Deep learning driven anomaly based intrusion detection system for IoT
Yue Guan, Naser Ezzati‐Jivan · 2022
We propose a hybrid framework of machine learning and deep learning networks for efficient classification of attacks and anomalies. The machine learning algorithms are adopted to distinguish between normal data and anomaly data. The deep networks, on the other hand, are used to perform anomaly type classification. The framework is optimally tuned by selecting the most efficient hyperparameter values. These values are selected experimentally for the proposed deep network for optimal and efficient training of the network. We further propose the use of the Synthetic Minority Oversampling Technique (SMOTE) to address the data imbalance problem and Particle swarm optimization (PSO) as a feature selection mechanism to improve accuracy as well as execution time.