pith. sign in

arxiv: 2605.22379 · v1 · pith:ICWY3C62new · submitted 2026-05-21 · 💻 cs.HC · cs.AI· cs.LG

Cross-Subject EEG Emotion Recognition Based on Temporal Asynchronous Alignment Contrastive Learning

Ying Xie , Yi Zheng , Zehui Xiao , Wenkai Lu , Mengting Liu This is my paper

Pith reviewed 2026-05-22 04:08 UTC · model grok-4.3

classification 💻 cs.HC cs.AIcs.LG
keywords EEG emotion recognitioncross-subjectcontrastive learningtemporal alignmentlocal matchingTA2CLlate interaction
0
0 comments X

The pith

A contrastive learning method aligns local EEG segments across subjects to reduce timing and individual differences in emotion recognition.

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

The paper introduces TA2CL to fix temporal misalignment between EEG signals recorded from different people. Instead of matching entire signals at once, the approach searches for and pairs smaller locally correlated segments between subjects. This shift from global hard alignment to fine-grained local matching is meant to lessen the effects of personal variations and response delays. If the idea holds, EEG emotion systems could work more reliably without heavy per-person retraining. Readers in brain-computer interfaces would care because cross-subject generalization remains a major barrier to practical use.

Core claim

The paper claims that adapting the late interaction mechanism from ColBERT turns the similarity calculation from global hard alignment into fine-grained local matching. This lets the model adaptively search for and align locally highly correlated segments between EEG signals from different subjects, thereby mitigating inter-subject differences and temporal delays. The resulting TA2CL framework delivers 64.5 percent accuracy on nine-class and 79.5 percent on binary classification for the FACED dataset, plus 86.4 percent on SEED and 70.1 percent on SEED-V.

What carries the argument

Temporal Asynchronous Alignment-based Contrastive Learning (TA2CL) framework that performs fine-grained local matching of EEG segments instead of global hard alignment.

If this is right

  • The local matching strategy improves generalization across subjects by focusing on correlated segments rather than entire signals.
  • Performance reaches 64.5 percent nine-class and 79.5 percent binary accuracy on FACED, 86.4 percent on SEED, and 70.1 percent on SEED-V.
  • The method reduces the impact of temporal delays without requiring perfect synchronization between recordings.
  • Contrastive learning with late interaction becomes a viable route for other variable-timing biosignal tasks.

Where Pith is reading between the lines

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

  • The same local-alignment idea could transfer to other time-series domains such as speech or wearable sensor data where sources have mismatched timing.
  • Real-time EEG applications might need less precise clock synchronization if local segment search tolerates small delays.
  • Combining this approach with larger unlabeled EEG corpora could support semi-supervised training for broader emotion categories.

Load-bearing premise

Adaptively searching for and aligning locally highly correlated segments between EEG signals from different subjects will effectively mitigate inter-subject differences and temporal delays.

What would settle it

An ablation experiment that replaces the local segment matching step with standard global hard alignment and finds no accuracy drop or even an increase on the same cross-subject tasks would falsify the claimed benefit of the mechanism.

Figures

Figures reproduced from arXiv: 2605.22379 by Mengting Liu, Wenkai Lu, Ying Xie, Yi Zheng, Zehui Xiao.

Figure 1
Figure 1. Figure 1: Comparison of our method with previous methods. Previous methods calculate sample similarity based on [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Schematic diagram of Async-InfoNCE based on temporal token matching. Given an anchor sequence [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: The performance of different K values across various datasets [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Analysis of variance for T op3 token similarity under different K settings. The figure compares the variance of T op3 (ui , V ) pairs for K=1 and K=3 under both poor and good performance conditions. (a) A larger variance indicates more pronounced differences among high-ranking matches, suggesting that additional matches may introduce unstable information, making a smaller K value more appropriate. (b) A sm… view at source ↗
Figure 5
Figure 5. Figure 5: Confusion matrices of the datasets FACED, SEED, and SEEDV [PITH_FULL_IMAGE:figures/full_fig_p011_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Visualization of attention responses of DAEST and TA2CL on FACED-9 and SEED.The curves are obtained [PITH_FULL_IMAGE:figures/full_fig_p012_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Comparison of tSNE plots for our model (TA2CL) and the DEAST model on the FACED, SEED, and SEEDV [PITH_FULL_IMAGE:figures/full_fig_p014_7.png] view at source ↗
read the original abstract

