Spatio-temporal feature extraction and classification of Event-Related Potentials
Résumé
Brain-Computer Interfaces (BCI) translate variations in the Electroencephalogram (EEG) into a set of particular commands, in order to control a real world machine. For this purpose, it is necessary to classify reliably EEG signals. Classifying EEG activities is a challenging task since EEG recordings exhibit distinct and individualized spatial and temporal characteristics correlated with noise and various physical and mental activities. To increase classification accuracy, it is thus crucial to enhance the Signal to Noise Ratio (SNR) and to identify relevant spatio-temporal features. This paper presents a method for denoising Event-Related Potential (ERP) data and for identifying discriminant spatio-temporal characteristics. First, a Blind Source Separation (BSS) strategy is used to denoise data and enhance SNR. Second, a resampling procedure based on Global Field Power (GFP) automatically selects temporal windows. Third, a spatially weighted SVM (sw-SVM) learns a spatial filter optimizing the classification performance for each temporal feature. Finally, the so obtained ensemble of sw-SVM classifiers are combined using a weighted combination of all sw-SVM outputs. Results indicate that denoising and identification of spatio-temporal features of ERP enhance the classification accuracy, yield a better understanding of the underlying physiology and provide useful insight about the spatio-temporal characteristics of the ERP.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...