Pith. sign in

REVIEW 5 major objections 5 minor 32 references

Synthetic Time Series Generation via Complex Networks

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

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

Pith's one-line read A quantile-graph inverse mapping generates synthetic time series that preserve marginal distributions and short-term dependencies, offering a training-free alternative to GAN-based generators.

desk verdict A clear but flawed empirical study of a known method: InvQG is a first-order Markov chain on quantile bins, and its fidelity evaluation is largely self-confirming. read the letter →

arxiv 2601.22879 v2 pith:O5MGFYUC submitted 2026-01-30 cs.LG

Jaime Vale , Vanessa Freitas Silva , Maria Eduarda Silva , Fernando Silva This is my paper · ORCID
classification cs.LG MSC 62M1005C82
keywords synthetictimeseriescomplexnetworksquantilegraphinversemappingMarkovtransitionmatrixdataaugmentationfidelitynetwork-basedgeneration
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 argues that a time series can be compressed into a small Markov transition matrix over quantile bins, and then resampled from that matrix to produce new series that resemble the original. The authors claim this Inverse Quantile Graph method preserves the distribution of values and the immediate lag-1 dependencies for a wide range of time series models, and that it matches or beats two GAN-based generators on fidelity, while requiring no training and only one hyperparameter (the number of quantiles). The known limitation is that long-range or higher-order temporal structure is not captured, because the underlying process is a first-order Markov chain. If true, practitioners get an interpretable, cheap, and easy-to-set-up tool for data augmentation and privacy-preserving data sharing.

What carries the argument

The central object is the Quantile Graph: a directed, weighted graph whose nodes are Q empirical quantile bins of the series and whose edge weights are the empirical transition counts between consecutive values, normalized into a Markov transition matrix W. The inverse mapping walks this Markov chain—at each step sampling a uniform value inside the current quantile's range, then moving to the next bin according to row probabilities—to emit a new series. The QG matrix is the single carrier of all information about the original series; the number of quantiles Q is the only tunable parameter. Its work is to convert the series' marginal distribution (bin boundaries) and its lag-1 dynamics (trans

What would settle it

A concrete check: take an original series, build the QG, generate many synthetic series, then use a separate held-out portion of the same underlying process (or a new series from the same model) as the comparison reference; if the synthetic series' statistical and network features deviate from the held-out series as much as they do from the training series, the fidelity claim fails. A second decisive test is to measure the autocorrelation function at lags beyond 1 on the synthetic versus original for an AR(2) series with strong pseudo-periodic peaks: the current method should fail to reproduce

Watch

Extended reading notes

Core claim

The central claim is that the inverse of the Quantile Graph mapping is a viable general-purpose synthetic time series generator. Concretely: split the range of the original series into Q quantiles; count transitions between consecutive values to form a Q×Q Markov matrix; then synthesize a new series by picking a starting quantile, drawing a value uniformly within that quantile's range, transitioning to the next quantile according to the matrix row, and repeating. The paper reports that across eleven simulated time series models and twenty-two real-world hourly electricity demand series, the synthetic series produced this way retain the statistical and network-topological features of the orig

Load-bearing premise

The load-bearing premise is that evaluating a generator on the very series used to build it is a valid test of fidelity: the transition matrix and quantile ranges are estimated from each original series, and the same series are then used as the reference for comparing statistical and topological features, so any error the generator makes is measured against the data it was tuned to reproduce.

Editorial extensions

If this is right

  • Synthetic series with the same marginal distribution and lag-1 autocorrelation can be produced for AR, GARCH, SETAR, INAR, count, and noise-type series, enabling data augmentation when real series are scarce.
  • Because the generator needs no training and has a single parameter (Q), it can be deployed in applications where GAN training is impractical, such as very long high-resolution series or privacy-sensitive settings.
  • The Markov structure makes the method interpretable: a user can inspect the transition matrix to see exactly what temporal dependencies are being preserved, and can see why long-range structure is lost.
  • In clustering tasks, networks built from synthetic series group with their original model in PCA and clustering evaluation, indicating the synthetic data retains utility for downstream analysis.
  • The known failure on AR(2), ARFIMA, and HMM-like series delimits the method: long-range or higher-order dependencies are not preserved, so users should not use InvQG for series with strong cycles or long memory.

