Table 3.

Comparison of classification performances using the conventional methods and the DSTCLN.

Methods	Alert State and Drowsy State (2-Class)				Drowsiness Levels (5-Class)
	Accuracy	Std	Sensitivity	Specificity	Accuracy	Std
PSD-SVM [57]	0.64	0.03	0.77	0.50	0.31	0.04
CCA-SVM [46]	0.78	0.02	0.73	0.84	0.47	0.05
CCNN [58]	0.52	0.01	0.68	0.34	0.33	0.03
ESTCNN [23]	0.78	0.01	0.73	0.85	0.56	0.01
LSTM-D [55]	0.74	0.01	0.55	0.92	0.45	0.02
Proposed DSTCLN	0.87	0.01	0.86	0.88	0.69	0.02

The text using bold type represents classification performances of the proposed DSTCLN.