EZSL-GAN: EEG-based Zero-Shot Learning approach using a Generative Adversarial Network

Sunhee Hwang, Kibeom Hong, Guiyoung Son, Hyeran Byun

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Recent studies show that deep neural network can be effective for learning EEG-based classification network. In particular, Recurrent Neural Networks (RNN) show competitive performance to learn the sequential information of the EEG signals. However, none of the previous approaches considers recognizing the unknown EEG signals which have never been seen in the training dataset. In this paper, we first propose a new scheme for Zero-Shot EEG signal classification. Our EZSL-GAN has three parts. The first part is an EEG encoder network that generates 128-dim of EEG features using a Gated Recurrent Unit (GRU). The second part is a Generative Adversarial Network (GAN) that can tackle the problem for recognizing unknown EEG labels with a knowledge base. The third part is a simple classification network to learn unseen EEG signals from the fake EEG features which are generated from the learned Generator. We evaluate our method on the EEG dataset evoked from 40 classes visual object stimuli. The experimental results show that our EEG encoder achieves an accuracy of 95.89%. Furthermore, our Zero-Shot EEG classification method reached an accuracy of 39.65% for the ten untrained EEG classes. Our experiments demonstrate that unseen EEG labels can be recognized by the knowledge base.

Original languageEnglish
Title of host publication7th International Winter Conference on Brain-Computer Interface, BCI 2019
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781538681169
DOIs
Publication statusPublished - 2019 Feb
Event7th International Winter Conference on Brain-Computer Interface, BCI 2019 - Gangwon, Korea, Republic of
Duration: 2019 Feb 182019 Feb 20

Publication series

Name7th International Winter Conference on Brain-Computer Interface, BCI 2019

Conference

Conference7th International Winter Conference on Brain-Computer Interface, BCI 2019
CountryKorea, Republic of
CityGangwon
Period19/2/1819/2/20

Fingerprint

Electroencephalography
Learning
Knowledge Bases
Labels
Recurrent neural networks

All Science Journal Classification (ASJC) codes

  • Artificial Intelligence
  • Human-Computer Interaction
  • Signal Processing
  • Neuroscience (miscellaneous)

Cite this

Hwang, S., Hong, K., Son, G., & Byun, H. (2019). EZSL-GAN: EEG-based Zero-Shot Learning approach using a Generative Adversarial Network. In 7th International Winter Conference on Brain-Computer Interface, BCI 2019 [8737322] (7th International Winter Conference on Brain-Computer Interface, BCI 2019). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/IWW-BCI.2019.8737322
Hwang, Sunhee ; Hong, Kibeom ; Son, Guiyoung ; Byun, Hyeran. / EZSL-GAN : EEG-based Zero-Shot Learning approach using a Generative Adversarial Network. 7th International Winter Conference on Brain-Computer Interface, BCI 2019. Institute of Electrical and Electronics Engineers Inc., 2019. (7th International Winter Conference on Brain-Computer Interface, BCI 2019).
@inproceedings{0a5b508d9657472fa391c537695781a2,
title = "EZSL-GAN: EEG-based Zero-Shot Learning approach using a Generative Adversarial Network",
abstract = "Recent studies show that deep neural network can be effective for learning EEG-based classification network. In particular, Recurrent Neural Networks (RNN) show competitive performance to learn the sequential information of the EEG signals. However, none of the previous approaches considers recognizing the unknown EEG signals which have never been seen in the training dataset. In this paper, we first propose a new scheme for Zero-Shot EEG signal classification. Our EZSL-GAN has three parts. The first part is an EEG encoder network that generates 128-dim of EEG features using a Gated Recurrent Unit (GRU). The second part is a Generative Adversarial Network (GAN) that can tackle the problem for recognizing unknown EEG labels with a knowledge base. The third part is a simple classification network to learn unseen EEG signals from the fake EEG features which are generated from the learned Generator. We evaluate our method on the EEG dataset evoked from 40 classes visual object stimuli. The experimental results show that our EEG encoder achieves an accuracy of 95.89{\%}. Furthermore, our Zero-Shot EEG classification method reached an accuracy of 39.65{\%} for the ten untrained EEG classes. Our experiments demonstrate that unseen EEG labels can be recognized by the knowledge base.",
author = "Sunhee Hwang and Kibeom Hong and Guiyoung Son and Hyeran Byun",
year = "2019",
month = "2",
doi = "10.1109/IWW-BCI.2019.8737322",
language = "English",
series = "7th International Winter Conference on Brain-Computer Interface, BCI 2019",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
booktitle = "7th International Winter Conference on Brain-Computer Interface, BCI 2019",
address = "United States",

