I'm still working on this section. Since 2006 I've worked on many different parts of our computer poker effort, from state space abstraction to solving to evaluation and adaptation, and there's a lot of ground to cover.

My PhD thesis (2016) is a comprehensive end-to-end look at the task of computing strong strategies for imperfect information games, like Texas hold'em poker. It includes details of how we beat human poker pros at heads-up limit Texas hold'em in our 2008 Man-vs-Machine Poker Championship, how we constructed our Annual Computer Poker Competition "Hyperborean" agent from 2008 to 2014, and how we (University of Alberta team, plus Oskari Tammelin) solved heads-up limit Texas hold'em in 2015.

My Masters thesis (2007) described how we first challenged poker pros in our 2007 Man-vs-Machine match, which we narrowly lost. It describes the core components in detail, including the CFR algorithm that we used to solve large games, simplify games to a tractable size, and adapt to opponents. All of these aspects have undergone quite a bit of evolution in the last 9 years, though, and the PhD thesis largely replaces this work.

Publications :: PhD Thesis
Publications :: MSc Thesis