2026-09论文中文摘要
蛋白质衍生知识能否改进病理学基础模型？
Can Protein-Derived Knowledge Improve Pathology Foundation Models?
arXiv 2609.33178 · 2026-09-29DeltaWAM：通过Delta令牌实现以变化为中心的视觉预见，用于高效的世界动作模型
DeltaWAM: Change-Centric Visual Foresight via Delta Tokens for an Efficient World-Action Model
arXiv 2609.33177 · 2026-09-29ABO-Med：用于少样本医学图像分类的加速双层优化
ABO-Med: Accelerated Bilevel Optimization for Few-Shot Medical Image Classification
arXiv 2609.33176 · 2026-09-29磁化非线性等离子体尾场用于正电子加速：机制与运行极限
Magnetizing nonlinear plasma wakefields for positron acceleration: mechanism and operating limits
arXiv 2609.33175 · 2026-09-29多地形掌握：双足运动综合控制器
Multi-Terrain Mastery: A Comprehensive Controller for Bipedal Locomotion
arXiv 2609.33174 · 2026-09-29Frobenius 抛物型李代数的主谱非单峰性
Nonunimodal principal spectra of Frobenius parabolic Lie algebras
arXiv 2609.33173 · 2026-09-29通过反事实规划的世界-动作模型动态操作
Dynamic Manipulation with World-Action Models via Counterfactual Planning
arXiv 2609.33172 · 2026-09-29扰动与求解：潜空间扩散逆问题的高效学习算子条件化
Perturb-and-Solve: Efficient Learned-Operator Conditioning for Latent Diffusion Inverse Problems
arXiv 2609.33171 · 2026-09-29爱因斯坦-麦克斯韦-带电标量场方程在3+1维中的带电裸奇点构造
A construction of charged naked singularities for the Einstein--Maxwell-charged scalar field equations in 3+1 dimensions
arXiv 2609.33168 · 2026-09-29面向下游的上下文选择用于在线上下文强化学习
Downstream-Aware Context Selection for Online In-Context Reinforcement Learning
arXiv 2609.33166 · 2026-09-29超越任务：利用现代机器人学习技术复现类动物行为基质的愿景
Beyond Tasks: A Vision for Reproducing an Animal-like Behavioral Substrate Using Modern Robot Learning Techniques
arXiv 2609.33165 · 2026-09-29关于生成建模在反馈控制与规划中的注释
Notes on Generative Modeling for Feedback Control and Planning
arXiv 2609.33164 · 2026-09-29JFS-CryoMem：一种采用电压控制超导器件和飞焦级写/读能量的低温存储器
JFS-CryoMem: A Cryogenic Memory with Voltage-Controlled Superconducting Devices and Femtojoule-Scale Write/Read Energies
arXiv 2609.33163 · 2026-09-29加权超曲面在有限域上的堆栈高度、熵与Zeta函数
Stacky Heights, Entropy, and Zeta Functions of Weighted Hypersurfaces over Finite Fields
arXiv 2609.33162 · 2026-09-29裂变矩阵方法的不确定性量化：一个严格的数学框架与计算高效的替代方案
Uncertainty Quantification for the Fission Matrix Method: A Rigorous Mathematical Framework and Computationally Efficient Alternatives
arXiv 2609.33161 · 2026-09-29PDFa11yMut：衡量PDF无障碍检查器中突变特异性检测
PDFa11yMut: Measuring Mutation-Specific Detection in PDF Accessibility Checkers
arXiv 2609.33160 · 2026-09-29调和映射的尖锐常数问题
Sharp constant problems for harmonic mappings
arXiv 2609.33159 · 2026-09-29FOCUS：从基础视觉编码器到多模态大语言模型的视网膜模型泛化基准测试
FOCUS: Benchmarking Retinal Model Generalization from Foundation Vision Encoders to Multimodal LLMs
arXiv 2609.33158 · 2026-09-29关于Max-k-CSP和有界度k-SAT的多样化解
On Diverse Solutions to Max-k-CSP and Bounded Degree k-SAT
arXiv 2609.33156 · 2026-09-29连续时间自适应控制中的尖锐临界极小极大定律与无学习阈值
Sharp Critical Minimax Laws and No-Learning Thresholds in Continuous-Time Adaptive Control
arXiv 2609.33154 · 2026-09-29实用Muon的收敛性
Convergence of Practical Muon
arXiv 2609.33152 · 2026-09-29静态力场中的应力与D项的符号
Stress in static force fields and the sign of the D-term
arXiv 2609.33151 · 2026-09-29语言模型预训练中的泛化动力学
Generalization Dynamics of LM Pre-training
arXiv 2609.33150 · 2026-09-29DroneWAM：面向无人机视觉导航的高效世界动作模型
DroneWAM: Efficient World Action Model for Drone Visual Navigation
arXiv 2609.33148 · 2026-09-29CFLoRA：基于互补因子的联邦大语言模型微调以实现无误差聚合
CFLoRA: Federated Fine-tuning of LLMs with Complementary Factors for Error-free Aggregation
arXiv 2609.33147 · 2026-09-29LiteEvo：面向未见任务泛化的自动化、低成本智能体框架演化
LiteEvo: Automated, Cost-Efficient Harness Evolution for Generalization to Unseen Tasks
