Domain Adaptation Techniques for EEG-based Emotion Recognition: A Comparative Study on Two Public Datasets

Zirui Lan, Olga Sourina, Lipo Wang, Reinhold Scherer, Gernot R. Muller-Putz

Research output: Contribution to journalArticleResearchpeer-review

Abstract

Affective brain-computer interface (aBCI) introduces personal affective factors to human-computer interaction. The state-of-the-art aBCI tailors its classifier to each individual user to achieve accurate emotion classification. A subject-independent classifier that is trained on pooled data from multiple subjects generally leads to inferior accuracy, due to the fact that encephalogram (EEG) patterns vary from subject to subject. Transfer learning or domain adaptation techniques have been leveraged to tackle this problem. Existing studies have reported successful applications of domain adaptation techniques on SEED dataset. However, little is known about the effectiveness of the domain adaptation techniques on other affective datasets or in a cross-dataset application. In this paper, we focus on a comparative study on several state-of-the-art domain adaptation techniques on two datasets: DEAP and SEED. We demonstrate that domain adaptation techniques can improve the classification accuracy on both datasets, but not so effective on DEAP as on SEED. Then, we explore the efficacy of domain adaptation in a cross-dataset setting when the data are collected under different environments using different devices and experimental protocols. Here, we propose to apply domain adaptation to reduce the inter-subject variance as well as technical discrepancies between datasets, and then train a subject-independent classifier on one dataset and test on the other. Experiment results show that using domain adaptation technique in a transductive adaptation setting can improve the accuracy significantly by 7.25% -13.40% compared to the baseline accuracy where no domain adaptation technique is used.

Original languageEnglish
Pages (from-to)85-94
JournalIEEE Transactions on Cognitive and Developmental Systems
Volume11
Issue number1
DOIs
Publication statusPublished - Mar 2019

Fingerprint

Brain computer interface
Classifiers
Human computer interaction
Network protocols
Experiments

Keywords

  • affective brain-computer interface (aBCI)
  • Brain-computer interfaces
  • cross dataset.
  • domain adaptation
  • Electroencephalogram (EEG)
  • Electroencephalography
  • Emotion recognition
  • emotion recognition
  • Feature extraction
  • Motion pictures
  • Task analysis
  • Training
  • transfer learning

ASJC Scopus subject areas

  • Software
  • Artificial Intelligence

Fields of Expertise

  • Human- & Biotechnology

Cite this

Domain Adaptation Techniques for EEG-based Emotion Recognition : A Comparative Study on Two Public Datasets. / Lan, Zirui; Sourina, Olga; Wang, Lipo; Scherer, Reinhold; Muller-Putz, Gernot R.

In: IEEE Transactions on Cognitive and Developmental Systems, Vol. 11, No. 1, 03.2019, p. 85-94.

Research output: Contribution to journalArticleResearchpeer-review

@article{e178d558ca8c4d9e8a4ab3125c7f8134,
title = "Domain Adaptation Techniques for EEG-based Emotion Recognition: A Comparative Study on Two Public Datasets",
abstract = "Affective brain-computer interface (aBCI) introduces personal affective factors to human-computer interaction. The state-of-the-art aBCI tailors its classifier to each individual user to achieve accurate emotion classification. A subject-independent classifier that is trained on pooled data from multiple subjects generally leads to inferior accuracy, due to the fact that encephalogram (EEG) patterns vary from subject to subject. Transfer learning or domain adaptation techniques have been leveraged to tackle this problem. Existing studies have reported successful applications of domain adaptation techniques on SEED dataset. However, little is known about the effectiveness of the domain adaptation techniques on other affective datasets or in a cross-dataset application. In this paper, we focus on a comparative study on several state-of-the-art domain adaptation techniques on two datasets: DEAP and SEED. We demonstrate that domain adaptation techniques can improve the classification accuracy on both datasets, but not so effective on DEAP as on SEED. Then, we explore the efficacy of domain adaptation in a cross-dataset setting when the data are collected under different environments using different devices and experimental protocols. Here, we propose to apply domain adaptation to reduce the inter-subject variance as well as technical discrepancies between datasets, and then train a subject-independent classifier on one dataset and test on the other. Experiment results show that using domain adaptation technique in a transductive adaptation setting can improve the accuracy significantly by 7.25{\%} -13.40{\%} compared to the baseline accuracy where no domain adaptation technique is used.",
keywords = "affective brain-computer interface (aBCI), Brain-computer interfaces, cross dataset., domain adaptation, Electroencephalogram (EEG), Electroencephalography, Emotion recognition, emotion recognition, Feature extraction, Motion pictures, Task analysis, Training, transfer learning",
author = "Zirui Lan and Olga Sourina and Lipo Wang and Reinhold Scherer and Muller-Putz, {Gernot R.}",
year = "2019",
month = "3",
doi = "10.1109/TCDS.2018.2826840",
language = "English",
volume = "11",
pages = "85--94",
journal = "IEEE Transactions on Cognitive and Developmental Systems",
issn = "2379-8920",
publisher = "IEEE Press",
number = "1",

