题　目：Algorithmic Foundations of Risk-averse Optimization for Trustworthy AI

嘉　宾：蓝光辉 教授 佐治亚理工学院

主持人：许欢 教授 上海交通大学安泰经济与管理学院

时　间：2025年6月13日（周五）10:00-11:30

地　点：安泰楼B207室

内容简介：

Over the past two decades, stochastic optimization has made remarkable strides, driving its widespread adoption in machine learning (ML) and artificial intelligence (AI). However, most existing models prioritize minimizing expected loss, often leaving AI-driven decisions vulnerable to costly or catastrophic failures and raising concerns about their trustworthiness in high-stakes applications. Risk-averse optimization provides a principled approach to mitigating such vulnerabilities, yet its adoption remains limited due to the lack of scalable and efficient solution methods. In this talk, I will present the algorithmic foundations of risk-averse optimization, focusing on an important class of  L_P risk measures. I will introduce novel lifted reformulations that enhance tractability, develop stochastic approximation algorithms with provable convergence guarantees, and establish fundamental complexity limits. These advances provide a deeper theoretical understanding of risk-aware decision-making, laying the groundwork for more robust and trustworthy AI systems.

演讲人简介：

Guanghui (George) Lan is an A. Russell Chandler III Chair and professor in the H. Milton Stewart School of Industrial and Systems Engineering at the Georgia Institute of Technology. Prior to returning to Georgia Tech, where he earned his Ph.D. in August 2009, Dr. Lan served on the faculty of the Department of Industrial and Systems Engineering at the University of Florida from 2009 to 2015. His primary research interests lie in optimization, machine learning, and reinforcement learning, with applications in sustainability and healthcare. His academic honors include the INFORMS Frederick W. Lanchester Prize (2023), the INFORMS Computing Society Prize (2022), the National Science Foundation CAREER Award (2013), First Place in the INFORMS Junior Faculty Interest Group Paper Competition (2012), and recognition as a finalist for the Mathematical Optimization Society Tucker Prize (2012).   Dr. Lan serves as an associate editor for Mathematical Programming, SIAM Journal on Optimization, Operations Research, and Computational Optimization and Applications. He is also an associate director of the Center for Machine Learning at Georgia Tech.  

欢迎广大师生参加！