DLGE: Dual Local-Global Encoding for Generalizable Cross-BCI-Paradigm

Jingyuan Wang; Junhua Li

REVIEW 3 major objections 4 minor 43 references

DLGE: Dual Local-Global Encoding for Generalizable Cross-BCI-Paradigm

T0 review · 3 major / 4 minor · reviewed 2026-08-05 · deepseek-v4-flash

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

Pith's one-line read This paper claims that a dual local-global encoder, with anatomical brain-region padding, lets one EEG model classify three BCI paradigms without retraining.

desk verdict A reasonable cross-paradigm architecture, but the evaluation never holds out a paradigm, so the headline claim is untested. read the letter →

arxiv 2509.07991 v1 pith:7AA2JZYE submitted 2025-08-25 q-bio.NC cs.AIcs.LG

Jingyuan Wang , Junhua Li This is my paper · ORCID

classification q-bio.NCcs.AIcs.LG

keywords cross-BCI-paradigmclassificationEEGdecodingbrain-regionpartitioningzero-paddingchannelstandardizationduallocal-globalencodermotorimageryrestingstatedrivingfatigue

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 show that one deep-learning model can classify EEG from different BCI paradigms—motor imagery, resting state, and driving fatigue—even though these paradigms were recorded with different channel montages and involve different mental tasks. To bridge the channel mismatch, the model groups electrodes into 11 anatomical brain regions and zero-pads each region to a fixed 9-channel layout. A local brain encoder learns features shared across paradigms within each region, and a global brain encoder combines them into task-specific representations. On a pooled, shuffled test set the authors report macro precision 60.16%, recall 59.88%, and F1 59.56%. If this holds, it would mean a single trained model could be reused across BCI paradigms without retraining or re-tuning.

What carries the argument

The key object is the anatomical brain-region standardization plus the dual encoder stack. First, each sample's EEG channels are grouped into 11 fixed brain regions, and each region is padded to exactly 9 channels, with padded channels masked in later attention. The local brain encoder treats each channel's time-frequency matrix as a token sequence, applies a single-head transformer per channel as a channel embedder, computes spatial attention between channels inside a region, and condenses valid channels into one region embedding. The global brain encoder then applies multi-head self-attention over the 11 region embeddings, with padded regions masked by setting their attention scores to neg

What would settle it

Train DLGE on motor imagery and resting-state data only, then test it on all driving-fatigue data without any fatigue samples in the training folds. If the held-out-paradigm F1 falls to near chance, the central claim fails. As a second check, compare the current shuffled-segment result with a subject-disjoint split; a large drop would indicate that the model exploited within-recording segment similarity.

Watch

Extended reading notes

Core claim

The central claim is that DLGE, a dual local-global encoder, achieves cross-BCI-paradigm classification without retraining. Channel heterogeneity is handled upstream of learning: electrodes from any montage are mapped to the same 11 brain regions, and missing positions are zero-padded, so the transformer-based feature extractor always sees a standardized regional input. Shared regional features are extracted by the local encoder using temporal attention per channel and spatial attention across channels; the global encoder then merges the 11 regional summaries into the high-level representation used by the classifier. Evaluated on the three paradigms together, the model reaches an average mac

Load-bearing premise

The load-bearing premise is that random 5-fold cross-validation on shuffled 3-second segments pooled from all three paradigms measures cross-paradigm generalization; because adjacent segments from the same recording are highly similar and no paradigm is ever held out, this premise is what must hold for the reported numbers to support the claim.

Editorial extensions

If this is right

A model trained once on several paradigms could accept EEG from a new device with a different channel montage, if its channels can be mapped into the 11 brain regions.
Cross-paradigm decoding would remove the need to store and maintain separate models for motor imagery, vigilance, and resting-state applications.
The small optimal depth (two attention layers) suggests the model can be kept light enough for portable BCI hardware.
The same regional-padding strategy could be applied to other EEG tasks with different channel densities, as long as an anatomical mapping is available.
The reported small fold-to-fold variation suggests stable training, which is a prerequisite for deploying a single universal decoder.

Reading between the lines

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

The paper's random shuffled 5-fold cross-validation is a weaker test than holding out an entire paradigm; the strongest version of the claim, 'without retraining on a new paradigm,' would require training on two paradigms and testing on the third.
Because adjacent 3-second segments from the same continuous recording are correlated, the current split may overestimate cross-paradigm generalization; a subject-disjoint or recording-disjoint split would give a more conservative estimate.
The zero-padding strategy means different paradigms effectively have different numbers of valid channels per region; comparing performance against a fixed common subset of channels would clarify how much of the gain comes from padding versus genuine shared representations.
The regional contributions shown by the visualization are averaged over a classification model that sees all paradigms together; a per-paradigm analysis with a held-out paradigm would show whether the model learns paradigm-specific shortcuts or stable neural signatures.

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.

