Classification of Food Additives Using UV Spectroscopy and One-Dimensional Convolutional Neural Network.

Area of Science:

• Analytical Chemistry
• Spectroscopy
• Food Science

Background:

• Food additives are integral to modern food products, influencing flavor, texture, and shelf life.
• Accurate identification and quantification of food additives are crucial for quality control and regulatory compliance.

Purpose of the Study:

• To develop an automated classification system for five distinct food additives.
• To leverage ultraviolet (UV) absorbance spectroscopy and deep learning for additive identification.

Main Methods:

• Preparation of simple and mixed solutions of five food additives at varying concentrations.
• Measurement of UV absorbance spectra for each sample between 190-360 nm.
• Classification using deep learning models, primarily Convolutional Neural Networks (CNNs), on spectral data.

Main Results:

• Each food additive exhibited characteristic absorbance peaks within the UV spectrum (190-360 nm).
• CNN models demonstrated high accuracy in classifying additive spectra.
• A CNN with three convolutional layers achieved a mean testing accuracy of 92.38% and validation accuracy of 93.43%.

Conclusions:

• UV absorbance spectroscopy provides a distinctive spectral fingerprint for identifying food additives.
• Deep learning, particularly CNNs, offers a robust and accurate method for the automated classification of food additives based on spectral data.