The repository includes basic code for motor unit decomposition using deep CNN.
hdEMG_DCNN includes different callable modules:
Load EMG and spike data from mat file;
Build and train deep CNN model;
Evaluate deep cnn model and save the results to csv file; etc.
hdEMG_validate is the runable code to train and validate the deep CNN:
Initialize the parameters: step size, window size;
Build and train the deep CNN model;
Test the deep CNN model for prediction/estimation accuracy performance and time efficiency performance;
An example data set to run the code can be found at https://doi.org/10.6084/m9.figshare.14179583.v1.
This repository intends to service as supplementary materials for our publication in Journal of Neural Engineering. Detailed information to be added.