2026-10论文中文摘要
评估 GPT-6 Astra 是否执行未经授权的供应链攻击
Evaluating Whether GPT-6 Astra Performs Unsanctioned Supply-Chain Attacks
arXiv 2609.38415 · 2026-10-01无需训练的合成：面向表格、时序与关系型合成数据的纯推理流水线
Synthesis Without Training: An Inference-Only Pipeline for Tabular, Temporal, and Relational Synthetic Data
arXiv 2609.38414 · 2026-10-01VidHarness：面向成本高效长视频理解的智能体框架演化
VidHarness: Evolving Agent Harnesses for Cost-Efficient Long Video Understanding
arXiv 2609.38413 · 2026-10-01局部多项式密度比估计
Local polynomial density ratio estimation
arXiv 2609.38412 · 2026-10-01自主智能体失控的竞争风险系统化分析
A Competing-Hazards Systematization of Loss of Control in Autonomous Agents
arXiv 2609.38411 · 2026-10-01弱极大性性质不被定义域的子空间继承
The weak maximizing property is not inherited by subspaces of the domain
arXiv 2609.38410 · 2026-10-01ArgGYM：一个用于结构化可废止推理的程序化、引擎验证基准
ArgGYM: A Procedural, Engine-Verified Benchmark for Structured Defeasible Reasoning
arXiv 2609.38409 · 2026-10-01Romanoff数之间的巨大间隙
Large gaps between Romanoff numbers
arXiv 2609.38408 · 2026-10-01基于高斯和Beta策略的基于图像的连续控制中的分歧正则化模仿学习
Disagreement-Regularized Imitation Learning for Image-Based Continuous Control with Gaussian and Beta Policies
arXiv 2609.38407 · 2026-10-01评估LLM能否可靠地连接DOTs？
Evaluating Whether LLMs Can Reliably Connect the DOTs?
arXiv 2609.38406 · 2026-10-01Draft：用于机器人设计探索的参数化工具
Draft: A Parametric Tool for Robot Design Exploration
arXiv 2609.38405 · 2026-10-01通过组分调控的掺锡GST实现节能非易失性光子开关
Energy-Efficient Non-Volatile Photonic Switching via Composition-Engineered Sn-Doped GST
arXiv 2609.38404 · 2026-10-01量子PAC学习中样本复杂度的优势需要逆访问态制备酉算子
Advantage of Sample Complexity in Quantum PAC Learning Requires Inverse Access to State-Preparation Unitaries
arXiv 2609.38403 · 2026-10-01从代码库到元凶（C2C）：利用语义检索与分层强化学习缩减缺陷搜索空间
From Codebase to Culprit (C2C): Reducing the Search Space for Bugs with Semantic Retrieval and Hierarchical Reinforcement Learning
arXiv 2609.38402 · 2026-10-01记忆、适应、忽略：训练数据变化下机器人学习机制的诊断
Memorize, Adapt, Ignore: Diagnosing Robot Learning Mechanisms under Training Data Variation
arXiv 2609.38401 · 2026-10-01GestAdapt：面向人形机器人的工作空间条件化共语手势生成
GestAdapt: Workspace-Conditioned Co-Speech Gesture Generation for Humanoid Robots
arXiv 2609.38400 · 2026-10-01Fisher 和 DALI 后验近似用于引力波引力检验
Fisher and DALI posterior approximations for testing gravity with gravitational waves
arXiv 2609.38399 · 2026-10-01Grassmannian 树状编织与簇变量
Grassmannian tree webs and cluster variables
arXiv 2609.38398 · 2026-10-01SimTrace：面向在线用户建模的基于真实环境的多模态用户轨迹生成
SimTrace: Grounded Multimodal User Trajectories Generation for Online User Modeling
arXiv 2609.38397 · 2026-10-01弱紧生成空间中的Lotz性质
The Lotz Property in Weakly Compactly Generated Spaces
arXiv 2609.38396 · 2026-10-01粗糙路径与Itô框架下随机最优控制的统一最优性条件
Unified Optimality Conditions for Stochastic Optimal Control in the Rough Path and Itô Frameworks
arXiv 2609.38395 · 2026-10-01Zong问题：包含单位球的$2d$个半空间
Zong's problem on $2d$ half-spaces containing the unit ball
arXiv 2609.38394 · 2026-10-01哪些任务能在自监督学习中存活？
Which Tasks Survive Self-Supervised Learning?
