ACO-Based Neural Network to Enhance the Efficiency of Network Controllability of Temporal Networks

The controllability of temporal networks has been one of the most important challenges in this type of network over the last decade. The main goal of network controllability processes is to find the minimum set of control nodes in such a way that all network nodes can be controlled by them. This problem is NP-hard in the temporal networks. In this paper, a controllability method is proposed to improve the efficiency of the controllability process on temporal networks. In the proposed method, a population method based on the ant colony optimization (ACO) algorithm is proposed, which is compatible with temporal networks. Due to the temporal nature of the controllability processes in temporal networks, the ACO algorithm is adapted temporally. Also, due to the time-consuming controllable processes in temporal networks and in order to increase the efficiency of the ACO algorithm, a backpropagation neural network has been used, which finds the minimum driver node set of the network based on the layered model in order to fully control the network nodes. The results of the implementation of the proposed method on real-world datasets demonstrate that the proposed ACO-BPNN method works stably and with high efficiency on high-volume datasets. By comparing the efficiency of the proposed method with conventional controllability methods, it is found that the proposed method has performed better in terms of the speed of execution and the length of the minimum driver node set.