Aperiodic and Low-Frequency Spectral Bias in Reconstruction based EEG Foundation Models

Aditya Kommineni; Andreas Peter Juhl Hansen; Emily Zhou; Jeppe Roden M\"unster; Kleanthis Avramidis; Richard Leahy; Shrikanth Narayanan; Simon Bock Segaard; Takfarinas Medani; Tiantian Feng

arxiv: 2605.26434 · v1 · pith:DCY5VDLUnew · submitted 2026-05-26 · 💻 cs.LG · cs.AI

Aperiodic and Low-Frequency Spectral Bias in Reconstruction based EEG Foundation Models

Aditya Kommineni , Emily Zhou , Kleanthis Avramidis , Simon Bock Segaard , Jeppe Roden M\"unster , Andreas Peter Juhl Hansen , Takfarinas Medani , Tiantian Feng

show 2 more authors

Richard Leahy Shrikanth Narayanan

This is my paper

Pith reviewed 2026-06-29 18:55 UTC · model grok-4.3

classification 💻 cs.LG cs.AI

keywords EEG foundation modelsspectral biasaperiodic componentsoscillatory componentsreconstruction objectivesbrain-computer interfacesembeddingspretext tasks

0 comments

The pith

Reconstruction-based EEG foundation models capture aperiodic signal components while under-representing high-frequency oscillatory ones.

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

The paper establishes that pretraining EEG foundation models via reconstruction on large unlabeled datasets creates a systematic bias: the learned embeddings favor the high-power aperiodic parts of the signal and under-represent the low-power oscillatory components, especially at higher frequencies. This mismatch arises because reconstruction objectives align more naturally with aperiodic structure than with the oscillatory elements that carry much task-relevant information in EEG. Linear probe tests on real BCI datasets show the embeddings encode subject identity more strongly than task labels, which explains the observed underperformance relative to smaller supervised models in low-resource settings. A reader would care because the bias points to a concrete, addressable limitation in current EEG foundation model design.

Core claim

Using controlled synthetic EEG inputs that separate aperiodic and oscillatory components, reconstruction-based EEG foundation model embeddings are shown to preferentially encode aperiodic structure while under-representing oscillatory activity, with the effect strongest at higher frequencies. On real-world BCI datasets, linear probes confirm that these embeddings represent subject identity more strongly than task-relevant features, thereby reinforcing the low-frequency and aperiodic bias induced by the reconstruction objective.

What carries the argument

The reconstruction pretext task, which aligns embeddings with high-power aperiodic EEG components at the expense of low-power oscillatory ones.

If this is right

Embeddings will show weaker performance on downstream tasks that depend on high-frequency oscillatory content.
Linear probes will continue to recover subject identity more readily than task labels across multiple BCI datasets.
The performance gap versus fully supervised models will remain largest in low-resource regimes where fine-tuning cannot overcome the spectral bias.
Adding explicit auxiliary losses for high-frequency oscillatory structure during pretraining would reduce the mismatch.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The same reconstruction-induced bias could appear in foundation models trained on other biosignals that exhibit strong aperiodic backgrounds.
Pretraining objectives that directly penalize loss of oscillatory power, such as frequency-specific contrastive terms, offer a testable route to more balanced representations.
Subject-identity dominance in embeddings suggests that current models may require explicit disentanglement steps before they can generalize across individuals.

Load-bearing premise

Synthetic EEG signals accurately reproduce the spectral decomposition and statistical properties of real EEG without introducing artifacts that exaggerate the observed bias.

What would settle it

An experiment that measures the power spectrum of signals reconstructed from the model embeddings on synthetic inputs with isolated high-frequency oscillations and finds equal or stronger representation of those oscillations compared with aperiodic components.

Figures

Figures reproduced from arXiv: 2605.26434 by Aditya Kommineni, Andreas Peter Juhl Hansen, Emily Zhou, Jeppe Roden M\"unster, Kleanthis Avramidis, Richard Leahy, Shrikanth Narayanan, Simon Bock Segaard, Takfarinas Medani, Tiantian Feng.

