Recently, researchers have made significant progress combining the advances
in deep learning for learning feature representations with reinforcement
learning. Some notable examples include training agents to play Atari games
based on raw pixel data and to acquire advanced manipulation