Comparison of adaptive features with linear discriminant classifier for Brain computer Interfaces

Author

Vidaurre, Carmen ; Schlögl, Alois

Author_Institution

Intelligent Data Analysis Group at FIRST, Fraunhofer Institute, Kekulestr. 7, Berlin 12489, Germany

fYear

2008

fDate

20-25 Aug. 2008

Firstpage

173

Lastpage

176

Abstract

Many Brain-computer Interfaces (BCI) use band-power estimates with more or less subject-specific optimization of the frequency bands. However, a number of alternative EEG features do not need to select the frequency bands; estimators for these features have been modified for an adaptive use. The popular band power estimates were compared with Adaptive AutoRegressive parameters, Hjorth, Barlow, Wackermann, Brain-Rate and a new feature type called Time Domain Parameter. The results from 21 subjects show that several features provide an equally good or even better performance, while no subject-specific optimization is needed, and they are also preferable to band power when the most discriminating frequency band of a subject is not known.

Keywords

Brain computer interfaces; Data processing; Delay effects; Electrodes; Electroencephalography; Feedback; Frequency estimation; Linear discriminant analysis; Position measurement; Signal processing algorithms; Algorithms; Brain; Computer Simulation; Discriminant Analysis; Electroencephalography; Evoked Potentials, Motor; Humans; Imagination; Linear Models; Models, Neurological; Models, Statistical; Pattern Recognition, Automated; Reproducibility of Results; Sensitivity and Specificity; User-Computer Interface;

fLanguage

English

Publisher

ieee

Conference_Titel

Engineering in Medicine and Biology Society, 2008. EMBS 2008. 30th Annual International Conference of the IEEE

Conference_Location

Vancouver, BC

ISSN

1557-170X

Print_ISBN

978-1-4244-1814-5

Electronic_ISBN

1557-170X

Type

conf

DOI

10.1109/IEMBS.2008.4649118

Filename

4649118