Pith. sign in

REVIEW 4 major objections 5 minor 54 references

Modeling Eye Gaze Velocity Trajectories using GANs with Spectral Loss for Enhanced Fidelity

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read A spectrally regularized LSTM-CNN GAN produces synthetic eye-gaze velocity trajectories that match real data far more closely than a hidden Markov model on the reported distributional and temporal metrics.

desk verdict Competent empirical study whose headline GAN-vs-HMM result is credible, but the spectral-loss improvement claim is under-supported by a circular metric and a missing held-out split. read the letter →

arxiv 2412.04184 v1 pith:V5EHBMKD submitted 2024-12-05 cs.NE cs.LG

Shailendra Bhandari , Pedro Lencastre , Rujeena Mathema , Alexander Szorkovszky , Anis Yazidi , Pedro Lind This is my paper · ORCID
classification cs.NEcs.LG
keywords GenerativeAdversarialNetworkseyegazevelocityspectrallossLSTMCNNHiddenMarkovModelsynthetictimeseriesJensen-Shannondivergence
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

This paper argues that a lightweight GAN pairing an LSTM generator with a CNN discriminator, trained with a spectral loss that compares the log Fourier magnitudes of real and generated sequences, can generate synthetic eye-gaze velocity trajectories that match recorded data in distribution and autocorrelation. The authors report a Jensen–Shannon divergence of $0.000322$ for this model, versus $0.01244$ for a four-state HMM, and show that spectral regularization improves fidelity over adversarial-only training. If correct, the result provides a practical generative tool for simulation environments, training systems, and human-computer interaction research that needs realistic gaze data, and it narrows a previously reported gap in which simple Markov models often beat GANs on stochastic trajectories.

What carries the argument

The load-bearing mechanism is the spectral loss $L_{\mathrm{spectral}} = \sum_{k=0}^{N-1} [\log(|F(x)_k|) - \log(|F(\hat{x})_k|)]^2$, which penalizes differences in the log magnitudes of the discrete Fourier transforms of real and generated velocity sequences, normalizing amplitude disparities and focusing on relative spectral energy. It is added to the non-saturating adversarial generator loss as $L_{\mathrm{final}} = L_G + \lambda L_{\mathrm{spectral}}$ with $\lambda = 0.1$. The generator and discriminator pair LSTM and CNN layers: the LSTM generator converts random noise into a sequence of hidden states that become velocity displacements, capturing temporal dependencies, while the CNN discriminator uses strided convolutions and a sigmoid output to classify trajectory realism. Training uses 200-point sequences, 500 epochs, a 0.0002 learning rate, and fidelity is measured with Jensen–Shannon divergence and autocorrelation.

What would settle it

Train the LSTM-CNN GAN on one set of gaze-velocity sequences and compute Jensen–Shannon divergence and autocorrelation on a disjoint set from different participants or recording sessions; if $D_{JS}$ on the held-out data rises to the HMM's level or above, the central fidelity claim is not established.

Watch

Extended reading notes

Core claim

The central claim is that a GAN trained to minimize $L_{\mathrm{final}} = L_G + \lambda L_{\mathrm{spectral}}$, where $L_{\mathrm{spectral}} = \sum_{k=0}^{N-1}[\log(|F(x)_k|) - \log(|F(\hat{x})_k|)]^2$ and $x$, $\hat{x}$ are real and generated velocity sequences, reproduces the statistical and spectral properties of real eye-gaze velocity time series. Among the four tested architectures, the LSTM-CNN model showed the closest alignment to the real data distribution, with $D_{JS}=0.000322$; the HMM diverged with $D_{JS}=0.01244$. The paper also reports that LSTM-CNN matches the mean, standard deviation, skewness, and kurtosis of the real data more closely than the HMM and captures the early-lag autocorrelation function. The gain is attributed to the LSTM generator's long-term temporal modeling, the CNN discriminator's local feature extraction, and spectral regularization's emphasis on high- and low-frequency components.

Load-bearing premise

The paper compares generated data against the same gaze-velocity sequences used to train the GAN, with no described held-out split, so the reported closeness could reflect overfitting rather than an ability to generate new-looking gaze trajectories.

