Deep reinforcement learning is a subfield of machine learning that combines reinforcement learning and deep learning. RL considers the problem of a computational agent learning to make decisions by trial and error. Wikipedia