Pith. sign in

REVIEW 5 major objections 5 minor 1 cited by

CSBrain: A Cross-scale Spatiotemporal Brain Foundation Model for EEG Decoding

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

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read CSBrain claims that modeling EEG across multiple temporal and spatial scales, rather than with fixed tokens and dense attention, yields a foundation model that beats existing task-specific and foundation baselines on 11 decoding tasks…

desk verdict A genuinely new cross-scale EEG architecture with an unusually broad benchmark, but the results are unverified until pretraining/test overlap and channel harmonization are addressed. read the letter →

arxiv 2506.23075 v1 pith:QEJZQYP2 submitted 2025-06-29 cs.HC cs.LGeess.SPq-bio.NC

Yuchen Zhou , Jiamin Wu , Zichen Ren , Zhouheng Yao , Weiheng Lu , Kunyu Peng , Qihao Zheng , Chunfeng Song
show 2 more authors
Wanli Ouyang Chao Gou
This is my paper
classification cs.HCcs.LGeess.SPq-bio.NC
keywords EEGfoundationmodelcross-scalespatiotemporalmodelingstructuredsparseattentionmaskedautoencodingbrain-computerinterfacedecodingmulti-scaletokenizationtransferlearning
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

CSBrain argues that EEG foundation models fall short when they treat brain signals as flat sequences of fixed-scale tokens processed by dense attention. The paper proposes that generalized EEG decoding should instead respect the cross-scale spatiotemporal structure of neural activity: brief bursts versus slow rhythms in time, and localized versus distributed patterns in space. Its architecture couples Cross-scale Spatiotemporal Tokenization (CST), which builds multi-scale tokens within temporal windows and anatomical brain regions, with Structured Sparse Attention (SSA), which restricts attention to grouped windows and regions. Pretrained by masked autoencoding on over 9,000 hours of unlabeled clinical EEG, CSBrain is evaluated on 11 tasks across 16 datasets and reports the highest macro-average score, beating the strongest foundation baselines CBraMod, LaBraM, and BIOT by 3.35%, 3.98%, and 7.73%, respectively. If the results hold, cross-scale structure is a useful inductive bias for generalized EEG decoding.

What carries the argument

The central objects are two interleaved modules. Cross-scale Spatiotemporal Tokenization (CST) applies multi-scale temporal convolution kernels (sizes 1, 3, 5) over local time windows and multi-scale spatial convolutions over electrode neighborhoods inside anatomically defined brain regions, concatenating the scales into compact tokens with embedding dimensions allocated by an exponentially decaying scheme $d_k \propto 1/2^k$. Structured Sparse Attention (SSA) replaces dense attention with inter-window attention, which groups tokens by their relative position across time windows, and inter-region attention, which forms a descriptor per brain region from a sampled token plus the mean-pooled regional feature and attends across regions; this yields roughly linear complexity $O(N \cdot k)$. CST and SSA are alternately stacked for 12 layers, so the model progressively integrates local multi-scale patterns into global structured dependencies.

What would settle it

Retrain and evaluate CSBrain on a held-out dataset such as CHB-MIT or HMC using the original electrode montage and channel count without projecting onto the 19-channel template, while keeping the baselines under the same protocol; if the cross-scale spatial grouping no longer yields a gain, or if a baseline retrained on the projected data closes the 3.35-7.73% macro-average gap, the paper's central claim would be refuted.

Watch

Extended reading notes

Core claim

The paper's central claim is that generalized EEG decoding requires modeling the cross-scale spatiotemporal structure of neural activity, and that a foundation model built on this principle outperforms both task-specific models and existing EEG foundation models that use fixed-scale tokenization with dense attention. On the paper's own terms, CSBrain's Cross-scale Spatiotemporal Tokenization and Structured Sparse Attention produce scale-aware tokens and structured long-range dependencies, and the reported macro-average across 11 tasks and 16 datasets is the highest, surpassing CBraMod, LaBraM, and BIOT by 3.35%, 3.98%, and 7.73%, respectively.

Load-bearing premise

The load-bearing premise is that every downstream dataset can be mapped onto the same fixed 19-channel 10-20 electrode template used in pretraining, even though the evaluated datasets have 4 to 64 channels and a mix of bipolar and referential montages; if that mapping is not coherent, the CST spatial regions and SSA inter-region attention are ill-defined for most datasets.

Editorial extensions

If this is right

  • If CSBrain is right, future EEG foundation models should replace fixed-scale tokenization with multi-scale, region-aware token construction.
  • Structure-aware sparse attention can match or beat dense attention on noisy EEG while reducing computational cost, making large-scale EEG modeling more practical.
  • Large-scale self-supervised pretraining is consistently beneficial across 16 datasets, and the gains are largest when downstream labeled data is scarce.
  • Cross-scale modeling becomes a reusable architectural template for other neurophysiological modalities, as the paper itself suggests for MEG and fMRI.

Reading between the lines

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

  • I infer that the strongest test of whether cross-scale structure matters is a controlled ablation that removes CST while keeping everything else identical; the paper's ablations vary tokenization together with attention, so the individual contribution of spatial grouping remains partly entangled.
  • The fixed 19-channel template means datasets with 64 channels or bipolar derivations must be down-projected or re-referenced before the CST spatial regions exist, so the montage-mapping step is the first place where gains could vanish across institutions.
  • If the architecture generalizes as claimed, the same CST and SSA recipe should transfer to MEG and high-density EEG without retraining the spatial grouping, because the anatomical region prior is montage-independent; this is a testable extension the paper does not run.
  • The scaling analysis suggests EEG foundation models follow neural scaling laws, but the plateau after about 5,000 hours of pretraining data implies that adding more unlabeled EEG alone will not keep improving accuracy; model capacity and data must grow together.
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

5 major / 5 minor

Summary. The paper proposes CSBrain, an EEG foundation model built on two new components: Cross-scale Spatiotemporal Tokenization (CST), which aggregates multi-scale features over local temporal windows and anatomically defined brain regions, and Structured Sparse Attention (SSA), which restricts attention to structured temporal and spatial groups. The model is pretrained by masked autoencoding on 1,109,545 segments from the Temple University Hospital EEG corpus (TUEG) and then fine-tuned on 11 tasks across 16 public datasets. The central claim is that explicitly modeling cross-scale spatiotemporal structure, as opposed to the scale-agnostic dense modeling of prior EEG foundation models, yields consistently better downstream decoding performance. The paper reports a macro-average gain of 3.35% over CBraMod, 3.98% over LaBraM, and 7.73% over BIOT, along with ablations on tokenization scale, attention mechanism, pretraining data volume, model capacity, and regional descriptor design.

