Measuring the Accuracy of SVM with Varying Kernel Function for Classification of Indonesian Wayang on Images
Muhathir Muhathir, Al-Khowarizmi Al-Khowarizmi
Abstract
Support vector machine (SVM) is a method that is often used in various studies to do pattern recognition of objects in the form of images. SVM is also a classification technique which is a characteristic of carrying out the training process and testing process. In classifying with data in the form of images, SVM is assisted by a feature extract technique where the process normalizes data so that a good training process can be carried out. however, SVM generally uses a linear kernel function in the testing phase. So that there is interest in this paper by designing a thought to make comparisons with several other kernel function techniques such as the cubic kernel function and the quadratic kernel function in classifying the Indonesian Wayang images which are the legacy of Indonesian ancestors. The results of this paper by varying the kernel function in SVM have an accuracy of the cubic kernel function of 83.4%.