Unsupervised State Representation Learning in Atari | Read Paper on Bytez