Off-policy Reinforcement Learning

Allows training using data generated by policies other than the current one, e.g., experience replay