CRoNoS & MDA 2019: Start Registration
View Submission - CRONOSMDA2019
A0251
Title: Stability of a network inference procedure in high-dimension Authors:  Emilie Devijver - CNRS (France) [presenting]
Melina Gallopin - Universite Paris Sud (France)
Remi Molinier - Universite Grenoble Alpes - Institut Fourier (France)
Abstract: Network inference is widely utilized to evaluate and represent dependencies between continuous variables. Gaussian graphical models have been developed the last years, tackling the high-dimension problem through several assumptions. The focus is on the stability of a procedure called shock, which infers a modular network represented by a block-diagonal covariance matrix. This structure has strong advantages, among such reducing the dimension, facilitating the interpretation and being stable. The stability of the procedure is supported by strong theoretical guarantees based on topological tools, intensive simulations and real data analysis.