GCN-ETA: High-Efficiency Encrypted Malicious Traffic Detection

Table 5

Test accuracy (%), (%), AUC (%), and after 5-fold cross-validation.

Model	Accuracy		AUC
LR	73.55 ± 0.39	76.18 ± 0.22	73.55 ± 0.26	58994
BernoulliNB	79.86 ± 0.44	80.31 ± 0.43	79.85 ± 0.47	129224
DT	83.26 ± 0.67	83.81 ± 0.68	83.24 ± 0.50	2061
XGBoost	86.39 ± 0.26	87.10 ± 0.24	86.39 ± 0.85	111
nnDPI	65.31 ± 0.35	69.03 ± 0.33	65.80 ± 0.16	<1
GCN-TC	83.12 ± 0.70	83.34 ± 0.46	83.55 ± 0.97	28
GCN + XGBoost	97.65 ± 0.33	97.63 ± 0.33	97.52 ± 0.80	68
GCN + RF	98.61 ± 0.11	98.59 ± 0.11	98.34 ± 0.85	70
GCN + KNN	97.37 ± 0.29	97.36 ± 0.29	96.94 ± 0.83	71
GCN + DT	98.38 ± 0.07	98.37 ± 0.17	98.02 ± 0.94	72
GCN-ETA	98.65 ± 0.09	98.63 ± 0.09	98.59 ± 0.65	1333