A comparison of classification methods for recognizing single-trial ERP in RSVP-based brain-computer interfaces

Abstract

Event-related potentials (ERPs) are one of the most popular control signals for brain-computer interfaces (BCIs). Fast classifying ERPs is vital for the good performance of ERP BCIs. However, due to noisy background electroencephalography (EEG) environments, current ERP-based BCI systems need to collect multiple trials for a reliable output, which is inefficient. This study compared a recently developed algorithm, i.e. discriminative canonical pattern matching (DCPM), with five traditional classification methods, i.e. linear discriminant analysis (LDA), four advanced methods of LDA included stepwise LDA, Bayesian LDA, shrinkage LDA and spatial-temporal discriminant analysis (STDA), for the detection of single-trial ERPs with a small number of training samples. Public dataset from RSVP-speller, which would induce ERPs contained N200 and P300 components in ERPs, was addressed in this study. Study results showed that the DCPM significantly outperformed the other traditional methods in single-trial ERP classification in RSVP-based BCI even with small training samples, suggesting the DCPM is a promising classification algorithm for the ERP-BCI.