Dimension Reduction Technique Based on Supervised Autoencoder for Intrusion Detection of Industrial Control Systems

Table 4

Hyperparameter settings in the experiments. Except for the hyperparameters of SupervisedAE, the hyperparameter names of machine learning techniques are consistent with the function of the scikit-learn library. The hyperparameter “None” of DT indicates that the default settings will be used.

Model	Hyperparameter	Value
PCA	n_components	[1, 16]
LDA	n_components	[1, 16]
KNN	n_neighbors	{1,3,5,7,9}
DT	max_depth	{2,4,6,8,10,None}
AdaBoost	n_estimators	{20,40,60}
Bagging	n_estimators	{20,40,60}