cs.LG（2026-10-15）

📊 共 54 篇论文 | 🔗 1 篇有代码

🎯 兴趣领域导航

支柱二：RL算法与架构 (RL & Architecture) (25) 支柱九：具身大模型 (Embodied Foundation Models) (19 🔗1) 支柱八：物理动画 (Physics-based Animation) (3) 支柱一：机器人控制 (Robot Control) (3) 支柱七：动作重定向 (Motion Retargeting) (1) 支柱三：空间感知与语义 (Perception & Semantics) (1) 支柱四：生成式动作 (Generative Motion) (1) 支柱六：视频提取与匹配 (Video Extraction) (1)

🔬 支柱二：RL算法与架构 (RL & Architecture) (25 篇)

#题目一句话要点标签🔗⭐
1 Physics-Aware Distillation of a Grid Foundation Model for Real-Time Edge Inference 提出物理感知蒸馏方法以解决边缘推理中的模型体积问题 distillation foundation model
2 Provably Efficient Risk-Sensitive Offline Reinforcement Learning via Optimized Certainty Equivalents 提出优化确定性等价方法以解决离线风险敏感强化学习问题 reinforcement learning offline reinforcement learning
3 Ranking and Partitioning of Attention Heads for Sparse and Continual Reinforcement Learning Fine-tuning 通过注意力头排名与分区提升稀疏与持续强化学习微调效果 reinforcement learning instruction following
4 Towards Continual Learning: Mitigating Forgetting With Subspace Tuning 提出子空间调优以解决大语言模型的遗忘问题 reinforcement learning distillation large language model
5 Waypoint-OPD: Transferring RL-Induced Policy Shifts via Gated On-Policy Distillation 提出Waypoint-OPD以解决策略蒸馏中的局部偏好一致性问题 reinforcement learning distillation
6 Does On-Policy Distillation Need to Complete Every Rollout? 提出随机截断的在线蒸馏方法以降低计算成本 teacher-student distillation
7 Compatibility Matters: A Framework for Token-Level Teacher Selection in On-Policy Distillation 提出基于兼容性的教师选择框架以优化在线蒸馏 teacher-student distillation
8 PathFM: Fast Generative Forecasting of Multivariate Trajectories 提出PathFM以解决多变量轨迹生成预测问题 flow matching foundation model
9 Learning to Steer LLM Agents and Multi-Agent Systems 提出引导代理以解决大语言模型系统的目标漂移问题 reinforcement learning large language model
10 Librarian Agents: Training LLMs to Curate Long-Context Workspaces that Anticipate Future Queries 提出图书管理员代理以解决长文档上下文推理问题 reinforcement learning large language model
11 Baselines and Score Centering in Policy Gradient Estimation 提出协方差视角以优化策略梯度估计中的基线与评分中心化 reinforcement learning large language model
12 Training-Free On-Policy Distillation 提出无训练的在线蒸馏方法以整合专家模型 distillation
13 Specification Gaming and a Constrained Reinforcement Learning Mitigation in a Cleaning Gridworld 提出约束强化学习以解决清洁机器人中的规范游戏问题 reinforcement learning
14 Router Drift Control: Mitigating Off-Policy Degradation in Mixture-of-Experts Reinforcement Learning 提出Router Drift Control以解决Mixture-of-Experts强化学习中的离策略退化问题 reinforcement learning
15 ReFlow: Replay-Driven Decoupling for Efficient RL Post-Training 提出ReFlow以解决RL后训练效率低下问题 reinforcement learning large language model
16 Finite-Time Guarantees for Constrained Reinforcement Learning under Partial Observability 提出有限时间保证的约束强化学习方法以解决部分可观测问题 reinforcement learning
17 The Tilted Sampling Problem: Optimism, Posterior Sampling, and Logarithmic Regret 提出倾斜采样框架以解决在线学习中的采样问题 reinforcement learning RLHF
18 V-SHARE: Provable Convergence Guarantees for Advantage-based Algorithms 提出V-SHARE以解决优势基算法的理论理解不足问题 reinforcement learning deep reinforcement learning
19 Predicting Privileged Information Effectiveness for On-Policy Self-Distillation 提出有效的特权信息评估方法以优化在线自蒸馏训练 distillation privileged information large language model
20 AlphaBet: Reinforcement Learning for LLM Event Forecasting Agents 提出AlphaBet框架以解决LLM事件预测中的证据获取问题 reinforcement learning large language model
21 Does On-Policy Distillation Need to Complete Every Rollout? 提出STOPD以加速在政策蒸馏中的回合生成 teacher-student distillation
22 Chance-Constrained Personalized LLM Routing with Online Preference Learning 提出个性化LLM路由的机会约束算法以解决用户偏好未知问题 preference learning
23 Progressive Misalignment Can Emerge from Benign Recursive Self-Distillation 揭示递归自蒸馏中的渐进性失调问题及其安全隐患 distillation
24 Lean4AgenticRL: Advancing Agentic RL with Verifiable Multi-level Process Rewards 提出Lean4AgenticRL以解决多层次过程奖励不足的问题 reinforcement learning large language model
25 Router Drift Control: Mitigating Off-Policy Degradation in Mixture-of-Experts Reinforcement Learning 提出Router Drift Control以解决Mixture-of-Experts强化学习中的离策略退化问题 reinforcement learning

🔬 支柱九：具身大模型 (Embodied Foundation Models) (19 篇)

