A Network Intrusion Detection Algorithm Based on Outlier Mining

A spectral clustering and Local Outlier Factor (LOF) based outlier mining algorithm, which is aiming to solve network intrusion detection problem, is proposed in this paper. First of all, the structure of similarity matrix method in spectral clustering is used for data preprocessing to find out the smaller similarity objects. During this process, the pruning of the outliers is completed, and a set of candidate outliers is obtained. Then, we calculate the local outlier factor of each data object in this set through LOF algorithm. And the final results of detection of outliers are acquired. The experimental results show that the proposed algorithm improves the accuracy of detecting outliers and the effectiveness of network intrusion detection.