pith. sign in

arxiv: 2606.00180 · v1 · pith:YQ4GSE5Jnew · submitted 2026-05-29 · 💻 cs.LG · cs.AI

Beyond Augmentation: Score-Guided Pathological Prior for EEG-based Depression Detection

Xiaojing Chen , Jingqi Cheng , Xu Zhao , Wan Jiang , Jingjing Wu This is my paper

Pith reviewed 2026-06-28 23:29 UTC · model grok-4.3

classification 💻 cs.LG cs.AI
keywords EEGDepression DetectionAnomaly ScorePathological PriorScore-Guided ClassificationFew-Shot LearningCross-Channel AdaptationMDD
0
0 comments X

The pith

An unsupervised generative network's anomaly scores form a pathological prior that fuses with EEG features to guide depression classification without any data augmentation.

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

The paper seeks to solve the small-sample problem in EEG-based major depressive disorder detection by rejecting the usual strategy of generating synthetic training data. It instead trains an unsupervised generative network to compute structural and statistical anomaly scores for each real sample; after normalization these scores become an explicit pathological prior. The prior is concatenated or otherwise fused with learned deep features so that the downstream classifier's decision boundary is steered by the anomaly information. A separate cross-channel spatial adaptation module maps features across mismatched electrode setups to support multi-center evaluation. On the Mumtaz2016 and MODMA datasets the resulting score-guided classifier reaches usable accuracy under a strict zero-augmentation regime.

Core claim

The core claim is that modeling per-sample anomaly degrees with an unsupervised generative network yields a normalized pathological prior that, when fused with deep feature representations, supplies sufficient guidance for the classifier to separate depressed from non-depressed EEG recordings on the Mumtaz2016 and MODMA collections without synthesizing any additional samples or incurring augmentation overhead.

What carries the argument

The Score-Guided Classification (SGC) framework: an unsupervised generative network produces anomaly scores that become the pathological prior; after robust normalization the prior is explicitly fused with deep features to steer the classifier decision boundary, augmented by a Cross-Channel Spatial Adaptation module for channel mismatch.

If this is right

  • MDD detection becomes feasible on small real-only EEG collections without the compute or noise risk of generative augmentation.
  • The same prior-fusion step can be applied to any downstream EEG classifier that already extracts deep features.
  • Cross-channel adaptation removes the need to retrain or discard data when electrode montages differ across recording sites.
  • Classification boundaries are explicitly regularized by a data-driven measure of deviation from normality rather than by volume of synthetic examples.

Where Pith is reading between the lines

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

  • The approach could be tested on other binary EEG classification tasks such as seizure detection or sleep staging to check whether anomaly-score priors transfer beyond depression.
  • If the generative network is replaced by a simpler density estimator the framework might become lighter while preserving the same fusion logic.
  • The normalized prior could serve as an interpretable per-subject score for clinical triage before the full classifier is run.

Load-bearing premise

The anomaly scores computed by the unsupervised generative network on the target EEG recordings faithfully encode pathological structure and, once normalized, improve the classifier without injecting dataset-specific bias or circular dependence on the classification samples themselves.

What would settle it

Training the same classifier architecture on Mumtaz2016 and MODMA with the pathological-prior fusion removed and measuring whether accuracy drops below the full SGC result by a statistically significant margin.

Figures

Figures reproduced from arXiv: 2606.00180 by Jingjing Wu, Jingqi Cheng, Wan Jiang, Xiaojing Chen, Xu Zhao.

