Artificial intelligence for spectral classification to identify the basal stem rot disease in oil palm using dielectric spectroscopy measurements

Basal stem rot (BSR) is one of the diseases that threaten the oil palm plantations in Southeast Asia, particularly in Malaysia and Indonesia. As the oil palm plantations continue to grow, there is a need for time-effective, non-destructive, and more precise techniques for detecting BSR. Dielectric spectroscopy has been proven to be an effective method for noninvasive classification of BSR in oil palm trees. However, due to the nature of the large spectral data for spectroscopy analysis, there is a need to reduce the data without losing the main features for more efficient computation. This study investigated the feasibility of applying genetic algorithm (GA) as a feature selection algorithm to select the most significant frequencies of dielectric spectral data for identifying BSR disease in oil palms. Then, the data at the most significant frequencies were used as the input of four classifiers: linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), k-nearest neighbors (kNN), and naive Bayes (NB). The results showed that the best classification accuracy was achieved using LDA classifier with the accuracy of 86.36%. Without implementing GA, the highest classification accuracy was obtained by using the QDA classifier with an accuracy of 82.22%. These results demonstrate the advantages of applying GA as a feature selection model to enhance spectral classification in the identification of BSR in oil palms using dielectric spectroscopy measurements.