pith. sign in

arxiv: 2605.00856 · v3 · pith:CVDVIONHnew · submitted 2026-04-21 · 📡 eess.SP · cs.AI· cs.HC· cs.LG

One-Block Transformer (1BT) for EEG-Based Cognitive Workload Assessment

Stefanos Gkikas , Christian Arzate Cruz , Thomas Kassiotis , Giorgos Giannakakis , Raul Fernandez Rojas , Randy Gomez This is my paper

Pith reviewed 2026-05-21 01:12 UTC · model grok-4.3

classification 📡 eess.SP cs.AIcs.HCcs.LG
keywords EEGcognitive workloadtransformerclassificationefficient modelreal-time monitoringbrain-computer interfaceworkload assessment
0
0 comments X

The pith

A minimal one-block transformer classifies cognitive workload from EEG with under 0.5 million parameters and 0.02 GFLOPs.

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

This paper introduces the One-Block Transformer as a compact model for continuous cognitive workload assessment from multi-channel EEG signals. It processes temporal sequences through a minimal latent bottleneck, a single cross-attention module, and lightweight self-attention to keep the architecture small. Experiments involved 11 participants completing abstract reasoning, numerical problem-solving, and interactive video game tasks at two workload levels. The design maintains strong classification accuracy while slashing computational demands, which matters for running such monitoring on everyday hardware without heavy processors. Systematic tests pinpoint the smallest configuration that still delivers reliable results.

Core claim

The paper establishes that the 1BT architecture aggregates multi-channel EEG temporal sequences via a minimal latent bottleneck using one cross-attention module followed by lightweight self-attention, achieving high workload classification performance with under 0.5 million parameters and 0.02 GFLOPs to support real-time monitoring in resource-constrained settings.

What carries the argument

The One-Block Transformer (1BT), which aggregates multi-channel temporal EEG sequences via a minimal latent bottleneck using a single cross-attention module followed by lightweight self-attention.

If this is right

  • The compact model enables real-time cognitive workload monitoring on devices with limited processing power.
  • Architectural analysis identifies the smallest viable configurations that retain high classification accuracy.
  • Substantially lower computational costs support practical deployment in adaptive human-machine systems.
  • Continuous EEG-based workload estimation becomes feasible without requiring high-end hardware.

Where Pith is reading between the lines

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

  • Similar lightweight designs could support wearable EEG headsets for ongoing monitoring during daily work or learning tasks.
  • The approach might extend to related brain-signal tasks such as detecting fatigue or sustained attention in operational settings.
  • Integration with other physiological signals could create multi-modal systems that improve robustness in varied conditions.

Load-bearing premise

The EEG patterns recorded from 11 participants performing three specific tasks at two workload levels are representative enough for the model to generalize to real-world continuous monitoring across diverse users and environments.

What would settle it

Testing the trained model on EEG data collected from a larger and more varied group of participants during uncontrolled everyday activities would reveal whether the reported classification performance generalizes.

Figures

Figures reproduced from arXiv: 2605.00856 by Christian Arzate Cruz, Giorgos Giannakakis, Randy Gomez, Raul Fernandez Rojas, Stefanos Gkikas, Thomas Kassiotis.

Figure 1
Figure 1. Figure 1: Overview of the study pipeline: (left) EEG data acquisition with the [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of the proposed transformer-based model and [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
read the original abstract

Accurate and continuous estimation of cognitive workload is fundamental to creating adaptive human-machine systems. However, designing architectures that balance representational capacity with computational efficiency has been challenging for practical deployment. This paper introduces 1BT, a One-Block Transformer for compact and efficient EEG-based cognitive workload assessment. The model aggregates multi-channel temporal sequences via a minimal latent bottleneck, using a single cross-attention module followed by lightweight self-attention. A controlled study involving 11 participants performing three cognitively diverse tasks (abstract reasoning, numerical problem-solving, and an interactive video game) was conducted with continuous EEG recordings across two workload levels. Systematic architectural analysis identifies the most compact configuration that preserves high performance, while substantially lowering computational cost. The final model achieves high workload classification performance with under 0.5 million parameters and 0.02 GFLOPs, paving the way for a design direction for real-time cognitive workload monitoring in resource-constrained settings.

