cs.LG（2026-10-02）

📊 共 51 篇论文 | 🔗 4 篇有代码

🎯 兴趣领域导航

支柱二：RL算法与架构 (RL & Architecture) (28 🔗2) 支柱九：具身大模型 (Embodied Foundation Models) (18 🔗1) 支柱一：机器人控制 (Robot Control) (4 🔗1) 支柱八：物理动画 (Physics-based Animation) (1)

🔬 支柱二：RL算法与架构 (RL & Architecture) (28 篇)

#题目一句话要点标签🔗⭐
1 ZeroMAG: Zero-Shot Multimodal Adapter Generation for Plug-and-Play EEG Foundation Models 提出ZeroMAG框架以解决EEG多模态适应问题 representation learning foundation model multimodal
2 Structural-Functional Brain Connectivity Generation via Multimodal Hypergraph-based Flow Matching 提出多模态超图流匹配框架以解决脑连接生成问题 flow matching dyadic interaction multimodal
3 Test-time Calibration Learning for Large Language Model Reasoning 提出无标签测试时校准学习以提升大语言模型推理能力 reinforcement learning large language model ✅
4 What Should World Models Forget? Stratified Retention for Continual Adaptation 提出分层保留机制以解决世界模型的持续适应问题 world model world models
5 Divergence controls entropy in distillation 研究蒸馏过程中发散性对熵的控制机制 distillation privileged information large language model
6 Understanding Trajectory Heterogeneity in Federated World Model Learning 提出针对联邦世界模型学习的轨迹异质性理解方法 world model world models
7 Counterfactual Action Evaluation, Observation Bottlenecks, and Representation Geometry in Joint-Embedding Predictive World Models 提出反事实行动评估方法以解决世界模型的观察瓶颈问题 world model world models
8 Measure Less, Know More: Self-Supervised Test-Time Feature Acquisition 提出ECHO-$k$以解决测试时特征获取成本高的问题 reinforcement learning foundation model multimodal
9 SCAD: Structured Credit Assignment and Distillation for Long-Horizon Agents 提出SCAD以解决长时间代理的信用分配问题 distillation multimodal
10 Follow the Winners: Conservative Policy Improvement with the Cross-Entropy Method for Critic-Free RFT 提出Follow the Winners以解决状态环境中的RFT问题 policy learning PPO DPO
11 Prospective Hindsight: Self-Calibrating Reinforcement Learning via Prediction-Reality Gaps 提出前瞻性回顾方法以解决强化学习中的自我校准问题 reinforcement learning distillation privileged information
12 Distributed Learning with Selective State Space Models: Architecture-Aware Convergence Analysis 提出选择性状态空间模型的分布式学习方法以解决现有算法不足问题 Mamba SSM state space model
13 Causal Representation Learning with Instantaneous and Lagged Relations via Nonstationarity 提出iCReN框架以解决时间序列因果关系建模问题 representation learning contrastive learning
14 Differentiable Koopman Operator for Contrastive Learning on Dynamic Graphs 提出KAIROS以解决动态图对比学习中的时序建模问题 representation learning contrastive learning
15 OPD Before RL: Warm-Starting Rubric-Based RL with On-Policy Distillation 提出双阶段训练框架以优化基于评分的强化学习 reinforcement learning distillation
16 Learning from Evolving Errors: Adaptive Iterative Repair for On-Policy Distillation 提出AIR-OPD框架以解决自蒸馏中的错误修复问题 reinforcement learning distillation
17 How to Find and Reuse Policies for Continuous Adaptation in Lifelong Reinforcement Learning 提出自适应掩码选择与组合方法以解决终身强化学习中的策略重用问题 reinforcement learning
18 Pivot-SD: Efficient Self-Distillation for Masked Diffusion Language Models 提出Pivot-SD以解决掩蔽扩散语言模型的信用分配问题 distillation
19 Getting Your Guidance Weights Right in diffusion and flow-matching posterior sampling 提出一种自动调节引导权重的方法以优化后验采样 flow matching
20 Single or Multiple Policies for Phase-Structured Reinforcement Learning? 提出基于阶段分解的方法以优化强化学习策略选择 reinforcement learning
21 AdaStep: Adaptive Step Credit Weighting for Agentic Reinforcement Learning 提出AdaStep以解决长时间跨度强化学习中的信用分配问题 reinforcement learning
22 NeuroLens: Learning Latent Embeddings of Neural Semantics from Chronic Recordings 提出NeuroLens以解决慢性神经记录中的表示塑性与记录不稳定性问题 JEPA Joint-Embedding Predictive Architecture joint-embedding predictive architecture
23 Turnover-Orthogonal Credit Assignment for Open-Team Multi-Agent Reinforcement Learning 提出TOCA以解决开放团队多智能体强化学习中的信用分配问题 reinforcement learning
24 Understanding Enrichment in Reinforcement Learning 提出序列蒙特卡洛权重校正机制以优化稀疏奖励的强化学习 reinforcement learning
25 All Work And No Play Makes Jack a Dull Boy: Understanding and Preventing Catastrophic Strategy Collapse in RLVR 提出Mesh Learning以解决RLVR中的灾难性策略崩溃问题 reinforcement learning large language model ✅
26 Gated Slot Attention-2: Two-Sided Associative Memory Correction in Linear Attention 提出Gated Slot Attention-2以解决线性注意力中的记忆管理问题 linear attention
27 AIGS: Adaptive Incremental Gating System for Online Representation Learning in Non-Stationary Data Streams 提出自适应增量门控系统以解决非平稳数据流的在线表示学习问题 representation learning
28 Mastering Atari 2600 Games with Discovered Options 提出Wayfarer以解决大规模强化学习中的选项发现问题 reinforcement learning representation learning