Editorial extensions

If this is right

  • Synthetic gaze-velocity data with realistic heavy tails and temporal autocorrelation can be produced by a comparatively small GAN, making it usable in simulation environments and eye-tracking training systems.
  • Adding spectral regularization stabilizes GAN training and improves distributional fidelity for eye-gaze trajectories relative to adversarial loss alone.
  • Among the four tested architectures, the LSTM-CNN combination gives the best balance between fidelity (low $D_{JS}$ and spectral loss) and per-epoch computation time.
  • A four-state HMM fitted by Baum-Welch does not reproduce the mean, standard deviation, skewness, kurtosis, or early-lag autocorrelation of real gaze velocity, whereas the spectrally regularized LSTM-CNN GAN does on the reported metrics.
  • The result directly addresses the earlier finding that Markov models can outperform GANs on stochastic trajectory data, by showing that a suitably regularized GAN closes that gap.

Reading between the lines

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

  • The reported fidelity statistics are computed on sequences used to train the GAN; a held-out evaluation on previously unseen participants or recording sessions would test whether the model generalizes or partly memorizes the training data.
  • The log-magnitude Fourier loss is scale-sensitive in a way that suits heavy-tailed velocity distributions; the same regularization could transfer to other non-stationary physiological time series, such as heart-rate variability or EEG, where spectral shape and tail behavior matter.
  • The manuscript reports two different HMM divergence values (0.00714 in the comparison text and 0.01244 in Table 2); the magnitude of the claimed gap depends on which value is used.
  • Because the GAN starts from uniform noise and produces 200-point segments, the generated trajectories are stationary in distribution; assessing whether the model captures longer-range, non-stationary gaze behavior would require comparing on multi-segment concatenated sequences.
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

4 major / 5 minor

Summary. The paper proposes a generative-adversarial-network (GAN) framework with LSTM and CNN generators/discriminators for synthetic eye-gaze velocity trajectories, augmented with a spectral loss term L_spectral (Eq. 8) that penalizes differences in log-magnitude Fourier spectra between real and generated sequences, combined as L_final = L_G + λ L_spectral (Eq. 10). Four GAN architectures (CNN-CNN, LSTM-CNN, CNN-LSTM, LSTM-LSTM) are compared under adversarial-only and spectral-regularized losses, and the best architecture (LSTM-CNN) is compared against a four-state HMM using Jensen-Shannon divergence, autocorrelation, and distribution moments. The central claim is that the spectrally regularized LSTM-CNN GAN 'closely matches the real data' (DJS = 0.000322) while the HMM 'significantly diverges' (DJS = 0.01244), positioning the GAN as a high-fidelity synthetic gaze data generator.

Significance. If the central claim were fully supported, the paper would offer a practical, lightweight GAN-based alternative to traditional Markov models for generating realistic eye-gaze trajectories, with applications in HCI, simulation, and diagnostic tool development. Strengths include the public release of code and data-generation package, the comparison against a standard HMM baseline, and the use of several evaluation metrics (DJS, ACF, moments) that are at least partly independent of the training objective. However, the significance is currently bounded by methodological gaps: the spectral-loss contribution is not established through independent quantitative comparison, the evaluation appears to lack a held-out test set, and no variability measures are reported. These issues are fixable but require additional experiments.

