We show that reinforcement learning agents that learn by surprise (surprisal) get stuck at abrupt environmental transition boundaries because these transitions are difficult to learn. We propose a counter-intuitive solution that we call Mutual Information Minimising Exploration (MIME) where an agent learns a latent representation of the environment without trying to predict the future states. We show that our agent performs significantly better over sharp transition boundaries while matching the performance of surprisal driven agents elsewhere. In particular, we show state-of-the-art performance on difficult learning games such as Gravitar, Montezuma's Revenge and Doom.
@article{arxiv.2001.05636,
title = {MIME: Mutual Information Minimisation Exploration},
author = {Haitao Xu and Brendan McCane and Lech Szymanski and Craig Atkinson},
journal= {arXiv preprint arXiv:2001.05636},
year = {2020}
}