cs.LG（2026-10-29）

📊 共 56 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) (26 🔗2) 支柱九：具身大模型 (Embodied Foundation Models) (17 🔗1) 支柱一：机器人控制 (Robot Control) (9) 支柱七：动作重定向 (Motion Retargeting) (2 🔗1) 支柱四：生成式动作 (Generative Motion) (1) 支柱三：空间感知与语义 (Perception & Semantics) (1)

🔬 支柱二：RL算法与架构 (RL & Architecture) (26 篇)

#题目一句话要点标签🔗⭐
1 world2vec: Mechanistic world models through latent dynamics learning 提出world2vec以实现机械世界模型的学习 world model world models dreamer
2 DYNAS: A Dynamic Network Penetration Testing Framework Using Deep Hierarchical Reinforcement Learning 提出DYNAS框架以解决动态网络渗透测试问题 reinforcement learning penetration
3 H$^2$R: Hierarchical Hint-Based Reward for Mathematical Reasoning via Reinforcement Learning 提出层次提示奖励以解决复杂数学推理问题 reinforcement learning reward design large language model
4 Training Foundation Models with a Single Algorithm 提出统一训练算法以简化基础模型训练过程 reinforcement learning foundation model
5 Reinforcement Learning Under Runtime Confounding: Policy Learning and Off-Policy Evaluation 提出强化学习算法以解决运行时混淆问题 reinforcement learning policy learning
6 From MS/MS evidence to adaptive spectroscopic characterization: Reinforcement learning guided by candidate molecular structures 提出结构信息驱动的深度Q网络以优化小分子鉴定 reinforcement learning multimodal
7 Reinforcement learning for control of partial differential equations 提出强化学习方法以控制偏微分方程系统 reinforcement learning
8 SCARL: Sequence Confidence-Aligned Reinforcement Learning 提出SCARL以解决序列信心对齐的强化学习问题 reinforcement learning ✅
9 Latent Prediction Yields Cross-Subject Alignable Representations in Sensory-Evoked fMRI 提出自监督学习方法以实现跨参与者fMRI数据对齐 JEPA Joint-Embedding Predictive Architecture joint-embedding predictive architecture
10 Confidence-Aware Speed-over-Ground Estimation from Doppler Spectra via Reconstruction-Based Representation Learning 提出基于重建学习的信心感知地面速度估计方法 representation learning
11 Escaping Path Mirages in Offline Goal-Conditioned Reinforcement Learning 提出分支感知图规划以解决离线目标条件强化学习中的路径幻影问题 reinforcement learning ✅
12 Turing LLM Post-Training: Calibrated Distillation and Efficient Dynamic-MoE Execution 提出动态路由混合专家模型的后训练方法以提升整合效率 distillation
13 LinQuant: Functional Invariances for Low-Precision Linear Recurrences 提出LinQuant框架以解决低精度线性递归推理问题 Mamba large language model
14 Safe Multi-Objective Reinforcement Learning: Zero Constraint Violation and Primal-Averaging Convergence 提出安全多目标强化学习算法以解决约束违反问题 reinforcement learning
15 IMRouter: Imbalance-aware Temporal Pattern Routing for LLM-based Time Series Forecasting 提出IMRouter以解决时间序列预测中的模式不平衡问题 representation learning large language model
16 Grounding Predictive Reliability through Masked Observability 提出MORSE以解决预测不确定性语义模糊问题 representation learning PULSE
17 Improving the Intrinsic Interpretability of Differentiable Decision Trees through Euclidean Feature-Concentration Regularisation 提出欧几里得集中可微决策树以提升特征解释性 reinforcement learning PPO
18 Density-Guided Bootstrap Boundaries for Latent World-Model Reinforcement Learning 提出密度引导的自举边界以解决模型依赖深度选择问题 reinforcement learning world model world models
19 CAMEL: Cross-Architecture Masked Embedding Learning for Ultra-Lightweight Retinal Foundation Models 提出CAMEL以解决轻量级视网膜模型的计算复杂性问题 representation learning distillation foundation model
20 On the Origin and Evolution of Training-Inference Mismatch in LLM Reinforcement Learning 提出差异-滞后校正方法以解决LLM强化学习中的训练推理不匹配问题 reinforcement learning large language model
21 Communication-Efficient Private Federated Reinforcement Learning with Post-Quantum Authentication for Vehicular Networks 提出一种通信高效的私有联邦强化学习方法以解决车载网络中的数据隐私问题 reinforcement learning
22 Meshless Latent Dynamics Models for Model Order Reduction of time-dependent Partial Differential Equations 提出无网格潜在动力学模型以解决时间相关偏微分方程的模型降阶问题 latent dynamics
23 Turing LLM Post-Training: Calibrated Distillation and Efficient Dynamic-MoE Execution 提出动态路由专家模型的后训练方法以提高整合效率 distillation
24 Bayesian Distribution System State Estimation with Flow Matching 提出PoSE以解决配电系统状态估计中的不确定性问题 flow matching
25 Steering Vectors as a Training Signal in LLM Post-Training 提出使用引导向量作为LLM后训练信号以提升性能 reinforcement learning distillation
26 TensorSSM: Memorizing Spatial Position in Dynamic State Space for Tensor Core Generation 提出TensorSSM以解决高维张量存储效率问题 SSM state space model

🔬 支柱九：具身大模型 (Embodied Foundation Models) (17 篇)

