Revealing effective classifiers through network comparison

Abstract

The ability to compare complex systems can provide new insight into the fundamental nature of the processes captured in ways that are otherwise inaccessible to observation. Here, we introduce the $n$-tangle method to directly compare two networks for structural similarity, based on the distribution of edge density in network subgraphs. We demonstrate that this method can efficiently introduce comparative analysis into network science and opens the road for many new applications. For example, we show how the construction of a phylogenetic tree across animal taxa according to their social structure can reveal commonalities in the behavioral ecology of the populations, or how students create similar networks according to the University size. Our method can be expanded to study a multitude of additional properties, such as network classification, changes during time evolution, convergence of growth models, and detection of structural changes during damage.

11 pages, 4 figures, includes Supplementary Information. For the source code, see http://www-levich.engr.ccny.cuny.edu/~lazaros/research_comparison.html

References in corpus (3)

• Biological network comparison using graphlet degree distribution
• Community structure of complex software systems: Analysis and applications
• Subgraph Ensembles and Motif Discovery Using a New Heuristic for Graph Isomorphism

Cited by in corpus (5)

• Analysis of heat kernel highlights the strongly modular and heat-preserving structure of proteins
• Propinquity drives the emergence of network structure and density
• Equation-free analysis of a dynamically evolving multigraph
• One Node at a Time: Node-Level Network Classification
• Network-ensemble comparisons with stochastic rewiring and von Neumann entropy