Artificial intelligence has seen several breakthroughs in recent years, with games such as checkers, chess, and go often serving as milestones of progress. Poker is another game entirely, with players having their own asymmetric information about what's happening in the game. In this talk, I'll describe a decade long research program to build AI that can cope with the hallmarks of poker — deception, bluffing, and manipulating what other players know. This research has culminated in two landmark results: Cepheus playing a perfect game of limit poker, and most recently DeepStack (in a collaboration with Czech researchers) beating poker pros at the game of no-limit poker. These two computer programs take very different approaches, and shed light on what is required to play a game at an expert-level and what is required to play it perfectly.
Michael is a full professor at the University of Alberta. His research focuses on machine learning, games and robotics, and he's particularly fascinated by the problem of how computers can learn to play games through experience. Michael leads the Computer Poker Research Group, which has built some of the best poker playing programs on the planet. The programs have won international AI competitions as well as being the first to beat top professional players in a meaningful competition. He is also a principal investigator in the Reinforcement Learning and Artificial Intelligence (RLAI) group and the Alberta Machine Intelligence Institute (Amii).