Pith. sign in

REVIEW 4 major objections 6 minor 53 references

Deep evolving semi-supervised anomaly detection

T0 review · 4 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read The paper formalises continual semi-supervised anomaly detection and shows that a variational autoencoder with outlier-rejected generative replay is a strong baseline, reaching the highest AUC-ROC on MNIST and Fashion MNIST among the…

desk verdict A genuinely new problem formalization is undermined by an evaluation that tunes the anomaly threshold on the test set; the paper is worth refereeing but the empirical claims need a redo. read the letter →

arxiv 2412.00860 v1 pith:7ZHR6YZP submitted 2024-12-01 cs.LG cs.AIstat.ML

Jack Belham , Aryan Bhosale , Samrat Mukherjee , Biplab Banerjee , Fabio Cuzzolin This is my paper · ORCID
classification cs.LGcs.AIstat.ML
keywords continualsemi-supervisedanomalydetectionvariationalautoencodergenerativereplayoutlierrejectionextremevaluetheorycatastrophicforgettingAUC-ROCdomain-incrementallearning
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 formalises continual semi-supervised anomaly detection (CSAD): a learner sees a stream of disjoint datasets, each with a few labelled normal examples and many unlabelled ones, and must flag anomalies without revisiting past data. It argues that this combination matches real-world monitoring better than anomaly detection studied in isolation, because anomalies are often unlabelled, new anomaly types appear over time, and storing all historical data is impractical. To make the task testable, it contributes a baseline: a variational autoencoder trained semi-supervised with a classifier, plus deep generative replay that samples only latent points judged to be inliers of the normal class. The claim is that this outlier-rejection replay achieves the best AUC-ROC among the compared strategies on MNIST (0.690) and Fashion MNIST (0.581), and remains competitive on CIFAR-10 (0.546).

What carries the argument

The load-bearing mechanism is a semi-supervised variational autoencoder (the M2 model with a beta-weighted KL term) whose classifier regularises the latent space into clustered normals, combined with generative replay filtered by extreme value theory. For replay, the latent distance of each generated sample from the normal-class mean is modelled by a per-class Weibull distribution, and samples whose outlier probability exceeds a validation-set-tuned threshold are rejected before decoding. Anomaly scores come from the ELBO reconstruction term, and evaluation is reported as AUC-ROC.

What would settle it

Recompute Tables 2 through 5 by choosing the threshold once on a held-out validation stream and then applying it to the test stream; if outlier rejection no longer attains the highest AUC on MNIST and Fashion MNIST, the paper's main empirical claim fails. A second check is to rerun joint training with the same encoder-decoder: if it still trails naive training on Fashion MNIST, the upper-bound assumption behind the comparison is invalid.

Watch

Extended reading notes

Core claim

The central discovery is that continual semi-supervised anomaly detection can be formalised as a distinct problem and that a comparatively simple VAE-based learner with outlier-rejected replay is a workable baseline for it. In the paper's experiments, the outlier-rejection method outperforms the other continual-learning strategies on most datasets, including the highest AUC-ROC on MNIST and Fashion MNIST. The ablation studies show that performance rises with more labelled normal data, that labelled anomalies help up to a point, and that unlabelled anomalies initially help then hurt. The paper also reports that its assumed upper bound, joint training, is not always the best, notably on Fashion MNIST where joint training scores 0.401 while naive training reaches 0.555.

Load-bearing premise

The reported AUC values are obtained by selecting the anomaly-detection threshold on the test set itself, so the comparison measures each method's best possible operating point rather than its predictive quality on unseen data.

Editorial extensions

If this is right

  • CSAD is usable as a labelled research problem with a reproducible benchmark setup, since the five data-splitting parameters define streams for MNIST, Fashion MNIST, and CIFAR-10.
  • A practitioner with mostly unlabelled streaming data can deploy the VAE baseline as a first continual anomaly detector without storing past data.
  • Because outlier rejection beats EWC on MNIST and Fashion MNIST, generative replay with inlier filtering is a stronger default than weight regularisation in this setting.
  • Performance is sensitive to the placement of labelled data: skewing labels toward later experiences reduces the need for forgetting mitigation and changes method rankings.

