A Bayesian Model for Opening Prediction in RTS Games with Application to StarCraft

Gabriel Synnaeve (1, 2) , Pierre Bessiere (3)

1 E-MOTION - Geometry and Probability for Motion and Action
2 LIG - Laboratoire d'Informatique de Grenoble
3 LPPA - Laboratoire de Physiologie de la Perception et de l'Action

This paper presents a Bayesian model to predict the opening (first strategy) of opponents in real-time strategy (RTS) games. Our model is general enough to be applied to any RTS game with the canonical gameplay of gathering resources to extend a technology tree and produce military units and we applied it to StarCraft1. This model can also predict the possible technology trees of the opponent, but we will focus on openings here. The parameters of this model are learned from replays (game logs), labeled with openings. We present a semi- supervised method of labeling replays with the expectation- maximization algorithm and key features, then we use these labels to learn our parameters and benchmark our method with cross-validation. Uses of such a model range from a commentary assistant (for competitive games) to a core component of a dynamic RTS bot/AI, as it will be part of our StarCraft AI competition entry bot.

Mots clés: Bayesian; RTS; machine learning; artificial intelligence; games; plan recognition

Domaines: Intelligence artificielle [cs.AI]; Machine Learning [stat.ML]

Gabriel Synnaeve, Pierre Bessiere. A Bayesian Model for Opening Prediction in RTS Games with Application to StarCraft. Computational Intelligence and Games, Aug 2011, Seoul, South Korea. pp.000. ⟨hal-00607277⟩