Desk Editor's Note

read the letter

The paper does one thing genuinely new: it tackles channel-configuration heterogeneity across BCI paradigms with an anatomical brain-region partition plus zero padding, and combines a local per-region encoder with a global encoder. That's a reasonable engineering contribution, and the ablation suggests the global encoder earns its keep—removing it drops macro F1 by ~7.7 points. The writing is clear, the method is described in reproducible detail, and the choice of three real paradigms (motor imagery, resting state, driving fatigue) is sensible.

The soft spot is the evaluation. The paper claims 'without retraining and retuning' for diverse paradigms, but no paradigm is ever held out. The 5-fold CV is run on samples shuffled across all three paradigms, so every training fold contains motor imagery, resting-state, and fatigue samples; the model never faces an unseen paradigm or an unseen channel layout. Worse, samples are 3-second non-overlapping segments from continuous recordings; random fold assignment leaks within-subject, within-session correlations into training, which can inflate macro F1 well above the 12.5% chance level. So the headline claim is untested. The reported 59.6% macro F1 may reflect low-level recording artifacts or subject identity rather than cross-paradigm representation.

There are also no baselines—no EEGNet, no Conformer, no within-paradigm comparison—so we can't tell whether the architecture adds anything over a generic transformer with padding. The parameter sensitivity and Grad-CAM analyses are internal consistency checks, not evidence of generalization.

That said, the problem is real and the padding strategy is a plausible building block. If the authors re-ran the evaluation with a leave-one-paradigm-out split, and ideally subject-wise folds within that, the paper would be worth taking seriously. As is, it's a promising proof-of-concept with a load-bearing evaluation flaw.

Recommendation: I'd send it to peer review rather than desk-reject, because the idea and ablation are substantive and the flaw is fixable. But I'd tell the authors the evaluation must be redone before acceptance.

Referee Report

3 major / 4 minor

Summary. The paper proposes DLGE, a dual local-global encoder for EEG classification across three BCI paradigms: motor imagery (4 classes), resting state (2 classes), and driving fatigue (2 classes), giving an 8-class pooled task. To handle heterogeneous channel configurations, the authors partition channels into 11 anatomical brain regions, standardize each region to 9 channels, and zero-pad missing channels. The local brain encoder applies temporal and spatial attention per region, the global brain encoder aggregates regional features, and a cross-task classifier outputs predictions. The authors report mean macro precision, recall, and F1 of 60.16%, 59.88%, and 59.56% from five-fold cross-validation after shuffling all samples from the three datasets, plus ablation, parameter sensitivity, and Grad-CAM analyses.

Significance. If the central claim were supported, the paper would make a useful contribution: a single model that handles multiple BCI paradigms with heterogeneous channel layouts, without per-paradigm retraining, would be relevant to practical brain-computer interfaces. The anatomical brain-region padding strategy and the local-global architecture are sensible design ideas, and the ablation study provides some insight into which components matter. However, the evaluation design does not test the headline claim. All three paradigms are present in every training fold, no paradigm is ever held out, and the random shuffle of continuous 3-second segments creates likely information leakage between training and test folds. The reported numbers are therefore not evidence of cross-paradigm generalization. The absence of baseline comparisons and statistical tests further limits the interpretability of the results.

major comments (3)

[Section II-B, II-D] The evaluation protocol does not test cross-paradigm generalization. The authors state 'Finally, we shuffle all the samples' (II-B) and then use five-fold CV (II-D) on the pooled eight-class data. Every fold contains training and test segments from all three paradigms, and no paradigm is ever held out. Consequently, the central claim in the Abstract ('processing diverse BCI paradigms without retraining and retuning') is untested: the model is only shown to classify a fixed set of eight paradigm-condition classes, with each paradigm represented in the training set. Moreover, because 3-second segments are cut from continuous recordings (5,184 MI; ~1,000 resting; 2,400 fatigue samples), adjacent non-overlapping segments from the same subject and session are highly correlated; random shuffling places correlated segments in both training and test folds, inflating all reported metrics. A valid
[Section III-A, Table II] The reported macro F1 of ~59.56% is presented without any baseline or comparison. With eight classes chance is 12.5%, but no comparison is made to (i) a single-paradigm model retrained per paradigm, (ii) a simple spectral or Riemannian baseline, or (iii) a model trained on shuffled segments with subject-disjoint folds. Because the results in Table II are obtained under the leaky protocol of Major Comment 1, the numbers cannot be interpreted as evidence of generalizable cross-paradigm representations. The absence of confidence intervals or significance tests further weakens the claim.
[Section II-C, Eqs. (6)-(9)] In Eq. (6), the second projection is written q(r)=W_k v(r); it should be k(r)=W_k v(r). More substantively, the spatial interaction in Eq. (7) computes softmax over all ~C=9 channels per region, including zero-padded channels. Padded channels contribute zero-valued keys, which can dilute attention weights assigned to real channels. The authors mask padded positions in the GBE (Section II-C.2) but do not describe an equivalent mask in the LBE spatial attention. This should be clarified or corrected, since it affects the validity of the regional feature extraction for paradigms with fewer than nine channels in a region.

