Beijing-Saint Petersburg Mathematics Colloquium (online)
Abstract
Biological networks often change under different environmental and genetic conditions. In this paper, we model the network change as the difference of two precision matrices and propose a novel loss function called the D-trace loss, which allows us to directly estimate the precision matrix difference without attempting to estimate precision matrices. Under a new irrepresentability condition, we show that the D-trace loss function with the lasso penalty can give consistent estimators in high-dimensional settings if the difference network is sparse. A very efficient algorithm is developed based on the alternating direction method of multipliers to minimize the penalized loss function. Simulation studies and a real data analysis show that the proposed method outperforms other methods.
Bio
Dr. Ruibin Xi is an associate professor at School of Mathematical Sciences, Peking University. He obtained his PhD from Washington University in St. Louis and received post doc training at Harvard Medical School. His main research interests include statistical analysis of big biological data, cancer genomics, network and graphical models, Bayesian analysis and high-dimensional statistics.