Reading between the lines

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

  • A natural, untested extension is to use higher-order Quantile Graphs (e.g., lag-2 or lag-k transition matrices) to capture the AR(2)/ARFIMA failures; the current paper only tests the lag-1 version, so its stated limitation is a property of this specific configuration, not of the quantile-graph idea in general.
  • Because within-bin values are drawn uniformly, the method cannot reproduce conditional shapes inside a quantile; pairing the transition matrix with a learned conditional density per bin (instead of uniform) is a testable modification that could fix bias in features like curvature and entropy.
  • If the fidelity holds under a true held-out evaluation, the method also suggests a compact, privacy-preserving representation: sharing only the quantile boundaries and transition matrix (plus Q) releases a usable data description without sharing raw values—though whether this genuinely protects privacy would need a separate test, since the synthetic series can come arbitrarily close to original va
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

5 major / 5 minor

Summary. The paper proposes a synthetic time-series generator, InvQG, built from the quantile-graph (QG) representation. For a given series Y, Algorithm 1 estimates Q empirical quantiles and the lag-1 Markov transition matrix W; Algorithm 2 generates a new series by sampling uniformly within each quantile bin and transitioning according to W. The authors evaluate fidelity with tsfeatures and NetF topological features and compare visually with TimeGAN and DoppelGANger on simulated and real smart-meter data. The stated claim is that InvQG preserves marginal distributions and short-term temporal dependencies across many models, with predictable failures on long-range/higher-order dynamics (AR2, ARFIMA, HMM), while remaining competitive with GAN baselines.

Significance. If the claims hold, InvQG would be a useful interpretable, training-free baseline for synthetic time series generation, requiring only one hyperparameter (Q) and no optimization. The method is simple, reproducible in principle (code is on GitHub), and the paper honestly identifies its limitations on long-memory and higher-order models. However, the empirical support is currently weakened by an in-sample evaluation design, missing reporting of Q, and a qualitative comparison against GANs. The core idea is sound and worth publishing after substantial revision; as written, the evidence does not yet establish that the method is competitive in a broad sense.

major comments (5)
  1. [Section 4.1, 4.3, 4.5] The evaluation is in-sample in a way that partly restates the algorithm. Algorithm 1 builds W and the quantile bins from each original Y_i, and Sections 4.3 and 4.5 compare tsfeatures and NetF features of the same Y_i against InvQG(Y_i). Therefore e_acf1, x_acf1, and quantile-graph topology are expected to match by construction. The paper does not report any held-out split or external validation. Please add an out-of-sample evaluation, e.g., estimate W on a training segment and evaluate on a test segment, or compare generated feature distributions against independent realizations from the same generating process.
  2. [Section 4 and Algorithm 1] The number of quantiles Q is never reported for any experiment, despite being the only hyperparameter and central to the method. Without Q, the experiments cannot be reproduced, and the sensitivity of the results to Q cannot be assessed. Please state Q for each dataset and ideally show results for several Q values.
  3. [Table 3 and Figures 5-6] The paper says the statistical features are 'nearly identical', but Table 3 shows large paired differences for several models: e_acf10 mean differences are 0.852 (SD 0.125) for AR2 and 0.854 (SD 0.244) for ARFIMA; ARIMA shows e_acf10 -1.017 (SD 1.193). The accompanying ACF plots confirm these are not negligible. Please quantify the deviations with effect sizes or confidence intervals, and state which models meet a pre-specified fidelity tolerance.
  4. [Section 4.4] The comparison with TimeGAN and DoppelGANger is not yet a rigorous benchmark. It relies on visual t-SNE inspection of one (or a few) selected series per model, and it is unclear whether the GANs were trained on a single series or on multiple series; the two GANs are known to need multiple training examples. The claim that InvQG 'surpasses' GAN methods is unsupported without quantitative metrics (e.g., feature-wise distances, MMD, or discriminator accuracy) and a defined training protocol for the baselines.
  5. [Section 3-4, global] InvQG is, by construction, a first-order Markov chain on empirical quantile bins with uniform within-bin sampling. The paper never compares against simpler baselines such as i.i.d. resampling of empirical marginals, a first-order Markov chain with raw values, or a Markov chain with a different binning. Such baselines would clarify whether the reported fidelity is due to the QG machinery or to the elementary Markov property. Without this comparison, the 'competitive alternative' claim is not yet established.
minor comments (5)
  1. [Algorithm 2] Line 7 uses quantiles[ni-1] when ni=1; quantiles[0] is likely out of bounds or undefined. The lower bound of the first quantile should be the minimum of the series, not an indexed array element.
  2. [Section 4.2.2] The text says InvQG was applied for imputation of missing smart-meter values, but no details or results of this imputation are given. Either provide the imputation protocol and evaluation, or remove the remark.
  3. [Table 3] Abbreviated column headers 'linea.' and 'curvat.' are not defined; spell out 'linearity' and 'curvature'.
  4. [Figures 9 and 13] The arrow lengths are said to indicate feature contribution to the PCs, but the mapping from arrow length to loadings is not explained. Add a sentence describing the construction.
  5. [Global] Minor typos: 'DopplerGANger' appears in several places instead of 'DoppelGANger'; 'data shortage due t privacy' in the Introduction; 'which can hinder' should be 'which can hinder'. Also consider defining 'fidelity' and 'utility' formally in Section 3.

