Pith. sign in

REVIEW 3 major objections 5 minor 40 references

Privatization of Synthetic Gaze: Attenuating State Signatures in Diffusion-Generated Eye Movements

T0 review · 3 major / 5 minor · reviewed 2026-08-03 · deepseek-v4-flash

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

Pith's one-line read Diffusion-generated gaze weakens fatigue and difficulty signals while keeping eye-motion realism.

desk verdict First evaluation of state-signal leakage in diffusion-generated gaze, but the attenuation claim depends on an indirect comparison to a prior baseline and needs same-pipeline recomputation before it holds. read the letter →

arxiv 2601.21057 v2 pith:3FGRT475 submitted 2026-01-28 cs.HC

Kamrul Hasan , Oleg V. Komogortsev This is my paper
classification cs.HC
keywords gazesynthesisdiffusionmodelseyemovementsprivacysubjectivestatefatigueSpearmancorrelationsbiometrics
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 asks whether synthetic eye-tracking data can serve as privatized data: realistic enough to stand in for real recordings, but stripped of the internal states that make gaze privacy-sensitive. Using a diffusion-based gaze synthesizer, the authors generate subject-specific eye-movement sequences and compare Spearman correlations between 58 oculomotor features and self-reported fatigue, mental tiredness, and task difficulty against correlations previously reported for real data. They find that synthetic sequences show much weaker, sparser, and less stable correlations across tasks and recording rounds, while preserving spatial accuracy, precision, and identity-related embedding similarity. The conclusion is that, under their training setup, gaze synthesis attenuates state-related signatures and can support privacy-aware gaze applications. The result matters because it offers a concrete path to sharing or publishing gaze data without exposing fatigue or stress.

What carries the argument

The mechanism is a correlation-based audit: a feature extractor reduces each synthetic gaze sequence to 58 interpretable eye-movement features (saccade rate, fixation drift, velocity profiles, etc.), and a correlation analyzer computes Spearman rank correlations between those features and three subjective ratings (overall difficulty, mental tiredness, eye tiredness), separately for each task and for pooled versus per-round subsets. The comparison against the real-data baseline from the earlier study carries the privacy claim: weaker and less stable correlations in synthetic data are read as attenuation of state signatures.

What would settle it

Re-run the same correlation analysis on the real eye-tracking recordings using the exact feature extractor, task subset, rounds, and Spearman pipeline applied to the synthetic data. If the real-data correlations come out as weak and unstable as the synthetic ones, the attenuation claim collapses; alternatively, if a state classifier trained on synthetic features decodes fatigue or difficulty above chance, the data are not privatized.

Watch

Extended reading notes

Core claim

The central claim is that a diffusion-based generative model trained on identity-labeled eye-movement velocity signals produces synthetic gaze that largely severs the statistical link between oculomotor features and subjective state. For horizontal saccades and random saccades, almost no feature–rating pairs reach significance, and the few that do fail to replicate across rounds; for reading, some fixation- and drift-related correlations survive but are far weaker than in real data. The authors argue this attenuation is evidence of 'state-level privatization' — the synthetic signal keeps the kinematics of real gaze (spatial accuracy, precision, embedding similarity) while dropping fatigue- a

Load-bearing premise

The load-bearing assumption is that the correlation patterns previously reported for real gaze are directly comparable to the new synthetic-gaze correlations, despite being computed in a separate study with possibly different feature extraction, participant subsets, and statistical handling.

Editorial extensions

If this is right

  • Synthetic gaze generated this way can be shared or published with lower risk of revealing a user's fatigue, mental load, or perceived difficulty.
  • Applications that depend on state decoding, such as fatigue monitoring or workload estimation, would lose accuracy on this synthetic data.
  • The same evaluation recipe — feature extraction plus subjective-report correlations — can be reused to audit other generative models for state leakage.
  • Training objectives could be extended to deliberately retain or suppress state-related features, for example via multitask losses or disentangled latent variables.

