Policy-Induced Self-Supervision Improves Representation Finetuning in Visual RL

Séb Arnold, Fei Sha

Keywords: visual reinforcement learning, representation learning, transfer learning

TL;DR: We study the transfer of visual representations in RL, show that they can be partially frozen, and propose a self-supervised method to accelerate their finetuning.

Abstract:

We study how to transfer representations pretrained on source tasks to target tasks in visual percept based RL. We analyze two popular approaches: freezing or finetuning the pretrained representations. Empirical studies on a set of popular tasks reveal several properties of pretrained representations. First, finetuning is required even when pretrained representations perfectly capture the information required to solve the target task. Second, finetuned representations improve learnability and are more robust to noise.
Third, pretrained bottom layers are task-agnostic and readily transferable to new tasks, while top layers encode task-specific information and require adaptation. Building on these insights, we propose a self-supervised objective that \emph{clusters representations according to the policy they induce}, as opposed to traditional representation similarity measures which are policy-agnostic (\eg Euclidean norm, cosine similarity). Together with freezing the bottom layers, this objective results in significantly better representation than frozen, finetuned, and self-supervised alternatives on a wide range of benchmarks.

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)

Community Implementations: [ 3 code implementations](https://www.catalyzex.com/paper/policy-induced-self-supervision-improves/code)