Pith. sign in

REVIEW 3 major objections 5 minor 3 cited by

Transformer-based EEG Decoding: A Survey

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Pith's one-line read This survey organizes Transformer-based EEG decoding into three architectural families and, for the first time, four categories of customized Transformer.

desk verdict The survey's taxonomy is a real contribution, but the performance tables carry too many transcription errors to support the paper's comparative claims as printed. read the letter →

arxiv 2507.02320 v1 pith:DMTXZCEU submitted 2025-07-03 cs.LG cs.HC

classification cs.LGcs.HC
keywords EEGdecodingTransformerbrain-computerinterfaceattentionmechanismhybriddeeplearningcustomizedMetaFormersurvey
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This survey sets out to bring order to the fast-growing literature on Transformer models for EEG decoding (using brain signals to infer a user's intent) and claims to be the most thorough overview of the area to date. It screened and compiled about 165 studies published since 2020 into a single structural map. Its organizing claim is that every approach belongs to one of three families: the Transformer backbone used directly, hybrids pairing the Transformer with other deep-learning techniques such as convolutional, graph, recurrent, and generative-adversarial networks, and customized Transformers with re-engineered internal structures. It further claims to be the first to split the customized family into four categories (multi-encoder designs, modified encoders, pyramid architectures, and reconstructed architectures), with the modified-encoder category itself organized by the MetaFormer framing of attention and feed-forward blocks as swappable pieces. A reader would care because the taxonomy converts a scattered literature into a navigable design space with named entry points for future work.

What carries the argument

The load-bearing object is the taxonomy itself: the survey's three-family scheme (backbone, hybrid, customized) and, inside the customized family, a four-way split that the authors present as new. The modified-encoder branch is organized by the MetaFormer idea, which redefines the multi-head attention block as an interchangeable 'Token Mixer' and the MLP block as a 'Token Enhancer'; that framing lets the survey group attention replacements (sparse, linear, spiking, retentive), MLP replacements (multi-branch or convolutional feed-forward networks), full replacements of both blocks, and add-on 'affiliate' modules such as gating, convolution, and distillation blocks. The scheme does the work of turning a scattered literature into named categories, each populated with example models, tasks, datasets, and reported results, so that a researcher can locate any approach and see the open design space.

What would settle it

Re-derive Tables II and III from the cited sources. Where the same accuracy value is listed for several different models on the same dataset (for example, 98.47±1.72% on SEED for ACTNN, DANet, and MV-SSTMA; 96.28% on SEED for both EeT and MV-SSTMA; and 74.50% on BCI-IV 2a for EEG Conformer and GAT in Table II while Table I credits GAT with 76.58% on the same benchmark), check what the original papers actually report. If the sources disagree with the survey's numbers, the comparative claims built on those tables cannot be trusted as stated.

Watch

Extended reading notes

Core claim

The paper argues that the roughly 165 Transformer-based EEG decoding studies it screened can be organized along the line of architectural evolution into three families (direct backbone usage, hybrids with other deep-learning networks, and customized Transformers), and that this organization is the most thorough overview of the field to date. Direct backbone models apply the standard Transformer encoder or a Vision Transformer to extract and fuse temporal, spatial, and spectral EEG features. Hybrid models pair the Transformer with convolutional, recurrent, graph, spiking, capsule, generative-adversarial, or diffusion networks so the partner handles local patterns and the Transformer handles global dependencies. The customized family, which the paper says it sorts for the first time, splits into four categories: multi-encoder architectures including TNT-like local-plus-global designs, modified encoders organized through the MetaFormer framing (Token Mixer replacements for multi-head attention, Token Enhancer replacements for the feed-forward block, Extended MetaFormer replacing both, and EM with Affiliate Modules), pyramid Transformers following PVT and Swin designs, and reconstructed architectures that reshape the whole network for a specific task. With the classification in place, the survey further claims to observe that most models fuse temporal-spatial-spectral features rather than mining one domain, and that the field's binding constraints are data scarcity, generalization limits, computational cost, and weak interpretability.

Load-bearing premise

The survey's comparative conclusions assume that the accuracy figures it copied from the cited papers are correct, yet its own tables list identical values for several different models on the same dataset (for instance, three models all at 98.47±1.72% on SEED), which points to transcription errors that would undercut any performance ranking drawn from those tables.

Editorial extensions

If this is right

  • A researcher can take any published Transformer-based EEG decoder, place it in one of the three families (and, for customized models, one of the four sub-categories), and immediately see which design space it occupies without reading all 165 source papers.
  • Because most surveyed models win by fusing temporal, spatial, and spectral features rather than specializing in one domain, the survey's own evidence points future work toward frequency-only modeling as the under-explored corner and toward multimodal fusion as the default strategy.
  • The four challenges the paper names (data scarcity, limited generalization, computational cost, and weak interpretability) define the field's research agenda, with brain foundation models, synthetic EEG generation, cross-domain transfer, and parameter-efficient customization as the proposed routes.
  • The compiled trends support concrete architecture-selection heuristics: CNN-Transformer hybrids appear across several benchmark tasks with top reported accuracies, GNN-Transformer hybrids cluster in emotion recognition, and multi-encoder TNT-like models are frequently applied to sleep staging.
  • The parameter survey (from roughly 1K-parameter EEGNet up to brain foundation models in the hundreds of millions to billions of parameters) gives practitioners a quantitative scale for trading expressiveness against deployability.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the taxonomy proves durable, it gives the field a stable reporting geometry: future architectures can be classified on entry, which would make cross-family performance comparisons meaningful and would let the community track how the design space fills in over time.
  • The four-way split of customized encoders is built from a signal-agnostic framing (MetaFormer), so the same template could organize surveys of Transformer use in other biosignals such as EMG, EOG, or ECG once those literatures grow large enough.
  • The survey's observation that temporal-spatial-spectral fusion is the common winning pattern is testable: a controlled study that holds the feature input fixed while varying architectural family would reveal whether structure or representation drives the reported accuracies.
  • The 'most thorough' claim is bounded by the survey's own screening criteria (studies without comparative experiments were excluded), so the map describes comparatively validated work rather than the entire field, and a novel architecture reported without benchmark comparisons would not appear in it.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. This manuscript surveys Transformer-based EEG decoding from roughly 165 papers published since 2020. Its organizing framework is architectural: backbone Transformers, hybrid models (CNN, GAN, diffusion, GNN, RNN/LSTM, SNN, capsule), and customized Transformers, with the latter subdivided into multi-encoder, modified-encoder, pyramid, and reconstructed architectures. The survey also compiles performance tables, dataset tables, feature-use tables, and parameter-count tables, and it derives model-task alignment observations from these tables in Section V. The main claimed contributions are the most thorough overview to date, a first-time sorting of customized Transformer intrinsic structures, and a discussion of challenges and future directions.

Significance. If the data extraction were reliable, this would be a useful reference for the EEG-decoding community: the taxonomy is coherent, the coverage is broad, and the separation of modified encoders into token-mixer, token-enhancer, extended MetaFormer, and affiliate-module variants is a genuinely helpful organizing scheme. The supplementary tables on datasets and parameter scales are also valuable resources. However, the central comparative claims in Section V rest on performance tables with widespread internal inconsistencies, so the empirical component of the contribution is not currently dependable. The taxonomy itself appears sound, but the paper cannot serve as a trustworthy reference survey until every table entry is reconciled with its primary source.

major comments (3)
  1. [Table II and Section V] Table II lists the SEED accuracy 98.47 ± 1.72% identically for ACTNN, DANet, and MV-SSTMA, and the CHB-MIT AUC 93.5% identically for MdC, HviT-DUL, and TGCNN. These are implausible as independent results and indicate transcription or copying errors. Because Section V explicitly derives model-task preferences (e.g., that CNN-Transformer performs well on SEED, BCI-IV 2a, Sleep-EDFx, and CHB-MIT; that GNN-Transformer suits emotion recognition) directly from this table, the comparative conclusions are not supported as printed. Every entry in Table II must be checked against the cited papers, and any genuine duplication should be explained or removed.
  2. [Table III] Table III contains a literal duplicate row for SECT (both rows list the same reference [152], tasks, model characteristics, and SEED/DEAP/HIED results), and the MEET and EEG-Deformer rows report identical SEED, SEED-IV, and WM results (99.18 ± 0.34, 95.18 ± 1.92, and 98.14 ± 1.97). In addition, the ST-DG row cites [78], the same reference used for MI-CAT in the same table, which cannot be correct for two different models. These errors directly affect the paper's credibility as a survey and its claims about customized Transformer variants, so the table must be regenerated from the primary sources.
  3. [Tables I and II cross-check] Cross-table inconsistencies make even individual model comparisons unreliable. GAT is reported as 76.58% accuracy with kappa 0.6877 on BCI-IV 2a in Table I, but Table II reports it as 74.50% with kappa 0.6877 on the same dataset. Similarly, the MdC row in Table II reports CHB-MIT accuracy 79.39%, while Table I for M-d-C reports CHB-MIT AUC 93.5%, SEN 97.8%, and FPR 0.059; the 79.39% value in Table II is the same number reported for ADFCNN on BCI-IV 2a. These discrepancies mean that a reader cannot use the comparison tables to infer the relative performance of any of these models, and all claims in Section V that depend on those tables need to be revisited.
minor comments (5)
  1. [Section V] Section V says 'Combined with Section III. B, GNN-Transformer models are more commonly applied in emotion recognition tasks,' but the GNN-Transformer discussion is in Section III.D, not Section III.B.
  2. [Page 1] The text contains a duplicated editorial fragment 'interfaces (BCIS/BMIS) have garnered much attention over the past decades...' immediately after the first introduction paragraph, which should be removed.
  3. [References] Reference [78] is used for both MI-CAT and ST-DG in Table III; these are different works and must be assigned distinct reference numbers, with the bibliography corrected accordingly.
  4. [Tables] Several entries in Table III use informal notation such as 'Acc about 96%' and 'Acc about 80%'; all numeric values should be reported with the same precision and explicit units so that the tables can be compared meaningfully.
  5. [Abstract and I-A] The claim to provide 'the most thorough overview to date' is difficult to substantiate and should be softened or supported by a precise comparison with the existing surveys cited in Section I.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the survey organizes existing literature and makes no prediction or derivation that reduces to its own inputs.

