Format

Send to

Choose Destination
Science. 2018 Dec 7;362(6419):1140-1144. doi: 10.1126/science.aar6404.

A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play.

Author information

1
DeepMind, 6 Pancras Square, London N1C 4AG, UK. davidsilver@google.com dhcontact@google.com.
2
University College London, Gower Street, London WC1E 6BT, UK.
3
DeepMind, 6 Pancras Square, London N1C 4AG, UK.

Abstract

The game of chess is the longest-studied domain in the history of artificial intelligence. The strongest programs are based on a combination of sophisticated search techniques, domain-specific adaptations, and handcrafted evaluation functions that have been refined by human experts over several decades. By contrast, the AlphaGo Zero program recently achieved superhuman performance in the game of Go by reinforcement learning from self-play. In this paper, we generalize this approach into a single AlphaZero algorithm that can achieve superhuman performance in many challenging games. Starting from random play and given no domain knowledge except the game rules, AlphaZero convincingly defeated a world champion program in the games of chess and shogi (Japanese chess), as well as Go.

Comment in

PMID:
30523106
DOI:
10.1126/science.aar6404

Supplemental Content

Full text links

Icon for HighWire
Loading ...
Support Center