February 22, 2023 - Faculty Candidate Seminar - Next-Generation Training Algorithms with Deterministic Global Optimality Guarantees

Dr. Kaixun Hua
University of Massachussetts
 

Wednesday, February 22, 2023 at 1:00 pm
Room EV003.309

Abstract

Due to the NP -hardness of many machine learning problems such as clustering, decision tree, and neural network, one primary belief is that solving ML problems to global optimality is computationally intractable. Based on this, the dominant methods in practice are based on either heuristics or local optimization algorithms, producing sub- optimal solutions. The other specious belief is that, in the era of big data, simple models with strong interpretability (e.g., decision trees) cannot have the same predictive power as black -box models. In this talk, we propose a new training framework with deterministic global optimality guarantees for large-scale machine learning problems. We illustrate its usage on two typical machine learning tasks, centroid-based clustering and sparse decision tree. Numerical results on extensive datasets demonstrate that our training framework challenges the prevailing mindsets with the following conclusions. 1) With the detection of problem structure, solving large-scale machine learning problems to global optimality is computationally feasible. 2) When the global optimality is obtained, the performance of highly interpretable ML models on large datasets can be substantially improved.

Biography

Kaixun Hua is a Postdoctoral Research Fellow at the Institute of Applied Mathematics and Department of Chemical and Biological Engineering, University of British Columbia. His research focuses on applying clusterability and deterministic global optimization tools to address challenges when building scalable trustworthy machine learning systems. He obtained his M.S. and Ph.D. degree in Computer Science from the University of Massachusetts Boston in 2016 and 2019, respectively, under the supervision of Dan A. Simovici. Before that, he received an M.Eng. degree in Systems Engineering from Cornell University in 2013 and a B.S. degree in Electrical and Computer Engineering from Shanghai Jiao Tong University in 2012.