An elitism-based multi-objective evolutionary algorithm for min-cost network disintegration

Abstract

Network disintegration or strengthening is a significant problem, which is widely used in infrastructure construction, social networks, infectious disease prevention and so on. But most studies assume that the cost of attacking anyone node is equal. In this paper, we investigate the robustness of complex networks under a more realistic assumption that costs are functions of degrees of nodes. A multi-objective, elitism-based, evolutionary algorithm (MOEEA) is proposed for the network disintegration problem with heterogeneous costs. By defining a new unit cost influence measure of the target attack node and combining with an elitism strategy, some combination nodes’ information can be retained. Through an ingenious update mechanism, this information is passed on to the next generation to guide the population to move to more promising regions, which can improve the rate of convergence of the proposed algorithm. A series of experiments have been carried out on four benchmark networks and some model networks, the results show that our method performs better than five other state-of-the-art attack strategies. MOEEA can usually find min-cost network disintegration solutions. Simultaneously, through testing different cost functions, we find that the stronger the cost heterogeneity, the better performance of our algorithm.