Pith. sign in

REVIEW 4 major objections 5 minor 50 references

CAPMix: Robust KPI Anomaly Detection for AIOps in Noisy and Dynamic Environments

T0 review · 4 major / 5 minor · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read CAPMix claims time-series anomaly detection fails when injected anomalies are too normal or too weird, and that a pipeline of prior-guided injection, DTW-based label revision, and dual-space mixup closes that gap, reporting RPA-F1 scores of

desk verdict CAPMix is a plausible incremental extension of CutAddPaste, but the oracle-threshold evaluation and an internally contradictory label-revision definition make the headline numbers unreliable as reported. read the letter →

arxiv 2509.06419 v2 pith:T6ICNIAG submitted 2025-09-08 cs.LG cs.AI

Xudong Mou , Rui Wang , Tiejun Wang , Zexin Wu , Fangda Guo , Jie Sun , Shiru Chen , Penghao Zhang
show 6 more authors
Tiezi Zhang Tianyu Wo Hao Peng Chunming Hu Xudong Liu Renyu Yang
This is my paper
classification cs.LGcs.AI
keywords anomalyshifttimeseriesdetectionCutAddPastelabelrevisiondual-spacemixupAIOpscontaminatedtrainingdataKPImonitoring
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

CAPMix claims that time-series anomaly detection can be made robust in noisy, contaminated KPI streams by generating controlled pseudo-anomalies and then correcting for their imperfections. It names the core failure 'Anomaly Shift': injected anomalies that look too much like normal data or too unlike real failures bend the learned decision boundary and cause false alarms or misses. The framework's three parts—CutAddPaste injection, DTW-distance label revision, and dual-space mixup inside a temporal convolutional network—are designed to keep synthetic anomalies close to the real anomaly distribution. Across AIOps, UCR, SWaT, WADI, and ESA, CAPMix reports RPA-F1 scores of 80.46, 71.89, 47.04, 34.08, and 84.46, each above the strongest baselines it compares against.

What carries the argument

The load-bearing mechanism is a three-stage pipeline: (1) CutAddPaste cuts a patch from a random source sample, adds a random linear trend to selected dimensions, and pastes it into a target sample, generating structurally diverse anomalies; (2) DTW label revision measures each generated sample's distance to the mean normal center and replaces the hard label 1 with soft label 1/γ for samples inside a γ-sigma boundary, preventing near-normal injections from teaching false alarms; (3) dual-space mixup applies convex interpolation at the input and after each of three TCN blocks, with labels mixed proportionally, so the classifier sees intermediate, coherent examples instead of extreme outliers.

What would settle it

A reader could run CAPMix and CutAddPaste on AIOps and ESA using one fixed threshold chosen from validation only, then compare RPA-F1. If the CAPMix margin vanishes or reverses, the paper's robustness claim is an artifact of test-set threshold search. Separately, an explicit distributional distance (MMD or Wasserstein) between synthetic and real anomaly embeddings before and after CAPMix would confirm or refute the claimed reduction in anomaly shift.

Watch

Extended reading notes

Core claim

The paper's central claim is that the distribution mismatch between synthetic and real anomalies—formalized as D_shift = Dist(D_real, D_syn)—is the main reason anomaly-assumption detectors underperform, and that it can be reduced without knowing real anomalies in advance. CutAddPaste injects five anomaly types (shape, correlation, seasonality, trend, point-wise global and contextual) in a targeted way; a DTW-based label revision assigns a soft label of 1/γ to synthetic samples that fall near the normality center; dual-space mixup interpolates samples and labels in input and latent spaces to prevent overly distant synthetic outliers. The paper argues this yields Dist(D_real, D_hat_syn) < Dist

Load-bearing premise

The reported margins assume that searching the test set for the best anomaly threshold (and, on UCR, taking the largest score as the anomaly) is a fair comparison; if those oracle choices are not available when the model is deployed, the advantage may shrink.

Editorial extensions

If this is right

  • Anomaly-assumption methods can be made safer for production: training sets with up to at least 8× injected real anomalies still yield stable F1 and precision on AIOps, and the full model outperforms its CutAddPaste base once contamination grows.
  • The framework does not require anomalies in training data: on UCR, SWaT, and WADI, where training anomalies are 0%, CAPMix still reaches the reported scores.
  • Because the major performance gain appears on ESA (84.46 vs 18.56 for the base CutAddPaste), the label-revision and mixup mechanism transfers to very long multivariate telemetry with sparse anomalies, not just short KPI windows.
  • The component ablations indicate the method is adaptive: label revision carries the univariate gains, dual-space mixup carries the multivariate gains, and the full model is best or comparable everywhere.

