Optimum Selection of Linear Features for Improving the Accuracy of BCI-Speller

Brain Computer Interface (BCI) is a communication system that enables control of devices or human interaction through brain signal activities without using motor activities.The P300 speller is aimed at helping patients to spell words by utilizing their brain activities. This paper demonstrated theperformance of different features based on their classification accuracies. These features were evaluated on a data set acquiredusing BCI 2000's P300 paradigm provided by the BCIcompetition III for two subjects. The aim of this paper is to investigate the performance of various linear features of P300for character recognition. For feature selection we use stepwise regression. Finally Discriminant Analysis (DA) and Support Vector Machine (SVM) were used for classification.