Classification of Minority Attacks using Machine Learning

Abstract

In the digital age, the mass adoption of edge devices or Internet of Things (IoT) devices pose serious challenges to cybersecurity. Today, various new types of attacks including minority attacks are increasing due to the presence of intruders in the network. Furthermore, due to the complex behavior in network or IoT networks, these attacks cannot be detected by traditional algorithms. Therefore, this paper proposes an effective intrusion detection system to detect these attacks in network or IoT networks. Machine learning algorithms Decision Trees, Extra Trees, Gradient Boosted Trees, k-Nearest Neighbors and Random Forest classifiers are used to estimate the benchmark dataset CICIDS2017. Furthermore, the RFE (recursive feature elimination technique) is utilized to select the most suitable or optimal set of features for detecting minority attack.