Reading between the lines

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

  • A deployment version needs a threshold rule that does not peek at test data; the reported numbers use a test-set search for τ, so a fixed validation-based threshold may close part of the gap.
  • The mechanism claim that anomaly shift is reduced is visual (UMAP), not quantitative; computing a divergence between synthetic and real anomaly embeddings before and after CAPMix would make the claim directly testable.
  • Since DTW label revision helps univariate but not multivariate data, a per-dimension or learned revision boundary might extend the gains to multivariate sets; the paper leaves adaptive thresholds as future work.
  • The label-revision and mixup stages act on labels and representations rather than on the injection operator, so they could likely harden other anomaly-assumption generators, not only CutAddPaste; this is a natural transfer test.
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 CAPMix, an anomaly-assumption time-series anomaly detection framework that extends the authors' earlier CutAddPaste method with two components: a DTW-distance-based label revision strategy and a dual-space (input and latent) Mixup integrated into a TCN. The goal is to reduce 'Anomaly Shift,' i.e., the mismatch between injected synthetic anomalies and real anomalies. CAPMix is evaluated on five datasets (AIOps, UCR, SWaT, WADI, ESA) using the RPA F1 metric and is reported to outperform existing baselines consistently. The authors also include ablation studies, contamination robustness experiments, UMAP visualizations, hyperparameter sensitivity analysis, and mention deployment at Kuaishou.

Significance. If the empirical claims are established, CAPMix would be a useful contribution to anomaly-assumption time-series anomaly detection, particularly in settings with contaminated training data. The paper's strengths include releasing code, evaluating on a wide range of public benchmarks, reporting a real deployment, and providing a clear conceptual discussion of why synthetic anomalies can distort decision boundaries. However, the central claim of consistent state-of-the-art performance is not yet supported by the evaluation as written: the threshold-selection protocol is a test-set oracle procedure whose uniform application to baselines is not documented, the theoretical claim in Eq. (10) is not actually proved, and several internal inconsistencies in the label-revision definition and ablation results need resolution.

major comments (4)
  1. [§IV-B and Table III] The main results are obtained under a test-set oracle threshold protocol: 'we search for the optimal anomaly threshold τ∈[-3,3]' and, for UCR, 'we choose the sample with the largest anomaly score as the location of the anomaly.' The paper does not state that every baseline received exactly the same oracle threshold treatment. RPA F1 is highly sensitive to the threshold, and CAPMix outputs probabilities in [0,1] while many baselines produce unbounded scores, so an oracle threshold choice can disproportionately help CAPMix. Please specify the threshold protocol used for each baseline, and ideally re-run all methods under a common protocol (e.g., threshold chosen on a validation set, or a fixed false-positive budget). Without this, the 'consistently outperforms' conclusion in Table III is not interpretable.
  2. [§III-D, Eq. (10), and §IV-F] Eq. (10) states a formal inequality: Dist(D_real, Dhat_syn) < Dist(D_real, D_syn), and the text says 'which will be proved in the visualization part of our experiments.' Section IV-F provides only UMAP plots of synthetic samples generated by the method itself and test samples. This is not a proof, and it is self-referential because the visualized samples come from the very pipeline whose distributional property is being asserted. No quantitative divergence measure is reported. Either provide a quantitative distribution-gap estimate (e.g., MMD or energy distance between real test anomalies and synthetic samples for CAP vs. CAPMix) or explicitly rephrase Eq. (10) as a motivating hypothesis rather than a theorem.
  3. [§III-D, Eq. (6), Eq. (7), and Algorithm 1] The label-revision rule is defined inconsistently in three places. Eq. (6) defines the soft-label zone as [0, μ_d − γ·σ_d] and says samples in this zone receive 1/γ. Eq. (7) instead assigns 1/γ when d(X_syn, C_n) ≤ μ_d + γσ_d and 1 only when d > μ_d + γσ_d. Algorithm 1, line 25, sets y'_i = 1/γ when d(X'_i) > μ_d + γσ_d. These conditions are mutually incompatible. This ambiguity affects the reproducibility of a core component and must be corrected in a revised version.
  4. [Table IV and §IV-D] The ablation table and main table are inconsistent and the ESA result is unexplained. Table III reports CAPMix on ESA as 84.46±5.50, while Table IV reports the full CAPMix as 84.46±6.67. More importantly, the entire ESA gain over CutAddPaste (18.56 → 84.46) is already present in the CAP-mix variant (84.46), with label revision contributing nothing on ESA, yet the text only says CAP performs poorly 'possibly due to the large window size' and does not explain why dual-space Mixup produces a ~66-point jump on this dataset while contributing little elsewhere. Please reconcile the inconsistent standard deviations and provide a mechanistic explanation or focused analysis of the ESA gain.
minor comments (5)
  1. [§III-B, Eq. (3)] Eq. (3) appears corrupted: it reuses Ltrain for the test-time objective and contains 'D∫†' in place of the anomaly distribution. Please correct the notation.
  2. [§IV-D, Table IV] The text says 'CAP-γ significantly improves the performance on AIOps, and the average of CAPMix rises 1.05% compared to CAP-mix,' but Table IV shows CAP-γ giving no improvement over CAP on UCR, SWaT, WADI, and ESA (identical values). The claim of improvement on UCR is not supported by the reported numbers.
  3. [§IV-G] The hyperparameter analysis sweeps γ and α, but the paper does not state how these values were selected for the main results (e.g., validation set or test set). Please clarify the selection procedure so the reader can assess the risk of overfitting to the test benchmark.
  4. [Algorithm 1] The pseudocode has syntax and logic issues: 'anomaly num = int(v×B))' has an extra parenthesis, and 'random.sample(X′, anomaly num)' selects a random subset but then X′ is used again; the relation between this selection and the earlier loop is unclear. Please revise the pseudocode.
  5. [§VI] The conclusion states the method 'consistently outperforms strong baselines' and the main-results section refers to 'all four datasets,' although five datasets are used. Please correct this inconsistency.