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

2 major / 2 minor

Summary. The paper introduces the One-Block Transformer (1BT), a compact EEG processing architecture that uses a single cross-attention module with a latent bottleneck followed by lightweight self-attention to perform binary cognitive workload classification. It reports results from a controlled study of 11 participants completing abstract reasoning, numerical problem-solving, and interactive game tasks at two workload levels, with systematic architectural search yielding a final model under 0.5 million parameters and 0.02 GFLOPs that achieves high classification performance.

Significance. If the reported performance holds under rigorous validation, the work offers a useful design direction for low-compute, real-time EEG monitoring on resource-constrained hardware. The focus on minimizing parameters and FLOPs while preserving accuracy addresses a practical bottleneck in deploying adaptive human-machine systems.

major comments (2)
  1. [Experimental Evaluation] Experimental Evaluation section: With only 11 participants and three discrete, short-duration tasks, the study does not report subject-independent (e.g., leave-one-subject-out) cross-validation or any out-of-distribution testing on continuous or naturalistic monitoring scenarios; this directly undermines the claim that the model paves the way for deployment across diverse users and environments.
  2. [Results] Results section, performance table: Specific accuracy, F1-score, and statistical significance values (with confidence intervals or p-values) must be provided alongside baseline comparisons (CNN, LSTM, or standard transformer variants) to substantiate that the efficiency gains do not come at the cost of meaningful performance degradation.
minor comments (2)
  1. [Model Architecture] The description of the latent bottleneck dimension and attention hyperparameters should include the exact values used in the final model and a brief ablation table showing sensitivity.
  2. Figure captions for the architectural diagram and performance plots should explicitly state the input EEG channel count, sampling rate, and window length to improve reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which have helped us improve the clarity and rigor of the manuscript. We provide point-by-point responses below and have revised the paper accordingly.

read point-by-point responses

  1. Referee: [Experimental Evaluation] Experimental Evaluation section: With only 11 participants and three discrete, short-duration tasks, the study does not report subject-independent (e.g., leave-one-subject-out) cross-validation or any out-of-distribution testing on continuous or naturalistic monitoring scenarios; this directly undermines the claim that the model paves the way for deployment across diverse users and environments.

    Authors: We appreciate the referee highlighting the importance of subject-independent validation for claims of broader applicability. The original study was a controlled experiment with 11 participants across three cognitively diverse tasks to enable systematic architectural search. In the revised manuscript we have added leave-one-subject-out cross-validation results to the Experimental Evaluation section, confirming consistent performance. We fully acknowledge the absence of out-of-distribution testing on continuous naturalistic data as a genuine limitation and have inserted a new Limitations subsection in the Discussion that explicitly discusses this gap and outlines future work. We have also moderated the language in the abstract and conclusion to emphasize the architectural design direction rather than immediate deployment readiness. revision: partial

  2. Referee: [Results] Results section, performance table: Specific accuracy, F1-score, and statistical significance values (with confidence intervals or p-values) must be provided alongside baseline comparisons (CNN, LSTM, or standard transformer variants) to substantiate that the efficiency gains do not come at the cost of meaningful performance degradation.

    Authors: We agree that explicit quantitative metrics and baseline comparisons are required to support the efficiency claims. The revised Results section now contains an expanded performance table that reports accuracy, F1-score, 95% confidence intervals, and p-values (from paired statistical tests) for the final 1BT model alongside CNN, LSTM, and standard transformer baselines. These additions demonstrate that the substantial reductions in parameters and FLOPs are achieved without meaningful degradation in classification performance relative to the baselines. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical architecture proposal with independent evaluation

full rationale

The paper introduces an empirical neural architecture (1BT) for EEG workload classification and reports its performance on a controlled study with 11 participants. No mathematical derivation chain, uniqueness theorem, or self-citation load-bearing premise is present. The model is defined by explicit architectural choices (single cross-attention + lightweight self-attention with latent bottleneck), then evaluated on held-out data; performance metrics are not redefined as predictions by construction, nor do any equations reduce to fitted inputs. The contribution remains a standard empirical proposal whose validity rests on external test-set results rather than internal redefinition.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The model rests on standard transformer components adapted with a custom single-block structure. No new physical entities are introduced. The main domain assumption is that the chosen tasks reliably induce measurable differences in cognitive workload via EEG.

