An EEG data fusion approach based on deep learning for brain motor image classification

Fuente: Zenodo
Saved in:
Bibliographic Details
Main Authors: Fedia, Arfaoui, Sellami, Akrem, Farah, Imed Riadh
Format: Recurso digital
Published: Zenodo 2022
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866901888356581376
author Fedia, Arfaoui
Sellami, Akrem
Farah, Imed Riadh
author_facet Fedia, Arfaoui
Sellami, Akrem
Farah, Imed Riadh
contents <p>This paper explores Deep Learning (DL) method to classify Motor Imagery (MI)<br>based on electroencephalogram (EEG) brainwaves data using the two feature extractors<br>Independent Component Analysis (ICA) and Discrete Wavelet Transform (DWT). The contribution of this paper is in the processing phase, more precisely in the intermediate level by<br>applying the fuzzy logic fusion method to combine the information from these two methods.<br>The objective is to predict the MI from EEG signals after applying several algorithms and<br>techniques.<br>Convolutional Neural Network (CNN) was applied in 2a dataset from the brain computer<br>interface (BCI) Competition IV during the classification phase, whence comparison between<br>CNN and different classifiers is made, while considering the comparison between the fusion<br>method and each feature extractor separately.<br>The results proved that CNN is the best classifier for each of the classes and for the overall data set. Thus they showed that the adapted fusion method performed well, hence it<br>improved the sense of class prediction by recording the best classification rates.<br>Keywords : Brain Computer Interface, Motor Imagery, Electroencephalography, Classification, Deep Learning</p>
format Recurso digital
id zenodo_https___doi_org_10_5281_zenodo_15585363
institution Zenodo
language
publishDate 2022
publisher Zenodo
record_format zenodo
spellingShingle An EEG data fusion approach based on deep learning for brain motor image classification
Fedia, Arfaoui
Sellami, Akrem
Farah, Imed Riadh
<p>This paper explores Deep Learning (DL) method to classify Motor Imagery (MI)<br>based on electroencephalogram (EEG) brainwaves data using the two feature extractors<br>Independent Component Analysis (ICA) and Discrete Wavelet Transform (DWT). The contribution of this paper is in the processing phase, more precisely in the intermediate level by<br>applying the fuzzy logic fusion method to combine the information from these two methods.<br>The objective is to predict the MI from EEG signals after applying several algorithms and<br>techniques.<br>Convolutional Neural Network (CNN) was applied in 2a dataset from the brain computer<br>interface (BCI) Competition IV during the classification phase, whence comparison between<br>CNN and different classifiers is made, while considering the comparison between the fusion<br>method and each feature extractor separately.<br>The results proved that CNN is the best classifier for each of the classes and for the overall data set. Thus they showed that the adapted fusion method performed well, hence it<br>improved the sense of class prediction by recording the best classification rates.<br>Keywords : Brain Computer Interface, Motor Imagery, Electroencephalography, Classification, Deep Learning</p>
title An EEG data fusion approach based on deep learning for brain motor image classification
url https://doi.org/10.5281/zenodo.15585363