Immune Network Based Anomaly Detection Algorithm

Xinlei Hu, Zhengxia Wang, Yunbing Hu

Abstract: Inspired by immune network theory, this paper proposes an immune network-based anomaly detection algorithm (INADA), which effectively solves the problems of detection based on group theory. INADA is divided into three stages: In the immune network generation stage, mature detectors are generated through the negative selection algorithm. Each mature detector is not isolated. In the detector training phase, the mature detector is not only stimulated by the antigen during the life cycle, but also stimulated and suppressed by other detectors in the network. In the detection stage, the UCI standard data set is used as the test sample to determine whether all mature detectors in the immune network match the test sample. The result shows that the INADA algorithm has high detection rate and low false alarm rate.