An Intelligent Intrusion Detection Framework Using Deep Learning and Unsupervised Feature Selection for Industry 4.0

Abstract

Manufacturing has been revolutionized by the integration of advanced technologies such as the Internet of Things, cyber-physical systems, and cloud computing, but it is also exposed to a variety of cyber-attacks because of these technologies. A system for detecting unsupervised intrusions is proposed as an answer to these growing security challenges, which we call the Unsupervised Intrusion Detection System for Industry 4.0. Using the isolation forest method, the framework identifies anomalous network traffic anomalies through random forest-based feature selection. Based on deep learning and unsupervised anomaly detection, the proposed system is more accurate, computationally efficient, and reduces false positives than traditional intrusion detection systems. With the proposed model, interconnected industrial systems will be protected from evolving cyber threats, demonstrating enhanced security for Industry 4.0 systems.