Detecting Cyber-Attacks on Wireless Mobile Networks Using Multicriterion Fuzzy Classifier with Genetic Attribute Selection

Table 3

Comparisons of accuracy for different approaches using 10-fold cross validation (results are approximated to two decimal digits). All model constructions have taken reasonable time except SVM and MLP.

Approach	KDD’99 dataset		WEP/WPA dataset		WPA2 dataset
	Acc (%)	Time (sec)	Acc (%)	Time (sec)	Acc (%)	Time (sec)
With attribute selection
Proposed	99.92	7	85.70	6	90.10	4
NB	94.57	3	77.38	2	68.82	2
SVM	97.61	27	83.23	21	80.24	19
MLP	96.15	31	84.32	29	88.46	23
Without attribute selection
Proposed	99.20	10	84.89	8	91.82	6
NB	93.28	3.8	77.24	4	76.41	3
SVM	97.58	33	83.02	28	83.26	22
MLP	96.24	48	78.73	34	90.44	29