Towards a behavioral network intrusion detection system based on the SVM model

A behavioral IDS (Intrusion Detection System) is an effective tool for the detection of computer network intrusions, especially the most recent ones. However, the behavioral IDS have a very high false alarm rate compared to traditional IDS that use a signature base for each intrusion. In this paper, we propose an original method of network intrusion detection using machine learning techniques. Our method is based on a behavioral IDS capable of identifying new attacks without using a signature database. We use the SVM (Support Vector Machine) classification model with two cores (Polynomial and Gaussian). This model is trained and tested with the UNSW-NB15 dataset. We have obtained interesting results in terms of detection rate (DR) in comparison with other classification models (ANN, RepTree, Random Forest, MLP).