cs.LG（2026-10-19）

📊 共 42 篇论文 | 🔗 3 篇有代码

🎯 兴趣领域导航

支柱二：RL算法与架构 (RL & Architecture) (22 🔗1) 支柱九：具身大模型 (Embodied Foundation Models) (15 🔗2) 支柱八：物理动画 (Physics-based Animation) (2) 支柱一：机器人控制 (Robot Control) (1) 支柱四：生成式动作 (Generative Motion) (1) 支柱五：交互与反应 (Interaction & Reaction) (1)

🔬 支柱二：RL算法与架构 (RL & Architecture) (22 篇)

#题目一句话要点标签🔗⭐
1 Simulation Priors for Rapid Adaptation in Model-Based RL for Autonomous Racing 提出Prior-Informed Adaptation以解决模型基强化学习在自主赛车中的快速适应问题 reinforcement learning world model world models
2 Next Best Action: Causal Constrained Offline RL for Business Action Targeting 提出基于因果约束的离线强化学习以优化商业行动选择 reinforcement learning offline RL offline reinforcement learning
3 Cooperative Multi-Agent Reinforcement Learning for Market Making on Prediction-Market Order Books 提出合作多智能体强化学习以解决预测市场做市问题 reinforcement learning reward design reward shaping
4 A Fisher-Geometric View of Multi-Teacher On-Policy Distillation 提出RGBM方法以优化多教师在线蒸馏过程 distillation foundation model
5 LOOP: Recovery-Aligned Iterative Layer Pruning via On-Policy Distillation 提出LOOP框架以解决层级剪枝中的恢复能力问题 distillation large language model
6 Zephyrus-R1: Agentic Reinforcement Learning for Weather Reporting 提出Zephyrus-R1以解决天气报告中的RL验证难题 reinforcement learning curriculum learning reward design
7 Curriculum Learning Teaches LLMs Long-Horizon Implicit Reasoning, Up To A Limit 通过课程学习提升LLM的长远隐式推理能力 curriculum learning chain-of-thought
8 Nash-MOPD: Optimizing Task Trade-offs in Multi-Teacher On-Policy Distillation 提出Nash-MOPD以优化多教师策略蒸馏中的任务权衡问题 teacher-student distillation
9 Correctness-Dependent Asymmetric Target Distributions for Self-Distillation Policy Optimization 提出不对称目标分布以优化自蒸馏策略 reinforcement learning distillation
10 Asynchronous On-Policy Self-Distillation 提出异步在线自蒸馏以提升训练效率 distillation privileged information
11 Statistical Learning Theory for Direct Preference Optimization 提出直接偏好优化方法以解决偏好建模不足问题 DPO direct preference optimization
12 Remember Less, Adapt Longer: Bounded-State In-Context Reinforcement Learning with Recurrent Summary Memory 提出RSM以解决长期任务中的记忆与适应问题 reinforcement learning transformer policy
13 Anatomy of Off-Policy Policy Gradient: Importance Sampling, KL Regularization, and Baselines 提出离线策略梯度方法以解决大语言模型后训练中的分布偏移问题 reinforcement learning offline RL large language model
14 Predictive Bandit Curriculum Reinforcement Learning via Gradient-Based Sample Group Discovery 提出预测性带式课程强化学习以提升训练效率 reinforcement learning curriculum learning
15 OPDArena: Understanding Learning and Forgetting in On-Policy Distillation 提出OPDArena框架以理解在政策蒸馏中的学习与遗忘 reinforcement learning distillation
16 Conditional Average Treatment Effect Distillation for Hypothetical State-Intervention Targeting Under Budget Constraints 提出条件平均治疗效应蒸馏以解决预算约束下的干预目标问题 teacher-student distillation
17 Kronecker Delta Rule for Long Context Linear Attention 提出KronLSTM以解决长上下文语言建模问题 linear attention ✅
18 Reinforcement Learning Controllers for Fusion Scenario Development 提出强化学习控制器以优化核聚变场景开发 reinforcement learning
19 PnP-CF: A Consensus Flow-Matching Solver for General Inverse Problems 提出PnP-CF以解决一般逆问题 flow matching
20 Learn the Interface, Not the Implementation: Output Representation Dominates Training Regime in LLM-Based RTL Generation 提出基于输出表示的训练策略以优化LLM的RTL生成 reinforcement learning large language model
21 Adaptive Decoding Methods for Token-Efficient Reinforcement Learning 提出自适应解码方法以提高强化学习的采样效率 reinforcement learning
22 VaDar: Implicit Goal Conditioning via Value Disaggregation 提出VaDar以解决长时间任务中的子目标选择问题 reinforcement learning behavior cloning

