Deep Encrypted Traffic Detection: An Anomaly Detection Framework for Encryption Traffic Based on Parallel Automatic Feature Extraction

Table 3

Performance of feature set filtered by six feature selection algorithms on Naive Bayes.

FS	Accuracy (%)	F1_score (%)	Recall (%)	AUC_score (%)
VT	79.050	73.333	86.347	80.872
Chi_2 test	76.329	68.446	76.959	76.487
RECFV	98.392	97.581	97.528	98.175
Decision tree	98.581	97.827	95.765	97.878
Random forest	99.559	99.335	98.697	99.344
L1_R	99.662	99.491	99.006	99.498

The bolded values represent the best-performing result.