Significance. If the results hold, CSBrain provides a genuinely useful architectural hypothesis for EEG foundation models: replacing fixed-scale, dense tokenization/attention with cross-scale, anatomically structured sparse operations is a principled inductive bias. The strengths of the paper are its unusually broad evaluation (16 datasets, 10 baselines), the inclusion of scaling analyses and ablations, and the stated intention to release code and models. The cross-scale tokenization and structured sparse attention are computationally plausible and offer a clear qualitative improvement over dense attention. However, the current manuscript leaves two load-bearing verification gaps: the absence of any stated protocol for excluding TUEG pretraining data from the TUEG-derived downstream test sets, and the absence of any description of how the fixed 19-channel pretrained model is adapted to datasets with 4 to 64 channels. Until these are resolved, the reported quantitative advantages cannot be treated as established.

major comments (5)
  1. [Section 3.1, Appendix D.9-D.11] The pretraining set and three downstream datasets are not shown to be disjoint. Section 3.1 states that CSBrain is pretrained on 1,109,545 EEG segments from TUEG, while Table 1 and Appendix D.9-D.11 evaluate on TUEV, TUAB, and TUSL, which are derived from the same Temple University Hospital EEG corpus (reference [71]). The paper never states that these datasets, or at least their test partitions, were excluded from pretraining. Because pretraining is unsupervised masked reconstruction, overlapping recordings could be memorized at the segment level, inflating exactly the tasks where CSBrain shows its largest margins (e.g., TUSL B-Acc of 0.8571 vs 0.7625 for LaBraM). Please state the exclusion protocol explicitly, and if overlap cannot be ruled out, rerun the affected downstream experiments with a disjoint pretraining set or report the overlap-adjusted results.
  2. [Section 2.2, Table 1, Table 3] The channel adaptation from the pretrained model to downstream datasets is unspecified. The pretraining configuration in Table 3 fixes the input to 19 channels, yet Table 1 lists datasets with 4, 6, 16, 17, 20, 22, 23, 29, 32, 62, and 64 channels, with mixed bipolar and referential montages. The paper does not state whether channels are selected, re-referenced, interpolated, or projected to the 19-channel 10-20 template, nor whether the input embedding layer is reinitialized for each dataset. Without this information, the 'anatomically defined brain regions' R used in Eq. (2) and the regional descriptors in Eq. (4) are not well defined for most downstream datasets, and the spatial tokenization cannot be reproduced. Please specify the exact per-dataset channel mapping and confirm that the pretrained weights used are compatible with that mapping.
  3. [Section 3.3, Table 2] The claim that CSBrain 'consistently achieves state-of-the-art performance across nearly all tasks and metrics' is not supported by Table 2. On the metrics shown, CSBrain is not the best on CHB-MIT B-Acc (0.7262 vs CBraMod 0.7398), SEED-VIG Corr (0.6314 vs LaBraM 0.6347), TUAB AUROC (0.8957 vs LaBraM 0.9022), TUEV F1-W (0.8333 vs CBraMod 0.8342), and ISRUC F1-W (0.7990 vs CBraMod 0.8011). Please either correct the wording to acknowledge these exceptions or present a more nuanced comparison. In addition, the macro-average in Table 2 is reported without standard deviations or significance tests; the paper should report whether the headline differences are statistically significant across the five seeds.
  4. [Section 2.3, Eq. (3)] The SSA inter-window grouping is under-specified. The text says that for each relative index g in {1,...,w}, the model collects 'all tokens that occupy the same position g in each window,' but the windows W(k)_t(i) are defined per location i with kernel size s(k)_t, and the paper does not define the window layout over the sequence, the membership of tokens in windows, or how w is chosen. Without a formal definition of the window partition, the claimed linear complexity O(N·k) and the exact attention pattern are not reproducible. Please provide a precise algorithmic description of the grouping, including how overlapping windows are handled.
  5. [Appendix C.2, Appendix D] The provenance of the baseline results is not fully specified. Appendix C.2 states that when baselines are 'unavailable or incompatible' they are re-implemented and re-trained, and it describes a channel projection for BIOT, but it does not state whether the foundation baselines (BIOT, LaBraM, CBraMod) were used with their official pretrained checkpoints or were retrained from scratch, nor does it give the per-dataset fine-tuning hyperparameters for each baseline. Because the main claim is superiority over these foundation models, the paper should state the source of every baseline result and the exact fine-tuning protocol used for each dataset.
minor comments (5)
  1. [Section 2.2, Eq. (2)] Equation (2) reuses the symbol Conv(k)_t for the spatial convolution kernels even though the temporal kernels in Eq. (1) use the same symbol; this should be Conv(k)_s or a distinct notation for the spatial kernels.
  2. [Section 2.4] The masking description says masked segments are 'represented by learnable embeddings and integrated alongside visible tokens during encoding,' but standard masked autoencoding omits masked tokens from the encoder. The text should clarify whether masked embeddings are added to the encoder input or only used at the reconstruction head, since this affects the meaning of the reconstruction loss in Eq. (7).
  3. [Appendix E.1, Figure 8] The comparison between CSBrain with and without pretraining is informative, but the figure does not state whether the non-pretrained model uses identical hyperparameters, initialization, or training epochs; please add these details.
  4. [Appendix E.2, Figures 9 and 10] The scaling analysis figures lack axis labels and error bars in the current version; please add them so the claims about plateauing and scaling trends can be evaluated.
  5. [General] The abstract and Section 3.3 use the phrases 'consistently outperforms' and 'consistently achieves state-of-the-art,' which are stronger than the results in Table 2 justify; please soften these to 'typically outperforms' or 'achieves the best macro-average' where appropriate.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: CSBrain's architecture and benchmark claims are self-contained, with only peripheral self-citations.

full rationale

CSBrain's derivation chain is constructive rather than definitional: CST tokenization (Eqs. 1-2) and SSA attention (Eqs. 3-5) are defined from raw EEG features and standard 10-20 spatial groupings, and the pretraining objective (Eq. 7) is a masked MSE reconstruction over those same segments. No fitted parameter is fed back as a prediction: hyperparameters (Table 3, Table 4) are fixed settings, and ablations (Fig. 4, Fig. 5) vary architectural choices without optimizing the reported benchmarks. The cited self-work ([4], [95], [97]) appears in general motivation and related-work context and is not load-bearing for the architecture or the headline numbers. The paper does not invoke a uniqueness theorem or smuggle an ansatz through a self-citation. The main substantive risk is experimental rather than circular: pretraining on TUEG (Section 3.1) with downstream TUEV/TUAB/TUSL (Section D.9-D.11) from the same TUH corpus [71] could inflate gains if subject/recording overlap is not excluded, but the paper provides no equation or definition that makes any reported result equivalent to its input by construction, so per the hard rules this is a correctness risk, not circularity.

