You've been logged out of GDC Vault since the maximum users allowed for this account has been reached. To access Members Only content on GDC Vault, please log out of GDC Vault from the computer which last accessed this account.

Click here to find out about GDC Vault Membership options for more users.

close

Session Name:

AI Summit: Every Day I'm Shufflin': Neural Network Variations for Digital Deck-Builders

Overview:

Multiplayer boardgame adaptations require competitive AI opponents for solo play. But what's the best approach to creating these opponents? Temple Gates has scripted a heuristic AI for 'Ascension' and integrated a neural network for 'Race for the Galaxy'. Since then, each new game has used new techniques to improve AI competitiveness. Deckbuilder 'Shards of Infinity's' AI identifies symmetries, such as symmetries between players or cards, to supercharge the speed of training. 'Roll for the Galaxy's' AI is designed with holes in the inputs to speed training without creating strategic blindspots. 'Dominion' introduces an embedding layer, so the neural network can learn not only the cards themselves, but the value of their constituent features. This helps handle the highly variant game state where one of a quadrillion of permutations of cards may be in play. Understanding what techniques are available in ml and which best apply to different game scenarios is a shortcut to getting the most of a nn AI.

Did you know free users get access to 30% of content from the last 2 years?


Get your team full access to the most up to date GDC content

  • Game Developers Conference 2024
  • Theresa Duringer
  • Temple Gates Games
  • free content
  • AI Summit
  • Programming