Picture of James

Contact

jrwright@microsoft.com

"That's your game theory? Rock Paper Scissors with statistics?"
— Peter Watts, Blindsight

James R. Wright

Hello, I'm James Wright. I am a postdoctoral researcher at Microsoft Research in New York City. In July 2018, I will start as an Assistant Professor at the University of Alberta. I completed my Ph.D. at UBC in 2016, advised by Kevin Leyton-Brown.

Research

My primary research interest is in using data-driven machine learning models to predict human strategic behavior; that is, behavior in interactions where each participant's rewards depend partially on the actions of other participants. My long-term research agenda is to build a general theory for optimally designing algorithms for mediating interactions involving humans or other realistically bounded agents rather than idealized, perfectly rational game theoretic agents.

Curriculum Vitae

My academic CV is available as both an HTML page and a PDF document. I also have a public Google Scholar citations page.

Publications

  1. Predicting Human Behavior in Unrepeated, Simultaneous-Move Games.
    James R. Wright and Kevin Leyton-Brown.
    Games and Economic Behavior, Volume 106, pages 16–37, November 2017.
    (supersedes Wright & Leyton-Brown [2010, 2012])
  2. Learning in the Repeated Secretary Problem.
    Daniel G. Goldstein, R. Preston McAfee, Siddarth Suri, and James R. Wright.
    ACM Conference on Economics and Computation (ACM-EC), 2017.
  3. Deep Learning for Predicting Human Strategic Behavior.
    Jason Hartford, James R. Wright, and Kevin Leyton-Brown.
    NIPS 2016: Thirtieth Annual Conference on Neural Information Processing Systems, 2016.
    Oral presentation.
  4. Incentivizing Evaluation via Limited Access to Ground Truth: Peer-Prediction Makes Things Worse.
    Xi Alice Gao, James R. Wright, and Kevin Leyton-Brown.
    Workshop on Algorithmic Game Theory and Data Science at ACM Conference on Economics and Computation, 2016.
  5. Mechanical TA: Partially Automated High-Stakes Peer Grading.
    James R. Wright, Chris Thornton, Kevin Leyton-Brown.
    ACM Technical Symposium on Computer Science Education (ACM-SIGCSE), 2015
  6. Level-0 Meta-Models for Predicting Human Behavior in Games.
    James R. Wright and Kevin Leyton-Brown.
    ACM Conference on Economics and Computation (ACM-EC), 2014.
  7. Behavioral Game-Theoretic Models: A Bayesian Framework For Parameter Analysis.
    James R. Wright and Kevin Leyton-Brown.
    Proceedings of the 11th International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2012), pages 921–928, 2012.
    Best student paper (runner up).
  8. Linear solvers for nonlinear games: using pivoting algorithms to find Nash equilibria in n-player games.
    James R. Wright, Albert Xin Jiang, and Kevin Leyton-Brown.
    SIGecom Exchanges, volume 10, number 1, pages 9–12, 2011.
  9. Beyond Equilibrium: Predicting Human Behavior in Normal Form Games.
    James R. Wright and Kevin Leyton-Brown.
    Proceedings of the Twenty-Fourth AAAI Conference on Artificial Intelligence (AAAI-10), pages 901–907, 2010.

Working Papers

  1. Models of Level-0 Behavior for Predicting Human Behavior in Games.
    James R. Wright and Kevin Leyton-Brown.
    Under review at Journal of Artificial Intelligence Research.
    (supersedes Wright & Leyton-Brown [2014])
  2. Incentivizing Evaluation via Limited Access to Ground Truth:
    Peer-Prediction Makes Things Worse.

    Xi Alice Gao, James R. Wright, and Kevin Leyton-Brown.
    Under review by Artificial Intelligence Journal.
    (supersedes Gao, Wright, and Leyton-Brown [2016])

Last update: Oct 10/2017