Using Classification Algorithms in Building Models for Network Intrusion Detection

Abstract

Network Intrusion Detection Systems are considered as one part of important and basic components of the security and protection system of computer networks. This issue encouraged researchers to engage in improving and enhancing the performance of these systems. Most of researches depend on KDDCup'99 Dataset, which contains structured data with four types of network attacks U2R, R2L, DOS and Probing. It has been recently focused on using Data Mining techniques in intrusion detection research since it employs a wide range of classification and clustering algorithms that can be used in building models to detect attacks accurately. Most previous studies and researches have shown that most classification detection models are based on one algorithm to detect all four types of attacks, and this in turn forms a starting point for aggregating several classification models to be used in detecting attacks and intrusions. In this paper we will compare between the performances of a set of classification algorithms implemented in WEKA to build detection models, and finally the constructed models will be employed in building a synthesis model that can detect attacks effectively at an acceptable level of accuracy.
Copyright © 2019 Praise Worthy Prize - All rights reserved.