Comparison of Classification Methods for P300 Brain-Computer Interface on Disabled Subjects
Figure 5
Distribution (percentage with respect to all typed characters) of the P300 speller outputs for intensifications. Cells with zero coordinates correspond to correctly spelled characters, while other cells show the results of mistyping. The coordinates of those cells indicate the relative positions of the mistyped and intended characters. The presented results are for the Bayesian linear discriminant analysis (BLDA), the Fisher’s linear discriminant analysis (LDA), the stepwise linear discriminant analysis (SWLDA), a method based on feature extraction (FE), the linear support vector machine (SVM), the multilayer perceptron (NN), and the Gaussian kernel support vector machine (nSVM).