arXiv 2609.38393 · 2026-10-01MetaPersona：基于经验社会科学的任务导向合成人群
MetaPersona: Task-Grounded Synthetic Populations from Empirical Social Science
arXiv 2609.38392 · 2026-10-01MSU GenText-Forensics 挑战赛 2026 技术报告
Team MSU GenText-Forensics Challenge 2026 Technical Report
arXiv 2609.38391 · 2026-10-01以行为为中心的恶意软件分类与细粒度恶意逻辑定位
Behavior-Centric Malware Classification with Fine-Grained Malicious Logic Localization
arXiv 2609.38390 · 2026-10-01多轮攻击中危害性的几何结构
The Geometry of Harmfulness in Multi-Turn Attacks
arXiv 2609.38389 · 2026-10-01PTED：用于科学推断和生成式机器学习的多维双样本检验
PTED: A multi-dimensional two-sample test for scientific inference and generative machine learning
arXiv 2609.38388 · 2026-10-01移除量子证明中的交互
On Removing Interaction from Quantum Proofs
arXiv 2609.38387 · 2026-10-01解码延迟反馈预填充：一种无模型控制器及其泛化极限
Decode-Latency Feedback Prefill: A Model-Free Controller and Its Generalization Limits
arXiv 2609.38386 · 2026-10-01微调扩散语言模型：上下文选择与目标加权
Fine-Tuning Diffusion Language Models with Context Selection and Target Weighting
arXiv 2609.38385 · 2026-10-01LeanPolish：Lean 证明压缩的验证监督
LeanPolish: Verified Supervision for Lean Proof Compression
arXiv 2609.38384 · 2026-10-01从随机探索中学习规划
Learning to Plan from Random Exploration
arXiv 2609.38383 · 2026-10-01Kolmogorov动力学下概率单纯形中的可达集
Reachable sets under Kolmogorov dynamics in the probability simplex
arXiv 2609.38382 · 2026-10-01LogiC-Diff：将安全属性嵌入人工智能赋能的网络物理系统
LogiC-Diff: Embedding Security Properties Into AI-Enabled Cyber-Physical Systems
arXiv 2609.38381 · 2026-10-01量子网络中的优先级感知路由：将相干时间约束集成到调度中
Priority-Aware Routing for Quantum Networks:Integrating Coherence-Time Constraints into Scheduling
arXiv 2609.38380 · 2026-10-01对齐数据可通过上下文混淆引发不对齐
Aligned Data Can Induce Misalignment via Context Confusion
arXiv 2609.38379 · 2026-10-01一种用于复杂几何域上波动方程的能量型间断伽辽金方法
An energy-based discontinuous Galerkin method for wave equations on complex geometric domains
arXiv 2609.38378 · 2026-10-01PhyProbe：重新思考生成视频中的物理一致性评估
PhyProbe: Rethinking Physical Consistency Evaluation in Generated Videos
arXiv 2609.38377 · 2026-10-01模顶点代数的共形块
Conformal blocks for modular vertex algebras
arXiv 2609.38376 · 2026-10-01线性预言机在线学习的下界
Lower Bounds for Linear-Oracle Online Learning
arXiv 2609.38375 · 2026-10-01Doc2LoRA 提供科学思想的可解码表示
Doc2LoRA Provides Decodable Representations of Scientific Ideas
arXiv 2609.38374 · 2026-10-01NEXUS：在 $z=5$ 处发现一颗明亮的年轻暂现源
NEXUS: Discovery of a Luminous Young Transient at $z=5$
arXiv 2609.38373 · 2026-10-01自进化框架：以智能体自身为优化器在多任务上的应用
Self-Evolving Harness on Multiple Tasks with the Agent as Its Own Optimizer
arXiv 2609.38372 · 2026-10-01TALK-Dem：痴呆相关沟通模式下的具身任务规划基准
TALK-Dem: Benchmarking Embodied Task Planning under Dementia-Associated Communication Patterns
arXiv 2609.38371 · 2026-10-01具有时间和空间依赖性属性的石墨烯等离激元建模的近似方法
An Approximate Method for Modeling Plasmons on Graphene with Time- and Space-dependent Properties
arXiv 2609.38370 · 2026-10-01AI代理能否重新发现Blaschke曲线不变量？
Can an AI Agent Rediscover a Blaschke-Curve Invariant?
arXiv 2609.38369 · 2026-10-01组合，而非对话：视觉语言模型丢失的是场景，不是线索
Composition, Not Conversation: VLMs Lose the Scene, Not the Thread
arXiv 2609.38368 · 2026-10-01可证明的近似最优性：统一（半）随机CSP搜索与反驳的简单框架
Certifiable Near-Optimality: A Simple Framework for Unifying Search and Refutation for (Semi)random CSPs
arXiv 2609.38367 · 2026-10-01关于有限群Reidemeister谱的一个推广
On a generalization of the Reidemeister spectrum of a finite group
arXiv 2609.38366 · 2026-10-01
