Invited Talk

It is common to assume that players in a game will adopt Nash equilibrium strategies. However, experimental studies have demonstrated that Nash equilibrium is often a poor description of human players’ behavior, even in unrepeated normal-form games. Nevertheless, human behavior in such settings is far from random. Drawing on data from real human play, the field of behavioral game theory has developed a variety of models that aim to capture these patterns. The current state of the art in that literature is a model called quantal cognitive hierarchy. It predicts that agents approximately best respond and explicitly model others’ beliefs to a finite depth, grounded in a uniform model of nonstrategic play. We have shown that even stronger models can be built by drawing on ideas from cognitive psychology to better describe nonstrategic behavior. However, this whole approach requires extensive expert knowledge and careful choice of functional form. Deep learning presents an alternative, offering the promise of automatic cognitive modeling. We introduce a novel architecture that allows a single network to generalize across different input and output dimensions by using matrix units rather than scalar units, and show that its performance significantly outperforms that of the previous state of the art.

Kevin’s bio:

Kevin Leyton-Brown is a professor of Computer Science at the University of British Columbia and an associate member of the Vancouver School of Economics. He holds a PhD and M.Sc. from Stanford University (2003; 2001) and a B.Sc. from McMaster University (1998). He studies the intersection of computer science and microeconomics, addressing computational problems in economic contexts and incentive issues in multiagent systems. He also applies machine learning to the automated design and analysis of algorithms for solving hard computational problems.

He has co-written two books, “Multiagent Systems” and “Essentials of Game Theory,” and over 100 peer-refereed technical articles; his work has received over 8,000 citations and an h-index of 37. He is the recipient of UBC’s2015 Charles A. McDowell Award for Excellence in Research, a 2014 NSERC E.W.R. Steacie Memorial Fellowship—previously given to a computer scientist only 10 times since its establishment in 1965—and a 2013Outstanding Young Computer Science Researcher Prize from the Canadian Association of Computer Science. He and his coauthors have received paper awards from JAIR, ACM-EC, AAMAS and LION, and numerous medals for the portfolio-based SAT solver SATzilla at international SAT solver competitions (2003–15).

He has co-taught two Coursera courses on “Game Theory” to over half a million students, and has received awards for his teaching at UBC—notably, a 2013/14 Killam Teaching Prize. He is chair of the ACM Special Interest Group on Electronic Commerce, which runs the annual Economics & Computation conference. He serves as an associate editor for the Artificial Intelligence Journal (AIJ), ACM Transactions on Economics and Computation (ACM-TEAC), and AI Access; serves as an advisory board member for the Journal of Artificial Intelligence Research (JAIR, after serving as associate editor for eight years); and was program chair for the ACM Conference on Electronic Commerce (ACM-EC) in 2012. In 2016 he was a visiting researcher at Microsoft Research New England and a visiting professor at Harvard’s EconCS group. Previously, he spent the fall of 2015 at the Simons Institute for the Theory of Computing at UC Berkeley, and split his 2010–11 sabbatical between Makerere University in Kampala, Uganda and the Institute for Advanced Studies at Hebrew University of Jerusalem, Israel. He currently advises Auctionomics, Inc. (and through them, the Federal Communications Commission) andQudos, Inc. He is a co-founder of Kudu.ug and Meta-Algorithmic Technologies. In the past, he served as a consultant for Zynga, Inc., Trading Dynamics Inc., Ariba Inc., Cariocas Inc., and was scientific advisor to UBC spinoff Zite Inc. until it was acquired by CNN in 2011.