Comput Stat (2013) 28:5–18
DOI 10.1007/s00180-011-0292-y
ORIGINAL PAPER
On the discovery of events in EEG data utilizing
information fusion
Martin Schels · Stefan Scherer · Michael Glodek ·
Hans A. Kestler · Günther Palm ·
Friedhelm Schwenker
Received: 18 October 2010 / Accepted: 2 November 2011 / Published online: 29 November 2011
© Springer-Verlag 2011
Abstract One way to tackle brain computer interfaces is to consider event related
potentials in electroencephalography, like the well established P300 phenomenon.
In this paper a multiple classifier approach to discover these events in the bioelec-
trical signal and with them whether or not a subject has recognized a particular
pattern, is employed. Dealing with noisy data as well as heavily imbalanced target
class distributions are among the difficulties encountered. Our approach utilizes par-
titions of electrodes to create robust and meaningful individual classifiers, which are
then subsequently combined using decision fusion. Furthermore, a classifier selection
approach using genetic algorithms is evaluated and used for optimization. The pro-
posed approach utilizing information fusion shows promising results (over 0.8 area
under the ROC curve).
Keywords Multiple classifier systems · EEG data · P300 event
M. Schels · S. Scherer · M. Glodek · H. A. Kestler · G. Palm · F. Schwenker (B )
Institute of Neural Information Processing, University of Ulm, Ulm, Germany
e-mail: friedhelm.schwenker@uni-ulm.de
M. Schels
e-mail: martin.schels@uni-ulm.de
M. Glodek
e-mail: michael.glodek@uni-ulm.de
G. Palm
e-mail: guenther.palm@uni-ulm.de
S. Scherer
Speech Communication Laboratory, Trinity College Dublin, Dublin, Ireland
e-mail: scherers@tcd.ie
H. A. Kestler
Internal Medicine I, University Hospital, Ulm, Germany
e-mail: hans.kestler@uni-ulm.de
123