cs.LG（2024-10-21）

📊 共 32 篇论文 | 🔗 4 篇有代码

🎯 兴趣领域导航

支柱二：RL算法与架构 (RL & Architecture) (14 🔗1) 支柱九：具身大模型 (Embodied Foundation Models) (14 🔗3) 支柱八：物理动画 (Physics-based Animation) (4)

🔬 支柱二：RL算法与架构 (RL & Architecture) (14 篇)

#题目一句话要点标签🔗⭐
1 Promoting cross-modal representations to improve multimodal foundation models for physiological signals 提出基于跨模态表征增强的多模态生理信号预训练模型，提升医疗健康应用性能。 contrastive learning foundation model multimodal
2 Pruning Foundation Models for High Accuracy without Retraining 提出APT免训练剪枝算法，实现大模型高精度压缩与加速 Mamba large language model foundation model ✅
3 Offline reinforcement learning for job-shop scheduling problems 提出一种离线强化学习方法，用于解决Job-Shop调度问题。 reinforcement learning deep reinforcement learning offline RL
4 On The Global Convergence Of Online RLHF With Neural Parametrization 提出双层优化框架以解决RLHF中的分布偏移问题 reinforcement learning policy learning RLHF
5 Understanding and Alleviating Memory Consumption in RLHF for LLMs 针对LLM的RLHF微调，提出内存优化方法以降低资源消耗 reinforcement learning RLHF large language model
6 In-Trajectory Inverse Reinforcement Learning: Learn Incrementally Before An Ongoing Trajectory Terminates 提出在线轨迹逆强化学习以解决增量学习问题 reinforcement learning inverse reinforcement learning
7 Solving Continual Offline RL through Selective Weights Activation on Aligned Spaces 提出VQ-CD，通过对齐空间上的选择性权重激活解决持续离线强化学习问题 reinforcement learning offline RL offline reinforcement learning
8 Do Audio-Language Models Understand Linguistic Variations? 提出RobustCLAP，增强音频语言模型对文本查询中语言变体的泛化能力 contrastive learning open-vocabulary open vocabulary
9 A Plug-and-Play Fully On-the-Job Real-Time Reinforcement Learning Algorithm for a Direct-Drive Tandem-Wing Experiment Platforms Under Multiple Random Operating Conditions 针对串联翼飞行器，提出一种即插即用、全流程实时强化学习算法CRL2E，解决多重随机工况下的运动控制难题。 reinforcement learning
10 RGMDT: Return-Gap-Minimizing Decision Tree Extraction in Non-Euclidean Metric Space RGMDT：非欧度量空间中基于回报差距最小化的决策树提取方法 reinforcement learning deep reinforcement learning DRL
11 Modeling Structured Data Learning with Restricted Boltzmann Machines in the Teacher-Student Setting 研究受限玻尔兹曼机在师生框架下的结构化数据学习能力 teacher-student
12 Information-Theoretic Minimax Regret Bounds for Reinforcement Learning based on Duality 基于对偶性的强化学习信息论Minimax遗憾界 reinforcement learning
13 Diverse Policies Recovering via Pointwise Mutual Information Weighted Imitation Learning 提出基于点互信息加权的模仿学习方法，用于恢复多样化策略。 imitation learning
14 Model Mimic Attack: Knowledge Distillation for Provably Transferable Adversarial Examples 提出基于知识蒸馏的模型模仿攻击，提升黑盒对抗样本的可迁移性并提供理论保证 distillation

🔬 支柱九：具身大模型 (Embodied Foundation Models) (14 篇)

#题目一句话要点标签🔗⭐
15 LiMTR: Time Series Motion Prediction for Diverse Road Users through Multimodal Feature Integration LiMTR：通过多模态特征融合预测道路用户时序运动轨迹 foundation model multimodal
16 Generative Design of Functional Metal Complexes Utilizing the Internal Knowledge of Large Language Models 提出LLM-EO框架，利用大语言模型内部知识进行功能性金属配合物生成设计。 large language model
17 Large Language Models in Computer Science Education: A Systematic Literature Review 系统性文献综述：大型语言模型在计算机科学教育中的应用与影响 large language model
18 An Interpretable N-gram Perplexity Threat Model for Large Language Model Jailbreaks 提出一种可解释的N-gram困惑度威胁模型以评估LLM越狱攻击 large language model
19 SeisLM: a Foundation Model for Seismic Waveforms SeisLM：用于地震波形分析的预训练基础模型 foundation model ✅
20 Residual vector quantization for KV cache compression in large language model 提出残差向量量化方法，用于压缩大语言模型中的KV缓存，实现高效推理。 large language model
21 SoftSRV: Learn to Generate Targeted Synthetic Data SoftSRV：学习生成目标合成数据，提升特定任务模型性能 large language model
22 Insights and Current Gaps in Open-Source LLM Vulnerability Scanners: A Comparative Analysis 对比分析开源LLM漏洞扫描器，揭示其局限性并提供改进方向 large language model
23 LLM-TS Integrator: Integrating LLM for Enhanced Time Series Modeling LLM-TS Integrator：融合LLM增强时间序列建模，提升预测、插补、分类和异常检测性能 large language model
24 In Search of the Successful Interpolation: On the Role of Sharpness in CLIP Generalization 探讨尖锐度在CLIP模型插值成功中的作用 foundation model ✅
25 Beyond 2:4: exploring V:N:M sparsity for efficient transformer inference on GPUs 探索V:N:M稀疏性，加速GPU上Transformer模型推理，实现精度与效率的平衡。 large language model
26 CartesianMoE: Boosting Knowledge Sharing among Experts via Cartesian Product Routing in Mixture-of-Experts 提出 CartesianMoE，通过笛卡尔积路由增强MoE模型中专家间的知识共享。 large language model
27 Mesa-Extrapolation: A Weave Position Encoding Method for Enhanced Extrapolation in LLMs 提出Mesa-Extrapolation，一种用于LLM的编织位置编码方法，增强外推能力。 large language model ✅
28 Bayesian Concept Bottleneck Models with LLM Priors 提出BC-LLM，利用大语言模型先验的贝叶斯概念瓶颈模型，提升可解释性和鲁棒性。 large language model

🔬 支柱八：物理动画 (Physics-based Animation) (4 篇)

#题目一句话要点标签🔗⭐
29 Extracting Spatiotemporal Data from Gradients with Large Language Models 提出ST-GIA及ST-GIA+，提升时空联邦学习梯度泄露攻击效果并设计自适应防御策略 spatiotemporal large language model
30 Building Conformal Prediction Intervals with Approximate Message Passing 提出基于近似消息传递的共形预测区间构建方法，加速高维广义线性回归的不确定性量化。 AMP
31 Exploring how deep learning decodes anomalous diffusion via Grad-CAM 利用Grad-CAM解释深度学习在反常扩散机制识别中的作用，并提升模型鲁棒性。 spatiotemporal
32 State Estimation Using Sparse DEIM and Recurrent Neural Networks 提出基于RNN的无方程S-DEIM方法，用于动态系统状态估计，无需系统方程知识。 spatiotemporal

⬅️ 返回 cs.LG 首页 · 🏠 返回主页