Reinforcement Learning with Augmented Data
This work addresses data-efficiency and generalization problems for reinforcement learning practitioners, offering an incremental but effective enhancement to existing algorithms.
The paper tackled data-efficiency and generalization challenges in reinforcement learning from visual observations by introducing RAD, a plug-and-play module using data augmentations like random translate and amplitude scale, which enabled simple RL algorithms to outperform complex state-of-the-art methods, achieving new SOTA on DeepMind Control Suite and OpenAI Gym benchmarks with significant improvements in test-time generalization on ProcGen benchmarks.
Learning from visual observations is a fundamental yet challenging problem in Reinforcement Learning (RL). Although algorithmic advances combined with convolutional neural networks have proved to be a recipe for success, current methods are still lacking on two fronts: (a) data-efficiency of learning and (b) generalization to new environments. To this end, we present Reinforcement Learning with Augmented Data (RAD), a simple plug-and-play module that can enhance most RL algorithms. We perform the first extensive study of general data augmentations for RL on both pixel-based and state-based inputs, and introduce two new data augmentations - random translate and random amplitude scale. We show that augmentations such as random translate, crop, color jitter, patch cutout, random convolutions, and amplitude scale can enable simple RL algorithms to outperform complex state-of-the-art methods across common benchmarks. RAD sets a new state-of-the-art in terms of data-efficiency and final performance on the DeepMind Control Suite benchmark for pixel-based control as well as OpenAI Gym benchmark for state-based control. We further demonstrate that RAD significantly improves test-time generalization over existing methods on several OpenAI ProcGen benchmarks. Our RAD module and training code are available at https://www.github.com/MishaLaskin/rad.