cs.LG（2025-07-12）

📊 共 12 篇论文 | 🔗 3 篇有代码

🎯 兴趣领域导航

支柱二：RL算法与架构 (RL & Architecture) (5 🔗2) 支柱八：物理动画 (Physics-based Animation) (3 🔗1) 支柱九：具身大模型 (Embodied Foundation Models) (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) (5 篇)

#题目一句话要点标签🔗⭐
1 Deep Reinforcement Learning with Gradient Eligibility Traces 提出多步信用分配的深度强化学习方法以解决收敛性问题 reinforcement learning deep reinforcement learning policy learning ✅
2 Adversarial Activation Patching: A Framework for Detecting and Mitigating Emergent Deception in Safety-Aligned Transformers 提出对抗激活修补框架，用于检测和缓解安全对齐Transformer中的涌现欺骗行为 reinforcement learning RLHF large language model
3 A Generalization Theory for Zero-Shot Prediction 提出零样本预测的泛化理论框架，分析其学习目标与泛化能力 contrastive learning foundation model multimodal
4 Semi-Supervised Federated Learning via Dual Contrastive Learning and Soft Labeling for Intelligent Fault Diagnosis 提出SSFL-DCSL框架，通过双重对比学习和软标签解决智能故障诊断中半监督联邦学习问题。 representation learning contrastive learning
5 Fair CCA for Fair Representation Learning: An ADNI Study 提出公平典型相关分析(Fair CCA)方法，用于提升表征学习的公平性，应用于ADNI研究。 representation learning ✅

🔬 支柱八：物理动画 (Physics-based Animation) (3 篇)

#题目一句话要点标签🔗⭐
6 Geometric Generative Modeling with Noise-Conditioned Graph Networks 提出噪声条件图网络，用于提升空间结构图的生成建模效果 spatiotemporal ✅
7 Controllable Patching for Compute-Adaptive Surrogate Modeling of Partial Differential Equations 提出可控Patching方法，实现偏微分方程代理模型计算自适应性 spatiotemporal
8 POIFormer: A Transformer-Based Framework for Accurate and Scalable Point-of-Interest Attribution 提出POIFormer，利用Transformer解决复杂场景下兴趣点归因难题 spatiotemporal

🔬 支柱九：具身大模型 (Embodied Foundation Models) (2 篇)

#题目一句话要点标签🔗⭐
9 Scaling Laws for Optimal Data Mixtures 提出基于缩放定律的数据混合优化方法，提升大模型在目标领域的性能。 large language model foundation model multimodal
10 XiChen: An observation-scalable fully AI-driven global weather forecasting system with 4D variational knowledge XiChen：一个可扩展观测数据的全AI全球天气预报系统 foundation model

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
11 Continual Reinforcement Learning by Planning with Online World Models 通过在线世界模型规划解决持续强化学习中的遗忘问题 model predictive control reinforcement learning world model

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

#题目一句话要点标签🔗⭐
12 Temporal Misalignment Attacks against Multimodal Perception in Autonomous Driving 提出DejaVu攻击以解决自动驾驶多模态感知的时间对齐问题 scene understanding multimodal

⬅️ 返回 cs.LG 首页 · 🏠 返回主页