CoBERL: Contrastive BERT for Reinforcement LearningDownload PDF

29 Sept 2021, 00:35 (modified: 12 Mar 2022, 21:22)ICLR 2022 SpotlightReaders: Everyone
Keywords: Reinforcement Learning, Contrastive Learning, Representation Learning, Transformer, Deep Reinforcement Learning
Abstract: Many reinforcement learning (RL) agents require a large amount of experience to solve tasks. We propose Contrastive BERT for RL (COBERL), an agent that combines a new contrastive loss and a hybrid LSTM-transformer architecture to tackle the challenge of improving data efficiency. COBERL enables efficient and robust learning from pixels across a wide variety of domains. We use bidirectional masked prediction in combination with a generalization of a recent contrastive method to learn better representations for RL, without the need of hand engineered data augmentations. We find that COBERL consistently improves data efficiency across the full Atari suite, a set of control tasks and a challenging 3D environment, and often it also increases final score performance.
One-sentence Summary: A new loss and an improved architecture to efficiently train attentional models in reinforcement learning.
Supplementary Material: zip
12 Replies