Intrusion detection using copula functions in computer networks

Today, with the rapidly developing technology, the interest in technology in the world is increasing day by day. These rapid developments in technology has brought many undesired attacks such as cyber attack, unauthorized access and digital piracy. Intrusion detection systems have been often used to prevent such attacks. In this study, attack detection has been carried out using the most used machine learning classifiers and copula-based classifiers. As machine learning classifiers; Decision Trees, Ensemble Learning and Support Vector Machines classifiers have been preferred. Classification has been performed on the KDD'99 data set using these three classification techniques. As Copula-based; gumbel, independent, clayton, gaussian, student's-t and frank classifiers have been preferred. Using these classifiers, the classification process has been carried out on the KDD'99 data set. In the classification stage, 10-fold cross-validation technique has been used, and the best performance rate has obtained gaussian copula based classifier with 99.41% on the KDD'99 data set. As a result, it has been observed that copula based classifiers have achieved an effective success in intrusion detection.