free parameters (1)
  • latent bottleneck dimension and attention hyperparameters
    The architecture description implies several design choices (bottleneck size, number of heads, embedding dimensions) that are typically selected or tuned during development.
axioms (1)
  • domain assumption EEG signals recorded during the three tasks accurately reflect two distinct levels of cognitive workload
    The study design assumes the abstract reasoning, numerical, and game tasks produce reliably separable workload states detectable in EEG.

pith-pipeline@v0.9.0 · 5715 in / 1162 out tokens · 36419 ms · 2026-05-21T01:12:57.301505+00:00 · methodology

Review history (2 revisions) →

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

49 extracted references · 49 canonical work pages

  1. [1]

    Data augmentation for 3dmm-based arousal-valence prediction for hri,

    C. A. Cruz, Y . Sechayk, T. Igarashi, and R. Gomez, “Data augmentation for 3dmm-based arousal-valence prediction for hri,” in2024 33rd IEEE International Conference on Robot and Human Interactive Communica- tion (ROMAN), 2024, pp. 2015–2022

    work page 2024

  2. [2]

    Clare: Cognitive load assessment in real-time with multimodal data,

    A. Bhatti, P. Angkan, B. Behinaein, Z. Mahmud, D. Rodenburg, H. Braund, P. J. Mclellan, A. Ruberto, G. Harrison, D. Wilson, A. Szulewski, D. Howes, A. Etemad, and P. Hungler, “Clare: Cognitive load assessment in real-time with multimodal data,”IEEE Transactions on Cognitive and Developmental Systems, vol. 17, no. 6, pp. 1337–1349, 2025

    work page 2025

  3. [3]

    Cognitive load estimation in the wild,

    L. Fridman, B. Reimer, B. Mehler, and W. T. Freeman, “Cognitive load estimation in the wild,” inProceedings of the 2018 CHI Conference on Human Factors in Computing Systems, ser. CHI ’18. New York, NY , USA: Association for Computing Machinery, 2018, p. 1–9

    work page 2018

  4. [4]

    Unobtrusive measurement of cognitive load and physiological signals in uncontrolled environments,

    C. Anders, S. Moontaha, S. Real, and B. Arnrich, “Unobtrusive measurement of cognitive load and physiological signals in uncontrolled environments,”Scientific Data, vol. 11, no. 1, p. 1000, 2024. TABLE II: Performance comparison for different numbers of self-attention heads across three tasks. Task Model Configuration Computational Cost Performance #Laten...

    work page 2024

  5. [5]

    Ecg and eeg based machine learning models for the classification of mental workload and stress levels for women in different menstrual phases, men, and mixed sexes,

    A. Hemakom, D. Atiwiwat, and P. Israsena, “Ecg and eeg based machine learning models for the classification of mental workload and stress levels for women in different menstrual phases, men, and mixed sexes,” Biomedical Signal Processing and Control, vol. 95, p. 106379, 2024

    work page 2024

  6. [6]

    Gendered mental labor: A systematic literature review on the cognitive dimension of unpaid work within the household and childcare,

    N. Reich-Stiebert, L. Froehlich, and J.-B. V oltmer, “Gendered mental labor: A systematic literature review on the cognitive dimension of unpaid work within the household and childcare,”Sex Roles, vol. 88, no. 11, pp. 475–494, 2023

    work page 2023

  7. [7]

    Twifly: A data analysis framework for twitter,

    P. Chatziadam, A. Dimitriadis, S. Gikas, I. Logothetis, M. Michalodim- itrakis, M. Neratzoulakis, A. Papadakis, V . Kontoulis, N. Siganos, D. Theodoropoulos, G. V ougioukalos, I. Hatzakis, G. Gerakis, N. Pa- padakis, and H. Kondylakis, “Twifly: A data analysis framework for twitter,”Information, vol. 11, no. 5, 2020

    work page 2020

  8. [8]

