title	author	date
PersonalizedClassifierSelection	Javad Rahimipour Anaraki, Antonina Kolokolova, and Tom Chau	29/02/2024

Notice

This code is related to the "Personalized Classifier Selection for EEG-based Brain-Computer Interfaces" paper LINK. If you need more details and explanations about the algorithm, please contact Javad Rahimipour Anaraki.

Use case

A systematic methodology for individual classifier selection, wherein structural characteristics of an EEG dataset are used to predict a classifier that will perform with high accuracy.

Code

The code has three parts:

• Convert the data
• Generate features
• Classify

Convert the data

This code reads BCI2000 EDF files, applies ICA and down-sampling (160Hz -> 10Hz), concatenates three sessions of performing Task 2 (i.e. 4, 8, and 12), and finally stores the results to a CSV file for each participant. The original data and the paper for the BCI2000 dataset can be downloaded from https://physionet.org/content/eegmmidb/1.0.0/ and https://pubmed.ncbi.nlm.nih.gov/15188875/, respectively.

Generate features

This code generates 41 structural features and forms a classifier dataset.

Classify

This code uses PCA to extract features from the classifier dataset and classify the reduced dataset using RF.

Statistical results

To generate statistical results, we used KEEL software to perform nonparametric statistical comparisons.

Requirements

Install the requirements using:

conda env create -f environment.yml

Run

To run the code, run each step using the following commands:

python edf2csv.py
python generate.py
python classifiy.py