Oct 15, 2020

Hsin-Hsiung Huang

University of Central Florida, Department of Statistics & Data Science

3:00 pm in Zoom (Meeting ID: 917 7737 6704)

Unsupervised Bayesian matrix-variate clustering, and robust discriminant analysis

In this talk, I will present my recent research studies supported by the NSF ATD grant. Our research team has developed a Bayesian clustering framework which uses the distance-based Chinese restaurant process, matrix-variate Gaussian, regression means, and a robustness weight against outliers from heavy tailed distribution identifies the groups of subjects in networks. The developed unsupervised clustering algorithms which have high accuracy with low computational complexity provide real-time solutions to detect dynamic pattern changes.