MLSP Competition, 2010: Description of third place method.

Author

Labbé, Benjamin ; Tian, Xilan ; Rakotomamonjy, Alain

Author_Institution

LITIS EA 4108, INSA de Rouen, St. Etienne du Rouvray, France

fYear

2010

fDate

Aug. 29 2010-Sept. 1 2010

Firstpage

116

Lastpage

117

Abstract

The MLSP competition (2010) purpose is to design a pattern recognition system for “mind reading”. This paper is a study of the EEG competition dataset and the crafting of the third place winning method. It shortly presents our signal processing methods for feature extraction, and channel selection. We accurately tuned all the parameters of these preprocessing stage before feeding a Gaussian SVM classifier. Without using any complex signal processing techniques nor strong neuroscience prior knowledge, we were able to build to a simple and fast agnostic classifier which achieves good performances.

Keywords

Gaussian processes; electroencephalography; feature extraction; pattern classification; pattern recognition; support vector machines; EEG competition dataset; Gaussian SVM classifier; MLSP competition; channel selection; feature extraction; pattern recognition system; Bandwidth; Electroencephalography; Feature extraction; Kernel; Signal processing; Support vector machines; Training;

fLanguage

English

Publisher

ieee

Conference_Titel

Machine Learning for Signal Processing (MLSP), 2010 IEEE International Workshop on

Conference_Location

Kittila

ISSN

1551-2541

Print_ISBN

978-1-4244-7875-0

Electronic_ISBN

1551-2541

Type

conf

DOI

10.1109/MLSP.2010.5589242

Filename

5589242