Circularity Check

1 steps flagged · score 2.0 of 10

Minor self-referential validation in the anomaly-shift claim; overall method is self-contained and benchmarked externally.

  1. other [Section III-D (Eq. 10) and Section IV-F (Visualization)]
    "Denoting the revised synthetic anomaly distribution as Dsyn after applying DTW-based soft labeling and dual-space Mixup, we claim that Dist(Dreal, Dsyn) < Dist(Dreal, Dsyn), which will be proved in the visualization part of our experiments."

    The paper's central theoretical claim—that CAPMix reduces anomaly shift—is not derived from the method's construction or from any established theorem. Instead, Eq. (10) is a bare assertion, and the promised proof is a UMAP visualization of CAPMix's own generated samples plotted against test anomalies. This is self-referential evidence: the 'proof' uses the method's outputs to validate the method's effectiveness, rather than an independent derivation from first principles. It does not make the result equivalent to the inputs by definition, because the method does not explicitly optimize for closeness to real anomalies, but it falls short of a genuine theoretical justification.

full rationale

The CAPMix method is largely self-contained: anomaly injection, label revision, and dual-space mixup are all described in the paper, and the experimental evaluation is conducted on external benchmarks (AIOps, UCR, SWaT, WADI, ESA) with comparisons to a wide range of published baselines. The method extends the authors' prior CutAddPaste work, but that is a normal incremental extension rather than a load-bearing citation. The main circularity concern is the anomaly-shift claim (Eq. 10), which is asserted and then 'proved' by the paper's own visualization—a self-referential validation. In addition, the evaluation protocol searches for an optimal anomaly threshold on the test set (Section IV-B), which could inflate reported performance if not applied identically to all baselines; the paper does not explicitly state that every baseline receives the same oracle threshold search, so this is an evaluation-fairness risk rather than a logical circularity. There is also an internal inconsistency between Eq. (6), Eq. (7), and Algorithm 1 regarding the assignment of soft labels, but that is a correctness bug, not circularity. Overall, the central empirical claims are based on external benchmarks and are not forced by construction; the issues found are secondary limitations.

