cs.LG(2024-02-14)

📊 共 24 篇论文 | 🔗 3 篇有代码

🎯 兴趣领域导航

支柱二:RL算法与架构 (RL & Architecture) (11 🔗2) 支柱九:具身大模型 (Embodied Foundation Models) (8 🔗1) 支柱一:机器人控制 (Robot Control) (3) 支柱五:交互与反应 (Interaction & Reaction) (1) 支柱八:物理动画 (Physics-based Animation) (1)

🔬 支柱二:RL算法与架构 (RL & Architecture) (11 篇)

#题目一句话要点标签🔗
1 Learning Interpretable Concepts: Unifying Causal Representation Learning and Foundation Models 提出统一因果表示学习与基础模型的方法以学习可解释概念 representation learning large language model foundation model
2 Reinforcement Learning from Human Feedback with Active Queries 提出基于主动查询的强化学习方法以提高人类反馈效率 reinforcement learning RLHF DPO
3 Exploiting Estimation Bias in Clipped Double Q-Learning for Continous Control Reinforcement Learning Tasks 提出偏差利用机制以解决连续控制强化学习中的估计偏差问题 reinforcement learning deep reinforcement learning policy learning
4 Graph Contrastive Learning with Low-Rank Regularization and Low-Rank Attention for Noisy Node Classification 提出GCL-LRR以解决图神经网络在噪声节点分类中的挑战 representation learning contrastive learning
5 InfoRM: Mitigating Reward Hacking in RLHF via Information-Theoretic Reward Modeling 提出InfoRM以解决强化学习中的奖励黑客问题 reinforcement learning RLHF
6 Learning Interpretable Policies in Hindsight-Observable POMDPs through Partially Supervised Reinforcement Learning 提出部分监督强化学习框架以提升POMDP中的可解释性 reinforcement learning deep reinforcement learning
7 Towards Robust Model-Based Reinforcement Learning Against Adversarial Corruption 提出CR-OMLE和CR-PMLE以解决模型基础强化学习中的对抗性腐蚀问题 reinforcement learning model-based RL
8 Dataset Clustering for Improved Offline Policy Learning 提出行为感知深度聚类以提升离线策略学习效果 policy learning
9 Active Disruption Avoidance and Trajectory Design for Tokamak Ramp-downs with Neural Differential Equations and Reinforcement Learning 提出基于强化学习的托卡马克等离子体安全降温策略 reinforcement learning
10 When Representations Align: Universality in Representation Learning Dynamics 提出有效理论以揭示表示学习动态的普遍性 representation learning
11 FedSiKD: Clients Similarity and Knowledge Distillation: Addressing Non-i.i.d. and Constraints in Federated Learning 提出FedSiKD以解决非独立同分布数据下的联邦学习问题 distillation

🔬 支柱九:具身大模型 (Embodied Foundation Models) (8 篇)

#题目一句话要点标签🔗
12 Embracing the black box: Heading towards foundation models for causal discovery from time series data 提出因果预训练方法以解决时间序列因果发现问题 foundation model
13 Attacking Large Language Models with Projected Gradient Descent 提出基于投影梯度下降的对抗攻击方法以提升LLM安全性 large language model
14 Get More with LESS: Synthesizing Recurrence with KV Cache Compression for Efficient LLM Inference 提出LESS以解决大语言模型推理中的KV缓存瓶颈问题 large language model
15 HiRE: High Recall Approximate Top-$k$ Estimation for Efficient LLM Inference 提出HiRE以解决LLM推理中的内存瓶颈问题 large language model
16 Trained Without My Consent: Detecting Code Inclusion In Language Models Trained on Code 提出TraWiC以解决代码审计中的版权问题 large language model
17 Instruction Backdoor Attacks Against Customized LLMs 提出指令后门攻击以解决定制LLM的安全问题 large language model
18 Leveraging the Context through Multi-Round Interactions for Jailbreaking Attacks 提出上下文多轮交互攻击以应对Jailbreaking攻击问题 large language model
19 Exploring Federated Deep Learning for Standardising Naming Conventions in Radiotherapy Data 提出联邦深度学习以标准化放射治疗数据命名规范 multimodal

🔬 支柱一:机器人控制 (Robot Control) (3 篇)

#题目一句话要点标签🔗
20 Under manipulations, are some AI models harder to audit? 研究模型容量与审计难度的关系 manipulation
21 Conformalized Adaptive Forecasting of Heterogeneous Trajectories 提出一种新型的保形自适应预测方法以解决异构轨迹预测问题 motion planning
22 Soft Prompt Threats: Attacking Safety Alignment and Unlearning in Open-Source LLMs through the Embedding Space 提出嵌入空间攻击以解决开源LLMs的安全对齐与遗忘问题 manipulation

🔬 支柱五:交互与反应 (Interaction & Reaction) (1 篇)

#题目一句话要点标签🔗
23 I can't see it but I can Fine-tune it: On Encrypted Fine-tuning of Transformers using Fully Homomorphic Encryption 提出BlindTuner以解决隐私保护下的Transformer微调问题 OMOMO

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

#题目一句话要点标签🔗
24 Deinterleaving of Discrete Renewal Process Mixtures with Application to Electronic Support Measures 提出新方法以解决离散更新过程混合的去交错问题 PULSE

⬅️ 返回 cs.LG 首页 · 🏠 返回主页