. 2022 Aug 16;2022:9813841. doi: 10.34133/2022/9813841

Table 5.

Regression results using fine-tuned CNN and conventional models.

Source/target domain	Pretreatment	Model	Calibration set^a		Validation set^b		Prediction set^c
Source/target domain	Pretreatment	Model	R²	RMSE	R²	RMSE	R²	RMSE
LMY24/XLZ53	None	PLS	0.719	4.792	0.663	5.596	0.696	5.235
		SVR	0.733	4.297	0.629	5.113	0.667	4.834
		Fine-tuned CNN	0.850	3.225	0.796	3.786	0.842	3.327
		Fine-tuned CNN using a smaller set	0.885	2.901	0.802	3.730	0.811	3.643
	FD + SNV	PLS	0.821	3.600	0.754	4.359	0.777	4.098
		SVR	0.761	4.065	0.662	4.877	0.761	4.090
		Fine-tuned CNN	0.909	2.505	0.850	3.248	0.870	3.020
		Fine-tuned CNN using a smaller set	0.914	2.451	0.815	3.613	0.853	3.207
XLZ53/LMY24	None	PLS	0.538	5.749	0.586	5.423	0.559	5.810
		SVR	0.537	5.537	0.561	5.429	0.506	5.759
		Fine-tuned CNN	0.850	3.156	0.746	4.129	0.757	4.036
		Fine-tuned CNN using a smaller set	0.734	4.235	0.672	4.691	0.689	4.568
	FD + SNV	PLS	0.637	5.140	0.630	5.086	0.636	5.300
		SVR	0.654	4.909	0.664	4.928	0.578	5.445
		Fine-tuned CNN	0.889	2.708	0.835	3.332	0.822	3.460
		Fine-tuned CNN using a smaller set	0.892	2.697	0.822	3.458	0.796	3.703

^aCalibration set means the calibration set of the target domain; ^bValidation set means the validation set of the target domain; ^cprediction set means the prediction set of the target domain; the numbers are bolded to highlight models with relatively good results.