IDC-insight: boosting intrusion detection accuracy in IoT networks with Naïve Bayes and multiple classifiers

In this paper, a novel hybrid architecture, combining a Machine Learning (ML) infrastructure, operates as an IDS on IoT networks for the purpose of detecting intrusions. IDS systems learn to understand peoples' network behavior and make delimitation of anomalies by machine learning and statistical methods providing more enhanced protection. The IDC-Insight champion Decision Tree and Naïve Bayes algorithms for detection purposes and five ML algorithms for the classification task from the Random Forest, AdaBoost, C4.5, MLP and SVM. An evaluation on NSL-KDD-Cup dataset after an improvement, show that Naïve Bayes model outperformed Decision Tree (0.8330) under the framework, and this confirms the framework's potential in classifying network intrusion attacks along with improving network intrusion detection accuracy with (0.9847).