Learn mahjong in five minutes. A visual guide for total beginners — tiles, winning hands, wall mechanics, and turn actions.

We present the first deep learning model to successfully learn control policies directly from high-dimensional sensory input using reinforcement learning.