Comparative study of data fusion algorithms in P300 Based BCI

Abstract

Brain-Computer interfaces (BCI) research aims at developing systems that help those disabled people communicating through the use of computers and their brain waves. The BCI researchers put most of their effort on developing new algorithms to improve the speed and accuracy of the prediction mechanisms in BCI applications. For that reason, this study is examine the four combination methods that used for aggregate information form several trials. These methods include Summing Scores, Ensemble Average, Bayesian theory and Dempster Shafer. The main purpose of this study is to improve the speed of prediction mechanism with keep a good classification accuracy. This study was applied on able and disable subjects. Our study result show that the performance of four methods is comparable on able subjects. But the Dempster shafer theory appears best in performance for disabled subjects.