cs.LG（2025-07-04）

📊 共 18 篇论文 | 🔗 1 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (10) 支柱二：RL算法与架构 (RL & Architecture) (5 🔗1) 支柱八：物理动画 (Physics-based Animation) (2) 支柱一：机器人控制 (Robot Control) (1)

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

#题目一句话要点标签🔗⭐
1 Conformal Information Pursuit for Interactively Guiding Large Language Models 提出C-IP方法以优化大语言模型的交互式问答性能 large language model chain-of-thought
2 MGAA: Multi-Granular Adaptive Allocation fof Low-Rank Compression of LLMs 提出MGAA：一种多粒度自适应分配方法，用于LLM的低秩压缩。 large language model multimodal
3 Importance-Aware Activation Space Reconstruction 提出IMPACT：一种重要性感知的激活空间重构方法，用于压缩大语言模型。 large language model
4 Skewed Score: A statistical framework to assess autograders 提出Skewed Score框架，用于统计评估LLM自动评分器并检测潜在偏差。 large language model
5 Predicting Business Angel Early-Stage Decision Making Using AI 利用AI预测商业天使的早期投资决策，提升效率与准确性 large language model
6 Graph Neural Networks for Electricity Load Forecasting 提出结合图神经网络、注意力机制和集成策略的电力负荷预测框架 foundation model
7 Re-Emergent Misalignment: How Narrow Fine-Tuning Erodes Safety Alignment in LLMs 研究表明，窄域微调通过侵蚀先验对齐导致LLM安全性下降 large language model
8 Generating Synthetic Relational Tabular Data via Structural Causal Models 提出基于结构因果模型的框架，用于生成合成关系型表格数据。 foundation model
9 LRM-1B: Towards Large Routing Model 提出大规模路由模型LRM-1B，解决车辆路径问题并达到SOTA large language model
10 Securing Transformer-based AI Execution via Unified TEEs and Crypto-protected Accelerators TwinShield：通过统一TEE和加密加速器保护Transformer模型安全执行 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (5 篇)

#题目一句话要点标签🔗⭐
11 Action Robust Reinforcement Learning via Optimal Adversary Aware Policy Optimization 提出OA-PI框架，增强强化学习策略在动作扰动下的鲁棒性 reinforcement learning DRL PPO
12 Reinforcement Learning-based Feature Generation Algorithm for Scientific Data 提出基于强化学习的多智能体特征生成框架，自动化提升科学数据挖掘任务性能。 reinforcement learning large language model
13 ObjectRL: An Object-Oriented Reinforcement Learning Codebase ObjectRL：一个面向对象深度强化学习研究的代码库 reinforcement learning deep reinforcement learning ✅
14 Adaptive Gate-Aware Mamba Networks for Magnetic Resonance Fingerprinting 提出GAST-Mamba网络，用于加速高欠采样磁共振指纹图谱重建。 Mamba
15 Degrees of Freedom for Linear Attention: Distilling Softmax Attention with Optimal Feature Efficiency 提出基于自由度的线性注意力蒸馏方法，优化特征维度选择。 linear attention

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

#题目一句话要点标签🔗⭐
16 Do Tensorized Large-Scale Spatiotemporal Dynamic Atmospheric Data Exhibit Low-Rank Properties? 利用低秩张量模型补全Sentinel-5P大气NO2时空数据缺失值 spatiotemporal
17 UWB TDoA Error Correction using Transformers: Patching and Positional Encoding Strategies 提出基于Transformer的UWB TDoA误差校正方法，提升复杂NLOS环境下的定位精度。 PULSE

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
18 Dyn-O: Building Structured World Models with Object-Centric Representations Dyn-O：构建基于对象中心表示的结构化世界模型，提升复杂场景泛化能力 manipulation world model dreamer

⬅️ 返回 cs.LG 首页 · 🏠 返回主页