rather than what it could write down. We introduce DMV-Bench (Code: https://github.com/yyyujintang/DMV-Bench), the first interactive benchmark for multimodal-agent visual memory. DMV-Bench is built on a controlled home-furnishing e-commerce catalogue of 1,000 product variants in which a text-leakage contract keeps the discriminative signal of each task in the pixels alone. Across a chain of autonomous shopping sessions, every visited product image carries a unique, pre-rendered incidental cue, and the agent is later asked to recall a particular cued product and navigate to its URL. Inspired by dual-coding theory, we propose DualMem, a memory architecture that maintains a visual and a verbal code in parallel. On DMV-Bench, DualMem outperforms a caption baseline and three recent multimodal agent-memory systems at every chain length J in {5, 10, 15, 50} on both Gemini 2.5 Flash and Qwen2.5-VL-7B, with the lead surviving controls for memory-bank size and encoding-position bias, and an asymmetric dual-coding regime in which vision carries the cue end-to-end while the verbal channel plays a smaller query-grounding role.

Chinese Translation

关于智能体记忆的研究迅速发展，但几乎完全集中在文本方面：现有的基准几乎没有在交互环境中询问智能体何时真正需要记住它所看到的内容，而不是它可以写下的内容。我们引入了 DMV-Bench（代码：https://github.com/yyyujintang/DMV-Bench），这是第一个针对多模态智能体视觉记忆的交互基准。DMV-Bench 基于一个受控的家居电商目录，包含 1000 种产品变体，其中文本泄漏合同仅在像素中保留每个任务的区分信号。在一系列自主购物会话中，每个访问的产品图像都携带一个独特的预渲染偶然线索，随后要求智能体回忆特定的提示产品并导航到其 URL。受到双重编码理论的启发，我们提出了 DualMem，这是一种同时维护视觉和语言编码的记忆架构。在 DMV-Bench 上，DualMem 在每个链长度 J ∈ {5, 10, 15, 50} 上均优于一个标题基线和三个近期的多模态智能体记忆系统，在 Gemini 2.5 Flash 和 Qwen2.5-VL-7B 上表现出色，且在记忆库大小和编码位置偏差的控制下，采用不对称的双重编码机制，其中视觉通道在端到端中传递线索，而语言通道则发挥较小的查询基础作用。

View on arXiv Download PDF AI Translation

cs.CV / 6 / 2606.27500

Aloe-Vision: Robust Vision-Language Models for Healthcare

Aloe-Vision：面向医疗保健的稳健视觉-语言模型

Guasch-Martí, Jaume, Lopez-Cuena, Enrique, Suárez-Fernández, Martín, Bayarri-Planas, Jordi, Arias-Duart, Anna, Garcia-Gasulla, Dario

Abstract

Large Vision-Language Models (LVLMs) specialized in healthcare are emerging as a promising research direction due to their potential impact in clinical and biomedical applications. However, progress is constrained by the scarcity of high-quality medical multimodal data, concerns about robustness in safety-critical settings, and the narrow and potentially contaminated evaluation benchmarks that limit reliable assessment. To address these issues, the field requires state-of-the-art solutions to be fully open and reproducible systems in which all components can be inspected, evaluated, and improved. This work introduces Aloe-Vision-Data, a large-scale, quality-filtered mixture which integrates both medical and general domains across multimodal and text-only sources, designed for direct use in model fine-tuning. Building on this dataset, we train the Aloe-Vision family of medical LVLMs, openly released with full weights, training recipes and data, in two scales (7B and 72B). Through comprehensive benchmarking, we demonstrate that high quality training mixtures produce balanced LVLMs which yield significant gains over the baseline models without compromising general capabilities, achieving competitive performance with respect to state-of-the-art alternatives. To support reliable evaluation, we introduce CareQA-Vision, a carefully curated vision benchmark derived from MIR and EIR exams, the residency entrance exams for medical and nursing specialists in Spain, offering novel vision questions with low likelihood of contamination. Finally, we show that current LVLMs remain vulnerable to adversarial and misleading inputs, underscoring reliability challenges in clinical contexts.

Chinese Translation

专注于医疗保健的大型视觉-语言模型（LVLMs）作为一种有前景的研究方向正在兴起，因为它们在临床和生物医学应用中的潜在影响。然而，进展受到高质量医疗多模态数据稀缺、安全关键环境中的稳健性问题以及狭窄且可能受到污染的评估基准的限制，这些因素限制了可靠评估的进行。为了解决这些问题，该领域需要最先进的解决方案，以实现完全开放和可重复的系统，其中所有组件均可被检查、评估和改进。本研究介绍了Aloe-Vision-Data，这是一个大规模的、经过质量筛选的混合数据集，整合了医疗和一般领域的多模态及文本源，旨在直接用于模型微调。在此数据集的基础上，我们训练了Aloe-Vision系列医疗LVLMs，并公开发布了完整的权重、训练配方和数据，分为两个规模（7B和72B）。通过全面的基准测试，我们证明高质量的训练混合物能够产生平衡的LVLMs，这些模型在不妥协通用能力的情况下，相较于基线模型取得了显著的提升，并在与最先进的替代方案相比时表现出竞争力。为了支持可靠的评估，我们引入了CareQA-Vision，这是一个经过精心策划的视觉基准，源自西班牙医学和护理专业的住院医师入学考试（MIR和EIR），提供了低污染可能性的创新视觉问题。最后，我们展示了当前的LVLMs仍然容易受到对抗性和误导性输入的影响，强调了临床环境中可靠性面临的挑战。

View on arXiv Download PDF AI Translation

cs.CV / 7 / 2606.27504

ReWorld: Learning Better Representations for World Action Models

ReWorld：为世界行动模型学习更好的表征

Xia, Tianze, Zhou, Lijun, Xiong, Kaixin, Yao, Jingfeng, Zhu, Yu, Zhu, Zhenxin, Wang, Bing, Chen, Guang, Ye, Hangjun, Liu, Wenyu, Sun, Haiyang, Wang, Xinggang

Abstract

World Action Models (WAMs) model future environment evolution under action conditioning, offering a scalable paradigm for autonomous driving. However, existing approaches focus largely on model architecture design, and how a WAM can efficiently learn better world representations for planning remains underexplored. To address this gap, we propose ReWorld, the first representation learning framework specifically designed for autonomous-driving world action models. In WAMs, standard training supervises only the output ends of the generation and planning modules, leaving the intermediate representations that carry world knowledge to be shaped only indirectly, as byproducts of fitting these outputs. The core idea of ReWorld is to treat intermediate representations as direct targets of optimization, shaping them along three complementary dimensions. On the Video DiT responsible for generation, we impose future-predictive supervision on its intermediate representations. On the Action DiT responsible for planning, we first align its intermediate representations cross-modally with the video world representation, then further shape them to be discriminative around safety-critical boundaries via hard-negative supervision. In addition, we systematically analyze the effectiveness of existing representation learning methods in video generation world models, and discuss why their performance is limited on this task. Experiments on nuScenes and NAVSIM show that ReWorld improves fine-tuned video generation by 23.9% in FVD (81.3 to 61.9), raises closed-loop PDMS from 89.1 to 90.4 without any post-training such as RL or post-processing, and accelerates from-scratch convergence by approximately 2x.

Chinese Translation

世界行动模型（WAMs）在行动条件下模拟未来环境演变，为自动驾驶提供了一种可扩展的范式。然而，现有的方法主要集中在模型架构设计上，而WAM如何有效学习更好的世界表征以进行规划仍然未被充分探索。为了解决这一问题，我们提出了ReWorld，这是第一个专门为自动驾驶世界行动模型设计的表征学习框架。在WAM中，标准训练仅对生成和规划模块的输出端进行监督，导致承载世界知识的中间表征仅通过适应这些输出的副产品间接形成。ReWorld的核心思想是将中间表征视为优化的直接目标，从三个互补维度对其进行塑造。在负责生成的Video DiT上，我们对其中间表征施加未来预测监督。在负责规划的Action DiT上，我们首先将其中间表征与视频世界表征进行跨模态对齐，然后通过困难负样本监督进一步塑造它们，使其在安全关键边界附近具有区分性。此外，我们系统地分析了现有表征学习方法在视频生成世界模型中的有效性，并讨论了其在该任务上表现受限的原因。在nuScenes和NAVSIM上的实验表明，ReWorld在FVD上提高了微调视频生成的性能，提升幅度达到23.9%（从81.3降至61.9），将闭环PDMS从89.1提高到90.4，而无需任何后训练（如强化学习或后处理），并将从头收敛的速度加快了约2倍。

View on arXiv Download PDF AI Translation

cs.CV / 8 / 2606.27505

TruEye: Fine-Grained Detection of AI-Generated Human Subjects in Images

TruEye：对图像中AI生成的人物进行细粒度检测

Barot, Jay, Lin, Dan

Abstract

AI generated images are proliferating across the Internet. While some are used for entertainment, others are weaponized for fraud and social engineering attacks on social media users. Existing detectors overfit to generators seen during training, treat detection as opaque binary classification, or rely on costly Large Language Models (LLMs) to explain their outputs. In this paper, we present TruEye, a novel model for fine grained detection and localization of AI manipulated or AI generated humans and scenes. Unlike conventional detectors that assign a single authenticity label, TruEye is the first to distinguish among five compositional categories of synthetic content, including the most challenging case in which a real human is composited into a real scene where they were never physically present. At its core is a mask conditioned dual stream transformer that separates human and scene tokens while preserving patch level spatial correspondence. Specialized reasoning within each stream and region gated cross attention enforce semantic coherence between subject and background, while token level supervision and global compositional classification yield robust, interpretable predictions without invoking an LLM. By restricting intra stream attention to semantically coherent tokens, TruEye also runs over $100\times$ faster than LLM based competitors. Experiments on 6 datasets and our newly curated FineSyn dataset, show that TruEye surpasses state of the art detectors with higher accuracy, faster inference, and stronger generalization to unseen AI generated or manipulated images.

Chinese Translation

AI生成的图像在互联网上迅速传播。虽然有些图像用于娱乐，但另一些则被用于社交媒体用户的欺诈和社会工程攻击。现有的检测器往往过拟合于训练中看到的生成器，将检测视为不透明的二元分类，或依赖于昂贵的大型语言模型（LLMs）来解释其输出。在本文中，我们提出了TruEye，这是一种新颖的模型，用于对AI操控或AI生成的人物和场景进行细粒度检测和定位。与传统检测器仅分配单一真实性标签不同，TruEye首次区分五种合成内容的组成类别，包括最具挑战性的情况，即真实人类被合成到他们从未实际存在的真实场景中。其核心是一个掩码条件的双流变换器，能够在保持补丁级空间对应性的同时分离人类和场景标记。每个流和区域门控交叉注意力中的专业推理强化了主体和背景之间的语义一致性，而标记级监督和全局组合分类则在不调用LLM的情况下产生稳健、可解释的预测。通过将流内注意力限制在语义一致的标记上，TruEye的运行速度比基于LLM的竞争对手快超过100倍。在6个数据集和我们新创建的FineSyn数据集上的实验表明，TruEye在准确性、更快的推理速度和对未见AI生成或操控图像的更强泛化能力方面超越了最先进的检测器。

View on arXiv Download PDF AI Translation

cs.CV / 9 / 2606.27509

Structured-Li-GS: Structured 3D Gaussians Splatting with LiDAR Incorporation and Spatial Constraints

结构化-Li-GS：结合LiDAR和空间约束的结构化3D高斯点云渲染

Weng, Huaiyuan, Li, Huibin, Yeum, Chul Min

Abstract

In this study, we develop a Structured framework for Gaussian Splatting (3DGS) with LiDAR integration (Structured-Li-GS). It is a lightweight Gaussian Splatting pipeline that leverages LiDAR-inertial-visual SLAM. Structured-Li-GS achieves high-quality 3D reconstructions with fewer Gaussians by training on accurate, dense, colorized point clouds. Gaussian primitives are anchored using sub-sampled point clouds, and their ellipsoidal parameters are initialized from local surface geometry. Our training strategy integrates a comprehensive set of loss terms, including photometric, flattening, offset, depth, and normal losses, guided by the dense point cloud, enabling accurate reconstruction without Gaussian densification. This approach produces up-to-scale, high-fidelity results with a moderate model size. For experimental validation, we develop a custom hardware-synchronized LiDAR-camera handheld scanner. Experiments on both benchmark datasets and our real-world in-house dataset demonstrate that Structured-Li-GS surpasses state-of-the-art methods while using fewer Gaussians.

Chinese Translation

在本研究中，我们开发了一种结合LiDAR集成的高斯点云渲染（3DGS）的结构化框架（Structured-Li-GS）。这是一种轻量级的高斯点云渲染管道，利用LiDAR-惯性-视觉SLAM。Structured-Li-GS通过在准确、密集、彩色的点云上进行训练，能够以更少的高斯实现高质量的3D重建。高斯原语通过子采样点云进行锚定，其椭球参数从局部表面几何形状初始化。我们的训练策略整合了一套全面的损失项，包括光度损失、平坦化损失、偏移损失、深度损失和法线损失，这些损失由密集点云引导，使得在不进行高斯密集化的情况下实现准确重建。这种方法在适中的模型大小下产生了按比例缩放的高保真结果。为了进行实验验证，我们开发了一种定制的硬件同步LiDAR-相机手持扫描仪。对基准数据集和我们真实世界内部数据集的实验表明，Structured-Li-GS在使用更少高斯的情况下超越了最先进的方法。

View on arXiv Download PDF AI Translation

cs.CV / 10 / 2606.27514

Tessellating The Earth

地球的镶嵌

Cher, Daniel, Iqbal, Hamza, Xing, Eric, Wei, Brian, Jacobs, Nathan

Abstract

Geolocation encoders, which map geographic coordinates to learned representations, are emerging as an effective means of capturing visual and non-visual characteristics from a latitude-longitude pair alone. However, existing approaches project coordinates onto fixed bases (e.g., spherical harmonics), allocating representational capacity uniformly and devoting equal resources to the open ocean and to a developing city. We introduce Tessellating the Earth (TTE), a location encoder built from learnable Spherical Voronoi partitions that concentrates representational capacity where it is needed in a fully differentiable, end-to-end manner. Each Voronoi site carries its own embedding and migrates during training toward discriminative areas. To bridge the gap between local spatial structure and global semantic understanding, we introduce \emph{global semantic tokens}: a set of shared learnable concept tokens that distill semantic knowledge from the satellite imagery into a compact vocabulary the location encoder can reference at inference, enabling geographically distant sites covering similar environments to share semantics. TTE sets a new state of the art for location encoders across a suite of geospatial classification and regression tasks, and achieves the strongest results when used as a geographic prior for fine-grained species classification on iNaturalist-2018. Code, and weights are available at https://github.com/mvrl/TTE.

Chinese Translation

地理位置编码器将地理坐标映射到学习到的表示中，正在成为一种有效的手段，仅通过经纬度对捕捉视觉和非视觉特征。然而，现有的方法将坐标投影到固定基底（例如，球面谐波），均匀分配表示能力，并对开放海洋和发展中的城市投入相同的资源。我们提出了地球镶嵌（Tessellating the Earth, TTE），这是一个基于可学习的球面Voronoi划分构建的位置编码器，能够在完全可微分的端到端方式中，将表示能力集中在需要的地方。每个Voronoi站点都有自己的嵌入，并在训练过程中向区分性区域迁移。为了弥合局部空间结构与全球语义理解之间的差距，我们引入了 extit{全球语义标记}：一组共享的可学习概念标记，从卫星图像中提炼语义知识，形成位置编码器在推理时可以参考的紧凑词汇，使得覆盖相似环境的地理上远离的站点能够共享语义。TTE在一系列地理空间分类和回归任务中设定了位置编码器的新状态，并在用作iNaturalist-2018上细粒度物种分类的地理先验时取得了最强的结果。代码和权重可在https://github.com/mvrl/TTE获取。

View on arXiv Download PDF AI Translation

cs.CV / 11 / 2606.27527

Large Language Model Teaches Visual Students: Cross-Modality Transfer of Fine-Grained Conceptual Knowledge

大型语言模型教视觉学生：细粒度概念知识的跨模态迁移

Liang, Thomas Shih-Chao, Yu, Zhuoran, Lee, Yong Jae

Abstract

Large Language Models (LLMs) possess broad conceptual knowledge acquired through large-scale text pretraining, yet their potential to supervise models in other modalities remains underexplored. In this work, we propose LaViD--Language-to-Visual Knowledge Distillation--a simple and effective framework for transferring high-level semantic knowledge from a language-only teacher to a vision-only student model. Instead of relying on paired multimodal data, LaViD elicits conceptual signals from an LLM by prompting it to generate multiple-choice questions (MCQs) that probe semantic distinctions between visual classes. Each class is mapped to a soft label distribution over these MCQs, forming a rich conceptual signature that guides the student through an auxiliary distillation loss. Notably, despite using a language-only teacher without access to image data, LaViD consistently outperforms recent methods like MaKD that distill from vision-language models across multiple fine-grained benchmarks. It also achieves competitive or superior performance compared to state-of-the-art visual distillation methods such as DKD and MLKD, with further gains when combined with logit standardization. On the Waterbirds dataset, LaViD substantially improves worst-group accuracy, demonstrating enhanced robustness to spurious correlations with distillation. Code is available at https://github.com/lliangthomas/lavid.

Chinese Translation

大型语言模型（LLMs）通过大规模文本预训练获得了广泛的概念知识，但其在监督其他模态模型方面的潜力仍未得到充分探索。在本研究中，我们提出了LaViD——语言到视觉知识蒸馏（Language-to-Visual Knowledge Distillation）——一个简单而有效的框架，用于将高层语义知识从仅使用语言的教师模型转移到仅使用视觉的学生模型。LaViD并不依赖于配对的多模态数据，而是通过提示LLM生成多项选择题（MCQs），以探测视觉类别之间的语义差异，从而引出概念信号。每个类别被映射到这些MCQs上的软标签分布，形成丰富的概念特征，通过辅助蒸馏损失引导学生模型。值得注意的是，尽管使用的是不接触图像数据的语言教师，LaViD在多个细粒度基准测试中始终优于最近的方法，如MaKD，这些方法是从视觉-语言模型中进行蒸馏的。与最先进的视觉蒸馏方法（如DKD和MLKD）相比，LaViD也实现了具有竞争力或更优的性能，并在与对数标准化结合时进一步提升。在Waterbirds数据集上，LaViD显著提高了最差组的准确性，展示了在蒸馏过程中对虚假相关性的增强鲁棒性。代码可在 https://github.com/lliangthomas/lavid 获取。

View on arXiv Download PDF AI Translation

cs.CV / 12 / 2606.27537

MemoBench: Benchmarking World Modeling in Dynamically Changing Environments

MemoBench：动态变化环境中的世界建模基准测试

Chen, Haoyu, Zhou, Kaichen, Hua, Hang, Zhang, Kaile, Qian, Jingwen, Ma, Wufei, Chen, Haonan, Liu, Chunjiang, Zhao, Yizhou, Wang, Xiaoyuan, Li, Weiyue, Yuille, Alan, Liang, Paul Pu, Du, Yilun

Abstract

Video generation models aspire to simulate dynamic environments, and several benchmarks now evaluate memory consistency across frames. However, most assess consistency only while the target remains in view, and the few that force objects out of view evaluate static scenes where nothing changes during occlusion. To bridge this gap, we introduce MemoBench, a diagnostic benchmark built around the disappear-and-reappear paradigm in dynamically changing environments: a target object undergoes a physical process, disappears from view, and must be correctly recovered in its updated state upon reappearance. We curate 360 ground-truth clips spanning synthetic and real-world scenes, and design an evaluation suite combining automated metrics with VQA-based assessment across four diagnostic pillars. Evaluation of eight state-of-the-art models reveals key insights and open challenges regarding memory consistency under the disappear-and-reappear paradigm.

Chinese Translation

视频生成模型旨在模拟动态环境，目前有多个基准测试评估帧间的记忆一致性。然而，大多数仅在目标物体可见时评估一致性，而少数强制物体移出视野的测试则评估静态场景，在遮挡期间没有任何变化。为填补这一空白，我们引入了MemoBench，这是一个围绕动态变化环境中的消失与重现范式构建的诊断基准：目标物体经历一个物理过程，从视野中消失，并在重新出现时必须正确恢复其更新状态。我们整理了360个真实场景和合成场景的真实数据片段，并设计了一个评估套件，结合自动化指标与基于视觉问答（VQA）的评估，涵盖四个诊断支柱。对八个最先进模型的评估揭示了关于消失与重现范式下记忆一致性的关键见解和开放挑战。

View on arXiv Download PDF AI Translation

cs.CV / 13 / 2606.27547

Beyond MoCap: Scaling Motion Tokenizers with Synthetic Human Motion for Generative Modeling

超越动作捕捉：利用合成人体运动扩展运动标记器以进行生成建模

Yan, Yiwen, He, Wanning, Tai, Yu-Wing

Abstract

Human motion generation models are fundamentally constrained by the limited diversity of motion capture datasets, which predominantly contain common, repetitive actions and fail to cover the long tail of complex human movements, resulting in a restricted motion vocabulary in learned latent representations and poor generalization to rare, compositional, and highly dynamic motions. In this work, we propose a framework for expanding the motion representation space by leveraging large-scale synthetic human motion, introducing a data generation pipeline that produces diverse, physically plausible motion sequences beyond the distribution of existing datasets and integrating it with a redesigned VQ-VAE tokenizer that adapts to this expanded motion space. Unlike conventional tokenizers trained on narrow data distributions, our approach jointly scales both the training distribution and the discrete codebook, enabling the model to capture a significantly richer set of motion primitives. We demonstrate that training with synthetic motion substantially improves the coverage and compositionality of the learned motion vocabulary, leading to consistent gains across motion generation tasks such as text-to-motion and motion continuation, while remaining fully compatible with existing frameworks including MotionGPT. Our results suggest that the primary bottleneck lies in the limited support of the learned motion representation, rather than model architecture alone. Scaling synthetic motion in tandem with representation learning offers a principled path toward more expressive, controllable, and generalizable human motion synthesis.

Chinese Translation

人体运动生成模型在根本上受到动作捕捉数据集有限多样性的限制，这些数据集主要包含常见的、重复的动作，未能覆盖复杂人类运动的长尾，导致学习到的潜在表示中的运动词汇受限，并且对稀有、组合性和高度动态的动作泛化能力较差。在本研究中，我们提出了一种框架，通过利用大规模合成人体运动来扩展运动表示空间，介绍了一种数据生成管道，能够生成超出现有数据集分布的多样化、物理上合理的运动序列，并将其与重新设计的 VQ-VAE 标记器相结合，以适应这一扩展的运动空间。与在狭窄数据分布上训练的传统标记器不同，我们的方法同时扩展训练分布和离散代码本，使模型能够捕捉到显著更丰富的运动原语集合。我们证明，使用合成运动进行训练显著提高了学习到的运动词汇的覆盖范围和组合性，在文本到运动和运动延续等运动生成任务中带来了持续的提升，同时与包括 MotionGPT 在内的现有框架完全兼容。我们的结果表明，主要瓶颈在于学习到的运动表示的有限支持，而不仅仅是模型架构。与表示学习同步扩展合成运动为更具表现力、可控性和可泛化的人体运动合成提供了一条有原则的路径。

View on arXiv Download PDF AI Translation

cs.CV / 14 / 2606.27554

Understanding Cross-Rig Generalization in Automotive Perception: a Multi-Rig Benchmark and Rig Variation Metrics

理解汽车感知中的跨设备泛化：多设备基准测试与设备变异度量

Bader, Tim Alexander, Eberhardt, Tim Dieter, Dillitzer, Maximilian, Stork, Wilhelm

Abstract

Camera-based perception systems for autonomous driving are typically developed and evaluated using fixed sensor rigs, while real-world vehicle fleets exhibit substantial variation in camera placement, orientation, field of view, and camera count. This mismatch introduces a cross-rig domain gap in which only the geometric observation process changes. To study this effect under controlled conditions, we introduce Plentiful CARLA Camera Rigs, a benchmark that renders identical driving scenes under 14 systematically designed camera rigs. This setup enables direct analysis of cross-rig generalization without confounding changes in scene content or appearance. Using the benchmark, we analyze cross-rig transfer behavior of representative multi-view perception architectures and observe substantial performance shifts induced by geometric rig variation. To facilitate structured analysis, we further introduce two calibration-based descriptors derived from rig metadata: Rig Variance, capturing internal rig diversity, and Rig Contrastive Distance, measuring geometric discrepancy between rigs. Our experiments show that geometric rig differences strongly correlate with relative cross-rig performance shifts and that Rig Contrastive Distance provides a reliable proxy for ranking transfer difficulty between sensor rigs.

Chinese Translation

基于摄像头的自动驾驶感知系统通常是在固定传感器设备下开发和评估的，而现实世界中的车辆车队在摄像头的放置、方向、视场和摄像头数量上存在显著差异。这种不匹配引入了一个跨设备领域差距，其中仅几何观察过程发生变化。为了在受控条件下研究这一效应，我们引入了丰富的CARLA摄像头设备（Plentiful CARLA Camera Rigs），这是一个在14个系统设计的摄像头设备下渲染相同驾驶场景的基准测试。该设置使得在不混淆场景内容或外观变化的情况下，能够直接分析跨设备泛化。利用该基准测试，我们分析了代表性的多视角感知架构的跨设备迁移行为，并观察到几何设备变异引起的显著性能变化。为了促进结构化分析，我们进一步引入了两个基于设备元数据的校准描述符：设备方差（Rig Variance），捕捉内部设备多样性，以及设备对比距离（Rig Contrastive Distance），测量设备之间的几何差异。我们的实验表明，几何设备差异与相对跨设备性能变化之间存在强相关性，且设备对比距离为排名传感器设备之间的迁移难度提供了可靠的代理。

View on arXiv Download PDF AI Translation

cs.CV / 15 / 2606.27556

Radar Guided Camera Verification for Automatic Emergency Braking Rethinking Object Detection in Radar Camera Fusion

雷达引导的摄像头验证用于自动紧急制动：重新思考雷达摄像头融合中的目标检测

Akula, Ram Charan, Kandhasamy, Sivanathan, Ganesan, Manikandan

Abstract

Radar camera fusion is widely used in Automatic Emergency Braking AEB systems because radar provides reliable range and velocity measurements while cameras provide a proper visual confirmation of the objects . Most of the deployed systems perform this confirmation using computationally intensive object detectors. However, if the radar has already localized a target, the camera may only need to verify the obstacles presence rather than solving a full problem by identifying the object. Our work proposes a radar scoped edge density gate that performs obstacle verification within radar guided image regions of interest. This method requires no training data, model weights, or GPU acceleration and was integrated into a complete radar camera fusion AEB system with brake by wire actuation. Evaluated on a real instrumented vehicle across 72 driving sessions and 131,603 camera frames, the proposed approach reduced the camera search space by up to 98.7 percentage, achieved a mean processing latency of 0.121 ms per ROI, an AUC of 0.898, and a recall of 0.994. Across 33 staged threat scenarios, the complete AEB system recorded zero missed brake events.

Chinese Translation

雷达摄像头融合在自动紧急制动（AEB）系统中被广泛应用，因为雷达提供可靠的距离和速度测量，而摄像头则提供对物体的适当视觉确认。大多数已部署的系统使用计算密集型目标检测器进行这一确认。然而，如果雷达已经定位了目标，摄像头可能只需要验证障碍物的存在，而不必通过识别物体来解决完整的问题。我们的工作提出了一种雷达范围的边缘密度门，能够在雷达引导的感兴趣图像区域内执行障碍物验证。该方法不需要训练数据、模型权重或GPU加速，并已集成到一个完整的雷达摄像头融合AEB系统中，采用线控制动。经过在一辆真实仪器化车辆上进行72次驾驶会话和131,603帧摄像头图像的评估，所提出的方法将摄像头搜索空间减少了多达98.7%，每个感兴趣区域的平均处理延迟为0.121毫秒，AUC值为0.898，召回率为0.994。在33个 staged 威胁场景中，完整的AEB系统记录了零次漏制动事件。

View on arXiv Download PDF AI Translation

cs.CV / 16 / 2606.27575

Perceptual 3D Simulation With Physical World Modeling

具有物理世界建模的感知3D模拟

Lee, Wanhee, Kotar, Klemen, Venkatesh, Rahul Mysore, Watrous, Jared, Yamins, Daniel L. K.

Abstract

Predicting how a scene will evolve after a desired 3D transformation from images is a central goal in vision, graphics, and robotics. Yet unlike ideal simulators with full access to 3D geometry and dynamics, real world systems must rely on perceptual inputs and local actions that are inherently partial and incomplete. In this work, we present P3Sim, a physical world modeling system that simulates future scene states under both partial observations and incomplete 3D transformation signals. P3Sim is composed of three interacting components: a learned physical world model, a geometric conditioning module, and a persistent scene memory. The world model interprets perception as probabilistic inference over multimodal scene variables, providing predictions of the distributions of any scene variable conditioned on any combination of others. The geometric conditioning module provides a partial 3D transform signal for conditioning the world model at inference time. The persistent scene memory integrates predictions over time, enabling online updates and consistency under uncertainty. By combining learned inference with explicit geometric structure, P3Sim balances data-driven flexibility with built-in inductive bias. This design yields a flexible perceptual simulator that generalizes across diverse 3D transformation tasks, such as novel view synthesis, object manipulation, and dynamic scene prediction, advancing toward general purpose 3D scene understanding and transformation.

Chinese Translation

预测场景在经过期望的3D变换后如何演变是视觉、图形学和机器人领域的一个核心目标。然而，与理想的模拟器（能够完全访问3D几何和动态）不同，现实世界系统必须依赖于感知输入和局部动作，这些输入和动作本质上是部分和不完整的。在本研究中，我们提出了P3Sim，一个物理世界建模系统，能够在部分观察和不完整的3D变换信号下模拟未来的场景状态。P3Sim由三个相互作用的组件组成：一个学习的物理世界模型、一个几何条件模块和一个持久的场景记忆。世界模型将感知解释为对多模态场景变量的概率推断，提供任何场景变量在其他变量任意组合条件下的分布预测。几何条件模块在推断时提供部分3D变换信号，以对世界模型进行条件化。持久的场景记忆在时间上整合预测，支持在线更新和在不确定性下的一致性。通过将学习的推断与显式的几何结构相结合，P3Sim在数据驱动的灵活性与内置的归纳偏差之间取得了平衡。这种设计产生了一个灵活的感知模拟器，能够在多种3D变换任务中进行泛化，如新视图合成、物体操控和动态场景预测，推动了通用3D场景理解和变换的进展。

View on arXiv Download PDF AI Translation

cs.CV / 17 / 2606.27576

DeLux: Cross-Modal Local Artifact Restoration in Video Using Neuromorphic Data

DeLux：基于神经形态数据的视频跨模态局部伪影恢复

Stachowiak, Bartosz, Brzezinski, Dariusz

Abstract

Conventional RGB cameras suffer from lighting artifacts such as flare, glare, flicker, and overexposure, leading to irrecoverable information loss that necessitates computational restoration. However, existing approaches treat these problems in isolation, failing to recover structural details completely obscured by complex spatially discrete image degradations. In this paper, we propose a novel cross-modal restoration paradigm and present DeLux, a modular proof-of-concept pipeline that leverages neuromorphic event streams as a structural prior to guide the targeted detection and inpainting of lighting artifacts in RGB video. Validation on synthetic benchmarks and real-world automotive footage demonstrates that DeLux effectively suppresses local artifacts and restores affected regions. The proposed approach outperforms existing RGB-only baselines and event-guided HDR models, achieving an average MS-SSIM of over 0.99 across all artifact types and demonstrating up to an 88% reduction in artifact severity in real-world automotive footage. The synthetic artifact generation tools and curated real-world evaluation datasets are made publicly available to foster future research on cross-modal restoration.

Chinese Translation

传统的RGB相机受到光照伪影的影响，如眩光、反射、闪烁和过度曝光，导致不可恢复的信息损失，迫使需要进行计算恢复。然而，现有的方法往往孤立地处理这些问题，未能完全恢复被复杂空间离散图像退化完全遮蔽的结构细节。本文提出了一种新颖的跨模态恢复范式，并展示了DeLux，一个模块化的概念验证管道，利用神经形态事件流作为结构先验，以指导RGB视频中光照伪影的针对性检测和修复。在合成基准和真实世界汽车视频上的验证表明，DeLux有效抑制局部伪影并恢复受影响区域。所提出的方法在所有伪影类型上均超越了现有的仅RGB基线和事件引导的HDR模型，平均MS-SSIM超过0.99，并在真实世界汽车视频中实现了高达88%的伪影严重性降低。合成伪影生成工具和精心策划的真实世界评估数据集已公开，以促进未来的跨模态恢复研究。

View on arXiv Download PDF AI Translation

cs.CV / 18 / 2606.27579

Distribution-based deep multiple instance learning for tumor proportion scoring in NSCLC

基于分布的深度多实例学习在非小细胞肺癌肿瘤比例评分中的应用

Pysz, Krzysztof, Bartczak, Artur, Kwiecień, Jarosław, Krajewski, Piotr, Dyrka, Witold

Abstract

Accurate assessment of tumor proportion score (TPS) in non-small cell lung cancer (NSCLC) is critical for treatment planning and prognosis. Key challenges include the tedious manual work required to annotate each slide, combined with the limited number of experts certified for this task. Multiple instance learning (MIL) has proven to be an effective approach for predicting TPS scores at the slide level; however, existing methods struggle with non-expressive (zero class) images. Our approach involves two models: (1) an embedding-extraction and multiclass-classification network that captures the histopathological features of individual patches, and (2) a MIL model that aggregates these embeddings to predict zero-inflated beta (ZIBeta) parameters representing the overall TPS probability distribution for the entire slide. Using only slide-level TPS scores as labels, we demonstrate how this end-to-end framework can leverage a novel distribution-based architecture to improve prediction accuracy and explainability. ZIBeta modeling significantly outperforms baseline linear and ridge regression while capturing expected accuracy through distribution concentration.

Chinese Translation

准确评估非小细胞肺癌（NSCLC）中的肿瘤比例评分（TPS）对治疗规划和预后至关重要。主要挑战包括需要对每个切片进行繁琐的手动标注，以及获得此任务认证的专家数量有限。多实例学习（MIL）已被证明是一种有效的方法，用于在切片级别预测TPS评分；然而，现有方法在处理非表现性（零类）图像时面临困难。我们的方法涉及两个模型：（1）一个嵌入提取和多类分类网络，用于捕捉单个补丁的组织病理特征；（2）一个MIL模型，将这些嵌入聚合以预测代表整个切片的总体TPS概率分布的零膨胀贝塔（ZIBeta）参数。仅使用切片级别的TPS评分作为标签，我们展示了该端到端框架如何利用一种新颖的基于分布的架构来提高预测准确性和可解释性。ZIBeta建模显著优于基线线性回归和岭回归，同时通过分布集中捕获预期的准确性。

View on arXiv Download PDF AI Translation

cs.CV / 19 / 2606.27582

Beyond Points: Spherical Distributional Part Prototypes for Interpretable Classification

超越点：用于可解释分类的球面分布式部分原型

Leão, Duarte, Araújo, Diogo Pereira, Barata, Catarina, Santiago, Carlos

Abstract

Prototype-based neural networks aim to provide intrinsic interpretability by grounding predictions in a small set of part prototypes. However, modern vision backbones typically operate in normalized, directional embedding spaces where each semantic part exhibits substantial intra-class variability. As a result, point prototypes often become redundant or unstable, hurting both explanation quality and robustness. We propose vMFProto, a distributional part-prototype framework that models each class as a mixture of von Mises-Fisher components on the hypersphere. Each prototype learns its own concentration, capturing part-specific variability, and we use entropic optimal transport (OT) to obtain structured patch-to-prototype assignments. A two-stage training schedule performs OT-driven prototype discovery followed by end-to-end refinement with patch-level distillation and distribution-aware diversity regularization. Experiments on CUB-200-2011, Stanford Dogs, and Stanford Cars with frozen DINO backbones show that vMFProto achieves state-of-the-art explanation quality (consistency, stability, and distinctiveness) with competitive accuracy. Qualitative results confirm that vMFProto yields localized, non-redundant part evidence.

Chinese Translation

基于原型的神经网络旨在通过将预测与一小组部分原型相结合来提供内在的可解释性。然而，现代视觉主干网络通常在归一化的方向嵌入空间中操作，其中每个语义部分表现出显著的类内变异性。因此，点原型往往变得冗余或不稳定，从而影响了解释质量和鲁棒性。我们提出了vMFProto，一种分布式部分原型框架，将每个类别建模为超球面上的von Mises-Fisher成分的混合。每个原型学习其自身的集中度，以捕捉部分特定的变异性，我们使用熵最优传输（OT）来获得结构化的补丁到原型的分配。一个两阶段的训练计划首先进行OT驱动的原型发现，然后通过补丁级蒸馏和分布感知的多样性正则化进行端到端的精炼。在CUB-200-2011、斯坦福狗和斯坦福汽车的数据集上，使用冻结的DINO主干网络的实验表明，vMFProto在解释质量（连贯性、稳定性和独特性）方面达到了最先进的水平，并且具有竞争力的准确性。定性结果确认vMFProto产生了局部的、非冗余的部分证据。

View on arXiv Download PDF AI Translation

cs.CV / 20 / 2606.27584

CoIn: Comprehensive 2D-3D Inpainting with Gaussian Splatting Guidance

CoIn：基于高斯溅射引导的综合性2D-3D修复

Kim, Hana, Kim, Minje, Kim, Tae-Kyun

Abstract

3D scene inpainting is essential for reconstructing areas corrupted by occlusions or limited viewpoints. While recent methods leverage Gaussian Splatting (GS) for efficient 3D editing, they often depend on precise multi-view segmentation masks and are inherently constrained to object removal tasks. We propose CoIn, a novel framework that bridges 2D inpainting models and 3DGS through a multi-stage consistency pipeline. Our approach first generates initial inpainted images using a diffusion model, enabling the use of arbitrary-shaped masks and diverse tasks like object insertion. We then introduce Reference Adaptive GS with Feature Attention to reconstruct a coarse 3D scene by adaptively weighing towards a reference view (2D -> 3D). This 3D representation provides geometric guidance to the diffusion process via GS-based Reference Feature Warping, ensuring multi-view consistency (3D -> 2D). Finally, a Texture-Enhancing Discriminator refines the 3D scene to achieve high photometric realism (2D -> 3D). Experiments show that CoIn, effectively leveraging bidirectional information flow, achieves state-of-the-art performance and effectively handles both object removal and object insertion with flexible mask input.

Chinese Translation

3D场景修复对于重建因遮挡或视角限制而受损的区域至关重要。尽管近期的方法利用高斯溅射（Gaussian Splatting, GS）进行高效的3D编辑，但它们通常依赖于精确的多视图分割掩膜，并且本质上仅限于物体移除任务。我们提出了CoIn，一个新颖的框架，通过多阶段一致性管道将2D修复模型与3D GS连接起来。我们的方法首先使用扩散模型生成初始修复图像，使得可以使用任意形状的掩膜以及执行诸如物体插入等多样化任务。接着，我们引入了具有特征注意力的参考自适应GS，通过自适应地加权参考视图（2D -> 3D）来重建粗略的3D场景。该3D表示通过基于GS的参考特征变形为扩散过程提供几何指导，确保多视图一致性（3D -> 2D）。最后，纹理增强判别器对3D场景进行细化，以实现高光度真实感（2D -> 3D）。实验表明，CoIn有效利用双向信息流，达到了最先进的性能，并有效处理物体移除和物体插入任务，同时支持灵活的掩膜输入。

View on arXiv Download PDF AI Translation

cs.CV / 21 / 2606.27596

Dismantling Pathological Shortcuts: A Causal Framework for Faithful LVLM Decoding

拆解病态捷径：忠实LVLM解码的因果框架

Yu, Liu, Chen, Can, Kuang, Ping, Feng, Zhikun, Zhou, Fan, Dobbie, Gillian

Abstract

Large Vision-Language Models (LVLMs) exhibit sophisticated reasoning but remain susceptible to object hallucination. Deviating from the prevailing attention intensity assumption, we reveal a deeper dynamic structural misalignment: hallucination is triggered at decision-critical steps where specific attention heads, acting as risky mediators, decouple from visual evidence to lock onto language priors. This establishes a pathological shortcut that bypasses visual grounding. To dismantle this, we propose Fox (Faithfulness and Observational-flow via eXpression-rectification), a training-free inference-time framework. Fox diagnoses structural misalignment using a visual attention entropy probe to localize risky mediators unsupervisedly. We then execute a targeted causal intervention via numerical logit saturation to physically sever the shortcut path. Finally, a conflict-gated cooperative decoding strategy reconciles interventional faithfulness with observational fluency. Extensive experiments demonstrate that Fox achieves SOTA performance, outperforming SID by 29.1% while preserving linguistic richness. Code is available at https://github.com/Cc2021start/Fox.

Chinese Translation

大型视觉语言模型（LVLMs）展现出复杂的推理能力，但仍然容易受到对象幻觉的影响。我们偏离了当前主流的注意力强度假设，揭示了一种更深层次的动态结构不一致性：幻觉在决策关键步骤被触发，此时特定的注意力头作为风险中介，从视觉证据中解耦，转而锁定语言先验。这建立了一条绕过视觉基础的病态捷径。为了解构这一现象，我们提出了Fox（通过表达修正实现的忠实性和观察流），这是一个无训练的推理时框架。Fox利用视觉注意力熵探测器诊断结构不一致性，以无监督的方式定位风险中介。然后，我们通过数值对数饱和执行有针对性的因果干预，以物理上切断捷径路径。最后，冲突门控合作解码策略调和了干预的忠实性与观察的流畅性。大量实验表明，Fox在性能上达到了最先进水平（SOTA），比SID提高了29.1%，同时保持了语言的丰富性。代码可在 https://github.com/Cc2021start/Fox 获取。

View on arXiv Download PDF AI Translation

cs.CV / 22 / 2606.27608

Qwen-Image-2.0-RL Technical Report

Qwen-Image-2.0-RL 技术报告

Xu, Yixian, Gao, Kaiyuan, Chen, Yuxiang, Chen, Yilei, Tang, Zecheng, Liu, Zihao, Zhou, Zikai, Li, Deqing, Meng, Hao, Cao, Kuan, Li, Jiahao, Zhang, Jie, Peng, Liang, Jiang, Lihan, Tang, Ningyuan, Yin, Shengming, Wu, Tianhe, Chen, Xiaoyue, Shu, Yan, Zhang, Yanran, Wang, Yi, Wu, Yu, Wu, Yujia, Zhang, Zekai, Wang, Zhendong, Xu, Xiao, Yan, Kun, Wu, Chenfei

Abstract

We present Qwen-Image-2.0-RL, a post-training pipeline that applies reinforcement learning from human feedback (RLHF) and on-policy distillation (OPD) to improve both the visual quality and instruction-following capability of the Qwen-Image-2.0 diffusion model. To provide reliable reward signals, we construct task-specific composite reward models by fine-tuning vision-language models with a pointwise scoring paradigm and chain-of-thought reasoning. For text-to-image generation, the reward models cover alignment, aesthetics, and portrait fidelity dimensions. For image editing tasks, the reward system addresses instruction-following accuracy and face identity preservation. Building on this reward system, we develop a scalable GRPO-based RL training framework, incorporating a hybrid classifier-free guidance (CFG) strategy to preserve pre-trained knowledge, prompt curation via intra-group reward range filtering, and per-category reward weight calibration. To merge the task-specialized RL policies for T2I and editing, we propose on-policy distillation as the final training stage, which consolidates multiple teachers into a single student model through trajectory-level velocity matching. Extensive evaluation shows that Qwen-Image-2.0-RL achieves 57.84 overall score on Qwen-Image-Bench (+2.61 over the base model), Elo ratings of 1193 in text-to-image arena (+78) and 1349 in image edit arena (+93), demonstrating consistent gains in aesthetic quality, prompt adherence, and editing accuracy.

Chinese Translation

我们提出了 Qwen-Image-2.0-RL，这是一个后训练管道，应用了基于人类反馈的强化学习（RLHF）和在线策略蒸馏（OPD），以提高 Qwen-Image-2.0 扩散模型的视觉质量和指令遵循能力。为了提供可靠的奖励信号，我们通过使用逐点评分范式和思维链推理对视觉-语言模型进行微调，构建了特定任务的复合奖励模型。在文本到图像生成方面，奖励模型涵盖了对齐、美学和肖像保真度等维度。在图像编辑任务中，奖励系统关注指令遵循的准确性和面部身份的保留。在此奖励系统的基础上，我们开发了一个可扩展的基于 GRPO 的强化学习训练框架，结合了混合无分类器引导（CFG）策略，以保留预训练知识，通过组内奖励范围过滤进行提示策划，以及按类别的奖励权重校准。为了合并 T2I 和编辑任务的特定强化学习策略，我们提出了在线策略蒸馏作为最终训练阶段，通过轨迹级别的速度匹配将多个教师模型整合为一个学生模型。广泛的评估表明，Qwen-Image-2.0-RL 在 Qwen-Image-Bench 上获得了 57.84 的整体得分（比基础模型提高了 2.61），在文本到图像领域的 Elo 评分为 1193（提高了 78），在图像编辑领域的 Elo 评分为 1349（提高了 93），显示出在美学质量、提示遵循和编辑准确性方面的一致提升。

View on arXiv Download PDF AI Translation

cs.CV / 23 / 2606.27635

Denoising ICF Images with Multiplicative Uniform Noise: A Self-Supervised Study Based on the Log-Domain Noisier2Inverse Framework

使用乘法均匀噪声去噪ICF图像：基于对数域Noisier2Inverse框架的自监督研究

Hwang, Gyeongha, Wolfe, Bradley Thomas, Naheed, Naima

Abstract

This paper documents the implementation and evaluation of a self-supervised denoising framework on Inertial Confinement Fusion (ICF) images corrupted by Multiplicative Uniform noise: the \emph{Log-Domain Noisier2Inverse} framework. This framework is developed and analysed in this work; the key theoretical result -- that minimising the log-domain self-supervised loss is equivalent to supervised learning in the transformed domain -- is presented with full proof. We document significant implementation challenges arising from the unique characteristics of ICF imagery, describe the fixes applied at each stage, and report final quantitative results. The log-domain approach with per-image JSON Uniform noise loading (Variant~B) achieves the best result: a mean PSNR of $21.41\db$ and SSIM of $0.8358$, a $+19.46\db$ improvement over the noisy input baseline of $1.95\db$, substantially outperforming BM3D log-domain ($4.47\db$, SSIM $0.5181$) and Noise2Self ($4.75\db$, SSIM $0.0177$). Variant~A, using fixed Gaussian noise loading, achieves $21.39\db$ PSNR and SSIM $0.8436$. Of the three evaluated methods, Log-Domain Noisier2Inverse and Noise2Self are entirely self-supervised during training, requiring no clean ground truth data; BM3D is a classical filter-based method requiring no training at all. The clean reference images are used solely for quantitative evaluation of all three methods.

Chinese Translation

本文记录了在受到乘法均匀噪声污染的惯性约束聚变（ICF）图像上实施和评估自监督去噪框架的过程： extit{Log-Domain Noisier2Inverse}框架。该框架在本研究中得到了开发和分析；关键理论结果——最小化对数域自监督损失等价于在变换域中的监督学习——提供了完整的证明。我们记录了由于ICF图像的独特特性而产生的显著实施挑战，描述了在每个阶段应用的修复措施，并报告了最终的定量结果。采用每图像JSON均匀噪声加载的对数域方法（变体B）取得了最佳结果：平均PSNR为$21.41 ext{dB}$，SSIM为$0.8358$，相比于噪声输入基线$1.95 ext{dB}$提高了$+19.46 ext{dB}$，显著优于BM3D对数域（$4.47 ext{dB}$，SSIM $0.5181$）和Noise2Self（$4.75 ext{dB}$，SSIM $0.0177$）。使用固定高斯噪声加载的变体A达到了$21.39 ext{dB}$的PSNR和SSIM $0.8436$。在评估的三种方法中，Log-Domain Noisier2Inverse和Noise2Self在训练过程中完全自监督，不需要干净的真实数据；BM3D是一种经典的基于滤波的方法，根本不需要训练。干净的参考图像仅用于对三种方法的定量评估。

View on arXiv Download PDF AI Translation

cs.CV / 24 / 2606.27637

AI-Generated Image Recognition via Fusion of CNNs and Vision Transformers

通过卷积神经网络与视觉变换器融合的AI生成图像识别

Mai, Xuan-Bach, Nguyen-Huu, Hoang-Minh, Nguyen, Quoc-Nghia, Vu, Hoang-Tung, Tran, Minh-Triet, Le, Trung-Nghia

Abstract

Recent advancements in synthetic data technology have opened a new era where images of remarkable quality are generated, blurring the lines between real-life images and those produced by Artificial Intelligence (AI). This evolution poses a significant challenge to ensuring the reliability and authenticity of data, underscoring the need for robust detection methods. In this paper, we present a robust approach aimed at addressing these pressing concerns. Our methodology revolves around leveraging fusion strategies, combining the strengths of multiple detection methods for identifying AI-generated images. Through extensive experimentation on the CIFAKE dataset, our model showcases remarkable performance, achieving an impressive accuracy rate of 97.32%. This accomplishment underscores the efficacy of our approach in accurately distinguishing between AI-generated images and real-life images, thus contributing to the advancement of data authentication techniques amidst the proliferation of synthetic data.

