Classification Of Hand And Wrist Movements Via Surface Electromyogram Using The Random Convolutional Kernels Transform

This article discusses the use of sEMG signals for movement classification and control of hand prostheses. It compares two approaches – feature engineering and deep learning – and reviews recent papers that incorporate ideas from deep learning into feature extraction based techniques. The work of Phinyomark et al. is highlighted for its examination of 37 time and frequency domain features and identification of strong and redundant features.