CoChat

Deep learning for electroencephalogram (EEG) classification tasks: a review

Open access
Craik, Alexander, He, Yongtian, Contreras-Vidal, Jose L · 2019 · Journal of Neural Engineering

The DOI matches its Crossref record; 123 of the 123 references it cites were found in a registry or an index.

Abstract

OBJECTIVE: Electroencephalography (EEG) analysis has been an important tool in neuroscience with applications in neuroscience, neural engineering (e.g. Brain-computer interfaces, BCI's), and even commercial applications. Many of the analytical tools used in EEG studies have used machine learning to uncover relevant information for neural classification and neuroimaging. Recently, the availability of large EEG data sets and advances in machine learning have both led to the deployment of deep learning architectures, especially in the analysis of EEG signals and in understanding the information it may contain for brain functionality. The robust automatic classification of these signals is an important step towards making the use of EEG more practical in many applications and less reliant on trained professionals. Towards this goal, a systematic review of the literature on deep learning applications to EEG classification was performed to address the following critical questions: (1) Which EEG classification tasks have been explored with deep learning? (2) What input formulations have been used for training the deep networks? (3) Are there specific deep learning network structures suitable for specific types of tasks? APPROACH: A systematic literature review of EEG classification using deep learning was performed on Web of Science and PubMed databases, resulting in 90 identified studies. Those studies were analyzed based on type of task, EEG preprocessing methods, input type, and deep learning architecture. MAIN RESULTS: For EEG classification tasks, convolutional neural networks, recurrent neural networks, deep belief networks outperform stacked auto-encoders and multi-layer perceptron neural networks in classification accuracy. The tasks that used deep learning fell into five general groups: emotion recognition, motor imagery, mental workload, seizure detection, event related potential detection, and sleep scoring. For each type of task, we describe the specific input formulation, major characteristics, and end classifier recommendations found through this review. SIGNIFICANCE: This review summarizes the current practices and performance outcomes in the use of deep learning for EEG classification. Practical suggestions on the selection of many hyperparameters are provided in the hope that they will promote or guide the deployment of deep learning to EEG datasets in future research.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1088/1741-2552/ab0ab5

Cites

123 of 123 references verified0 citation errors · 0 not checkable

Statistics

Figures

Figure 1 PRISMA study selection diagram. The diagram moves through the four stages of PRISMA study selection: identification, screening, eligibility, and included. This process led this review from an original count of 349 studies to a final count of 90 studies.
Figure 1 PRISMA study selection diagram. The diagram moves through the four stages of PRISMA study selection: identification, screening, eligibility, and included. This process led this review from an original count of 349 studies to a final count of 90 studies.CC-BY · From the PDF
Image 2
CC-BY · From the PDF
Image 3
CC-BY · From the PDF
Figure 3 (A) Input formulations across all studies. The inner circle shows the general input formulation while the outer circle shows more specific choices. (B) General input formulation compared across different tasks. Most tasks had calculated features as inputs, with seizure detection studies instead having a much higher proportion of signal values. Key -CVT: complex value transformation, CSP: common spatial pattern, DE: dynamic energy, FFT: fast Fourier transform, MAD: mean absolute difference, PSD: power spectral density, STFT: short time Fourier transformation, SVD: singular value decomposition, SWD: swarm decomposition
Figure 3 (A) Input formulations across all studies. The inner circle shows the general input formulation while the outer circle shows more specific choices. (B) General input formulation compared across different tasks. Most tasks had calculated features as inputs, with seizure detection studies instead having a much higher proportion of signal values. Key -CVT: complex value transformation, CSP: common spatial pattern, DE: dynamic energy, FFT: fast Fourier transform, MAD: mean absolute difference, PSD: power spectral density, STFT: short time Fourier transformation, SVD: singular value decomposition, SWD: swarm decompositionCC-BY · From the PDF
Image 5
CC-BY · From the PDF
Image 6
CC-BY · From the PDF
Figure 6 The proportions of input formulations by deep learning architecture types. While there was not a clear consensus when looking at all studies together, studies that employed either MLP ' s or SAE ' s showed a clear preference towards using calculated features. DBN ' s were solely split between using signal values and calculated features, whereas both CNN and RNN studies saw instances of all three input formulations.
Figure 6 The proportions of input formulations by deep learning architecture types. While there was not a clear consensus when looking at all studies together, studies that employed either MLP ' s or SAE ' s showed a clear preference towards using calculated features. DBN ' s were solely split between using signal values and calculated features, whereas both CNN and RNN studies saw instances of all three input formulations.CC-BY · From the PDF
Figure 7 Task-specific deep learning recommendation diagram. The workflow begins with task type (with connected boxes indicating the task ' s general deep learning architecture recommendation) and leads into deep learning architecture characteristic recommendations, which can serve as the starting point for designing deep learning architectures in future research.
Figure 7 Task-specific deep learning recommendation diagram. The workflow begins with task type (with connected boxes indicating the task ' s general deep learning architecture recommendation) and leads into deep learning architecture characteristic recommendations, which can serve as the starting point for designing deep learning architectures in future research.CC-BY · From the PDF