minor comments (4)

[Section III-A] The text reports 'approximately 1 2%' for the standard deviation across folds; this should be '1-2%' or the exact value should be given. Table II also lacks standard deviations and confidence intervals.
[Fig. 3] The y-axis in the ablation figure starts at 47.5%, which visually exaggerates differences. Reporting effect sizes with error bars would be more informative.
[General] No code or data availability statement is provided, which hinders reproducibility. Given the novelty of the claim, releasing the preprocessing and evaluation scripts would be valuable.
[Section III-D] The Grad-CAM visualizations are computed from a model that has access to paradigm-specific channel layouts and padding patterns; the authors should discuss whether these maps might reflect dataset artifacts rather than neurophysiological mechanisms.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: DLGE is trained end-to-end and evaluated on held-out folds of pooled labeled EEG; no fitted parameter is renamed as a prediction.

full rationale

The paper's derivation chain is empirical. Section II-C defines the model (STFT, brain-region partitioning, Local Brain Encoder, Global Brain Encoder, cross-task classifier; Eqs. 1-12) with learnable parameters trained by class-weighted cross-entropy via five-fold cross-validation (Section II-D). The reported macro F1, precision, and recall in Table II are measured test-fold statistics, not quantities constructed from the model's definitions. No target result appears among the inputs: no parameter is fitted to macro F1 or to cross-paradigm performance, and no equation defines performance in terms of a fitted constant. The self-citations (e.g., [10], [30], [37]) are contextual and not load-bearing; none is invoked as an external uniqueness theorem or to forbid alternative model choices. The anatomical partitioning and zero-padding are modeling choices, not renamed empirical patterns. The main weakness is the evaluation protocol: Section II-B states 'Finally, we shuffle all the samples' and then five-fold CV is used, so no paradigm is ever held out and adjacent non-overlapping segments from the same continuous recording may leak into both training and test folds. This means the 'without retraining' claim is untested and possibly inflated, but that is an external-validity/correctness concern, not circularity. Per the instructions, lack of external validity is not circularity; therefore the circularity score is 0.

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

The model introduces no new physical entities. The load-bearing ledger entries are the author-defined region partition, the 9-channel padding scheme, segmentation and STFT choices, and the evaluation split assumption. These choices are not externally benchmarked and directly determine the reported performance.

free parameters (6)

Number of LBE/GBE self-attention layers = 2
Chosen from sensitivity analysis on the same pooled data; deeper stacks reduced macro F1.
Number of GBE attention heads = 8
Chosen from sensitivity analysis over 1, 2, 4, 8, 16 heads on the same pooled data.
STFT window length = 250 samples (1 second at 250 Hz)
Selected without stated justification; changes time-frequency resolution and all downstream features.
STFT overlap rate = 50%
Selected without stated justification; affects temporal granularity of time-frequency features.
Segment length = 3 seconds, non-overlapping
Chosen without justification; all samples and evaluation folds depend on this segmentation.
Standardized channels per brain region = 9
Set by the author-defined anatomical partition; missing channels are zero-padded to reach this number.

assumptions (5)

domain assumption EEG channels can be mapped into 11 fixed anatomical regions and every region standardized to 9 channels while preserving meaning across 10-20 and 10-10 montages.
Core to the channel-heterogeneity solution; if region boundaries or padding distort signals, the shared-feature extraction is compromised. Introduced in Section II-C and Figure 2.
domain assumption Zero-padding missing channels and masking them in attention removes heterogeneity without introducing noise.
Assumed in Section II-C; padded zeros are masked in the global encoder, but the local encoder processes padded channels before masking is applied.
ad hoc to paper Random shuffling of 3-second segments and 5-fold cross-validation is a valid performance estimate for cross-paradigm generalization.
Adopted in Sections II-B and II-D; this is fragile because adjacent segments from the same recording are correlated and no paradigm is held out.
domain assumption STFT time-frequency features contain discriminative information for all three paradigms under the same 0.5-45 Hz band and 250 Hz resampling.
All model inputs use STFT on resampled, filtered signals; no paradigm-specific feature analysis is provided to justify this shared representation.
domain assumption A single label set of 4+2+2 classes from three paradigms is an appropriate target for 'cross-BCI-paradigm' classification.
The paper defines the task as an 8-class pooled problem; this equates classifying within a paradigm with generalizing across paradigms.

how reviews work

0 comments

Cite this review

Pith. "Pith review of DLGE: Dual Local-Global Encoding for Generalizable Cross-BCI-Paradigm." pith.science (2026). https://pith.science/paper/7AA2JZYE

