Disentangling causal effects for hierarchical reinforcement learning

Oriol Corcoll, Raul Vicente

Exploration and credit assignment under sparse rewards are still challenging problems. We argue that these challenges arise in part due to the intrinsic rigidity of operating at th…