major comments (4)
  1. [Section 4.1 and Eq. (10)] The central novelty claim—that spectral regularization improves fidelity—is not quantitatively demonstrated. The only quantitative scores in Fig. 4 are the 'Spectral Score' (essentially L_spectral from Eq. 8) and integrals of L_final and DJS for models trained with L_final. Because L_final directly minimizes L_spectral, reporting a lower L_spectral for L_final-trained models is circular and does not show that the generated distribution is closer to the real data. The independent metric DJS is shown in Fig. 4 only for L_final-trained models; no DJS, moment, or ACF values are reported for the LG-only variants of each architecture. To support the abstract's claim, the authors should report independent metrics (DJS, moments, ACF) for both LG-only and L_final conditions for every architecture, with confidence intervals over repeated runs.
  2. [Sections 3 and 4] The evaluation does not describe a train/test split. Section 3 states that velocity data were segmented into sequences of 200 points and supplied to the discriminator, but it never specifies a held-out set of real sequences for evaluation. The DJS and ACF calculations appear to use the same sequences that were used for GAN training. If the comparison set overlaps the training set, the reported DJS = 0.000322 may reflect overfitting or memorization rather than generalization to new gaze trajectories. The authors must separate training and test sequences, describe how synthetic sequences are generated (length, number, initialization), and report DJS/ACF on the held-out real data.
  3. [Table 2, Section 4.2] The reported HMM divergence values are internally inconsistent. Table 2 lists the HMM average DJS as 0.01244, while Section 4.2 states the four-state HMM has DJS = 0.0131 and the five-state HMM has DJS = 0.013245. A later paragraph in the same section refers to a 'higher JS divergence (0.00714)'. These three numbers cannot all be correct for the same model and evaluation. The authors should correct the inconsistencies and specify exactly which DJS value corresponds to the four-state HMM used in the comparison, along with the log base used in Eq. (11) (the claimed bound 0 ≤ DJS ≤ 1 requires base-2 logarithms).
  4. [Section 4.1 and Table 2] The article reports no repeated-seed variability or statistical significance tests. Single-run point estimates for DJS, moments, and ACF do not support ranking of architectures or the claim that the HMM 'significantly diverges' from the real data. Given the stochasticity of GAN training, the authors should train each configuration with multiple random seeds, report means and standard deviations (or confidence intervals), and perform a statistical test (e.g., bootstrap or paired test) for the LSTM-CNN versus HMM differences.
minor comments (5)
  1. [Section 4.2] The text says 'The CNN-LSTM GAN utilized a random noise vector...' but the model being evaluated is the LSTM-CNN architecture; this naming inconsistency should be fixed.
  2. [Fig. 4(a)] The caption and text refer to a 'red trajectory' for the spectral loss and a 'blue line' for DJS, but the figure legend and axis labels are unclear. Please ensure the plot clearly distinguishes the loss components and DJS, and define what 'Spectral Score' means relative to Eq. (8).
  3. [Section 3] The data description lacks participant information: number of participants, total recording duration, number of sequences, and how left/right eye data were combined or treated. This information is needed for reproducibility.
  4. [Section 2.4 and Appendix A] The HMM implementation details are incomplete: no mention of the number of training sequences, initial parameter values for the EM algorithm, convergence criteria, or how synthetic HMM sequences were generated (e.g., from the fitted emission distributions).
  5. [Introduction] The reference to 'reproducing stochastic processes' in Lencastre et al. (2023) is relevant, but the introduction does not clearly state which concrete limitations of prior GANs (e.g., mode collapse, poor tail capture) the spectral loss specifically targets. A more precise motivation would strengthen the paper.

Circularity Check

1 steps flagged · score 6.0 of 10

Spectral-score evaluation is the spectral-loss objective; the claimed LG-vs-Lfinal benefit lacks an independent quantitative DJS comparison.

  1. self definitional [Section 2.2 (Eqs. 8 and 10), Section 3 ('Data and the Statistical Measure'), Fig. 4 and Table 2]
    "Complementing the DJS, the spectral loss Lspectral introduces a frequency domain perspective to the evaluation of generative models. ... The top left panel illustrates the Spectral Score (in log scale), showing how CNN-based models achieve more stable and lower scores compared to LSTM-based models. ... Spectral regularization is then defined by combining the generator loss LG with the spectral loss Lspectral: Lfinal = LG + λ Lspectral, (10)."

    The 'Spectral Score' used as the quantitative architecture-comparison metric is exactly the Lspectral term that Eq. (10) adds to LG and that the Lfinal-trained models are explicitly optimized to minimize. Reporting lower Lspectral for Lfinal-trained models is therefore partly self-referential: the quantity being reported is the quantity being minimized by construction. The independent metric DJS is reported only for Lfinal-trained models in the quantitative tables and figures; no DJS values for the LG-only variants are given, so the claimed advantage of spectral regularization over plain adversarial training rests on the self-defined spectral score plus visual inspection of Fig. 3.

