Pith. sign in

REVIEW 3 major objections 5 minor 37 references

Frequency-Decorrelated Temporal Ensembles for EEG--fNIRS Imagined-Handwriting Decoding

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

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

Pith's one-line read The paper establishes that frequency-diverse EEG ensembles and a protocol-matched block decoder, rather than the fNIRS stream, drive imagined-handwriting decoding to 79.5 percent overall accuracy (0.7718 on the private split).

desk verdict A well-engineered, honestly reported BCI challenge paper whose frequency-decorrelation result is real, but whose largest gain depends on an unverified test-set block-balance assumption. read the letter →

arxiv 2608.03176 v1 pith:OZ4W7TB4 submitted 2026-08-04 cs.CV cs.HC

Xiao Fan , Hongbin Guo , Yubo Han , Yi Zhang This is my paper · ORCID
classification cs.CVcs.HC
keywords brain–computerinterfaceimaginedhandwritingEEG–fNIRSfrequency-decorrelatedensemblemulti-scaletemporalnetworkHungarianassignmentblockdecodingsubject-independent
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

The paper tries to establish that sparse-montage EEG can decode four imagined Chinese characters in unseen participants when two task-grounded choices are made. First, a compact multi-scale temporal network trained separately on three EEG frequency views (broadband 0.5–40 Hz, 4–38 Hz, and 8–30 Hz) forms an ensemble whose members make complementary errors: error correlation is 0.29 across bands versus 0.82 across same-band seed replicas. Second, because every 12-trial acquisition block is known to contain exactly three trials of each class, the aggregated posteriors are decoded by a quota-constrained Hungarian assignment rather than trial-wise argmax; on one fixed posterior pool this raises overall accuracy from 0.7600 to 0.7952. The complete submitted system reaches 0.8498/0.7718/0.7952 on public/private/overall partitions, placing fourth on the private ranking. A modality audit finds fNIRS-only decoding at chance (0.2511) and a fusion gain of only +0.0025, and the authors report the clean ensemble separately from the transductive and protocol-constrained stages so the final score is not mistaken for a standalone trial classifier.

What carries the argument

Two mechanisms carry the argument. (1) Frequency-decorrelated ensemble: one compact multi-scale temporal network (≈4.2M parameters; five parallel temporal kernels of lengths 31, 21, 15, 11, 7 with attention over scales; no global temporal pooling) is trained independently on three FIR-filtered EEG views — broadband 0.5–40 Hz, 4–38 Hz, 8–30 Hz mu–beta — with a low-capacity fNIRS branch; posteriors are combined by geometric mean. The key fact: across-band members have error correlation 0.29, versus 0.82 for same-band seed replicas. (2) Paradigm-aware block decoding: each 36-trial session is split by its two longest rest gaps into three 12-trial blocks, and since every complete block holds exac

What would settle it

Permute the true labels within each 12-trial block while keeping the posteriors and the [3,3,3,3] quota decoder fixed: if block-constrained decoding still improves over argmax when the quota no longer matches the true composition, the reported +0.035 gain is not evidence of protocol matching. A direct counterpart is to compare the timestamp-derived block boundaries against the experimenter's logged boundaries session by session; any segmentation error bounds the real gain below +0.035.

Watch

Extended reading notes

Core claim

FRED's central claim: for EEG decoding, ensemble diversity is tied to the frequency view, and the balanced acquisition protocol can be turned into an exact decoding constraint. The same multi-scale temporal network, trained on three EEG frequency views with three seeds each, forms a nine-member ensemble scoring 0.8076/0.7242/0.7492 (public/private/overall) with no test-set adaptation; cross-band error correlation is 0.29 versus 0.82 for same-band replicas. Pseudo-label training, Conformer members, and block-quota decoding bring the submitted system to 0.7952 overall; block decoding alone adds 0.035 over argmax on fixed posteriors. fNIRS-only decoding is at chance (0.2511) and fusion adds onl

Load-bearing premise

The block-decoding gain rests on the assumption that each test session can be segmented into 12-trial blocks that each truly contain exactly three trials of every class; if the two-longest-rest-gap segmentation misplaces a boundary, or the hidden cohort violates the balanced protocol, the quota decoder is applied where its prior does not hold.

