Table 3.

Comparing network intrusion detection results for the six datasets (%).

Dataset	Model	Normal	DoS	Probe	U2R	R2L	Acc	Recall	ER
Dataset 1	SVM	98.21	83	66.01	0.88	3.14	81.52	77.72	18.48
	BP	96.51	89.49	46.18	9.21	1.93	85.66	83.48	14.34
	RF	93.65	96.62	59.27	0	0	90.44	91.08	9.56
	Bayes	91.51	95.59	61.35	4.39	3.56	89.48	92.57	10.52
	SCDNN	97.21	96.87	80.32	11.4	6.88	91.97	91.68	8.03
Dataset 2	SVM	96.22	97.1	65.84	0	0.05	91.39	90.52	8.61
	BP	91.44	97.42	62.69	7.02	5.41	90.93	92.88	9.07
	RF	98.23	96.48	38.26	0	0	90.95	89.51	9.05
	Bayes	95.92	95.98	62.55	4.82	4.38	90.69	91.07	9.31
	SCDNN	98.42	97.2	70.64	3.51	1.57	92.03	91.35	7.97
Dataset 3	SVM	95.87	97.23	64.86	0	0.06	91.41	90.59	8.59
	BP	81.53	96.95	8.81	6.14	7.26	88.03	90.05	11.97
	RF	99.57	96.57	0	0	0	90.76	89.37	9.24
	Bayes	96.38	96.29	59.15	7.02	7.46	91.12	90.95	8.88
	SCDNN	97.61	97.23	65.96	4.39	6.59	92.1	92.23	7.9
Dataset 4	SVM	95.54	70.18	57.37	0	1.63	70.73	53.26	29.27
	BP	96.35	71.17	65.55	0	0.58	72.16	57.79	27.84
	RF	99.63	63.11	7.23	0	0	64.57	40.45	35.43
	Bayes	93.9	72.18	41.02	0	0	68.73	52.78	31.27
	SCDNN	96.17	75.84	53.37	3	3.01	72.64	57.48	27.36
Dataset 5	SVM	98.57	18.93	49.89	0	0.11	54.1	20.45	45.9
	BP	91.79	7.63	66.58	1.5	2.43	49.53	27.56	50.47
	RF	99.69	62.64	48.99	0	0	68.93	46.43	31.07
	Bayes	99.06	61.65	35.4	0	0	66.87	44.28	33.13
	SCDNN	97.19	74.51	48.37	5	0.62	71.83	55.08	28.17
Dataset 6	SVM	95.81	41.5	43.67	0	0	41.46	30.6	58.54
	BP	74.72	4.61	88.67	0	1.53	33.59	30.6	66.41
	RF	99.72	36.15	6.74	0	0	32.73	18.9	67.27
	Bayes	82.16	48.25	28.52	0	0	38.37	30.08	61.63
	SCDNN	84.2	50.02	52.66	1.5	0.98	44.55	37.85	55.45