Adversarial Games
David Banks
Dept. of Statistical Science, Duke University, Durham, NC, United States

Adversarial risk analysis attempts to combine game theoretic perspectives with Bayesian risk analysis. This presentation applies that approach in two classical applications: La Relance (the “Borel Game”) and Blotto games (convoy routing in hostile territory). We introduce a “Borel Game” that enables the analyst to create a model for the decision-making of the opponent, while accounting for the fact that the opponent is modeling the the decision-maker. The results are interestingly and usefully different from the traditional solutions obtained through the minimax solution concept.

Keywords: Game theory; Risk; Blotto games; La Relance

Biography: Davd Banks is a professor of statistics at Duke University. He is past-editor of the Journal of the American Statistical Association, and is a fellow of the ASA. He works on risk analysis, data mining, and metabolomics.