A Robust and Efficient Spatio-Temporal Feature Selection for Interpretation of EEG Single Trials

Abstract

Interpretation of brain states from EEG single trials, multiple electrodes and time points, is addressed. A computationally efficient and robust framework for spatio-temporal feature selection is introduced. The framework is generic and can be applied to different classification tasks. Here, it is applied to a visual task of distinguishing between faces and houses. The framework includes training of regularized logistic regression classifier with cross-validation and the usage of a wrapper approach to find the optimal model. It was compared with two other methods for selection of time points. The spatial-temporal information of brain activity obtained using this framework, can give an indication to correlated activity of regions in the brain (spatial) as well as temporal activity correlations between and within EEG electrodes. This spatial-temporal analysis can render a far more holistic interpretability for visual perception mechanism without any a priori bias on certain time periods or scalp locations.