Statistics and Its Interface

Volume 8 (2015)

Number 3

Exponential random graph models for networks resilient to targeted attacks

Pages: 267 – 276

DOI: https://dx.doi.org/10.4310/SII.2015.v8.n3.a2

Authors

Jingfei Zhang (Department of Statistics, University of Illinois at Urbana-Champaign, Champaign, Il., U.S.A.)

Yuguo Chen (Department of Statistics, University of Illinois at Urbana-Champaign, Champaign, Il., U.S.A.)

Abstract

One important question for complex networks is how the network’s connectivity will be affected if the network is under targeted attacks, i.e., the nodes with the most links are attacked. In this paper, we fit an exponential random graph model to a dolphin network which is known to be resilient to targeted attacks. The fitted model characterizes network resiliency and identifies local structures that can reproduce the global resilience property. Such a statistical model can be used to build the Internet and other networks to increase the attack tolerance of those networks.

Keywords

exponential random graph model, global efficiency, Markov chain Monte Carlo, maximum likelihood estimation, network robustness, random graphs

Published 17 April 2015