cs.LG（2025-02-19）

📊 共 26 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) (17 🔗3) 支柱二：RL算法与架构 (RL & Architecture) (8) 支柱六：视频提取与匹配 (Video Extraction) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (17 篇)

#题目一句话要点标签🔗⭐
1 Refining embeddings with fill-tuning: data-efficient generalised performance improvements for materials foundation models 提出Fill-tuning方法，通过数据高效地改进材料领域预训练模型泛化性能 foundation model
2 Smaller But Better: Unifying Layout Generation with Smaller Large Language Models 提出LGGPT，一种基于小型LLM的统一布局生成模型，在效率和性能间取得平衡。 large language model ✅
3 Are Large Language Models In-Context Graph Learners? 提出基于RAG的框架，提升大语言模型在图学习任务中的上下文学习能力 large language model
4 Train Small, Infer Large: Memory-Efficient LoRA Training for Large Language Models LoRAM：通过训练小模型、推理大模型，实现大语言模型的高效LoRA训练 large language model ✅
5 GeLLMO: Generalizing Large Language Models for Multi-property Molecule Optimization 提出GeLLMO：一种基于指令调优的大语言模型，用于多属性分子优化。 large language model ✅
6 Exploring Code Language Models for Automated HLS-based Hardware Generation: Benchmark, Infrastructure and Analysis 探索代码语言模型在自动化HLS硬件生成中的应用：基准、基础设施与分析 large language model chain-of-thought
7 Quantifying Memorization and Parametric Response Rates in Retrieval-Augmented Vision-Language Models 量化检索增强视觉-语言模型中的记忆与参数响应率，揭示模态差异。 large language model multimodal
8 Where's the Bug? Attention Probing for Scalable Fault Localization 提出Bug Attention Probe (BAP)，无需标注实现可扩展的缺陷定位。 large language model
9 SPEX: Scaling Feature Interaction Explanations for LLMs SPEX：扩展LLM特征交互解释，高效处理长输入 large language model
10 Evaluation of EAS directions based on TAIGA HiSCORE data using fully connected neural networks 利用全连接神经网络评估TAIGA HiSCORE数据中的EAS方向 multimodal
11 LESA: Learnable LLM Layer Scaling-Up 提出LESA以解决大规模语言模型训练成本高的问题 large language model
12 Which Attention Heads Matter for In-Context Learning? 揭示上下文学习的关键：功能向量头而非归纳头驱动LLM的ICL能力 large language model
13 Concept Layers: Enhancing Interpretability and Intervenability via LLM Conceptualization 提出Concept Layers，通过LLM概念化增强LLM的可解释性和可干预性 large language model
14 Unraveling the Localized Latents: Learning Stratified Manifold Structures in LLM Embedding Space with Sparse Mixture-of-Experts 提出基于稀疏混合专家模型的分析框架，揭示LLM嵌入空间的分层流形结构。 large language model
15 LSR-Adapt: Ultra-Efficient Parameter Tuning with Matrix Low Separation Rank Kernel Adaptation LSR-Adapt：利用矩阵低分离秩核自适应的超高效参数调优 large language model
16 Megrez-Omni Technical Report Megrez系列模型：软硬件协同设计，实现快速、紧凑、鲁棒的端侧智能 multimodal
17 An explainable transformer circuit for compositional generalization 揭示Transformer组合泛化能力：构建可解释的电路并实现模型行为精确控制 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (8 篇)

#题目一句话要点标签🔗⭐
18 Multi-Target Radar Search and Track Using Sequence-Capable Deep Reinforcement Learning 提出基于序列深度强化学习的多目标雷达搜索与跟踪方法 reinforcement learning deep reinforcement learning behavior cloning
19 Democratizing Large Language Model-Based Graph Data Augmentation via Latent Knowledge Graphs DemoGraph：利用潜在知识图谱，实现基于大语言模型的图数据增强 representation learning large language model
20 AS-GCL: Asymmetric Spectral Augmentation on Graph Contrastive Learning 提出AS-GCL，通过非对称谱增强提升图对比学习的泛化能力 contrastive learning ReMoS
21 Optimistically Optimistic Exploration for Provably Efficient Infinite-Horizon Reinforcement and Imitation Learning 提出高效算法解决无限期强化学习与模仿学习问题 reinforcement learning imitation learning
22 Dynamic Activation with Knowledge Distillation for Energy-Efficient Spiking NN Ensembles 提出基于知识蒸馏的动态激活Spiking神经网络集成，提升能效并保持精度。 distillation
23 Playing Hex and Counter Wargames using Reinforcement Learning and Recurrent Neural Networks 提出基于RNN和强化学习的AlphaZero算法，解决六角棋和兵棋推演的复杂决策问题 reinforcement learning
24 Contrastive Learning-Based privacy metrics in Tabular Synthetic Datasets 提出基于对比学习的隐私度量方法，用于评估表格合成数据集的隐私保护能力 contrastive learning
25 An Empirical Risk Minimization Approach for Offline Inverse RL and Dynamic Discrete Choice Model 提出基于经验风险最小化的离线逆强化学习方法 reinforcement learning inverse reinforcement learning

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

#题目一句话要点标签🔗⭐
26 InsightVision: A Comprehensive, Multi-Level Chinese-based Benchmark for Evaluating Implicit Visual Semantics in Large Vision Language Models 提出InsightVision，用于评估大型视觉语言模型对图像隐式语义的理解能力。 HuMoR multimodal

⬅️ 返回 cs.LG 首页 · 🏠 返回主页