cs.LG（2025-07-14）

📊 共 29 篇论文 | 🔗 6 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) (10 🔗3) 支柱八：物理动画 (Physics-based Animation) (2) 支柱七：动作重定向 (Motion Retargeting) (2) 支柱一：机器人控制 (Robot Control) (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) (13 篇)

#题目一句话要点标签🔗⭐
1 Pimba: A Processing-in-Memory Acceleration for Post-Transformer Large Language Model Serving Pimba：面向后Transformer大语言模型服务的存内计算加速方案 SSM state space model linear attention
2 Offline Reinforcement Learning with Wasserstein Regularization via Optimal Transport Maps 提出基于最优传输映射和Wasserstein正则化的离线强化学习方法，解决分布偏移问题。 reinforcement learning offline RL offline reinforcement learning ✅
3 GHPO: Adaptive Guidance for Stable and Efficient LLM Reinforcement Learning 提出GHPO：自适应引导的稳定高效LLM强化学习框架 reinforcement learning imitation learning curriculum learning
4 Graph World Model 提出图世界模型GWM，统一处理非结构化和图结构数据，支持多模态任务。 world model foundation model ✅
5 Reasoning or Memorization? Unreliable Results of Reinforcement Learning Due to Data Contamination 揭示数据污染对RL微调大模型数学推理能力评估的影响，提出清洁数据集RandomCalculation。 reinforcement learning large language model
6 MoCap-Impute: A Comprehensive Benchmark and Comparative Analysis of Imputation Methods for IMU-based Motion Capture Data MoCap-Impute：针对IMU运动捕捉数据缺失值插补的综合基准与对比分析 MAE IMU-based motion
7 Recognizing Dementia from Neuropsychological Tests with State Space Models 提出基于状态空间模型的Demenba框架，用于神经心理学测试的痴呆症自动识别。 state space model large language model
8 A Generalizable Physics-Enhanced State Space Model for Long-Term Dynamics Forecasting in Complex Environments 提出Phy-SSM，融合物理知识的状态空间模型，用于复杂环境下的长期动态预测。 SSM state space model ✅
9 Compression Method for Deep Diagonal State Space Model Based on $H^2$ Optimal Reduction 提出基于$H^2$最优降阶的深对角状态空间模型压缩方法 SSM state space model
10 FusionFactory: Fusing LLM Capabilities with Multi-LLM Log Data FusionFactory：融合多LLM日志数据，提升LLM在不同任务上的性能。 distillation large language model
11 Feature Distillation is the Better Choice for Model-Heterogeneous Federated Learning 提出FedFD：一种基于特征蒸馏的模型异构联邦学习方法 distillation
12 Text-Driven Causal Representation Learning for Source-Free Domain Generalization 提出TDCRL，通过文本驱动的因果表示学习解决无源域泛化问题 representation learning
13 Multi-Armed Sampling Problem and the End of Exploration 提出多臂采样框架，证明采样无需探索，为熵正则化强化学习等提供理论基础。 reinforcement learning RLHF

🔬 支柱九：具身大模型 (Embodied Foundation Models) (10 篇)

#题目一句话要点标签🔗⭐
14 Towards Applying Large Language Models to Complement Single-Cell Foundation Models 提出scMPT模型，融合单细胞Foundation模型与LLM，提升单细胞分析性能。 large language model foundation model
15 ElasticMM: Efficient Multimodal LLMs Serving with Elastic Multimodal Parallelism ElasticMM：通过弹性多模态并行加速多模态LLM服务 large language model multimodal
16 LaCache: Ladder-Shaped KV Caching for Efficient Long-Context Modeling of Large Language Models LaCache：一种梯形KV缓存方法，用于高效的大语言模型长文本建模 large language model ✅
17 TolerantECG: A Foundation Model for Imperfect Electrocardiogram TolerantECG：一种对噪声和导联缺失具有鲁棒性的心电图（ECG）基础模型 foundation model
18 AdaBrain-Bench: Benchmarking Brain Foundation Models for Brain-Computer Interface Applications 提出AdaBrain-Bench，用于评估脑机接口应用中脑基础模型的性能 foundation model
19 Semantic Context for Tool Orchestration 提出基于语义上下文的工具编排方法，提升LLM在复杂任务中的性能 large language model
20 Iceberg: Enhancing HLS Modeling with Synthetic Data Iceberg：通过合成数据增强HLS建模，提升泛化能力 large language model ✅
21 Memorization Sinks: Isolating Memorization during LLM Training 提出MemSinks，通过隔离记忆神经元解决LLM训练中的隐私和版权问题。 large language model ✅
22 Mechanistic Interpretability of LoRA-Adapted Language Models for Nuclear Reactor Safety Applications 针对核反应堆安全应用，提出LoRA微调语言模型的可解释性分析方法 large language model
23 Rethinking Prompt Optimization: Reinforcement, Diversification, and Migration in Blackbox LLMs 提出一种新型的提示优化框架以提升LLM性能 large language model

🔬 支柱八：物理动画 (Physics-based Animation) (2 篇)

#题目一句话要点标签🔗⭐
24 Boosted Enhanced Quantile Regression Neural Networks with Spatiotemporal Permutation Entropy for Complex System Prognostics 提出基于时空排列熵与增强分位数回归神经网络的复杂系统预测框架 spatiotemporal multimodal
25 HEIMDALL: a grapH-based sEIsMic Detector And Locator for microseismicity 提出基于图神经网络的地震检测与定位模型HEIMDALL，用于微震监测和地震目录生成。 spatiotemporal

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

#题目一句话要点标签🔗⭐
26 ZClassifier: Temperature Tuning and Manifold Approximation via KL Divergence on Logit Space ZClassifier：通过KL散度在Logit空间进行温度调整和流形逼近 geometric consistency
27 Rethinking Inductive Bias in Geographically Neural Network Weighted Regression GNNWR的归纳偏置再思考：融合CNN、RNN和Transformer以提升空间回归性能 spatial relationship

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
28 MTF-Grasp: A Multi-tier Federated Learning Approach for Robotic Grasping MTF-Grasp：一种用于机器人抓取的的多层联邦学习方法 manipulation

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

#题目一句话要点标签🔗⭐
29 Algorithm Development in Neural Networks: Insights from the Streaming Parity Task 通过流式奇偶校验任务，揭示神经网络算法涌现的机制 implicit representation

⬅️ 返回 cs.LG 首页 · 🏠 返回主页