full rationale

The one genuine circular step is the use of Lspectral as both the training objective and the reported 'Spectral Score' evaluation metric for the architecture-selection experiment. Since Lfinal = LG + λLspectral, comparing Lfinal-trained models on the spectral loss measures how well each model minimized its own objective, not independently how close the generated distribution is to the real data. The strongest independent evidence, DJS = 0.000322 for LSTM-CNN versus DJS = 0.01244 for HMM and the moments in Table 2, compares the Lfinal-trained GAN with an HMM; it does not isolate the contribution of Lspectral against an LG-only version, because no LG-only DJS values are provided. The absence of a held-out test split (Section 3 only describes segmentation into training batches, and DJS/ACF appear to be computed on the same sequences) is a generalization and validation weakness, but it is not itself a definitional circularity, so it is noted but not scored as a circular step. The self-citations, Lencastre et al. 2023 and Bhandari et al. 2024, are used for motivation and comparison rather than as load-bearing justifications of the derivation. Overall, one evaluation metric reduces by construction while the central comparison retains independent content, giving a moderate partial-circularity score of 6.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central claim rests on two modeling choices: fixed-length windowing without a held-out evaluation set, and a spectral loss that is also used as the evaluation metric. These are not independent evidence of fidelity.

free parameters (4)
  • Spectral loss weight lambda = 0.1
    Set to 0.1 with no sensitivity analysis; the central claim that spectral regularization improves fidelity depends on this single value. Table 1 and Eq. 10.
  • Number of HMM hidden states = 4
    Selected by minimizing DJS between real and HMM-generated data on the same dataset used for the final comparison. Section 4.2, Fig. 5c.
  • Integration start epoch for Fig. 4 = 100
    The integral volume comparison in Fig. 4 integrates L_final and DJS from epoch 100 to 500; this arbitrary cutoff affects the architecture ranking. Section 4.1, Fig. 4.
  • Sequence length for data segmentation = 200
    Data are cut into fixed 200-point windows for both training and evaluation; no sensitivity analysis is provided for this choice. Section 3.
assumptions (4)
  • domain assumption Eye gaze velocity sequences can be treated as stationary, time-homogeneous stochastic processes and segmented into 200-point windows for both training and evaluation.
    Section 3 cuts the data into fixed-length sequences without modeling non-stationarity or between-participant variation.
  • domain assumption The real sequences used for DJS and ACF evaluation are an unbiased reference, independent of GAN training; no train/test split is described.
    Sections 3 and 4 evaluate against 'real data' without specifying a held-out subset, which is a load-bearing assumption for the fidelity claim.
  • ad hoc to paper Log-magnitude Fourier spectra, ignoring phase, are sufficient to capture the spectral characteristics relevant for gaze trajectories.
    Eq. 8 defines L_spectral on |F(x)| only, dropping phase; this quantity is used both as a training loss and as the Spectral Score in Fig. 4.
  • standard math The non-saturating GAN objective approximates minimization of the Jensen-Shannon divergence between real and generated distributions.
    Standard GAN theory (Goodfellow et al., 2014), used in Eqs. 2-7 and invoked for interpreting DJS.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Modeling Eye Gaze Velocity Trajectories using GANs with Spectral Loss for Enhanced Fidelity." pith.science (2026). https://pith.science/paper/V5EHBMKD

@misc{pith2026241204184,
  author       = {Pith},
  title        = {Pith review of: Modeling Eye Gaze Velocity Trajectories using GANs with Spectral Loss for Enhanced Fidelity},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/V5EHBMKD}},
  note         = {Machine review of arXiv:2412.04184}
}
read the original abstract

Accurate modeling of eye gaze dynamics is essential for advancement in human-computer interaction, neurological diagnostics, and cognitive research. Traditional generative models like Markov models often fail to capture the complex temporal dependencies and distributional nuance inherent in eye gaze trajectories data. This study introduces a GAN framework employing LSTM and CNN generators and discriminators to generate high-fidelity synthetic eye gaze velocity trajectories. We conducted a comprehensive evaluation of four GAN architectures: CNN-CNN, LSTM-CNN, CNN-LSTM, and LSTM-LSTM trained under two conditions: using only adversarial loss and using a weighted combination of adversarial and spectral losses. Our findings reveal that the LSTM-CNN architecture trained with this new loss function exhibits the closest alignment to the real data distribution, effectively capturing both the distribution tails and the intricate temporal dependencies. The inclusion of spectral regularization significantly enhances the GANs ability to replicate the spectral characteristics of eye gaze movements, leading to a more stable learning process and improved data fidelity. Comparative analysis with an HMM optimized to four hidden states further highlights the advantages of the LSTM-CNN GAN. Statistical metrics show that the HMM-generated data significantly diverges from the real data in terms of mean, standard deviation, skewness, and kurtosis. In contrast, the LSTM-CNN model closely matches the real data across these statistics, affirming its capacity to model the complexity of eye gaze dynamics effectively. These results position the spectrally regularized LSTM-CNN GAN as a robust tool for generating synthetic eye gaze velocity data with high fidelity.