@misc{pith2026250907991,
  author       = {Pith},
  title        = {Pith review of: DLGE: Dual Local-Global Encoding for Generalizable Cross-BCI-Paradigm},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7AA2JZYE}},
  note         = {Machine review of arXiv:2509.07991}
}

read the original abstract

Deep learning models have been frequently used to decode a single brain-computer interface (BCI) paradigm based on electroencephalography (EEG). It is challenging to decode multiple BCI paradigms using one model due to diverse barriers, such as different channel configurations and disparate task-related representations. In this study, we propose Dual Local-Global Encoder (DLGE), enabling the classification across different BCI paradigms. To address the heterogeneity in EEG channel configurations across paradigms, we employ an anatomically inspired brain-region partitioning and padding strategy to standardize EEG channel configuration. In the proposed model, the local encoder is designed to learn shared features across BCI paradigms within each brain region based on time-frequency information, which integrates temporal attention on individual channels with spatial attention among channels for each brain region. These shared features are subsequently aggregated in the global encoder to form respective paradigm-specific feature representations. Three BCI paradigms (motor imagery, resting state, and driving fatigue) were used to evaluate the proposed model. The results demonstrate that our model is capable of processing diverse BCI paradigms without retraining and retuning, achieving average macro precision, recall, and F1-score of 60.16\%, 59.88\%, and 59.56\%, respectively. We made an initial attempt to develop a general model for cross-BCI-paradigm classification, avoiding retraining or redevelopment for each paradigm. This study paves the way for the development of an effective but simple model for cross-BCI-paradigm decoding, which might benefit the design of portable devices for universal BCI decoding.

Figures

Figures reproduced from arXiv: 2509.07991 by the authors.

**Figure 1.** The schematic of the proposed Dual Local-Global Encoder (DLGE) model for cross-BCI-paradigm classification. Each padded brain region is used as the input to the local brain encoder (LBE) that will extract fundamental shared feature within each brain by channel embedder and spatial attention. The input of channel embedder is the time–frequency features of each channel. As for the spatial attention, the input is spati… view at source ↗

**Figure 2.** Illustration of the recorded electrode layouts and the padded electrodes for each paradigm, as well as the brain region partition in our study. Solid circles represent electrodes actually used for data recording, while dashed circles represent the missing electrodes that are padded in our study. Background colors indicate the partitioned brain regions. TABLE II THE RESULTS OF DLGE ON CROSS-BCI-PARADIGM CLASSIFICATIO… view at source ↗

**Figure 3.** Ablation study results of the proposed Dual Local-Global Encoder (DLGE) model across three BCI paradigms. We evaluate the impact of removing key architectural components, which are global brain encoder (GBE), local brain encoder (LBE) and its internal spatial interaction (SI) module, SI module, and both GBE and SI module. The removal of GBE leads to the most significant performance drop across all evaluation metrics… view at source ↗

Figures from the paper (2 more)

**Figure 4.** Figure 4: Parameter sensitivity analysis of the proposed Dual LocalGlobal Encoder (DLGE) model on classification performance. (a) Effect of varying the number of self-attention layers in both the local brain encoder (LBE) and global brain encoder (GBE). The model performs best…

**Figure 5.** Figure 5: Topographical distribution of task-specific brain region contributions based on Gradient-weighted Class Activation Mapping (Grad-CAM). The leftmost column shows the average contribution map across all classes, while the remaining columns are class-specific contribution…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 43 canonical work pages

[1]

MIN2Net: End-to-End Multi-Task Learning for Sub ject- Independent Motor Imagery EEG Classiﬁcation

Phairot Autthasan, Rattanaphon Chaisaen, Thapanun Sudhaw iyangkul, Phurin Rangpong, Suktipol Kiatthaveephong, Nat Dilokthan akul, Gun Bhakdisongkhram, Huy Phan, Cuntai Guan, and Theerawit Wila ipr- asitporn. MIN2Net: End-to-End Multi-Task Learning for Sub ject- Independent Motor Imagery EEG Classiﬁcation. IEEE Transactions on Biomedical Engineering , 69(6...

work page 2022
[2]

Barry, Adam R

Robert J. Barry, Adam R. Clarke, Stuart J. Johnstone, Christ opher A. Magee, and Jacqueline A. Rushby. EEG differences between ey es- closed and eyes-open resting conditions. Clinical Neurophysiology , 118(12):2765–2773, December 2007. Publisher: Elsevier BV

work page 2007
[3]

Hybrid Net- work Using Dynamic Graph Convolution and Temporal Self-Att ention for EEG-Based Emotion Recognition

Cheng Cheng, Zikang Y u, Y ong Zhang, and Lin Feng. Hybrid Net- work Using Dynamic Graph Convolution and Temporal Self-Att ention for EEG-Based Emotion Recognition. IEEE Transactions on Neural Networks and Learning Systems , 35(12):18565–18575, December 2024

work page 2024
[4]

EEG-based drowsiness estimat ion for safety driving using independent component analysis

Chin-Teng Lin, Ruei-Cheng Wu, Sheng-Fu Liang, Wen-Hung Cha o, Y u- Jie Chen, and Tzyy-Ping Jung. EEG-based drowsiness estimat ion for safety driving using independent component analysis. IEEE Transac- tions on Circuits and Systems I: Regular Papers , 52(12):2726–2738, December 2005

work page 2005
[5]

