Closing the Gap between TD Learning and Supervised Learning -- A Generalisation Point of View.

Published: 03 Nov 2023, Last Modified: 27 Nov 2023GCRL WorkshopEveryoneRevisionsBibTeX
Confirmation: I have read and confirm that at least one author will be attending the workshop in person if the submission is accepted
Keywords: reinforcement learning, data augmentation, stitching
TL;DR: We show that the stitching property in RL is a form of generalization that SL-based methods can not have, however we show how a new type of data augmentation facilitates this type of generalization.
Abstract: Some reinforcement learning (RL) algorithms have the capability of recombining together pieces of previously seen experience to solve a task never seen before during training. This oft-sought property is one of the few ways in which dynamic programming based RL algorithms are considered different from supervised learning (SL) based RL algorithms. Yet, recent RL methods based on off-the-shelf SL algorithms achieve excellent results without an explicit mechanism for stitching; it remains unclear whether those methods forgo this important stitching property. This paper studies this question in the setting of goal-reaching problems. We show that the desirable stitching property corresponds to a form of generalization: after training on a distribution of (state, goal) pairs, one would like to evaluate on (state, goal) pairs not seen \emph{together} in the training data. Our analysis shows that this sort of generalization is different from \emph{i.i.d.} generalization. This connection between stitching and generalization reveals why we should not expect existing RL methods based on SL to perform stitching, even in the limit of large datasets and models. We experimentally validate this result on carefully constructed datasets. This connection suggests a simple remedy, the same remedy for improving generalization in supervised learning: data augmentation. We propose a naive \emph{temporal} data augmentation approach and demonstrate that adding it to RL methods based on SL enables them to stitch together experience so that they succeed in navigating between states and goals unseen together during training.
Supplementary Material: zip
Submission Number: 26