Figures

Figures reproduced from arXiv: 2412.04184 by the authors.

Figure 1
Figure 1. Overview of the GAN process for generating eye-tracking data, showing the interaction between the generator, discriminator, and adversarial/spectral loss feedback. Detailed architectures of LSTM and CNN-based generators and discriminators, illustrating possible combinations for generating and discriminating gaze trajectory data from noise input. 2.2 Spectral Loss To enhance the training of the generator, we compleme… view at source ↗
Figure 2
Figure 2. (a, b, c, d) X-Y positions and velocity plots for left and right eye movements. The top two plots show the eye movement trajectories (left and right eyes) in terms of X and Y positions, while the bottom two plots show the respective velocity profiles over time. (e, f) Log-scaled velocity distributions for left and right eye movements for one participant. The histograms represent the distribution of logarithmic veloc… view at source ↗
Figure 3
Figure 3. Comparison of velocity distributions for real data (dark blue) and synthetic data generated by GAN models (a-d) with adversarial loss as in Equation (6) and (e-h) with (Lfinal) as in Equation (8). adversarial loss (LG, [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Comparison of neural network architectures (CNN-CNN, LSTM-CNN, CNN-LSTM, LSTM-LSTM) across performance and computational metrics. (a) Total loss (Lfinal) over epochs for LSTM-CNN, where Lfinal combine GAN loss and spectral loss. The red line represents the spectral los…
Figure 5
Figure 5. Figure 5: Comparison of real and generated data using GAN and HMM models. (a): The time series plot shows the differences between real and generated data over time for the LSTM-CNN model. (b): The histogram (blue) represents the real data distribution while the red and black cur…
Figure 6
Figure 6. Figure 6: (a) Histogram of log-transformed velocities in real data with the density estimation from a Markov model, illustrating the similarity in velocity distributions. The real data (blue bars) aligns closely with the Markov model’s Kernel Density Estimation (KDE) curve (dash…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

54 extracted references · 29 canonical work pages

  1. [1]

    W., Hayhoe, M

    Tatler, B. W., Hayhoe, M. M., Land, M. F. & Ballard, D. H. Eye guidance in natural vision: Reinterpreting salience. J. Vis. 11, 5–5, DOI: 10.1167/11.5.5 (2011)

    work page doi:10.1167/11.5.5 2011

  2. [2]

    E., Langrock, R., Walters, J

    McKellar, A. E., Langrock, R., Walters, J. R. & Kesler, D. C. Using mixed hidden markov models to examine behavioral states in a cooperatively breeding bird. BEHA VIORAL ECOLOGY26, 148–157, DOI: 10.1093/beheco/aru171 (2015)

    work page doi:10.1093/beheco/aru171 2015

  3. [3]

    & Stockman, G

    Shapiro, L. & Stockman, G. Computer Vision (Prentice Hall, the University of California, 2001). 11/16

    work page 2001

  4. [4]

    Ball, L. J. & Richardson, B. H. Eye Movement in User Experience and Human–Computer Interaction Research , 165–183 (Springer US, New York, NY , 2022)

    work page 2022

  5. [5]

    A., Thomas, L., Wilcox, C., Ovaskainen, O

    Patterson, T. A., Thomas, L., Wilcox, C., Ovaskainen, O. & Matthiopoulos, J. State–space models of individual animal movement. Trends Ecol. Evol. 23, 87–94, DOI: https://doi.org/10.1016/j.tree.2007.10.009 (2008)

    doi:10.1016/j.tree.2007.10.009 2008

  6. [6]

    C., Kello, C

    Rhodes, S. C., Kello, C. T. & Kerster, B. S. A comparison of visual search and foraging in humans. Vis. Cogn. 22, 704–728 (2014)

    work page 2014

  7. [7]