**Figure 1.** Figure 1: Simulated and real-world EEG experiments (A) Pipeline for obtaining embeddings from multi-channel EEG signals. Embeddings are extracted from the last layer of the encoder for three foundation models (LaBraM, CBraMod and CSBrain) (B) Pipeline for computing linear decodability through controlled synthetic single channel EEG generation (In the figure, aperiodic exponent β is sampled between [θmin, θmax] to cr… view at source ↗

**Figure 2.** Figure 2: Linear decodability R2 values for Cz channel across three foundation models (CBraMod, CSBrain, LaBraM) for Aperiodic Exponent (β), Aperiodic Offset (Aap) and Oscillation frequency (fosc). The linear decodability is computed through synthetic single channel EEG with the respective variables varied across the ranges mentioned in [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

**Figure 3.** Figure 3: Linear decodability comparison of Cz channel across oscillatory frequencies ( [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

**Figure 4.** Figure 4: Euclidean distance for task and subject based clusters. As can be seen in the plots, for all [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗

**Figure 5.** Figure 5: t-SNE embeddings of pre-trained LaBraM, CBraMod, and CSBrain models across the BCIC IV 2a and Sleep-EDF datasets. t-SNE plots indicate that models generally learn representations that tend to cluster by subject identity rather than by task label. Embeddings from a maximum of 15 subjects are shown for clarity. model representations through t-SNE plots with for both subject and task labels indicate clusters … view at source ↗

**Figure 6.** Figure 6: (Top) Time series of generated EEG signal for [PITH_FULL_IMAGE:figures/full_fig_p015_6.png] view at source ↗

**Figure 7.** Figure 7: Linear decodability for Fz channel across three foundation models (CBraMod, CSBrain, [PITH_FULL_IMAGE:figures/full_fig_p015_7.png] view at source ↗

**Figure 8.** Figure 8: Linear decodability comparison of Fz channel across oscillatory frequencies ( [PITH_FULL_IMAGE:figures/full_fig_p016_8.png] view at source ↗

**Figure 9.** Figure 9: Linear decodability for Pz channel across three foundation models (CBraMod, CSBrain, [PITH_FULL_IMAGE:figures/full_fig_p016_9.png] view at source ↗

**Figure 10.** Figure 10: Linear decodability comparison of Pz channel across oscillatory frequencies ( [PITH_FULL_IMAGE:figures/full_fig_p017_10.png] view at source ↗

**Figure 11.** Figure 11: Linear decodability for Oz channel across three foundation models (CBraMod, CSBrain, [PITH_FULL_IMAGE:figures/full_fig_p017_11.png] view at source ↗

**Figure 12.** Figure 12: Linear decodability comparison of Oz channel across oscillatory frequencies ( [PITH_FULL_IMAGE:figures/full_fig_p018_12.png] view at source ↗

**Figure 13.** Figure 13: t-SNE embeddings of pre-trained LaBraM, CBraMod, and CSBrain models across the PhysioNet-MI and Kaggle-ERN datasets. 18 [PITH_FULL_IMAGE:figures/full_fig_p018_13.png] view at source ↗

read the original abstract

EEG foundation models, pre-trained on large-scale unlabelled EEG data, have emerged as a promising direction towards learning generalizable EEG representations. Despite showing positive results in data-rich regimes, they often fail to outperform significantly smaller supervised models in low-resource settings compared to fully supervised models. We provide a mechanistic account of this shortcoming, attributing it to a fundamental mismatch between reconstruction-based pretext tasks and the idiosyncratic spectral structure of EEG signals, which decompose into distinct high-power aperiodic and low-power oscillatory components. Using controlled, synthetically-generated EEG inputs, we demonstrate that EEG foundation model embeddings are biased to capture the aperiodic components of the EEG signal while under-representing oscillatory components, particularly at higher frequencies. Additionally, linear probe evaluations on real-world BCI datasets further reveal that embeddings encode subject identity more strongly than task-relevant information, thereby reinforcing the low-frequency and aperiodic component bias in foundation model embeddings trained primarily on reconstruction based objectives. Together, these findings elucidate a failure mode in reconstruction based EEG foundation models and motivate future work to incorporate auxiliary losses explicitly targeting high-frequency oscillatory structure as a path toward more capable and generalizable EEG representations.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper gives a concrete mechanistic story for why reconstruction-based EEG foundation models underperform in low-resource settings, but the synthetic data controls need more scrutiny to pin the bias on the objective itself.

