cs.LG（2025-05-28）

📊 共 44 篇论文 | 🔗 7 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (22 🔗4) 支柱二：RL算法与架构 (RL & Architecture) (20 🔗3) 支柱八：物理动画 (Physics-based Animation) (1) 支柱一：机器人控制 (Robot Control) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (22 篇)

#题目一句话要点标签🔗⭐
1 Defining Foundation Models for Computational Science: A Call for Clarity and Rigor 为计算科学定义基础模型：呼吁清晰性和严谨性 foundation model
2 On Learning Verifiers for Chain-of-Thought Reasoning 提出学习可信验证器框架，用于自然语言思维链推理的正确性验证。 chain-of-thought
3 DES-LOC: Desynced Low Communication Adaptive Optimizers for Training Foundation Models DES-LOC：面向大规模模型训练的解耦低通信自适应优化器 foundation model
4 EnsemW2S: Enhancing Weak-to-Strong Generalization with Large Language Model Ensembles EnsemW2S：利用大语言模型集成提升弱到强泛化能力 large language model
5 SlimLLM: Accurate Structured Pruning for Large Language Models SlimLLM：面向大语言模型的精确结构化剪枝方法 large language model
6 Revisiting Bayesian Model Averaging in the Era of Foundation Models 提出基于贝叶斯模型平均（BMA）和可优化模型平均（OMA）的集成方法，提升图像和文本分类任务性能。 foundation model
7 Investigating the effectiveness of multimodal data in forecasting SARS-COV-2 case surges 利用多模态数据预测SARS-COV-2病例激增，揭示国家和阶段异质性。 multimodal
8 NOCL: Node-Oriented Conceptualization LLM for Graph Tasks without Message Passing 提出NOCL，一种无需消息传递的面向节点概念化的大语言模型，用于图任务。 large language model foundation model
9 SimuGen: Multi-modal Agentic Framework for Constructing Block Diagram-Based Simulation Models SimuGen：多模态Agent框架，用于构建基于框图的Simulink仿真模型 large language model multimodal ✅
10 Scalable Parameter and Memory Efficient Pretraining for LLM: Recent Algorithmic Advances and Benchmarking 针对LLM预训练，提出权重重构和动量重置技术，提升参数效率和降低内存需求。 large language model
11 BLUR: A Benchmark for LLM Unlearning Robust to Forget-Retain Overlap BLUR：一个针对LLM非学习的基准测试，对遗忘-保留重叠具有鲁棒性 large language model ✅
12 Highly Efficient and Effective LLMs with Multi-Boolean Architectures 提出基于多核布尔架构的高效LLM微调方法，无需全精度潜在权重。 large language model
13 Navigating the Latent Space Dynamics of Neural Models 提出基于隐空间动力系统的神经网络分析方法，用于分析模型泛化能力和提取先验知识。 foundation model
14 Multivariate de Bruijn Graphs: A Symbolic Graph Framework for Time Series Forecasting 提出DRAGON，利用多变量de Bruijn图解决时间序列预测中符号结构缺失问题 foundation model ✅
15 FlashFormer: Whole-Model Kernels for Efficient Low-Batch Inference FlashFormer：用于高效低批量推理的全模型融合Kernel large language model
16 Update Your Transformer to the Latest Release: Re-Basin of Task Vectors 提出TransFusion，通过重构任务向量实现Transformer模型微调知识的无数据迁移。 foundation model ✅
17 Adaptive Budget Allocation for Orthogonal-Subspace Adapter Tuning in LLMs Continual Learning 提出OA-Adapter以解决LLMs持续学习中的预算分配问题 large language model
18 MoRE: A Mixture of Low-Rank Experts for Adaptive Multi-Task Learning 提出MoRE：一种低秩专家混合模型，用于自适应多任务学习。 large language model
19 Detecting Undesired Process Behavior by Means of Retrieval Augmented Generation 提出基于检索增强生成(RAG)的方法，无需微调即可检测流程中不期望的行为。 large language model
20 ACE: Exploring Activation Cosine Similarity and Variance for Accurate and Calibration-Efficient LLM Pruning ACE：探索激活余弦相似性和方差，实现LLM精确高效的校准剪枝 large language model
21 LLMs Judging LLMs: A Simplex Perspective 提出几何贝叶斯方法以评估大型语言模型的输出质量 large language model
22 FALCON: An ML Framework for Fully Automated Layout-Constrained Analog Circuit Design FALCON：全自动布局约束模拟电路设计的机器学习框架 foundation model

🔬 支柱二：RL算法与架构 (RL & Architecture) (20 篇)

