2026-10论文中文摘要
犹豫具有几何结构：基于熵训练的双曲探针用于稀疏激活引导
Hesitation Has a Geometry: Entropy-Trained Hyperbolic Probes for Sparse Activation Steering
arXiv 2610.02391 · 2026-10-05关于正则锦标赛中缺失的近生成平衡反定向树的注记
Note on near-spanning balanced antidirected trees missing from regular tournaments
arXiv 2610.02390 · 2026-10-05第一代恒星与星系中坍缩加速的小尺度发电机
Collapse-accelerated small-scale dynamos in the first stars and galaxies
arXiv 2610.02389 · 2026-10-05八叉树作为一种显式3D语言
Octrees as an Explicit 3D Language
arXiv 2610.02388 · 2026-10-05ChatGPT时代的社交机器人检测：挑战与机遇
Social bot detection in the age of ChatGPT: Challenges and opportunities
arXiv 2610.02386 · 2026-10-05基于SIMD的哈希表实现的细粒度分析
Fine-Grained Analysis of SIMD-Based Hash Table Implementations
arXiv 2610.02385 · 2026-10-05网络作战中的连通性、认知负荷与人机监督
Connectedness, Cognitive Load, and Human-AI Oversight in Cyber Operations
arXiv 2610.02384 · 2026-10-05循环Transformer中共享内存的惊人有效性
The Surprising Effectiveness of Shared Memory in Looped Transformers
arXiv 2610.02383 · 2026-10-05FactorSplat：用于医学体绘制的外观可控高斯代理
FactorSplat: Appearance-Controllable Gaussian Proxies for Medical Volume Rendering
arXiv 2610.02382 · 2026-10-05Latent-MOPD：潜在多教师在线策略蒸馏
Latent-MOPD: Latent Multi-Teacher On-Policy Distillation
arXiv 2610.02381 · 2026-10-05CuCrP2S6 中室温下厚度控制的铁电到离子输运转变
Thickness-Controlled Ferroelectric-to-Ionic Transport in CuCrP2S6 at Room Temperature
arXiv 2610.02380 · 2026-10-05三个局部 $\pi$-特征
Three local $π$-characters
arXiv 2610.02379 · 2026-10-05THPL：面向RAS中虹鳟投喂管理的视觉到语言决策支持框架
THPL: A Vision-to-Language Decision Support Framework for Rainbow Trout Feeding Management in RAS
arXiv 2610.02378 · 2026-10-05流匹配用于贝叶斯逆问题中的快速后验采样
Flow Matching for Fast Posterior Sampling in Bayesian Inverse Problems
arXiv 2610.02377 · 2026-10-05Coco：硬件-软件协同设计生命周期的智能体副驾驶
Coco: An Agentic Copilot for the Hardware--Software Co-Design Lifecycle
arXiv 2610.02376 · 2026-10-05EviDent-CBCT：在非穷尽报告监督下从牙科CBCT生成基于证据瓶颈的报告
EviDent-CBCT: Evidence-Bottlenecked Report Generation from Dental CBCT under Non-Exhaustive Report Supervision
arXiv 2610.02375 · 2026-10-05“我尽量不让自己被黑”：智力和发育障碍成年人是如何应对安全与隐私通知的
"I'm trying not to get hacked:" How Adults with Intellectual and Developmental Disabilities Navigate Security and Privacy Notifications
arXiv 2610.02374 · 2026-10-05Hop-Decayed Influence: GraphRAG 流水线中结构辅助索引的新漏洞（LLM）
Hop-Decayed Influence: New Vulnerabilities of Structural Auxiliary Indexing in GraphRAG Pipelines with LLM
arXiv 2610.02373 · 2026-10-05遍历文本到图像扩散中的满意度-多样性前沿
Traversing the Satisfaction-Diversity Frontier in Text-to-Image Diffusion
arXiv 2610.02372 · 2026-10-05平面图中多源最短路径算法的一个简单分析
Simple analysis of an algorithm for multiple-source shortest paths in planar graphs
arXiv 2610.02371 · 2026-10-05大规模网络在环：GPU批处理5G仿真用于大规模并行机器人学习
Network-in-the-Loop at Scale: GPU-Batched 5G Simulation for Massively Parallel Robot Learning
arXiv 2610.02370 · 2026-10-05自动化应用人机交互原则：按需用户界面构建的技能、人机思考空间与人机交互的未来
Automating the Application of HCI Principles: Skills for On-Demand UI Construction, the Human-AI Space to Think, and the Future of HCI
arXiv 2610.02369 · 2026-10-05重新思考用于组合式与上下文内机器人操作的世界-动作模型
Rethinking World-Action Model for Compositional and In-Context Robotic Manipulation
arXiv 2610.02368 · 2026-10-05新发现壳Be星+sdOB双星AN Col的完整轨道解与动力学质量
Full orbital solution and dynamical masses for a new shell-Be + sdOB binary AN Col
arXiv 2610.02367 · 2026-10-05协同设计健身房：一个用于体现-策略协同优化的统一基准
Co-design Gym: A Unified Benchmark for Embodiment-Policy Co-optimization
