Pith. sign in

REVIEW 4 major objections 5 minor 46 references

Towards Foundation Auto-Encoders for Time-Series Anomaly Detection

T0 review · 4 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The paper argues that a modest dilated-convolutional variational auto-encoder, pretrained on normal univariate time-series, can track and flag anomalies on previously unseen series in a zero-shot fashion, and that its latent space…

desk verdict Univariate VAE sketch with honest intentions, but the decoder as written cannot produce the time-varying reconstructions shown. read the letter →

arxiv 2507.01875 v1 pith:VV3EAOVK submitted 2025-07-02 cs.LG cs.AI

Gastón García González , Pedro Casas , Emilio Martínez , Alicia Fernández This is my paper
classification cs.LGcs.AI
keywords time-seriesanomalydetectionvariationalauto-encoderdilatedconvolutionszero-shotlearningfoundationmodelsunivariatelatentspaceinterpretation
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

FAE is a variational auto-encoder with dilated causal convolutions that is trained on multiple univariate time-series of normal behavior and then applied to previously unseen series. The paper argues that such a model can track the expected mean and variance of normal operation well enough to flag anomalies on unseen series without retraining. If that holds, anomaly detection for a new monitoring system could be bootstrapped by a single pretrained model instead of bespoke per-series models. The paper supports the claim qualitatively with TELCO mobile-network series, a domain-shifted TELCO2, a zero-shot TS12 experiment, and 12 KDD2021 series.

What carries the argument

The central object is the dilated-convolutional variational auto-encoder: each encoder layer dilates by a factor $F^h$ (with $F$ the filter length and $h$ the layer depth), so the final output at position $T-1$ has seen the whole input $\mathbf{X}$. The number of layers $N$ is the minimum integer satisfying $T \le 2F^{N-1}$, tying architecture size directly to window length. The encoder's two parallel $J$-filter convolutional layers map this output to $\boldsymbol{\mu}_\mathbf{Z}$ and $\boldsymbol{\sigma}_\mathbf{Z}$; the reparameterization trick samples $\mathbf{Z}$, and the decoder, symmetric to the encoder, expands $\mathbf{Z}$ back to $\boldsymbol{\mu}_\mathbf{X}$, $\boldsymbol{\sigma}_\mathbf{X}$. This mechanism is what lets one model observe many series at once and produce an interpretable, reconstruction-based normality region.

What would settle it

Run FAE on the labeled TELCO test set and compute the precision-recall curve obtained by sweeping $\alpha$ in the normality rule; if no single $\alpha$ yields detection performance comparable to a per-series calibrated threshold (or to a simple residual z-score baseline), then the fixed-rule zero-shot claim is falsified. Applying the pretrained model to all 250 KDD2021 series and checking whether its normality regions align with the dataset's known anomalies would reveal whether the 'foundation' behavior extends beyond the 12 series shown.

Watch

Extended reading notes

Core claim

FAE adapts the DC-VAE architecture to univariate series: a stack of causal dilated convolutions expands the receptive field exponentially, the encoder compresses the last temporal position into a low-dimensional latent vector $\mathbf{Z}$, and a symmetric decoder repeats that vector across time to output per-timestep parameters $\boldsymbol{\mu}_\mathbf{X}$ and $\boldsymbol{\sigma}_\mathbf{X}$. A window $\mathbf{X}_t$ is normal when it stays within $\alpha$ times the predicted standard deviation of the predicted mean. The paper reports that when the model is trained on several normal series, the latent space organizes by hour of the day, by weekday versus weekend, and by day of the month, and that the decoder tracks seasonality and trends on held-out series; in the zero-shot TS12 experiment, removing TS12 from training leaves predictions almost unchanged as long as a series with similar seasonal and trend behavior is present.