Figure 1
Figure 1. Figure 1: Generative flaws in standard diffusion-based aug [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overall architecture of the Score-Guided Classification (SGC) framework. The [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Probability density distribution of the normalized [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: Confusion matrices on a representative fold. SGC (c) [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 4
Figure 4. Figure 4: t-SNE visualization. SGC (b) achieves explicit feature [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
read the original abstract

Deep learning-based Major Depressive Disorder (MDD) detection using Electroencephalography (EEG) is fundamentally constrained by the "small-sample dilemma." Prevailing generative data augmentation methods not only incur heavy computational overhead but also risk introducing synthetic noise, thereby blurring classification boundaries. To challenge the traditional "data quantity first" convention, we propose a novel framework "Beyond Augmentation": Score-Guided Classification (SGC). SGC does not synthesize pseudo-samples; instead, it utilizes an unsupervised generative network architecture to model the structural and statistical anomaly degrees of samples, serving as the core "Pathological Prior". This prior, after robust normalization, is explicitly fused with deep feature representations, thereby precisely guiding the classifier's decision boundary. Furthermore, to dynamically adapt to varying channel configurations, we propose a Cross-Channel Spatial Adaptation module, utilizing a spatial mapping mechanism to effectively resolve the hardware heterogeneity of mismatched channels in multi-center datasets. Extensive experiments on the Mumtaz2016 and high-density MODMA datasets demonstrate the effectiveness and exceptional generalizability of our method under the challenging "zero data augmentation" setting and at "zero sample synthesis cost". Keywords: Electroencephalography (EEG), Depression Detection, Anomaly Score, Diffusion Models, Few-Shot Learning

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.

Referee Report

3 major / 2 minor

Summary. The paper proposes a Score-Guided Classification (SGC) framework for EEG-based MDD detection that avoids data augmentation. It employs an unsupervised generative network (diffusion-based per keywords) to derive a 'Pathological Prior' from anomaly scores capturing structural and statistical deviations; after normalization this prior is fused with deep features to guide the classifier decision boundary. A Cross-Channel Spatial Adaptation module handles channel mismatches across datasets. Experiments claim strong performance on Mumtaz2016 and MODMA under zero-augmentation conditions.

Significance. If the anomaly scores reliably encode MDD-specific pathology independent of the classification task and the fusion demonstrably improves boundaries without circular dependence, the approach could offer a lower-cost alternative to generative augmentation for small-sample EEG problems while addressing hardware heterogeneity.

major comments (3)
  1. [Abstract, §3] Abstract and §3 (Pathological Prior construction): the unsupervised generative network is described as modeling anomaly degrees on the target samples, yet no equations, training split, or separation of healthy vs. MDD cohorts during prior computation are supplied; this leaves open whether scores reflect general reconstruction error on the full dataset rather than MDD-specific structure, directly undermining the claim of non-circular guidance.
  2. [§4] §4 (Experiments): the abstract asserts effectiveness on Mumtaz2016 and MODMA but supplies no equations for the fusion step, no ablation removing the prior, no error bars, and no baseline comparisons; without these the central performance claim cannot be checked against the 'zero augmentation' setting.
  3. [§3.2] §3.2 (Cross-Channel Spatial Adaptation): the module is introduced to resolve channel heterogeneity, but no formal definition of the spatial mapping or proof that it preserves the independence of the pathological prior is given, making it unclear whether adaptation interacts with or biases the anomaly scores.
minor comments (2)
  1. [Abstract] The abstract contains several run-on sentences that obscure the precise fusion mechanism; rephrasing for clarity would aid readability.
  2. [Keywords, §3] Keywords list 'Diffusion Models' but the main text does not explicitly confirm the generative architecture; consistent terminology would help.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the constructive comments. We address each major comment below and will revise the manuscript to incorporate the requested clarifications and additions.

read point-by-point responses

  1. Referee: [Abstract, §3] Abstract and §3 (Pathological Prior construction): the unsupervised generative network is described as modeling anomaly degrees on the target samples, yet no equations, training split, or separation of healthy vs. MDD cohorts during prior computation are supplied; this leaves open whether scores reflect general reconstruction error on the full dataset rather than MDD-specific structure, directly undermining the claim of non-circular guidance.

