Correcting Data Distribution Mismatch in Offline Meta-Reinforcement Learning with Few-Shot Online Adaptation

Jianhao Wang, Jin Zhang, Haozhe Jiang, Junyu Zhang, Liwei Wang, Chongjie Zhang

Keywords: offline meta reinforcement learning, offline reinforcement learning, meta-reinforcement learning, few-shot online adaptation, data distribution mismatch correction

Abstract: Offline meta-reinforcement learning (offline meta-RL) extracts knowledge from a given dataset of multiple tasks and achieves fast adaptation to new tasks. Recent offline meta-RL methods typically use task-dependent behavior policies (e.g., training RL agents on each individual task) to collect a multi-task dataset and learn an offline meta-policy. However, these methods always require extra information for fast adaptation, such as offline context for testing tasks or oracle reward functions. Offline meta-RL with few-shot online adaptation remains an open problem. In this paper, we first formally characterize a unique challenge under this setting: data distribution mismatch between offline training and online adaptation. This distribution mismatch may lead to unreliable offline policy evaluation and the regular adaptation methods of online meta-RL will suffer. To address this challenge, we introduce a novel mechanism of data distribution correction, which ensures the consistency between offline and online evaluation by filtering out out-of-distribution episodes in online adaptation. As few-shot out-of-distribution episodes usually have lower returns, we propose a Greedy Context-based data distribution Correction approach, called GCC, which greedily infers how to solve new tasks. GCC diversely samples “task hypotheses” from the current posterior belief and selects a greedy hypothesis with the highest return to update the task belief. Our method is the first to provide an effective online adaptation without additional information, and can be combined with off-the-shelf context-based offline meta-training algorithms. Empirical experiments show that GCC achieves state-of-the-art performance on the Meta-World ML1 benchmark compared to baselines with/without offline adaptation.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics

Submission Guidelines: Yes

Please Choose The Closest Area That Your Submission Falls Into: Reinforcement Learning (eg, decision and control, planning, hierarchical RL, robotics)

TL;DR: This paper formalizes the data distribution mismatch between offline meta-training and online adaptation, and proposes a novel data correction algorithm for effective online adaptation.