Chinese Translation

近期合成数据技术的进步开启了一个新的时代，生成的图像质量显著，模糊了现实图像与人工智能（AI）生成图像之间的界限。这一演变对确保数据的可靠性和真实性提出了重大挑战，强调了强大检测方法的必要性。本文提出了一种旨在解决这些紧迫问题的稳健方法。我们的方法围绕融合策略展开，结合多种检测方法的优势，以识别AI生成的图像。通过对CIFAKE数据集的广泛实验，我们的模型展示了卓越的性能，达到了令人印象深刻的97.32%的准确率。这一成就突显了我们的方法在准确区分AI生成图像与现实图像方面的有效性，从而为在合成数据泛滥的背景下推进数据认证技术做出了贡献。

View on arXiv Download PDF AI Translation

cs.CV / 25 / 2606.27644

CascadeOcc: Rethinking 3D Occupancy World Models with Cascaded VQ Representations

CascadeOcc：重新思考具有级联向量量化表示的 3D 占用世界模型

Hwang, Kyumin, Choi, Wonhyeok, Kim, Jaeyeul, Park, Jihun, Park, Daehee, Im, Sunghoon

Abstract

This letter proposes CascadeOcc, a novel occupancy world model that prioritizes intrinsic structural hierarchy over extrinsic auxiliary modalities for autonomous driving. Occupancy world models -- forecasting the future driving environment and planning the driving trajectory -- effectively bridge perception and planning, but current approaches often heavily rely on external modalities or large language models, failing to fully exploit the inherent structural potential of occupancy representations themselves. To enhance representational capacity for complex 3D scenes, we integrate a cascaded Vector Quantized (VQ) mechanism into an autoregressive framework. Following a coarse-to-fine principle, CascadeOcc progressively refines fine-grained details from global structures through a multi-scale architecture. Additionally, we incorporate a TimeMixer to capture multi-scale temporal dependencies, establishing a dual-hierarchy mechanism in both space and time. Experimental results on 4D occupancy forecasting and motion planning benchmarks demonstrate that CascadeOcc achieves superior performance among vision-centric approaches, validating that optimizing inherent representations is a powerful alternative to relying on external foundation models.

Chinese Translation

本文提出了 CascadeOcc，一种新颖的占用世界模型，优先考虑内在结构层次而非外在辅助模态，以用于自动驾驶。占用世界模型——预测未来驾驶环境并规划驾驶轨迹——有效地桥接了感知与规划，但当前的方法往往过于依赖外部模态或大型语言模型，未能充分利用占用表示自身的内在结构潜力。为了增强对复杂 3D 场景的表征能力，我们将级联向量量化（Vector Quantized, VQ）机制集成到自回归框架中。遵循粗到细的原则，CascadeOcc 通过多尺度架构逐步从全局结构中提炼细粒度细节。此外，我们还引入了 TimeMixer 来捕捉多尺度时间依赖性，在空间和时间上建立了双层次机制。在 4D 占用预测和运动规划基准测试中的实验结果表明，CascadeOcc 在以视觉为中心的方法中实现了优越的性能，验证了优化内在表示是依赖外部基础模型的强有力替代方案。

View on arXiv Download PDF AI Translation

cs.CV / 26 / 2606.27646

VLM-Aware Meta-Optic Front-End Design for Frozen Vision-Language Models

面向冻结视觉-语言模型的VLM感知元光学前端设计

Kang, Chanik, Pestourie, Raphaël, Chung, Haejun

Abstract

Conventional machine-vision pipelines typically rely on high-quality optics that produce clean, human-interpretable images, and optical design has therefore been driven by image-level criteria such as resolution, aberration correction, and pixel fidelity. However, such optics are often impractical for size-, cost-, or form-factor-constrained applications, where compact meta-optics offer an attractive alternative but operate under strict physical efficiency limits. We propose CODA, a co-design framework that optimizes a continuous-density meta-optic front-end for frozen-model recognition using differentiable image formation and adjoint-gradient updates of Maxwell-based simulations. CODA directly optimizes the cross-entropy loss of a fixed zero-shot CLIP classifier without learned reconstruction, image signal processing, or image-fidelity auxiliary objectives. In a two-dimensional simulated imaging benchmark on ImageNet-100, CODA improves CLIP ViT-L/14 zero-shot accuracy from 53.75 $\pm$ 3.57$\%$ with a focal-concentration baseline to 65.41 $\pm$ 3.99$\%$. The optimized optics further transfer without re-optimization across CLIP, SigLIP, and DINOv2 on ImageNet-100, CIFAR-100, and Food-101. These results demonstrate that, under constrained meta-optic imaging, downstream recognition can be improved by aligning optical design with frozen vision-model objectives rather than conventional image-formation criteria.

Chinese Translation

传统的机器视觉管道通常依赖于高质量的光学系统，以生成清晰且易于人类理解的图像，因此光学设计主要受到图像级标准的驱动，如分辨率、像差校正和像素保真度。然而，这种光学系统在尺寸、成本或形状因素受限的应用中往往不切实际，而紧凑的元光学则提供了一个有吸引力的替代方案，但在严格的物理效率限制下运行。我们提出了CODA，一个共同设计框架，优化用于冻结模型识别的连续密度元光学前端，采用可微分的图像形成和基于麦克斯韦模拟的伴随梯度更新。CODA直接优化固定零样本CLIP分类器的交叉熵损失，而无需学习重建、图像信号处理或图像保真度辅助目标。在ImageNet-100的二维模拟成像基准测试中，CODA将CLIP ViT-L/14的零样本准确率从53.75 $ ext{±}$ 3.57$ ext{ extperthousand}$的焦点集中基线提高到65.41 $ ext{±}$ 3.99$ ext{ extperthousand}$。优化后的光学系统在ImageNet-100、CIFAR-100和Food-101上，进一步在CLIP、SigLIP和DINOv2之间转移，无需重新优化。这些结果表明，在受限的元光学成像条件下，下游识别可以通过将光学设计与冻结视觉模型目标对齐，而不是依赖传统的图像形成标准来改善。

View on arXiv Download PDF AI Translation

cs.CV / 27 / 2606.27655

Temporal-Emerged Prompting for Segment Anything in Multiframe Infrared Small Target Detection

多帧红外小目标检测中的时序显现提示方法

Xing, Yinghui, Chu, Donghao, Zhang, Shizhou, Xu, Di

Abstract

Accurately localizing and segmenting small targets in low signal-to-noise ratio (SNR) infrared sequences remains a challenging task. Since targets are often indistinguishable from the background in individual frames, existing methods, even when equipped with advanced foundation model and powerful inter-frame association mechanisms, still fail to detect them. Motivated by the observation that targets tend to emerge gradually from the background over time and become distinguishable, we propose Temporal-Emerged Prompting for Segment Anything Model (TEP-SAM), a principled framework designed to explicitly exploit such temporal-emerged cues to modulate and prompt SAM. TEP-SAM operates by jointly modeling global motion patterns and local motion deviations to locate potential targets. It further enhances target region features by leveraging motion discrepancy, thereby generating temporal-emerged cues for SAM and enabling non-interactive segmentation. By bridging large-scale semantic pretraining with task-specific temporal modeling, TEP-SAM effectively adapts SAM to the challenging multiframe infrared small target detection task. Extensive experiments demonstrate the effectiveness of our approach, particularly under severely low-SNR conditions and in complex dynamic background.

Chinese Translation

在低信噪比（SNR）红外序列中准确定位和分割小目标仍然是一项具有挑战性的任务。由于目标在单帧中常常与背景难以区分，即使现有方法配备了先进的基础模型和强大的帧间关联机制，仍然无法有效检测这些目标。基于目标往往随着时间的推移逐渐从背景中显现并变得可区分的观察，我们提出了时序显现提示方法（Temporal-Emerged Prompting for Segment Anything Model, TEP-SAM），这是一个旨在明确利用这种时序显现线索来调节和提示SAM的原则性框架。TEP-SAM通过联合建模全局运动模式和局部运动偏差来定位潜在目标，并通过利用运动差异进一步增强目标区域特征，从而为SAM生成时序显现线索，实现非交互式分割。通过将大规模语义预训练与任务特定的时序建模相结合，TEP-SAM有效地将SAM适应于具有挑战性的多帧红外小目标检测任务。大量实验表明，我们的方法在极低SNR条件下以及复杂动态背景中表现出色。

View on arXiv Download PDF AI Translation

cs.CV / 28 / 2606.27659

GeoFace: Consistent Multi-View Face Generation with Geometry-Constrained Diffusion

GeoFace：基于几何约束的多视角一致性人脸生成

Choi, Yeji, Choi, Jinhyeok, Min, Jaewon, Kwon, Minkyung, Kim, Jin Hyeon, Kim, Seungryong

Abstract

We present GeoFace, a geometry-constrained multi-view diffusion framework for consistent face generation from a single input. % While recent multi-view diffusion models achieve photorealistic synthesis at the per-view level, they lack an explicit mechanism to enforce a shared 3D structure across views, often leading to inconsistent geometry across viewpoints. To address this, GeoFace proposes a unified dual-stream framework for joint generation of multi-view RGB images and 3D face geometry, where the appearance and geometry streams interact through shared attention layers. To encourage the two streams to mutually constrain each other, we introduce a geometry-guided attention alignment loss that supervises the cross-attention between appearance and geometry tokens with 3D-consistent correspondences, enabling the appearance stream to correctly reference pose-invariant geometric cues for robust alignment across viewpoints. Geometry is represented as a canonical UV position map, derived from a FLAME mesh fitted to multi-view observations, serving as a view-invariant shared constraint across all generated views. Experiments on RenderMe-360 and NeRSemble demonstrate that GeoFace consistently outperforms existing methods in both visual quality and cross-view geometric consistency, facilitating more efficient 3D reconstruction.

Chinese Translation

我们提出了GeoFace，一个基于几何约束的多视角扩散框架，用于从单一输入中生成一致的人脸。尽管近期的多视角扩散模型在每个视角层面上实现了逼真的合成，但它们缺乏强制执行视角间共享3D结构的明确机制，常常导致视角间几何不一致。为了解决这个问题，GeoFace提出了一个统一的双流框架，用于多视角RGB图像和3D人脸几何的联合生成，其中外观流和几何流通过共享注意力层进行交互。为了鼓励这两个流相互约束，我们引入了一种几何引导的注意力对齐损失，该损失监督外观和几何标记之间的交叉注意力，确保3D一致的对应关系，使得外观流能够正确引用与姿态无关的几何线索，以实现视角间的稳健对齐。几何以规范的UV位置图表示，该图源自适配于多视角观察的FLAME网格，作为所有生成视角间的视角不变共享约束。对RenderMe-360和NeRSemble的实验表明，GeoFace在视觉质量和视角间几何一致性方面始终优于现有方法，从而促进了更高效的3D重建。

View on arXiv Download PDF AI Translation

cs.CV / 29 / 2606.27660

MVPruner: Dynamic Token Pruning for Accelerating Multi-view Vision-Language Models in Autonomous Driving

MVPruner：用于加速自主驾驶中的多视角视觉-语言模型的动态标记剪枝

Yang, Nan, Liu, Zhanwen, Zhang, Linfeng, Xie, Shangyu, Wang, Yang, Zhou, Wenzhuo, Zhao, Xiangmo

Abstract

Vision-Language Models (VLMs) improve generalization and interpretability in autonomous driving but suffer from efficiency issues due to long visual token sequences, particularly in standard multi-view settings. Existing token pruning methods employ fixed pruning rate allocation and static importance metrics, ignoring dynamic inter-view importance differences and the evolving information importance during inference. Our analysis reveals that multi-view VLMs inherently encode task-related view priors in deeper layers and exhibit dynamic information requirements. Motivated by these findings, we propose MVPruner, a two-stage adaptive token pruning method that aligns pruning behavior with the model's dynamic information requirements. The first stage allocates pruning budgets based on the information diversity of each view, and retains tokens with consistent contribution across stages, ensuring semantic representational capacity. The second stage allocates budgets and selects tokens guided by instruction text to guarantee task alignment. Experimental results on four benchmarks demonstrate the superior performance of our method. For example, DriveMM equipped with MVPruner achieves 87.3% reduction in FLOPs, 4.97* speedup in prefilling phase while retaining 98.5% accuracy on DriveLM benchmark.

Chinese Translation

视觉-语言模型（VLMs）在自主驾驶中提高了泛化能力和可解释性，但由于长视觉标记序列，特别是在标准多视角设置中，效率问题严重。现有的标记剪枝方法采用固定的剪枝率分配和静态重要性度量，忽略了视角间动态重要性差异以及推理过程中信息重要性的演变。我们的分析表明，多视角VLMs在更深层次上固有地编码了与任务相关的视角先验，并表现出动态信息需求。基于这些发现，我们提出了MVPruner，一种两阶段自适应标记剪枝方法，使剪枝行为与模型的动态信息需求相一致。第一阶段根据每个视角的信息多样性分配剪枝预算，并保留在各阶段中贡献一致的标记，以确保语义表示能力。第二阶段根据指令文本分配预算并选择标记，以保证任务对齐。在四个基准测试上的实验结果表明我们方法的优越性能。例如，配备MVPruner的DriveMM在DriveLM基准测试中实现了87.3%的FLOPs减少，预填充阶段速度提升4.97倍，同时保持98.5%的准确率。

View on arXiv Download PDF AI Translation

cs.CV / 30 / 2606.27667

Explainable AI for Biodiversity Monitoring and Ecological Image Analysis

可解释人工智能在生物多样性监测和生态图像分析中的应用

Bent, Brinnae, Houliston, Holly R., Zhou, Jiayi, Aghakishiyeva, Günel, Johnston, David W.

Abstract

Artificial intelligence is transforming biodiversity monitoring by enabling automated analysis of ecological imagery collected from camera traps, drones, satellites, underwater platforms, and other sensing systems. These tools can expand the scale and speed of conservation assessments, yet many computer vision models remain difficult to inspect, making it challenging to determine whether predictions are based on ecologically meaningful signals or on spurious correlations, sampling biases, and other artifacts that may undermine conservation decisions. We argue that explainable artificial intelligence (XAI) should become a standard component of ecological model validation because conservation practitioners increasingly depend on understanding not only whether a model is accurate, but why it is accurate. We provide practical guidance for applying XAI to three common ecological computer vision tasks: image classification, object detection, and image segmentation. To illustrate how XAI can support ecological model auditing, refinement, and deployment, we present two case studies using aerial imagery: harbor seal detection and cetacean anatomical segmentation. These examples demonstrate how explanation methods can identify biologically meaningful cues, reveal false positives driven by background and shape confounds, uncover edge and occlusion effects, and guide data collection, augmentation, and retraining strategies. More broadly, they show how explainability can help assess whether model reasoning aligns with ecological understanding. We conclude by identifying key challenges and opportunities. By making model behavior more transparent and scientifically interrogable, XAI can help ensure that AI-supported ecological evidence is more reliable, understandable, and actionable for biodiversity conservation.

Chinese Translation

人工智能正在通过实现对来自相机陷阱、无人机、卫星、水下平台及其他传感系统收集的生态图像的自动分析，改变生物多样性监测。这些工具可以扩大保护评估的规模和速度，但许多计算机视觉模型仍然难以检查，这使得判断预测是否基于生态上有意义的信号，或是基于虚假的相关性、采样偏差和其他可能削弱保护决策的伪影变得具有挑战性。我们认为，可解释人工智能（XAI）应成为生态模型验证的标准组成部分，因为保护实践者越来越依赖于理解模型不仅是否准确，而且为什么准确。我们为将XAI应用于三种常见的生态计算机视觉任务提供了实用指导：图像分类、目标检测和图像分割。为了说明XAI如何支持生态模型的审计、改进和部署，我们展示了两个使用航空图像的案例研究：港湾海豹检测和鲸类解剖分割。这些例子展示了解释方法如何识别生物学上有意义的线索，揭示由背景和形状混淆引起的假阳性，发现边缘和遮挡效应，并指导数据收集、增强和再训练策略。更广泛地说，它们展示了可解释性如何帮助评估模型推理是否与生态理解相一致。我们最后识别了关键挑战和机遇。通过使模型行为更加透明和科学可检验，XAI可以帮助确保支持人工智能的生态证据在生物多样性保护中更加可靠、易懂和可操作。

View on arXiv Download PDF AI Translation

cs.CV / 31 / 2606.27671

Multi-Modal Conditioned High-Resolution Transformer for Urban Electromagnetic Field Map Prediction Download PDF

用于城市电磁场地图预测的多模态条件高分辨率变换器

Kim, Do-Eon, Park, Dongryul, Ahn, Seungyoung, Kang, Namwoo, Kim, Seong-heum, Kim, Seongsin

Abstract

Predicting electromagnetic field (EMF) strength in urban environments is essential for cellular network planning but computationally expensive with physics-based simulators. We propose a multi-conditioned dense prediction framework that generates 500 500 EMF maps from building layout images and antenna configurations. Our architecture uses a High-Resolution Transformer (HRFormer) backbone with two complementary conditioning mechanisms: Feature-wise Linear Modulation (FiLM) injects scalar antenna parameters into all backbone stages, while cross-attention fuses 1-D radiation pattern tokens with spatial features at the deepest stage. We further introduce transmitter-relative spatial channels encoding distance, proximity, and bearing from the antenna, enabling coordinate-consistent test-time augmentation (TTA) that reduces test MAE by 6.3%. To address the prediction difficulty imbalance across EMF maps, we design a composite loss combining masked L1, multi-scale structural similarity (MS-SSIM), and a focal L1 term that upweights high-signal pixels, outperforming individual loss components in all metrics. Our best model achieves a test MAE of 0.0461, a 25.2% improvement over a plain UNet baseline and 31.8% over an HRFormer-only baseline.Do-

Chinese Translation

在城市环境中预测电磁场（EMF）强度对于蜂窝网络规划至关重要，但使用基于物理的模拟器计算成本高昂。我们提出了一种多条件密集预测框架，该框架能够从建筑布局图像和天线配置生成500x500的EMF地图。我们的架构采用高分辨率变换器（HRFormer）作为主干，并结合了两种互补的条件机制：特征线性调制（FiLM）将标量天线参数注入到所有主干阶段，而交叉注意力在最深层融合了一维辐射模式标记与空间特征。我们进一步引入了相对于发射器的空间通道，编码距离、接近度和天线的方位，从而实现坐标一致的测试时间增强（TTA），将测试平均绝对误差（MAE）降低了6.3%。为了解决EMF地图之间的预测难度不平衡问题，我们设计了一种复合损失函数，结合了掩蔽L1、多尺度结构相似性（MS-SSIM）和一个聚焦L1项，后者对高信号像素进行加权，在所有指标上均优于单一损失组件。我们的最佳模型在测试中实现了0.0461的MAE，相较于普通UNet基线提高了25.2%，相较于仅使用HRFormer的基线提高了31.8%。

View on arXiv Download PDF AI Translation

cs.CV / 32 / 2606.27678

Two-Stage Cross-Domain Cervical Abnormality Screening with Cytopathological Image Synthesis and Knowledge Distillation

基于细胞病理图像合成和知识蒸馏的两阶段跨域宫颈异常筛查

Li, Jincheng, He, Yuzhi, Zhan, Yihui, Zhang, Xinmei, Sun, Yifei, Liu, Zelin, Zhang, Lichi, Shao, Minye, Zhao, Lili

Abstract

Cross-domain diagnosis remains a major challenge in cervical cell pathology due to pronounced domain shifts across institutions and the subtle visual differences among disease stages, which jointly impair model generalization. To address these issues, this paper proposes a two-stage framework for cross-domain cervical cell detection. In the first stage, we propose the Spatially-Continuous Unpaired Neural Schr\"odinger Bridge (SC-UNSB), which constructs a synthetic intermediate domain to mitigate cross-domain distribution shifts by modeling image translation as an entropy-regularized optimal transport process. In the second stage, we propose a dual-level feature alignment strategy within a knowledge distillation, which progressively aligns shallow structural features and deep semantic representations to facilitate the transfer of domain-invariant knowledge from the source to the target model. Experimental results demonstrate that the proposed method effectively mitigates domain shift and category ambiguity, improving the cross-domain detection performance.

Chinese Translation

跨域诊断在宫颈细胞病理学中仍然是一个主要挑战，因为不同机构之间存在明显的领域转变，以及疾病阶段之间微妙的视觉差异，这共同影响了模型的泛化能力。为了解决这些问题，本文提出了一种用于跨域宫颈细胞检测的两阶段框架。在第一阶段，我们提出了空间连续无配对神经薛定谔桥（SC-UNSB），该方法通过将图像翻译建模为熵正则化的最优传输过程，构建合成的中间域，以减轻跨域分布转变。在第二阶段，我们提出了一种在知识蒸馏中进行双层特征对齐的策略，该策略逐步对齐浅层结构特征和深层语义表示，以促进从源模型到目标模型的领域不变知识的转移。实验结果表明，所提方法有效减轻了领域转变和类别模糊，提高了跨域检测性能。

View on arXiv Download PDF AI Translation

cs.CV / 33 / 2606.27700

Joint Transcription and Decryption of Images of Encrypted Handwritten Documents: A Comparison with the Traditional Pipeline

加密手写文档图像的联合转录与解密：与传统流程的比较

Oliveros-Blanco, Marino, Kang, Lei, Fornés, Alicia, Megyesi, Beáta

Abstract

Historical encrypted manuscripts present a challenging problem at the intersection of cryptology, linguistics, paleography, and computer vision. Current automatic decipherment approaches usually rely on a two-stage pipeline: transcription of cipher symbols from manuscript images, followed by decryption into plaintext. However, this design is sensitive to transcription errors, which propagate to the final output. We present Direct Image Decryption, an end-to-end approach that directly maps encrypted manuscript images to plaintext, bypassing the intermediate transcription stage. Using the Copiale cipher as a case study, we build a synthetic data generation pipeline to create large-scale cipher-like training data and compare the traditional pipeline with the proposed joint architecture. Results show that joint image-to-plaintext modeling is a promising alternative to traditional transcription-based pipelines.

Chinese Translation

历史加密手稿在密码学、语言学、古文字学和计算机视觉的交叉领域中提出了一个具有挑战性的问题。目前的自动解密方法通常依赖于两阶段流程：首先从手稿图像中转录密码符号，然后将其解密为明文。然而，这种设计对转录错误非常敏感，这些错误会传播到最终输出。我们提出了一种直接图像解密（Direct Image Decryption）的方法，这是一种端到端的方案，直接将加密手稿图像映射到明文，绕过中间的转录阶段。以Copiale密码为案例研究，我们建立了一个合成数据生成流程，以创建大规模的类密码训练数据，并将传统流程与所提出的联合架构进行比较。结果表明，联合图像到明文建模是传统基于转录流程的有希望的替代方案。

View on arXiv Download PDF AI Translation

cs.CV / 34 / 2606.27708

ZooClaw-FashionSigLIP2: Distilled Fine-tuning for Robust Fashion Retrieval

ZooClaw-FashionSigLIP2：用于稳健时尚检索的蒸馏微调

Xue, Siqiao, Xu, Chunxue

Abstract

Adapting a foundation vision-language encoder to a specialized retrieval task creates a fundamental tradeoff: gains on the target distribution come at the cost of the foundation model's broad generalization, and fashion retrieval is a stringent instance of this problem. We present ZooClaw-FashionSigLIP2, a fashion-specialized SigLIP2-base model that resolves this tradeoff with a simple recipe -- full fine-tuning with knowledge distillation on curated in-domain data, followed by \wiseft~\citep{wortsman2022wiseft} weight interpolation with the base model -- and outperforms LoRA, larger backbones (up to 1B parameters), and external training data. Under fair evaluation, ZooClaw-FashionSigLIP2 outperforms all baselines on every benchmark in our suite. In addition, we release ZooClaw-Fashion, a new high-quality fashion retrieval benchmark, and a systematic quality analysis of widely-used benchmarks that exposes and mitigates structural biases in their public ground truth. We open-source the model weights and all evaluation artifacts to facilitate future research.

Chinese Translation

将基础视觉-语言编码器适应于特定的检索任务会产生一个基本的权衡：在目标分布上的收益是以基础模型的广泛泛化能力为代价的，而时尚检索是这一问题的严格实例。我们提出了ZooClaw-FashionSigLIP2，一个专门针对时尚的SigLIP2基础模型，通过一个简单的方案解决了这一权衡——在精心挑选的领域内数据上进行全面微调和知识蒸馏，随后与基础模型进行 extit{wiseft}~ extup{(Wortsman et al., 2022)}权重插值——并且超越了LoRA、更大的骨干网络（参数高达10亿）和外部训练数据。在公平评估下，ZooClaw-FashionSigLIP2在我们套件中的每个基准测试上都超越了所有基线。此外，我们发布了ZooClaw-Fashion，一个新的高质量时尚检索基准，以及对广泛使用的基准的系统质量分析，揭示并缓解了其公共真值中的结构性偏见。我们开源了模型权重和所有评估文档，以促进未来的研究。

View on arXiv Download PDF AI Translation

cs.CV / 35 / 2606.27718

MASS: Motion-Aligned Selective Scan for Refinement in Flow-Based Video Frame Interpolation

MASS：基于运动对齐的选择性扫描用于流动视频帧插值的精细化

Yoo, Jun-Sang, Jung, Seung-Won

Abstract

Video frame interpolation (VFI) remains a challenging task, particularly when dealing with large, non-linear motions and complex occlusions. While flow-based methods are prevalent, they often struggle with ambiguous correspondences. Recent VFI methods based on selective State Space Models (SSMs) are still limited by static grid-based scanning that misaligns with physical motion. In this paper, we propose Motion-Aligned Selective Scan (MASS), a novel framework that reformulates feature scanning from static spatial grids to dynamic motion trajectories. MASS builds a feature sequence along each pixel's flow-guided trajectory and aggregates it with an SSM. Specifically, we introduce a learnable non-linear path integration to approximate complex curved trajectories via residual velocity updates, and a velocity-aware SSM that dynamically adjusts the sampling budget and step size based on motion magnitude. This adaptive strategy allocates denser sampling to fast-motion regions while keeping static regions efficient. Furthermore, the aggregated states guide a refinement module to rectify intermediate flows and masks in an end-to-end manner. Extensive experiments indicate that MASS achieves highly competitive overall performance on standard benchmarks, establishing state-of-the-art results particularly in challenging scenarios with large displacements and complex dynamics.

Chinese Translation

视频帧插值（VFI）仍然是一项具有挑战性的任务，尤其是在处理大规模非线性运动和复杂遮挡时。尽管基于流的方法广泛应用，但它们常常在模糊对应关系上遇到困难。近期基于选择性状态空间模型（SSMs）的VFI方法仍然受到静态网格扫描的限制，这与物理运动不对齐。本文提出了一种新的框架——运动对齐选择性扫描（MASS），将特征扫描从静态空间网格重新构造为动态运动轨迹。MASS沿着每个像素的流引导轨迹构建特征序列，并与SSM进行聚合。具体而言，我们引入了一种可学习的非线性路径积分，通过残余速度更新来近似复杂的曲线路径，以及一种速度感知的SSM，根据运动幅度动态调整采样预算和步长。这种自适应策略为快速运动区域分配更密集的采样，同时保持静态区域的高效性。此外，聚合的状态指导一个精细化模块，以端到端的方式修正中间流和掩膜。大量实验表明，MASS在标准基准测试中实现了高度竞争的整体性能，特别是在大位移和复杂动态的挑战场景中建立了最先进的结果。

View on arXiv Download PDF AI Translation

cs.CV / 36 / 2606.27720

Scene and Human in One World: Reconstruction in a Feedforward Pass

场景与人类共存于一个世界：前馈传递中的重建

Shi, Boao, Feng, Qiao, Huang, Yiming, Liu, Lingjie

Abstract

Reconstructing humans in dynamic scenes from moving monocular cameras remains challenging due to scale ambiguity, human-scene misalignment, and occlusion interference. Rather than treating human mesh recovery and scene reconstruction as separate tasks, we believe that accurate human-scene reconstruction requires the two tasks to mutually inform each other: parametric human models offer semantic structure and metric-scale priors, while scene geometry provides spatial context for human localization and alignment. Built on this insight, we introduce SHOW, a mask-promptable human mesh recovery framework that couples feed-forward 3D scene reconstruction with Human Mesh Recovery in a unified metric space. SHOW injects human semantics and scale priors from parametric human models into normalized point-map prediction, enabling metric-scale scene reconstruction from inherently scale-ambiguous monocular input. In turn, the recovered scene geometry constrains human mesh estimation, encouraging spatially consistent human placement and improved human-scene alignment. To handle complex multi-person and cluttered scenes, SHOW further incorporates a promptable masking mechanism that enables flexible target-human selection while suppressing background distractions and occlusion interference. Through joint training, the model learns both human-aware geometric features and geometry-constrained human features, producing aligned metric-scale reconstructions from monocular human-centric videos. Extensive experiments demonstrate that SHOW improves metric-scale consistency, human-scene alignment, and reconstruction accuracy under challenging camera motion, occlusion, and cluttered backgrounds.

Chinese Translation

从移动的单目摄像机中重建动态场景中的人类仍然面临挑战，主要由于尺度模糊、人类与场景的不对齐以及遮挡干扰。我们认为，准确的人类-场景重建需要将人类网格恢复和场景重建视为相互关联的任务：参数化的人类模型提供语义结构和度量尺度先验，而场景几何则为人类定位和对齐提供空间上下文。基于这一见解，我们提出了SHOW，一个可通过掩模提示的人类网格恢复框架，将前馈3D场景重建与人类网格恢复结合在一个统一的度量空间中。SHOW将来自参数化人类模型的人类语义和尺度先验注入到归一化点图预测中，使得能够从本质上尺度模糊的单目输入中进行度量尺度的场景重建。反过来，恢复的场景几何约束了人类网格估计，促进了空间一致的人类放置和改善的人类-场景对齐。为了处理复杂的多人物和杂乱场景，SHOW进一步结合了可提示的掩模机制，使得在抑制背景干扰和遮挡干扰的同时，能够灵活选择目标人类。通过联合训练，模型学习到人类感知的几何特征和几何约束的人类特征，从单目以人类为中心的视频中生成对齐的度量尺度重建。大量实验表明，SHOW在挑战性的摄像机运动、遮挡和杂乱背景下提高了度量尺度的一致性、人类-场景对齐和重建精度。

View on arXiv Download PDF AI Translation

cs.CV / 37 / 2606.27729

Learning 1-Bit LiDAR-based Localization with Auxiliary Objective

基于1位LiDAR的定位学习与辅助目标

Yin, Kaijie, Zhang, Zhiyuan, Gao, Tian, Zhu, Wentao, Xu, Cheng-zhong, Kong, Hui

Abstract

6-DoF LiDAR-based localization is a fundamental capability for autonomous systems operating in large-scale outdoor environments. Many deep-learning-based localization methods have achieved promising performance so far. However, as one of the always-on modules competing for limited on-board computational resources, the localization module is expected to consume only a small portion of the overall compute budget. Most existing learning-based methods are still too heavy for this purpose. In contrast, binary neural networks (BNNs) offer an appealing solution, but the 1-bit compression causes severe information loss and performance drop. In this paper, we address this challenge by proposing Binarized LiDAR-based Localization (BiLoc), the first binary neural network framework for 6-DoF LiDAR localization. Specifically, we reinterpret the training of BNNs from the perspective of the information-bottleneck principle, aiming at retaining minimal yet sufficient representations for pose estimation while suppressing redundant variations. And we introduce an auxiliary objective that adaptively regulates information retention in the binary encoder, effectively mitigating the information loss caused by binarization. This auxiliary objective provides additional optimization signals that compensate for the limited representational capacity and the gradient mismatch inherent in BNNs. Extensive experiments on large-scale outdoor LiDAR datasets demonstrate that BiLoc establishes a new state of the art for LiDAR localization with BNNs.

Chinese Translation

六自由度（6-DoF）基于LiDAR的定位是自主系统在大规模户外环境中操作的基本能力。迄今为止，许多基于深度学习的定位方法已经取得了令人鼓舞的性能。然而，作为一个始终在线的模块，它与有限的车载计算资源竞争，定位模块预计只会消耗整体计算预算的一小部分。大多数现有的基于学习的方法仍然过于庞大，无法满足这一需求。相比之下，二进制神经网络（BNNs）提供了一个吸引人的解决方案，但1位压缩导致严重的信息损失和性能下降。本文通过提出二进制LiDAR定位（BiLoc），解决了这一挑战，这是第一个用于6-DoF LiDAR定位的二进制神经网络框架。具体而言，我们从信息瓶颈原理的角度重新解释BNNs的训练，旨在保留最小但足够的姿态估计表示，同时抑制冗余变化。我们引入了一个辅助目标，能够自适应地调节二进制编码器中的信息保留，有效减轻由二进制化引起的信息损失。这个辅助目标提供了额外的优化信号，以补偿BNNs固有的有限表示能力和梯度不匹配。对大规模户外LiDAR数据集的广泛实验表明，BiLoc在基于BNNs的LiDAR定位中建立了新的最先进水平。

View on arXiv Download PDF AI Translation

cs.CV / 38 / 2606.27741

SIFT: Self-Imagination Fine-Tuning for Physically Plausible Motion in Video Diffusion Models

SIFT：用于视频扩散模型中物理合理运动的自我想象微调

Wang, Ruoyu, Liu, Jialun, Huang, Huayang, Huang, Haibin, Wang, Jiepeng, Zhang, Chi, Li, Xuelong, Wu, Yu

Abstract

Recent advances in video diffusion models have greatly improved visual fidelity, yet their generated motions often violate physical plausibility. We observe a common kinematic failure, "motion entanglement", the unintended coupling of independent motion sources, such as camera movement and object motion. We identify that this issue stems from data bias and the reconstruction-based training design of diffusion models. Training on noisy videos that still retain coarse motion cues inadvertently encourages the model to replicate existing motion without an incentive to learn how to model kinematically-grounded motions. To address this, we propose a Self-Imagination Fine-Tuning (SIFT) paradigm, which enables the model to learn from its own generated videos rather than directly reconstructing real ones, breaking the reconstruction shortcut. We further employ motion-aware discriminative supervision and a progressive hard-case replay strategy to stabilize and accelerate learning. By leveraging freely-generated text prompts, our method can densely cover a broad motion space, including rare or finely-disentangled scenarios that would be costly to collect as video data. Extensive experiments demonstrate that our approach substantially improves the physical realism, motion disentanglement, and controllability of generated videos.

Chinese Translation

最近，视频扩散模型的进展大大提高了视觉逼真度，但其生成的运动往往违反物理合理性。我们观察到一种常见的运动学失败，即“运动纠缠”，即独立运动源（如相机运动和物体运动）之间的意外耦合。我们确定这个问题源于数据偏差和扩散模型的基于重建的训练设计。在噪声视频上训练，这些视频仍然保留粗略的运动线索，无意中促使模型复制现有运动，而没有激励其学习如何建模基于运动学的运动。为了解决这个问题，我们提出了一种自我想象微调（Self-Imagination Fine-Tuning，SIFT）范式，使模型能够从自身生成的视频中学习，而不是直接重建真实视频，从而打破重建捷径。我们进一步采用运动感知的判别监督和渐进式难例重放策略，以稳定和加速学习。通过利用自由生成的文本提示，我们的方法可以密集覆盖广泛的运动空间，包括稀有或细致分离的场景，这些场景作为视频数据收集成本高昂。大量实验表明，我们的方法显著提高了生成视频的物理真实感、运动解耦和可控性。

View on arXiv Download PDF AI Translation

cs.CV / 39 / 2606.27745

Panoramic Scene Analysis: A Survey from Distortion-Aware Engineering to Sphere-Native Foundation Modeling

全景场景分析：从考虑畸变的工程到球面原生基础建模的综述

Zhu, Qinfeng, Fan, Lei

Abstract

Panoramic images capture the complete visual sphere in a single frame, providing spatial context unattainable by conventional cameras. Yet this completeness comes at a geometric cost: the 2-sphere cannot be faithfully mapped to the plane, and every planar representation introduces distortions that violate the assumptions underlying standard vision architectures. This survey traces the evolution of panoramic scene analysis along a methodological trajectory, from projection-based adaptation, through distortion-aware engineering, to sphere-native modeling and geometry-aware tokenization for foundation models, and argues that this evolution reflects a progressive deepening of geometric commitment rather than a simple accumulation of techniques. We organize the literature along two orthogonal dimensions: architectural design (how operators interact with spherical geometry) and training paradigm (how knowledge is transferred across domains). Covering dense prediction (semantic segmentation, depth estimation, and room layout estimation), unified multi-task understanding, open-world perception, vision-language reasoning, and dynamic video analysis, we identify a central unresolved tension: among the methods surveyed, none simultaneously delivers strict spherical equivariance and full reuse of perspective-pretrained foundation-model weights, and we argue that this is a structural rather than incidental gap. We further expose five systematic gaps in current evaluation protocols, namely the absence of spherical-area-weighted metrics, seam-consistency testing, polar-robustness stratification, cross-projection generalization, and open-world protocol standardization, and propose a six-point research roadmap toward general-purpose panoramic intelligence. The corresponding repository is publicly available at: https://github.com/zhuqinfeng1999/Awesome-Panoramic-Scene-Analysis.

Chinese Translation

全景图像在单帧中捕捉完整的视觉球体，提供了传统相机无法获得的空间上下文。然而，这种完整性带来了几何上的代价：二维球体无法忠实地映射到平面，每一种平面表示都会引入违反标准视觉架构假设的畸变。本文综述了全景场景分析沿着方法论轨迹的发展，从基于投影的适应、经过考虑畸变的工程，到球面原生建模和针对基础模型的几何感知标记化，认为这一演变反映了几何承诺的逐步深化，而不仅仅是技术的简单积累。我们从两个正交维度组织文献：架构设计（操作符如何与球面几何相互作用）和训练范式（知识如何跨领域转移）。涵盖密集预测（语义分割、深度估计和房间布局估计）、统一的多任务理解、开放世界感知、视觉-语言推理和动态视频分析，我们识别出一个核心未解决的紧张关系：在所调查的方法中，没有一种方法能够同时实现严格的球面等变性和完全重用经过透视预训练的基础模型权重，我们认为这是一个结构性而非偶然的缺口。我们进一步揭示了当前评估协议中的五个系统性缺口，即缺乏球面面积加权指标、接缝一致性测试、极坐标鲁棒性分层、跨投影泛化和开放世界协议标准化，并提出了一个六点研究路线图，以实现通用的全景智能。相应的代码库已公开可用，网址为：https://github.com/zhuqinfeng1999/Awesome-Panoramic-Scene-Analysis。

View on arXiv Download PDF AI Translation

cs.CV / 40 / 2606.27760

PixelU: A U-Shaped Transformer for Efficient End-to-End Pixel Diffusion

PixelU：一种用于高效端到端像素扩散的U形变换器

Guo, Zipeng, Ma, Lichen, He, Yu, Fu, Xiaolong, Fu, Jingling, Huang, Junshi, Li, Yan

Abstract

End-to-end pixel-space diffusion models bypass the lossy compression of Latent Diffusion Models (LDMs) but struggle to jointly model low-frequency semantics and high-frequency signals in high-dimensional space. Existing works heavily rely on complex pixel decoders to alleviate this issue. In this paper, we challenge this trend by revealing that these decoders primarily compensate for the optimization difficulties inherent to velocity prediction ($v$-prediction). Under the clean data paradigm ($x$-prediction), they are redundant. Motivated by this insight, we advocate for simplicity over complexity and introduce PixelU, a minimalist, single-stage U-shaped Diffusion Transformer tailored for pixel space. PixelU abandons auxiliary decoders in favor of zero-cost skip connections, which provide an "information highway" that directly routes uncorrupted high-frequency spatial details from shallow to deep layers. To further enable the backbone to focus exclusively on modeling low-frequency semantics, we introduce a constant-channel spatial down-sampling mechanism as a natural low-pass filter, which compresses deep features into a compact, low-frequency semantic manifold. Extensive experiments demonstrate that this decoupling of frequencies could outperform the strong baseline (JiT-G) with only about 1/3 of its computation cost. On ImageNet 256$\times$256 and 512$\times$512, PixelU achieves FID of 1.63 and 1.92 respectively, surpassing recent pixel-space methods and establishing a simple yet powerful new paradigm for end-to-end diffusion models.

Chinese Translation

端到端像素空间扩散模型绕过了潜在扩散模型（Latent Diffusion Models, LDMs）的有损压缩，但在高维空间中共同建模低频语义和高频信号方面面临困难。现有研究在很大程度上依赖复杂的像素解码器来缓解这一问题。本文挑战这一趋势，揭示这些解码器主要是为了补偿速度预测（$v$-预测）固有的优化困难。在干净数据范式下（$x$-预测），它们是多余的。基于这一见解，我们提倡简单胜于复杂，提出了PixelU，一种为像素空间量身定制的极简单阶段U形扩散变换器。PixelU放弃了辅助解码器，采用零成本跳跃连接，提供了一条“信息高速公路”，直接将未损坏的高频空间细节从浅层路由到深层。为了进一步使主干网络专注于建模低频语义，我们引入了一种常量通道空间下采样机制，作为一种自然的低通滤波器，将深层特征压缩为紧凑的低频语义流形。大量实验表明，这种频率解耦能够以约三分之一的计算成本超越强基线（JiT-G）。在ImageNet 256$ imes$256和512$ imes$512上，PixelU分别达到了1.63和1.92的FID，超越了近期的像素空间方法，并为端到端扩散模型建立了一个简单而强大的新范式。

View on arXiv Download PDF AI Translation

cs.CV / 41 / 2606.27772

An Embedded Real-Time License Plate Recognition System for Complex Traffic Scenes

复杂交通场景下的嵌入式实时车牌识别系统

Pasqual, Anuki, Lokugeegana, Dulan, Thiriloganathan, Manimohan, Rathnayake, Nuthya, Samarasinghe, Kithsiri, Thanthrige, Udaya S. K. P. Miriya

Abstract

Vehicle license plate recognition is an integral component of intelligent transportation systems. In this work, we present an embedded real-time license plate recognition system customized for developing countries. We address the challenge of handling complex, unstructured traffic scenes with diverse vehicle types while implementing the system on an embedded platform for low-cost deployment. Our method consists of license plate detection on a multi-vehicle image, followed by character recognition on the detected license plates. Both steps use lightweight convolutional neural networks to balance accuracy and efficiency. We also introduce the SL-LPR dataset of Sri Lankan road images, which contains a variety of vehicle types and traffic conditions typically seen in developing countries. On this dataset, the license plate detection and character recognition models achieved 93.6% mAP and 87.88% accuracy, respectively, and were competitive against larger models on several public datasets. To achieve real-time performance in a resource-constrained embedded environment, we applied low-bitwidth quantization using the Brevitas library and implemented FPGA acceleration for the models using the FINN framework. The end-to-end system can operate at 11.5~FPS when implemented on the Xilinx Kria KV260 platform. These results demonstrate that our system is effective for real-time license plate recognition on an embedded device, even in complex traffic scenarios. The SL-LPR dataset is available for research use at: https://github.com/sl-lpr-uom/SL-LPR.git.

Chinese Translation

车辆车牌识别是智能交通系统的重要组成部分。在本研究中，我们提出了一种针对发展中国家的嵌入式实时车牌识别系统。我们解决了在复杂、非结构化交通场景中处理多种车辆类型的挑战，同时在嵌入式平台上实现该系统，以便于低成本部署。我们的方法包括在多车辆图像上进行车牌检测，随后对检测到的车牌进行字符识别。这两个步骤均使用轻量级卷积神经网络，以平衡准确性和效率。我们还引入了斯里兰卡道路图像的SL-LPR数据集，其中包含发展中国家常见的多种车辆类型和交通条件。在该数据集上，车牌检测和字符识别模型分别达到了93.6%的平均精度均值（mAP）和87.88%的准确率，并且在多个公共数据集上与更大模型的表现具有竞争力。为了在资源受限的嵌入式环境中实现实时性能，我们使用Brevitas库进行了低位宽量化，并利用FINN框架对模型进行了FPGA加速。该端到端系统在Xilinx Kria KV260平台上运行时可达到11.5帧每秒（FPS）。这些结果表明，我们的系统在复杂交通场景中能够有效地在嵌入式设备上实现实时车牌识别。SL-LPR数据集可供研究使用，网址为：https://github.com/sl-lpr-uom/SL-LPR.git。

View on arXiv Download PDF AI Translation

cs.CV / 42 / 2606.27773

ModaFlow: Modality-Aware Flow Matching for High-Fidelity Virtual Try-On

ModaFlow：面向模态的流匹配用于高保真虚拟试穿

Sai, Xiangyu, Madadi, Meysam, Escalera, Sergio, Xu, Yong

Abstract

Image-based virtual try-on has emerged as a compelling task in e-commerce and augmented reality, yet existing methods struggle to simultaneously preserve fine garment semantics and adapt to diverse person body geometries under large clothing-body deformations. We present ModaFlow, a modality-aware flow-matching based framework for high-fidelity virtual try-on that achieves precise alignment between textual descriptions and garment appearance. Unlike prior methods that treat multimodal conditions uniformly, ModaFlow introduces a modality-aware guidance scheme: visual garment embeddings extracted by a pretrained image prompt adapter provide deterministic, persistent structural guidance, while textual embeddings generated from garment descriptions are controlled via classifier-free guidance (CFG) with adaptive scaling and zero-initialized velocity. To further enhance flow field accuracy, we propose two regularization losses, cosine similarity and perceptual flow discrimination, that jointly improve directional consistency and perceptual realism of the velocity field. Additionally, a mask manipulation strategy stochastically samples among box, transparent, and relaxed masks during training, simulating diverse occlusion scenarios and enabling robust inference under unpaired settings where only a box mask is available. Experiments show that ModaFlow achieves state-of-the-art results in both qualitative and quantitative evaluations, reducing FID by approximately 30% on paired and 20% on unpaired benchmarks.

Chinese Translation

基于图像的虚拟试穿在电子商务和增强现实中已成为一项引人注目的任务，但现有方法在同时保留精细服装语义和适应在大规模服装-身体变形下的多样化人体几何形状方面面临挑战。我们提出了ModaFlow，一种基于面向模态的流匹配框架，用于高保真虚拟试穿，能够实现文本描述与服装外观之间的精确对齐。与以往将多模态条件统一处理的方法不同，ModaFlow引入了一种面向模态的引导机制：通过预训练的图像提示适配器提取的视觉服装嵌入提供确定性、持久的结构引导，而从服装描述生成的文本嵌入则通过无分类器引导（CFG）进行控制，具有自适应缩放和零初始化速度。为了进一步提高流场的准确性，我们提出了两种正则化损失：余弦相似性和感知流区分，这两者共同改善了速度场的方向一致性和感知真实感。此外，一种掩模操作策略在训练过程中随机选择盒状、透明和放松掩模，模拟多样的遮挡场景，并在仅提供盒状掩模的无配对设置下实现稳健推理。实验表明，ModaFlow在定性和定量评估中均达到了最先进的结果，在配对基准上将FID降低了约30%，在无配对基准上降低了20%。

View on arXiv Download PDF AI Translation

cs.CV / 43 / 2606.27777

TRUST: Efficient Abdominal Trauma Recognition via Image-to-Ultrasound-Video Transfer Learning

TRUST：通过图像到超声视频的迁移学习实现高效的腹部创伤识别

Wang, Enguang, Zhou, Hao, Gao, Shuo, Liu, Tuo, Zhou, Guangquan

Abstract

Abdominal ultrasound is indispensable for rapid, noninvasive trauma triage. However, interpreting the subtle dynamic cues embedded in continuous scanning is time-intensive and operator-dependent. Parameter-Efficient Image-to-Video Transfer Learning (PEIVTL), which efficiently adapts pre-trained image models to the video domain, notably through visual-textual alignment, offers a promising paradigm for ultrasound video analysis. Nevertheless, substantial spatiotemporal and semantic variations arising from physician-dependent scanning practices continue to limit the effectiveness and generalizability of this framework. We propose TRUST, a scan-aware PEIVTL framework that explicitly models fine-grained spatiotemporal variations to enable reliable ultrasound video understanding. First, we introduce a Cross-Frequency Collaborative Adapter (CFCA) that establishes mutual constraints between low- and high-frequency components, enhancing discriminative spatial feature extraction under heavy speckle corruption. Second, we design a Multi-Granularity Motion-Aware (MGMA) module that integrates local temporal convolutions with motion-prior-guided global self-attention, jointly capturing stable intra-view patterns and abrupt inter-view transitions to characterize complex scanning dynamics. Third, a Visual Query Semantic Aggregation (VQSA) module dynamically generates text prototypes conditioned on visual features, enabling adaptive visual-textual alignment robust to intra-class variability under diverse scanning conditions. Experiments on in-house ultrasound trauma datasets demonstrate that TRUST outperforms state-of-the-art methods by 9.63% with superior computational efficiency.

