cs.LG（2025-12-19）

📊 共 19 篇论文

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (9) 支柱二：RL算法与架构 (RL & Architecture) (6) 支柱八：物理动画 (Physics-based Animation) (3) 支柱一：机器人控制 (Robot Control) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (9 篇)

#题目一句话要点标签🔗⭐
1 Wireless Traffic Prediction with Large Language Model 提出TIDES框架，利用大语言模型进行城市无线流量预测，显著提升预测精度和鲁棒性。 large language model foundation model
2 Towards Benchmarking Privacy Vulnerabilities in Selective Forgetting with Large Language Models 构建选择性遗忘隐私漏洞基准，评估大语言模型的隐私泄露风险 large language model
3 Sampling from multimodal distributions with warm starts: Non-asymptotic bounds for the Reweighted Annealed Leap-Point Sampler 提出Re-ALPS算法，加速多模态分布采样，无需高斯近似。 multimodal
4 Disentangling Fact from Sentiment: A Dynamic Conflict-Consensus Framework for Multimodal Fake News Detection 提出动态冲突-共识框架DCCF，用于增强多模态假新闻检测中矛盾信息的利用。 multimodal
5 Enabling Disaggregated Multi-Stage MLLM Inference via GPU-Internal Scheduling and Resource Sharing 提出FlashCodec和UnifiedServe，通过GPU内调度和资源共享加速多阶段MLLM推理。 large language model multimodal
6 Graph-based Nearest Neighbors with Dynamic Updates via Random Walks 提出基于随机游走的动态更新图最近邻搜索算法，支持高效删除操作 large language model
7 A Dataset and Benchmarks for Atrial Fibrillation Detection from Electrocardiograms of Intensive Care Unit Patients 发布ICU心电图房颤检测数据集与基准，验证ECG基础模型有效性 foundation model
8 Weighted Stochastic Differential Equation to Implement Wasserstein-Fisher-Rao Gradient Flow 提出基于加权随机微分方程的Wasserstein-Fisher-Rao梯度流方法，提升生成模型采样效率。 multimodal
9 Hierarchical Sparse Plus Low Rank Compression of LLM 提出分层稀疏加低秩压缩（HSS）方法，用于压缩LLM并保持性能。 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🔗⭐
10 Shuttling Compiler for Trapped-Ion Quantum Computers Based on Large Language Models 提出基于大语言模型的穿梭编译器，优化囚禁离子量子计算机的量子比特路由 DPO direct preference optimization large language model
11 Probabilistic Digital Twins of Users: Latent Representation Learning with Statistically Validated Semantics 提出基于概率数字孪生的用户建模框架，实现可解释的用户表征学习。 predictive model representation learning
12 Trust-Region Adaptive Policy Optimization 提出TRAPO框架，交错SFT与RL优化LLM推理能力，提升探索与稳定性。 reinforcement learning large language model
13 Assessing Long-Term Electricity Market Design for Ambitious Decarbonization Targets using Multi-Agent Reinforcement Learning 提出基于多智能体强化学习的电力市场长期设计评估框架，助力实现深度脱碳目标。 reinforcement learning
14 AdvJudge-Zero: Binary Decision Flips in LLM-as-a-Judge via Adversarial Control Tokens AdvJudge-Zero：通过对抗控制令牌翻转LLM评判器的二元决策 RLHF DPO
15 A Theoretical Analysis of State Similarity Between Markov Decision Processes 提出广义双模拟度量以解决多马尔可夫决策过程间状态相似性问题 reinforcement learning representation learning

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

#题目一句话要点标签🔗⭐
16 MINPO: Memory-Informed Neural Pseudo-Operator to Resolve Nonlocal Spatiotemporal Dynamics 提出MINPO：一种记忆增强神经伪算子，用于求解非局部时空动力学问题 spatiotemporal
17 Perfect reconstruction of sparse signals using nonconvexity control and one-step RSB message passing 提出基于非凸性控制和一步RSB消息传递的稀疏信号完美重构方法 AMP
18 Learning solution operator of dynamical systems with diffusion maps kernel ridge regression 提出基于扩散映射核岭回归的动力系统解算子学习方法，实现长期预测。 spatiotemporal

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
19 Biosecurity-Aware AI: Agentic Risk Auditing of Soft Prompt Attacks on ESM-Based Variant Predictors 提出SAGE框架，用于评估ESM等基因组模型在软提示攻击下的安全性。 manipulation large language model foundation model

⬅️ 返回 cs.LG 首页 · 🏠 返回主页