THE WINDOWING EFFECT IN CEREBRAL PATTERN CLASSIFICATION: AN APPLICATION TO BRAIN COMPUTER INTERFACE TECHNOLOGY

摘要

The Brain Computer Interface technology allows the communication between people and mechanical devices controlled by microprocessors. It translates the human mental activity into device commands. The kernel of this technology is an algorithm that takes samples, filters and classifies the electro-encephalographic signal. In this paper, different types of filtering windows are considered, the main objective is to determine the type of window with best results in the discrimination stage, when the user is thinking about different activities; as secondary result the most relevant features are extracted. With an earlier and better discrimination the classifier would be easier to implement, faster and more reliable.