An Intrusion Detection Method Based on Outlier Ensemble Detection

摘要

In this paper, we try to bring the concept of Ensemble into Outlier Detection. Two Outlier mining algorithms are ensembled: one based on similar coefficient sum and the other based on Kernel Density. An anomaly detection approach based on Voting Mechanism is proposed and applied into Intrusion Detection. We convert the character feature into numerical value by code mapping and use Principal Components Analysis(PCA) to reduce dimension. We apply this technique on KDD99 data set and get satisfactory results.