Multiple classification algorithms for the BCI P300 speller diagram using ensemble of SVMs

Brain computer interface is one of the most recent and controversial field in Computer Science which emerged in order to help some handicapped people. This paper investigates different classification algorithms dealing with the BCI P300 speller diagram. The system used is composed of an ensemble of Support vector machines. Three different methods are used namely weighted ensemble of SVM, row & column based SVM ensemble and channel selection with optimized SVM's. Experimental results show that proposed methods obtain better results than published results of competition III dataset II.