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cites

Show all 123 references

Research works printed but not linked: 24

  1. He Y, 2018, J. Neural Eng., 15
  2. Karpathy A, 2014, Proc. IEEE Computer Society Conf. Computer Vision Pattern Recognition, 1725
  3. Sutskever I, 2011, Proc. of the 28th Int. Conf. on Machine Learning, 1017
  4. Zhao Y, 2014, 2014 Asian Conf. Computer Visison, 1
  5. Roy S, 2018, 40th Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. (EMBC), 2756
  6. Teplan M, 2002, Meas. Sci., 2, 1
  7. Pfurtscheller G, 2001, 89, 1123
  8. Kuanar S, 2018 IEEE Int. Conf. on Acoustics, Speech and Signal Processing (ICASSP), 352
  9. Abbas W, 2018, 40th Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. (EMBC), 219
  10. Wei X, 2018, 18, 111
  11. Xu H, 2016, 2016 IEEE 18th Int. Workshop Multimedia Signal Processing, 1
  12. Vareka L, 2017, Application of Stacked Autoencoders to P300 Experimental Data (Int. Conf. on Artificial Intelligence and Soft Computing)
  13. Yuen T C, 2009, Int. J. Integr. Eng., 1, 71
  14. Alhagry S, 2017, Emotion, 8, 8
  15. Yin Z, 2016, Chinese Control Conf. CCC, 2016, 3907
  16. Pereira A, 2018, Preprint
  17. Ang K K, 2019, J. Neural Eng., 16
  18. Liu J, 2015, Chinese Control Conf. CCC, 2015, 3518
  19. Ansari M L A, 2018, J. Neural Eng., 15
  20. Kobler R J, 2016, 2016 IEEE Int. Conf. Syst. Man, Cybernetics SMC 2016—Conf. Proc., 469
  21. An X, 2014, Int. Conf. on Intelligent Computing, 203
  22. Gao Y, 2015, 2015 IEEE Int. Conf. Multimedia Expo Workshops ICMEW 2015, 2, 1
  23. Bablani A, 2018, 2018 9th Int. Conf. Computing Communication Networking Technologies, 1
  24. Luo T, 2018, Bioinformatics, 19, 344

Cited by

See all 1,758

Metadata

Authors
Craik, A., He, Y., & Contreras-Vidal, J. L.
Affiliations
University of Houston (US)
Corresponding author
Craik, Alexander
Venue
Journal of Neural Engineering
Published
2019-04-09
Volume
16
Issue
3
Pages
031001
ISSN
  • 1741-2552 (ISSN-L)
  • 1741-2560
OpenAlex
W2915893085
Cited by
1,758
References
122
Citations by year
2026: 308 · 2025: 287 · 2024: 273 · 2023: 290 · 2022: 246 · 2021: 213 · 2020: 116 · 2019: 23
FWCI
18.93
Type
journal article
Language
en
Open access
Licence
Locations
  • Journal of Neural Engineering (publishedVersion, cc-by)
  • PubMed (publishedVersion)
Funders
Center for Advanced Computing and Data Science
Topics
  • Life Sciences › Neuroscience › Cognitive Neuroscience › EEG and Brain-Computer Interfaces
  • Physical Sciences › Computer Science › Artificial Intelligence › Emotion and Mood Recognition
  • Health Sciences › Medicine › Cardiology and Cardiovascular Medicine › ECG Monitoring and Analysis
Keywords
EEG, EEG classification, deep learning, convolutional neural networks, recurrent neural networks, deep belief network, stacked autoencoder, motor imagery, emotion recognition, mental workload, seizure detection, sleep stage classification, EEG preprocessing
MeSH terms
Deep Learning, Animals, Brain, Electroencephalography, Humans, Psychomotor Performance, Neural Networks, Computer, Brain-Computer Interfaces

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

How to cite

Cite CoChat for the checks on this page, and the paper for its own findings.

This page
CoChat. (2026). Citation check: Deep learning for electroencephalogram (EEG) classification tasks: a review. https://app.cochat.ai/papers/10.1088/1741-2552/ab0ab5
The paper
Craik, A., He, Y., & Contreras-Vidal, J. L. (2019). Deep learning for electroencephalogram (EEG) classification tasks: a review. Journal of Neural Engineering. https://doi.org/10.1088/1741-2552/ab0ab5

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.