    Empatheticexchanges: Toward understanding the cues for empathy in dyadic conversations,

    E. C. Montiel-Vazquez, C. Arzate Cruz, J. A. R. Uresti, and R. Gomez, “Empatheticexchanges: Toward understanding the cues for empathy in dyadic conversations,”IEEE Access, vol. 12, pp. 195 097–195 110, 2024

    work page 2024

  9. [9]

    The cost of work-related stress to society: A systematic review

    J. Hassard, K. R. Teoh, G. Visockaite, P. Dewe, and T. Cox, “The cost of work-related stress to society: A systematic review.”Journal of occupational health psychology, vol. 23, no. 1, p. 1, 2018

    work page 2018

  10. [10]

    Painformer: A vision foundation model for automatic pain assessment,

    S. Gkikas, R. F. Rojas, and M. Tsiknakis, “Painformer: A vision foundation model for automatic pain assessment,”IEEE Transactions on Affective Computing, vol. 16, no. 4, pp. 3369–3386, 2025

    work page 2025

  11. [11]

    Mul- timodal automatic assessment of acute pain through facial videos and heart rate signals utilizing transformer-based architectures,

    S. Gkikas, N. S. Tachos, S. Andreadis, V . C. Pezoulas, D. Zaridis, G. Gkois, A. Matonaki, T. G. Stavropoulos, and D. I. Fotiadis, “Mul- timodal automatic assessment of acute pain through facial videos and heart rate signals utilizing transformer-based architectures,”Frontiers in Pain Research, vol. 5, 2024

    work page 2024

  12. [12]

    A pain assessment framework based on multimodal data and deep machine learning methods,

    S. Gkikas, “A pain assessment framework based on multimodal data and deep machine learning methods,” 2025, arXiv preprint arXiv:2505.05396. [Online]. Available: https://arxiv.org/abs/2505.05396

    work page arXiv 2025

  13. [13]

    A lightweight transformer for pain recognition from brain activity,

    S. Gkikas, C. A. Cruz, Y . Fang, L. Cao, M. U. Khan, T. Kassiotis, G. Giannakakis, R. F. Rojas, and R. Gomez, “A lightweight transformer for pain recognition from brain activity,” 2026

    work page 2026

  14. [14]

    Electroencephalographic workload indicators during teleoperation of an unmanned aerial vehicle shepherding a swarm of unmanned ground vehicles in contested environments,

    R. Fernandez Rojas, E. Debie, J. Fidock, M. Barlow, K. Kasmarik, S. Ana- vatti, M. Garratt, and H. Abbass, “Electroencephalographic workload indicators during teleoperation of an unmanned aerial vehicle shepherding a swarm of unmanned ground vehicles in contested environments,” Frontiers in Neuroscience, vol. V olume 14 - 2020, 2020

    work page 2020

  15. [15]

    Measuring mental workload with the nasa-tlx needs to examine each dimension rather than relying on the global score: an example with driving,

    E. Galy, J. Paxion, and C. Berthelon, “Measuring mental workload with the nasa-tlx needs to examine each dimension rather than relying on the global score: an example with driving,”Ergonomics, vol. 61, no. 4, pp. 517–527, 2018

    work page 2018

  16. [16]

    Measuring cognitive workload using multimodal sensors,

    N. Hirachan, A. Mathews, J. Romero, and R. F. Rojas, “Measuring cognitive workload using multimodal sensors,” in2022 44th annual international conference of the IEEE engineering in medicine & biology society (EMBC). IEEE, 2022, pp. 4921–4924

    work page 2022

  17. [17]

    Measuring mental workload using physiological measures: A systematic review,

    R. L. Charles and J. Nixon, “Measuring mental workload using physiological measures: A systematic review,”Applied Ergonomics, vol. 74, pp. 221–232, 2019

    work page 2019

  18. [18]

    Electroencephalogram-based mental workload prediction for using augmented reality head mounted display in construc- tion assembly: A deep learning approach,

    Y . Qin and T. Bulbul, “Electroencephalogram-based mental workload prediction for using augmented reality head mounted display in construc- tion assembly: A deep learning approach,”Automation in Construction, vol. 152, p. 104892, 2023

    work page 2023

  19. [19]