    Authors: The pathological prior is constructed via fully unsupervised training of the generative network on the target samples without labels or cohort separation, which is the standard anomaly detection setup and ensures non-circularity since no classification labels are used. We will add the explicit equations for anomaly score computation, detail the training procedure, and discuss how deviations align with MDD-related EEG pathology. revision: yes

  2. Referee: [§4] §4 (Experiments): the abstract asserts effectiveness on Mumtaz2016 and MODMA but supplies no equations for the fusion step, no ablation removing the prior, no error bars, and no baseline comparisons; without these the central performance claim cannot be checked against the 'zero augmentation' setting.

    Authors: We agree additional experimental details are needed. The revised manuscript will include the fusion equations, ablation studies removing the prior, results with error bars, and baseline comparisons under zero-augmentation conditions. revision: yes

  3. Referee: [§3.2] §3.2 (Cross-Channel Spatial Adaptation): the module is introduced to resolve channel heterogeneity, but no formal definition of the spatial mapping or proof that it preserves the independence of the pathological prior is given, making it unclear whether adaptation interacts with or biases the anomaly scores.

    Authors: We will add a formal mathematical definition of the spatial mapping. We will also include analysis demonstrating that the mapping preserves prior independence, as it is label-agnostic and applied before anomaly scoring. revision: yes

Circularity Check

0 steps flagged

No circularity: unsupervised prior is label-independent and does not reduce to fitted inputs or self-definition

full rationale

The derivation computes anomaly scores via an unsupervised generative network (diffusion-based) on the EEG samples without reference to class labels, normalizes them, and fuses the result as an auxiliary input to a supervised classifier. This chain is self-contained: the prior is derived from data statistics alone, the fusion step is an explicit architectural choice, and evaluation occurs on held-out labeled data. No equation or step equates the prior to the classification target by construction, no self-citation is load-bearing for the core claim, and no fitted parameter is relabeled as a prediction. The method therefore satisfies the default expectation of non-circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 1 invented entities

Abstract-only review supplies insufficient detail to enumerate concrete free parameters or background axioms; the central mechanism rests on the unstated premise that anomaly scores from the generative network are meaningful and normalizable.

invented entities (1)
  • Pathological Prior no independent evidence
    purpose: Anomaly-score representation that guides the classifier decision boundary
    Introduced as the core of SGC; no independent falsifiable handle supplied in abstract.

pith-pipeline@v0.9.1-grok · 5763 in / 1267 out tokens · 29245 ms · 2026-06-28T23:29:06.617029+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

74 extracted references · 10 canonical work pages · 3 internal anchors

  1. [1]

    Jinwon An and Sungzoon Cho. 2015. Variational autoencoder based anomaly detection using reconstruction probability.Special lecture on IE2, 1 (2015), 1–18

    2015

  2. [2]

    Betul Ay, Ozal Yildirim, Muhammed Talo, Ulas Baran Baloglu, Galip Aydin, Subha D Puthankattil, and U Rajendra Acharya. 2019. Automated depression detection using deep representation and sequence learning with EEG signals. Journal of medical systems43, 7 (2019), 205

    2019

  3. [3]

    Christoph Baur, Benedikt Wiestler, Shadi Albarqouni, and Nassir Navab. 2018. Deep autoencoding models for unsupervised anomaly segmentation in brain MR images. InInternational MICCAI brainlesion workshop. Springer, 161–169

    2018

  4. [4]

    Hanshu Cai, Zhenqin Yuan, Yiwen Gao, Shuting Sun, Na Li, Fuze Tian, Han Xiao, Jianxiu Li, Zhengwu Yang, Xiaowei Li, et al. 2022. A multi-modal open dataset for mental-disorder analysis.Scientific data9, 1 (2022), 178

    2022

  5. [5]

    Yu Chen and Chunfeng Yang. 2025. STGE-Former: Spatial-Temporal Graph- Enhanced Transformer for EEG-Based Major Depressive Disorder Detection. In ICASSP 2025-2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 1–5

    2025

  6. [6]

