Abstract
This study introduces a unique framework for an Intrusion Detection System (IDS) that employs an advanced machine learning approach to improve the Internet of Things (IoT) networks. IoT devices, now increasingly prevalent, rely on data which is a subject of interest to hackers/attackers who explore the present rise in network security vulnerabilities. There is therefore the need for a more robust and accurate intrusion detection system. The integration of Random Forest (RF) algorithms with Deep Neural Networks (DNNs) provides a significant increase in model evaluation metrics and robustness. A comprehensive CICIoT2023 dataset was adopted and used to meticulously train and evaluate the IDS model, resulting in an exceptional and effective system of identifying and preventing potential threats. Also, the study analysis highlights areas of improvement, particularly in detecting specific attack types such as SQL injection. Whilst these findings push the boundaries of IoT security using state-of-the-art machine learning techniques, they have also underlined the need for further studies to address the obvious gaps.