Dinstinguishing brain signals

In this project I will use an open dataset (from the LIMO task - Guillaume A. Rousselet. LIMO EEG dataset. 2016. doi:10.7488/ds/1556.).

In this dataset, participants were subjected to a task in which they would observe two different faces (face A/face B) across trials. In each trial, the images could be 'less coherent' - meaning more blurred. Then, participants had to report which face they had just seen.

However, the question I'm trying to investigate is: can the EEG signal determine which face the participant was observing?

To this end, I built this script to read the EEG data via MNE and test different decoders to find which more accurately can distinguish from the EEG signal if the participant was subjected to seeing Face A or Face B.

Run Locally

Clone the project

  git clone https://github.com/joaohnp/EEG_decoding

Go to the project directory

  cd my-project

Install dependencies

  pip install -r requeriments.txt

Run EDA.py

  python EDA.py

After EDA has been ran, we now have accuracy scores for different models. With this we can investigate which classifier works best with our data.

  python best_model.py

Now we can check in detail which decoder worked best in which electrode.

Tech Stack

Data preparation and analysis: MNE, numpy

Machine Learning and Deep Learning: TensorFlow, XGBoost, LDA, SVM

Authors

• @joaohnp

Support

For support, email [email protected].