}

Hwang, S, Hong, K, Son, G & Byun, H 2019, EZSL-GAN: EEG-based Zero-Shot Learning approach using a Generative Adversarial Network. in 7th International Winter Conference on Brain-Computer Interface, BCI 2019., 8737322, 7th International Winter Conference on Brain-Computer Interface, BCI 2019, Institute of Electrical and Electronics Engineers Inc., 7th International Winter Conference on Brain-Computer Interface, BCI 2019, Gangwon, Korea, Republic of, 19/2/18. https://doi.org/10.1109/IWW-BCI.2019.8737322

EZSL-GAN : EEG-based Zero-Shot Learning approach using a Generative Adversarial Network. / Hwang, Sunhee; Hong, Kibeom; Son, Guiyoung; Byun, Hyeran.

7th International Winter Conference on Brain-Computer Interface, BCI 2019. Institute of Electrical and Electronics Engineers Inc., 2019. 8737322 (7th International Winter Conference on Brain-Computer Interface, BCI 2019).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - EZSL-GAN

T2 - EEG-based Zero-Shot Learning approach using a Generative Adversarial Network

AU - Hwang, Sunhee

AU - Hong, Kibeom

AU - Son, Guiyoung

AU - Byun, Hyeran

PY - 2019/2

Y1 - 2019/2

N2 - Recent studies show that deep neural network can be effective for learning EEG-based classification network. In particular, Recurrent Neural Networks (RNN) show competitive performance to learn the sequential information of the EEG signals. However, none of the previous approaches considers recognizing the unknown EEG signals which have never been seen in the training dataset. In this paper, we first propose a new scheme for Zero-Shot EEG signal classification. Our EZSL-GAN has three parts. The first part is an EEG encoder network that generates 128-dim of EEG features using a Gated Recurrent Unit (GRU). The second part is a Generative Adversarial Network (GAN) that can tackle the problem for recognizing unknown EEG labels with a knowledge base. The third part is a simple classification network to learn unseen EEG signals from the fake EEG features which are generated from the learned Generator. We evaluate our method on the EEG dataset evoked from 40 classes visual object stimuli. The experimental results show that our EEG encoder achieves an accuracy of 95.89%. Furthermore, our Zero-Shot EEG classification method reached an accuracy of 39.65% for the ten untrained EEG classes. Our experiments demonstrate that unseen EEG labels can be recognized by the knowledge base.

AB - Recent studies show that deep neural network can be effective for learning EEG-based classification network. In particular, Recurrent Neural Networks (RNN) show competitive performance to learn the sequential information of the EEG signals. However, none of the previous approaches considers recognizing the unknown EEG signals which have never been seen in the training dataset. In this paper, we first propose a new scheme for Zero-Shot EEG signal classification. Our EZSL-GAN has three parts. The first part is an EEG encoder network that generates 128-dim of EEG features using a Gated Recurrent Unit (GRU). The second part is a Generative Adversarial Network (GAN) that can tackle the problem for recognizing unknown EEG labels with a knowledge base. The third part is a simple classification network to learn unseen EEG signals from the fake EEG features which are generated from the learned Generator. We evaluate our method on the EEG dataset evoked from 40 classes visual object stimuli. The experimental results show that our EEG encoder achieves an accuracy of 95.89%. Furthermore, our Zero-Shot EEG classification method reached an accuracy of 39.65% for the ten untrained EEG classes. Our experiments demonstrate that unseen EEG labels can be recognized by the knowledge base.

UR - http://www.scopus.com/inward/record.url?scp=85068328315&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85068328315&partnerID=8YFLogxK

U2 - 10.1109/IWW-BCI.2019.8737322

DO - 10.1109/IWW-BCI.2019.8737322

M3 - Conference contribution

AN - SCOPUS:85068328315

T3 - 7th International Winter Conference on Brain-Computer Interface, BCI 2019

BT - 7th International Winter Conference on Brain-Computer Interface, BCI 2019

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Hwang S, Hong K, Son G, Byun H. EZSL-GAN: EEG-based Zero-Shot Learning approach using a Generative Adversarial Network. In 7th International Winter Conference on Brain-Computer Interface, BCI 2019. Institute of Electrical and Electronics Engineers Inc. 2019. 8737322. (7th International Winter Conference on Brain-Computer Interface, BCI 2019). https://doi.org/10.1109/IWW-BCI.2019.8737322