2026-09论文中文摘要
GLIDE：面向异构LLM智能体的广义逐层内在分布评估
GLIDE: Generalized Layer-wise Intrinsic Distributional Evaluation for Heterogeneous LLM Agents
arXiv 2609.32295 · 2026-09-29丰富群胚扩张的全局与局部平凡性：一种代数方法
Global and local triviality for extensions of ample groupoids: an algebraic approach
arXiv 2609.32294 · 2026-09-29TRAP：理解并缓解语言模型中的隐私记忆
TRAP: Understanding and Mitigating Privacy Memorization in Language Models
arXiv 2609.32293 · 2026-09-29通往稳定性边缘的旅程
A Journey to the Edge of Stability
arXiv 2609.32290 · 2026-09-29神经ODE遇见并发学习：具有Lyapunov保证的稳定在线学习
Neural ODEs Meet Concurrent Learning: Stable Online Learning with Lyapunov Guarantees
arXiv 2609.32289 · 2026-09-29预训练数据投毒的可解理论：状态依赖的标度指数
A Solvable Theory of Pre-training Data Poisoning: Regime-Dependent Scaling Exponents
arXiv 2609.32288 · 2026-09-29Bernstein--Sato 理想可能不是线性的
Bernstein--Sato ideals might not be linear
arXiv 2609.32287 · 2026-09-29Single-or-Sample：面向组合型智能体的在线公平分配
Single-or-Sample: Online Fair Allocation for Combinatorial Agents
arXiv 2609.32286 · 2026-09-29音频预处理对口吃检测的影响：类别特定分析
Audio Preprocessing Effects on Stuttering Detection: A Class-Specific Analysis
arXiv 2609.32285 · 2026-09-29恒星形成星系尘埃消光标度关系中弥散的物理起源：恒星—尘埃几何
The Physical Origins of Scatter in the Dust Attenuation Scaling Relation of Star-Forming Galaxies: Star--Dust Geometry
arXiv 2609.32284 · 2026-09-29AgentReplay：令牌级轨迹重放对于公平的服务系统性能基准测试至关重要
AgentReplay: Token-Wise Trace Replay Is Essential for Fair Serving System Performance Benchmarking
arXiv 2609.32283 · 2026-09-29认证的贝叶斯最优实验设计：从期望信息增益到信噪比
Certified Bayesian optimal experimental design: from Expected Information Gain to Signal-to-Noise Ratio
arXiv 2609.32282 · 2026-09-29从内部支持与践行文化
Supporting and Performing Culture from the Inside
arXiv 2609.32281 · 2026-09-29HeroFrame-Bench：基于评分标准-排名协同演化的参考锚定电影英雄帧选择评估基准
HeroFrame-Bench: Reference-Anchored Evaluation via Rubric--Ranking Co-Evolution for Movie Hero Frame Selection
arXiv 2609.32280 · 2026-09-29SIMANF：通过归一化流中的模拟退火实现无样本学习非归一化分布
SIMANF: Sample Free Learning of Unnormalized Distributions via Simulated Annealing in Normalizing Flows
arXiv 2609.32279 · 2026-09-29RR-Evict：面向智能体LLM服务中超越LRU的细粒度前缀缓存逐出策略
RR-Evict: Fine-Grained Prefix Cache Eviction beyond LRU for Agentic LLM Serving
arXiv 2609.32278 · 2026-09-29规范声学黑洞的稳定性分析
Stability Analysis of Canonical Acoustic Black Hole
arXiv 2609.32277 · 2026-09-29HyperLabel：基于超图标签相关性建模的多标签分类
HyperLabel: Multi-Label Classification via Hypergraph-Based Label Correlation Modeling
arXiv 2609.32276 · 2026-09-29拓扑自适应双曲图注意力网络：由双曲Sombor指数引导
Topology-Adaptive Hyperbolic Graph Attention Networks Guided by the Hyperbolic Sombor Index
arXiv 2609.32275 · 2026-09-29技能何时增加价值？面向选择性技能使用的任务条件增益预测
When Does a Skill Add Value? Task-Conditional Gain Prediction for Selective Skill Use
arXiv 2609.32274 · 2026-09-29FSS-UBrain：多区域少样本脑肿瘤MRI分割
FSS-UBrain: Multi-region Few-Shot Brain Tumor MRI Segmentation
arXiv 2609.32273 · 2026-09-29通过游戏学习：表格知识的不对称迁移以增强图像模型
Learning Through Game: Skewed Transfer of Tabular Knowledge to Strengthen Image Model
arXiv 2609.32272 · 2026-09-29高斯LoRA的认证前沿：独立先验、后验风险与预测保持平衡
Certification Frontiers for Gaussian LoRA: Independent Priors, Posterior Risk, and Prediction-Preserving Balancing
arXiv 2609.32271 · 2026-09-29谱梯度下降的过拟合：矩阵几何如何塑造泛化与隐式偏差
Overfitting of Spectral Gradient Descent: How Matrix Geometry shapes Generalization and Implicit Bias
