AUTOMATIC SELECTION KERNEL WITH ENSEMBLE CONCEPT IN SUPPORT VECTOR MACHINE (SVM) FOR CLASSIFICATION OF SOYBEAN PLANT DISEASE
Fendy Yulianto, Erfan Nurkholis Efendi, Wayan Firdaus Mahmudy · 2022
Soybean is one of the plants with high vegetable protein value as a substitute for animal protein. In the process of cultivating soybeans, soybeans are often exposed to diseases caused by pathogens caused by viruses, bacteria and pests. To find out the disease usually requires direct observation to the field which requires an expert, so we need a way to be able to identify the disease automatically without having to directly monitor the field with an expert. SVM method is one method that can be used to classify the types of diseases that attack soybean plants. The SVM method has a lot of Kernel functions that can be used, where the Kernel is the core of the SVM method process, there are many kernels that can be used so that if you choose the wrong Kernel will have an impact on the results obtained. To overcome this problem, this research applies 3 Ensemble Kernel concepts which are used to combine several Kernels into 1, from the tests carried out using the Ensemble concept, better results are obtained than the classical SVM method. There are 3 concepts that can be used, namely bagging, boosting and stacking, Ensemble Kernel SVM with the stacking concept and the ANOVA kernel for the testing process to get the best accuracy value of 76.62%.