Reading between the lines

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

  • The reported AUCs should be read as an upper envelope, since the threshold is chosen on the test set itself; a held-out threshold selection would give a fairer comparison and might reorder the methods.
  • The outlier-rejection idea transfers to other generative replay models: any generator whose latent space has a well-estimated normal-class mean could use the same Weibull filter.
  • A natural next experiment, not run here, is to use the outlier-rejection probability itself as the anomaly score instead of the ELBO, which the paper notes is currently infeasible with its label-conditioned encoder.
  • If joint training is not actually an upper bound in this semi-supervised continual setting, the standard upper-bound comparison used in continual-learning evaluation needs rethinking.
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. This paper formalizes the task of continual semi-supervised anomaly detection (CSAD), in which a model receives a stream of experiences each containing a mixture of labelled and unlabelled data, and must detect anomalies while mitigating catastrophic forgetting. The authors propose a variational autoencoder (VAE) baseline that combines semi-supervised training with generative replay and an outlier-rejection mechanism based on extreme value theory. They evaluate this method (OR) against naive training, joint training, and Elastic Weight Consolidation on MNIST, Fashion-MNIST, and CIFAR-10, and report per-episode AUC-ROC scores and ablation studies. The main empirical claim is that OR outperforms other continual learning strategies on most datasets.

Significance. If the results held, the paper would provide a useful new problem formulation (CSAD) and a simple baseline for future work. The dataset configurations and ablation experiments are described in enough detail to be reproduced, which is a practical contribution. However, the current experimental evidence is undermined by the use of the test set to select the anomaly detection threshold, the failure of the joint-training upper bound on Fashion-MNIST, and the absence of error bars. The conceptual contribution is therefore not yet matched by a reliable empirical demonstration.

major comments (4)
  1. [Section 3.3] The evaluation protocol leaks test information: the text states that the anomaly detection metric 'will use the AUC calculated over the test set to find the optimal threshold for these particular metrics which maximises the AUC.' This tunes the threshold on the same test set used to report the AUC values in Table 2 and the per-episode tables. Consequently, the reported AUCs are fitted quantities rather than unbiased estimates. The threshold should be selected on the validation stream defined in Section 4.2 (or via a nested cross-validation on the training stream) and then evaluated on the test stream. This is load-bearing because the central claim that OR 'outperforms other continual learning strategies across most datasets' rests on these numbers.
  2. [Section 4.4 / Section 5] The paper treats joint training as the upper bound, but the results contradict this: on Fashion-MNIST, Joint scores 0.401, well below Naive (0.555) and OR (0.581). The paper itself admits in Section 5 that 'There were issues in joint training which meant that it was not in fact the upper bound for all experiments.' The authors need to diagnose and repair the joint-training baseline, or explicitly remove the upper-bound claim and reinterpret the comparisons accordingly.
  3. [Section 4.4 / Tables 3-5] No multiple seeds or error bars are reported; all AUC values are from single runs. The per-episode scores show large swings (e.g., MNIST OR ranges from 0.60 to 0.84 in Table 3), so the aggregate differences in Table 2 (0.690 vs 0.655 for MNIST) may not be statistically significant. The paper should report mean and standard deviation over several seeds and, ideally, a paired significance test for the method comparisons.
  4. [Section 5] The CIFAR-10 experiments use a dense encoder/decoder, and the paper acknowledges that 'simple dense layers are unlikely to be able to correctly reconstruct CIFAR-10 images.' As a result, the CIFAR-10 results reflect an architecture mismatch rather than the properties of the CSAD methods. A convolutional VAE (or an explicit statement that CIFAR-10 is a degenerate stress test) is needed before any cross-dataset conclusions can be drawn.
minor comments (6)
  1. [Section 3.2] Equation (8) appears to be missing parentheses in the Weibull CDF; it should read 1 - exp(-(|d|/lambda)^kappa) or similar, and the notation z-bar versus z should be clarified.
  2. [Section 3.3] The sentence 'which will use the AUC calculated over the test set to find the optimal threshold for these particular metrics which maximises the AUC' is grammatically ambiguous; please rewrite it to state exactly which score is used and how thresholds are chosen.
  3. [Throughout] There are numerous typographical and grammatical errors ('reproducability', 'discriminte', 'the the goal', 'hyperparamter'); the manuscript needs careful proofreading.
  4. [Section 4.4 / Table 2] The table caption describes bold and underline formatting for best and second-best results, but the formatting is not visible in the text version; please ensure the final PDF displays it.
  5. [Section 2.4] The two definitions of CSSL discussed in [7] and [35] could be presented in a table to make the adopted setting clearer.
  6. [Section 5] The narrative in the ablation analysis refers to 'Figure 2', 'Figure 9', etc., but the ordering of figures in the text is repetitive; consider renumbering or summarizing the figures to improve readability.

