Table 6. Comparison results between three boosting and different methods (%).
| Model | KDDCup99 | UNSW NB15 | CICIDS 2017 | |||
|---|---|---|---|---|---|---|
| Accuracy | F1 Score | Accuracy | F1 Score | Accuracy | F1 Score | |
| Naïve Bayes | 73.55 | 72.31 | 61.8 | 65.27 | 93.90 | 93.53 |
| Decicision Tree | 77.89 | 75.25 | 73.25 | 76.36 | 99.62 | 99.57 |
| Random Forest | 77.20 | 73.23 | 74.35 | 77.28 | 99.79 | 99.78 |
| SVM | 72.85 | 68.84 | 68.49 | 70.13 | 96.97 | 96.99 |
| MLP | 78.97 | 75.40 | 78.32 | 76.98 | 99.48 | 99.39 |
| RUS + SVM | 73.57 | 70.11 | 67.16 | 70.45 | 96.45 | 96.55 |
| RUS + MLP | 76.66 | 72.38 | 77.27 | 76.21 | 99.46 | 99.42 |
| ROS + SVM | 73.34 | 69.90 | 68.32 | 70.00 | 96.98 | 97.04 |
| ROS + MLP | 78.10 | 74.18 | 76.13 | 76.97 | 99.55 | 99.55 |
| SMOTE + SVM | 79.23 | 78.36 | 71.5 | 73.77 | 97.00 | 97.04 |
| SMOTE + MLP | 77.47 | 75.18 | 79.59 | 80.10 | 99.33 | 99.34 |
| CNN | 78.33 | 74.75 | 80.52 | 76.61 | 99.48 | 99.44 |
| Fuzziness-based NN | 75.33 | 70.58 | 81.21 | 78.58 | 99.61 | 99.57 |
| LSSVM + MIFS (β = 0.3) | 78.20 | 72.76 | 76.83 | 77.43 | 98.76 | 98.67 |
| LSSVM + FMIFS | 75.67 | 73.67 | 77.18 | 77.65 | 99.51 | 99.48 |
| IGAN-IDS | 84.45 | 84.17 | 82.53 | 82.86 | 99.79 | 99.98 |
| SMOTE + LightGBM | 97.68 | 97.61 | 89.32 | 89.54 | 99.63 | 99.62 |
| SMOTE + XGBoost | 99.92 | 99.91 | 89.32 | 89.64 | 99.46 | 99.44 |
| SMOTE + CatBoost | 99.90 | 99.89 | 88.66 | 88.73 | 99.57 | 99.56 |