Title: Strategic Reasoning in Complex Zero-Sum Computer Games

To succeed in complex computer games, human players often resort to previously trained strategies and excel at generalizing responses by analogy between unforeseen and familiar situations. This thesis presents a computational version of this behavior: we replace the human repertoire of strategies by a portfolio of game-playing algorithms, and we use known generalization schemes to promote similar responses to similar game states. We make use of existing algorithms, aiming to discover the best in each game situation. Synthetic experiments demonstrate the advantages of algorithm selection according to the problem complexity and the portfolio characteristics. We instantiate our framework in real-time strategy games, where a player must strategically develop its economy and quickly maneuver its units in combat. Our framework allows the discussion of game-theoretic aspects of algorithm selection and consistently outperforms state-of-the-art game-tree search approaches.