cs.LG（2026-10-07）

📊 共 49 篇论文 | 🔗 2 篇有代码

🎯 兴趣领域导航

支柱二：RL算法与架构 (RL & Architecture) (27 🔗2) 支柱九：具身大模型 (Embodied Foundation Models) (18) 支柱一：机器人控制 (Robot Control) (4)

🔬 支柱二：RL算法与架构 (RL & Architecture) (27 篇)

#题目一句话要点标签🔗⭐
1 Beyond Policy Support: Interaction Constrained Offline Reinforcement Learning for Autonomous Driving 提出交互约束离线强化学习以解决自动驾驶中的分布偏移问题 reinforcement learning offline reinforcement learning world model ✅
2 MUNITE: Unified Multimodal Latent Inference for Any-to-Any Multimodal Generation 提出MUNITE以解决多模态生成中的推理问题 flow matching distillation multimodal
3 LeCuration: A Tiny World Model as a Data Curation Multi-Tool 提出LeCuration以解决物理AI数据整理问题 world model worldmodel world models
4 DSReg: Provably Recovering Individual World Latents without Reconstruction 提出DSReg以无重建方式恢复个体世界潜变量 world model world models JEPA
5 A Good Self-Teacher Meets the Student Where They Are: Joint On-Policy Learning and Teaching 提出JOLT方法以解决稀疏监督下的强化学习问题 reinforcement learning policy learning distillation
6 Sparse Planning in Visual World Models via Cost Gradients 提出COSTGRAD以解决稀疏规划中的高计算成本问题 world model world models ✅
7 COPC: Coupled Off-Policy Correction for Asynchronous LLM Reinforcement Learning 提出COPC以解决异步LLM强化学习中的策略与优势估计不匹配问题 reinforcement learning PPO large language model
8 Multi-Agent Coordination via Support-Preserving Distillation 提出MoSDOT以解决离线多智能体强化学习中的教师训练问题 distillation multimodal
9 MorphCL: Morphological Contrastive Learning for Inertial-based Human Activity Recognition 提出MorphCL以解决惯性传感器人类活动识别中的数据结构问题 contrastive learning foundation model
10 RollVerify: Bridging Efficiency and Accuracy in Long-Tail Rollout Reinforcement Learning 提出RollVerify以解决长尾回滚强化学习中的效率与准确性问题 reinforcement learning large language model
11 Collaborative Reasoning Distillation via Cross-Feedback and Coherent Curation 提出协作推理蒸馏方法以提升小型模型的推理能力 distillation large language model
12 Composing What Each Teacher Learned: Multi-Teacher On-Policy Distillation through Teacher-Relative Shifts 提出Δ-MOPD以优化多教师在线蒸馏过程 distillation
13 Conditional Flow Matching for Generation of 3D Multi-variable Instantaneous Urban Microclimate Fields 提出条件流匹配方法以快速生成城市微气候场 flow matching
14 Executing Causal Structure Learning with Linear-Attention Transformers 提出线性注意力变换器以执行因果结构学习 linear attention
15 Average-Reward Reinforcement Learning for Multichain MDPs: A Hierarchical Decomposition Approach 提出一种层次分解方法解决多链MDP的平均回报强化学习问题 reinforcement learning
16 Continual Graph Multi-Agent Reinforcement Learning 提出CGMARL框架以解决多任务学习中的遗忘问题 reinforcement learning
17 Shared Gaussianization: What Gaussian Regularizers Certify About Contrastive Learning, and What They Miss 提出共享高斯化方法以优化对比学习中的分布匹配 contrastive learning
18 A Closed-Loop Non-Asymptotic Convergence Analysis of PPO with Learned Critics and Clipping 提出闭环非渐近收敛分析以优化PPO-Clip算法 PPO
19 Policy Learning with Weak Signals 提出基于线性平滑的策略学习方法以应对弱信号挑战 policy learning
20 BoT-GRPO: Efficient Process-Reward RL for Reasoning via Bag-of-Token Aggregation 提出BoT-GRPO以提高强化学习中推理效率 reinforcement learning large language model
21 EC-EarthFlow: Probabilistic emulation of daily transient global climate model simulations with flow matching 提出EC-EarthFlow以高效模拟全球气候模型EC-Earth3 flow matching
22 Boundary-aware Reinforcement Learning for Hypercube State Spaces via Deterministic Policy Gradient 提出边界感知强化学习框架以解决超立方体状态空间问题 reinforcement learning
23 GeoPrior-Mamba: Structured Process Priors with Mamba for Fine-Resolution XCO2 Reconstruction 提出GeoPrior-Mamba以解决稀疏卫星观测下的XCO2重建问题 Mamba
24 Kuration SDK: Addressing the Virtual2Real Gap via Data Curation 提出Kuration SDK以解决虚拟与现实之间的差距问题 world model world models
25 Decoupling Exploration from Optimization in RLVR 提出探索-蒸馏框架以解决RLVR中的优化与探索问题 reinforcement learning distillation
26 Distilling Graph Geometry: Knowledge Gap from GNNs to MLPs 提出G^2MLP以解决GNN到MLP蒸馏中的几何知识缺失问题 teacher-student distillation
27 Unrolled Flow Models for Reasoning 提出基于展开流模型的推理方法以解决图可达性问题 flow matching latent dynamics