With the advancement of science and technology, the importance of emotion research has become increasingly evident. Electroencephalography (EEG)-based emotion recognition has emerged as an active research area in recent years, owing to its objectivity and high temporal resolution. However, most existing methods focus on optimizing encoder structures to enhance feature extraction capabilities, while paying relatively little attention to similarity calculation strategies, particularly overlooking the potential temporal misalignment of responses among different subjects. To address these shortcomings, this paper draws inspiration from the late interaction mechanism of ColBERT in natural language processing (NLP) and proposes a Temporal Asynchronous Alignment-based Contrastive Learning (TA2CL) framework. This method transforms the traditional global "hard alignment" similarity calculation approach into a fine-grained local matching mechanism, enabling the model to adaptively search for and align "locally highly correlated" segments between two EEG signals, thereby effectively mitigating the effects of inter-subject differences and temporal delays. Experimental results demonstrate that the proposed method achieves strong performance across multiple public datasets. Specifically, on the FACED dataset, it achieves an accuracy of 64.5% for the nine-class classification task and 79.5% for the binary classification task, while on the SEED and SEED-V datasets, it achieves accuracies of 86.4% and 70.1%, respectively, validating the method's effectiveness and generalization capability.

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

1 major / 1 minor

Summary. The manuscript proposes a Temporal Asynchronous Alignment-based Contrastive Learning (TA2CL) framework for cross-subject EEG emotion recognition. Inspired by the late-interaction mechanism in ColBERT, the method replaces global hard alignment with adaptive fine-grained local matching of highly correlated EEG segments to mitigate inter-subject variability and temporal asynchrony. It reports accuracies of 64.5% (9-class) and 79.5% (binary) on FACED, 86.4% on SEED, and 70.1% on SEED-V, claiming improved generalization.

Significance. If the reported gains prove robust, the adaptation of local max-similarity matching from information retrieval to EEG signals offers a concrete, falsifiable mechanism for handling temporal misalignment in cross-subject settings. This could influence future work on physiological signal alignment beyond emotion recognition.

major comments (1)
  1. [Abstract] Abstract: The abstract states specific accuracies (64.5% 9-class and 79.5% binary on FACED; 86.4% on SEED; 70.1% on SEED-V) as evidence that local alignment mitigates inter-subject differences, yet supplies no information on data splits, baselines, statistical significance, error bars, or subject counts. These numbers are load-bearing for the central claim that the ColBERT-style mechanism drives the improvement.
minor comments (1)
  1. The abstract would be clearer if it briefly noted the number of subjects or recording conditions in each dataset to contextualize the cross-subject results.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We address the major comment point by point below and describe the planned revisions.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The abstract states specific accuracies (64.5% 9-class and 79.5% binary on FACED; 86.4% on SEED; 70.1% on SEED-V) as evidence that local alignment mitigates inter-subject differences, yet supplies no information on data splits, baselines, statistical significance, error bars, or subject counts. These numbers are load-bearing for the central claim that the ColBERT-style mechanism drives the improvement.

    Authors: We agree that the abstract would benefit from additional context to better support the reported results and the central claim regarding the temporal asynchronous alignment mechanism. In the revised version, we will expand the abstract to note the subject counts (123 subjects in FACED, 15 in SEED, 16 in SEED-V), the use of leave-one-subject-out cross-validation for subject-independent evaluation, and that the accuracies represent means across folds with standard deviations and statistical comparisons to baselines (including recent contrastive methods) provided in the experimental section. These details will clarify that the gains are evaluated under rigorous cross-subject protocols. The full experimental setup, including data splits, baselines, and significance testing, is already described in Sections 3 and 4; we will ensure the abstract references this context more explicitly. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper introduces the TA2CL framework by adapting ColBERT-style late interaction for local segment alignment in cross-subject EEG signals. All reported results (e.g., 64.5% 9-class accuracy on FACED) are presented as empirical experimental outcomes on public datasets rather than quantities obtained by fitting parameters inside the same equations or by self-referential definitions. No derivation step reduces the claimed performance to an input by construction, and the central mechanism is a falsifiable extension of an external retrieval technique without load-bearing self-citation chains or ansatz smuggling.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review provides no explicit free parameters, axioms, or invented entities. The central claim rests on the unstated assumption that local segment correlation exists and can be reliably identified in EEG data across subjects, plus standard supervised learning assumptions about labeled emotion classes.

