报告地点：腾讯会议 605844523
报告摘要： We study the newsvendor problem in the presence of contextual information that enables personalized decisions. Our goal is to seek an optimal data-driven and end-to-end policy among all policies that map each context to an ordering decision. We show that this challenging problem can be solved exactly, by first learning an in-sample robust policy defined only on observed contexts using linear programming, and then extend it to unseen contexts using Shapley policy---a novel family of policies based on the pure equilibrium of a matrix. The policy can be interpreted as a regularized critical fractile of the empirical conditional demand distribution, in which the regularization penalizes large policy variation with respect to contexts. To the best of our knowledge, our approach provides the first tractable policy optimization framework for the feature-based newsvendor which produces a robust optimal policy without restricting the policy class. We compare our method with several benchmarks using real data and demonstrate its superior empirical performance.
This is a joint work with Jincheng Yang and Luhao Zhang.
报告人简介: Rui Gao is an Assistant Professor in the Department of Information, Risk, and Operations Management at the McCombs School of Business at the University of Texas at Austin. His main research lies in developing robust and computationally efficient methodology with strong performance guarantees for learning and decision-making problems involving offline/online/contextual/observational data, arising in the area of machine learning, risk management and revenue management. His works on distributionally robust optimization have been recognized by several INFORMS awards, including a winner in Junior Faculty Interest Group Paper Competition, a winner in Data Mining Best Paper Award, a runner-up in Computing Society Student Paper Competition, and a finalist in Nicholson Student Paper Competition; as well as a Best Paper Award in IEEE Transactions on Power Systems. He received a Ph.D. in Operations Research from Georgia Institute of Technology in 2018 and a bachelor's degree in Mathematics (honors program) from Xi'an Jiaotong University in 2013.