I am an undergraduate student who is trying to classify motor imagery from EEG data!

I have no experience working with EEG or any neuroscience background, I only have a very basic knowledge of how EEG, ERD/ERS and frequency domain work from the papers I've tried to read but don't understand and the basic idea of what they are doing. I also have a reasonably basic knowledge of supervised Machine Learning.

I plan to use ERD/ERS to classify Left foot, Right hand and rest to classify motor imagery in real time!

What is the easiest way to do feature selection on EEG data for ERD/ERS?

So far I have bandpassed the data between 7-30 Hz for Mu and Beta wavelengths as they are the ones related to the motor imagery as well as removing the baseline and just plotted the Power Spectral Density. I passed this data through a basic SVM; linear and gaussian, optimised parameters with poor results (60% accuracy on average). From this graph, I'm not sure how I can extract the features that define each catergory.

Power Spectral Density

I don't know how to do feature extraction and where to start learning the basics.

If someone could guide me and provide the steps on how to proceed I would extremely grateful!

I want to work in the field of Brain Computer Interfaces as I find it fascinating, this is my first step :)

Here is the link to the dataset and its details BCI competition: IVc

  • 2
    $\begingroup$ The narrative of this question is very ambitious. You ask "what is the easiest way to do this?" but at the same time state that you start from zero knowledge about classification, despite the fact that you have already tried an SVM. Do you think you can focus the question a bit? Have you looked at any basic introduction to classification? (e.g. nearest neighbour). $\endgroup$ – A_A Apr 17 at 15:56
  • 1
    $\begingroup$ Yes I have @A_A, let me rephrase, I know the basic supervised classification models such as SVM, KNN, simple neural networks. However I don't know how to do feature selection, the papers I have read on them are too complicated for me due to my limited math background. Will edit my question :) $\endgroup$ – Po Chen Liu Apr 19 at 5:32

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.