pith-pipeline@v0.9.0 · 5786 in / 1371 out tokens · 28583 ms · 2026-05-22T04:08:50.064893+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

46 extracted references · 46 canonical work pages

  1. [1]

    Affective computing: challenges,

    R. W. Picard, “Affective computing: challenges,”International journal of human-computer studies, vol. 59, no. 1-2, pp. 55–64, 2003

    work page 2003

  2. [2]

    Mental health monitoring with multimodal sensing and machine learning: A survey,

    E. Garcia-Ceja, M. Riegler, T. Nordgreen, P. Jakobsen, K. J. Oedegaard, and J. Tørresen, “Mental health monitoring with multimodal sensing and machine learning: A survey,”Pervasive and Mobile Computing, vol. 51, pp. 1–26, 2018

    work page 2018

  3. [3]

    Deep learning for electroencephalogram (eeg) classification tasks: a review,

    A. Craik, Y . He, and J. L. Contreras-Vidal, “Deep learning for electroencephalogram (eeg) classification tasks: a review,”Journal of neural engineering, vol. 16, no. 3, p. 031001, 2019

    work page 2019

  4. [4]

    Accelerating 3d convolutional neural network with channel bottleneck module for eeg-based emotion recognition,

    S. Kim, T.-S. Kim, and W. H. Lee, “Accelerating 3d convolutional neural network with channel bottleneck module for eeg-based emotion recognition,”Sensors, vol. 22, no. 18, p. 6813, 2022

    work page 2022

  5. [5]

    Eeg-based emotion recognition using transfer learning based feature extraction and convolutional neural network,

    V . Jadhav, N. Tiwari, and M. Chawla, “Eeg-based emotion recognition using transfer learning based feature extraction and convolutional neural network,” inITM web of conferences, vol. 53. EDP Sciences, 2023, p. 02011

    work page 2023

  6. [6]

    Emotion recognition from multi-channel eeg through parallel convolutional recurrent neural network,

    Y . Yang, Q. Wu, M. Qiu, Y . Wang, and X. Chen, “Emotion recognition from multi-channel eeg through parallel convolutional recurrent neural network,” in2018 international joint conference on neural networks (IJCNN). IEEE, 2018, pp. 1–7

    work page 2018

  7. [7]

    Multi-modal dimensional emotion recognition using recurrent neural networks,

    S. Chen and Q. Jin, “Multi-modal dimensional emotion recognition using recurrent neural networks,” inProceed- ings of the 5th International Workshop on Audio/Visual Emotion Challenge, 2015, pp. 49–56. 14 Xieet al.: TA2CL

    work page 2015

  8. [8]

    Context matters: situational stress impedes functional reorganization of intrinsic brain connectivity during problem-solving,

    M. Liu, R. A. Backer, R. C. Amey, E. E. Splan, A. Magerman, and C. E. Forbes, “Context matters: situational stress impedes functional reorganization of intrinsic brain connectivity during problem-solving,”Cerebral Cortex, vol. 31, no. 4, pp. 2111–2124, 2021

    work page 2021

  9. [9]

    How the brain negotiates divergent executive processing demands: evidence of network reorganization in fleeting brain states,

    M. Liu, R. A. Backer, R. C. Amey, and C. E. Forbes, “How the brain negotiates divergent executive processing demands: evidence of network reorganization in fleeting brain states,”Neuroimage, vol. 245, p. 118653, 2021

    work page 2021

  10. [10]

    Surfgnn: A robust surface-based prediction model with interpretability for coactivation maps of spatial and cortical features,

    Z. Li, J. Zhang, Y . Zeng, J. Lin, D. Zhang, J. Zhang, D. Xu, H. Kim, B. Liu, and M. Liu, “Surfgnn: A robust surface-based prediction model with interpretability for coactivation maps of spatial and cortical features,”Medical Image Analysis, p. 103793, 2025

    work page 2025

  11. [11]