Editorial extensions

If this is right

  • A purely EEG, subject-independent imagined-handwriting decoder can exceed 0.80 public accuracy with no test-set adaptation if within-trial temporal structure and frequency views are preserved.
  • Ensemble design for EEG should sample frequency views, not just seeds: same-architecture members on different bands improve accuracy by 3.5–4.3 percentage points at equal ensemble size.
  • Known balanced acquisition protocols can be exploited as exact constraints: block-quota Hungarian decoding corrects 235 predictions and breaks 121 relative to argmax on the same posteriors.
  • In this sparse four-channel montage, fNIRS contributes no measurable trial-level information, so hybrid EEG–fNIRS gain is montage- and protocol-dependent rather than automatic.
  • Per-participant accuracy varies widely (0.534–0.926 in the clean system), and protocol constraints refine informative posteriors but cannot compensate for weak trial-level representations.

Reading between the lines

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

  • If the error-correlation result generalizes, frequency-view sampling could serve as a general recipe for EEG ensembles beyond handwriting, such as fine-grained motor-imagery or speech-imagery decoding, where band-split features already exist but are rarely quantified as error decorrelation.
  • The block-decoding gain suggests any balanced-trial BCI protocol carries exploitable structure; a testable extension is to check whether the same Hungarian quota decoder transfers to other balanced-by-design challenge datasets, where the gain should reappear when posteriors are informative but locally imbalanced.
  • Because the clean ensemble is the only fully subject-independent stage, the 0.046 overall gap between clean (0.7492) and submitted (0.7952) systems is not fully attributed; the paper's label-free stopping rule for pseudo-label rounds (histogram drift, prediction agreement) is a proxy, not a certified selection, so the true contribution of transductive adaptation remains an open measurement.
  • The fNIRS-at-chance result is specific to four optical channels at two locations; it does not contradict hybrid gains reported with denser optode coverage, and a testable extension would be a montage-density sweep to identify the coverage level at which the hemodynamic stream starts to add trial-level information.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

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. The paper presents FRED, a system for the four-class EEG–fNIRS imagined-handwriting challenge. A compact multi-scale temporal network is trained on three FIR-filtered EEG views (broadband, 4–38 Hz, 8–30 Hz), with three seeds per view; the nine-member clean ensemble achieves 0.8076/0.7242/0.7492 on public/private/overall test partitions without test-set adaptation or output constraints. The submitted pipeline adds pseudo-label training, three EEG-Conformer members, posterior aggregation, and a paradigm-aware decoder that enforces a [3,3,3,3] class quota on each reconstructed 12-trial block via Hungarian assignment, reaching 0.8498/0.7718/0.7952 and fourth place on the private split. Controlled analyses on a fixed posterior pool show session-level and block-level constraints add +0.0158 and +0.0352 overall, and a frequency-diversity analysis finds cross-band members have error correlation 0.29 versus 0.82 for same-band seed replicas. An fNIRS audit reports chance-level performance and negligible fusion gain. The paper is carefully structured into clean, transductive, and submitted levels, with participant-bootstrap CIs for the main controlled comparisons and a clear separation of post-competition analyses.

Significance. If its assumptions hold, the paper makes a useful practical and methodological contribution: it cleanly separates trial-level, transductive, and protocol-aware components; provides controlled comparisons on a fixed posterior pool with participant-bootstrap CIs; and gives a falsifiable negative result for fNIRS in this sparse montage. The code-release commitment and explicit post-competition audit are strengths. The main scientific claims—frequency diversity in EEG ensembles and protocol-matched structured inference—are relevant to BCI challenges. However, the headline gain from block decoding rests on an unverified structural assumption about the test cohort, which as written tempers the significance of the submitted result.

