Electrodes selection for microsleeps detection by classification of EEG spectrograms
David Coufal · 2008
In the paper we focus on discrimination between electrodes which are used for EEG signals recording and subsequent detection of car drivers' vigilance levels. The detection is based on the classification of EEG spectrograms which are recorded from 19 electrodes spread over driver's head. We are interested in the selection of a small subset of the most discriminating electrodes for each level. Especially, we are aimed on detection of hazardous states of microsleeps. The selection of electrodes was performed by means of GUHA data mining method and for classification we employed C4.5 classification algorithm.