Assumptions & free parameters 6 free parameters · 5 assumptions · 0 invented entities

The central empirical comparison rests on the preprocessing and region-mapping assumptions above, plus the pretraining-transfer assumption. None of these are derived; they are standard or ad hoc domain choices. The main unstated free quantities are the segment length t and the exact region mapping for partial montages, which are needed to replicate the model.

free parameters (6)
  • CST kernel sizes {1, 3, 5} = {1, 3, 5}
    Multi-scale temporal and spatial convolution kernel sizes are chosen by hand (Section 2.2); the paper varies the number of kernels K but not which sizes are used.
  • Number of CST/SSA layers L = 12
    The final model stacks 12 layers; the scaling study in Section E.2 varies this from 2 to 12, so 12 is a chosen configuration.
  • Hidden dimension d = 200
    Set to 200 in Table 3; used for all token embeddings and attention.
  • Mask ratio r = 0.5
    Pretraining masks 50% of segments (Section 2.4, Table 3); chosen without a sensitivity study.
  • Segment length t = not stated
    The temporal segment length in preprocessing (Ep in R^{C x n x t}) is never specified, yet it determines token granularity and masking.
  • Brain region partition = Frontal/Central/Parietal/Temporal/Occipital
    Electrodes are assigned to five anatomical regions from 10-20 labels (Section D), but the exact mapping for datasets with 4-6 channels or bipolar montages is not given.
assumptions (5)
  • domain assumption Standardized preprocessing (band-pass, notch, resample to 200 Hz, scaling to 100 µV) preserves task-relevant EEG information across all datasets.
    Section 2.1; every result depends on this.
  • domain assumption The 10-20 electrode labels are valid proxies for anatomical brain regions, and the same region grouping is meaningful for every downstream dataset.
    Section 2.2 and D; datasets range from 4 to 64 channels with mixed montages, making this non-trivial.
  • domain assumption Pretraining on 1.1M TUEG segments transfers to all downstream tasks.
    Section 3.1; no task-specific pretraining or domain adaptation is described.
  • domain assumption Masked autoencoding learns representations useful for arbitrary downstream classification and regression tasks.
    Section 2.4; standard SSL assumption inherited from MAE.
  • ad hoc to paper The structured sparse attention grouping rules (same relative position within each window; one representative token per region) capture the dependencies needed for EEG decoding.
    Section 2.3; the specific grouping is introduced for this model without independent justification.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CSBrain: A Cross-scale Spatiotemporal Brain Foundation Model for EEG Decoding." pith.science (2026). https://pith.science/paper/QEJZQYP2

@misc{pith2026250623075,
  author       = {Pith},
  title        = {Pith review of: CSBrain: A Cross-scale Spatiotemporal Brain Foundation Model for EEG Decoding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QEJZQYP2}},
  note         = {Machine review of arXiv:2506.23075}
}
read the original abstract

Understanding and decoding brain activity from electroencephalography (EEG) signals is a fundamental challenge in neuroscience and AI, with applications in cognition, emotion recognition, diagnosis, and brain-computer interfaces. While recent EEG foundation models advance generalized decoding via unified architectures and large-scale pretraining, they adopt a scale-agnostic dense modeling paradigm inherited from NLP and vision. This design neglects a core property of neural activity: cross-scale spatiotemporal structure. EEG task patterns span a wide range of temporal and spatial scales, from short bursts to slow rhythms, and from localized cortical responses to distributed interactions. Ignoring this diversity leads to suboptimal representations and weak generalization. We propose CSBrain, a Cross-scale Spatiotemporal Brain foundation model for generalized EEG decoding. CSBrain introduces: (i) Cross-scale Spatiotemporal Tokenization (CST), which aggregates multi-scale features from localized temporal windows and anatomical brain regions into compact scale-aware tokens; and (ii) Structured Sparse Attention (SSA), which captures cross-window and cross-region dependencies, enhancing scale diversity while removing spurious correlations. CST and SSA are alternately stacked to progressively integrate multi-scale dependencies. Experiments on 11 EEG tasks across 16 datasets show that CSBrain consistently outperforms task-specific and foundation model baselines. These results establish cross-scale modeling as a key inductive bias and position CSBrain as a robust backbone for future brain-AI research.

Figures

Figures reproduced from arXiv: 2506.23075 by the authors.

Figure 1
Figure 1. Comparison between previous brain foundation models and our proposed CSBrain. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Overview of the CSBrain architecture. After EEG signal preprocessing, the Cross [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. The overview of CSBrain pre-training process. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (16 more)
Figure 4
Figure 4. Figure 4: The performance comparisons of tokenization. [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: The performance comparisons of attention mechanism. [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Topography visualization on four different tasks. More in the supplementary materials. [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: Training loss curve during CSBrain pretraining. [PITH_FULL_IMAGE:figures/full_fig_p019_7.png]
Figure 8
Figure 8. Figure 8: Performance comparison between CSBrain without and with pretraining. [PITH_FULL_IMAGE:figures/full_fig_p031_8.png]
Figure 9
Figure 9. Figure 9: Effect of pretraining data volume on CSBrain performance. [PITH_FULL_IMAGE:figures/full_fig_p032_9.png]
Figure 10
Figure 10. Figure 10: Effect of model parameter size on CSBrain performance. [PITH_FULL_IMAGE:figures/full_fig_p032_10.png]
Figure 11
Figure 11. Figure 11: Ablation on regional descriptor design [PITH_FULL_IMAGE:figures/full_fig_p032_11.png]
Figure 12
Figure 12. Figure 12: Topography visualization of SEED-VIG. For the BCIC-IV-2a dataset, as shown in [PITH_FULL_IMAGE:figures/full_fig_p033_12.png]
Figure 13
Figure 13. Figure 13: Topography visualization of BCIC-IV-2a. For the SEED-V dataset, as shown in [PITH_FULL_IMAGE:figures/full_fig_p033_13.png]
Figure 14
Figure 14. Figure 14: Topography visualization of SEED-V. ‘thank you’ ‘yes’ ‘hello’ ‘help me’ ‘stop’ [PITH_FULL_IMAGE:figures/full_fig_p034_14.png]
Figure 15
Figure 15. Figure 15: Topography visualization of BCIC2020-3. elicit distinct activation regions and spatial scales, reinforcing the importance of modeling cross-scale neural dynamics. By explicitly capturing such multi-scale patterns, CSBrain effectively adapts to the heterogeneous spatia…
Figure 16
Figure 16. Figure 16: t-SNE visualization of feature representations on BCIC-IV-2a. Left: raw EEG signals; [PITH_FULL_IMAGE:figures/full_fig_p035_16.png]
Figure 17
Figure 17. Figure 17: t-SNE visualization of feature representations on BCIC2020-3. Left: raw EEG signals; [PITH_FULL_IMAGE:figures/full_fig_p035_17.png]
Figure 18
Figure 18. Figure 18: Channel similarity relationships of BCIC-IV-2a. [PITH_FULL_IMAGE:figures/full_fig_p036_18.png]
Figure 19
Figure 19. Figure 19: Channel similarity relationships of BCIC2020-3. [PITH_FULL_IMAGE:figures/full_fig_p037_19.png]

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

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

  1. Large-Scale AI and Foundation Models for Neuroscience: A Comprehensive Review

    cs.AI 2025-10 conditional novelty 1.0 of 10

    This paper is a survey: it organizes existing foundation-model work in neuroscience into five application domains and lists public datasets, without presenting new experiments.

