cs.LG（2026-02-17）

📊 共 17 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) (4 🔗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 On the Out-of-Distribution Generalization of Reasoning in Multimodal LLMs for Simple Visual Planning Tasks 评估多模态LLM在简单视觉规划任务中的推理泛化能力 large language model multimodal chain-of-thought
2 MRC-GAT: A Meta-Relational Copula-Based Graph Attention Network for Interpretable Multimodal Alzheimer's Disease Diagnosis 提出基于Meta关系Copula图注意力网络(MRC-GAT)，用于可解释的多模态阿尔茨海默病诊断。 multimodal
3 ER-MIA: Black-Box Adversarial Memory Injection Attacks on Long-Term Memory-Augmented Large Language Models 提出ER-MIA框架，针对长期记忆增强的大语言模型进行黑盒对抗性记忆注入攻击。 large language model
4 Discovering Implicit Large Language Model Alignment Objectives 提出Obj-Disco框架以解决LLM对齐目标不明确问题 large language model
5 Operationalising the Superficial Alignment Hypothesis via Task Complexity 通过任务复杂度量化，揭示大语言模型中的表层对齐假设 large language model instruction following
6 Neural Scaling Laws for Boosted Jet Tagging 研究喷注标记任务的神经标度律，揭示算力、数据与性能间的关系。 large language model foundation model
7 CrispEdit: Low-Curvature Projections for Scalable Non-Destructive LLM Editing 提出CrispEdit以解决大语言模型编辑中的能力保持问题 large language model
8 LLM-as-Judge on a Budget 提出基于多臂赌博机理论的LLM评估方法以优化查询分配 large language model
9 Prescriptive Scaling Reveals the Evolution of Language Model Capabilities 提出Prescriptive Scaling方法，揭示语言模型能力随算力演进规律，并评估其稳定性。 foundation model
10 On Surprising Effectiveness of Masking Updates in Adaptive Optimizers 提出Magma，通过掩码更新优化LLM训练，显著提升性能。 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (4 篇)

#题目一句话要点标签🔗⭐
11 Latency-aware Human-in-the-Loop Reinforcement Learning for Semantic Communications 提出时延感知的人在环强化学习框架，用于语义通信中的资源调度。 reinforcement learning PPO reward shaping
12 Solving Parameter-Robust Avoid Problems with Unknown Feasibility using Reinforcement Learning 提出可行性引导探索(FGE)算法，解决参数鲁棒避障问题中未知可行域的挑战。 reinforcement learning deep reinforcement learning
13 CDRL: A Reinforcement Learning Framework Inspired by Cerebellar Circuits and Dendritic Computational Strategies 提出CDRL：一种受小脑电路和树突计算策略启发的强化学习框架，提升样本效率和泛化能力。 reinforcement learning representation learning
14 GLM-5: from Vibe Coding to Agentic Engineering GLM-5：从Vibe Coding到Agentic Engineering的范式转变，提升软件工程能力 reinforcement learning foundation model ✅

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

#题目一句话要点标签🔗⭐
15 Guided Diffusion by Optimized Loss Functions on Relaxed Parameters for Inverse Material Design 提出基于优化损失函数和松弛参数的引导扩散方法，用于逆向材料设计。 differentiable simulation
16 Neural-POD: A Plug-and-Play Neural Operator Framework for Infinite-Dimensional Functional Nonlinear Proper Orthogonal Decomposition 提出Neural-POD，一种即插即用的神经算子框架，用于无限维函数非线性本征正交分解。 spatiotemporal

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
17 The Information Geometry of Softmax: Probing and Steering 利用信息几何探究Softmax表征，提出双重引导方法实现概念操控 manipulation

⬅️ 返回 cs.LG 首页 · 🏠 返回主页