cs.LG（2025-09-14）

📊 共 18 篇论文 | 🔗 3 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (10 🔗3) 支柱二：RL算法与架构 (RL & Architecture) (6) 支柱七：动作重定向 (Motion Retargeting) (2)

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

#题目一句话要点标签🔗⭐
1 Intelligent Reservoir Decision Support: An Integrated Framework Combining Large Language Models, Advanced Prompt Engineering, and Multimodal Data Fusion for Real-Time Petroleum Operations 提出融合大语言模型、提示工程和多模态数据的智能油藏决策支持框架，提升石油作业效率。 large language model multimodal chain-of-thought
2 MatQnA: A Benchmark Dataset for Multi-modal Large Language Models in Materials Characterization and Analysis 提出MatQnA：用于材料表征与分析的多模态大语言模型基准数据集 large language model ✅
3 Agentic Username Suggestion and Multimodal Gender Detection in Online Platforms: Introducing the PNGT-26K Dataset 提出PNGT-26K波斯语姓名数据集，用于提升在线平台性别检测和用户名推荐。 multimodal
4 Harnessing Optimization Dynamics for Curvature-Informed Model Merging 提出OTA+FFG，利用优化动态信息进行曲率感知的模型合并，提升SFT模型性能。 large language model instruction following ✅
5 Decoding Musical Origins: Distinguishing Human and AI Composers 提出YNote音乐表示法，并构建分类模型以区分人类和AI作曲 large language model
6 From PowerSGD to PowerSGD+: Low-Rank Gradient Compression for Distributed Optimization with Convergence Guarantees 提出PowerSGD+算法，解决低秩梯度压缩分布式优化收敛性问题 large language model
7 From Parameters to Performance: A Data-Driven Study on LLM Structure and Development 构建大规模LLM结构-性能数据集，揭示模型结构对性能的影响。 large language model ✅
8 Predictable Compression Failures: Why Language Models Actually Hallucinate 提出可预测的压缩失败以解决语言模型幻觉问题 large language model
9 AQUA: Attention via QUery mAgnitudes for Memory and Compute Efficient Inference in LLMs AQUA：通过查询幅度注意力机制，提升LLM推理的内存和计算效率 large language model
10 Self-Evolving LLMs via Continual Instruction Tuning 提出MoE-CL框架，通过持续指令调优实现LLM在工业场景下的自进化。 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (6 篇)

#题目一句话要点标签🔗⭐
11 PersonaX: Multimodal Datasets with LLM-Inferred Behavior Traits PersonaX：构建包含LLM推断行为特征的多模态数据集，促进行为分析与因果推理。 representation learning large language model multimodal
12 Gradient Free Deep Reinforcement Learning With TabPFN 提出TabPFN RL，一种利用预训练Transformer进行免梯度深度强化学习的框架。 reinforcement learning deep reinforcement learning
13 Opal: An Operator Algebra View of RLHF Opal：提出RLHF的算子代数视角，并构建通用表示框架GKPO reinforcement learning RLHF DPO
14 Contrastive Network Representation Learning 提出ACERL，用于解决高维稀疏网络边表示学习问题，尤其适用于脑连接数据分析。 representation learning contrastive learning
15 Learning to Optimize Multi-Objective Alignment Through Dynamic Reward Weighting 提出动态奖励权重调整方法，优化多目标对齐问题 reinforcement learning large language model
16 Tabular Data with Class Imbalance: Predicting Electric Vehicle Crash Severity with Pretrained Transformers (TabPFN) and Mamba-Based Models 利用预训练Transformer和Mamba模型预测电动汽车碰撞严重程度 Mamba

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

#题目一句话要点标签🔗⭐
17 BIGNet: Pretrained Graph Neural Network for Embedding Semantic, Spatial, and Topological Data in BIM Models 提出BIGNet，用于BIM模型中语义、空间和拓扑数据的图神经网络预训练。 spatial relationship foundation model
18 GCN-TULHOR: Trajectory-User Linking Leveraging GCNs and Higher-Order Spatial Representations 提出GCN-TULHOR以解决轨迹用户关联问题 spatial relationship TAMP

⬅️ 返回 cs.LG 首页 · 🏠 返回主页