2026-10论文中文摘要
U-Space：揭示语言模型中不确定性何时及为何出现
U-Space: Uncovering When and Why Uncertainty Arises in Language Models
arXiv 2610.09087 · 2026-10-08可容C序列与L中的不可言喻性
Amenable C-sequences and ineffability in L
arXiv 2610.09086 · 2026-10-08Anaximander：在任何计算后端上交互式运行地理空间深度学习模型
Anaximander: Interactively Running Geospatial Deep Learning Models on Any Compute Backend
arXiv 2610.09085 · 2026-10-08单步移动标记令牌路由问题的复杂性
On the complexity of the single-move labeled token routing problem
arXiv 2610.09084 · 2026-10-08JAXIGA - 一个支持GPU和可微分的等几何分析软件框架
JAXIGA - A software framework for GPU-enabled and differentiable isogeometric analysis
arXiv 2610.09083 · 2026-10-08快速行动，修补问题：利用社交媒体评论追踪不断演变的AI危害
Move Fast and Mend Things: Keeping Up with Evolving AI Harms Using Social Media Commentary
arXiv 2610.09082 · 2026-10-08非线性谱计算
Nonlinear Spectral Computing
arXiv 2610.09081 · 2026-10-08低空无线网络中基于NOMA的ISAC用户-目标配对
User-Target Pairing for NOMA-Enabled ISAC in Low-Altitude Wireless Networks
arXiv 2610.09080 · 2026-10-08大规模仓库工程：基于智能体原生可复用代码原语
Large-scale Repository Engineering via Agent-Native Reusable Code Primitives
arXiv 2610.09079 · 2026-10-08三维及以上空间中双物种交叉吸引系统的尖锐爆破条件
Sharp blow-up condition for a two-species cross attraction system in 3 or more dimensions
arXiv 2610.09078 · 2026-10-08DISRQAD：扩散图像超分辨率质量评估数据集与基准
DISRQAD: Diffusion Image Super-Resolution Quality Assessment Dataset and Benchmark
arXiv 2610.09077 · 2026-10-08Laplacian随机矩阵极端特征值的普适性
Universality for the extreme eigenvalues of Laplacian random matrices
arXiv 2610.09076 · 2026-10-08面向AI生成音乐抄袭检测：作为版本识别问题
Towards AI-Generated Music Plagiarism Detection as a Version Identification Problem
arXiv 2610.09075 · 2026-10-08TAP：通过轨迹锚定恢复实现高效的长时程智能体剪枝
TAP: Efficient Long-Horizon Agent Pruning via Trajectory-Anchored Recovery
arXiv 2610.09074 · 2026-10-08多模囚禁原子系综中相互作用辅助的自旋相干性
Interaction-Assisted Spin Coherence in a Multimode Trapped Atomic Ensemble
arXiv 2610.09073 · 2026-10-08鲁棒扩张图中扩张与定向哈密顿圈的继承：一个免正则引理的证明
Inheritance of expansion and oriented Hamilton cycles in robust expanders: a regularity-free proof
arXiv 2610.09072 · 2026-10-08OverLay++：密集重叠布局到图像生成数据集
OverLay++: Dense-Overlap Layout-to-Image Generation Dataset
arXiv 2610.09071 · 2026-10-08协变量相关的多元序数偏好联合建模及其与比较模型的联系
Covariate-dependent Joint Modeling of Multivariate Ordinal Preferences and Its Connections with Comparison Models
arXiv 2610.09070 · 2026-10-08感应加热微机器人的局部热管理用于热疗应用
Localized Thermal Management of Induction-Heated Microrobots for Hyperthermia Applications
arXiv 2610.09069 · 2026-10-08GUARD：面向机器人硬盘驱动器拆解的几何不确定性感知点云去噪与分割
GUARD: Geometric Uncertainty-Aware Point Cloud Denoising and Segmentation for Robotic Hard Disk Drive Disassembly
arXiv 2610.09068 · 2026-10-08从模拟到缝合线：计算流体动力学模拟引导的复杂双心室修复挡板构建
From Simulation to Suture Lines: Computational Fluid Dynamic Simulation-Guided Baffle Construction for Complex Biventricular Repair
arXiv 2610.09067 · 2026-10-08用于流行病过程分析与最优缓解的局部再生数
Local Reproduction Numbers for Analysis and Optimal Mitigation of Epidemic Processes
arXiv 2610.09066 · 2026-10-08分离梳状不等式是NP难的
Separating comb inequalities is NP-hard
arXiv 2610.09065 · 2026-10-08与语言模型交谈
Talking with Language Models
arXiv 2610.09064 · 2026-10-08通过LLM蒸馏进行多标签主题分配：生成式与判别式学生模型的比较分析
Multi-Label Topic Assignment via LLM Distillation: A Comparative Analysis of Generative vs. Discriminative Student Models
arXiv 2610.09063 · 2026-10-08量子处理器上的几何热泵
