Few-Shot Learning-Based Network Intrusion Detection through an Enhanced Parallelized Triplet Network

Table 4

The performance results based on different datasets.

Dataset	Algorithm	TP	FP	TN	FN
CICIDS-2017	CNN	10,189	521	9,085	915
	ResNet	9,893	817	8,817	1,183
	XGBoost	10,132	578	9,252	748
	Proposed	10,301	409	9,409	591
UNSW_NB15	CNN	5,823	839	9,391	709
	ResNet	5,605	1,057	9,281	719
	XGBoost	6,154	508	8,243	1757
	Proposed	6,475	187	9,671	329

The bold values represent the optimal values for the different evaluation criteria in the comparison experiment. For example, proposed method achieved the best value under the evaluation criterion TP on CICIDS-2017 dataset.