EEG-Deformer: A Dense Convo - lutional Transformer for Brain-Computer Interfaces

Yi Ding, Y ong Li, Hao Sun, Rui Liu, Chengxuan Tong, Chenyu Liu , Xinliang Zhou, and Cuntai Guan. EEG-Deformer: A Dense Convo - lutional Transformer for Brain-Computer Interfaces. IEEE Journal of Biomedical and Health Informatics , 29(3):1909–1918, March 2025

work page 1909
[6]

TSception: Capturing Temporal Dynamics and Spatial Asymme try From EEG for Emotion Recognition

Yi Ding, Neethu Robinson, Su Zhang, Qiuhao Zeng, and Cuntai G uan. TSception: Capturing Temporal Dynamics and Spatial Asymme try From EEG for Emotion Recognition. IEEE Transactions on Affective Com- puting, 14(3):2238–2250, July 2023

work page 2023
[7]

EmT: A Novel Transformer for Generalized Cross-subject EEG Emotion Recognition

Yi Ding, Chengxuan Tong, Shuailei Zhang, Muyun Jiang, Y ong L i, 8 Kevin Lim Jun Liang, and Cuntai Guan. EmT: A Novel Transforme r for Generalized Cross-subject EEG Emotion Recognition. IEEE Trans- actions on Neural Networks and Learning Systems , pages 1–13, 2025. arXiv:2406.18345 [cs]

work page Pith review arXiv 2025
[8]

Ehrlich, Kat R

Stefan K. Ehrlich, Kat R. Agres, Cuntai Guan, and Gordon Chen g. A closed-loop, music-based brain-computer interface for em otion media- tion. PLOS ONE , 14(3):e0213516, March 2019

work page 2019

Show all 43 references

[9]

Flesher, John E

Sharlene N. Flesher, John E. Downey, Jeffrey M. Weiss, Chris topher L. Hughes, Angelica J. Herrera, Elizabeth C. Tyler-Kabara, Mi chael L. Boninger, Jennifer L. Collinger, and Robert A. Gaunt. A brai n-computer interface that evokes tactile sensations improves robotic arm cont...

2021
[10]

Abbass, Kay Chen Tan, Abdullah Al- Mamun, Nitish Thakor, Anastasios Bezerianos, and Junhua Li

Sim Kuan Goh, Hussein A. Abbass, Kay Chen Tan, Abdullah Al- Mamun, Nitish Thakor, Anastasios Bezerianos, and Junhua Li . Spa- tio–Spectral Representation Learning for Electroencepha lographic Gait- Pattern Classiﬁcation. IEEE Transactions on Neural Systems and Re- habilitation...

2018
[11]

Deep Learning in EEG: Advance of the Last Ten-Y ear Critical Perio d

Shu Gong, Kaibo Xing, Andrzej Cichocki, and Junhua Li. Deep Learning in EEG: Advance of the Last Ten-Y ear Critical Perio d. IEEE Transactions on Cognitive and Developmental Systems , 14(2):348–365, June 2022. Publisher: Institute of Electrical and Electron ics Engineers (IEEE)

2022
[12]

Driver Vigilance State Esti- mation Based on Multisource Data

Zizheng Guo, Ming Zhou, and Guofa Li. Driver Vigilance State Esti- mation Based on Multisource Data. IEEE Sensors Journal , 23(3):2592– 2604, February 2023

2023
[13]

Reli ability of EEG Measures in Driving Fatigue

Jonathan Harvy, Anastasios Bezerianos, and Junhua Li. Reli ability of EEG Measures in Driving Fatigue. IEEE Transactions on Neural Sys- tems and Rehabilitation Engineering , 30:2743–2753, 2022. Publisher: Institute of Electrical and Electronics Engineers (IEEE)

2022
[14]

EEG-Channel-Temporal-Spect ral- Attention Correlation for Motor Imagery EEG Classiﬁcation

Wei-Y en Hsu and Y a-Wen Cheng. EEG-Channel-Temporal-Spect ral- Attention Correlation for Motor Imagery EEG Classiﬁcation . IEEE Transactions on Neural Systems and Rehabilitation Enginee ring, 31:1659–1669, 2023

2023
[15]

EEGNet: a compact convolutional neural network for EEG-based brain–compute r interfaces

V ernon J Lawhern, Amelia J Solon, Nicholas R Waytowich, Step hen M Gordon, Chou P Hung, and Brent J Lance. EEGNet: a compact convolutional neural network for EEG-based brain–compute r interfaces. Journal of Neural Engineering , 15(5):056013, October 2018

2018
[16]

Self-Regulation Phenomenon Emerge d During Prolonged Fatigue Driving: An EEG Connectivity Study

