Geographic Origin Discrimination of Millet Using Vis-NIR Spectroscopy Combined with Machine Learning Techniques

Area of Science:

• Agricultural Science
• Analytical Chemistry
• Food Science

Background:

• Millet is a crucial food source in arid regions globally.
• Ensuring food originality through Protected Geographical Indication (PGI) is a global priority.
• Accurate crop origin tracing is vital for food supply integrity.

Purpose of the Study:

• To develop a method for discriminating millet varieties based on origin.
• To evaluate the effectiveness of Visible and Near-Infrared Spectroscopy (Vis-NIR) combined with machine learning for millet authentication.
• To assess the performance of various machine learning algorithms in origin tracing.

Main Methods:

• Visible and Near-Infrared Spectroscopy (Vis-NIR) was employed to analyze 480 millet samples from 16 varieties.
• Five machine learning algorithms (K-NN, LDA, LR, RF, SVM) were utilized to analyze spectral data.
• Principal Component Analysis (PCA) visualized spectral clustering, and cross-validation optimized model performance.

Main Results:

• All tested machine learning algorithms achieved high discrimination performance, with F-Scores above 98.8%.
• Support Vector Machine (SVM), Random Forest (RF), and Linear Discriminant Analysis (LDA) achieved the highest F-Score of 99.5%.
• Non-linear algorithms showed a slight performance advantage over linear ones.

Conclusions:

• Vis-NIR spectroscopy coupled with machine learning offers a rapid, cost-effective, and non-destructive method for tracing millet origins.
• This approach supports food authentication and contributes to maintaining the integrity of the food supply chain.
• The study demonstrates the potential for safeguarding agricultural product originality and supporting PGI initiatives.