Direct comparison of supervised and semi-supervised retraining approaches for co-adaptive BCIs: A.S. and J.B. equally contributed to this study and share first authorship.

Research output: Contribution to journalArticleResearchpeer-review

Abstract

For Brain-Computer interfaces (BCIs), system calibration is a lengthy but necessary process for successful operation. Co-adaptive BCIs aim to shorten training and imply positive motivation to users by presenting feedback already at early stages: After just 5 min of gathering calibration data, the systems are able to provide feedback and engage users in a mutual learning process. In this work, we investigate whether the retraining stage of co-adaptive BCIs can be adapted to a semi-supervised concept, where only a small amount of labeled data is available and all additional data needs to be labeled by the BCI itself. The aim of the current work was to evaluate whether a semi-supervised co-adaptive BCI could successfully compete with a supervised co-adaptive BCI model. In a supporting two-class (190 trials per condition) BCI study based on motor imagery tasks, we evaluated both approaches in two separate groups of 10 participants online, while we simulated the other approach in each group offline. Our results indicate that despite the lack of true labeled data, the semi-supervised driven BCI did not perform significantly worse (p > 0.05) than the supervised counterpart. We believe that these findings contribute to developing BCIs for long-term use, where continuous adaptation becomes imperative for maintaining meaningful BCI performance.
Original languageEnglish
Pages (from-to)1-11
Number of pages11
JournalMedical & biological engineering & computing
DOIs
Publication statusE-pub ahead of print - 14 Sep 2019

Fingerprint

Brain computer interface
Calibration
Feedback

Keywords

  • Brain–computer interface (BCI)
  • Motor imagery
  • Co-adaptive BCI
  • Semi-supervised learning
  • Supervised learning

Fields of Expertise

  • Human- & Biotechnology
  • Information, Communication & Computing

Cite this

@article{4ee181c40c4540a7801ff011a917bf41,
title = "Direct comparison of supervised and semi-supervised retraining approaches for co-adaptive BCIs: A.S. and J.B. equally contributed to this study and share first authorship.",
abstract = "For Brain-Computer interfaces (BCIs), system calibration is a lengthy but necessary process for successful operation. Co-adaptive BCIs aim to shorten training and imply positive motivation to users by presenting feedback already at early stages: After just 5 min of gathering calibration data, the systems are able to provide feedback and engage users in a mutual learning process. In this work, we investigate whether the retraining stage of co-adaptive BCIs can be adapted to a semi-supervised concept, where only a small amount of labeled data is available and all additional data needs to be labeled by the BCI itself. The aim of the current work was to evaluate whether a semi-supervised co-adaptive BCI could successfully compete with a supervised co-adaptive BCI model. In a supporting two-class (190 trials per condition) BCI study based on motor imagery tasks, we evaluated both approaches in two separate groups of 10 participants online, while we simulated the other approach in each group offline. Our results indicate that despite the lack of true labeled data, the semi-supervised driven BCI did not perform significantly worse (p > 0.05) than the supervised counterpart. We believe that these findings contribute to developing BCIs for long-term use, where continuous adaptation becomes imperative for maintaining meaningful BCI performance.",
keywords = "Brain–computer interface (BCI), Motor imagery, Co-adaptive BCI, Semi-supervised learning, Supervised learning",
author = "Andreas Schwarz and Julia Brandstetter and Joana Pereira and Gernot M{\"u}ller-Putz",
year = "2019",
month = "9",
day = "14",
doi = "10.1007/s11517-019-02047-1",
language = "English",
pages = "1--11",
journal = "Medical & biological engineering & computing",
issn = "0140-0118",
publisher = "Springer Verlag",

}

TY - JOUR

T1 - Direct comparison of supervised and semi-supervised retraining approaches for co-adaptive BCIs

T2 - A.S. and J.B. equally contributed to this study and share first authorship.

AU - Schwarz, Andreas

AU - Brandstetter, Julia

AU - Pereira, Joana

AU - Müller-Putz, Gernot

PY - 2019/9/14

Y1 - 2019/9/14

N2 - For Brain-Computer interfaces (BCIs), system calibration is a lengthy but necessary process for successful operation. Co-adaptive BCIs aim to shorten training and imply positive motivation to users by presenting feedback already at early stages: After just 5 min of gathering calibration data, the systems are able to provide feedback and engage users in a mutual learning process. In this work, we investigate whether the retraining stage of co-adaptive BCIs can be adapted to a semi-supervised concept, where only a small amount of labeled data is available and all additional data needs to be labeled by the BCI itself. The aim of the current work was to evaluate whether a semi-supervised co-adaptive BCI could successfully compete with a supervised co-adaptive BCI model. In a supporting two-class (190 trials per condition) BCI study based on motor imagery tasks, we evaluated both approaches in two separate groups of 10 participants online, while we simulated the other approach in each group offline. Our results indicate that despite the lack of true labeled data, the semi-supervised driven BCI did not perform significantly worse (p > 0.05) than the supervised counterpart. We believe that these findings contribute to developing BCIs for long-term use, where continuous adaptation becomes imperative for maintaining meaningful BCI performance.

AB - For Brain-Computer interfaces (BCIs), system calibration is a lengthy but necessary process for successful operation. Co-adaptive BCIs aim to shorten training and imply positive motivation to users by presenting feedback already at early stages: After just 5 min of gathering calibration data, the systems are able to provide feedback and engage users in a mutual learning process. In this work, we investigate whether the retraining stage of co-adaptive BCIs can be adapted to a semi-supervised concept, where only a small amount of labeled data is available and all additional data needs to be labeled by the BCI itself. The aim of the current work was to evaluate whether a semi-supervised co-adaptive BCI could successfully compete with a supervised co-adaptive BCI model. In a supporting two-class (190 trials per condition) BCI study based on motor imagery tasks, we evaluated both approaches in two separate groups of 10 participants online, while we simulated the other approach in each group offline. Our results indicate that despite the lack of true labeled data, the semi-supervised driven BCI did not perform significantly worse (p > 0.05) than the supervised counterpart. We believe that these findings contribute to developing BCIs for long-term use, where continuous adaptation becomes imperative for maintaining meaningful BCI performance.

KW - Brain–computer interface (BCI)

KW - Motor imagery

KW - Co-adaptive BCI

KW - Semi-supervised learning

KW - Supervised learning

U2 - 10.1007/s11517-019-02047-1

DO - 10.1007/s11517-019-02047-1

M3 - Article

SP - 1

EP - 11

JO - Medical & biological engineering & computing

JF - Medical & biological engineering & computing

SN - 0140-0118

ER -