2026-10论文中文摘要
基于计划条件模仿的密集杂乱场景自遮挡下鲁棒物体抓取
Plan-Conditioned Imitation for Robust Object Retrieval under Self-Occlusion in Dense Clutter
arXiv 2609.38857 · 2026-10-01解耦球面推理与稠密预测以实现360度深度估计
Decoupling Spherical Reasoning from Dense Prediction for 360 Depth Estimation
arXiv 2609.38856 · 2026-10-01在线演化策略用于基于流匹配的视觉-语言-动作策略，通过自监督轨迹分布优化
Online Evolution Strategy for Flow-Matching VLA Policies via Self-Supervised Trajectory Distribution Optimization
arXiv 2609.38855 · 2026-10-01缓解长度扩展税：在线蒸馏方法
Mitigating the Length-Scaling Tax with Online Distillation
arXiv 2609.38854 · 2026-10-01生成扩散模型中的分布覆盖可视化
Visualizing Distribution Coverage in Generative Diffusion Models
arXiv 2609.38853 · 2026-10-01基于运动的地面人形足球：多方向踢球库的任务门控强化学习
Locomotion-Grounded Humanoid Soccer: Task-Gated Reinforcement Learning of a Multi-Directional Kicking Library
arXiv 2609.38852 · 2026-10-01多模态大语言模型为何失败及如何失败：基于因果任务分解的能力缺陷诊断
Where MLLMs Fail and Why: Causal Task Decomposition for Capability Failure Diagnosis
arXiv 2609.38851 · 2026-10-01OpenJev-RLCD：一个可运行的RLCD实现
OpenJev-RLCD: A Working RLCD Implementation
arXiv 2609.38850 · 2026-10-01临界椭圆矩条件下非散度型方程的定量均匀化与大尺度正则性
Quantitative homogenization and large-scale regularity for nondivergence-form equations under a critical ellipticity moment
arXiv 2609.38849 · 2026-10-01Affleck-Kennedy-Lieb-Tasaki模型中疤痕态的量子淬火：基于Clifford增强张量网络模拟
Quantum quenches of scar states in the Affleck-Kennedy-Lieb-Tasaki model via Clifford augmented tensor network simulation
arXiv 2609.38848 · 2026-10-01得分更高，回答更差：通过协议级评分标准缓解基于评分标准的强化学习中的奖励黑客行为
Scoring Higher, Answering Worse: Mitigating Reward Hacking in Rubric-Based RL via Protocol-Level Rubrics
arXiv 2609.38847 · 2026-10-01珊瑚育成机器人评估系统（CGRAS）：通过机器人和计算机视觉扩展珊瑚幼体监测规模
Coral Grow-out Robotic Assessment System (CGRAS): Scaling Coral Recruit Monitoring Through Robotics and Computer Vision
arXiv 2609.38846 · 2026-10-01IRS辅助无线系统中的盲干扰抑制：一种统计信道比估计方法
Blind Interference Suppression in IRS-Aided Wireless Systems: A Statistical Channel Ratio Estimation Approach
arXiv 2609.38845 · 2026-10-01$(k,t)$-Fibonacci 与 $(k,t)$-Lucas 多项式不可约因子的单项性
Monogenity of the irreducible factors of $(k,t)$-Fibonacci and $(k,t)$-Lucas polynomials
arXiv 2609.38844 · 2026-10-01$^{229}$Th:CaF$_2$晶体中的缺陷动力学与同核异能素淬灭
Defect Dynamics and Isomer Quenching in $^{229}$Th:CaF$_2$ Crystals
arXiv 2609.38843 · 2026-10-01先学习后微分的梯度估计
Learn-Then-Differentiate Gradient Estimation
arXiv 2609.38842 · 2026-10-01双曲守恒律稳态问题的一种带WENO-JS局部求解器的全收敛不动点快速扫描方法
A fully convergent fixed-point fast sweeping method with the WENO-JS local solver for steady state of hyperbolic conservation laws
arXiv 2609.38841 · 2026-10-01FrameMorrow：面向长时程视频生成的未来引导帧选择与前瞻令牌
FrameMorrow: Future-guided Frame Selection with Prospective Tokens for Long-Horizon Video Generation
arXiv 2609.38839 · 2026-10-01正 Bakry-Émery 曲率下的格林函数单调性
Green-Function Monotonicity under Positive Bakry-Émery Curvature
arXiv 2609.38838 · 2026-10-01脉冲星应力达到阈值时由星震触发的自转突变
Pulsar glitches triggered by quakes when stress reaches a threshold
arXiv 2609.38837 · 2026-10-01共面POVM的联合可测性
Joint measurability of coplanar POVMs
arXiv 2609.38836 · 2026-10-01量子零和博弈中乐观矩阵镜像近端的平均迭代与最后迭代下界
Average-and Last-Iterate Lower Bounds for Optimistic Matrix Mirror-Prox in Quantum Zero-Sum Games
arXiv 2609.38835 · 2026-10-01带间隔的对比学习的 VC 维最优界
Optimal VC Dimension of Contrastive Learning with Margin
arXiv 2609.38834 · 2026-10-01ReSCENE：联邦持续学习中灾难性遗忘的结构性缓解的服务器端重放
ReSCENE: Server-Side Replay for Structural Mitigation of Catastrophic Forgetting in Federated Continual Learning
arXiv 2609.38833 · 2026-10-01缩放注意力中的参数与上下文：来自多头混合的原生稀疏注意力