Reference graph

Works this paper leans on

109 extracted references · 56 canonical work pages · cited by 1 Pith paper

  1. [71]

    The temple university hospital eeg data corpus

    Iyad Obeid and Joseph Picone. The temple university hospital eeg data corpus. Frontiers in neuroscience, 10:196, 2016

    work page 2016

  2. [1]

    Motor imagery and direct brain-computer communication

    Gert Pfurtscheller and Christa Neuper. Motor imagery and direct brain-computer communication. Proceedings of the IEEE, 89(7):1123–1134, 2001

    2001

  3. [2]

    Decoding mental states from brain activity in humans

    John-Dylan Haynes and Geraint Rees. Decoding mental states from brain activity in humans. Nature reviews neuroscience, 7(7):523–534, 2006

    2006

  4. [3]

    Beyond the brain-computer interface: Decoding brain activity as a tool to understand neuronal mechanisms subtending cognition and behavior

    Célia Loriette, Julian L Amengual, and Suliann Ben Hamed. Beyond the brain-computer interface: Decoding brain activity as a tool to understand neuronal mechanisms subtending cognition and behavior. Frontiers in Neuroscience, 16:811736, 2022

    2022

  5. [4]

    Mindaligner: Explicit brain functional alignment for cross-subject visual decoding from limited fmri data

    Yuqin Dai, Zhouheng Yao, Chunfeng Song, Qihao Zheng, Weijian Mai, Kunyu Peng, Shuai Lu, Wanli Ouyang, Jian Yang, and Jiamin Wu. Mindaligner: Explicit brain functional alignment for cross-subject visual decoding from limited fmri data. arXiv preprint arXiv:2502.05034, 2025

    arXiv 2025

  6. [5]

    Electroencephalography

    Gernot R Müller-Putz. Electroencephalography. Handbook of clinical neurology, 168:249–262, 2020

    2020

  7. [6]

    A brain- computer interface that evokes tactile sensations improves robotic arm control

    Sharlene N Flesher, John E Downey, Jeffrey M Weiss, Christopher L Hughes, Angelica J Herrera, Elizabeth C Tyler-Kabara, Michael L Boninger, Jennifer L Collinger, and Robert A Gaunt. A brain- computer interface that evokes tactile sensations improves robotic arm control. Science, 372(6544):831– 836, 2021

    2021

  8. [7]

    Combining vr with electroencephalography as a frontier of brain-computer interfaces

    Hongbian Li, Hyonyoung Shin, Luis Sentis, Ka-Chun Siu, José del R Millán, and Nanshu Lu. Combining vr with electroencephalography as a frontier of brain-computer interfaces. Device, 2(6), 2024

    2024

Show all 109 references
  1. [8]

    Chrononet: A deep recurrent neural network for abnormal eeg identification

    Subhrajit Roy, Isabell Kiral-Kornek, and Stefan Harrer. Chrononet: A deep recurrent neural network for abnormal eeg identification. In Artificial Intelligence in Medicine: Conference on Artificial Intelligence in Medicine, pages 47–56. Springer, 2019

    2019

  2. [9]

    Automatic signal abnormality detection using time-frequency features and machine learning: A newborn eeg seizure case study

    Boualem Boashash and Samir Ouelha. Automatic signal abnormality detection using time-frequency features and machine learning: A newborn eeg seizure case study. Knowledge-Based Systems, 106:38–50, 2016

    2016

  3. [10]

    Deep learning for motor imagery eeg-based classification: A review

    Ali Al-Saegh, Shefa A Dawwd, and Jassim M Abdul-Jabbar. Deep learning for motor imagery eeg-based classification: A review. Biomedical Signal Processing and Control, 63:102172, 2021

    2021

  4. [11]

    A spatial filter temporal graph convolutional network for decoding motor imagery eeg signals

    Xianlun Tang, Jing Zhang, Yidan Qi, Ke Liu, Rui Li, and Huiming Wang. A spatial filter temporal graph convolutional network for decoding motor imagery eeg signals. Expert Systems with Applications, 238:121915, 2024

    2024

  5. [12]

    Eeg-based emotion recognition: a state- of-the-art review of current trends and opportunities

    Nazmi Sofian Suhaimi, James Mountstephens, and Jason Teo. Eeg-based emotion recognition: a state- of-the-art review of current trends and opportunities. Computational intelligence and neuroscience , 2020(1):8875426, 2020

    2020

  6. [13]

    Contrastive learning of eeg representation of brain area for emotion recognition

    Sheng Dai, Ming Li, Xu Wu, Xiangyu Ju, Xinyu Li, Jun Yang, and Dewen Hu. Contrastive learning of eeg representation of brain area for emotion recognition. IEEE Transactions on Instrumentation and Measurement, 2025. 10

    2025

  7. [14]

    Self-supervised eeg representation learning for automatic sleep staging

    Chaoqi Yang, Danica Xiao, M Brandon Westover, and Jimeng Sun. Self-supervised eeg representation learning for automatic sleep staging. JMIR AI, 2023

    2023

  8. [15]

    Generalizable sleep staging via multi-level domain alignment

    Jiquan Wang, Sha Zhao, Haiteng Jiang, Shijian Li, Tao Li, and Gang Pan. Generalizable sleep staging via multi-level domain alignment. In Proceedings of the AAAI Conference on Artificial Intelligence , volume 38, pages 265–273, 2024

    2024

  9. [16]

    A hybrid deep learning approach for epileptic seizure detection in eeg signals

    Ijaz Ahmad, Xin Wang, Danish Javeed, Prabhat Kumar, Oluwarotimi Williams Samuel, and Shixiong Chen. A hybrid deep learning approach for epileptic seizure detection in eeg signals. IEEE Journal of Biomedical and Health Informatics, 2023

    2023

  10. [17]

    Eeg datasets for seizure detection and prediction—a review

    Sheng Wong, Anj Simmons, Jessica Rivera-Villicana, Scott Barnett, Shobi Sivathamboo, Piero Perucca, Zongyuan Ge, Patrick Kwan, Levin Kuhlmann, Rajesh Vasa, et al. Eeg datasets for seizure detection and prediction—a review. Epilepsia Open, 8(2):252–267, 2023

    2023

  11. [18]

    Automatic and efficient framework for identifying multiple neurological disorders from eeg signals

    Md Nurul Ahad Tawhid, Siuly Siuly, Kate Wang, and Hua Wang. Automatic and efficient framework for identifying multiple neurological disorders from eeg signals. IEEE Transactions on Technology and Society, 4(1):76–86, 2023

    2023

  12. [19]

    A review of graph theory-based diagnosis of neurological disorders based on eeg and mri

    Ying Yan, Guanting Liu, Haoyang Cai, Edmond Qi Wu, Jun Cai, Adrian David Cheok, Na Liu, Tao Li, and Zhiyong Fan. A review of graph theory-based diagnosis of neurological disorders based on eeg and mri. Neurocomputing, page 128098, 2024

    2024

  13. [20]

    Medformer: A multi-granularity patching transformer for medical time-series classification

    Yihe Wang, Nan Huang, Taida Li, Yujun Yan, and Xiang Zhang. Medformer: A multi-granularity patching transformer for medical time-series classification. In The Annual Conference on Neural Information Processing Systems, 2024

    2024

  14. [21]

    Eegnet: a compact convolutional neural network for eeg-based brain–computer interfaces

    Vernon J Lawhern, Amelia J Solon, Nicholas R Waytowich, Stephen M Gordon, Chou P Hung, and Brent J Lance. Eegnet: a compact convolutional neural network for eeg-based brain–computer interfaces. Journal of neural engineering, 15(5):056013, 2018

    2018

  15. [22]

    Development of expert-level classification of seizures and rhythmic and periodic patterns during eeg interpretation

    Jin Jing, Wendong Ge, Shenda Hong, Marta Bento Fernandes, Zhen Lin, Chaoqi Yang, Sungtae An, Aaron F Struck, Aline Herlopian, Ioannis Karakis, et al. Development of expert-level classification of seizures and rhythmic and periodic patterns during eeg interpretation. Neurology,...