Assumptions & free parameters 7 free parameters · 6 assumptions · 1 invented entities

The method rests on several heuristic assumptions: that the training data is mostly normal, that DTW distance to a mean sequence captures anomaly likelihood, that the five CutAddPaste anomaly types span real failure modes, and that the test-set threshold search does not bias results. The injected anomaly types and soft-label boundary are governed by hyperparameters tuned per dataset, and the claimed distributional improvement (Eq. 10) is not derived.

free parameters (7)
  • gamma (label revision width) = 1.00 to 1.08 on AIOps; 1.4 to 2.2 on WADI
    Tuned per dataset; controls the soft-label boundary in Eq. (7).
  • alpha (mixup ratio) = swept across ranges per dataset
    Controls interpolation strength in dual-space mixup.
  • rho (trend degree) = from prior work, not reported here
    Controls slope magnitude in CutAddPaste trend addition.
  • zeta (minimum patch length) = from prior work
    Minimum length of the cut patch in CutAddPaste.
  • e (number of dimensions for trend) = set in Algorithm 1
    Number of dimensions selected to receive additive trends.
  • v (fraction of samples converted to anomalies) = set in Algorithm 1
    Controls how many pseudo-anomalies are generated per batch.
  • tau (detection threshold) = searched in [-3,3] on test set Z-scores
    Optimal threshold selected on test data per 3-sigma rule, a post-hoc oracle.
assumptions (6)
  • domain assumption Training data is predominantly normal, so the mean sequence C_n is a valid normality center.
    Used in Eq. (5) to compute C_n; if training data is heavily contaminated, the center is biased.
  • domain assumption Time series can be decomposed as X = Gamma(2*pi*omega*T) + Theta(T) with shapelet, seasonality, and trend components.
    Adopted from Lai et al. [11] to define five anomaly types.
  • domain assumption DTW distance to C_n is a valid measure of anomaly likelihood.
    Used in label revision; the paper admits it fails to characterize multivariate anomaly semantics.
  • domain assumption The five injected anomaly types (shape, correlation, seasonality, trend, point) cover the real anomaly distribution.
    Core design assumption of CutAddPaste.
  • ad hoc to paper Searching the threshold tau on the test set gives an unbiased evaluation.
    Section IV-B: 'we search for the optimal anomaly threshold tau in [-3,3]' on test data; if applied only to CAPMix, comparisons are biased.
  • ad hoc to paper Equation (10) holds, i.e., label revision and mixup reduce the distribution gap to real anomalies.
    Claimed and 'proved' only by UMAP visualization, not derived.
invented entities (1)
  • Anomaly Shift concept
    purpose: Frames the problem that synthetic anomalies may be too close or too far from real anomalies, motivating label revision and mixup.
    Formalized in Eq. (1) as a distributional gap, but not measured directly; evidence of its alleviation is qualitative UMAP.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CAPMix: Robust KPI Anomaly Detection for AIOps in Noisy and Dynamic Environments." pith.science (2026). https://pith.science/paper/T6ICNIAG

@misc{pith2026250906419,
  author       = {Pith},
  title        = {Pith review of: CAPMix: Robust KPI Anomaly Detection for AIOps in Noisy and Dynamic Environments},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/T6ICNIAG}},
  note         = {Machine review of arXiv:2509.06419}
}
read the original abstract

Time-series anomaly detection is crucial in AIOps for maintaining large-scale service reliability. In production, streams of Key Performance Indicators (KPI) are high-dimensional, non-stationary, and affected by noise, deployment changes, and latent anomalies, making real failures hard to distinguish from benign variation. Most existing methods assume either normality (learning from "normal" history) or rely on injected anomalies for training. Yet injected patterns often misalign with real failure modes, skewing decision boundaries -- aka. Anomaly Shift. We propose CAPMix, a controllable anomaly augmentation framework with prior-guided injection for realistic temporal behaviors. CAPMix combines label revision and dual-space mixup to enhance robustness under contaminated and mixed data. CAPMix consistently outperforms state-of-the-art methods on public AIOps and time-series benchmarks. It has been deployed in Kuaishou's large-scale production system, reducing false alarms and improving monitoring reliability. A real-world dataset is also released to enrich the research on robust KPI anomaly detection.

Figures

Figures reproduced from arXiv: 2509.06419 by the authors.

Figure 1
Figure 1. Schematic of normality and anomaly assumptions used in AD [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. In (a), anomalies are overly similar to the normal data, leading to misclassification as normal. In (b), anomalies are injected too far from the real data distribution, resulting in poor boundary generalization. However, it is often not so easy to evaluate the rationality of anomaly injection. To formalize this problem, let Xnorm ∼ Dnorm and X a real ∼ Dreal denote the normal and real anomalous samples, respectively… view at source ↗
Figure 3
Figure 3. The overview of CAPMix, composed of three main components: a) Generation with CutAddPaste, which produces diverse synthetic anomalies by [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Example of generating an anomalous sample over sine waves. The [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Robustness evaluation of CAPMix and CutAddPaste (CAP) under [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]
Figure 6
Figure 6. Figure 6: The visualization of CAPMix AD results on the five datasets. The x-axis is timestamps, and the y-axis is signal values. The original data is present [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]
Figure 7
Figure 7. Figure 7: UMAP comparison between synthetic samples generated by CAP (a-e) and CAPMix (f-j), with normal (blue) and real anomalous (red) test samples. [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]
Figure 8
Figure 8. Figure 8: Effect of different γ values (five lines) and α settings on the RPA F1 score. Each line represents one γ; points along each line correspond to different α values. baselines and more fine-grained visualizations. Such an anal￾ysis would provide deeper insight into how th…

