From random failures to targeted attacks in network dismantling

Abstract

It is well-known that real-world systems, modeled as complex networks, are mostly robust against random failures but susceptible to targeted attacks. In this study, we propose a novel perspective to solve the network dismantling problem. Instead of designing an effective attack from scratch, we show how knowledge extracted from random failures in the network leads to extremely effective attacks. This observed connection between random failures and targeted attacks is striking on its own. Experiments on a wide range of networks show the efficacy of our novel method for network dismantling, providing an excellent trade-off between attack quality and scalability. We believe that our contribution also stimulates research in related domains, including social network influence analysis, spreading dynamics in networks, and efficiency considerations.