cs.LG（2026-08-05）

📊 共 28 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) (13 🔗3) 支柱九：具身大模型 (Embodied Foundation Models) (12 🔗1) 支柱一：机器人控制 (Robot Control) (2) 支柱八：物理动画 (Physics-based Animation) (1)

🔬 支柱二：RL算法与架构 (RL & Architecture) (13 篇)

#题目一句话要点标签🔗⭐
1 Multimodal Spatiotemporal Atmospheric Data Assimilation with Latent Flow-matching 提出基于潜在流匹配的多模态时空大气数据同化方法 flow matching spatiotemporal multimodal
2 Agentic Reinforcement Learning with Observation-Calibrated Self-Distillation 提出观察校准自蒸馏以解决稀疏奖励问题 reinforcement learning distillation privileged information ✅
3 Helping Music Co-Creation Agents 'Listen' Well: Hierarchical Self-Supervised World Models for Understanding and Generation 提出层次自监督世界模型以增强音乐共创代理的理解与生成能力 flow matching world model world models
4 SpecRoll: Fast-Slow Verifier-Feedback Adaptation for Speculative Reinforcement Learning Rollouts 提出SpecRoll以解决强化学习中生成效率瓶颈问题 reinforcement learning large language model
5 NodeJEPA: Structure-Conditioned Latent Prediction for Node-Level Graph Self-Supervised Learning 提出NodeJEPA以解决节点级图自监督学习的结构依赖问题 JEPA Joint-Embedding Predictive Architecture joint-embedding predictive architecture ✅
6 Learning When to Stop: Prefix-Optimal Dynamic Diffusion Policies for Continuous Control 提出Prefix-Optimal生成策略以解决动态扩散策略的计算瓶颈问题 diffusion policy
7 Optimizing What Policies Learn From: Recoverability-aware Rollout Intervention Learning 提出Recoverability-Aware Intervention Learning以优化策略学习过程 reinforcement learning large language model
8 Consistency-Driven Co-Evolution for Self-Supervised Cross-Representation Learning 提出CoCoEvolve以解决跨表示学习中的一致性问题 representation learning ✅
9 Attention, Anomalies! Handling Attention Layers in Unsupervised Federated Outlier Detection 提出新聚合方法以解决联邦学习中的异常检测问题 representation learning large language model
10 Differentiating Through Dual Prices: End-to-End Policy Learning Under Capacity Constraints 提出双价格区分方法以解决资源分配中的容量约束问题 policy learning
11 Discretization and Statistical Consistency of Functional Flow Matching 提出功能流匹配的离散化与统计一致性方法以解决传感器极限问题 flow matching
12 SPOT: Sparse Probing and Outcome Calibration for On-Policy Distillation 提出SPOT以解决在策略蒸馏中的不确定性问题 distillation
13 Not Every Divergence Should Be Suppressed: Counterfactual Recoverability in On-Policy Distillation 提出反事实可恢复性方法以优化在线蒸馏训练 distillation

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

#题目一句话要点标签🔗⭐
14 Deltoris: Enabling Real-time VLA Inference in Embodied AI via Bit-level Sparsity and Speculative Inference 提出Deltoris以解决边缘设备上VLA推理效率问题 embodied AI vision-language-action VLA
15 MultiPathFormer: Towards a Foundation Model for Multipath Wireless Propagation 提出MultiPathFormer以解决无线传播建模问题 foundation model
16 Above-ground Biomass Estimation with Geospatial Foundation Models 利用地理基础模型提高地上生物量估计的准确性 foundation model
17 Personalized Federated Sparse Adaptation of Time-Series Foundation Models 提出个性化联邦稀疏适应框架以解决时间序列模型适应问题 foundation model
18 EvtGraph: Event-Adaptive Compression for Sparse Temporal Graph Learning in Multimodal Time Series 提出EvtGraph以解决多模态时间序列稀疏图学习中的信息不均问题 multimodal
19 A Model Merging Approach for Continual MLLM Unlearning 提出MCU以解决多模态大语言模型的持续去学习问题 large language model multimodal
20 DASyR-LLM: Domain-Aware Symbolic Regression with LLMs for Kinetic Model Discovery 提出DASyR-LLM框架以解决动量模型发现问题 large language model
21 SVI-DAG: A Structured Variational Inference Approach to Bayesian Causal Discovery 提出SVI-DAG以解决贝叶斯因果发现中的图识别问题 multimodal
22 On MUON optimization: From non-convergence to an error analysis with Polar Express and the Newton-Schulz polynomial from implementations 提出MUON优化器的广义变体以解决收敛性问题 large language model
23 Training-Free Hashing-Based Attention via Binary Principal Components 提出BinaryPC以解决长上下文LLMs的自注意力效率瓶颈 large language model ✅
24 Looking in the Mirror: Introspecting Side-Effect Misalignments Induced by Fine-Tuning 提出侧效应内省以解决微调引发的对齐偏差问题 large language model
25 Cost-Aware Multi-Objective Bandits: Theory and Application to Budgeted LLM Configuration Evaluation 提出成本感知多目标赌博模型以解决预算有限的LLM配置评估问题 large language model

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
26 Reward Structure Shapes the Interaction Between Episodic Exploration and Neural Memory in Reinforcement Learning 提出奖励结构以优化部分可观察强化学习中的探索与记忆交互 manipulation reinforcement learning
27 Manipulation-Proof Oblivious Audits against Deceptive Model Providers 提出一种新审计协议以解决模型提供者操控问题 manipulation

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

#题目一句话要点标签🔗⭐
28 The Neural Echo: A Signal Processing Perspective for Understanding Neural Networks 提出神经回声以理解神经网络行为 PULSE

⬅️ 返回 cs.LG 首页 · 🏠 返回主页