Circularity Check

1 steps flagged · score 6.0 of 10

InvQG's fidelity for marginal and lag-1 properties is largely a restatement of its construction, because the transition matrix and quantile bins used to generate Y* are estimated from the same Y used as the fidelity benchmark.

  1. fitted input called prediction [Section 3 (Algorithms 1-2, Appendix A) and Section 4.3]
    "Algorithm 1: 'W←Array(Q,Q); ... q←quantile(Y, quantiles); ... W[vi][vj]←W[vi][vj]+1' ... 'return W, quantiles'. Algorithm 2: 'W, quantiles←QG(Y,Q)' ... 'X[t]←runif(min range, maxrange); ni←select next quantil(W[ni])'. Section 4.3: 'InvQG predominantly captures short-term dependencies... by focusing on lag 1 probability transitions' and 'This suggests that, as predicted, InvQG effectively captures the immediate relationships between consecutive observations'."

    The generator's parameters W and the quantile bins are fitted to the original series Y in Algorithm 1: W is the empirical lag-1 quantile transition matrix and the bins are the empirical quantiles of Y. Algorithm 2 then generates Y* by sampling uniformly inside those same bins and moving among bins with W. Therefore, the synthetic series reproduces the original's quantile occupancy and lag-1 quantile transition frequencies in expectation. The fidelity evidence in Section 4.3 measures exactly these properties—marginal distribution features and e_acf1/x_acf1—on the same Y used for estimation. A positive result is thus not an independent confirmation of a generative capability; it is a restatement of the fitted transition structure. The paper's own admission that lag-10 and entropy features di

full rationale

The main circularity is in the fidelity evaluation. InvQG is built from the empirical quantile bins and the lag-1 transition matrix W of each original Y (Algorithms 1-2). The evaluation then scores the synthetic Y* against the same Y using features dominated by marginal distribution and lag-1 autocorrelation (Section 4.3), and by quantile-graph topology (Section 4.5). Consequently, the headline result that InvQG 'preserves marginal distributions and short-term temporal dependencies' is largely an in-sample restatement: the model is asked to reproduce the very quantities used to define it. The paper's discussion of AR2, ARFIMA, and HMM limitations shows that only lag-1 structure is encoded, making the 'predictable limitations' the complement of the fitted input. This warrants a score of 6. The self-citations are not the main problem. The QG/InvQG mapping is attributed to Campanharo et al. (2011), and the NetF feature set and preliminary claims that network mappings capture dynamics are used as evaluation tools or background context rather than as a substitute for the empirical comparison. No uniqueness theorem or ansatz is smuggled in via self-citation. The comparison against GANs is weakened by the absence of a trivial first-order Markov-chain baseline and by evaluating on the same data, but that is a fairness/validation concern more than a definitional circularity. The central circular step is the fidelity claim itself, which reduces by construction to the fitted transition matrix and quantile bins.

