Variable down-selection for brain-computer interfaces

Nuno S. Dias, Mst Kamrunnahar, Paulo M. Mendes, Steven Schiff, Jose H. Correia

Research output: Chapter in Book/Report/Conference proceedingChapter

2 Citations (Scopus)

Abstract

A new formulation of principal component analysis (PCA) that considers group structure in the data is proposed as a variable down-selection method. Optimization of electrode channels is a key problem in brain-computer interfaces (BCI). BCI experiments generate large feature spaces compared to the sample size due to time limitations in EEG sessions. It is essential to understand the importance of the features in terms of physical electrode channels in order to design a high performance yet realistic BCI. The proposed algorithm produces a ranked list of original variables (electrode channels or features), according to their ability to discriminate movement imagery tasks. A linear discrimination analysis (LDA) classifier is applied to the selected variable subset. Evaluation of the down-selection method using synthetic datasets selected more than 83% of relevant variables. Classification of imagery tasks using real BCI datasets resulted in less than 19% classification error. Across-Group Variance (AGV) showed the best classification performance with the largest dimensionality reduction in comparison with other algorithms in common use.

Original languageEnglish (US)
Title of host publicationBiomedical Engineering Systems and Technologies
Subtitle of host publicationInternational Joint Conference, BIOSTEC 2009 Porto, Portugal, January 14-17, 2009, Revised Selected Papers
EditorsAna Fred, Hugo Gamboa, Joaquim Filipe
Pages158-172
Number of pages15
DOIs
StatePublished - Apr 8 2010

Publication series

NameCommunications in Computer and Information Science
Volume52
ISSN (Print)1865-0929

Fingerprint

Brain computer interface
Electrodes
Electroencephalography
Principal component analysis
Classifiers
Experiments

All Science Journal Classification (ASJC) codes

  • Computer Science(all)

Cite this

Dias, N. S., Kamrunnahar, M., Mendes, P. M., Schiff, S., & Correia, J. H. (2010). Variable down-selection for brain-computer interfaces. In A. Fred, H. Gamboa, & J. Filipe (Eds.), Biomedical Engineering Systems and Technologies: International Joint Conference, BIOSTEC 2009 Porto, Portugal, January 14-17, 2009, Revised Selected Papers (pp. 158-172). (Communications in Computer and Information Science; Vol. 52). https://doi.org/10.1007/978-3-642-11721-3_12
Dias, Nuno S. ; Kamrunnahar, Mst ; Mendes, Paulo M. ; Schiff, Steven ; Correia, Jose H. / Variable down-selection for brain-computer interfaces. Biomedical Engineering Systems and Technologies: International Joint Conference, BIOSTEC 2009 Porto, Portugal, January 14-17, 2009, Revised Selected Papers. editor / Ana Fred ; Hugo Gamboa ; Joaquim Filipe. 2010. pp. 158-172 (Communications in Computer and Information Science).
@inbook{ad5a06b87b6049de80d8a50eff49c4d1,
title = "Variable down-selection for brain-computer interfaces",
abstract = "A new formulation of principal component analysis (PCA) that considers group structure in the data is proposed as a variable down-selection method. Optimization of electrode channels is a key problem in brain-computer interfaces (BCI). BCI experiments generate large feature spaces compared to the sample size due to time limitations in EEG sessions. It is essential to understand the importance of the features in terms of physical electrode channels in order to design a high performance yet realistic BCI. The proposed algorithm produces a ranked list of original variables (electrode channels or features), according to their ability to discriminate movement imagery tasks. A linear discrimination analysis (LDA) classifier is applied to the selected variable subset. Evaluation of the down-selection method using synthetic datasets selected more than 83{\%} of relevant variables. Classification of imagery tasks using real BCI datasets resulted in less than 19{\%} classification error. Across-Group Variance (AGV) showed the best classification performance with the largest dimensionality reduction in comparison with other algorithms in common use.",
author = "Dias, {Nuno S.} and Mst Kamrunnahar and Mendes, {Paulo M.} and Steven Schiff and Correia, {Jose H.}",
year = "2010",
month = "4",
day = "8",
doi = "10.1007/978-3-642-11721-3_12",
language = "English (US)",
isbn = "9783642117206",
series = "Communications in Computer and Information Science",
pages = "158--172",
editor = "Ana Fred and Hugo Gamboa and Joaquim Filipe",
booktitle = "Biomedical Engineering Systems and Technologies",