🔬 支柱九：具身大模型 (Embodied Foundation Models) (15 篇)

#题目一句话要点标签🔗⭐
23 SleepPairNet: Learning Multimodal Pediatric Sleep Representations with Random-Pair Cross-Attention 提出SleepPairNet以解决多模态儿童睡眠信号分析问题 multimodal
24 Bombadil: A Foundation Model for Plant Electrophysiology 提出Bombadil模型以解决植物电生理学任务的迁移学习问题 foundation model
25 From Retrieval to Recognition:Knowledge-Enhanced Post-training of Time Series Foundation Models for Classification 提出R2R方法以解决时间序列分类中的知识利用问题 foundation model
26 Internalizing Chain-of-thought into Looped Transformers 提出循环变换器以优化链式思维计算资源 chain-of-thought
27 Computation, Not Memory: Chain-of-Thought and State Tracking in LLM Agents 提出基于推理生成与状态跟踪的LLM代理方法以解决动态状态下的决策问题 chain-of-thought
28 Adversarial Post-Training of Foundation Models for High-Frequency Trading 提出对抗性后训练方法以提升高频交易模型的市场模拟能力 foundation model
29 Scaling Parameter-Efficient Fine-Tuning for Brain Encoding 提出参数高效微调方法以优化脑编码模型 multimodal
30 On Data Scaling in Tabular ML: More Data, Better Generalization? 研究数据扩展在表格机器学习中的影响与挑战 foundation model
31 Training with ML-labeled Data via Stochastic First- and Zeroth-Order Optimization 提出随机一阶和零阶优化方法以解决ML标注数据训练问题 large language model
32 Parallel Language Models via Variational Decoding 提出并行语言模型以解决自回归模型推理延迟问题 instruction following ✅
33 Rethinking LLM Fine-Tuning via Weight Space Reparameterization: Preserving Safety during Downstream Adaptation 提出WSR-Tune以解决大语言模型微调中的安全性问题 large language model
34 SPLASH: Contrastive Audit Pressure for Detecting and Localizing Security Hallucinations in Generated Code 提出SPLASH以检测和定位生成代码中的安全幻觉问题 large language model
35 NADS: Navigator-Guided Data Selection for Mitigating Catastrophic Forgetting in Fine-Tuning 提出NADS以解决大语言模型微调中的灾难性遗忘问题 large language model ✅
36 How Much Does One Wrong Answer Cost? Error Propagation in Constraint Pruning with Noisy LLM Oracles 提出错误传播模型以优化约束修剪中的噪声LLM预言机 large language model
37 Knowledge Manifold Optimization for Scientific Discovery 提出知识流形优化方法以促进科学发现 large language model

🔬 支柱八：物理动画 (Physics-based Animation) (2 篇)

#题目一句话要点标签🔗⭐
38 OceanTensorCube and ST-MMT: A Multimodal Data and Transformer System for Seven-Day Forecasting of Pyropia Bleaching 提出OceanTensorCube和ST-MMT以解决紫菜漂白预测问题 spatiotemporal multimodal
39 Variance-Aware Non-Stationary Bandits:All-Budget Regret Bounds and Adaptation Costs 提出方差感知非平稳多臂赌博机以优化预算悔恨界限 PULSE

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
40 Progress-Guided Preference Generation for Aligning Generative Control Policies 提出进度引导偏好生成方法以解决生成控制策略对专家示范依赖的问题 manipulation imitation learning flow matching

🔬 支柱四：生成式动作 (Generative Motion) (1 篇)

#题目一句话要点标签🔗⭐
41 Wet-Biased Data Pipelines Mislead the Training and Evaluation of Generative Precipitation Nowcasting 揭示湿偏数据管道对生成降水预报训练与评估的误导 classifier-free guidance

🔬 支柱五：交互与反应 (Interaction & Reaction) (1 篇)

#题目一句话要点标签🔗⭐
42 Maestro: Agentic Formal Proof Development by Coordination 提出Maestro以解决形式证明开发中的失败协调问题 IMoS

⬅️ 返回 cs.LG 首页 · 🏠 返回主页