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

50 extracted references · 36 canonical work pages

  1. [1]

    Lstm-based encoder-decoder for multi-sensor anomaly de- tection,

    P. Malhotra, A. Ramakrishnan, G. Anand, L. Vig, P. Agarwal, and G. Shroff, “Lstm-based encoder-decoder for multi-sensor anomaly de- tection,”arXiv preprint arXiv:1607.00148, 2016

    arXiv 2016

  2. [2]

    Deep one-class classification,

    L. Ruff, R. Vandermeulen, N. Goernitz, L. Deecke, S. A. Siddiqui, A. Binder, E. M ¨uller, and M. Kloft, “Deep one-class classification,” inInternational conference on machine learning. PMLR, 2018, pp. 4393–4402

    2018

  3. [3]

    Deep contrastive one-class time series anomaly detection,

    R. Wang, C. Liu, X. Mou, K. Gao, X. Guo, P. Liu, T. Wo, and X. Liu, “Deep contrastive one-class time series anomaly detection,” in Proceedings of the 2023 SIAM International Conference on Data Mining (SDM). SIAM, 2023, pp. 694–702

    work page 2023

  4. [4]

    Deep autoencoding one-class time series anomaly detection,

    X. Mou, R. Wang, T. Wang, J. Sun, B. Li, T. Wo, and X. Liu, “Deep autoencoding one-class time series anomaly detection,” inICASSP 2023- 2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 2023, pp. 1–5

    work page 2023

  5. [5]

    RoCA: Robust Contrastive One-class Time Series Anomaly Detection with Contaminated Data

    X. Mou, R. Wang, B. Li, T. Wo, J. Sun, H. Wang, and X. Liu, “Roca: Robust contrastive one-class time series anomaly detection with contaminated data,”arXiv preprint arXiv:2503.18385, 2025

    work page Pith review arXiv 2025

  6. [6]

    Deep anomaly detec- tion with outlier exposure,

    D. Hendrycks, M. Mazeika, and T. Dietterich, “Deep anomaly detec- tion with outlier exposure,” inInternational Conference on Learning Representations, 2018

    work page 2018

  7. [7]

    Cutpaste: Self-supervised learning for anomaly detection and localization,

    C.-L. Li, K. Sohn, J. Yoon, and T. Pfister, “Cutpaste: Self-supervised learning for anomaly detection and localization,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, 2021, pp. 9664–9674

    2021

  8. [8]

    Deep semi-supervised anomaly detection,

    L. Ruff, R. A. Vandermeulen, N. G ¨ornitz, A. Binder, E. M ¨uller, K.-R. M¨uller, and M. Kloft, “Deep semi-supervised anomaly detection,”arXiv preprint arXiv:1906.02694, 2019

    arXiv 1906

Show all 50 references
  1. [9]

    Neural contextual anomaly detection for time series,

    C. U. Carmona, F.-X. Aubet, V . Flunkert, and J. Gasthaus, “Neural contextual anomaly detection for time series,”IJCAI, 2022

    2022

  2. [10]

    Anomalybert: Self-supervised transformer for time series anomaly detection using data degradation scheme,

    Y . Jeong, E. Yang, J. H. Ryu, I. Park, and M. Kang, “Anomalybert: Self-supervised transformer for time series anomaly detection using data degradation scheme,”arXiv preprint arXiv:2305.04468, 2023

    2023 arXiv

  3. [11]

    Revisiting time series outlier detection: Definitions and benchmarks,

    K.-H. Lai, D. Zha, J. Xu, Y . Zhao, G. Wang, and X. Hu, “Revisiting time series outlier detection: Definitions and benchmarks,” inThirty- fifth conference on neural information processing systems datasets and benchmarks track (round 1), 2021

    2021

  4. [12]

