Towards zero training for brain-computer interfacing

Matthias Krauledat, Michael Tangermann, Benjamin Blankertz, Klaus Muller

Research output: Contribution to journalArticle

157 Citations (Scopus)

Abstract

Electroencephalogram (EEG) signals are highly subject-specific and vary considerably even between recording sessions of the some user within the same experimental paradigm. This challenges a stable operation of Brain-Computer Interface (BCI) systems. The classical approach is to train users by neurofeedback to produce fixed stereotypical patterns of brain activity. In the machine learning approach, a widely adapted method for dealing with those variances is to record a so called calibration measurement on the beginning of each session in order to optimize spatial filters and classifiers specifically for each subject and each day. This adaptation of the system to the individual brain signature of each user relieves from the need of extensive user training. In this paper we suggest a new method that overcomes the requirement of these time-consuming calibration recordings for long-term BCI users. The method takes advantage of knowledge collected in previous sessions: By a novel technique, prototypical spatial filters are determined which have better generalization properties compared to single-session filters. In particular, they can be used in follow-up sessions without the need to recalibrate the system. This way the calibration periods can be dramatically shortened or even completely omitted for these 'experienced' BCI users. The feasibility of our novel approach is demonstrated with a series of online BCI experiments. Although performed without any calibration measurement at all, no loss of classification performance was observed.

Original languageEnglish
Article numbere2967
JournalPLoS One
Volume3
Issue number8
DOIs
Publication statusPublished - 2008 Aug 13
Externally publishedYes

Fingerprint

Brain-Computer Interfaces
Brain computer interface
Calibration
Brain
brain
calibration
user interface
Neurofeedback
Computer Systems
Electroencephalography
Learning systems
electroencephalography
Classifiers
artificial intelligence
methodology
Experiments

ASJC Scopus subject areas

  • Agricultural and Biological Sciences(all)
  • Biochemistry, Genetics and Molecular Biology(all)
  • Medicine(all)

Cite this

Krauledat, M., Tangermann, M., Blankertz, B., & Muller, K. (2008). Towards zero training for brain-computer interfacing. PLoS One, 3(8), [e2967]. https://doi.org/10.1371/journal.pone.0002967

Towards zero training for brain-computer interfacing. / Krauledat, Matthias; Tangermann, Michael; Blankertz, Benjamin; Muller, Klaus.

In: PLoS One, Vol. 3, No. 8, e2967, 13.08.2008.

Research output: Contribution to journalArticle

Krauledat, M, Tangermann, M, Blankertz, B & Muller, K 2008, 'Towards zero training for brain-computer interfacing', PLoS One, vol. 3, no. 8, e2967. https://doi.org/10.1371/journal.pone.0002967
Krauledat M, Tangermann M, Blankertz B, Muller K. Towards zero training for brain-computer interfacing. PLoS One. 2008 Aug 13;3(8). e2967. https://doi.org/10.1371/journal.pone.0002967
Krauledat, Matthias ; Tangermann, Michael ; Blankertz, Benjamin ; Muller, Klaus. / Towards zero training for brain-computer interfacing. In: PLoS One. 2008 ; Vol. 3, No. 8.
@article{002f95304ade4ed7a8dfffba532c1876,
title = "Towards zero training for brain-computer interfacing",
abstract = "Electroencephalogram (EEG) signals are highly subject-specific and vary considerably even between recording sessions of the some user within the same experimental paradigm. This challenges a stable operation of Brain-Computer Interface (BCI) systems. The classical approach is to train users by neurofeedback to produce fixed stereotypical patterns of brain activity. In the machine learning approach, a widely adapted method for dealing with those variances is to record a so called calibration measurement on the beginning of each session in order to optimize spatial filters and classifiers specifically for each subject and each day. This adaptation of the system to the individual brain signature of each user relieves from the need of extensive user training. In this paper we suggest a new method that overcomes the requirement of these time-consuming calibration recordings for long-term BCI users. The method takes advantage of knowledge collected in previous sessions: By a novel technique, prototypical spatial filters are determined which have better generalization properties compared to single-session filters. In particular, they can be used in follow-up sessions without the need to recalibrate the system. This way the calibration periods can be dramatically shortened or even completely omitted for these 'experienced' BCI users. The feasibility of our novel approach is demonstrated with a series of online BCI experiments. Although performed without any calibration measurement at all, no loss of classification performance was observed.",
author = "Matthias Krauledat and Michael Tangermann and Benjamin Blankertz and Klaus Muller",
year = "2008",
month = "8",
day = "13",
doi = "10.1371/journal.pone.0002967",
language = "English",
volume = "3",
journal = "PLoS One",
issn = "1932-6203",
publisher = "Public Library of Science",
number = "8",

}

TY - JOUR

T1 - Towards zero training for brain-computer interfacing

AU - Krauledat, Matthias

AU - Tangermann, Michael

AU - Blankertz, Benjamin

AU - Muller, Klaus

PY - 2008/8/13

Y1 - 2008/8/13

N2 - Electroencephalogram (EEG) signals are highly subject-specific and vary considerably even between recording sessions of the some user within the same experimental paradigm. This challenges a stable operation of Brain-Computer Interface (BCI) systems. The classical approach is to train users by neurofeedback to produce fixed stereotypical patterns of brain activity. In the machine learning approach, a widely adapted method for dealing with those variances is to record a so called calibration measurement on the beginning of each session in order to optimize spatial filters and classifiers specifically for each subject and each day. This adaptation of the system to the individual brain signature of each user relieves from the need of extensive user training. In this paper we suggest a new method that overcomes the requirement of these time-consuming calibration recordings for long-term BCI users. The method takes advantage of knowledge collected in previous sessions: By a novel technique, prototypical spatial filters are determined which have better generalization properties compared to single-session filters. In particular, they can be used in follow-up sessions without the need to recalibrate the system. This way the calibration periods can be dramatically shortened or even completely omitted for these 'experienced' BCI users. The feasibility of our novel approach is demonstrated with a series of online BCI experiments. Although performed without any calibration measurement at all, no loss of classification performance was observed.

AB - Electroencephalogram (EEG) signals are highly subject-specific and vary considerably even between recording sessions of the some user within the same experimental paradigm. This challenges a stable operation of Brain-Computer Interface (BCI) systems. The classical approach is to train users by neurofeedback to produce fixed stereotypical patterns of brain activity. In the machine learning approach, a widely adapted method for dealing with those variances is to record a so called calibration measurement on the beginning of each session in order to optimize spatial filters and classifiers specifically for each subject and each day. This adaptation of the system to the individual brain signature of each user relieves from the need of extensive user training. In this paper we suggest a new method that overcomes the requirement of these time-consuming calibration recordings for long-term BCI users. The method takes advantage of knowledge collected in previous sessions: By a novel technique, prototypical spatial filters are determined which have better generalization properties compared to single-session filters. In particular, they can be used in follow-up sessions without the need to recalibrate the system. This way the calibration periods can be dramatically shortened or even completely omitted for these 'experienced' BCI users. The feasibility of our novel approach is demonstrated with a series of online BCI experiments. Although performed without any calibration measurement at all, no loss of classification performance was observed.

UR - http://www.scopus.com/inward/record.url?scp=51849087749&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=51849087749&partnerID=8YFLogxK

U2 - 10.1371/journal.pone.0002967

DO - 10.1371/journal.pone.0002967

M3 - Article

VL - 3

JO - PLoS One

JF - PLoS One

SN - 1932-6203

IS - 8

M1 - e2967

ER -