WineNet: A Wine Classification Model Based on Electronic Nose

Red wine is a popular beverage in daily life. However, the issue of low-quality red wine being passed off as high-quality red wine has long plagued the market, greatly affecting consumer interests. Therefore, it is crucial to identify and categorize the quality of red wine. Thankfully, the electronic nose offers an efficient means of collecting data on red wine aromas. In this study, I utilized a neural network to analyze a dataset obtained from the electronic nose. Furthermore, I developed a lightweight network called WineN et, which incorporates channel shuffle, to effectively classify red wines based on their level of corruption. WineN et demonstrated exceptional performance in the red wine classification task, achieving impressive evaluation metrics including accuracy (98.89<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>), recall (95.24<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>), precision (95.00<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>), f1-score (93.65<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup>), and kappa factor(86.11 %). My proposed method surpasses traditional approaches in recognizing red wine, providing a novel and effective means of assessing and classifying red wine quality using electronic nose technology.