One important property of DIstribution Correction Estimation (DICE) methods is that the solution is the optimal stationary distribution ratio between the optimized and data collection policy. In this work, we show that DICE-based methods can be viewed as a transformation from the behavior distribution to the optimal policy distribution. Based on this, we propose a novel approach, Diffusion-DICE, that directly performs this transformation using diffusion models. We find that the optimal policy's score function can be decomposed into two terms: the behavior policy's score function and the gradient of a guidance term which depends on the optimal distribution ratio. The first term can be obtained from a diffusion model trained on the dataset and we propose an in-sample learning objective to learn the second term. Due to the multi-modality contained in the optimal policy distribution, the transformation in Diffusion-DICE may guide towards those local-optimal modes. We thus generate a few candidate actions and carefully select from them to approach global-optimum. Different from all other diffusion-based offline RL methods, the guide-then-select paradigm in Diffusion-DICE only uses in-sample actions for training and brings minimal error exploitation in the value function. We use a didatic toycase example to show how previous diffusion-based methods fail to generate optimal actions due to leveraging these errors and how Diffusion-DICE successfully avoids that. We then conduct extensive experiments on benchmark datasets to show the strong performance of Diffusion-DICE. Project page at https://ryanxhr.github.io/Diffusion-DICE/.
A major challenge in real-world reinforcement learning (RL) is the sparsity of reward feedback. Often, what is available is an intuitive but sparse reward function that only indicates whether the task is completed partially or fully. However, the lack of carefully designed, fine grain feedback implies that most existing RL algorithms fail to learn an acceptable policy in a reasonable time frame. This is because of the large number of exploration actions that the policy has to perform before it gets any useful feedback that it can learn from. In this work, we address this challenging problem by…
Diffusion Self-Weighted Guidance for Offline Reinforcement Learning
Initiative for Data & AI, Universidad de Chile（数据与人工智能倡议，智利大学）；AMTC & Dept. of Electrical Eng., Universidad de Chile（AMTC与电气工程系，智利大学）；Department of Mathematics, Imperial College London（数学系，伦敦帝国学院）
FlowQ: Energy-Guided Flow Policies for Offline Reinforcement Learning
Machine Learning Research Lab, Volkswagen Group（大众集团机器学习研究实验室）；Technical University of Munich（慕尼黑技术大学）；Foundation Robotics Labs, Munich（慕尼黑基础机器人实验室）；Eötvös Loránd University（劳尔·艾尔多迪大学）
King Abdullah University of Science and Technology（阿卜杜拉国王科技大学）；The Swiss AI Lab IDSIA/USI/SUPSI（瑞士人工智能实验室 IDSIA/USI/SUPSI）；NNAISENSE
Offline RL methods have been shown to reduce the need for environment interaction by training agents using offline collected episodes. However, these methods typically require action information to be logged during data collection, which can be difficult or even impossible in some practical cases. In this paper, we investigate the potential of using action-free offline datasets to improve online reinforcement learning, name this problem Reinforcement Learning with Action-Free Offline Pretraining (AFP-RL). We introduce Action-Free Guide (AF-Guide), a method that guides online training by extrac…
Wuhan University（武汉大学）；JD Explore Academy（京东探索研究院）；University of Maryland, College Park（马里兰大学学院市分校）
Deep reinforcement learning (RL) has shown remarkable success in specific offline decision-making scenarios, yet its theoretical guarantees are still under development. Existing works on offline RL theory primarily emphasize a few trivial settings, such as linear MDP or general function approximation with strong assumptions and independent data, which lack guidance for practical use. The coupling of deep learning and Bellman residuals makes this problem challenging, in addition to the difficulty of data dependence. In this paper, we establish a non-asymptotic estimation error of pessimistic of…
OGER: A Robust Offline-Guided Exploration Reward for Hybrid Reinforcement Learning
Institute of Computing and Intelligence, Harbin Institute of Technology, Shenzhen, China（哈尔滨工业大学深圳研究院）；Hithink RoyalFlush Information Network, Hangzhou, China（杭州Hithink RoyalFlush信息网络）；Computer and Information Science, University of Macau, Macau, China（澳门大学计算机与信息科学学院）
JEF-Hinter: Leveraging Offline Knowledge for Improving Web Agents Adaptation
ServiceNow AI Research（ServiceNow AI研究机构）；Chandar Research Lab（Chandar研究实验室）；Mila -- Quebec AI Institute（魁北克人工智能研究院）；Dalhousie University（达尔豪斯大学）；Canada CIFAR AI Chair（加拿大CIFAR人工智能主席）
Offline reinforcement learning (RL) enables effective learning from previously collected data without exploration, which shows great promise in real-world applications when exploration is expensive or even infeasible. The discount factor, $γ$, plays a vital role in improving online RL sample efficiency and estimation accuracy, but the role of the discount factor in offline RL is not well explored. This paper examines two distinct effects of $γ$ in offline RL with theoretical analysis, namely the regularization effect and the pessimism effect. On the one hand, $γ$ is a regulator to trade-off op…
Boosting Continuous Control with Consistency Policy
Institute of Automation, Chinese Academy of Sciences（中国科学院自动化研究所）；School of Artificial Intelligence, University of Chinese Academy of Sciences（中国科学院大学人工智能学院）
本文提出 CPQL，借助一致性模型将反向扩散轨迹映射为单步动作策略，并用 Q-Learning 实现准确策略改进，在离线与在线连续控制任务上达到 SOTA 且推理较 Diffusion-QL 提速近 45 倍。
Learning Diverse Policies with Soft Self-Generated Guidance
School of Mathematical Sciences, Beihang University（北京航空航天大学数学科学学院）；Institute of Artificial Intelligence, Beihang University（北京航空航天大学人工智能研究院）；Zhongguancun Laboratory（中关村实验室）
Diffusion Actor-Critic: Formulating Constrained Policy Iteration as Diffusion Noise Regression for Offline Reinforcement Learning
Hong Kong University of Science and Technology（香港科技大学）；Hong Kong University of Science and Technology (Guangzhou)（香港科技大学（广州））；Southern University of Science and Technology（南方科技大学）
FOSP: Fine-tuning Offline Safe Policy through World Models
Shenzhen International Graduate School, Tsinghua University（清华大学深圳国际研究生院）；College of Design and Engineering, National University of Singapore（新加坡国立大学设计与工程学院）
MAGE: Multi-scale Autoregressive Generation for Offline Reinforcement Learning
Xiamen University（厦门大学）；Hong Kong University of Science and Technology（香港科技大学）；National University of Defense Technology（国防科技大学）；Jimei University（集美大学）
