Objective Abstract: A Brain-Computer-Interface (BCI) transforms brain activity into control signal. The aim is to improve the performance. In order to improve the performance, well investigate multivariate and non-linear parameters of the EEG. In order to get under control the curse of dimensionality, a twofold approach will be used. First, parametric autoregressive (AR) models including multivariate and non-linear extensions will be applied. AR parameters are known to be a so-called maximum entropy spectral estimator, which minimizes the number of parameters. In order words, the same number of parameters allows to describe the EEG in more detail. Second, support vector machines (SVM) will be applied, since SVM's are able to handle high-dimensional feature space. Fields of science natural sciencescomputer and information sciencessoftwareengineering and technologyelectrical engineering, electronic engineering, information engineeringelectronic engineeringsignal processing Keywords Brain Computer interface Electroencephalogram biomedical signal processing machine learning Programme(s) FP6-MOBILITY - Human resources and Mobility in the specific programme for research, technological development and demonstration "Structuring the European Research Area" under the Sixth Framework Programme 2002-2006 Topic(s) MOBILITY-2.1 - Marie Curie Intra-European Fellowships (EIF) Call for proposal FP6-2004-MOBILITY-5 See other projects for this call Funding Scheme EIF - Marie Curie actions-Intra-European Fellowships Coordinator FRAUNHOFER INSTITUT FIRST Address Hansastr. 27c Munchen Germany See on map Links Website Opens in new window EU contribution € 0,00
