Improving on a deep-learning method pioneered for Pong, Space Invaders, and other Atari games, Stanford University computer science student Kevin Chen has created an algorithm that’s quite good ...