Classification of same arm movements using different kernels learning support vector machine

A robust model is sought for the identification of electroencephalographic (EEG) signals including movements of three distinct parts of the user's arm, namely hand, elbow and shoulder. This study investigates the classification performances of the same upper limb motor movements using various kernel functions of the support vector machine (SVM). Polynomial, linear and radial basis (RBF) functions were utilized on a dataset of seven able-bodied individuals. The results obtained showed that the RBF kernel function provided superior classification performance compared with the other kernel functions for these specific tasks. The presented approach is promising for implementing mental task classification, especially for identification of the same upper limb movements.