Classification of Dried Moringa Leaf Quality Using Extreme Gradient Boosting with Hyperparameter Optimization

Abstract

Classification using deep learning models has shown superior predictive performance compared to conventional methods. Deep learning enables the automatic extraction of complex patterns from data, thereby reducing the need for manual feature engineering and enhancing consistency in prediction outcomes. Its capacity to learn directly from raw or structured inputs makes it highly suitable for tasks such as quality classification, where subtle variations may be complex to detect manually. This study investigates the impact of different optimization algorithms on CNN performance, including Stochastic Gradient Descent (SGD), SGD with Momentum, Adam, RMSProp, and Adagrad. Our goal is to find an optimizer that enhances accuracy while maintaining reasonable training time. We found that CNN optimized with the Adam optimizer achieved the highest test accuracy of 85.83%, outperforming the default CNN model (83.33%), with a training time of 146 seconds. This demonstrates the importance of optimizer selection in deep learning applications, especially when dealing with real-world agricultural data. To validate our findings, we used 5-fold cross-validation, confusion matrix analysis, and comparison of training durations. The results suggest that Adam provides a balanced trade-off between speed and classification performance.