Chinese Translation

腹部超声在快速、非侵入性创伤分诊中不可或缺。然而，解读连续扫描中嵌入的微妙动态线索既耗时又依赖操作员。参数高效的图像到视频迁移学习（PEIVTL）通过视觉-文本对齐有效地将预训练的图像模型适应于视频领域，为超声视频分析提供了一个有前景的范式。然而，由于医生依赖的扫描实践所产生的显著时空和语义变化，仍然限制了该框架的有效性和普适性。我们提出了TRUST，一个扫描感知的PEIVTL框架，明确建模细粒度的时空变化，以实现可靠的超声视频理解。首先，我们引入了跨频率协作适配器（CFCA），在低频和高频成分之间建立相互约束，在严重斑点干扰下增强判别空间特征提取。其次，我们设计了多粒度运动感知（MGMA）模块，将局部时间卷积与运动先验引导的全局自注意力相结合，共同捕捉稳定的视图内模式和突发的视图间转换，以表征复杂的扫描动态。第三，视觉查询语义聚合（VQSA）模块动态生成基于视觉特征的文本原型，使得在多样化扫描条件下能够适应性地进行视觉-文本对齐，增强对类内变异性的鲁棒性。在内部超声创伤数据集上的实验表明，TRUST在计算效率上优于最先进的方法，提升了9.63%。

View on arXiv Download PDF AI Translation

cs.CV / 44 / 2606.27779

MindFlow: Harmonizing Cognitive Semantics and Acoustic Dynamics for Facial Animation Generation in Dyadic Conversations

MindFlow：协调认知语义与声学动态以生成双人对话中的面部动画

Chen, Hejia, Zhang, Haoxian, He, Xu, Liu, Xiaoqiang, Wan, Pengfei, Zhang, Shoulong, Li, Shuai

Abstract

Generating lifelike facial animation for dyadic conversations requires reconciling high-level cognitive intent with precise low-level motor reflexes, yet existing methods fall short in the semantic understanding of dialogue context and in precise dynamic control. In this paper, we propose MindFlow, a dual-pathway generative framework inspired by the Ventral-Dorsal pathway model in neuroscience, which decouples generation into two collaborative streams, thereby harmonizing deep semantic reasoning with fine-grained control. In the Ventral module, we transform the conventional Sentence-Action approach into a novel Chunk-State approach that models raw acoustic streams as a context-aware, evolving emotional state chain, capturing subtle paralinguistic nuances and mid-utterance emotional shifts missed by sentence-level modeling. The Dorsal module features a conditional autoregressive flow matching network for high-fidelity facial motion, driven by high-frequency acoustic cues and modulated by emotion states, plus a Selective Acoustic Injector for adaptive audio gating to ensure robustness in talking-and-listening dynamics without interference. Extensive experiments demonstrate that MindFlow achieves superior semantic appropriateness and motion naturalness compared to state-of-the-art baselines.

Chinese Translation

为双人对话生成逼真的面部动画需要将高层次的认知意图与精确的低层次运动反射相结合，但现有方法在对话语境的语义理解和精确动态控制方面存在不足。本文提出了MindFlow，一种受神经科学中腹侧-背侧通路模型启发的双通道生成框架，该框架将生成过程解耦为两个协作流，从而协调深层语义推理与细粒度控制。在腹侧模块中，我们将传统的句子-动作方法转变为一种新颖的块-状态方法，该方法将原始声学流建模为一种上下文感知的、不断演变的情感状态链，捕捉句子级建模所遗漏的细微副语言细节和中间话语的情感变化。背侧模块则采用条件自回归流匹配网络以实现高保真度的面部运动，该运动由高频声学线索驱动，并由情感状态调制，同时配备选择性声学注入器以实现自适应音频门控，确保在对话和倾听动态中保持鲁棒性而不产生干扰。大量实验表明，MindFlow在语义适宜性和运动自然性方面优于最先进的基线方法。

View on arXiv Download PDF AI Translation

cs.CV / 45 / 2606.27784

Improving Adversarial Robustness via Activation Amplification and Attenuation

通过激活放大与衰减提高对抗鲁棒性

Gonçalves, Taïga, Huang, Yongsong, Miyazaki, Tomo, Omachi, Shinichiro

Abstract

The existence of adversarial attacks is often attributed to the presence of non-robust features in neural networks. While prior defenses reduce their impact via pruning, masking, or feature recalibration, we instead propose to jointly learn to amplify and attenuate these signals through a simple activation scaling mechanism. To this end, we introduce Activation Amplification and Attenuation (A3), a lightweight plug-in module that enhances adversarial robustness with minimal modifications of the activations. A3 dynamically rescales the activations using a learnable mask and a scaling factor derived from the original activation magnitudes. The influence of adversarial perturbations can be amplified or attenuated using the same learnable parameters by simply flipping the sign of the scaling operation. The amplified signals serve as negative references to construct novel contrastive and ranking loss functions. Experimental analysis shows that learning to degrade the predictions in amplification mode simultaneously improves adversarial robustness in attenuation mode. Moreover, A3 relies on only a small number of learnable parameters, with most of its behavior being determined by the scaling mechanism rather than additional network capacity. Extensive experiments demonstrate that integrating A3 into different backbones, datasets, and training methods consistently improves adversarial robustness while introducing negligible computational and memory overhead compared to existing plug-in modules. Code is available at: https://github.com/tgoncalv/A3.

Chinese Translation

对抗攻击的存在通常归因于神经网络中非鲁棒特征的存在。虽然之前的防御方法通过剪枝、遮蔽或特征重新校准来减少其影响，我们则提出通过简单的激活缩放机制共同学习放大和衰减这些信号。为此，我们引入了激活放大与衰减（Activation Amplification and Attenuation, A3），这是一个轻量级的插件模块，能够在对激活进行最小修改的情况下增强对抗鲁棒性。A3使用可学习的掩码和源自原始激活幅度的缩放因子动态地重新缩放激活。通过简单地翻转缩放操作的符号，可以使用相同的可学习参数放大或衰减对抗扰动的影响。放大的信号作为负参考，用于构建新颖的对比损失和排名损失函数。实验分析表明，在放大模式下学习降低预测值同时能在衰减模式下提高对抗鲁棒性。此外，A3仅依赖少量可学习参数，其大部分行为由缩放机制决定，而非额外的网络容量。大量实验表明，将A3集成到不同的主干网络、数据集和训练方法中，始终能够提高对抗鲁棒性，同时与现有插件模块相比，带来的计算和内存开销微乎其微。代码可在以下链接获取：https://github.com/tgoncalv/A3。

View on arXiv Download PDF AI Translation

cs.CV / 46 / 2606.27794

Text as Illumination: Spatial Contrastive Retinex Learning for Language-guided Medical Image Segmentation

文本作为照明：基于空间对比Retinex学习的语言引导医学图像分割

Shi, Jian, Zhen, Cheng, Zhang, Pingping, Xu, Rui, Lv, Yanan, Ma, Yili, Bi, Huan, Li, Haojie, Lu, Huchuan

Abstract

Language-guided Medical Image Segmentation (LMIS) has shown great potential to improve the delineation of anatomical structures and lesions by integrating clinical textual information. Existing methods generally rely on either implicit interaction between textual and visual features or auxiliary coarse-grained supervision for cross-modal alignment. However, these methods lack explicit and fine-grained constraints to ensure semantic consistency, causing a mismatch between language and the segmentation outputs. To address this issue, we propose Text-as-Illumination Retinex Network (TIRNet), a novel Retinex-inspired framework that treats text embeddings as semantic illumination for feature modulation, thereby improving semantic consistency in LMIS. TIRNet introduces two key blocks integrated at each decoder stage: (1) the Retinex-inspired Text Modulation Block (RTMB), which employs positive and negative illumination maps to enhance text-relevant foreground features and suppress background interference; and (2) the Consistent Detail Compensation Block (CDCB), which selectively recovers high-frequency details via a consistency-gated mechanism conditioned on illumination reliability. Furthermore, we propose a Multi-Scale Illumination Supervision Loss (MSIS-Loss), comprising a Region-Grounded Contrastive Loss (RGC-Loss) that enforces cross-modal similarity to be concentrated in text-relevant foreground regions and suppressed in background regions, and a Background Suppression Loss (BS-Loss) that provides pixel-level supervision for negative illumination maps, jointly ensuring a precise cross-modal alignment at each decoder stage. Extensive experiments on the MosMedData+ and QaTa-COV19 datasets demonstrate that TIRNet achieves state-of-the-art performance in LMIS. The code is available at: https://github.com/anaanaa/TIRNet.

Chinese Translation

语言引导医学图像分割（LMIS）通过整合临床文本信息，展现出改善解剖结构和病变轮廓描绘的巨大潜力。现有方法通常依赖于文本特征与视觉特征之间的隐式交互或辅助的粗粒度监督进行跨模态对齐。然而，这些方法缺乏明确且细粒度的约束，以确保语义一致性，导致语言与分割输出之间的不匹配。为了解决这一问题，我们提出了文本作为照明的Retinex网络（TIRNet），这是一个新颖的受Retinex启发的框架，将文本嵌入视为特征调制的语义照明，从而提高LMIS中的语义一致性。TIRNet在每个解码器阶段引入了两个关键模块：（1）受Retinex启发的文本调制模块（RTMB），利用正负照明图增强与文本相关的前景特征并抑制背景干扰；（2）一致性细节补偿模块（CDCB），通过基于照明可靠性条件的一致性门控机制选择性地恢复高频细节。此外，我们提出了一种多尺度照明监督损失（MSIS-Loss），包括区域基础对比损失（RGC-Loss），该损失强制跨模态相似性集中在与文本相关的前景区域，并在背景区域被抑制，以及背景抑制损失（BS-Loss），为负照明图提供像素级监督，联合确保每个解码器阶段的精确跨模态对齐。在MosMedData+和QaTa-COV19数据集上的大量实验表明，TIRNet在LMIS中实现了最先进的性能。代码可在以下链接获取：https://github.com/anaanaa/TIRNet。

View on arXiv Download PDF AI Translation

cs.CV / 47 / 2606.27818

Scalable and Differentiable Point-Cloud Registration Using Maximum Mean Discrepancy

基于最大均值差异的可扩展且可微分的点云配准

Crane, Rixon, Maken, Fahira Afzal, Lawrance, Nicholas, Funiak, Stanislav, Khosoussi, Kasra, Xu, Ming, Tsuchida, Russell

Abstract

We present MMD-Reg, a novel correspondence-free approach to point-cloud registration that is differentiable and has linear computational complexity in the number of points. We model registration as a nonlinear least-squares problem based on the Maximum Mean Discrepancy, approximated using random Fourier features. The resulting objective can be solved efficiently with standard methods such as Levenberg-Marquardt, and the solution is differentiable via the implicit function theorem. This allows MMD-Reg to be used as a differentiable optimization layer within end-to-end trainable models, supporting registration under challenging conditions such as poor initial alignment and partial overlap. We demonstrate this Neural MMD-Reg formulation by integrating the layer with a set transformer, training the resulting model in supervised and unsupervised settings, and comparing its performance against recent learning-based methods. We also evaluate standalone MMD-Reg, comparing its accuracy and scalability against widely used non-learning-based registration methods.

Chinese Translation

我们提出了MMD-Reg，一种新颖的无对应点云配准方法，该方法是可微分的，并且在点数上具有线性计算复杂度。我们将配准建模为基于最大均值差异的非线性最小二乘问题，该问题通过随机傅里叶特征进行近似。得到的目标可以通过标准方法（如Levenberg-Marquardt）有效求解，并且根据隐函数定理，解是可微分的。这使得MMD-Reg可以作为可微分优化层在端到端可训练模型中使用，支持在诸如初始对齐不佳和部分重叠等挑战性条件下的配准。我们通过将该层与集合变换器集成来展示这种神经MMD-Reg的形式，在监督和无监督设置中训练得到的模型，并将其性能与近期的基于学习的方法进行比较。我们还评估了独立的MMD-Reg，比较其准确性和可扩展性与广泛使用的非学习基准配准方法。

View on arXiv Download PDF AI Translation

cs.CV / 48 / 2606.27828

Video-MME-Logical: A Controlled Diagnostic Benchmark for Video Temporal-Logical Reasoning

视频-MME-逻辑：一个用于视频时间逻辑推理的受控诊断基准

Kwan, Hohin, Li, Hongyu, Zhang, Ray, Zhang, Manyuan, Kong, Xianghao, Rao, Anyi, Xie, Jiahao, Liu, Si

Abstract

Recent interest in multimodal large language models (MLLMs) raises a central question: can they reason over dynamic visual evidence rather than merely recognize objects or events in individual frames? This ability, which we refer to as video temporal-logical reasoning, requires models to maintain, update, and compose evidence as visual states evolve across frames. Existing video benchmarks often conflate this capability with scene complexity, static recognition, or uncontrolled temporal variation. To isolate this capability, we introduce Video-MME-Logical, a controlled benchmark organized around five temporal-logical operations: state tracking, sequential counting, temporal ordering, dynamic spatiality, and structural composition. The benchmark contains 25 fine-grained task categories generated with controlled object states, transitions, temporal dependencies, and logical compositions. It enables difficulty-controlled final-answer evaluation by varying temporal horizon and reasoning complexity, and supports intermediate-state diagnostics by verifying whether models recover the required logical reasoning trace before producing the final answer. Experiments with state-of-the-art MLLMs reveal a substantial human-model gap, especially as temporal-logical complexity increases. Supervised fine-tuning on up to 500K generated samples improves performance but remains insufficient to close the reasoning gap, positioning Video-MME-Logical as a scalable testbed for analyzing and improving temporal-logical reasoning in MLLMs.

Chinese Translation

近年来对多模态大型语言模型（MLLMs）的关注引发了一个核心问题：它们能否在动态视觉证据上进行推理，而不仅仅是在单个帧中识别对象或事件？我们称这种能力为视频时间逻辑推理，它要求模型在视觉状态跨帧演变的过程中维护、更新和组合证据。现有的视频基准往往将这一能力与场景复杂性、静态识别或不受控制的时间变化混为一谈。为了孤立这一能力，我们引入了视频-MME-逻辑，这是一个围绕五种时间逻辑操作组织的受控基准：状态跟踪、顺序计数、时间排序、动态空间性和结构组合。该基准包含25个细粒度任务类别，这些类别是通过控制对象状态、转变、时间依赖关系和逻辑组合生成的。它通过改变时间范围和推理复杂性，实现了难度受控的最终答案评估，并通过验证模型在生成最终答案之前是否恢复所需的逻辑推理轨迹，支持中间状态诊断。与最先进的MLLMs的实验揭示了显著的人类-模型差距，尤其是在时间逻辑复杂性增加时。对多达50万生成样本的监督微调提高了性能，但仍不足以缩小推理差距，使得视频-MME-逻辑成为分析和改善MLLMs中时间逻辑推理的可扩展测试平台。

View on arXiv Download PDF AI Translation

cs.CV / 49 / 2606.27829

CSD: Content-aware Speculative Decoding for Efficient Image Generation

CSD：内容感知的推测解码用于高效图像生成

Wang, Mingcheng, Qiao, Junbo, Li, Yunchen, Jiang, Lingfu, Li, Wei, Hu, Jie, Xie, Jiao, Yu, Zhou, Chen, Xinghao, Zhang, Guixu, Lin, Shaohui

Abstract

Speculative decoding (SD) has emerged as a key solution to accelerate the inference of autoregressive models. However, in the field of image generation, it faces the challenge of low acceptance rates, and directly relaxing its criteria leads to degradation in image quality. In this paper, we propose a novel content-aware speculative decoding algorithm, termed CSD, which integrates an entropy-based probability relaxation mechanism with an optimal resampling strategy to enhance the inference efficiency for autoregressive image generation. By leveraging the informational uncertainty inherent in different regions of an image, CSD dynamically adjusts the acceptance probability of candidate tokens, increasing the acceptance rate in low-detail areas to accelerate generation. Moreover, a distribution alignment filter is introduced to ensure the output distribution to be aligned with the target model, which significantly improves the generative quality. Experiments conducted on Lumina-mGPT and Janus-Pro demonstrate that the superiority of the proposed CSD. Our source code is available at https://github.com/aderfebr/CSD.

Chinese Translation

推测解码（Speculative Decoding, SD）已成为加速自回归模型推理的关键解决方案。然而，在图像生成领域，它面临着低接受率的挑战，直接放宽其标准会导致图像质量下降。本文提出了一种新颖的内容感知推测解码算法，称为CSD（Content-aware Speculative Decoding），该算法将基于熵的概率放宽机制与最优重采样策略相结合，以提高自回归图像生成的推理效率。通过利用图像不同区域固有的信息不确定性，CSD动态调整候选标记的接受概率，在低细节区域提高接受率，从而加速生成。此外，引入了一种分布对齐过滤器，以确保输出分布与目标模型对齐，从而显著提高生成质量。在Lumina-mGPT和Janus-Pro上的实验表明了所提出的CSD的优越性。我们的源代码可在https://github.com/aderfebr/CSD获取。

View on arXiv Download PDF AI Translation

cs.CV / 50 / 2606.27831

Hippocampus-DETR: An Explicit Memory Object Detection Framework Based on Hippocampus Modeling

海马-DETR：基于海马建模的显式记忆目标检测框架

Shi, Zhaoning, Ma, Bo, Xu, Hao, Yang, Zepeng, Liang, Bo

Abstract

This paper addresses the lack of explicit memory mechanisms in current object detection models and proposes Hippocampus-DETR, a novel detection framework based on biological hippocampal memory modeling. This framework integrates a hippocampal memory network module, HipNet, into the DETR architecture and systematically simulates the anatomical structure and functional organization of hippocampal subregions, including the entorhinal cortex, dentate gyrus, CA3, CA1, and subiculum. Through this design, Hippocampus-DETR realizes pattern separation, pattern completion, importance filtering, and information integration of visual encoding features. During training, different memory submodules are optimized using a layer-wise training strategy, ultimately forming a memory system with memory retrieval and completion capabilities. Experimental results demonstrate that Hippocampus-DETR achieves higher detection accuracy than current mainstream models. More importantly, models equipped with this framework also exhibit excellent generalization ability and data efficiency in tasks such as few-shot image classification, multimodal feature construction, and image restoration. Subsequent experiments further validate the functional necessity and internal interpretability of each memory submodule. This study not only provides a novel object detection framework, but also offers a feasible technical pathway for integrating neurocognitive mechanisms with deep learning models, highlighting its significant value in improving model learning efficiency and task robustness. The project is available at https://github.com/2186cloud/hipnet.

Chinese Translation

本文解决了当前目标检测模型中缺乏显式记忆机制的问题，提出了海马-DETR，一种基于生物海马记忆建模的新型检测框架。该框架将海马记忆网络模块 HipNet 集成到 DETR 架构中，系统地模拟了海马亚区的解剖结构和功能组织，包括内嗅皮层、齿状回、CA3、CA1 和海马下部。通过这一设计，海马-DETR 实现了视觉编码特征的模式分离、模式完成、重要性过滤和信息整合。在训练过程中，采用逐层训练策略优化不同的记忆子模块，最终形成具有记忆检索和完成能力的记忆系统。实验结果表明，海马-DETR 的检测准确率高于当前主流模型。更重要的是，配备该框架的模型在少样本图像分类、多模态特征构建和图像恢复等任务中也表现出优异的泛化能力和数据效率。后续实验进一步验证了每个记忆子模块的功能必要性和内部可解释性。本研究不仅提供了一种新型的目标检测框架，还为将神经认知机制与深度学习模型相结合提供了一条可行的技术路径，突显了其在提高模型学习效率和任务鲁棒性方面的重要价值。该项目可在 https://github.com/2186cloud/hipnet 获取。

View on arXiv Download PDF AI Translation

cs.CV / 51 / 2606.27862

ScaLe-INR: Scale and Learn Implicit Neural Representations

ScaLe-INR：尺度与学习隐式神经表示

Epakanda, Buwaneka, Ratnayake, Athulya, Thennakoon, Pandula, De Silva, Mario, Ranasinghe, Avishka, Godaliyadda, Roshan, Ekanayake, Parakrama

Abstract

Implicit Neural Representations (INRs) parameterized by multilayer perceptrons excel at modeling continuous signals. However, a key challenge persists as INRs fundamentally suffer from spectral bias and information cross-talk. When a single network attempts to capture multi-scale phenomena, high-frequency weight updates destructively interfere with the underlying low-frequency structural approximation. We introduce Scale and Learn INR (ScaLe-INR), a novel multi-branch architecture that resolves these limitations by explicitly matching the signal's frequency spectrum with the optimal operating region of the INR. Drawing upon the Fourier inverse scaling theorem we demonstrate that applying directional coordinate scaling expands a network's representational bandwidth along specific spatial axes. To mathematically enforce functional disentanglement and minimize task-specific information leakage between branches, we propose a Directional Edge Guidance Loss, a spatially-conditioned sparsity prior derived from ground-truth gradients. By constraining the high-frequency branches to act as strict, localized edge-filters, ScaLe-INR eliminates spectral cross-talk, accelerates convergence, and achieves high-fidelity signal reconstruction on complex multi-scale topologies. We evaluate ScaLe-INR across diverse reconstruction and inverse tasks, demonstrating substantial performance gains over existing state-of-the-art (SOTA) methods. The proposed architecture improves upon the nearest baselines by +5.16 dB in image reconstruction and +0.65 dB in image denoising. Furthermore, it achieve an impressive figure of 50.02 dB on audio reconstruction and 0.999 IOU(Intersection Over Union) on 3D reconstruction which beats the all SOTA models.

Chinese Translation

隐式神经表示（INRs）由多层感知器参数化，在建模连续信号方面表现出色。然而，一个关键挑战依然存在，因为INRs在本质上受到谱偏差和信息串扰的影响。当单个网络试图捕捉多尺度现象时，高频权重更新会对底层低频结构近似产生破坏性干扰。我们提出了尺度与学习INR（ScaLe-INR），一种新颖的多分支架构，通过明确匹配信号的频谱与INR的最佳操作区域来解决这些局限性。基于傅里叶逆缩放定理，我们证明了应用方向性坐标缩放可以沿特定空间轴扩展网络的表示带宽。为了在数学上强制功能解耦并最小化分支之间的任务特定信息泄漏，我们提出了一种方向性边缘引导损失（Directional Edge Guidance Loss），这是一种基于真实梯度推导的空间条件稀疏先验。通过限制高频分支作为严格的局部边缘滤波器，ScaLe-INR消除了谱串扰，加速了收敛，并在复杂的多尺度拓扑上实现了高保真信号重建。我们在多种重建和逆任务中评估了ScaLe-INR，展示了其相较于现有最先进（SOTA）方法的显著性能提升。所提架构在图像重建中比最近的基线提高了+5.16 dB，在图像去噪中提高了+0.65 dB。此外，在音频重建中取得了令人印象深刻的50.02 dB，在3D重建中实现了0.999的IOU（交并比），超越了所有SOTA模型。

View on arXiv Download PDF AI Translation

cs.CV / 52 / 2606.27864

A Unified Framework for Vision Transformers Equivariant to Discrete Subgroups of $\mathrm{O}(2)$

一个统一框架的视觉变换器，对离散子群 $ ext{O}(2)$ 具有等变性

Ông, T\=ıkun, Bökman, Georg

Abstract

Vision transformers have become a dominant architecture for visual recognition. However, standard models do not explicitly encode the planar symmetries that arise in many vision domains. We introduce a family of vision transformers equivariant to arbitrary discrete subgroups of $\mathrm{O}(2)$, providing a unified framework that generalizes prior flipping- and $D_4$-equivariant transformer architectures. Our construction yields equivariant analogues of the core transformer components, together with expressivity guarantees for the resulting layers. In particular, we show that whenever $H \le G$, the class of $G$-equivariant ViTs embeds naturally into the class of $H$-equivariant ViTs. We also prove that, in the single-head setting, the corresponding equivariant self-attention layer realizes every $G$-equivariant self-attention map representable by ordinary self-attention. We further construct a $D_6$-equivariant model based on hexagonal patches, making the architecture compatible with six-fold rotational symmetries. We evaluate the resulting models on the PatternNet aerial image dataset in artificially data-scarce regimes across subgroups of $D_4$ and $D_6$. Our experiments compare two equivariant attention mechanisms and analyze how the choice of homogeneous-space configurations used in the nonlinearities affects performance. Preliminary results under matched parameter budgets indicate that equivariance can improve recognition accuracy, motivating further study of how discrete symmetry groups shape transformer-based visual recognition models.

Chinese Translation

视觉变换器已成为视觉识别的主导架构。然而，标准模型并未明确编码在许多视觉领域中出现的平面对称性。我们引入了一类对任意离散子群 $ ext{O}(2)$ 等变的视觉变换器，提供了一个统一框架，推广了先前的翻转和 $D_4$ 等变变换器架构。我们的构造产生了核心变换器组件的等变类比，并为生成的层提供了表达能力保证。特别地，我们展示了每当 $H le G$ 时，$G$-等变的 ViTs 类自然嵌入到 $H$-等变的 ViTs 类中。我们还证明，在单头设置中，相应的等变自注意力层实现了由普通自注意力可表示的每个 $G$-等变自注意力映射。我们进一步基于六边形补丁构建了一个 $D_6$-等变模型，使得该架构与六重旋转对称性兼容。我们在 PatternNet 航空图像数据集上评估了生成的模型，在 $D_4$ 和 $D_6$ 的离散子群中进行人工数据稀缺的实验。我们的实验比较了两种等变注意力机制，并分析了在非线性中使用的均匀空间配置的选择如何影响性能。在匹配参数预算下的初步结果表明，等变性可以提高识别准确性，激励进一步研究离散对称群如何塑造基于变换器的视觉识别模型。

View on arXiv Download PDF AI Translation

cs.CV / 53 / 2606.27876

SpatialUAV: Benchmarking Spatial Intelligence for Low-Altitude UAV Perception, Collaboration, and Motion

SpatialUAV：低空无人机感知、协作与运动的空间智能基准测试

Zhang, Haoyu, Liu, Meng, Xiang, Qianlong, Wang, Kun, Wang, Yaowei, Nie, Liqiang

Abstract

Spatial intelligence is essential for low-altitude unmanned aerial vehicle (UAV) perception, collaboration, and navigation. However, existing UAV benchmarks often emphasize image-level recognition, single-view understanding, or narrow answer formats, leaving 3D spatial inference, multi-view collaboration, scene dynamics, and diverse task formulations insufficiently evaluated. To address these gaps, we introduce SpatialUAV, a real low-altitude UAV benchmark comprising 4,331 curated instances across 14 fine-grained task types, covering semantic discrimination, spatial relation, aerial--aerial collaboration, aerial--ground collaboration, and motion understanding. SpatialUAV organizes all samples into a unified visual-input--question--answer schema, while supporting seven input configurations and nine answer formats, including option labels, region identifiers, geometric values, cross-view correspondences, and free-form motion descriptions. To ensure reliable and grounded evaluation, our data construction pipeline integrates detector-assisted regions, depth supervision, metadata-derived rules, extensive manual annotation, blind filtering, and multi-turn human validation, together with task-specific metrics for heterogeneous outputs. Evaluating representative vision-language models across three categories, we show that current models remain far from human-level performance, with pronounced bottlenecks in cross-view association, structured grounding, geometric reasoning, and temporal viewpoint understanding. These results offer empirical guidance for advancing low-altitude UAV spatial intelligence. Code and data are available at https://github.com/Hyu-Zhang/SpatialUAV.

Chinese Translation

空间智能对于低空无人机（UAV）的感知、协作和导航至关重要。然而，现有的无人机基准测试往往强调图像级识别、单视角理解或狭窄的答案格式，导致3D空间推理、多视角协作、场景动态以及多样化任务形式的评估不足。为了解决这些问题，我们引入了SpatialUAV，一个真实的低空无人机基准测试，包含4,331个精心策划的实例，涵盖14种细粒度任务类型，包括语义区分、空间关系、空中-空中协作、空中-地面协作和运动理解。SpatialUAV将所有样本组织为统一的视觉输入-问题-答案模式，同时支持七种输入配置和九种答案格式，包括选项标签、区域标识符、几何值、跨视角对应关系和自由形式的运动描述。为了确保可靠且有依据的评估，我们的数据构建流程整合了检测器辅助区域、深度监督、元数据派生规则、大量人工标注、盲过滤和多轮人工验证，以及针对异构输出的任务特定指标。在三个类别中评估代表性的视觉-语言模型，我们发现当前模型的性能远未达到人类水平，在跨视角关联、结构化基础、几何推理和时间视角理解等方面存在明显瓶颈。这些结果为推动低空无人机空间智能的发展提供了经验指导。代码和数据可在 https://github.com/Hyu-Zhang/SpatialUAV 获取。

View on arXiv Download PDF AI Translation

cs.CV / 54 / 2606.27880

OrthoTryOn: Geometric Orthogonalization for Conflict-Free Unified Fashion Generation

OrthoTryOn：用于无冲突统一时尚生成的几何正交化

Yang, Zhaotong, Tai, Ying, Zhan, Jiahui, Zheng, Yu, Qian, Jianjun, Yang, Jian

Abstract

Unified fashion generation integrates tasks like virtual try-on and garment reconstruction into a single model to reduce task-specific adaptation costs. However, naive parameter sharing across semantically distinct tasks induces negative transfer through severe inter-task gradient conflict. We propose OrthoTryOn, a unified framework mitigating this interference within a shared Low-Rank Adaptation (LoRA) module. Its Orthogonal Subspace Projection (OSP) applies task-specific orthogonal rotations to bottleneck features, mapping them into decorrelated coordinate frames. To address residual semantic coupling at inference time, we further propose Fisher-guided Negative Guidance (FNG), a parameter-free strategy that utilizes diagonal Fisher information to quantify inter-task sensitivity overlap and explicitly repels generation trajectories from the most confusable task via Classifier-Free Guidance. Extensive experiments demonstrate that OrthoTryOn avoids the severe performance degradation typical of naive unified training and even surpasses independently trained task-specific models, achieving state-of-the-art results across multiple benchmarks while generalizing robustly across diverse diffusion backbones. Code is available at https://github.com/NJU-PCALab/OrthoTryOn.

Chinese Translation

统一时尚生成将虚拟试穿和服装重建等任务整合到一个模型中，以减少特定任务的适应成本。然而，在语义上截然不同的任务之间进行简单的参数共享会通过严重的任务间梯度冲突引发负迁移。我们提出了OrthoTryOn，一个统一框架，通过共享的低秩适应（Low-Rank Adaptation, LoRA）模块来减轻这种干扰。其正交子空间投影（Orthogonal Subspace Projection, OSP）对瓶颈特征应用任务特定的正交旋转，将其映射到去相关的坐标框架中。为了在推理时解决残余的语义耦合问题，我们进一步提出了基于Fisher的负引导（Fisher-guided Negative Guidance, FNG），这是一种无参数策略，利用对角Fisher信息来量化任务间的敏感性重叠，并通过无分类器引导（Classifier-Free Guidance）明确地排斥生成轨迹来自最具混淆性的任务。大量实验表明，OrthoTryOn避免了简单统一训练中典型的严重性能下降，甚至超越了独立训练的任务特定模型，在多个基准上实现了最先进的结果，并在多样的扩散骨干网络中表现出强大的泛化能力。代码可在 https://github.com/NJU-PCALab/OrthoTryOn 获取。

View on arXiv Download PDF AI Translation

cs.CV / 55 / 2606.27897

A Multi-Attribute Latent Space for Visual Analysis of Watches

用于手表视觉分析的多属性潜在空间

Lawonn, Kai, Günther, Tobias, Meuschke, Monique

Abstract

We present a design rationale, embedding model, and interactive visual-analysis system for exploring large wristwatch collections through heterogeneous visual and semantic attributes. The system addresses a common limitation of catalog and e-commerce interfaces: users can filter by metadata, but they receive little support for open-ended exploration of visual similarity, stylistic alternatives, and mixed aesthetic-functional criteria. We therefore represent watches with separate attribute graphs for dial color and dial design, while using watch type as an explicit semantic organizer. Dials are segmented with a U-Net, watch types are predicted with a Vision Transformer, colors are represented through a shared CIELAB reference palette, and dial structure is described with a gradient-based image descriptor. We extend UMAP by combining attribute-specific neighborhood graphs in a unified probabilistic objective and by adding a class-aware layout term that separates global type structure from local visual neighborhoods. The resulting map is exposed in an interactive interface with spatial navigation, metadata filtering, detail inspection, and search-by-example insertion. We evaluate the approach through parameter analysis, runtime measurements, and a qualitative pilot study with watch experts and novices. The results suggest that the system supports discovery and comparison, while also revealing limitations in scalability assessment, search-by-example validation, and the need for broader domain studies. We explicitly discuss these limitations and derive design implications for multi-attribute latent-space visualization across heterogeneous visual collections.

Chinese Translation

我们提出了一种设计理念、嵌入模型和交互式视觉分析系统，用于通过异构视觉和语义属性探索大型手表收藏。该系统解决了目录和电子商务界面的一个常见限制：用户可以通过元数据进行筛选，但在开放式探索视觉相似性、风格替代品和混合美学-功能标准方面支持不足。因此，我们用单独的属性图表示手表的表盘颜色和表盘设计，同时使用手表类型作为明确的语义组织者。表盘通过 U-Net 进行分割，手表类型通过 Vision Transformer 进行预测，颜色通过共享的 CIELAB 参考调色板表示，表盘结构则通过基于梯度的图像描述符进行描述。我们通过在统一的概率目标中结合属性特定的邻域图，并添加一个类感知布局项，将全局类型结构与局部视觉邻域分离，从而扩展了 UMAP。最终生成的地图在一个交互式界面中展示，支持空间导航、元数据过滤、细节检查和示例搜索插入。我们通过参数分析、运行时测量以及与手表专家和新手的定性初步研究来评估该方法。结果表明，该系统支持发现和比较，同时也揭示了在可扩展性评估、示例搜索验证以及对更广泛领域研究需求方面的局限性。我们明确讨论了这些局限性，并推导出针对异构视觉集合中多属性潜在空间可视化的设计启示。

View on arXiv Download PDF AI Translation

cs.CV / 56 / 2606.27900

Long-Term Prediction of Local and Global Human Motion with Occlusion Recovery

局部和全局人类运动的长期预测与遮挡恢复

Yang, Qiaoyue, Heutger, Sven, Niemann, Christopher, Jung, Magnus, Al-Hamadi, Ayoub, Wachsmuth, Sven

Abstract

Human motion describes the three-dimensional full-body movement of a person. Anticipating such motion holds significant relevance across a wide range of application domains such as human-robot interaction, autonomous driving, animation, and healthcare. In recent research, spatial and temporal dependencies are modeled by bidirectional attention mechanisms. These typically anticipate human motion in an autoregressive manner which could cause an accumulation of errors over time. As a consequence, they solely focus on local pose forecasting. To address these limitations, we propose a non-autoregressive transformer based on spatio-temporal attention, and train it not only for local pose anticipation, but also for global motion prediction in space. Furthermore, to enhance its applicability in real-world scenarios, our model is also trained to recover missing joints due to occlusions, and is capable of processing varying lengths of history observations. Our code is publicly available at https://github.com/Q-Y-Yang/Prediction-of-Local-and-Global-Human-Motion.

Chinese Translation

人类运动描述了一个人的三维全身运动。预测这种运动在许多应用领域中具有重要意义，例如人机交互、自动驾驶、动画和医疗保健。在近期的研究中，空间和时间依赖性通过双向注意机制进行建模。这些机制通常以自回归的方式预测人类运动，这可能导致随时间积累的误差。因此，它们仅专注于局部姿态预测。为了解决这些限制，我们提出了一种基于时空注意力的非自回归变换器，并将其训练用于局部姿态预测和空间中的全局运动预测。此外，为了增强其在现实场景中的适用性，我们的模型还被训练以恢复由于遮挡而缺失的关节，并能够处理不同长度的历史观测数据。我们的代码已公开发布在 https://github.com/Q-Y-Yang/Prediction-of-Local-and-Global-Human-Motion。

View on arXiv Download PDF AI Translation

cs.CV / 57 / 2606.27905

There and Back Again: A Flexible-Frame Transformer for Multi-Exposure Fusion

往返之旅：一种用于多曝光融合的灵活框架变换器

Qu, Lishen, Liu, Yao, Zhou, Shihao, Liang, Jie, Zeng, Hui, Zhang, Lei, Yang, Jufeng

Abstract

Multi-exposure fusion (MEF) brings the dynamic range of conventional cameras closer to that of human vision, producing images with rich scene content. Given the large variability in scene luminance, exposure strategies often require different numbers of frames to capture the full radiance range faithfully. However, conventional MEF techniques are typically designed for a fixed number of inputs, forcing deployment systems to maintain separate models for different frame-count requirements, which undermines deployment efficiency. To address this limitation, we propose FreeMEF, the first flexible-frame transformer for MEF that seamlessly accommodates varying numbers of input exposures without retraining or architectural changes. The proposed approach consists of two key modules. First, we introduce a recurrent state space module (RSSM) that sequentially fuses features from arbitrary sequences via adaptive alignment and state-space recurrent modeling, thereby providing global information guidance for the subsequent restoration. Second, we devise a global feature guided block (GFGB) incorporating an extremity-aware hybrid attention (EAHA) and an affine-injection feed-forward network (AFFN), which effectively resolves the similarity paradox while simultaneously optimizing contrast and brightness regulation. Extensive experiments on three benchmark datasets demonstrate the effectiveness of our method, which performs favorably against state-of-the-art methods both quantitatively and qualitatively.

Chinese Translation

多曝光融合（MEF）使传统相机的动态范围更接近人类视觉，生成具有丰富场景内容的图像。由于场景亮度的巨大变异性，曝光策略通常需要不同数量的帧来忠实捕捉完整的辐射范围。然而，传统的MEF技术通常设计为固定数量的输入，这迫使部署系统为不同的帧数需求维护单独的模型，从而降低了部署效率。为了解决这一限制，我们提出了FreeMEF，这是首个灵活帧变换器，能够无缝适应不同数量的输入曝光，而无需重新训练或架构更改。所提出的方法由两个关键模块组成。首先，我们引入了一个递归状态空间模块（RSSM），通过自适应对齐和状态空间递归建模，顺序融合来自任意序列的特征，从而为后续恢复提供全局信息指导。其次，我们设计了一个全局特征引导块（GFGB），结合了极值感知混合注意力（EAHA）和仿射注入前馈网络（AFFN），有效解决了相似性悖论，同时优化对比度和亮度调节。在三个基准数据集上的大量实验表明，我们的方法在定量和定性上均优于最先进的方法，证明了其有效性。

View on arXiv Download PDF AI Translation

cs.CV / 58 / 2606.27918

Every Step of the Way: Video-based Parkinsonian Turning Step Counting

每一步：基于视频的帕金森病转身步数计数

Cheng, Qiushuo, Liu, Jingjing, Morgan, Catherine, Whone, Alan, Mirmehdi, Majid

Abstract

As a prominent symptom of Parkinson's disease (PD), turning impairment is evaluated through parameters such as turning angle, duration, and particularly, the number of steps required to complete a turn, which directly reflects motor dysfunction. Accurate step counting is challenging due to variability in real-world turning movements and atypical shuffling patterns in parkinsonian gait. Existing methods are predominantly wearable-based, requiring users to wear and manage dedicated devices, which can be inconvenient for continuous daily use. To address this, we propose a passive, video-based framework that estimates step count in a coarse-to-fine manner using diverse motion representations. Specifically, an initial step count is estimated from foot movement signals derived from 3D human mesh recovery, providing high-level motion structures. To incorporate fine-grained motion details, a motion encoder learns complementary gait dynamics from mesh and optical flow to refine the initial estimate. In this process, coarse foot movement signals query the pixel-level motion cues via cross attention to capture subtle parkinsonian gait dynamics. To handle varying video lengths, we partition each video into clips and integrate clip-wise motion embeddings via multiple instance learning (MIL) for step count residual prediction. Extensive experiments show our method consistently outperforms existing step counting methods on real-world PD turning datasets.

Chinese Translation

作为帕金森病（PD）的显著症状，转身障碍通过转身角度、持续时间以及特别是完成转身所需的步数等参数进行评估，这直接反映了运动功能障碍。由于现实世界转身动作的变异性以及帕金森病步态中不典型的拖步模式，准确的步数计数面临挑战。现有方法主要基于可穿戴设备，要求用户佩戴和管理专用设备，这在日常持续使用中可能不够方便。为了解决这一问题，我们提出了一种被动的基于视频的框架，该框架通过多样的运动表示以粗到细的方式估计步数。具体而言，初步步数是通过从3D人类网格恢复中提取的脚部运动信号进行估计，从而提供高层次的运动结构。为了融入细粒度的运动细节，运动编码器从网格和光流中学习互补的步态动态，以细化初步估计。在此过程中，粗略的脚部运动信号通过交叉注意力查询像素级运动线索，以捕捉微妙的帕金森病步态动态。为了处理不同长度的视频，我们将每个视频划分为多个片段，并通过多实例学习（MIL）整合片段级运动嵌入以进行步数残差预测。大量实验表明，我们的方法在真实世界的帕金森病转身数据集上始终优于现有的步数计数方法。

View on arXiv Download PDF AI Translation

cs.CV / 59 / 2606.27922

Reflect-R1: Evidence-Driven Reflection for Self-Correction in Long Video Understanding

Reflect-R1：基于证据驱动的长视频理解自我修正反思机制

Chen, Shuimu, Chen, Yuteng, Guan, Yuanshen, Cheng, Zebang, Zhang, Zeyu, Qin, Shengqian, Xia, Bin, Li, Jiaran, Yang, Wenming, Ma, Fei

Abstract

Current multimodal reflection mechanisms for long video understanding predominantly rely on closed-loop self-reflection within internal parameters. Lacking objective external evidence, models are frequently trapped in blind confidence and often fail to correct errors. Furthermore, applying reinforcement learning to multi-stage reflection pipelines introduces severe policy coupling, which is exacerbated by a critical scarcity of dedicated training data. To address these limitations, this work proposes Reflect-R1, the first Evidence-Driven self-correction framework for long video understanding. The framework constructs a three-stage pipeline consisting of intuition, verification, and arbitration. By dynamically retrieving objective visual evidence to verify initial intuitions and autonomously executing multiple temporal searches to resolve conflicts, it completely breaks the hallucination loop. To overcome policy coupling, we design a stage-decoupled reinforcement learning algorithm named SD-GRPO that independently computes advantage functions across different reasoning stages. Concurrently, we construct a dataset of 120K samples to bridge the training data gap. Extensive experiments on benchmarks such as VideoMME and LongVideoBench demonstrate that Reflect-R1 achieves state-of-the-art performance. Our method significantly improves the genuine rectification rate and enables authentic self-correction strictly grounded in objective evidence.

Chinese Translation

当前针对长视频理解的多模态反思机制主要依赖于内部参数的闭环自我反思。由于缺乏客观的外部证据，模型常常陷入盲目的自信，难以纠正错误。此外，将强化学习应用于多阶段反思管道会引入严重的策略耦合，而这一问题因专用训练数据的严重匮乏而加剧。为了解决这些局限性，本文提出了Reflect-R1，这是首个基于证据驱动的长视频理解自我修正框架。该框架构建了一个由直觉、验证和仲裁组成的三阶段管道。通过动态检索客观视觉证据来验证初步直觉，并自主执行多次时间搜索以解决冲突，它完全打破了幻觉循环。为克服策略耦合，我们设计了一种名为SD-GRPO的阶段解耦强化学习算法，该算法在不同推理阶段独立计算优势函数。同时，我们构建了一个包含12万样本的数据集，以填补训练数据的空白。在VideoMME和LongVideoBench等基准上的广泛实验表明，Reflect-R1实现了最先进的性能。我们的方法显著提高了真实修正率，并实现了严格基于客观证据的真实自我修正。

View on arXiv Download PDF AI Translation

cs.CV / 60 / 2606.27923

Home3D 1.0: A High-Fidelity Image-to-3D Asset Generation System for Interior Design

Home3D 1.0：用于室内设计的高保真图像到3D资产生成系统

Fei, Yiyun, Li, Guoqiu, Song, Jin, Wu, Chuqiao, Wu, Delong, Wu, Hong, Zeng, Ziru, Chen, Haohui, Kong, YinDong, Li, Jing, Wu, Qi, Zhang, Feng

Abstract

We present Home3D 1.0, a modular image-to-3D generation system that produces high-quality 3D assets from a single reference image, targeting interior design and e-commerce applications. Given a photograph of a furniture or decor item, the system outputs a mesh with physically-based rendering (PBR) materials, and the mesh can be decomposed into material-specific components. The pipeline is organized into four tightly coupled modules: Geometry reconstructs a watertight mesh through latent SDF modelling with a geometry VAE and a coarse-to-fine flow-matching DiT; Texture predicts multiview albedo observations, reprojects them onto the mesh, and completes unseen surface regions with a 3D texture field; Material uses MatWeaver to obtain component masks through video-based segmentation and UV-space voting, then retrieves and bakes PBR maps from a curated material library through hierarchical multi-modal matching; and Parts generates material-editable semantic part meshes with a PartVAE and PartDiT, decoding multi-head part-specific SDF fields in one pass. Each module is evaluated independently with dedicated metrics, highlighting both the current system capability and the remaining gaps toward broader deployment.

Chinese Translation

我们提出了Home3D 1.0，这是一个模块化的图像到3D生成系统，能够从单一参考图像生成高质量的3D资产，主要针对室内设计和电子商务应用。给定一张家具或装饰物品的照片，该系统输出一个具有物理基础渲染（PBR）材料的网格，并且该网格可以分解为特定材料的组件。该流程组织为四个紧密耦合的模块：几何模块通过潜在的SDF建模与几何变分自编码器（VAE）以及粗到细的流匹配DiT重建一个密闭网格；纹理模块预测多视角的反照率观测，将其重新投影到网格上，并使用3D纹理场填补未见的表面区域；材料模块利用MatWeaver通过基于视频的分割和UV空间投票获取组件掩膜，然后通过层次多模态匹配从策划的材料库中检索和烘焙PBR贴图；部件模块使用PartVAE和PartDiT生成可编辑材料的语义部件网格，一次解码多头特定部件的SDF场。每个模块都使用专门的指标进行独立评估，突显了当前系统的能力以及在更广泛应用中仍存在的差距。

View on arXiv Download PDF AI Translation

cs.CV / 61 / 2606.27935

Controllable Histopathology Image Synthesis with Training-free Structural Initialization and Textural Modulation

可控的组织病理图像合成：无训练的结构初始化与纹理调制

Qiu, Yuheng, Luo, Jingyi, Ye, Chenfei, Ma, Ting, Cao, Jianfeng

Abstract

Deep learning has demonstrated remarkable success in high-throughput histopathology image analysis. However, the performance of learning-based models critically depends on the quality and size of annotations by expert pathologists, which is a resource-intensive and time-consuming process. To address the limitations of data scarcity and annotation burden, several methods have been proposed to synthesize paired histopathology data. Nevertheless, these frameworks typically still require annotation data, albeit in reduced quantities, to impose structural constraints during training. In this work, we present CHIS, a plug-in framework that guides the sampling trajectory of a pretrained diffusion model through two key stages: structural initialization at the start and textural modulation during generation. The initial noise state is refined by fusing the phase information from a prior mask with the amplitude of Gaussian noise in the frequency domain, yielding a structurally informed starting point. During the reverse diffusion process, we adaptively modulate both coarse-grained and fine-grained textures at different wavelet decomposition levels. This enables a diffusion model pretrained solely on unlabeled images to generate outputs that align with prior structural masks while preserving the reference tissue style. We conducted extensive experiments demonstrating the superiority of CHIS in generation fidelity and its substantial benefits for downstream segmentation tasks. Code is available at https://github.com/IBIL-Code/CHIS.