major comments (3)
  1. [§3.5 / Eq. (10) / abstract / Table 4] The block decoder enforces a strict [3,3,3,3] quota on every reconstructed 12-trial test block. The paper verifies this balance for all 536 complete training blocks, but provides no analogous verification for the 270 reconstructed test blocks. Reconstruction from 'the two longest rest gaps' is plausible, but extra breaks, missing trials, or incomplete sessions could yield mis-segmented blocks whose class counts are not balanced; the enumerated feasible-count handling for incomplete blocks does not by itself validate the class distribution. Because block decoding is the largest single improvement (+0.024 overall in Table 1; +0.0352 over argmax on a fixed posterior in Table 4), and the abstract's headline claim is 'because every 12-trial randomization block contains three instances of each class,' this unverified assumption is load-bearing. The authors now have access to test labels; pleas
  2. [§3.4] The pseudo-label round selection uses label-free diagnostic (i): the predicted class histogram 'should remain close to the balanced protocol.' This diagnostic presumes the same test-block balance that the block decoder requires, so it cannot independently confirm the test protocol. If test blocks are unbalanced, the pseudo-label round selection and the final quota decoder are affected together. Please separate verification of the protocol assumption from the pseudo-label diagnostics, or state explicitly that the round selection relies on the same unverified assumption.
  3. [§3.4 / Table 1] The submitted result combines pseudo-label-trained temporal members and three EEG-Conformer members, but the manuscript does not provide sufficient detail to reproduce these components: number of Conformer layers, heads, tokenization, dropout, and the exact pseudo-label refresh procedure (e.g., whether confidence is applied to the ensemble posterior or to individual members, and how many trials are retained per round). This is not central to the controlled claims, but it is necessary for the reproducibility promised by the code release.
minor comments (5)
  1. [Table 3] The error-correlation point estimates 0.82 and 0.29 lack uncertainty intervals. A participant-bootstrap interval or a per-pair range would strengthen the frequency-diversity claim.
  2. [Tables 1 and 4] Headline accuracies are point estimates; adding participant-bootstrap CIs for the clean and submitted overall accuracies would help calibrate the comparisons.
  3. [§3.1 / §3.5] The 'two longest rest gaps' criterion is not quantitative. Please specify the minimum gap duration, percentile, or threshold used so the block reconstruction is reproducible.
  4. [§3.2 / §4.3] The choices of cosine temperature τ=10, contrastive weight λ=1.0, noise σ=0.1, and kernel lengths are stated without sensitivity analysis; one sentence on their robustness or development-fold behavior would be useful.
  5. [§2 / §3.4] The Conformer members are introduced only by reference ([32]); since they contribute to the submitted result, a short architectural description would improve self-containedness.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central results are evaluated on held-out labels and the block quota is an external protocol constraint, not a fitted parameter.

full rationale

The paper's derivation chain is self-contained against external benchmarks. The clean ensemble is trained only on the 20 labelled participants and evaluated on held-out test labels; the transductive extension uses unlabelled test inputs and is reported separately from the clean result. Pseudo-label training is self-training, not label-based fitting, and round selection uses label-free diagnostics. The block decoder in Sec. 3.5 maximizes the aggregated posterior subject to the known 3/3/3/3 class quota; the quota is an external acquisition-protocol constraint verified on all 536 complete training blocks, not a quantity fitted to test labels. Table 4's decoding comparison reuses one fixed posterior pool, so the differences are purely decision-rule effects. The frequency-diversity analysis compares equal-size pools on held-out labels and reports bootstrap CIs; the error-correlation numbers are measurements, not definitions. There are no load-bearing self-citations: references to the authors' own prior work are absent, and standard methods (Hungarian algorithm, EEG-Conformer, SupCon, pseudo-label) are cited to external sources. Two limitations are flagged but non-circular: (i) the test cohort's block balance is asserted from the protocol and verified only for training blocks (Sec. 3.1/3.5), which could affect the block-decoder gain if test blocks are mis-segmented or unbalanced; (ii) the challenge reference [6] is marked 'Forthcoming; verify final bibliographic metadata before submission', leaving the benchmark definition not fully citable. Neither limitation makes any reported 'prediction' equivalent to an input by construction.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The central claims rest on a handful of hand-chosen bands and thresholds, plus domain assumptions about the acquisition protocol and about pseudo-label informativeness. No fitted constants are used to fabricate the block decoder, which is an external protocol constraint.

