Artificial intelligence has seen several breakthroughs in recent years. Games such as checkers, chess and others have often served as milestones for progress. Poker is an entirely different game, players having their own asymmetrical information about what is going on in the game. In this talk, I will describe a research program of ten or so years to create a artificial intelligence capable of handling the characteristics of poker: deception, bluffing and manipulating the knowledge of other players. This search resulted in two landmark results: Cepheus playing a perfect limit poker game and, more recently, DeepStack (in collaboration with Czech researchers) beating poker pros in no-limit poker. These two computer programs take very different approaches and highlight what is needed to play an expert level game and what is required to play it perfectly.
Michael is a tenured professor at the University of Alberta. His research focuses on machine learning, games and robotics. He is particularly fascinated by the problem of how computers can learn to play games through experience. Michael leads the computer poker research group, which has developed some of the best poker programs on the planet. The programs have won international competitions of artificial intelligence and were the first to beat the best professional players during a significant competition. He is also a Senior Research Scientist in the Reinforcement Learning and Artificial Intelligence (RLAI) group and the Alberta Machine Intelligence Institute (Amii).