In value-based deep reinforcement learning, a pruned network is a good network.

Johan S. Obando-Ceron, Aaron C. Courville, Pablo Samuel Castro

Browse the full ICML paper archive.