Abstract:
|
In this paper, we propose a procedure to find differential networks between two graphs from high-dimensional multivariate data. We estimate the precision matrices and their differences by solving a penalized regression-based least square problem. We assume sparsity on differences between two graphs, not graphs themselves. Thus, we impose L2 penalty on partial correlations from elements of precision matrices while L1 penalty on their differences. We apply the proposed procedure to finding differential functional connectivity in the resting brain between normal and Alzheimer's disease patients.
|