Clustering Algorithm Based on Outlier Detection for Anomaly Intrusion Detection

Abstract

Many experiments show that outliers have important implications for clustering. However, Most of the clustering algorithm ignores to compute outliers, or does not detect outliers well. In this paper, we present a local deviation factor graph-based (LDFGB) algorithm. We measure the effectiveness of the algorithm by detection rate, false positive rate, false negative rate, time overhead, and so on. This algorithm can accurately detect outliers by calculating the relative distance between the data nodes. It can detect any shape of the cluster and still keep high detection rate for detecting known and unknown attacks. Using KDD CUP99 data sets, the experimental results show that this method is effective for improving the detection rates and false positive rates.