We’re holding a tournament at UC Irvine for students to train AI agents that compete against each other in multiplayer games using reinforcement learning.
Our focus is to raise interest in tackling complex games with more than two players competing against each another in a free-for-all scenario. These games can pose a greater challenge to learn than previous popular 1v1 games such as chess and Go because there is no single best course of action for any given state. Groups of players can temporarily form arbitrary coalitions against other players to get them out of the picture, and strategies that might work well against certain combinations of play styles can totally fail in the presence of a one or more opponents that play the game slightly differently than expected.
The first game we are holding a tournament for is Blokus, a 4-player free-for-all turn-based game in which players place pieces on a board to claim space and strategically block opponents from placing their own pieces.
Currently we’re in the beta phase of testing our multiplayer game API with Blokus. If you’d like to get involved and start working on your own agent, you can get started using our API now with the help in our Documentation.