Load-bearing premise

FAE's anomaly detection depends on the rule that a sample is normal only if its deviation from the predicted mean stays within $\alpha$ times the predicted standard deviation; the paper never gives the $\alpha$ used in its figures and provides no quantitative evidence that one fixed $\alpha$ works across heterogeneous series. If this rule needs per-series calibration, the zero-shot detection claim collapses even if the reconstruction tracking is accurate.

Editorial extensions

If this is right

  • A single FAE model can monitor all series in a multivariate system one at a time, relying only on univariate temporal structure.
  • When the model has seen a temporal pattern in one series, it can track the same pattern zero-shot in a related unseen series, as shown for TS12 after training on TS1 to TS11.
  • When the monitored domain shifts from TELCO to TELCO2, the pretrained univariate model keeps tracking, whereas the multivariate DC-VAE requires recalibration.
  • Projections of the latent space expose hour-of-day, weekday-versus-weekend, and monthly-trend structure, making the model's normality judgments interpretable.

Reading between the lines

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

  • The architecture's direct tie between window length $T$ and layer count suggests a simple scaling rule for building FAE variants at different temporal resolutions.
  • Because the univariate model discards cross-correlations, it may miss anomalies that appear only as spatial deviations; an ensemble that routes between univariate and multivariate detectors could combine both strengths.
  • The latent-space 'clock' suggests the VAE encodes the phase of daily seasonality; this phase structure could be exploited for explicit forecasting or for explaining why a point is anomalous.
  • The foundation-model claim would be tested more convincingly on the full 250-series KDD2021 set, since the reported experiments use only 12 of the 250 series and do not quantify detection performance on them.
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 FAE, a VAE with causal dilated convolutional encoder/decoder for univariate time-series modeling, intended as a pretrained 'foundation' model for zero-shot anomaly detection. The authors present qualitative results on the TELCO dataset and on 12 series from the KDD2021 dataset, including a partial zero-shot experiment on TS12, and argue that FAE tracks seasonal and trend behavior and yields interpretable latent representations.

Significance. If the architecture as described is actually able to produce the time-varying reconstructions shown, FAE would be a small-scale, interpretable foundation model for anomaly detection, and the latent-space analysis is a nice demonstration of VAE transparency. The paper also provides a useful departure point for applying VAEs to univariate time series. However, the current manuscript does not provide quantitative validation, and the architectural description contains a serious internal inconsistency; the significance of the contribution therefore cannot be assessed as it stands.

