PROJECT TITLE :

Phase Transitions and a Model Order Selection Criterion for Spectral Graph Clustering - 2018

ABSTRACT:

One in every of the longstanding open issues in spectral graph clustering (SGC) is the thus-called model order choice problem: automated selection of the correct range of clusters. This is reminiscent of the matter of finding the quantity of connected elements or communities in an undirected graph. We propose an automatic model order selection (AMOS), a resolution to the SGC model selection downside below a random interconnection model employing a novel selection criterion that's based on an asymptotic section transition analysis. AMOS will additional usually be applied to discovering hidden block diagonal structure in symmetric non-negative matrices. Numerical experiments on simulated graphs validate the part transition analysis, and real-world network data are used to validate the performance of the proposed model choice procedure.