Intrusion Detection Systems Based on Logarithmic Autoencoder and XGBoost

Table 6

Comparison of the proposed model and other excellent classifiers for CICIDS2017.

Classifier	Accuracy	Precision	Recall	F1-score	Time(s)
NB-SVM [24]	0.9892	0.9946	0.9700	0.9821	—
T-SNERF [25]	0.9978	0.9980	0.9980	0.9980	—
MTH-IDS [26]	0.9989	0.9981	0.9160	0.9989	478.2
LMDRT-SVM2 [27]	0.9928	0.9916	0.9939	0.9927	—
DT + rules-based model [28]	0.9666	0.9447	0.9886	0.9662	160.07
KNN [18]	0.9630	0.9620	0.9370	0.9630	15,243.6
RF [18]	0.9882	0.9880	0.9985	0.9880	1,848.3
LogAE-XGBoost	0.9992	0.9971	0.9986	0.9979	1,092.35

It is used to distinguish the metrics of our proposed model from those of other models.