major comments (4)
  1. [Section 3, Figure 1, Figures 3, 9, 10] The decoder as described cannot produce time-varying reconstructions. The input to the decoder is the latent vector Z repeated T times, forming a constant sequence along the time axis. Since all decoder layers are causal dilated convolutions, a constant input produces a constant output for all time steps (except for boundary effects). Yet Figures 3, 9, and 10 show strongly time-varying μ_X and σ_X that track daily seasonality and trends. This is a load-bearing discrepancy: without a mechanism that breaks the temporal symmetry (e.g., per-timestep conditioning, a non-constant latent input, or autoregressive connections), the reported results cannot be reproduced by the architecture. The authors must either provide the actual decoder mechanism or correct the description.
  2. [Section 3 (anomaly rule) and Section 4] The paper defines the anomaly detection rule using a threshold α—a sample is normal only if its deviation from μ_X stays within α times σ_X—but never reports the α value used in Figures 7 and 8, and provides no quantitative anomaly detection results (e.g., precision, recall, F1, or ROC). The evaluation is entirely qualitative, based on visual inspection, with no error bars and no baselines beyond the earlier DC-VAE, which is also not quantified. Without a specified threshold and quantitative metrics, the claim that FAE detects anomalies on unseen data is unsupported.
  3. [Section 4.5] The zero-shot experiment for TS12 is confounded by the high correlation between TS12 and TS11: training on TS1–TS11 gives the model direct evidence of TS12's seasonal and trend patterns, so this is not a clean test of zero-shot generalization. The more challenging setting (training on TS1–TS10) shows degraded performance, but is only evaluated qualitatively. To substantiate zero-shot generalization, the authors need quantitative comparisons across multiple held-out series and domains, with baselines and statistical significance.
  4. [Section 4.6] The KDD2021-12 experiment trains on the train partition and tests on the test partition of the same 12 series, which is within-series transfer, not zero-shot or foundation-model generalization. The paper's stated goal is to model 'previously unseen datasets,' yet the only partially held-out evaluation is the TS12 experiment in TELCO. A proper foundation-model evaluation should pretrain on a large, diverse corpus and test on completely unseen series from different domains without any retraining, which is not done here.
minor comments (5)
  1. [Section 3] The description of the receptive field expansion ('the dilation of each layer at the encoder increases exponentially as the network deepens') is imprecise; consider giving an explicit formula, e.g., receptive field size after N layers with kernel F and base dilation d.
  2. [Figure 4] The abbreviation 'ZPC' is used for the principal components of the latent vectors, but it is not defined in the caption or text; please write out 'principal component of Z' or explain the notation.
  3. [Abstract and Section 1] The abstract mentions 'preliminary results in different multi-dimensional time-series datasets,' but FAE is a univariate model; please clarify that the datasets are multivariate but the model analyzes each series separately, or rephrase to avoid confusion.
  4. [References] Reference [22] for the KDD2021 dataset points to a competition URL; a proper citation to the dataset itself (e.g., the UCR archive or the competition organizers' publication) would be more appropriate.
  5. [Figures 7 and 8] The captions mention 'flagged anomalies,' but the anomaly markers and the threshold used to flag them are not described in the captions or the main text; adding a legend and the α value would greatly improve interpretability.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: FAE's held-out and zero-shot evaluations do not reduce to fitted inputs or self-citation chains.

full rationale

The paper's central claims are empirical: FAE, a VAE with dilated convolutional encoder/decoder, is pretrained on TELCO series and evaluated on held-out TELCO testing samples, TELCO2, and KDD2021-12. The zero-shot experiment with TS12 is a genuine holdout: the model is trained on TS1-TS11 or TS1-TS10 and then applied to TS12 (Section 4.5). No parameter is fitted to the target series; the alpha threshold rule in Section 3 is a decision rule, not a fitted quantity, and although alpha is not reported, that is a completeness/correctness gap rather than circularity. The architecture is taken from the authors' prior DC-VAE work [12,13], but this self-citation is provenance, not a load-bearing uniqueness or optimality argument; the experiments here independently test the resulting model on unseen data. The KDD2021 evaluation trains on the provided train partition and reports on test partition (Section 4.6), which is standard. The discussion honestly states limitations (small number of datasets/domains, Section 5). The skeptic's concern that a decoder fed a time-repeated latent vector cannot produce time-varying output is a mechanistic correctness question about the described architecture; it does not exhibit a circular equivalence between reported predictions and inputs, so it is outside the circularity score.

Assumptions & free parameters 6 free parameters · 3 assumptions · 0 invented entities

No invented physical entities; the claims rely on standard VAE machinery, a reconstruction-deviation decision rule, and hyperparameters selected by search, with the anomaly threshold alpha unreported.

