Discrimination of unknown complex network based on the information of local nodes

At present, the study on complex network and game theory has been spread across the fields of science and engineering. The research contents on these topics mainly focus on the evolution results of structured populations based on the determined network structures and game types, which may not be open information to the interacting players. Therefore, how to obtain the maximum benefit in an unfamiliar game environment for an individual new to the population is a topic that has received little attention but is of great significance. In this paper, we explore how to use the minimum amount of known information available to realize the identification of large-scale unknown networks. We simulate real large structured populations with five classical large complex networks and hide their real structures. Then, we propose an improved k-nearest neighbor (KNN) algorithm, which can infer the real network structures with high accuracy only based on the degree and clustering coefficient information of a few local nodes on the network. The results and method presented here provide some hints about the study on the evolutionary game process and swarm intelligence in networked populations.