    Driver emotion recognition for safe driving: A comprehensive survey,

    T. Gamage, E. Sandamali, and P. Kalansooriya, “Driver emotion recognition for safe driving: A comprehensive survey,” in15th International Research Conference of General Sir John Kotelawala Defence University, 2022, pp. 95–101

    work page 2022

  12. [12]

    Emotions recognition using eeg signals: A survey,

    S. M. Alarcao and M. J. Fonseca, “Emotions recognition using eeg signals: A survey,”IEEE transactions on affective computing, vol. 10, no. 3, pp. 374–393, 2017

    work page 2017

  13. [13]

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

    N. S. Suhaimi, J. Mountstephens, and J. Teo, “Eeg-based emotion recognition: a state-of-the-art review of current trends and opportunities,”Computational intelligence and neuroscience, vol. 2020, no. 1, p. 8875426, 2020

    work page 2020

  14. [14]

    Deep learning-based eeg emotion recognition: Current trends and future perspectives,

    X. Wang, Y . Ren, Z. Luo, W. He, J. Hong, and Y . Huang, “Deep learning-based eeg emotion recognition: Current trends and future perspectives,”Frontiers in psychology, vol. 14, p. 1126994, 2023

    work page 2023

  15. [15]

    Contrastive representation learning for electroencephalogram classification,

    M. N. Mohsenvand, M. R. Izadi, and P. Maes, “Contrastive representation learning for electroencephalogram classification,” inMachine learning for health. PMLR, 2020, pp. 238–253

    work page 2020

  16. [16]

    Contrastive learning of subject-invariant eeg representations for cross-subject emotion recognition,

    X. Shen, X. Liu, X. Hu, D. Zhang, and S. Song, “Contrastive learning of subject-invariant eeg representations for cross-subject emotion recognition,”IEEE Transactions on Affective Computing, vol. 14, no. 3, pp. 2496–2511, 2022

    work page 2022

  17. [17]

    A simple framework for contrastive learning of visual representations,

    T. Chen, S. Kornblith, M. Norouzi, and G. Hinton, “A simple framework for contrastive learning of visual representations,” inInternational conference on machine learning. PmLR, 2020, pp. 1597–1607

    work page 2020

  18. [18]

    Spatial-temporal transformer with curriculum learning for eeg-based emotion recognition,

    X. Lin, T. Peng, P. Dai, Y . Liang, and W. Wu, “Spatial-temporal transformer with curriculum learning for eeg-based emotion recognition,”arXiv preprint arXiv:2507.14698, 2025

    work page arXiv 2025

  19. [19]

    Dynamic-attention-based eeg state transition modeling for emotion recognition,

    X. Shen, R. Gan, K. Wang, S. Yang, Q. Zhang, Q. Liu, D. Zhang, and S. Song, “Dynamic-attention-based eeg state transition modeling for emotion recognition,”IEEE Transactions on Affective Computing, 2025

    work page 2025

  20. [20]

    Time-series representation learning via temporal and contextual contrasting.arXiv preprint arXiv:2106.14112,

    E. Eldele, M. Ragab, Z. Chen, M. Wu, C. K. Kwoh, X. Li, and C. Guan, “Time-series representation learning via temporal and contextual contrasting,”arXiv preprint arXiv:2106.14112, 2021

    work page arXiv 2021

  21. [21]

    Sentence-bert: Sentence embeddings using siamese bert-networks,

    N. Reimers and I. Gurevych, “Sentence-bert: Sentence embeddings using siamese bert-networks,” inProceedings of the 2019 conference on empirical methods in natural language processing and the 9th international joint conference on natural language processing (EMNLP-IJCNLP), 2019, pp. 3982–3992

    work page 2019

  22. [22]

    Poly-encoders: Transformer architectures and pre-training strategies for fast and accurate multi-sentence scoring,

    S. Humeau, K. Shuster, M.-A. Lachaux, and J. Weston, “Poly-encoders: Transformer architectures and pre-training strategies for fast and accurate multi-sentence scoring,”arXiv preprint arXiv:1905.01969, 2019

    work page arXiv 1905

  23. [23]