free parameters (6)
  • T (sequence length) = 256
    Selected by TPE grid search (range 128 to 512), five-minute window covering 21h20m of data.
  • J (latent dimension) = 48
    Selected by TPE grid search (range 16 to T/4), controls compression and expressiveness.
  • gamma (learning rate) = 6e-5
    Selected by TPE grid search (range 1e-5 to 5e-4).
  • m (mini-batch size) = 32
    Selected by TPE grid search (range 16 to 96).
  • U (number of filters) = 128
    Selected by TPE grid search (range 16 to 128).
  • alpha (anomaly threshold) = unreported
    Section 3 defines a normal region as alpha times the predicted sigma, but the paper never reports the alpha value used to flag anomalies in Figures 7 and 8.
assumptions (3)
  • domain assumption The VAE latent space is a centered isotropic Gaussian with approximately diagonal-covariance posteriors.
    Standard VAE assumption invoked in Section 2 and used by the architecture; the paper itself notes in Section 5 that this may limit expressiveness for highly heterogeneous data.
  • domain assumption A fixed integer alpha times the predicted standard deviation defines the normal region for every time-series.
    Stated in Section 3 and used for all anomaly flagging; no alpha value or quantitative validation is provided.
  • standard math Causal dilated convolutional layers with exponential dilation give each output time an effective receptive field covering the whole window.
    Section 3 claims that layer depth N satisfies T <= 2 F^(N-1), ensuring full temporal coverage; this is a standard property of causal dilated convolutions, not experimentally verified here.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Towards Foundation Auto-Encoders for Time-Series Anomaly Detection." pith.science (2026). https://pith.science/paper/VV3EAOVK

@misc{pith2026250701875,
  author       = {Pith},
  title        = {Pith review of: Towards Foundation Auto-Encoders for Time-Series Anomaly Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VV3EAOVK}},
  note         = {Machine review of arXiv:2507.01875}
}
read the original abstract

We investigate a novel approach to time-series modeling, inspired by the successes of large pretrained foundation models. We introduce FAE (Foundation Auto-Encoders), a foundation generative-AI model for anomaly detection in time-series data, based on Variational Auto-Encoders (VAEs). By foundation, we mean a model pretrained on massive amounts of time-series data which can learn complex temporal patterns useful for accurate modeling, forecasting, and detection of anomalies on previously unseen datasets. FAE leverages VAEs and Dilated Convolutional Neural Networks (DCNNs) to build a generic model for univariate time-series modeling, which could eventually perform properly in out-of-the-box, zero-shot anomaly detection applications. We introduce the main concepts of FAE, and present preliminary results in different multi-dimensional time-series datasets from various domains, including a real dataset from an operational mobile ISP, and the well known KDD 2021 Anomaly Detection dataset.

Figures

Figures reproduced from arXiv: 2507.01875 by the authors.

Figure 1
Figure 1. FAE’s encoder/decoder architecture using causal [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. TELCO time-series, for one month worth of data (March 2021), sampled at a five minutes rate. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Predictions with fully-trained FAE and multivariate [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (6 more)
Figure 5
Figure 5. Figure 5: Latent space representation, (a) per different time-series (TS [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: Latent space representation for TS12, in a daily basis – from day 1 in purple to day 31 in yellow, for a full month. the different times of the analysis. Recall that the latent space set for FAE in this analysis is 𝐽 = 48; to easily visualize 𝒁 as a two- or three￾dimen…
Figure 7
Figure 7. Figure 7: FAE vs DC-VAE in TELCO. DC-VAE FAE [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]
Figure 8
Figure 8. Figure 8: FAE vs DC-VAE in TELCO2. 4.4 FAE vs DC-VAE in TELCO2 What happens when we modify the application domain? In [PITH_FULL_IMAGE:figures/full_fig_p006_8.png]
Figure 9
Figure 9. Figure 9: Zero-shot modeling experimentation, predicting [PITH_FULL_IMAGE:figures/full_fig_p007_9.png]
Figure 11
Figure 11. Figure 11: FAE’s latent space for the 12 KDD2021 time-series. [PITH_FULL_IMAGE:figures/full_fig_p008_11.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

46 extracted references · 26 canonical work pages

  1. [1]

    Justin Bayer and Christian Osendorfer. 2014. Learning stochastic recurrent networks. arXiv preprint arXiv:1411.7610 (2014)

    arXiv 2014

  2. [2]

    James Bergstra, Rémi Bardenet, Yoshua Bengio, and Balázs Kégl. 2011. Algorithms for Hyper-parameter Optimization. Advances in neural information processing systems 24 (2011)

    work page 2011

  3. [3]