free parameters (7)
  • EEG frequency band edges = {0.5-40, 4-38, 8-30} Hz
    Chosen by hand from physiological priors; the central 'frequency-decorrelated' claim rests on these specific bands producing complementary errors.
  • Cosine softmax temperature tau = 10
    Chosen by hand in Eq. (6); sharpens the posterior used in ensemble aggregation and block decoding.
  • Contrastive loss weight lambda = 1.0
    Chosen in Eq. (7); balances cross-entropy and supervised contrastive term during training.
  • Gaussian augmentation noise sigma = 0.1
    Chosen for training only; compensates for the small labelled set (6,443 trials).
  • Pseudo-label confidence threshold = 0.90
    Chosen by hand in Sec. 3.4; controls which test trials enter transductive retraining.
  • Pseudo-label round count = 3
    Selected by monitoring label-free test diagnostics (histogram drift and prediction agreement), not by true labels; affects full-system accuracy.
  • Multi-scale kernel lengths and channel count = k={31,21,15,11,7}, C0=32
    Architecture hyperparameters chosen during development on subject-disjoint folds.
assumptions (5)
  • domain assumption Every complete 12-trial randomization block in the challenge contains exactly three trials of each of the four classes.
    Invoked in Sec. 3.5 to formulate the Hungarian assignment (Eq. 10); the paper verifies it for 536 training blocks and assumes it for test blocks.
  • domain assumption The two longest within-session rest gaps segment each session into the three 12-trial blocks, for both training and test cohorts.
    Used in Sec. 3.5 to reconstruct block boundaries from timestamps; mis-segmentation would misapply the quota decoder.
  • domain assumption The three frequency views (0.5-40, 4-38, 8-30 Hz) capture complementary task-relevant sensorimotor activity.
    Motivates the ensemble design in Sec. 3.3; the observed decorrelation is measured post hoc on test labels.
  • domain assumption Supervised contrastive learning on g improves subject-independent representation.
    Used in the training objective, Eq. (7); no independent ablation isolates its contribution in the final pipeline.
  • domain assumption High-confidence pseudo-labels on the unlabeled test set improve test-cohort accuracy without label leakage.
    Central to the transductive extension in Sec. 3.4; relies on the model's own confidence being informative.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Frequency-Decorrelated Temporal Ensembles for EEG--fNIRS Imagined-Handwriting Decoding." pith.science (2026). https://pith.science/paper/OZ4W7TB4

@misc{pith2026260803176,
  author       = {Pith},
  title        = {Pith review of: Frequency-Decorrelated Temporal Ensembles for EEG--fNIRS Imagined-Handwriting Decoding},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OZ4W7TB4}},
  note         = {Machine review of arXiv:2608.03176}
}
read the original abstract

Imagined handwriting offers a temporally rich paradigm for non-invasive neural decoding, yet reliable recognition across unseen participants remains difficult because scalp EEG is noisy and internally generated stroke sequences vary across individuals. The Multimodal Brain-Computer Interface Grand Challenge provides synchronized EEG and fNIRS for four-class subject-independent handwriting-trajectory classification. We propose FRED, a task-adapted system that models imagined handwriting as a multi-second motor sequence and trains a compact multi-scale temporal network on three complementary EEG frequency views. With three seeds per view, cross-band members produce substantially less-correlated errors than same-band replicas, yielding a clean nine-member ensemble accuracy of 0.8076/0.7242/0.7492 on the public/private/overall test partitions without test-set adaptation or output constraints. The submitted pipeline further incorporates transductive pseudo-label training, three EEG-Conformer members, posterior aggregation, and a paradigm-aware decoder. Because every 12-trial randomization block contains three instances of each class, the final predictions are obtained by Hungarian assignment under the known block quota. On one fixed posterior pool, independent, session-constrained, and block-constrained decoding achieve 0.7600, 0.7758, and 0.7952 overall accuracy, respectively. The complete system reaches 0.8498/0.7718/0.7952, ranking fourth on the private split. A modality audit finds fNIRS-only decoding at chance (0.2511 overall), while adding fNIRS to EEG changes accuracy by only +0.0025. These results identify frequency-diverse temporal EEG modeling and protocol-matched structured inference as the principal sources of performance in this sparse-montage EEG--fNIRS setting. The source code is available at https://github.com/XiuFan719/EEG-fNIRS-fuse-method-for-MM-challenge.

Figures

Figures reproduced from arXiv: 2608.03176 by the authors.

