Acquisition of Chess Knowledge in AlphaZero

11/17/2021 ∙ by Thomas McGrath, et al. ∙ 80

What is being learned by superhuman neural network agents such as AlphaZero? This question is of both scientific and practical interest. If the representations of strong neural networks bear no resemblance to human concepts, our ability to understand faithful explanations of their decisions will be restricted, ultimately limiting what we can achieve with neural network interpretability. In this work we provide evidence that human knowledge is acquired by the AlphaZero neural network as it trains on the game of chess. By probing for a broad range of human chess concepts we show when and where these concepts are represented in the AlphaZero network. We also provide a behavioural analysis focusing on opening play, including qualitative analysis from chess Grandmaster Vladimir Kramnik. Finally, we carry out a preliminary investigation looking at the low-level details of AlphaZero's representations, and make the resulting behavioural and representational analyses available online.

READ FULL TEXT
POST COMMENT

Comments

There are no comments yet.

Authors

page 17

page 19

page 23

page 37

page 38

page 39

page 40

page 42

This week in AI

Get the week's most popular data science and artificial intelligence research sent straight to your inbox every Saturday.