    Dalca, Luca Saglietti, Jennifer Listgarten, and Nicoló Fusi

    Francesco Paolo Casale, Adrian V. Dalca, Luca Saglietti, Jennifer Listgarten, and Nicoló Fusi. 2018. Gaussian Process Prior Variational Autoencoders. In Advances in Neural Information Processing Systems

    work page 2018

  4. [4]

    Chen, Lee Dicker, Carson Eisenach, and Dhruv Madeka

    Kevin C. Chen, Lee Dicker, Carson Eisenach, and Dhruv Madeka. 2022. MQTrans- former: Multi-horizon Forecasts with Context Dependent Attention and Optimal Bregman Volatility. InKDD 2022 Workshop on Mining and Learning from Time Series – Deep Forecasting: Models, Interpretability, and Applications

    work page 2022

  5. [5]

    Run-Qing Chen, Guang-Hui Shi, Wanlei Zhao, and Chang-Hui Liang. 2021. A joint model for IT operation series prediction and anomaly detection. Neurocomputing 448 (2021), 130–139

    work page 2021

  6. [6]

    Junyoung Chung, Kyle Kastner, Laurent Dinh, Kratarth Goel, Aaron C Courville, and Yoshua Bengio. 2015. A recurrent latent variable model for sequential data. Advances in Neural Information Processing Systems 28 (2015)

    work page 2015

  7. [7]

    Abhimanyu Das, Weihao Kong, Rajat Sen, and Yichen Zhou. 2024. A Decoder-only Foundation Model for Time-series Forecasting. arXiv:2310.10688 [cs.CL]

    arXiv 2024

  8. [8]

    Carl Doersch. 2016. Tutorial on variational autoencoders. arXiv preprint arXiv:1606.05908 (2016)

    arXiv 2016

Show all 46 references
  1. [9]

    Jeff Donahue, Philipp Krähenbühl, and Trevor Darrell. 2016. Adversarial feature learning. arXiv preprint arXiv:1605.09782 (2016)

    2016 arXiv

  2. [10]

    Vincent Fortuin, Dmitry Baranchuk, Gunnar Rätsch, and Stephan Mandt. 2020. GP-VAE: Deep Probabilistic Time Series Imputation. InInternational conference on artificial intelligence and statistics . PMLR, 1651–1661

    2020

  3. [11]

    Gastón García González, Pedro Casas, Alicia Fernández, and Gabriel Gómez

  4. [12]

    Gastón García González, Sergio Martinez Tagliafico, Alicia Fernández, Gabriel Gómez, José Acuña, and Pedro Casas. 2022. DC-VAE, Fine-grained Anomaly Detection in Multivariate Time-Series with Dilated Convolutions and Varia- tional Auto Encoders. In 2022 IEEE European Symposium...

    2022

  5. [13]

    Gastón García González, Sergio Martinez Tagliafico, Alicia Fernández, Gabriel Gómez, José Acuña, and Pedro Casas. 2023. One Model to Find Them All – Deep Learning for Multivariate Time-Series Anomaly Detection in Mobile Network Data. IEEE Transactions on Network and Service Ma...

    2023

  6. [14]

    Gastón García González, Pedro Casas, and Alicia Fernández. 2023. Fake it till you Detect it: Continual Anomaly Detection in Multivariate Time-Series using Gen- erative AI. In 2023 IEEE European Symposium on Security and Privacy Workshops (EuroS&PW). 558–566. https://doi.org/10...