Reading between the lines

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

  • The privacy conclusion depends on comparing against correlations previously reported for real data; recomputing real-data correlations with the exact same feature pipeline, participant subset, and aggregation would settle whether the attenuation is real or a pipeline artifact.
  • Attenuating state signatures does not necessarily imply attenuating identity, since the same features can encode both; a fuller privacy audit should test whether identity verification performance also drops alongside state attenuation.
  • A stronger test of privatization would train a classifier on subjective labels using synthetic gaze features and check whether decoding accuracy falls to chance; correlations are a necessary but not sufficient probe of state leakage.
  • The finding is conditional on the evaluated tasks (saccades and reading), the 5-second window length, and the diffusion training configuration; other tasks or longer windows may retain more state information.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper evaluates whether diffusion-generated synthetic gaze attenuates internal-state information. Using GazeBase HSS, RAN, and TEX tasks, the authors generate subject-specific synthetic gaze with an updated DiffEyeSyn model, extract 58 eye-movement features, compute Spearman correlations with three subjective ratings (OverDiff, Mentally, TiredEyes) for pooled rounds and for rounds 2–4 individually, and compare the resulting sparse and weak correlations with previously published real-data correlations from the baseline [16]. The paper concludes that synthetic gaze largely disrupts the mapping from subjective fatigue and difficulty to eye-movement features, thereby privatizing state-related information while preserving signal realism.

Significance. The question is timely and important: whether synthetic gaze can serve as a privacy-preserving substitute for real eye-tracking data depends on whether state-related information is attenuated. To my knowledge, this is the first study to directly examine subjective-state correlations in diffusion-synthesized gaze, and the authors use a public dataset and promise code release. If the attenuation result were established against a same-pipeline real-data baseline with proper statistical inference, this would be a useful contribution to privacy-aware gaze synthesis. However, the current evidence is not self-contained, and the central claim rests on an indirect comparison that is not fully controlled.

major comments (3)
  1. [§5.2, Figs. 2–4] The central comparison is to correlations reported in the baseline [16] rather than to correlations recomputed on real GazeBase with the exact same feature-extraction, event-detection, outlier-handling, participant/round subset, and statistical code. Differences in event-detection algorithms [37–39] alone can materially change feature values; if the baseline used a different version, the sparse synthetic correlations could be an artifact of pipeline mismatch rather than attenuation. The manuscript says the evaluation uses 'the same rounds, tasks, and subjects as the baseline,' but does not demonstrate that the real-data feature values are comparable. Please recompute the real-data correlations with the same pipeline used for synthetic data and report them side-by-side, or provide a formal statistical comparison of the real and synthetic correlation distributions.
  2. [§5.2, uncorrected p<0.05] All attenuation conclusions rest on uncorrected p<0.05 across 58 features × 3 ratings × multiple rounds/tasks, with no sample sizes, effect sizes, confidence intervals, or multiple-comparison adjustment. In this setting, 'no features display significant correlations' (HSS) and 'few effects' (RAN) are not evidence of attenuation; they are equally consistent with low power or noise. The paper should report the number of subjects per round and task, use corrected intervals, equivalence tests, or a bootstrap comparison between real and synthetic correlations, and state explicitly which differences are statistically meaningful rather than visually weaker.
  3. [§7] The conclusion states that gaze synthesis 'privatizes internal state information,' but the analysis only measures marginal Spearman correlations between features and self-reports. This is a reasonable proxy, but it is not a direct test of decodability: an adversary or downstream classifier could in principle recover state information from nonlinear combinations of features even when marginal correlations are weak. Either narrow the privacy claim to the correlation-based framework inherited from [16], or add a classifier-based state-decoding probe on real and synthetic data to support the broader privacy wording.
minor comments (5)
  1. [Abstract, §3.1] The terms 'DiffEyeSyn' and 'Updated DiffEyeSyn' are used interchangeably; define the exact model once and use a single name thereafter.
  2. [Figures 2–4] The heatmaps are difficult to read in the present form: axis labels and feature names are small, and no colorbar or significance legend is described. Add readable labels and a color scale in the final version.
  3. [References] Ref. [25] contains a typo ('super resolutin'); Ref. [40] is incomplete as formatted; check all bibliographic entries against the journal style.
  4. [§4.2] The authors state that source code will be released, but no repository URL is provided. For reproducibility, include a URL or an anonymized repository in the revision.
  5. [Keywords/abstract] Minor formatting issues: 'V AEs' and 'Y ou' appear with unintended spacing in the abstract and related work; proofread for spacing artifacts.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the attenuation finding is an independently computed empirical result; the self-cited baseline creates a comparability risk but not a definitional reduction.