read the letter

The main thing here is that reconstruction pretraining on EEG pushes embeddings to over-weight aperiodic and low-frequency content while under-representing higher-frequency oscillations, and that this shows up as stronger subject-identity encoding than task information in linear probes.

What is new is the direct link from the reconstruction loss to that spectral mismatch, backed by controlled synthetic inputs plus real BCI probe results. The abstract frames this as a fixable limitation rather than a generic complaint about foundation models, and it points to auxiliary losses for oscillatory structure as a next step. That diagnosis is more specific than most prior work on EEG pretraining failures.

The experiments sound like a reasonable way to isolate the effect, and the motivation for better low-resource performance is practical. If the synthetic generation actually reproduces the joint statistics of real EEG, the causal attribution to the pretext task would hold up.

The soft spot is exactly the one in the stress-test note. Without details on how the synthetic signals were built—whether they preserve non-stationarity, phase-amplitude coupling, or channel correlations—the observed bias could partly reflect the input construction rather than the loss. The abstract does not report effect sizes or statistical tests, so it is hard to gauge how large or consistent the mismatch is. That makes the central claim provisional until the methods section is checked.

This is for people working on EEG foundation models or BCI transfer who need a handle on why current pretraining falls short. It is worth sending to referees because the question is well-posed and the suggested direction is concrete, even if the current evidence is thin on the synthetic controls.

Referee Report

2 major / 1 minor

Summary. The paper claims that reconstruction-based pretraining in EEG foundation models induces a spectral bias favoring high-power aperiodic components and low-frequency content while under-representing low-power high-frequency oscillations. This is shown via controlled synthetic EEG inputs that isolate the effect, plus linear probe experiments on real BCI datasets demonstrating stronger encoding of subject identity than task information; the authors conclude this mismatch explains poor low-resource performance and motivate auxiliary losses targeting oscillatory structure.

Significance. If the central attribution holds, the work supplies a concrete mechanistic account of why reconstruction objectives are mismatched to EEG statistics and supplies an actionable path (auxiliary losses) for improving foundation-model pretraining. The controlled synthetic setup, if shown to preserve real EEG joint statistics, would be a strength for causal isolation; the linear-probe results on subject vs. task encoding would further ground the practical relevance.

major comments (2)

[Experiments / synthetic EEG generation] Synthetic EEG generation subsection (Experiments section): the central claim that the observed aperiodic bias is caused by the reconstruction objective requires that the synthetic inputs reproduce the joint spectral statistics of real EEG (non-stationarity, phase-amplitude coupling, channel correlations). The manuscript must supply the precise generation procedure and quantitative comparisons (e.g., PSD, coherence, PAC metrics) between synthetic and real data; absent these controls, the mismatch could originate in the synthetic construction itself rather than the pretext task.
[Results / linear probe evaluation] Linear probe evaluation paragraph (Results section): the claim that embeddings encode subject identity more strongly than task-relevant information is load-bearing for the practical implication. The manuscript should report the exact probe accuracies, the number of subjects/tasks, cross-validation scheme, and statistical comparison (e.g., paired t-test or effect size) between subject and task probes; without these numbers the strength of the bias remains unquantified.

minor comments (1)

[Abstract] Abstract, final sentence: the phrasing 'thereby reinforcing the low-frequency and aperiodic component bias' is circular; reword to separate the empirical observation from the interpretive claim.

