GMM-Based Classification Method for Continuous Prediction in Brain-Computer Interface

Brain-Computer Interface (BCI) requires effective classification algorithms for Electroencephalogram (EEG) signal processing. To train a classifier for continuous prediction, trials in training dataset are first divided into segments. The difficulty here is how to combine the predictions across time to make the final decision of a whole trial as early and as accurately as possible. In this paper, we propose a novel statistical approach based on Gaussian Mixture Models (GMM) to classify the EEG trials by combining the predictions of segments according to the discriminative powers at individual time intervals during a trial. We evaluate the proposed method on two datasets of BCI competition 2003 and 2005. The experimental results have shown that the performance of the proposed method is among the best.