    Eeg-based estimation of cognitive workload across multiple tasks,

    A. S. Mathews, N. Hirachan, C. Joseph, M. Ghahramani, J. Lopez- Aparicio, and R. F. Rojas, “Eeg-based estimation of cognitive workload across multiple tasks,” in2024 46th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC). IEEE, 2024, pp. 1–4

    work page 2024

  20. [20]

    Neural networks meet neural activity: Utilizing eeg for mental workload estimation,

    G. Siddhad, P. P. Roy, and B.-G. Kim, “Neural networks meet neural activity: Utilizing eeg for mental workload estimation,” inPattern Recognition: 27th International Conference, ICPR 2024, Kolkata, India, December 1–5, 2024, Proceedings, Part XI. Berlin, Heidelberg: Springer- Verlag, 2024, p. 325–339

    work page 2024

  21. [21]

    A full transformer-based framework for automatic pain estimation using videos,

    S. Gkikas and M. Tsiknakis, “A full transformer-based framework for automatic pain estimation using videos,” in2023 45th Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), 2023, pp. 1–6

    work page 2023

  22. [22]

    Twins-painvit: Towards a modality-agnostic vision transformer framework for multimodal automatic pain assessment using facial videos and fnirs,

    ——, “Twins-painvit: Towards a modality-agnostic vision transformer framework for multimodal automatic pain assessment using facial videos and fnirs,” in2024 12th International Conference on Affective Computing and Intelligent Interaction Workshops and Demos (ACIIW), 2024, pp. 13–21

    work page 2024

  23. [23]

    Synthetic thermal and rgb videos for automatic pain assessment utilizing a vision-mlp architecture,

    ——, “Synthetic thermal and rgb videos for automatic pain assessment utilizing a vision-mlp architecture,” in2024 12th International Conference on Affective Computing and Intelligent Interaction Workshops and Demos (ACIIW), 2024, pp. 4–12

    work page 2024

  24. [24]

    Automatic pain intensity estimation based on electrocardiogram and demographic factors

    S. Gkikas., C. Chatzaki., E. Pavlidou., F. Verigou., K. Kalkanis., and M. Tsiknakis., “Automatic pain intensity estimation based on electrocardiogram and demographic factors.” SciTePress, 2022, pp. 155–162

    work page 2022

  25. [25]

    Multi-task neural networks for pain intensity estimation using electrocardiogram and demographic factors,

    S. Gkikas, C. Chatzaki, and M. Tsiknakis, “Multi-task neural networks for pain intensity estimation using electrocardiogram and demographic factors,” inInformation and Communication Technologies for Ageing Well and e-Health. Springer Nature Switzerland, 2023, pp. 324–337

    work page 2023

  26. [26]

    Multi-representation diagrams for pain recognition: Integrating various electrodermal activity signals into a single image,

    S. Gkikas, I. Kyprakis, and M. Tsiknakis, “Multi-representation diagrams for pain recognition: Integrating various electrodermal activity signals into a single image,” inCompanion Proceedings of the 27th International Conference on Multimodal Interaction, ser. ICMI Companion ’25. New York, NY , USA: Association for Computing Machinery, 2025, p. 162–171

    work page 2025

  27. [27]

    Mental workload assessment using deep learning models from eeg signals: A systematic review,

    K. Kingphai and Y . Moshfeghi, “Mental workload assessment using deep learning models from eeg signals: A systematic review,”IEEE Transactions on Cognitive and Developmental Systems, vol. 17, no. 1, pp. 40–60, 2025

    work page 2025

  28. [28]

    Automatic assessment of pain based on deep learning methods: A systematic review,

    S. Gkikas and M. Tsiknakis, “Automatic assessment of pain based on deep learning methods: A systematic review,”Computer Methods and Programs in Biomedicine, vol. 231, p. 107365, 2023

    work page 2023

  29. [29]

    Mental workload classification using convolutional neural networks based on fnirs-derived prefrontal activity,

    J.-H. Park, “Mental workload classification using convolutional neural networks based on fnirs-derived prefrontal activity,”BMC Neurology, vol. 23, no. 1, p. 442, 2023

    work page 2023

  30. [30]