Gang Li, Jie Wang, Wanxiu Xu, Kuijun Wu, Y ufeng Liu, Anastasi os Bezerianos, and Y u Sun. Self-Regulation Phenomenon Emerge d During Prolonged Fatigue Driving: An EEG Connectivity Study. IEEE Transac- tions on Neural Systems and Rehabilitation Engineering , 31:4895–4906, 2023

2023
[17]

Design of Assistive Wheelchair System Directly Stee red by Human Thoughts

Junhua Li, Jianyi Liang, Qibin Zhao, Jie Li, Kan Hong, and Liq ing Zhang. Design of Assistive Wheelchair System Directly Stee red by Human Thoughts. International Journal of Neural Systems , 23(03):1350013, June 2013. Publisher: World Scientiﬁc Pub Co Pte Ltd

2013
[18]

Exploring EEG Features in Cross-Subject Emotion Rec ognition

Xiang Li, Dawei Song, Peng Zhang, Y azhou Zhang, Y uexian Hou, and Bin Hu. Exploring EEG Features in Cross-Subject Emotion Rec ognition. Frontiers in Neuroscience, 12:162, March 2018

2018
[19]

Fine-Gra ined Spatial-Frequency-Time Framework for Motor Imagery Brain -Computer Interface

Guoyang Liu, Rui Zhang, Lan Tian, and Weidong Zhou. Fine-Gra ined Spatial-Frequency-Time Framework for Motor Imagery Brain -Computer Interface. IEEE Journal of Biomedical and Health Informatics , pages 1–13, 2025

2025
[20]

DA-CapsNet: A multi-branch capsule network b ased on adversarial domain adaption for cross-subject EEG emotion recognition

Shuaiqi Liu, Zeyao Wang, Y anling An, Bing Li, Xinrui Wang, an d Y udong Zhang. DA-CapsNet: A multi-branch capsule network b ased on adversarial domain adaption for cross-subject EEG emotion recognition. Knowledge-Based Systems , 283:111137, January 2024

2024
[21]

Driver fatigue de tection through multiple entropy fusion analysis in an EEG-based sy stem

Jianliang Min, Ping Wang, and Jianfeng Hu. Driver fatigue de tection through multiple entropy fusion analysis in an EEG-based sy stem. PLOS ONE, 12(12):e0188756, December 2017

2017
[22]

Cross-Session Emotion Recognition by Joint L abel- Common and Label-Speciﬁc EEG Features Exploration

Y ong Peng, Honggang Liu, Junhua Li, Jun Huang, Bao-Liang Lu, and Wanzeng Kong. Cross-Session Emotion Recognition by Joint L abel- Common and Label-Speciﬁc EEG Features Exploration. IEEE Transac- tions on Neural Systems and Rehabilitation Engineering , 31:759–768, 2023

2023
[23]

Motor imagery activ ates primary sensorimotor area in humans

Gert Pfurtscheller and Christa Neuper. Motor imagery activ ates primary sensorimotor area in humans. Neuroscience Letters , 239(2-3):65–68, December 1997

1997
[24]

Deep learning with convolutional neural networks for EEG de coding and visualization

Robin Tibor Schirrmeister, Jost Tobias Springenberg, Luka s Do- minique Josef Fiederer, Martin Glasstetter, Katharina Egg ensperger, Michael Tangermann, Frank Hutter, Wolfram Burgard, and Ton io Ball. Deep learning with convolutional neural networks for EEG de coding and visua...

2017
[25]

Grad-CAM: Vis ual Explanations From Deep Networks via Gradient-Based Locali zation

Ramprasaath R Selvaraju, Michael Cogswell, Abhishek Das, R amakr- ishna V edantam, Devi Parikh, and Dhruv Batra. Grad-CAM: Vis ual Explanations From Deep Networks via Gradient-Based Locali zation
[26]

Simultaneous Eye Blink Characterization and Elim ination From Low-Channel Prefrontal EEG Signals Enhances Driver Dr owsi- ness Detection

Mohammad Shahbakhti, Matin Beiramvand, Izabela Rejer, Pio tr Au- gustyniak, Anna Broniec-Wojcik, Michal Wierzchon, and V ai dotas Marozas. Simultaneous Eye Blink Characterization and Elim ination From Low-Channel Prefrontal EEG Signals Enhances Driver Dr owsi- ness Detection. ...

2022
[27]

Multisource Associate Domain Adaptation for Cross- Subject and Cross-Session EEG Emotion Recognition

Qingshan She, Chenqi Zhang, Feng Fang, Y uliang Ma, and Yingc hun Zhang. Multisource Associate Domain Adaptation for Cross- Subject and Cross-Session EEG Emotion Recognition. IEEE Transactions on Instrumentation and Measurement , 72:1–12, 2023

2023
[28]

Contrastive Learning of Subject-Invariant EEG Representa tions for Cross-Subject Emotion Recognition

Xinke Shen, Xianggen Liu, Xin Hu, Dan Zhang, and Sen Song. Contrastive Learning of Subject-Invariant EEG Representa tions for Cross-Subject Emotion Recognition. IEEE Transactions on Affective Computing, 14(3):2496–2511, July 2023

