Enhanced IoMT security framework using group teaching optimized auto-encoder for intrusion detection.

Providing security to Internet of Medical Things (IoMT) is significant worldwide problem for future generations its implementation to be successful. The traditional security methodologies developed for IoMT struggles with the specific issues of high false positives and lower detection rate. Therefore, the proposed work aims to develop a ground-breaking intrusion detection model, named as, Group Teaching Optimized Probabilistic Deep Auto-Encoder (GTPDA) for increasing the security of IoMT networks. Here, the data transformation and normalization processes are applied to balance the dataset's properties. Then, an Intriguing Group Teaching Optimization (IGTO) algorithm is applied to choose the most correlated and essential traits from the normalized dataset for effective intrusion detection. Consequently, a Conditional Probabilistic Deep Auto-Encoder (CPDAE) model is used to more accurately classify the type of intrusion with system complexity. This study uses the BoT-IoT, Kaggle invasion dataset, and ToN-IoT open benchmarking datasets for evaluation and performance assessments. Among all, the proposed GTPDA with its various performance metrics presented, achieves an impressive 98.8% precision, 99% recall, 98.8% F1-score, and 99% accuracy, showing its significant performance in ensuring IoMT network security.