#题目一句话要点标签🔗⭐
23 LLM-ODDR: A Large Language Model Framework for Joint Order Dispatching and Driver Repositioning 提出LLM-ODDR框架，利用大语言模型解决网约车订单分配与司机调度联合优化问题 reinforcement learning spatiotemporal large language model
24 A Closer Look at Multimodal Representation Collapse 揭示多模态表征坍塌机理，提出显式基向量重分配算法以提升多模态融合性能。 distillation multimodal ✅
25 Scaling Offline RL via Efficient and Expressive Shortcut Models 提出SORL算法，利用高效且富有表现力的捷径模型扩展离线强化学习。 reinforcement learning offline RL offline reinforcement learning
26 SOReL and TOReL: Two Methods for Fully Offline Reinforcement Learning 提出SOReL和TOReL，解决离线强化学习中超参数调优和性能评估难题。 reinforcement learning offline RL offline reinforcement learning ✅
27 Estimating the Effects of Sample Training Orders for Large Language Models without Retraining 提出一种免重训练框架，用于评估大语言模型训练样本顺序的影响 curriculum learning large language model
28 Preference Learning with Response Time: Robust Losses and Guarantees 提出基于响应时间的偏好学习方法，提升奖励模型学习的样本效率与理论保证。 preference learning foundation model
29 Skywork Open Reasoner 1 Technical Report Skywork-OR1：通过强化学习提升长CoT模型推理能力，显著超越同规模模型。 reinforcement learning large language model chain-of-thought
30 Reinforcement Learning for Out-of-Distribution Reasoning in LLMs: An Empirical Study on Diagnosis-Related Group Coding 提出DRG-Sapphire，利用强化学习解决LLM在DRG编码中的分布外推理难题。 reinforcement learning large language model
31 SDPO: Importance-Sampled Direct Preference Optimization for Stable Diffusion Training 提出SDPO以解决扩散模型训练中的偏差与不稳定问题 preference learning DPO direct preference optimization
32 Scaling Reasoning without Attention 提出无注意力语言模型以解决推理效率低下问题 Mamba state space model large language model
33 Two-Stage Feature Generation with Transformer and Reinforcement Learning 提出基于Transformer和强化学习的两阶段特征生成框架，提升模型性能和适应性。 reinforcement learning PPO
34 A Provable Approach for End-to-End Safe Reinforcement Learning 提出PLS：一种可证明的端到端安全强化学习方法，确保学习和部署全过程的安全性。 reinforcement learning
35 Contraction Actor-Critic: Contraction Metric-Guided Reinforcement Learning for Robust Path Tracking 提出Contraction Actor-Critic算法，用于未知动力学下的鲁棒路径跟踪。 reinforcement learning
36 The Entropy Mechanism of Reinforcement Learning for Reasoning Language Models 针对推理语言模型，提出基于熵机制的强化学习方法，解决策略熵坍塌问题。 reinforcement learning
37 Physics-Informed Distillation of Diffusion Models for PDE-Constrained Generation 提出物理信息蒸馏方法以解决扩散模型中的PDE约束问题 distillation
38 When Does Neuroevolution Outcompete Reinforcement Learning in Transfer Learning Tasks? 探讨神经进化在迁移学习任务中超越强化学习的能力 reinforcement learning
39 An Augmentation-Aware Theory for Self-Supervised Contrastive Learning 提出一种数据增强感知的自监督对比学习理论框架，显式建模数据增强的影响。 contrastive learning
40 Weakly-Supervised Contrastive Learning for Imprecise Class Labels 提出基于图的弱监督对比学习框架，解决标签不准确情况下的表征学习问题 contrastive learning ✅
41 FNOPE: Simulation-based inference on function spaces with Fourier Neural Operators FNOPE：利用傅里叶神经算子进行函数空间上的模拟推断，提升时空过程建模效率。 flow matching spatiotemporal
42 Revisiting Group Relative Policy Optimization: Insights into On-Policy and Off-Policy Training 改进群体相对策略优化：探索其在On-Policy和Off-Policy训练中的应用 reinforcement learning PPO

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

#题目一句话要点标签🔗⭐
43 Forecasting Multivariate Urban Data via Decomposition and Spatio-Temporal Graph Analysis 提出DST模型，通过分解和时空图分析预测多元城市数据 spatiotemporal

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
44 Practical Adversarial Attacks on Stochastic Bandits via Fake Data Injection 提出基于伪数据注入的随机Bandit算法对抗攻击方法 manipulation

⬅️ 返回 cs.LG 首页 · 🏠 返回主页