    Detecting spacecraft anomalies using lstms and nonparametric dynamic thresholding,

    K. Hundman, V . Constantinou, C. Laporte, I. Colwell, and T. Soder- strom, “Detecting spacecraft anomalies using lstms and nonparametric dynamic thresholding,” inProceedings of the 24th ACM SIGKDD international conference on knowledge discovery & data mining, 2018, pp. 387–395

    2018

  5. [13]

    Cutaddpaste: Time series anomaly detection by exploiting abnormal knowledge,

    R. Wang, X. Mou, R. Yang, K. Gao, P. Liu, C. Liu, T. Wo, and X. Liu, “Cutaddpaste: Time series anomaly detection by exploiting abnormal knowledge,” inProceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 2024, pp. 3176–3187

    2024

  6. [14]

    Deep learning for anomaly detection: A review,

    G. Pang, C. Shen, L. Cao, and A. V . D. Hengel, “Deep learning for anomaly detection: A review,”ACM Computing Surveys (CSUR), vol. 54, no. 2, pp. 1–38, 2021

    2021

  7. [15]

    Unsupervised anomaly detection with generative adversarial networks to guide marker discovery,

    T. Schlegl, P. Seeb ¨ock, S. M. Waldstein, U. Schmidt-Erfurth, and G. Langs, “Unsupervised anomaly detection with generative adversarial networks to guide marker discovery,” inInternational conference on information processing in medical imaging. Springer, 2017, pp. 146– 157

    2017

  8. [16]

    Gan-based anomaly detection: A review,

    X. Xia, X. Pan, N. Li, X. He, L. Ma, X. Zhang, and N. Ding, “Gan-based anomaly detection: A review,”Neurocomputing, vol. 493, pp. 497–535, 2022

    2022

  9. [17]

    Deep autoencoding gaussian mixture model for unsupervised anomaly detection,

    B. Zong, Q. Song, M. R. Min, W. Cheng, C. Lumezanu, D. Cho, and H. Chen, “Deep autoencoding gaussian mixture model for unsupervised anomaly detection,” inICLR, 2018

    2018

  10. [18]

    Learning and evaluating representations for deep one-class classification,

    K. Sohn, C.-L. Li, J. Yoon, M. Jin, and T. Pfister, “Learning and evaluating representations for deep one-class classification,”ICLR, 2021

    2021

  11. [19]

    Genias: Generator for instantiating anomalies in time series,

    Z. Z. Darban, Q. Wang, G. I. Webb, S. Pan, C. C. Aggarwal, and M. Salehi, “Genias: Generator for instantiating anomalies in time series,” arXiv preprint arXiv:2502.08262, 2025

    2025

  12. [20]

    Robust and explainable detector of time series anomaly via augmenting multiclass pseudo- anomalies,

    K. Obata, Y . Matsubara, and Y . Sakurai, “Robust and explainable detector of time series anomaly via augmenting multiclass pseudo- anomalies,” 2025

    2025

  13. [21]

    Rethinking assumptions in deep anomaly detection,

    L. Ruff, R. A. Vandermeulen, B. J. Franks, K.-R. M ¨uller, and M. Kloft, “Rethinking assumptions in deep anomaly detection,”arXiv preprint arXiv:2006.00339, 2020. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2021 14

    2006 arXiv

  14. [22]

    mixup: Beyond empirical risk minimization,

    H. Zhang, M. Cisse, Y . N. Dauphin, and D. Lopez-Paz, “mixup: Beyond empirical risk minimization,”arXiv preprint arXiv:1710.09412, 2017

    2017 arXiv

  15. [23]

    Cutmix: Reg- ularization strategy to train strong classifiers with localizable features,

    S. Yun, D. Han, S. J. Oh, S. Chun, J. Choe, and Y . Yoo, “Cutmix: Reg- ularization strategy to train strong classifiers with localizable features,” inProceedings of the IEEE/CVF international conference on computer vision, 2019, pp. 6023–6032

    2019

  16. [24]

    Snapmix: Semantically proportional mixing for augmenting fine-grained data,

    S. Huang, X. Wang, and D. Tao, “Snapmix: Semantically proportional mixing for augmenting fine-grained data,” inProceedings of the AAAI conference on artificial intelligence, vol. 35, no. 2, 2021, pp. 1628–1636

    2021

  17. [25]

    Smoothmix: Training confidence-calibrated smoothed classifiers for certified robustness,

