Classification of Hand Movements by Surface Myoelectric Signal Using Artificial-Spiking Neural Network Model

Real-time classification of the myoelectric signal has applications in the field of neuro-rehabilitation systems such as prosthesis. The classifier which is a human-computer-interaction (HCI) controller should be ideally fast and computationally less intensive. In this work, we have done a simulation-based study to estimate the performance of a deep artificial/spiking neural network (ANN) model for classification. The model parameters are tuned for a subject to get a 93.33 % and 89.39 % classification accuracy using the ANN and SNN classifiers respectively. A comparison between the two classifiers is studied in terms of computational complexity, external noise effect and trained parameters approximation.