Network Intrusion Detection Using Genetic Clustering

Abstract. We apply the Unsupervised Niche Clustering (UNC), a genetic niching technique for robust and unsupervised clustering, to the intrusion detection problem. Using the normal samples, UNC generates clusters sumarizing the normal space. These clusters can be characterized by fuzzy membership functions, that are later aggregated to determine a level of normality. Anomalies are identified by their low normality levels.

• Type: Conference
• Year: 2004
• Where: GECCO
• Authors: Elizabeth Leon, Olfa Nasraoui, Jonatan Gómez
• Keywords: GECCO 2004, Intrusion Detection Problem, UNC Generates Clusters, Unsupervised Clustering