Geometric heat pumping on a quantum processor
arXiv 2610.09062 · 2026-10-08支持虚拟协作中的人工智能盟友关系：一项基于场景的研究
Supporting Allyship in Virtual Collaboration with Artificial Intelligence: A Scenario-Based Study
arXiv 2610.09061 · 2026-10-08AI数据中心与电网：电力挑战与协同解决方案综述
AI Data Centers Meet Electrical Grids: A Review of Power Challenges and Coordinated Solutions
arXiv 2610.09060 · 2026-10-08EDiS：用于图神经网络的边不相交子图稀疏化框架
EDiS: Edge Disjoint Subgraph Sparsification Framework for Graph Neural Networks
arXiv 2610.09059 · 2026-10-08学习跨模型激活对齐与显式多对多层映射
Learning Cross-Model Activation Alignments with Explicit Many-to-Many Layer Maps
arXiv 2610.09058 · 2026-10-08面向医疗器械制造质量控制的认知不确定性感知缺陷检测
Epistemic Uncertainty-Aware Defect Detection for Quality Control in Medical Device Manufacturing
arXiv 2610.09057 · 2026-10-08基于几何的有限特征联想记忆容量理论
A Geometry-Based Capacity Theory for Finite-Feature Associative Memory
arXiv 2610.09056 · 2026-10-08MimicX：策略在环监督细化用于视频驱动的人形运动跟踪
MimicX: Policy-in-the-Loop Supervision Refinement for Video-Driven Humanoid Motion Tracking
arXiv 2610.09055 · 2026-10-08超越转录组：染色质信息指导的细胞状态依赖性扰动响应预测
Beyond the Transcriptome: Chromatin-Informed Prediction of Cell-State-Dependent Perturbation Responses
arXiv 2610.09054 · 2026-10-08身份之后的正义：大语言模型与“无处不在的视角”
Justice After Identity: Large Language Models and the View from Everywhere
arXiv 2610.09053 · 2026-10-08BeatFlow-ECG：用于从间接可穿戴信号重建心电图的整流流模型
BeatFlow-ECG: Rectified Flow for ECG Reconstruction from Indirect Wearable Signals
arXiv 2610.09052 · 2026-10-08自剪枝Transformer：基于通用注意力的极致KV缓存压缩
A Self-Pruning Transformer: Extreme KV-Cache Compression with Universal Attention
arXiv 2610.09051 · 2026-10-08在线次可加组合分配的高效算法
Efficient Algorithms for Online Subadditive Combinatorial Allocations
arXiv 2610.09050 · 2026-10-08具有非齐次 $SU(2)$ 和 $SL(2,\mathbb{R})$ 非线性 $\sigma$-模型的不均匀宇宙学
Anisotropic Cosmology with Inhomogeneous $SU(2) $ and $SL(2,\mathbb{R})$ Nonlinear $σ$-Models
arXiv 2610.09049 · 2026-10-08迈向金融世界建模
Towards Financial World Modeling
arXiv 2610.09048 · 2026-10-08受限原子在有色噪声作用下的加热与逃逸
Heating and Escape of Confined Atoms Subject to Colored Noise
arXiv 2610.09047 · 2026-10-08Schrödinger算子扰动在Anosov能级附近的长时间Gutzwiller迹公式
A long time Gutzwiller trace formula for perturbations of Schrödinger operators near Anosov energy levels
arXiv 2610.09046 · 2026-10-08利用条件扩散模型学习跳跃扩散过程的转移核
Learning Transition Kernels of Jump-Diffusion Processes with Conditional Diffusion Models
arXiv 2610.09045 · 2026-10-08随机特征网络中基于随机矩阵理论的二次弱到强泛化
Quadratic Weak-to-Strong Generalization in Random Feature Networks via Random Matrix Theory
arXiv 2610.09044 · 2026-10-08谨慎裁判：安全高效的人机协同决策
Careful Judge: Safe and Efficient Human-AI Collaborative Decision Making
arXiv 2610.09043 · 2026-10-08通过同态计数区分具有简单谱的图
Distinguishing graphs with simple spectrum by homomorphism counts
arXiv 2610.09042 · 2026-10-08无需搜索的三阶段可组合导航图构建
Building Navigable Graphs Without Search in Three Composable Stages
arXiv 2610.09041 · 2026-10-08ORACLE：面向约束学习的优化器相对对齐
ORACLE: Optimizer-Relative Alignment for Constrained LEarning
arXiv 2610.09040 · 2026-10-08从高召回到高效用：面向LLM生成客户意图的数据集自适应后处理
From High Recall to High Utility: Dataset-Adaptive Post-Processing of LLM-Generated Customer Intents
arXiv 2610.09039 · 2026-10-08平均场颗粒介质方程的长时大偏差渐近
Long-Time Large Deviation Asymptotics for Mean-Field Granular Media Equations
arXiv 2610.09038 · 2026-10-08