arXiv 2609.32270 · 2026-09-29在回答之前：大小匹配记忆构建下的证据充分性
Before Answering: Evidence Sufficiency under Size-Matched Memory Construction
arXiv 2609.32269 · 2026-09-29自重建动力学用于自编码器重建精化
Self-Reconstruction Dynamics for Autoencoder Reconstruction Refinement
arXiv 2609.32268 · 2026-09-29旧评估何时能认证新模型？评估器漂移下的标签高效发布决策
When Can Old Evaluations Certify a New Model? Label-Efficient Release Decisions under Evaluator Drift
arXiv 2609.32267 · 2026-09-29对称置换纠缠器用于非可逆对称保护拓扑相
Symmetry-permuting entanglers for non-invertible symmetry-protected topological phases
arXiv 2609.32266 · 2026-09-29LHD-ECRH活动回顾及展望
Review of LHD-ECRH activities and beyond
arXiv 2609.32265 · 2026-09-29LANTERN：照亮语言模型中的隐藏数学知识
LANTERN: Illuminating Hidden Mathematical Knowledge in Language Models
arXiv 2609.32264 · 2026-09-29多模态测试时自适应的表示编辑
Representation Editing for Multimodal Test-Time Adaptation
arXiv 2609.32263 · 2026-09-29带分布预测的树搜索
Tree Search With Distributional Predictions
arXiv 2609.32260 · 2026-09-29多用户夹持天线系统中端口选择的离散扩散方法
Discrete Diffusion for Port Selection in Multiuser Pinching-Antenna Systems
arXiv 2609.32258 · 2026-09-29四元数代数及其他赋范可除环上的非交换花环图案
Non-commutative frieze patterns over quaternion algebras and other normed division rings
arXiv 2609.32257 · 2026-09-29LAM：具有检索分数误差界的高效有损智能体记忆框架
LAM: Efficient Lossy Agent Memory Framework With A Retrieval-Score Error Bound
arXiv 2609.32256 · 2026-09-29澄清用户还是验证世界？面向主动式智能体的不确定性路由
Clarify the User or Verify the World? Uncertainty Routing for Proactive Agents
arXiv 2609.32255 · 2026-09-29为什么直接学习周期轨迹可能失败
Why Directly Learning Periodic Trajectories Can Fail
arXiv 2609.32254 · 2026-09-29DS-VLA：一种树突启发的视觉-语言-动作模型，用于鲁棒动作控制
DS-VLA: A Dendritic-inspired Vision-Language-Action Model for Robust Action Control
arXiv 2609.32253 · 2026-09-29通过离散扩散先验利用辅助信息进行主动数据采集
Active Data Acquisition with Side Information via Discrete Diffusion Priors
arXiv 2609.32252 · 2026-09-29同与不同：光化学理论的机器验证谱系
The Same and Not the Same: A Machine-Checked Genealogy of Photochemical Theory
arXiv 2609.32251 · 2026-09-29RoboSTAR：面向人形机器人的下一代自回归手语翻译
RoboSTAR: Next-Scale Autoregressive Sign Language Translation for Humanoid Robots
arXiv 2609.32250 · 2026-09-29可压缩Navier-Stokes方程中的等温壁：仿射熵、弹道能量与熵稳定离散格式
Isothermal walls in the compressible Navier-Stokes equations: affine entropy, ballistic energy, and entropy-stable discrete formulations
arXiv 2609.32249 · 2026-09-29任意时刻有效的大语言模型排行榜：基于基准加权与块因子化 e-过程
Anytime-Valid LLM Leaderboards via Benchmark-weighted and Block-Factorized e-Processes
arXiv 2609.32248 · 2026-09-29认证观测等价因果模型间的干预一致性
Certifying Interventional Agreement Among Observationally Equivalent Causal Models
arXiv 2609.32247 · 2026-09-29无Gabor对偶的Gabor框架
Gabor frames with no Gabor duals
arXiv 2609.32246 · 2026-09-29AutoPDEBench：面向神经PDE求解器设计的LLM自动研究基准
AutoPDEBench: Benchmarking LLM Auto-Research for Neural PDE Solver Design
arXiv 2609.32245 · 2026-09-29两阶段多视角步态识别与重嵌入网络
Two-Stage Multi-View Gait Recognition with a Re-Embedding Network
arXiv 2609.32244 · 2026-09-29引力瞬子与双曲空间
Gravitational instantons and hyperbolic space
arXiv 2609.32243 · 2026-09-29Curv-Tail：基于离散包长度编码和Lorentz原型的高效长尾加密流量分类
Curv-Tail: Lightweight Long-Tailed Encrypted Traffic Classification with Discrete Packet-Length Encoding and Lorentz Prototypes
arXiv 2609.32242 · 2026-09-29神经图像水印中的残差可迁移性
Residual Transferability in Neural Image Watermarking
arXiv 2609.32241 · 2026-09-29
