Rob Fergus
@rob_fergus
Director, AI Research and Head of FAIR at Meta
Happy to share my new work -- Proto-RL, a task-agnostic pre-training scheme that reconciles exploration and representation learning in image-based RL! with: @rob_fergus, Alessandro Lazaric, and @LerrelPinto. paper: arxiv.org/pdf/2102.11271… code: github.com/denisyarats/pr… [1/N]
@ylecun @AravSrinivas recent Proto-RL (arxiv.org/pdf/2102.11271…) that uses ideas from both CURL and DrQ gets us closer to the LeCake, as it's contrastively learned representations are **fully detached** from RL, besides they can be learned with just the MaxEnt objective.
Excited to share an update to our work on evolutionary-scale modeling (ESM)! Over the past year, we rewrote our paper with better pretraining and downstream models, leading to state-of-the-art results across multiple benchmarks. (1/8) biorxiv.org/content/10.110…
Excited to share our new paper “Automatic Data Augmentation for Generalization in Deep Reinforcement Learning” w/ @maxagoldstein8, @denisyarats, @ikostrikov, and @rob_fergus! Paper: arxiv.org/pdf/2006.12862… Code: github.com/rraileanu/auto… Website: sites.google.com/view/ucb-drac
Great work from my PhD students @denisyarats and @ikostrikov !
Exciting to announce our new work together with @ikostrikov and @rob_fergus: Image Augmentation Is All You Need: Regularizing Deep Reinforcement Learning from Pixels. Paper: arxiv.org/abs/2004.13649 Code: github.com/denisyarats/drq Website: sites.google.com/view/data-regu… [1/N]