Assumptions & free parameters 2 free parameters · 4 assumptions · 1 invented entities

The central claim rests on the Markov-chain-in-quantile-bins model and the in-sample evaluation design. There are no new physical entities; the model is a reparameterized first-order Markov chain with an unspecified granularity Q.

free parameters (2)
  • Q (number of quantiles) = not reported
    Algorithms 1 and 2 take Q as input; the paper never states the value used for the experiment, and no sensitivity analysis is provided. This single hyperparameter controls bin granularity and therefore the fidelity of the synthetic series.
  • First-order transition matrix W (empirical) = estimated from each original series
    The generative model is the QxQ Markov transition matrix extracted from the original series (Algorithm 1). It is fitted to the very data used later for fidelity evaluation, which is the source of the in-sample circularity. Because W is data-derived, the 'prediction' of short-term dependencies is not independent.
assumptions (4)
  • domain assumption First-order Markov dependence is sufficient to capture the temporal structure that matters for the claimed fidelity.
    Algorithm 2 Step 4.3 transitions using only the current quantile; no higher-order or longer-range information is represented. The paper acknowledges this limits AR2/ARFIMA/HMM cases but still uses the method as a general generator.
  • ad hoc to paper Uniform sampling within each quantile range faithfully represents the original within-bin distribution.
    Algorithm 2 Step 4.2 samples Y_t uniformly between quantile boundaries; if the original values within a bin are not uniform, the synthetic marginal distribution is only a stepwise-uniform approximation. This is not tested.
  • ad hoc to paper Fidelity can be validly assessed on the same data used to construct the generator (in-sample evaluation).
    Sections 4.3-4.5 compare synthetic series against the original series used to build W and quantiles. No held-out split or external validation is performed.
  • domain assumption The selected feature sets (tsfeatures statistics and NetF topological metrics) are sufficient proxies for all practically relevant properties of time series.
    The fidelity verdict is based entirely on these features; any property not captured by them would be invisible to the evaluation.
invented entities (1)
  • none
    purpose: No new entities are introduced.
    The method reuses known quantile-graph concepts; no new particles, forces, or dimensions are postulated.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Synthetic Time Series Generation via Complex Networks." pith.science (2026). https://pith.science/paper/O5MGFYUC

@misc{pith2026260122879,
  author       = {Pith},
  title        = {Pith review of: Synthetic Time Series Generation via Complex Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/O5MGFYUC}},
  note         = {Machine review of arXiv:2601.22879}
}
read the original abstract

Time series data are essential for a wide range of applications, yet access to high-quality datasets is often constrained by privacy concerns, acquisition costs, and labelling challenges. Synthetic time series generation has emerged as a promising approach to address these limitations. In this work, we investigate the use of complex network mappings for synthetic time series generation, focusing on the Quantile Graph (QG) representation and its inverse. While the inverse QG mapping has been previously proposed, its potential as a general-purpose data generator has not been systematically evaluated. We address this gap through a comprehensive empirical study assessing both the fidelity and utility of synthetic time series generated by the Inverse Quantile Graph (InvQG) framework. The evaluation combines statistical feature analysis, network-based topological characteristics, and performance in downstream clustering and classification tasks, using simulated and real-world datasets. The results show that InvQG effectively preserves marginal distributions and short-term temporal dependencies across a wide range of models, while exhibiting predictable limitations in capturing long-range or higher-order dynamics.

Figures

Figures reproduced from arXiv: 2601.22879 by the authors.