    Pain assessment using multi-kernel-fcn-lstm and haemoglobin difference in fnirs,

    G. Bargshady, S. Aziz, S. Gkikas, M. Tsiknakis, R. Goecke, and R. Fernandez Rojas, “Pain assessment using multi-kernel-fcn-lstm and haemoglobin difference in fnirs,”ACM Trans. Comput. Healthcare, 2025

    work page 2025

  31. [31]

    Enhancing cognitive workload classification using integrated lstm layers and cnns for fnirs data analysis,

    M. A. Khan, H. Asadi, M. R. C. Qazani, A. Arogbonlo, S. Pedrammehr, A. Anwar, H. Zhou, L. Wei, A. Bhatti, S. Oladazimi, B. Khan, and S. Nahavandi, “Enhancing cognitive workload classification using integrated lstm layers and cnns for fnirs data analysis,”Computers, vol. 14, no. 2, 2025

    work page 2025

  32. [32]

    Multimodal fusion for objective assessment of cognitive workload: A review,

    E. Debie, R. Fernandez Rojas, J. Fidock, M. Barlow, K. Kasmarik, S. Anavatti, M. Garratt, and H. A. Abbass, “Multimodal fusion for objective assessment of cognitive workload: A review,”IEEE Transactions on Cybernetics, vol. 51, no. 3, pp. 1542–1555, 2021

    work page 2021

  33. [33]

    Convolutional neural network for hybrid fnirs-eeg mental workload classification,

    M. Saadati, J. Nelson, and H. Ayaz, “Convolutional neural network for hybrid fnirs-eeg mental workload classification,” inAdvances in Neuroergonomics and Cognitive Engineering, H. Ayaz, Ed. Springer International Publishing, 2020, pp. 221–232

    work page 2020

  34. [34]

    Ef-net: Mental state recognition by analyzing multimodal eeg-fnirs via cnn,

    A. Arif, Y . Wang, R. Yin, X. Zhang, and A. Helmy, “Ef-net: Mental state recognition by analyzing multimodal eeg-fnirs via cnn,”Sensors, vol. 24, no. 6, 2024

    work page 2024

  35. [35]

    Simultaneous eeg-fnirs data classification through selective channel representation and spectrogram imaging,

    C. Bunterngchit, J. Wang, and Z.-G. Hou, “Simultaneous eeg-fnirs data classification through selective channel representation and spectrogram imaging,”IEEE Journal of Translational Engineering in Health and Medicine, vol. 12, pp. 600–612, 2024

    work page 2024

  36. [36]

    Multimodal mbc-att: cross-modality attentional fusion of eeg-fnirs for cognitive state decoding,

    Y . Li, L. Zhu, A. Huang, J. Zhang, and P. Yuan, “Multimodal mbc-att: cross-modality attentional fusion of eeg-fnirs for cognitive state decoding,” Frontiers in Human Neuroscience, vol. V olume 19 - 2025, 2025

    work page 2025

  37. [37]

    Multimodal fusion of eeg-fnirs: a mutual information-based hybrid classification framework,

    R. J. Deligani, S. B. Borgheai, J. McLinden, and Y . Shahriari, “Multimodal fusion of eeg-fnirs: a mutual information-based hybrid classification framework,”Biomed. Opt. Express, vol. 12, no. 3, pp. 1635–1650, Mar 2021

    work page 2021

  38. [38]

    A crossmod-transformer deep learning framework for multi-modal pain detection through eda and ecg fusion,

    J. Farmani, G. Bargshady, S. Gkikas, M. Tsiknakis, and R. Fernandez Ro- jas, “A crossmod-transformer deep learning framework for multi-modal pain detection through eda and ecg fusion,”Scientific Reports, vol. 15, no. 1, p. 29467, 2025

    work page 2025

  39. [39]

    Giaformer: A gradient-infused attention and transformer for pain assessment with eda-fnirs fusion,

    M. U. Khan, G. Chetty, S. Gkikas, M. Tsiknakis, R. Goecke, and R. Fernandez-Rojas, “Giaformer: A gradient-infused attention and transformer for pain assessment with eda-fnirs fusion,”Information Fusion, vol. 131, p. 104173, 2026

    work page 2026

  40. [40]

