A channel selection method for EEG classification in emotion assessment based on synchronization likelihood

When assessing human emotion using EEG classification, one of the critical problems is to deal with the very large number of features to be classified. In this paper, we address this problem using synchronization likelihood as a new channel selection method. Applying this method, we could significantly reduce the number of EEG channels to be used in emotion assessment, with only slight (if any) loss of classification performance depending on the used feature. We report and compare the results obtained by employing a linear classifier on different features extracted either from all channels or from the selected subset of channels. These features include synchronization likelihood, Hjorth parameters, and fractal dimension.