full rationale

This manuscript is a literature survey, not a derivation chain. Its central claims are taxonomic: it sorts Transformer-based EEG models into backbone, hybrid, and customized categories and, for the first time, subdivides customized intrinsic structures. These claims are supported by cited primary studies and by the paper's own descriptive figures, not by any equation or fitted parameter that is equivalent to the conclusion. Equations (1)-(4) restate the standard attention and multi-head attention formulas from Vaswani et al., which are external, well-established results and are not used to derive any new conclusion. The self-citations ([3], [81]) appear as ordinary references to the authors' prior work and are not load-bearing: the survey's comparative statements about model performance are drawn from the cited primary papers, and removing the self-citations would not change the taxonomy or the discussion. The MetaFormer-based classification in Section IV-B and Figure 9 is an organizing framework proposed by the authors, but it is presented as a labeling of existing model families (Token Mixer, Token Enhancer, Extended MetaFormer, EMA) rather than as a result derived from an authority, so it is not circular. The duplicated and internally inconsistent accuracy values in Tables I-III, such as the same SEED accuracy 98.47 ± 1.72% attributed to four different models, are data-integrity concerns that undermine the reliability of the comparative discussion; they are transcription or compilation errors, not evidence that a prediction was constructed from its own target. Accordingly, no circular step can be exhibited, and the paper is best assessed as a survey with independent organizational content and correctness risks outside the scope of circularity.

Assumptions & free parameters 0 free parameters · 2 assumptions · 0 invented entities

The survey introduces no free parameters or new entities. It makes two domain assumptions about comparability and completeness of the surveyed literature, both of which are standard for qualitative reviews but are load-bearing for its 'thorough overview' claim.

assumptions (2)
  • domain assumption Reported accuracy/F1 values from surveyed papers are comparable across datasets and protocols.
    The survey's comparative tables (Tables I-III) place metrics from heterogeneous studies side by side, implicitly assuming protocol comparability despite different preprocessing, splits, and subject pools.
  • domain assumption The literature search (IEEE Xplore, Science Direct, Web of Science, Google Scholar, arXiv, cutoff Jun 29 2025, keywords) captured the relevant field.
    The central claim of 'thorough overview' depends on the completeness of the search; no systematic meta-analysis or PRISMA-style flow is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Transformer-based EEG Decoding: A Survey." pith.science (2026). https://pith.science/paper/DMTXZCEU

@misc{pith2026250702320,
  author       = {Pith},
  title        = {Pith review of: Transformer-based EEG Decoding: A Survey},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DMTXZCEU}},
  note         = {Machine review of arXiv:2507.02320}
}
read the original abstract

Electroencephalography (EEG) is one of the most common signals used to capture the electrical activity of the brain, and the decoding of EEG, to acquire the user intents, has been at the forefront of brain-computer/machine interfaces (BCIs/BMIs) research. Compared to traditional EEG analysis methods with machine learning, the advent of deep learning approaches have gradually revolutionized the field by providing an end-to-end long-cascaded architecture, which can learn more discriminative features automatically. Among these, Transformer is renowned for its strong handling capability of sequential data by the attention mechanism, and the application of Transformers in various EEG processing tasks is increasingly prevalent. This article delves into a relevant survey, summarizing the latest application of Transformer models in EEG decoding since it appeared. The evolution of the model architecture is followed to sort and organize the related advances, in which we first elucidate the fundamentals of the Transformer that benefits EEG decoding and its direct application. Then, the common hybrid architectures by integrating basic Transformer with other deep learning techniques (convolutional/recurrent/graph/spiking neural netwo-rks, generative adversarial networks, diffusion models, etc.) is overviewed in detail. The research advances of applying the modified intrinsic structures of customized Transformer have also been introduced. Finally, the current challenges and future development prospects in this rapidly evolving field are discussed. This paper aims to help readers gain a clear understanding of the current state of Transformer applications in EEG decoding and to provide valuable insights for future research endeavors.

Discussion (0). Sign in to comment.

Forward citations

Cited by 3 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. EVA-Net: Subject-Independent EEG Motor Decoding with Video-Derived Motor Priors

    cs.AI 2026-06 unverdicted novelty 5.0 of 10

    EVA-Net improves subject-independent EEG motor decoding by using video action priors via cross-modal contrastive alignment and knowledge distillation, reporting an 8.66% LOSO accuracy gain on EEGMMI.

  2. WorldWeaver: Generating Long-Horizon Video Worlds via Rich Perception

    cs.CV 2025-08 unverdicted novelty 5.0 of 10

    WorldWeaver reduces temporal drift in long-horizon video generation by jointly modeling RGB and depth perceptual conditions with segmented noise scheduling.

  3. Foundation Models for Cross-Domain EEG Analysis Application: A Survey

    cs.HC 2025-08 conditional novelty 4.0 of 10

    A survey that organizes EEG foundation-model research into five output-modality categories: native EEG, text, vision, audio, and multimodal fusion, with a claim to be the first such comprehensive taxonomy.

Reference graph

Works this paper leans on

184 extracted references · 62 canonical work pages · cited by 3 Pith papers

  1. [152]

    SECT: A method of shifted EEG channel transformer for emotion recognition,

    Z. Bai et al. , “SECT: A method of shifted EEG channel transformer for emotion recognition,” IEEE J. Biomed. Health Inform. , vol. 27, no. 10, pp. 4758-4767, Oct. 2023

  2. [78]

    MI-CAT: A transformer-based domain adaptation network for motor imagery classification,

    D. Zhang, H. Li and J. Xie, “MI-CAT: A transformer-based domain adaptation network for motor imagery classification,” Neural Netw. , vol. 165, pp. 451-462, Aug. 2023

  3. [1]

    Emotions recognition using EEG signals: a survey,

    S. M. Alarc ˜ao and M. J. Fonseca, “Emotions recognition using EEG signals: a survey,”IEEE Trans. Affect. Comput., vol. 10, no. 3, pp. 374- 393, Jun. 2017

  4. [2]

    A human-centric metaverse enabled by brain- computer interface: a survey,

    H. Y . Zhu, et al. , “A human-centric metaverse enabled by brain- computer interface: a survey,” IEEE Commun. Surv. Tutor. , vol. 26, no. 3, pp. 2120-2145, 2024

  5. [3]

    Noninvasive EEG-based intelligent mobile robots: a systematic review,

    H. Li, X. Li and J. d. R. Mill ´an, “Noninvasive EEG-based intelligent mobile robots: a systematic review,” IEEE Trans. Autom. Sci. Eng., vol. 22, pp. 6291-6315, 2025

  6. [4]

    Attention is all you need,

    A. Vaswani et al., “Attention is all you need,” in Proc. Adv. Neural Inf. Process. Syst., 2017, pp. 5998-6008

  7. [5]

    A survey of transformers,

    T. Lin et al., “A survey of transformers,” AI Open, vol. 3, pp. 111–132, Jan. 2022

  8. [6]

    EEG based Continuous Speech Recognition using Transformers

    G. Krishna et al. , “EEG based continuous speech recognition using Transformers,” 2019, arXiv:2001.00501

Show all 184 references
  1. [7]

    TABS: Transformer based seizure detection,

    J. Pedoeem et al. , “TABS: Transformer based seizure detection,” in Proc. IEEE Signal Process. Med. Biol. Symp. , Jul. 2020, pp. 1-6

  2. [8]

