Eigenvalue Spectra of Functional Networks in fMRI Data and Artificial Models

Abstract

In this work we provide a spectral comparison of functional networks in fMRI data of brain activity and artificial energy-based neural model. The spectra (set of eigenvalues of the graph adjacency matrix) of both networks turn out to obey similar decay rate and characteristic power-law scaling in their middle parts. This extends the set of statistics, which are already confirmed to be similar for both neural models and medical data, by the graph spectrum.