full rationale

The derivation is not circular by construction. The synthetic signals are produced by the pre-trained DiffEyeSyn model and then passed through a fixed feature-extraction and Spearman-correlation pipeline; none of the model's weights or hyperparameters is fitted to the subjective reports or to the observed attenuation. The real-data side of the comparison is imported from [16], a prior empirical study that shares two authors with the present paper. This is a self-citation, and it is the main comparator for the attenuation claim, so the missing recomputation of real correlations with the identical pipeline is a genuine correctness/comparability risk. However, this is a validity threat, not a circularity: the synthetic correlations do not reduce by definition to the [16] values, and the attenuation is not a fitted parameter renamed as a prediction. The self-citation supplies an external empirical baseline, while the paper's own analysis contributes newly computed observations; there is no equation that reduces to itself or to the input data.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

The central claim rests on the comparability of synthetic results to prior real-data correlations and on the sufficiency of correlation analysis as a privacy measure. No new entities are introduced, and no parameters are fitted in the correlation analysis itself.

assumptions (3)
  • domain assumption Subjective self-reports (Likert ratings) are valid behavioral readouts of internal states such as fatigue, mental tiredness, and task difficulty.
    The entire analysis treats self-report ratings as ground-truth state labels. If the ratings are noisy, biased, or not linearly related to the intended states, then the correlation analysis may not measure state encoding at all. Introduced in Section 3.2.
  • domain assumption Spearman correlations between eye-movement features and subjective ratings are a sufficient measure of privacy-relevant state information.
    The paper equates low bivariate correlations with state attenuation, but multivariate/nonlinear decoders could still recover state information from the feature set. This assumption is load-bearing for the privacy claim. See Sections 5.2 and 7.
  • domain assumption The real-data correlation patterns reported in the baseline paper [16] are directly comparable to the synthetic-data results obtained here.
    The paper does not recompute real-data correlations with the exact same pipeline, participant subset, and statistical choices; the comparison to [16] is the primary evidence for attenuation. If those real-data results are not directly comparable, the observed difference may be an artifact. Assumed throughout Section 5.2.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Privatization of Synthetic Gaze: Attenuating State Signatures in Diffusion-Generated Eye Movements." pith.science (2026). https://pith.science/paper/3FGRT475

@misc{pith2026260121057,
  author       = {Pith},
  title        = {Pith review of: Privatization of Synthetic Gaze: Attenuating State Signatures in Diffusion-Generated Eye Movements},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3FGRT475}},
  note         = {Machine review of arXiv:2601.21057}
}
read the original abstract

The recent success of deep learning (DL) has enabled the generation of high-quality synthetic data, advancing the development of data-driven biometric applications. Among various biometric modalities, eye movement sequences have emerged as a promising behavioral biometric. However, gaze data also raises privacy concerns because it may encode individuals' internal states, such as fatigue, emotional load, and stress. Ideally, synthetic gaze data should preserve the signal quality of real recordings, including identity features, while removing or attenuating privacy-sensitive, state-related attributes to reduce risks of personal state exposure. Many recent DL-based generative models focus on replicating real gaze trajectories but do not explicitly evaluate whether generated signals retain subjective-state information. In this work, we examine a recent diffusion-based gaze synthesis approach by analyzing the correlations between synthetic gaze features and subjective reports, including fatigue and other self-reported states. Our results show that these correlations are weaker and less stable in synthetic gaze than in real gaze, suggesting the attenuation of state-related signatures under the evaluated protocol. At the same time, synthetic gaze preserves essential signal characteristics similar to real data, supporting its potential use in privacy-aware gaze-based applications.

Figures

Figures reproduced from arXiv: 2601.21057 by the authors.

Figure 1
Figure 1. Overview of the architecture for the correlation between subjective reports and objective features: The real [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Spearman correlations between synthetic gaze-derived features and subjective ratings for the HSS task. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Spearman correlations between synthetic gaze-derived features and subjective ratings for the RAN task. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Spearman correlations between synthetic gaze-derived features and subjective ratings for the TEX task. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

40 extracted references · 1 canonical work pages

  1. [16]

    Why do we need high-fidelity synthetic eye movement data and how should they look like?bioRxiv, pages 2025–12, 2025

    C Stella Qian, Samantha Aziz, Kamrul Hasan, and Oleg V Komogortsev. Why do we need high-fidelity synthetic eye movement data and how should they look like?bioRxiv, pages 2025–12, 2025

    2025

  2. [1]

    Eye movements in natural behavior.Trends in cognitive sciences, 9(4):188–194, 2005

    Mary Hayhoe and Dana Ballard. Eye movements in natural behavior.Trends in cognitive sciences, 9(4):188–194, 2005

    2005

  3. [2]

    The effectiveness of eye tracking in the diagnosis of cognitive disorders: A systematic review and meta-analysis.PloS one, 16(7):e0254059, 2021