    EEG classification with transformer-based models,

    J. Sun, J. Xie and H. Zhou, “EEG classification with transformer-based models,” in Proc. IEEE 3rd Global Conf. Life Sci. Technol., Mar. 2021, pp. 92-93

  3. [9]

    Understanding the brain with attention: A survey of Transformers in brain sciences,

    C. Chen et al. , “Understanding the brain with attention: A survey of Transformers in brain sciences,” Brain-X, vol. 1, no. 3, p. e29, 2023

  4. [10]

    Deep learning in EEG- based BCIs: A comprehensive review of transformer models advantages challenges and applications,

    B. Abibullaev, A. Keutayeva and A. Zollanvari, “Deep learning in EEG- based BCIs: A comprehensive review of transformer models advantages challenges and applications,” IEEE Access, vol. 11, pp. 127271-127301, 2023

  5. [11]

    Data constraints and performance op- timization for transformer-based models in EEG-based brain-computer interfaces: A survey,

    A. Keutayeva and B. Abibullaev, “Data constraints and performance op- timization for transformer-based models in EEG-based brain-computer interfaces: A survey,” IEEE Access, vol. 12, pp. 62628-62647, 2024

  6. [12]

    A comprehensive survey on applications of transform- ers for deep learning tasks,

    S. Islam et al., “A comprehensive survey on applications of transform- ers for deep learning tasks,” Exp. Syst. Appl., vol. 241, May 2024, Art. no. 122666

  7. [13]

    EEG tempor–spatial transformer for person identifica- tion,

    Y . Du et al. , “EEG tempor–spatial transformer for person identifica- tion,” Sci. Rep., vol. 12, no. 1, Aug. 2022, Art. no. 14378

  8. [14]

    Classification of EEG signals using Transformer based deep learning and ensemble models,

    M. Zeynali, H. Seyedarabi and R. Afrouzian, “Classification of EEG signals using Transformer based deep learning and ensemble models,” Biomed. Signal Process. Control , vol. 86, Sep. 2023, Art. no. 105130

  9. [15]

    A transformer-based approach combining deep learning network and spatial-temporal information for raw EEG classification,

    J. Xie et al., “A transformer-based approach combining deep learning network and spatial-temporal information for raw EEG classification,” IEEE Trans. Neural Syst. Rehabil. Eng. , vol. 30, pp. 2126-2136, 2022

  10. [16]

    MMPosE: Movie-induced multi-label positive emotion classification through EEG signals,

    X. Du et al. , “MMPosE: Movie-induced multi-label positive emotion classification through EEG signals,” IEEE Trans. Affect. Comput. , vol. 14, no. 4, pp. 2925-2938, 2023

  11. [17]

    Seizure prediction based on transformer using scalp electroencephalogram,

    J. Yan et al. , “Seizure prediction based on transformer using scalp electroencephalogram,” Appl. Sci., vol. 12, no. 9, pp. 4158, 2022

  12. [18]

    Spatio-temporal analysis of transformer based architecture for attention estimation from EEG,

    V . Delvigne et al. , “Spatio-temporal analysis of transformer based architecture for attention estimation from EEG,” inProc. 26th Int. Conf. Pattern Recognit., Montreal, QC, Canada, 2022, pp. 1076-1082

  13. [19]

    AMDET: Attention based multiple dimensions EEG transformer for emotion recognition,

    Y . Xu et al. , “AMDET: Attention based multiple dimensions EEG transformer for emotion recognition,” IEEE Trans. Affect. Comput., vol. 15, no. 3, pp. 1067-1077, 2023

  14. [20]

    EEG source imaging based on a transformer encoder network,

    T. Zheng and Z. Guan, “EEG source imaging based on a transformer encoder network,” in Proc. 3rd Int. Conf. Neural Networks Inf. Com- mun. Eng., Guangzhou, China, 2023, pp. 208-212

  15. [21]

    Semi-supervised domain adaptation for EEG- based sleep stage classification,

    S. Zheng and D. Wu, “Semi-supervised domain adaptation for EEG- based sleep stage classification,” in Proc. IEEE Int. Conf. Acoust. Speech Signal Process. , Seoul, Korea, Republic of, 2024, pp. 1776- 1780

  16. [22]

    EEG-based cognitive load classification using feature masked autoencoding and emotion transfer learning

    D. Pulver et al., “EEG-based cognitive load classification using feature masked autoencoding and emotion transfer learning.” in Proc. 25th Int. Conf. Multimodal Interact. , Paris, France, 2023, pp. 190-197

  17. [23]

    Decoding the continuous motion imagery trajectories of upper limb skeleton points for EEG-based brain–computer interface,

    P. Wang et al., “Decoding the continuous motion imagery trajectories of upper limb skeleton points for EEG-based brain–computer interface,” IEEE Trans. Instrum. Meas. , vol. 72, pp. 1-12, 2023

  18. [24]

    An image is worth 16x16 words: Transformers for image recognition at scale,

    A. Dosovitskiy et al., “An image is worth 16x16 words: Transformers for image recognition at scale,” 2020, arXiv:2010.11929

  19. [25]

    Patient-specific seizure prediction from scalp EEG using vision transformer,

    X. Zhang and H. Li, “Patient-specific seizure prediction from scalp EEG using vision transformer,” in Proc. IEEE 6th Inf. Technol. Mecha- tronics Eng. Conf. , Chongqing, China, 2022, pp. 1663-1667

  20. [26]

    Exploring feasibility of truth-involved automatic sleep staging combined with transformer,

    Z. Yang et al., “Exploring feasibility of truth-involved automatic sleep staging combined with transformer,” in Proc. IEEE Int. Conf. Bioinf. Biomed., Houston, TX, USA, 2021, pp. 2920-2923

  21. [27]

    Automated sleep staging via parallel frequency-cut attention,

    Z. Chen et al. , “Automated sleep staging via parallel frequency-cut attention,” IEEE Trans. Neural Syst. Rehabil. Eng. , vol. 31, pp. 1974- 1985, 2023

  22. [28]

    Trust recognition in human-robot cooperation using EEG,

    C. Xu et al. , “Trust recognition in human-robot cooperation using EEG,” in Proc. IEEE Int. Conf. Robot. Autom., Yokohama, Japan, 2024, pp. 7827-7833

  23. [29]

    BrainBERT: Self-supervised representation learning for intracranial recordings,

    C. Wang et al. , “BrainBERT: Self-supervised representation learning for intracranial recordings,” 2023, arXiv:2302.14367

  24. [30]

    Large transformers are better EEG learners,

    B. Wang et al. , “Large transformers are better EEG learners,” 2023, arXiv:2308.11654

  25. [31]

    Introducing attention mechanism for EEG signals: Emotion recognition with vision transformers,

    A. Arjun, A. S. Rajpoot and M. Raveendranatha Panicker, “Introducing attention mechanism for EEG signals: Emotion recognition with vision transformers,” in Proc. 43rd Annu. Int. Conf. IEEE Eng. Med. Biol. Soc., Mexico, 2021, pp. 5723-5726

  26. [32]

    DeepEnsemble: A novel brain wave classification in MI-BCI using ensemble of deep learners,

    A. Mehtiyev et al., “DeepEnsemble: A novel brain wave classification in MI-BCI using ensemble of deep learners,” in Proc. IEEE Int. Conf. Consum. Electron., Las Vegas, NV , USA, 2023, pp. 1-5

  27. [33]

    Clustering sparse swarm decom- position for automated recognition of upper limb movements from nonhomogeneous cross-channel EEG signals,

    S. V . Bhalerao and R. B. Pachori, “Clustering sparse swarm decom- position for automated recognition of upper limb movements from nonhomogeneous cross-channel EEG signals,” IEEE Sensors Lett., vol. 8, no. 1, pp. 1-4, Jan. 2024

  28. [34]

    EEG-based emotion recognition via transformer neural architecture search,

    C. Li et al. , “EEG-based emotion recognition via transformer neural architecture search,” IEEE Trans. Ind. Inform., vol. 19, no. 4, pp. 6016- 6025, Apr. 2023

  29. [35]

    Distilling invariant representations with domain adversarial learning for cross-subject children seizure prediction,

    Z. Zhang et al. , “Distilling invariant representations with domain adversarial learning for cross-subject children seizure prediction,” IEEE Trans. Cogn. Develop. Syst. , vol. 16, no. 1, pp. 202-211, Feb. 2024. JOURNAL OF LATEX CLASS FILES, VOL. 18, NO. 9, SEPTEMBER 2020 13

  30. [36]

    Scaling up your kernels to 31x31: Revisiting large kernel design in CNNs

    X. Ding et al. , “Scaling up your kernels to 31x31: Revisiting large kernel design in CNNs.” in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit., Jun. 2022, pp. 11953-11965

  31. [37]

    A CNN-transformer deep learning model for real-time sleep stage classification in an energy-constrained wireless device,

