Optimal sequential decision making is a fundamental problem to many diverse fields. In recent years, Reinforcement Learning (RL) methods have experienced unprecedented success, largely enabled by the use of deep learning ...