MuZero by DeepMind learns rules autonomously
DeepMind (a subsidiary of Alphabet) has made groundbreaking advancements using reinforcement learning to teach programs to master the board game Go and the strategy game Shogi, as well as chess and challenging Atari video games. In all those cases, the computers were given the rules of the game. A new report says that DeepMind’s MuZero has accomplished the same feats, and in some cases, surpassed previous programs without learning the rules first. The DeepMind programmers relied on a principle called lookahead. With that approach, MuZero evaluates a series of potential moves based on how an opponent would respond. While there are likely to be a staggering number of potential moves in complex games like chess, MuZero prioritizes the most relevant and likely moves, learning from successful tactics and avoiding the ones that failed.