Chinese Translation

深度学习在高通量组织病理图像分析中展现了显著的成功。然而，基于学习的模型性能严重依赖于专家病理学家的注释质量和数量，这一过程资源密集且耗时。为了解决数据稀缺和注释负担的限制，已经提出了几种合成配对组织病理数据的方法。然而，这些框架通常仍然需要注释数据，尽管数量有所减少，以在训练过程中施加结构约束。在本研究中，我们提出了CHIS，一个插件框架，通过两个关键阶段引导预训练扩散模型的采样轨迹：开始时的结构初始化和生成过程中的纹理调制。初始噪声状态通过将先前掩膜的相位信息与频域中的高斯噪声幅度融合来精炼，从而产生一个结构信息丰富的起始点。在反向扩散过程中，我们在不同的小波分解层次上自适应地调制粗粒度和细粒度的纹理。这使得仅在未标记图像上预训练的扩散模型能够生成与先前结构掩膜一致的输出，同时保持参考组织风格。我们进行了广泛的实验，证明了CHIS在生成保真度方面的优越性及其对下游分割任务的显著益处。代码可在 https://github.com/IBIL-Code/CHIS 获取。

View on arXiv Download PDF AI Translation

cs.CV / 62 / 2606.27947

Understanding How MLLMs Describe Artworks Using Token Activation Maps

理解多模态大型语言模型如何描述艺术作品的标记激活图

Fanelli, Nicola, De Marinis, Pasquale, Scaringi, Raffaele, Cetinic, Eva, Vessio, Gennaro, Castellano, Giovanna

Abstract

Multimodal Large Language Models (MLLMs) describe artworks with remarkable fluency, yet the visual reasoning behind their outputs remains opaque. When an MLLM names a style, identifies a subject, or recognizes an iconographic symbol, does it ground each claim in the relevant region of the canvas, draw on an undifferentiated visual signal, or rely primarily on textual priors? We study this using the Token Activation Map (TAM), which produces, for each generated token, a heatmap isolating the visual evidence specific to that token from prior-context interference. Applying TAM to a curated set of paintings spanning multiple periods and genres, we analyze grounding patterns across five semantically distinct token categories: common visual objects, style descriptors, metadata, iconographic tokens, and affective expressions. We find that visual grounding varies substantially with token semantics. We further show that MLLMs attempt to identify artworks and artists, achieving higher accuracy in artist attribution than in title prediction, where hallucinations are more frequent. Finally, we compare TAM with SAM~3 open-vocabulary segmentation. To ensure reproducibility, we release our code, experimental configurations, prompts, and qualitative results on the project page at https://nicolafan.github.io/tamart/.

Chinese Translation

多模态大型语言模型（MLLMs）以显著的流畅性描述艺术作品，但其输出背后的视觉推理仍然不透明。当一个MLLM命名一种风格、识别一个主题或识别一个图像符号时，它是否将每个主张与画布的相关区域联系起来，还是依赖于未区分的视觉信号，或者主要依赖于文本先验？我们使用标记激活图（Token Activation Map, TAM）来研究这一问题，该图为每个生成的标记生成一个热图，隔离出与该标记相关的视觉证据，排除先前上下文的干扰。通过将TAM应用于跨越多个时期和流派的策划绘画集，我们分析了五个语义上不同的标记类别的基础模式：常见视觉对象、风格描述符、元数据、图像标记和情感表达。我们发现视觉基础在标记语义上有显著差异。我们进一步表明，MLLMs试图识别艺术作品和艺术家，在艺术家归属方面的准确性高于标题预测，而后者的幻觉现象更为频繁。最后，我们将TAM与SAM~3开放词汇分割进行比较。为了确保可重复性，我们在项目页面（https://nicolafan.github.io/tamart/）上发布了我们的代码、实验配置、提示和定性结果。

View on arXiv Download PDF AI Translation

cs.CV / 63 / 2606.27964

Directing the World: Fast Autoregressive Video Generation with Compositional Human-Camera Control

引导世界：具有组合人机控制的快速自回归视频生成

Wang, Haoyuan, Chen, Yabo, Huang, Haibin, Zhang, Chi, Li, Xuelong

Abstract

Building interactive world models requires generating realistic videos while maintaining controllable dynamics over long horizons. Autoregressive video generation offers a scalable foundation, but suffers from error accumulation and temporal degradation during extended rollouts. This issue is further amplified under heterogeneous controls such as human motion and camera trajectories, which may interfere and destabilize a pretrained video prior, while existing methods often trade off controllability and visual quality. We propose "Directing the World", a fast autoregressive framework for controllable world-model video generation with compositional human-motion and camera-trajectory control. Our key idea is to decouple control learning while preserving a unified autoregressive video prior. We introduce a Fast-Slow Memory training strategy to stabilize long-horizon rollout learning and improve convergence. For human motion control, we design a t-guided Dynamic Projection mechanism and a refined Motion-CFG strategy, enabling temporally smooth and accurate motion alignment without degrading visual fidelity, and supporting multi-person control.After learning a robust motion prior, we introduce a second-stage camera-trajectory control module to compose human dynamics with viewpoint changes for coherent world exploration. We further construct a large-scale dataset with synchronized video, text, human-motion, and camera-trajectory annotations, organized into motion-centric and camera-centric subsets for decoupled training. Extensive experiments show stable long-horizon generation with precise controllability and high visual quality. See more at https://whydahuzi.github.io/Directing-the-World.github.io/.

Chinese Translation

构建互动世界模型需要生成逼真的视频，同时在长时间范围内保持可控的动态。自回归视频生成提供了一个可扩展的基础，但在延长的生成过程中会遭遇误差积累和时间降解的问题。在异构控制（如人类运动和相机轨迹）下，这一问题进一步加剧，因为这些控制可能会干扰并使预训练的视频先验不稳定，而现有方法往往在可控性和视觉质量之间进行权衡。我们提出了“引导世界”这一快速自回归框架，用于具有组合人类运动和相机轨迹控制的可控世界模型视频生成。我们的关键思想是解耦控制学习，同时保持统一的自回归视频先验。我们引入了一种快速-慢速记忆训练策略，以稳定长时间范围的生成学习并改善收敛性。对于人类运动控制，我们设计了一种 t 引导的动态投影机制和一种改进的运动-条件生成框架（Motion-CFG）策略，使得运动对齐在时间上平滑且准确，同时不降低视觉保真度，并支持多人控制。在学习到稳健的运动先验后，我们引入了第二阶段的相机轨迹控制模块，以将人类动态与视点变化组合，从而实现连贯的世界探索。我们进一步构建了一个大规模数据集，包含同步的视频、文本、人类运动和相机轨迹注释，组织成以运动为中心和以相机为中心的子集，以便于解耦训练。大量实验表明，该方法在长时间范围内生成稳定，具有精确的可控性和高视觉质量。更多信息请访问 https://whydahuzi.github.io/Directing-the-World.github.io/。

View on arXiv Download PDF AI Translation

cs.CV / 64 / 2606.27974

ProMSA:Progressive Multimodal Search Agents for Knowledge-Based Visual Question Answering

ProMSA：用于基于知识的视觉问答的渐进式多模态搜索代理

Wu, ZhengXian, Xu, Hangrui, Shi, Kai, Chen, Zhuohong, Yu, Yunyao, Zhang, Chuanrui, Liao, Zirui, Yang, Jun, Yang, Zhenyu, Lu, Haonan, Wang, Haoqian

Abstract

Knowledge-based Visual Question Answering (KB-VQA) requires models to combine image understanding with external knowledge. Most prior methods use a fixed retrieve-then-generate pipeline with a pre-selected retriever and a static top-k setting, which is not adaptive during reasoning. We propose ProMSA, a progressive multimodal search agent for KB-VQA. Given an image-question pair, the agent iteratively chooses image search, text search, or stop, under explicit tool-call budgets and with deduplication to avoid redundant retrieval. For training, we first use rejection-sampling SFT to learn valid tool-use formats, then optimize the agent with TN-GSPO, a sequence-level RL objective that normalizes updates by both generation length and tool-interaction depth. Experiments on E-VQA and InfoSeek show consistent gains over strong RAG and agent baselines, and improved retrieval and end-to-end accuracy. The code is available at https://github.com/DingWu1021/Promsa.

Chinese Translation

基于知识的视觉问答（KB-VQA）要求模型将图像理解与外部知识相结合。大多数先前的方法使用固定的检索-生成管道，配备预选的检索器和静态的前k设置，这在推理过程中并不具备自适应性。我们提出了ProMSA，一种用于KB-VQA的渐进式多模态搜索代理。给定一个图像-问题对，代理在明确的工具调用预算下，迭代选择图像搜索、文本搜索或停止，并进行去重以避免冗余检索。在训练过程中，我们首先使用拒绝采样的SFT学习有效的工具使用格式，然后通过TN-GSPO优化代理，这是一种序列级的强化学习目标，通过生成长度和工具交互深度来规范化更新。在E-VQA和InfoSeek上的实验表明，相较于强大的RAG和代理基线，ProMSA在检索和端到端准确性上均有一致的提升。代码可在https://github.com/DingWu1021/Promsa获取。

View on arXiv Download PDF AI Translation

cs.CV / 65 / 2606.27978

Parallel Rollout Approximation for Pixel-Space Autoregressive Image Generation

像素空间自回归图像生成的并行展开近似

Xu, Jiayi, He, Di, Ke, Guolin

Abstract

Pixel-space continuous-token autoregressive (AR) generation directly models images as sequences of raw pixel patches, avoiding discrete tokenization or a separately pretrained tokenizer. However, it faces coupled challenges: high-dimensional patch generation causes large single-step errors, and teacher-forced training creates a train--inference gap that makes these errors accumulate across AR steps. Existing fixes such as $x$-prediction and input noise injection only partially mitigate these issues. Exact rollout training better matches inference-time conditions, but is impractical due to prohibitively slow sequential sampling. We propose \emph{Parallel Rollout Approximation} (PRA), a scalable framework that addresses both challenges jointly. PRA generates low-dimensional intermediate states instead of high-dimensional pixel patches, then maps them back to pixel-space tokens with a pixel decoder, preserving a pixel-in, pixel-out AR interface. It also constructs inference-like pixel inputs through the same intermediate-state-to-pixel path used at inference, independently across positions, approximating the pixel-feedback interface encountered during inference-time rollout while retaining parallel teacher-forced training. On class-conditional ImageNet-1K generation at $256\times256$ resolution, PRA-S with 135M parameters achieves an FID of 2.58, surpassing the previous billion-scale pixel-space AR result of 3.60. Scaling to PRA-L with 511M parameters further improves FID to 1.94, establishing a new state of the art among pixel-space AR models. Beyond generation, PRA achieves higher ImageNet classification probing accuracy than other AR and diffusion baselines, suggesting its potential for unified pixel-space image generation and understanding.

Chinese Translation

像素空间连续标记自回归（AR）生成直接将图像建模为原始像素块的序列，避免了离散标记化或单独预训练的标记器。然而，它面临着耦合的挑战：高维块生成导致大的单步误差，而教师强制训练则产生了训练与推理之间的差距，使这些误差在AR步骤中累积。现有的修复方法，如$x$-预测和输入噪声注入，仅部分缓解了这些问题。精确的展开训练更好地匹配推理时的条件，但由于序列采样速度过慢而不切实际。我们提出了 extit{并行展开近似}（PRA），这是一个可扩展的框架，能够共同解决这两个挑战。PRA生成低维中间状态，而不是高维像素块，然后通过像素解码器将其映射回像素空间标记，保持了像素输入和像素输出的AR接口。它还通过在推理时使用的相同中间状态到像素的路径独立地构建类似推理的像素输入，从而近似推理时展开过程中遇到的像素反馈接口，同时保留并行的教师强制训练。在$256 imes256$分辨率的类条件ImageNet-1K生成中，具有1.35亿参数的PRA-S实现了2.58的FID，超越了之前十亿规模像素空间AR结果的3.60。扩展到具有5.11亿参数的PRA-L进一步将FID提高至1.94，确立了像素空间AR模型中的新最优表现。除了生成之外，PRA在ImageNet分类探测准确性上也超过了其他AR和扩散基线，表明其在统一像素空间图像生成和理解方面的潜力。

View on arXiv Download PDF AI Translation

cs.CV / 66 / 2606.27988

Latent Visual Diffusion Reasoning with Monte Carlo Tree Search

基于蒙特卡洛树搜索的潜在视觉扩散推理

Teng, Xirui, Xi, Nan, Yuan, Junsong

Abstract

Analyzing fine-grained skill activities (e.g., sports, surgery) requires not only recognizing visual patterns but also performing step-by-step visual reasoning that leads to the final judgment. While recent advances in action quality assessment have achieved remarkable progress in evaluating performance, existing models remain black boxes, where they lack the ability to explicitly reveal the reasoning processes underlying their judgments. To address this limitation, we propose Latent Visual Diffusion Reasoning (LVDR), a novel framework that integrates keypoint-guided Monte Carlo Tree Search (MCTS) to model and visualize the latent visual reasoning process. LVDR not only produces more accurate skill assessments but also uncovers the critical visual reasoning sequences that contribute to the final evaluation. Extensive experiments across four datasets spanning diverse sports and surgical domains demonstrate that LVDR achieves competitive quantitative performance while providing interpretable visual reasoning trajectories leading to the final predictions. Source codes and models can be found through the following link: https://github.com/XiruiTeng/LVDR_Official.git.

Chinese Translation

分析细粒度技能活动（例如，体育、外科手术）不仅需要识别视觉模式，还需要进行逐步的视觉推理，以得出最终判断。尽管近期在动作质量评估方面取得了显著进展，但现有模型仍然是黑箱，缺乏明确揭示其判断背后推理过程的能力。为了解决这一局限性，我们提出了潜在视觉扩散推理（Latent Visual Diffusion Reasoning, LVDR），这是一个新颖的框架，结合了关键点引导的蒙特卡洛树搜索（Monte Carlo Tree Search, MCTS），以建模和可视化潜在的视觉推理过程。LVDR不仅能够产生更准确的技能评估，还能揭示对最终评估起关键作用的视觉推理序列。在涵盖多种体育和外科领域的四个数据集上进行的广泛实验表明，LVDR在提供可解释的视觉推理轨迹的同时，达到了具有竞争力的定量性能。源代码和模型可以通过以下链接找到：https://github.com/XiruiTeng/LVDR_Official.git。

View on arXiv Download PDF AI Translation

cs.CV / 67 / 2606.27999

HumanMoveVQA: Can Video MLLMs reason about human movement in videos?

HumanMoveVQA：视频中的多模态大型语言模型能否推理人类运动？

Gera, Pulkit, Sardari, Faegheh, Nadeem, Asmar, Bono, Valentina, Boulton, Padraig, Hilton, Adrian, Mustafa, Armin

Abstract

Despite the rapid advance of Multimodal Large Language Models (MLLMs) in high-level video understanding, a fundamental bottleneck remains: these models collapse complex human motion into coarse semantic labels. Existing benchmarks mostly focus on scene-centric events or local joint articulations, failing to probe global human motion in space over time (trajectory and orientation changes). We introduce HumanMoveVQA, the first comprehensive benchmark designed to evaluate global trajectory and orientation reasoning from an exocentric perspective. Our benchmark utilizes a first-frame anchored world coordinate system, preserving translation and rotation relative to a fixed starting point. We propose a scalable, multi-stage pipeline that lifts 2D video observations into world-consistent 3D motion tracks to generate over 10K structured question-answer pairs across seven reasoning categories, including motion aggregation, sequential ordering, and trajectory-level inference. Our extensive evaluation reveals a critical capability gap in state-of-the-art proprietary models on deep human motion understanding. However, we demonstrate that this is a learnable problem; by fine-tuning an open-source baseline with our targeted, world-consistent supervision, we achieve a significant improvement.HumanMoveVQA establishes a rigorous geometric foundation for developing next-generation, movement-aware video understanding models.

Chinese Translation

尽管多模态大型语言模型（MLLMs）在高层次视频理解方面迅速发展，但仍存在一个根本瓶颈：这些模型将复杂的人类运动简化为粗略的语义标签。现有基准主要集中于场景中心事件或局部关节运动，未能探讨空间随时间变化的全球人类运动（轨迹和方向变化）。我们提出了HumanMoveVQA，这是第一个旨在从外部视角评估全球轨迹和方向推理的综合基准。我们的基准利用以第一帧为锚的世界坐标系统，保持相对于固定起点的平移和旋转。我们提出了一种可扩展的多阶段流程，将2D视频观察提升为世界一致的3D运动轨迹，以生成超过10,000个结构化问答对，涵盖运动聚合、顺序排序和轨迹级推理等七个推理类别。我们的广泛评估揭示了当前最先进的专有模型在深度人类运动理解方面存在的关键能力差距。然而，我们证明这是一个可学习的问题；通过用我们针对性的、世界一致的监督对开源基线进行微调，我们实现了显著的改进。HumanMoveVQA为开发下一代运动感知视频理解模型奠定了严格的几何基础。

View on arXiv Download PDF AI Translation

cs.CV / 68 / 2606.28012

Curriculum-guided Change Detection Training: Toward Accurate Serac Fall Monitoring

课程引导的变化检测训练：朝向准确的冰川崩塌监测

Dérédel, Arthur, Crispim-Junior, Carlos, Lemaire, Pierre, Berthet, Johan, Rodet, Laure Tougne

Abstract

Change Detection (CD) aims to identify semantic or structural changes from nearly registered multi-temporal images. While recent advances in training methodologies have largely focused on semi-supervised learning and consistency regularization, alternative training paradigms remain underexplored. In particular, most deep CD methods rely on uniform sampling during training, implicitly assuming that all training samples contribute equally to the optimization process. However, such naive sampling can introduce noisy gradients and hinder robust representation learning. To address this limitation, we propose a curriculum learning framework tailored for change detection. Our approach investigates two complementary difficulty measures: the Solar Angular Gap (SAG), a physically grounded proxy for acquisition-condition variability, and the Structural Similarity Index Measure (SSIM), which evaluates appearance similarity between image pairs. Based on these criteria, the framework progressively introduces challenging samples during training, enabling models to learn robust representations in a coarse-to-fine manner. We evaluate our method on the challenging SeracFallDet benchmark, where results demonstrate consistent improvements of the proposed approach over standard uniform-sampling strategies for both pixel-based and object-based approaches. These results highlight the potential of curriculum learning to improve robustness in deep change detection. Importantly, our training framework is orthogonal to existing CD architectures, making it readily applicable to a broad range of methods.

Chinese Translation

变化检测（Change Detection, CD）旨在从几乎注册的多时相图像中识别语义或结构变化。尽管近期在训练方法上的进展主要集中于半监督学习和一致性正则化，但其他训练范式仍然未被充分探索。特别是，大多数深度变化检测方法在训练过程中依赖于均匀采样，隐含地假设所有训练样本对优化过程的贡献是相等的。然而，这种简单的采样方式可能引入噪声梯度，从而妨碍稳健的表征学习。为了解决这一局限性，我们提出了一种针对变化检测的课程学习框架。我们的方法研究了两种互补的难度度量：太阳角间隙（Solar Angular Gap, SAG），这是一个基于物理的获取条件变异性的代理，以及结构相似性指数度量（Structural Similarity Index Measure, SSIM），用于评估图像对之间的外观相似性。基于这些标准，该框架在训练过程中逐步引入具有挑战性的样本，使模型能够以粗到细的方式学习稳健的表征。我们在具有挑战性的SeracFallDet基准上评估了我们的方法，结果表明所提方法在基于像素和基于对象的方法上均优于标准均匀采样策略。这些结果突显了课程学习在提高深度变化检测稳健性方面的潜力。重要的是，我们的训练框架与现有的变化检测架构是正交的，使其能够广泛应用于多种方法。

View on arXiv Download PDF AI Translation

cs.CV / 69 / 2606.28016

TempAct: Advancing Temporal Plausibility in Autoregressive Video Generation via Planner-Executor RL

TempAct：通过规划者-执行者强化学习推进自回归视频生成中的时间合理性

Wang, Jing, Zhou, Xiangxin, Liang, Jiajun, Liu, Kaiqi, Pang, Wanyun, Xie, Zhenyu, Pang, Tianyu, Liang, Xiaodan

Abstract

Autoregressive (AR) video diffusion models enable low-latency streaming generation by synthesizing videos chunk by chunk with cached visual context, but this chunk-wise formulation makes temporal instruction following ambiguous. A single global prompt does not specify which sub-event should be realized in each chunk, while naively switching to step-wise prompts often leads to delayed reactions, blended step semantics, and error propagation across prompt transitions. These failures are difficult to address with supervised fine-tuning or distillation alone: SFT suffers from exposure bias, while rollout-based distillation still optimizes low-level denoising or teacher-distribution matching rather than directly enforcing action ordering and prompt-transition correctness. We address these challenges with TempAct, a planner--executor reinforcement learning framework that jointly optimizes temporal decomposition and step-conditioned execution for temporally plausible AR video generation. TempAct uses an LLM planner to explore span-aware step prompts that are executable by the video model, and trains an AR diffusion executor to follow these prompts under its own generated histories. Its key mechanism is hierarchical group exploration: candidate plans form planning groups, and each plan induces an execution group of multiple continuations from a shared visual context, enabling plan-level credit assignment for long-horizon temporal outcomes and executor-level credit assignment for prompt-switch behavior. We further design hierarchical rewards that combine plan-quality and full-video temporal feedback for the planner with local transition-level step-following rewards, aesthetic regularization, and KL constraints for the executor. Experiments on Self-Forcing and LongLive show that TempAct improves temporal consistency while preserving overall visual quality.

Chinese Translation

自回归（AR）视频扩散模型通过逐块合成视频并利用缓存的视觉上下文，实现低延迟流式生成，但这种逐块的形式使得时间指令的跟随变得模糊。单一的全局提示并未明确规定每个块中应实现哪个子事件，而简单地切换到逐步提示往往会导致反应延迟、步骤语义混合以及在提示转换中的错误传播。这些问题仅通过监督微调或蒸馏很难解决：监督微调（SFT）受到曝光偏差的影响，而基于回滚的蒸馏仍然优化低级去噪或教师分布匹配，而不是直接强制执行动作顺序和提示转换的正确性。我们通过TempAct解决这些挑战，这是一种规划者-执行者强化学习框架，联合优化时间分解和步骤条件执行，以实现时间合理的自回归视频生成。TempAct使用大型语言模型（LLM）规划者探索可由视频模型执行的跨度感知步骤提示，并训练自回归扩散执行者在其自身生成的历史下遵循这些提示。其关键机制是层次化组探索：候选计划形成规划组，每个计划诱导出多个来自共享视觉上下文的延续执行组，使得能够对长期时间结果进行计划级的信用分配，并对提示切换行为进行执行者级的信用分配。我们进一步设计了层次化奖励，将计划质量和完整视频时间反馈结合起来，为规划者提供局部转换级的步骤跟随奖励、美学正则化和KL约束。Self-Forcing和LongLive上的实验表明，TempAct在保持整体视觉质量的同时，提高了时间一致性。

View on arXiv Download PDF AI Translation

cs.CV / 70 / 2606.28026

EMOSH: Expressive Motion and Shape Disentanglement for Human Animation

EMOSH：用于人类动画的表现性运动与形状解耦

Zhang, Dongbin, Liu, Hao, Dai, Binquan, Chen, Kangjie, Wang, Chuming, Li, Chen, Lyu, Jing, Wang, Haoqian

Abstract

High-fidelity and expressive controllable human animation is essential for content creation and digital avatar applications. However, existing methods face a dilemma between expressiveness and disentanglement. Mainstream 2D pose-conditioned approaches suffer from "motion-shape entanglement", leading to the leakage of the driving subject's body shape. Conversely, methods relying on 3D priors (e.g., SMPL) achieve geometric disentanglement but struggle to capture facial expressions and complex gestures, resulting in rigid animations. To this end, we propose EMOSH, a novel framework for high-fidelity controllable human video generation. First, an Expressive Human Model (EHM) is introduced as the core control representation. By explicitly disentangling shape and pose parameters, we fundamentally resolve the body shape leakage issue. Alongside this, a robust motion tracker is designed to accurately estimate EHM parameters from video. Second, we propose a Coarse-to-Fine Hybrid Motion Injection strategy, enabling more fine-grained control over expressions and gestures. Furthermore, we introduce a Spatially-Aligned Conditioning mechanism to bridge the domain gap between training and inference, improving identity consistency. Extensive experiments demonstrate that EMOSH outperforms previous methods in both self-driven and cross-driven scenarios, producing high-fidelity videos with vivid expressions while maintaining shape disentanglement.

Chinese Translation

高保真且可控的表现性人类动画对于内容创作和数字化身应用至关重要。然而，现有方法在表现性与解耦之间面临困境。主流的基于2D姿态的条件方法遭受“运动-形状纠缠”的问题，导致驱动主体的身体形状泄漏。相反，依赖于3D先验（例如，SMPL）的方法能够实现几何解耦，但在捕捉面部表情和复杂手势方面存在困难，导致动画僵硬。为此，我们提出了EMOSH，一个用于高保真可控人类视频生成的新框架。首先，引入表现性人类模型（EHM）作为核心控制表示。通过明确解耦形状和姿态参数，我们从根本上解决了身体形状泄漏的问题。同时，设计了一种强大的运动追踪器，以准确估计视频中的EHM参数。其次，我们提出了一种粗到细的混合运动注入策略，使得对表情和手势的控制更加细致。此外，我们引入了一种空间对齐条件机制，以弥合训练与推理之间的领域差距，提高身份一致性。大量实验表明，EMOSH在自驱动和交叉驱动场景中均优于之前的方法，生成高保真的视频，展现生动的表情，同时保持形状解耦。

View on arXiv Download PDF AI Translation

cs.CV / 71 / 2606.28039

Mind the Gap: Quantifying the Domain Gap in Cross-Sensor Diffusion Super-Resolution

注意差距：量化跨传感器扩散超分辨率中的领域差距

Kopeć, Dawid, Jabłońska, Katarzyna, Kozłowski, Wojciech, Zięba, Maciej

Abstract

Demand for high-resolution satellite imagery has increased interest in super-resolution (SR) to bridge the spatial resolution gap between freely available missions such as Sentinel-2 and commercial systems like PlanetScope. Because no sensor provides true paired low- and high-resolution observations, SR models are usually trained on synthetically degraded data, creating a domain gap on real cross-sensor imagery. In this work, we provide the first systematic study of how this synthetic-to-real mismatch affects the performance of modern diffusion-based SR models. Using a large, geometrically and temporally aligned dataset of Sentinel-2 and PlanetScope imagery, we evaluate five state-of-the-art diffusion architectures under controlled experimental settings. We also introduce LPIPS-Sat, a domain-adapted perceptual metric based on Sentinel-2 self-supervised features. Our results show two persistent challenges: synthetically trained models degrade sharply on real pairs, while models trained on real cross-sensor data exhibit optimisation difficulties and struggle to adapt to the physical and radiometric diversity. These findings highlight a key limitation of current SR and motivate methods that disentangle super-resolution from domain adaptation.

Chinese Translation

对高分辨率卫星影像的需求增加了对超分辨率（SR）的关注，以弥补如Sentinel-2等免费任务与PlanetScope等商业系统之间的空间分辨率差距。由于没有传感器提供真实的低分辨率和高分辨率观测，SR模型通常在合成降级数据上进行训练，这在真实的跨传感器影像上造成了领域差距。在本研究中，我们首次系统性地研究了这种合成与真实不匹配如何影响现代基于扩散的SR模型的性能。我们使用了一个大型的几何和时间对齐的Sentinel-2与PlanetScope影像数据集，在受控实验条件下评估了五种最先进的扩散架构。我们还引入了LPIPS-Sat，这是一种基于Sentinel-2自监督特征的领域适应感知度量。我们的结果显示出两个持续的挑战：合成训练的模型在真实配对上急剧降级，而在真实跨传感器数据上训练的模型则表现出优化困难，难以适应物理和辐射多样性。这些发现突显了当前SR的一个关键限制，并激励了将超分辨率与领域适应解耦的方法。

View on arXiv Download PDF AI Translation

cs.CV / 72 / 2606.28049

AirGroundBench: Probing Spatial Intelligence in Multimodal Large Models under Heterogeneous Multi-View Embodied Collaboration

AirGroundBench：在异构多视角具身协作中探测多模态大模型的空间智能

Li, Haotian, Wang, Yida, Wang, Leyuan, Lai, Jinshan, Wang, Keyang, Guo, Zonghao, Ma, Qiang, Xiang, Liuyu, Hu, Jianwei, He, Zhaofeng

Abstract

In recent years, multimodal large language models (MLLMs) have shown strong potential for embodied intelligence, yet their ability to maintain geometrically consistent spatial understanding across heterogeneous views remains under-evaluated. Existing benchmarks largely focus on single-agent, single-view perception, leaving a gap in the systematic assessment of collaborative air-ground settings, where multi-scale observations are complementary but introduce scale mismatch, asymmetric occlusion, and reference-frame inconsistencies. We present AirGroundBench, a diagnostic benchmark for evaluating multi-view spatial intelligence in heterogeneous UAV-UGV collaboration. AirGroundBench is built from 11 high-fidelity simulated environments with 1,021 synchronized air-ground observation pairs, yielding approximately 62,000 dual-view, four-option single-choice visual question answering instances and 115 closed-loop vision-language navigation episodes. It covers 10 task types organized into four progressively demanding capability dimensions: spatial perception, cross-view alignment, spatial transformation and reasoning, and embodied decision-making. To support geometry-grounded evaluation and analysis, we provide structured spatial annotations, including cross-view object identities and metric 2D and 3D bounding boxes. Evaluations of 13 representative MLLMs under UAV-only, UGV-only, and dual-view input settings reveal consistent bottlenecks: models perform relatively well on spatial perception but struggle with cross-view alignment and transformation-intensive reasoning, and these deficits propagate to sequential decision-making in vision-language navigation. Although dual-view inputs provide measurable gains over single-view variants, a persistent gap from human performance remains, highlighting geometric consistency as a key limitation of current embodied MLLMs.

Chinese Translation

近年来，多模态大语言模型（MLLMs）展现出强大的具身智能潜力，但它们在异构视角下保持几何一致的空间理解能力仍然未得到充分评估。现有基准测试主要集中于单代理、单视角感知，导致在系统评估协作空地环境方面存在空白，在这些环境中，多尺度观察是互补的，但引入了尺度不匹配、非对称遮挡和参考框架不一致等问题。我们提出了AirGroundBench，这是一个用于评估异构无人机-地面车辆（UAV-UGV）协作中的多视角空间智能的诊断基准。AirGroundBench基于11个高保真模拟环境构建，包含1,021对同步的空地观察，生成约62,000个双视角、四选一的单项选择视觉问答实例和115个闭环视觉-语言导航情节。它涵盖了10种任务类型，组织为四个逐步提升的能力维度：空间感知、跨视角对齐、空间变换与推理，以及具身决策。为了支持基于几何的评估和分析，我们提供了结构化的空间注释，包括跨视角的物体身份和度量2D及3D边界框。在仅使用无人机、仅使用地面车辆和双视角输入设置下对13个代表性MLLM的评估揭示了一致的瓶颈：模型在空间感知上表现相对良好，但在跨视角对齐和变换密集的推理上存在困难，这些缺陷在视觉-语言导航中的顺序决策中得以延续。尽管双视角输入相较于单视角变体提供了可测量的提升，但与人类表现之间仍存在持续的差距，突显了几何一致性是当前具身MLLMs的一个关键限制。

View on arXiv Download PDF AI Translation

cs.CV / 73 / 2606.28060

ReScene: Structured Indoor Scene Reconstruction from Multi-View Captures

ReScene：基于多视角捕捉的结构化室内场景重建

Xu, Haoran, Zhang, Lechao, Dong, Daoguo, Gao, Yan, Tan, Xin

Abstract

Constructing simulation-ready 3D scenes from multi-view captures is a key bottleneck for Embodied Artificial Intelligence, as downstream tasks require object-level structure, explicit inter-object relations, and physical plausibility. Existing approaches either rely on specialized capture hardware, suffer from single-view bias in object reconstruction, or yield layouts that are geometrically reasonable but physically inconsistent. We identify that the problem is not single-object reconstruction but cross-view relation fusion and physically plausible scene assembly. To address this challenge, we present ReScene, a framework that threads multi-view geometry throughout the pipeline as a unifying prior. Our method consists of two main components: HierView prioritizes reconstruction views based on semantic consistency and 3D coverage completeness, replacing the largest-mask heuristic that conflates image occupancy with object coverage; and Relation-Aware Assembly fuses multi-frame relation predictions from a vision-language model with geometric and room-shell priors into a confidence-weighted scene graph, enabling physically consistent scene assembly. ReScene sets a new state of the art across geometry, rendering, and perceptual quality on a set of ScanNet scenes, achieving a 17% reduction in Chamfer Distance and 26% in LPIPS over the strongest prior baseline, while running up to 10x faster than prior multi-view methods. Based on the reconstructed scenes, we also generate an embodied visual question answering dataset, on which fine-tuned Qwen-VL approaches the performance of strong closed-source models on several spatial reasoning tasks.

Chinese Translation

从多视角捕捉构建适用于仿真的3D场景是具身人工智能的一个关键瓶颈，因为下游任务需要对象级结构、明确的对象间关系和物理合理性。现有的方法要么依赖于专用捕捉硬件，要么在对象重建中受到单视角偏差的影响，或者产生几何上合理但物理上不一致的布局。我们发现问题不在于单个对象的重建，而在于跨视角关系的融合和物理上合理的场景组装。为了解决这一挑战，我们提出了ReScene，一个在整个流程中将多视角几何作为统一先验的框架。我们的方法由两个主要组件组成：HierView根据语义一致性和3D覆盖完整性优先选择重建视图，替代了将图像占用与对象覆盖混淆的最大掩模启发式；而Relation-Aware Assembly则将来自视觉-语言模型的多帧关系预测与几何和房间外壳先验融合成一个加权信心场景图，从而实现物理一致的场景组装。ReScene在一组ScanNet场景上设定了几何、渲染和感知质量的新技术水平，相较于最强的先前基线，Chamfer距离减少了17%，LPIPS减少了26%，同时运行速度比先前的多视角方法快了最多10倍。基于重建的场景，我们还生成了一个具身视觉问答数据集，在该数据集上微调的Qwen-VL在多个空间推理任务上接近强闭源模型的性能。

View on arXiv Download PDF AI Translation

cs.CV / 74 / 2606.28077

TextDS: Parameter-Efficient Representation Alignment for Scene Text Detection under Distribution Shifts

TextDS：在分布变化下进行场景文本检测的参数高效表示对齐

Chen, Boyuan, Dang, Zichen, Yang, Chuang, Chau, Lap-pui, Wang, Yi

Abstract

In real-world deployments, scene text detectors inevitably face distribution shifts beyond the training distribution. Prior work often depends on large-scale scene-text pretraining, yet evaluation under cross-domain changes and real-world imaging degradations remains limited. We propose TextDS, an efficient framework for scene text detection under distribution shifts. First, we propose a data-efficient dual-encoder design with visual foundation models, eliminating the reliance on large-scale scene-text pretraining. Second, we introduce Step-wise LoRA adaptation (SWLoRA), which performs progressive low-rank refinement with a dynamic early-exit mechanism for effective feature adaptation. Third, we propose Common Subspace Fusion (CSF) to align and fuse the two branches in a shared subspace while retaining complementary, shift-robust information. Finally, we construct adverse-condition scene text detection datasets to address the gap in evaluating under imaging degradation. Experiments show that TextDS achieves competitive performance in scene text detection, demonstrating robustness across domains and adverse imaging conditions with only 4.9M trainable parameters.

Chinese Translation

在实际应用中，场景文本检测器不可避免地面临超出训练分布的分布变化。以往的研究通常依赖于大规模的场景文本预训练，但在跨域变化和现实世界成像退化下的评估仍然有限。我们提出了TextDS，一个高效的场景文本检测框架，旨在应对分布变化。首先，我们提出了一种数据高效的双编码器设计，结合视觉基础模型，消除了对大规模场景文本预训练的依赖。其次，我们引入了逐步低秩适应（Step-wise LoRA adaptation，SWLoRA），通过动态早期退出机制进行渐进式低秩细化，以实现有效的特征适应。第三，我们提出了共同子空间融合（Common Subspace Fusion，CSF），在共享子空间中对齐和融合两个分支，同时保留互补的、抗变化的信息。最后，我们构建了不利条件下的场景文本检测数据集，以填补在成像退化下评估的空白。实验表明，TextDS在场景文本检测中实现了具有竞争力的性能，展示了在不同领域和不利成像条件下的鲁棒性，仅需4.9M可训练参数。

View on arXiv Download PDF AI Translation

cs.CV / 75 / 2606.28083

STAG: Spatio-temporal Evolving Structural Representation of Action Units for Micro-expression Recognition

STAG：用于微表情识别的时空演变结构表示的动作单元

Sharma, Nandani, Sharma, Varun, Singh, Dinesh

Abstract

Micro-expression recognition is challenging due to subtle and short-lived facial muscle movements. Existing methods rely heavily on apex-onset frames, overlook fine-grained inter-frame dynamics, and separately model spatial and temporal information, limiting generalization across datasets. To address these challenges, we propose STAG, a dynamic ROI-AU-coupled spatial-temporal network that jointly models motion flow and adaptive facial connectivity. The framework extracts optical flow from discriminative frames using magnitude-based selection and temporal attention. A dual-branch architecture combines an enhanced graph attention network for structured spatial reasoning with a transformer encoder for temporal modeling. A bidirectional cross-attention module enables mutual refinement of spatial and temporal features, while AU-guided dynamic connectivity adapts facial region interactions according to muscle activation patterns. The transformer captures subtle temporal dynamics beyond apex-based approaches, improving semantic consistency and interpretability for explainable micro-expression recognition. The fused representation is optimized using focal loss and evaluated on CASME II, 4DME, DFME, NaME, SAMM, and SMIC-HS. Extensive experiments demonstrate improved robustness, generalization, interpretability, and computational efficiency, confirming the effectiveness of adaptive relational reasoning, AU-guided dynamic connectivity, and deep spatial-temporal feature fusion for accurate cross-dataset micro-expression recognition.

Chinese Translation

微表情识别因面部肌肉运动的细微和短暂而具有挑战性。现有方法过于依赖顶点起始帧，忽视了细粒度的帧间动态，并分别建模空间和时间信息，限制了跨数据集的泛化能力。为了解决这些挑战，我们提出了STAG，一种动态ROI-AU耦合的时空网络，联合建模运动流和自适应面部连接。该框架通过基于幅度的选择和时间注意力从判别帧中提取光流。双分支架构结合了增强的图注意力网络用于结构化空间推理，以及用于时间建模的变换器编码器。双向交叉注意力模块实现了空间和时间特征的相互精炼，而AU引导的动态连接根据肌肉激活模式调整面部区域交互。变换器捕捉超越顶点基础方法的细微时间动态，提高了语义一致性和可解释性，以实现可解释的微表情识别。融合表示通过聚焦损失进行优化，并在CASME II、4DME、DFME、NaME、SAMM和SMIC-HS上进行评估。大量实验表明，适应性关系推理、AU引导的动态连接和深度时空特征融合在准确的跨数据集微表情识别中显著提高了鲁棒性、泛化能力、可解释性和计算效率，证实了其有效性。

View on arXiv Download PDF AI Translation

cs.CV / 76 / 2606.28089

RPM-Distill: Physiology-guided Adaptive Cross-modal Distillation for Robust Remote Physiological Measurement

RPM-Distill：基于生理学指导的自适应跨模态蒸馏用于稳健的远程生理测量

Wang, Jiyao, Hu, Qingyong, Tang, Duoxun, Yang, Xiao, Wu, Kaishun, Yu, Jiangbo

Abstract

Video-based remote physiological measurement (RPM) is highly accessible but remains fragile under varying illumination, skin tones, and motion. Radio frequency (RF) radar is largely invariant to illumination and appearance, providing complementary cardio-respiratory micro-motion cues; however, requiring radar at inference is often impractical due to its limited ubiquity and deployment overhead. We propose RPM-Distill, a physiology-guided cross-modal distillation framework that leverages synchronized radar only during training while retaining video-only inference. Our key observation is that although RGB and RF waveforms differ in sensing physics and time-domain morphology, they share similar latent periodic rhythm in the frequency domain. We thus distill physiology-structured spectral evidence to improve robustness, via losses that (i) anchor the fundamental peak, (ii) match the off-peak background distribution, and (iii) preserve spectral morphology and sharpness. To avoid negative transfer under sample-level teacher quality and alignment uncertainty, a spectral policy network predicts sample-level distillation gates and component weights from the student--teacher spectral relation map, learned with a meta bilevel objective on a small labeled validation split. Through extensive experiments in challenging conditions and cross-dataset settings, RPM-Distill brings 81\% MAE and 21\% correlation improvement over unimodal baselines. Code is at https://github.com/WJULYW/RPM-Distill.

Chinese Translation

基于视频的远程生理测量（RPM）具有高度可及性，但在不同的光照、肤色和运动条件下仍然脆弱。射频（RF）雷达对光照和外观的变化具有较强的不变性，提供互补的心肺微运动线索；然而，由于其有限的普遍性和部署开销，在推理时使用雷达往往不切实际。我们提出了RPM-Distill，一个基于生理学指导的跨模态蒸馏框架，该框架在训练期间仅利用同步雷达，同时保留视频推理。我们的关键观察是，尽管RGB和RF波形在感知物理和时域形态上存在差异，但它们在频域中共享相似的潜在周期性节奏。因此，我们蒸馏出结构化的生理光谱证据以提高鲁棒性，通过损失函数实现（i）锚定基本峰值，（ii）匹配非峰值背景分布，以及（iii）保持光谱形态和清晰度。为了避免在样本级教师质量和对齐不确定性下的负迁移，一个光谱策略网络从学生-教师光谱关系图中预测样本级蒸馏门和组件权重，该关系图通过在小型标注验证集上学习的元双层目标获得。通过在挑战性条件和跨数据集设置中的广泛实验，RPM-Distill在单模态基线之上带来了81 ext{%}的平均绝对误差（MAE）和21 ext{%}的相关性提升。代码可在https://github.com/WJULYW/RPM-Distill获取。

View on arXiv Download PDF AI Translation

cs.CV / 77 / 2606.28092

Diffusion Model Attribution via Spectral Coupling of Denoiser Responses

通过去噪响应的谱耦合进行扩散模型归属

Meshram, Pragati Shuddhodhan, Chandrasekaran, Varun

Abstract

Attributing a generated image to its source diffusion model is a fundamental challenge in provenance verification and intellectual property protection. This problem is particularly difficult because diffusion models trained on different datasets can converge to similar score functions and thus similar output distributions, making the generated images themselves unreliable as attribution evidence. Existing non-invasive methods either fail on architecturally similar variants or rely on signals that vanish when models share the same autoencoder. We propose Spectral Denoising Signatures (SDS), a non-invasive attribution method that identifies the source model by fingerprinting each candidate model's denoising behavior. Our key insight is that a model's denoising score function exhibits a distinctive spectral geometry, reflected in how it redistributes energy across spatial frequency bands during denoising. By probing this behavior with frequency-controlled perturbations, SDS extracts a stable signature that is intrinsic to the model, requiring only standard forward passes with no inversion, optimization, or generation-time enrollment. Our results demonstrate that SDS achieves approximately 99.9% accuracy across eight diverse diffusion models and 96.2% under cross-domain prompt shift, outperforming non-invasive baselines across variations in training data, architecture, and training procedure, establishing spectral geometry as a principled and practical basis for diffusion model attribution. Code is available at: https://github.com/Pragati-Meshram/SGS

Chinese Translation

将生成的图像归属到其源扩散模型是来源验证和知识产权保护中的一个基本挑战。这个问题尤其困难，因为在不同数据集上训练的扩散模型可能会收敛到相似的评分函数，从而导致相似的输出分布，使得生成的图像本身作为归属证据变得不可靠。现有的非侵入性方法要么在架构相似的变体上失败，要么依赖于在模型共享相同自编码器时消失的信号。我们提出了谱去噪签名（Spectral Denoising Signatures, SDS），这是一种非侵入性归属方法，通过指纹识别每个候选模型的去噪行为来识别源模型。我们的关键见解是，模型的去噪评分函数表现出独特的谱几何特征，这体现在其在去噪过程中如何在空间频率带之间重新分配能量。通过使用频率控制的扰动来探测这种行为，SDS提取出一种稳定的签名，该签名是模型固有的，仅需标准的前向传递，无需反演、优化或生成时的注册。我们的结果表明，SDS在八个不同的扩散模型上实现了约99.9%的准确率，在跨域提示转移下为96.2%，在训练数据、架构和训练过程的变化中超越了非侵入性基线，确立了谱几何作为扩散模型归属的原则性和实用基础。代码可在以下链接获取：https://github.com/Pragati-Meshram/SGS

View on arXiv Download PDF AI Translation

cs.CV / 78 / 2606.28094

OSOR: One-Step Diffusion Inpainting for Effect-Aware Object Removal

OSOR：一种一步扩散修复的效果感知对象移除方法

Zhou, Qinming, Sun, Chenxi, Kong, Deyang, He, Junhao, Tang, Xiangheng, Yu, Peike, Wu, Haotian, Cao, Leilei, Zhang, Linfeng

Abstract

Real-world object removal is challenging due to two key difficulties: the target object's non-local effects, such as shadows and reflections, which are difficult to model, and the fact that user-provided masks are often inaccurate or incomplete. With billions of parameters and tens of denoising steps, diffusion-based models achieve strong removal performance at the expense of substantial computational cost, limiting their use in interactive applications and on edge devices. To address these challenges, we present OSOR (One-Step Object Removal), which simultaneously achieves efficient, effect-aware, and mask-robust object removal. Concretely, OSOR introduces: (1) an occupancy-guided discriminator for precise boundary supervision, enabling stable single-step diffusion training; (2) an alpha head that leverages knowledge from pretrained diffusion models to predict appropriate removal regions with minimal overhead, thereby handling imperfect masks; and (3) a semantic-anchored verification pipeline (SAVP) that filters noisy instruction-based triplets to produce effect-aware supervision at scale. Using SAVP, we curate CORNE, which contains 280K verified removal pairs, and further annotate AnimeEraseBench and TextEraseBench to evaluate performance on more complex removal tasks. Experiments show that OSOR surpasses strong multi-step diffusion baselines in perceptual quality while achieving $4\times$ to $30\times$ faster inference.

Chinese Translation

现实世界中的对象移除面临两个主要挑战：目标对象的非局部效应（如阴影和反射）难以建模，以及用户提供的掩码往往不准确或不完整。基于扩散的模型具有数十亿个参数和数十个去噪步骤，能够在巨大的计算成本下实现强大的移除性能，这限制了它们在交互式应用和边缘设备上的使用。为了解决这些挑战，我们提出了OSOR（一步对象移除），它同时实现了高效、效果感知和掩码鲁棒的对象移除。具体而言，OSOR引入了：(1) 一个基于占用率的鉴别器，用于精确的边界监督，从而实现稳定的单步扩散训练；(2) 一个α头，利用预训练扩散模型的知识，以最小的开销预测适当的移除区域，从而处理不完美的掩码；(3) 一个语义锚定的验证管道（SAVP），用于过滤噪声指令三元组，以大规模生成效果感知的监督。通过SAVP，我们策划了CORNE，包含28万个经过验证的移除对，并进一步注释了AnimeEraseBench和TextEraseBench，以评估在更复杂移除任务上的性能。实验表明，OSOR在感知质量上超越了强大的多步扩散基线，同时实现了$4 imes$到$30 imes$的推理加速。

View on arXiv Download PDF AI Translation

cs.CV / 79 / 2606.28104

Cross-view Multimodal Vision-Based Assessment Framework for Traditional Chinese Medicine Rehabilitation Training

基于跨视角多模态视觉的传统中医康复训练评估框架

Zhang, Francis Xiatian, Yao, Hao, Chen, Shengxuan, Zhu, Hong, Jia, Hongxiao, Zheng, Sisi, Shum, Hubert P. H.

