Information spectrum and its application to EEG-based brain-computer interface

Abstract

A typical goal in signal processing is to find a representation in which certain attributes of the signal are made explicit. The most important variables for identifying signal certain attributes are time and features extracted from the signal. In this paper, a novel method has been proposed for simultaneous selection of optimal features and time window for processing EEG signals. In this method, at first the amount of useful information for separating the classes is obtained on a time-feature plane, then by using it, the optimal time window and features containing maximal information about class label are selected simultaneously. The effectiveness of the proposed method is evaluated by using the classification of EEG signals. The tasks to be discriminated are the imaginative hand movement and the resting state. The results demonstrate that the proposed method performed well in several experiments on different subjects and can improve the classification accuracy in the BCI systems.