Seminar 3

Speaker: Dr. Chul Ho-Lee, Computer Science and Dr. Young Ju Lee, Mathematics

Abstract: Graph clustering is a fundamental problem in machine learning and has found a wide range of applications. In particular, spectral clustering has been the most popular unsupervised graph-clustering algorithm, and it has still been actively extended to various graph clustering problems to improve the quality and fairness of clustering. However, the current literature is lopsided as the research has been mostly limited to under the framework of spectral clustering that requires computing the eigenvectors of graph Laplacian matrices, which can be computationally expensive. Thus motivated, in this project, we aim to develop a computationally efficient framework and scalable algorithms for a class of recent graph clustering problems, which do not require computing the eigenvectors but achieve scalable and high-quality fair clustering. To this end, we focus on the following inter-related research tasks: (1) to develop an efficient framework to solve constrained graph clustering problems, (2) to devise a novel method of incorporating the notion of fairness into the problems, and (3) to develop parallel algorithms and implementations for scalable fair clustering

Date: November 1st, 2024, Friday, 11:30 -13:00.
11:30-11:45: Registration and lunch
11:45-12:00: Opening Keynote: AI and Trust, Emily LaRosa, Michigan State University 
12:05-12:35: Innovation Seed Award Seminar by Drs. Chul Ho-Lee and Young Ju Lee, TXST
12:40-13:00: Funding Opportunities Overview, Evy Gonzales and Open Discussion