    Z. Yao and X. Liu, “A CNN-transformer deep learning model for real-time sleep stage classification in an energy-constrained wireless device,” in Proc. 11th Int. IEEE/EMBS Conf. Neural Eng. , Baltimore, MD, USA, pp. 1-4, 2023

  32. [38]

    EEG2vec: Self-supervised electroencephalographic representation learning,

    Q. Zhu et al. , “EEG2vec: Self-supervised electroencephalographic representation learning,” 2023, arXiv:2305.13957

  33. [39]

    Expert knowledge inspired contrastive learning for sleep staging,

    H. Zhang et al. , “Expert knowledge inspired contrastive learning for sleep staging,” in Proc. Int. Joint Conf. Neural Networks , Padua, Italy, 2022, pp. 1-6

  34. [40]

    Neuro2vec: Masked fourier spectrum prediction for neurophysiological representation learning,

    D. Wu et al. , “Neuro2vec: Masked fourier spectrum prediction for neurophysiological representation learning,” IEEE J. Biomed. Health Inform., 2024, doi: 10.1109/JBHI.2024.3415959

  35. [41]

    EEGformer: Transformer-based epilepsy detection on raw EEG traces for low-channel-count wearable continuous monitoring devices,

    P. Busia et al., “EEGformer: Transformer-based epilepsy detection on raw EEG traces for low-channel-count wearable continuous monitoring devices,” in Proc. IEEE Biomed. Circuits Syst. Conf. , Taipei, Taiwan, 2022, pp. 640-644

  36. [42]

    EEG emotion recognition using attention-based convo- lutional transformer neural network,

    L. Gong et al., “EEG emotion recognition using attention-based convo- lutional transformer neural network,” Biomed. Signal Process. Control, vol. 84, Jul. 2023, Art. no. 104835

  37. [43]

    A residual based attention model for EEG based sleep staging,

    W. Qu et al., “A residual based attention model for EEG based sleep staging,” IEEE J. Biomed. Health Inform. , vol. 24, no. 10, pp. 2833- 2843, Oct. 2020

  38. [44]

    High-density electroencephalography and speech signal based deep framework for clinical depression diagnosis,

    A. Qayyum et al. , “High-density electroencephalography and speech signal based deep framework for clinical depression diagnosis,” IEEE/ACM Trans. Comput. Biol. Bioinf., vol. 20, no. 4, pp. 2587-2597, Jul. 2023

  39. [45]

    BENDR: Using trans- formers and a contrastive self-supervised learning task to learn from massive amounts of EEG data,

    D. Kostas, S. Aroca-Ouellette and F. Rudzicz, “BENDR: Using trans- formers and a contrastive self-supervised learning task to learn from massive amounts of EEG data,” Front. Hum. Neurosci. , vol. 15, pp. 253, Jun. 2021

  40. [46]

    Temporal aware mixed attention-based convolution and transformer network (MACTN) for EEG emotion recognition,

    X. Si et al. , “Temporal aware mixed attention-based convolution and transformer network (MACTN) for EEG emotion recognition,” 2023, arXiv:2305.18234

  41. [47]

    An efficient temporal network with dual self-distillation for electroencephalography signal classification,

    Z. Xiao et al., “An efficient temporal network with dual self-distillation for electroencephalography signal classification,” in Proc. IEEE Int. Conf. Bioinf. Biomed. , Las Vegas, NV , USA, 2022, pp. 1759-1762

  42. [48]

    Exploring the applicability of transfer learning and feature engineering in epilepsy prediction using hybrid transformer model,

    S. Hu et al., “Exploring the applicability of transfer learning and feature engineering in epilepsy prediction using hybrid transformer model,” IEEE Trans. Neural Syst. Rehabil. Eng. , vol. 31, pp. 1321-1332, 2023

  43. [49]

    A novel data augmentation approach using mask encoding for deep learning-based asynchronous SSVEP-BCI,

    W. Ding et al. , “A novel data augmentation approach using mask encoding for deep learning-based asynchronous SSVEP-BCI,” IEEE Trans. Neural Syst. Rehabil. Eng. , vol. 32, pp. 875-886, 2024

  44. [50]

    An attention-based deep learning approach for sleep stage classification with single-channel EEG,

    Eldele et al. , “An attention-based deep learning approach for sleep stage classification with single-channel EEG,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 29, pp. 809-818, 2021

  45. [51]

    A temporal-spectral fused and attention-based deep model for automatic sleep staging,

    G. Fu et al., “A temporal-spectral fused and attention-based deep model for automatic sleep staging,” IEEE Trans. Neural Syst. Rehabil. Eng. , vol. 31, pp. 1008-1018, 2023

  46. [52]

    A cross-scale transformer and triple-view attention based domain-rectified transfer learning for EEG classification in RSVP tasks,

    J. Luo et al. , “A cross-scale transformer and triple-view attention based domain-rectified transfer learning for EEG classification in RSVP tasks,” IEEE Trans. Neural Syst. Rehabil. Eng. , vol. 32, pp. 672-683, 2024

  47. [53]

    Denosieformer: A transformer-based approach for single-channel EEG artifact removal,

    J. Chen et al. , “Denosieformer: A transformer-based approach for single-channel EEG artifact removal,” IEEE Trans. Instrum. Meas., vol. 73, pp. 1-16, 2024

  48. [54]

    A transformer-based spatial-temporal sleep staging model through raw EEG,

    G. Shi, Z. Chen and R. Zhang, “A transformer-based spatial-temporal sleep staging model through raw EEG,” in Proc. Int. Conf. High Perform. Big Data Intell. Syst. , Macau, China, 2021, pp. 110-115

  49. [55]

    EEG-based seizure prediction via hybrid vision transformer and data uncertainty learning,

    Z. Deng et al. , “EEG-based seizure prediction via hybrid vision transformer and data uncertainty learning,” Eng. Appl. Artif. Intell. , vol. 123, Aug. 2023

  50. [56]

    Dual deep learning for sleep classification,

    M. Abdelmagid et al. , “Dual deep learning for sleep classification,” in Proc. 11th Int. Japan-Africa Conf. Electron. Commun. Comput. , Alexandria, Egypt, 2023, pp. 190-193

  51. [57]

    A multi-view spectral-spatial-temporal masked autoen- coder for decoding emotions with self-supervised learning,

    R. Li et al. , “A multi-view spectral-spatial-temporal masked autoen- coder for decoding emotions with self-supervised learning,” in Proc. 30th ACM Int. Conf. Multimedia. , Oct. 2022, pp. 6-14

  52. [58]

    Temporal-spatial prediction: Pre-training on diverse datasets for EEG classification,

    Z. Li et al. , “Temporal-spatial prediction: Pre-training on diverse datasets for EEG classification,” in Proc. IEEE Int. Conf. Acoust. Speech Signal Process. , Seoul, Korea, Republic of, 2024, pp. 1806- 1810

  53. [59]

    Context-aware EEG-based perceived stress recognition based on emotion transition paradigm,

    J. Liu et al. , “Context-aware EEG-based perceived stress recognition based on emotion transition paradigm,” in Proc. 11th Int. Conf. Affect. Comput. Intell. Interact. Workshops Demos , Cambridge, MA, USA, 2023, pp. 1-8

  54. [60]

    Consformer: consciousness detection using transformer networks with correntropy-based measures,

    X. Sun et al., “Consformer: consciousness detection using transformer networks with correntropy-based measures,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 31, pp. 2933-2943, 2023

  55. [61]

    Multimodal neurophysiological transformer for emotion recognition,

    S. Koorathota et al. , “Multimodal neurophysiological transformer for emotion recognition,” in Proc. 44th Annu. Int. Conf. IEEE Eng. Med. Biol. Soc., Glasgow, Scotland, United Kingdom, 2022, pp. 3563-3567

  56. [62]

    Multimodal vigilance estimation with modality- pairwise contrastive loss,

    M. Zhang et al. , “Multimodal vigilance estimation with modality- pairwise contrastive loss,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 71, no. 4, pp. 1139-1150, Apr. 2024

  57. [63]

    MaEEG: Masked auto-encoder for EEG repre- sentation learning,

    H.-Y . S. Chien et al., “MaEEG: Masked auto-encoder for EEG repre- sentation learning,” 2022, arXiv:2211.02625

  58. [64]

    Large brain model for learning generic representations with tremendous EEG data in BCI,

    W. B. Jiang, L. M. Zhao and B. L. Lu, “Large brain model for learning generic representations with tremendous EEG data in BCI,” 2024, arXiv:2405.18765

  59. [65]

    FoME: A foundation model for EEG using adaptive temporal-lateral attention scaling,

    E. Shi, et al. , “FoME: A foundation model for EEG using adaptive temporal-lateral attention scaling,” 2024, arXiv:2409.12454

  60. [66]

    Inducing inductive bias in vision transformer for EEG classification,

    R. Khadka et al. , “Inducing inductive bias in vision transformer for EEG classification,” in Proc. ICASSP IEEE Int. Conf. Acoust. Speech Signal Process., Seoul, Korea, Republic of, 2024, pp. 2096-2100

