2025年第74期(总第1115期)

演讲主题：Learning-driven optimization

主讲人：Jin-Kao Hao 法国昂热大学特级教授

主持人：吴庆华 管理科学系教授

活动时间：2025年11月03日（周一）14:00-15:30

活动地点：管院大楼119教室

主讲人简介：

Dr. Jin-Kao Hao is a professor of computer science at the University of Angers in France. His research focuses on designing effective algorithms and intelligent computational methods for solving large-scale combinatorial search problems. Alongside his collaborators, he has developed some of the most effective solutions to several classic NP-hard problems, such as graph coloring, knapsack problems and quadratic optimization, as well as real-world applications in mobile networks and transportation. He has co-authored more than 330 peer-reviewed publications and has served on the editorial boards of some 10 international journals and the program committees of over 250 international conferences.

活动简介：

This lecture provides an introduction to the latest trends in learning-driven optimization. Unlike the end-to-end approach, which considers optimization as a machine learning problem, learning-driven optimization uses learning methods to guide the search process, helping the algorithm to make informed decisions during the search. We will illustrate this approach through three methods developed in recent years: frequent pattern-based search with application to quadratic assignment, probabilistic learning based search with application to grouping problems, and deep learning guided memetic search with application to graph coloring. We will also discuss recent advances in GPU-accelerated combinatorial optimization techniques.