Circularity Check

1 steps flagged · score 6.0 of 10

Test-set threshold selection makes reported AUCs fitted; the central comparison is not supported by a clean evaluation.

  1. fitted input called prediction [Section 3.3, Approach to Anomaly Detection]
    "Another natural choice for anomaly detection would be the outlier rejection probability [29]. This could be used in a similar way to the reconstruction probability or ELBO which will use the AUC calculated over the test set to find the optimal threshold for these particular metrics which maximises the AUC."

    The reported AUC-ROC scores in Table 2 are presented as the performance of each continual-learning strategy, yet the anomaly-detection threshold is chosen by directly optimizing on the test set: the method 'will use the AUC calculated over the test set to find the optimal threshold ... which maximises the AUC.' The same test set is then used to compute the reported AUC, so the numbers are fitted maxima over thresholds rather than unbiased estimates on unseen data. Because the test set serves as both the tuning set and the evaluation set, the headline claim that Outlier Rejection 'outperforms other continual learning strategies across most datasets' (Section 4.4) is not supported by a clean protocol.

full rationale

The paper's central empirical contribution is the comparison in Table 2, where Outlier Rejection (OR) is claimed to outperform EWC, Joint, and Naive on MNIST and Fashion MNIST. The evaluation protocol, however, tunes the anomaly-detection threshold on the test set: Section 3.3 states that the method 'will use the AUC calculated over the test set to find the optimal threshold for these particular metrics which maximises the AUC.' Since the same test set is then used to compute the reported AUC, the numbers in Table 2 are fitted maxima rather than unbiased estimates on unseen data. This does not make the formalization of CSAD itself circular, but it means the headline superiority claim rests on a test-set-tuned quantity. The paper's own Section 5 admits that joint training was not a reliable upper bound, and per-episode AUCs in Tables 3-5 vary widely (e.g., MNIST OR from 0.59 to 0.84), so the ranking could change under a clean validation-based threshold selection. Minor self-citations ([7], [51]) are not load-bearing and do not contribute to circularity. Overall circularity is partial, focused on the evaluation protocol rather than on the problem definition.

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

The paper introduces no new theoretical entities. Its contributions are a problem formalization and an empirical combination of existing methods. The central free parameters are the Weibull fit and the anomaly threshold, with the latter fitted on the test set, which is a significant circularity concern.

