This comprehensive review examines the application of machine learning (ML) techniques to the detection and prediction of mycotoxin contamination in food. The authors provide a systematic overview of the literature, focusing on the most commonly used ML algorithms, including neural networks (NNs), random forests (RFs), gradient boosting (GB), and support vector machines (SVMs).
The review begins by introducing the problem of mycotoxin contamination in food and the limitations of traditional detection methods. It then delves into the ML process, explaining key concepts such as training, validation, and test data, as well as common performance metrics used to evaluate the models.
The main body of the review is dedicated to discussing the application of each ML algorithm to different types of mycotoxin data, including spatio-temporal, spectral, and treatment-related data. For NNs, the authors highlight their use in analyzing hyperspectral imaging data and electronic nose data for mycotoxin detection. RFs are shown to be effective in classifying mycotoxin contamination levels using spectral data and predicting the impact of treatments on fungal growth and toxin production. GB models demonstrate strong predictive capabilities in forecasting aflatoxin and fumonisin contamination in crops based on spatio-temporal data.
The review also identifies critical areas that warrant further attention, such as the need for more detailed reporting on hyperparameter selection and the importance of ensuring reproducibility through open access to data and code. Overall, the review provides a comprehensive understanding of the current state of ML applications in mycotoxin detection and highlights the significant potential of these techniques to enhance food safety and quality control.
To Another Language
from source content
arxiv.org
Key Insights Distilled From
by Alan Inglis,... at arxiv.org 04-25-2024
https://arxiv.org/pdf/2404.15387.pdfDeeper Inquiries