A0742
Title: Joint graphical lasso with regularized aggregation for high-dimensional time series with long-memory
Authors: Jongik Chung - University of Central Florida (United States) [presenting]
Qihu Zhang - Fei Tian College - Middletown (United States)
Cheolwoo Park - Korea Advanced Institute of Science and Technology (Korea, South)
Abstract: The purpose is to outline methods for estimating multiple precision matrices for high-dimensional long-memory time series within the framework of Gaussian graphical models, particularly focusing on analyzing functional magnetic resonance imaging (fMRI) data collected from multiple subjects. The aim is to estimate individual brain networks and a collective structure representing a group of subjects. A method is proposed that utilizes regularized aggregation to simultaneously estimate individual and group precision matrices, assigning varying weights to each individual based on their outlier status within the group. The convergence rates of the precision matrix estimators are examined across different norms and expectations, evaluating their performance under sub-Gaussian and heavy-tailed assumptions. The efficacy of the methods is demonstrated through simulations and real fMRI data.