free parameters (4)
  • Weibull distribution parameters (tau, kappa, lambda) for normal class latent distances = not reported
    Used in Equation 8 to compute the outlier probability of generated samples; fitted to the distances of correctly identified normal instances in each experience (Section 3.2).
  • Anomaly detection threshold on ELBO = maximizes AUC on the test set
    The threshold for classifying a sample as anomalous is chosen by scanning the test set to maximize AUC (Section 3.3), which makes the reported AUC a fitted value.
  • Beta hyperparameter for KL term in VAE loss = not specified
    Introduced in Equation 6 as a beta-VAE modification; no value or tuning procedure is reported (Section 3.1.1).
  • Outlier rejection probability threshold = not specified
    Used to decide which generated samples to reject in replay; determined using a validation set (Section 3.2).
assumptions (4)
  • standard math Variational inference and the ELBO derivation for the M2 VAE are valid and standard.
    The loss functions in Equations 1-5 follow Kingma et al. [19] and are used without independent verification.
  • domain assumption The Weibull distribution accurately models the distribution of distances from latent representations to the per-class mean.
    Equation 8 treats the distance variable as heavy-tailed Weibull, following Mundt et al. [29]; no goodness-of-fit is reported.
  • domain assumption ELBO (or reconstruction probability) is a reliable anomaly score.
    The paper uses ELBO as the anomaly detection metric, citing An and Cho [44] and Wiewel and Yang [11], but does not validate it in this setting.
  • domain assumption The cluster and manifold assumptions of semi-supervised learning hold for the benchmark datasets.
    Used to justify the semi-supervised VAE setup in Section 2.1, following Ouali et al. [17].

how reviews work

0 comments
Cite this review

Pith. "Pith review of Deep evolving semi-supervised anomaly detection." pith.science (2026). https://pith.science/paper/7ZHR6YZP

@misc{pith2026241200860,
  author       = {Pith},
  title        = {Pith review of: Deep evolving semi-supervised anomaly detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7ZHR6YZP}},
  note         = {Machine review of arXiv:2412.00860}
}
read the original abstract

The aim of this paper is to formalise the task of continual semi-supervised anomaly detection (CSAD), with the aim of highlighting the importance of such a problem formulation which assumes as close to real-world conditions as possible. After an overview of the relevant definitions of continual semi-supervised learning, its components, anomaly detection extension, and the training protocols; the paper introduces a baseline model of a variational autoencoder (VAE) to work with semi-supervised data along with a continual learning method of deep generative replay with outlier rejection. The results show that such a use of extreme value theory (EVT) applied to anomaly detection can provide promising results even in comparison to an upper baseline of joint training. The results explore the effects of how much labelled and unlabelled data is present, of which class, and where it is located in the data stream. Outlier rejection shows promising initial results where it often surpasses a baseline method of Elastic Weight Consolidation (EWC). A baseline for CSAD is put forward along with the specific dataset setups used for reproducability and testability for other practitioners. Future research directions include other CSAD settings and further research into efficient continual hyperparameter tuning.

Figures

Figures reproduced from arXiv: 2412.00860 by the authors.