Abstract

Vision-based assessment can provide convenient and cost-effective evaluation in Traditional Chinese Medicine (TCM) rehabilitation training, where action quality assessment (AQA) from computer vision offers a promising solution. Existing automatic AQA frameworks for physical therapy typically rely on skeletal data captured from a single viewpoint, which is inefficient for TCM techniques such as acupuncture or Tuina that involve dense hand self-occlusion and complex hand-object interactions. To address these challenges, we propose CME-AQA, a cross-view, multimodal vision-based assessment framework that integrates visual-pose fusion to enhance understanding of environmental context and leverages both first-person and third-person videos during training to improve inference robustness. We collected two dual-view datasets, TCM-AQA61-A (Acupuncture) and TCM-AQA61-T (Tuina), each containing synchronized first-person and third-person recordings of 61 subjects with expert annotations. Experimental results show that our approach achieves superior or comparable mean performance against competitive baselines, achieving over 10% relative improvement in weighted F1 over the best competing method on key rating tasks such as Needle Depth and Quick Needle Insertion, while also reducing mean absolute error in quantitative measures such as insertion time and manipulation frequency. Testing on a CPR dataset further demonstrates comparable performance on several posture-based criteria, suggesting applicability to related structured simulated clinical skill assessments where participant motion is central to evaluation. Overall, CME-AQA enhances assessment accuracy for structured TCM rehabilitation training and facilitates more convenient and effective training-oriented skill evaluation.

Chinese Translation

基于视觉的评估可以为传统中医（TCM）康复训练提供便捷且具有成本效益的评价，其中计算机视觉的动作质量评估（AQA）提供了一个有前景的解决方案。现有的物理治疗自动AQA框架通常依赖于从单一视角捕获的骨骼数据，这对于涉及密集手部自遮挡和复杂手物体交互的中医技术（如针灸或推拿）而言效率较低。为了解决这些挑战，我们提出了CME-AQA，一个跨视角的多模态视觉评估框架，集成了视觉-姿态融合，以增强对环境上下文的理解，并在训练过程中利用第一人称和第三人称视频来提高推理的鲁棒性。我们收集了两个双视角数据集，TCM-AQA61-A（针灸）和TCM-AQA61-T（推拿），每个数据集包含61名受试者的同步第一人称和第三人称录音，并附有专家标注。实验结果表明，我们的方法在关键评分任务（如针深度和快速针刺）上相较于竞争基线取得了优越或可比的平均表现，在加权F1上相较于最佳竞争方法实现了超过10%的相对改善，同时在插入时间和操作频率等定量指标上减少了平均绝对误差。对CPR数据集的测试进一步表明在多个基于姿势的标准上表现相当，表明其在相关结构化模拟临床技能评估中的适用性，其中参与者的运动是评估的核心。总体而言，CME-AQA提高了结构化中医康复训练的评估准确性，并促进了更便捷和有效的以训练为导向的技能评估。

View on arXiv Download PDF AI Translation

cs.CV / 80 / 2606.28112

BiDeMem: Bidirectional Degradation Memory for Explainable Image Restoration

BiDeMem：用于可解释图像恢复的双向退化记忆

Wu, Xinrui, Huang, Lichen

Abstract

Degradation-aware prompts, conditions, and latent priors are increasingly used in image restoration, yet they are usually judged by a single endpoint: whether the restored image obtains higher PSNR. This is a weak test of semantics. A condition can help by adding capacity, acting as a global correction bias, or exploiting dataset shortcuts, without becoming an interpretable degradation prior. We propose BiDeMem, a bidirectional degradation memory for explainable image restoration. A query built from restoration features and input statistics retrieves a compact top-k subset of memory slots. The same selected slot identity supports the restoration path at inference time and a training-only forward-degradation explanation path. The study centers on verifiability in a controlled multi-degradation NAFNet setting. New controls separate the gain from a correction head alone, a dense query prior, and a static global prior: these variants are 0.2588, 0.2586, and 0.2839 dB below BiRank, respectively. Strong residual supervision and a wider degradation head also remain below the full bidirectional memory model. Intervention probes show that BiRank preserves restoration quality while increasing wrong-prior and native-prior sensitivity, framing degradation memory as both a restoration module and a falsifiable explanation mechanism.

Chinese Translation

在图像恢复中，退化感知的提示、条件和潜在先验越来越多地被使用，但通常仅通过一个单一的终点来评判：恢复后的图像是否获得更高的PSNR。这是对语义的一个薄弱测试。条件可以通过增加容量、充当全局修正偏差或利用数据集捷径来提供帮助，而不成为可解释的退化先验。我们提出了BiDeMem，一种用于可解释图像恢复的双向退化记忆。由恢复特征和输入统计构建的查询检索出紧凑的前k个记忆槽子集。相同的选定槽身份在推理时支持恢复路径，并且在仅训练的前向退化解释路径中也得到支持。本研究集中在可控的多退化NAFNet设置中的可验证性。新的控制措施将来自单一修正头、密集查询先验和静态全局先验的增益分开：这些变体分别比BiRank低0.2588、0.2586和0.2839 dB。强大的残差监督和更宽的退化头也仍然低于完整的双向记忆模型。干预探针显示，BiRank在提高错误先验和本地先验敏感性的同时保持恢复质量，将退化记忆框架视为恢复模块和可证伪的解释机制。

View on arXiv Download PDF AI Translation

cs.CV / 81 / 2606.28128

PhysisForcing: Physics Reinforced World Simulator for Robotic Manipulation

PhysisForcing：强化物理的机器人操作世界模拟器

Zhang, Peiwen, Deng, Yufan, Sun, Shangkun, Ma, Juncheng, Wang, Duomin, Du, Jonas, Pan, Zilin, Huang, Ye, Liang, Hao, Huang, Songyan, Zhang, Ruihua, Xie, Enze, Liu, Ming-Yu, Zhou, Daquan

Abstract

Video generation models have emerged as a promising paradigm for embodied world simulation. However, both general-domain video generators and robot-specific data fine-tuned models can still produce physically implausible manipulations, including discontinuous motion trajectories and inconsistent robot-object interactions, which limits their reliability as world simulators. Through extensive experiments, we find that such physical instability mainly arises from two factors: deformation of moving objects and implausible spatio-temporal correlations among interacting entities, particularly during contact. Building on this observation, we propose PhysisForcing, a scalable training framework that strengthens physical consistency by focusing supervision on physics-informative regions through joint optimization of pixel-level and semantic-level features. The framework consists of a pixel-level trajectory alignment loss, which supervises DiT features using reference point trajectories, and a semantic-level relational alignment loss, which aligns DiT features with inter-region relations extracted from a frozen video understanding encoder. Extensive experiments on R-Bench, PAI-Bench, and EZS-Bench show that PhysisForcing consistently improves embodied video generation over strong baselines, improving the Wan2.2-I2V-A14B and Cosmos3-Nano base models on R-Bench by 22.3\% and 9.2\% (7.1\% and 3.7\% over vanilla finetuning), with the Cosmos3-Nano variant attaining the best overall score. Beyond generation, as a world model under the WorldArena action-planner protocol it raises the closed-loop success rate from 16.0\% to 24.0\% and further improves downstream policy success, indicating that physically aligned video models yield stronger representations for robotic manipulation.

Chinese Translation

视频生成模型已成为具身世界模拟的有希望的范式。然而，无论是通用领域的视频生成器还是针对机器人特定数据进行微调的模型，仍然可能产生不符合物理规律的操作，包括不连续的运动轨迹和不一致的机器人与物体的交互，这限制了它们作为世界模拟器的可靠性。通过大量实验，我们发现这种物理不稳定性主要源于两个因素：移动物体的变形以及交互实体之间不合理的时空关联，特别是在接触期间。基于这一观察，我们提出了PhysisForcing，一个可扩展的训练框架，通过对物理信息区域的重点监督，结合像素级和语义级特征的联合优化，增强物理一致性。该框架包括一个像素级轨迹对齐损失，它使用参考点轨迹监督DiT特征，以及一个语义级关系对齐损失，它将DiT特征与从冻结的视频理解编码器提取的区域间关系对齐。在R-Bench、PAI-Bench和EZS-Bench上的大量实验表明，PhysisForcing在强基线之上始终提高了具身视频生成，在R-Bench上将Wan2.2-I2V-A14B和Cosmos3-Nano基础模型的性能分别提高了22.3\%和9.2\%（相较于普通微调提高了7.1\\%和3.7\\%），其中Cosmos3-Nano变体获得了最佳整体得分。超越生成，作为WorldArena行动规划协议下的世界模型，它将闭环成功率从16.0\\%提高到24.0\\%，并进一步改善下游策略的成功率，这表明物理对齐的视频模型为机器人操作提供了更强的表征。

View on arXiv Download PDF AI Translation

cs.CV / 82 / 2606.28144

Monocular Avatar Reconstruction via Cascaded Diffusion Priors and UV-Space Differentiable Shading

通过级联扩散先验和UV空间可微分阴影的单目头像重建

Li, Hong, Meng, Minqi, Liang, Yanjun, Ye, Chongjie, Chen, Houyuan, Xiao, Weiqing, Guo, Xianda, Lei, Guojun, Liu, Xuhui, Yang, Chaojie, Peng, Yanlun, Zhao, Hao, Zhang, Baochang

Abstract

Reconstructing high-fidelity, relightable 3D avatars from a single in-the-wild image is a challenging ill-posed problem, primarily hindered by the scarcity of high-quality PBR data and the complexity of disentangling illumination from intrinsic materials. In this paper, we present a data-efficient framework that leverages the robust priors of a unified pre-trained diffusion backbone to sequentially address texture completion, delighting, and material decomposition. Unlike existing methods that rely on fragmented pipelines or extensive proprietary datasets, we utilize cascaded Low-Rank Adaptations (LoRAs) to adapt the strong generative prior of the diffusion model for each sub-task in UV space. Specifically, we first employ an Inpainting LoRA to complete missing UV textures caused by occlusion, leveraging the model's semantic understanding to generate semantically and photometrically coherent details. Subsequently, a Light-Homogenization LoRA and a novel Cross-Intrinsic Attention mechanism are introduced to remove baked-in lighting and collaboratively synthesize pixel-aligned PBR maps (Albedo, Normal, Roughness, Specular, and Displacement). To ensure physical plausibility, we impose a UV-space differentiable BRDF shading loss during the decomposition stage, forcing the generative process to adhere to the rendering equation without the artifacts typical of rasterization-based supervision. Extensive experiments demonstrate that our method, trained on fewer than 100 real 3D scans, generates comprehensive, 4K-resolution PBR assets with superior realism and generalization compared to state-of-the-art methods, and all training code and model weights will be released upon acceptance.

Chinese Translation

从单张自然图像中重建高保真、可重光照的3D头像是一个具有挑战性的病态问题，主要受到高质量PBR数据稀缺和从内在材料中解开光照复杂性的阻碍。在本文中，我们提出了一种数据高效的框架，利用统一的预训练扩散骨干网络的强大先验，顺序解决纹理补全、光照处理和材料分解。与依赖于碎片化管道或广泛专有数据集的现有方法不同，我们利用级联低秩适应（Low-Rank Adaptations, LoRAs）来为UV空间中的每个子任务适应扩散模型的强生成先验。具体而言，我们首先采用图像修复LoRA来补全因遮挡而缺失的UV纹理，利用模型的语义理解生成语义和光度上连贯的细节。随后，引入光照均匀化LoRA和一种新颖的跨内在注意机制，以去除嵌入光照并协同合成像素对齐的PBR贴图（反照率、法线、粗糙度、镜面反射和位移）。为了确保物理上的合理性，我们在分解阶段施加了UV空间可微分BRDF阴影损失，强迫生成过程遵循渲染方程，而不产生基于光栅化监督的典型伪影。大量实验表明，我们的方法在少于100个真实3D扫描的训练下，生成了全面的4K分辨率PBR资产，其真实感和泛化能力优于最先进的方法，所有训练代码和模型权重将在接受后发布。

View on arXiv Download PDF AI Translation

cs.CV / 83 / 2606.28149

Toward Robust In-Context Segmentation via Concept Guidance

通过概念引导实现稳健的上下文分割

Chen, Zhigang, Zheng, Xiawu, Ji, Rongrong

Abstract

In-context segmentation (ICS) requires a model to segment target regions in a query image using only a few reference images and their corresponding masks, without updating any parameters. Despite recent progress, prior ICS studies have largely overlooked a critical aspect: system robustness, ie, whether the model can produce stable segmentation results for the same query under different references. In this work, we revisit ICS from the robustness perspective and introduce a novel paradigm, Concept-Guided In-Context Segmentation (CG-ICS), which performs segmentation by extracting high-level semantic concepts from references rather than relying solely on low-level visual matching. Specifically, CG-ICS introduces a concept reasoning module that uses an MLLM to propose candidates and a SAM3-driven scoring function with tree-search refinement to select reliable textual concepts, together with a parallel visual exemplar route that provides query-side spatial grounding via a simple context construction. Both the textual concept and the visual exemplar are then used to activate the segmentation capability of a frozen SAM3 backbone. Extensive experiments on standard ICS benchmarks demonstrate that CG-ICS not only achieves state-of-the-art accuracy but also substantially improves robustness, yielding a more reliable ICS system with significantly reduced variance across diverse reference choices.

Chinese Translation

上下文分割（ICS）要求模型在不更新任何参数的情况下，仅使用少量参考图像及其对应的掩膜来对查询图像中的目标区域进行分割。尽管近期取得了一些进展，但以往的ICS研究在很大程度上忽视了一个关键方面：系统的稳健性，即模型在不同参考下是否能够为相同查询产生稳定的分割结果。在本研究中，我们从稳健性的角度重新审视ICS，并提出了一种新颖的范式——概念引导上下文分割（CG-ICS），该方法通过从参考中提取高层语义概念而非仅依赖低层视觉匹配来进行分割。具体而言，CG-ICS引入了一个概念推理模块，该模块使用多模态大语言模型（MLLM）提出候选概念，并结合树搜索优化的评分函数来选择可靠的文本概念，同时提供一个并行的视觉示例路径，通过简单的上下文构建为查询侧提供空间基础。文本概念和视觉示例随后被用来激活冻结的SAM3主干的分割能力。在标准ICS基准上的大量实验表明，CG-ICS不仅实现了最先进的准确性，而且显著提高了稳健性，提供了一个在多样化参考选择中具有显著降低方差的更可靠的ICS系统。

View on arXiv Download PDF AI Translation

cs.CV / 84 / 2606.28164

EchoSonar-R: A Multi-View Reasoning-Enabled Model for Disease Classification and Report Generation in Echocardiography

EchoSonar-R：一种支持多视角推理的心脏病分类与报告生成模型

Taratynova, Darya, Aly, Ahmed, Saeed, Numan, Yaqub, Mohammad

Abstract

Echocardiography is the most widely used non-invasive cardiac imaging modality, providing essential information for cardiovascular diagnosis. Interpreting an echocardiogram requires synthesizing complementary evidence across multiple heart views to identify abnormalities and produce structured clinical reports. While recent efforts focus on improving classification performance, most models lack explicit diagnostic reasoning and spatially grounded anatomical evidence, limiting clinician trust. We present EchoSonar-R, a multi-view reasoning-enabled vision-language model that jointly performs multi-label disease classification and report generation from echocardiography studies. EchoSonar-R combines a spatiotemporal video encoder with a structure-aware cardiac detector that provides spatially grounded anatomical cues to improve interpretability and clinician trust during cross-view reasoning. EchoSonar-R is trained in two stages: supervised fine-tuning (SFT) on reasoning-annotated targets, followed by Group Relative Policy Optimization (GRPO) with task-specific rewards that jointly align classification and report generation within a unified reinforcement-learning framework. Across a private multi-view dataset and two public benchmarks, EchoSonar-R improves macro balanced accuracy by 17.1% on the private set and 6.1% on MIMICEchoQA over the strongest baseline, achieves a GREEN clinical faithfulness score of 0.800, and produces interpretable reasoning traces grounded in multi-view visual evidence.

Chinese Translation

超声心动图是最广泛使用的非侵入性心脏成像方式，为心血管诊断提供了重要信息。解读超声心动图需要综合多个心脏视角的互补证据，以识别异常并生成结构化的临床报告。尽管最近的研究集中在提高分类性能上，但大多数模型缺乏明确的诊断推理和空间基础的解剖证据，从而限制了临床医生的信任。我们提出了EchoSonar-R，这是一种支持多视角推理的视觉-语言模型，能够从超声心动图研究中共同执行多标签疾病分类和报告生成。EchoSonar-R结合了时空视频编码器和结构感知心脏检测器，提供空间基础的解剖线索，以提高跨视角推理过程中的可解释性和临床医生的信任。EchoSonar-R的训练分为两个阶段：在带有推理注释的目标上进行监督微调（SFT），然后进行带有任务特定奖励的群体相对策略优化（GRPO），在统一的强化学习框架内共同对齐分类和报告生成。在一个私有多视角数据集和两个公共基准测试中，EchoSonar-R在私有数据集上提高了17.1%的宏观平衡准确率，在MIMICEchoQA上提高了6.1%，达到了0.800的GREEN临床可信度评分，并生成了基于多视角视觉证据的可解释推理轨迹。

View on arXiv Download PDF AI Translation

cs.CV / 85 / 2606.28215

HAT-4D: Lifting Monocular Video for 4D Multi-Object Interactions via Human-Agent Collaboration

HAT-4D：通过人机协作提升单目视频中的4D多物体交互

Li, Jiaxin, Wu, Yuxiang, Zhang, Zhenkai, Shi, Xinrui, Wang, Haoyuan, Zhao, Yichen, Linxiang, Su, Yu, Chenyang, Zhang, Mingyu, Ding, Yifan, Wen, Boran, Zhang, Li, Liu, Ruiyang, Li, Yong-Lu

Abstract

Extracting dynamic 4D object interactions from massive, in-the-wild monocular videos offers a highly efficient data collection pathway for scaling Embodied AI and training VLAs. However, existing monocular 4D reconstruction methods primarily focus on isolated objects, often failing under the severe occlusions and complex dynamics inherent in multi-object interactions. To bridge this gap, we propose HAT-4D, the first agentic framework designed to reconstruct the 3D geometry, temporal dynamics, and physical interactions of multiple objects from a single video. By integrating VLMs with a multi-level human-in-the-loop feedback mechanism, HAT-4D efficiently resolves depth ambiguities and interaction-induced occlusions during 3D generation and 4D propagation, yielding physically plausible assets without relying on expensive multicamera rigs. As a scalable data engine, HAT-4D facilitates the creation of MVOIK-4D, an open-world benchmark for monocular 4D interaction reconstruction, accompanied by a novel multi-dimensional evaluation protocol focused on physical plausibility and temporal consistency. Extensive experiments demonstrate that HAT-4D achieves SOTA performance on most evaluation metrics, while maintaining competitive semantic alignment. Ablation studies show that introducing a small amount of human feedback improves interaction reconstruction. Moreover, the data produced by HAT-4D effectively improves baseline performance when used for fine-tuning. Our data and code are available at https://lijiaxin0111.github.io/HAT4D/

Chinese Translation

从大量自然环境中的单目视频中提取动态4D物体交互，为扩展具身人工智能（Embodied AI）和训练视觉语言代理（VLA）提供了一条高效的数据收集途径。然而，现有的单目4D重建方法主要集中于孤立物体，往往在多物体交互中固有的严重遮挡和复杂动态下表现不佳。为了解决这一问题，我们提出了HAT-4D，这是第一个旨在从单个视频重建多个物体的3D几何形状、时间动态和物理交互的智能框架。通过将视觉语言模型（VLM）与多层次的人机反馈机制相结合，HAT-4D在3D生成和4D传播过程中有效解决了深度模糊和交互引起的遮挡问题，生成物理上合理的资产，而无需依赖昂贵的多摄像头设备。作为一个可扩展的数据引擎，HAT-4D促进了MVOIK-4D的创建，这是一个针对单目4D交互重建的开放世界基准，并附带了一种新的多维评估协议，重点关注物理合理性和时间一致性。大量实验表明，HAT-4D在大多数评估指标上实现了最先进的性能，同时保持了竞争性的语义对齐。消融研究表明，引入少量人类反馈可以改善交互重建。此外，HAT-4D生成的数据在用于微调时有效提高了基线性能。我们的数据和代码可在 https://lijiaxin0111.github.io/HAT4D/ 获取。

View on arXiv Download PDF AI Translation

cs.CV / 86 / 2606.28226

Exposure Bias Can Alleviate Itself via Directional and Frequency Rectification in Flow Matching

曝光偏差可以通过方向和频率校正在流匹配中自我缓解

Huang, Guanbo, Mao, Jingjia, Huang, Fanding, Liu, Fengkai, Luo, Xiangyang, Liang, Yaoyuan, Lu, Jiasheng, Wang, Xiaoe, Liu, Pei, Fu, Ruiliu, Huang, Ruqi, Huang, Shao-Lun

Abstract

Flow Matching (FM) has achieved remarkable generative performance, yet it suffers from exposure bias due to discrepancies between training and inference. Existing mitigation strategies typically rely on static constraints or external heuristics. In this work, we propose that exposure bias itself inherently contains dynamic signals that can guide its own rectification. To leverage this, we introduce DEFAR (DirEctional-Frequency Adaptive Rectification). This framework simulates the single-step inference process during training to identify exposure bias. It utilizes directional and frequency-adaptive feedback signals from the bias itself to enhance the model's bias tolerance. It consists of two key components: (1) Anti-Drift Rectification (ADR). ADR treats inference-time drift as a signal to learn the direction to steer deviated states back toward the target. ADR endows the model with intrinsic active self-rectification capabilities; (2) Frequency Compensation (FC). Empirically, we observe that accumulated bias often stems from a lack of low-frequency components in high-noise stages, and exposure bias carries the missing frequency. FC leverages the bias itself as a self-feedback weighting factor to reinforce the missing frequency components. Experiments on CIFAR-10, CelebA-64, and ImageNet-256/512 show that DEFAR outperforms prior baselines and further demonstrates favorable scalability, compatibility, and inference robustness.

Chinese Translation

流匹配（Flow Matching, FM）已实现显著的生成性能，但由于训练与推理之间的差异，它面临曝光偏差的问题。现有的缓解策略通常依赖于静态约束或外部启发式方法。在本研究中，我们提出曝光偏差本身内在地包含动态信号，可以指导其自我校正。为此，我们引入了DEFAR（方向-频率自适应校正）。该框架在训练期间模拟单步推理过程，以识别曝光偏差。它利用来自偏差本身的方向性和频率自适应反馈信号来增强模型的偏差容忍度。该框架由两个关键组件组成：（1）抗漂移校正（Anti-Drift Rectification, ADR）。ADR将推理时的漂移视为信号，以学习将偏离状态引导回目标的方向。ADR赋予模型内在的主动自我校正能力；（2）频率补偿（Frequency Compensation, FC）。经验表明，累积偏差通常源于高噪声阶段缺乏低频成分，而曝光偏差携带着缺失的频率。FC利用偏差本身作为自反馈加权因子，以增强缺失的频率成分。在CIFAR-10、CelebA-64和ImageNet-256/512上的实验表明，DEFAR的表现优于之前的基线，并进一步展示了良好的可扩展性、兼容性和推理鲁棒性。

View on arXiv Download PDF AI Translation

cs.CV / 87 / 2606.28266

RSICCLLM: A Multimodal Large Language Model for Remote Sensing Image Change Captioning

RSICCLLM：一种用于遥感图像变化描述的多模态大语言模型

Wang, Yelin, Song, Zijia, Ye, Shuo, Yang, Chuanguang, Wang, Miaoyu, Xu, Yong, An, Zhulin, Xu, Yongjun, Yu, Zitong

Abstract

Remote Sensing Image Change Captioning (RSICC) aims to describe changes between bi-temporal remote sensing images and holds significant research and application value. However, most existing methods rely on conventional deep learning architectures, and the limited model capacity constrains performance. Although large-model post-training techniques have achieved great success in general domains, their direct transfer to RSICC remains challenging due to data scarcity and the need for fine-grained change understanding. To address this, we propose RSICCLLM, the first post-training framework for large vision-language models in RSICC. Specifically, we design a data generation paradigm, release the instruction dataset RSICI, and establish a task-specific RSICC benchmark. We further introduce Difference-aware Supervised Fine-tuning to explicitly extract change representations and guide the model in perceiving and understanding temporal differences. In addition, we propose Dual-Negative Preference Optimization (DNPO), which employs two complementary negative-sample construction strategies to construct the preference dataset RSICP and further refine model performance. Extensive experiments validate the superior capability of RSICCLLM, which achieves outstanding results with only 7B parameters, surpassing models of substantially larger scales. The code and dataset will be made publicly available at https://github.com/keaill/RSICCLLM.

Chinese Translation

遥感图像变化描述（RSICC）旨在描述双时相遥感图像之间的变化，具有重要的研究和应用价值。然而，大多数现有方法依赖于传统的深度学习架构，有限的模型容量限制了性能。尽管大模型后训练技术在一般领域取得了巨大成功，但由于数据稀缺和对细粒度变化理解的需求，其直接转移到RSICC仍然具有挑战性。为了解决这一问题，我们提出了RSICCLLM，这是RSICC中首个针对大规模视觉-语言模型的后训练框架。具体而言，我们设计了一种数据生成范式，发布了指令数据集RSICI，并建立了任务特定的RSICC基准。我们进一步引入了差异感知监督微调（Difference-aware Supervised Fine-tuning），以明确提取变化表示并指导模型感知和理解时间差异。此外，我们提出了双负偏好优化（Dual-Negative Preference Optimization，DNPO），采用两种互补的负样本构建策略来构建偏好数据集RSICP，并进一步优化模型性能。大量实验验证了RSICCLLM的卓越能力，其在仅有70亿参数的情况下取得了优异的结果，超过了规模更大的模型。代码和数据集将公开发布在 https://github.com/keaill/RSICCLLM。

View on arXiv Download PDF AI Translation

cs.CV / 88 / 2606.28268

Learning Topology-Aware Representations via Test-Time Adaptation for Anomaly Segmentation

通过测试时适应学习拓扑感知表示用于异常分割

Zia, Ali, Ali, Usman, Rehman, Abdul, Ramzan, Umer, Han, Kang, Faheem, Muhammad, Qureshi, Shahnawaz, Xiang, Wei

Abstract

Test-time adaptation (TTA) has emerged as a promising paradigm for mitigating distribution shifts in deep models. However, existing TTA approaches for anomaly segmentation remain limited by their reliance on pixel-level heuristics, such as confidence thresholding or entropy minimisation, which fail to preserve structural consistency under noise and texture variation. Moreover, they typically treat anomaly maps as flat intensity fields, ignoring the higher-order spatial relationships that characterise complex defect geometries. We introduce TopoTTA (Topological Test-Time Adaptation), a novel framework that integrates persistent homology, a tool from topological data analysis, into the TTA pipeline to enforce geometric and structural coherence during adaptation. By applying multi-level cubical complex filtration to anomaly score maps, TopoTTA derives robust topological pseudo-labels that guide a lightweight test-time classifier, enhancing segmentation quality without retraining the backbone model. The approach avoids reliance on method-specific raw-score thresholding for mask binarisation, preserves connectivity, and generalises across both 2D and 3D modalities. Extensive experiments across six standard benchmarks (MVTec AD, VisA, Real-IAD, MVTec 3D-AD, AnomalyShapeNet, and MVTec LOCO) demonstrate an average 15% F1 improvement over state-of-the-art unsupervised anomaly detection and segmentation methods, with the largest gains on anomalies exhibiting complex geometric or structural variations. These findings suggest that integrating topological reasoning into test-time adaptation provides a principled route to structure-aware generalisation, bridging the gap between geometric learning and robust adaptation.

Chinese Translation

测试时适应（TTA）作为一种有前景的范式，已被提出用于缓解深度模型中的分布转变。然而，现有的用于异常分割的 TTA 方法仍然受到依赖于像素级启发式方法的限制，例如置信度阈值或熵最小化，这些方法无法在噪声和纹理变化下保持结构一致性。此外，它们通常将异常图视为平坦的强度场，忽略了表征复杂缺陷几何形状的高阶空间关系。我们提出了 TopoTTA（拓扑测试时适应），这是一个新颖的框架，将持久同调这一拓扑数据分析工具整合到 TTA 流程中，以在适应过程中强制执行几何和结构一致性。通过对异常得分图应用多级立方复合体滤波，TopoTTA 推导出稳健的拓扑伪标签，以指导轻量级的测试时分类器，从而提高分割质量而无需重新训练主干模型。该方法避免了对特定方法的原始得分阈值进行掩膜二值化的依赖，保持了连通性，并在 2D 和 3D 模态之间具有良好的泛化能力。在六个标准基准（MVTec AD、VisA、Real-IAD、MVTec 3D-AD、AnomalyShapeNet 和 MVTec LOCO）上的广泛实验表明，相较于最先进的无监督异常检测和分割方法，平均 F1 值提升了 15%，在表现出复杂几何或结构变化的异常上获得了最大的提升。这些发现表明，将拓扑推理整合到测试时适应中为结构感知泛化提供了一条有原则的途径，弥合了几何学习与稳健适应之间的差距。

View on arXiv Download PDF AI Translation

cs.CV / 89 / 2606.28321

StructSplat: Generalizable 3D Gaussian Splatting from Uncalibrated Sparse Views

StructSplat：来自未校准稀疏视图的可泛化3D高斯溅射

Zhao, Jia-Chen, Chen, Beiqi, Chen, Xinyang, Wang, Guangcong, Nie, Liqiang

Abstract

We present StructSplat, a feed-forward and generalizable 3D Gaussian reconstruction framework that operates directly on uncalibrated images without requiring camera parameters. Existing methods either rely on per-scene optimization or assume known camera poses, and often entangle geometry and appearance within a unified backbone, limiting reconstruction fidelity and generalization. Our key idea is to adopt a structured representation that organizes geometry, semantic, and texture cues with explicit roles in the reconstruction process. Specifically, we introduce a pixel-aligned feature injection mechanism to enable accurate texture modeling from 2D observations, incorporate semantic-aware priors to improve global consistency, and design a camera alignment strategy to prevent information leakage and improve generalization. Experiments show that our method significantly outperforms prior approaches on challenging benchmarks. On DL3DV, our method achieves 28.045 PSNR, surpassing AnySplat (22.377) by +5.67 dB. In cross-dataset evaluation, our method achieves +1.94 dB over AnySplat on ACID and +1.72 dB on RealEstate10K. Project page: https://structsplat.github.io Code: https://github.com/J-C-Zhao/StructSplat

Chinese Translation

我们提出了StructSplat，一个前馈且可泛化的3D高斯重建框架，能够直接在未校准的图像上操作，而无需相机参数。现有方法要么依赖于每个场景的优化，要么假设已知的相机姿态，通常将几何和外观纠缠在一个统一的骨干网络中，从而限制了重建的保真度和泛化能力。我们的关键思想是采用一种结构化表示，组织几何、语义和纹理线索，在重建过程中具有明确的角色。具体而言，我们引入了一种像素对齐特征注入机制，以实现从2D观测中准确建模纹理，结合语义感知的先验以改善全局一致性，并设计了一种相机对齐策略，以防止信息泄漏并提高泛化能力。实验表明，我们的方法在具有挑战性的基准测试中显著优于之前的方法。在DL3DV上，我们的方法达到了28.045的PSNR，超越了AnySplat（22.377）5.67 dB。在跨数据集评估中，我们的方法在ACID上比AnySplat提高了1.94 dB，在RealEstate10K上提高了1.72 dB。项目页面：https://structsplat.github.io 代码：https://github.com/J-C-Zhao/StructSplat

View on arXiv Download PDF AI Translation

cs.CV / 90 / 2606.28322

PerceptionRubrics: Calibrating Multimodal Evaluation to Human Perception

感知评分标准：将多模态评估校准至人类感知

Wei, Yana, Peng, Hongbo, Lai, Yanlin, Zhao, Liang, Lin, Kangheng, Yu, En, Lv, Keyu, Zhou, Han, Tang, Yin, Li, Haodong, Huang, Mitt, Guo, Hangyu, Sun, Jianjian, Ge, Zheng, Zhang, Xiangyu, Jiang, Daxin, Patel, Vishal M.

Abstract

We introduce PerceptionRubrics, a rubric-based evaluation framework that addresses the gap between saturated benchmark scores and real-world brittleness. Shifting evaluation from holistic semantic matching to rigorous atomic auditing, PerceptionRubrics pairs 1,038 information-dense images with over 12,000 instance-specific rubrics. These criteria are derived from golden captions constructed via a novel Circular Peer-Review consensus pipeline and then distilled into a dual-stream system of Must-Right (essential facts) and Easy-Wrong (fine-grained details) rubrics. Crucially, PerceptionRubrics implements a Gated Scoring mechanism: unlike linear averages, failure on mandatory visual facts triggers sharp binary penalties. Extensive evaluation yields critical insights: (1) The Reliability Gap: models often verify fragmented elements correctly yet fail strict conjunctive constraints, exposing brittleness in dense domains; (2) Open-Closed Stratification: contrary to reasoning trends, we reveal a persistent 8% perception deficit between open-source and proprietary frontiers; and (3) Human-Aligned Rigor: our gated metrics substantially out-align conventional benchmarks, validating that strict perceptual fidelity is the prerequisite for reliable generation.

Chinese Translation

我们提出了感知评分标准（PerceptionRubrics），这是一个基于评分标准的评估框架，旨在弥合饱和基准分数与现实世界脆弱性之间的差距。感知评分标准将评估从整体语义匹配转变为严格的原子审计，配对了1,038幅信息密集的图像和超过12,000个特定实例的评分标准。这些标准源自通过一种新颖的循环同行评审共识流程构建的黄金标题，并被提炼为一个双流系统，包括必须正确（Must-Right，基本事实）和易错（Easy-Wrong，细节）评分标准。至关重要的是，感知评分标准实施了一种门控评分机制：与线性平均不同，未能满足强制视觉事实会触发严格的二元惩罚。广泛的评估提供了关键的见解：（1）可靠性差距：模型通常能够正确验证碎片化元素，但未能满足严格的联结约束，暴露了在密集领域中的脆弱性；（2）开放-封闭分层：与推理趋势相反，我们揭示了开放源和专有前沿之间持续存在的8%感知缺口；（3）人类对齐的严格性：我们的门控指标显著优于传统基准，验证了严格的感知保真度是可靠生成的前提。

View on arXiv Download PDF AI Translation

人工智能 (Artificial Intelligence)

cs.AI / 1 / 2606.27382

AI-Model Network: Concept, Current State and Future

人工智能模型网络：概念、现状与未来

Zhetao, Li, Xiyu, Zeng, Jianhui, Wang, Yong, Xiao, Zhongren, Liu, Junru, Wu, Junjie, Lai, Jijun, Huang, Saiqin, Long

Abstract

While the primary function of computers lies in computation and processing, the core value of the Internet is rooted in sharing and collaboration. Computers create the Internet, and the Internet empowers the value of computers. The rapid development of the Internet, cloud computing, and big data is pushing artificial intelligence into the era of large models (LMs). However, the practical application of LMs is currently hindered by high training costs and deployment complexities, driving a shift toward lightweight, private, and domain-specific models. With the rapid proliferation and wide distribution of heterogeneous models, enabling effective interaction and collaboration among them has emerged as a critical bottleneck that urgently needs to be addressed in LM development. Drawing inspiration from the development of the Internet, this paper proposes the concept, vision, and system architecture of world wide AI-model network (AI-ModelNet). It is a novel paradigm that achieves interconnection, capability sharing, and collaborative reasoning by establishing pathways between models. We first briefly review the current state of single-model and multi-model research. Subsequently, the systemic vision and hierarchical architecture of AI-ModelNet are articulated, followed by validation of the framework's feasibility through a prototype system and diverse application cases. Finally, key directions for future research are discussed preliminarily.

Chinese Translation

计算机的主要功能在于计算和处理，而互联网的核心价值则根植于共享与协作。计算机创造了互联网，而互联网又赋予了计算机更大的价值。互联网、云计算和大数据的快速发展正在推动人工智能进入大模型（LMs）时代。然而，LMs的实际应用目前受到高昂的训练成本和复杂的部署问题的制约，促使人们转向轻量级、私有化和特定领域的模型。随着异构模型的快速扩散和广泛分布，促进它们之间的有效互动与协作已成为LM开发中亟待解决的关键瓶颈。本文借鉴互联网的发展，提出了全球人工智能模型网络（AI-ModelNet）的概念、愿景和系统架构。这是一种新颖的范式，通过在模型之间建立通道，实现互联互通、能力共享和协同推理。我们首先简要回顾了单模型和多模型研究的现状。随后，阐述了AI-ModelNet的系统愿景和分层架构，并通过原型系统及多样化应用案例验证了该框架的可行性。最后，初步讨论了未来研究的关键方向。

View on arXiv Download PDF AI Translation

cs.AI / 2 / 2606.27443

When Does Personality Composition Matter for Multi-Agent LLM Teams?

个性构成何时对多智能体大语言模型团队重要？

Keluskar, Aryan, Bhattacharjee, Amrita, Liu, Huan

Abstract

Personality prompting shapes how large language models communicate, yet whether these behavioral shifts affect objective task outcomes remains under-explored. Prior work shows that agents prompted with low agreeableness produce adversarial language, while those prompted with high agreeableness become cooperative, but the relationship between communication style and task performance has not been systematically examined across multiple domains. In this work, we investigate whether personality composition matters for multi-agent team performance by manipulating personality traits across frontier LLMs on three task domains: structured coding, open-ended research collaboration, and competitive bargaining. We find that personality effects depend critically on task structure. In coding tasks, low agreeableness leads to large communication shifts that have little effect on milestone completion. In open-ended collaboration and bargaining, the same manipulation substantially degrades performance. We discuss implications for multi-agent system design and the limits of personality manipulation.

Chinese Translation

个性提示影响大型语言模型的沟通方式，但这些行为变化是否会影响客观任务结果仍然未被充分探讨。先前的研究表明，低宜人性提示的智能体会产生对抗性语言，而高宜人性提示的智能体则变得合作，但沟通风格与任务表现之间的关系尚未在多个领域系统性地研究。在本研究中，我们通过操控前沿大型语言模型（LLMs）中的个性特征，探讨个性构成对多智能体团队表现的重要性，研究涉及三个任务领域：结构化编码、开放式研究合作和竞争性谈判。我们发现，个性效应在很大程度上依赖于任务结构。在编码任务中，低宜人性导致显著的沟通变化，但对里程碑完成的影响很小。在开放式合作和谈判中，同样的操控会显著降低表现。我们讨论了对多智能体系统设计的影响以及个性操控的局限性。

View on arXiv Download PDF AI Translation

cs.AI / 3 / 2606.27483

Internalizing the Future: A Unified Agentic Training Paradigm for World Model Planning

内化未来：一种统一的自主训练范式用于世界模型规划

Zhang, Xuan, Zhou, Zhijian, Qiao, Lingfeng, Qin, Yulei, Li, Ke, Sun, Xing, Tan, Xiaoyu, Qu, Chao, Qi, Yuan

Abstract

Large language model (LLM) agents have demonstrated strong capability in sequential decision-making, yet they remains fundamentally reactive in long-horizon tasks. Unlike humans who employ "what-if" reasoning to evaluate potential plans before commitment, standard agents lack an internal world model to simulate future outcomes. Therefore, we propose to internalize future-aware planning by training a single autoregressive model to verbalize both a prospective state rollout and a plan-conditioned success estimate-a textual analogue of the Q-value. Crucially, we identify a format-capability gap: simply fine-tuning agents on look-ahead traces during post-training leads to superficial mimicry of foresight without genuine predictive grounding. To bridge this gap, we introduce a three-stage training paradigm: (i) World Model Agentic Mid-Training (WM-AMT) to inject latent predictive capabilities into the policy; (ii) Format-Eliciting SFT (FE-SFT) to structure this injected capability; and (iii) Foresight-Conditioned Reinforcement Learning (FC-RL) to refine the calibration and utility of the generated simulations. Evaluated on search and mathematical reasoning tasks, our approach consistently outperforms other training baselines. Our results demonstrate that effective internal world modeling in LLM agents requires a capability-first training pipeline to achieve grounded and calibrated foresight.

Chinese Translation

大型语言模型（LLM）代理在顺序决策中表现出强大的能力，但在长时间跨度的任务中仍然基本上是反应性的。与人类通过“假设”推理在承诺之前评估潜在计划不同，标准代理缺乏内部世界模型来模拟未来结果。因此，我们提出通过训练一个单一的自回归模型来内化未来感知规划，该模型能够同时表达预期状态的展开和基于计划的成功估计——一个文本类比于 Q 值。关键是，我们识别出一种格式-能力差距：在后期训练中仅仅对代理进行前瞻性轨迹的微调，会导致对前瞻性的肤浅模仿，而没有真正的预测基础。为了解决这一差距，我们引入了一种三阶段的训练范式：（i）世界模型自主中期训练（WM-AMT），以将潜在的预测能力注入策略；（ii）格式引导的微调（FE-SFT），以构建这种注入的能力；（iii）前瞻性条件强化学习（FC-RL），以精炼生成模拟的校准和效用。在搜索和数学推理任务上的评估表明，我们的方法始终优于其他训练基准。我们的结果表明，在 LLM 代理中有效的内部世界建模需要一种以能力为先的训练流程，以实现有根据和经过校准的前瞻性。

View on arXiv Download PDF AI Translation

cs.AI / 4 / 2606.27593

Odyssey: Constructing Verifiable Local Truth-Preserving Foundation Models

奥德赛：构建可验证的局部真实保持基础模型

Mahadevan, Sridhar

Abstract

We introduce a categorical framework called ODYSSEY for constructing verifiable, local truth-preserving foundation models as compositions of foundries: building-block architectural components that specify a cover of local contexts, local representation families, restriction maps, gluing rules, obstruction policies, update obligations, and human-facing views. A foundry is an organized sheaf of knowledge that carries within it an argumentation component. Concrete foundries are built from generic foundries such as evidence/argument, operational decision, institutional/financial, market meaning, scientific challenge, research-program, assistant-build, and evaluation-harness foundries. Universal Foundry Learning (UFL) formalizes foundry construction as a composition of left and right Kan extensions, with left Kan extension rolling local artifacts into candidate foundries and right Kan extension enforcing the restriction, gluing, obstruction, and argumentation conditions required for promotion. Foundry SQL (FSQL) is a small typed query surface for slicing maintained foundry artifacts that uses TICKET (Topos Integration using Causal Kan Extension Transformers) certification for admitting external or pre-built models into durable ODYSSEY state. ODYSSEY is fully implemented and tested across a wide spectrum of concrete foundries, showing that the same categorical machinery supports domain construction, artifact replay, sheaf diagnostics, grounded Toulmin/local-LLM scrutiny, residual-obstruction ledgers, and optimized TICKET-compatible causal-claim extraction across heterogeneous sources. This paper is to be presented as a 2.5 hour tutorial at ICML 2026. The tutorial home page is at https://bit.ly/4ajS0nA.

Chinese Translation

我们引入了一种称为ODYSSEY的范畴框架，用于构建可验证的局部真实保持基础模型，这些模型是由铸造厂的组合构成的：铸造厂是建筑块架构组件，指定了局部上下文的覆盖、局部表示族、限制映射、粘合规则、障碍政策、更新义务和面向人类的视图。铸造厂是一个组织化的知识层，它内部包含一个论证组件。具体的铸造厂是由通用铸造厂构建而成，例如证据/论证、操作决策、机构/金融、市场意义、科学挑战、研究计划、助手构建和评估工具铸造厂。通用铸造学习（Universal Foundry Learning, UFL）将铸造厂的构建形式化为左Kan扩展和右Kan扩展的组合，其中左Kan扩展将局部工件整合为候选铸造厂，而右Kan扩展则强制执行提升所需的限制、粘合、障碍和论证条件。铸造厂SQL（Foundry SQL, FSQL）是一个小型类型查询界面，用于切片维护的铸造厂工件，使用TICKET（因果Kan扩展变换器的拓扑集成）认证，以允许外部或预构建模型进入持久的ODYSSEY状态。ODYSSEY已在广泛的具体铸造厂中全面实施和测试，显示出相同的范畴机制支持领域构建、工件重放、层诊断、基于Toulmin的局部LLM审查、残余障碍账本，以及跨异构来源的优化TICKET兼容因果声明提取。本文将作为2026年ICML会议的2.5小时教程进行展示。教程主页为https://bit.ly/4ajS0nA。

View on arXiv Download PDF AI Translation

cs.AI / 5 / 2606.27619

DysLexLens: A Low-Resource LLM Framework for Analysing Dyslexic Learners Insights from Online Forums

DysLexLens：一种低资源大型语言模型框架，用于分析来自在线论坛的阅读障碍学习者的见解

Rezazadegan, Dana, Kia, Atie, Nandavong, Phongpadid, Carlon, Dominique, Nguyen, Jeremy, Banerjee, Abhik, Marshall, James, McCosker, Anthony, Kang, Yong-Bin

Abstract

Dyslexic learners increasingly use artificial intelligence (AI) tools to support reading, writing, organisation, and study-related tasks. However, their lived experiences with these tools remain largely underexamined. This paper proposes DysLexLens, a low-resource LLM framework, designed to analyse dyslexic learners experience with AI through online forum discussions. DysLexLens is designed as an end-to-end, evidence-traceable architecture which transforms noisy social media posts into a dictionary-driven corpora, provides knowledge-graph (KG)-based question reasoning, generates verifiable query responses, and enables response evaluation through quantitative and human-grounded assessment. DysLexLens has four key features. First, it employs a dictionary-driven filtering method to construct a more focused Reddit corpus on dyslexia and AI, filtering out noisy and weakly related posts to improve the relevance of data collected from low-resource forum contexts. Second, it integrates LLM-assisted semantic analysis with KG-based query reasoning to uncover meaningful patterns. Third, it has quantitative evaluation metrics (RAGAS and Query Robustness) to measure LLM-generated response performance. Fourth, it provides structured qualitative validation guidelines for assessing response quality, with a specific focus on hallucination and evidence alignment. We demonstrate the effectiveness of DysLexLens using dyslexia-related Reddit forum data and 30 questions. The results show its potential generalisability to other low-resource forum data contexts. DysLexLens, sample data, questions and evaluation results are available at Github to support reproducibility.

Chinese Translation

阅读障碍学习者越来越多地使用人工智能（AI）工具来支持阅读、写作、组织和学习相关任务。然而，他们与这些工具的实际体验仍然在很大程度上未被深入研究。本文提出了DysLexLens，一种低资源大型语言模型（LLM）框架，旨在通过在线论坛讨论分析阅读障碍学习者与AI的体验。DysLexLens被设计为一个端到端、可追溯证据的架构，将嘈杂的社交媒体帖子转化为基于词典的语料库，提供基于知识图谱（KG）的问答推理，生成可验证的查询响应，并通过定量和基于人类的评估实现响应评估。DysLexLens具有四个关键特征。首先，它采用基于词典的过滤方法，构建一个更聚焦于阅读障碍和AI的Reddit语料库，过滤掉嘈杂和弱相关的帖子，以提高从低资源论坛上下文中收集的数据的相关性。其次，它将LLM辅助的语义分析与基于KG的查询推理相结合，以揭示有意义的模式。第三，它具有定量评估指标（RAGAS和查询鲁棒性），用于衡量LLM生成的响应性能。第四，它提供结构化的定性验证指南，以评估响应质量，特别关注幻觉和证据一致性。我们使用与阅读障碍相关的Reddit论坛数据和30个问题展示了DysLexLens的有效性。结果表明其在其他低资源论坛数据上下文中的潜在普遍适用性。DysLexLens、样本数据、问题和评估结果可在Github上获取，以支持可重复性。

View on arXiv Download PDF AI Translation

cs.AI / 6 / 2606.27652

MER-R1: Multimodal Emotion Reasoning via Slow-Fast Thinking Synergy

MER-R1：通过慢速-快速思维协同进行多模态情感推理

Han, Zhiyuan, Zhu, Beier, Tong, Wenwen, Qin, Chengwei, Wang, Xinyi, Zhang, Jiayu, Chen, Jiangnan, Guo, Hewei, Ran, Dongchuan, Lu, Lewei, Yang, Xun

Abstract

We find that explicit reasoning does not necessarily translate into better multimodal emotion recognition (MER) accuracy, even though it makes predictions more interpretable. Specifically, for reasoning-based MLLMs, fast thinking by triggering direct answers often outperforms slow thinking after deliberative reasoning. Our empirical analyses show that fast thinking improves recall with broader and more confident predictions, whereas slow thinking favors precision through conservative filtering of incorrect categories. Building on these insights, we propose MER-R1, a reinforcement learning framework that turns slow-fast complementarity into explicit optimization. Dual-objective disentanglement separates recall and precision into two optimization signals, allowing them to be jointly optimized rather than traded off against each other. Slow-fast confidence calibration further aligns the final slow-thinking answer with fast-thinking intuition, strengthening correct emotions while suppressing incorrect ones. In this way, MER-R1 unifies the recall-oriented intuition of fast thinking with the precision-oriented selectivity of slow thinking. We further provide theoretical justification for this synergy, showing that it mitigates variance-induced interference during optimization. Extensive experiments on MER-UniBench and MME-Emotion show that MER-R1 achieves state-of-the-art performance and makes reasoning genuinely benefit emotion recognition.

