EMG Signal Classification for Human Hand Rehabilitation via Two Machine Learning Techniques: kNN and SVM

In the last few years, electromyography (EMG) has shown a lot of potential in therapy and rehabilitation applications of the human limbs though exoskeletons and prosthesis. The use of Machine Learning (ML) techniques has made huge contributions to biomedical signals classification. Many ML methods have been adopted and the results were very promising. In this paper, we choose two different ML classifiers: the k-Nearest Neighbors (kNN) and the Support Vector Machine (SVM). The main goal is to compare their performance using different combinations of time-domain features. This crucial strategy allows to choose the adequate features in order to obtain good model performance. The experimental results demonstrate that the SVM classifier is more efficient where it gives a higher accuracy compared to the kNN technique.