Figure 1
Figure 1. Working of CSSAD-VAE. training and no access to past data is allowed. The aim is to mitigate catastrophic forgetting in a set-up similar to continual semi-supervised domain incremental learning [7]. Whilst domain-incremental continual learning, semi-supervised learning and anomaly detection have all been researched as separate research areas [7],[36],[4], no paper has yet defined the problem of continual semi-superv… view at source ↗
Figure 2
Figure 2. AUC Scores for Different Methods on MNIST, CIFAR10, and Fashion MNIST [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗
Figure 3
Figure 3. AUC Scores for Different Methods on MNIST, CIFAR10, and Fashion MNIST [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗
Figures from the paper (13 more)
Figure 4
Figure 4. Figure 4: AUC Scores for Different Methods on MNIST, CIFAR10, and Fashion MNIST [PITH_FULL_IMAGE:figures/full_fig_p012_4.png]
Figure 5
Figure 5. Figure 5: AUC Scores for Different Methods on MNIST, CIFAR10, and Fashion MNIST [PITH_FULL_IMAGE:figures/full_fig_p013_5.png]
Figure 6
Figure 6. Figure 6: AUC Scores for Different Methods on MNIST, CIFAR10, and Fashion MNIST [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]
Figure 7
Figure 7. Figure 7: AUC Scores for Different Methods on MNIST, CIFAR10, and Fashion MNIST [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]
Figure 8
Figure 8. Figure 8: AUC Scores for Different Methods on MNIST, CIFAR10, and Fashion MNIST [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]
Figure 9
Figure 9. Figure 9: AUC Scores (Exp 8) MNIST CIFAR10 Fashion MNIST 0.4 0.5 0.6 0.7 0.8 AUC Score Naive Joint Training EWC Outlier Rejection (a) AUC Scores (Exp 9) [PITH_FULL_IMAGE:figures/full_fig_p015_9.png]
Figure 10
Figure 10. Figure 10: AUC Scores for Different Methods on MNIST, CIFAR10, and Fashion MNIST [PITH_FULL_IMAGE:figures/full_fig_p015_10.png]
Figure 11
Figure 11. Figure 11: AUC Scores for Different Methods on MNIST, CIFAR10, and Fashion MNIST [PITH_FULL_IMAGE:figures/full_fig_p016_11.png]
Figure 12
Figure 12. Figure 12: AUC Scores for Different Methods on MNIST, CIFAR10, and Fashion MNIST [PITH_FULL_IMAGE:figures/full_fig_p016_12.png]
Figure 13
Figure 13. Figure 13: AUC Scores for Different Methods on MNIST, CIFAR10, and Fashion MNIST [PITH_FULL_IMAGE:figures/full_fig_p017_13.png]
Figure 14
Figure 14. Figure 14: AUC Scores for Different Methods on MNIST, CIFAR10, and Fashion MNIST [PITH_FULL_IMAGE:figures/full_fig_p017_14.png]
Figure 15
Figure 15. Figure 15: AUC Scores for Different Methods on MNIST, CIFAR10, and Fashion MNIST [PITH_FULL_IMAGE:figures/full_fig_p018_15.png]
Figure 16
Figure 16. Figure 16: AUC Scores for Different Methods on MNIST, CIFAR10, and Fashion MNIST [PITH_FULL_IMAGE:figures/full_fig_p018_16.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 29 canonical work pages

  1. [1]

    Anomaly detection: A survey

    Varun Chandola, Arindam Banerjee, and Vipin Kumar. Anomaly detection: A survey. ACM computing surveys (CSUR), 41(3):1–58, 2009. 16 MNIST CIFAR10 Fashion MNIST 0.4 0.5 0.6 0.7 0.8 AUC Score Naive Joint Training EWC Outlier Rejection (a) AUC Scores (Exp 12) Figure 13: AUC Scores for Different Methods on MNIST, CIFAR10, and Fashion MNIST MNIST CIFAR10 Fashio...

    work page 2009

  2. [2]

    Robust one-class svm for fault detection

    Yingchao Xiao, Huangang Wang, Wenli Xu, and Junwu Zhou. Robust one-class svm for fault detection. Chemometrics and Intelligent Laboratory Systems, 151:15–25, 2016

    work page 2016

  3. [3]

    One-class classification for heart disease diagnosis

    George Gomes Cabral and Adriano Lorena Inácio de Oliveira. One-class classification for heart disease diagnosis. In 2014 IEEE International conference on systems, man, and cybernetics (SMC), pages 2551–2556. IEEE, 2014

    work page 2014

  4. [4]

    Lifelong machine learning: a paradigm for continuous learning.Frontiers of Computer Science, 11(3):359–361, 2017

    Bing Liu. Lifelong machine learning: a paradigm for continuous learning.Frontiers of Computer Science, 11(3):359–361, 2017

    work page 2017

  5. [5]

    Deep semi-supervised anomaly detection

    Lukas Ruff, Robert A Vandermeulen, Nico Görnitz, Alexander Binder, Emmanuel Müller, Klaus-Robert Müller, and Marius Kloft. Deep semi-supervised anomaly detection. arXiv preprint arXiv:1906.02694, 2019

    arXiv 1906

  6. [6]

    Semi-supervised learning (chapelle, o

    Olivier Chapelle, Bernhard Scholkopf, and Alexander Zien. Semi-supervised learning (chapelle, o. et al., eds.; 2006)[book reviews]. IEEE Transactions on Neural Networks, 20(3):542–542, 2009. 17 MNIST CIFAR10 Fashion MNIST 0.4 0.5 0.6 0.7 0.8 AUC Score Naive Joint Training EWC Outlier Rejection (a) AUC Scores (Exp 14) Figure 15: AUC Scores for Different Me...