Chinese Translation

我们发现，尽管显式推理使预测更具可解释性，但并不一定能转化为更好的多模态情感识别（MER）准确性。具体而言，对于基于推理的多模态大语言模型（MLLMs），通过直接回答触发的快速思维往往优于经过深思熟虑的慢速思维。我们的实证分析表明，快速思维通过更广泛和更自信的预测提高了召回率，而慢速思维则通过对错误类别的保守过滤来提高精确度。基于这些洞察，我们提出了MER-R1，一个强化学习框架，将慢速-快速互补性转化为显式优化。双目标解耦将召回率和精确度分离为两个优化信号，使它们能够联合优化，而不是相互权衡。慢速-快速置信度校准进一步将最终的慢速思维答案与快速思维直觉对齐，增强正确情感的同时抑制错误情感。通过这种方式，MER-R1将快速思维的召回导向直觉与慢速思维的精确导向选择性统一起来。我们进一步提供了这种协同作用的理论依据，表明它减轻了优化过程中由方差引起的干扰。在MER-UniBench和MME-Emotion上的大量实验表明，MER-R1实现了最先进的性能，并使推理真正惠及情感识别。

View on arXiv Download PDF AI Translation

cs.AI / 7 / 2606.27736

ToE: A Hierarchical and Explainable Claim Verification Framework with Dynamic Multi-source Evidence Retrieval and Aggregation

ToE：一种具有动态多源证据检索与聚合的分层可解释声明验证框架

Wang, Zhaoqi, Zhang, Zijian, Zheng, Kun, Li, Zhen, Li, Xin, Li, Chunlei, Liu, Jiamou

Abstract

The rapid spread of fake news poses increasing threats to information ecosystems, especially as AI-generated misinformation under Generative Engine Optimization (GEO) poisoning allows adversarially crafted content to be systematically surfaced by retrieval systems, contaminating LLM reasoning. In this paper, we propose Tree of Evidence (ToE), a hierarchical evidence reasoning framework for automated fact-checking that models each claim as a dynamically expanding argument tree. ToE integrates a reinforcement learning-driven multi-source retrieval agent, an evidence evaluation agent, and an argument tree aggregation algorithm to iteratively decompose, retrieve, and verify claims through an explainable evidence chain. We further provide a theoretical analysis of the retrieval process, deriving a formal error bound that guarantees the learned policy converges to a neighborhood of the information-theoretically optimal policy. Experiments across multiple datasets and backbone LLMs demonstrate that ToE achieves improvements ranging from 4 to 24 percentage points over competitive baselines, with particularly pronounced gains on adversarially poisoned inputs.

Chinese Translation

假新闻的快速传播对信息生态系统构成了日益严重的威胁，尤其是在生成引擎优化（Generative Engine Optimization, GEO）下，人工智能生成的虚假信息使得对抗性内容能够被检索系统系统性地呈现，从而污染大型语言模型（LLM）的推理。在本文中，我们提出了证据树（Tree of Evidence, ToE），这是一种用于自动事实核查的分层证据推理框架，将每个声明建模为动态扩展的论证树。ToE集成了一个基于强化学习的多源检索代理、一个证据评估代理以及一个论证树聚合算法，通过可解释的证据链迭代地分解、检索和验证声明。我们进一步提供了检索过程的理论分析，推导出一个形式化的误差界限，保证所学习的策略收敛到信息论上最优策略的邻域。针对多个数据集和基础大型语言模型的实验表明，ToE在竞争基线之上实现了4到24个百分点的改进，尤其在对抗性污染输入上取得了显著的提升。

View on arXiv Download PDF AI Translation

cs.AI / 8 / 2606.27757

Towards Reliable and Robust LLM Planning: Symbolic Feedback-Driven Iterative Self-Refinement Framework

迈向可靠且稳健的大型语言模型规划：基于符号反馈的迭代自我精炼框架

Zhang, Jiajing, Jiang, Jiamei, Zhang, Chenyang, Mo, Feifei, Li, Linjing, Zeng, Daniel

Abstract

Large language models (LLMs) have attracted widespread attention from academia and industry, yet their deployment raises critical security concerns regarding robustness and reliability. Planning, a core component of intelligent behavior, remains challenging for LLMs, which often produce infeasible or incorrect solutions in long-horizon decision-making tasks due to inherent complexity. In this paper, we propose a symbolic feedback-driven iterative self-refinement framework to enhance the robustness and reliability of LLMs in long-horizon planning. Specifically, a natural language prompting mechanism is introduced to map logical symbols into natural language descriptions, enabling LLMs to better capture task constraints and semantics. We further design a symbolic verifier that identifies errors and converts them into corrective instructions interpretable by the LLM, thereby guiding self-refinement. In addition, we leverage a plan recognizer to infer goal reachability, facilitating more effective guidance toward desired goals. Empirical results demonstrate that the proposed framework consistently improves both feasibility and correctness in long-horizon planning tasks. This highlights its effectiveness in enhancing the reliability of LLM-based planning and potential to enable more trustworthy AI systems.

Chinese Translation

大型语言模型（LLMs）已引起学术界和工业界的广泛关注，但其部署引发了关于稳健性和可靠性的重大安全隐患。规划作为智能行为的核心组成部分，对于LLMs而言仍然具有挑战性，因为它们在长时间决策任务中常常产生不可行或不正确的解决方案，这主要是由于固有的复杂性。本文提出了一种基于符号反馈的迭代自我精炼框架，以增强LLMs在长时间规划中的稳健性和可靠性。具体而言，我们引入了一种自然语言提示机制，将逻辑符号映射为自然语言描述，使LLMs能够更好地捕捉任务约束和语义。我们进一步设计了一种符号验证器，识别错误并将其转换为LLM可解释的纠正指令，从而指导自我精炼。此外，我们利用计划识别器推断目标可达性，以便更有效地引导实现期望目标。实证结果表明，所提出的框架在长时间规划任务中始终提高了可行性和正确性。这突显了其在增强基于LLM的规划可靠性方面的有效性，并具有促进更可信赖的人工智能系统的潜力。

View on arXiv Download PDF AI Translation

cs.AI / 9 / 2606.27780

Understanding Rollout Error in Graph World Models

理解图世界模型中的展开误差

Song, Xinyuan, Cai, Zekun

Abstract

World models are often used for planning by rolling learned dynamics forward. Many planning environments, however, are not vectors or images; they are graphs of agents, tools, skills, routes, and dependencies. In these settings, a local prediction error may stay local or spread through the graph, and the failure mode changes again when edges are predicted rather than fixed. This paper studies long-horizon rollout error in Graph World Models (GWMs). We formulate a unified fixed-edge and dynamic-edge GWM framework with action nodes for node-, edge-, and graph-level decisions. We develop graph-valued rollout bounds that separate topology-induced amplification from model-induced amplification, and we introduce a joint node-edge operator for dynamic-edge rollouts. Guided by the analysis, we propose Error-Aware GWM, which combines spectral regularization, rollout consistency, and critical-node weighting. Across synthetic topologies and heterogeneous agent-graph testbeds, rollout error and planning regret grow with horizon, dynamic-edge training is needed when structure evolves, and Error-Aware GWM prevents long-horizon divergence while preserving prediction accuracy. Real-world graph benchmarks clarify the scope of GWMs: they are most useful for dynamic graph rollout and agent planning, while specialized graph models remain strong on static or sparse prediction tasks.

Chinese Translation

世界模型通常通过向前滚动学习到的动态来进行规划。然而，许多规划环境并不是向量或图像；它们是由代理、工具、技能、路线和依赖关系构成的图。在这些环境中，局部预测误差可能保持局部或通过图传播，当边缘被预测而不是固定时，失败模式又会发生变化。本文研究了图世界模型（Graph World Models, GWM）中的长时间展望展开误差。我们提出了一个统一的固定边缘和动态边缘GWM框架，包含用于节点、边缘和图级决策的动作节点。我们开发了图值展开界限，将拓扑引起的放大与模型引起的放大区分开，并引入了一个用于动态边缘展开的联合节点-边缘算子。在分析的指导下，我们提出了错误感知GWM（Error-Aware GWM），该模型结合了谱正则化、展开一致性和关键节点加权。在合成拓扑和异构代理图测试平台上，展开误差和规划遗憾随着时间展望的增加而增长，当结构演变时需要动态边缘训练，而错误感知GWM在保持预测准确性的同时防止了长时间展望的发散。真实世界的图基准测试阐明了GWM的适用范围：它们在动态图展开和代理规划中最为有用，而专门的图模型在静态或稀疏预测任务中仍然表现强劲。

View on arXiv Download PDF AI Translation

cs.AI / 10 / 2606.27806

Grounded Iterative Language Planning: How Parameterized World Models Reduce Hallucination Propagation in LLM Agents

基于参数化世界模型的迭代语言规划：如何减少大型语言模型代理中的幻觉传播

Song, Xinyuan, Cai, Zekun

Abstract

World models for language agents come in two useful forms. An agent-based world model calls an LLM API and reasons flexibly in language, but its errors appear as hallucinated state changes that are hard to score with ordinary regression losses. A parameterized world model is a trained transition predictor; its errors are easier to measure with quantities such as NodeMSE, delta accuracy, and validity accuracy, but it is usually weaker as a standalone planner. We compare these two families on four graph-structured planning benchmarks and introduce operational hallucination metrics for the agent-based case. The comparison motivates \textbf{Grounded Iterative Language Planning} (GILP), which trains only a small parameterized backbone and combines it with API-based agent reasoning. The backbone supplies valid actions, predicted state deltas, risk, and value; the LLM drafts an action and imagined delta; and a consistency gate asks for revision when the two disagree. On real GPT-4o-mini calls, GILP reduces hallucinated-state rate from 0.176 to 0.035. In calibrated simulator ablations, it raises success from 0.668 to 0.838 while adding only ~22% extra LLM calls.

Chinese Translation

语言代理的世界模型有两种有用的形式。基于代理的世界模型调用大型语言模型（LLM）API，并以灵活的方式进行语言推理，但其错误表现为难以用普通回归损失评分的幻觉状态变化。参数化世界模型是一个训练好的转移预测器；其错误可以通过节点均方误差（NodeMSE）、增量准确性（delta accuracy）和有效性准确性（validity accuracy）等量度更容易地进行测量，但作为独立规划器通常较弱。我们在四个图结构规划基准上比较这两种模型，并为基于代理的情况引入了操作性幻觉度量。比较结果激励了 extbf{基于参数化的迭代语言规划}（GILP），该方法仅训练一个小型的参数化骨干，并将其与基于API的代理推理相结合。骨干提供有效的动作、预测的状态增量、风险和价值；LLM起草一个动作和想象的增量；当两者不一致时，一致性门控会要求修正。在真实的GPT-4o-mini调用中，GILP将幻觉状态率从0.176降低到0.035。在经过校准的模拟器消融实验中，它将成功率从0.668提高到0.838，同时仅增加约22%的LLM调用。

View on arXiv Download PDF AI Translation

cs.AI / 11 / 2606.27814

ATOD: Annealed Turn-aware On-policy Distillation for Multi-turn Autonomous Agents

ATOD：用于多轮自主智能体的退火轮次感知在线蒸馏

Tan, Qitai, Zong, Zefang, Li, Yang, Chen, Peng

Abstract

Training small language-model agents for long-horizon interactive tasks requires both fast imitation and reward-driven improvement. On-policy distillation (OPD) provides dense teacher guidance and typically improves rapidly in the early stage, but its gains saturate once the student approaches the teacher, limiting the final performance ceiling. Reinforcement learning (RL) directly optimizes environment rewards and encourages exploratory improvement toward a higher reward-defined ceiling, but sparse and delayed feedback makes early-stage learning much less efficient than OPD. In this paper, we propose ATOD (Annealed Turn-aware On-policy Distillation), a hybrid online distillation algorithm that explicitly exploits this complementarity. (1) ATOD uses an annealed OPD-RL schedule: OPD dominates early training to approach teacher-level behavior, while RL is gradually strengthened to drive reward-based exploration. (2) ATOD introduces Turn-level Disagreement-Uncertainty Reweighting (T-DUR), which softly amplifies high-utility turns and improves dense supervision in long trajectories. Experiments on ALFWorld, WebShop, and Search-QA show that ATOD consistently outperforms competing post-training baselines: across the three student sizes, ATOD improves average success rate by 3.03 points over OPD and 23.62 points over GRPO, while surpassing the corresponding teacher models by 2.16 points.

Chinese Translation

训练小型语言模型智能体以应对长期交互任务需要快速模仿和基于奖励的改进。在线政策蒸馏（OPD）提供了密集的教师指导，通常在早期阶段迅速改善，但一旦学生接近教师，其收益便会饱和，限制了最终的性能上限。强化学习（RL）直接优化环境奖励，并鼓励朝向更高奖励定义的上限进行探索性改进，但稀疏和延迟的反馈使得早期学习的效率远低于OPD。在本文中，我们提出了ATOD（Annealed Turn-aware On-policy Distillation），一种混合在线蒸馏算法，明确利用了这种互补性。(1) ATOD使用退火的OPD-RL调度：OPD在早期训练中占主导地位，以接近教师级别的行为，而RL逐渐增强以推动基于奖励的探索。(2) ATOD引入了轮次级不一致性-不确定性重加权（T-DUR），该方法柔和地放大高效用的轮次，并改善长轨迹中的密集监督。在ALFWorld、WebShop和Search-QA上的实验表明，ATOD始终优于竞争的后训练基线：在三种学生规模中，ATOD的平均成功率比OPD提高了3.03个百分点，比GRPO提高了23.62个百分点，同时超越了相应的教师模型2.16个百分点。

View on arXiv Download PDF AI Translation

cs.AI / 12 / 2606.27826

NormAct: A Benchmark for Hidden Social Norm Compliance in Embodied Planning

NormAct：隐性社会规范遵从的具身规划基准

Zhao, Shiyun, Song, Xinwei, Guo, Tianyu, Gao, Xiaomeng, Liu, Mingyuan, Han, Xu, Zhang, Yuanyuan, Zhang, Zhenliang, Feng, Xue, Dai, Bo

Abstract

Multimodal large language models (MLLMs) are increasingly deployed as embodied planners in egocentric environments, where task success requires not only achieving instructed goals but also acting in socially appropriate ways. While explicit goals may render certain actions optimal, implicit social norms often impose hidden constraints. Existing evaluations typically focus on explicit goal achievement or direct norm knowledge, seldom assessing whether planners can infer and apply these hidden constraints within action sequences. We introduce NormAct, a benchmark for embodied social-norm interactions that evaluates plans on Goal Achievement, Norm Compliance, and overall Task Success. NormAct uniquely embeds hidden norms within ordinary tasks, testing whether models can realize them without explicit instruction. Experiments with state-of-the-art MLLMs (GPT-5.4, Claude Opus 4.7, Gemini 3 Pro) reveal a significant gap: models achieve explicit goals in 67.3\% of cases, but comply with hidden norms in only 26.4\%. Cue-condition experiments indicate that this gap stems not from a lack of general social knowledge, but from challenges in activating and grounding relevant norms in context. To address this, we propose NormPerceptor, a context-conditioned cue generator that infers scene-relevant norms prior to planning, increasing Task Success from 24.2\% to 46.7\%. Our results underscore the importance of enabling embodied agents to proactively detect hidden norms, ground them in visual evidence, and integrate them as action-planning constraints. Our benchmark is publicly available at https://huggingface.co/datasets/Caleb196x/NormAct.

Chinese Translation

多模态大型语言模型（MLLMs）越来越多地被应用于自我中心环境中的具身规划，其中任务成功不仅需要实现指令目标，还需要以社会适当的方式行动。虽然显性目标可能使某些行为变得最优，但隐性社会规范往往施加隐藏约束。现有评估通常集中在显性目标的实现或直接的规范知识上，鲜有评估规划者是否能够推断并应用这些隐藏约束于行动序列中。我们提出了NormAct，这是一个评估具身社会规范交互的基准，评估计划的目标实现、规范遵从和整体任务成功。NormAct独特地将隐藏规范嵌入普通任务中，测试模型是否能够在没有显性指令的情况下实现这些规范。与最先进的MLLM（GPT-5.4、Claude Opus 4.7、Gemini 3 Pro）的实验显示出显著差距：模型在67.3%的情况下实现显性目标，但仅在26.4%的情况下遵从隐性规范。线索条件实验表明，这一差距并非源于缺乏一般社会知识，而是由于在上下文中激活和落实相关规范的挑战。为了解决这一问题，我们提出了NormPerceptor，一个上下文条件的线索生成器，它在规划之前推断场景相关的规范，将任务成功率从24.2%提高到46.7%。我们的结果强调了使具身代理能够主动检测隐藏规范、将其与视觉证据结合并将其整合为行动规划约束的重要性。我们的基准数据集已在 https://huggingface.co/datasets/Caleb196x/NormAct 上公开发布。

View on arXiv Download PDF AI Translation

cs.AI / 13 / 2606.27926

Verifiable Geometry Problem Solving: Solver-Driven Autoformalization and Theorem Proposing

可验证的几何问题求解：求解器驱动的自动形式化与定理提出

Li, Can, Zhang, Ting, Zhao, Junbo, Huang, Hua

Abstract

Geometry Problem Solving have increasingly adopt the neuro-symbolic paradigm, combining neural intuition with symbolic rigor. However, current frameworks suffer from severe bottlenecks in two core stages: autoformalization, which treats multimodal translation as a static task decoupled from downstream solver compatibility, and theorem prediction, where solvers frequently hit a deductive impasse due to fixed rule libraries. To address these, we propose SD-GPS, a solver-driven framework that treats the symbolic solver as an execution oracle throughout both formalization and deduction. First, Solver-Driven Autoformalization unifies supervised formal-language adaptation and solvability-guided reinforcement learning into a single module built on QwenVL3-2B, making executability the central training signal. Second, Verified Theorem Proposing introduces an impasse-aware agent that proposes local auxiliary lemmas from current proof states, ensuring soundness by filtering all proposals through symbolic verification. Empirical evaluations on Geometry3K and PGPS9K demonstrate that SD-GPS consistently outperforms existing MLLM, neural, and neuro-symbolic methods across standard completion, multiple-choice, and cross-modal reference regimes, proving that closing the loop between multimodal perception and symbolic execution significantly improves geometric reasoning, offering profound insights into how neural agents can be grounded by formal systems to achieve verifiable problem-solving capabilities.

Chinese Translation

几何问题求解越来越多地采用神经符号范式，将神经直觉与符号严谨性相结合。然而，当前框架在两个核心阶段存在严重瓶颈：自动形式化将多模态翻译视为与下游求解器兼容性解耦的静态任务，以及定理预测中，由于固定规则库，求解器经常陷入演绎僵局。为了解决这些问题，我们提出了SD-GPS，一个求解器驱动的框架，将符号求解器视为形式化和推理过程中的执行神谕。首先，求解器驱动的自动形式化将监督形式语言适应和可解性引导的强化学习统一为一个基于QwenVL3-2B构建的单一模块，使可执行性成为中心训练信号。其次，验证的定理提出引入了一个意识到僵局的代理，从当前证明状态中提出局部辅助引理，通过符号验证过滤所有提案以确保健全性。在Geometry3K和PGPS9K上的实证评估表明，SD-GPS在标准完成、多项选择和跨模态参考机制中始终优于现有的MLLM、神经和神经符号方法，证明了多模态感知与符号执行之间的闭环显著提高了几何推理，为神经代理如何通过形式系统实现可验证的问题解决能力提供了深刻的见解。

View on arXiv Download PDF AI Translation

cs.AI / 14 / 2606.27967

RelBall: Relation Ball with Quaternion Rotation for Knowledge Graph Completion

RelBall：基于四元数旋转的关系球用于知识图谱补全

Liu, Yike, Xie, Peijia, He, Chao, Zhu, Huiling

Abstract

Real-world knowledge graphs are often incomplete, lacking many valid facts. Knowledge Graph Completion (KGC) aims to predict missing links using known triples, thereby enhancing graph coverage. A key challenge is modeling diverse relational patterns such as symmetry, antisymmetry, inversion, composition and semantic hierarchy. Existing models such as RotatE can capture symmetric, antisymmetric, inverse, and commutative composition patterns, yet struggle with non-commutative composition. Rotate3D addresses this by introducing non-commutativity via three-dimensional rotations, but still fails to capture the semantic hierarchies prevalent in knowledge graphs. Moreover, both models cannot effectively model one-to-many relations. To overcome these limitations, we propose RelBall, which extends Rotate3D with two innovations. First, our model introduces modulus transformation to model hierarchies, driving abstract concepts toward smaller moduli and concrete instances toward larger ones. Second, it introduces a tail-centric relation ball to model one-to-one, one-to-many, many-to-one, and many-to-many relations. RelBall offers the following advantages: (1) coverage of all relational patterns, including the ones mentioned above; (2) an interpretable hierarchical representation where the modulus directly reflect semantic levels; (3) support for one-to-one, one-to-many, many-to-one, and many-to-many relations. Experiments on multiple datasets demonstrate RelBall's competitive link prediction performance against various baselines.

Chinese Translation

现实世界中的知识图谱往往是不完整的，缺乏许多有效的事实。知识图谱补全（Knowledge Graph Completion, KGC）旨在利用已知三元组预测缺失的链接，从而增强图谱的覆盖率。一个关键挑战是建模多样的关系模式，如对称性、反对称性、反转、组合和语义层次。现有模型如 RotatE 能够捕捉对称、反对称、逆和可交换组合模式，但在处理非可交换组合时表现不佳。Rotate3D 通过引入三维旋转来解决这一问题，增加了非可交换性，但仍未能有效捕捉知识图谱中普遍存在的语义层次。此外，这两种模型都无法有效建模一对多关系。为克服这些局限性，我们提出了 RelBall，它通过两项创新扩展了 Rotate3D。首先，我们的模型引入模变换以建模层次结构，将抽象概念驱动向较小的模，而将具体实例驱动向较大的模。其次，它引入了以尾部为中心的关系球，以建模一对一、一对多、多对一和多对多关系。RelBall 提供了以下优势：（1）覆盖所有关系模式，包括上述模式；（2）可解释的层次表示，其中模直接反映语义层次；（3）支持一对一、一对多、多对一和多对多关系。在多个数据集上的实验表明，RelBall 在链接预测性能上与各种基线相比具有竞争力。

View on arXiv Download PDF AI Translation

cs.AI / 15 / 2606.28024

Lifted Causal Inference

提升因果推断

Luttermann, Malte, Braun, Tanya, Möller, Ralf, Gehrke, Marcel

Abstract

Lifted inference exploits indistinguishabilities in probabilistic graphical models by using a representative for indistinguishable objects, thereby speeding up query answering while maintaining exact answers. In this article, we show how lifting can be applied to efficiently compute causal effects in relational domains. More specifically, we introduce parametric causal factor graphs (PCFGs) to incorporate causal knowledge in lifted models and give a formal semantics of interventions therein. We further present the Lifted Causal Inference (LCI) algorithm to compute causal effects on a lifted level, thereby drastically speeding up causal inference compared to propositional inference, e.g., in causal Bayesian networks. In addition, we present partially directed parametric causal factor graphs (PD-PCFGs) as a generalisation of PCFGs to handle partial causal knowledge and extend LCI to perform lifted causal inference in a PD-PCFG, thereby extending the applicability of lifted causal inference to a broader range of models requiring less prior knowledge about causal relationships.

Chinese Translation

提升推断通过使用不可区分对象的代表，利用概率图模型中的不可区分性，从而加速查询回答，同时保持精确答案。在本文中，我们展示了如何将提升应用于有效计算关系领域中的因果效应。更具体地，我们引入了参数化因果因子图（Parametric Causal Factor Graphs, PCFGs），以在提升模型中融入因果知识，并给出了干预的形式语义。我们进一步提出了提升因果推断（Lifted Causal Inference, LCI）算法，以在提升层面计算因果效应，从而与命题推断相比，显著加快了因果推断的速度，例如在因果贝叶斯网络中。此外，我们提出了部分定向参数化因果因子图（Partially Directed Parametric Causal Factor Graphs, PD-PCFGs），作为PCFGs的推广，以处理部分因果知识，并扩展LCI以在PD-PCFG中执行提升因果推断，从而将提升因果推断的适用性扩展到更广泛的模型，减少对因果关系的先验知识要求。

View on arXiv Download PDF AI Translation

cs.AI / 16 / 2606.28070

JD Oxygen AI Item Center (Oxygen AIIC) V1: An Industrial-Scale LLM/VLM-Centric Solution for Item Understanding, Management, and Applications

JD氧气人工智能商品中心（Oxygen AIIC）V1：一种面向商品理解、管理和应用的工业规模LLM/VLM中心解决方案

AIIC, Oxygen, Long, Chan, Liu, Chao, Chen, Chaofan, Dong, Chaohui, Guo, Chunyuan, Liu, Danping, Liu, Debin, Xiang, Deping, Xu, Fulai, Liu, Guangyue, Li, Hao, Hu, Huichun, Yang, Jian, Wang, Jianan, Zhao, Jianbo, Li, Jiaoyang, Wang, Jiaxing, Li, Jinglong, Guo, Jinjin, Fang, Jun, Liu, Jun, Zhou, Kai, Wang, Li, Gao, Lili, Chen, Liying, Yang, Luning, Zhou, Mengdi, Liu, Pengzhang, Lv, Qi, Wang, Qianyun, Jiang, Qixia, Li, Ruyue, Liang, Shimu, Wang, Shuxing, Zhang, Sijie, Li, Siqi, Gao, Tianhao, Ke, Wang, Huang, Weihu, Lai, Wencan, Zhang, Wenjie, Zhang, Xiaohui, Dong, Xiaojing, Liu, Ya, Zhang, Yifeng, Wang, Yixiang, Zhang, Yongtai, Liao, Yongyi, Chen, Zhaoru, Chen, Zhen, Ma, Zhiyong, Liu, Zhiyuan, Liu, Zhongwei, Xing, Ziyan

Abstract

JD.com, one of the world's largest e-commerce platforms, serves over 700 million active users and millions of merchants, with a catalog of tens of billions of SKUs. At this scale, high-quality, structured item knowledge underpins a better consumer experience, lower management costs, and higher operational efficiency-yet producing and serving it poses three industrial-scale challenges: fast-emerging concepts, high-quality knowledge production for massive SKUs, and diverse downstream requirements. To address these challenges, we present the JD Oxygen AI Item Center (Oxygen AIIC), an industrial-scale platform built on LLMs/VLMs for item-knowledge production and service. Oxygen AIIC is built around four core pillars: (i) ontology engineering driven by efficient human-AI collaboration, which supports the dynamic evolution and agile expansion of an ontology with millions of entries; (ii) a "Semantic Search then Discrimination"(S2D) knowledge identification architecture that, combined with throughput improvement strategies, enables scalable, extensible, and high-throughput AI Item Library production for tens of billions of SKUs; (iii) self-evolving item-understanding LLMs/VLMs that improve in a stable and controllable manner, enabling knowledge production with 94.2% precision and 82.8% recall; and (iv) a unified item tunnel that serves as the data and service hub. Oxygen AIIC now covers tens of thousands of JD categories and processes hundreds of millions of item updates per day on Huawei Ascend NPUs. It has accumulated hundreds of billions of item-knowledge assets. Deployed across core business scenarios-including search, recommendation, operations, category planning-Oxygen AIIC has delivered measurable gains at scale. Search-traffic coverage reaches 80.4%, item-information quality issues drop by 37%, the automated fill rate of core attributes during item listing exceeds 80%.

Chinese Translation

京东（JD.com）是全球最大的电子商务平台之一，服务超过7亿活跃用户和数百万商家，拥有数十亿SKU的商品目录。在如此规模下，高质量、结构化的商品知识是提升消费者体验、降低管理成本和提高运营效率的基础，但生产和提供这些知识面临三个工业规模的挑战：快速出现的新概念、大规模SKU的高质量知识生产以及多样化的下游需求。为了解决这些挑战，我们提出了JD氧气人工智能商品中心（Oxygen AIIC），这是一个基于LLM/VLM的工业规模平台，用于商品知识的生产和服务。Oxygen AIIC围绕四个核心支柱构建：（i）由高效的人机协作驱动的本体工程，支持具有数百万条目的本体的动态演变和敏捷扩展；（ii）一种“语义搜索后鉴别”（Semantic Search then Discrimination, S2D）知识识别架构，结合吞吐量提升策略，使得数十亿SKU的AI商品库生产具备可扩展性、可延展性和高吞吐量；（iii）自我演化的商品理解LLM/VLM，能够以稳定和可控的方式提升，知识生产的精确率达到94.2%，召回率达到82.8%；（iv）统一的商品通道，作为数据和服务的中心。Oxygen AIIC目前覆盖了数万个JD类目，每天处理数亿次商品更新，运行在华为Ascend NPU上。它已积累了数百亿的商品知识资产。Oxygen AIIC已在核心业务场景中部署，包括搜索、推荐、运营和类目规划，并在规模上实现了可测量的收益。搜索流量覆盖率达到80.4%，商品信息质量问题下降37%，商品上架时核心属性的自动填充率超过80%。

View on arXiv Download PDF AI Translation

cs.AI / 17 / 2606.28076

Ontology-Guided Evidence Path Inference for Multi-hop Knowledge Graph Question Answering

基于本体引导的多跳知识图谱问答证据路径推理

Shan, Yongxue, Wu, Meihan, Fang, Cundi, Peng, Jie, Wang, Xiaodong

Abstract

Knowledge graph question answering (KGQA) aims to answer natural-language questions by reasoning over structured facts. Existing multi-hop KGQA methods mainly rely on topic-centered expansion, which faces two key challenges: the search space rapidly grows with noisy mixed-type paths, and retrieved paths may fail to satisfy the semantic constraints of complex questions. To address these challenges, we propose OPI, an ontology-guided evidence path inference framework for multi-hop KGQA. OPI introduces a relation-centric ontology graph to capture the head-tail type constraints of relations, providing a compact interface for answer-side constraints. Based on this ontology graph, OPI first introduces a bidirectional retrieval mechanism by mapping the predicted answer type to compatible final-hop relations and combining topic-side prefix expansion with answer-side final-hop matching, thereby suppressing noisy mixed-type expansion. OPI further adopts an iterative refinement strategy to reassess retrieved paths and candidate answers under the question context, filtering type-compatible but question-irrelevant evidence for more reliable answer prediction. Experiments on WebQSP, CWQ, and MetaQA show that OPI substantially reduces the search space, improves Hit@1/F1 by 4.6/5.0 points on WebQSP and 8.9/3.3 points on CWQ over the strongest prior results, and achieves near-saturated Hit@1 on MetaQA with the retrieval module alone.

Chinese Translation

知识图谱问答（KGQA）旨在通过对结构化事实的推理来回答自然语言问题。现有的多跳 KGQA 方法主要依赖于以主题为中心的扩展，这面临两个关键挑战：搜索空间随着噪声混合路径的增加而迅速扩大，且检索到的路径可能无法满足复杂问题的语义约束。为了解决这些挑战，我们提出了 OPI，一个基于本体引导的多跳 KGQA 证据路径推理框架。OPI 引入了一个以关系为中心的本体图，以捕捉关系的头尾类型约束，为答案侧约束提供了一个紧凑的接口。基于该本体图，OPI 首先通过将预测的答案类型映射到兼容的最终跳跃关系，并结合主题侧前缀扩展与答案侧最终跳跃匹配，引入了双向检索机制，从而抑制噪声混合扩展。OPI 进一步采用迭代精炼策略，在问题上下文中重新评估检索路径和候选答案，过滤类型兼容但与问题无关的证据，以实现更可靠的答案预测。在 WebQSP、CWQ 和 MetaQA 上的实验表明，OPI 显著减少了搜索空间，在 WebQSP 上将 Hit@1/F1 提高了 4.6/5.0 分，在 CWQ 上提高了 8.9/3.3 分，相较于最强的先前结果，并在仅使用检索模块的情况下在 MetaQA 上达到了近饱和的 Hit@1。

View on arXiv Download PDF AI Translation

cs.AI / 18 / 2606.28126

AI-Driven Synthesis for High-Tech System Design: Automating Innovation

基于人工智能的高科技系统设计合成：创新的自动化

Oerlemans, Luuk, Westerhof, Steven, Hofman, Theo

Abstract

This article addresses the combinatorial complexity inherent in modern high-tech system design by presenting automation-in-design (AiD) as a transformative paradigm. We propose computational design synthesis (CDS), a framework utilising deep learning and generative AI to automate the creation of novel systems. Two case studies (e-drive system design and spatial dimensioning problem) serve as proof-points for this approach. The AI-driven methods used in the case studies represent a fundamental shift in engineering, advancing from simulation-based optimisation towards autonomous design with minimal human supervision.

Chinese Translation

本文通过提出设计自动化（AiD）作为一种变革性范式，解决了现代高科技系统设计中固有的组合复杂性。我们提出了计算设计合成（CDS）框架，利用深度学习和生成式人工智能来自动化新系统的创建。两个案例研究（电驱动系统设计和空间尺寸问题）作为该方法的证明点。案例研究中使用的基于人工智能的方法代表了工程领域的根本转变，从基于仿真的优化进展到在最小人类监督下的自主设计。

View on arXiv Download PDF AI Translation

cs.AI / 19 / 2606.28166

Tandem Reinforcement Learning with Verifiable Rewards

具有可验证奖励的串联强化学习

Jiao, Difan, Singhal, Raghav, West, Robert, Anderson, Ashton

Abstract

Reinforcement learning with verifiable rewards (RLVR) has significantly improved the reasoning capability of large language models, reaching expert or even superhuman performance in domains such as competition math. However, whether weaker agents and humans can actually harness this capability is far less certain, with RLVR documented to drift reasoning toward idiosyncratic patterns such as poor readability and language mixing. Tandem training is a recently introduced paradigm that targets this compatibility problem: a trained, stronger senior co-generates each rollout with a frozen, weaker junior, and the two are rewarded as a team, so the senior is pushed to reason in ways the junior can follow. Yet this paradigm has so far been demonstrated only in proof-of-concept settings, leaving open whether it scales to the long chains of thought of the modern RLVR pipeline. In this work, we propose Tandem Reinforcement Learning (TRL), which carries the tandem training paradigm into RLVR. In TRL, the senior and a frozen junior alternate stochastically to co-generate the reasoning, the resulting generation is rewarded, and the standard GRPO loss is applied to the senior. Training Qwen3-4B-Instruct on competition math, we find that TRL matches vanilla GRPO on solo reasoning capability while three properties emerge together from the same rollout structure: stronger handoff robustness with the junior, reduced distributional drift from the junior, and a chain-of-thought more legible to the junior. Our results demonstrate a promising route for RLVR with practical payoffs in multi-model communication and human compatibility.

Chinese Translation

具有可验证奖励的强化学习（RLVR）显著提升了大型语言模型的推理能力，在竞争数学等领域达到了专家甚至超人水平的表现。然而，较弱的智能体和人类是否能够真正利用这一能力仍然不确定，RLVR被记录为倾向于向个性化模式漂移，例如可读性差和语言混合。串联训练是一种最近提出的范式，旨在解决这一兼容性问题：经过训练的强大高级智能体与一个被冻结的较弱初级智能体共同生成每次推理，并作为一个团队获得奖励，从而推动高级智能体以初级智能体能够理解的方式进行推理。然而，这一范式迄今仅在概念验证环境中得到展示，尚不清楚它是否能够扩展到现代RLVR管道的长推理链。在本研究中，我们提出了串联强化学习（TRL），将串联训练范式引入RLVR。在TRL中，高级智能体和一个被冻结的初级智能体以随机方式交替共同生成推理，生成的结果获得奖励，并对高级智能体应用标准的GRPO损失。在竞争数学上训练Qwen3-4B-Instruct时，我们发现TRL在单独推理能力上与普通GRPO相当，同时从相同的推理结构中出现了三个特性：与初级智能体的更强交接鲁棒性、来自初级智能体的分布漂移减少，以及更易于初级智能体理解的推理链。我们的结果展示了RLVR在多模型通信和人类兼容性方面的一个有前景的途径，具有实际收益。

View on arXiv Download PDF AI Translation

cs.AI / 20 / 2606.28270

Agent-Native Immune System: Architecture, Taxonomy, and Engineering

代理原生免疫系统：架构、分类与工程

Shen, Bo, Chang, Lifeng, Wei, Tianyuan, Li, Yunpeng, Shi, Feng, Han, Yichen, Gao, Peijie, Kuang, Shiyi, Chang, Xin, Li, Dehui

Abstract

The transition from static chat bots to autonomous agents--equipped with persistent memory, tool-use protocols, and multi-agent collaboration--has fundamentally expanded the AI threat landscape. Current defense mechanisms, such as perimeter security and training-time alignment, remain external to the agent's active reasoning loop. Consequently, they fall short: a fully aligned agent remains highly vulnerable to runtime hijacking via memory poisoning, tool-chain manipulation, or multi-agent protocol attacks. To address this critical gap, we introduce the Agent-Native Immune System (ANIS), the first biologically inspired, endogenous defense architecture embedded directly within the agent's cognitive loop. Our framework presents four primary contributions. First, we design a six-layer Immune Tower (L0-L5), distinctly incorporating Barrier Immunity (L1) as a non-cognitive, physical-and-logical isolation layer. Second, we establish a unified taxonomy of Agent Viruses and Agent Vaccines, formalizing the critical distinction between superficial non-parametric defenses and robust parametric vaccines. Third, we conceptualize the Harness Triad--Meta, Self, and Auto--a self-monitoring, meta-cognitive automation backbone that drives Continual Immune Learning (CIL), enabling vaccines to dynamically adapt to novel threats. Finally, we establish a rigorous theoretical demarcation between model alignment and agent immunity: while alignment provides a static "constitutional" value foundation during training, ANIS serves as the dynamic "law enforcement" mechanism during runtime. We conclude by framing open challenges for the field, including immune protocol standardization, novel evaluation metrics such as the Autoimmunity Rate (false-positive intervention rate), and the co-evolutionary dynamics between pathogens and vaccines within collective intelligence ecosystems.

Chinese Translation

从静态聊天机器人到具备持久记忆、工具使用协议和多代理协作的自主代理的转变，根本上扩展了人工智能的威胁格局。当前的防御机制，如周界安全和训练时对齐，仍然处于代理的主动推理循环之外。因此，它们无法有效应对：一个完全对齐的代理仍然高度易受运行时劫持的影响，例如内存中毒、工具链操控或多代理协议攻击。为了解决这一关键缺口，我们提出了代理原生免疫系统（Agent-Native Immune System, ANIS），这是第一个生物启发的、内生的防御架构，直接嵌入代理的认知循环中。我们的框架提出了四个主要贡献。首先，我们设计了一个六层免疫塔（Immune Tower, L0-L5），明确将屏障免疫（Barrier Immunity, L1）作为非认知的、物理与逻辑隔离层。其次，我们建立了一个统一的代理病毒和代理疫苗分类法，正式区分了表面非参数防御和稳健的参数疫苗之间的关键区别。第三，我们构思了三重支架（Harness Triad）——元（Meta）、自我（Self）和自动（Auto）——一个自我监控的元认知自动化基础，驱动持续免疫学习（Continual Immune Learning, CIL），使疫苗能够动态适应新威胁。最后，我们建立了模型对齐与代理免疫之间的严格理论划分：虽然对齐在训练期间提供了静态的“宪法”价值基础，但ANIS在运行时则充当动态的“执法”机制。我们最后提出了该领域的开放挑战，包括免疫协议标准化、新的评估指标如自身免疫率（Autoimmunity Rate，假阳性干预率），以及在集体智能生态系统中病原体与疫苗之间的共同进化动态。

View on arXiv Download PDF AI Translation

计算语言学 (Computation and Language)

cs.CL / 1 / 2606.27378

Formalizing Latent Thoughts: Four Axioms of Thought Representation in LLMs

形式化潜在思想：大型语言模型中思想表征的四个公理

Seddik, Fahd, Fard, Fatemeh

Abstract

We introduce an axiomatic evaluation framework for latent thought representations in LLMs, comprising metrics that are independent of downstream benchmark scores and reveal representational failures that benchmark accuracy masks. Existing evaluations conflate representation quality with model capacity. Therefore, failures cannot be attributed to the representation rather than to the model that processes it. We formalize four functional axioms (Causality, Minimality, Separability, and Stability) and define a quantitative measure for each, computed directly on the representation independently of downstream accuracy. We audit open-weight LLMs across 23 reasoning tasks (e.g., Spatial Reasoning, Factual QA). We find that no candidate satisfies all four axioms simultaneously, that the representations distinguish task type reliably but cannot distinguish between two questions within the same task, and that the representations encode little information beyond what is already present in the input embedding. The failure is consistent across dense, reasoning-distilled, and RL-trained model families, indicating that the gap is structural rather than a property of model size or training procedure.

Chinese Translation

我们提出了一个用于大型语言模型（LLMs）中潜在思想表征的公理评估框架，该框架包含独立于下游基准分数的度量，揭示了基准准确性掩盖的表征失败。现有评估将表征质量与模型能力混为一谈。因此，失败无法归因于表征本身，而是归因于处理它的模型。我们形式化了四个功能公理（因果性、最小性、可分离性和稳定性），并为每个公理定义了一个定量度量，该度量直接在表征上计算，与下游准确性无关。我们在23个推理任务（例如，空间推理、事实问答）中审计了开放权重的LLMs。我们发现没有候选模型能够同时满足所有四个公理，表征能够可靠地区分任务类型，但无法区分同一任务中的两个问题，并且表征编码的信息远不及输入嵌入中已有的信息。该失败在稠密模型、推理提炼模型和强化学习训练的模型家族中是一致的，表明这一差距是结构性的，而不是模型规模或训练过程的属性。

View on arXiv Download PDF AI Translation

cs.CL / 2 / 2606.27379

Position: The Term "Machine Unlearning" Is Overused in LLMs

立场：术语“机器遗忘”在大语言模型中被过度使用

Yoon, Sangyeon, Jun, Yeachan, No, Albert

Abstract

Large language models increasingly face demands to "forget" training data, knowledge, or behaviors due to regulatory deletion obligations, copyright/licensing disputes, and safety or product-policy requirements. This position paper argues that machine unlearning is overused as a term in LLM research and should be reserved for dataset-defined deletion: removing the training influence of a precisely specified forget set such that the resulting model is approximately indistinguishable from retraining without that data. We contend that many tasks currently labeled "unlearning" (e.g., refusal for harmful requests, entity/knowledge removal, or targeted suppression) pursue different, often policy-dependent objectives and therefore require different terminology and baselines (e.g., alignment, suppression, editing, obfuscation). We further argue that this confusion is not cosmetic: because papers make different implicit guarantees under the same label, metrics and benchmarks are frequently reused outside their intended scope, rewarding surface-level non-disclosure (e.g., low ROUGE/forget accuracy) even when retraining-equivalence is not tested and derived capabilities remain. We conclude by calling for stricter terminology tied to explicit guarantees and reference models, and for evaluations that match the claimed objective.

Chinese Translation

由于监管删除义务、版权/许可争议以及安全或产品政策要求，大型语言模型日益面临“遗忘”训练数据、知识或行为的需求。本文立场论文认为，在大语言模型研究中，术语机器遗忘被过度使用，应当仅保留用于数据集定义的删除：移除一个精确指定的遗忘集的训练影响，以使得结果模型与在没有该数据的情况下重新训练的模型几乎无法区分。我们主张，目前被标记为“遗忘”的许多任务（例如，拒绝有害请求、实体/知识移除或有针对性的抑制）追求不同的、通常依赖政策的目标，因此需要不同的术语和基准（例如，协调、抑制、编辑、模糊化）。我们进一步认为，这种混淆并非表面现象：由于论文在同一标签下做出不同的隐含保证，度量标准和基准经常在其预期范围之外被重复使用，奖励表面层次的非披露（例如，低 ROUGE/遗忘准确率），即使在没有测试重新训练等价性且派生能力仍然存在的情况下。我们最后呼吁对与明确保证和参考模型相关的术语进行更严格的定义，并进行与所声称目标相匹配的评估。

View on arXiv Download PDF AI Translation

cs.CL / 3 / 2606.27380

A Survey of Automated Presentation Coaching: Systems, Methods, and Open Challenges

自动化演示辅导的综述：系统、方法与开放挑战

Liang, Wen, Siyan, Li, Rackauckas, Zackary, Hirschberg, Julia

Abstract

Automated coaching for oral presentations sits at the intersection of computer-assisted pronunciation training (CAPT), prosody modeling, and speech synthesis, yet no prior work has systematically surveyed and compared existing systems along these dimensions. This survey reviews and categorizes automated presentation coaching systems, spanning pronunciation tutors, fluency and prosody coaches, multimodal trainers, and conference Q&A practice tools. We introduce a five-dimensional task taxonomy - covering segmental pronunciation, lexical stress, suprasegmental prosody, pacing, and content faithfulness - and explicitly map surveyed systems onto it to reveal coverage gaps. We further review the core technical methods these systems employ: TTS-based exemplar generation and diagnostic methods for pronunciation, prosody, and fluency assessment. Key open challenges include the scarcity of annotated presentation corpora, achieving accent-fair feedback across diverse L1 backgrounds, and delivering low-latency diagnostics for real-time rehearsal.

Chinese Translation

自动化口头演示辅导位于计算机辅助发音训练（CAPT）、韵律建模和语音合成的交叉点，但之前的研究尚未系统性地调查和比较现有系统在这些维度上的表现。本综述回顾并分类了自动化演示辅导系统，涵盖了发音辅导、流利度与韵律辅导、多模态训练器以及会议问答练习工具。我们引入了一个五维任务分类法——涵盖音段发音、词汇重音、超音段韵律、节奏和内容忠实度——并明确将调查的系统映射到该分类法上，以揭示覆盖的空白。此外，我们还回顾了这些系统所采用的核心技术方法：基于文本到语音（TTS）的示例生成和发音、韵律与流利度评估的诊断方法。主要的开放挑战包括标注演示语料库的稀缺、在不同母语背景下实现口音公平反馈，以及为实时排练提供低延迟的诊断。

View on arXiv Download PDF AI Translation

cs.CL / 4 / 2606.27446

Causal Connections: Leveraging Multilingual Fine-Tuning for Financial QA@FinCausal 2026

因果关系：利用多语言微调进行金融问答@FinCausal 2026

Gautam, Akash Kumar, Hamotskyi, Serhii, Hänig, Christian

Abstract

This paper describes team HSA_CORAL's submission to the FinCausal 2026 shared task on extracting cause-effect relations from financial narratives via extractive question answering in English and Spanish. We compare three modeling families: (i) encoder-only token tagging with multilingual BERT, (ii) encoder-decoder generation with multilingual BART, and (iii) decoder-only LLMs (Llama 3.1 and GPT variants) using prompt refinement, few-shot demonstrations, and supervised fine-tuning. Across settings, prompting and few-shot examples yield competitive performance, while supervised fine-tuning provides the largest gains. Our best system, GPT-4.1 Mini fine-tuned on combined English and Spanish training data, achieves a tied highest score on the English subtask (score 4.8140) and ranks third on Spanish (score 4.7753) under the shared task's LLM-as-a-judge metric. Overall, the results highlight the value of task-specific adaptation and multilingual fine-tuning for cross-lingual transfer in financial causality QA.

