A Modified Gray Wolf Optimizer-Based Negative Selection Algorithm for Network Anomaly Detection

Table 4

Classification accuracy for different datasets with different methods.

Dataset	Algorithm	Training time (s)	Classification accuracy (%)
NSL-KDD	Full dataset	724	77.75
	Chi2	249	76.40
	ANOVA-F	153	73.18
	Mutual info	233	77.68
	Random forest	216	74.61
	RFE	240	75.63
	CBFS	186	78.48
	DP-SUMIC	140	79.12
CICIDS-2017	Full dataset	123	96.61
	Chi2	47	94.67
	ANOVA-F	18.6	90.82
	Mutual info	69.8	97.00
	Random forest	76.9	96.87
	RFE	77.1	96.50
	CBFS	42.1	96.50
	DP-SUMIC	38.5	97.83
UNSW-NB15	Full dataset	1250	89.60
	Chi2	701	89.24
	ANOVA-F	199	85.59
	Mutual info	486	90.10
	Random forest	332	88.69
	RFE	271	89.48
	CBFS	256	90.15
	DP-SUMIC	208	91.06