cs.LG（2025-12-14）

📊 共 13 篇论文

🎯 兴趣领域导航

支柱二：RL算法与架构 (RL & Architecture) (9) 支柱八：物理动画 (Physics-based Animation) (2) 支柱九：具身大模型 (Embodied Foundation Models) (1) 支柱一：机器人控制 (Robot Control) (1)

🔬 支柱二：RL算法与架构 (RL & Architecture) (9 篇)

#题目一句话要点标签🔗⭐
1 The Laminar Flow Hypothesis: Detecting Jailbreaks via Semantic Turbulence in Large Language Models 提出层流假设，通过语义湍流检测大语言模型的越狱攻击 RLHF large language model instruction following
2 Error-Free Linear Attention is a Free Lunch: Exact Solution from Continuous-Time Dynamics 提出无误差线性注意力EFLA，通过连续时间动力学实现精确解，解决长文本建模的二次复杂度问题。 SSM state space model linear attention
3 Information-Consistent Language Model Recommendations through Group Relative Policy Optimization 提出基于GRPO的信息一致性语言模型推荐方法，解决企业场景下LLM推荐结果不一致问题。 reinforcement learning large language model
4 On the continuity of flows 研究表明Flow Matching在拓扑不匹配分布间可能产生速度场不连续性 flow matching multimodal
5 Flow matching Operators for Residual-Augmented Probabilistic Learning of Partial Differential Equations 提出残差增强概率学习的Flow Matching算子，解决数据稀缺下PDE学习难题。 flow matching
6 Self-Motivated Growing Neural Network for Adaptive Architecture via Local Structural Plasticity 提出自驱动增长神经网络SMGrNN，通过局部结构可塑性实现自适应架构 reinforcement learning deep reinforcement learning distillation
7 PIS: A Generalized Physical Inversion Solver for Arbitrary Sparse Observations via Set Conditioned Flow Matching PIS：基于集合条件流匹配的通用物理反演求解器，适用于任意稀疏观测 flow matching
8 Noise-robust Contrastive Learning for Critical Transition Detection in Dynamical Systems 提出基于SVD和半正交约束的噪声鲁棒对比学习方法，用于动态系统临界跃迁检测。 contrastive learning
9 COBRA: Catastrophic Bit-flip Reliability Analysis of State-Space Models COBRA：针对状态空间模型的灾难性位翻转可靠性分析框架 Mamba SSM

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

#题目一句话要点标签🔗⭐
10 Unsupervised learning of multiscale switching dynamical system models from multimodal neural data 提出一种无监督多尺度切换动态系统模型，用于融合多模态神经数据并解码行为。 spatiotemporal multimodal
11 DARTs: A Dual-Path Robust Framework for Anomaly Detection in High-Dimensional Multivariate Time Series 提出DARTs：一种双路径鲁棒框架，用于高维多元时间序列异常检测。 spatiotemporal

🔬 支柱九：具身大模型 (Embodied Foundation Models) (1 篇)

#题目一句话要点标签🔗⭐
12 Resting Neurons, Active Insights: Improving Input Sparsification for Large Language Models 引入可训练自发神经元，提升大语言模型输入稀疏化的性能 large language model

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
13 ceLLMate: Sandboxing Browser AI Agents ceLLMate：提出浏览器级沙箱框架，防御浏览器AI代理的提示注入攻击。 manipulation

⬅️ 返回 cs.LG 首页 · 🏠 返回主页