#题目一句话要点标签🔗⭐
27 TaxoMFM: A Taxonomy-Aware Foundation Model for the Human Gut Microbiome 提出TaxoMFM以解决肠道微生物组模型的层次结构利用问题 foundation model zero-shot transfer
28 Quantum-Inspired Convolutional Networks for High-Resolution Pseudo-Label Generation from Multimodal Earth Observation Data 提出量子启发卷积网络以解决多模态地球观测数据伪标签生成问题 multimodal
29 TRACER, a time-to-event EHR foundation model for disease detection across health systems 提出TRACER模型以解决电子健康记录中的疾病检测问题 foundation model
30 IMPACT: Inverter-Informed Multimodal Learning for Photovoltaic Generation Forecasting 提出物理信息多模态学习框架以提高光伏发电预测精度 multimodal
31 TabCUQ: Context-Uncertainty Quantification for Tabular Foundation Models 提出TabCUQ以解决表格基础模型的不确定性量化问题 foundation model
32 A Pretrained Model for Bayesian Filtering in Heterogeneous Dynamical Systems 提出FilterFM以解决异构动态系统的贝叶斯滤波问题 foundation model
33 A Unified Hessian-Guided Framework for Post-Training Quantization 提出统一的Hessian引导框架以优化后训练量化 large language model
34 DolphinRouting: Cluster-Level Binary Routing for Disaggregated LLM Inference Serving 提出DolphinRouting以优化大规模语言模型推理服务 large language model
35 Attention, Dynamic Wave and Hybrid Sequence Mixers in Language Models 提出动态波动层以优化长序列建模效率 large language model
36 DWISE: Denoising Workload-Informed Scheduling Engine 提出DWISE以解决扩散大语言模型调度效率问题 large language model
37 Surprise Alone Learns Lies: Evidence-Gated Continual Learning for LLMs from Streaming Claims 提出证据门控持续学习方法以解决大语言模型中的谎言学习问题 large language model
38 Uncertainty-Guided Cross-Modal Alignment for Diffusion-Based IoT Intrusion Detection 提出基于不确定性引导的跨模态对齐方法以解决物联网入侵检测问题 multimodal
39 TaxoMFM: A Taxonomy-Aware Foundation Model for the Human Gut Microbiome 提出TaxoMFM以解决肠道微生物组模型的分类层次问题 foundation model zero-shot transfer
40 A Theoretical Framework for Understanding and Building "Any-task inference" Capable Foundation Models 提出理论框架以构建具备任何任务推理能力的基础模型 foundation model
41 Towards Understanding Score-Seeking in LLMs 探讨大型语言模型中的得分追求行为及其动机 large language model
42 FARO: Function-Space Adaptive Return-to-Risk Optimization via Cross-Block Equal-Risk-Contribution Budget Allocation 提出FARO优化器以提升大语言模型训练效率 large language model ✅
43 Certified Behavioral Retention for Continual Adaptation with Held-Out Anchors 提出认证锚点基础适应方法以解决持续适应问题 foundation model

🔬 支柱一：机器人控制 (Robot Control) (9 篇)

#题目一句话要点标签🔗⭐
44 Is It the Curriculum or Just the Easy Start? Benchmarking Unsupervised Environment Design on Quadrupeds 提出统一基准以评估无监督环境设计在四足机器人中的应用 quadruped legged robot bipedal
45 Smooth Action Chunking Via Movement-Primitive Representations for Imitation Learning 通过运动原语表示提出平滑动作分块以解决模仿学习中的抖动问题 bi-manual imitation learning diffusion policy
46 Generating Critic-Guided Synthetic Experience in Latent World Models 提出CGSE方法以提升离线策略的合成经验生成 manipulation world model world models
47 Learned Terminal Costs for Short-Horizon Model Predictive Control of Autonomous Trucks in Interactive Highway Traffic 提出混合MPC与RL框架以解决自主卡车高速公路驾驶问题 MPC model predictive control reinforcement learning
48 Coupled Guidance for Flow Matching 提出耦合引导机制以解决流匹配中的指导集成问题 manipulation dexterous manipulation flow matching
49 Frozen in Time: How a Physics Foundation Model Forgets Under Sim-to-Real Adaptation 研究表明SciFM在模拟到真实适应中存在灾难性遗忘问题 sim-to-real foundation model
50 Before We Adapt: The Zero-Shot Generalization Bottleneck in Continual World Models 提出基于上下文马尔可夫决策过程的零-shot泛化方法以解决持续世界模型中的瓶颈问题 MPC reinforcement learning world model
51 Characterizing Recoverable Expert-Utilization Signatures in Fine-Tuned Mixture-of-Experts LLMs 提出结构化记忆漏洞检测方法以解决MoE模型安全问题 manipulation large language model
52 Model-Agnostic Search for New Physics in the Presence of Imperfect Background by Smearing the Data 提出MASSIB以解决高能物理中背景不完美问题 manipulation

🔬 支柱七：动作重定向 (Motion Retargeting) (2 篇)

#题目一句话要点标签🔗⭐
53 LatentWeave: Test-Time Latent Optimization for Multimodal Reasoning 提出LatentWeave以解决多模态推理中的潜在优化问题 latent optimization multimodal ✅
54 Observability-Aligned Evidential Masked Pretraining for Reliable VRU Motion Prediction 提出eCoMAE以解决VRU运动预测中的不确定性问题 motion prediction

🔬 支柱四：生成式动作 (Generative Motion) (1 篇)

#题目一句话要点标签🔗⭐
55 End-to-End Guarantees for Masked Diffusion Models under Conditional Mixing 提出端到端保证的条件混合下的掩蔽扩散模型 MDM

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

#题目一句话要点标签🔗⭐
56 Event-SAMA: Language-Grounded Video Object Segmentation on Neuromorphic Event Streams 提出Event-SAMA以解决事件流视频物体分割问题 optical flow

⬅️ 返回 cs.LG 首页 · 🏠 返回主页