    Colbert: Efficient and effective passage search via contextualized late interaction over bert,

    O. Khattab and M. Zaharia, “Colbert: Efficient and effective passage search via contextualized late interaction over bert,” inProceedings of the 43rd International ACM SIGIR conference on research and development in Information Retrieval, 2020, pp. 39–48

    work page 2020

  24. [24]

    Eeg conformer: Convolutional transformer for eeg decoding and visualization,

    Y . Song, Q. Zheng, B. Liu, and X. Gao, “Eeg conformer: Convolutional transformer for eeg decoding and visualization,”IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. 31, pp. 710–719, 2022

    work page 2022

  25. [25]

    Analysis and compensation of the reaction lag of evaluators in continuous emotional annotations,

    S. Mariooryad and C. Busso, “Analysis and compensation of the reaction lag of evaluators in continuous emotional annotations,” in2013 Humaine Association Conference on Affective Computing and Intelligent Interaction. IEEE, 2013, pp. 85–90

    work page 2013

  26. [26]

    Unsupervised scalable representation learning for multivariate time series,

    J.-Y . Franceschi, A. Dieuleveut, and M. Jaggi, “Unsupervised scalable representation learning for multivariate time series,”Advances in neural information processing systems, vol. 32, 2019

    work page 2019

  27. [27]

    Soft-dtw: a differentiable loss function for time-series,

    M. Cuturi and M. Blondel, “Soft-dtw: a differentiable loss function for time-series,” inInternational conference on machine learning. PMLR, 2017, pp. 894–903

    work page 2017

  28. [28]

    Personalizing eeg-based affective models with transfer learning,

    W.-L. Zheng and B.-L. Lu, “Personalizing eeg-based affective models with transfer learning,” inProceedings of the twenty-fifth international joint conference on artificial intelligence, 2016, pp. 2732–2738

    work page 2016

  29. [29]

    Domain adaptation for eeg emotion recognition based on latent representation similarity,

    J. Li, S. Qiu, C. Du, Y . Wang, and H. He, “Domain adaptation for eeg emotion recognition based on latent representation similarity,”IEEE Transactions on Cognitive and Developmental Systems, vol. 12, no. 2, pp. 344–353, 2019. 15 Xieet al.: TA2CL

    work page 2019

  30. [30]

    Domain-generalized deep learning for improved subject-independent emotion recognition based on electroencephalography,

    J.-H. Kim, H. Nam, D. Won, and C.-H. Im, “Domain-generalized deep learning for improved subject-independent emotion recognition based on electroencephalography,”Experimental Neurobiology, vol. 34, no. 3, p. 119, 2025

    work page 2025

  31. [31]

    A transformer framework based on self-supervised contrastive learning for eeg-based emotion recognition,

    H. Wen and T. Yu, “A transformer framework based on self-supervised contrastive learning for eeg-based emotion recognition,” in2024 7th International Conference on Advanced Algorithms and Control Engineering (ICAACE). IEEE, 2024, pp. 313–317

    work page 2024

  32. [32]

    Supervised contrastive learning for eeg-based cross-subject emotion recognition,

    H. Wang and L. Song, “Supervised contrastive learning for eeg-based cross-subject emotion recognition,” in2023 3rd International Conference on Networking Systems of AI (INSAI). IEEE, 2023, pp. 401–405

    work page 2023

  33. [33]

    Gmss: Graph-based multi-task self-supervised learning for eeg emotion recognition,

    Y . Li, J. Chen, F. Li, B. Fu, H. Wu, Y . Ji, Y . Zhou, Y . Niu, G. Shi, and W. Zheng, “Gmss: Graph-based multi-task self-supervised learning for eeg emotion recognition,”IEEE Transactions on Affective Computing, vol. 14, no. 3, pp. 2512–2525, 2022

    work page 2022

  34. [34]

    One fits all: Power general time series analysis by pretrained lm,

    T. Zhou, P. Niu, L. Sun, R. Jinet al., “One fits all: Power general time series analysis by pretrained lm,”Advances in neural information processing systems, vol. 36, pp. 43 322–43 355, 2023

    work page 2023

  35. [35]

