Performance evaluation of machine learning algorithms for detecting abnormal data traffic in computer networks

Abstract

Detection and classification of abnormal data in computer network traffic is a very important cyber-security problem. In this study, the aim is to determine the methods with a high rate of success in classifying the data to minimize the processing power required to detect abnormal data traffic and increase the performance in classification. For this reason, the study includes a performance evaluation of machine learning algorithms used for detecting harmful data traffic. NSL-KDD dataset was used for performance tests and evaluations. The classification performance of the methods used for data test was compared. As a result, the Random Forest method achieved the highest classification accuracy. © 2020 IEEE.