2026-09论文中文摘要
在时空系统中先分解动态再学习依赖关系
Decompose Dynamics Before Learning Dependencies in Spatiotemporal Systems
arXiv 2609.36637 · 2026-09-30什么使得循环在循环语言模型中有效？
What Makes Recurrence Effective in Looped Language Models?
arXiv 2609.36636 · 2026-09-30WitnessGym：在错误见证构建任务上对编码智能体进行基准测试
WitnessGym: Benchmarking Coding Agents on the Construction of Bug Witnesses
arXiv 2609.36635 · 2026-09-30查找表无损压缩用于硬件应用
Lossless Compression of Lookup Tables for Hardware Applications
arXiv 2609.36634 · 2026-09-30面向信息物理系统的物理感知机器遗忘
Physics-Aware Machine Unlearning for Cyber-Physical Systems
arXiv 2609.36633 · 2026-09-30宇宙中微子漫射通量在10^16 eV至10^26 eV范围内的联合约束
Combined constraints on the diffuse flux of cosmic neutrinos between 10^16 eV and 10^26 eV
arXiv 2609.36632 · 2026-09-30一种用于最优买价和卖价动态的价差门控霍克斯-群聚模型，及其在限价单放置中的应用
A Spread-Gated Hawkes-Flocking Model for Best Bid and Ask Dynamics, with an Application to Limit Order Placement
arXiv 2609.36631 · 2026-09-30智能体系统蒸馏：跨模型、工件与执行框架的路线图
Distilling Agentic Systems: A Roadmap across Models, Artifacts, and Harnesses
arXiv 2609.36630 · 2026-09-30Picard群的Farey符号
Farey Symbols for the Picard Group
arXiv 2609.36629 · 2026-09-30超越二元偏好：面向肢体动作描述的分级偏好优化
Beyond Binary Preferences: Graded Preference Optimization for Limb-Motion Captioning
arXiv 2609.36628 · 2026-09-30有损信道中的同步量子设备
Synchronized Quantum Devices in Lossy Channels
arXiv 2609.36627 · 2026-09-30语义投影用于语言代理的持续自我进化
Semantic Projection for Continual Self-Evolution of Language Agents
arXiv 2609.36626 · 2026-09-30核形状演化中四极-十六极关联的半经典描述
Semiclassical description of quadrupole-hexadecapole correlation in nuclear shape evolution
arXiv 2609.36625 · 2026-09-30DualTrack：通过预训练先验的对称耦合实现同步语音-手势生成
DualTrack: Synchronized speech-gesture generation via symmetric coupling of pretrained priors
arXiv 2609.36624 · 2026-09-30测地线条件下的端点传递图分类
The classification of endpoint-transitive graphs with geodesic condition
arXiv 2609.36622 · 2026-09-30阿秒圆二色光谱探测空穴环电流
Attosecond circular-dichroism spectroscopy of hole ring currents
arXiv 2609.36621 · 2026-09-30神经结构推理器：一种受大脑启发的结构化知识推理架构
Neural Structural Reasoner: A Brain-inspired Architecture for Reasoning over Structured Knowledge
arXiv 2609.36620 · 2026-09-30SemPSG：一种面向多导睡眠图分析的语义通道感知基础模型
SemPSG: A Semantic Channel-Aware Foundation Model for Polysomnography Analysis
arXiv 2609.36619 · 2026-09-30RobotEQ 3.0：迈向具身智能体中的个性化社会主动智能
RobotEQ 3.0: Towards Personalized Social Proactive Intelligence in Embodied Agents
arXiv 2609.36618 · 2026-09-30CrossTimeEdit：跨十年跨视角数据集与奖励引导编辑用于历史街景生成
CrossTimeEdit: A Decade-Spanning Cross-View Dataset and Reward-Guided Editing for Historical Street-View Generation
arXiv 2609.36616 · 2026-09-30CI-PINN：用于求解演化方程的因果积分物理信息神经网络
CI-PINN: Causal Integral Physics-Informed Neural Network for Solving Evolution Equations
arXiv 2609.36615 · 2026-09-30选择性引导作为商业影响渠道：一个可复现的合成购物智能体压力测试
Selective Elicitation as a Commercial Influence Channel: A Reproducible Synthetic Shopping-Agent Stress Test
arXiv 2609.36614 · 2026-09-30区组传递的 $5$-$(v,k,2)$ 设计与 $k$ 整除 $v$
Block-Transitive $5$-$(v,k,2)$ Designs with $k$ divides $v$
arXiv 2609.36613 · 2026-09-30大型语言模型真的会遗忘吗？模型遗忘中的隐藏状态泄漏及其修复方法
Do LLMs Really Forget? Hidden-State Leakage in Model Unlearning and How to Fix it
arXiv 2609.36612 · 2026-09-30多通道缓解事实核查强化学习智能体中的来源信任捷径