    Daniel Choi, Cordelia Yip, Andrew Choi, and Junho Park. 2026. Fail closed trust gated synthetic augmentation governs tail risk under subject shift in EEG.bioRxiv (2026), 2026–01

    2026

  7. [7]

    Kushan Choudhury, Shubhrodeep Roy, Ankur Chanda, Shubhajit Biswas, and Somenath Kuiry. 2025. Improving Predictive Confidence in Medical Imaging via Online Label Smoothing. InBIO Web of Conferences, Vol. 204. EDP Sciences, 01019

    2025

  8. [8]

    Yi Ding, Neethu Robinson, Qiuhao Zeng, Duo Chen, Aung Aung Phyo Wai, Tih- Shih Lee, and Cuntai Guan. 2020. TSception: A deep learning framework for emotion detection using EEG. In2020 international joint conference on neural networks (IJCNN). IEEE, 1–7

    2020

  9. [9]

    Mohamed El Kerdawy, Mohamed El Halaby, Afnan Hassan, Mohamed Maher, Hatem Fayed, Doaa Shawky, and Ashraf Badawi. 2020. The automatic detection of cognition using eeg and facial expressions.Sensors20, 12 (2020), 3516

    2020

  10. [10]

    Tharindu Fernando, Harshala Gammulle, Simon Denman, Sridha Sridharan, and Clinton Fookes. 2021. Deep learning for medical anomaly detection–a survey. ACM computing surveys (CSUR)54, 7 (2021), 1–37

    2021

  11. [11]

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. 2014. Generative adversarial nets. InAdvances in neural information processing systems, Vol. 27

    2014

  12. [12]

    Louisa Hallal, Jason Rhinelander, Ramesh Venkat, and Aaron Newman. 2026. Efficient Feature Extraction for EEG-Based Classification: A Comparative Review of Deep Learning Models.AI7, 2 (2026), 50

    2026

  13. [13]

    Amr M Hamed, Abdel-Fattah A Heliel, and Heba El-Behery. 2026. Explainable EEG Analysis of Major Psychiatric Disorders: Power Spectra, Functional Connec- tivity, and SHAP Interpretation.Journal of Contemporary Technology and Applied Engineering5, 1 (2026), 32–48

    2026

  14. [14]

    Kay Gregor Hartmann, Robin Tibor Schirrmeister, and Tonio Ball. 2018. EEG- GAN: Generative adversarial networks for electroencephalograhic (EEG) brain signals.arXiv preprint arXiv:1806.01875(2018)

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  15. [15]

    Marwa Hassan and Naima Kaabouch. 2024. Impact of feature selection techniques on the performance of machine learning models for depression detection using EEG data.Applied Sciences14, 22 (2024), 10532

    2024

  16. [16]

    Chao He, Jialu Liu, Yuesheng Zhu, and Wencai Du. 2021. Data augmentation for deep neural networks model in EEG classification task: a review.Frontiers in Human Neuroscience15 (2021), 765525

    2021

  17. [17]

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. 2020. Denoising diffusion probabilistic models.Advances in neural information processing systems33 (2020), 6840–6851

    2020

  18. [18]

    Pengfei Hou, Xiaowei Li, Jing Zhu, and Bin Hu. 2025. A lightweight convolutional transformer neural network for EEG-based depression recognition.Biomedical Signal Processing and Control100 (2025), 107112

    2025

  19. [19]

    Sergey Ioffe and Christian Szegedy. 2015. Batch normalization: Accelerating deep network training by reducing internal covariate shift. InInternational conference on machine learning. pmlr, 448–456

    2015

  20. [20]

    Jung-Hwan Kim, Hyerin Nam, Doyeon Won, and Chang-Hwan Im. 2025. Domain- generalized deep learning for improved subject-independent emotion recognition based on electroencephalography.Experimental Neurobiology34, 3 (2025), 119

    2025

  21. [21]

    Diederik P Kingma and Max Welling. 2013. Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114(2013)

    work page internal anchor Pith review Pith/arXiv arXiv 2013

  22. [22]

