WoodStock is a constraint-based general game player for the General Game Playing (GGP). The first component is a network player, called spy-ggp, used to communicate with the game manager from the GGP server. The second component is the core of our program, which includes: - a parser that translates a GDL input into a Stochastic Constraint Satisfaction Problem (SCSP), and - a stochastic constraint solver that finds winning (maximin) strategies. Our solver, MAC-UCB-SYM, is based on three key ideas : - Constraint-based search propagation (the MAC component), and - Bandit-based stochastic sampling (the UCB component). - Constraint-based symmetry detection for effectively filtering equivalent positions and moves (the SYM component). WoodStock is developped as part of Eric Piette's phd thesis supervised by Frédéric Koriche, Sylvain Lagrue and Sébastien Tabary.
August 25, 2016 : WoodStock won the 2016 International General Game Playing Competition (IGGPC'16). WoodStock is champion !
Since March 2016 : WoodStock consistently leads the GGP Tiltyard continuous tournament.
December 6, 2015 : WoodStock has finished third on the 2015 Tiltyard Open and second during the qualification round.
Frédéric Koriche, Sylvain Lagrue, Eric Piette and Sébastien Tabary, « Constraint-Based Symmetry Detection in General Game Playing ». 26th International Joint Conference on Artificial Intelligence (IJCAI'17). see cite
Frédéric Koriche, Sylvain Lagrue, Eric Piette and Sébastien Tabary, « General Game Playing with Stochastic CSP ». Constraints, 21(1) :95–114, 2016. see cite
Frédéric Koriche, Sylvain Lagrue, Eric Piette and Sébastien Tabary, « Stochastic Constraint Programming for General Game Playing with Imperfect Information ». In General Intelligence in Game-Playing Agents (GIGA'16) at the 25th International Joint Conference on Artificial Intelligence on workshop (ws-IJCAI'16). see cite
Frédéric Koriche, Sylvain Lagrue, Eric Piette and Sébastien Tabary, « General Game Playing with Stochastic CSP (abstract version) ». In the 21th International Conference on Principles and Practice of Constraint Programming (CP'15). see cite
Frédéric Koriche, Sylvain Lagrue, Eric Piette and Sébastien Tabary, « Compiling Strategic Games with Complete Information into Stochastic CSPs ». In the Association for the Advancement of Artificial Intelligence on workshop : Planning, Search, and Optimization (ws-AAAI'15). see cite
Frédéric Koriche, Sylvain Lagrue, Eric Piette et Sébastien Tabary, « WoodStock : un programme-joueur générique ». Dans Revue d'Intelligence Artificielle (RIA'17), Vol 31/3 - 2017 - pp.281-310. see cite
Frédéric Koriche, Sylvain Lagrue, Eric Piette et Sébastien Tabary, « Programmation par contraintes stochastiques pour le General Game Playing avec informations incomplètes ». Dans Douzièmes Journées Francophones de la Programmation par Contraintes (JPFC'16). see cite
Frédéric Koriche, Sylvain Lagrue, Eric Piette et Sébastien Tabary, « Résolution de SCSP avec borne de confiance pour les jeux de stratégie ». Dans Onzièmes Journées Francophones de la Programmation par Contraintes (JPFC'15). see cite
Frédéric Koriche, Sylvain Lagrue, Eric Piette et Sébastien Tabary, « Traduction de jeux à information incertaine en réseaux de contraintes stochastiques ». Dans Dizièmes Journées Francophones de la Programmation par Contraintes (JPFC'14). see cite