Modelling Abstractions in the Context of Deep Reinforcement Learning and Rate-distortion Theory
project page