2026-09论文中文摘要
BaLEEN：利用潜在编码实体进行上下文感知自动语音识别的偏置方法
BaLEEN: Biasing with Latent Encoded Entities for Context-Aware ASR
arXiv 2609.36913 · 2026-09-30量子态的普适归纳推理
Universal Inductive Inference of Quantum States
arXiv 2609.36912 · 2026-09-30变分混合与多边缘流匹配：推进统计推断及其生物学应用
Variational Mixtures and Multi-Marginal Flow Matching: Advancing Statistical Inference with Biological Applications
arXiv 2609.36911 · 2026-09-30圆柱分拆中的部分大小计数与小轮廓的双射
Part Size Count in Cylindric Partitions and Bijections for Small Profiles
arXiv 2609.36910 · 2026-09-30有理连通性与K3曲面的O'Grady广义Franchetta猜想
Rationally connectedness and O'Grady's generalized Franchetta conjecture for K3 surfaces
arXiv 2609.36909 · 2026-09-30精确搜索多集合交集的量子算法
Exact quantum algorithm for searching of multiple sets intersection
arXiv 2609.36908 · 2026-09-30三方场景中量子导引性的设备无关量化
Device-independent quantification of steerability in tripartite scenario
arXiv 2609.36907 · 2026-09-30基于事件视觉传感器的三元可见光通信
Ternary Visible Light Communication Using Event-Based Vision Sensors
arXiv 2609.36905 · 2026-09-30不可约 $\mathbb Z_+$-模在 $\mathbb Z[\sqrt[m]{d_1},\sqrt[m]{d_2},\ldots, \sqrt[m]{d_r}]$ 上
Irreducible $\mathbb Z_+$-modules over $\mathbb Z[\sqrt[m]{d_1},\sqrt[m]{d_2},\ldots, \sqrt[m]{d_r}]$
arXiv 2609.36904 · 2026-09-30MultiTalk：将全双工语音模型扩展到长时、多方、双语对话
MultiTalk: Scaling Full-Duplex Speech Models to Long, Multi-Party, Bilingual Conversation
arXiv 2609.36903 · 2026-09-30RAEGNet: 关系感知证据图网络用于危害感知的多模态假新闻检测
RAEGNet: Relation-Aware Evidence Graph Network for Harm-Aware Multimodal Fake News Detection
arXiv 2609.36902 · 2026-09-30量子动力系统的数字孪生建模：耗散量子储备计算
Digital Twin Modeling of Quantum Dynamical Systems: Dissipative Quantum Reservoir Computing
arXiv 2609.36901 · 2026-09-30STAR-GRPO：针对表示依赖的奖励黑客的规范锚定与可靠性优先优势
STAR-GRPO: Canonical Anchoring and Reliability-First Advantages against Representation-Dependent Reward Hacking
arXiv 2609.36900 · 2026-09-30重塑异步强化学习后训练在异构加速器上的回放工作负载
Reshaping Rollout Workloads for Asynchronous RL Post-Training on Heterogeneous Accelerators
arXiv 2609.36899 · 2026-09-30Volcanite：面向连接组学及其他领域的商品硬件分割体可视化
Volcanite: Commodity-Hardware Segmentation Volume Visualization for Connectomics and Beyond
arXiv 2609.36898 · 2026-09-30任意复单李代数的Toda系统的约化解
Reduced Solutions of Toda Systems for Arbitrary Complex Simple Lie Algebras
arXiv 2609.36897 · 2026-09-30HorizonFlow：离线目标条件强化学习的变长规划
HorizonFlow: Variable-Length Planning for Offline Goal-Conditioned RL
arXiv 2609.36896 · 2026-09-30DiffReID：用于目标重识别的判别式扩散模型
DiffReID: Discriminative Diffusion Model for Object Re-Identification
arXiv 2609.36894 · 2026-09-30动量耦合的评分标准自适应用于细粒度图像描述生成
Momentum-Coupled Rubric Adaptation for Detailed Image Captioning
arXiv 2609.36893 · 2026-09-30驾驭演化即学习：自我改进型个人智能体的逼近、泛化与优化极限
Harness Evolution as Learning: Approximation, Generalization, and Optimization Limits of Self-Improving Personal Agents
arXiv 2609.36892 · 2026-09-30ProGuT：面向森林场景的标签高效全景分割
ProGuT: Label-Efficient Panoptic Segmentation for Forest Scenes
arXiv 2609.36891 · 2026-09-30SINO：尺度不变神经算子
SINO: Scale-Invariant Neural Operator
arXiv 2609.36890 · 2026-09-30条条大路通罗马：面向开放环境主动三维建图的流驱动多锚点探索
All Roads Lead to Rome: Flow-driven Multi-Anchor Exploration for Open-Environment Active 3D Mapping
arXiv 2609.36889 · 2026-09-30超越次高斯检测器分数：面向人机文本分割的鲁棒加权轮廓损失变点检测
Beyond Sub-Gaussian Detector Scores: Robust Weighted Profile-Loss Change Point Detection for Human-LLM Text Segmentation
arXiv 2609.36888 · 2026-09-30WEFT：面向通用智能体的工具使用后训练规模化
WEFT: Scaling Tool-Use Post-Training for General-Purpose Agents
arXiv 2609.36887 · 2026-09-30站点稀释kagome伊辛模型中的静态与动态普适性
