Recursive Fisher Linear Discriminant for BCI Applications

Author

Huang, D. ; Xiang, C. ; Ge, S.S.

Author_Institution

Nat. Univ. of Singapore, Singapore

fYear

2007

fDate

3-6 Dec. 2007

Firstpage

383

Lastpage

388

Abstract

A novel recursive procedure for extracting discriminant features, termed Recursive Fisher Linear Discriminant (RFLD), is applied to brain-computer interface (BCI) problems. Compared to traditional Fisher Linear Discriminant (FLD), RFLD relaxes the constraint on the total number of features that can be extracted. The new RFLD has been tested on motor imagery classification with the electrocorticography (ECoG) signals. The resulting improvement of performance by the new feature extraction scheme suggests the effectiveness of our method.

Keywords

electroencephalography; feature extraction; handicapped aids; iterative methods; medical signal processing; pattern classification; principal component analysis; brain-computer interface problem; electrocorticography signal; handicapped aids; motor imagery classification; pattern classification; principal component analysis; recursive fisher linear discriminant feature extraction; Application software; Brain computer interfaces; Data mining; Electrodes; Electroencephalography; Feature extraction; Microelectrodes; Muscles; Principal component analysis; Signal to noise ratio;

fLanguage

English

Publisher

ieee

Conference_Titel

Intelligent Sensors, Sensor Networks and Information, 2007. ISSNIP 2007. 3rd International Conference on

Conference_Location

Melbourne, Qld.

Print_ISBN

978-1-4244-1501-4

Electronic_ISBN

978-1-4244-1502-1

Type

conf

DOI

10.1109/ISSNIP.2007.4496874

Filename

4496874