Concurrent Credit Assignment for Data-efficient Reinforcement Learning

Emmanuel Daucé

The capability to widely sample the state and action spaces is a key ingredient toward building effective reinforcement learning algorithms. The variational optimization principles…