  61. [67]

    AITST—affective EEG-based person identification via interrelated temporal-spatial transformer,

    H. Cai et al., “AITST—affective EEG-based person identification via interrelated temporal-spatial transformer,” Pattern Recognit. Lett., vol. 174, pp. 32-38, Oct. 2023

  62. [68]

    A transformer convolutional network with the method of image segmentation for EEG-based emotion recognition,

    X. Zhang and X. Cheng, “A transformer convolutional network with the method of image segmentation for EEG-based emotion recognition,” IEEE Signal Process. Lett. , vol. 31, pp. 401-405, 2024

  63. [69]

    DICE-net: A novel convolution-transformer architecture for Alzheimer detection in EEG signals,

    A. Miltiadous et al. , “DICE-net: A novel convolution-transformer architecture for Alzheimer detection in EEG signals,” IEEE Access , vol. 11, pp. 71840-71858, 2023

  64. [70]

    Multi-domain encoding of spatiotemporal dynamics in EEG for emotion recognition,

    C. Cheng et al., “Multi-domain encoding of spatiotemporal dynamics in EEG for emotion recognition,” IEEE J. Biomed. Health Informat. , vol. 27, no. 3, pp. 1342-1353, Mar. 2023

  65. [71]

    MEEG-transformer: Transformer network based on multi-domain EEG for emotion recognition,

    H. Sun et al. , “MEEG-transformer: Transformer network based on multi-domain EEG for emotion recognition,” in Proc. IEEE Int. Conf. Bioinf. Biomed., Istanbul, Turkiye, 2023, pp. 3332-3339

  66. [72]

    A novel hybrid CNN-transformer model for EEG motor imagery classification,

    Y . Ma, Y . Song and F. Gao, “A novel hybrid CNN-transformer model for EEG motor imagery classification,” in Proc. Int. Joint Conf. Neural Networks, Padua, Italy, 2022, pp. 1-8

  67. [73]

    Coupling convolution, transformer and graph embedding for motor imagery brain-computer interfaces,

    Z. Wu, B. Sun and X. Zhu, “Coupling convolution, transformer and graph embedding for motor imagery brain-computer interfaces,” in Proc. IEEE Int. Symp.Circuits Syst. , Austin, TX, USA, 2022, pp. 404- 408

  68. [74]

    Transformer-based spatial-temporal feature learning for P300,

    Q. Li et al. , “Transformer-based spatial-temporal feature learning for P300,” in Proc. 16th ICME Int. Conf. Complex Med. Eng. , Zhongshan, China, 2022, pp. 310-313

  69. [75]

    A shallow mirror transformer for subject-independent motor imagery BCI,

    J. Luo et al. , “A shallow mirror transformer for subject-independent motor imagery BCI,” Comput. Biol. Med. , vol. 164, Sep. 2023, Art. no. 107254

  70. [76]

    ADFCNN: Attention-based dual-scale fusion convolu- tional neural network for motor imagery brain–computer interface,

    W. Tao et al., “ADFCNN: Attention-based dual-scale fusion convolu- tional neural network for motor imagery brain–computer interface,” IEEE Trans. Neural Syst. Rehabil. Eng. , vol. 32, pp. 154-165, 2024

  71. [77]

    EEG-deformer: A dense convolutional transformer for brain-computer interfaces,

    Y . Ding et al., “EEG-deformer: A dense convolutional transformer for brain-computer interfaces,” IEEE J. Biomed. Health Informat. , vol. 29, no. 3, pp. 1909-1918, Mar. 2025

  72. [79]

    Neuro-gpt: Developing a foundation model for EEG,

    W. Cui et al., “Neuro-gpt: Developing a foundation model for EEG,” 2023, arXiv:2311.03764

  73. [80]

    Decoding natural images from EEG for object recognition,

    Y . Song, “Decoding natural images from EEG for object recognition,” in Proc. 12th Int. Conf. Learn. Representant , 2024. pp. 1-6

  74. [81]

    Dual-TSST: A dual-branch temporal- spectral-spatial Transformer model for EEG decoding,

    H. Li, H. Zhang and Y . Chen, “Dual-TSST: A dual-branch temporal- spectral-spatial Transformer model for EEG decoding,” IEEE J. Biomed. Health Inform. , 2025, doi: 10.1109/JBHI.2025.3577611

  75. [82]

    A transformer encoder and convolutional neural network combined method for classification of error-related potentials,

    G. Ren et al., “A transformer encoder and convolutional neural network combined method for classification of error-related potentials,” in Proc. IEEE Biomed. Circuits Syst. Conf. , Toronto, ON, Canada, 2023, pp. 1- 5

  76. [83]

    Global adaptive transformer for cross-subject enhanced EEG classification,

    Y . Song et al., “Global adaptive transformer for cross-subject enhanced EEG classification,” IEEE Trans. Neural Syst. Rehabil. Eng. , vol. 31, pp. 2767-2777, 2023

  77. [84]

    Transformer-based network with optimization for cross-subject motor imagery identification,

    X. Tan “Transformer-based network with optimization for cross-subject motor imagery identification,” Bioengineering, vol. 10, no. 5, pp. 609, May 2023. JOURNAL OF LATEX CLASS FILES, VOL. 18, NO. 9, SEPTEMBER 2020 14

  78. [85]

    EEG-transformer: Self-attention from transformer architecture for decoding EEG of imagined speech,

    Y . E. Lee and S. H. Lee, “EEG-transformer: Self-attention from transformer architecture for decoding EEG of imagined speech,” in Proc. 10th Int. Winter Conf. Brain-Computer Interface , Gangwon-do, Korea, Republic of, 2022, pp. 1-4

  79. [86]

    EEG conformer: Convolutional transformer for EEG decoding and visualization,

    Y . Song et al. , “EEG conformer: Convolutional transformer for EEG decoding and visualization,” IEEE Trans. Neural Syst. Rehabil. Eng. , vol. 31, pp. 710-719, 2023

  80. [87]

    Multiscale convolutional transformer for EEG classification of mental imagery in different modalities,

    H. J. Ahn et al. , “Multiscale convolutional transformer for EEG classification of mental imagery in different modalities,” IEEE Trans. Neural Syst. Rehabil. Eng. , vol. 31, pp. 646-656, 2023

  81. [88]

    MST-net: A multi-scale swin transformer network for EEG-based cognitive load assessment,

    Z. Li et al. , “MST-net: A multi-scale swin transformer network for EEG-based cognitive load assessment,” Brain Res. Bull. , vol. 206, 2024, Art. no. 110834

  82. [89]

    EEG-ConvTransformer for single-trial EEG-based visual stimulus classification,

    S. Bagchi and D. R. Bathula, “EEG-ConvTransformer for single-trial EEG-based visual stimulus classification,” Pattern Recognit., vol. 129, Sep. 2022. Art. no. 108757

  83. [90]

    Multi-scale masked autoencoders for cross-session emotion recognition,

    M. Pang et al. , “Multi-scale masked autoencoders for cross-session emotion recognition,” IEEE Trans. Neural Syst. Rehabil. Eng. , vol. 32, pp. 1637-1646, 2024

  84. [91]

    Extracting temporal-spectral- spatial representation of EEG using self-supervised learning for the identification of children with ASD,

    H. Chen, O. Gaoxiang and X. Li, “Extracting temporal-spectral- spatial representation of EEG using self-supervised learning for the identification of children with ASD,” in Proc. IEEE 13th Int. Conf. on Cyber Technol. Automat., Control, Intelli. Syst., (CYBER) , Qinhuang- dao,...