Figure 1
Figure 1. Illustration of Quantile Graph mapping method with a toy time series. [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Illustration of the Inverse Quantile Graph ( [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Diagram of the methodology implemented in this work. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (22 more)
Figure 4
Figure 4. Figure 4: Boxplots of the paired samples statistical features: [PITH_FULL_IMAGE:figures/full_fig_p009_4.png]
Figure 5
Figure 5. Figure 5: Boxplots of the paired samples statistical features: [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: (a) Autocorrelation function (ACF) plot of an original [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 7
Figure 7. Figure 7: (a) Autocorrelation function (ACF) plot of an original [PITH_FULL_IMAGE:figures/full_fig_p011_7.png]
Figure 8
Figure 8. Figure 8: Comparison of an original and synthetic time series instance for (a) [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]
Figure 9
Figure 9. Figure 9: Plot of the first two principal components (PCs) for the original and synthetic time series data. [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]
Figure 10
Figure 10. Figure 10: t-SNE comparison between InvQG (first row), TimeGAN (second row) and DoppelGANger (third row) methods for WN, AR -0.5, AR 0.5, AR 0.9 and AR2 models. 13 [PITH_FULL_IMAGE:figures/full_fig_p013_10.png]
Figure 11
Figure 11. Figure 11: (continued) t-SNE comparison between InvQG (first row), TimeGAN (second row) and Doppel￾GANger (third row) methods for ARFIMA, ARIMA, GARCH, INAR, SETAR and HMM models. Based on the t-SNE visualizations, we observe notable differences in the ability of each method to …
Figure 12
Figure 12. Figure 12: t-SNE comparison between InvQG (first row), TimeGAN (second row) and DoppelGANger (third row) methods for the smart meters of houses 5, 6, 7, 8, 14, and 17 (see Appendix D for remaining houses). 4.5 Utility Analysis via Topological Network Features In this section, we…
Figure 13
Figure 13. Figure 13: Plot of the first two principal components (PCs) for the original and synthetic time series network [PITH_FULL_IMAGE:figures/full_fig_p016_13.png]
Figure 14
Figure 14. Figure 14: (a) Autocorrelation function (ACF) plot of an original [PITH_FULL_IMAGE:figures/full_fig_p025_14.png]
Figure 15
Figure 15. Figure 15: (a) Autocorrelation function (ACF) plot of an original [PITH_FULL_IMAGE:figures/full_fig_p025_15.png]
Figure 16
Figure 16. Figure 16: (a) Autocorrelation function (ACF) plot of an original [PITH_FULL_IMAGE:figures/full_fig_p026_16.png]
Figure 17
Figure 17. Figure 17: (a) Autocorrelation function (ACF) plot of an original [PITH_FULL_IMAGE:figures/full_fig_p026_17.png]
Figure 18
Figure 18. Figure 18: (a) Autocorrelation function (ACF) plot of an original [PITH_FULL_IMAGE:figures/full_fig_p027_18.png]
Figure 19
Figure 19. Figure 19: (a) Autocorrelation function (ACF) plot of an original [PITH_FULL_IMAGE:figures/full_fig_p027_19.png]
Figure 20
Figure 20. Figure 20: (a) Autocorrelation function (ACF) plot of an original [PITH_FULL_IMAGE:figures/full_fig_p028_20.png]
Figure 21
Figure 21. Figure 21: (a) Autocorrelation function (ACF) plot of an original [PITH_FULL_IMAGE:figures/full_fig_p028_21.png]
Figure 22
Figure 22. Figure 22: (a) Autocorrelation function (ACF) plot of an original [PITH_FULL_IMAGE:figures/full_fig_p029_22.png]
Figure 23
Figure 23. Figure 23: t-SNE comparison between InvQG (first row), TimeGAN (second row) and DoppelGANger (third row) methods for the smart meters of houses 1, 2, 3, 4, 9, and 10 (continued). 29 [PITH_FULL_IMAGE:figures/full_fig_p029_23.png]
Figure 24
Figure 24. Figure 24: t-SNE comparison between InvQG (first row), TimeGAN (second row) and DoppelGANger (third row) methods for the smart meters of houses 11, 12, 13, 15, 16, and 18 (continued). House 22 [PITH_FULL_IMAGE:figures/full_fig_p030_24.png]
Figure 25
Figure 25. Figure 25: t-SNE comparison between InvQG (first row), TimeGAN (second row) and DoppelGANger (third row) methods for the smart meters of houses 19, 20, 21, and 22 (continued). 30 [PITH_FULL_IMAGE:figures/full_fig_p030_25.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

32 extracted references · 1 canonical work pages

  1. [1]

    R., D \'a n, G., Miorandi, D., and Chlamtac, I

    Asghar, M. R., D \'a n, G., Miorandi, D., and Chlamtac, I. (2017). Smart meter data privacy: A survey. IEEE Communications Surveys & Tutorials , 19(4):2820--2835