    George H Klem. 1999. The ten-twenty electrode system of the international federation. The international federation of clinical neurophysiology.Electroen- cephalogr. Clin. Neurophysiol. Suppl.52 (1999), 3–6

    1999

  23. [23]

    Elnaz Lashgari, Dehua Liang, and Uri Maoz. 2020. Data augmentation for deep- learning-based electroencephalography.Journal of Neuroscience Methods346 (2020), 108885

    2020

  24. [24]

    Mustapha Abdulrahman Lawal, Abdultaofeek Abayomi, Abubakar Abba Salihu, Anes Aihong Aliyu, Muazu Aminu Aliyu, and Ismail Zahraddeen Yakubu. 2026. Transfer learning and domain adaptation in neuroinformatics. InDeep Learning Applications in Neuroinformatics. Elsevier, 287–310

    2026

  25. [25]

    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

    2018

  26. [26]

    Ivana Leccisotti, Anita Mollica, Rossana Laurello, Maria Claudia Moretti, Mario Altamura, Antonello Bellomo, Francesco Panza, and Madia Lozupone. 2026. Ma- chine learning-assisted resting-state electroencephalography improves diagnostic accuracy in psychiatric disorders: A narrative review.Advanced Technology in Neuroscience3, 1 (2026), 21–33

    2026

  27. [27]

    Cheol-Hui Lee, Hwa-Yeon Lee, and Dong-Joo Kim. 2026. RL-BioAug: Label- Efficient Reinforcement Learning for Self-Supervised EEG Representation Learn- ing.arXiv preprint arXiv:2601.13964(2026)

    work page arXiv 2026

  28. [28]

    Wei Li, Siyi Wang, Shitong Shao, and Kaizhu Huang. 2024. Distillation-based domain generalization for cross-dataset EEG-based emotion recognition.IEEE Transactions on Emerging Topics in Computational Intelligence9, 3 (2024), 2474– 2490

    2024

  29. [29]

    Chuncheng Liao, Shiyu Zhao, Xiangcun Wang, Jiacai Zhang, Yongzhong Liao, and Xia Wu. 2025. EEG Data Augmentation Method Based on the Gaussian Mixture Model.Mathematics13, 5 (2025), 729

    2025

  30. [30]

    Lequan Lin, Zhengkun Li, Ruikun Li, Xuliang Li, and Junbin Gao. 2024. Diffusion models for time-series applications: a survey.Frontiers of Information Technology & Electronic Engineering25, 1 (2024), 19–41

    2024

  31. [31]

    Yuan Liu, Changqin Pu, Shan Xia, Dingyu Deng, Xing Wang, and Mengqian Li

  32. [32]

    Machine learning approaches for diagnosing depression using EEG: A review.Translational Neuroscience13, 1 (2022), 224–235

    2022

  33. [33]

    Ilya Loshchilov and Frank Hutter. 2016. Sgdr: Stochastic gradient descent with warm restarts.arXiv preprint arXiv:1608.03983(2016)

    work page internal anchor Pith review Pith/arXiv arXiv 2016

  34. [34]

    Haifeng Lu, Zhiyang You, Yi Guo, and Xiping Hu. 2024. Mast-gcn: Multi-scale adaptive spatial-temporal graph convolutional network for eeg-based depression recognition.IEEE Transactions on Affective Computing15, 4 (2024), 1985–1996

    2024

  35. [35]

    Gang Luo, Hong Rao, Panfeng An, Yunxia Li, Ruiyun Hong, Wenwu Chen, and Shengbo Chen. 2023. Exploring adaptive graph topologies and temporal graph networks for EEG-based depression detection.IEEE Transactions on Neural Systems and Rehabilitation Engineering31 (2023), 3947–3957

    2023

  36. [36]

    RuiFang Lyu. 2026. Deep learning approaches for EEG-based healthcare appli- cations: a comprehensive review.Frontiers in Human Neuroscience19 (2026), 1689073