    work page 2006

  7. [7]

    International workshop on continual semi-supervised learning: Introduction, benchmarks and baselines

    Ajmal Shahbaz, Salman Khan, Mohammad Asiful Hossain, Vincenzo Lomonaco, Kevin Can- nons, Zhan Xu, and Fabio Cuzzolin. International workshop on continual semi-supervised learning: Introduction, benchmarks and baselines. arXiv preprint arXiv:2110.14613, 2021

    arXiv

  8. [8]

    Learning on the job: Online lifelong and continual learning

    Bing Liu. Learning on the job: Online lifelong and continual learning. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 34, pages 13544–13549, 2020

    work page 2020

Show all 53 references
  1. [9]

    Catastrophic forgetting in connectionist networks

    Robert M French. Catastrophic forgetting in connectionist networks. Trends in cognitive sciences, 3(4):128–135, 1999

    1999

  2. [10]

    Overcoming catastrophic forgetting in neural networks

    James Kirkpatrick, Razvan Pascanu, Neil Rabinowitz, Joel Veness, Guillaume Desjardins, Andrei A Rusu, Kieran Milan, John Quan, Tiago Ramalho, Agnieszka Grabska-Barwinska, et al. Overcoming catastrophic forgetting in neural networks. Proceedings of the national academy of scien...

    2017

  3. [11]

    Continual learning for anomaly detection with variational autoen- coder

    Felix Wiewel and Bin Yang. Continual learning for anomaly detection with variational autoen- coder. In ICASSP 2019-2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 3837–3841. IEEE, 2019. 18

    2019

  4. [12]

    Beyond supervised continual learning: a review

    Benedikt Bagus, Alexander Gepperth, and Timothée Lesort. Beyond supervised continual learning: a review. arXiv preprint arXiv:2208.14307, 2022

    2022 arXiv

  5. [13]

    Avalanche: an end-to-end library for continual learning

    Vincenzo Lomonaco, Lorenzo Pellegrini, Andrea Cossu, Antonio Carta, Gabriele Graffieti, Tyler L Hayes, Matthias De Lange, Marc Masana, Jary Pomponi, Gido M Van de Ven, et al. Avalanche: an end-to-end library for continual learning. In Proceedings of the IEEE/CVF Conference on ...

    2021

  6. [14]

    Continuum: Simple management of complex continual learning scenarios

    Arthur Douillard and Timothée Lesort. Continuum: Simple management of complex continual learning scenarios. arXiv preprint arXiv:2102.06253, 2021

    2021 arXiv

  7. [15]

    Sequoia: A software framework to unify continual learning research

    Fabrice Normandin, Florian Golemo, Oleksiy Ostapenko, Pau Rodriguez, Matthew D Riemer, Julio Hurtado, Khimya Khetarpal, Dominic Zhao, Ryan Lindeborg, Timothée Lesort, et al. Sequoia: A software framework to unify continual learning research. arXiv preprint arXiv:2108.01005, 2021

    2021 arXiv

  8. [16]

    The rise of ’big data’ on cloud computing: Review and open research issues

    Ibrahim Abaker Targio Hashem, Ibrar Yaqoob, Nor Badrul Anuar, Salimah Mokhtar, Abdullah Gani, and Samee Ullah Khan. The rise of ’big data’ on cloud computing: Review and open research issues. Information systems, 47:98–115, 2015

    2015

  9. [17]

    An overview of deep semi-supervised learning

    Yassine Ouali, Céline Hudelot, and Myriam Tami. An overview of deep semi-supervised learning. arXiv preprint arXiv:2006.05278, 2020

    2006 arXiv

  10. [18]

    Information theory and statistics

    Solomon Kullback. Information theory and statistics. Courier Corporation, 1997

    1997

  11. [19]

