Table 4: Maximum mean accuracy comparison performance of four networks with six PSD feature extraction methods.

S.No	Features	Maximum Mean Accuracy of Neural Network Classifiers (%)
		FFNN	PNN	CFBNN	FMSVM
1	AR Burg	93.81	92.16	92.34	94.75
2	AR Yule – Walker	93.25	91.88	90.67	93.92
3	AR Covariance	92.20	91.29	91.50	91.92
4	AR Modified Covariance	93.08	91.21	90.67	91.92
5	Levinson Durbin Recursion	91.73	90.54	91.50	91.92
6	Linear Prediction Coefficient	93.08	90.42	90.67	91.09