题　目：From Conditional Stochastic Optimization to Multistage Stochastic Programming: Breaking the Curse of Dimension

嘉　宾：Yifan Hu，Assistant Professor，Rutgers University

主持人：邓琪，副教授，上海交通大学安泰经济与管理学院

时　间：2026年6月26日（周五）14:00-15:30

地　点：上海交通大学徐汇校区安泰包图A403室

内容简介:

This talk discusses the curse of dimensionality in multistage stochastic programming (MSP). It is widely known that SAA methods for T-stage MSP admit a high sample complexity, which scales exponentially with respect to number of stages. We introduce a new approximation method and establish a sample complexity that scales polynomially with respect to T (ignoring the log factors). The talk will discuss the trajectory of the development from conditional stochastic optimization and contextual stochastic bilevel optimization to multistage stochastic programming.

演讲人简介：

Yifan Hu is an assistant professor in the Department of Statistics, Rutgers University (New Brunswick, NJ). His research interests lie in decision-making under uncertainty, with an intersection of optimization, RL, statistics, and causality, aiming to build new models and develop simple-to-implement methods with provable guarantees. He serves as an area chair for ICML and ICLR. He was a postdoc researcher at EPFL and ETH Zurich with Prof. Daniel Kuhn and Prof. Andreas Krause. I obtained PhD in Operations Research from University of Illinois Urbana-Champaign with Prof. Xin Chen and Prof. Niao He. 

欢迎广大师生参加!