Pith. sign in

REVIEW 4 major objections 6 minor 50 references

Anomaly Detection for Non-stationary Time Series using Recurrent Wavelet Probabilistic Neural Network

T0 review · 4 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Modeling a recurrent net's latent space with wavelet densities beats six baselines on non-stationary time-series anomaly detection.

desk verdict A coherent latent-space wavelet density model with broad experiments, but the non-stationarity claim rests on frozen test-time coefficients and a static noise simulation, so the paper needs revision before it supports its headline. read the letter →

arxiv 2505.11321 v1 pith:ULYVG2AC submitted 2025-05-16 cs.LG eess.SP

Pu Yang , J. A. Barria This is my paper · ORCID
classification cs.LGeess.SP
keywords waveletprobabilisticnetworkrecurrentneuralnetworkstimeseriesanomalydetectionnon-stationaryenvironmentdensityestimatorunsupervisedlearningconceptdriftensembleestimation
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 tries to show that time-series anomaly detection in non-stationary settings is better served by modeling the probability density of a recurrent network's latent space than by reconstruction error or a parametric density model. To do this it builds RWPNN, an unsupervised framework that compresses windows with a stacked recurrent encoder-decoder and then estimates the density of the resulting latent vectors with an ensemble of wavelet density estimators, each tuned to a different forgetting factor. The ensemble is meant to track different rates of data variation without assuming a Gaussian or other parametric form, and the paper reports that on 45 real-world datasets RWPNN improves F1 by up to roughly 6 points over reconstruction- and latent-space baselines, with the largest gains under scarce training data and injected drift. If correct, this makes wavelet density estimation practical inside deep sequence models and offers a path to early warnings by watching the latent density shift before anomalies become visible in the signal.

What carries the argument

The load-bearing object is the Multi-Receptive-field Wavelet Probabilistic Network (MRWPN): an ensemble of wavelet density estimators built from radial B-spline scaling functions, one estimator per forgetting factor $\alpha$. Each estimator keeps a coefficient vector $\hat{w}$ updated online by $\hat{w} = (1-\Gamma)\hat{w} + \Gamma\,2^{nj_0/2}\phi(\|2^{j_0}h^E_L - k\| + m/2)$ (written elementwise for the whole set of forgetting factors), so the module sees the latent vector through many time scales at once. Because the density estimate has a closed form and only touches the low-dimensional latent code, it avoids the curse of dimensionality that made classical wavelet density estimators impractical in deep networks. The SREnc-Dec is the other half: stacked LSTM layers compress a sliding window into a latent vector and reconstruct it, giving MRWPN a compact representation to score. The decision rule is simply thresholding the selected view's density, so the whole pipeline is unsupervised apart from choosing which ensemble member and threshold to use.

What would settle it

A reader could rerun the 45-dataset suite with drift generated by gradually changing the data-generating process (for example, shifting mean and variance over time or swapping seasonal regimes) instead of adding Gaussian noise to 30% of the test set, and also restrict model selection to unlabeled validation data; if RWPNN's F1 margin over DIF disappears in that setting, the claimed non-stationary advantage is an artifact of the perturbation protocol.

Watch

Extended reading notes

Core claim

The central claim is that replacing reconstruction-based and parametric-latent anomaly scoring with nonparametric wavelet density estimation on the latent code of an encoder-decoder gives a more accurate and more stable detector when the data distribution shifts and training data are limited. Specifically, RWPNN learns latent features with a stacked LSTM encoder-decoder, then the MRWPN module maintains, for every forgetting factor $\alpha \in \Gamma$, a set of radial B-spline wavelet coefficients updated recursively from the latent vectors; at test time each view returns a density $\hat{p}_i(h^E_x)$, and the input is anomalous if that density falls below threshold $\beta$. The view and threshold are chosen by maximizing F1 on a validation set (Equation 10). Across 45 real-world datasets drawn from standard public time-series archives and the SMD server dataset, the paper reports average F1 of 0.7400 at $P=0.8$ (0.7167 with injected drift) and 0.9063 at $P=0.2$ (0.9008 with drift), beating LAD, LED, AML, GE, DIF, and TED by up to about 6 points, with the biggest margins when training data are scarce; an ablation replacing MRWPN with OCSVM, LOF, isolation forest, DIF, GOAD, or DSVDD shows lower F1, and the multi-receptive-field ensemble also outperforms a single-wavelet WPNN while running faster than training separate WPNNs. The paper frames this as the first wavelet density estimator embedded in an unsupervised deep network.

