CMStatistics 2016: Start Registration
View Submission - CMStatistics
B1472
Title: Zoom-in/out joint graphical lasso for different coarseness scales Authors:  Eugen Pircalabelu - Université catholique de Louvain (Belgium) [presenting]
Abstract: Graphical models are estimated from data obtained at $K$ different coarseness scales. Starting from a predefined scale $k^{*}\leq K$ the method zooms in or out over scales on particular edges, thus estimating graphs with similar structures, but different levels of sparsity. The graphs are jointly estimated at all coarseness scales and evaluate the evolution of the graphs from the coarsest to the finest scale or vice-versa. We select an optimal coarseness scale to be used for further analyses. The method is motivated by fMRI datasets that do not all contain measurements on the same set of brain regions. For certain datasets some of the regions have been split in smaller subregions and this gives rise to the framework of mixed scale measurements where the purpose is to estimate sparse graphical models. We accomplish this by pooling information from all subjects in order to estimate a common undirected and directed graph at each coarseness scale, accounting for time dependencies and multiple coarseness scales and by jointly estimating the graphs at all coarseness scales. The applicability of the method goes beyond fMRI data, to other areas where data on different scales are observed and where the joint estimation of graphs is desired.