Evaluating Performance of Different Machine Learning Algorithms for the Acute EMG Hand Gesture Datasets

Abstract

In this paper, different machine learning and tabular learning classification algorithms have been studied and compared on the acute hand-gesture Electromyogram dataset. The comparative study between different models such as KNN, RandomForest, TabNet, etc. depicts that small datasets can achieve high-level accuracy along with the intuition of high-performing neural net architectures through tabular learning approaches like TabNet. The performed analysis produced an accuracy of 99.9% through TabNet while other conventional classifiers also gave satisfactory results with KNN being at highest achieving accuracy of 97.8 %.