LS-SVM: a new chemometric tool for multivariate regression. Comparison of LS-SVM and pls regression for determination of common adulterants in powdered milk by nir spectroscopy

Quimica nova

关键词: support vector machines;    multivariateregression;    powdered milk.;   

DOI  :  10.1590/S0100-40422007000400018

学科分类：化学（综合）

来源: Sociedade Brasileira de Quimica

【 摘 要 】

Least-squares support vector machines (LS-SVM)were used as an alternative multivariate calibration method for the simultaneousquantification of some common adulterants found in powdered milk samples, usingnear-infrared spectroscopy. Excellent models were built using LS-SVM for determiningR2, RMSECV and RMSEP values. LS-SVMs show superior performance forquantifying starch, whey and sucrose in powdered milk samples in relation toPLSR. This study shows that it is possible to determine precisely the amountof one and two common adulterants simultaneously in powdered milk samples usingLS-SVM and NIR spectra.

【 授权许可】

CC BY