Multi-Objective Optimization of Critical Node Detection Based on Both Cascading and Non-Cascading Scenarios in Complex Networks

Critical node detection is an important task to analyze network vulnerability and measure network security. Most existing methods either focus on non-cascading scenario from the perspective of network static properties or focus on cascading scenario from the perspective of network dynamic behavior. The problem of critical node detection is unexplored in both scenarios simultaneously. To this end, we formalize the critical node detection problem into a multi-objective problem in the two scenarios at the same time. Then, a multi-objective critical node detection algorithm named MOBSCND is proposed to effectively solve the transformed problem, which can provide multiple critical node solutions for decision-makers in both cascading and non-cascading scenarios. In MOBSCND, we design an effective initialization strategy, which can produce better initial individuals while ensuring population diversity. In addition, a local search strategy is proposed to accelerate the convergence of the population and improve the quality of population individuals. The experimental results on 12 datasets show that the convergence and diversity of the solution set obtained by the proposed method are all superior to the baseline algorithms.