    E., Rhodes, T

    Kerster, B. E., Rhodes, T. & Kello, C. T. Spatial memory in foraging games. Cognition 148, 85–96, DOI: https: //doi.org/10.1016/j.cognition.2015.12.015 (2016)

    work page doi:10.1016/j.cognition.2015.12.015 2016

  8. [8]

    & Khanna, R

    Awad, M. & Khanna, R. Hidden Markov Model, 81–104 (Apress, Berkeley, CA, 2015)

    work page 2015

Show all 54 references
  1. [9]

    Langrock, R. et al. Flexible and practical modeling of animal telemetry data: hidden markov models and extensions. Ecology 93, 2336–2342, DOI: 10.1890/11-2241.1 (2012)

    2012 doi

  2. [10]

    Langrock, R. et al. Modelling group dynamic animal movement. METHODS IN ECOLOGY AND EVOLUTION 5, 190–199, DOI: 10.1111/2041-210X.12155 (2014)

    2014 doi

  3. [11]

    Analysis of animal accelerometer data using hidden markov models

    Leos-Barajas, V .et al. Analysis of animal accelerometer data using hidden markov models. Methods Ecol. Evol. 8, 161–173, DOI: https://doi.org/10.1111/2041-210X.12657 (2017)

    2017 doi

  4. [12]

    Michelot, T. et al. Estimation and simulation of foraging trips in land-based marine predators. Ecology 98, 1932–1944, DOI: https://doi.org/10.1002/ecy.1880 (2017)

    2017 doi

  5. [13]

    & Gourisaria, M

    GM, H., Sahu, A. & Gourisaria, M. Gm score: Incorporating inter-class and intra-class generator diversity, dis- criminability of latent space, and sample fidelity for evaluating gans. Arab. J. for Sci. Eng. 48, 2207–2230, DOI: 10.1007/s13369-022-07054-3 (2023)

    2023 doi

  6. [14]

    & Elhassouny, A

    Oussidi, A. & Elhassouny, A. Deep generative models: Survey. In 2018 International Conference on Intelligent Systems and Computer Vision (ISCV), 1–8, DOI: 10.1109/ISACV .2018.8354080 (IEEE, Morocco, 2018)

    2018

  7. [15]

    Variational autoencoder for deep learning of images, labels and captions

    Pu, Y .et al. Variational autoencoder for deep learning of images, labels and captions. In Proceedings of the 30th International Conference on Neural Information Processing Systems , NIPS’16, 2360–2368 (Curran Associates Inc., Red Hook, NY , USA, 2016)

    2016

  8. [16]

    Goodfellow, I. J. et al. Generative adversarial networks (2014). 1406.2661

    2014 arXiv

  9. [17]

    Goodfellow, I. et al. Generative adversarial networks. Commun. ACM 63, 139–144, DOI: 10.1145/3422622 (2020)

    2020 doi

  10. [18]

    & Zhou, Y

    Gonog, L. & Zhou, Y . A review: Generative adversarial networks. In2019 14th IEEE Conference on Industrial Electronics and Applications (ICIEA), 505–510, DOI: 10.1109/ICIEA.2019.8833686 (IEEE, Chaina, 2019)

    2019

  11. [19]

    V ., Gerget, O

    Laptev, V . V ., Gerget, O. M. & Markova, N. A. Generative Models Based on VAE and GAN for New Medical Data Synthesis, 217–226 (Springer International Publishing, Cham, 2021)

    2021

  12. [20]

    & Simonyan, K

    Brock, A., Donahue, J. & Simonyan, K. Large scale gan training for high fidelity natural image synthesis (2019). 1809.11096

    2019 arXiv

  13. [21]

    & Efros, A

    Pathak, D., Krähenbühl, P., Donahue, J., Darrell, T. & Efros, A. A. Context encoders: Feature learning by inpainting. IEE 1, 2536–2544, DOI: 10.1109/CVPR.2016.278 (2016)

    2016 doi

  14. [22]

    & Ward, T

    Brophy, E., Wang, Z., She, Q. & Ward, T. Generative adversarial networks in time series: A systematic literature review. ACM Comput. Surv. 55, DOI: 10.1145/3559540 (2023)