2023
[29]

EEG Conformer: Convolutional Transformer for EEG Decoding and Visualization

Y onghao Song, Qingqing Zheng, Bingchuan Liu, and Xiaorong G ao. EEG Conformer: Convolutional Transformer for EEG Decoding and Visualization. IEEE Transactions on Neural Systems and Rehabilitation Engineering, 31:710–719, 2023

2023
[30]

AdaptEEG: A Deep Subdomain Adaptation Network With Class Confusion Loss for Cross-Subject Mental Workload Classiﬁcation

Wu Sun and Junhua Li. AdaptEEG: A Deep Subdomain Adaptation Network With Class Confusion Loss for Cross-Subject Mental Workload Classiﬁcation. IEEE Journal of Biomedical and Health Informatics , 29(3):1940–1949, March 2025

1940
[31]

Func- tional cortical connectivity analysis of mental fatigue un masks hemi- spheric asymmetry and changes in small-world networks

Y u Sun, Julian Lim, Kenneth Kwok, and Anastasios Bezerianos . Func- tional cortical connectivity analysis of mental fatigue un masks hemi- spheric asymmetry and changes in small-world networks. Brain and Cognition, 85:220–230, March 2014. Publisher: Elsevier BV

2014
[32]

Miller, Gernot R

Michael Tangermann, Klaus-Robert M¨ uller, Ad Aertsen, Nie ls Bir- baumer, Christoph Braun, Clemens Brunner, Robert Leeb, Car sten Mehring, Kai J. Miller, Gernot R. M¨ uller-Putz, Guido Nolte , Gert Pfurtscheller, Hubert Preissl, Gerwin Schalk, Alois Schl¨ ogl, Carmen Vidaurre...

2012
[33]

Prediction of Reaction Time and Vigilance V ar iability From Spatio-Spectral Features of Resting-State EEG in a Lon g Sustained Attention Task

Mastaneh Torkamani-Azar, Sumeyra Demir Kanik, Serap Aydin , and Mujdat Cetin. Prediction of Reaction Time and Vigilance V ar iability From Spatio-Spectral Features of Resting-State EEG in a Lon g Sustained Attention Task. IEEE Journal of Biomedical and Health Informatics , 24(...

2020
[34]

Past, Pres ent, and Future of EEG-Based BCI Applications

Kaido V¨ arbu, Naveed Muhammad, and Y ar Muhammad. Past, Pres ent, and Future of EEG-Based BCI Applications. Sensors, 22(9):3331, April 2022

2022
[35]

Thakor, and Anastasios Bezerianos

Hongtao Wang, Andrei Dragomir, Nida Itrat Abbasi, Junhua Li , Nitish V . Thakor, and Anastasios Bezerianos. A novel real-time drivi ng fatigue detection system based on wireless dry EEG. Cognitive Neurodynamics, 12(4):365–376, August 2018

2018
[36]

Driving Fatigue Recognition With Funct ional Connectivity Based on Phase Synchronization

Hongtao Wang, Xucheng Liu, Junhua Li, Tao Xu, Anastasios Bezerianos, Y u Sun, and Feng Wan. Driving Fatigue Recognition With Funct ional Connectivity Based on Phase Synchronization. IEEE Transactions on Cognitive and Developmental Systems , 13(3):668–678, September 2021. Publi...

2021
[37]

Multikernel Capsule Network for Schizophrenia Identiﬁcat ion

Tian Wang, Anastasios Bezerianos, Andrzej Cichocki, and Ju nhua Li. Multikernel Capsule Network for Schizophrenia Identiﬁcat ion. IEEE Transactions on Cybernetics , 52(6):4741–4750, June 2022

2022
[38]

Phase Synchrony Measurement in Motor Cortex for Classifying Sing le-trial EEG during Motor Imagery

Yijun Wang, Bo Hong, Xiaorong Gao, and Shangkai Gao. Phase Synchrony Measurement in Motor Cortex for Classifying Sing le-trial EEG during Motor Imagery. In 2006 International Conference of the IEEE Engineering in Medicine and Biology Society , pages 75–78, New Y ork, NY , Augu...

2006
[39]

Effects of Motor Imagery Tasks on Brain Functional Networks Based on EE G Mu/Beta Rhythm

Hongli Y u, Sidi Ba, Y uxue Guo, Lei Guo, and Guizhi Xu. Effects of Motor Imagery Tasks on Brain Functional Networks Based on EE G Mu/Beta Rhythm. Brain Sciences , 12(2):194, January 2022

2022
[40]

Deep Neural Network-Based Empirical Mode De com- position for Motor Imagery EEG Classiﬁcation

Hyunsoo Y u, Suwhan Baek, Jiwoon Lee, Illsoo Sohn, Bosun Hwang, and Cheolsoo Park. Deep Neural Network-Based Empirical Mode De com- position for Motor Imagery EEG Classiﬁcation. IEEE Transactions on Neural Systems and Rehabilitation Engineering , 32:3647–3656, 2024

