cs.LG（2025-12-30）

📊 共 19 篇论文 | 🔗 2 篇有代码

🎯 兴趣领域导航

支柱二：RL算法与架构 (RL & Architecture) (10 🔗1) 支柱九：具身大模型 (Embodied Foundation Models) (6 🔗1) 支柱一：机器人控制 (Robot Control) (2) 支柱八：物理动画 (Physics-based Animation) (1)

🔬 支柱二：RL算法与架构 (RL & Architecture) (10 篇)

#题目一句话要点标签🔗⭐
1 Universal Battery Degradation Forecasting Driven by Foundation Model Across Diverse Chemistries and Conditions 提出统一电池衰退预测框架以解决多化学成分挑战 representation learning foundation model
2 Efficient Inference for Inverse Reinforcement Learning and Dynamic Discrete Choice Models 提出一种半参数逆强化学习框架，实现高效且有统计保证的奖励函数推断。 reinforcement learning inverse reinforcement learning
3 SmartFlow Reinforcement Learning and Agentic AI for Bike-Sharing Optimisation SmartFlow：融合强化学习与Agentic AI优化共享单车动态再平衡 reinforcement learning large language model
4 GRADE: Replacing Policy Gradients with Backpropagation for LLM Alignment GRADE：用反向传播替代策略梯度，实现LLM对齐 reinforcement learning PPO RLHF
5 Lifting Vision: Ground to Aerial Localization with Reasoning Guided Planning 提出ViReLoc框架，利用视觉推理进行地面到空中定位与规划 contrastive learning multimodal
6 How and Why LLMs Generalize: A Fine-Grained Analysis of LLM Reasoning from Cognitive Behaviors to Low-Level Patterns 提出细粒度LLM推理基准，揭示SFT与RL微调泛化能力差异的深层原因 reinforcement learning large language model
7 Hyperspherical Graph Representation Learning via Adaptive Neighbor-Mean Alignment and Uniformity HyperGRL：基于超球面表示学习的图神经网络统一框架 representation learning
8 Implicit geometric regularization in flow matching via density weighted Stein operators 提出γ-Flow Matching，通过密度加权Stein算子实现Flow Matching的几何正则化。 flow matching
9 Physics-informed Graph Neural Networks for Operational Flood Modeling 提出DUALFloodGNN，融合物理信息的图神经网络用于快速洪水模拟 curriculum learning spatiotemporal ✅
10 Stationary Reweighting Yields Local Convergence of Soft Fitted Q-Iteration 提出基于平稳重加权的Soft FQI算法，解决离线强化学习中的局部收敛问题 reinforcement learning offline reinforcement learning

🔬 支柱九：具身大模型 (Embodied Foundation Models) (6 篇)

#题目一句话要点标签🔗⭐
11 HOLOGRAPH: Active Causal Discovery via Sheaf-Theoretic Alignment of Large Language Model Priors HOLOGRAPH：通过层论对齐大语言模型先验知识进行主动因果发现 large language model ✅
12 LLMize: A Framework for Large Language Model-Based Numerical Optimization LLMize：基于大语言模型的数值优化开源框架，简化复杂约束问题求解。 large language model
13 Harvesting AlphaEarth: Benchmarking the Geospatial Foundation Model for Agricultural Downstream Tasks 评估AlphaEarth GFM在农业下游任务中的性能，并与传统遥感模型对比 foundation model
14 Towards mechanistic understanding in a data-driven weather model: internal activations reveal interpretable physical features 利用稀疏自编码器，揭示GraphCast内部激活中可解释的物理特征 large language model
15 OptRot: Mitigating Weight Outliers via Data-Free Rotations for Post-Training Quantization OptRot：通过免数据旋转缓解权重异常值，用于训练后量化 large language model
16 RepetitionCurse: Measuring and Understanding Router Imbalance in Mixture-of-Experts LLMs under DoS Stress RepetitionCurse：揭示并利用MoE LLM在DoS攻击下的路由失衡问题 large language model

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
17 MaRCA: Multi-Agent Reinforcement Learning for Dynamic Computation Allocation in Large-Scale Recommender Systems MaRCA：基于多智能体强化学习的大规模推荐系统动态计算资源分配 MPC model predictive control reinforcement learning
18 Tensor Computing Interface: An Application-Oriented, Lightweight Interface for Portable High-Performance Tensor Network Applications 提出TCI：一种面向应用、轻量级的张量计算接口，提升张量网络应用的可移植性。 manipulation

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

#题目一句话要点标签🔗⭐
19 OptiVote: Non-Coherent FSO Over-the-Air Majority Vote for Communication-Efficient Distributed Federated Learning in Space Data Centers OptiVote：面向空间数据中心，通信高效的非相干FSO空中多数投票联邦学习 PULSE

⬅️ 返回 cs.LG 首页 · 🏠 返回主页