Publication

Abstract : Data mining techniques are commonly used for designing intrusion detection systems. The rule-based supervised classifiers play a prominent role in the intrusion detection process and infact, empower the detectors for quick discovery of intrusions in a network of computing devices. However, the design architecture of these classifiers has a significant impact on the speed and accuracy of the detection process. This paper analyzes various rule-based classifiers for designing effective intrusion detection systems. The rule-based classifiers are explored extensively on the basis of detection accuracy, false-positive rate, and ROC value. Three widely used intrusion datasets such as NSL-KDD, ISCXIDS2012, and CICIDS2017 datasets have been considered for effective analysis of rule-based classifiers. Subsequently, the article proposed an architecture considering the best classifier for designing intrusion detection systems.