Load-bearing premise

The central empirical claim rests on treating randomly injected Gaussian noise as a proxy for concept drift and on using labeled anomalies in a validation set to select the best density view and threshold, so if either of those is not a fair test of non-stationary unsupervised detection, the reported advantage may not transfer.

Editorial extensions

If this is right

  • Non-stationary time series can be scored by latent density rather than reconstruction error, so a detector does not fail when reconstruction quality stays high under drift.
  • A single RWPNN model monitors several rates of variation at once because the ensemble of forgetting factors is evaluated in parallel, avoiding the cost of training separate wavelet networks.
  • The framework supports early warnings: shifts in the latent density from one timestamp to the next expose anomaly precursors before they are visible in the raw signal, as shown on ECG data.
  • Because the density module acts on the low-dimensional latent code, the approach can handle multivariate and high-dimensional series without the computational blow-up of classical wavelet density estimation.
  • The reported improvement concentrates in the scarce-data, drift-heavy setting: about 5.6 F1 points over the best baseline at $P=0.8$ with injected drift, versus about 1.7 points at $P=0.2$ without drift.

Reading between the lines

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

  • A plausible extension is to swap the RNN encoder for a Transformer or CNN encoder and repeat the benchmark, since the MRWPN module only consumes the latent vector and should be agnostic to how it was produced.
  • The paper's drift simulation injects static Gaussian noise into part of the test set; a fairer test of non-stationary robustness would use gradual or regime-switching distribution shifts and measure detection latency, which the paper leaves open.
  • Because the best ensemble view and threshold are chosen with labeled anomalies on a validation set, a fully label-free deployment would need a different calibration rule, such as extreme-value statistics on normal latent densities.
  • The ECG early-warning result invites a multi-patient test of whether latent-density precursors appear earlier and more consistently than reconstruction-error-based precursors, rather than on a single dataset.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes RWPNN, an unsupervised TSAD method for non-stationary environments. A stacked recurrent encoder-decoder (SREnc-Dec) compresses input windows into a latent representation, and a multi-receptive-field wavelet probabilistic network (MRWPN) estimates the density of that latent space with an ensemble of wavelet density estimators indexed by forgetting factors Γ. Anomalies are declared when the density under the selected ensemble view falls below a threshold. The method is evaluated on 45 univariate and multivariate datasets under two training-data regimes (P=0.8 and P=0.2), with concept drift simulated by adding Gaussian noise to 30% of the test set. The paper reports higher F1 scores than LAD, LED, AML, GE, DIF, and TED, and includes ablations replacing MRWPN with classical and deep AD methods, an early-warning ECG case study, and a runtime comparison against WPNN.

Significance. The construction is coherent and builds on a published wavelet density estimator [22,24], so the core density-estimation step is inherited from prior work rather than derived here; the contribution is the application of that estimator to an encoder-decoder latent space with an ensemble over forgetting factors. If the empirical results are reproducible, this is a useful nonparametric alternative to GMM/VAE latent-space models for TSAD, particularly in data-scarce regimes. Strengths include the breadth of the evaluation (45 datasets, two data-availability regimes, ablations against several detectors) and the explicit runtime comparison showing a computational advantage of the ensemble MRWPN over training separate WPNNs. The main weaknesses are that the concept-drift setup is a static noise perturbation applied to a test set whose MRWPN coefficients are frozen, that the 'unsupervised' claim is weakened by selecting the threshold and ensemble view on labeled validation data, and that several reported ablation results are internally inconsistent. These issues must be addressed before the central claims about non-stationarity can be accepted.

