A Recipe for Unbounded Data Augmentation in Visual Reinforcement Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Almuzairee, Abdulaziz, Hansen, Nicklas, Christensen, Henrik I.
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916326513049600
author Almuzairee, Abdulaziz
Hansen, Nicklas
Christensen, Henrik I.
author_facet Almuzairee, Abdulaziz
Hansen, Nicklas
Christensen, Henrik I.
contents Q-learning algorithms are appealing for real-world applications due to their data-efficiency, but they are very prone to overfitting and training instabilities when trained from visual observations. Prior work, namely SVEA, finds that selective application of data augmentation can improve the visual generalization of RL agents without destabilizing training. We revisit its recipe for data augmentation, and find an assumption that limits its effectiveness to augmentations of a photometric nature. Addressing these limitations, we propose a generalized recipe, SADA, that works with wider varieties of augmentations. We benchmark its effectiveness on DMC-GB2 - our proposed extension of the popular DMControl Generalization Benchmark - as well as tasks from Meta-World and the Distracting Control Suite, and find that our method, SADA, greatly improves training stability and generalization of RL agents across a diverse set of augmentations. For visualizations, code and benchmark: see https://aalmuzairee.github.io/SADA/
format Preprint
id arxiv_https___arxiv_org_abs_2405_17416
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle A Recipe for Unbounded Data Augmentation in Visual Reinforcement Learning
Almuzairee, Abdulaziz
Hansen, Nicklas
Christensen, Henrik I.
Machine Learning
Computer Vision and Pattern Recognition
Robotics
Q-learning algorithms are appealing for real-world applications due to their data-efficiency, but they are very prone to overfitting and training instabilities when trained from visual observations. Prior work, namely SVEA, finds that selective application of data augmentation can improve the visual generalization of RL agents without destabilizing training. We revisit its recipe for data augmentation, and find an assumption that limits its effectiveness to augmentations of a photometric nature. Addressing these limitations, we propose a generalized recipe, SADA, that works with wider varieties of augmentations. We benchmark its effectiveness on DMC-GB2 - our proposed extension of the popular DMControl Generalization Benchmark - as well as tasks from Meta-World and the Distracting Control Suite, and find that our method, SADA, greatly improves training stability and generalization of RL agents across a diverse set of augmentations. For visualizations, code and benchmark: see https://aalmuzairee.github.io/SADA/
title A Recipe for Unbounded Data Augmentation in Visual Reinforcement Learning
topic Machine Learning
Computer Vision and Pattern Recognition
Robotics
url https://arxiv.org/abs/2405.17416