Implementing a network intrusion detection system using semi-supervised support vector machine and random forest

Abstract

Network security is an important aspect for any organization to keep their information systems secure. A Network Intrusion Detection System (NIDS) is an aid to secure the network by detecting abnormal or malicious traffic. In this paper, we applied a Semi-supervised machine learning approach to design a NIDS. We implemented semi-supervised Support Vector Machine (SVM) and semi-supervised Random Forest (RF) classifiers to classify the NSL-KDD dataset. We have classified the dataset in both binary and multiclass. We have also implemented a Genetic Algorithm (GA) approach to select the optimal features from the original features set. Results show that the random forest algorithm produces a better result than SVM using semi-supervised learning method. Also, the results show that applying the GA in SVM produces a better result than without using GA, and so does using GA in Semi-supervised Random Forest.