Figure 1
Figure 1. Overview of FRED. Three EEG frequency views share the same temporal architecture design and are trained with [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Per-participant results on the true test labels. [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 27 canonical work pages

  1. [1]

    Sangtae Ahn and Sung Chan Jun. 2017. Multi-modal integration of EEG-fNIRS for brain–computer interfaces: a review.Frontiers in Human Neuroscience11 (2017), 503. doi:10.3389/fnhum.2017.00503

    work page arXiv 2017

  2. [2]

    Kai Keng Ang, Zheng Yang Chin, Haihong Zhang, and Cuntai Guan. 2012. Filter Bank Common Spatial Pattern Algorithm on BCI Competition IV Datasets 2a and 2b.Frontiers in Neuroscience6 (2012), 39. doi:10.3389/fnins.2012.00039

    arXiv 2012

  3. [3]

    Alexandre Barachant, Stéphane Bonnet, Marco Congedo, and Christian Jutten

    work page

  4. [4]

    Alessio Paolo Buccino, Hasan Onur Keles, and Ahmet Omurtag. 2016. Hybrid EEG-fNIRS asynchronous brain-computer interface for multiple motor tasks. PLOS ONE11, 1 (2016), e0146610

    work page 2016

  5. [5]

    Ujwal Chaudhary, Niels Birbaumer, and Ander Ramos-Murguialday. 2016. Brain– computer interfaces for communication and rehabilitation.Nature Reviews Neu- rology12, 9 (2016), 513–525

    work page 2016

  6. [6]

    Badong Chen, Ziyu Jia, Roger Mark, Jing Wang, Tianzi Jiang, Hengguan Huang, Xinliang Zhou, and Jingying Ma. 2026. Multimodal Brain–Computer Interface Grand Challenge: EEG–fNIRS-based Handwriting-Trajectory Classification. In Proceedings of the ACM International Conference on Multimedia. Forthcoming; verify final bibliographic metadata before submission

    work page 2026

  7. [7]

    Antonio Maria Chiarelli, Pierpaolo Croce, Arcangelo Merla, and Filippo Zap- pasodi. 2018. Deep learning for hybrid EEG-fNIRS brain–computer interface: application to motor imagery classification.Journal of Neural Engineering15, 3 (2018), 036028. doi:10.1088/1741-2552/aaaf82

    work page doi:10.1088/1741-2552/aaaf82 2018

  8. [8]

    Siamac Fazli, Jan Mehnert, Jens Steinbrink, Gabriel Curio, Arno Villringer, Klaus- Robert Müller, and Benjamin Blankertz. 2012. Enhanced performance by a hybrid NIRS–EEG brain computer interface.NeuroImage59, 1 (2012), 519–529. doi:10.1016/j.neuroimage.2011.07.084

    work page doi:10.1016/j.neuroimage.2011.07.084 2012

Show all 37 references
  1. [9]

    He He and Dongrui Wu. 2020. Transfer Learning for Brain–Computer Interfaces: A Euclidean Space Data Alignment Approach.IEEE Transactions on Biomedical Engineering67, 2 (2020), 399–410. doi:10.1109/TBME.2019.2913914

    2020

  2. [10]

    Vinay Jayaram, Morteza Alamgir, Yasemin Altun, Bernhard Schölkopf, and Moritz Grosse-Wentrup. 2016. Transfer learning in brain–computer interfaces.IEEE Computational Intelligence Magazine11, 1 (2016), 20–31

    2016

  3. [11]

    Prannay Khosla, Piotr Teterwak, Chen Wang, Aaron Sarna, Yonglong Tian, Phillip Isola, Aaron Maschinot, Ce Liu, and Dilip Krishnan. 2020. Supervised Contrastive Learning. InAdvances in Neural Information Processing Systems, Vol. 33. 18661– 18673

    2020

  4. [12]

    Harold W. Kuhn. 1955. The Hungarian Method for the Assignment Prob- lem.Naval Research Logistics Quarterly2, 1–2 (1955), 83–97. doi:10.1002/nav. 3800020109 FRED: Frequency-Decorrelated Temporal Ensembles for EEG–fNIRS Imagined-Handwriting Decoding ACM MM ’26, 2026, BCI Challenge

    1955 doi

  5. [13]