    2023

  16. [23]

    Self-supervised electroencephalogram representation learning for automatic sleep staging: Model development and evaluation study

    Chaoqi Yang, Cao Xiao, M Brandon Westover, and Jimeng Sun. Self-supervised electroencephalogram representation learning for automatic sleep staging: Model development and evaluation study. JMIR AI, 2, 2023

    2023

  17. [24]

    Motor imagery eeg classification algorithm based on cnn-lstm feature fusion network

    Hongli Li, Man Ding, Ronghua Zhang, and Chunbo Xiu. Motor imagery eeg classification algorithm based on cnn-lstm feature fusion network. Biomedical signal processing and control, 72:103342, 2022

    2022

  18. [25]

    Emotion recognition based on eeg using lstm recurrent neural network

    Salma Alhagry, Aly Aly Fahmy, and Reda A El-Khoribi. Emotion recognition based on eeg using lstm recurrent neural network. International Journal of Advanced Computer Science and Applications, 8(10), 2017

    2017

  19. [26]

    An efficient lstm network for emotion recognition from multichannel eeg signals

    Xiaobing Du, Cuixia Ma, Guanhua Zhang, Jinyao Li, Yu-Kun Lai, Guozhen Zhao, Xiaoming Deng, Yong-Jin Liu, and Hongan Wang. An efficient lstm network for emotion recognition from multichannel eeg signals. IEEE Transactions on Affective Computing, 13(3):1528–1540, 2020

    2020

  20. [27]

    Graph neural network-based eeg classification: A survey

    Dominik Klepl, Min Wu, and Fei He. Graph neural network-based eeg classification: A survey. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 32:493–503, 2024

    2024

  21. [28]

    Eeg-gnn: Graph neural networks for classification of electroencephalogram (eeg) signals

    Andac Demir, Toshiaki Koike-Akino, Ye Wang, Masaki Haruna, and Deniz Erdogmus. Eeg-gnn: Graph neural networks for classification of electroencephalogram (eeg) signals. In Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), pages 1061–1...

    2021

  22. [29]

    Eeg-based emotion recognition using regularized graph neural networks

    Peixiang Zhong, Di Wang, and Chunyan Miao. Eeg-based emotion recognition using regularized graph neural networks. IEEE Transactions on Affective Computing, 13(3):1290–1301, 2020

    2020

  23. [30]

    Eeg conformer: Convolutional transformer for eeg decoding and visualization

    Yonghao Song, Qingqing Zheng, Bingchuan Liu, and Xiaorong Gao. Eeg conformer: Convolutional transformer for eeg decoding and visualization. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 31:710–719, 2022

    2022

  24. [31]

    Transformer-based spatial-temporal feature learning for eeg decoding

    Yonghao Song, Xueyu Jia, Lie Yang, and Longhan Xie. Transformer-based spatial-temporal feature learning for eeg decoding. arXiv preprint arXiv:2106.11170, 2021. 11

    2021 arXiv

  25. [32]

    Transformer convolutional neural networks for automated artifact detection in scalp eeg

    Wei Yan Peh, Yuanyuan Yao, and Justin Dauwels. Transformer convolutional neural networks for automated artifact detection in scalp eeg. In Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), pages 3599–3602. IEEE, 2022

    2022

  26. [33]

    Toward reliable signals decoding for electroencephalogram: A benchmark study to eegnex

    Xia Chen, Xiangbin Teng, Han Chen, Yafeng Pan, and Philipp Geyer. Toward reliable signals decoding for electroencephalogram: A benchmark study to eegnex. Biomedical Signal Processing and Control, 87:105475, 2024

    2024

  27. [34]

    Biot: Biosignal transformer for cross-data learning in the wild

    Chaoqi Yang, M Westover, and Jimeng Sun. Biot: Biosignal transformer for cross-data learning in the wild. Advances in Neural Information Processing Systems, 36:78240–78260, 2023

    2023

  28. [35]

    Large brain model for learning generic representations with tremendous eeg data in bci

    Wei-Bang Jiang, Li-Ming Zhao, and Bao-Liang Lu. Large brain model for learning generic representations with tremendous eeg data in bci. In The International Conference on Learning Representations, 2024

    2024

  29. [36]

    Eegpt: Pretrained transformer for universal and reliable representation of eeg signals

    Guangyu Wang, Wenchao Liu, Yuhong He, Cong Xu, Lin Ma, and Haifeng Li. Eegpt: Pretrained transformer for universal and reliable representation of eeg signals. In Advances in Neural Information Processing Systems, volume 37, pages 39249–39280, 2024

    2024

  30. [37]

    Neuro-gpt: Towards a foundation model for eeg

    Wenhui Cui, Woojae Jeong, Philipp Thölke, Takfarinas Medani, Karim Jerbi, Anand A Joshi, and Richard M Leahy. Neuro-gpt: Towards a foundation model for eeg. In IEEE International Symposium on Biomedical Imaging (ISBI), pages 1–5. IEEE, 2024

    2024

  31. [38]

    Brainwave: A brain signal foundation model for clinical applications

    Zhizhang Yuan, Fanqi Shen, Meng Li, Yuguo Yu, Chenhao Tan, and Yang Yang. Brainwave: A brain signal foundation model for clinical applications. arXiv preprint arXiv:2402.10251, 2024

    2024 arXiv

  32. [39]

    A simple review of eeg foundation models: Datasets, advancements and future perspectives

    Junhong Lai, Jiyu Wei, Lin Yao, and Yueming Wang. A simple review of eeg foundation models: Datasets, advancements and future perspectives. arXiv preprint arXiv:2504.20069, 2025

    2025

  33. [40]

    CBramod: A criss-cross brain foundation model for EEG decoding

    Jiquan Wang, Sha Zhao, Zhiling Luo, Yangxuan Zhou, Haiteng Jiang, Shijian Li, Tao Li, and Gang Pan. CBramod: A criss-cross brain foundation model for EEG decoding. In The International Conference on Learning Representations, 2025

    2025

  34. [41]

    Attention is all you need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. Attention is all you need. Advances in neural information processing systems, 30, 2017

    2017

  35. [42]

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

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, Jakob Uszkoreit, and Neil Houlsby. An image is worth 16x16 words: Transformers for image recognition at...

