A novel GAN-based regression model for predicting frying oil deterioration.

Area of Science:

• Food Science and Technology
• Machine Learning Applications
• Chemical Engineering

Background:

• Deep frying is a popular cooking method, but the process deteriorates frying oil, posing health risks.
• Current methods for detecting oil deterioration require professional expertise and expensive equipment, limiting accessibility.
• Existing methods often have limitations, such as fixed temperature requirements.

Purpose of the Study:

• To develop a novel, accessible method for predicting frying oil deterioration.
• To overcome the limitations of existing detection techniques using advanced machine learning.
• To establish a regression model capable of accurately assessing oil quality.

Main Methods:

• Conducting deep frying experiments to collect data on oil deterioration indexes under varying temperatures and times.
• Developing and training a generative adversarial network (GAN)-based regression model on the experimental dataset.
• Validating the model's predictive performance using a dedicated test set.

Main Results:

• The proposed GAN-based regression model accurately predicts frying oil deterioration.
• The model demonstrates the capability to assess oil quality without the need for direct experimentation.
• Experimental results confirm the model's effectiveness and reliability.

Conclusions:

• The developed GAN model provides an efficient and cost-effective solution for monitoring frying oil quality.
• This machine learning approach can be generalized to various regression problems beyond food science.
• The model offers a promising tool for applications in price forecasting and trend analysis.