A Stacking Ensemble for Network Intrusion Detection Using Heterogeneous Datasets

Table 3

Critical hyperparameters.

Model	UNSW NB-15	UGR’16
Random forest	Estimators = 100	Estimators = 50
	Criterion = gini	Criterion = entropy
K nearest neighbor (KNN)	Neighbors = 5	Neighbors = 6
	Metric = Minkowski	Metric = Euclidean
Logistic regression	Penalty = L2	Penalty = L2
Support vector machine	C = 1.0	C = 1.0
	Kernel = rbf	Kernel = rbf