    2026

  37. [37]

    Andre F Marquand, Iead Rezek, Jan Buitelaar, and Christian F Beckmann. 2016. Understanding heterogeneity in clinical cohorts using normative models: beyond case-control studies.Biological psychiatry80, 7 (2016), 552–561

    2016

  38. [38]

    Przemysław Mirowski and Anna Fabijańska. 2026. Diffusion model-based syn- thesis of brain images for data augmentation.Biomedical Signal Processing and Control113 (2026), 108940

    2026

  39. [39]

    Rafael Müller, Simon Kornblith, and Geoffrey E Hinton. 2019. When does label smoothing help?Advances in neural information processing systems32 (2019)

    2019

  40. [40]

    Wajid Mumtaz, Likun Xia, Syed Saad Azhar Ali, Mohd Azhar Mohd Yasin, Muham- mad Hussain, and Aamir Saeed Malik. 2017. Electroencephalogram (EEG)-based computer-aided technique to diagnose major depressive disorder (MDD).Biomed- ical Signal Processing and Control31 (2017), 108–115

    2017

  41. [41]

    Sebastian Olbrich, Natalia Jaworska, Sara de la Salle, Verner Knott, Pierre Blier, Martin Brunovsky, Tobias Welt, Mateo de Bardeci, and Cheng Teng-Ip. 2026. Deep learning using electroencephalogram (EEG) data for diagnosing and predicting SSRI response in major depressive disorder.Communications Medicine6, 1 (2026), 159

    2026

  42. [42]

    Dan Peng, Wei-Long Zheng, and Bao-Liang Lu. 2025. Enhancing Depression Detection from Emotion EEG with Temporal-Spatial-Spectral Representation Learning. In2025 47th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). IEEE, 1–5

    2025

  43. [43]

    François Perrin, Jacques Pernier, Olivier Bertrand, and Jean Francois Echallier

  44. [44]

    Spherical splines for scalp potential and current density mapping.Elec- troencephalography and clinical neurophysiology72, 2 (1989), 184–187

    1989

  45. [45]

    Walter HL Pinaya, Mark S Graham, Robert Gray, Pedro F Da Costa, Petru-Daniel Tudosiu, Paul Wright, Yee H Mah, Andrew D MacKinnon, James T Teo, Rolf Jager, et al. 2022. Fast unsupervised brain anomaly detection and segmentation with diffusion models. InInternational Conference on Medical Image Computing and Computer-Assisted Intervention. Springer, 705–714

    2022

  46. [46]

    U Raghavendra, Anjan Gudigar, Yashas Chakole, Praneet Kasula, DP Subha, Nahrizul Adib Kadri, Edward J Ciaccio, and U Rajendra Acharya. 2023. Automated detection and screening of depression using continuous wavelet transform with electroencephalogram signals.Expert Systems40, 4 (2023), e12803

    2023

  47. [47]

    Cédric Rommel, Joseph Paillard, Thomas Moreau, and Alexandre Gramfort. 2022. Data augmentation for learning predictive models on EEG: a systematic compar- ison.Journal of Neural Engineering19, 6 (2022), 066020

    2022

  48. [48]

    Saige Rutherford, Seyed Mostafa Kia, Thomas Wolfers, Charlotte Fraza, Mariam Zabihi, Richard Dinga, Pierre Berthet, Amanda Worker, Serena Verdi, Henri- cus G Ruhe, et al. 2022. The normative modeling framework for computational psychiatry.Nature protocols17, 7 (2022), 1711–1734. Beyond Augmentation: Score-Guided Pathological Prior for EEG-based Depression...