Simulated Author's Rebuttal

2 responses · 0 unresolved

Thank you for the constructive feedback on our manuscript. We address each major comment below and will incorporate revisions to strengthen the work.

read point-by-point responses

Referee: [Experiments / synthetic EEG generation] Synthetic EEG generation subsection (Experiments section): the central claim that the observed aperiodic bias is caused by the reconstruction objective requires that the synthetic inputs reproduce the joint spectral statistics of real EEG (non-stationarity, phase-amplitude coupling, channel correlations). The manuscript must supply the precise generation procedure and quantitative comparisons (e.g., PSD, coherence, PAC metrics) between synthetic and real data; absent these controls, the mismatch could originate in the synthetic construction itself rather than the pretext task.

Authors: We agree that quantitative validation of the synthetic data against real EEG is necessary to attribute the observed bias specifically to the reconstruction objective. In the revised manuscript, we will expand the synthetic EEG generation subsection to provide the precise generation procedure along with direct comparisons of power spectral density (PSD), coherence, and phase-amplitude coupling (PAC) metrics between synthetic and real datasets. This addition will address the concern and reinforce the causal interpretation. revision: yes
Referee: [Results / linear probe evaluation] Linear probe evaluation paragraph (Results section): the claim that embeddings encode subject identity more strongly than task-relevant information is load-bearing for the practical implication. The manuscript should report the exact probe accuracies, the number of subjects/tasks, cross-validation scheme, and statistical comparison (e.g., paired t-test or effect size) between subject and task probes; without these numbers the strength of the bias remains unquantified.

Authors: We acknowledge that detailed quantitative reporting is required to substantiate the relative encoding of subject identity versus task information. In the revision, we will update the linear probe evaluation paragraph to include the exact probe accuracies, the number of subjects and tasks, the cross-validation scheme, and statistical comparisons (e.g., paired t-tests or effect sizes) between the subject and task probes. These additions will quantify the bias and support the practical implications. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical observations from synthetic and real-data experiments, no derivation chain

full rationale

The paper advances its central claims through controlled experiments on synthetically generated EEG signals and linear-probe evaluations on real BCI datasets. No mathematical derivation, uniqueness theorem, or first-principles result is presented that reduces to fitted parameters, self-definitions, or self-citations. The abstract and described methodology rely on direct measurement of spectral bias and subject/task encoding, which are falsifiable against external benchmarks rather than tautological. This matches the default expectation for non-circular empirical work.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the domain assumption that EEG can be cleanly decomposed into aperiodic and oscillatory parts and that synthetic signals preserve this decomposition; no free parameters or invented entities are introduced in the abstract.

axioms (1)

domain assumption EEG signals decompose into distinct high-power aperiodic and low-power oscillatory components
Invoked throughout the abstract as the basis for the bias diagnosis.

pith-pipeline@v0.9.1-grok · 5780 in / 1233 out tokens · 33304 ms · 2026-06-29T18:55:35.251577+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

15 extracted references · 9 canonical work pages · 2 internal anchors

[1]

GPT-4 Technical Report

Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, et al. Gpt-4 technical report. arXiv preprint arXiv:2303.08774,

work page internal anchor Pith review Pith/arXiv arXiv
[2]

Neural codecs as biosignal tokenizers.arXiv preprint arXiv:2510.09095,

Kleanthis Avramidis, Tiantian Feng, Woojae Jeong, Jihwan Lee, Wenhui Cui, Richard M Leahy, and Shrikanth Narayanan. Neural codecs as biosignal tokenizers.arXiv preprint arXiv:2510.09095,

work page arXiv
[3]

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, Sandhini Agarwal, Ariel Herbert-V oss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel Ziegler, Jeffrey Wu, Clemens Winter, Chris Hesse, Mark Chen, Eric Sigler, Mateusz Litwin, Scott G...

