Background Planning and Variational State Tabulation.

DeepRL2.3A, CS-456 Artificial Neural Networks

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.