Performance Assessment of Machine Learning Algorithms and Ensemble Techniques for Hand Gesture Recognition using Electromyographic Signals

The measuring of electrical activity produced by muscular contractions is the objective of electromyography (EMG). In contrast, hand gesture recognition focuses on recognizing a specific gesture made with the hand. The methods for classifying and recognizing six hand motions based on electromyography (EMG) data are discussed in this paper. The experiment uses samples from 5 people from the UCI benchmark dataset of static hand motions. Two series are performed by the subject, each consisting of six (or seven) fundamental motions. Each motion was made for three seconds, followed by a three-second break. In order to recognize the hand movements, performance is assessed using machine learning agorithms. Some machine learning algorithms and ensemble approaches have been examined here for this study. A comparison of ensemble techniques and machine learning algorithms reveals that the bagging tree classifier-based system achieves the greatest recognition rate. With an average classification accuracy of 98.56%, the Bagged Trees model, among Ensemble algorithms was shown to have the best classification performance.