Analysis of Machine Learning based Techniques for Intrusion Detection System

Abstract

One of the solutions used against malicious threats is the Intrusion Detection System (IDS). In addition, attackers still keep adjusting their instruments and tactics. It is still a difficult job to incorporate an agreed IDS scheme, however. Several studies have been carried out and tested in this paper to test different machine learning classifiers based on the KDD intrusion dataset. In order to test the chosen classifiers, multiple output parameters were successfully computed. In order to increase the intrusion detection system's detection rate, the emphasis was on false negative and false positive performance indicators. The tests carried out found that the decision table classifier obtained the lowest false negative score, while the highest average accuracy rating was achieved by the random forest classifier.