DeepRL2.3A, Background Planning and Variational State Tabulation.

views comments

Model based reinforcement learning enables planning in the background. Here we show a modern version of this idea where the moded of the environment is learned from the data.