E-nose-based Optimized Ensemble Learning for Meat Quality Classification

Abstract

One of the most important sustainable development goals is eliminating hunger. Meat is an essential source of protein for the human body, which aids in its health. Because it is a perishable product, it was vital to keep an eye on meat quality. In this paper, a dataset has been obtained that expresses the meat's quality. This dataset represents the measurement of several sensors that measure the gases emitted from meat, which we consider as an electronic nose (E-nose). Several single machine learning algorithms have been used to classify meat quality. These algorithms are Logistic regression (LR), Random Forest (RF), K-Nearest Neighbor (KNN), and Decision Tree (DT). The complex voting ensemble learning algorithm was employed in conjunction with the E-nose. E-nose's ensemble learning accuracy using complex voting ensemble learning techniques was 99.57%, which is superior to the average performance of the other single machine learning classifiers. Grid search is used to tune the ensemble algorithm's hyperparameters, better results are obtained, and this outcome is reached when the ensemble is soft. The result was 99.9%.