Table 3.

Classification results without and with fine-tuning.

S-Ta	Non-fine-tuning			Fine-tuning			Finetuning with a smaller set e
	Trb(%)	Valc(%)	Ted(%)	Tr(%)	Val(%)	Te(%)	Tr(%)	Val(%)	Te(%)
01–02	60.23	52.94	68.75	98.32	88.24	87.50	96.25	88.24	87.50
01–03	76.44	84.21	88.89	100.00	89.47	88.89	95.00	89.47	88.89
01–04	52.04	73.68	57.89	92.35	89.47	89.47	91.25	89.47	89.47
02–01	62.03	56.25	66.67	99.36	93.75	86.66	96.25	93.75	86.67
02–03	67.02	89.47	61.11	91.62	89.47	88.89	100.00	89.43	83.33
02–04	53.57	42.11	52.63	96.43	94.74	89.47	91.25	89.48	78.95
03–01	70.89	75.00	60.00	94.30	93.75	93.33	95.00	93.75	93.34
03–02	63.74	47.06	50.00	98.83	88.24	87.50	95.00	88.23	87.50
03–04	58.67	78.95	52.63	97.96	100.00	89.47	100.00	94.74	84.21
04–01	51.27	50.00	53.33	97.47	93.75	86.67	98.75	87.50	86.67
04–02	43.27	47.06	37.50	100.00	88.24	87.50	98.75	88.24	87.50
04–03	72.77	57.89	77.78	97.91	100.00	88.89	100.00	94.74	88.89

^aS-T means the source domain-the target domain; ^bTr means the training set of the target domain; ^cVal means the validation set of the target domain; ^dTe means the test set of the target domain; ^ea small set only consisted of 20 samples of each class; the test set remained the same as in Table 1. The bold values means the accuracy rate of the test set was beyond 86%.