Recursive Feature Elimination and Clustering Technique for Network Anomaly Detection

Nowadays the number of people using the network has increased with the increase in security threats. So, anomaly detection technique Recursive feature elimination and clustering technique for network anomaly detection is framed with two phases: training and detection. The training phase works on three algorithms computed one after another. The three algorithms are Recursive feature elimination method, Cuckoo search optimization and K-means clustering. The multi-objective functions such as Silhouette index and the mean square error are employed. Then classification measure and anomaly detection measure is calculated. In the detection phase, a fuzzy logic toolbox is used to detect the anomaly data objects. Detection Rate and False Alarm Rate are the claiming factors in construction of the Anomaly Detection System. A agreeable system should have a high Detection Rate and a low False Alarm Rate in order to efficiently identify the anomaly data objects. The metrics accuracy, false positive-rate, F-measure and detection rate are evaluated. The proposed system achieves high Detection Rate with Low False Alarm Rate.