Application Research for Fusion Model of Pseudolabel and Cross Network

Table 3

Experimental results of multitask and single task.

(, , , )	Multitask		Multitask		Single task
	Facial beauty prediction (%)	Gender recognition (%)	Facial beauty prediction (%)	Expression recognition (%)	Facial beauty prediction (%)
(0, 1, 1, 0)	60.94	97.64	63.35	23.99	58.75
(0.1, 0.9, 0.9, 0.1)	62.85	96.91	62.55	45.91
(0.2, 0.8, 0.8, 0.2)	64.06	95.64	62.45	49.85
(0.3, 0.7, 0.7, 0.3)	62.65	97.82	62.95	38.95
(0.4, 0.6, 0.6, 0.4)	—	55.82	61.75	44.85
(0.5, 0.5, 0.5, 0.5)	61.65	95.45	62.75	32.80
(0.6, 0.4, 0.4, 0.6)	62.35	94.73	62.65	45.67
(0.7, 0.3, 0.3, 0.7)	61.55	95.82	63.45	46.21
(0.8, 0.2, 0.2, 0.8)	62.85	96.55	61.35	48.91
(0.9, 0.1, 0.1, 0.9)	63.45	96.55	62.35	47.94
(1, 0, 0, 1)	59.35	97.45	63.05	23.99

Better results in terms of recognition rates were highlighted in bold.