    2021

  36. [43]

    Self-supervised learning: Generative or contrastive

    Xiao Liu, Fanjin Zhang, Zhenyu Hou, Li Mian, Zhaoyu Wang, Jing Zhang, and Jie Tang. Self-supervised learning: Generative or contrastive. IEEE transactions on knowledge and data engineering, 35(1):857– 876, 2021

    2021

  37. [44]

    Bendr: Using transformers and a contrastive self-supervised learning task to learn from massive amounts of eeg data

    Demetres Kostas, Stephane Aroca-Ouellette, and Frank Rudzicz. Bendr: Using transformers and a contrastive self-supervised learning task to learn from massive amounts of eeg data. Frontiers in Human Neuroscience, 15:653659, 2021

    2021

  38. [45]

    Masked autoencoders are scalable vision learners

    Kaiming He, Xinlei Chen, Saining Xie, Yanghao Li, Piotr Dollár, and Ross Girshick. Masked autoencoders are scalable vision learners. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 16000–16009, 2022

    2022

  39. [46]

    Cswin transformer: A general vision transformer backbone with cross-shaped windows

    Xiaoyi Dong, Jianmin Bao, Dongdong Chen, Weiming Zhang, Nenghai Yu, Lu Yuan, Dong Chen, and Baining Guo. Cswin transformer: A general vision transformer backbone with cross-shaped windows. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pa...

    2022

  40. [47]

    Principles of neural science, volume 4

    Eric R Kandel, James H Schwartz, Thomas M Jessell, Steven Siegelbaum, A James Hudspeth, Sarah Mack, et al. Principles of neural science, volume 4. McGraw-hill New York, 2000

    2000

  41. [48]

    Rhythms of the Brain

    György Buzsáki. Rhythms of the Brain. Oxford university press, 2006

    2006

  42. [49]

    A weighted and directed interareal connectivity matrix for macaque cerebral cortex

    Nikola T Markov, Maria M Ercsey-Ravasz, AR Ribeiro Gomes, Camille Lamy, Loic Magrou, Julien Vezoli, Pierre Misery, Arnaud Falchier, Rene Quilodran, Marie-Alice Gariel, et al. A weighted and directed interareal connectivity matrix for macaque cerebral cortex. Cerebral cortex, 2...

    2014

  43. [50]

    2020 international brain–computer interface competition: A review

    Ji-Hoon Jeong, Jeong-Hyun Cho, Young-Eun Lee, Seo-Hyun Lee, Gi-Hwan Shin, Young-Seok Kweon, José del R Millán, Klaus-Robert Müller, and Seong-Whan Lee. 2020 international brain–computer interface competition: A review. Frontiers in human neuroscience, 16:898300, 2022. 12

    2020

  44. [51]

    The sleep slow oscillation as a traveling wave

    Marcello Massimini, Reto Huber, Fabio Ferrarelli, Sean Hill, and Giulio Tononi. The sleep slow oscillation as a traveling wave. Journal of Neuroscience, 24(31):6862–6870, 2004

    2004

  45. [52]

    The wu-minn human connectome project: an overview

    David C Van Essen, Stephen M Smith, Deanna M Barch, Timothy EJ Behrens, Essa Yacoub, Kamil Ugurbil, Wu-Minn HCP Consortium, et al. The wu-minn human connectome project: an overview. Neuroimage, 80:62–79, 2013

    2013

  46. [53]

    Cognitive motor processes: the role of motor imagery in the study of motor representations

    Jörn Munzert, Britta Lorey, and Karen Zentgraf. Cognitive motor processes: the role of motor imagery in the study of motor representations. Brain research reviews, 60(2):306–326, 2009

    2009

  47. [54]

    Eeg based emotion recognition: A tutorial and review

    Xiang Li, Yazhou Zhang, Prayag Tiwari, Dawei Song, Bin Hu, Meihong Yang, Zhigang Zhao, Neeraj Kumar, and Pekka Marttinen. Eeg based emotion recognition: A tutorial and review. ACM Computing Surveys, 55(4):1–57, 2022

    2022

  48. [55]

    Eeg artifact removal—state-of-the-art and guidelines

    Jose Antonio Urigüen and Begoña Garcia-Zapirain. Eeg artifact removal—state-of-the-art and guidelines. Journal of neural engineering, 12(3):031001, 2015

    2015

  49. [56]

    Simmim: A simple framework for masked image modeling

    Zhenda Xie, Zheng Zhang, Yue Cao, Yutong Lin, Jianmin Bao, Zhuliang Yao, Qi Dai, and Han Hu. Simmim: A simple framework for masked image modeling. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 9653–9663, 2022

    2022

  50. [58]

    Bci competi- tion 2008–graz data set a

    Clemens Brunner, Robert Leeb, Gernot Müller-Putz, Alois Schlögl, and Gert Pfurtscheller. Bci competi- tion 2008–graz data set a. Institute for knowledge discovery (laboratory of brain-computer interfaces), Graz University of Technology, 16(1-6):1, 2008

    2008

  51. [59]

    Bci2000: a general-purpose brain-computer interface (bci) system

    Gerwin Schalk, Dennis J McFarland, Thilo Hinterberger, Niels Birbaumer, and Jonathan R Wolpaw. Bci2000: a general-purpose brain-computer interface (bci) system. IEEE Transactions on biomedical engineering, 51(6):1034–1043, 2004

    2004

  52. [60]

    A large eeg dataset for studying cross-session variability in motor imagery brain-computer interface.Scientific Data, 9(1):531, 2022

    Jun Ma, Banghua Yang, Wenzheng Qiu, Yunzhe Li, Shouwei Gao, and Xinxing Xia. A large eeg dataset for studying cross-session variability in motor imagery brain-computer interface.Scientific Data, 9(1):531, 2022

    2022

  53. [61]

    Gnn4eeg: A benchmark and toolkit for electroencephalography classification with graph neural network

    Kaiyuan Zhang, Ziyi Ye, Qingyao Ai, Xiaohui Xie, and Yiqun Liu. Gnn4eeg: A benchmark and toolkit for electroencephalography classification with graph neural network. In Companion of the ACM International Joint Conference on Pervasive and Ubiquitous Computing, pages 612–617, 2024

    2024

  54. [62]

    Comparing recognition performance and robustness of multimodal deep learning models for multimodal emotion recognition

    Wei Liu, Jie-Lin Qiu, Wei-Long Zheng, and Bao-Liang Lu. Comparing recognition performance and robustness of multimodal deep learning models for multimodal emotion recognition. IEEE Transactions on Cognitive and Developmental Systems, 2021

    2021

  55. [63]

    Physiobank, physiotoolkit, and physionet: components of a new research resource for complex physiologic signals

    Ary L Goldberger, Luis AN Amaral, Leon Glass, Jeffrey M Hausdorff, Plamen Ch Ivanov, Roger G Mark, Joseph E Mietus, George B Moody, Chung-Kang Peng, and H Eugene Stanley. Physiobank, physiotoolkit, and physionet: components of a new research resource for complex physiologic si...

