Human-level Trick-Taking Card Game AI
Leveraging reinforcement, imitation, and supervised learning to beat humans at Judgement, a playing card game which I’ve played far, far too much of (repository here - try your hand at beating the bot!), with applications to general trick-taking games.
Over my college career, I played a LOT of the trick-taking card game called Judgement/Kachufool (rules here). An embarassing amount, almost. So much so that I developed an entire software based infastructure around the game, including automated scoring:
and an entire league system complete with seeding and playoffs, where eight of my friends and I played dozens of games of Judgement at locations as varied as coffee shops, the laundry room, and of course IHOP:
After all this effort and the heights of competition that were reached in the Judgement League, it was only natural that I would try to use computers to defeat my friends at the game.
Judgment, and trick-taking games in general, can mostly be separated into two distinct phases - the pre-round phase, and the trick-taking phase.
.jpg)