1901
[4]

Clemens Brunner, Robert Leeb, Gernot Müller-Putz, Alois Schlögl, and Gert Pfurtscheller

URL https://proceedings.neurips.cc/ paper_files/paper/2020/file/1457c0d6bfcb4967418bfb8ac142f64a-Paper.pdf. Clemens Brunner, Robert Leeb, Gernot Müller-Putz, Alois Schlögl, and Gert Pfurtscheller. Bci competition 2008–graz data set a.Institute for knowledge discovery (laboratory of brain-computer interfaces), Graz University of Technology, 16(1-6):1,

2020
[5]

10 Hsiang-Yun Sherry Chien, Hanlin Goh, Christopher Michael Sandino, and Joseph Yitan Cheng

URLhttps://openreview.net/forum?id=oUMiuYHW21. 10 Hsiang-Yun Sherry Chien, Hanlin Goh, Christopher Michael Sandino, and Joseph Yitan Cheng. MAEEG: Masked auto-encoder for EEG representation learning. InNeurIPS 2022 Workshop on Learning from Time Series for Health,

2022
[6]

Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities

URL https://openreview.net/forum?id= kttuLV59ZuJ. Gheorghe Comanici, Eric Bieber, Mike Schaekermann, Ice Pasupat, Noveen Sachdeva, Inderjit Dhillon, Marcel Blistein, Ori Ram, Dan Zhang, Evan Rosen, et al. Gemini 2.5: Pushing the frontier with advanced reasoning, multimodality, long context, and next generation agentic capabilities. arXiv preprint arXiv:25...

work page internal anchor Pith review Pith/arXiv arXiv
[7]

doi: https://doi.org/10.1016/j.neuroimage.2022.119034

ISSN 1053-8119. doi: https://doi.org/10.1016/j.neuroimage.2022.119034. URL https://www.sciencedirect.com/ science/article/pii/S105381192200163X. Weibang Jiang, Liming Zhao, and Bao liang Lu. Large brain model for learning generic representa- tions with tremendous EEG data in BCI. InThe Twelfth International Conference on Learning Representations,

work page doi:10.1016/j.neuroimage.2022.119034 2022
[8]

2024.10782310

doi: 10.1109/EMBC53108. 2024.10782310. 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,

work page doi:10.1109/embc53108 2024
[9]

Gayal Kuruppu, Neeraj Wagh, Vaclav Kremen, and Yogatheesan Varatharajah

doi: 10.1109/TIFS.2021.3067998. Gayal Kuruppu, Neeraj Wagh, Vaclav Kremen, and Yogatheesan Varatharajah. Eeg foundation models: a critical review of current progress and future directions.Journal of neural engineering, 23(2):021001,

work page doi:10.1109/tifs.2021.3067998 2021
[10]

EEG foundation models: Progresses, benchmarki ng, and open problems

Chenyu Liu, Yuqiu Deng, Tianyu Liu, Jinan Zhou, Xinliang Zhou, Ziyu Jia, and Yi Ding. ECHO: Toward contextual seq2seq paradigms in large EEG models. InThe F ourteenth International Conference on Learning Representations, 2026a. URL https://openreview.net/forum?id= ClLQ6cLkoR. 11 Dingkun Liu, Yuheng Chen, Zhu Chen, Zhenyao Cui, Yaozhi Wen, Jiayu An, Jingwe...

work page arXiv
[11]

Aditya Mishra, Ahnaf Mozib Samin, Ali Etemad, and Javad Hashemi

Kaggle. Aditya Mishra, Ahnaf Mozib Samin, Ali Etemad, and Javad Hashemi. Subject representation learning from eeg using graph convolutional variational autoencoders. InICASSP 2025-2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 1–5. IEEE,

2025
[12]

Schalk, D.J

G. Schalk, D.J. McFarland, T. Hinterberger, N. Birbaumer, and J.R. Wolpaw. Bci2000: a general- purpose brain-computer interface (bci) system.IEEE Transactions on Biomedical Engineering, 51 (6):1034–1043, 2004a. doi: 10.1109/TBME.2004.827072. Gerwin Schalk, Dennis J McFarland, Thilo Hinterberger, Niels Birbaumer, and Jonathan R Wol- paw. Bci2000: a general...

work page doi:10.1109/tbme.2004.827072 2004
[13]