    Zicai Liu, Zhen Yang, Yueming Gu, Huiyu Liu, and Pu Wang. The effectiveness of eye tracking in the diagnosis of cognitive disorders: A systematic review and meta-analysis.PloS one, 16(7):e0254059, 2021

    2021

  4. [3]

    Eye-tracking based classification of mandarin chinese readers with and without dyslexia using neural sequence models.arXiv preprint arXiv:2210.09819, 2022

    Patrick Haller, Andreas Säuberli, Sarah Elisabeth Kiener, Jinger Pan, Ming Yan, and Lena Jäger. Eye-tracking based classification of mandarin chinese readers with and without dyslexia using neural sequence models.arXiv preprint arXiv:2210.09819, 2022

    arXiv 2022

  5. [4]

    Eye-tracking based autism spectrum disorder diagnosis using chaotic butterfly optimization with deep learning model.Computers, Materials & Continua, 76(2), 2023

    Tamilvizhi Thanarajan, Youseef Alotaibi, Surendran Rajendran, and Krishnaraj Nagappan. Eye-tracking based autism spectrum disorder diagnosis using chaotic butterfly optimization with deep learning model.Computers, Materials & Continua, 76(2), 2023

    2023

  6. [5]

    The application of eye tracking in education

    Yuyang Sun, Qingzhong Li, Honggen Zhang, and Jiancheng Zou. The application of eye tracking in education. InInternational Conference on Intelligent Information Hiding and Multimedia Signal Processing, pages 27–33. Springer, 2017

    2017

  7. [6]

    Eye know you too: Toward viable end-to-end eye movement biometrics for user authentication.IEEE Transactions on Information Forensics and Security, 17:3151–3164, 2022

    Dillon Lohr and Oleg V Komogortsev. Eye know you too: Toward viable end-to-end eye movement biometrics for user authentication.IEEE Transactions on Information Forensics and Security, 17:3151–3164, 2022

    2022

  8. [7]

    Gaze authentication: Factors influencing authentication performance.arXiv preprint arXiv:2509.10969, 2025

    Dillon Lohr, Michael J Proulx, Mehedi Hasan Raju, and Oleg V Komogortsev. Gaze authentication: Factors influencing authentication performance.arXiv preprint arXiv:2509.10969, 2025

    arXiv 2025

Show all 40 references
  1. [8]

    Eye-tracking in ar/vr: A technological review and future directions.IEEE Open Journal on Immersive Displays, 2024

    Xin Jin, Suyu Chai, Jie Tang, Xianda Zhou, and Kai Wang. Eye-tracking in ar/vr: A technological review and future directions.IEEE Open Journal on Immersive Displays, 2024

    2024

  2. [9]

    Gazebase, a large-scale, multi-stimulus, longitudinal eye movement dataset.Scientific Data, 8(1):184, 2021

    Henry Griffith, Dillon Lohr, Evgeny Abdulin, and Oleg Komogortsev. Gazebase, a large-scale, multi-stimulus, longitudinal eye movement dataset.Scientific Data, 8(1):184, 2021

    2021

  3. [10]

    Privacy-aware eye tracking using differential privacy

    Julian Steil, Inken Hagestedt, Michael Xuelin Huang, and Andreas Bulling. Privacy-aware eye tracking using differential privacy. InProceedings of the 11th ACM Symposium on Eye Tracking Research & Applications, pages 1–9, 2019. 7 Privatization of Synthetic Gaze: Attenuating Sta...

    2019

  4. [11]

    Sp-eyegan: Generating synthetic eye movement data with generative adversarial networks

    Paul Prasse, David Robert Reich, Silvia Makowski, Seoyoung Ahn, Tobias Scheffer, and Lena A Jäger. Sp-eyegan: Generating synthetic eye movement data with generative adversarial networks. InProceedings of the 2023 symposium on eye tracking research and applications, pages 1–9, 2023

    2023

  5. [12]

    Diffeyesyn: Diffusion-based user-specific eye movement synthesis.arXiv preprint arXiv:2409.01240, 2024

    Chuhan Jiao, Guanhua Zhang, Yeonjoo Cho, Zhiming Hu, and Andreas Bulling. Diffeyesyn: Diffusion-based user-specific eye movement synthesis.arXiv preprint arXiv:2409.01240, 2024

    2024 arXiv

  6. [13]

    Denoising diffusion probabilistic models.Advances in neural information processing systems, 33:6840–6851, 2020