  85. [92]

    Transformer-based multiscale 3-D convolutional network for motor imagery classification,

    J. Su et al., “Transformer-based multiscale 3-D convolutional network for motor imagery classification,” IEEE Sensors J. , vol. 25, no. 5, pp. 8621-8630, 2025

  86. [93]

    BiCCT: A compact convolutional transformer for EEG emotion recognition,

    G. Cao et al. , “BiCCT: A compact convolutional transformer for EEG emotion recognition,” in Proc. IEEE Int. Conf. Bioinf. Biomed. , Istanbul, Turkiye, pp. 4792-4799, 2023

  87. [94]

    Spatial-temporal transformers for EEG emotion recogni- tion,

    J. Liu et al., “Spatial-temporal transformers for EEG emotion recogni- tion,” in Proc. Int. Conf. Adv. Artif. Intell. , 2022, pp. 116-120

  88. [95]

    Continual learning of a transformer-based deep learning classifier using an initial model from action observation EEG data to online motor imagery classification,

    P. L. Lee et al. , “Continual learning of a transformer-based deep learning classifier using an initial model from action observation EEG data to online motor imagery classification,” Bioengineering, vol. 10, no. 2, pp. 186, Feb. 2023

  89. [96]

    Multi-modal network based on spatio-temporal and attention for emotion recognition,

    Y . Zhang, W. Chen and C. Cheng, “Multi-modal network based on spatio-temporal and attention for emotion recognition,” in Proc. IEEE Int. Conf. Bioinf. Biomedicine, Istanbul, Turkiye, 2023, pp. 2364-2367

  90. [97]

    DREAM: Domain invariant and contrastive representation for sleep dynamics,

    S. Lee, T. H. Pham and P. Zhang, “DREAM: Domain invariant and contrastive representation for sleep dynamics,” in Proc. IEEE Int. Conf. Data Mining, Orlando, FL, USA, 2022, pp. 1029-1034

  91. [98]

    LGSleepNet: An automatic sleep staging model based on local and global representation learning,

    Q. Shen et al., “LGSleepNet: An automatic sleep staging model based on local and global representation learning,” IEEE Trans. Instrum. Meas., vol. 72, pp. 1-14, 2023

  92. [99]

    Adversarial discriminative domain adaptation and transformers for EEG-based cross-subject emotion recognition,

    S. Sartipi and M. Cetin, “Adversarial discriminative domain adaptation and transformers for EEG-based cross-subject emotion recognition,” in Proc. 11th Int. IEEE/EMBS Conf. Neural Engineer. , Baltimore, MD, USA, 2023, pp. 1-4

  93. [100]

    Denoise enhanced neural network with efficient data generation for automatic sleep stage classification of class imbalance,

    L. Yu et al. , “Denoise enhanced neural network with efficient data generation for automatic sleep stage classification of class imbalance,” in Proc. Int. Joint Conf. Neural Networks, Gold Coast, Australia, 2023, pp. 1-8

  94. [101]

    Multi-class image generation from EEG features with conditional generative adversarial networks,

    Y . Duan et al. , “Multi-class image generation from EEG features with conditional generative adversarial networks,” in Proc. Int. Conf. Wireless Commun. Signal Process. , Hangzhou, China, 2023, pp. 534- 539

  95. [102]

    Dual AxAtGAN: A feature intregrate BCI model for image reconstruction,

    X. Zhao and Y . Guo, “Dual AxAtGAN: A feature intregrate BCI model for image reconstruction,” in Proc. 13th Int. Conf. Inf. Technol. Med. Educ., Wuyishan, China, 2023, pp. 135-139

  96. [103]

    Scalable diffusion models with Transformers,

    W. Peebles and S. Xie, “Scalable diffusion models with Transformers,” in Proc. IEEE/CVF Int. Conf. on Comput. Vis. (ICCV) , Oct. 2023, pp. 4172-4182

  97. [104]

    DiffMDD: A diffusion-based deep learning framework for MDD diagnosis using EEG,

    Y . Wang, et al., “DiffMDD: A diffusion-based deep learning framework for MDD diagnosis using EEG, ” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 32, pp. 728-738,2024

  98. [105]

    Generative AI enables EEG super-resolution via spatio-temporal adaptive diffusion learning,

    S. Wang, et al. , “Generative AI enables EEG super-resolution via spatio-temporal adaptive diffusion learning,” IEEE Trans. Consum. Electron., vol.71, no.1, pp.1034-1045, Feb. 2025

  99. [106]

    Improving EEG classification through randomly reassembling original and generated data with transformer-based dif- fusion models,

    M. Chen, et al. ., “Improving EEG classification through randomly reassembling original and generated data with transformer-based dif- fusion models,” 2024, arXiv:2407.20253

  100. [107]

    Visual decoding and reconstruction via EEG embeddings with guided diffusion,

    D. Li, et al., “Visual decoding and reconstruction via EEG embeddings with guided diffusion,” 2024, arXiv:2403.07721

  101. [108]

    A dual-branch dynamic graph convolution based adap- tive transformer feature fusion network for EEG emotion recognition,

    M. Sun et al., “A dual-branch dynamic graph convolution based adap- tive transformer feature fusion network for EEG emotion recognition,” IEEE Trans. Affect. Comput. , vol. 13, no. 4, pp. 2218-2228, 2022

  102. [109]

    Elastic graph transformer networks for EEG-based emotion recognition,

    W. B. Jiang et al., “Elastic graph transformer networks for EEG-based emotion recognition,” in Proc. IEEE Int. Conf. Acoust. Speech Signal Process., Rhodes Island, Greece, 2023, pp. 1-5

  103. [110]

    Semi-supervised EEG emotion recognition with discriminative graph transformer model,

    H. Kim and Y . S. Choi, “Semi-supervised EEG emotion recognition with discriminative graph transformer model,” in Proc. IEEE Int. Conf. Consum. Electronics-Asia, Busan, Korea, Republic of, 2023, pp. 1-4

  104. [111]

    Dynamic multi-graph convolution-based channel- weighted transformer feature fusion network for epileptic seizure prediction,

    Y . Wang et al. , “Dynamic multi-graph convolution-based channel- weighted transformer feature fusion network for epileptic seizure prediction,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 31, pp. 4266- 4277, 2023

  105. [112]

    A comprehensive interaction in multiscale multichannel EEG signals for emotion recognition,

    Y . Guo et al., “A comprehensive interaction in multiscale multichannel EEG signals for emotion recognition,” Mathematics, vol. 12, no. 8, p. 1180, Apr. 2024

  106. [113]

    A dual-branch spatio-temporal-spectral transformer feature fusion network for EEG-based visual recognition,

    J. Luo et al. , “A dual-branch spatio-temporal-spectral transformer feature fusion network for EEG-based visual recognition,” IEEE Trans. Ind. Informat., vol. 20, no. 2, pp. 1721-1731, Feb. 2024

  107. [114]

    R-transformer: Recurrent neural network enhanced transformer,

    Z. Wang et al. , “R-transformer: Recurrent neural network enhanced transformer,” 2019, arXiv:1907.05572

  108. [115]

    EEGAlzheimer’sNet: Development of transformer-based attention long short term memory network for de- tecting Alzheimer disease using EEG signal,

    D. K. Ravikanti and S. Saravanan, “EEGAlzheimer’sNet: Development of transformer-based attention long short term memory network for de- tecting Alzheimer disease using EEG signal,” Biomed. Signal Process. Control, vol. 86, Sep. 2023. Art. no. 105318

  109. [116]

    Automatic sleep stage classification by CNN-transformer-LSTM using single-channel EEG signal,

    D. T. Pham and R. Mou ˇcek, “Automatic sleep stage classification by CNN-transformer-LSTM using single-channel EEG signal,” in Proc. IEEE Int. Conf. Bioinf. Biomed. , Istanbul, Turkiye, 2023, pp. 2559- 2563

  110. [117]

    EmoTVR: A hybrid model to estimate continuous- time and continuous-level emotion from electroencephalography,

    X. Zhou et al. , “EmoTVR: A hybrid model to estimate continuous- time and continuous-level emotion from electroencephalography,” in Proc. IEEE Int. Conf. Acoust. Speech Signal Process. , Seoul, Korea, Republic of, 2024, pp. 2021-2025

  111. [118]

    TBEEG: A two-branch manifold domain enhanced transformer algorithm for learning EEG decoding,

    Y . Qin, W. Zhang and X. Tao, “TBEEG: A two-branch manifold domain enhanced transformer algorithm for learning EEG decoding,” IEEE Trans. Neural Syst. Rehabil. Eng. , vol. 32, pp. 1466-1476, 2024

  112. [119]

    Band-level adaptive fusion network for cross-subject EEG emotion recognition,

    Y . Wang, L. Zhang and Y . Zhang, “Band-level adaptive fusion network for cross-subject EEG emotion recognition,” IEEE Trans. Instrum. Meas., vol. 74, pp. 1-12, 2025

  113. [120]

    Deep learning in spiking neural networks,

    A. Tavanaei et al., “Deep learning in spiking neural networks,” Neural Networks, vol. 111, pp. 47-63, 2019

  114. [121]

    Epilepsy seizure detection and prediction using an approximate spiking convolutional transformer,

    Q. Chen et al. , “Epilepsy seizure detection and prediction using an approximate spiking convolutional transformer,” 2024, arXiv:2402. 09424

  115. [122]

    Spikformer: When spiking neural network meets transformer,

    Z. Zhou et al. , “Spikformer: When spiking neural network meets transformer,” 2022, arXiv:2209.15425

  116. [123]

    Capsule networks–a survey,

    M. K. Patrick et al., “Capsule networks–a survey,” J. King Saud Univ.- Comput. Inf. Sci. , vol. 34, no. 1, pp. 1295-1310, 2022

  117. [124]

    TC-Net: A transformer capsule network for EEG-based emotion recognition,

    Y . Wei et al., “TC-Net: A transformer capsule network for EEG-based emotion recognition,” Comput. Biol. Med. , vol. 152, Jan. 2023. Art. no. 106463

  118. [125]

    MES-CTNet: A novel capsule transformer network base on a multi-domain feature map for electroencephalogram-based emotion recognition,

    Y . Du et al. , “MES-CTNet: A novel capsule transformer network base on a multi-domain feature map for electroencephalogram-based emotion recognition,” Brain Sci., vol. 14, no. 4, p. 344, 2024

  119. [126]

    Bi-branch vision transformer network for EEG emotion recognition,

    W. Lu, T. P. Tan and H. Ma, “Bi-branch vision transformer network for EEG emotion recognition,” IEEE Access, vol. 11, pp. 36233-36243, 2023

  120. [127]

    Time-series representation learning via temporal and contextual contrasting,

    E. Eldele et al., “Time-series representation learning via temporal and contextual contrasting,” 2021, arXiv:2106.14112

  121. [128]

    Can brain signals reveal inner alignment with human languages?

