Table 5.

The classification accuracy through the 4 methods using 27 features and the CFS feature set.

Methods	Feature set	Electrode	Recognition rate (%)
SVM	27 original features	Pz	47.68
		T3	51.22
		T4	57.49
	Feature set selected by CFS	Pz	75.42
		T3	55.83
		T4	63.74
C4.5	27 original features	Pz	72.24
		T3	52.56
		T4	73.32
	Feature set selected by CFS	Pz	85.46
		T3	63.78
		T4	80.95
BP	27 original features	Pz	79.63
		T3	54.64
		T4	72.27
	Feature set selected by CFS	Pz	82.55
		T3	62.96
		T4	78.71
LDA	27 original features	Pz	92.23
		T3	63.54
		T4	88.65
	Feature set selected by CFS	Pz	89.78
		T3	56.89
		T4	70.43