ENHANCING IOT SECURITY WITH MACHINE LEARNING: A RANDOM FOREST-BASED APPROACH FOR CYBER-ATTACK DETECTION
Ilyas Haman, Zahra Oughannou, Nour El Houda Chaoui, Habiba Chaoui · Proceedings on Engineering Sciences · 2025
A few industries that have undergone significant changes as a result of the Internet of Things' (IoT) rapid growth are smart cities, manufacturing, and healthcare.But the quick spread of IoT devices has led to serious security flaws, making them vulnerable to several kinds of cyberattacks.This paper presents a robust machine learning methodology employing the Random Forest (RF) model to effectively predict and classify cyber-attacks in IoT networks.Comprehensive data preprocessing techniques were utilized, including data cleaning, feature transformation, balancing using the Synthetic Minority Oversampling Technique (SMOTE), and standard scaling to maintain the integrity and uniformity of data.The suggested method's high accuracy (99%), precision, recall, and F1 score in detecting and classifying cyberattacks were assessed using the CIC-IoT2023 dataset.The results demonstrate how cutting-edge machine learning approaches can improve Internet of Things security and help with the critical challenge of defending IoT networks from sophisticated cyberattacks.This work contributes to the development of more reliable and efficient security solutions for IoT components and establishes the foundation for proactive security actions.