arXiv 2610.02366 · 2026-10-05流形与图上的通过格林势得到的严格正最优Hardy权
Strictly positive optimal Hardy weights on manifolds and graphs via Green potential
arXiv 2610.02365 · 2026-10-05域偏移下行人检测的置信度受控XAI审计
Confidence-Controlled XAI Auditing for Pedestrian Detection under Domain Shift
arXiv 2610.02364 · 2026-10-05ArrivalBench：智能体生成的数据管道一次正确，随时间推移出错
ArrivalBench: Agent-Generated Data Pipelines Are Correct Once and Wrong Under Time
arXiv 2610.02363 · 2026-10-05SEDIMA：进化搜索智能体的跨运行分层洞察记忆
SEDIMA: Cross-Run Hierarchical Insight Memory for Evolutionary Search Agents
arXiv 2610.02361 · 2026-10-05SocialVLA：面向VLA操作中基于人类反应失败检测与恢复的社会感知网关
SocialVLA: A Social Perception Gateway for Human-Reaction-Based Failure Detection and Recovery in VLA Manipulation
arXiv 2610.02360 · 2026-10-05字典序多目标在策略蒸馏
Lexicographic Multi-Objective On-Policy Distillation
arXiv 2610.02359 · 2026-10-05可积散焦Ablowitz-Ladik方程的爆破
Blow-up for the integrable defocusing Ablowitz-Ladik equation
arXiv 2610.02358 · 2026-10-05深度序列模型的时间序列保形预测
Conformal Prediction for Time Series with Deep Sequence Models
arXiv 2610.02357 · 2026-10-05半经典极限下质量型势的Dirac共振下界
Lower bound on Dirac resonances with mass-type potentials in the semi-classical limit
arXiv 2610.02356 · 2026-10-05为什么自适应批处理有助于大语言模型预训练？来自无界方差的视角
Why Does Adaptive Batching Help LLM Pretraining? A Perspective from Unbounded Variance
arXiv 2610.02355 · 2026-10-05与一个泛二值代数群相关的量子Yang–Baxter方程的解
A solution to the quantum Yang--Baxter equation associated with a universal two-valued algebraic group
arXiv 2610.02354 · 2026-10-05是否每个用户都需要私有LoRA？将个性化与逐用户适配解耦
Does Every User Need a Private LoRA? Decoupling Personalization from Per-User Adaptation
arXiv 2610.02353 · 2026-10-05F_2 上具有正正交重数的多项式比例
The proportion of polynomials over F_2 with positive orthogonal multiplicity
arXiv 2610.02352 · 2026-10-05DeReAct：面向可靠AI智能体的分解推理与行动
DeReAct: Decomposed Reasoning and Acting for Reliable AI Agents
arXiv 2610.02351 · 2026-10-05Van Douwen 族、可乘性与超滤极大性
Van Douwen Families, Productivity and Ultrafilter Maximality
arXiv 2610.02350 · 2026-10-05MIRROR：用于LLM多智能体通信的多路径法定人数完整性
MIRROR: Multipath Quorum Integrity for LLM Multi-Agent Communication
arXiv 2610.02349 · 2026-10-05自同态的部分交换半群中的算术增长
Arithmetic growth in partially commutative semigroups of endomorphisms
arXiv 2610.02348 · 2026-10-05从行为到溯源：将表格基础模型归因于合成预训练数据
From Behavior to Provenance: Attributing Tabular Foundation Models to Synthetic Pretraining Data
arXiv 2610.02347 · 2026-10-05活动阶梯模型中的行波、相分离与图案形成
Traveling waves, phase separation and pattern formation in the active stepping stone model
arXiv 2610.02346 · 2026-10-05通过核锚定局部性正则化缓解固定目标异常检测器中的收敛坍缩
Mitigating Convergence Collapse in Fixed-Target Anomaly Detectors via Kernel-Anchored Locality Regularization
arXiv 2610.02345 · 2026-10-05驱动与衰减：联合嵌入预测架构的训练动力学
Drive vs. Decay: On the Training Dynamics of Joint-Embedding Predictive Architectures
arXiv 2610.02344 · 2026-10-05SCOPE-4D：内窥镜4D几何基础模型
SCOPE-4D: Endoscopic 4D Geometry Foundation Models
arXiv 2610.02343 · 2026-10-05基于自然可见图的网络攻击检测中拓扑度量的多方法重要性与性能效率分析
A Multi Method Importance and Performance Efficiency Analysis of Topological Metrics for Natural Visibility Graph Based Cyber Attack Detection
arXiv 2610.02342 · 2026-10-05面向安全人形全身跟踪的滤波器感知微调
Filter-Aware Fine-Tuning for Safe Humanoid Whole-Body Tracking
arXiv 2610.02341 · 2026-10-05有限能量下的全息热传播子：模性与三元性
Holographic thermal propagator at finite energy from modularity and triality
arXiv 2610.02340 · 2026-10-05
