Feature Reduction and Anomaly Detection in IoT Using Machine Learning Algorithms

Anomaly detection in IoT is a hot topic in cybersecurity. Also, there is no doubt that the increased volume of IoT trading technology increases the challenges it faces. This paper explores several machine-learning algorithms for IoT anomaly detection. The algorithms used are Naïve Bayesian (NB), Support Vector Machine (SVM), Decision Tree (DT), XGBoost, Random Forest (RF), and K-nearest Neighbor (K-NN). Besides that, this research uses three techniques for feature reduction (FR). The dataset used in this study is RT-IoT2022, which is considered a new dataset. Feature reduction methods used in this study are Principal Component Analysis (PCA), Particle Swarm Optimization (PSO), and Gray Wolf Optimizer (GWO). Several assessment metrics are applied, such as Precision (P), Recall(R), F-measures, and accuracy. The results demonstrate that most machine learning algorithms perform well in IoT anomaly detection. The best results are shown in SVM with approximately 99.99% accuracy.