2024
[41]

FMLAN: A novel framework for cross-subject and cross-session EEG emotion recognition

Peng Y u, Xiaopeng He, Haoyu Li, Haowen Dou, Y eyu Tan, Hao Wu, and Badong Chen. FMLAN: A novel framework for cross-subject and cross-session EEG emotion recognition. Biomedical Signal Processing and Control, 100:106912, February 2025

2025
[42]

Hanzhong Zhang, Tienyu Zuo, Zhiyang Chen, Xin Wang, and Poly Z.H. Sun. Evolutionary Ensemble Learning for EEG-Based Cross-S ubject Emotion Recognition. IEEE Journal of Biomedical and Health Infor- matics, 28(7):3872–3881, July 2024

2024
[43]

Bleich- ner, Ingmar Gutberlet, and Maarten De V os

Catharina Zich, Stefan Debener, Cornelia Kranczioch, Mart in G. Bleich- ner, Ingmar Gutberlet, and Maarten De V os. Real-time EEG fee dback 9 during simultaneous EEG–fMRI identiﬁes the cortical signa ture of motor imagery. NeuroImage, 114:438–447, July 2015

2015

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [1]

MIN2Net: End-to-End Multi-Task Learning for Sub ject- Independent Motor Imagery EEG Classiﬁcation

Phairot Autthasan, Rattanaphon Chaisaen, Thapanun Sudhaw iyangkul, Phurin Rangpong, Suktipol Kiatthaveephong, Nat Dilokthan akul, Gun Bhakdisongkhram, Huy Phan, Cuntai Guan, and Theerawit Wila ipr- asitporn. MIN2Net: End-to-End Multi-Task Learning for Sub ject- Independent Motor Imagery EEG Classiﬁcation. IEEE Transactions on Biomedical Engineering , 69(6...

work page 2022

[2] [2]

Barry, Adam R

Robert J. Barry, Adam R. Clarke, Stuart J. Johnstone, Christ opher A. Magee, and Jacqueline A. Rushby. EEG differences between ey es- closed and eyes-open resting conditions. Clinical Neurophysiology , 118(12):2765–2773, December 2007. Publisher: Elsevier BV

work page 2007

[3] [3]

Hybrid Net- work Using Dynamic Graph Convolution and Temporal Self-Att ention for EEG-Based Emotion Recognition

Cheng Cheng, Zikang Y u, Y ong Zhang, and Lin Feng. Hybrid Net- work Using Dynamic Graph Convolution and Temporal Self-Att ention for EEG-Based Emotion Recognition. IEEE Transactions on Neural Networks and Learning Systems , 35(12):18565–18575, December 2024

work page 2024

[4] [4]

EEG-based drowsiness estimat ion for safety driving using independent component analysis

Chin-Teng Lin, Ruei-Cheng Wu, Sheng-Fu Liang, Wen-Hung Cha o, Y u- Jie Chen, and Tzyy-Ping Jung. EEG-based drowsiness estimat ion for safety driving using independent component analysis. IEEE Transac- tions on Circuits and Systems I: Regular Papers , 52(12):2726–2738, December 2005

work page 2005

[5] [5]

EEG-Deformer: A Dense Convo - lutional Transformer for Brain-Computer Interfaces

Yi Ding, Y ong Li, Hao Sun, Rui Liu, Chengxuan Tong, Chenyu Liu , Xinliang Zhou, and Cuntai Guan. EEG-Deformer: A Dense Convo - lutional Transformer for Brain-Computer Interfaces. IEEE Journal of Biomedical and Health Informatics , 29(3):1909–1918, March 2025

work page 1909

[6] [6]

TSception: Capturing Temporal Dynamics and Spatial Asymme try From EEG for Emotion Recognition

Yi Ding, Neethu Robinson, Su Zhang, Qiuhao Zeng, and Cuntai G uan. TSception: Capturing Temporal Dynamics and Spatial Asymme try From EEG for Emotion Recognition. IEEE Transactions on Affective Com- puting, 14(3):2238–2250, July 2023

work page 2023

[7] [7]

EmT: A Novel Transformer for Generalized Cross-subject EEG Emotion Recognition

Yi Ding, Chengxuan Tong, Shuailei Zhang, Muyun Jiang, Y ong L i, 8 Kevin Lim Jun Liang, and Cuntai Guan. EmT: A Novel Transforme r for Generalized Cross-subject EEG Emotion Recognition. IEEE Trans- actions on Neural Networks and Learning Systems , pages 1–13, 2025. arXiv:2406.18345 [cs]

work page Pith review arXiv 2025

[8] [8]

Ehrlich, Kat R

Stefan K. Ehrlich, Kat R. Agres, Cuntai Guan, and Gordon Chen g. A closed-loop, music-based brain-computer interface for em otion media- tion. PLOS ONE , 14(3):e0213516, March 2019

work page 2019