Multi-Channel Mitigation of Source-Trust Shortcuts in Fact-Checking RL Agents
arXiv 2609.36611 · 2026-09-30通过更快收敛与压缩实现通信高效的不可知联邦学习
Communication-Efficient Agnostic Federated Learning via Faster Convergence and Compression
arXiv 2609.36610 · 2026-09-30脑电图分析的最佳实践：预处理、建模与机器学习
Best Practices in EEG Analysis: Preprocessing, Modeling, and Machine Learning
arXiv 2609.36609 · 2026-09-30先行动，后推理：通过参考条件逆动力学加速多轮智能体的在线策略蒸馏
Act First, Reason Later: Accelerating On-Policy Distillation for Multi-Turn Agents via Reference-Conditioned Inverse Dynamics
arXiv 2609.36608 · 2026-09-30像素级曝光用于高鲁棒性车载远程光电容积描记
Pixel-wise Exposure for Highly Robust In-Vehicle Remote-PPG
arXiv 2609.36607 · 2026-09-30带隙节线半金属中由能带几何激活的磁场驱动输运
Magnetic-field-activated transport from band geometry in gapped nodal-line semimetals
arXiv 2609.36606 · 2026-09-30RoboChrono：用于流式任务理解的真实机器人基准
RoboChrono: A Real Robot Benchmark for Streaming Task Understanding
arXiv 2609.36605 · 2026-09-30颜色关系与离壳双拷贝：理论塔
Color Relations and Off-Shell Double-Copy for Towers of Theories
arXiv 2609.36604 · 2026-09-30自进化防御：面向LLM智能体的持续安全策略学习
Self-Evolving Defense: Continual Security Policy Learning for LLM Agents
arXiv 2609.36603 · 2026-09-30OTRetarget：通过最优传输实现机器人与物体运动的联合重定向
OTRetarget: Joint Robot and Object Motion Retargeting via Optimal Transport
arXiv 2609.36602 · 2026-09-30SAKI：面向在线策略蒸馏的最大耦合路由教师监督
SAKI: Maximal-Coupling-Routed Teacher Supervision for On-Policy Distillation
arXiv 2609.36601 · 2026-09-30二阶矩随机逼近方法
Second-Moment Stochastic Approximation Methods
arXiv 2609.36600 · 2026-09-30扩展视频生成用于推理：代价是什么？
Scaling Video Generation for Reasoning: At What Cost?
arXiv 2609.36599 · 2026-09-30通过可行性证书的执行器感知候选选择
Executor-aware Candidate Selection via a Feasibility Certificate
arXiv 2609.36597 · 2026-09-30HACo：学习力觉感知灵巧操作中的触觉主动柔顺
HACo: Learning Haptic Active Compliance for Force-Aware Dexterous Manipulation
arXiv 2609.36596 · 2026-09-30简单智能体记忆用于通用机器人策略
Simple Agentic Memory for Generalist Robot Policies
arXiv 2609.36595 · 2026-09-30一般矩变化的优化检测：均值与协方差变化检测及其扩展
Optimal detection of general moment changes: Simultaneous mean and covariance change detection and beyond
arXiv 2609.36594 · 2026-09-30隐藏在激光拍频统计中的两条线宽的精确恢复
Accurate recovery of the two linewidths hidden in laser beatnote statistics
arXiv 2609.36592 · 2026-09-30化学驱动浴的力学特征
Mechanical Signature of a Chemically Driven Bath
arXiv 2609.36591 · 2026-09-30SEED：通过隐式编码器-解码器实现自推测解码
SEED: Self-Speculative Decoding via Implicit Encoder-Decoder
arXiv 2609.36590 · 2026-09-30非线性定价恢复数据卖方的可计算性
Non-Linear Pricing Restores Tractability for a Data Seller
arXiv 2609.36589 · 2026-09-30协作多智能体视觉-语言-动作模型：基于强化微调
Cooperative Multi-Agent Vision-Language-Action Models via Reinforced Fine Tuning
arXiv 2609.36588 · 2026-09-30RLVR 中学得的报告偏好可能与当前请求冲突
Learned Reporting Preferences in RLVR Can Conflict with the Current Request
arXiv 2609.36587 · 2026-09-30Transformer过早停止思考，一个微小的LoRA即可修复
Transformers Stop Thinking Too Early, and a Tiny LoRA Fixes It
arXiv 2609.36585 · 2026-09-30让模拟训练规模化：协同设计映射、优化器与转换器
Making Analog Training Scale: Co-Designing Mapping, Optimizer, and Converters
arXiv 2609.36584 · 2026-09-30智能电网广域控制系统的攻击韧性分析
Attack-Resiliency Analytics for Wide-Area Control Systems in Smart Grids
arXiv 2609.36583 · 2026-09-30
