A COMPARATIVE PERFORMANCE STUDY OF MACHINE LEARNING ALGORITHMS, FOR EFFICIENT DATA MINING MANAGEMENT OF INTRUSION DETECTION SYSTEMS
Salihu Alhasan, Ebenezer Akinyemi Ajayi, Daniel Dauda Wisdom · International Journal of Engineering Applied Sciences and Technology · 2020
Data mining provide decision support for intrusion management, and also help in Intrusion DetectionSystem (IDS) in detecting of new vulnerabilities and intrusions by discovering unknown patterns of attacks or intrusions.In this paper, we have compared four algorithms of Machine Learning models which are namely: Naives Bayes (NB), Decision trees (J48), Support Vector machines (SVN), and Sequential Machine optimization (SMO).The realistic models were evaluated and compared using data sets as obtained from NSL-KDDCup.The method takes into consideration the relative sizes of the classes to each other in the dataset.which allows the user of the IDS to evaluate how well they will predict the classes given the distribution of the dataset.In addition, graphs were plotted in order to efficiently analyze the results obtained of the various models depicted in Figures.The simulation results were obtained using WEKA.The simulation parameters were filtered into various tables as depicted also in Figures so as to achieve a visual conception.Finally, we carried out various computational analyses to give us semblance of graphical constructions that are related to some parameters (time, kappa characteristics, ROC etc.) of our experiments respectively.