Enhancing Classification by Perceptual Characteristic for the P300 Speller Paradigm

We find that in P300 speller paradigm, there exists difference between electroencephalogram (EEG) signals evoked by flashing adjacent to the row (column) containing the target and those evoked by other non-target stimuli. This difference might arise out of users' perceptual confusion. We use a machine learning approach to learn the difference of this perceptual characteristic, and employ the classifier to aid classification in P300 speller. Experiments show that our method remarkably improves the performance