EVALUATION OF THE EFFECTIVENESS OF ANOMALY IDS BASED ON THE CLUSTERING ALGORITHM AND DATA MINING TECHNIQUES

Abstract

The purpose of this paper is to examine the feasibility of clustering-based approach to anomaly-based intrusion detection systems (IDS). The examined methodology includes a 2-means clustering algorithm with and without data mining techniques, i.e. classification trees. With purpose to evaluate the effectiveness of the methodology, Jaccard index was applied. Davies-Bouldin index, Dunn index and C-index were applied in order to compare the performance results of the two models.