Identificação da concentração dos diferentes tipos de adulterantes do leite por meio da reflectância hiperespetral aplicada a aprendizagem de máquinas

Abstract

Milk has been adulterated with different solutions in order to produce more milk with less pure milk. Milk adulteration is a dangerous practice in the dairy industry and extremely harmful to consumers, as milk is one of the most consumed food products. Thus, the use of spectral data together with machine learning can be an essential tool in detecting milk fraud from different sources, especially as it is an innovative methodology, since there have been no reports of anything similar in the literature. The aim of this work was to identify which concentrations of the different types of milk adulterants the spectral distinction and the ML model (support vector machine (SVM)) can identify. Samples of pasteurized milk were used, to which samples of the same milk adulterated with different substances were compared. Specifically, the milk pasteurized was adulterated with whey, water and starch in concentrations of 5%, 10%, 15%, 50% and 75% for each adulterant. The samples were read twice each in a spectroradiometer. The data was then tabulated for analysis using machine learning (ML). The following algorithm was used: support vector machine (SVM) and logistic regression (LR), which was used as a control model. The methodology proposed in the work was that it would be possible to identify a difference in the spectral behavior of milk samples adulterated with different concentrations of adulterants compared to pasteurized milk, making it possible to more clearly identify the distinction between curves in the VIS and NIR regions. Another objective of the work was to verify the accuracy of the machine learning algorithm in classifying milk adulterations and their concentrations, achieving accuracies of 80 for correct classification (CC), close to 1 for F-score and above 0.8 for kappa coefficient