    Tiny-biomoe: a lightweight embedding model for biosignal analysis,

    S. Gkikas, I. Kyprakis, and M. Tsiknakis, “Tiny-biomoe: a lightweight embedding model for biosignal analysis,” inCompanion Proceedings of the 27th International Conference on Multimodal Interaction, ser. ICMI Companion ’25. New York, NY , USA: Association for Computing Machinery, 2025, p. 117–126

    work page 2025

  41. [41]

    Efficient pain recognition via respiration signals: A single cross- attention transformer multi-window fusion pipeline,

    ——, “Efficient pain recognition via respiration signals: A single cross- attention transformer multi-window fusion pipeline,” inCompanion Pro- ceedings of the 27th International Conference on Multimodal Interaction, ser. ICMI Companion ’25. New York, NY , USA: Association for Computing Machinery, 2025, p. 70–79

    work page 2025

  42. [42]

    A view on edge caching applications,

    D. Antonogiorgakis, A. Britzolakis, P. Chatziadam, A. Dimitriadis, S. Gikas, E. Michalodimitrakis, M. Oikonomakis, N. Siganos, E. Tza- gkarakis, Y . Nikoloudakis, S. Panagiotakis, E. Pallis, and E. K. Markakis, “A view on edge caching applications,” 2019

    work page 2019

  43. [43]

    When and how to express empathy in human-robot interaction scenarios,

    C. Arzate Cruz, E. C. Montiel-Vazquez, C. Maeda, and R. Gomez, “When and how to express empathy in human-robot interaction scenarios,” in2025 34th IEEE International Conference on Robot and Human Interactive Communication (RO-MAN), 2025, pp. 1070–1077

    work page 2025

  44. [44]

    Empathetic robots using empathy classifiers in hri settings,

    C. Arzate Cruz, E. C. Montiel-Vazquez, C. Maeda, D. Lam, and R. Gomez, “Empathetic robots using empathy classifiers in hri settings,” in2025 20th ACM/IEEE International Conference on Human-Robot Interaction (HRI), 2025, pp. 1211–1215

    work page 2025

  45. [45]

    Enhancing social robot’s direct gaze expression through vestibulo-ocular movements,

    Y . Fang, J. M. P ´erez-Moler´on, L. Merino, S.-L. Yeh, S. Nishina, and R. Gomez, “Enhancing social robot’s direct gaze expression through vestibulo-ocular movements,”Advanced Robotics, vol. 38, no. 19-20, pp. 1457–1469, 2024

    work page 2024

  46. [46]

    Virtual reflections on a dynamic 2-d eye model improve spatial reference identification,

    M. Kruger, Y . Oshima, and Y . Fang, “Virtual reflections on a dynamic 2-d eye model improve spatial reference identification,”IEEE Transactions on Human-Machine Systems, vol. 56, no. 2, pp. 203–212, 2026

    work page 2026

  47. [47]

    A visual perceptual perspective on gaze in social robotics,

    R. S. Hessels and Y . Fang, “A visual perceptual perspective on gaze in social robotics,”Psychonomic Bulletin & Review, vol. 33, no. 4, p. 131, 2026

    work page 2026

  48. [48]

    Efficient emotion-aware iconic gesture prediction for robot co-speech,

    E. C. Montiel-Vazquez, C. A. Cruz, S. Gkikas, T. Kassiotis, G. Gi- annakakis, and R. Gomez, “Efficient emotion-aware iconic gesture prediction for robot co-speech,” 2026

    work page 2026

  49. [49]

    Eegnet: a compact convolutional neural network for eeg-based brain–computer interfaces,

    V . J. Lawhern, A. J. Solon, N. R. Waytowich, S. M. Gordon, C. P. Hung, and B. J. Lance, “Eegnet: a compact convolutional neural network for eeg-based brain–computer interfaces,”Journal of Neural Engineering, vol. 15, no. 5, 2018. APPENDIX A. Complementary Experiments In this section, we present additional experiments using EEGNet[49].EEGNetis a compact c...

    work page 2018