Intrusion Detection System Using Feature Selection and Machine Learning Techniques

Abstract

Intrusion is a serious problem in computer network security. With the rapid rise in various applications in the networking domain, intrusion attacks on networks increase that cannot be detected by humans effectively. To prevent such threats a system called Intrusion Detection System is designed. In this paper, five Machine Learning techniques i.e., K-Nearest Neighbours, Multiple Layer Perceptron, Decision Tree, Naïve Bayes, and Random Forest classifiers are used for Intrusion Detection System. For the proposed work, the NSL-KDD dataset with Random Forest Feature Selection Technique has been used for the training and testing of the Intrusion Detection System. Results demonstrated that Random Forest attains the classification accuracy of 99.60% which is the highest in comparison to other machine learning models and the least accuracy of 87.53% has been achieved by Naïve Bayes. Our results also demonstrate that 37.87 seconds is the highest training time required by the Multiple Layer Perceptron whereas the least training time of 0.01 seconds is required by the Naïve Bayes Machine Learning algorithm.