    2023 doi

  15. [23]

    & Alahari, K

    Shmelkov, K., Schmid, C. & Alahari, K. How good is my gan? In Ferrari, V ., Hebert, M., Sminchisescu, C. & Weiss, Y . (eds.) Computer Vision – ECCV 2018 , 218–234 (Springer International Publishing, Cham, 2018)

    2018

  16. [24]

    & Wang, H

    Cai, L., Chen, Y ., Cai, N., Cheng, W. & Wang, H. Utilizing amari-alpha divergence to stabilize the training of generative adversarial networks. Entropy 22, DOI: 10.3390/e22040410 (2020)

    2020 doi

  17. [25]

    Goodfellow, I. et al. Generative adversarial nets. In Ghahramani, Z., Welling, M., Cortes, C., Lawrence, N. & Weinberger, K. (eds.) Advances in Neural Information Processing Systems , vol. 27 (Curran Associates, Inc., Cambridge, MA, 2014)

    2014

  18. [26]

    Kumar, M. R. P. & Jayagopal, P. Generative adversarial networks: a survey on applications and challenges.Int. J. Multimed. Inf. Retr. 10, 1 – 24 (2020)

    2020

  19. [27]

    Gerych, W. et al. Stabilizing adversarial training for generative networks. In 2023 IEEE International Conference on Big Data (BigData), 5223–5232, DOI: 10.1109/BigData59044.2023.10386654 (IEEE Computer Society, Los Alamitos, CA, USA, 2023). 12/16

    2023

  20. [28]

    Computational complexity of neural networks: a survey

    Orponen, P. Computational complexity of neural networks: a survey. Nord. J. Comput. 1, 94–110 (1994)

    1994

  21. [29]

    & Bottou, L

    Arjovsky, M., Chintala, S. & Bottou, L. Wasserstein gan (2017). 1701.07875

    2017 arXiv

  22. [30]

    & Ramakrishnan, G

    Garg, S. & Ramakrishnan, G. Advances in quantum deep learning: An overview (2020). 2005.04316

    2020 arXiv

  23. [31]

    & Svore, K

    Wiebe, N., Kapoor, A. & Svore, K. M. Quantum deep learning (2015). 1412.3489

    2015 arXiv

  24. [32]

    R., Yazidi, A

    Lencastre, P., Gjersdal, M., Gorjão, L. R., Yazidi, A. & Lind, P. G. Modern ai versus century-old mathematical models: How far can we go with generative adversarial networks to reproduce stochastic processes? Phys. D: Nonlinear Phenom. 453, 133831, DOI: https://doi.org/10.1016...

    2023

  25. [33]

    & Zheng, S

    Situ, H., He, Z., Wang, Y ., Li, L. & Zheng, S. Quantum generative adversarial network for generating discrete distribution. Inf. Sci. 538, 193–208, DOI: https://doi.org/10.1016/j.ins.2020.05.127 (2020)

    2020 doi

  26. [34]

    Fedus, W. et al. Many paths to equilibrium: Gans do not need to decrease a divergence at every step (2018). 1710.08446

    2018 arXiv

  27. [35]

    Ledig, C. et al. Photo-Realistic Single Image Super-Resolution Using a Generative Adversarial Network . In 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , 105–114, DOI: 10.1109/CVPR.2017.19 (IEEE Computer Society, Los Alamitos, CA, USA, 2017)

    2017 doi

  28. [36]

    & Keuper, J

    Durall, R., Keuper, M. & Keuper, J. Watch Your Up-Convolution: CNN Based Generative Deep Neural Networks Are Failing to Reproduce Spectral Distributions . In 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 7887–7896, DOI: 10.1109/CVPR42600.2020.0079...