Chinese Translation

本文描述了团队 HSA_CORAL 在 FinCausal 2026 共享任务中的提交，该任务旨在通过提取式问答从金融叙述中提取因果关系，支持英语和西班牙语。我们比较了三种建模家族：(i) 使用多语言 BERT 的编码器仅标记，(ii) 使用多语言 BART 的编码器-解码器生成，以及 (iii) 使用提示优化、少量示例和监督微调的解码器仅 LLM（Llama 3.1 和 GPT 变体）。在各种设置中，提示和少量示例表现出竞争力，而监督微调提供了最大的提升。我们最佳的系统是基于结合英语和西班牙语训练数据微调的 GPT-4.1 Mini，在英语子任务中获得并列最高分（得分 4.8140），在西班牙语中排名第三（得分 4.7753），根据共享任务的 LLM-as-a-judge 指标。总体而言，结果强调了特定任务适应和多语言微调在金融因果问答中的跨语言转移价值。

View on arXiv Download PDF AI Translation

cs.CL / 5 / 2606.27460

Developmental approach reveals the statistical learning of Neural Language Models: Transformers generalize from the most abstract statistical patterns

发展性方法揭示神经语言模型的统计学习：变换器从最抽象的统计模式中进行概括

Bojun, Wang, Jenkins, Holly, Wonnacott, Elizabeth

Abstract

In this study, we use a developmental approach to investigate the statistical learning and mental representation of neural language models (NLM). A series of Generative Transformer models are trained on a synthetic grammar. The model states are saved at multiple stages in the course of training. Through analyzing how the internal representations of these models change in the developmental path, we found that NLMs acquire the most abstract global statistical knowledge at the beginning of learning and later acquire the relatively local statistical dependencies. This learning path contains many over-generalizations from the very beginning and these over-generalizations are gradually constrained in the later stage of learning. Based on this observation, we propose a new framework to explain the statistical learning and language cognition of NLMs.

Chinese Translation

在本研究中，我们采用发展性方法探讨神经语言模型（NLM）的统计学习和心理表征。我们对一系列生成性变换器模型进行了训练，使用的是一种合成语法。在训练过程中，我们在多个阶段保存了模型状态。通过分析这些模型在发展路径中内部表征的变化，我们发现NLM在学习初期获得了最抽象的全局统计知识，随后才获得相对局部的统计依赖关系。这一学习路径从一开始就包含了许多过度概括，而这些过度概括在学习的后期逐渐受到限制。基于这一观察，我们提出了一个新的框架来解释NLM的统计学习和语言认知。

View on arXiv Download PDF AI Translation

cs.CL / 6 / 2606.27472

Supersede: Diagnosing and Training the Memory-Update Gap in LLM Agents

Supersede：诊断和训练大型语言模型代理中的记忆更新差距

Patel, Vedant

Abstract

Large language model (LLM) agents operate over long, multi-session interactions in which facts change: a user moves, a price updates, a plan is revised. Acting correctly requires using the current value of a fact and discarding values that have been superseded. We isolate this ability on real conversational data and show that it is a distinct, unsolved failure. On the knowledge-update subset of LongMemEval, replacing an agent's full context with a bounded, self-maintained memory drops accuracy from 92% to 77% even on a frontier model (gpt-5.4), a gap that is statistically significant (paired McNemar p<0.005) and persists across model scale while full-context accuracy saturates near 92%. The bottleneck is therefore memory maintenance, not comprehension, and is not closed by a stronger model. We then ask whether this is merely an undersized memory, and find it is not: as the conversation grows 24x, accuracy falls further (from 68% to 28%), and granting the agent proportionally more memory yields no detectable recovery (28% to 28%, n=25). The failure scales with the length of the conversation, not the compression ratio. We release Supersede, an open reinforcement-learning environment (on the verifiers / prime-rl stack) that turns this measurement into a training signal: agents are rewarded for answering from the current value and penalized for stale ones. Finally, we close the loop and show the gap is trainable: GRPO fine-tuning a small open model (Qwen2.5-3B) on this environment nearly doubles its held-out supersession accuracy on real, unseen conversations (9.0% to 16.7%, a single run), along a monotonic checkpoint curve indicating the learned policy, not the harness, carries the gain. To our knowledge this is the first trainable environment whose reward targets temporal fact-currency, and the first evidence the supersession gap can be trained down, not only measured.

Chinese Translation

大型语言模型（LLM）代理在长时间的多会话交互中运作，其中事实会发生变化：用户移动、价格更新、计划修订。正确的行为需要使用事实的当前值并丢弃已被替代的值。我们在真实的对话数据中孤立出这种能力，并表明这是一个独特的、尚未解决的失败。在 LongMemEval 的知识更新子集上，将代理的完整上下文替换为有限的自我维护记忆，准确率从 92% 降至 77%，即使在前沿模型（gpt-5.4）上，这一差距在统计上是显著的（配对 McNemar p<0.005），并且在模型规模上持续存在，而完整上下文的准确率接近 92% 达到饱和。因此，瓶颈在于记忆维护，而非理解，且更强的模型并不能弥补这一点。接着我们询问这是否仅仅是记忆不足，结果发现并非如此：随着对话增长 24 倍，准确率进一步下降（从 68% 降至 28%），而给予代理成比例更多的记忆并未带来可检测的恢复（28% 到 28%，n=25）。这一失败与对话的长度成比例，而非压缩比。我们发布了 Supersede，一个开放的强化学习环境（基于 verifiers / prime-rl 堆栈），将这一测量转化为训练信号：代理因使用当前值回答而获得奖励，因使用过时值回答而受到惩罚。最后，我们闭合循环，表明这一差距是可训练的：在该环境中对一个小型开放模型（Qwen2.5-3B）进行 GRPO 微调，几乎将其在真实、未见对话中的超越准确率几乎翻倍（9.0% 到 16.7%，单次运行），沿着单调的检查点曲线表明学习到的策略，而非环境本身，带来了收益。据我们所知，这是第一个奖励目标为时间性事实货币的可训练环境，也是第一个证据表明超越差距可以被训练缩小，而不仅仅是被测量。

View on arXiv Download PDF AI Translation

cs.CL / 7 / 2606.27538

The Context-Ready Transformer

上下文准备型变换器

Godavarti, Mahesh

Abstract

We introduce the context-ready transformer, a new recurrent neural network architecture built from a D-layer transformer block that pre-contextualizes each token before it enters the block. During left-to-right generation, a correction network combines the previous position's block output -- a cached summary of past context -- with the current token embedding, so the tokenenters the block already contextualized rather than as a raw embedding. At sequential inference, the correction chain makes the architecture a recurrent neural network. For training, we unroll the correction process K times over the full sequence, processing all positions in parallel at each step. A pretrained transformer can also be converted to a context-ready model by adding a zero-initialized correction FFN and fine-tuning. We evaluate across widths, depths, block sizes, and two datasets, with all comparisons against standard transformers, variants, and ablations. A D=5 model beats a 12-layer transformer while generating 1.7x faster on an A100. With K=10, a single-layermodel (D=1) beats a 6-layer transformer with a 2.6x inference speedup, and sequential inference matches parallel K=10 to within 0.01 PPL. The architecture benefits most from wide representations and long contexts. On a pointer-chasing task, D=1 trained with BPTT solves all 10 composition levels, while standard transformers exhibit staircase-like depth dependence.

Chinese Translation

我们提出了一种上下文准备型变换器，这是一种新的递归神经网络架构，由一个 D 层变换器模块构成，该模块在每个标记进入模块之前对其进行预上下文化处理。在从左到右的生成过程中，一个校正网络将前一个位置的模块输出（过去上下文的缓存摘要）与当前标记的嵌入结合，从而使标记在进入模块时已经具有上下文，而不是作为原始嵌入。在顺序推理中，校正链使得该架构成为递归神经网络。在训练时，我们在整个序列上展开校正过程 K 次，在每一步并行处理所有位置。预训练的变换器也可以通过添加一个零初始化的校正前馈网络（FFN）并进行微调来转换为上下文准备型模型。我们在宽度、深度、模块大小和两个数据集上进行了评估，所有比较均与标准变换器、变体和消融实验进行对比。D=5 的模型在 A100 上生成速度比 12 层变换器快 1.7 倍。对于 K=10，单层模型（D=1）在推理速度上比 6 层变换器快 2.6 倍，而顺序推理的结果与并行 K=10 的结果相差不超过 0.01 PPL。该架构最受益于宽表示和长上下文。在指针追踪任务中，使用 BPTT 训练的 D=1 模型解决了所有 10 个组合层次，而标准变换器则表现出阶梯状的深度依赖性。

View on arXiv Download PDF AI Translation

cs.CL / 8 / 2606.27550

EntMTP: Accelerating LLM Inference with Entropy Guided Multi Token Prediction

EntMTP：通过熵引导的多标记预测加速大规模语言模型推理

Chen, Carrie

Abstract

Multi-token prediction has been shown to increase data density during training, improve downstream text-generation quality, and serves as the defacto approach for self-speculative decoding. Existing foundation and open source models that use MTP heads commit to a static tree-based attention topology throughout the entire generation sequence, meaning the speculation depth, and thus the compute required during verification, stays constant regardless of the context. This is fundamentally misaligned with the entropy patterns of natural language where low-entropy regions often support reliable multi-step drafting, while high-entropy regions require more conservative speculation. To address this, we propose Entropy-guided Multi-Token Prediction (EntMTP), a training-free scheduler that toggles between tree-based attention topologies from a set of task-specific pareto-optimal trees conditioned on a running estimate of local generation entropy. By matching speculation depth to context predictability, EntMTP maximizes expected accepted-token throughput across the full distribution of generated text without sacrificing generation quality. When evaluated across Humaneval, ShareGPT, GSM8k, and Litbench benchmarks, EntMTP consistently achieves a 1.15x speedup against Hydra and peak speedup of 1.36x against Medusa baselines respectively.

Chinese Translation

多标记预测已被证明在训练过程中增加数据密度，提高下游文本生成质量，并作为自我推测解码的事实标准方法。现有的基础模型和开源模型使用 MTP（Multi-Token Prediction）头部在整个生成序列中承诺采用静态树状注意力拓扑，这意味着推测深度，因此在验证过程中所需的计算量保持不变，无论上下文如何。这与自然语言的熵模式根本不一致，其中低熵区域通常支持可靠的多步草拟，而高熵区域则需要更保守的推测。为了解决这个问题，我们提出了熵引导的多标记预测（Entropy-guided Multi-Token Prediction，EntMTP），这是一种无训练调度器，它在一组特定任务的帕累托最优树之间切换树状注意力拓扑，条件是基于局部生成熵的运行估计。通过将推测深度与上下文可预测性相匹配，EntMTP 在不牺牲生成质量的情况下，最大化了生成文本全分布中的预期接受标记吞吐量。在 Humaneval、ShareGPT、GSM8k 和 Litbench 基准测试中评估时，EntMTP 相较于 Hydra 一致性地实现了 1.15 倍的加速，并在 Medusa 基准测试中达到了最高 1.36 倍的加速。

View on arXiv Download PDF AI Translation

cs.CL / 9 / 2606.27595

Ko-WideSearch: A Korean Breadth-Search Benchmark for Exhaustive Set Enumeration by Web Agents

Ko-WideSearch：一个针对网络代理的韩语广度搜索基准，用于全面集合枚举

Jeong, Minbyul

Abstract

Web-agent benchmarks overwhelmingly measure depth -- pinning one obscure answer behind a chain of constraints -- while breadth, exhaustively enumerating a closed set and filling each item's attributes, is barely evaluated, especially outside English. Breadth is also hard to build: certifying that a gold set is complete and every cell correct is far costlier than checking a single answer. I introduce \textsc{Ko-WideSearch}, a Korean breadth-search benchmark built by an automated synthesize-and-verify pipeline. Each task names a set-parent entity -- a TV season, a dynasty, a league, an administrative region, an election -- and asks for its full membership plus a per-item attribute table, graded by Item-, Column-, and Row-F1. It spans 228 tables over 190 entities and sixteen categories across three difficulty tiers, set by two structural knobs I dial independently -- table width and a 2-D composite key -- so cross-product membership climbs from 0\% to 100\% across the tiers. A single normalization-aware comparator is shared between gold construction and grading, so stable date and count columns are not over-dropped on formatting alone. Across twenty web agents, the failure is consistent: agents recover the set but not the rows (e.g.\ Item-F1 92.8 against Row-F1 53.7), accuracy falls steadily as the knobs harden, and neither more search nor more spend closes the gap. Broken down by cell, the hard part is finding the right value, not formatting it: open-ended free-text cells fail most, while cells with a standard answer such as a date or a name usually come out right.

Chinese Translation

网络代理基准主要测量深度——将一个晦涩的答案固定在一系列约束条件后面——而广度，即全面枚举一个封闭集合并填充每个项目的属性，几乎没有得到评估，尤其是在英语以外的语言中。构建广度也很困难：验证一个黄金集合是否完整以及每个单元是否正确的成本远高于检查单个答案。我介绍了 extsc{Ko-WideSearch}，这是一个通过自动合成和验证管道构建的韩语广度搜索基准。每个任务命名一个集合父实体——一个电视季、一段王朝、一个联盟、一个行政区域、一场选举——并要求提供其完整成员资格以及每个项目的属性表，评分标准为项目、列和行的 F1 值。该基准涵盖了 228 个表格，涉及 190 个实体和十六个类别，分为三个难度级别，由我独立调节的两个结构参数设定——表格宽度和一个二维复合键——因此交叉产品成员资格在各个级别之间从 0\% 上升到 100\%。一个单一的考虑归一化的比较器在黄金构建和评分之间共享，因此稳定的日期和计数列不会因格式问题而被过度丢弃。在二十个网络代理中，失败是一致的：代理能够恢复集合但无法恢复行（例如，项目 F1 为 92.8，而行 F1 为 53.7），随着参数的收紧，准确性稳步下降，更多的搜索或更多的支出都无法缩小差距。按单元划分，困难的部分是找到正确的值，而不是格式化：开放式自由文本单元失败最多，而具有标准答案的单元，如日期或名称，通常能够正确输出。

View on arXiv Download PDF AI Translation

cs.CL / 10 / 2606.27598

Narrative-UFET: Narrative Generation for Ultra-Fine Entity Typing

叙事-UFET：超细粒度实体类型生成

Gupta, Mreedul, Deshmukh, Advait, Umadi, Ashwin, Pauk, Matt, Pacheco, Maria Leonor

Abstract

Ultra-fine entity typing (UFET) assigns highly specific types to entity mentions, but current approaches struggle with types in the long tail. We hypothesize that a key limitation is the reliance on sentence-level context, since disambiguating evidence is often spread across multiple sentences. Testing this has been difficult because all existing UFET resources are sentence-level. We present Narrative-UFET, a controlled extension of UFET in which each entity mention is paired with an automatically generated short, coherent narrative. Synthesizing narratives lets us isolate the effect of specific discourse properties. We experiment with two paired variants: one in which the entity's type is held constant across the narrative (Maintain) and one in which it shifts (Change). We show that narrative context yields consistent improvements on long-tail types over sentence-level baselines, with the Change variant providing the stronger signal. A comparison against naturally occurring contexts shows that synthetic narratives yield stronger gains, indicating that controlled discourse construction can surface signals that real text leaves implicit. Substantial room for improvement remains, suggesting open directions in both discourse modeling and narrative construction.

Chinese Translation

超细粒度实体类型（UFET）为实体提及分配高度特定的类型，但当前的方法在处理长尾类型时面临困难。我们假设一个关键限制是对句子级上下文的依赖，因为消歧证据通常分散在多个句子中。测试这一假设一直很困难，因为所有现有的UFET资源都是句子级的。我们提出了叙事-UFET，这是UFET的一个受控扩展，其中每个实体提及都与自动生成的简短连贯叙事配对。合成叙事使我们能够隔离特定话语属性的影响。我们实验了两种配对变体：一种在叙事中保持实体类型不变（Maintain），另一种则发生变化（Change）。我们展示了叙事上下文在长尾类型上相较于句子级基线的一致性提升，其中Change变体提供了更强的信号。与自然发生的上下文比较表明，合成叙事带来了更强的增益，表明受控的话语构建能够揭示真实文本隐含的信号。仍有相当大的改进空间，这为话语建模和叙事构建的开放方向提供了建议。

View on arXiv Download PDF AI Translation

cs.CL / 11 / 2606.27617

Masked Language Flow Models

掩蔽语言流模型

Azangulov, Iskander, Ashouritaklimi, Kianoosh, Zhang, Leo, Vary, Simon, Rebeschini, Patrick

Abstract

Masked Diffusion Models (MDMs) promise fast, parallel language generation, but their reverse transition factorises across token positions -- an approximation that breaks down in the few-step sampling regime where parallel generation ought to provide the greatest efficiency gains. Flow Language Models (FLMs) sidestep this limitation by learning a continuous flow that transports noise toward clean sequences represented in Euclidean space, inducing a flow map that can be distilled for single-step generation. However, this makes complex tasks requiring multi-step reasoning problematic for FLMs, as FLMs are forced to decode every token during generation. To address this, we introduce Masked Language Flow Models (MLFMs), which incorporate masking into FLMs using a continuous stochastic interpolant to bridge partially masked and clean sequences. This design enables conditional generation via continuous flows and allows pretrained MDMs to be converted into MLFMs through a simple, lightweight adaptation. Leveraging this flexibility, we propose a novel sampler that alternates continuous denoising with the discrete unmasking of confident tokens to better support multi-step reasoning. We evaluate our approach on GSM8K and MT-Bench and find, for the first time, that flow-based language models can be scaled to solve downstream reasoning and instruction-following tasks.

Chinese Translation

掩蔽扩散模型（Masked Diffusion Models, MDMs）承诺实现快速、并行的语言生成，但其反向过渡在标记位置上进行因式分解——这一近似在少步采样模式下失效，而此时并行生成应提供最大的效率提升。流语言模型（Flow Language Models, FLMs）通过学习一种连续流来绕过这一限制，该流将噪声传输到在欧几里得空间中表示的干净序列，从而诱导出可以用于单步生成的流映射。然而，这使得需要多步推理的复杂任务对FLMs而言变得棘手，因为FLMs在生成过程中被迫解码每一个标记。为了解决这一问题，我们引入了掩蔽语言流模型（Masked Language Flow Models, MLFMs），该模型通过使用连续随机插值将掩蔽和干净序列连接起来，将掩蔽引入FLMs。这一设计使得通过连续流进行条件生成成为可能，并允许预训练的MDMs通过简单、轻量的适配转换为MLFMs。利用这一灵活性，我们提出了一种新颖的采样器，该采样器交替进行连续去噪和对自信标记的离散解掩蔽，以更好地支持多步推理。我们在GSM8K和MT-Bench上评估了我们的方法，首次发现基于流的语言模型可以扩展到解决下游推理和遵循指令的任务。

View on arXiv Download PDF AI Translation

cs.CL / 12 / 2606.27629

Cross-Platform Chinese Offensive Comment Detection via Dual-Threshold Hard Example Mining

基于双阈值困难样本挖掘的跨平台中文攻击性评论检测

Ren, Ruixing, Zhao, Junhui, Wang, Fangfang

Abstract

Cross-platform deployment of offensive comment detection for Chinese social media suffers performance degradation. The paper proposes a dual-threshold hard mining method to address this. First, the clean-Chinese-base RoBERTa is finetuned on COLD to establish a binary baseline for fair comparison. Second, a three-class fine-labeled test set covering Weibo, Xiaohongshu, Tieba, and Zhihu is constructed, domain distances from the source are quantified using Jaccard and Proxy-A Distance, as well as the degradation bottleneck of the baseline under domain shift is systematically revealed. Herein, a dual threshold hard example mining strategy is proposed. High- and low-confidence error-prone samples are filtered from unlabeled corpora by prediction confidence. The model is secondarily finetuned under implicit contexts with merely a small set of manually labeled hard examples, realizing low-cost cross-platform domain adaptation. Experiments reveal significant performance gains of the optimized model across four platforms.

Chinese Translation

针对中文社交媒体的跨平台攻击性评论检测存在性能下降的问题。本文提出了一种双阈值困难样本挖掘方法来解决这一问题。首先，在COLD数据集上对clean-Chinese-base RoBERTa进行微调，以建立一个二分类基线，确保公平比较。其次，构建了一个涵盖微博、小红书、贴吧和知乎的三类精细标注测试集，使用Jaccard和Proxy-A距离量化源域之间的距离，并系统性揭示了基线在领域转移下的性能瓶颈。在此基础上，提出了一种双阈值困难样本挖掘策略。通过预测置信度，从未标注语料中筛选出高置信度和低置信度的易错样本。模型在隐式上下文中进行二次微调，仅需少量手动标注的困难样本，从而实现低成本的跨平台领域适应。实验结果显示，优化后的模型在四个平台上均显著提升了性能。

View on arXiv Download PDF AI Translation

cs.CL / 13 / 2606.27632

Yuvion LLM: An Adversarially-Aware Large Language Model for Content And AI Safety

Yuvion LLM：一种关注对抗性的安全大型语言模型

Ma, Ting, Huang, Xiufeng, Cui, Benlei, Xu, Xiaowen, Qiu, Shikai, Jian, Ruijie, Li, Hongxing, Wang, Guanghui, Huang, Longtao, Hong, Haiwen, Xu, Haolei, Jiang, Wenjing, Xu, Ziwen, Fan, Zhaoyu, He, Shaoxuan, Xiao, Chuxi, Li, Yujian, Chen, Xinyue, Chai, Chunyang, Liu, Wenxuan, Wang, Ziheng, Zhang, Dongjie, Zhou, Yangfan, Dong, Libin, Cao, Yupeng, Xia, Xiaoqian, Wang, Jing, Jiang, Zhe, Ye, Zhenan, Yang, Guang, Liu, Bin, Peng, Wei, Zhu, Ziqiang, Lian, Meihui, Kacuila, Kaiwen Lv, Ding, Haidong, Zhu, Bingyu, Wang, Yan, Zhao, Hai, Jin, Xuan, Zhao, Wei, Sun, Pengfei, Wang, Wei, Zhang, Huiming, Li, Bin, Xue, Hui

Abstract

As large language models are increasingly deployed in real-world systems, safety failures can still lead to harmful outputs and dangerous misuse. We argue that the essence of safety is adversarial: many failures arise not from natural inputs alone, but from strategic attempts to evade model policies and safeguards. However, existing general-purpose model development largely overlook this adversarial nature, and often remain insufficient for realistic safety scenarios involving planning, tool use, and multi-step reasoning, causing measured safety performance to overestimate real deployment robustness. To address this gap, we present Yuvion LLM, a large language model built for adversarially robust content safety and broader AI safety. Yuvion LLM treats adversarial robustness and agentic capability as first-class objectives. Its pipeline combines adversarially aware data construction, knowledge-enhanced continued pretraining, and policy-grounded multi-task safety post-training, including risk-aware supervised fine-tuning and reinforcement learning-based policy optimization, together with safety-aware agentic reinforcement learning for tool use and multi-step reasoning in complex safety scenarios. We further introduce the Yuvion LLM RiskEval (YLRE), a collection of 93 benchmarks across four evaluation categories, covering diverse open and internal evaluations with a focus on safety, adversarial robustness, and real-world capability requirements. Across these evaluations, Yuvion LLM demonstrates clear advantages on safety-focused benchmarks and particularly strong robustness under adversarial conditions, while maintaining solid overall capability. Notably, Yuvion-8B outperforms most state-of-the-art baselines, including substantially larger models such as GPT-5.4 and Qwen3-MAX, on several safety tasks.

Chinese Translation

随着大型语言模型在现实系统中的日益应用，安全失效仍可能导致有害输出和危险误用。我们认为安全的本质是对抗性的：许多失效并非仅源于自然输入，而是来自于规避模型政策和保护措施的战略性尝试。然而，现有的通用模型开发在很大程度上忽视了这种对抗性特征，往往不足以应对涉及规划、工具使用和多步骤推理的现实安全场景，导致测量的安全性能高估了实际部署的稳健性。为了解决这一问题，我们提出了Yuvion LLM，这是一种为对抗性稳健内容安全和更广泛的人工智能安全而构建的大型语言模型。Yuvion LLM将对抗性稳健性和代理能力视为首要目标。其流程结合了对抗性意识的数据构建、知识增强的持续预训练和基于政策的多任务安全后训练，包括风险意识的监督微调和基于强化学习的政策优化，以及针对复杂安全场景中的工具使用和多步骤推理的安全意识代理强化学习。我们进一步介绍了Yuvion LLM风险评估（YLRE），这是一个涵盖四个评估类别的93个基准的集合，涉及多样的开放和内部评估，重点关注安全性、对抗性稳健性和现实世界能力要求。在这些评估中，Yuvion LLM在以安全为重点的基准上表现出明显优势，并在对抗条件下展现出特别强的稳健性，同时保持了整体能力的稳定性。值得注意的是，Yuvion-8B在多个安全任务上超越了大多数最先进的基线，包括像GPT-5.4和Qwen3-MAX这样的大型模型。

View on arXiv Download PDF AI Translation

cs.CL / 14 / 2606.27669

When Search Agents Should Ask: DiscoBench for Clarification-Aware Deep Search

搜索代理何时应提问：面向澄清意识的深度搜索基准DiscoBench

Tao, Yiling, Deng, Shihan, Tao, Meiling, Wei, Pengzhi, Hu, Zhichao, Zhu, Zhihao

Abstract

Search agents powered by large language models (LLMs) are increasingly used to solve complex information-seeking tasks, requiring multi-step retrieval and reasoning to fulfill user goals. However, existing benchmarks often assume that user queries are complete and explicit, overlooking the fact that real-world search requests are frequently vague, underspecified, or even factually incorrect. In deep search scenarios, such ambiguity can propagate along multi-step reasoning chains and lead agents toward incorrect search trajectories. To address this gap, we introduce DiscoBench, a benchmark for clarification-aware deep search, designed to evaluate whether search agents can proactively identify ambiguity, ask effective clarification questions, and recover correct reasoning paths through user interaction. DiscoBench contains 211 samples and 463 ambiguity instances across 11 real-world domains, covering four ambiguity types. We further design a user simulator for multi-turn interaction and evaluate model performance from four perspectives: task utility, ambiguity detection, interaction strategy, and cost efficiency. Experiments on representative LLMs show that ambiguity detection and effective clarification are distinct capabilities, and that repeatedly searching instead of asking for clarification often performs worse than direct guessing, highlighting a critical gap between retrieval ability and interactive problem-solving in current search agents.

Chinese Translation

由大型语言模型（LLMs）驱动的搜索代理越来越多地被用于解决复杂的信息检索任务，这些任务需要多步骤的检索和推理以实现用户目标。然而，现有基准通常假设用户查询是完整和明确的，忽视了现实世界中的搜索请求常常模糊、不明确，甚至事实错误这一事实。在深度搜索场景中，这种模糊性可能沿着多步骤推理链传播，导致代理走向错误的搜索轨迹。为了解决这一问题，我们引入了DiscoBench，一个面向澄清意识的深度搜索基准，旨在评估搜索代理是否能够主动识别模糊性、提出有效的澄清问题，并通过用户互动恢复正确的推理路径。DiscoBench包含211个样本和463个模糊性实例，涵盖11个真实世界领域，涉及四种模糊性类型。我们进一步设计了一个用户模拟器用于多轮互动，并从四个角度评估模型性能：任务效用、模糊性检测、互动策略和成本效率。对代表性LLM的实验表明，模糊性检测和有效澄清是不同的能力，而反复搜索而不是请求澄清往往表现得比直接猜测更差，突显了当前搜索代理在检索能力与互动问题解决之间的关键差距。

View on arXiv Download PDF AI Translation

cs.CL / 15 / 2606.27679

From Signals to Transfer: A Factorised Study of Probe-Based Uncertainty Estimation in Large Language Models

从信号到转移：基于探针的不确定性估计在大型语言模型中的分解研究

Srey, Ponhvoan, Wu, Xiaobao, Nguyen, Cong-Duy, Nguyen, Quang Minh, Vu, Duc Anh, Luu, Anh Tuan

Abstract

Probe-based uncertainty estimation (UE) has emerged as a prominent approach to detect hallucinations in Large Language Models (LLMs) by learning uncertainty from internal model signals. Yet, recent methods vary simultaneously across feature design, training data construction, and evaluation setting, obscuring what actually drives performance. To address this issue, we propose a factorised study of probe-based UE under matched conditions. Our results show that raw hidden states and attention features are difficult to outperform in-domain. However, under distribution shift, structured and compressed features are more robust, suggesting that in-domain performance alone is insufficient to measure progress. Furthermore, prompting and label construction significantly affect probe behaviour. Building on these best-practice findings, we train benchmark-based pretrained probes that transfer reasonably well to open-ended factual generation, providing a stable off-the-shelf baseline. Our work encourages more deployment-oriented evaluation of probe-based uncertainty estimators. The code repository is available at https://github.com/ponhvoan/ProbeUE.

Chinese Translation

基于探针的不确定性估计（UE）已成为检测大型语言模型（LLMs）幻觉的一个重要方法，通过学习内部模型信号的不确定性。然而，近期的方法在特征设计、训练数据构建和评估设置上存在显著差异，模糊了实际驱动性能的因素。为了解决这个问题，我们提出在匹配条件下对基于探针的UE进行分解研究。我们的结果表明，原始隐藏状态和注意力特征在领域内难以被超越。然而，在分布转移的情况下，结构化和压缩特征更具鲁棒性，这表明仅依赖领域内的性能不足以衡量进展。此外，提示和标签构建显著影响探针的行为。基于这些最佳实践发现，我们训练了基于基准的预训练探针，这些探针在开放式事实生成中表现良好，提供了一个稳定的现成基线。我们的工作鼓励对基于探针的不确定性估计器进行更多面向部署的评估。代码库可在 https://github.com/ponhvoan/ProbeUE 获取。

View on arXiv Download PDF AI Translation

cs.CL / 16 / 2606.27687

Mitigating LLM-based p-Hacking by Preregistering for the Next LLM

通过预注册下一个大型语言模型来缓解基于大型语言模型的p-hacking

Thomas, Maria, Gligoric, Kristina, Shah, Nihar B.

Abstract

Large language models (LLMs) are increasingly used to generate, classify, and annotate data whose outputs feed downstream hypothesis tests. However, LLM-based research is easy to p-hack: a researcher can tune the prompts, decoding parameters, or output format until a desired result is reached. We propose a protocol to mitigate p-hacking in LLM-based research: preregistering the experiment and eligible models, and then running it on the first eligible LLM that is released after the preregistration. The researcher finalizes the procedure on current models, preregisters the analysis plan together with a set of eligible future models, and runs the confirmatory analysis on the first eligible model released afterward. Because this model does not exist at commitment time, it cannot be hacked against; furthermore, configurations that hack one model frequently do not transfer to the next. We evaluate the protocol on two tasks whose true values are known. Across 20 models from four providers and 11 LLM-analysis configurations, the protocol would have blocked successful transfer of the p-hack in 73.9% and 72.7% of cases in the two tasks. Additional analyses reveal that mitigation remains substantial under several stress tests. Finally, putting money where our mouth is, we followed our own protocol and preregistered our experiment. The preregistered experiment confirmed the protocol's effectiveness: out of the 7 configurations that hacked the prior model, the hacking failed to carry over in 6 configurations on the first eligible model released afterward.

Chinese Translation

大型语言模型（LLMs）越来越多地用于生成、分类和注释数据，其输出用于下游假设检验。然而，基于LLM的研究容易受到p-hacking的影响：研究者可以调整提示、解码参数或输出格式，直到达到所需结果。我们提出了一种协议，以缓解基于LLM的研究中的p-hacking：预注册实验和合格模型，然后在预注册后发布的第一个合格LLM上运行实验。研究者在当前模型上最终确定程序，预注册分析计划以及一组合格的未来模型，并在随后发布的第一个合格模型上进行确认性分析。由于该模型在承诺时并不存在，因此无法进行p-hacking；此外，针对一个模型的配置通常不会转移到下一个模型。我们在两个已知真实值的任务上评估了该协议。在来自四个提供者的20个模型和11种LLM分析配置中，该协议在两个任务中分别阻止了73.9%和72.7%的p-hacking成功转移。额外分析表明，在多个压力测试下，缓解效果仍然显著。最后，我们遵循自己的协议并预注册了实验。预注册的实验证实了该协议的有效性：在7种对先前模型进行p-hacking的配置中，6种在随后发布的第一个合格模型上未能转移p-hacking。

View on arXiv Download PDF AI Translation

cs.CL / 17 / 2606.27705

Mitigating Position Bias in Transformers via Layer-Specific Positional Embedding Scaling

通过层特定的位置信息嵌入缩放来减轻变换器中的位置偏差

Lv, Changze, Wang, Zhenghua, Ding, Yiran, Wu, Yixin, Li, Tianlong, Xu, Zhibo, Wu, Muling, Shi, Tianyuan, Li, Shizheng, Qian, Qi, Huang, Xuanjing, Zheng, Xiaoqing

Abstract

Large Language Models (LLMs) still struggle with the ``lost-in-the-middle'' problem, where critical information located in the middle of long-context inputs is often underrepresented or lost. While existing methods attempt to address this by combining multi-scale rotary position embeddings (RoPE), they typically suffer from high latency or rely on suboptimal hand-crafted scaling strategies. To overcome these limitations, we introduce a layer-specific positional embedding scaling~(LPES) method that assigns distinct scaling factors to each layer. LPES achieves a more balanced attention distribution without fine-tuning model parameters or increasing inference delay. A specially designed genetic algorithm is employed to efficiently select the optimal scaling factors for each layer by incorporating B\'{e}zier curves to significantly reduce the search space. Extensive experiments demonstrate that LPES effectively mitigates positional attention bias and delivers consistent improvements across multiple long-context benchmarks, yielding up to an $11.2$\% accuracy gain on the key-value retrieval dataset.

Chinese Translation

大型语言模型（LLMs）仍然面临“迷失在中间”问题，即长上下文输入中位于中间的关键信息往往被低估或丢失。尽管现有方法试图通过结合多尺度旋转位置嵌入（RoPE）来解决此问题，但它们通常存在高延迟或依赖于次优的手工缩放策略。为了克服这些限制，我们提出了一种层特定的位置信息嵌入缩放（LPES）方法，该方法为每一层分配不同的缩放因子。LPES在不微调模型参数或增加推理延迟的情况下，实现了更平衡的注意力分布。我们采用了一种专门设计的遗传算法，通过结合贝塞尔曲线有效选择每层的最佳缩放因子，从而显著减少搜索空间。大量实验表明，LPES有效减轻了位置注意力偏差，并在多个长上下文基准测试中提供了一致的改进，在关键值检索数据集上实现了高达11.2%的准确率提升。

View on arXiv Download PDF AI Translation

cs.CL / 18 / 2606.27709

Low-Agreeableness Persona Conditioning for Safe LLM Fine-Tuning

低宜人性人格条件下的安全大型语言模型微调

Cheung, Austin MY, Yang, Yi

Abstract

Recent work has shown that fine-tuning large language models (LLMs) for social warmth degrades factual reliability and increases sycophancy. We investigate a related but distinct failure mode: warmth fine-tuning also weakens adversarial safety, making models more susceptible to jailbreaks and harmful output generation. We examine whether this reflects an inherent consequence of empathetic adaptation or an artifact of data construction. To address this, we introduce a persona-driven rewriting pipeline that conditions user turns on low agreeableness and pairs this with warm, de-escalating assistant responses. Across three experiments on four models, our approach reduces jailbreak susceptibility and harmful output rates relative to generic warmth fine-tuning baselines, while preserving conversational warmth. Representational probing provides suggestive evidence that this conditioning reduces the geometric alignment between warmth and compliance directions in latent space. These results show that safer empathetic fine-tuning is achievable through data design alone, without safety labels, harm detectors, or changes to the training objective.

Chinese Translation

近期的研究表明，为了增强社交温暖而对大型语言模型（LLMs）进行微调会降低其事实可靠性并增加谄媚性。我们探讨了一种相关但不同的失败模式：温暖微调还削弱了对抗安全性，使模型更容易受到越狱攻击和有害输出生成的影响。我们考察这是否反映了同情适应的内在结果，或是数据构建的副作用。为了解决这个问题，我们引入了一种基于人格驱动的重写管道，该管道将用户的发言条件设定为低宜人性，并与温暖、缓和的助手回应相结合。在对四个模型进行的三项实验中，我们的方法相较于通用温暖微调基线，降低了越狱易感性和有害输出率，同时保持了对话的温暖性。表征探测提供了暗示性证据，表明这种条件化减少了潜在空间中温暖与顺从方向之间的几何对齐。这些结果表明，通过数据设计，安全的同情微调是可以实现的，而无需安全标签、伤害检测器或对训练目标的改变。

View on arXiv Download PDF AI Translation

cs.CL / 19 / 2606.27717

Do Speech Emphasis Models Generalize across Languages and Emotions?

语音强调模型在不同语言和情感中的泛化能力研究

Wei, Megan, Aneja, Deepali, Su, Jiaqi, Wang, Yunyun, Chen, Haonan, Jin, Zeyu

Abstract

Prosodic emphasis varies across languages, emotions, and speaking styles, yet existing emphasis detection models are largely trained and evaluated on monolingual neutral read speech. We introduce MMEE (Multilingual Multi-Emotion Emphasis), a corpus of 10,000 professionally recorded expressive utterances (14.13 hours) across 7 languages and 34 emotion/style categories, with three-level perceptual labels (10 annotations per sample). We benchmark two state-of-the-art architectures under monolingual, cross-lingual, multilingual, cross-emotion, cross-dataset, and data-scale settings. Monolingual models show limited zero-shot transfer, degrading across typologically distant languages, while multilingual training substantially improves robustness. Models transfer robustly between high- and low-arousal emotions; bidirectional transfer between synthetic and perceptual benchmarks suggests shared prosodic structure; and performance stays robust even at smaller training scales.

Chinese Translation

韵律强调在不同语言、情感和说话风格中存在差异，但现有的强调检测模型主要是在单语言中性朗读语音上进行训练和评估。我们引入了MMEE（多语言多情感强调）语料库，该语料库包含10,000个专业录制的富有表现力的语句（14.13小时），涵盖7种语言和34种情感/风格类别，并提供三层感知标签（每个样本10个注释）。我们在单语言、跨语言、多语言、跨情感、跨数据集和数据规模设置下，对两种最先进的架构进行了基准测试。单语言模型在零样本迁移方面表现有限，在类型学上相距较远的语言之间性能下降，而多语言训练显著提高了模型的鲁棒性。模型在高唤起和低唤起情感之间具有良好的迁移能力；合成基准和感知基准之间的双向迁移表明共享的韵律结构；即使在较小的训练规模下，模型的性能依然保持稳健。

View on arXiv Download PDF AI Translation

cs.CL / 20 / 2606.27731

Enhancing Numerical Prediction in LLMs via Smooth MMD Alignment

通过平滑最大均值差异对齐增强大语言模型中的数值预测

Zuo, Zhuo, Yue, Li, Zheng, Wenhao, Wang, Chenpeng, Liu, Xianggen

Abstract

Despite their strong general capabilities, large language models (LLMs) often remain unreliable when outputs must be numerically precise. A key reason is the training objective: standard cross-entropy treats numeric tokens as unstructured categories and ignores the metric structure of their values. We address this mismatch with Smooth Maximum Mean Discrepancy (SMMD), which builds on the classic MMD by incorporating value-distance kernels over numeric tokens and graph-based smoothness. With this kernel defined over a numeric sub-vocabulary, SMMD aligns the predicted numeric distribution to the target via kernel matching and smooths the prediction-target residual over the induced kernel graph to encourage local consistency. We evaluate SMMD on four numeric-target tasks: mathematical reasoning, arithmetic calculation, clock-time recognition, and chart question answering, across multiple open-weight LLM and VLM backbones. SMMD consistently improves accuracy over both cross-entropy and recent numeric-target losses; analyses show complementary effects between MMD and smoothness and underscore the importance of distance-based kernel design. Code is available at https://github.com/Zuozhuo/smmd-loss.

Chinese Translation

尽管大语言模型（LLMs）具有强大的通用能力，但在需要数值精确的输出时，它们往往仍然不可靠。一个关键原因在于训练目标：标准的交叉熵将数值标记视为非结构化类别，并忽视了其值的度量结构。我们通过平滑最大均值差异（Smooth Maximum Mean Discrepancy, SMMD）来解决这一不匹配问题，该方法在经典的最大均值差异（MMD）基础上，结合了数值标记的值距离核和基于图的平滑性。通过在数值子词汇表上定义该核，SMMD通过核匹配将预测的数值分布与目标对齐，并在诱导的核图上平滑预测与目标的残差，以促进局部一致性。我们在四个数值目标任务上评估SMMD：数学推理、算术计算、时钟时间识别和图表问答，涵盖多个开放权重的LLM和VLM骨干网络。SMMD在准确性上始终优于交叉熵和近期的数值目标损失；分析表明MMD和平滑性之间存在互补效应，并强调了基于距离的核设计的重要性。代码可在 https://github.com/Zuozhuo/smmd-loss 获取。

View on arXiv Download PDF AI Translation

cs.CL / 21 / 2606.27742

KG2Cypher: Data-Centric Pipeline for Building Enterprise Text-to-Cypher Systems

KG2Cypher：构建企业文本到Cypher系统的数据驱动管道

Choi, Minjun, Kim, Yerin, Seo, Junghyuk, Mo, Sujin, Lee, Hyemin, Ko, Youngjoong

Abstract

Enterprise Knowledge Graphs (KGs) are increasingly used for internal search, analytics, and question answering, but building natural-language interfaces for private enterprise graphs remains costly. We present KG2Cypher, a data-centric pipeline for building enterprise text-to-Cypher systems from existing KGs. KG2Cypher first constructs an executable Cypher query from observed graph facts and then uses LLMs to generate its associated natural-language question. The resulting Text-Cypher pairs are validated with an LLM judge and human validation, and are converted into candidate-aware SFT data. The trained generator is served with class-conditioned schema prompting, entity retrieval, and LoRA-based inference. We evaluate KG2Cypher in Korean enterprise settings, where short search-style queries and schema paraphrases make language grounding difficult. LoRA SFT improves execution-result F1 from 0.806 to 0.950 on broadcast-program queries and from 0.70 to 0.92 on company queries. In an 11-class setting, KG2Cypher achieves 95.2% exact match, 99.9% execution rate, and 0.964 execution-result F1.

Chinese Translation

企业知识图谱（KGs）越来越多地用于内部搜索、分析和问答，但为私有企业图谱构建自然语言接口仍然成本高昂。我们提出了KG2Cypher，这是一个从现有KGs构建企业文本到Cypher系统的数据驱动管道。KG2Cypher首先从观察到的图谱事实构建可执行的Cypher查询，然后使用大型语言模型（LLMs）生成其相关的自然语言问题。生成的文本-Cypher对通过LLM评估和人工验证进行验证，并转换为候选感知的SFT数据。训练后的生成器通过类别条件的模式提示、实体检索和基于LoRA的推理进行服务。我们在韩国企业环境中评估KG2Cypher，在这些环境中，短搜索风格的查询和模式释义使得语言基础变得困难。LoRA SFT将广播节目查询的执行结果F1从0.806提高到0.950，将公司查询的执行结果F1从0.70提高到0.92。在11类设置中，KG2Cypher实现了95.2%的精确匹配率、99.9%的执行率和0.964的执行结果F1。

View on arXiv Download PDF AI Translation

cs.CL / 22 / 2606.27785

Output-Space Allocation Costs for Calibration-Guided LLM Compression: An Empirical Study

基于校准指导的LLM压缩的输出空间分配成本：一项实证研究

Tang, Qiong, Hu, Xiangkun, Liu, Xiangyang, Chen, Yiran, Shao, Yunfan

Abstract

Training-free compression methods for large language models (LLMs) often use calibration data to guide compression decisions. ROCKET, a recent method combining sparse-dictionary factorization with multi-choice knapsack problem (MCKP) allocation, derives its per-layer factorization from an output reconstruction objective but uses weight-space Frobenius error as the MCKP allocation cost. We investigate whether aligning the allocation cost with the output-space objective improves compressed model fidelity. On Qwen3-8B at 50\% compression, our ROCKET-ActCost achieves +0.8 percentage points higher average accuracy across 8 zero-shot benchmarks (53.1\% vs 52.3\%), but increases WikiText perplexity by 16\% (61.46 vs 52.98). This accuracy-perplexity tradeoff reveals that different allocation objectives favor different downstream metrics. The high correlation ($>$0.99) between weight-space and output-space errors limits allocation divergence, explaining the modest effect size. On Llama-3.2-1B at 20\% compression, the two methods produce near-identical results (53.3\% vs 53.5\% accuracy, 14.45 vs 14.66 PPL), suggesting that the effect of the cost function is minor at lower compression ratios.

Chinese Translation

无训练压缩方法用于大型语言模型（LLMs）通常利用校准数据来指导压缩决策。ROCKET是一种新近提出的方法，它结合了稀疏字典分解与多选择背包问题（MCKP）分配，其每层的分解源于输出重构目标，但使用权重空间的Frobenius误差作为MCKP分配成本。我们研究了将分配成本与输出空间目标对齐是否能提高压缩模型的保真度。在Qwen3-8B模型上以50%的压缩率，我们的ROCKET-ActCost在8个零样本基准测试中实现了平均准确率提高0.8个百分点（53.1%对比52.3%），但WikiText的困惑度增加了16%（61.46对比52.98）。这种准确率与困惑度的权衡揭示了不同的分配目标有利于不同的下游指标。权重空间与输出空间误差之间的高相关性（>0.99）限制了分配的差异，解释了效果大小的适度。在Llama-3.2-1B模型上以20%的压缩率，两种方法产生了几乎相同的结果（准确率53.3%对比53.5%，困惑度14.45对比14.66），这表明在较低压缩比下成本函数的影响较小。

View on arXiv Download PDF AI Translation

cs.CL / 23 / 2606.27786

SHIFT: Gate-Modulated Activation Steering for Knowledge Conflict Mitigation in Retrieval-Augmented Generation

SHIFT：用于缓解检索增强生成中的知识冲突的门控调制激活引导

Li, Ruochang, Huang, Pengcheng, Liu, Zhenghao, Yan, Yukun, Xie, Huiyuan, Gu, Yu, Yu, Ge, Sun, Maosong

Abstract

Retrieval-augmented generation (RAG) enhances LLMs by incorporating external knowledge to support response generation. However, conflicts between retrieved context and parametric knowledge have emerged as a critical challenge in RAG systems. To mitigate such conflicts, numerous studies have attempted to identify and edit knowledge-related internal neurons, aiming to improve the ability of LLMs to rely on contextual evidence during generation. However, these neuron-level approaches may introduce unintended cascading effects that compromise the general capabilities of LLMs, as the modified neurons are often entangled with broader model behaviors and functionalities. In this paper, we introduce SHIFT, a novel framework that reformulates neuron-level modification as learnable gate modulation, allowing LLMs to adaptively regulate internal activations for knowledge conflict resolution. Technically, our SHIFT equips LLMs with a lightweight gate module and optimizes fewer than 0.01% trainable parameters while keeping the backbone model frozen. During generation, the gate module adjusts the model's internal representations to adaptively leverage contextual and parametric knowledge. Extensive experiments on six datasets validate the effectiveness of our SHIFT in comparison with various competing baselines. All datasets and code are available at https://github.com/OpenBMB/SHIFT.

Chinese Translation

检索增强生成（RAG）通过结合外部知识来增强大型语言模型（LLMs），以支持响应生成。然而，检索到的上下文与参数知识之间的冲突已成为RAG系统中的一个关键挑战。为了缓解这些冲突，许多研究尝试识别和编辑与知识相关的内部神经元，旨在提高LLMs在生成过程中依赖上下文证据的能力。然而，这些神经元级别的方法可能引入意想不到的级联效应，从而损害LLMs的整体能力，因为被修改的神经元通常与更广泛的模型行为和功能交织在一起。在本文中，我们介绍了SHIFT，一个新颖的框架，将神经元级别的修改重新构建为可学习的门控调制，使LLMs能够自适应地调节内部激活以解决知识冲突。从技术上讲，我们的SHIFT为LLMs配备了一个轻量级的门控模块，并优化了不到0.01%的可训练参数，同时保持主干模型不变。在生成过程中，门控模块调整模型的内部表示，以自适应地利用上下文和参数知识。在六个数据集上的大量实验验证了我们SHIFT的有效性，并与各种竞争基线进行了比较。所有数据集和代码均可在https://github.com/OpenBMB/SHIFT获取。

View on arXiv Download PDF AI Translation

cs.CL / 24 / 2606.27791

NLL-Guided Full-Attention Layer Selection for Training-Free Sliding-Window Adaptation

基于NLL引导的全注意力层选择用于无训练的滑动窗口适应

Tang, Qiong, Hu, Xiangkun, Liu, Xiangyang, Chen, Yiran, Shao, Yunfan

Abstract

Hybrid attention models that mix full and sliding-window attention across layers offer a promising approach to efficient long-context inference, but the critical question of \emph{which layers} should retain full attention remains unsolved. Existing methods use either fixed periodic patterns or attention-based heuristics that may not capture what matters for downstream accuracy. We propose NLL-guided layer selection, a training-free method that directly measures each layer's importance by computing the negative log-likelihood degradation on answer tokens when that layer uses sliding-window instead of full attention. On LongMemEval with Qwen3-4B, our method achieves 64.6\% accuracy using only 1/4 full-attention layers, matching the 1/2-FA periodic baseline (65.0\%) while halving the computational budget. NLL-guided selection outperforms the SWAA-reported periodic 1/4-FA baseline by 10.4 percentage points and a matched LightTransfer-style baseline by 26.4 percentage points. De-confounding analysis shows the signal is consistent with long-range attention needs rather than generic layer sensitivity. The method requires only $\sim$15 minutes of one-time calibration, advancing the efficiency-accuracy Pareto frontier for long-context LLM deployment.

