EEG Channel Selection Algorithm for Classification Problem

I'm working on a classification problem using eeg signals, each record is 62 channel * number of timesteps. I want to reuce the number of chaneels before training the deep network, is it the right approach? and if so any ideas about channel reduction algorithms.

* I tried to train the model with each channel and after that perform the majority voting on different combinations of the predicted outputs to get the final predictions but how can I know the number of combinations that can give me the optimum train and test results?

Answer 1

[Drew]

There are many ways to classify EEG signals.

• Here is an example from the matlab documentation: https://www.mathworks.com/help/wavelet/ug/time-frequency-convolutional-network-for-eeg-data-classification.html
• Here is a file exchange entry: https://www.mathworks.com/matlabcentral/fileexchange/155822-eeglab-matlab-deep-learning-demo
• And more can be found ...

If this answer helps you, please remember to accept the answer.