    2020

  29. [37]

    & Bertrand, S

    Roy, A., Fablet, R. & Bertrand, S. L. Using generative adversarial networks (gan) to simulate central-place foraging trajectories. Methods Ecol. Evol. 13, 1275–1287, DOI: https://doi.org/10.1111/2041-210X.13853 (2022)

    2022 doi

  30. [38]

    Alom, M. Z. et al. A state-of-the-art survey on deep learning theory and architectures. Electronics 8, DOI: 10.3390/ electronics8030292 (2019)

    2019

  31. [39]

    & Ren, J

    Wang, X., Zhao, X. & Ren, J. A new type of eye movement model based on recurrent neural networks for simulating the gaze behavior of human reading. Complexity 2019, 8641074, DOI: https://doi.org/10.1155/2019/8641074 (2019)

    2019 doi

  32. [40]

    & Lecomte, N

    Christin, S., Hervet, E. & Lecomte, N. Applications for deep learning in ecology. Methods Ecol. Evol. 10, 1632–1644, DOI: https://doi.org/10.1111/2041-210X.13256 (2019)

    2019 doi

  33. [41]

    Esteban, C., Hyland, S. L. & Rätsch, G. Real-valued (medical) time series generation with recurrent conditional gans (2017). 1706.02633

    2017 arXiv

  34. [42]

    Gao, N. et al. Generative adversarial networks for spatio-temporal data: A survey. ACM Trans. Intell. Syst. Technol. 13, DOI: 10.1145/3474838 (2022)

    2022 doi

  35. [43]

    & Chintala, S

    Radford, A., Metz, L. & Chintala, S. Unsupervised representation learning with deep convolutional generative adversarial networks. CoRR abs/1511.06434 (2015)

    2015 arXiv

  36. [44]

    Discrete-Time Markov Chains, 63–109 (Springer International Publishing, Cham, 2020)

    Brémaud, P. Discrete-Time Markov Chains, 63–109 (Springer International Publishing, Cham, 2020)

    2020

  37. [45]

    A tutorial on hidden markov models and selected applications in speech recognition

    Rabiner, L. A tutorial on hidden markov models and selected applications in speech recognition. Proc. IEEE 77, 257–286, DOI: 10.1109/5.18626 (1989)

    1989 doi

  38. [46]

    E., Petrie, T., Soules, G

    Baum, L. E., Petrie, T., Soules, G. & Weiss, N. A maximization technique occurring in the statistical analysis of probabilistic functions of markov chains. The Annals Math. Stat. 41, 164–171, DOI: 10.1214/aoms/1177697196 (1970)

    1970

  39. [47]

    EyeLink 1000 Manual (2024)

    SR Research Ltd. EyeLink 1000 Manual (2024). Accessed: October 21, 2024

    2024

  40. [48]

    Where’s Waldo (Candlewick Press, Somerville, MA, USA, 2007)

    Handford, M. Where’s Waldo (Candlewick Press, Somerville, MA, USA, 2007)

    2007

  41. [49]

    B., Cavalcanti, G

    de Amorim, L. B., Cavalcanti, G. D. & Cruz, R. M. The choice of scaling technique matters for classification performance. Appl. Soft Comput. 133, 109924, DOI: https://doi.org/10.1016/j.asoc.2022.109924 (2023)

    2023

  42. [50]

    On the jensen–shannon symmetrization of distances relying on abstract means

    Nielsen, F. On the jensen–shannon symmetrization of distances relying on abstract means. Entropy 21, 485, DOI: 10.3390/e21050485 (2019)

    2019 doi

  43. [51]

    From GAN to WGAN (2019)

    Weng, L. From GAN to WGAN (2019). 1904.08994

    2019 arXiv

  44. [52]

    & Lind, P

    Bhandari, S., Lencastre, P. & Lind, P. Modeling stochastic eye tracking data: A comparison of quantum generative adversarial networks and markov models. In Proceedings of the Genetic and Evolutionary Computation Conference Companion, GECCO ’24 Companion, 1934–1941, DOI: https:...

    1934

  45. [53]

    Density Estimation for Statistics and Data Analysis (Routledge, New York, NY , USA, 1998), 1st edn

    Silverman, B. Density Estimation for Statistics and Data Analysis (Routledge, New York, NY , USA, 1998), 1st edn

    1998

  46. [54]

    Chapter 8 - variants of hsmms

    Yu, S.-Z. Chapter 8 - variants of hsmms. In Yu, S.-Z. (ed.) Hidden Semi-Markov Models , 143–161, DOI: https: //doi.org/10.1016/B978-0-12-802767-7.00008-5 (Elsevier, Boston, 2016). 13/16 A Mathematical Foundation for Markov and Hidden Markov Models A sequence of random variable...

    2016 doi

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.