CBE/ENGR 225 Seminar Presents: Song-Yun Oh, Ph.D., Assistant Professor, Department of Statistics and Applied Probability
Sang-Yun Oh, Ph.D.
Dept. of Statistics and Applied Probability
University of California, Santa Barbara
Tuesday, May 2nd, 2017
ESB, Room #2001
*Cookies and Coffee will be provided*
Extreme Scale Graphical Model Selection and Applications in Data-Driven Discovery
Abstract: Data-driven scientific discovery (DDD) approaches are becoming increasingly popular in various research communities. In DDD, unsupervised learning methods are often used as hypothesis generating tool by detecting patterns of interest in the data. Graphical models framework is an unsupervised learning approach useful for capturing relationships in high dimensional data. In particular, undirected graphical models, also known as Markov Random Fields, can find most significant (pairwise) associations accounting for other variables. This talk will give an overview of our projects in graphical model selection methodology and applications. The first part of the talk will describe our proposed graphical model selection method for extremely high dimensional data. The second part will describe two applications of undirected graphical models in neuroscience and genomics.