    Balaji Lakshminarayanan, Alexander Pritzel, and Charles Blundell. 2017. Sim- ple and Scalable Predictive Uncertainty Estimation Using Deep Ensembles. In Advances in Neural Information Processing Systems, Vol. 30

    2017

  6. [14]

    Lawhern, Amelia J

    Vernon J. Lawhern, Amelia J. Solon, Nicholas R. Waytowich, Stephen M. Gordon, Chou P. Hung, and Brent J. Lance. 2018. EEGNet: a compact convolutional neural network for EEG-based brain–computer interfaces.Journal of Neural Engineering 15, 5 (2018), 056013. doi:10.1088/1741-2552/aace8c

    2018 doi

  7. [15]

    Dong-Hyun Lee. 2013. Pseudo-label: The Simple and Efficient Semi-supervised Learning Method for Deep Neural Networks. InICML Workshop on Challenges in Representation Learning

    2013

  8. [16]

    Zhe Liu, Jared Shore, Miao Wang, Fengyu Yuan, Aaron Buss, and Xiaopeng Zhao. 2021. A review of the functional near-infrared spectroscopy and electroen- cephalography hybrid brain–computer interface.Biomedical Signal Processing and Control68 (2021), 102595

    2021

  9. [17]

    Fabien Lotte, Laurent Bougrain, Andrzej Cichocki, Maureen Clerc, Marco Con- gedo, Alain Rakotomamonjy, and Florian Yger. 2018. A review of classification algorithms for EEG-based brain–computer interfaces: a 10 year update.Journal of Neural Engineering15, 3 (2018), 031005. doi...

    2018 doi

  10. [18]

    Scott Makeig, Christian Kothe, Tim Mullen, Nima Bigdely-Shamlo, Zhilin Zhang, and Kenneth Kreutz-Delgado. 2012. Evolving signal processing for brain– computer interfaces.Proc. IEEE100, Special Centennial Issue (2012), 1567–1584

    2012

  11. [19]

    Lopes da Silva

    Gert Pfurtscheller and Fernando H. Lopes da Silva. 1999. Event-related EEG/MEG synchronization and desynchronization: basic principles.Clinical Neurophysiol- ogy110, 11 (1999), 1842–1857. doi:10.1016/S1388-2457(99)00141-8

    1999 doi

  12. [20]

    Gert Pfurtscheller and Christa Neuper. 2001. Motor imagery and direct brain– computer communication.Proc. IEEE89, 7 (2001), 1123–1134

    2001

  13. [21]

    Paola Pinti, Ilias Tachtsidis, Antonia Hamilton, Joy Hirsch, Clarisse Aichelburg, Sam Gilbert, and Paul W. Burgess. 2020. The present and future use of functional near-infrared spectroscopy (fNIRS) for cognitive neuroscience.Annals of the New York Academy of Sciences1464, 1 (2...

    2020

  14. [22]

    handwriting brain

    Stéphanie Planton, Mélanie Jucla, Franck-Emmanuel Roux, and Jean-François Démonet. 2013. The “handwriting brain”: A meta-analysis of neuroimaging studies of motor versus orthographic processes.Cortex49, 10 (2013), 2772–2787. doi:10.1016/j.cortex.2013.05.011

    2013 doi

  15. [23]

    Allison, Jing Jin, Yu Zhang, Xingyu Wang, Wei Li, and Andrzej Cichocki

    Zhaoyang Qiu, Brendan Z. Allison, Jing Jin, Yu Zhang, Xingyu Wang, Wei Li, and Andrzej Cichocki. 2016. Optimized motor imagery paradigm based on imagining Chinese characters writing movement. arXiv:1609.09578 [cs.HC]

    2016 arXiv

  16. [24]

    Srinivas Ravishankar, Nora Zajzon, and Virginia R. de Sa. 2025. Decoding Imag- ined Handwriting from EEG. arXiv:2503.11202 [cs.HC]

    2025 arXiv

  17. [25]

    Falk, and Jocelyn Faubert

    Yannick Roy, Hubert Banville, Isabela Albuquerque, Alexandre Gramfort, Tiago H. Falk, and Jocelyn Faubert. 2019. Deep learning-based electroencephalography analysis: a systematic review.Journal of Neural Engineering16, 5 (2019), 051001

    2019

  18. [26]

    Simanto Saha and Mathias Baumert. 2020. Intra- and inter-subject variability in EEG-based sensorimotor brain computer interface: a review.Frontiers in Computational Neuroscience13 (2020), 87

    2020

  19. [27]

    Robin Tibor Schirrmeister, Jost Tobias Springenberg, Lukas Dominique Josef Fiederer, Martin Glasstetter, Katharina Eggensperger, Michael Tangermann, Frank Hutter, Wolfram Burgard, and Tonio Ball. 2017. Deep learning with convolutional neural networks for EEG decoding and visua...

