Neural network and principal component regression in non-destructive soluble solids content assessment: a comparison

Abstract

Abstract:Visible and near infrared spectroscopy is a non-destructive, green, and rapid technology that can be utilized to estimate the components of interest without conditioning it, as compared with classical analytical methods. The objective of this paper is to compare the performance of artificial neural network (ANN) (a nonlinear model) and principal component regression (PCR) (a linear model) based on visible and shortwave near infrared (VIS-SWNIR) (400-1000 nm) spectra in the non-destructive soluble solids content measurement of an apple. First, we used multiplicative scattering correction to pre-process the spectral data. Second, PCR was applied to estimate the optimal number of input variables. Third, the input variables with an optimal amount were used as the inputs of both multiple linear regression and ANN models. The initial weights and the number of hidden neurons were adjusted to optimize the performance of ANN. Findings suggest that the predictive performance of ANN with two hidden neurons outperforms that of PCR.

Prediction model for amylose content of Malaysian rice grains via visible-shortwave NIR spectroscopy

“…Firstly, since there were big scattering effects at the beginning and the end of the spectrum, the spectrum has been set for the range between 680 nm to 1000 nm only, and the wavelength beyond the range has been removed due to a high signal to noise ratio (S/N) [18]. Secondly, the VIS-SWNIR Spectroscopy data was transformed from reflectance value (R) into the most commonly used absorbance term (-log (R)).…”

“…In spectroscopy technique, calibration is important to replace slow, complex, expensive measurement of 'y' property (AC reference values) of interest by a faster spectroscopic feature [18]. Multivariate calibration is a " process to develop a model 'f'' that relates the sample properties 'y' to the intensities or absorbance 'X' at more than one wavelength or frequency of a set of known reference samples" [18].…”

“…Multivariate calibration is a " process to develop a model 'f'' that relates the sample properties 'y' to the intensities or absorbance 'X' at more than one wavelength or frequency of a set of known reference samples" [18]. There are several linear multivariate calibration (LMVC) models which are accepted by scientists in this area for developing the regression model, such as multiple linear regression (MLR), principal component regression (PCR), and partial least square regression (PLSR).…”

“…There are several linear multivariate calibration (LMVC) models which are accepted by scientists in this area for developing the regression model, such as multiple linear regression (MLR), principal component regression (PCR), and partial least square regression (PLSR). The calibration and validation processes can be referred to studies conducted by [18]. Figure 1 shows the flow diagram summary of the modelling processes.…”

“…PCR is a widely used regression model; it is a combination of principal component analysis (PCA) and multiple linear regressions (MLR). There are two types of decomposition technique; PCA and Singular Value Decomposition (SVD), which is used to reduce the dimensionality of a large data set or spectrum data into a smaller number of variables called principal components (PCs) scores [18] . PCA has a close connection with SVD from numerical linear algebra, but in this study, we used SVD because it is a well-established, stable and numerically accurate technique [23].…”