GDC is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.

View, browse and sort the ever-growing list of GDC sessions by time, pass type, topic, and format. With GDC Session Scheduler, you can build your schedule in advance and access it during the show via export or with the GDC Mobile App, once live.

If you've registered for GDC, you can use the same login credentials for your Session Scheduler account. Adding a session to your schedule does not guarantee you a seat. Sessions do fill up, so please arrive early to sessions that you would like to attend.

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

Theresa Duringer (CEO, Temple Gates Games)

Location: Room 2002, West Hall

Date: Monday, March 16

Time: 1:20pm - 1:50pm

Pass Type: All Access Pass, Core+Summits Pass, Summits Pass - Get your pass now!

Topic: Programming

Format: Session

Vault Recording: TBD

Audience Level: Intermediate

Multiplayer boardgame adaptations require competitive AI opponents for solo play. But what's the best approach to creating these opponents? At Temple Gates, we've found neural network AI to be a good fit for boardgames. In this session, we'll give an overview of our AI development strategy each of our games: Ascension, Race for the Galaxy, Shards of Infinity, Roll for the Galaxy, and our upcoming deckbuilder release. We'll discuss input representations (and pitfalls), taking advantage of symmetry, use of an embedding layer, and training and debugging techniques. Tricks like these are all aimed at improving the speed of the nn's learning and improving competence.


Attendees will learn about designing reinforcement learning AI and adapting it for production across a range of boardgames. They'll find out about performance improving features such as identifying symmetries, strategically designing input gaps, and adding embedding layers to better handle large permutations of game states.

Intended Audience

This talk is intended for AI designers and programmers who want to offer highly replayable solo content in turn-based games. It can help studio leads determine which games are good candidates for reinforcement learning AI techniques. Finally, it can inform boardgame designers know what features to avoid or to include based on the feasibility of AI solutions.