VentureBeat reports: “DeepMind — the research division of Alphabet whose work chiefly involves reinforcement learning, an area of AI concerned with how software agents ought to take actions to maximize some reward — published a preprint paper in the past week that greatly improves upon existing work.”