}

TY - JOUR

T1 - Domain Adaptation Techniques for EEG-based Emotion Recognition

T2 - A Comparative Study on Two Public Datasets

AU - Lan, Zirui

AU - Sourina, Olga

AU - Wang, Lipo

AU - Scherer, Reinhold

AU - Muller-Putz, Gernot R.

PY - 2019/3

Y1 - 2019/3

N2 - Affective brain-computer interface (aBCI) introduces personal affective factors to human-computer interaction. The state-of-the-art aBCI tailors its classifier to each individual user to achieve accurate emotion classification. A subject-independent classifier that is trained on pooled data from multiple subjects generally leads to inferior accuracy, due to the fact that encephalogram (EEG) patterns vary from subject to subject. Transfer learning or domain adaptation techniques have been leveraged to tackle this problem. Existing studies have reported successful applications of domain adaptation techniques on SEED dataset. However, little is known about the effectiveness of the domain adaptation techniques on other affective datasets or in a cross-dataset application. In this paper, we focus on a comparative study on several state-of-the-art domain adaptation techniques on two datasets: DEAP and SEED. We demonstrate that domain adaptation techniques can improve the classification accuracy on both datasets, but not so effective on DEAP as on SEED. Then, we explore the efficacy of domain adaptation in a cross-dataset setting when the data are collected under different environments using different devices and experimental protocols. Here, we propose to apply domain adaptation to reduce the inter-subject variance as well as technical discrepancies between datasets, and then train a subject-independent classifier on one dataset and test on the other. Experiment results show that using domain adaptation technique in a transductive adaptation setting can improve the accuracy significantly by 7.25% -13.40% compared to the baseline accuracy where no domain adaptation technique is used.

AB - Affective brain-computer interface (aBCI) introduces personal affective factors to human-computer interaction. The state-of-the-art aBCI tailors its classifier to each individual user to achieve accurate emotion classification. A subject-independent classifier that is trained on pooled data from multiple subjects generally leads to inferior accuracy, due to the fact that encephalogram (EEG) patterns vary from subject to subject. Transfer learning or domain adaptation techniques have been leveraged to tackle this problem. Existing studies have reported successful applications of domain adaptation techniques on SEED dataset. However, little is known about the effectiveness of the domain adaptation techniques on other affective datasets or in a cross-dataset application. In this paper, we focus on a comparative study on several state-of-the-art domain adaptation techniques on two datasets: DEAP and SEED. We demonstrate that domain adaptation techniques can improve the classification accuracy on both datasets, but not so effective on DEAP as on SEED. Then, we explore the efficacy of domain adaptation in a cross-dataset setting when the data are collected under different environments using different devices and experimental protocols. Here, we propose to apply domain adaptation to reduce the inter-subject variance as well as technical discrepancies between datasets, and then train a subject-independent classifier on one dataset and test on the other. Experiment results show that using domain adaptation technique in a transductive adaptation setting can improve the accuracy significantly by 7.25% -13.40% compared to the baseline accuracy where no domain adaptation technique is used.

KW - affective brain-computer interface (aBCI)

KW - Brain-computer interfaces

KW - cross dataset.

KW - domain adaptation

KW - Electroencephalogram (EEG)

KW - Electroencephalography

KW - Emotion recognition

KW - emotion recognition

KW - Feature extraction

KW - Motion pictures

KW - Task analysis

KW - Training

KW - transfer learning

UR - http://www.scopus.com/inward/record.url?scp=85048330084&partnerID=8YFLogxK

U2 - 10.1109/TCDS.2018.2826840

DO - 10.1109/TCDS.2018.2826840

M3 - Article

VL - 11

SP - 85

EP - 94

JO - IEEE Transactions on Cognitive and Developmental Systems

JF - IEEE Transactions on Cognitive and Developmental Systems

SN - 2379-8920

IS - 1

ER -