Table 8.

Normalized confusion matrix of the proposed network with an asymmetric parameter.

		Predicted Class
Actual class		(L = 1, S = 2)			(L = 1, S = 4)			(L = 1, S = 8)
		Normal	Drowsy		Normal	Drowsy		Normal	Drowsy
Normal		0.897	0.103		0.910	0.090		0.918	0.082
Drowsy		0.105	0.895		0.089	0.911		0.082	0.918
Actual class		(L = 2, S = 1)			(L = 4, S = 1)			(L = 8, S = 1)
		Normal	Drowsy		Normal	Drowsy		Normal	Drowsy
Normal		0.884	0.116		0.916	0.084		0.924	0.082
Drowsy		0.118	0.882		0.079	0.921		0.081	0.919