FTIR Spectroscopy and Machine Learning Based Method for Accurate Classification of Honey Purity

Résumé fourni par la source

Natural sweeteners like honey are frequently utilized in medical treatments or as a sugar replacement in a variety of dishes. Due to frequent adulteration with other compounds brought on by its high market value, consumers find it challenging to confirm its purity. Our goal is to provide consumers and ethical producers with protection by creating a practical and effective honey classification system. Our tests show that this approach can reliably identify many major honey types in Bangladesh by using machine learning algorithms and FTIR spectroscopy to analyze the infrared spectra of honey samples. This instantaneous, low-labor, non-destructive method efficiently filters honey products. PCA is more effective than LDA or FA approaches at differentiating between pure and adulterated honey samples, as demonstrated by visual classification plots. The validated classification results achieved a $94.11 \%$ accuracy, surpassing other state-of-the-art methods. These results indicate that preprocessing and the applied machine learning model significantly improve the classification of pure and adulterated honey samples.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.