#题目一句话要点标签🔗⭐
26 SurvICL: a Foundation Model for in-context Survival Analysis 提出SurvICL以解决生存分析中的模型性能问题 foundation model
27 Decoding Naturalistic Speech from fMRI with Temporal Modeling and Multimodal Alignment 提出NOEMA以解决fMRI自然语言理解中的信息恢复问题 multimodal
28 N-Gram Memories at Two Bits: Quantizing Lookup Tables in Large Language Models 提出条件程序以压缩大语言模型的查找表 large language model
29 Not All Change Is Damage: Update-Matched Unlearning for Large Language Models 提出更新匹配的遗忘方法以解决大语言模型的知识删除问题 large language model
30 From Perception to Prediction and Intervention: Benchmarking Proactive Video Assistants in Real-Life Scenarios 提出P$^3$IBench以解决视频助手主动性不足的问题 large language model multimodal
31 TimeCap: A Scalable Open-Source Pipeline for Captioning Wearable Health Time-Series Data 提出TimeCap以解决可穿戴健康时间序列数据标注问题 large language model multimodal
32 OS-Helix: Unifying GUI Interaction and Code Execution in Native Computer-Using Agents 提出OS-Helix以解决GUI交互与代码执行统一的问题 multimodal ✅
33 CAT-alyzing Honesty: Continuous Adversarial Training for Transferable LLM Honesty 提出CATCH以解决大型语言模型的诚实性问题 large language model
34 BitSparse: Compact Activation Storage for ReLU Feed-Forward Networks 提出BitSparse以解决大语言模型训练中的内存使用问题 large language model
35 ABNAT: Attention-Based Noise-Aware Training for Robust Transformers on Analog Compute-in-Memory Systems 提出ABNAT以解决模拟计算内存系统中的噪声影响问题 multimodal
36 SelectPFN: Amortized Bayesian Variable Selection for Sparse Nonlinear Regression with Prior-Data Fitted Networks 提出SelectPFN以解决稀疏非线性回归中的变量选择问题 foundation model
37 Each Reward to Its Relevant Tokens: Reward-Specific Token-Level Credit Assignment for Multi-Reward RL 提出MRTPO以解决多奖励强化学习中的信用分配问题 instruction following
38 Two Sample Sizes in Test-Time LLM Ensembling: Buffered ERM and Minimax Theory 提出缓冲ERM与极小极大理论以优化测试时LLM集成 large language model
39 Fractional Power-of-Two Quantization for Efficient and Effective Multiplier-Free LLM Inference 提出分数二次幂量化以解决低位宽LLM推理效率问题 large language model
40 Resisting Harmful Instructions through Training-Free Weight Editing 提出无训练权重编辑方法以抵抗有害指令 instruction following
41 Language Models Share a Readout Bottleneck in Working Memory 探讨大语言模型在工作记忆中的读取瓶颈 large language model
42 INCONVENIENT: A Benchmark to Assess LLM Generalization Capabilities from Convenience Samples 提出INCONVENIENT基准以评估LLM对便利样本偏差的鲁棒性 large language model
43 N-Gram Memories at Two Bits: Quantizing Lookup Tables in Large Language Models 提出条件程序以优化大语言模型的查找表存储 large language model
44 Can LLMs predict the outcomes of machine learning experiments? 评估大型语言模型在机器学习实验结果预测中的有效性 large language model

🔬 支柱八：物理动画 (Physics-based Animation) (3 篇)

#题目一句话要点标签🔗⭐
45 Learning to Forecast on Multiple Spatiotemporal Hierarchies 提出图深度学习框架以提升多层次时空预测精度 spatiotemporal
46 Learning and Forecasting Multiscale Spatiotemporal Epidemic Dynamics with Lagrangian-Eulerian Graph Neural ODEs 提出LE-EpiGNODE以解决流行病动态预测问题 spatiotemporal
47 Reducing the Cost of Qudit State Preparation by Learning a Sparse Structure for SNAP Gates 提出稀疏结构学习方法以降低量子态准备成本 PULSE

🔬 支柱一：机器人控制 (Robot Control) (3 篇)

#题目一句话要点标签🔗⭐
48 Honesty is the Best Policy ... or is it? Demand Management for Probabilistic Service Options 提出动态概率信息披露以优化服务选择管理 manipulation reinforcement learning
49 Augment the Cache, Replace the Activation: A 2x2 Study of Internal-State Sharing 提出共享内部状态机制以优化多智能体系统通信 shared control
50 Source-Aware Dynamic Pricing with Strategic Buyers 提出源感知动态定价框架以应对战略买家问题 manipulation

🔬 支柱七：动作重定向 (Motion Retargeting) (1 篇)

#题目一句话要点标签🔗⭐
51 Local Temporal Contrasts on Human Motion Data Induce a Latent Geometry 提出SemTAC以解决人类运动数据表示中的几何结构问题 human motion

🔬 支柱三：空间感知与语义 (Perception & Semantics) (1 篇)

#题目一句话要点标签🔗⭐
52 Scene Reconstruction Using Ubiquitous Wireless Signals via Diffusion Posterior Inference Over physics-Grounded Representations 提出基于无线信号的场景重建方法以解决传感器局限性问题 scene reconstruction foundation model

🔬 支柱四：生成式动作 (Generative Motion) (1 篇)

#题目一句话要点标签🔗⭐
53 Query-Efficient Rejection Sampling for Parallel Masked Diffusion 提出单增量拒绝采样器以提高并行掩蔽扩散模型的采样效率 MDM

🔬 支柱六：视频提取与匹配 (Video Extraction) (1 篇)

#题目一句话要点标签🔗⭐
54 SALTO: Solver-Anchored Lifted Transfer Operators for Nonlinear Event Inference 提出SALTO框架以解决非线性事件推断问题 sparse sensors

⬅️ 返回 cs.LG 首页 · 🏠 返回主页