    2000

  56. [64]

    Application of machine learning to epileptic seizure onset detection and treatment

    Ali Hossam Shoeb. Application of machine learning to epileptic seizure onset detection and treatment. PhD thesis, Massachusetts Institute of Technology, 2009

    2009

  57. [65]

    Eeg synchronization analysis for seizure prediction: A study on data of noninvasive recordings

    Paolo Detti, Giampaolo Vatti, and Garazi Zabalo Manrique de Lara. Eeg synchronization analysis for seizure prediction: A study on data of noninvasive recordings. Processes, 8(7), 2020

    2020

  58. [66]

    Isruc-sleep: A comprehensive public dataset for sleep researchers

    Sirvan Khalighi, Teresa Sousa, José Moutinho Santos, and Urbano Nunes. Isruc-sleep: A comprehensive public dataset for sleep researchers. Computer methods and programs in biomedicine , 124:180–192, 2016

    2016

  59. [67]

    Inter-database validation of a deep learning approach for automatic sleep scoring

    Diego Alvarez-Estevez and Roselyne M Rijsman. Inter-database validation of a deep learning approach for automatic sleep scoring. PloS one, 16(8):e0256111, 2021

    2021

  60. [68]

    A multimodal approach to estimating vigilance using eeg and forehead eog

    Wei-Long Zheng and Bao-Liang Lu. A multimodal approach to estimating vigilance using eeg and forehead eog. Journal of Neural Engineering, 14(2):026017, 2017

    2017

  61. [69]

    Electroencephalograms during mental arithmetic task performance

    Igor Zyma, Sergii Tukaev, Ivan Seleznov, Ken Kiyono, Anton Popov, Mariia Chernykh, and Oleksii Shpenkov. Electroencephalograms during mental arithmetic task performance. Data, 4(1):14, 2019. 13

    2019

  62. [70]

    MDD Patients and Healthy Controls EEG Data (New)

    Wajid Mumtaz. MDD Patients and Healthy Controls EEG Data (New). https://figshare.com/ articles/dataset/EEG_Data_New/4244171, 2016. Figshare

    2016

  63. [72]

    Grad-cam: Visual explanations from deep networks via gradient-based localization

    Ramprasaath R Selvaraju, Michael Cogswell, Abhishek Das, Ramakrishna Vedantam, Devi Parikh, and Dhruv Batra. Grad-cam: Visual explanations from deep networks via gradient-based localization. In Proceedings of the IEEE international conference on computer vision, pages 618–626, 2017

    2017

  64. [73]

    Event-related cortical desynchronization detected by power measure- ments of scalp eeg

    Gert Pfurtscheller and A Aranibar. Event-related cortical desynchronization detected by power measure- ments of scalp eeg. Electroencephalography and clinical neurophysiology, 42(6):817–826, 1977

    1977

  65. [74]

    Speech production: Wernicke, broca and beyond

    S Catrin Blank, Sophie K Scott, Kevin Murphy, Elizabeth Warburton, and Richard JS Wise. Speech production: Wernicke, broca and beyond. Brain, 125(8):1829–1838, 2002

    2002

  66. [75]

    Revealing the spatiotemporal brain dynamics of covert speech compared with overt speech: A simultaneous eeg-fmri study

    Wei Zhang, Muyun Jiang, Kok Ann Colin Teo, Raghavan Bhuvanakantham, LaiGuan Fong, Wei Khang Jeremy Sim, Zhiwei Guo, Chuan Huat Vince Foo, Rong Hui Jonathan Chua, Parasuraman Padmanabhan, et al. Revealing the spatiotemporal brain dynamics of covert speech compared with overt sp...

    2024

  67. [76]

    Fft-based deep feature learning method for eeg classification

    Mingyang Li and Wanzhong Chen. Fft-based deep feature learning method for eeg classification. Biomedical Signal Processing and Control, 66:102492, 2021

    2021

  68. [77]

    Epileptic seizure classification of eeg time-series using rational discrete short-time fourier transform

    Kaveh Samiee, Peter Kovacs, and Moncef Gabbouj. Epileptic seizure classification of eeg time-series using rational discrete short-time fourier transform. IEEE transactions on Biomedical Engineering , 62(2):541–552, 2014

    2014

  69. [78]

    An eeg based real-time epilepsy seizure detection approach using discrete wavelet transform and machine learning methods

    Mingkan Shen, Peng Wen, Bo Song, and Yan Li. An eeg based real-time epilepsy seizure detection approach using discrete wavelet transform and machine learning methods. Biomedical Signal Processing and Control, 77:103820, 2022

    2022

  70. [79]

    Optimizing spatial filters for robust eeg single-trial analysis

    Benjamin Blankertz, Ryota Tomioka, Steven Lemm, Motoaki Kawanabe, and Klaus-Robert Muller. Optimizing spatial filters for robust eeg single-trial analysis. IEEE Signal processing magazine, 25(1):41– 56, 2007

    2007

  71. [80]

    Multiclass brain– computer interface classification by riemannian geometry

    Alexandre Barachant, Stéphane Bonnet, Marco Congedo, and Christian Jutten. Multiclass brain– computer interface classification by riemannian geometry. IEEE Transactions on Biomedical Engineering, 59(4):920–928, 2011

    2011

  72. [81]

    Differential evolution algorithm as a tool for optimal feature subset selection in motor imagery eeg

    Muhammad Zeeshan Baig, Nauman Aslam, Hubert PH Shum, and Li Zhang. Differential evolution algorithm as a tool for optimal feature subset selection in motor imagery eeg. Expert Systems with Applications, 90:184–195, 2017

    2017

  73. [82]

    Multiclass support vector machines for eeg-signals classification

    Inan Guler and Elif Derya Ubeyli. Multiclass support vector machines for eeg-signals classification. IEEE transactions on information technology in biomedicine, 11(2):117–126, 2007

    2007

  74. [83]

    Comparison of signal decomposition methods in classification of eeg signals for motor-imagery bci system

    Jasmin Kevric and Abdulhamit Subasi. Comparison of signal decomposition methods in classification of eeg signals for motor-imagery bci system. Biomedical Signal Processing and Control, 31:398–406, 2017

    2017

  75. [84]

    Logistic regression for single trial eeg classification

    Ryota Tomioka, Kazuyuki Aihara, and Klaus-Robert Müller. Logistic regression for single trial eeg classification. Advances in neural information processing systems, 19, 2006

    2006

  76. [85]

    Deep learning with convolutional neural networks for eeg decoding and visualization

    Robin Tibor Schirrmeister, Jost Tobias Springenberg, Lukas Dominique Josef Fiederer, Martin Glasstetter, Katharina Eggensperger, Michael Tangermann, Frank Hutter, Wolfram Burgard, and Tonio Ball. Deep learning with convolutional neural networks for eeg decoding and visualizati...

    2017

  77. [86]

    Lmda-net: A lightweight multi-dimensional attention network for general eeg-based brain-computer interfaces and interpretability

    Zhengqing Miao, Meirong Zhao, Xin Zhang, and Dong Ming. Lmda-net: A lightweight multi-dimensional attention network for general eeg-based brain-computer interfaces and interpretability. NeuroImage, page 120209, 2023

    2023

  78. [87]

    Lstm-based eeg classification in motor imagery tasks

    Ping Wang, Aimin Jiang, Xiaofeng Liu, Jing Shang, and Li Zhang. Lstm-based eeg classification in motor imagery tasks. IEEE transactions on neural systems and rehabilitation engineering, 26(11):2086–2095, 2018

    2018

  79. [88]

    An efficient lstm network for emotion recognition from multichannel eeg signals

    Xiaobing Du, Cuixia Ma, Guanhua Zhang, Jinyao Li, Yu-Kun Lai, Guozhen Zhao, Xiaoming Deng, Yong-Jin Liu, and Hongan Wang. An efficient lstm network for emotion recognition from multichannel eeg signals. IEEE Transactions on Affective Computing, 13(3):1528–1540, 2020. 14