    J. Qiu, et al. , “Can brain signals reveal inner alignment with human languages?” in Proc. Findings Assoc. Comput. Linguistics: EMNLP ,

  122. [129]

    W A VELET2VEC: A filter bank masked autoencoder for EEG-based seizure subtype classification,

    R. Peng et al. , “W A VELET2VEC: A filter bank masked autoencoder for EEG-based seizure subtype classification,” in Proc. IEEE Int. Conf. Acoust. Speech Signal Process. , Rhodes Island, Greece, 2023, pp. 1-5

  123. [130]

    Linear–nonlinear feature reconstruction network for emotion recognition from brain functional connectivity,

    B. Fu, X. Yu and Y . Liu, “Linear–nonlinear feature reconstruction network for emotion recognition from brain functional connectivity,” IEEE Trans. Instrum. Meas. , vol. 74, pp. 1-10, 2025

  124. [131]

    Emotion transformer fusion: complementary repre- sentation properties of EEG and eye movements on recognizing anger and surprise,

    Y . Wang et al. , “Emotion transformer fusion: complementary repre- sentation properties of EEG and eye movements on recognizing anger and surprise,” in Proc. IEEE Int. Conf. Bioinf. Biomed. , Houston, TX, USA, 2021, pp. 1575-1578

  125. [132]

    START: Automatic sleep staging with attention-based cross-modal learning transformer,

    J. Sun et al. , “START: Automatic sleep staging with attention-based cross-modal learning transformer,” in Proc. IEEE Int. Conf. Bioinf. Biomed., Istanbul, Turkiye, 2023, pp. 1421-1428

  126. [133]

    Transformer in transformer,

    K. Han et al., “Transformer in transformer,” in Proc. Adv. Neural Inf. Process. Syst., 2021, pp. 15908-15919

  127. [134]

    Transformers for EEG-based emotion recognition: A hierarchical spatial information learning model,

    Z. Wang et al., “Transformers for EEG-based emotion recognition: A hierarchical spatial information learning model,” IEEE Sensors J., vol. 22, no. 5, pp. 4359-4368, Mar. 2022. JOURNAL OF LATEX CLASS FILES, VOL. 18, NO. 9, SEPTEMBER 2020 15

  128. [135]

    SleepTransformer: Automatic sleep staging with interpretability and uncertainty quantification,

    H. Phan et al. ., “SleepTransformer: Automatic sleep staging with interpretability and uncertainty quantification,” IEEE Trans. Biomed. Eng., vol. 69, no. 8, pp. 2456-2467, Aug. 2022

  129. [136]

    A spatial-temporal transformer architecture using multi- channel signals for sleep stage classification,

    H. Yao et al., “A spatial-temporal transformer architecture using multi- channel signals for sleep stage classification,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 31, pp. 3353-3362, 2023

  130. [137]

    EEG2TEXT: Open vocabulary EEG-to-text de- coding with EEG pre-Training and multi-view transformer,

    H. Liu et al. , “EEG2TEXT: Open vocabulary EEG-to-text de- coding with EEG pre-Training and multi-view transformer,” 2024 arXiv:2405.02165

  131. [138]

    Automatic sleep stage classification method based on transformer-in-transformer,

    M. Kim, K. Jung and W. Chung, “Automatic sleep stage classification method based on transformer-in-transformer,” in Proc. 11th Int. Winter Conf. Brain-Computer Interface , Gangwon, Korea, Republic of, 2023, pp. 1-5

  132. [139]

    Convolutional transformer-in-transformer for automatic sleep stage classification,

    M. Kim and W. Chung, “Convolutional transformer-in-transformer for automatic sleep stage classification,” in Proc. 12th Int. Winter Conf. Brain-Computer Interface , Gangwon, Korea, Republic of, 2024, pp. 1-5

  133. [140]

    Structure-preserving transformers for sequences of SPD matrices,

    M. Seraphim et al. , “Structure-preserving transformers for sequences of SPD matrices,” 2023, arXiv:2309.07579

  134. [141]

    Hierarchical transformer for brain computer interface,

    P. Deny and K. W. Choi, “Hierarchical transformer for brain computer interface,” in Proc. 11th Int. Winter Conf. Brain-Computer Interface, Gangwon, Korea, Republic of, 2023, pp. 1-5

  135. [142]

    Automatic sleep staging method based on multi- channel fusion,

    J. Fan et al. , “Automatic sleep staging method based on multi- channel fusion,” in Proc. 3rd Int. Conf. on Electron. Inf. Eng. Comput. Commun., Wuhan, China, 2023, pp. 1-5

  136. [143]

    MultiChannelSleepNet: A transformer-based model for automatic sleep stage classification with PSG,

    Y . Dai et al., “MultiChannelSleepNet: A transformer-based model for automatic sleep stage classification with PSG,” IEEE J. Biomed. Health Inform., vol. 27, no. 9, pp. 4204-4215, Sept. 2023

  137. [144]

    B2-ViT net: Broad vision transformer network with broad attention for seizure prediction,

    S. Shi and W. Liu, “B2-ViT net: Broad vision transformer network with broad attention for seizure prediction,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 32, pp. 178-188, 2024

  138. [145]

    Metaformer is actually what you need for vision,

    W.Yu, et al. , “Metaformer is actually what you need for vision,” in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. (CVPR) , 2022. pp. 10819-10829

  139. [146]

    EEGDiR: Electroencephalogram de- noising network for temporal information storage and global modeling through Retentive Network,

    B. Wang, D. Fei and P. Jiang, “EEGDiR: Electroencephalogram de- noising network for temporal information storage and global modeling through Retentive Network,” Comput. Biol. Med. , vol. 177, 2024, Art. no. 108626

  140. [147]

    Unified convolutional sparse transformer for disease diag- nosis, monitoring, drug development, and therapeutic effect prediction from EEG raw data,

    Z. He et al., “Unified convolutional sparse transformer for disease diag- nosis, monitoring, drug development, and therapeutic effect prediction from EEG raw data,” Biology, vol. 13, no. 4, 2024

  141. [148]

    Clinical grade prediction of therapeutic dosage for electroconvulsive therapy (ECT) based on patient’s pre-ictal EEG using fuzzy causal transformers,

    C. Kalantarpour et al., “Clinical grade prediction of therapeutic dosage for electroconvulsive therapy (ECT) based on patient’s pre-ictal EEG using fuzzy causal transformers,” in Proc. 3rd Int. Conf. Elect. Comput. Commun. Mechatronics Eng. , Tenerife, Canary Islands, Spain, 2...

  142. [149]

    An efficient spatial-temporal representation method for EEG emotion recognition,

    W. Weng et al. , “An efficient spatial-temporal representation method for EEG emotion recognition,” in Proc. IEEE Smartworld Ubiquitous Intell. Comput., Scalable Comput. , Haikou, China, 2022, pp. 458-467

  143. [150]

    BIOT: Cross-data biosignal learning in the wild,

    C. Yang, M. B. Westover and J. Sun, “BIOT: Cross-data biosignal learning in the wild,” 2023, arXiv:2305.10351

  144. [151]

    Spatiotemporal pooling on appropri- ate topological maps represented as two-dimensional images for EEG classification,

    T. Fukushima and R. Miyamoto, “Spatiotemporal pooling on appropri- ate topological maps represented as two-dimensional images for EEG classification,” 2024, arXiv:2403.04353

  145. [153]

    Epoch-level and sequence-level multi-head self-attention-based sleep stage classification,

    K. Jung, M. Kim and W. Chung, “Epoch-level and sequence-level multi-head self-attention-based sleep stage classification,” in Proc. 11th Int. Winter Conf. Brain-Computer Interface, Gangwon, Korea, Republic of, 2023, pp. 1-6

  146. [154]

    A spatial-temporal transformer based on domain gen- eralization for motor imagery classification,

    S. Liu et al. , “A spatial-temporal transformer based on domain gen- eralization for motor imagery classification,” in Proc. IEEE Int. Conf. Syst. Man Cybern. , Honolulu, Oahu, HI, USA, 2023, pp. 3789-3794

  147. [155]

