A new complex network robustness attack algorithm

Complex networks have been widely used in many systems of bank, social networks and smart grid, etc. A vital quantitative criteria for complex networks is to measure the robust performance in those systems. It observes the response of the networks when nodes or links are removed from potential threats. Most of the existing works focus on the enhancement of the robustness itself, but not considering the completeness of the possible removal attacks from the viewpoint of adversary. We first put forward a new ensemble learning based critical node removal attack (ECNRA) algorithm, and prove that the damage of the robustness using the proposed algorithm is worse than that using the degree attack and random node attack. In order to solve the uncertainty of the network robustness evaluation algorithm proposed before, we propose a new complex network robustness evaluation algorithm. Finally we apply our attack algorithm to a Bitcoin OTC network. The result shows that our algorithm is better than the other two algorithm.