    2020

  80. [89]

    Sleeptransformer: Automatic sleep staging with interpretability and uncertainty quantification

    Huy Phan, Kaare Mikkelsen, Oliver Y Chén, Philipp Koch, Alfred Mertins, and Maarten De V os. Sleeptransformer: Automatic sleep staging with interpretability and uncertainty quantification. IEEE Transactions on Biomedical Engineering, 69(8):2456–2467, 2022

    2022

  81. [90]

    Graphsleepnet: Adaptive spatial-temporal graph convolutional networks for sleep stage classification

    Ziyu Jia, Youfang Lin, Jing Wang, Ronghao Zhou, Xiaojun Ning, Yuanlai He, and Yaoshuai Zhao. Graphsleepnet: Adaptive spatial-temporal graph convolutional networks for sleep stage classification. In Ijcai, volume 2021, pages 1324–1330, 2020

    2021

  82. [91]

    Bert: Pre-training of deep bidirectional transformers for language understanding

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. Bert: Pre-training of deep bidirectional transformers for language understanding. In Proceedings of the 2019 conference of the North American chapter of the association for computational linguistics: human langu...

    2019

  83. [92]

    Language models are few-shot learners

    Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. Language models are few-shot learners. Advances in neural information processing systems, 33:1877–1901, 2020

    1901

  84. [93]

    Learning transferable visual models from natural language supervision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. Learning transferable visual models from natural language supervision. In International conference on machine learning, pa...

    2021

  85. [94]

    FlexCAD: Unified and versatile controllable CAD generation with fine-tuned large language models

    Zhanwei Zhang, Shizhao Sun, Wenxiao Wang, Deng Cai, and Jiang Bian. FlexCAD: Unified and versatile controllable CAD generation with fine-tuned large language models. In The Thirteenth International Conference on Learning Representations, 2025

    2025

  86. [95]

    Learning from observer gaze: Zero-shot attention prediction oriented by human-object interaction recognition

    Yuchen Zhou, Linkai Liu, and Chao Gou. Learning from observer gaze: Zero-shot attention prediction oriented by human-object interaction recognition. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 28390–28400, 2024

    2024

  87. [96]

    Skeleton-in- context: Unified skeleton sequence modeling with in-context learning

    Xinshun Wang, Zhongbin Fang, Xia Li, Xiangtai Li, Chen Chen, and Mengyuan Liu. Skeleton-in- context: Unified skeleton sequence modeling with in-context learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 2436–2446, 2024

    2024

  88. [97]

    Drivinggen: Efficient safety-critical driving video generation with latent diffusion models

    Zipeng Guo, Yuchen Zhou, and Chao Gou. Drivinggen: Efficient safety-critical driving video generation with latent diffusion models. In 2024 IEEE International Conference on Multimedia and Expo (ICME), pages 1–6. IEEE, 2024

    2024

  89. [98]

    Brainbert: Self-supervised representation learning for intracranial recordings

    Christopher Wang, Vighnesh Subramaniam, Adam Uri Yaari, Gabriel Kreiman, Boris Katz, Ignacio Cases, and Andrei Barbu. Brainbert: Self-supervised representation learning for intracranial recordings. arXiv preprint arXiv:2302.14367, 2023

    2023 arXiv

  90. [99]

    Brant: Foundation model for intracranial neural signal

    Daoze Zhang, Zhizhang Yuan, Yang Yang, Junru Chen, Jingjing Wang, and Yafeng Li. Brant: Foundation model for intracranial neural signal. Advances in Neural Information Processing Systems, 36:26304– 26321, 2023

    2023

  91. [100]

    Eeg2rep: enhancing self-supervised eeg representation through informative masked inputs

    Navid Mohammadi Foumani, Geoffrey Mackellar, Soheila Ghane, Saad Irtza, Nam Nguyen, and Mahsa Salehi. Eeg2rep: enhancing self-supervised eeg representation through informative masked inputs. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Minin...

    2024

  92. [101]

    Ten-twenty electrode system of the international federation

    Herbert H Jasper. Ten-twenty electrode system of the international federation. Electroencephalogr Clin Neurophysiol, 10:371–375, 1958

    1958

  93. [102]

    Ten percent electrode system for topographic studies of spontaneous and evoked eeg activities

    Gian Emilio Chatrian, Ettore Lettich, and Paula L Nelson. Ten percent electrode system for topographic studies of spontaneous and evoked eeg activities. American Journal of EEG technology, 25(2):83–92, 1985

    1985

  94. [103]

    Guideline thirteen: guidelines for standard electrode position nomenclature

    American Electroencephalographic Society. Guideline thirteen: guidelines for standard electrode position nomenclature. J Clin Neurophysiol, 11(1):111–113, 1994

    1994

  95. [104]

    Three-dimensional probabilistic anatomical cranio-cerebral correlation via the international 10–20 system oriented for transcranial functional brain mapping

    Masako Okamoto, Haruka Dan, Kuniko Sakamoto, Kazuhiro Takeo, Koji Shimizu, Satoru Kohno, Ichiro Oda, Seiichiro Isobe, Tateo Suzuki, Kaoru Kohyama, et al. Three-dimensional probabilistic anatomical cranio-cerebral correlation via the international 10–20 system oriented for tran...

    2004

  96. [105]

    Electroencephalography: basic principles, clinical applications, and related fields

    J Craig Henry. Electroencephalography: basic principles, clinical applications, and related fields. Neurol- ogy, 67(11):2092–2092, 2006. 15

    2006

  97. [106]

    The american academy of sleep medicine manual for the scoring of sleep and associated events

    C Iber, S Ancoli-Israel, C Chesson, and S Quan. The american academy of sleep medicine manual for the scoring of sleep and associated events. Westchester, IL: American Academy of Sleep Medicine, 2007

    2007

  98. [107]

    Automatic sleep staging of eeg signals: recent development, challenges, and future directions

    Huy Phan and Kaare Mikkelsen. Automatic sleep staging of eeg signals: recent development, challenges, and future directions. Physiological Measurement, 43(4):04TR01, 2022

    2022

  99. [108]

    NeuroLM: A universal multi-task foun- dation model for bridging the gap between language and EEG signals

    Weibang Jiang, Yansen Wang, Bao liang Lu, and Dongsheng Li. NeuroLM: A universal multi-task foun- dation model for bridging the gap between language and EEG signals. In The International Conference on Learning Representations, 2025

    2025

  100. [109]

    Scaling laws for neural language models

    Jared Kaplan, Sam McCandlish, Tom Henighan, Tom B Brown, Benjamin Chess, Rewon Child, Scott Gray, Alec Radford, Jeffrey Wu, and Dario Amodei. Scaling laws for neural language models. arXiv preprint arXiv:2001.08361, 2020

    2001 arXiv

  101. [110]

    hello,” “help me,

    Laurens Van der Maaten and Geoffrey Hinton. Visualizing data using t-sne. Journal of machine learning research, 9(11), 2008. 16 A Related Work Electroencephalography (EEG) provides a direct, non-invasive window into human brain activity, and plays a pivotal role in cognitive s...

    2008

Pith tools

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