Chinese Translation

混合注意力模型通过在层间混合全注意力和滑动窗口注意力，为高效的长上下文推理提供了一种有前景的方法，但关键问题是 extit{哪些层}应保留全注意力仍未解决。现有方法使用固定的周期模式或基于注意力的启发式方法，这可能无法捕捉对下游准确性至关重要的因素。我们提出了基于NLL引导的层选择，这是一种无训练的方法，通过计算在该层使用滑动窗口而非全注意力时答案标记的负对数似然降级，直接衡量每层的重要性。在LongMemEval上使用Qwen3-4B，我们的方法仅使用1/4的全注意力层便达到了64.6\%的准确率，匹配了1/2-FA周期基线（65.0\%），同时将计算预算减半。NLL引导的选择比SWAA报告的周期1/4-FA基线高出10.4个百分点，比匹配的LightTransfer风格基线高出26.4个百分点。去混淆分析表明，该信号与长距离注意力需求一致，而非通用层敏感性。该方法仅需约15分钟的一次性校准，推动了长上下文LLM部署的效率-准确性帕累托前沿。

View on arXiv Download PDF AI Translation

cs.CL / 25 / 2606.27793

Position Bias Correction is Insufficient for One-Pass Attention Sorting

位置偏差校正不足以实现一次性注意力排序

Tang, Qiong, Hu, Xiangkun, Liu, Xiangyang, Chen, Yiran, Shao, Yunfan

Abstract

Long-context language models suffer from position bias, where information in middle positions is underutilized. Attention Sorting addresses this by iteratively reordering documents based on attention patterns, but its multiple sort-and-generate cycles increase deployment cost. We hypothesize that position bias is the primary bottleneck and propose Debiased One-Pass Attention Sorting, which estimates a per-prompt position-bias curve from the low-attention majority of documents and uses it to correct raw attention scores (via subtraction or division) to enable single-pass sorting. Our experiments on two models refute this hypothesis in the tested setting: on LLaMA-2-7B-32K-Instruct, debiasing produces identical results to uncalibrated single-pass sorting (94.83\% containment accuracy), while on YaRN-Llama-2-7b-64k, debiasing improves accuracy by 8.67 percentage points but remains 14.84pp behind iterative sorting, closing only 37\% of the gap. These results suggest that position-bias correction is insufficient to match iterative sorting, and that repeated reordering provides additional benefits beyond bias correction.

Chinese Translation

长上下文语言模型存在位置偏差的问题，其中中间位置的信息被利用不足。注意力排序通过基于注意力模式迭代地重新排序文档来解决这一问题，但其多次排序和生成周期增加了部署成本。我们假设位置偏差是主要瓶颈，并提出了去偏置的一次性注意力排序（Debiased One-Pass Attention Sorting），该方法从低注意力的多数文档中估计每个提示的位置信息偏差曲线，并利用该曲线通过减法或除法来校正原始注意力分数，以实现单次排序。我们在两个模型上的实验反驳了这一假设：在 LLaMA-2-7B-32K-Instruct 上，去偏置的结果与未校准的一次性排序相同（94.83% 的包含准确率），而在 YaRN-Llama-2-7b-64k 上，去偏置提高了准确率 8.67 个百分点，但仍比迭代排序落后 14.84 个百分点，仅缩小了 37% 的差距。这些结果表明，位置偏差校正不足以与迭代排序相匹配，而重复的重新排序提供了超出偏差校正的额外好处。

View on arXiv Download PDF AI Translation

cs.CL / 26 / 2606.27808

Learning Complementary Action Modeling from Automotive Maintenance Instructions

从汽车维护说明中学习互补动作建模

Wu, Jiaqi, Li, Bai, Hartmann, Jochen, Gaedke, Martin, Stuijk, Sander

Abstract

A minute lexical variation can reverse the procedural meaning of an instruction even when the rest of the sentence remains unchanged. In automotive maintenance instructions, this pattern often appears when an action phrase turns an instruction into its procedural counterpart. The entities, modifiers, and surrounding context remain largely invariant, while the action phrase determines the procedural relation. We define this task as Complementary Action Modeling (CAM). Given a maintenance instruction, the goal is to identify or generate its procedural counterpart by modifying the action phrase while preserving the remaining sentence context. This task focuses on three aspects: distinguishing complementarity from surface similarity, controlling generation at the action-phrase level, and evaluating relational correctness using retrieval, overlap-based, and human evaluation. Using a German automotive maintenance dataset, we examine these questions through candidate matching and controlled Seq2Seq generation. The results show that complementary maintenance instructions are best modeled as procedural associations grounded in subtle lexical cues. They should therefore not be treated as ordinary cases of sentence similarity or synonym-based paraphrasing.

Chinese Translation

一个微小的词汇变化可以逆转指令的程序意义，即使句子的其余部分保持不变。在汽车维护说明中，当一个动作短语将指令转变为其程序对应物时，这种模式经常出现。实体、修饰语和周围的上下文基本保持不变，而动作短语决定了程序关系。我们将这一任务定义为互补动作建模（Complementary Action Modeling, CAM）。给定一条维护指令，目标是通过修改动作短语来识别或生成其程序对应物，同时保留剩余的句子上下文。该任务关注三个方面：区分互补性与表面相似性、在动作短语层面控制生成，以及使用检索、重叠基础和人工评估来评估关系的正确性。通过使用德国汽车维护数据集，我们通过候选匹配和受控的Seq2Seq生成来研究这些问题。结果表明，互补的维护指令最好被建模为基于细微词汇线索的程序关联。因此，它们不应被视为普通的句子相似性或基于同义词的释义案例。

View on arXiv Download PDF AI Translation

cs.CL / 27 / 2606.27881

A Study of Temporal Fusion Strategies for Named Entity Recognition in Historical Texts

历史文本中命名实体识别的时间融合策略研究

Boros, Emanuela

Abstract

Temporal variation poses a unique challenge for named entity recognition (NER) in historical texts, where entities drift in surface form and salience across time. While language models (LMs) have made progress in various NLP tasks, their ability to reason about temporality, especially in diachronic contexts, remains limited or at least, questionable. In this paper, we systematically study how temporal metadata can be structurally embedded into NER models using a range of lightweight fusion strategies. We experiment with both absolute and relative temporal representations, injected into Transformer-based architectures via early or late fusion mechanisms such as cross-attention, adapters, and concatenation. Our evaluations on French and German historical datasets reveal that late fusion strategies yield more robust and temporally generalisable performance, particularly in early and noisy periods.

Chinese Translation

时间变异对历史文本中的命名实体识别（NER）提出了独特的挑战，其中实体在表面形式和显著性上随时间漂移。尽管语言模型（LMs）在各种自然语言处理（NLP）任务中取得了进展，但它们在推理时间性方面的能力，尤其是在历时语境中，仍然有限或至少值得怀疑。本文系统研究了如何通过一系列轻量级融合策略将时间元数据结构性地嵌入到NER模型中。我们实验了绝对和相对时间表示，通过早期或晚期融合机制（如交叉注意力、适配器和拼接）注入到基于Transformer的架构中。我们在法语和德语历史数据集上的评估表明，晚期融合策略在早期和噪声较大的时期表现出更强的鲁棒性和时间可推广性。

View on arXiv Download PDF AI Translation

cs.CL / 28 / 2606.27909

Triadic Werewolf: A Jester Role for Multi-Hop Theory of Mind in LLMs

三方狼人：大型语言模型中的小丑角色与多跳心智理论

Mittal, Avni

Abstract

Theory-of-mind evaluations of large language models typically use dyadic social-deduction games, where every observable cue points to a single hidden side, so a model with strong language priors can score well without ever simulating opponents' incentives. We extend the Werewolf game with a Jester, a third faction whose utility on peer suspicion is inverted because it wins by being voted out, so optimal play requires reasoning across three opposing utility functions. Across 60 games on GPT-4.1, DeepSeek-V3.1, and Llama-3.3-70B with Jester self-learning on and off, the Jester wins 60-70% of games while Werewolves never exceed 20%, and GPT-4.1 wolves vote the Jester out on day 1 in 60-70% of games, a strictly self-defeating action. Self-learning helps DeepSeek and Llama but hurts GPT-4.1, with the cost landing on Villagers rather than Werewolves. Only DeepSeek learns the subtle strategy of looking suspicious without looking intentionally suspicious, and it gains the most from the loop. Triadic incentive structure exposes a layer of multi-agent reasoning that dyadic deduction games leave invisible.

Chinese Translation

大型语言模型的心智理论评估通常使用二元社交推理游戏，其中每个可观察的线索都指向一个单一的隐藏面，因此具有强语言先验的模型可以在不模拟对手动机的情况下表现良好。我们扩展了狼人游戏，引入了小丑这一第三方势力，其在同伴怀疑中的效用是反向的，因为它通过被投票淘汰而获胜，因此最佳策略需要在三个相互对立的效用函数之间进行推理。在对GPT-4.1、DeepSeek-V3.1和Llama-3.3-70B进行的60场游戏中，无论小丑自我学习是否开启，小丑的胜率为60-70%，而狼人从未超过20%。在60-70%的游戏中，GPT-4.1的狼人会在第一天将小丑投票淘汰，这是一种严格的自我挫败行为。自我学习对DeepSeek和Llama有帮助，但对GPT-4.1则有害，成本落在村民而非狼人身上。只有DeepSeek学会了在不显得故意可疑的情况下看起来可疑的微妙策略，并从循环中获得了最多的收益。三方激励结构揭示了二元推理游戏所隐含的多智能体推理层面。

View on arXiv Download PDF AI Translation

cs.CL / 29 / 2606.27941

VASAE: Naming SAE Dictionary Directions with Vocabulary-Aligned Anchoring

VASAE：通过词汇对齐锚定命名稀疏自编码器字典方向

Zhang, Kairui, Yu, Ziwen, Abdallah, Zahraa S., Lewis, Martha

Abstract

Sparse autoencoders (SAEs) provide useful decompositions of Transformer residual streams, but their learned features are usually named post hoc rather than directly connected to the Transformer's token vocabulary. We introduce Vocabulary-Aligned Sparse Autoencoder (VASAE), a method that trains SAE features under vocabulary-aligned anchoring and assigns each feature an intrinsic token name: the token string whose embedding is nearest to that feature. Without reducing reconstruction quality compared with a standard SAE, VASAE produces dictionaries with vocabulary-aligned features. Using a 0.8 cutoff on the nearest-token alignment score, dictionaries trained on GPT-2-small post-residual streams align about 90% of features in layers 0--10. In Llama-3.1-8B, representative shallow and middle-layer dictionaries contain strongly aligned features, including 92.8% in the shallow layer, while the representative final-layer dictionary shows limited alignment. After subtracting the sentence-level mean sparse code, case studies show that many remaining intrinsic token names are relevant to nearby input tokens. These results suggest that vocabulary-aligned anchoring can connect learned features to intrinsic token names during training, complementing post hoc interpretation of learned dictionaries.

Chinese Translation

稀疏自编码器（SAEs）为Transformer残差流提供了有用的分解，但它们学习到的特征通常是事后命名，而不是直接与Transformer的标记词汇相关联。我们提出了词汇对齐稀疏自编码器（VASAE），一种在词汇对齐锚定下训练SAE特征的方法，并为每个特征分配一个内在的标记名称：与该特征的嵌入最接近的标记字符串。在与标准SAE相比不降低重构质量的情况下，VASAE生成了具有词汇对齐特征的字典。使用0.8的最近标记对齐分数截止值，在GPT-2-small的残差流上训练的字典在层0-10中对齐了约90%的特征。在Llama-3.1-8B中，代表性的浅层和中层字典包含强对齐特征，包括浅层中的92.8%，而代表性的最终层字典显示出有限的对齐。在减去句子级均值稀疏编码后，案例研究表明，许多剩余的内在标记名称与附近的输入标记相关。这些结果表明，词汇对齐锚定可以在训练期间将学习到的特征与内在标记名称连接起来，补充了对学习字典的事后解释。

View on arXiv Download PDF AI Translation

cs.CL / 30 / 2606.27959

An Empirical Analysis of Factual Errors in Human-Written Text and its Application

人类撰写文本中事实错误的实证分析及其应用

Iwamoto, Kazuma, Omura, Kazumasa, Ishihara, Shotaro

Abstract

Factual Error Detection (FED), which is the task of identifying factually incorrect spans in a given text, has long been recognized as an important research problem. However, with the rapid rise of large language models (LLMs), research attention has shifted toward factual errors specific to LLM-generated text (hallucinations) and their detection. As a result, the detection of factual errors in human-written text has been relatively neglected. To address this gap, we first distill a taxonomy of human-induced factual errors by analyzing corrections of newspaper articles, a representative source of text that is guaranteed to be human-written and contains few grammatical errors. Our analysis revealed that there are characteristic categories such as kanji misconversions and numeral classifier errors, which are not focused in existing hallucination benchmarks. Based on the taxonomy, we then evaluate the FED capability of vanilla LLMs on synthesized realistic test cases and real corrections. Experimental results demonstrated that even high-performance LLMs such as GPT-5.4 achieved only word-level F1 score of 52% on the synthetic evaluation data, highlighting the task difficulty. Furthermore, a detailed analysis by detection difficulty revealed the current state of FED.

Chinese Translation

事实错误检测（Factual Error Detection, FED）是识别给定文本中事实不正确部分的任务，长期以来被认为是一个重要的研究问题。然而，随着大型语言模型（Large Language Models, LLMs）的快速崛起，研究的关注点已转向特定于LLM生成文本的事实错误（幻觉）及其检测。因此，人类撰写文本中的事实错误检测相对被忽视。为了解决这一空白，我们首先通过分析报纸文章的修正，提炼出人类引发的事实错误的分类法，这些文章是保证人类撰写且语法错误较少的代表性文本来源。我们的分析揭示了诸如汉字误转换和数字分类错误等特征类别，这些类别在现有的幻觉基准中并未受到关注。基于该分类法，我们评估了普通LLMs在合成的现实测试案例和真实修正上的FED能力。实验结果表明，即使是像GPT-5.4这样高性能的LLMs，在合成评估数据上的词级F1得分仅为52%，突显了该任务的难度。此外，通过检测难度的详细分析揭示了当前FED的状态。

View on arXiv Download PDF AI Translation

cs.CL / 31 / 2606.27973

From Black-Box to Clinical Insight: A Multi-Stage Explainable Framework for Speech-Based Cognitive Impairment Detection

从黑箱到临床洞察：一种基于语音的认知障碍检测的多阶段可解释框架

Haghbin, Yasaman, Rashidi, Sina, Zolnour, Ali, Taherinezhad, Fatemeh, Fartoot, Ali, Azadmaleki, Hossein, Noble, James M, Dadkhah, Maryam, Zolnoori, Maryam

Abstract

Speech-based cognitive impairment detection offers a noninvasive, accessible alternative to costly biomarker assays, yet transformer-based models remain clinically uninterpretable. We propose a multi-stage explainability framework that translates black-box transformer predictions into clinically grounded narratives by integrating SHapley Additive exPlanations (SHAP)-based token attribution, theory-informed linguistic features, and a four-stage LLM reasoning pipeline using LLaMA-3.1-70B-Instruct. Built on the SpeechCARE-Adaptive Gating Network multimodal screening model (F1 = 72.11% on the NIA PREPARE benchmark), the framework maps model outputs to four cognitive-linguistic dimensions, including lexical richness, syntactic complexity, and semantic coherence. Physician evaluation on 70 stratified English samples demonstrated strong alignment with patient-level cognitive profiles, and a System Usability Scale score of 82/100 indicated high potential for clinical workflow integration.

Chinese Translation

基于语音的认知障碍检测提供了一种非侵入性、可获取的替代方案，取代了昂贵的生物标志物检测，然而基于变换器的模型在临床上仍然缺乏可解释性。我们提出了一种多阶段可解释性框架，通过整合基于SHapley加法解释（SHAP）的标记归因、理论驱动的语言特征以及使用LLaMA-3.1-70B-Instruct的四阶段大型语言模型（LLM）推理管道，将黑箱变换器的预测转化为临床基础的叙述。该框架基于SpeechCARE-自适应门控网络多模态筛查模型（在NIA PREPARE基准上的F1得分为72.11%），将模型输出映射到四个认知语言维度，包括词汇丰富度、句法复杂性和语义连贯性。对70个分层英语样本的医生评估显示与患者级别的认知特征高度一致，系统可用性评分为82/100，表明在临床工作流程整合方面具有很高的潜力。

View on arXiv Download PDF AI Translation

cs.CL / 32 / 2606.27981

ToxiREX: A Dataset on Toxic REasoning in ConteXt

ToxiREX：上下文中的有毒推理数据集

Schouten, Stefan F., Markov, Ilia, Vossen, Piek

Abstract

We introduce a new, contextual, multilingual dataset called ToxiREX: Toxic REasoning in ConteXt. The dataset consists of threads of Reddit comments and structured characterizations of what the comments imply, following a systematic toxic reasoning schema developed in a previous paper. Using the schema allows us to capture and explain implicit and context-dependent toxicity, while supporting mappings to existing toxicity taxonomies. The dataset includes comments in six languages (English, Arabic, Turkish, Spanish, German, and Dutch), collected from posts connected to specific major events (e.g. the 2023 Turkey earthquakes; the Russian invasion of Ukraine). We describe the context-preserving preprocessing of the threads. We create a training set of 125 thousand comments which is annotated by a commercially available LLM, and a test set of just under three thousand comments that is annotated by native speakers. We show that apparent disagreements in the test set annotations often reflect defensible alternative interpretations rather than noise. Finally, we provide baseline results by prompting and fine-tuning language models. To produce these results, we develop evaluation strategies for our hierarchical, schema-based predictions. While models perform better than random, there remains a lot of room for improvement, showing the task to be challenging. ToxiREX is the first dataset to simultaneously incorporate multiple languages, conversational context, and implicit toxicity, while using the toxic reasoning schema for rich, structured annotations. Dataset available at: https://github.com/cltl/toxirex

Chinese Translation

我们介绍了一个新的上下文多语言数据集，称为 ToxiREX：上下文中的有毒推理。该数据集由 Reddit 评论线程及其所隐含内容的结构化特征组成，遵循在之前论文中开发的系统性有毒推理框架。使用该框架使我们能够捕捉和解释隐含的、依赖于上下文的有毒性，同时支持与现有有毒性分类法的映射。数据集包含六种语言的评论（英语、阿拉伯语、土耳其语、西班牙语、德语和荷兰语），这些评论来自与特定重大事件相关的帖子（例如，2023年土耳其地震；俄罗斯入侵乌克兰）。我们描述了线程的上下文保留预处理过程。我们创建了一个包含125,000条评论的训练集，该训练集由商业可用的语言模型（LLM）进行标注，以及一个包含不到3,000条评论的测试集，该测试集由母语者进行标注。我们展示了测试集标注中的明显分歧往往反映出合理的替代解释，而非噪声。最后，我们通过提示和微调语言模型提供了基线结果。为了产生这些结果，我们为我们的层次化、基于框架的预测开发了评估策略。尽管模型的表现优于随机选择，但仍有很大的改进空间，显示出该任务的挑战性。ToxiREX 是第一个同时结合多种语言、对话上下文和隐含有毒性的数据库，同时使用有毒推理框架进行丰富的结构化标注。数据集可在以下链接获取：https://github.com/cltl/toxirex

View on arXiv Download PDF AI Translation

cs.CL / 33 / 2606.28002

Dialogue to Detection: A Multimodal Hybrid NLP Pipeline for Insurance Fraud Detection

对话到检测：一种用于保险欺诈检测的多模态混合自然语言处理管道

Akram, Muhammad Shakeel, Htait, Amal, Sadka, Abdul Hamid, Meisingseth, Emma, Jaitly, Karishma

Abstract

Insurance fraud imposes substantial financial losses and operational inefficiencies, raising premiums and impacting trust among legitimate policyholders. Early detection at FNOL remains a persistent challenge. Existing approaches rely largely on private, text-only datasets, limiting progress on multimodal methods that integrate linguistic, behavioural, and speaker-based indicators. We introduce a synthetic multimodal framework that replicates FNOL conditions. It generates agent-customer dialogue transcripts and two-speaker audios, performs ASR and diarisation. Downstream modules combine NER, regex-based feature extraction, LLM-RAG retrieval, and speaker embeddings in a rule-based risk score to flag narrative reuse, structural inconsistencies, and cross-case voice repetition while balancing sensitivity and false positives. Dataset validation and component-level evaluations show stability and transfer potential, offering a reproducible baseline beyond text-only fraud detection.

Chinese Translation

保险欺诈造成了巨大的财务损失和运营效率低下，导致保费上涨并影响合法投保人的信任。早期检测在首次通知索赔（FNOL）阶段仍然是一个持续的挑战。现有方法主要依赖于私有的文本数据集，这限制了在整合语言、行为和说话者指标的多模态方法上的进展。我们提出了一种合成的多模态框架，复制了FNOL条件。该框架生成代理-客户对话记录和双说话者音频，执行自动语音识别（ASR）和说话者分离。下游模块结合了命名实体识别（NER）、基于正则表达式的特征提取、基于大语言模型的检索（LLM-RAG）和说话者嵌入，采用基于规则的风险评分来标记叙述重用、结构不一致和跨案例语音重复，同时平衡敏感性和假阳性。数据集验证和组件级评估显示出稳定性和迁移潜力，提供了一个超越仅文本欺诈检测的可重复基线。

View on arXiv Download PDF AI Translation

cs.CL / 34 / 2606.28013

The Signal-Coverage Matrix: Stratifying Type and Semantic Errors in Statement Autoformalization

信号覆盖矩阵：在语句自动形式化中分层类型和语义错误

Dai, Chengxiao, Yan, Zhaokun, Lin, Zhanhui

Abstract

Headline type-correctness (TC\%) of LLM autoformalization has climbed from $\sim$53\% to $\sim$76\% in two years, yet this scalar conceals which errors each method resolves. We propose a signal-coverage matrix that crosses the Lean elaborator (pass/fail) with a semantic-equivalence judgment (equivalent/not), sorting every output into one of four cells: true success (TS), type-only (TO), semantic-only (SO), or both fail (BF). On ProofNet\# and MiniF2F-test with DeepSeek V4-Pro across Vanilla, Lean-Retry, Sample-Filter, and Stratified Autoformalization (SAF): (1) the +34 to +36 TS gain across the three elab-feedback methods is $\sim$64\% type-stratum recovery, with SO flat on net (87.5\% of original semantic errors rescued, 8 newly created). (2) The TO-to-TS rate is 23/61 for each method (Wilson 95\% CI [26.6\%, 50.3\%]), and this stratum-level recovery rate predicts $\Delta$TS on held-out methods to within 2/186 and renders $\Delta$TC linear in the Vanilla elab-fail rate across six (model, dataset) cells ($R^2=0.96$). (3) The two judges disagree by 26 to 37 pp on elab-feedback outputs (vs. 7 pp on Vanilla), with 30 to 56\% of symbolic-judge false negatives traceable to elaborator-forced rewrites. The persistent residual reduces to two gold-formalization errors. TC\% gains should be credited by which cell moved, not by the scalar alone.

Chinese Translation

在过去两年中，大型语言模型（LLM）自动形式化的标题类型正确率（TC\%）从约53\%上升至约76\%，然而这一标量掩盖了每种方法解决的具体错误。我们提出了一种信号覆盖矩阵，将Lean elaborator（通过/失败）与语义等价判断（等价/不等价）相交，将每个输出分类到四个单元格之一：真正成功（TS）、仅类型（TO）、仅语义（SO）或均失败（BF）。在ProofNet\#和MiniF2F-test上，使用DeepSeek V4-Pro进行Vanilla、Lean-Retry、Sample-Filter和分层自动形式化（SAF）的实验结果显示：（1）在三种elab反馈方法中，TS的增益为+34到+36，约为64\%的类型分层恢复，SO的净变化持平（87.5\%的原始语义错误被挽救，新增8个错误）。（2）每种方法的TO到TS的比率为23/61（Wilson 95\%置信区间[26.6\%, 50.3\%]），这一分层恢复率可预测在保留方法上的$ abla$TS误差为2/186，并使$ abla$TC在六个（模型，数据集）单元格的Vanilla elab失败率中呈线性关系（$R^2=0.96$）。（3）两位评审在elab反馈输出上的意见分歧为26到37个百分点（而在Vanilla上为7个百分点），其中30到56\%的符号评审假阴性可追溯至elaborator强制重写。持久的残差减少到两个黄金形式化错误。TC\%的增益应根据移动的单元格进行归因，而不仅仅依赖于标量。

View on arXiv Download PDF AI Translation

cs.CL / 35 / 2606.28044

A Tree-of-Thoughts Inspired Hybrid Approach for Legal Case Judgement Summarization using LLMs

一种受思维树启发的混合方法用于法律案例判决摘要生成，基于大型语言模型（LLMs）

Deroy, Aniket, Ghosh, Kripabandhu, Ghosh, Saptarshi

Abstract

In recent times, Large Language Models (LLMs) are increasingly being used for legal case judgement summarization. Most prior works have tried traditional extractive and abstractive summarization of case judgements. However, hybrid or extractive-abstractive techniques have not been explored much. In this work, we propose a novel tree-of-thoughts inspired extractive-abstractive summarization approach for legal judgement summarization. We conduct experiments using two popular LLMs, DeepSeek and LLama, and compare among extractive, abstractive and extractive-abstractive summarization. Our experiments show that the proposed extractive-abstractive prompt provides better summaries compared to other types of LLM prompts.

Chinese Translation

近年来，大型语言模型（LLMs）在法律案例判决摘要生成中的应用日益增多。大多数先前的研究尝试了传统的提取式和生成式摘要方法。然而，混合或提取-生成技术尚未得到充分探索。在本研究中，我们提出了一种新颖的受思维树启发的提取-生成摘要方法用于法律判决摘要生成。我们使用两种流行的LLMs，DeepSeek和LLama，进行实验，并比较提取式、生成式和提取-生成摘要的效果。实验结果表明，所提出的提取-生成提示相比其他类型的LLM提示提供了更优质的摘要。

View on arXiv Download PDF AI Translation

cs.CL / 36 / 2606.28050

Can LLMs Judge Better Than They Generate? Evaluating Task Asymmetry, Mechanistic Interpretability and Transferability for In-Context QA

大型语言模型能否比生成更好地进行判断？评估上下文问答中的任务不对称性、机制可解释性和迁移性

Bandyopadhyay, Sambaran

Abstract

LLM-as-a-Judge and self-evaluation pipelines implicitly assume that evaluation is easier than generation. We test this in a controlled in-context QA setting where a context passage is the sole information source and each model judges the answer it generated, removing the parametric-knowledge confound of open-domain comparisons. Across four benchmarks (SQuAD 2.0, DROP, HotpotQA, MuSiQue) and two models, evaluation is not uniformly easier: generation accuracy exceeds self-evaluation on three of four, with multi-hop MuSiQue the exception. Attention analysis reveals why: evaluation attends to context 3--5x less than generation does and barely reads the candidate answer. LoRA fine-tuning confirms the asymmetry is not a training artifact: generation fine-tuning induces over-acceptance and evaluation fine-tuning degrades generation. These findings challenge core assumptions in self-evaluation pipelines.

Chinese Translation

将大型语言模型（LLM）作为判断者和自我评估流程隐含地假设评估比生成更容易。我们在一个受控的上下文问答环境中测试这一假设，其中上下文段落是唯一的信息来源，每个模型评估其生成的答案，从而消除了开放领域比较中的参数知识混淆。在四个基准测试（SQuAD 2.0、DROP、HotpotQA、MuSiQue）和两个模型中，评估并不总是更容易：在四个基准中，生成准确性在三项中超过自我评估，只有多跳的MuSiQue是例外。注意力分析揭示了原因：评估对上下文的关注程度比生成少3到5倍，并且几乎没有阅读候选答案。LoRA微调确认这种不对称性并非训练伪影：生成微调导致过度接受，而评估微调则降低了生成效果。这些发现挑战了自我评估流程中的核心假设。

View on arXiv Download PDF AI Translation

cs.CL / 37 / 2606.28057

MultiHashFormer: Hash-based Generative Language Models

MultiHashFormer：基于哈希的生成语言模型

Xue, Huiyin, Yamaguchi, Atsuki, Aletras, Nikolaos

Abstract

Language models (LMs) represent tokens using embedding matrices that scale linearly with the vocabulary size. To constrain the parameter footprint, prior work proposes hashing many tokens into a single vector within encoder-only models. While this offers parameter efficiency, many-to-one collisions prevent its use in causal LMs. In this paper, we propose MultiHashFormer, a new framework that allows hash-based autoregression. Each token is represented as a unique hash signature, a short sequence of discrete hash IDs, generated by multiple independent hash functions. A Hash Encoder compresses this signature into a single latent vector for processing by a Transformer decoder. Then, a Hash Decoder generates the hash signature of the next token, which is then mapped back to text. We evaluate our approach at the 100M, 1B and 3B parameter scales, demonstrating that MultiHashFormer consistently outperforms standard Transformer LMs across multiple benchmarks. Furthermore, we show that our model handles multilingual vocabulary expansion with a constant parameter footprint without any modifications.

Chinese Translation

语言模型（LMs）使用嵌入矩阵表示标记，其规模与词汇表大小线性相关。为了限制参数占用，之前的研究提出在仅编码器模型中将多个标记哈希为单个向量。尽管这提供了参数效率，但多对一的冲突阻止了其在因果语言模型中的使用。本文提出了MultiHashFormer，一个允许基于哈希的自回归的新框架。每个标记被表示为一个独特的哈希签名，即由多个独立哈希函数生成的一短序列离散哈希ID。哈希编码器将该签名压缩为一个单一的潜在向量，以供Transformer解码器处理。然后，哈希解码器生成下一个标记的哈希签名，并将其映射回文本。我们在100M、1B和3B参数规模下评估了我们的方法，结果表明MultiHashFormer在多个基准测试中始终优于标准Transformer语言模型。此外，我们还展示了我们的模型在不进行任何修改的情况下，能够以恒定的参数占用处理多语言词汇扩展。

View on arXiv Download PDF AI Translation

cs.CL / 38 / 2606.28116

Mechanism-Driven Monitors for Preemptive Detection of LLM Training Instability

基于机制驱动的监测器用于预防性检测大语言模型训练不稳定性

Huang, Ruixuan, Wang, Yipei, Fang, Wenyi, Huang, Hantao, Huang, Yifan, You, Ansheng, Zhang, Zhenxing, Wang, Shuai, Wu, Fan, Zheng, Yang

Abstract

Frontier large language model training consumes massive accelerator fleets and long wall-clock computation, making stability failures costly when they occur. After a numerical or a hyperparameter fault has already destabilized the training dynamics, it may continue for thousands of steps while loss and gradient norms still appear normal. We study mechanism-driven detection of training instability by deriving internal monitors from the functional role of each critical module and from the earliest computational sites where failures are expected to produce measurable signatures. For low-precision flash attention, we monitor the spectral entropy of a QK bilinear decomposition, whose first-order term becomes abnormal before the loss fully collapses. For MoE routers, we derive indicators from their role in expert selection. Our fault-injection experiments on low-precision attention, large learning-rate, and combined faults show that these signals provide distinct signatures for different failures, triggering thousands of steps before loss divergence.

Chinese Translation

前沿的大语言模型训练消耗大量的加速器集群和长时间的计算，导致稳定性故障发生时成本高昂。在数值或超参数故障已经破坏训练动态后，训练可能会持续数千步，而损失和梯度范数仍然看似正常。我们通过从每个关键模块的功能角色和预期产生可测量特征的最早计算位置推导内部监测器，研究基于机制驱动的训练不稳定性检测。对于低精度的闪存注意力，我们监测QK双线性分解的谱熵，其一阶项在损失完全崩溃之前就会变得异常。对于MoE路由器，我们从其在专家选择中的角色推导指标。我们在低精度注意力、大学习率和组合故障上的故障注入实验表明，这些信号为不同的故障提供了独特的特征，能够在损失发散之前触发数千步。

View on arXiv Download PDF AI Translation

cs.CL / 39 / 2606.28127

From Tokens to States: LLMs as a Special Case of World Models and the Continuous Path Beyond

从标记到状态：大型语言模型作为世界模型的特例及其持续发展路径

Dubois, Paul

Abstract

The AI community has framed the relationship between large language models (LLMs) and world models as a dichotomy: LLMs predict tokens; world models simulate reality. Yann LeCun argues in 2022 that reaching general intelligence requires abandoning autoregressive token prediction in favour of latent-space architectures. This framing is unnecessarily binary. Two claims will be defended. First, LLMs are a degenerate special case of world models: the state space is the set of all token sequences, the only action is appending one token, and world models are therefore a strict generalisation of LLMs, not a replacement. Second, there is a natural continuous spectrum from NTP to JEPA, with multi-token prediction, future-summary prediction, and next-latent prediction as intermediate stations already populated by current research. Moving along this spectrum relaxes the LLM constraints one by one. It also progressively surrenders the two practical advantages that make LLMs trainable at scale: internet-scale self-supervised data, and a transformer architecture co-designed for discrete token prediction. Both are examined as open research questions: the data question (the cliff from self-supervised text to instrumented action-labelled environments) and the architecture question (whether the transformer generalises to continuous-state prediction, or whether a new primitive is needed).

Chinese Translation

人工智能界将大型语言模型（LLMs）与世界模型之间的关系框定为一种二元对立：LLMs 预测标记；世界模型模拟现实。Yann LeCun 在 2022 年指出，实现通用智能需要放弃自回归标记预测，转而采用潜在空间架构。这种框架是不必要的二元化。本文将辩护两个观点。首先，LLMs 是世界模型的一个退化特例：状态空间是所有标记序列的集合，唯一的动作是附加一个标记，因此世界模型是 LLMs 的严格推广，而不是替代。其次，从 NTP 到 JEPA 存在一个自然的连续谱，其中多标记预测、未来摘要预测和下一个潜在预测作为中间站，已经被当前研究所填充。沿着这个谱移动逐步放宽 LLM 的约束。同时，它也逐渐放弃了使 LLM 可大规模训练的两个实际优势：互联网规模的自监督数据，以及为离散标记预测共同设计的变换器架构。这两者都被视为开放的研究问题：数据问题（从自监督文本到带有工具化动作标签环境的悬崖）和架构问题（变换器是否能够推广到连续状态预测，或者是否需要新的原始构件）。

View on arXiv Download PDF AI Translation

cs.CL / 40 / 2606.28186

Cognitive Episodes in LLM Reasoning Traces Enable Interpretable Human Item Difficulty Prediction

大型语言模型推理轨迹中的认知事件促进可解释的人类项目难度预测

Wang, Chenguang, Li, Ming, Zeng, Xinyue, Li, Zhuochun, Jiao, Hong, Zhou, Tianyi, Zhou, Dawei

Abstract

Predicting human item difficulty is central to educational assessment, where reliable estimates support fairness and effective test construction. Existing methods often depend on costly human calibration or item-level textual representations, providing limited evidence about the cognitive processes that make items difficult. We argue that difficulty should be viewed not only as a property of item text, but also as an observable consequence of the problem-solving burden an item induces. Large Reasoning Models (LRMs) offer scalable process evidence through reasoning traces, but such evidence must be structured to support interpretable modeling. To this end, we introduce Epi2Diff (Episode to Difficulty), a framework that maps LRM reasoning traces into cognitively grounded episode sequences. These episodes group trace segments into functional problem-solving states, enabling difficulty to be modeled through reasoning scale, effort allocation, and state transitions. Epi2Diff extracts compact episode-dynamic features and combines them with semantic item representations for human difficulty prediction. Experiments on four real-world human difficulty datasets show that Epi2Diff consistently outperforms strong baselines, including fine-tuned small language models, LLM in-context learning, and supervised LLM adaptation. On SAT-derived classification benchmarks, Epi2Diff achieves an 8.1% average relative gain over supervised LLM fine-tuning baselines. Further analyses show that harder items induce more effortful, iterative, and implementation-centered episode dynamics, rather than merely longer responses. These results demonstrate that cognitive episodes in LRM reasoning traces provide a predictive and interpretable process representation for human item difficulty, offering a new lens for educational measurement with reasoning models.

Chinese Translation

预测人类项目难度是教育评估的核心，可靠的估计支持公平性和有效的测试构建。现有方法通常依赖于昂贵的人为校准或项目级文本表示，提供关于使项目变得困难的认知过程的有限证据。我们认为，难度不仅应被视为项目文本的属性，还应被视为项目所引发的问题解决负担的可观察结果。大型推理模型（Large Reasoning Models, LRM）通过推理轨迹提供可扩展的过程证据，但这些证据必须结构化以支持可解释的建模。为此，我们提出了Epi2Diff（Episode to Difficulty），一个将LRM推理轨迹映射为认知基础的事件序列的框架。这些事件将轨迹片段分组为功能性的问题解决状态，使得难度可以通过推理规模、努力分配和状态转变进行建模。Epi2Diff提取紧凑的事件动态特征，并将其与语义项目表示结合，以进行人类难度预测。在四个真实世界的人类难度数据集上的实验表明，Epi2Diff始终优于强基线，包括微调的小型语言模型、LRM上下文学习和监督的LRM适应。在基于SAT的分类基准上，Epi2Diff相较于监督的LRM微调基线实现了8.1%的平均相对增益。进一步分析表明，更难的项目引发了更费力、迭代和以实现为中心的事件动态，而不仅仅是更长的响应。这些结果表明，LRM推理轨迹中的认知事件为人类项目难度提供了可预测和可解释的过程表示，为使用推理模型进行教育测量提供了新的视角。

View on arXiv Download PDF AI Translation

cs.CL / 41 / 2606.28273

Vision-Default, Prior-Override: Causal Mechanisms of Perception-Knowledge Conflict in Vision-Language Models

视觉-默认，先验-覆盖：视觉-语言模型中感知-知识冲突的因果机制

Lietzow, Niclas, Bitterman, Danielle, Eickhoff, Carsten, Rudman, William, Golovanevsky, Michal

Abstract

Vision-language models must reconcile visual evidence with memorized world knowledge when the two conflict. How they resolve this conflict shapes the reliability of multimodal systems, yet prior work characterizes it behaviorally without a component-level causal account. We combine activation patching across three granularities (residual stream, attention heads, and MLP sublayers) with model-component ablation studies and mechanistic analysis. Across three VLM families, we find that visual grounding emerges by default, whereas prior grounding depends on a small set of causally necessary attention heads (2.5-4.8%) concentrated in the second half of the network. These heads enable answers from stored world knowledge (e.g., "red" for a strawberry) despite conflicting visual input. Ablating them flips predictions from knowledge-grounded to visually grounded answers in 68-96% of cases under prior-knowledge prompts, but changes only 0.8-7.5% of visually grounded predictions, establishing an asymmetric causal structure. The identified heads decompose into routing heads, which modulate information flow, and writing heads, which directly project answer tokens into the residual stream. This structure is consistent across model families and scales, revealing a sparse causal circuit underlying perception-knowledge conflict in VLMs.

Chinese Translation

视觉-语言模型在视觉证据与记忆中的世界知识发生冲突时，必须调和这两者。它们如何解决这一冲突影响多模态系统的可靠性，然而以往的研究仅从行为层面进行描述，而缺乏组件级的因果解释。我们结合了在三个粒度（残差流、注意力头和多层感知器子层）上的激活修补与模型组件消融研究和机制分析。在三种视觉-语言模型（VLM）家族中，我们发现视觉基础是默认出现的，而先验基础则依赖于一小组因果必要的注意力头（占比2.5%-4.8%），这些注意力头集中在网络的后半部分。这些注意力头使得即使在视觉输入冲突的情况下，仍能从存储的世界知识中得到答案（例如，“红色”对应草莓）。消融这些注意力头会使得在先验知识提示下，68%-96%的情况下预测从知识基础转变为视觉基础，但对视觉基础预测的改变仅为0.8%-7.5%，确立了不对称的因果结构。识别出的注意力头可分解为路由头（调节信息流）和写入头（直接将答案标记投射到残差流中）。这一结构在不同模型家族和规模中保持一致，揭示了在视觉-语言模型中感知-知识冲突的稀疏因果电路。

View on arXiv Download PDF AI Translation

arXiv Papers

Support-Constrained RL Enables Real-World Policy Improvement without Real-World Experience

AO-ARC: Almost-Surely Asymptotically Optimal Multi-Robot Motion Planning with ARC

Spacecraft Fiducial Marker for Autonomous Rendezvous, Proximity Operations, and Docking

SceneBot: Contact-Prompted General Humanoid Whole Body Tracking with Scene-Interaction

Learning to Throw: Agile and Accurate Cable-Suspended Payload Delivery with a Quadrotor

Physics-Guided Robotic Radiation Source Localization along Arbitrary Measurement Paths in Unstructured Environments

P-ARC: Exploiting Subproblem Independence for Parallel Multi-Robot Motion Planning

Direct Action-Head Injection of A Grounded 3D Point Unlocks Spatial and Task Generalization

CWI: Composite Humanoid Whole-Body Imitation System for Loco-manipulation

DIM-WAM: World-Action Modeling with Diverse Historical Event Memory

Drop-Then-Recovery: How Redundant Are Vision-Language-Action Models?

SpikeVLA: Vision-Language-Action Models with Spiking Neural Networks

LXD-SLAM: LiDAR+X Dense SLAM with $\sum_{i=0}^{5}C_5^i$ Configurable Sensor Combinations

Booster Lab: A Data-Centric Pipeline for Learning Deployable Humanoid Locomotion Policies

PPO-EAL: Exact Augmented Lagrangian Proximal Policy Optimization for Safe Robotic Control

LocalNav: Distilling Frontier VLMs and Embodied RL for On-Device Object Goal Navigation

S$^2$-VLA: State-Space Guided Vision-Language-Action Models for Long-Horizon Manipulation

Swarm sign language: motion-based communication between drones

Drifting in the Future: Stabilizing Path Following Drifting on High-Latency Vehicle Systems

When Multi-Robot Systems Meet Agentic AI:Towards Embodied Collective Intelligence

Building a Scalable, Reproducible, Evaluatable, and Closed-Loop Simulation Environment Foundation for Embodied Intelligence Cloud-Native Simulation Infrastructure for Embodied Intelligence Training, Evaluation, and Data Collection

Translation as a Bridging Action: Transferring Manipulation Skills from Humans to Robots

PA-BiCoop: A Primary-Auxiliary Cooperative Framework for General Bimanual Manipulation

Learning Stable In-Grasp Manipulation in a Non-Dropping Action Space

Unleashing Infinite Motion: Scaling Expressive Quadrupedal Motion via Generative Video Priors

SimFoundry: Modular and Automated Scene Generation for Policy Learning and Evaluation

CacheMPC: Certified Cached Model Predictive Control for Quadruped Locomotion

WARP-RM: A Warp-Augmented Relative Progress Reward Model for Data Curation

DexCompose: Reusing Dexterous Policies for Multi-Task Manipulation with a Single Hand

Not All Relations Rotate Alike: Transformation-Aware Decoupling for Viewpoint-Robust 3D Scene Graph Generation

SemCityLoc: Aerial 6DoF Localization Using Semantic 3D City Models

Fine-tuning a multimodal large language model for clinician-grade autism behavioral scoring from short home videos

SelectAnyTree: A Promptable Instance Segmentation Model for 3D Forest LiDAR Point Clouds

DMV-Bench: Diagnosing Long-Horizon Multimodal Agents' Visual Memory with Incidental Cue Injection

Aloe-Vision: Robust Vision-Language Models for Healthcare

ReWorld: Learning Better Representations for World Action Models

TruEye: Fine-Grained Detection of AI-Generated Human Subjects in Images

Structured-Li-GS: Structured 3D Gaussians Splatting with LiDAR Incorporation and Spatial Constraints

Tessellating The Earth

Large Language Model Teaches Visual Students: Cross-Modality Transfer of Fine-Grained Conceptual Knowledge

MemoBench: Benchmarking World Modeling in Dynamically Changing Environments

Beyond MoCap: Scaling Motion Tokenizers with Synthetic Human Motion for Generative Modeling

Understanding Cross-Rig Generalization in Automotive Perception: a Multi-Rig Benchmark and Rig Variation Metrics

Radar Guided Camera Verification for Automatic Emergency Braking Rethinking Object Detection in Radar Camera Fusion

Perceptual 3D Simulation With Physical World Modeling

DeLux: Cross-Modal Local Artifact Restoration in Video Using Neuromorphic Data

Distribution-based deep multiple instance learning for tumor proportion scoring in NSCLC

Beyond Points: Spherical Distributional Part Prototypes for Interpretable Classification

CoIn: Comprehensive 2D-3D Inpainting with Gaussian Splatting Guidance

Dismantling Pathological Shortcuts: A Causal Framework for Faithful LVLM Decoding

Qwen-Image-2.0-RL Technical Report

Denoising ICF Images with Multiplicative Uniform Noise: A Self-Supervised Study Based on the Log-Domain Noisier2Inverse Framework

AI-Generated Image Recognition via Fusion of CNNs and Vision Transformers

CascadeOcc: Rethinking 3D Occupancy World Models with Cascaded VQ Representations

VLM-Aware Meta-Optic Front-End Design for Frozen Vision-Language Models

Temporal-Emerged Prompting for Segment Anything in Multiframe Infrared Small Target Detection

GeoFace: Consistent Multi-View Face Generation with Geometry-Constrained Diffusion

MVPruner: Dynamic Token Pruning for Accelerating Multi-view Vision-Language Models in Autonomous Driving

Explainable AI for Biodiversity Monitoring and Ecological Image Analysis

Multi-Modal Conditioned High-Resolution Transformer for Urban Electromagnetic Field Map Prediction Download PDF

Two-Stage Cross-Domain Cervical Abnormality Screening with Cytopathological Image Synthesis and Knowledge Distillation

Joint Transcription and Decryption of Images of Encrypted Handwritten Documents: A Comparison with the Traditional Pipeline

ZooClaw-FashionSigLIP2: Distilled Fine-tuning for Robust Fashion Retrieval

MASS: Motion-Aligned Selective Scan for Refinement in Flow-Based Video Frame Interpolation

Scene and Human in One World: Reconstruction in a Feedforward Pass

Learning 1-Bit LiDAR-based Localization with Auxiliary Objective

SIFT: Self-Imagination Fine-Tuning for Physically Plausible Motion in Video Diffusion Models

Panoramic Scene Analysis: A Survey from Distortion-Aware Engineering to Sphere-Native Foundation Modeling

PixelU: A U-Shaped Transformer for Efficient End-to-End Pixel Diffusion

An Embedded Real-Time License Plate Recognition System for Complex Traffic Scenes

ModaFlow: Modality-Aware Flow Matching for High-Fidelity Virtual Try-On

TRUST: Efficient Abdominal Trauma Recognition via Image-to-Ultrasound-Video Transfer Learning

MindFlow: Harmonizing Cognitive Semantics and Acoustic Dynamics for Facial Animation Generation in Dyadic Conversations

Improving Adversarial Robustness via Activation Amplification and Attenuation

Text as Illumination: Spatial Contrastive Retinex Learning for Language-guided Medical Image Segmentation

Scalable and Differentiable Point-Cloud Registration Using Maximum Mean Discrepancy

Video-MME-Logical: A Controlled Diagnostic Benchmark for Video Temporal-Logical Reasoning

CSD: Content-aware Speculative Decoding for Efficient Image Generation

Hippocampus-DETR: An Explicit Memory Object Detection Framework Based on Hippocampus Modeling