    Physiosync: Temporal and cross-modal contrastive learning inspired by physiological synchronization for eeg-based emotion recognition,

    K. Cui, J. Li, Y . Liu, X. Zhang, Z. Hu, and M. Wang, “Physiosync: Temporal and cross-modal contrastive learning inspired by physiological synchronization for eeg-based emotion recognition,”IEEE Transactions on Computational Social Systems, 2025

    work page 2025

  36. [36]

    Estimating workload using eeg spectral power and erps in the n-back task,

    A.-M. Brouwer, M. A. Hogervorst, J. B. Van Erp, T. Heffelaar, P. H. Zimmerman, and R. Oostenveld, “Estimating workload using eeg spectral power and erps in the n-back task,”Journal of neural engineering, vol. 9, no. 4, p. 045008, 2012

    work page 2012

  37. [37]

    A large finer-grained affective computing eeg dataset,

    J. Chen, X. Wang, C. Huang, X. Hu, X. Shen, and D. Zhang, “A large finer-grained affective computing eeg dataset,”Scientific Data, vol. 10, no. 1, p. 740, 2023

    work page 2023

  38. [38]

    Investigating critical frequency bands and channels for eeg-based emotion recognition with deep neural networks,

    W.-L. Zheng and B.-L. Lu, “Investigating critical frequency bands and channels for eeg-based emotion recognition with deep neural networks,”IEEE Transactions on autonomous mental development, vol. 7, no. 3, pp. 162–175, 2015

    work page 2015

  39. [39]

    Differential entropy feature for eeg-based emotion classification,

    R.-N. Duan, J.-Y . Zhu, and B.-L. Lu, “Differential entropy feature for eeg-based emotion classification,” in2013 6th international IEEE/EMBS conference on neural engineering (NER). IEEE, 2013, pp. 81–84

    work page 2013

  40. [40]

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

    W. Liu, J.-L. Qiu, W.-L. Zheng, and B.-L. Lu, “Comparing recognition performance and robustness of multimodal deep learning models for multimodal emotion recognition,”IEEE Transactions on Cognitive and Developmental Systems, vol. 14, no. 2, pp. 715–729, 2021

    work page 2021

  41. [41]

    Hierarchical multi-label classification networks,

    J. Wehrmann, R. Cerri, and R. Barros, “Hierarchical multi-label classification networks,” inInternational conference on machine learning. PMLR, 2018, pp. 5075–5084

    work page 2018

  42. [42]

    A circumplex model of affect

    J. A. Russell, “A circumplex model of affect.”Journal of personality and social psychology, vol. 39, no. 6, p. 1161, 1980

    work page 1980

  43. [43]

    Plug-and-play domain adaptation for cross-subject eeg-based emotion recognition,

    L.-M. Zhao, X. Yan, and B.-L. Lu, “Plug-and-play domain adaptation for cross-subject eeg-based emotion recognition,” inProceedings of the AAAI conference on artificial intelligence, vol. 35, no. 1, 2021, pp. 863–870

    work page 2021

  44. [44]

    Transfer components between subjects for eeg-based emotion recognition,

    W.-L. Zheng, Y .-Q. Zhang, J.-Y . Zhu, and B.-L. Lu, “Transfer components between subjects for eeg-based emotion recognition,” in2015 international conference on affective computing and intelligent interaction (ACII). IEEE, 2015, pp. 917–922

    work page 2015

  45. [45]

    Application of functional connectivity to cognitive neurosciences,

    C. Forbes and M. Liu, “Application of functional connectivity to cognitive neurosciences,” inFunctional Connec- tivity of the Human Brain. Elsevier, 2026, pp. 139–166

    work page 2026

  46. [46]

    Adaptive node feature extraction in graph-based neural networks for brain diseases diagnosis using self-supervised learning,

    Y . Zeng, J. Lin, Z. Li, Z. Xiao, C. Wang, X. Ge, C. Wang, G. Huang, and M. Liu, “Adaptive node feature extraction in graph-based neural networks for brain diseases diagnosis using self-supervised learning,”NeuroImage, vol. 297, p. 120750, 2024. 16

    work page 2024