Edge correlations in spatial networks

TitleEdge correlations in spatial networks
Publication TypeJournal Article
Year of Publication2016
AuthorsFriedman, E., Landsberg A. S., Owen J., Hsieh W., Kam L., & Mukherjee P.
Published inJournal of Complex Networks
Page(s)1-14
Date Published06/2016
Abstract

In this paper, we show that many real-world spatial networks—including demographic migration networks, social networks and human brain networks—possess ‘spatial edge correlations’ that profoundly affect the network's underlying structure but which go largely undetected by standard network analysis methods. Even methods designed to characterize the spatial attributes like distribution of edge lengths in a network or the costliness of edge lengths (i.e. long-distance edges tend to be more costly in terms of efficiency, energy, price, etc. than shorter edges) miss the presence of spatial edge correlations. We introduce two network measures designed to detect and quantify these edge correlations, and discuss their implications. We show that even highly simplified models that explicitly take account of these edge correlations can achieve dramatic (potentially order-of-magnitude) improvements over standard network models. We demonstrate the utility of this approach by explaining certain anomalies seen in newly available, high-resolution human brain networks. The methods introduced here are very general, with potentially broad applicability to spatial networks across diverse branches of network science.

ICSI Research Group

Research Initiatives