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffusion probabilistic models.Advances in neural information processing systems, 33:6840–6851, 2020

    2020

  7. [14]

    Generative adversarial networks.Communications of the ACM, 63(11):139–144, 2020

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adversarial networks.Communications of the ACM, 63(11):139–144, 2020

    2020

  8. [15]

    Evaluation of eye tracking signal quality for virtual reality applications: A case study in the meta quest pro

    Samantha Aziz, Dillon J Lohr, Lee Friedman, and Oleg Komogortsev. Evaluation of eye tracking signal quality for virtual reality applications: A case study in the meta quest pro. InProceedings of the 2024 Symposium on Eye Tracking Research and Applications, pages 1–8, 2024

    2024

  9. [17]

    Quantitative and quali- tative comparison of generative models for subject-specific gaze synthesis: Diffusion vs gan.arXiv preprint arXiv:2511.09867, 2025

    Kamrul Hasan, Dmytro Katrychuk, Mehedi Hasan Raju, and Oleg V Komogortsev. Quantitative and quali- tative comparison of generative models for subject-specific gaze synthesis: Diffusion vs gan.arXiv preprint arXiv:2511.09867, 2025

    2025 arXiv

  10. [18]

    Eyes alive

    Sooha Park Lee, Jeremy B Badler, and Norman I Badler. Eyes alive. InProceedings of the 29th annual conference on Computer graphics and interactive techniques, pages 637–644, 2002

    2002

  11. [19]

    Modeling physiologically plausible eye rotations

    Andrew T Duchowski and Sophie Jörg. Modeling physiologically plausible eye rotations. InProceedings of computer graphics international, 2015

    2015

  12. [20]

    Eye movement synthesis

    Andrew T Duchowski, Sophie Jörg, Tyler N Allen, Ioannis Giannopoulos, and Krzysztof Krejtz. Eye movement synthesis. InProceedings of the ninth biennial ACM symposium on eye tracking research & applications, pages 147–154, 2016

    2016

  13. [21]

    Eyecatch: Simulating visuomotor coordination for object interception.ACM Transactions on Graphics (TOG), 31(4):1–10, 2012

    Sang Hoon Yeo, Martin Lesmana, Debanga R Neog, and Dinesh K Pai. Eyecatch: Simulating visuomotor coordination for object interception.ACM Transactions on Graphics (TOG), 31(4):1–10, 2012

    2012

  14. [22]

    Automatic scanpath generation with deep recurrent neural networks

    Daniel Simon, Srinivas Sridharan, Shagan Sah, Raymond Ptucha, Chris Kanan, and Reynold Bailey. Automatic scanpath generation with deep recurrent neural networks. InProceedings of the ACM symposium on applied perception, pages 130–130, 2016

    2016

  15. [23]

    Pathgan: Visual scanpath prediction with generative adversarial networks

    Marc Assens, Xavier Giro-i Nieto, Kevin McGuinness, and Noel E O’Connor. Pathgan: Visual scanpath prediction with generative adversarial networks. InProceedings of the European Conference on Computer Vision (ECCV) Workshops, pages 0–0, 2018

    2018

  16. [24]

    Eyegan: Gaze-preserving, mask-mediated eye image synthesis

    Harsimran Kaur and Roberto Manduchi. Eyegan: Gaze-preserving, mask-mediated eye image synthesis. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 310–319, 2020

    2020

  17. [25]

    Supreyes: Super resolutin for eyes using implicit neural representation learning

    Chuhan Jiao, Zhiming Hu, Mihai Bâce, and Andreas Bulling. Supreyes: Super resolutin for eyes using implicit neural representation learning. InProceedings of the 36th Annual ACM Symposium on User Interface Software and Technology, pages 1–13, 2023

    2023

  18. [26]

    Diffgaze: A diffusion model for modelling fine-grained human gaze behaviour on 360° images.ACM Transactions on Interactive Intelligent Systems, 2025

    Chuhan Jiao, Yao Wang, Guanhua Zhang, Mihai Bâce, Zhiming Hu, and Andreas Bulling. Diffgaze: A diffusion model for modelling fine-grained human gaze behaviour on 360° images.ACM Transactions on Interactive Intelligent Systems, 2025

    2025

  19. [27]

    Assessing the privacy risk of cross-platform identity linkage using eye movement biometrics

    Samantha Aziz and Oleg Komogortsev. Assessing the privacy risk of cross-platform identity linkage using eye movement biometrics. In2023 IEEE International Joint Conference on Biometrics (IJCB), pages 1–9. IEEE, 2023

