Machine-Learning Approach to Optimize SMOTE Ratio in Class Imbalance Dataset for Intrusion Detection

Table 13

Recall metrics of SVM, decision tree, and RNN-LSTM tests.

	Train B + Validation			Train B + Test			Train A + Test
Classes	SVM	DT	LSTM	SVM	DT	LSTM	SVM	DT	LSTM
Normal	0.961	0.999	0.967	0.977	0.993	0.947	0.977	0.994	0.982
U2R	0.808	0.769	0.769	0.641	0.462	0.641	0.564	0.256	0.615
R2L	0.982	0.961	0.975	0.275	0.235	0.260	0.302	0.261	0.274
DoS	0.999	1.000	0.999	0.855	0.999	0.998	0.871	0.996	0.999
Probe	0.924	0.992	0.974	0.928	0.988	0.977	0.941	0.997	0.996