AI and Machine Learning Approaches for Enhancing Cyber-Security in Internet of Things Systems

Abstract

The rapid proliferation of IoT networks has heightened the need for effective and reliable cyber threat detection systems. This study evaluates multiple models, including traditional machine learning techniques, deep neural networks, and a proposed Hybrid CNN-LSTM model, for detecting cyber threats in IoT environments. The Hybrid CNN-LSTM model achieved superior performance across all metrics, with an accuracy of 99.1%, precision of 98.8%, recall of 98.6%, F1-score of 98.7%, and ROC-AUC of 99.0%, significantly outperforming the other approaches. By combining CNN's spatial feature extraction capabilities with LSTM's temporal sequence processing strengths, the Hybrid CNN-LSTM effectively addresses the complexity of IoT datasets. This model demonstrates exceptional potential for real-time and resource-efficient deployment in IoT networks, ensuring robust and reliable threat detection.