cs.LG（2026-06-29）

📊 共 33 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) (17) 支柱九：具身大模型 (Embodied Foundation Models) (12 🔗1) 支柱一：机器人控制 (Robot Control) (2) 支柱三：空间感知与语义 (Perception & Semantics) (1) 支柱四：生成式动作 (Generative Motion) (1)

🔬 支柱二：RL算法与架构 (RL & Architecture) (17 篇)

#题目一句话要点标签🔗⭐
1 PS-PPO: Prefix-Sampling PPO for Critic-Free RLHF 提出PS-PPO以解决RLHF中的优化成本问题 reinforcement learning PPO RLHF
2 DreamForge-World 0.1 Preview: A Low-Compute Real-Time Controllable World Model 提出DreamForge-World以解决低计算实时世界建模问题 world model world models multimodal
3 DRIFT: Difficulty Routing Self-DIstillation with Rhythm-Gated Exploration and Success BuFfer Training 提出DRIFT框架以解决大语言模型自我改进中的学习进度跟踪问题 reinforcement learning curriculum learning distillation
4 Toward an Energy-Optimized Operation of Data Centers Located in Wind Farms Using Reinforcement Learning 提出基于强化学习的风电场数据中心能效优化方案 reinforcement learning PPO SAC
5 RoAd-RL: A Unified Library and Benchmark for Robust Adversarial Reinforcement Learning 提出RoAd-RL以解决对抗性强化学习的评估与重现性问题 reinforcement learning deep reinforcement learning DRL
6 Staged Hybridisation for Visual Quantum Reinforcement Learning via Knowledge Distillation 提出分阶段混合化策略以解决视觉量子强化学习问题 reinforcement learning distillation
7 DuoMem: Towards Capable On-Device Memory Agents via Dual-Space Distillation 提出DuoMem以解决资源受限设备上的记忆增强代理问题 distillation large language model
8 MuonSSM: Orthogonalizing State Space Models for Sequence Modeling 提出MuonSSM以解决状态空间模型不稳定性问题 SSM state space model
9 When Does Online Imitation Learning Help in LLM Post-Training? The Role of (Non-)Realizability Beyond Horizon 提出在线模仿学习以提升LLM后训练效果 imitation learning distillation
10 Diffusion Fine-tuning with Rewarded Moment Matching Distillation 提出奖励时刻匹配蒸馏以优化扩散模型训练 reinforcement learning distillation
11 Discovering Collaboration from Novelty: Random Network Distillation for Clustered Federated Learning 提出随机网络蒸馏以解决聚类联邦学习中的数据异质性问题 distillation
12 Experience Augmented Policy Optimization for LLM Reasoning 提出经验增强策略优化以提升大语言模型推理能力 reinforcement learning large language model
13 Notes on generative modeling: flow matching, diffusion, optimal transport and Schr{ö}dinger bridge 探讨生成建模中的流匹配与最优传输技术 flow matching
14 Building Multi-Task Agentic LLMs via Two-Phase Distillation 提出双阶段蒸馏方法以构建多任务智能大模型 distillation
15 Simplifying Flow Matching Transformations with Low-Rank Mixture Models 提出低秩混合模型以简化流匹配变换 flow matching
16 FlowAWR: Online Adaptive Flow Reinforcement via Advantage-Weighted Rectification 提出FlowAWR以解决生成流模型在线适应性问题 reinforcement learning classifier-free guidance
17 Neural Subspace Reallocation: Continual Learning as Retrieval-Based Subspace Memory Management 提出神经子空间重新分配以解决持续学习中的记忆管理问题 reinforcement learning distillation

🔬 支柱九：具身大模型 (Embodied Foundation Models) (12 篇)

#题目一句话要点标签🔗⭐
18 Arko-T: A Foundation Model for Text-to-Structured 3D Generation 提出Arko-T以解决文本到结构化3D生成的设计编辑问题 foundation model
19 Beyond IID: How General Are Tabular Foundation Models, Really? 提出BeyondArena以解决表格数据基准测试碎片化问题 foundation model
20 KnowsTFM: Knowledge-Informed Fine-Tuning of Small Tabular Foundation Models 提出知识驱动的小型表格基础模型微调方法以解决数据稀缺问题 foundation model
21 Towards Generalizable and Evidential Nuclear Magnetic Resonance-Based Molecular Structure Elucidation via Large Language Model Agent 提出NMRAgent以解决NMR分子结构解析中的可解释性问题 large language model
22 From Failure Taxonomy to Intervention: A Diagnostic Methodology for Industry-Scale AVLM in Video and Live-Streaming Platform Moderation 提出诊断方法以解决行业规模AVLM开发中的失败问题 foundation model multimodal
23 C$^{2}$R: Cross-sample Consistency Regularization Mitigates Feature Splitting and Absorption in Sparse Autoencoders 提出C$^{2}$R以解决稀疏自编码器中的特征分裂与吸收问题 large language model ✅
24 Attractor States Emerge in Multi-Turn LLM Conversations 研究多轮对话中的吸引子状态以理解LLM互动行为 large language model
25 HSAP: A Hierachical Sequence-aware Parallelism for Hybrid-Context Generative Models 提出HSAP以解决混合上下文序列并行计算问题 large language model
26 Online Data Selection for Instruction Tuning via Gaussian Processes 提出GAIA框架以解决在线数据选择问题 large language model
27 NeuReasoner: Theory-grounded Mapping of Reasoning Elicitation Boundaries 提出NeuReasoner以解决推理引发边界问题 large language model
28 Theory of Continual Learning Against Data Poisoning Attacks 提出理论框架以应对数据中毒攻击的持续学习问题 large language model
29 A Diagnostic Framework and Multi-Evaluator Audit of Evaluator-Driven Preference Dynamics in Self-Adapting LLM Agents 提出EPC框架以解决LLM评估者动态偏好失效问题 multimodal

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
30 Dual-Flow Reinforcement Learning with State-Aware Exploration 提出双流强化学习以解决多模态探索与价值估计问题 humanoid reinforcement learning flow matching
31 Robust Strategic Classification under Decision-Dependent Cost Uncertainty 提出决策依赖成本不确定性的稳健战略分类方法 manipulation

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

#题目一句话要点标签🔗⭐
32 MemLeak: Diagnosing Information Leaks in Multimodal Agent Memory 提出MemLeak以诊断多模态智能体记忆中的信息泄露问题 affordance multimodal

🔬 支柱四：生成式动作 (Generative Motion) (1 篇)

#题目一句话要点标签🔗⭐
33 Towards Continual Motion-Language Agents: LoRA Variants for Incremental Motion Understanding and Generation 提出LoRA变体以解决持续运动语言理解与生成问题 text-to-motion large language model

⬅️ 返回 cs.LG 首页 · 🏠 返回主页