major comments (4)
  1. [§5.1, §4.4, Algorithm 3] The concept-drift condition does not exercise the adaptation mechanism that motivates the method. In §5.1, drift is simulated by randomly adding Gaussian noise with μ=0.3 and Σ=0.2 to 30% of Dtest, which is a static, label-preserving perturbation rather than a temporal or gradual distribution shift. More importantly, §4.4 states that the MRWPN coefficients are updated only on hE_train, and Algorithm 3 evaluates the test set with fixed coefficients. With frozen coefficients, the ensemble of α values is not adapting to a changing test distribution; it is effectively a multi-resolution density estimate of the training latent distribution. The reported F1 gains may therefore reflect a flexible density estimator under a noise-shifted test set, not robustness to non-stationarity or concept drift. I request either an online-update evaluation on test data or temporal-drift benchmarks (e.g., gradually shifted windows, rotating concepts, abrupt regime changes) that actually change the data-generating process over time.
  2. [§4.4, Eq. (10)] Using labeled anomalies for model selection conflicts with the 'unsupervised' framing. Equation (10) selects the ensemble view i and threshold β by maximizing F1 on a validation set, and Section 5 confirms that Dv1+v2 contains both normal and anomaly subsequences. Thus anomaly labels are used for threshold calibration and ensemble-view selection. This is standard in TSAD evaluation, but it should be reported as supervision: without this step the reported F1 values are not attainable in a label-free deployment. I recommend either adopting a label-free threshold rule (e.g., a quantile of the training-set density) and reporting its performance, or explicitly reclassifying the method as 'unsupervised training with supervised threshold selection' and discussing the practical availability of such labels.
  3. [Table 6] The ablation table contains inconsistencies that need correction. DIF is reported with F1=0.7364 at P=0.8 without CD in Table 4 but with F1=0.6470 in Table 6 under the same named setting, and similarly at P=0.2 without CD (0.8892 in Table 5 versus 0.8410 in Table 6). The text does not state that Table 6 uses a different dataset subset, so the reader cannot tell whether the ablation is comparable to the main benchmark. Additionally, several Precision/Recall/F1 triples in Table 6 (e.g., OCSVM at P=0.2 without CD: Precision 0.5883, Recall 0.6244, F1 0.7695) are not consistent with the standard F1 formula unless F1 is macro-averaged over datasets before aggregation, which should be stated explicitly. Please recompute the table and clarify the aggregation procedure.
  4. [§5.1, Tables 4-5] The superiority claims lack statistical evidence. The tables report means and standard deviations across 45 datasets, but no paired significance tests or per-dataset win rates are provided. Given that the standard deviations overlap substantially (e.g., P=0.8 without CD: RWPNN F1=0.7400±0.11 versus DIF 0.7364±0.14; P=0.8 with CD: RWPNN 0.7167±0.11 versus DIF 0.6604±0.16), the claim that RWPNN outperforms all benchmarks is not supported beyond point estimates. I recommend adding Wilcoxon signed-rank tests or equivalent paired tests and reporting the number of datasets on which RWPNN wins, loses, or ties relative to each baseline.
minor comments (6)
  1. [§4.3.2, Eq. (9)] The notation in the coefficient update mixes scalar and vector quantities: Γ is introduced as a vector of forgetting factors, but the update is written as if each coefficient is updated by all Γ simultaneously; please clarify that the update is applied separately for each α-view.
  2. [Algorithm 3] The final line of the density computation contains garbled characters in the manuscript, making that step unreadable; please provide clean pseudocode.
  3. [§5.4] The phrase 'the seven UEA datasets used in Fig. 4' is ambiguous because Fig. 4 lists many datasets and does not mark which are UEA; please specify the seven dataset names.
  4. [§5.3] The early-warning trigger is described as 'a predefined threshold' on the deviation of p̂(yE_x), but the threshold value is never given; this makes the early-warning case study non-reproducible.
  5. [§3.1, Eq. (2)] The use of n for the data dimension in Eq. (2) conflicts with later uses of n as the number of features; consistent notation would improve readability.
  6. [§3.4, Eq. (8)] Equation (8) uses a class index c in a setting where only the normal class is modeled; this is confusing and should be simplified to the unsupervised single-class case.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: RWPNN's derivation is self-contained; the WDE estimator is imported from prior published work and test F1 is externally benchmarked.

full rationale

Walking the derivation chain, the anomaly-density estimator is not defined in terms of the target detection result. Equations (2)-(9) implement a radial B-spline wavelet density estimator taken from the authors' earlier peer-reviewed work [22,24] and a recursive coefficient update with a forgetting factor; these are fixed mathematical constructions whose assumptions (B-spline order, dilation j0, translation grid k, forgetting factors Gamma) do not include the anomaly labels or the reported F1. The SREnc-Dec latent space is trained by reconstruction (MAE) and the MRWPN then estimates the density of the normal-class latent features. At test time, Algorithm 3 only evaluates the trained density; it does not refit coefficients to test labels. The only label-dependent element is Equation (10), which selects the ensemble view i and threshold beta by maximizing F1 on a validation set built from both classes. This is disclosed and is standard calibration/model selection; because the reported metrics are computed on a separate test set, the test F1 is not equal to the selection objective by construction. The mismatch between the claimed online adaptation and the frozen-coefficient evaluation (Algorithm 3) is a validity concern about whether the experiments exercise non-stationarity, not a circularity. The paper is evaluated against external benchmarks on 45 real-world datasets, so the central empirical claim has independent content.

