2026-09论文中文摘要
Video2Skill：从流式经验到可复用的具身技能
Video2Skill: From Streaming Experience to Reusable Embodied Skills
arXiv 2609.36691 · 2026-09-30硅微环调制器每通道200 Gb/s以上协同设计：器件物理、工作点与面向扩展和横向扩展光I/O的紧凑模型
Co-design of Silicon Microring Modulator beyond 200 Gb/s per Lane: Device Physics, Operating Point, and Compact Models for Scale-Up and Scale-Out Optical I/O
arXiv 2609.36690 · 2026-09-30CHAIN：通过因果-时序超图推断实现校准的LLM预测
CHAIN: Calibrated LLM Forecasting via Causal-Temporal Hypergraph Inference
arXiv 2609.36689 · 2026-09-30GRP v0.1 技术报告
GRP v0.1 Technical Report
arXiv 2609.36688 · 2026-09-30MyoCodec：一种用于肌电信号的流式神经编解码器
MyoCodec: A Streaming Neural Codec for Electromyography
arXiv 2609.36687 · 2026-09-30根因分析失效之处：检索-重排序分解
Where Root Cause Analysis Fails: A Retrieval-Reranking Decomposition
arXiv 2609.36686 · 2026-09-30当语义重要时：共语手势生成的可靠性感知语义节奏控制
When Semantics Matter: Reliability-Aware Semantic-Rhythm Control for Co-Speech Gesture Generation
arXiv 2609.36685 · 2026-09-30ProgressCompass：具身进度奖励模型在缺乏正确上下文时会迷失方向
ProgressCompass: Embodied Progress Reward Models Are Lost Without the Right Context
arXiv 2609.36684 · 2026-09-30MARCO：用于条件分子优化的多轮智能体强化学习
MARCO: Multi-Round Agentic Reinforcement for Conditional Molecular Optimization
arXiv 2609.36683 · 2026-09-30极性域体积作为离子液体输运的统一描述符
Polar-Domain Volume as a Unified Descriptor of Transport in Ionic Liquids
arXiv 2609.36682 · 2026-09-30热力学位错理论中的平均化：一般宏观均匀应力与应变状态
Averaging in thermodynamic dislocation theory: general macroscopically uniform stress and strain states
arXiv 2609.36681 · 2026-09-30通过结构化提示重参数化重新编程视觉-语言模型
Reprogramming Vision-Language Models via Structured Prompt Reparameterization
arXiv 2609.36680 · 2026-09-30理解私有进化作为学习增强的聚类
Understanding Private Evolution as Learning-Augmented Clustering
arXiv 2609.36678 · 2026-09-30运动学非线性时空轨迹扭曲用于接触丰富的灵巧操作演示
Kinematic Nonlinear Spatio-Temporal Trajectory Warping for Contact-Rich Dexterous Manipulation Demonstrations
arXiv 2609.36676 · 2026-09-30Gödel Forest：面向数据为中心的递归自我改进的搜索深度与广度平衡
Gödel Forest: Balancing Search Depth and Breadth for Data-Centric Recursive Self-Improvement
arXiv 2609.36675 · 2026-09-30与图相关的顶点覆盖 Artinian 代数的弱和强 Lefschetz 性质
Weak and strong Lefschetz properties for vertex cover Artinian algebras associated to graphs
arXiv 2609.36674 · 2026-09-30二元透镜事件贝叶斯推断框架：完全纳入高阶效应
A Bayesian Inference Framework for Binary Lens Events Fully Incorporating Higher-Order Effects
arXiv 2609.36673 · 2026-09-30FairDiff：缓解扩散推荐模型中的自我强化马太效应
FairDiff: Mitigating the Self-Reinforcing Matthew Effect in Diffusion Recommender Models
arXiv 2609.36671 · 2026-09-30FineSID：面向生成式推荐的可扩展且高效的语义标识符学习
FineSID: Scalable and Efficient Semantic Identifier Learning for Generative Recommendation
arXiv 2609.36670 · 2026-09-30穿孔和多孔插入件的前缘降噪：无拟合常数的可压缩有限弦预测
Leading-edge noise reduction by perforated and porous inserts: a compressible finite-chord prediction without fitted constants
arXiv 2609.36669 · 2026-09-30带Polyak步长和Armijo线搜索的随机重球法：一般收敛性分析
Stochastic Heavy Ball with Polyak Step Size and Armijo Line Search: A General Convergence Analysis
arXiv 2609.36668 · 2026-09-30当一次泄露永久获利：上下文绑定与威慑共谋的代价
When One Leak Pays Forever: Context Binding and the Price of Deterring Collusion
arXiv 2609.36667 · 2026-09-30一种具有自适应自线性化的鲁棒赫兹线宽量子点相干扫频源
A Robust Hertz-Linewidth Quantum-Dot Coherent Swept Source with Adaptive Self-Linearization
arXiv 2609.36666 · 2026-09-30关于钩子分区的单项式隐式量的非负性
On the nonnegativity of monomial immanants for hook partitions
arXiv 2609.36665 · 2026-09-30拓扑节线半金属的线性和非线性输运响应
Linear and nonlinear transport responses of topological nodal-line semimetals
