Intrusion Detection Using Principal Component Analysis and Support Vector Machines
Anukriti Mishra, Albert M. K. Cheng, Yunpeng Zhang
Abstract
Intrusion detection nowadays is an integral part of network security. With the advancement in machine learning technologies, it has become easier than ever to construct a model to detect intrusions. While the accuracy of already existing models is quite high, another aspect of intrusion detection is the computation time of the model. As the speed of the network is increasing rapidly, the intrusion detection system (IDS) should be able to keep up with the high influx of network connections, and with them the potential attacks. In this paper, we present a supervised machine learning model to detect intrusion in the network. We have created a supervised classification model using principal component analysis (PCA) for dimensionality reduction in combination with support vector machines (SVM) for improved attack detection and faster computation time. Evaluation of the model is done using the UNSW-NB15 data set. Test study shows that the proposed model was able to improve model training and testing time by 33.75% for binary classification and 33.91% for multi-class classification with an overall accuracy of 99.99% and 99.97% respectively. Classification result compared to other model have also been presented.