Anomaly Detection in Multiplex Networks

Detecting anomalies in social is a vital task, with numerous high impacted social networks such as WWW, Facebook, Twitter and so on. There are multiple of techniques have been developed for detecting outliers and anomalies in graph data. More recently, the area of multiplex networks has extended a considerable attention among researchers for more concrete results. A Multiplex network is a network, which contains multiple systems of the same set of nodes and there exists various types of the relationship among nodes. In this paper, we discover the anomalies across numerous multiplex networks. By anomalies or outliers means nodes, which behave abnormal or suspicious in the system. Compared to single layer networks, the outliers' nodes may found into many layers of the multiplex network and find anomalies in the multiplex network is still untouched. From this study, we propose a new metric called cross-layer anomaly detection (CAD). The CAD is a measure, which detects the anomalies in the multiplex network. For experiments, we make use of two real-world multiplex networks. We compare the results of our proposed metric with other similar methods, and we get encouraging and similar results.