Analyzing Meat Samples: A Comparative Evaluation of Linear versus Nonlinear Machine Learning Techniques

Articles in Press

Document Type : Original Article

Authors

1 Halal Research Center of IRI, Food and Drug Administration, Ministry of Health and Medical Education, Tehran, Iran

2 Cosmetic Products Research Center,, Food and Drug Administration, Ministry of Health and Medical Education, Tehran, Iran

10.30502/h.2024.447237.1135

Abstract

Abstract
Background and Objective: The moisture, protein, and fat content in meat are among its most important features in determining its quality. Traditional methods of measuring these features are usually time-consuming, destructive, and costly. Substituting these methods with spectroscopic methods can be a significant step in innovating rapid methods in this field. Combining linear or nonlinear machine learning methods with near-infrared spectroscopic measurements enables the quantitative determination of these components in meat samples. The aim of this study is to compare the efficiencies of two linear and nonlinear machine learning methods in determining these three features in meat samples.
Results and Conclusion: In this research, partial least squares (PLS) and radial basis function artificial neural network (RBF-ANN) methods were employed to analyze spectral data from 240 meat samples. The variables of each of the learning machine learning methods in the calibration step using near-infrared spectra from 140 meat samples were adjusted at their optimal positions, and corresponding linear and nonlinear models were used to predict the concentration of three analytes in 70 independent meat samples. Although the nonlinear RBF-ANN machine demonstrated a more efficient performance in predicting concentration, the linear PLS machine also provided an acceptable performance in determining fat content in meat samples.

Keywords

Main Subjects

Journal of Halal Research

Articles in Press, Accepted Manuscript
Available Online from 17 March 2024

Files

Share

How to cite

Statistics