}

Dias, NS, Kamrunnahar, M, Mendes, PM, Schiff, S & Correia, JH 2010, Variable down-selection for brain-computer interfaces. in A Fred, H Gamboa & J Filipe (eds), Biomedical Engineering Systems and Technologies: International Joint Conference, BIOSTEC 2009 Porto, Portugal, January 14-17, 2009, Revised Selected Papers. Communications in Computer and Information Science, vol. 52, pp. 158-172. https://doi.org/10.1007/978-3-642-11721-3_12

Variable down-selection for brain-computer interfaces. / Dias, Nuno S.; Kamrunnahar, Mst; Mendes, Paulo M.; Schiff, Steven; Correia, Jose H.

Biomedical Engineering Systems and Technologies: International Joint Conference, BIOSTEC 2009 Porto, Portugal, January 14-17, 2009, Revised Selected Papers. ed. / Ana Fred; Hugo Gamboa; Joaquim Filipe. 2010. p. 158-172 (Communications in Computer and Information Science; Vol. 52).

Research output: Chapter in Book/Report/Conference proceedingChapter

TY - CHAP

T1 - Variable down-selection for brain-computer interfaces

AU - Dias, Nuno S.

AU - Kamrunnahar, Mst

AU - Mendes, Paulo M.

AU - Schiff, Steven

AU - Correia, Jose H.

PY - 2010/4/8

Y1 - 2010/4/8

N2 - A new formulation of principal component analysis (PCA) that considers group structure in the data is proposed as a variable down-selection method. Optimization of electrode channels is a key problem in brain-computer interfaces (BCI). BCI experiments generate large feature spaces compared to the sample size due to time limitations in EEG sessions. It is essential to understand the importance of the features in terms of physical electrode channels in order to design a high performance yet realistic BCI. The proposed algorithm produces a ranked list of original variables (electrode channels or features), according to their ability to discriminate movement imagery tasks. A linear discrimination analysis (LDA) classifier is applied to the selected variable subset. Evaluation of the down-selection method using synthetic datasets selected more than 83% of relevant variables. Classification of imagery tasks using real BCI datasets resulted in less than 19% classification error. Across-Group Variance (AGV) showed the best classification performance with the largest dimensionality reduction in comparison with other algorithms in common use.

AB - A new formulation of principal component analysis (PCA) that considers group structure in the data is proposed as a variable down-selection method. Optimization of electrode channels is a key problem in brain-computer interfaces (BCI). BCI experiments generate large feature spaces compared to the sample size due to time limitations in EEG sessions. It is essential to understand the importance of the features in terms of physical electrode channels in order to design a high performance yet realistic BCI. The proposed algorithm produces a ranked list of original variables (electrode channels or features), according to their ability to discriminate movement imagery tasks. A linear discrimination analysis (LDA) classifier is applied to the selected variable subset. Evaluation of the down-selection method using synthetic datasets selected more than 83% of relevant variables. Classification of imagery tasks using real BCI datasets resulted in less than 19% classification error. Across-Group Variance (AGV) showed the best classification performance with the largest dimensionality reduction in comparison with other algorithms in common use.

UR - http://www.scopus.com/inward/record.url?scp=77950347204&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77950347204&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-11721-3_12

DO - 10.1007/978-3-642-11721-3_12

M3 - Chapter

AN - SCOPUS:77950347204

SN - 9783642117206

T3 - Communications in Computer and Information Science

SP - 158

EP - 172

BT - Biomedical Engineering Systems and Technologies

A2 - Fred, Ana

A2 - Gamboa, Hugo

A2 - Filipe, Joaquim

ER -

Dias NS, Kamrunnahar M, Mendes PM, Schiff S, Correia JH. Variable down-selection for brain-computer interfaces. In Fred A, Gamboa H, Filipe J, editors, Biomedical Engineering Systems and Technologies: International Joint Conference, BIOSTEC 2009 Porto, Portugal, January 14-17, 2009, Revised Selected Papers. 2010. p. 158-172. (Communications in Computer and Information Science). https://doi.org/10.1007/978-3-642-11721-3_12