Static versus dynamic universality in the site-diluted kagome Ising model
arXiv 2609.36886 · 2026-09-30RNA设计：基于条件流匹配与有限策略强化学习
RNA Design via Conditioned Flow Matching and Finite-Policy Reinforcement Learning
arXiv 2609.36885 · 2026-09-30曲线的分支覆盖模空间的稳定有理上同调分解
Stable rational cohomology splitting of moduli of branched covers of curves
arXiv 2609.36884 · 2026-09-30表格基础模型中的稀疏先验架构对齐
Architecture Alignment With Sparse Priors in Tabular Foundation Models
arXiv 2609.36883 · 2026-09-30你观察什么决定你如何识别因果效应：跨观测视图评估因果模型
What You Observe Determines How You Identify Causal Effects: Evaluating Causal Models across Observational Views
arXiv 2609.36881 · 2026-09-30Weyl几何中的高曲率引力：无鬼与无快子条件
Higher-curvature gravity in Weyl geometry: No-ghost and no-tachyon conditions
arXiv 2609.36880 · 2026-09-30SKILLLITE：基于证据引导的恶意技能审计与紧凑型大语言模型
SKILLLITE: Evidence-Guided Malicious Skill Auditing with Compact LLMs
arXiv 2609.36879 · 2026-09-30NIDAR：用于可扩展场景无关的LiDAR强度重建的近红外引导的内在分解
NIDAR: NIR-Guided Intrinsic Decomposition for Scalable Scene-Agnostic LiDAR Intensity Reconstruction
arXiv 2609.36878 · 2026-09-30多项式的合成指数及其应用
The Index of Composition of polynomials and its applications
arXiv 2609.36877 · 2026-09-30核质量预测中核回归的Wigner与镜像关联
Wigner and mirror correlations in nuclear mass predictions with kernel ridge regression
arXiv 2609.36876 · 2026-09-30用双副本测量弥合到Holevo界的单副本差距
Closing a single-copy gap to the Holevo bound with two-copy measurements
arXiv 2609.36874 · 2026-09-30看见时间：面向可解释时间序列聚类的视觉-时间表征学习
Seeing Time: Visual-Temporal Representation Learning for Interpretable Time Series Clustering
arXiv 2609.36873 · 2026-09-30PreferenceFlow：基于人类干预的流匹配机器人策略测试时引导
PreferenceFlow: Test-Time Guidance of Flow-Matching Robot Policies from Human Interventions
arXiv 2609.36872 · 2026-09-30取向工程是铁电体横向切换的通用策略
Orientation engineering is a universal strategy for ferroelectric trans-switching
arXiv 2609.36871 · 2026-09-30VidAct：利用以对象为中心的3D感知从真实世界视频中学习操作
VidAct: Learning Manipulation from In-the-Wild Videos with Object-Centric 3D Awareness
arXiv 2609.36870 · 2026-09-30自治变分问题中有界函数的范数与能量逼近
Approximation in norm and energy with bounded functions for autonomous variational problems
arXiv 2609.36869 · 2026-09-30PrPb$_3$ 中植入 $\mu^+$ 位点处的量子化超精细场：局域 $f$ 电子与间隙带电粒子之间的相互作用
Quantized Hyperfine Field at an Implanted $μ^+$ Site in PrPb$_3$: Interplay between Localized $f$ Electrons and an Interstitial Charged Particle
arXiv 2609.36868 · 2026-09-30状态轨迹理由作为强化学习中的辅助任务
State Trace Rationale As Auxiliary Task in Reinforcement Learning
arXiv 2609.36867 · 2026-09-30S2T-Unet：一种用于跨模态MRI翻译的结构到风格框架
S2T-Unet: A Structure-to-Style Framework for Inter-Modality MRI Translation
arXiv 2609.36866 · 2026-09-30超越常数截断：有界Sobolev逼近间隙
Beyond Constant Truncations: The Bounded Sobolev Approximation Gap
arXiv 2609.36865 · 2026-09-30模型改变主意之处：用于可验证奖励强化学习的高效事后分歧定位
Where the Model Changes Its Mind: Hindsight-Divergence Localization for Efficient Reinforcement Learning with Verifiable Rewards
arXiv 2609.36864 · 2026-09-30平方kSZ与HI强度场的互相关：ACT--MeerKAT的图级预测
Cross-correlating Squared kSZ and HI Intensity Fields: A Map-level ACT--MeerKAT Forecast
arXiv 2609.36863 · 2026-09-30更安全的内容还是更坚定的拒绝？一种针对有害微调的对齐混合扰动防御
Safer Content or Firmer Refusals? A Hybrid Perturbation Defense for Alignment under Harmful Fine-tuning
arXiv 2609.36862 · 2026-09-30IronLLM：锻造面向实时具身智能的紧凑边缘原生语言模型
IronLLM: Forging Compact Edge-Native Language Models for Real-Time Embodied Intelligence
arXiv 2609.36860 · 2026-09-30用于强化学习的马尔可夫非凸ADMM：超越光滑块的Bellman-预解稳定性
Markovian Nonconvex ADMM for Reinforcement Learning: Bellman-Resolvent Stability Beyond Smooth Blocks
arXiv 2609.36859 · 2026-09-30
