Identification of Critical Nodes in Dynamic Systems Based on Graph Convolutional Networks

Identifying the critical nodes in a dynamic complex system is of great significance for an in-depth understanding of the interactions between system components, and is conducive to influence and control the dynamic behavior of complex systems. For complex dynamic systems containing time-series evolutionary structures and nonlinear behaviors, the traditional critical node identification algorithm that only considers the static topology structure is no longer applicable. In this work, we introduce deep learning technology to this challenging problem and propose a critical node identification model based on graph convolutional networks. The model comprehensively considers the dynamic evolution of the network structure, the nodes' attribute information, and interaction behavior information. The application results on the real datasets show the effectiveness of the proposed model.