About me
Name: Jie Wang
Talk Title: Stochastic Optimization with Decisions Truncated by Random Variables
Affiliation: Georgia Institute of Technology
Abstract: Stochastic optimization with decisions truncated by random variables has broad applications in revenue management. However, the training objective is generally nonconvex, even when the transformation function is convex. To address this challenge, we develop a convex hull relaxation that provides an exact reformulation under the assumption of positive dependence in the distribution of the random variable. This leads to an infinite-dimensional convex optimization problem over high-dimensional probability measures, where conventional methods suffer from the curse of dimensionality. To ensure computational tractability, we exploit the problem structure and show that it is equivalent to optimizing over marginal distributions. Based on this insight, we propose a finite-particle approximation that breaks the curse of dimensionality and establish its global convergence guarantees. Finally, we demonstrate the superior performance of our algorithm on a booking limit control problem.
Bio: Dr. Jie Wang received his Ph.D in Industrial Engineering at Georgia Institute of Technology in 2025. His research focuses on decision-making under uncertainty, through the lens of statistics and optimization, with practical applications in machine learning.
