Uncertainty quantification and rank selection for low-rank matrices from incomplete and noisy observations

Mon, May 6, 2019, 12:30 pm
Location: 
B205 Engineering Quadrangle
Speaker(s): 

Abstract: 

In this talk, Professor Xie will first present a new statistical framework to quantify uncertainty (UQ) for recovering low-rank matrices from incomplete and noisy observations. The proposed framework reveals several theoretical insights on the connection of coherence with UQ and code design (e.g., Latin squares and Kerdock codes). Using such insights, we develop an efficient posterior sampler for UQ and active sampling guided by UQ. She will then talk about a sequential statistical testing procedure to determine the "true" rank of a low-rank matrix. Joint works with Simon Mak, Alexander Shapiro, Shaowu Yuchi, and Rui Zhang at Georgia Institute of Technology.

Bio:

Yao Xie is an Assistant Professor at Georgia Institute of Technology in the H. Milton Stewart School of Industrial and Systems Engineering since 2013 (she has been promoted to Associate Professor with tenure effective of August 15, 2019). She received her Ph.D. in Electrical Engineering (minor in Mathematics) from Stanford University in January 2012, M.Sc. in Electrical and Computer Engineering from the University of Florida in 2006, and B.Sc. in Electrical Engineering and Computer Science from University of Science and Technology of China (USTC) in 2004. She was a Research Scientist at Duke University from 2012 to 2013. Her research interests are statistics, machine learning, and signal processing, in providing the theoretical foundation as well as developing computationally efficient and statistically powerful algorithms. She received the National Science Foundation (NSF) CAREER Award in 2017.

This seminar is supported with funds from the Korhammer Lecture Series.