🔬 支柱九：具身大模型 (Embodied Foundation Models) (18 篇)

#题目一句话要点标签🔗⭐
29 Toward Omni Multimodal Graph Foundation Model: A Topology-Driven Binding Approach 提出GraphBind以解决多模态图表示学习中的节点属性不完整问题 foundation model multimodal
30 Predicting and Repairing Merge Collapse in Large Language Models 提出PRISM以解决大语言模型合并崩溃问题 large language model ✅
31 Zephon: Elastic Determinism for Online, Stateful Foundation Model Data Loading Pipelines 提出Zephon以解决在线状态基础模型数据加载的弹性确定性问题 foundation model
32 Balancing Multimodal Learning via Functional Progress 提出功能空间引导的多模态优化方法以解决模态不平衡问题 multimodal
33 No-Free-Graph: Learning When Multimodal Data Should Be Graphified 提出MAG-SCOUT框架以优化多模态图学习中的图构建决策 multimodal
34 WebFovea: When the Model Is Right but the Click Is Wrong -- Reliable Round Trips for Vision-Based Web Agents on Live Websites 提出WebFovea以解决视觉基础网络代理在实时网站上的交互问题 large language model multimodal
35 Architecture-Dependent Fusion Pathways in MLLMs 提出多模态大语言模型架构依赖的融合路径以理解信息融合机制 large language model multimodal
36 LESSER: Post-Training Data Selection with Output-Layer Gradients 提出LESSER以降低后训练数据选择的计算成本 large language model
37 Objects Without Morphisms: What LLMs for Mathematics Do Not Represent 提出一种新方法以揭示大语言模型在数学翻译中的局限性 large language model
38 Metropolis-Hastings Dominates Importance Resampling for Policy Composition 提出独立Metropolis-Hastings方法以优化政策组合 large language model
39 When Is Accuracy Evidence? A Unified Theory of Generalisation, Validation, and Information Fusion 提出统一理论以改进K折交叉验证的风险评估 multimodal
40 SPEAR: A Spectral-Disentangled MoE Neural Operator with Knowledge-Guided Expert Aggregation for Large-Scale PDE Pretraining 提出SPEAR以解决PDE预训练中的知识干扰问题 foundation model
41 D2K-Bench: Can LLM Agents Turn Expert Designs into Efficient GPU Kernels? 提出D2K-Bench以评估LLM代理生成高效GPU内核的能力 large language model
42 Tailoring the Quantization Space for 1-Bit KV Cache Compression 提出TaSQ以解决1位KV缓存压缩中的性能瓶颈问题 chain-of-thought
43 Neural Data Needs Semantic Tokenization: Behavioral Events as Boundaries of Session-Transferable Tokens 提出基于状态的标记化方法以解决神经数据泛化问题 foundation model
44 DyRA: Dynamic Residual Approximation for Efficient Matrix Multiplication in DNNs 提出DyRA以解决深度神经网络中矩阵乘法效率问题 foundation model
45 Online Verification of Language Model Responses Under Cost Constraints 提出OMVV算法以解决语言模型输出验证的成本约束问题 large language model
46 What Is Lost in Post-Training? Default Collapse and the Loss of In-Context Steerability Across Diverse Perspectives 提出新目标以解决后训练导致的视角丧失问题 large language model

🔬 支柱一：机器人控制 (Robot Control) (4 篇)

#题目一句话要点标签🔗⭐
47 Forecasting from Counterfactual Simulator Rollouts: A Sim2Real Evaluation 提出基于反事实模拟器的预测方法以解决冷启动问题 sim2real zero-shot transfer
48 UniIntervene++: An Adaptive Intervention Agent for Efficient Real-World Reinforcement Learning 提出UniIntervene++以解决在线强化学习中的干预效率问题 manipulation reinforcement learning ✅
49 Bidirectional Voronoi-biased Exploration Curriculum for Reinforcement Learning 提出双向Voronoi偏置探索课程以解决强化学习中的探索瓶颈问题 quadruped reinforcement learning
50 Learning Transferable Policies from Action-free Time Series Through Dynamical Embeddings 提出层次化模型强化学习框架以解决无动作时间序列控制问题 manipulation reinforcement learning

🔬 支柱八：物理动画 (Physics-based Animation) (1 篇)

#题目一句话要点标签🔗⭐
51 TSGuard: A Real-Time Framework for Detecting and Imputing Missing Data in Streaming Time Series 提出TSGuard以解决流式时间序列数据缺失问题 spatiotemporal

⬅️ 返回 cs.LG 首页 · 🏠 返回主页