    J. Jeong, S. Park, M. Kim, H.-C. Lee, D.-G. Kim, and J. Shin, “Smoothmix: Training confidence-calibrated smoothed classifiers for certified robustness,”Advances in Neural Information Processing Sys- tems, vol. 34, pp. 30 153–30 168, 2021

    2021

  18. [26]

    Manifold mixup: Better representations by interpolating hidden states,

    V . Verma, A. Lamb, C. Beckham, A. Najafi, I. Mitliagkas, D. Lopez-Paz, and Y . Bengio, “Manifold mixup: Better representations by interpolating hidden states,” inInternational conference on machine learning. PMLR, 2019, pp. 6438–6447

    2019

  19. [27]

    Finding order in chaos: A novel data augmentation method for time series in contrastive learning,

    B. U. Demirel and C. Holz, “Finding order in chaos: A novel data augmentation method for time series in contrastive learning,”Advances in Neural Information Processing Systems, vol. 36, pp. 30 750–30 783, 2023

    2023

  20. [28]

    Timemixer++: A general time series pattern machine for universal predictive analysis,

    S. Wang, J. Li, X. Shi, Z. Ye, B. Mo, W. Lin, S. Ju, Z. Chu, and M. Jin, “Timemixer++: A general time series pattern machine for universal predictive analysis,”arXiv preprint arXiv:2410.16032, 2024

    2024 arXiv

  21. [29]

    Current time series anomaly detection bench- marks are flawed and are creating the illusion of progress,

    R. Wu and E. Keogh, “Current time series anomaly detection bench- marks are flawed and are creating the illusion of progress,”IEEE Transactions on Knowledge and Data Engineering, 2021

    2021

  22. [30]

    Evaluating real-time anomaly detection algorithms–the numenta anomaly benchmark,

    A. Lavin and S. Ahmad, “Evaluating real-time anomaly detection algorithms–the numenta anomaly benchmark,” in2015 IEEE 14th in- ternational conference on machine learning and applications (ICMLA). IEEE, 2015, pp. 38–44

    2015

  23. [31]

    S5 - a labeled anomaly detection dataset, version 1.0 (16m),

    Yahoo, “S5 - a labeled anomaly detection dataset, version 1.0 (16m),” https://webscope.sandbox.yahoo.com/catalog.php?datatype=s& did=70, 2015, accessed: 2023-04-14

    2015

  24. [32]

    Robust anomaly detection for multivariate time series through stochastic recurrent neural network,

    Y . Su, Y . Zhao, C. Niu, R. Liu, W. Sun, and D. Pei, “Robust anomaly detection for multivariate time series through stochastic recurrent neural network,” inProceedings of the 25th ACM SIGKDD international conference on knowledge discovery & data mining, 2019, pp. 2828– 2837

    2019

  25. [33]

    The 1st match for aiops,

    AIOps Challenge., “The 1st match for aiops,” https://github.com/ NetManAIOps/KPI-Anomaly-Detection, 2018, accessed: 2023-04-14

    2018

  26. [34]

    Swat: A water treatment testbed for research and training on ics security,

    A. P. Mathur and N. O. Tippenhauer, “Swat: A water treatment testbed for research and training on ics security,” in2016 international workshop on cyber-physical systems for smart water networks (CySWater). IEEE, 2016, pp. 31–36

    2016

  27. [35]

    Wadi: a water distribution testbed for research in the design of secure cyber physical systems,

    C. M. Ahmed, V . R. Palleti, and A. P. Mathur, “Wadi: a water distribution testbed for research in the design of secure cyber physical systems,” in Proceedings of the 3rd international workshop on cyber-physical systems for smart water networks, 2017, pp. 25–28

    2017

  28. [36]

    European space agency benchmark for anomaly detection in satellite telemetry,

    K. Kotowski, C. Haskamp, J. Andrzejewski, B. Ruszczak, J. Nalepa, D. Lakey, P. Collins, A. Kolmas, M. Bartesaghi, J. Martinez-Heraset al., “European space agency benchmark for anomaly detection in satellite telemetry,”arXiv preprint arXiv:2406.17826, 2024

    2024 arXiv

  29. [37]

    Unsupervised anomaly detection via variational auto- encoder for seasonal kpis in web applications,

    H. Xu, W. Chen, N. Zhao, Z. Li, J. Bu, Z. Li, Y . Liu, Y . Zhao, D. Pei, Y . Fenget al., “Unsupervised anomaly detection via variational auto- encoder for seasonal kpis in web applications,” inProceedings of the 2018 World Wide Web Conference, 2018, pp. 187–196