    2017 doi

  20. [28]

    Felix Scholkmann, Stefan Kleiser, Andreas Jaakko Metz, Raphael Zimmermann, Juan Mata Pavia, Ursula Wolf, and Martin Wolf. 2014. A review on continuous wave functional near-infrared spectroscopy and imaging instrumentation and methodology.NeuroImage85 (2014), 6–27

    2014

  21. [29]

    Jaeyoung Shin, Alexander von Lühmann, Do-Won Kim, Jan Mehnert, Han-Jeong Hwang, and Klaus-Robert Müller. 2017. Open access dataset for EEG+NIRS single-trial classification.IEEE Transactions on Neural Systems and Rehabilitation Engineering25, 10 (2017), 1735–1745

    2017

  22. [30]

    Jake Snell, Kevin Swersky, and Richard Zemel. 2017. Prototypical Networks for Few-shot Learning. InAdvances in Neural Information Processing Systems, Vol. 30

    2017

  23. [31]

    Cubuk, Alex Kurakin, Han Zhang, and Colin Raffel

    Kihyuk Sohn, David Berthelot, Chun-Liang Li, Zizhao Zhang, Nicholas Carlini, Ekin D. Cubuk, Alex Kurakin, Han Zhang, and Colin Raffel. 2020. FixMatch: Simplifying Semi-supervised Learning with Consistency and Confidence. In Advances in Neural Information Processing Systems

    2020

  24. [32]

    Yonghao Song, Qingqing Zheng, Bingchuan Liu, and Xiaorong Gao. 2023. EEG Conformer: Convolutional Transformer for EEG Decoding and Visualization. IEEE Transactions on Neural Systems and Rehabilitation Engineering31 (2023), 710–719. doi:10.1109/TNSRE.2022.3230250

    2023

  25. [33]

    Willett, Donald T

    Francis R. Willett, Donald T. Avansino, Leigh R. Hochberg, Jaimie M. Henderson, and Krishna V. Shenoy. 2021. High-performance brain-to-text communication via handwriting.Nature593, 7858 (2021), 249–254. doi:10.1038/s41586-021-03506-2

    2021 doi

  26. [34]

    Willett, Erin M

    Francis R. Willett, Erin M. Kunz, Chaofei Fan, Donald T. Avansino, Guy H. Wilson, Eun Young Choi, Foram Kamdar, Matthew F. Glasser, Leigh R. Hochberg, Shaul Druckmann, et al. 2023. A high-performance speech neuroprosthesis.Nature 620, 7976 (2023), 1031–1036

    2023

  27. [35]

    Wolpaw, Niels Birbaumer, Dennis J

    Jonathan R. Wolpaw, Niels Birbaumer, Dennis J. McFarland, Gert Pfurtscheller, and Theresa M. Vaughan. 2002. Brain–computer interfaces for communication and control.Clinical Neurophysiology113, 6 (2002), 767–791

    2002

  28. [36]

    Yücel, Alexander von Lühmann, Felix Scholkmann, Judit Gervain, Ippeita Dan, Hasan Ayaz, David A

    Meryem A. Yücel, Alexander von Lühmann, Felix Scholkmann, Judit Gervain, Ippeita Dan, Hasan Ayaz, David A. Boas, Robert J. Cooper, Joseph Culver, Clare E. Elwell, et al. 2021. Best practices for fNIRS publications.Neurophotonics8, 1 (2021), 012101

    2021

  29. [2012]

    IEEE Transactions on Biomedical Engineering59, 4 (2012), 920–928

    Multiclass brain–computer interface classification by Riemannian geometry. IEEE Transactions on Biomedical Engineering59, 4 (2012), 920–928

    2012

Pith tools

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