    2022

  49. [49]

    Ayan Seal, Rishabh Bajpai, Jagriti Agnihotri, Anis Yazidi, Enrique Herrera-Viedma, and Ondrej Krejcar. 2021. DeprNet: A deep convolution neural network frame- work for detecting depression using EEG.IEEE Transactions on Instrumentation and Measurement70 (2021), 1–13

    2021

  50. [50]

    Geetanjali Sharma, Amit M Joshi, Richa Gupta, and Linga Reddy Cenkeramaddi

  51. [51]

    DepCap: a smart healthcare framework for EEG based depression detection using time-frequency response and deep neural network.IEEE Access11 (2023), 52327–52338

    2023

  52. [52]

    Jian Shen, Kang Wang, Zeguang Zhao, Yanan Zhang, Fuze Tian, Xiaowei Zhang, Qunxi Dong, and Bin Hu. 2025. WDANet: Wasserstein Distribution Inspired Dynamic Adversarial Network for EEG-Based Cross-Domain Depression Recog- nition.IEEE Transactions on Affective Computing(2025)

    2025

  53. [53]

    Anurag Singh, Vivek Tiwari, Harshita Patel, GN Vivekananda, Dharmendra Singh Rajput, et al. 2024. Slitranet: an EEG-based automated diagnosis framework for major depressive disorder monitoring using a novel LGCN and transformer-based hybrid deep learning approach.IEEE Access12 (2024), 173109–173126

    2024

  54. [54]

    Yubing Sun, Jiaqi Sun, Zijian Zhou, Jing Cai, Wenjie Cui, and Guangda Liu. 2026. A Novel Spatial-Temporal Graph Neural Network for Major Depressive Disorder Detection Based on Resting State EEG Signals.IEEE Sensors Journal26, 6 (2026), 8660–8671. doi:10.1109/JSEN.2026.3657750

    work page doi:10.1109/jsen.2026.3657750 2026

  55. [55]

    Christian Szegedy, Wei Liu, Yangqing Jia, Pierre Sermanet, Scott Reed, Dragomir Anguelov, Dumitru Erhan, Vincent Vanhoucke, and Andrew Rabinovich. 2015. Going deeper with convolutions. InProceedings of the IEEE conference on computer vision and pattern recognition. 1–9

    2015

  56. [56]

    Christian Szegedy, Vincent Vanhoucke, Sergey Ioffe, Jon Shlens, and Zbigniew Wojna. 2016. Rethinking the inception architecture for computer vision. In Proceedings of the IEEE conference on computer vision and pattern recognition. 2818–2826

    2016

  57. [57]

    Dung Truong and Arnaud Delorme. 2025. Data Normalization Strategies for EEG Deep Learning.arXiv preprint arXiv:2506.22455(2025)

    work page arXiv 2025

  58. [58]

    Aaron Van Den Oord, Oriol Vinyals, et al. 2017. Neural discrete representation learning.Advances in neural information processing systems30 (2017)

    2017

  59. [59]

    Shruthi Narayanan Vaniya, Ahsan Habib, Maia Angelova, and Chandan Karmakar

  60. [60]

    Simplifying Depression Diagnosis: Single-Channel EEG and Deep Learning Approaches.IEEE Journal of Biomedical and Health Informatics(2026)

    2026

  61. [61]

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need.Advances in neural information processing systems30 (2017)

    2017

  62. [62]

    Hui Wang, Jinghui Yin, Siyuan Gao, Ju Liu, and Qiang Wu. 2024. Major depressive disorder detection using graph domain adaptation with global message-passing based on eeg signals.IEEE Transactions on Affective Computing16, 3 (2024), 1500–1513

    2024

  63. [63]

    Cheung, Jiangbo Pu, Sheng-Hua Zhong, Raymond Kai-Yu Tong, Ye Li, Michael Kwok-Po Ng, Kim-Fung Tsang, and Guanhua Ren

    Shuqiang Wang, Yi Guo, Yihang Dong, Yanyan Shen, Zhiguo Zhang, Albert C. Cheung, Jiangbo Pu, Sheng-Hua Zhong, Raymond Kai-Yu Tong, Ye Li, Michael Kwok-Po Ng, Kim-Fung Tsang, and Guanhua Ren. 2026. Generative AI Empowers Brain-Computer Interfaces: A Review-Perspective on Technical Realities and Future Visions.IEEE Transactions on Consumer Electronics72, 1 ...

