Table 6.

Performance of different methods on the sleep-edf and sleep-edfx datasets.

	Methods	Samples	Per-Class F1-Score					Overall Performances
			Wake	N1	N2	N3	REM	Accuracy	MF1
Sleep-edfx	Ref. [17]	37,022	71.6	47.0	84.6	84.0	81.4	78.9	73.7
	Ref. [40]	37,022	65.4	43.7	80.6	84.9	74.5	74.9	69.8
	Ref. [29]	41,950	84.7	46.6	85.9	84.8	82.4	82.0	76.9
	Ref. [26]	46,236	89.8	33.2	86.7	86.0	82.6	82.6	74.2
	Proposed	42,269	90.3	47.1	86.0	82.1	83.2	82.8	77.8
Sleep-edf	Ref. [19]	15,188	96.9	49.1	88.9	84.2	81.2	90.8	80.1
	Ref. [41]	15,136	97.8	30.4	89.0	85.5	82.5	91.3	77.0
	Ref. [25]	15,188	97.5	24.8	89.4	87.0	80.8	91.2	75.9
	Proposed	15,188	98.6	52.5	92.1	87.2	91.8	93.7	84.5