Coupling Machine Learning with Clusterization-Triggered Emission for Geographical Origin Tracing of Rice.

Area of Science:

• Analytical Chemistry
• Food Science
• Machine Learning

Background:

• Geographical origin tracing of rice is crucial for food safety and consumer protection.
• Subtle differences in rice emission properties exist but are challenging to detect directly.
• Machine learning (ML) offers a potential solution for analyzing these minor variations.

Purpose of the Study:

• To develop and validate a novel method for identifying the geographical origin of rice.
• To explore the efficacy of combining clustering-triggered emission (CTE) with machine learning models.
• To assess the accuracy and universality of the proposed approach.

Main Methods:

• Rice samples from various origins were analyzed using clustering-triggered emission (CTE) to obtain fluorescence and phosphorescence data.
• Emission properties (wavelengths and lifetimes) were used as features for machine learning models.
• Artificial neural network (ANN) was trained and tested for classification accuracy.

Main Results:

• Artificial neural network (ANN) demonstrated superior performance among evaluated ML models.
• The combined CTE + ANN approach achieved 96.4% classification accuracy on a test set of rice samples.
• The model showed an 84.6% identification accuracy for unknown rice samples and was extended to wheat.

Conclusions:

• The integration of CTE and ANN provides a robust method for geographical origin tracing of rice.
• This approach effectively leverages subtle emission property differences for accurate identification.
• The CTE + ANN method shows promise for application in other agricultural products like wheat.