Assumptions & free parameters 8 free parameters · 5 assumptions · 0 invented entities

The load-bearing inputs are the frame construction from the authors' prior papers [22, 24], the empirical assumption that anomalies live in low-density latent regions, and a validation protocol that uses labeled anomalies to select the ensemble view and threshold. The concept drift experiment is an ad hoc noise injection rather than a temporal shift. No new physical or mathematical entities are proposed.

free parameters (8)
  • B-spline order m = linear, quadratic, or cubic (searched per dataset)
    Controls smoothness of the density estimate; the paper searches three orders.
  • Resolution parameter j0 = 1 to 5 (searched per dataset)
    Sets the scale of the wavelet scaling functions; larger j0 adds detail and overfitting risk.
  • Forgetting factor set Gamma = {1, 1/10, 1/100, 1/500, 1/1000}
    Defines the multiple receptive fields; the winning view i is chosen by Equation (10) on labeled validation data.
  • Decision threshold beta = Maximizer of F1 on Dv1+v2
    Anomaly alarm threshold is fit to validation labels rather than derived from the density model alone.
  • Encoder and decoder hidden sizes = Searched over [128, 64, 32, 16, 8, 4, 2]
    Architecture capacity is tuned per dataset.
  • Learning rate = Cyclical search in [1e-5, 1e-1]
    Optimizer hyperparameter chosen by the cyclical learning rate method.
  • Batch size = Largest of {1, 2, 4, 8, 16, 32, 64, 128, 256, 512} bounded by split sizes
    Training hyperparameter selected from the given range.
  • Concept drift noise parameters = mu=0.3, Sigma=0.2, on 30% of test set
    Hand-chosen perturbation used as the CD condition; not fitted to data but not a realistic drift model.
assumptions (5)
  • standard math The radial B-spline scaling functions form a frame, so the density expansion in Equation (6) is valid and the coefficient update in Equation (7) converges.
    Inherited from [22] and [24] without re-derivation; it is a published analytical result, not established in this paper.
  • domain assumption Anomalous latent vectors from a test window are located in low-density regions of the normal latent-space density.
    Section 4.4 and Figure 3 observe this empirically for one dataset; no theory links reconstruction error or latent geometry to density separability.
  • ad hoc to paper Adding Gaussian noise to 30% of the test set is a valid proxy for concept drift in non-stationary environments.
    Section 5.1 introduces this condition as 'with CD'; random label-independent noise is not a temporal change in the data distribution.
  • ad hoc to paper Using labeled anomaly samples in the validation set to select beta and i is compatible with claiming an unsupervised detector.
    Section 4.4 Equation (10) maximizes F1 on Dv1+v2, where Dv2 contains anomaly-class labels; this is supervised calibration and model selection.
  • domain assumption The SREnc-Dec reconstruction objective yields a latent space whose normal density can be represented by the fixed B-spline frame at reasonable resolution.
    No analysis proves the latent distribution is compatible with the frame; it is assumed by construction.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Anomaly Detection for Non-stationary Time Series using Recurrent Wavelet Probabilistic Neural Network." pith.science (2026). https://pith.science/paper/ULYVG2AC

@misc{pith2026250511321,
  author       = {Pith},
  title        = {Pith review of: Anomaly Detection for Non-stationary Time Series using Recurrent Wavelet Probabilistic Neural Network},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ULYVG2AC}},
  note         = {Machine review of arXiv:2505.11321}
}
read the original abstract

