An Efficient Intrusion Detection System Using Machine Learning Model

Abstract

Evolving network technologies and increase in reliance on internet applications, cyber attacks are on the rise. More people and devices get connected across internet and generate large volume of traffic data. Utilizing such connected networks, exploiters attack using fraudulent activities for many reasons. To mitigate such malicious activities, an intelligent system capable of analyzing traffic and attacks becomes essential. Machine learning techniques are increasingly been utilized to detect anomalies and intrusions in network traffic. The performance of machine learning techniques largely depends on the dataset dimensions and the type of information present inside the data. This paper proposes a tree based ensemble model to detect intrusions in network traffic which addresses the feature dimension problem and model performance on detecting intrusions. The performance of the proposed model is compared against different machine learning techniques such as LDA, Logistic Regression, NN and SVM.