Table 9 Performance metrics for 7 ML algorithms using the UNSW-NB15 dataset and SMOTE.
From: Using machine learning algorithms to enhance IoT system security
Ser. | Classifier | Accuracy (%) | Precision (%) | Specificity (%) | Recall (%) | F1-score (%) | AUC |
|---|---|---|---|---|---|---|---|
1 | Random forest | 99.9 | 99.9 | 99.8 | 99.9 | 99.9 | 1.0 |
2 | Naive Bayes | 73.1 | 73.8 | 65.6 | 79.0 | 76.3 | 0.8 |
3 | Decision Tree | 99.9 | 100.0 | 100.0 | 99.1 | 99.5 | 0.8 |
4 | Back propagation NN | 67.6 | 66.7 | 49.9 | 82.0 | 73.6 | 0.6 |
5 | XGBoost | 99.9 | 100.0 | 100.0 | 99.1 | 99.9 | 1.0 |
6 | AdaBoost | 99.9 | 99.9.0 | 99.9 | 99.9 | 99.9 | 1.0 |
7 | Ensembled RF-BPNN | 99.9 | 99.9 | 99.8 | 99.9 | 99.9 | 1.0 |
