Table 7.

Classification Accuracy Analysis of four feature evaluation measures on NSL-KDD.

Attack class	ALL features				Correlation				Consistency
	KNN	RF	SVM	DBN	KNN	RF	SVM	DBN	KNN	RF	SVM	DBN
DoS	99.88	99.96	98.2	99.44	97.7	97.94	96.0	98.4	99.8	99.96	97.5	98.3
Prob	99.63	99.84	97.8	98.4	99.67	99.78	97.3	96.9	99.47	99.91	96.8	97.27
R2L	99.77	99.90	98.46	98.46	99.01	99.01	98.46	98.4	99.90	99.94	98.46	98.46
U2R	99.9	99.93	98.94	98.4	99.93	99.92	99.8	99.85	99.93	99.94	99.9	99.9

Attack class	Information				Distance
	KNN	RF	SVM	DBN	KNN	RF	SVM	DBN
DoS	99.91	99.97	98.4	99.5	98.9	99.0	93.7	92.0
Prob	99.69	99.90	98.8	98.8	99.79	99.80	98.7	98.7
R2L	99.72	99.91	98.9	98.46	99.73	99.84	98.7	98.84
U2R	99.93	99.96	99.9	99.85	99.93	99.93	99.9	99.85