In this paper, an unsupervised Recurrent Wavelet Probabilistic Neural Network (RWPNN) is proposed, which aims at detecting anomalies in non-stationary environments by modelling the temporal features using a nonparametric density estimation network. The novel framework consists of two components, a Stacked Recurrent Encoder-Decoder (SREnc-Dec) module that captures temporal features in a latent space, and a Multi-Receptive-field Wavelet Probabilistic Network (MRWPN) that creates an ensemble probabilistic model to characterise the latent space. This formulation extends the standard wavelet probabilistic networks to wavelet deep probabilistic networks, which can handle higher data dimensionality. The MRWPN module can adapt to different rates of data variation in different datasets without imposing strong distribution assumptions, resulting in a more robust and accurate detection for Time Series Anomaly Detection (TSAD) tasks in the non-stationary environment. We carry out the assessment on 45 real-world time series datasets from various domains, verify the performance of RWPNN in TSAD tasks with several constraints, and show its ability to provide early warnings for anomalous events.

Figures

Figures reproduced from arXiv: 2505.11321 by the authors.

Figure 1
Figure 1. The flowchart of the proposed RWPNN. The temporal feature extractor block [PITH_FULL_IMAGE:figures/full_fig_p014_1.png] view at source ↗
Figure 2
Figure 2. Visualisation of the MRWPN module, which provides |Γ| views at the same time. Each dashed module refers to an MRWPN with a specific value of α, and different values of α are used to construct the ensemble view. 4.3. Wavelet Optimisation 4.3.1. Initialisation As discussed in Section 3.3, four parameters are needed for MRWPN : (i) m, the order of Φ(x) from Equation 2. (ii-iii) k and k, the translation parameter and th… view at source ↗
Figure 3
Figure 3. Distribution of the estimated ˆp [PITH_FULL_IMAGE:figures/full_fig_p018_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: F1-Score performance heatmap for the models and datasets considering [PITH_FULL_IMAGE:figures/full_fig_p027_4.png]
Figure 5
Figure 5. Figure 5: The estimated PDFs for the ECG5000 dataset are depicted. (a) shows the mean and highlights the standard deviation of the normal and anomaly classes. (b) is the anomaly scores a(x) for LED. (c) shows the scores generated ˆp(y E x ) by RWPNN. (d-e) are the pattern variat…
Figure 6
Figure 6. Figure 6: A sliding window-based Early Warning System monitoring the change in ˆp [PITH_FULL_IMAGE:figures/full_fig_p029_6.png]
Figure 7
Figure 7. Figure 7: Summary of the performance difference between RWPNN and the benchmark [PITH_FULL_IMAGE:figures/full_fig_p030_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

50 extracted references · 24 canonical work pages

  1. [1]

    A. C. Atkinson, D. M. Hawkins, Identification of Outliers., Vol. 37, Chapman and Hall, London, 1981. doi:10.2307/2530182

    work page doi:10.2307/2530182 1981

  2. [2]

    C. C. Aggarwal, Outlier analysis, Vol. 9781461463, New York, 2013. doi: 10.1007/978-1-4614-6396-2

    doi:10.1007/978-1-4614-6396-2 2013

  3. [3]

    Alarcon-Aquino, J

    V. Alarcon-Aquino, J. A. Barria, Anomaly detection in communication networks using wavelets, IEE Proceedings: Communications 148 (6) (2001) 355–362. doi: 10.1049/ip-com:20010659

    work page doi:10.1049/ip-com:20010659 2001

  4. [4]

    Alarcon-Aquino, J

    V. Alarcon-Aquino, J. A. Barria, Multiresolution FIR neural-network-based learning algorithm applied to network traffic prediction, IEEE Transactions on Systems, Man and Cybernetics Part C: Applications and Reviews 36 (2) (2006) 208–220. doi:10.1109/TSMCC.2004.843217

    work page arXiv 2006

  5. [5]

    H. Wang, M. J. Bah, M. Hammad, Progress in Outlier Detection Techniques: A Survey, IEEE Access 7 (2019) 107964–108000. doi:10.1109/ACCESS.2019. 2932769

    doi:10.1109/access.2019 2019

  6. [6]

    Chalapathy, S

    R. Chalapathy, S. Chawla, Deep Learning for Anomaly Detection: A Survey, arXiv (2019). URL http://arxiv.org/abs/1901.03407

    arXiv 2019

  7. [7]

    H. Ren, B. Xu, Y. Wang, C. Yi, C. Huang, X. Kou, T. Xing, M. Yang, J. Tong, Q. Zhang, Time-series anomaly detection service at Microsoft, Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining 3330680 (c) (2019) 3009–3017

    work page 2019

  8. [8]

    Y. Liu, S. Garg, J. Nie, Y. Zhang, Z. Xiong, J. Kang, M. S. Hossain, Deep Anomaly Detection for Time-Series Data in Industrial IoT: A Communication- Efficient On-Device Federated Learning Approach, IEEE Internet of Things Journal 8 (8) (2021) 6348–6358. doi:10.1109/JIOT.2020.3011726

    arXiv 2021

Show all 50 references
  1. [9]

    L. Vu, V. L. Cao, Q. U. Nguyen, D. N. Nguyen, D. T. Hoang, E. Dutkiewicz, Learning Latent Representation for IoT Anomaly Detection, IEEE Transactions on Cybernetics 52 (5) (2022) 3769–3782. doi:10.1109/TCYB.2020.3013416

    2022

  2. [10]

    Braei, S

    M. Braei, S. Wagner, Anomaly Detection in Univariate Time-series: A Survey on the State-of-the-Art, arXiv (2020). URL http://arxiv.org/abs/2004.00433 32

    2020 arXiv

  3. [11]

    Q. Wen, T. Zhou, C. Zhang, W. Chen, Z. Ma, J. Yan, L. Sun, Transformers in Time Series: A Survey, IJCAI International Joint Conference on Artificial Intelligence 2023-Augus (2023) 6778–6786. doi:10.24963/ijcai.2023/759

    2023 doi

  4. [12]

    Sutskever, O

    I. Sutskever, O. Vinyals, Q. V. Le, Sequence to sequence learning with neu- ral networks, Advances in Neural Information Processing Systems 4 (January) (2014) 3104–3112

    2014

  5. [13]

    Malhotra, L

    P. Malhotra, L. Vig, G. Shroff, P. Agarwal, Long Short Term Memory networks for anomaly detection in time series, 23rd European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, ESANN 2015 - Proceedings (2015) 89–94

    2015

  6. [14]

    Malhotra, A

    P. Malhotra, A. Ramakrishnan, G. Anand, L. Vig, P. Agarwal, G. Shroff, LSTM- based Encoder-Decoder for Multi-sensor Anomaly Detection, arXiv (2016). URL http://arxiv.org/abs/1607.00148

    2016 arXiv

  7. [15]

    Y. Guo, W. Liao, Q. Wang, L. Yu, T. Ji, P. Li, J. Zhu, I. Takeuchi, Multi- dimensional Time Series Anomaly Detection: A GRU-based Gaussian Mixture Variational Autoencoder Approach, Proceedings of Machine Learning Research 95 (2018) 97–112

    2018

  8. [16]

    B. Xue, X. Gao, B. Li, F. Zhai, J. Lu, J. Yu, S. Fu, C. Xiao, A robust multi-scale feature extraction framework with dual memory module for multivariate time series anomaly detection, Neural Networks 177 (2024) 106395. doi:10.1016/ j.neunet.2024.106395

    2024

  9. [17]

    Y. Yao, J. Ma, S. Feng, Y. Ye, Svd-ae: An asymmetric autoencoder with svd regularization for multivariate time series anomaly detection, Neural Networks 170 (2024) 535–547. doi:10.1016/j.neunet.2023.11.023

    2024 doi

  10. [18]

    B. Zong, Q. Song, M. R. Min, W. Cheng, C. Lumezanu, D. Cho, H. Chen, Deep autoencoding Gaussian mixture model for unsupervised anomaly detection, 6th International Conference on Learning Representations, ICLR 2018 - Conference Track Proceedings (2018) 1–19

    2018

  11. [19]

    X. Wang, D. Pi, X. Zhang, H. Liu, C. Guo, Variational transformer-based anomaly detection approach for multivariate time series, Measurement: Journal of the International Measurement Confederation 191 (2022). doi:10.1016/j. measurement.2022.110791. 33

    2022

  12. [20]

    A. R. Webb, Statistical pattern recognition, 2nd Edition, Wiley, Chichester, 2002

    2002

  13. [21]

    Vidakovic, Statistical Modeling by Wavelets (Wiley Series in Probability and Statistics), Wiley-Interscience, 1999

    B. Vidakovic, Statistical Modeling by Wavelets (Wiley Series in Probability and Statistics), Wiley-Interscience, 1999. URL http://www.amazon.com/Statistical-Modeling-Wavelets-Probabi lity-Statistics/dp/0471293652

    1999

  14. [22]

    E. S. Garc´ ıa Trevi˜ no, V. Alarc´ on Aquino, J. A. Barria, The radial wavelet frame density estimator, Computational Statistics and Data Analysis 130 (2019) 111–

    2019

  15. [23]

    Mallat, A Wavelet Tour of Signal Processing, A Wavelet Tour of Signal Pro- cessing (2009)

    S. Mallat, A Wavelet Tour of Signal Processing, A Wavelet Tour of Signal Pro- cessing (2009). doi:10.1016/B978-0-12-374370-1.X0001-8

    2009 doi

  16. [24]

    E. S. Garcia-Trevino, P. Yang, J. A. Barria, Wavelet Probabilistic Neural Net- works, IEEE Transactions on Neural Networks and Learning Systems 35 (1) (2024) 376–389. doi:10.1109/TNNLS.2022.3174705

    2024

  17. [25]

    K. Choi, J. Yi, C. Park, S. Yoon, Deep Learning for Anomaly Detection in Time-Series Data: Review, Analysis, and Guidelines, IEEE Access 9 (2021) 120043–120065. doi:10.1109/ACCESS.2021.3107975

    2021

  18. [26]

    Y. H. Yoo, U. H. Kim, J. H. Kim, Recurrent Reconstructive Network for Se- quential Anomaly Detection, IEEE Transactions on Cybernetics 51 (3) (2021) 1704–1715. doi:10.1109/TCYB.2019.2933548

    2021

  19. [27]

    S. Tuli, G. Casale, N. R. Jennings, TranAD: Deep Transformer Networks for Anomaly Detection in Multivariate Time Series Data, Proceedings of the VLDB Endowment 15 (6) (2022) 1201–1214. doi:10.14778/3514061.3514067. URL http://arxiv.org/abs/2201.07284

    2022

  20. [28]

    J. Xu, H. Wu, J. Wang, M. Long, Anomaly Transformer: Time Series Anomaly Detection With Association Discrepancy, ICLR 2022 - 10th International Con- ference on Learning Representations (2022)

    2022

  21. [29]

    H. Zhao, S. Qiu, J. Yang, J. Guo, M. Liu, X. Cao, Satellite Early Anomaly Detec- tion Using an Advanced Transformer Architecture for Non-Stationary Teleme- try Data, IEEE Transactions on Consumer Electronics 70 (1) (2024) 4213–4225. doi:10.1109/TCE.2024.3357881. 34

    2024

  22. [30]

    Zhang, A

    Q. Zhang, A. Benveniste, Wavelet Networks, IEEE Transactions on Neural Net- works 3 (6) (1992) 889–898. doi:10.1109/72.165591. URL https://www.ncbi.nlm.nih.gov/pubmed/18276486

    1992

  23. [31]

    G. G. Walter, Y. Miao, J. Zhang, W. N. Wayne Lee, Wavelet Neural Networks for Function Learning, IEEE Transactions on Signal Processing 43 (6) (1995) 1485–1497. doi:10.1109/78.388860

    1995 doi

  24. [32]

    Y. Chen, B. Yang, J. Dong, Time-series prediction using a local linear wavelet neural network, Neurocomputing 69 (4-6) (2006) 449–465. doi:10.1016/j.ne ucom.2005.02.006

    2006 doi

  25. [33]

    M. Liu, A. Zeng, Q. Lai, R. Gao, M. Li, J. Qin, Q. Xu, T-Wavenet: a Tree- Structured Wavelet Neu- Ral Network for Time Series Signal Analysis, Iclr (2022) 1–12

    2022

  26. [34]

    D. Chen, L. Chen, Y. Zhang, B. Wen, C. Yang, A Multiscale Interactive Recur- rent Network for Time-Series Forecasting, IEEE Transactions on Cybernetics 52 (9) (2022) 8793–8803. doi:10.1109/TCYB.2021.3055951

    2022

  27. [35]

    D. F. Specht, Probabilistic neural networks for classification, mapping, or asso- ciative memory (1988) 525–532 doi:10.1109/icnn.1988.23887

    1988

  28. [36]