    2023

  7. [15]

    Gastón García González, Sergio Martínez Tagliafico, Alicia Fernández, Gabriel Gómez, José Acuña, and Pedro Casas. 2023. TELCO – a new Multivariate Time- Series Dataset for Anomaly Detection in Mobile Networks. https://doi.org/10. 21227/skpg-0539

    2023

  8. [16]

    Azul Garza and Max Mergenthaler-Canseco. 2023. TimeGPT-1. arXiv:2310.03589 [cs.LG]

    2023 arXiv

  9. [17]

    Alexander Geiger, Dongyu Liu, Sarah Alnegheimish, Alfredo Cuesta-Infante, and Kalyan Veeramachaneni. 2020. TadGAN: Time series anomaly detection using generative adversarial networks. In 2020 IEEE International Conference on Big Data (Big Data). IEEE, 33–43

    2020

  10. [18]

    Laurent Girin, Fanny Roche, Thomas Hueber, and Simon Leglaive. 2019. Notes on the use of variational autoencoders for speech and audio spectrogram modeling. In DAFx 2019-22nd International Conference on Digital Audio Effects . 1–8

    2019

  11. [19]

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. 2014. Generative Adversarial Nets. In Advances in Neural Information Processing Systems , Vol. 27

    2014

  12. [20]

    Nate Gruver, Marc Finzi, Shikai Qiu, and Andrew Gordon Wilson. 2023. Large Lan- guage Models Are Zero-Shot Time Series Forecasters. arXiv:2310.07820 [cs.LG]

    2023 arXiv

  13. [21]

    Ming Jin, Shiyu Wang, Lintao Ma, Zhixuan Chu, James Y Zhang, Xiaoming Shi, Pin-Yu Chen, Yuxuan Liang, Yuan-Fang Li, Shirui Pan, and Qingsong Wen. 2024. Time-LLM: Time series forecasting by reprogramming large language models. In International Conference on Learning Representat...

    2024

  14. [22]

    Keogh, T

    E. Keogh, T. Dutta Roy, U. Naik, and A. Agrawal. 2021. Multi-dataset Time-Series Anomaly Detection Competition, SIGKDD 2021. https://compete.hexagon- ml.com/practice/competition/39/

    2021

  15. [23]

    Diederik P Kingma and Max Welling. 2013. Auto-encoding Variational Bayes. CoRR abs/1312.6114 (2013). arXiv:1312.6114 https://arxiv.org/abs/1312.6114

    2013 arXiv

  16. [24]

    Diederik P Kingma and Max Welling. 2019. An introduction to variational autoencoders. arXiv preprint arXiv:1906.02691 (2019)

    2019 arXiv

  17. [25]

    Guokun Lai, Bohan Li, Guoqing Zheng, and Yiming Yang. 2018. Stochastic WaveNet: A Generative Latent Variable Model for Sequential Data.arXiv preprint arXiv:1806.06116 (2018)

    2018 arXiv

  18. [26]

    Dan Li, Dacheng Chen, Baihong Jin, Lei Shi, Jonathan Goh, and See-Kiong Ng

  19. [27]

    Bryan Lim, Sercan Arik, Nicolas Loeff, and Tomas Pfister. 2021. Temporal Fusion Transformers for Interpretable Multi-horizon Time Series Forecasting. Interna- tional Journal of Forecasting 37, 4 (2021), 1748–1764. 8 Towards Foundation Auto-Encoders for Time-Series Anomaly Dete...

    2021

  20. [28]

    Guansong Pang, Chunhua Shen, Longbing Cao, and Anton Van Den Hengel

  21. [29]

    Siddharth Ramchandran, Gleb Tikhonov, Kalle Kujanpää, Miika Koskinen, and Harri Lähdesmäki. 2021. Longitudinal variational autoencoder. In International Conference on Artificial Intelligence and Statistics . PMLR, 3898–3906

    2021

  22. [30]

    Kashif Rasul, Arjun Ashok, Andrew Robert Williams, Hena Ghonia, Rishika Bhag- watkar, Arian Khorasani, Mohammad Javad Darvishi Bayazi, George Adamopou- los, Roland Riachi, Nadhir Hassen, Marin Biloš, Sahil Garg, Anderson Schneider, Nicolas Chapados, Alexandre Drouin, Valentina...