Jiquan Wang, Sha Zhao, Yangxuan Zhou, Yiming Kang, Shijian Li, and Gang Pan

URL https://openreview.net/ forum?id=NPNUHgHF2w. Jiquan Wang, Sha Zhao, Yangxuan Zhou, Yiming Kang, Shijian Li, and Gang Pan. Deeperbrain: A neuro-grounded eeg foundation model towards universal bci.arXiv preprint arXiv:2601.06134,

work page arXiv
[14]

Liuyin Yang, Qiang Sun, Ang Li, and Marc M

URL https://proceedings.neurips.cc/paper_files/paper/2023/ file/f6b30f3e2dd9cb53bbf2024402d02295-Paper-Conference.pdf. Liuyin Yang, Qiang Sun, Ang Li, and Marc M. Van Hulle. Are EEG foundation models worth it? comparative evaluation with traditional decoders in diverse BCI tasks. InThe F ourteenth International Conference on Learning Representations,

2023
[15]

FOOOF python package was used for generating the frequency spectrum of single channel EEG [Donoghue et al., 2020]. Fig. 6 illustrates the examples generated by sweep of exponent values from 1.0 to 2.0. D Linear Decodability Results Additional linear decodability plots for channels across different EEG montage lobes, Frontal (Fz) [Fig 7, Fig 8], Parietal (...

2020

[1] [1]

GPT-4 Technical Report

Josh Achiam, Steven Adler, Sandhini Agarwal, Lama Ahmad, Ilge Akkaya, Florencia Leoni Aleman, Diogo Almeida, Janko Altenschmidt, Sam Altman, Shyamal Anadkat, et al. Gpt-4 technical report. arXiv preprint arXiv:2303.08774,

work page internal anchor Pith review Pith/arXiv arXiv

[2] [2]

Neural codecs as biosignal tokenizers.arXiv preprint arXiv:2510.09095,

Kleanthis Avramidis, Tiantian Feng, Woojae Jeong, Jihwan Lee, Wenhui Cui, Richard M Leahy, and Shrikanth Narayanan. Neural codecs as biosignal tokenizers.arXiv preprint arXiv:2510.09095,

work page arXiv

[3] [3]

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, Sandhini Agarwal, Ariel Herbert-V oss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel Ziegler, Jeffrey Wu, Clemens Winter, Chris Hesse, Mark Chen, Eric Sigler, Mateusz Litwin, Scott G...

1901

[4] [4]

Clemens Brunner, Robert Leeb, Gernot Müller-Putz, Alois Schlögl, and Gert Pfurtscheller

URL https://proceedings.neurips.cc/ paper_files/paper/2020/file/1457c0d6bfcb4967418bfb8ac142f64a-Paper.pdf. Clemens Brunner, Robert Leeb, Gernot Müller-Putz, Alois Schlögl, and Gert Pfurtscheller. Bci competition 2008–graz data set a.Institute for knowledge discovery (laboratory of brain-computer interfaces), Graz University of Technology, 16(1-6):1,

2020

[5] [5]

10 Hsiang-Yun Sherry Chien, Hanlin Goh, Christopher Michael Sandino, and Joseph Yitan Cheng

URLhttps://openreview.net/forum?id=oUMiuYHW21. 10 Hsiang-Yun Sherry Chien, Hanlin Goh, Christopher Michael Sandino, and Joseph Yitan Cheng. MAEEG: Masked auto-encoder for EEG representation learning. InNeurIPS 2022 Workshop on Learning from Time Series for Health,

2022

[6] [6]

Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities

URL https://openreview.net/forum?id= kttuLV59ZuJ. Gheorghe Comanici, Eric Bieber, Mike Schaekermann, Ice Pasupat, Noveen Sachdeva, Inderjit Dhillon, Marcel Blistein, Ori Ram, Dan Zhang, Evan Rosen, et al. Gemini 2.5: Pushing the frontier with advanced reasoning, multimodality, long context, and next generation agentic capabilities. arXiv preprint arXiv:25...

