Explainable Deep Learning Model Accurately Detects Industrial IoT Cybersecurity Threats

Written by

in

The lightweight intrusion detection model built by the research team combines 1D CNN, LSTM, and temporal attention mechanisms, achieving 99.66% accuracy, 96.95% precision, 98.68% recall, and an F1 score of 97.73% in multiclass testing for industrial IoT. The model incorporates Shapley Additive Explanations, allowing it to directly process raw multivariate data streams without manual feature engineering. Experimental results support its use for threat detection in real-world industrial IoT environments.

Source:BIOENGINEER.ORG · Read Original

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *