Pith. sign in

REVIEW 7 major objections 5 minor 39 references

Tabular Data Adapters: Improving Outlier Detection for Unlabeled Private Data

T0 review · 7 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read The paper claims that unlabeled private tabular data can be soft-labeled for outlier detection by finding a statistically similar public dataset, transforming the private rows into that dataset's format with a shared-latent autoencoder…

desk verdict A practical soft-labeling pipeline with real gains on three of four metrics; the key transfer assumption is under-validated and the abstract overclaims, but it deserves a serious referee. read the letter →

arxiv 2504.20862 v1 pith:BX7RDACQ submitted 2025-04-29 cs.LG cs.AI

Dayananda Herurkar , Jörn Hees , Vesselin Tzvetkov , Andreas Dengel This is my paper · ORCID
classification cs.LGcs.AI
keywords outlierdetectionsoftlabelstabulardataweaksupervisiondatasettransformationautoencodersimilaritycoldstart
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

Tabular Data Adapters is a method for labeling, without manual annotation, the outlier versus normal status of rows in a private tabular dataset. The idea is to find a public dataset that is statistically similar, train a shared autoencoder that rewrites private rows into the public dataset's format, and then run that public dataset's tuned outlier detection models on the rewritten rows. The paper argues that this generates more accurate soft labels than running standard unsupervised outlier detectors directly on the private data, while using far less computation time. If true, the method lets data owners escape the cold-start problem by bootstrapping downstream models from weak labels produced by public models instead of paying for labeling.

What carries the argument

The load-bearing object is the Dataset Transformation autoencoder: two dataset-specific encoder-decoder pairs joined by shared layers $m_\theta$. Training alternates private and public batches and minimizes a reconstruction loss (Equations 1-3); inference produces crossover samples $\hat{x}^{co} = \mathrm{Dec}_{pub}(m_\theta(\mathrm{Enc}_{prv}(x^{prv})))$ (Equation 4), which put private rows into the public format so a model trained on the public dataset can score them. Its success is checked only by the DS Diff inequality (Equation 5), which compares aggregate PCA reconstruction-error curves before and after transformation. The companion Dataset Similarity Measure ranks public datasets by the sum of absolute differences between the private and public reconstruction-error curves, and the final soft labels come from majority voting over the selected public models' predictions on the crossover samples.

What would settle it

Take a private dataset whose outliers are defined by a rare combination of two features, train the Dataset Transformation autoencoder, and check whether TDA's soft labels track the true outliers; if the crossover projection compresses those two features into one, balanced accuracy should fall to chance while the DS Diff inequality still holds, and that outcome would falsify the claim that successful format alignment preserves outlier structure.

Watch

Extended reading notes

Core claim

The paper's central claim is that you can generate useful soft labels for a private, unlabeled tabular dataset by ranking public tabular datasets with a PCA reconstruction-error similarity measure, training a shared-latent autoencoder that rewrites private samples in the chosen public dataset's format, and then applying that public dataset's already-tuned outlier detection models to the rewritten samples and fusing their votes. In a leave-one-out study over 50 tabular datasets, the authors report that both variants, Top1-DS and TopN-DS, beat the averaged and default unsupervised baselines on balanced accuracy, F1-score, and PR-AUC while using far less compute: Top1-DS averages 0.639 balanced accuracy and 9.94 minutes, versus 0.578 and 282.64 minutes for the averaged baseline. They also report that ROC-AUC remains at baseline level and that both methods stay below the supervised Best OD upper bound, which assumes labels that the method is designed to avoid needing.

Load-bearing premise

The one premise everything rests on is that the neural network that compresses private rows and rebuilds them in the public format keeps whatever makes private outliers anomalous; if that reshaping smooths away the anomaly, the labels are random no matter how similar the chosen public dataset looks.

Editorial extensions

If this is right

  • If the claim is right, an organization with an unlabeled private table can obtain outlier and normal weak labels without manual annotation, by pointing the pipeline at a repository of public datasets with tuned outlier detection models.
  • Method1's 9.94-minute average runtime, versus 282.64 minutes for the averaged baseline, makes per-dataset weak labeling cheap enough to rerun as private data shifts over time.
  • The reported upward trend between dataset similarity and balanced accuracy implies that maintaining a diverse index of public datasets directly improves label quality, so the similarity ranking is not a detail but a driver of performance.
  • Because ROC-AUC stays at baseline level while F1 and PR-AUC improve, the method's gains are concentrated in precision-oriented metrics, meaning it is most useful when outlier flags are used to prioritize candidates rather than to produce exact counts.
  • Both the single-dataset variant and the multi-dataset fusion variant beat the unsupervised baselines, so the approach does not depend on knowing in advance which public dataset is the perfect match.

Reading between the lines

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

  • The paper does not test whether the soft labels actually improve a downstream classifier trained on the private data; a natural next experiment is to train a model on TDA-soft-labeled private rows and compare it against models trained on the baseline labels or on unlabeled data with self-training.
  • Because the similarity measure only compares reconstruction-error curves, it could rank two datasets as similar even when their outlier-generating mechanisms differ; a sharper similarity test could compare per-sample reconstruction patterns or use a held-out probe of public outlier labels.
  • If the transformation is faithful enough to preserve outlier structure, the same pipeline could transfer other public-model capabilities to private tabular data, such as imputation or concept-drift detection, not just outlier flags.
  • The privacy angle is implicit: the private data never needs to be shared, but the trained transformation model is itself a function of private data, and whether crossover samples or the shared layers leak private information is not analyzed.
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

7 major / 5 minor

Summary. The paper proposes Tabular Data Adapters (TDA), a pipeline for generating soft labels for unlabeled private tabular data in outlier detection. A PCA-based reconstruction-error curve is used to rank public datasets by statistical similarity to the private dataset; a shared-latent autoencoder pair (the Dataset Transformation module) then maps private samples into the selected public dataset's feature space; publicly pre-trained outlier-detection models are applied to these 'crossover' samples, and their predictions are fused by majority voting. Two variants are introduced: Top1-DS, which uses the single most similar public dataset with its m best models, and TopN-DS, which uses the top n public datasets with one best model each. The method is evaluated in a leave-one-out fashion over 50 tabular datasets from ADBench and three financial datasets, with Table 1 reporting balanced accuracy, F1, PR-AUC, ROC-AUC, and runtime against three baselines: Best OD (supervised upper bound), Avg OD, and Default OD.

Significance. If the central claim holds, the contribution is practically useful: it offers a cold-start path to weak labels for tabular outlier detection without manual labeling, and the reported compute savings (9.94 and 27.28 minutes versus 282.64 for Avg OD) are substantial. The paper deserves credit for a broad 50-dataset leave-one-out evaluation, two clearly described method variants, a domain-level ablation, and an explicit analysis of where computation time is spent. However, the current evidence does not yet establish the load-bearing claim that the Dataset Transformation preserves the per-sample outlier structure needed for meaningful label transfer. The missing validation of that step, combined with the unresolved ROC-AUC shortfall and under-specified hyperparameters, means the headline 'more accurate' claim is not yet substantiated as written.

major comments (7)
  1. [Section 3.1, Eq. (5)] The success criterion for Dataset Transformation is the inequality DS_Diff(Xprv, Xpub) > DS_Diff(Xprv, Xhat_co), but DS_Diff is never defined in this manuscript, no numerical DS_Diff values are reported, and the criterion is aggregate: it compares reconstruction-error similarity between whole datasets, not whether individual private outliers remain outliers after crossover. Because Eq. (3) is plain MSE reconstruction on mixed batches, rare outliers contribute negligibly to the training loss, so there is no mechanism that explicitly preserves outlier-specific directions. This is load-bearing: without evidence that outlier/inlier separability survives the crossover mapping, the public models' scores on crossover data may be near-random even when the selected public dataset is highly similar. I ask the authors to provide a direct per-sample validation, e.g., reporting, for each held-out private dataset, the ROC-AUC or rank correlation between the public model's crossover scores and the private ground-truth labels, or an outlier-score distribution comparison before and after transformation. The paper's own §5.1 remark that the ROC-AUC drop is due to 'inherent noise introduced during dataset transformation' makes this test necessary rather than optional.
  2. [Table 1 and Abstract] The abstract and Section 5.1 state that the method provides 'more accurate annotations' than baseline approaches, but Table 1 shows both methods have lower ROC-AUC (0.689 and 0.683) than Avg OD (0.691) and Default OD (0.690). Since ROC-AUC is a standard threshold-independent metric for outlier-detection quality, the unqualified 'more accurate' claim is contradicted on one of the four reported metrics. The authors should either qualify the claim to the metrics where the method wins, or provide an explanation and supporting analysis for the ROC-AUC shortfall. Additionally, no standard deviations, confidence intervals, significance tests, or number of random seeds are reported, so it is unclear whether the differences in balanced accuracy, F1, and PR-AUC are robust across runs and dataset subgroups.
  3. [Sections 3.2 and 4.3] The parameters m (number of best public models in Method1) and n (number of top public datasets in Method2) are never given concrete values for the experiments in Table 1, even though they are the key free parameters of the method. Figure 5 plots m and n up to 9 or 10, but the text does not state which values were used for the headline results. Similarly, the Dataset Similarity Measure is described as varying the number of principal components 'from 1 to 100,' but several datasets in Table 2 have fewer than 100 features; the handling of this boundary case is not specified. Without these details the experiments are not reproducible and the sensitivity of the reported gains to m and n cannot be assessed.
  4. [Section 3.1, Dataset Similarity Measure] The text says 'To find the most similar public datasets, we rank them descending by their SAD to the private dataset,' which is inconsistent with the definition of SAD as a sum of absolute differences where lower values indicate higher similarity. Figure 8 plots a 'similarity score' between 0.2 and 1.0 but does not define how that score is derived from SAD. Because the selection of the public dataset is the first step of the pipeline, this ambiguity should be resolved by defining the similarity score precisely and correcting the ranking direction.
  5. [Section 4.3 and Table 1] The evaluation metrics include PR-AUC and ROC-AUC, but the description of the method says soft labels are produced by majority voting, which yields binary predictions rather than a continuous score. It is not specified whether PR-AUC and ROC-AUC are computed from vote proportions, from averaged anomaly scores of the individual public models, or from some other ranking. This matters because the method's PR-AUC gain over the baselines could in part reflect how the score is constructed. The authors should state exactly what quantity is used to rank samples for these two metrics.
  6. [Section 5.2 and Figure 5] The compute-time comparison, while informative, does not account for the cost of training the Dataset Transformation model for each new private dataset, nor for the presumed pre-computation of the public model repository. The claim of 'reducing computational time' is relative to baselines that run all 11 OD algorithms with multiple hyperparameter configurations, which is a fair baseline for a cold-start scenario, but the paper should state whether the reported Method1/Method2 times include transformation training and model inference, and should clarify what is assumed to be available as pre-existing assets.
  7. [Section 5.1, Figure 4] The box plots in Figure 4 are described as showing 'consistent superiority' and 'comparable performance,' but the ROC-AUC panel shows the methods slightly below the baselines on average, and no statistical testing is reported. The visual claim of consistency should be backed by per-dataset counts of wins/losses against each baseline for each metric, or by paired tests; otherwise the figure overstates the strength of the evidence.
minor comments (5)
  1. [Throughout] There are several typographical issues, including 'V esselin' in the author list, 'Leave one out' used as a compound adjective, and inconsistent hyphenation of 'leave-one-out'; these should be cleaned up in revision.
  2. [Section 3.1, Eq. (3)] The loss notation uses nprv and npub before defining them; please state explicitly that these are the batch sizes for the private and public batches, respectively.
  3. [Section 4.1] The text says 47 datasets were selected from ADBench and 10 were excluded, but ADBench is stated to contain 57 datasets; please verify the arithmetic and clarify whether the three additional financial datasets are counted within the 50 used in Table 1.
  4. [Figure 2 caption] The caption says the y-axis shows 'Mean Reconstruction Error' but the figure is described as a similarity plot; please ensure the axes and the SAD computation are described consistently in the caption and the text.
  5. [References] Reference [14] is the self-cited source for Dataset Transformation and is central to the method, but the paper does not summarize the DS_Diff metric from that work; a brief definition in the text or appendix would make this manuscript self-contained.

Circularity Check

1 steps flagged · score 3.0 of 10

Dataset Transformation's preservation of per-sample outlier structure is the load-bearing premise; it is imported from self-cited [14] and validated only by an undefined aggregate DS Diff metric.

  1. self citation load bearing [Section 3.1 (Dataset Transformation), Eqs. 4-5; §5.1]
    "Dataset Transformation: This component (Figure 3) leverages an autoencoder-based neural network to transform samples from private datasets into public datasets for outlier detection, following [14]. ... To evaluate the success of Dataset Transformation, we use the “DS Diff” Metric [14]. ... This demonstrates that the transformation successfully aligns the crossover dataset with the private dataset while preserving public dataset characteristics."

    The pipeline's only mechanism for transferring private outlier structure into the public model's input space is Eq. 4, whose architecture and success criterion are both imported from the authors' own prior work [14]. Eq. 5 checks only aggregate DS Diff similarity, not per-sample outlier separation, and no DS Diff values are reported; §5.1 concedes ROC-AUC loss to 'inherent noise introduced during dataset transformation.' Thus the claim that transformation preserves outlier structure -- and hence that soft labels are more accurate -- is justified by a self-citation rather than by an independent validation or derivation in this paper.

full rationale

The evaluation itself is not circular: soft labels are compared against external ADBench ground-truth labels, and no private-label information is used to fit the transformation or select public models, so the headline numbers in Table 1 are independent evidence that could have refuted the method. The circularity burden is concentrated in the Dataset Transformation component: its architecture and its DS Diff success criterion are taken verbatim from the authors' prior [14], and the paper does not define DS Diff, report its values, or test whether per-sample outlier rankings survive crossover. The paper's own ROC-AUC shortfall is attributed to transformation noise, which is exactly the unvalidated failure mode. Because the central empirical claim still has independent content, the score is 3 rather than higher.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central contribution is a pipeline built on three load-bearing assumptions: the PCA similarity heuristic selects transferable datasets, the autoencoder transformation preserves outlier structure (inherited from self-cited [14]), and majority voting over public-model outputs improves label quality. The paper provides no formal guarantee or external validation for any of these, and no code is released.

free parameters (4)
  • m (number of best public models in Method1) = not reported
    Method1 applies the best m optimized models from the selected public dataset and fuses via majority voting. The value of m is not specified in the paper, and soft-label accuracy depends on it (Figure 5 varies Top1_OD through Top9_OD).
  • n (number of top public datasets in Method2) = not reported
    Method2 transforms private data into the top n similar public datasets. n is never fixed in the experimental section; Figure 5 shows variants up to Top9, but the main Table 1 results do not state which n was used.
  • PCA component range for similarity = 1 to 100 PCs
    The Dataset Similarity Measure computes mean reconstruction errors from 1 to 100 principal components. This range is chosen by hand; datasets with fewer than 100 features are capped implicitly, which is not discussed.
  • Dataset Transformation training hyperparameters = Adam lr=0.001, beta1=0.9, beta2=0.999, 1000 epochs, LeakyReLU, architecture [128,64]/[64,32,64]/[64,128]
    These are chosen by hand and affect transformation quality. The paper does not study sensitivity to these choices or report random seeds.
assumptions (5)
  • domain assumption If a public dataset has high PCA reconstruction-error similarity to a private dataset, a model that works on the public dataset is likely to work on the private dataset.
    Stated in Section 1 and Section 5.3 (Figure 8); this assumption underlies the whole similarity-based selection. The paper provides a correlation trend but not a causal justification.
  • domain assumption The autoencoder-based Dataset Transformation (Equation 4) maps private samples into public format while preserving the information needed to separate outliers from inliers.
    Borrowed from self-cited [14]; checked only with the DS Diff aggregate distribution metric (Equation 5), which does not verify label-level transfer. This is the load-bearing premise.
  • domain assumption Reconstruction error curves from PCA, computed on arbitrarily scaled features, are comparable across datasets with different dimensions and feature types.
    Section 3.1; no feature standardization or normalization procedure is described, so the SAD measure may be sensitive to scaling choices.
  • domain assumption Majority voting over predictions from several public models improves soft-label quality.
    Used in both Method1 and Method2 without theoretical or empirical analysis of alternative fusion strategies.
  • domain assumption ADBench leave-one-out simulation is representative of a real private-data deployment where the private dataset is not in the public repository.
    Section 5.1. All 50 datasets come from the same benchmark family, which may overstate how well similarity-based selection works in genuinely heterogeneous private deployments.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Tabular Data Adapters: Improving Outlier Detection for Unlabeled Private Data." pith.science (2026). https://pith.science/paper/BX7RDACQ

@misc{pith2026250420862,
  author       = {Pith},
  title        = {Pith review of: Tabular Data Adapters: Improving Outlier Detection for Unlabeled Private Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BX7RDACQ}},
  note         = {Machine review of arXiv:2504.20862}
}
read the original abstract

The remarkable success of Deep Learning approaches is often based and demonstrated on large public datasets. However, when applying such approaches to internal, private datasets, one frequently faces challenges arising from structural differences in the datasets, domain shift, and the lack of labels. In this work, we introduce Tabular Data Adapters (TDA), a novel method for generating soft labels for unlabeled tabular data in outlier detection tasks. By identifying statistically similar public datasets and transforming private data (based on a shared autoencoder) into a format compatible with state-of-the-art public models, our approach enables the generation of weak labels. It thereby can help to mitigate the cold start problem of labeling by basing on existing outlier detection models for public datasets. In experiments on 50 tabular datasets across different domains, we demonstrate that our method is able to provide more accurate annotations than baseline approaches while reducing computational time. Our approach offers a scalable, efficient, and cost-effective solution, to bridge the gap between public research models and real-world industrial applications.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 19 canonical work pages

  1. [1]

    Abdi and L

    H. Abdi and L. J. Williams. Principal component analysis. Wiley inter- disciplinary reviews: computational statistics, 2(4):433–459, 2010

    work page 2010

  2. [2]

    Ahmed, A

    M. Ahmed, A. N. Mahmood, and M. R. Islam. A survey of anomaly de- tection techniques in financial domain. Future Gener. Comput. Syst., 55 (C):278–288, feb 2016. ISSN 0167-739X. doi: 10.1016/j.future.2015. 01.001. URL https://doi.org/10.1016/j.future.2015.01.001

    work page doi:10.1016/j.future.2015 2016

  3. [3]

    Becker and R

    B. Becker and R. Kohavi. Adult. UCI Machine Learning Repository,

  4. [4]

    Borisov, T

    V . Borisov, T. Leemann, K. Sessler, J. Haug, M. Pawelczyk, and G. Kas- neci. Deep neural networks and tabular data: A survey. IEEE Trans- actions on Neural Networks and Learning Systems , page 1–21, 2022. ISSN 2162-2388. doi: 10.1109/tnnls.2022.3229161. URL http://dx.doi. org/10.1109/TNNLS.2022.3229161

    arXiv 2022

  5. [5]

    L. Cao. Ai in finance: A review. SSRN Electronic Journal, 01 2020. doi: 10.2139/ssrn.3647625

    work page doi:10.2139/ssrn.3647625 2020

  6. [6]

    Chalapathy and S

    R. Chalapathy and S. Chawla. Deep learning for anomaly detection: A survey, 2019

    work page 2019

  7. [7]

    Dua and C

    D. Dua and C. Graff. UCI machine learning repository, 2017. URL http://archive.ics.uci.edu/ml

    work page 2017

  8. [8]

    Dwork and A

    C. Dwork and A. Roth. The algorithmic foundations of differential privacy. Foundations and Trends in Theoretical Computer Science, 9(3- 4):211–407, 2014. URL http://dblp.uni-trier.de/db/journals/fttcs/fttcs9. html#DworkR14

    work page 2014

Show all 39 references
  1. [9]

    A. Fang, S. Kornblith, and L. Schmidt. Does progress on imagenet transfer to real-world datasets?, 2023. URL https://arxiv.org/abs/2301. 04644

    2023

  2. [10]

    Goldstein and S

    M. Goldstein and S. Uchida. A comparative evaluation of unsupervised anomaly detection algorithms for multivariate data. PLOS ONE, 11(4): 1–31, 04 2016. doi: 10.1371/journal.pone.0152173

    2016 doi

  3. [11]

    S. Han, X. Hu, H. Huang, M. Jiang, and Y . Zhao. Adbench: Anomaly detection benchmark. Advances in Neural Information Processing Sys- tems, 35:32142–32159, 2022

    2022

  4. [12]

    J. A. Hartigan and M. A. Wong. A k-means clustering algorithm. JS- TOR: Applied Statistics, 28(1):100–108, 1979

    1979

  5. [13]

    Herurkar, M

    D. Herurkar, M. Meier, and J. Hees. Recol: Reconstruction er- ror columns for outlier detection. In KI 2023: Advances in Ar- tificial Intelligence: 46th German Conference on AI, Berlin, Ger- many, September 26–29, 2023, Proceedings , page 60–74, Berlin, Heidelberg, 2023. Sprin...

    2023 doi

  6. [14]

    Herurkar, T

    D. Herurkar, T. Sattarov, J. Hees, S. Palacio, F. Raue, and A. Dengel. Cross-domain transformation for outlier detection on tabular datasets. In International Joint Conference on Neural Networks, IJCNN 2023, Gold Coast, Australia, June 18-23, 2023, pages 1–8. IEEE, 2023. doi: ...

    2023

  7. [15]

    Herurkar, S

    D. Herurkar, S. Palacio, A. Anwar, J. Hees, and A. Dengel. Fin-fed- od: Federated outlier detection on financial tabular data, 2024. URL https://arxiv.org/abs/2404.14933

    2024 arXiv

  8. [16]

    Herurkar, F

    D. Herurkar, F. Raue, and A. Dengel. Tab-distillation: Impacts of dataset distillation on tabular data for outlier detection. In Proceedings of the 5th ACM International Conference on AI in Finance , ICAIF ’24, page 804–812, New York, NY , USA, 2024. Association for Computing ...

    2024

  9. [17]

    Hilal, S

    W. Hilal, S. A. Gadsden, and J. Yawney. Financial fraud: A review of anomaly detection techniques and recent advances. Expert Systems with Applications, 193:116429, 2022. ISSN 0957-4174. doi: https:// doi.org/10.1016/j.eswa.2021.116429. URL https://www.sciencedirect. com/scien...

    2022

  10. [18]

    Kazemi and H

    Z. Kazemi and H. Zarrabi. Using deep networks for fraud detection in the credit card transactions. In 2017 IEEE 4th International Conference on Knowledge-Based Engineering and Innovation (KBEI), pages 0630– 0633, 2017. doi: 10.1109/KBEI.2017.8324876

    2017

  11. [19]

    D. P. Kingma and J. Ba. Adam: A method for stochastic optimization,

  12. [20]

    Narayanan and V

    A. Narayanan and V . Shmatikov. Robust de-anonymization of large sparse datasets. In Proc. of the 29th IEEE Symposium on Security and Privacy, pages 111–125. IEEE Computer Society, May 2008. doi: 10.1109/SP.2008.33. URL http://www.cs.utexas.edu/~shmat/shmat_ oak08netflix.pdf

    2008 doi

  13. [21]

    Narayanan and V

    A. Narayanan and V . Shmatikov. Robust de-anonymization of large sparse datasets : a decade later. 2019

    2019

  14. [22]

    Nicholls, A

    J. Nicholls, A. Kuppa, and N.-A. Le-Khac. Financial cybercrime: A comprehensive survey of deep learning approaches to tackle the evolv- ing financial crime landscape. IEEE Access, 9:163965–163986, 2021. doi: 10.1109/ACCESS.2021.3134076

    2021

  15. [23]

    A. M. Ozbayoglu, M. U. Gudelek, and O. B. Sezer. Deep learning for financial applications : A survey. Applied Soft Computing , 93: 106384, 2020. ISSN 1568-4946. doi: https://doi.org/10.1016/j.asoc. 2020.106384. URL https://www.sciencedirect.com/science/article/pii/ S1568494620303240

    2020

  16. [24]

    Paszke, S

    A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga, A. Desmaison, A. Kopf, E. Yang, Z. DeVito, M. Raison, A. Tejani, S. Chilamkurthy, B. Steiner, L. Fang, J. Bai, and S. Chintala. Pytorch: An imperative style, high-per...

    2019

  17. [25]

    E. L. Paula, M. Ladeira, R. N. Carvalho, and T. Marzagão. Deep learn- ing anomaly detection as support fraud investigation in brazilian exports and anti-money laundering. In 2016 15th IEEE International Confer- ence on Machine Learning and Applications (ICMLA), pages 954–960,

    2016

  18. [26]

    Pei and O

    Y . Pei and O. R. Zaiane. A synthetic data generator for clustering and outlier analysis. 01 2006

    2006

  19. [27]

    Pumsirirat and L

    A. Pumsirirat and L. Yan. Credit card fraud detection using deep learn- ing based on auto-encoder and restricted boltzmann machine. Interna- tional Journal of Advanced Computer Science and Applications , 9(1),

  20. [28]

    Ramirez Rivera, A

    A. Ramirez Rivera, A. Khan, I. E. I. Bekkouch, and T. S. Sheikh. Anomaly detection based on zero-shot outlier synthesis and hierarchi- cal feature distillation. IEEE Transactions on Neural Networks and Learning Systems, 33(1):281–291, Jan 2022. ISSN 2162-2388. doi: 10.1109/tnn...

    2022

  21. [29]

    L. Ruff, R. Vandermeulen, N. Goernitz, L. Deecke, S. A. Siddiqui, A. Binder, E. Müller, and M. Kloft. Deep one-class classification. In J. Dy and A. Krause, editors,Proceedings of the 35th International Con- ference on Machine Learning , volume 80 of Proceedings of Machine Lea...

    2018

  22. [30]

    Sattarov, D

    T. Sattarov, D. Herurkar, and J. Hees. Explaining anomalies using de- noising autoencoders for financial tabular data. CoRR, abs/2209.10658,

    arXiv

  23. [31]

    Schreyer, T

    M. Schreyer, T. Sattarov, D. Borth, A. Dengel, and B. Reimer. Detec- tion of anomalies in large scale accounting data using deep autoencoder networks, 2017. URL https://arxiv.org/abs/1709.05254

    2017 arXiv

  24. [32]

    false positives

    R. Wedge, J. M. Kanter, S. M. Rubio, S. I. Perez, and K. Veeramacha- neni. Solving the “false positives” problem in fraud prediction, 2017

    2017

  25. [33]

    I.-C. Yeh. Default of Credit Card Clients. UCI Machine Learning Repository, 2016. DOI: https://doi.org/10.24432/C55S3H

    2016 doi

  26. [34]

    B. Zong, Q. Song, M. R. Min, W. Cheng, C. Lumezanu, D. ki Cho, and H. Chen. Deep autoencoding gaussian mixture model for unsu- pervised anomaly detection. In International Conference on Learning Representations, 2018. URL https://api.semanticscholar.org/CorpusID: 51805340. App...

    2018

  27. [1996]

    DOI: https://doi.org/10.24432/C5XW20

    doi

  28. [2014]

    URL https://arxiv.org/abs/1412.6980

    arXiv

  29. [2016]

    doi: 10.1109/ICMLA.2016.0172

    2016

  30. [2018]

    URL http://dx.doi.org/10

    doi: 10.14569/IJACSA.2018.090103. URL http://dx.doi.org/10. 14569/IJACSA.2018.090103

    2018

  31. [2022]

    URL https://doi.org/10

    doi: 10.48550/ARXIV .2209.10658. URL https://doi.org/10. 48550/arXiv.2209.10658

    doi arXiv

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.