CoBERL: Contrastive BERT for 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.

Cite

Text

Banino et al. "CoBERL: Contrastive BERT for Reinforcement Learning." International Conference on Learning Representations, 2022.

Markdown

[Banino et al. "CoBERL: Contrastive BERT for Reinforcement Learning." International Conference on Learning Representations, 2022.](https://mlanthology.org/iclr/2022/banino2022iclr-coberl/)

BibTeX

@inproceedings{banino2022iclr-coberl,
  title     = {{CoBERL: Contrastive BERT for Reinforcement Learning}},
  author    = {Banino, Andrea and Badia, Adria Puigdomenech and Walker, Jacob C and Scholtes, Tim and Mitrovic, Jovana and Blundell, Charles},
  booktitle = {International Conference on Learning Representations},
  year      = {2022},
  url       = {https://mlanthology.org/iclr/2022/banino2022iclr-coberl/}
}