SVM for Decoding the Human Activity Mode from sEMG Signals

Abstract

Nowadays, the relationship between muscles' electrical activity and body movements has been investigated in many medical applications. This Paper proposes the classification of activity mode of healthy human subjects based on surface Electromyography (sEMG) signals. Support vector machine (SVM) methodology is used to predict human activity mode, using the sEMG signals recorded from four main muscles in flexion and extension of the left leg. The presented method shows promising results with classification accuracies of up to 93%. This method provides a reliable solution for the classification of human activity modes, required in many applications like control of exoskeleton robots.