arXiv 2609.33146 · 2026-09-29超越训练视界：循环Transformer中长度泛化的机制与极限
Beyond the Training Horizon: Mechanisms and Limits of Length Generalization in Looped Transformers
arXiv 2609.33144 · 2026-09-29CAME：面向可解释的季度前瞻营收预测的公司感知证据-记忆专家模型
CAME: Company-Aware Evidence-Memory Experts for Interpretable Quarter-Ahead Revenue Forecasting
arXiv 2609.33143 · 2026-09-29设备端智能体操作缓存——以分类器为中心的从自然语言到动作生成
On Device Agentic Operation Caches -- Classifier-Centric NL-to-Action Generation
arXiv 2609.33141 · 2026-09-29关于伪双曲度量下的尖锐Lipschitz连续性问题
The sharp Lipschitz continuity problem with respect to the pseudo hyperbolic metric
arXiv 2609.33140 · 2026-09-29面向前传受限网络化ISAC的以通信为中心的符号级预编码
Communication-Centric Symbol-Level Precoding for Fronthaul-Limited Networked ISAC
arXiv 2609.33139 · 2026-09-29多模态非抓取式物理参数估计：基于按压-拉动倾翻的方法
Multi-Modal Non-Prehensile Estimation of Physical Parameters via Press-and-Pull Tipping
arXiv 2609.33138 · 2026-09-29具有经典和连续321避免的排列的最大Lehmer码
Maximal Lehmer Codes for Permutations with Classical and Consecutive 321-Avoidance
arXiv 2609.33137 · 2026-09-29泄漏的学生：针对同策略蒸馏的成员推断
Leaky Students: Membership Inference against On-Policy Distillation
arXiv 2609.33136 · 2026-09-29MpFA：Blackwell GPU 上硬件高效的免训练 QK4V8 FlashAttention 内核
MpFA: Hardware-Efficient Train-Free QK4V8 FlashAttention Kernels on Blackwell GPUs
arXiv 2609.33135 · 2026-09-29任务的极限：Transformer 何时能在没有思维链的情况下成功？
Ceiling of a Task: When Can a Transformer Succeed Without Its Chain of Thought?
arXiv 2609.33134 · 2026-09-29将提示词预填充与响应重放分离，用于上下文并行的长上下文LLM后训练
Splitting Prompt Prefill from Response Replay for Context-Parallel Long-Context LLM Post-Training
arXiv 2609.33133 · 2026-09-29固定置信度的量子蒙特卡洛树搜索
Quantum Monte Carlo Tree Search with Fixed Confidence
arXiv 2609.33132 · 2026-09-29ILP-BO：基于整数线性规划的黑箱优化
ILP-BO: Integer Linear Programming-Based Black-Box Optimization
arXiv 2609.33131 · 2026-09-29正交铝磁钙钛矿中自旋劈裂与霍尔输运的对称性指导设计原则
Symmetry-guided Design Principles for Spin Splitting and Hall Transport in Orthorhombic Altermagnetic Perovskites
arXiv 2609.33130 · 2026-09-29基于流匹配的蛋白质结构分词器，高效且易用
Flow-Matching-Based Protein Structure Tokenizer Made Efficient and Easy
arXiv 2609.33129 · 2026-09-29最优和极小的 $p$ 元线性码：来自分层偏序集的广义序理想
Optimal and minimal $p$-ary linear codes from generalized order ideals of hierarchical posets
arXiv 2609.33128 · 2026-09-29策略可塑性在离线到在线强化学习中至关重要：为在线适应重新拟合离线策略
Policy Plasticity Matters in Offline-to-Online Reinforcement Learning: Refitting Offline Policies for Online Adaptation
arXiv 2609.33127 · 2026-09-29保存你的饱和数据：在基于组的强化学习中超越奖励饱和进行学习
Save Your Saturated Data: Learning Beyond Reward Saturation in Group-Based RL
arXiv 2609.33126 · 2026-09-29一起训练还是之后合并？通过共享动作接口统一VLA专家
Train Together or Merge Later? Unifying VLA Experts via a Shared Action Interface
arXiv 2609.33125 · 2026-09-29真圆弧图是 $e$-正的
Proper circular arc graphs are $e$-positive
arXiv 2609.33124 · 2026-09-29PHL：用于蛋白质-蛋白质结合亲和力预测的持久超有向图学习
PHL: Persistent Hyperdigraph Learning for Protein-Protein Binding Affinity Prediction
arXiv 2609.33122 · 2026-09-29无需预训练文本嵌入的主时态取向分类：一种新颖的形态句法库存向量方法
Classifying Dominant Temporal Orientation without Pretrained Text Embeddings: A Novel Morphosyntactic Inventory Vector Approach
arXiv 2609.33121 · 2026-09-29Mycelium：一种面向配电系统的可泛化跨网格多任务模型
Mycelium: A Generalizable Cross-Grid Multi-Task Model for Electrical Distribution Systems
arXiv 2609.33120 · 2026-09-29MedRouter：揭示医学大语言模型间的知识差异以实现基于路由的推理
MedRouter: Demystifying Knowledge Differences Across Medical LLMs for Routing-Based Reasoning
arXiv 2609.33119 · 2026-09-29
