Transferring subspaces between subjects in brain - Computer interfacing

Wojciech Samek, Frank C. Meinecke, Klaus Muller

Research output: Contribution to journalArticle

89 Citations (Scopus)

Abstract

Compensating changes between a subjects' training and testing session in brain-computer interfacing (BCI) is challenging but of great importance for a robust BCI operation. We show that such changes are very similar between subjects, and thus can be reliably estimated using data from other users and utilized to construct an invariant feature space. This novel approach to learning from other subjects aims to reduce the adverse effects of common nonstationarities, but does not transfer discriminative information. This is an important conceptual difference to standard multisubject methods that, e.g., improve the covariance matrix estimation by shrinking it toward the average of other users or construct a global feature space. These methods do not reduces the shift between training and test data and may produce poor results when subjects have very different signal characteristics. In this paper, we compare our approach to two state-of-the-art multisubject methods on toy data and two datasets of EEG recordings from subjects performing motor imagery. We show that it can not only achieve a significant increase in performance, but also that the extracted change patterns allow for a neurophysiologically meaningful interpretation.

Original languageEnglish
Article number6482603
Pages (from-to)2289-2298
Number of pages10
JournalIEEE Transactions on Biomedical Engineering
Volume60
Issue number8
DOIs
Publication statusPublished - 2013 Aug 5

Fingerprint

Brain
Computer operating procedures
Electroencephalography
Covariance matrix
Play and Playthings
Imagery (Psychotherapy)
Testing
Learning
Datasets

Keywords

  • Brain-computer interface (BCI)
  • common spatial patterns (CSP)
  • nonstationarity
  • transfer learning

ASJC Scopus subject areas

  • Biomedical Engineering

Cite this

Transferring subspaces between subjects in brain - Computer interfacing. / Samek, Wojciech; Meinecke, Frank C.; Muller, Klaus.

In: IEEE Transactions on Biomedical Engineering, Vol. 60, No. 8, 6482603, 05.08.2013, p. 2289-2298.

Research output: Contribution to journalArticle

Samek, Wojciech ; Meinecke, Frank C. ; Muller, Klaus. / Transferring subspaces between subjects in brain - Computer interfacing. In: IEEE Transactions on Biomedical Engineering. 2013 ; Vol. 60, No. 8. pp. 2289-2298.
@article{492b94cd3f204f74a6e2b5fcf782e5ef,
title = "Transferring subspaces between subjects in brain - Computer interfacing",
abstract = "Compensating changes between a subjects' training and testing session in brain-computer interfacing (BCI) is challenging but of great importance for a robust BCI operation. We show that such changes are very similar between subjects, and thus can be reliably estimated using data from other users and utilized to construct an invariant feature space. This novel approach to learning from other subjects aims to reduce the adverse effects of common nonstationarities, but does not transfer discriminative information. This is an important conceptual difference to standard multisubject methods that, e.g., improve the covariance matrix estimation by shrinking it toward the average of other users or construct a global feature space. These methods do not reduces the shift between training and test data and may produce poor results when subjects have very different signal characteristics. In this paper, we compare our approach to two state-of-the-art multisubject methods on toy data and two datasets of EEG recordings from subjects performing motor imagery. We show that it can not only achieve a significant increase in performance, but also that the extracted change patterns allow for a neurophysiologically meaningful interpretation.",
keywords = "Brain-computer interface (BCI), common spatial patterns (CSP), nonstationarity, transfer learning",
author = "Wojciech Samek and Meinecke, {Frank C.} and Klaus Muller",
year = "2013",
month = "8",
day = "5",
doi = "10.1109/TBME.2013.2253608",
language = "English",
volume = "60",
pages = "2289--2298",
journal = "IEEE Transactions on Biomedical Engineering",
issn = "0018-9294",
publisher = "IEEE Computer Society",
number = "8",

}

TY - JOUR

T1 - Transferring subspaces between subjects in brain - Computer interfacing

AU - Samek, Wojciech

AU - Meinecke, Frank C.

AU - Muller, Klaus

PY - 2013/8/5

Y1 - 2013/8/5

N2 - Compensating changes between a subjects' training and testing session in brain-computer interfacing (BCI) is challenging but of great importance for a robust BCI operation. We show that such changes are very similar between subjects, and thus can be reliably estimated using data from other users and utilized to construct an invariant feature space. This novel approach to learning from other subjects aims to reduce the adverse effects of common nonstationarities, but does not transfer discriminative information. This is an important conceptual difference to standard multisubject methods that, e.g., improve the covariance matrix estimation by shrinking it toward the average of other users or construct a global feature space. These methods do not reduces the shift between training and test data and may produce poor results when subjects have very different signal characteristics. In this paper, we compare our approach to two state-of-the-art multisubject methods on toy data and two datasets of EEG recordings from subjects performing motor imagery. We show that it can not only achieve a significant increase in performance, but also that the extracted change patterns allow for a neurophysiologically meaningful interpretation.

AB - Compensating changes between a subjects' training and testing session in brain-computer interfacing (BCI) is challenging but of great importance for a robust BCI operation. We show that such changes are very similar between subjects, and thus can be reliably estimated using data from other users and utilized to construct an invariant feature space. This novel approach to learning from other subjects aims to reduce the adverse effects of common nonstationarities, but does not transfer discriminative information. This is an important conceptual difference to standard multisubject methods that, e.g., improve the covariance matrix estimation by shrinking it toward the average of other users or construct a global feature space. These methods do not reduces the shift between training and test data and may produce poor results when subjects have very different signal characteristics. In this paper, we compare our approach to two state-of-the-art multisubject methods on toy data and two datasets of EEG recordings from subjects performing motor imagery. We show that it can not only achieve a significant increase in performance, but also that the extracted change patterns allow for a neurophysiologically meaningful interpretation.

KW - Brain-computer interface (BCI)

KW - common spatial patterns (CSP)

KW - nonstationarity

KW - transfer learning

UR - http://www.scopus.com/inward/record.url?scp=84880856659&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84880856659&partnerID=8YFLogxK

U2 - 10.1109/TBME.2013.2253608

DO - 10.1109/TBME.2013.2253608

M3 - Article

C2 - 23529075

AN - SCOPUS:84880856659

VL - 60

SP - 2289

EP - 2298

JO - IEEE Transactions on Biomedical Engineering

JF - IEEE Transactions on Biomedical Engineering

SN - 0018-9294

IS - 8

M1 - 6482603

ER -