Support vector channel selection in BCI

Thomas Navin Lal, Michael Schröder, Thilo Hinterberger, Jason Weston, Martin Bogdan, Niels Birbaumer, Bernhard Schölkopf

Research output: Contribution to journalArticle

334 Citations (Scopus)

Abstract

Designing a brain computer interface (BCI) system one can choose from a variety of features that may be useful for classifying brain activity during a mental task. For the special case of classifying electroencephalogram (EEG) signals we propose the usage of the state of the art feature selection algorithms Recursive Feature Elimination [3] and Zero-Norm Optimization [13] which are based on the training of support vector machines (SVM) [11]. These algorithms can provide more accurate solutions than standard filter methods for feature selection [14]. We adapt the methods for the purpose of selecting EEG channels. For a motor imagery paradigm we show that the number of used channels can be reduced significantly without increasing the classification error. The resulting best channels agree well with the expected underlying cortical activity patterns during the mental tasks. Furthermore we show how time dependent task specific information can be visualized.

Original languageEnglish
Pages (from-to)1003-1010
Number of pages8
JournalIEEE Transactions on Biomedical Engineering
Volume51
Issue number6
DOIs
Publication statusPublished - Jun 2004

Fingerprint

Brain computer interface
Electroencephalography
Feature extraction
Support vector machines
Brain

Keywords

  • Brain computer interface (BCI)
  • Channel relevance
  • Channel selection
  • Electroencephalography (EEG)
  • Feature relevance
  • Feature selection
  • Recursive Feature Elimination (RFE)
  • Support vector machine (SVM)
  • Zero Norm Optimization (10-Opt)

ASJC Scopus subject areas

  • Biomedical Engineering

Cite this

Lal, T. N., Schröder, M., Hinterberger, T., Weston, J., Bogdan, M., Birbaumer, N., & Schölkopf, B. (2004). Support vector channel selection in BCI. IEEE Transactions on Biomedical Engineering, 51(6), 1003-1010. https://doi.org/10.1109/TBME.2004.827827

Support vector channel selection in BCI. / Lal, Thomas Navin; Schröder, Michael; Hinterberger, Thilo; Weston, Jason; Bogdan, Martin; Birbaumer, Niels; Schölkopf, Bernhard.

In: IEEE Transactions on Biomedical Engineering, Vol. 51, No. 6, 06.2004, p. 1003-1010.

Research output: Contribution to journalArticle

Lal, TN, Schröder, M, Hinterberger, T, Weston, J, Bogdan, M, Birbaumer, N & Schölkopf, B 2004, 'Support vector channel selection in BCI', IEEE Transactions on Biomedical Engineering, vol. 51, no. 6, pp. 1003-1010. https://doi.org/10.1109/TBME.2004.827827
Lal TN, Schröder M, Hinterberger T, Weston J, Bogdan M, Birbaumer N et al. Support vector channel selection in BCI. IEEE Transactions on Biomedical Engineering. 2004 Jun;51(6):1003-1010. https://doi.org/10.1109/TBME.2004.827827
Lal, Thomas Navin ; Schröder, Michael ; Hinterberger, Thilo ; Weston, Jason ; Bogdan, Martin ; Birbaumer, Niels ; Schölkopf, Bernhard. / Support vector channel selection in BCI. In: IEEE Transactions on Biomedical Engineering. 2004 ; Vol. 51, No. 6. pp. 1003-1010.
@article{4645553afbf24c87819426accf5a6e57,
title = "Support vector channel selection in BCI",
abstract = "Designing a brain computer interface (BCI) system one can choose from a variety of features that may be useful for classifying brain activity during a mental task. For the special case of classifying electroencephalogram (EEG) signals we propose the usage of the state of the art feature selection algorithms Recursive Feature Elimination [3] and Zero-Norm Optimization [13] which are based on the training of support vector machines (SVM) [11]. These algorithms can provide more accurate solutions than standard filter methods for feature selection [14]. We adapt the methods for the purpose of selecting EEG channels. For a motor imagery paradigm we show that the number of used channels can be reduced significantly without increasing the classification error. The resulting best channels agree well with the expected underlying cortical activity patterns during the mental tasks. Furthermore we show how time dependent task specific information can be visualized.",
keywords = "Brain computer interface (BCI), Channel relevance, Channel selection, Electroencephalography (EEG), Feature relevance, Feature selection, Recursive Feature Elimination (RFE), Support vector machine (SVM), Zero Norm Optimization (10-Opt)",
author = "Lal, {Thomas Navin} and Michael Schr{\"o}der and Thilo Hinterberger and Jason Weston and Martin Bogdan and Niels Birbaumer and Bernhard Sch{\"o}lkopf",
year = "2004",
month = "6",
doi = "10.1109/TBME.2004.827827",
language = "English",
volume = "51",
pages = "1003--1010",
journal = "IEEE Transactions on Biomedical Engineering",
issn = "0018-9294",
publisher = "IEEE Computer Society",
number = "6",

}

TY - JOUR

T1 - Support vector channel selection in BCI

AU - Lal, Thomas Navin

AU - Schröder, Michael

AU - Hinterberger, Thilo

AU - Weston, Jason

AU - Bogdan, Martin

AU - Birbaumer, Niels

AU - Schölkopf, Bernhard

PY - 2004/6

Y1 - 2004/6

N2 - Designing a brain computer interface (BCI) system one can choose from a variety of features that may be useful for classifying brain activity during a mental task. For the special case of classifying electroencephalogram (EEG) signals we propose the usage of the state of the art feature selection algorithms Recursive Feature Elimination [3] and Zero-Norm Optimization [13] which are based on the training of support vector machines (SVM) [11]. These algorithms can provide more accurate solutions than standard filter methods for feature selection [14]. We adapt the methods for the purpose of selecting EEG channels. For a motor imagery paradigm we show that the number of used channels can be reduced significantly without increasing the classification error. The resulting best channels agree well with the expected underlying cortical activity patterns during the mental tasks. Furthermore we show how time dependent task specific information can be visualized.

AB - Designing a brain computer interface (BCI) system one can choose from a variety of features that may be useful for classifying brain activity during a mental task. For the special case of classifying electroencephalogram (EEG) signals we propose the usage of the state of the art feature selection algorithms Recursive Feature Elimination [3] and Zero-Norm Optimization [13] which are based on the training of support vector machines (SVM) [11]. These algorithms can provide more accurate solutions than standard filter methods for feature selection [14]. We adapt the methods for the purpose of selecting EEG channels. For a motor imagery paradigm we show that the number of used channels can be reduced significantly without increasing the classification error. The resulting best channels agree well with the expected underlying cortical activity patterns during the mental tasks. Furthermore we show how time dependent task specific information can be visualized.

KW - Brain computer interface (BCI)

KW - Channel relevance

KW - Channel selection

KW - Electroencephalography (EEG)

KW - Feature relevance

KW - Feature selection

KW - Recursive Feature Elimination (RFE)

KW - Support vector machine (SVM)

KW - Zero Norm Optimization (10-Opt)

UR - http://www.scopus.com/inward/record.url?scp=2442670435&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=2442670435&partnerID=8YFLogxK

U2 - 10.1109/TBME.2004.827827

DO - 10.1109/TBME.2004.827827

M3 - Article

C2 - 15188871

AN - SCOPUS:2442670435

VL - 51

SP - 1003

EP - 1010

JO - IEEE Transactions on Biomedical Engineering

JF - IEEE Transactions on Biomedical Engineering

SN - 0018-9294

IS - 6

ER -