A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play
At a glance
AlphaZero learned chess, shogi and Go through self-play rather than human examples.
In one paragraph
Introduced AlphaZero, which learned chess, shogi and Go by playing against itself, using the same learning approach for each game. Given the rules but no examples of human play, it defeated leading specialist programs in all three games.
Authors 13
Organizations
Credited for this work at publication.