Factors that affect classification performance in EEG based brain-computer interfaces

Warning The system is temporarily closed to updates for reporting purpose.

Argunşah, Ali Özgür and Çürüklü, Ali Baran and Çetin, Müjdat and Erçil, Aytül (2007) Factors that affect classification performance in EEG based brain-computer interfaces. In: IEEE Conference on Signal Processing and Communications Applications, Eskişehir, Turkey

PDF - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader

Official URL: http://dx.doi.org/10.1109/SIU.2007.4298842


In this paper, some of the factors that affect classification performance of EEG based Brain-Computer Interfaces (BCI) is studied. Study is specified on P300 speller system which is also an EEG based BCI system. P300 is a physiological signal that represents a response of brain to a given stimulus which occurs right 300ms after the stimulus onset. When this signal occurs, it changes the continuous EEG some micro volts. Since this is not a very distinguished change, some other physiological signals (movement of muscles and heart, blinking or other neural activities) may distort this signal. In order to understand if there is really a P300 component in the signal, consecutive P300 epochs are averaged over trials. In this study, we have been tried two different multi channel data handling methods with two different frequency windows. Resulted data have been classified using Support Vector Machines (SVM). It has been shown that proposed method has a better classification performance.

Item Type:Papers in Conference Proceedings
Subjects:T Technology > TK Electrical engineering. Electronics Nuclear engineering
R Medicine > RC Internal medicine > RC0321 Neuroscience. Biological psychiatry. Neuropsychiatry
ID Code:6542
Deposited By:Müjdat Çetin
Deposited On:29 Oct 2007 19:26
Last Modified:22 May 2019 12:08

Repository Staff Only: item control page