cs.LG（2026-07-27）

📊 共 20 篇论文

🎯 兴趣领域导航

支柱二：RL算法与架构 (RL & Architecture) (10) 支柱九：具身大模型 (Embodied Foundation Models) (5) 支柱一：机器人控制 (Robot Control) (3) 支柱八：物理动画 (Physics-based Animation) (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) (10 篇)

#题目一句话要点标签🔗⭐
1 Explainable Reinforcement Learning via Physics-Aware Policy Distillation 通过物理感知的策略蒸馏提升深度强化学习可解释性 reinforcement learning deep reinforcement learning DRL
2 MEGA-CL: A Molecular Foundation Model for Generalizable ADMET Prediction through Graph External Attention and Contrastive Learning 提出MEGA-CL以解决小分子ADMET预测的挑战 contrastive learning foundation model
3 Stress-Testing EEG Foundation Models for Clinical Decoding: Dataset Identity and Targeted Negative Controls 评估EEG基础模型在临床解码中的有效性与鲁棒性 Mamba foundation model
4 ACRL: Adaptive Control of Training-Inference Discrepancy for Stable Reinforcement Learning 提出自适应控制方法以解决强化学习训练不稳定问题 reinforcement learning large language model
5 Context Is King: How In-Context Specification Shapes the Geometry of Concepts 提出上下文规范以重塑概念几何结构 world model world models large language model
6 FlowCTS: On-policy Continuous Trajectory Supervision of Flow Models 提出FlowCTS以解决流模型的稀疏奖励和偏见问题 distillation large language model
7 Unsupervised Graph Representation Learning with Complementary View Alignment 提出AlignGAE以解决异质图表示学习问题 representation learning
8 ML-based Predictive Models for Power Consumption in Virtualised O-RANs 提出基于机器学习的模型以预测虚拟化O-RAN中的功耗 predictive model
9 Constrained Reinforcement Learning Using Successor Representations 提出SafeDSR以解决强化学习中的安全约束问题 reinforcement learning
10 WorldDiT: A Unified Diffusion Architecture for World and Action Modeling 提出WorldDiT以解决机器人控制中的视觉与动作建模问题 world model world models

🔬 支柱九：具身大模型 (Embodied Foundation Models) (5 篇)

#题目一句话要点标签🔗⭐
11 LOCKS: Page-Local Compact Key Summaries for Efficient Long-Context Decoding 提出LOCKS以解决长上下文解码中的KV缓存瓶颈问题 large language model
12 When LLM Defenses Backfire: Characterizing Safety, Performance, and Cost Trade-offs 系统研究LLM防御的安全性、性能与成本权衡 large language model
13 DynaCalKV: Key-Value Cache Compression via Head Grouping and Adaptive Rank Allocation 提出DynaCalKV以解决长上下文窗口中的KV缓存瓶颈问题 large language model
14 KAP: Bridging the Knowledge Selection-Runtime Consumption Gap in LLM Systems 提出KAP以解决LLM系统中的知识选择与运行消耗差距问题 multimodal
15 HydroAgent: Formalizing Forecaster Expertise into Skill-Orchestrated Flood Forecasting Workflows 提出HydroAgent以解决洪水预报中专家经验难以形式化的问题 large language model

🔬 支柱一：机器人控制 (Robot Control) (3 篇)

#题目一句话要点标签🔗⭐
16 Learning Reusable Hybrid Motion Priors for Humanoid Locomotion from Motion Imitation 提出可重用的混合运动先验以解决类人机器人运动控制问题 humanoid humanoid control humanoid locomotion
17 Evaluating Fuzz Testing for Reinforcement Learning Agents 提出全面评估方法以优化强化学习代理的模糊测试 bipedal biped reinforcement learning
18 The balance between compactness and forecast accuracy of data-driven latent-space reduced-order models in controlled wake flows 提出数据驱动的低阶模型以平衡紧凑性与预测精度 model predictive control reinforcement learning latent dynamics

🔬 支柱八：物理动画 (Physics-based Animation) (1 篇)

#题目一句话要点标签🔗⭐
19 Frequency-Based Reservoir computing 提出基于频率的水库计算以优化时间序列预测 spatiotemporal

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

#题目一句话要点标签🔗⭐
20 Capacity-Aware Deep Learning for Generalizable Traffic Volume Estimation Across Links and Cities 提出容量感知深度学习以解决城市间交通流量估计问题 sparse sensors

⬅️ 返回 cs.LG 首页 · 🏠 返回主页