2026-10论文中文摘要
门控记忆：面向对话AI的准入控制式记忆形成机制
Gated Memory: Admission-Controlled Memory Formation for Conversational AI
arXiv 2610.11270 · 2026-10-09芬斯勒-兰德斯-佐佐木（FRS）宇宙学的晚期背景膨胀与宇宙学特征
Late-Time Background Expansion and Cosmographic Signatures of Finsler-Randers-Sasaki Cosmology
arXiv 2610.11269 · 2026-10-09TaReD：面向长程任务的工具感知递归分解
TaReD: Tool-Aware Recursive Decomposition for Long-Horizon Tasks
arXiv 2610.11268 · 2026-10-09基于连续衍射调制的片上表面声波整形
On-Chip Shaping of Surface Acoustic Waves via Continuous Diffractive Modulation
arXiv 2610.11267 · 2026-10-09带图拉普拉斯基准先验的正则化小区域估计
Regularized Small Area Estimation with Graph Laplacian Benchmarking priors
arXiv 2610.11266 · 2026-10-09路径图的2-令牌图的控制数
The domination number of the $2$-token graph of path graphs
arXiv 2610.11265 · 2026-10-09LR-V2X：低带宽通信下的抗损协同感知
LR-V2X: Loss-Resilient Collaborative Perception under Low-Bandwidth Communication
arXiv 2610.11264 · 2026-10-09有限密度下QGP的热光子产生率：解析截断独立性
Thermal photon production rate from QGP at finite density: Analytic cutoff independence
arXiv 2610.11263 · 2026-10-09稀疏约束保持型量子退火中的间隙敏感性与瓶颈迁移
Gap Sensitivity and Bottleneck Migration in Sparse Constraint-Preserving Quantum Annealing
arXiv 2610.11262 · 2026-10-09Q-Capsule：一种用于缓解贫瘠高原的基于局域化胶囊的量子神经架构
Q-Capsule: A Localized Capsule-Based Quantum Neural Architecture for Barren Plateau Mitigation
arXiv 2610.11261 · 2026-10-09作为弹性工程师的大语言模型智能体在科学应用中的应用
LLM Agents as Resilience Engineers for Scientific Applications
arXiv 2610.11260 · 2026-10-09超导线圈实现大超电流整流的高效率
High Efficiency in Large Supercurrent Rectification with Superconducting Coil
arXiv 2610.11259 · 2026-10-09树级下TMD因子化破缺的量化
Quantifying TMD factorization breaking at tree level
arXiv 2610.11258 · 2026-10-09表征学习中的残余谱不稳定性
Residual spectral instabilities in representation learning
arXiv 2610.11257 · 2026-10-09谁主导谁收集：异构语言模型市场中的算法合谋
Who Leads and Who Collects:Algorithmic Collusion in Markets of Heterogeneous Language Models
arXiv 2610.11256 · 2026-10-09魔法笔：用于文档标注的自动笔模式切换
Magic Pen: Automatic Pen Mode Switching for Document Annotation
arXiv 2610.11255 · 2026-10-09NIST第三轮格候选族的可证明次指数算法
Provable Subexponential Algorithms for NIST Third-Round Lattice Families
arXiv 2610.11254 · 2026-10-09LLM-IDEA：用于自主发现机制世界模型的可识别性驱动实验智能体
LLM-IDEA: Identifiability-Driven Experimental Agent for Autonomous Discovery of Mechanistic World Models
arXiv 2610.11253 · 2026-10-09用于类人跟车行为模拟的神经记忆模糊推理系统
Neuro-Memory Fuzzy Inference System for Mimicking Human-like Car Following Behavior
arXiv 2610.11252 · 2026-10-09V-CoLA：基于线性注意力的视觉令牌压缩
V-CoLA: Vision Token Compression with Linear Attention
arXiv 2610.11251 · 2026-10-09基于对数高斯 Cox 过程的连续脑连接贝叶斯估计
Bayesian Estimation of Continuous Brain Connectivity with Log-Gaussian Cox Processes
arXiv 2610.11250 · 2026-10-09保留零假设的变换用于高维两样本协方差检验
High-Dimensional Two-Sample Covariance Testing with Null-Preserving Transformations
arXiv 2610.11249 · 2026-10-09SimVLA：面向移动操作的零样本仿真到真实VLA学习
SimVLA: Zero-Shot Sim-to-Real VLA Learning for Mobile Manipulation
arXiv 2610.11248 · 2026-10-09为什么在线策略蒸馏有时会失败：学习信号消失
Why On-Policy Distillation Sometimes Fails: Vanishing Learning Signals
arXiv 2610.11247 · 2026-10-09评估基础机器学习势能在评估晶体学开放数据库真实性方面的可靠性
Assessing the Reliability of Foundational Machine Learning Potentials for Evaluating the Veracity of the Crystallography Open Database
arXiv 2610.11246 · 2026-10-09读取关键信息：面向KV缓存的查询自适应量化
