Optimization of E-Nose Technology Using Learning Methods based on Feature Selection Algorithm

E-nose technology has become an important tool in various applications due to its fast, cost-effective, and non-invasive capabilities. Recent research has shown that optimizing e-nose systems through feature selection can reduce the number of sensors required to achieve high performance. Several challenges arise when using a large number of sensors, such as increased system complexity, longer processing times, and higher power consumption. These factors can reduce the efficiency and effectiveness of e-nose technology, even though a greater number of sensors may improve accuracy. Therefore, optimizing the number of sensors through appropriate feature selection is crucial to maintain a balance between high performance and system efficiency. This study aims to optimize the sensor array in e-nose technology by employing machine learning algorithms, specifically through the Gini Index feature selection method, to assess the freshness of catfish. The results demonstrate that the feature selection approach effectively identifies optimal and significant feature combinations, significantly reducing the number of sensors required. The model achieved a high accuracy of 94% in detecting catfish freshness, reducing the number of sensors by more than half of the initial number used.