Improving IoT Intrusion Detection Through SMOTE-Based Oversampling and Extended Multi-Model Evaluation on Side-Channel Power Data
Researchers have developed a new method to improve intrusion detection in IoT networks by addressing class imbalance in datasets. They applied the Synthetic Minority Oversampling Technique (SMOTE) to balance the data, achieving an imbalance ratio of 1.1. This approach significantly improved the detection of minority attack classes, particularly those with combined infections, as revealed by macro-F1 scores and confusion matrices. Random Forest achieved a micro-averaged F1 score of 0.9989 and a macro F1 of 0.9794, outperforming previous methods. AI
IMPACT Enhances AI model robustness in cybersecurity by addressing class imbalance, crucial for detecting rare attack vectors.