    2017

  2. [2]

    Benhamouda, F., Joye, M., and Libert, B. (2016). A new framework for privacy-preserving aggregation of time-series data. ACM Transactions on Information and System Security (TISSEC) , 18(3):1--21

    2016

  3. [3]

    Brophy, E., Wang, Z., She, Q., and Ward, T. (2023). Generative adversarial networks in time series: A systematic literature review. ACM Computing Surveys , 55(10):1--31

    2023

  4. [4]

    S., Doescher, E., and Ramos, F

    Campanharo, A. S., Doescher, E., and Ramos, F. M. (2020). Application of quantile graphs to the automated analysis of eeg signals. Neural Processing Letters , 52(1):5--20

    2020

  5. [5]

    S., Sirer, M

    Campanharo, A. S., Sirer, M. I., Malmgren, R. D., Ramos, F. M., and Amaral, L. A. N. (2011). Duality between time series and networks. PloS one , 6(8):e23378

    2011

  6. [6]

    Chowdhury, M. J. M., Colman, A., Kabir, M. A., Han, J., and Sarda, P. (2019). Continuous authorization in subject-driven data sharing using wearable devices. In 2019 18th IEEE International Conference On Trust, Security And Privacy In Computing And Communications/13th IEEE International Conference On Big Data Science And Engineering (TrustCom/BigDataSE) ,...

    2019

  7. [7]

    K., Prabowo, A., Rahaman, M

    Gao, N., Xue, H., Shao, W., Zhao, S., Qin, K. K., Prabowo, A., Rahaman, M. S., and Salim, F. D. (2022). Generative adversarial networks for spatio-temporal data: A survey. ACM Transactions on Intelligent Systems and Technology (TIST) , 13(2):1--25

    2022

  8. [8]

    and Newe, T

    Halder, S. and Newe, T. (2022). Enabling secure time-series data sharing via homomorphic encryption in cloud-assisted iiot. Future Generation Computer Systems , 133:351--363

    2022

Show all 32 references
  1. [9]

    Hu, Z., Wu, T., Zhang, Y., Li, J., and Jiang, L. (2020). Time series anomaly detection based on graph convolutional networks. In 2020 2nd International Conference on Applied Machine Learning (ICAML) , pages 138--145. IEEE

    2020

  2. [10]

    Huang, Y., Mao, X., and Deng, Y. (2021). Natural visibility encoding for time series and its application in stock trend prediction. Knowledge-Based Systems , 232:107478

    2021

  3. [11]

    L., Esteban, C., and R \"a tsch, G

    Hyland, S. L., Esteban, C., and R \"a tsch, G. (2018). Real-valued (medical) time series generation with recurrent conditional GAN s

    2018

  4. [12]

    B., Hanqing, C., Lake, D

    Hyndman, R., Kang, Y., Montero-Manso, P., O'Hara-Wild, M., Talagala, T., Wang, E., Yang, Y., Taieb, S. B., Hanqing, C., Lake, D. K., Laptev, N., Moorman, J. R., and Zhang, B. (2023). tsfeatures: Time series feature extraction. Accessed: January 15, 2024

    2023

  5. [13]

    Iglesias, G., Talavera, E., Gonz \'a lez-Prieto, \'A ., Mozo, A., and G \'o mez-Canaval, S. (2023). Data augmentation techniques in time series domain: a survey and taxonomy. Neural Computing and Applications , 35(14):10123--10145

    2023

  6. [14]

    Iwana, B. K. and Uchida, S. (2021). An empirical survey of data augmentation for time series classification with neural networks. Plos one , 16(7):e0254841

    2021

  7. [15]

    J., and Li, F

    Kang, Y., Hyndman, R. J., and Li, F. (2020). GRATIS : Generating time series with diverse and controllable characteristics. Statistical Analysis and Data Mining: The ASA Data Science Journal , 13(4):354--376

    2020

  8. [16]

    Kegel, L., Hahmann, M., and Lehner, W. (2018). Feature-based comparison and generation of time series. In Proceedings of the 30th international conference on scientific and statistical database management , pages 1--12

    2018

  9. [17]