Scaling Parameter and Context in Attention: Native Sparse Attention from Mixture-of-Head
arXiv 2609.38832 · 2026-10-01通过定向改写伪造LLM作者指纹
Forging LLM Authorship Fingerprints with Targeted Rewriting
arXiv 2609.38831 · 2026-10-01SparLeak：共享GPU上LLM推理中稀疏注意力导致的隐私泄露
SparLeak: Privacy Leakage from Sparse Attention in LLM Inference on Shared GPUs
arXiv 2609.38830 · 2026-10-01多路径LLM推理的多样性合并
Diversity Combining for Multi-Path LLM Reasoning
arXiv 2609.38829 · 2026-10-01有序Ramsey数的森林和有界度图的上界
Upper bounds for ordered Ramsey numbers of forests and bounded-degree graphs
arXiv 2609.38828 · 2026-10-01更多选择，更少决策：JEV类直接决策模型中的序数尺度偏差
More Choices, Fewer Decisions: Ordinal-Scale Bias in JEV-like Direct-Decision Models
arXiv 2609.38827 · 2026-10-01时间反演对称性破缺实现手性连续域束缚态的实验演示
Experimental Demonstration of Chiral Bound States in the Continuum Enabled by Time-reversal Symmetry Breaking
arXiv 2609.38826 · 2026-10-01层间自旋关联的光学观测
Optical observation of interlayer spin correlation
arXiv 2609.38825 · 2026-10-01PhaseSync-Exo：基于人体时钟锚定的参考自适应动态步态跟踪
PhaseSync-Exo: Human Clock Anchored Reference Adaptation for Dynamic Gait Tracking
arXiv 2609.38824 · 2026-10-01DecoMoE：解耦视觉传播与专家计算以实现高效多模态MoE推理
DecoMoE: Decoupling Visual Propagation and Expert Computation for Efficient Multimodal MoE Inference
arXiv 2609.38823 · 2026-10-01SkillSeek：在市场规模的智能体技能检索中重新审视
SkillSeek: Revisiting Agent Skill Retrieval at Marketplace Scale
arXiv 2609.38822 · 2026-10-01BARRAC：一种英文基于方面的情感分析方法对阿拉伯方言分类任务的适配
BARRAC: Adaptation of an English Aspect-based Sentiment Analysis Approach for Classification Tasks in Arabic Dialects
arXiv 2609.38820 · 2026-10-01未来视频生成比观察视频更符合人类视觉皮层
Future Video Generation Better Aligns with the Human Visual Cortex than Observed Video
arXiv 2609.38819 · 2026-10-01谁的呼声能在摘要中幸存？LLM员工倾听中的声音保留审计
Whose Voice Survives the Summary? A Voice-Retention Audit of LLM Employee Listening
arXiv 2609.38818 · 2026-10-01当推理偏离正轨：不受控推理的注意力动态
When Reasoning Goes Astray: Attention Dynamics of Uncontrolled Reasoning
arXiv 2609.38817 · 2026-10-01你被录用了：面向大语言模型协作的战略性模型选择
You're Hired: Strategic Model Selection for LLM Collaboration
arXiv 2609.38816 · 2026-10-01纠缠的相对熵与三方极小曲面
Relative entropy of entanglement and tripartite minimal surface
arXiv 2609.38815 · 2026-10-01无见混沌而学混沌：自回归Transformer中全局动力学的外推
Learning Chaos Without Seeing Chaos: Extrapolation of Global Dynamics in Autoregressive Transformers
arXiv 2609.38814 · 2026-10-01心理健康支持热线作为容量有限的服务：需求、评估与未评估呼叫的联合建模
Mental Health Support Hotlines as Capacity-Limited Services: Joint Modeling of Demand, Assessment, and Unassessed Calls
arXiv 2609.38813 · 2026-10-01终端智能体能否信任自身的验证？诊断与改进自我验证
Can Terminal Agents Trust Their Own Verification? Diagnosing and Improving Self-Verification
arXiv 2609.38812 · 2026-10-01DCM-SAM：面向NPU部署的增材制造缺陷分割的缺陷条件LoRA专家混合模型
DCM-SAM: Defect-Conditioned Mixture of LoRA Experts for NPU-Deployed AM Defect Segmentation
arXiv 2609.38811 · 2026-10-01StateTree：通过强化学习增强长期对话推理
StateTree: Enhancing Long-Term Dialogue Reasoning via Reinforcement Learning
arXiv 2609.38809 · 2026-10-01Halpern加速的带退化预条件的最优非遍历收敛的majorized ADMM
Halpern-Accelerated Majorized ADMM with Optimal Non-Ergodic Convergence under Degenerate Preconditioning
arXiv 2609.38808 · 2026-10-01PatchHolmes：通过列表式选择进行智能体补丁检索
PatchHolmes: Agentic Patch Retrieval via Listwise Selection
arXiv 2609.38807 · 2026-10-01面向LLM智能体基于强化学习的后训练的显式轨迹多样性
Explicit Trajectory Diversity for RL-Based Post-Training of LLM Agents
arXiv 2609.38805 · 2026-10-01具有二次约束的状态相关不确定性的鲁棒规划
Robust Planning with Quadratically Constrained State-Dependent Uncertainties
arXiv 2609.38803 · 2026-10-01
