Linear Discriminant Analysis on Brain Computer Interface

Abstract

This report analyses the application of Linear Discriminant Analysis in Brain Computer Interface technology. It is aimed to obtain an objective evaluation of the discrimination capability achieved when different filtering windows are considered in order to differentiate between three different cerebral activities. In this report the following issues are discussed: quantification of the discrimination capability between the employed cerebral activities, identification of the frequency bands with the highest discrimination power, methodology to weight the amplitude of the previous frequency bands in order to reduce the dimensionality of the feature space and facilitate posterior analysis, without lost of intrinsic characteristics of each cerebral activity, determination of the best preprocess window. Linear Discrimination Analysis is employed in order to reduce the dimensionality of the input feature space; bilateral contrasts between features, inferred from each cerebral activity, are used to determine the discrimination power.