A Comparative Study of Machine Learning Algorithms for IoT Cybersecurity

Abstract—The explosive growth of the Internet of Things (IoT) has introduced unprecedented convenience and efficiency into our daily lives and industries. However, this proliferation has also created a massive attack surface, making IoT networks prime targets for a wide range of cyber threats. Traditional security mechanisms, often reliant on static signatures, are ill-equipped to handle the dynamic and sophisticated nature of modern attacks on heterogeneous IoT devices. Machine Learning (ML) has emerged as a powerful paradigm for developing intelligent and adaptive security solutions. This paper presents a comparative study of several prominent machine learning algorithms for detecting cyber attacks in an IoT environment. We evaluate the performance of Logistic Regression, Support Vector Machine (SVM), Decision Tree, Random Forest, XGBoost, and a simple Artificial Neural Network (ANN) on the widely-used Bot-IoT dataset. Our evaluation is based on key performance metrics including accuracy, precision, recall, and F1-score. The results demonstrate that ensemble methods, particularly Random Forest and XGBoost, and deep learning models like ANNs, achieve superior performance, with accuracies exceeding 99.9%. This study provides valuable insights into the efficacy of different ML models, aiding researchers and practitioners in selecting appropriate algorithms for robust IoT intrusion detection systems. Keywords—Internet of Things (IoT), Cybersecurity, Machine Learning, Intrusion Detection System (IDS), Anomaly Detection, Botnet.