    work page doi:10.1109/tce.2025.3650654 2026

  64. [64]

    Yuwen Wang, Yudan Peng, Mingxiu Han, Xinyi Liu, Haijun Niu, Jian Cheng, Suhua Chang, and Tao Liu. 2024. GCTNet: a graph convolutional transformer network for major depressive disorder detection based on EEG signals.Journal of Neural Engineering21, 3 (2024), 036042

    2024

  65. [65]

    Yilin Wang, Sha Zhao, Haiteng Jiang, Shijian Li, Benyan Luo, Tao Li, and Gang Pan. 2024. Diffmdd: A diffusion-based deep learning framework for mdd diagnosis using eeg.IEEE Transactions on Neural Systems and Rehabilitation Engineering32 (2024), 728–738

    2024

  66. [66]

    Xiaobin Wong, Zhonghua Zhao, Haoran Guo, Zhengyi Liu, Yu Wu, Feng Yan, Zhiren Wang, and Sen Song. 2026. CRCC: Contrast-Based Robust Cross-Subject and Cross-Site Representation Learning for EEG.arXiv preprint arXiv:2602.19138 (2026)

    work page arXiv 2026

  67. [67]

    Chenyue Xu, Longzhu Zhu, Jianbo Lai, Zeyu Luo, Jiayao Ying, Shaohua Hu, Peige Song, and Jianzhong Yang. 2026. Global and regional quality of care index in major depressive disorder: the global burden of disease study 2021.International Journal for Equity in Health(2026)

    2026

  68. [68]

    Guixun Xu, Wenhui Guo, and Yanjiang Wang. 2023. Subject-independent EEG emotion recognition with hybrid spatio-temporal GRU-Conv architecture.Medi- cal & Biological Engineering & Computing61, 1 (2023), 61–73

    2023

  69. [69]

    Wenchao Yang, Weidong Yan, Wenkang Liu, Yulan Ma, and Yang Li. 2025. THD- BAR: Topology Hierarchical Derived Brain Autoregressive Modeling for EEG Generic Representations.arXiv preprint arXiv:2511.13733(2025)

    work page arXiv 2025

  70. [70]

    Jintao Zhang, Zirui Liu, Mingyue Cheng, Xianquan Wang, Zhiding Liu, and Qi Liu. 2026. StaTS: Spectral Trajectory Schedule Learning for Adaptive Time Series Forecasting with Frequency Guided Denoiser.arXiv preprint arXiv:2603.00037 (2026)

    work page arXiv 2026

  71. [71]

    Jichang Zhang, Yuanjie Zheng, and Yunfeng Shi. 2023. A soft label method for medical image segmentation with multirater annotations.Computational Intelligence and Neuroscience2023, 1 (2023), 1883597

    2023

  72. [72]

    Yi-Dong Zhao, Yan-Kai Liu, Wei-Long Zheng, and Bao-Liang Lu. 2024. EEG data augmentation for emotion recognition using diffusion model. In2024 46th annual international conference of the ieee engineering in medicine and biology society (embc). IEEE, 1–4

    2024

  73. [73]

    Tong Zhou, Xuhang Chen, Yanyan Shen, Martin Nieuwoudt, Chi-Man Pun, and Shuqiang Wang. 2023. Generative ai enables eeg data augmentation for alzheimer’s disease detection via diffusion model. In2023 IEEE International Symposium on Product Compliance Engineering-Asia (ISPCE-ASIA). IEEE, 1–6

    2023

  74. [74]

    Ruslan Zhulduzbayev, Arian Ashourvan, Diana Arman, Alibek Bissembayev, and Almira Kustubayeva. 2026. A Structured Review of EEG-Based Machine Learning Approaches for Brain Age Prediction.Algorithms19, 1 (2026), 91

    2026