    H. Anh, E. Keogh, K. Kamgar, C.-C. Michael Yeh, Y. Zhu, S. Gharghabi, C. Ann Ratanamahatana, Y. Chen, B. Hu, N. Begum, A. Bagnall, A. Mueen, G. Batista, The UCR Time Series Classification Archive (2018) 1–148

    2018

  29. [37]

    J. An, S. Cho, Variational autoencoder based anomaly detection using recon- struction probability, Special Lecture on IE 2 (1) (2015) 1–18

    2015

  30. [38]

    Bagnall, H

    A. Bagnall, H. A. Dau, J. Lines, M. Flynn, J. Large, A. Bostrom, P. Southam, E. Keogh, The UEA multivariate time series classification archive, 2018 (2018). URL http://arxiv.org/abs/1811.00075

    2018 arXiv

  31. [39]

    Y. Su, R. Liu, Y. Zhao, W. Sun, C. Niu, D. Pei, Robust anomaly detection for multivariate time series through stochastic recurrent neural network, Proceed- ings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (2019) 2828–2837 doi:10.1145/32925...

    2019

  32. [40]

    H. Xu, G. Pang, Y. Wang, Y. Wang, Deep Isolation Forest for Anomaly De- tection, IEEE Transactions on Knowledge and Data Engineering 35 (12) (2023) 12591–12604. doi:10.1109/TKDE.2023.3270293. 35

    2023

  33. [41]

    A. Garg, W. Zhang, J. Samaran, R. Savitha, C. S. Foo, An Evaluation of Anomaly Detection and Diagnosis in Multivariate Time Series, IEEE Trans- actions on Neural Networks and Learning Systems 33 (6) (2022) 2508–2517. doi:10.1109/TNNLS.2021.3105827

    2022

  34. [42]

    L. N. Smith, Cyclical learning rates for training neural networks, Proceedings - 2017 IEEE Winter Conference on Applications of Computer Vision, WACV 2017 (2017) 464–472 doi:10.1109/WACV.2017.58

    2017 doi

  35. [43]

    J. L. Ba, D. P. Kingma, Adam: A method for stochastic optimization, 3rd International Conference on Learning Representations, ICLR 2015 - Conference Track Proceedings (2015) 1–15

    2015

  36. [44]

    Sch¨ olkopf, J

    B. Sch¨ olkopf, J. C. Platt, J. Shawe-Taylor, A. J. Smola, R. C. Williamson, Estimating the support of a high-dimensional distribution, Neural Computation 13 (7) (2001) 1443–1471. doi:10.1162/089976601750264965

    2001 doi

  37. [45]

    M. M. Breunig, H. P. Kriegel, R. T. Ng, J. Sander, LOF: Identifying Density- Based Local Outliers, SIGMOD 2000 - Proceedings of the 2000 ACM SIGMOD International Conference on Management of Data (2000) 93–104 doi:10.1145/ 342009.335388

    2000

  38. [46]

    F. T. Liu, K. M. Ting, Z. H. Zhou, Isolation forest, Proceedings - IEEE Inter- national Conference on Data Mining, ICDM (2008) 413–422 doi:10.1109/ICDM .2008.17

    2008 doi

  39. [47]

    Pedregosa, G

    F. Pedregosa, G. Varoquaux, A. Gramfort, V. Michel, B. Thirion, O. Grisel, M. Blondel, P. Prettenhofer, R. Weiss, V. Dubourg, J. Vanderplas, A. Passos, D. Cournapeau, M. Brucher, M. Perrot, E. Duchesnay, Scikit-learn: Machine learning in Python, Journal of Machine Learning Res...

    2011

  40. [48]

    Bergman, Y

    L. Bergman, Y. Hoshen, Classification-Based Anomaly Detection for General Data, 8th International Conference on Learning Representations, ICLR 2020 (2020)

    2020

  41. [49]

    L. Ruff, R. A. Vandermeulen, N. G¨ ornitz, L. Deecke, S. A. Siddiqui, A. Binder, E. M¨ uller, M. Kloft, Deep one-class classification, 35th International Conference on Machine Learning, ICML 2018 10 (2018) 6981–6996. 36

    2018

  42. [139]

    URL http://www.sciencedirect.com/science/article/pii/S01679473183 02044

    doi:10.1016/j.csda.2018.08.021. URL http://www.sciencedirect.com/science/article/pii/S01679473183 02044

    2018 doi

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.