    2024 arXiv

  23. [31]

    ACM Comput

    Deep Learning for Anomaly Detection: A Review. ACM Comput. Surv. 54, 2, Article 38 (March 2021), 38 pages

    2021

  24. [32]

    Samira Shabanian, Devansh Arpit, Adam Trischler, and Yoshua Bengio. 2017. Variational bi-LSTMs. arXiv preprint arXiv:1711.05717 (2017)

    2017 arXiv

  25. [33]

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan Gomez, Lukasz Kaiser, and Illia Polosukhin. 2017. Attention is All you Need. In Advances in Neural Information Processing Systems , Vol. 30

    2017

  26. [34]

    Sebastian Schmidl, Phillip Wenig, and Thorsten Papenbrock. 2022. Anomaly detection in Time Series: a Comprehensive Evaluation. Proc. VLDB Endow. 15, 9 (may 2022), 1779–1797. https://doi.org/10.14778/3538598.3538602

    2022

  27. [35]

    Hao Xue and Flora D. Salim. 2023. PromptCast: A New Prompt-Based Learning Paradigm for Time Series Forecasting. IEEE Transactions on Knowledge and Data Engineering (2023), 1–14. https://doi.org/10.1109/TKDE.2023.3342137

    2023

  28. [36]

    Zichao Yang, Zhiting Hu, Ruslan Salakhutdinov, and Taylor Berg-Kirkpatrick

  29. [37]

    Haixu Wu, Jiehui Xu, Jianmin Wang, and Mingsheng Long. 2021. Autoformer: Decomposition Transformers with Auto-Correlation for Long-Term Series Fore- casting. CoRR abs/2106.13008 (2021). arXiv:2106.13008 https://arxiv.org/abs/ 2106.13008

    2021 arXiv

  30. [38]

    Sultan Zavrak and Murat Iskefiyeli. 2020. Anomaly-based intrusion detection from network flow features using variational autoencoder. IEEE Access 8 (2020), 108346–108358

    2020

  31. [39]

    Houssam Zenati, Chuan Sheng Foo, Bruno Lecouat, Gaurav Manek, and Vijay Ra- maseshan Chandrasekhar. 2018. Efficient GAN-based anomaly detection. arXiv preprint arXiv:1802.06222 (2018)

    2018 arXiv

  32. [40]

    Haoyi Zhou, Shanghang Zhang, Jieqi Peng, Shuai Zhang, Jianxin Li, Hui Xiong, and Wancai Zhang. 2021. Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting. In Thirty-Fifth AAAI Conference on Artificial Intelligence. AAAI Press, 11106–11115. https://d...

    2021 doi

  33. [41]

    Jinsung Yoon, Daniel Jarrett, and Mihaela van der Schaar. 2019. Time-series Generative Adversarial Networks. In Advances in Neural Information Processing Systems, Vol. 32

    2019

  34. [45]

    Tian Zhou, Ziqing Ma, Qingsong Wen, Xue Wang, Liang Sun, and Rong Jin

  35. [2017]

    In International conference on machine learning

    Improved Variational Autoencoders for Text Modeling using Dilated Con- volutions. In International conference on machine learning . PMLR, 3881–3890

  36. [2019]

    In International Conference on Artificial Neural Networks

    MAD-GAN: Multivariate anomaly detection for time series data with generative adversarial networks. In International Conference on Artificial Neural Networks. Springer, 703–716

  37. [2021]

    SIGMETRICS Perform

    On the Usage of Generative Models for Network Anomaly Detection in Multivariate Time-Series. SIGMETRICS Perform. Eval. Rev. 48, 4 (may 2021), 49–52. https://doi.org/10.1145/3466826.3466843

    2021

  38. [2022]

    CoRR abs/2201.12740 (2022)

    FEDformer: Frequency Enhanced Decomposed Transformer for Long- term Series Forecasting. CoRR abs/2201.12740 (2022). arXiv:2201.12740 https: //arxiv.org/abs/2201.12740 9

    2022 arXiv

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.