work page internal anchor Pith review Pith/arXiv arXiv

[7] [7]

doi: https://doi.org/10.1016/j.neuroimage.2022.119034

ISSN 1053-8119. doi: https://doi.org/10.1016/j.neuroimage.2022.119034. URL https://www.sciencedirect.com/ science/article/pii/S105381192200163X. Weibang Jiang, Liming Zhao, and Bao liang Lu. Large brain model for learning generic representa- tions with tremendous EEG data in BCI. InThe Twelfth International Conference on Learning Representations,

work page doi:10.1016/j.neuroimage.2022.119034 2022

[8] [8]

2024.10782310

doi: 10.1109/EMBC53108. 2024.10782310. 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,

work page doi:10.1109/embc53108 2024

[9] [9]

Gayal Kuruppu, Neeraj Wagh, Vaclav Kremen, and Yogatheesan Varatharajah

doi: 10.1109/TIFS.2021.3067998. Gayal Kuruppu, Neeraj Wagh, Vaclav Kremen, and Yogatheesan Varatharajah. Eeg foundation models: a critical review of current progress and future directions.Journal of neural engineering, 23(2):021001,

work page doi:10.1109/tifs.2021.3067998 2021

[10] [10]

EEG foundation models: Progresses, benchmarki ng, and open problems

Chenyu Liu, Yuqiu Deng, Tianyu Liu, Jinan Zhou, Xinliang Zhou, Ziyu Jia, and Yi Ding. ECHO: Toward contextual seq2seq paradigms in large EEG models. InThe F ourteenth International Conference on Learning Representations, 2026a. URL https://openreview.net/forum?id= ClLQ6cLkoR. 11 Dingkun Liu, Yuheng Chen, Zhu Chen, Zhenyao Cui, Yaozhi Wen, Jiayu An, Jingwe...

work page arXiv

[11] [11]

Aditya Mishra, Ahnaf Mozib Samin, Ali Etemad, and Javad Hashemi

Kaggle. Aditya Mishra, Ahnaf Mozib Samin, Ali Etemad, and Javad Hashemi. Subject representation learning from eeg using graph convolutional variational autoencoders. InICASSP 2025-2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 1–5. IEEE,

2025

[12] [12]

Schalk, D.J

G. Schalk, D.J. McFarland, T. Hinterberger, N. Birbaumer, and J.R. Wolpaw. Bci2000: a general- purpose brain-computer interface (bci) system.IEEE Transactions on Biomedical Engineering, 51 (6):1034–1043, 2004a. doi: 10.1109/TBME.2004.827072. Gerwin Schalk, Dennis J McFarland, Thilo Hinterberger, Niels Birbaumer, and Jonathan R Wol- paw. Bci2000: a general...

work page doi:10.1109/tbme.2004.827072 2004

[13] [13]

Jiquan Wang, Sha Zhao, Yangxuan Zhou, Yiming Kang, Shijian Li, and Gang Pan

URL https://openreview.net/ forum?id=NPNUHgHF2w. Jiquan Wang, Sha Zhao, Yangxuan Zhou, Yiming Kang, Shijian Li, and Gang Pan. Deeperbrain: A neuro-grounded eeg foundation model towards universal bci.arXiv preprint arXiv:2601.06134,

work page arXiv

[14] [14]

Liuyin Yang, Qiang Sun, Ang Li, and Marc M

URL https://proceedings.neurips.cc/paper_files/paper/2023/ file/f6b30f3e2dd9cb53bbf2024402d02295-Paper-Conference.pdf. Liuyin Yang, Qiang Sun, Ang Li, and Marc M. Van Hulle. Are EEG foundation models worth it? comparative evaluation with traditional decoders in diverse BCI tasks. InThe F ourteenth International Conference on Learning Representations,

2023

[15] [15]

FOOOF python package was used for generating the frequency spectrum of single channel EEG [Donoghue et al., 2020]. Fig. 6 illustrates the examples generated by sweep of exponent values from 1.0 to 2.0. D Linear Decodability Results Additional linear decodability plots for channels across different EEG montage lobes, Frontal (Fz) [Fig 7, Fig 8], Parietal (...

2020