Read What Matters: Query-Adaptive Quantization for KV Caches
arXiv 2610.11245 · 2026-10-09通过调控斯格明子的尺寸与形状增强拓扑霍尔效应
Enhancement of the Topological Hall Effect through Engineering the Skyrmion Size and Shape
arXiv 2610.11244 · 2026-10-09双接触塞与动力双曲性
Bi-contact plugs and dynamical hyperbolicity
arXiv 2610.11243 · 2026-10-09基于参考滤波器驱动转移概率的交互式多模型（IMM）非合作卫星机动检测
Reference-Filter-Driven Transition Probabilities for IMM-Based Satellite Maneuver Detection
arXiv 2610.11242 · 2026-10-09TAP3D：热辅助三维人体点云
TAP3D: Thermal-Assisted 3D Human Point Clouds
arXiv 2610.11241 · 2026-10-09函数域上除数函数的统计特性
Statistics of divisor functions over function fields
arXiv 2610.11240 · 2026-10-09β-ensemble 边缘特征值的最优上尾估计
Optimal upper tail estimates for the edge eigenvalues of $β$-ensembles
arXiv 2610.11239 · 2026-10-09归因于「如何」，而非仅「哪个」：扩散模型的反事实响应轨迹
Attributing HOW, Not Just WHICH: Counterfactual Response Trajectories for Diffusion Models
arXiv 2610.11238 · 2026-10-09打破群体规模壁垒：基于舞者链的参数高效群体舞蹈生成
Breaking the Group Size Barrier: Parameter-Efficient Group Dance Generation with Chain-of-Dancers
arXiv 2610.11237 · 2026-10-09用于室内空气质量监测的低成本传感器校准：数据集、评估场景与轻量级模型
Low-Cost Sensor Calibration for Indoor Air Quality Monitoring: A Dataset, Evaluation Scenarios, and a Lightweight Model
arXiv 2610.11236 · 2026-10-09在具有空间马尔可夫普适物理的系统中，马尔可夫长度可以发散
Markov length can diverge in systems whose universal physics is spatially Markovian
arXiv 2610.11235 · 2026-10-09通过非线性斯塔克耦合获得精确的极化激元凝聚态
Exact polariton-condensate states via nonlinear Stark coupling
arXiv 2610.11234 · 2026-10-09MiniVer-V：识别短视频核查的最小充分证据
MiniVer-V: Identifying Minimal Sufficient Evidence for Short Video Verification
arXiv 2610.11233 · 2026-10-09反射高能电子衍射（RHEED）图像中候选菊池区域的定位：可见性与标注边界
Localization of Candidate Kikuchi Regions in RHEED Images: Visibility and Annotation Boundaries
arXiv 2610.11232 · 2026-10-09利用编译：小型视觉语言模型应保留哪些决策？
Harness Compilation: Which Decisions Should a Small Vision-Language Model Keep?
arXiv 2610.11231 · 2026-10-09CORAL：基于增量图构建的大规模跨模态向量检索
CORAL: Cross-modal Vector Retrieval via Incremental Graph Construction at Scale
arXiv 2610.11230 · 2026-10-09SteerCast：面向仅解码器时间序列预测的基于检索的潜在引导方法
SteerCast: Retrieval-Based Latent Steering for Decoder-Only Time Series Forecasting
arXiv 2610.11229 · 2026-10-09基于协同过滤路径的多模态图检索增强序列推荐
Multimodal Graph Retrieval-Augmented Sequential Recommendation via Collaborative Filtering Paths
arXiv 2610.11228 · 2026-10-09超越维数数据的傅里叶限制：多尺度矩准则
Fourier Restriction Beyond Dimension Data: A Multiscale Moment Criterion
arXiv 2610.11227 · 2026-10-09当更低的重构损失反而有害：针对低比特大语言模型（LLM）量化的分布鲁棒优化
When Lower Reconstruction Loss Hurts: Distributionally Robust Refinement for Low-Bit LLM Quantization
arXiv 2610.11226 · 2026-10-09用于量子位元的高速级联量子纠错码
High-Rate Concatenated Quantum Error-Correcting Codes for Qudits
arXiv 2610.11225 · 2026-10-09协同边缘计算中的在线自适应计算复用：一种双时间尺度方法
Online Adaptive Computation Reuse in Collaborative Edge Computing: A Two-Timescale Approach
arXiv 2610.11224 · 2026-10-09SafeInferCom：用于机器人任务规划的、基于验证器引导的生成中干预的安全推理时计算
SafeInferCom: Safe Inference-Time Compute via Verifier-Guided Mid-Generation Intervention for Robotic Task Planning
arXiv 2610.11223 · 2026-10-09跨物种表征学习对齐小鼠与人类神经动力学并追踪临床药物疗效
Cross-species representation learning aligns mouse and human neural dynamics and tracks clinical drug efficacy
arXiv 2610.11222 · 2026-10-09抓取力度应多大？
How Firm Should a Grasp Be?
arXiv 2610.11221 · 2026-10-09
