Various neurological diseases can disrupt the neuromuscular channels through which the brain communicates with the external world. In certain cases like hemorrhage in the anterior brain stem or degenerative neuromuscular diseases like amyotrophic lateral scleriosis (ALS), the patients suffer from a total motor paralysis [5]. This results in a condition known aslocked-in syndrome, wherein the patient is awake and fully aware but cannot communicate with the outside world due to complete paralysis. For such "locked-in" patients, there is a need for an assistive technology that needs no muscular activity whatsoever. A brain-computer interface (BCI) is a device that uses brain signals to provide a direct, non-muscular communication channel between brain and the outside world [32, 31, 29]. The idea underlying BCIs is to measure electric, magnetic, or other physical manifestations of the brain activity and to translate these into commands for a computer or other devices [21, 15]. For patients with locked-in syndrome,the P300 event-related potential (ERP), evoked in scalp-recorded electroencephalography (EEG) by external stimuli, has proven to be a reliable response for controlling a BCI [9]. In this study we present comparison of some classification methods to classify an EEG signal based on the presence of P300 component.

We adopted an approach of Farwell & Donchin [4], which we tried to improve in several aspects. The main objective was to improve the transfer rates based on offline analysis of EEGdata but within a more realistic setup closer to an online realization than in the original studies. The objective was achieved along two different tracks: on the one hand we used state-of-the-art machine learning techniques for signal classification and on the other hand we augmented the data space by using more electrodes for the interface. For the classification task we utilized SVMs and, as motivated by recent findings on the learning of discriminative densities, we accumulated the values of the classification function in order to combine several classifications, which finally lead to significantly improved rates as compared with techniques applied in the original work. In combination with the data space augmentation, we achieved competitive transfer rates at an average of 50.5 bits/min and with a maximum of 84.7 bits/min.

We adopted an approach of Farwell & Donchin [4], which we tried to improve in several aspects. The main objective was to improve the transfer rates based on offline analysis of EEGdata but within a more realistic setup closer to an online realization than in the original studies. The objective was achieved along two different tracks: on the one hand we used state-of-the-art machine learning techniques for signal classification and on the other hand we augmented the data space by using more electrodes for the interface. For the classification task we utilized SVMs and, as motivated by recent findings on the learning of discriminative densities, we accumulated the values of the classification function in order to combine several classifications, which finally lead to significantly improved rates as compared with techniques applied in the original work. In combination with the data space augmentation, we achieved competitive transfer rates at an average of 50.5 bits/min and with a maximum of 84.7 bits/min.

We adopted an approach of Farwell & Donchin [4], which we tried to improve in several aspects. The main objective was to improve the transfer ratesbased on offline analysis of EEGdata but within a more realistic setup closer to an online realization than in the original studies. The objective wasachieved along two different tracks: on the one hand we used state-of-the-art machine learning techniques for signal classification and on the other hand we augmented the data space by using more electrodes for the interface. For the classification task we utilized SVMs and, as motivated byrecent findings on the learning of discriminative densities, we accumulated the values of the classification function in order to combine several classifications, which finally lead to significantly improved rates as compared with techniques applied in the original work. In combination withthe data space augmentation, we achieved competitive transfer rates at an average of 50.5 bits/min and with a maximum of 84.7 bits/min.