    2023

  20. [28]

    For your eyes only: Privacy-preserving eye-tracking datasets

    Brendan David-John, Kevin Butler, and Eakta Jain. For your eyes only: Privacy-preserving eye-tracking datasets. In2022 symposium on eye tracking research and applications, pages 1–6, 2022

    2022

  21. [29]

    What does your gaze reveal about you? on the privacy implications of eye tracking

    Jacob Leon Kröger, Otto Hans-Martin Lutz, and Florian Müller. What does your gaze reveal about you? on the privacy implications of eye tracking. InIFIP International Summer School on Privacy and Identity Management, pages 226–241. Springer, 2020

    2020

  22. [30]

    Differential privacy for eye tracking with temporal correlations.Plos one, 16(8):e0255979, 2021

    Efe Bozkir, Onur Günlü, Wolfgang Fuhl, Rafael F Schaefer, and Enkelejda Kasneci. Differential privacy for eye tracking with temporal correlations.Plos one, 16(8):e0255979, 2021

    2021

  23. [31]

    Differential privacy for eye-tracking data

    Ao Liu, Lirong Xia, Andrew Duchowski, Reynold Bailey, Kenneth Holmqvist, and Eakta Jain. Differential privacy for eye-tracking data. InProceedings of the 11th ACM Symposium on Eye Tracking Research & Applications, pages 1–10, 2019. 8 Privatization of Synthetic Gaze: Attenuatin...

    2019

  24. [32]

    Real-time lightweight gaze privacy-preservation techniques validated via offline gaze-based interaction simulation.arXiv preprint arXiv:2511.09846, 2025

    Mehedi Hasan Raju and Oleg V Komogortsev. Real-time lightweight gaze privacy-preservation techniques validated via offline gaze-based interaction simulation.arXiv preprint arXiv:2511.09846, 2025

    2025

  25. [33]

    Privacy enhancement for gaze data using a noise-infused autoencoder

    Samantha Aziz and Oleg Komogortsev. Privacy enhancement for gaze data using a noise-infused autoencoder. arXiv preprint arXiv:2508.10918, 2025

    2025 arXiv

  26. [34]

    Privacy preserving gaze estimation using synthetic images via a randomized encoding based framework

    Efe Bozkir, Ali Burak Ünal, Mete Akgün, Enkelejda Kasneci, and Nico Pfeifer. Privacy preserving gaze estimation using synthetic images via a randomized encoding based framework. InACM symposium on eye tracking research and applications, pages 1–5, 2020

    2020

  27. [35]

    Dpgazesynth: enhancing eye-tracking virtual reality privacy with differentially private data synthesis.Information Sciences, 675:120720, 2024

    Xiaojun Ren, Jiluan Fan, Ning Xu, Shaowei Wang, Changyu Dong, and Zikai Wen. Dpgazesynth: enhancing eye-tracking virtual reality privacy with differentially private data synthesis.Information Sciences, 675:120720, 2024

    2024

  28. [36]

    Adding conditional control to text-to-image diffusion models

    Lvmin Zhang, Anyi Rao, and Maneesh Agrawala. Adding conditional control to text-to-image diffusion models. InProceedings of the IEEE/CVF international conference on computer vision, pages 3836–3847, 2023

    2023

  29. [37]

    Study of an extensive set of eye movement features: Extraction methods and statistical analysis.Journal of Eye Movement Research, 11(1):10–16910, 2018

    Ioannis Rigas, Lee Friedman, and Oleg Komogortsev. Study of an extensive set of eye movement features: Extraction methods and statistical analysis.Journal of Eye Movement Research, 11(1):10–16910, 2018

    2018

  30. [38]

    Komogortsev

    Lee Friedman, Ioannis Rigas, Evgeny Abdulin, and Oleg V . Komogortsev. A novel evaluation of two related and two independent algorithms for eye movement classification during reading. 50(4):1374–1397. ISSN 1554-3528. doi:10.3758/s13428-018-1050-7. URLhttps://doi.org/10.3758/s1...

    doi

  31. [39]

    Eye movement classification dataset

    Lee Friedman. Eye movement classification dataset. URLhttps://hdl.handle.net/10877/16339

  32. [40]

    D. P. Kingma and J. Ba. Adam: A method for stochastic optimization.CoRR, abs/1412.6980, 2014. 9

    2014 arXiv

Pith tools

Reviewed August 3, 2026 · model on record in the stance chip above.