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

Table 14

Comparisons with other works by detection rate.

		Normal	U2R	R2L	DoS	Probe	Acc.	FP
These methods	SVM	97.7	56.4	30.2	87.1	94.1	88.2	2.4
	LSTM	98.2	61.5	27.4	99.9	99.6	98.1	0.4
SVM and clustering [22]		99.3	19.7	28.8	99.5	97.5	95.7	0.7
ESC-IDS [24]		98.2	14.1	31.5	99.5	84.1	95.3	1.9
KDD'99 winner [25]		99.5	13.2	8.4	97.1	83.3	91.8	0.6
KDD'99 runner-up [26]		99.4	11.8	7.3	97.5	84.5	91.5	0.6
Multiclassifier [27]		N/A	29.8	9.6	97.3	88.7	N/A	N/A
Association rule [28]		99.5	3.8	7.9	96.8	74.9	N/A	N/A