    P., Welling, M., et al

    Kingma, D. P., Welling, M., et al. (2013). Auto-encoding variational bayes

    2013

  10. [18]

    and Schwung, A

    Klopries, H. and Schwung, A. (2024). ITF-GAN : Synthetic time series dataset generation and manipulation by interpretable features. Knowledge-Based Systems , 283:111131

    2024

  11. [19]

    Lacasa, L., Luque, B., Ballesteros, F., Luque, J., and Nuno, J. C. (2008). From time series to complex networks: The visibility graph. Proceedings of the National Academy of Sciences , 105(13):4972--4975

    2008

  12. [20]

    Leukam Lako, F., Lajoie-Mazenc, P., and Laurent, M. (2021). Privacy-preserving publication of time-series data in smart grid. Security and Communication Networks , 2021(1):6643566

    2021

  13. [21]

    Lin, Z., Jain, A., Wang, C., Fanti, G., and Sekar, V. (2019). Generating high-fidelity, synthetic time series datasets with doppelganger. arXiv preprint arXiv:1909.13403

    2019 arXiv

  14. [22]

    Lin, Z., Jain, A., Wang, C., Fanti, G., and Sekar, V. (2020). Using gans for sharing networked time series data: Challenges, initial promise, and open questions. In Proceedings of the ACM Internet Measurement Conference , pages 464--483

    2020

  15. [23]

    B., Holmstr m, O., Bak-Jensen, B., and S rensen, P

    Negra, N. B., Holmstr m, O., Bak-Jensen, B., and S rensen, P. (2008). Model of a synthetic wind speed time series generator. Wind Energy: An International Journal for Progress and Applications in Wind Power Conversion Technology , 11(2):193--209

    2008

  16. [24]

    C., Zaridis, D

    Pezoulas, V. C., Zaridis, D. I., Mylona, E., Androutsos, C., Apostolidis, K., Tachos, N. S., and Fotiadis, D. I. (2024). Synthetic data generation methods in healthcare: A review on open-source tools and methods. Computational and structural biotechnology journal , 23:2892--2910

    2024

  17. [25]

    and Thomson, M

    Richardson, I. and Thomson, M. (2010). One-minute resolution domestic electricity use data, 2008-2009. http://doi.org/10.5255/UKDA-SN-6583-1. [data collection]

    2010 doi

  18. [26]

    Roman, A.-S. (2023). Evaluating the privacy and utility of time-series data perturbation algorithms. Mathematics , 11(5):1260

    2023

  19. [27]

    Shi, E., Chan, H., Rieffel, E., Chow, R., and Song, D. (2011). Privacy-preserving aggregation of time-series data. In Annual Network & Distributed System Security Symposium (NDSS) . Internet Society

    2011

  20. [28]

    F., Silva, M

    Silva, V. F., Silva, M. E., Ribeiro, P., and Silva, F. (2021). Time series analysis via network science: Concepts and algorithms. Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery , 11(3):e1404

    2021

  21. [29]

    F., Silva, M

    Silva, V. F., Silva, M. E., Ribeiro, P., and Silva, F. (2022). Novel features for time series analysis: a complex networks approach. Data Mining and Knowledge Discovery , 36(3):1062--1101

    2022

  22. [30]

    F., Silva, M

    Silva, V. F., Silva, M. E., Ribeiro, P., and Silva, F. (2025). Multilayer quantile graph for multivariate time series analysis and dimensionality reduction. International Journal of Data Science and Analytics , 20(3):1795--1807

    2025

  23. [31]

    Wen, Q., Sun, L., Yang, F., Song, X., Gao, J., Wang, X., and Xu, H. (2021). Time series data augmentation for deep learning: A survey. In Proceedings of the Thirtieth International Joint Conference on Artificial Intelligence, IJCAI-21 , pages 4653--4660. Survey Track

    2021

  24. [32]

    Yoon, J., Jarrett, D., and Van der Schaar, M. (2019). Time-series generative adversarial networks. In Wallach, H., Larochelle, H., Beygelzimer, A., d Alch\' e -Buc, F., Fox, E., and Garnett, R., editors, Advances in neural information processing systems , volume 32

    2019

Pith tools

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