    2018

  30. [38]

    Towards a rigorous evaluation of time-series anomaly detection,

    S. Kim, K. Choi, H.-S. Choi, B. Lee, and S. Yoon, “Towards a rigorous evaluation of time-series anomaly detection,” inProceedings of the AAAI Conference on Artificial Intelligence, vol. 36, 2022, pp. 7194–7201

    2022

  31. [39]

    Local evaluation of time series anomaly detection algorithms,

    A. Huet, J. M. Navarro, and D. Rossi, “Local evaluation of time series anomaly detection algorithms,” inACM SIGKDD, 2022, pp. 635–645

    2022

  32. [40]

    Support vector method for novelty detection

    B. Sch ¨olkopf, R. C. Williamson, A. J. Smola, J. Shawe-Taylor, J. C. Platt et al., “Support vector method for novelty detection.” inNIPS, vol. 12. Citeseer, 1999, pp. 582–588

    1999

  33. [41]

    Isolation-based anomaly detection,

    F. T. Liu, K. M. Ting, and Z.-H. Zhou, “Isolation-based anomaly detection,”ACM Transactions on Knowledge Discovery from Data (TKDD), vol. 6, no. 1, pp. 1–39, 2012

    2012

  34. [42]

    Robust random cut forest based anomaly detection on streams,

    S. Guha, N. Mishra, G. Roy, and O. Schrijvers, “Robust random cut forest based anomaly detection on streams,” inICML. PMLR, 2016, pp. 2712–2721

    2016

  35. [43]

    Time-series anomaly detection service at microsoft,

    H. Ren, B. Xu, Y . Wang, C. Yi, C. Huang, X. Kou, T. Xing, M. Yang, J. Tong, and Q. Zhang, “Time-series anomaly detection service at microsoft,” inACM SIGKDD, 2019, pp. 3009–3017

    2019

  36. [44]

    Matrix profile xxiv: scaling time series anomaly detection to trillions of datapoints and ultra-fast arriving data streams,

    Y . Lu, R. Wu, A. Mueen, M. A. Zuluaga, and E. Keogh, “Matrix profile xxiv: scaling time series anomaly detection to trillions of datapoints and ultra-fast arriving data streams,” inProceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 2022, pp....

    2022

  37. [45]

    Time-series representation learning via temporal and contextual con- trasting,

    E. Eldele, M. Ragab, Z. Chen, M. Wu, C. K. Kwoh, X. Li, and C. Guan, “Time-series representation learning via temporal and contextual con- trasting,”IJCAI, 2021

    2021

  38. [46]

    Anomaly transformer: Time series anomaly detection with association discrepancy,

    J. Xu, H. Wu, J. Wang, and M. Long, “Anomaly transformer: Time series anomaly detection with association discrepancy,” inInternational Conference on Learning Representations, 2021

    2021

  39. [47]

    Mixmamba: Time series modeling with adaptive expertise,

    K. Alkilane, Y . He, and D.-H. Lee, “Mixmamba: Time series modeling with adaptive expertise,”Information Fusion, vol. 112, p. 102589, 2024

    2024

  40. [48]

    Beyond sharing: Conflict-aware multivariate time series anomaly detection,

    H. Si, C. Pei, Z. Li, Y . Zhao, J. Li, H. Zhang, Z. Diao, J. Li, G. Xie, and D. Pei, “Beyond sharing: Conflict-aware multivariate time series anomaly detection,” inProceedings of the 31st ACM Joint European Software Engineering Conference and Symposium on the Foundations of So...

    2023

  41. [49]

    Sensitivehue: Multivariate time series anomaly detection by enhancing the sensitivity to normal patterns,

    Y . Feng, W. Zhang, Y . Fu, W. Jiang, J. Zhu, and W. Ren, “Sensitivehue: Multivariate time series anomaly detection by enhancing the sensitivity to normal patterns,” inProceedings of the 30th ACM SIGKDD Confer- ence on Knowledge Discovery and Data Mining, 2024, pp. 782–793

    2024

  42. [50]

    Merlion: A machine learning library for time series,

    A. Bhatnagar, P. Kassianik, C. Liu, T. Lan, W. Yang, R. Cassius, D. Sahoo, D. Arpit, S. Subramanian, G. Wooet al., “Merlion: A machine learning library for time series,”arXiv preprint arXiv:2109.09265, 2021

    2021 arXiv

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.