Abstract: This talk discusses the role of stochastic dominance in optimization and control as a tool for risk management. We first develop duality theory and computational methods for optimization problems with stochastic dominance constraints. Then, we extend these ideas to the sequential setting to manage risk in MDPs with stochastic dominance via the convex analytic approach.
Bio: Will Haskell got his B.S. Mathematics from the University of Massachusetts Amherst in 2006, and his Ph.D. in Operations Research from the University of California in 2011. He taught in the industrial systems engineering department here at USC for two years, and now he is a postdoc for Rahul Jain and Milind Tambe. His research emphasizes risk management and simulation-based/data-driven optimization.