/P300-BCI

P300 Classification for EEG-based BCI system with Bayes LDA, SVM, LassoGLM and a Deep CNN methods

Primary LanguageMATLABGNU General Public License v3.0GPL-3.0

P300-classification-methods

A MATLAB toolbox for P300 Classification in EEG-based BCI system with Bayes LDA, SVM, LassoGLM and a Deep CNN methods

Codes and data for the following paper are extended to different methods:

An efficient P300-based brain-computer interface for disabled subjects

1. Introduction.

This package includes the prototype MATLAB codes for P300-based brain-computer interfaces.

The implemented methods include:

  1. Bayesian linear discriminant analysis (Bayes-LDA)
  2. Support-vector machines (SVMs)
  3. Penalized generalized linear models (LassoGLM)
  4. Deep Convolutional Neural Networks (Deep-CNNs)

2. Usage & Dependency.

Dependency:

 https://www.epfl.ch/labs/mmspg/research/page-58317-en-html/bci-2/bci_datasets/
 
 https://github.com/lrkrol/plot_erp

Usage:

It is recommended that you create a /dataset folder for the EPFL dataset without any new and extra codes like the below image.

Run "p300_pattern.m" to analyze the P300 ERP over baseline events.

Run "p300_classifiers.m" to check P300-BCI systems with different classifiers and performances.

image