    MEET: A multi-band EEG transformer for brain states decoding,

    E. Shi et al., “MEET: A multi-band EEG transformer for brain states decoding,” IEEE Trans. Biomed. Eng. , vol. 71, no. 5, pp. 1442-1453, May 2024

  148. [156]

    Toward domain-free transformer for generalized EEG pre-training,

    S. J. Kim et al., “Toward domain-free transformer for generalized EEG pre-training,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 32, pp. 482- 492, 2024

  149. [157]

    Improving generalized zero-shot learning SSVEP classification performance from data-efficient perspective,

    X. Wang et al. , “Improving generalized zero-shot learning SSVEP classification performance from data-efficient perspective,”IEEE Trans. Neural Syst. Rehabil. Eng. , vol. 31, pp. 4135-4145, 2023

  150. [158]

    TcT: Temporal and channel transformer for EEG-based emotion recognition,

    Y . Liu, Y . Zhou and D. Zhang, “TcT: Temporal and channel transformer for EEG-based emotion recognition,” in Proc. IEEE 35th Int. Symp. Computer-Based Med. Syst. , Shenzen, China, 2022, pp. 366-371

  151. [159]

    ESTformer: Transformer utilizing spatiotemporal depen- dencies for EEG super-resolution,

    D. Li et al., “ESTformer: Transformer utilizing spatiotemporal depen- dencies for EEG super-resolution,” 2023, arXiv:2312.10052

  152. [160]

    TACOformer: Token-channel compounded cross attention for multimodal emotion recognition,

    X. Li, “TACOformer: Token-channel compounded cross attention for multimodal emotion recognition,” 2023, arXiv:2306.13592

  153. [161]

    Multi-branch mutual-distillation transformer for EEG- based seizure subtype classification,

    R. Peng et al., “Multi-branch mutual-distillation transformer for EEG- based seizure subtype classification,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 32, pp. 831-839, 2024

  154. [162]

    EMPT: A sparsity transformer for EEG-based motor imagery recognition,

    M. Liu et al. , “EMPT: A sparsity transformer for EEG-based motor imagery recognition,” Front. Neurosci., vol. 18, 2024. Art. no. 1366294

  155. [163]

    IEEG-HCT: A hierarchical CNN-transformer com- bined network for intracranial EEG signal identification,

    M. Yang et al. , “IEEG-HCT: A hierarchical CNN-transformer com- bined network for intracranial EEG signal identification,” IEEE Sensors Lett., vol. 8, no. 2, pp. 1-4, Feb. 2024

  156. [164]

    EEG-based seizure prediction via transformer guided CNN,

    C. Li et al. , “EEG-based seizure prediction via transformer guided CNN,” Measurement, vol. 203, Nov. 2022. Art. no. 111948

  157. [165]

    A transformer-based deep neural network model for SSVEP classification,

    J. Chen et al. , “A transformer-based deep neural network model for SSVEP classification,” Neural Netw., vol. 164, pp. 521-534, Jul. 2023

  158. [166]

    Gated transformer for decoding human brain EEG signals,

    Y . Tao et al. , “Gated transformer for decoding human brain EEG signals,” in Proc. 43rd Annu. Int. Conf. IEEE Eng. Med. Biol. Soc. , Mexico, 2021, pp. 125-130

  159. [167]

    EENED: End-to-end neural epilepsy detection based on convolutional transformer,

    Liu, X. Zhou and Y . Liu, “EENED: End-to-end neural epilepsy detection based on convolutional transformer,” in Proc. IEEE Conf. Artif. Intell., Santa Clara, CA, USA, 2023, pp. 368-371

  160. [168]

    Pyramid vision transformer: A versatile backbone for dense prediction without convolutions,

    W. Wang, et al., “Pyramid vision transformer: A versatile backbone for dense prediction without convolutions,” in Proc. IEEE/CVF Int. Conf. on Comput. Vis. (ICCV) , 2021, pp. 568-578

  161. [169]

    ScatterFormer: Locally-invariant scattering trans- former for patient-independent multispectral detection of epileptiform discharges,

    R. Zheng et al. , “ScatterFormer: Locally-invariant scattering trans- former for patient-independent multispectral detection of epileptiform discharges,” Proc. AAAI Conf. Artif. Intell., vol. 37, pp. 148-158, 2023

  162. [170]

    JDAT: Joint-dimension-aware transformer with strong flexibility for EEG emotion recognition,

    Z. Wang et al., “JDAT: Joint-dimension-aware transformer with strong flexibility for EEG emotion recognition,” in Int. Conf. Electr. Electron. Inf. Eng. (EEIE) , vol. 13512, pp. 42-52, Jan. 2025

  163. [171]

    Swin transformer: Hierarchical vision transformer using shifted windows,

    Z. Liu et al., “Swin transformer: Hierarchical vision transformer using shifted windows,” in Proc. IEEE Int. Conf. Comput. Vis. , 2021, pp. 10012-10022

  164. [172]

    A novel algorithmic structure of EEG channel attention combined with swin transformer for motor patterns classification,

    H. Wang et al., “A novel algorithmic structure of EEG channel attention combined with swin transformer for motor patterns classification,” IEEE Trans. Neural Syst. Rehabil. Eng. , vol. 31, pp. 3132-3141, 2023

  165. [173]

    Spatio-temporal swin transformer-based 4- D EEG emotion recognition,

    Z. Chen, J. Jin and J. Pan, “Spatio-temporal swin transformer-based 4- D EEG emotion recognition,” in Proc. IEEE Int. Conf. Bioinf. Biomed., Istanbul, Turkiye, 2023, pp. 1850-1855

  166. [174]

    A temporal-spectral fusion transformer with subject-specific adapter for enhancing RSVP-BCI decoding,

    X. Li et al. , “A temporal-spectral fusion transformer with subject-specific adapter for enhancing RSVP-BCI decoding,” 2024, arXiv:2401.06340

  167. [175]

    Functional emotion transformer for EEG-assisted cross-modal emotion recognition,

    W. B. Jiang et al.., “Functional emotion transformer for EEG-assisted cross-modal emotion recognition,” in Proc. IEEE Int. Conf. Acoust. Speech Signal Process. , Seoul, Korea, Republic of, 2024, pp. 1841- 1845

  168. [176]

    Research on multimodal emotion recognition based on fusion of electroencephalogram and electrooculography,

    J. Yin et al. , “Research on multimodal emotion recognition based on fusion of electroencephalogram and electrooculography,” IEEE Trans. Instrum. Meas., vol. 73, pp. 1-12, 2024

  169. [177]

    CoRe-sleep: A multimodal fusion framework for time series robust to imperfect modalities,

    K. Kontras et al. , “CoRe-sleep: A multimodal fusion framework for time series robust to imperfect modalities,” IEEE Trans. Neural Syst. Rehabil. Eng., vol. 32, pp. 840-849, 2024

  170. [178]

    BNMTrans: A brain network sequence-driven manifold- based transformer for cognitive impairment detection using EEG,

    R. Qin et al., “BNMTrans: A brain network sequence-driven manifold- based transformer for cognitive impairment detection using EEG,” in Proc. IEEE Int. Conf. Acoust. Speech Signal Process. , Seoul, Korea, Republic of, 2024, pp. 2016-2020

  171. [179]

    Self-supervised learning for electroencephalogram: A systematic survey

    W. Weng, et al. , “Self-supervised learning for electroencephalogram: A systematic survey.” 2024, arXiv:2401.05446

  172. [180]

    Brain foundation models: A survey on advancements in neural signal processing and brain discovery,

    X. Zhou et al., “Brain foundation models: A survey on advancements in neural signal processing and brain discovery,” 2025,arXiv:2503:00580

  173. [181]

    The explainability of transformers: current status and directions

    P. Fantozzi and M. Naldi. “The explainability of transformers: current status and directions.” Computers, vol.13, no. 4, p. 92, 2024

  174. [182]

    A resource-efficient algorithm–hardware co-design to- ward semi-supervised neurological symptoms prediction,

    S. Zhao et al., “A resource-efficient algorithm–hardware co-design to- ward semi-supervised neurological symptoms prediction,” IEEE Trans. Instrum. Meas., vol. 74, pp. 1-12, 2025

  175. [183]

    Application of transformer with auto-encoder in motor imagery EEG signals,

    R. Jiang et al., “Application of transformer with auto-encoder in motor imagery EEG signals,” in Proc. 14th Int. Conf. Wireless Commun. Signal Process., Nanjing, China, 2022, pp. 1-7

  176. [184]

    A survey on bridging EEG signals and generative AI: from image and text to beyond,

    S. Shukla et al. , “A survey on bridging EEG signals and generative AI: from image and text to beyond,” 2025, arXiv:2502.12048. JOURNAL OF LATEX CLASS FILES, VOL. 18, NO. 9, SEPTEMBER 2020 16 Supplementary Material for “Transformer Based EEG Decoding: A Survey” Haodong Zhang, ...

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.