Infering Complex Network’s Topology Based On Point Processes

The dynamics and function of a network are influenced by the topology of the network. So it’s becoming important to develop a effective method of inferring network structure. In the past several years, topology identification of complex networks has received intensive interest and made great progress. Based on the theory of stochastic process, dynamic correlation analysis and conditional Granger-Causality, a new method to detect the underlying topology of a network from the point processes of each node is proposed. The superiority of the proposed method is justified by the90%above identification rate generated in the simulation experiments of Virus Spreading Model and Integrate-and-Fire Model. In the end, this method is applied to analyse the fMCI data for discussion of neural network’s topology.