This paper introduces R2D3, an agent that makes efficient use of demonstrations to solve hard exploration problems in partially observable environments with highly variable initial conditions. We also introduce a suite of eight tasks that combine these three properties, and show that R2D3 can solve several of the tasks where other state of the art methods (both with and without demonstrations) fail to see even a single successful trajectory after tens of billions of steps of exploration.
Cite
@article{arxiv.1909.01387,
title = {Making Efficient Use of Demonstrations to Solve Hard Exploration Problems},
author = {Tom Le Paine and Caglar Gulcehre and Bobak Shahriari and Misha Denil and Matt Hoffman and Hubert Soyer and Richard Tanburn and Steven Kapturowski and Neil Rabinowitz and Duncan Williams and Gabriel Barth-Maron and Ziyu Wang and Nando de Freitas and Worlds Team},
journal= {arXiv preprint arXiv:1909.01387},
year = {2019}
}