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

#题目一句话要点标签🔗⭐
28 Multimodal LLMs Can Learn to Read Brain Signals: A Vision--Language Model for Unified Multi-Task EEG Decoding 提出BraVista框架以解决EEG解码多任务学习问题 foundation model multimodal
29 Thinking in Depth: Retrospective Inference for Tabular Foundation Models 提出Retro模型以优化表格基础模型的预测精度 foundation model
30 Fault-tolerant foundation models 提出容错基础模型以应对不可靠硬件问题 foundation model
31 Efficient Provably Private Classification with a Tabular Foundation Model 提出PrivTab以解决敏感数据分类中的隐私保护问题 foundation model
32 WxFM-XL: Adapting Univariate Foundation Models to Multi-Station Weather Forecasting 提出WxFM-XL以解决多站点天气预报中的信息缺失问题 foundation model
33 Backdooring Acoustic Foundation Models for Physically Realizable Triggers 提出FAB攻击以解决声学基础模型的安全隐患问题 foundation model
34 Pretraining Shapes Spectral Structure: Architecture- and Strategy-Conditional Prediction of OOD Robustness in Foundation Models 提出基于谱结构的预训练权重分析以评估模型的OOD鲁棒性 foundation model
35 Training Parallel Speculative Draft Models by Directly Minimizing Expected Decoding Rounds 通过直接最小化期望解码轮次训练并行推测草稿模型 large language model
36 SemanticFold: Latent Sequence Compression SeparatesLanguage Modeling, Decodability, and Reasoning 提出SemanticFold以解决大语言模型推理能力压缩问题 large language model
37 Neural Sampling with Reweighted Normalizing Flows via the Wasserstein--Fisher--Rao JKO Scheme 提出基于WFR JKO方案的神经采样算法以解决无规范化Boltzmann分布问题 multimodal
38 From Prompts to Trees: Effective LLM-Guided Tree Generation for Few-Shot Tabular Classification 提出一种新框架以实现少样本表格分类中的决策树生成 large language model
39 Robust Decentralized Fairness Auditing 提出Auditopus以解决去中心化公平性审计问题 large language model
40 Identifiability of a dissipative knowledge-dynamics model: exact recovery under designed excitation, degeneration on observational data 提出非线性耗散系统模型以解决人类学习参数可识别性问题 TAMP
41 Reproducible LLM Inference Benchmarking: A Sequential Isolation Protocol for Regression Testing 提出顺序隔离方法以解决大语言模型推理基准测试的可重复性问题 large language model
42 A Strength-Monotonic Law for Domain Alignment in Frozen-Embedding Bioacoustic Classification 提出强度单调法则以优化生物声学分类中的领域对齐问题 foundation model
43 Sequential Pretraining Favors Large Models 提出Exposure Therapy以解决小模型在序列预训练中的学习效率问题 foundation model
44 Scalable Logistic Gaussian Process Density Regression with Kinetic Langevin Sampling 提出可扩展的逻辑高斯过程密度回归以解决条件密度估计问题 foundation model
45 Shared Low-rank Basis Factorization for Data-free Mixture-of-Experts Compression 提出共享低秩基因分解以解决数据无关的MoE压缩问题 large language model

🔬 支柱一：机器人控制 (Robot Control) (4 篇)

#题目一句话要点标签🔗⭐
46 An Informational Curse of Horizon in Goal-Conditioned Policy Learning 提出信息性地平线诅咒以解决目标条件策略学习问题 manipulation reinforcement learning policy learning
47 Q-Learning with Scalar Adjoint Matching 提出标量伴随匹配的Q学习以优化流政策 bi-manual vision-language-action
48 Safe on Average, Unsafe in the Tail: When Is the Episodic-Cost Tail Controllable? 提出基于CVaR的尾部安全性评估以解决强化学习中的成本约束问题 locomotion reinforcement learning
49 Beyond Reward Suppression: Near-Optimal Offline Attacks on Warm-Start Bandits with Bounded Rewards 提出近最优离线攻击方法以解决温启动强盗问题 manipulation

⬅️ 返回 cs.LG 首页 · 🏠 返回主页