arXiv 2609.36664 · 2026-09-30GenLimitLib：极限语言生成与AI辅助数学研究的正式库
GenLimitLib: A Formal Library for Language Generation in the Limit and AI-Assisted Mathematical Research
arXiv 2609.36663 · 2026-09-30AutoLoCo：通过自适应同步实现通信高效的大规模分布式LLM训练
AutoLoCo: Communication Efficient Distributed LLM Training via Adaptive Synchronization
arXiv 2609.36662 · 2026-09-30你只需重新编程一次：重新思考视觉重新编程的长时间训练
You Only Reprogram Once: Rethinking Prolonged Training for Visual Reprogramming
arXiv 2609.36661 · 2026-09-30拜占庭鲁棒的联邦表示学习
Byzantine-Robust Federated Representation Learning
arXiv 2609.36660 · 2026-09-30策略内参数更新方向是LLM后训练泛化能力的基础
On-Policy Parameter Update Direction Underlies Generalization in LLM Post-Training
arXiv 2609.36659 · 2026-09-30纳米束与量子点耦合产生的单光子非线性
Single-Photon Nonlinearity from a Nanobeam with a Quantum Dot
arXiv 2609.36658 · 2026-09-30宪法适配器：针对错位与滥用的推理时干预
Constitutional adapters: Inference-time interventions for misalignment and misuse
arXiv 2609.36657 · 2026-09-30宇宙黎明时期的电离超结构：天体物理研究基础模型的发现
An Ionized Superstructure at Cosmic Dawn Revealed by a Foundation Model for Astrophysical Research
arXiv 2609.36656 · 2026-09-30并非每次修正都有益：增益引导的持续测试时自适应
Not Every Correction Helps: Gain-Guided Continual Test-Time Adaptation
arXiv 2609.36655 · 2026-09-30重放曲率：面向大语言模型推理的准确且可扩展的NVFP4量化方法
Replay the Curvature: Accurate and Scalable NVFP4 Quantization for Large Language Model Inference
arXiv 2609.36654 · 2026-09-30循环Transformer中的递归推理调度
Scheduling Recursive Reasoning in Looped Transformers
arXiv 2609.36653 · 2026-09-30RankBuffer：面向开放式生成的基于排序的高效奖励
RankBuffer: Efficient Ranking-Based Rewards for Open-Ended Generation
arXiv 2609.36652 · 2026-09-30FocusVTC：自适应分辨率的高效高性能视觉文本压缩
FocusVTC: Efficient and High-Performance Visual Text Compression with Adaptive Resolution
arXiv 2609.36651 · 2026-09-30黑洞微分可观测量的阈值尾部
Threshold Tails of Black hole Differential Observables
arXiv 2609.36650 · 2026-09-30膨胀且旋转火球中的椭圆流
Elliptic flow in an expanding and rotating fireball
arXiv 2609.36649 · 2026-09-30VLM4Cluster：视觉语言预训练时代的深度聚类基准
VLM4Cluster: Benchmarking Deep Clustering In the Era of Vision-Language Pre-training
arXiv 2609.36648 · 2026-09-30CTE-Bench：有状态软件模拟器的反事实轨迹评估
CTE-Bench: Counterfactual Trace Evaluation for Stateful Software Simulators
arXiv 2609.36647 · 2026-09-30TRACE：面向可重构天线的三混合架构的三线性信道估计框架
TRACE: A Trilinear Channel Estimation Framework for Tri-Hybrid Architectures with Reconfigurable Antennas
arXiv 2609.36646 · 2026-09-30OCA：基于ODE驱动的交叉注意力用于图像到点云配准
OCA: ODE-Driven Cross-Attention for Image-to-Point-Cloud Registration
arXiv 2609.36644 · 2026-09-30基于大规模整车测量的车载天线辐射方向图真实表征与建模
Realistic Characterization and Modeling of Vehicle-Mounted Antenna Radiation Patterns Based on Large-Scale Full-Vehicle Measurements
arXiv 2609.36643 · 2026-09-30PR-OPD：用于智能体强化学习的特权表示在线策略自蒸馏
PR-OPD: Privileged Representation On-policy Self-Distillation for Agentic Reinforcement Learning
arXiv 2609.36642 · 2026-09-30从仅结果强化学习中诱导过程监督
Inducing Process Supervision from Outcome-Only Reinforcement Learning
arXiv 2609.36641 · 2026-09-30基础模型引导的拓扑感知语义风险场用于操作
Foundation-Model-Guided Topology-Aware Semantic Risk Fields for Manipulation
arXiv 2609.36640 · 2026-09-30用于基于物理生成真实实验扫描隧道显微镜图像的数字仿真工具包
A Digital Simulation Toolkit for Physics-Based Generation of Realistic Experimental Scanning Tunneling Microscopy Images
arXiv 2609.36639 · 2026-09-30PE-OPSD：通过在线策略自蒸馏将提示增强内化到流匹配模型中
PE-OPSD: Internalizing Prompt Enhancement into Flow-matching Models via On-Policy Self-Distillation
arXiv 2609.36638 · 2026-09-30
