Subject-independent mental state classification in single trials

Siamac Fazli, Florin Popescu, Márton Danóczy, Benjamin Blankertz, Klaus Muller, Cristian Grozea

Research output: Contribution to journalArticle

127 Citations (Scopus)

Abstract

Current state-of-the-art in Brain Computer Interfacing (BCI) involves tuning classifiers to subject-specific training data acquired from calibration sessions prior to functional BCI use. Using a large database of EEG recordings from 45 subjects, who took part in movement imagination task experiments, we construct an ensemble of classifiers derived from subject-specific temporal and spatial filters. The ensemble is then sparsified using quadratic regression with ℓ 1 regularization such that the final classifier generalizes reliably to data of subjects not included in the ensemble. Our offline results indicate that BCI-naïve users could start real-time BCI use without any prior calibration at only very limited loss of performance.

Original languageEnglish
Pages (from-to)1305-1312
Number of pages8
JournalNeural Networks
Volume22
Issue number9
DOIs
Publication statusPublished - 2009 Nov 1
Externally publishedYes

Fingerprint

Brain
Classifiers
Calibration
Imagination
Electroencephalography
Tuning
Databases
Experiments

Keywords

  • BCI
  • Machine learning
  • Zero-training

ASJC Scopus subject areas

  • Artificial Intelligence
  • Cognitive Neuroscience

Cite this

Fazli, S., Popescu, F., Danóczy, M., Blankertz, B., Muller, K., & Grozea, C. (2009). Subject-independent mental state classification in single trials. Neural Networks, 22(9), 1305-1312. https://doi.org/10.1016/j.neunet.2009.06.003

Subject-independent mental state classification in single trials. / Fazli, Siamac; Popescu, Florin; Danóczy, Márton; Blankertz, Benjamin; Muller, Klaus; Grozea, Cristian.

In: Neural Networks, Vol. 22, No. 9, 01.11.2009, p. 1305-1312.

Research output: Contribution to journalArticle

Fazli, S, Popescu, F, Danóczy, M, Blankertz, B, Muller, K & Grozea, C 2009, 'Subject-independent mental state classification in single trials', Neural Networks, vol. 22, no. 9, pp. 1305-1312. https://doi.org/10.1016/j.neunet.2009.06.003
Fazli S, Popescu F, Danóczy M, Blankertz B, Muller K, Grozea C. Subject-independent mental state classification in single trials. Neural Networks. 2009 Nov 1;22(9):1305-1312. https://doi.org/10.1016/j.neunet.2009.06.003
Fazli, Siamac ; Popescu, Florin ; Danóczy, Márton ; Blankertz, Benjamin ; Muller, Klaus ; Grozea, Cristian. / Subject-independent mental state classification in single trials. In: Neural Networks. 2009 ; Vol. 22, No. 9. pp. 1305-1312.
@article{6af057da8bc84a1592da1cfe79383f40,
title = "Subject-independent mental state classification in single trials",
abstract = "Current state-of-the-art in Brain Computer Interfacing (BCI) involves tuning classifiers to subject-specific training data acquired from calibration sessions prior to functional BCI use. Using a large database of EEG recordings from 45 subjects, who took part in movement imagination task experiments, we construct an ensemble of classifiers derived from subject-specific temporal and spatial filters. The ensemble is then sparsified using quadratic regression with ℓ 1 regularization such that the final classifier generalizes reliably to data of subjects not included in the ensemble. Our offline results indicate that BCI-na{\"i}ve users could start real-time BCI use without any prior calibration at only very limited loss of performance.",
keywords = "BCI, Machine learning, Zero-training",
author = "Siamac Fazli and Florin Popescu and M{\'a}rton Dan{\'o}czy and Benjamin Blankertz and Klaus Muller and Cristian Grozea",
year = "2009",
month = "11",
day = "1",
doi = "10.1016/j.neunet.2009.06.003",
language = "English",
volume = "22",
pages = "1305--1312",
journal = "Neural Networks",
issn = "0893-6080",
publisher = "Elsevier Limited",
number = "9",

}

TY - JOUR

T1 - Subject-independent mental state classification in single trials

AU - Fazli, Siamac

AU - Popescu, Florin

AU - Danóczy, Márton

AU - Blankertz, Benjamin

AU - Muller, Klaus

AU - Grozea, Cristian

PY - 2009/11/1

Y1 - 2009/11/1

N2 - Current state-of-the-art in Brain Computer Interfacing (BCI) involves tuning classifiers to subject-specific training data acquired from calibration sessions prior to functional BCI use. Using a large database of EEG recordings from 45 subjects, who took part in movement imagination task experiments, we construct an ensemble of classifiers derived from subject-specific temporal and spatial filters. The ensemble is then sparsified using quadratic regression with ℓ 1 regularization such that the final classifier generalizes reliably to data of subjects not included in the ensemble. Our offline results indicate that BCI-naïve users could start real-time BCI use without any prior calibration at only very limited loss of performance.

AB - Current state-of-the-art in Brain Computer Interfacing (BCI) involves tuning classifiers to subject-specific training data acquired from calibration sessions prior to functional BCI use. Using a large database of EEG recordings from 45 subjects, who took part in movement imagination task experiments, we construct an ensemble of classifiers derived from subject-specific temporal and spatial filters. The ensemble is then sparsified using quadratic regression with ℓ 1 regularization such that the final classifier generalizes reliably to data of subjects not included in the ensemble. Our offline results indicate that BCI-naïve users could start real-time BCI use without any prior calibration at only very limited loss of performance.

KW - BCI

KW - Machine learning

KW - Zero-training

UR - http://www.scopus.com/inward/record.url?scp=70350221669&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=70350221669&partnerID=8YFLogxK

U2 - 10.1016/j.neunet.2009.06.003

DO - 10.1016/j.neunet.2009.06.003

M3 - Article

VL - 22

SP - 1305

EP - 1312

JO - Neural Networks

JF - Neural Networks

SN - 0893-6080

IS - 9

ER -