Discrimination of Fresh Tobacco Leaves with Different Maturity Levels by Near-Infrared (NIR) Spectroscopy and Deep Learning

Table 7

The optimal parameters for KNN, BPNN, SVM, and ELM.

Data sets	Parameters
	KNN	BPNN	SVM		ELM
	Number of neighbors	Number of hidden layer nodes	Penalty parameter	Kernel function parameter	Number of hidden layer nodes
Upper leaves	6	19.6 ± 8.15	32	0.0313	161.1 ± 28.38
Middle leaves	5	23.3 ± 6.25	8	0.0625	145.9 ± 30.34
Lower leaves	8	16.3 ± 10.27	16	0.0313	138 ± 43.89