    An introduction to variational autoencoders

    Diederik P Kingma, Max Welling, et al. An introduction to variational autoencoders. Founda- tions and Trends® in Machine Learning, 12(4):307–392, 2019

    2019

  12. [20]

    Generative adversarial networks

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adversarial networks. Communications of the ACM, 63(11):139–144, 2020

    2020

  13. [21]

    Semi-supervised learning with generative adversarial networks.arXiv preprint arXiv:1606.01583, 2016

    Augustus Odena. Semi-supervised learning with generative adversarial networks.arXiv preprint arXiv:1606.01583, 2016

    2016 arXiv

  14. [22]

    Experi- ence replay for continual learning

    David Rolnick, Arun Ahuja, Jonathan Schwarz, Timothy Lillicrap, and Gregory Wayne. Experi- ence replay for continual learning. Advances in Neural Information Processing Systems, 32, 2019

    2019

  15. [23]

    Continual learning through one-class classifica- tion using vae

    Felix Wiewel, Andreas Brendle, and Bin Yang. Continual learning through one-class classifica- tion using vae. In ICASSP 2020-2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 3307–3311. IEEE, 2020

    2020

  16. [24]

    Boovae: Boosting approach for continual learning of vae

    Evgenii Egorov, Anna Kuzina, and Evgeny Burnaev. Boovae: Boosting approach for continual learning of vae. Advances in Neural Information Processing Systems, 34:17889–17901, 2021

    2021

  17. [25]

    Continual learning with deep generative replay

    Hanul Shin, Jung Kwon Lee, Jaehong Kim, and Jiwon Kim. Continual learning with deep generative replay. Advances in neural information processing systems, 30, 2017

    2017

  18. [26]

    Coresets for nonparametric estimation-the case of dp-means

    Olivier Bachem, Mario Lucic, and Andreas Krause. Coresets for nonparametric estimation-the case of dp-means. In International Conference on Machine Learning, pages 209–217. PMLR, 2015

    2015

  19. [27]

    Binplay: A binary latent autoencoder for generative replay continual learning

    Kamil Deja, Paweł Wawrzy ´nski, Daniel Marczak, Wojciech Masarczyk, and Tomasz Trz- ci´nski. Binplay: A binary latent autoencoder for generative replay continual learning. In 2021 International Joint Conference on Neural Networks (IJCNN), pages 1–8. IEEE, 2021

    2021

  20. [28]

    Metric learning for large scale image classification: Generalizing to new classes at near-zero cost

    Thomas Mensink, Jakob Verbeek, Florent Perronnin, and Gabriela Csurka. Metric learning for large scale image classification: Generalizing to new classes at near-zero cost. In European Conference on Computer Vision, pages 488–501. Springer, 2012

    2012

  21. [29]

    Unified probabilistic deep continual learning through generative replay and open set recognition

    Martin Mundt, Iuliia Pliushch, Sagnik Majumder, Yongwon Hong, and Visvanathan Ramesh. Unified probabilistic deep continual learning through generative replay and open set recognition. Journal of Imaging, 8(4):93, 2022. 19

    2022

  22. [30]

    Semi-supervised anomaly detection algorithms: A comparative summary and future research directions

    Miryam Elizabeth Villa-Pérez, Miguel A Alvarez-Carmona, Octavio Loyola-González, Miguel Angel Medina-Pérez, Juan Carlos Velazco-Rossell, and Kim-Kwang Raymond Choo. Semi-supervised anomaly detection algorithms: A comparative summary and future research directions. Knowledge-Ba...

    2021

  23. [31]

    Bagging-randomminer: A one-class classifier for file access-based masquerade detection.Machine Vision and Applications, 30(5):959–974, 2019

    José Benito Camiña, Miguel Angel Medina-Pérez, Raúl Monroy, Octavio Loyola-González, Luis Angel Pereyra Villanueva, and Luis Carlos González Gurrola. Bagging-randomminer: A one-class classifier for file access-based masquerade detection.Machine Vision and Applications, 30(5):9...

    2019

  24. [32]

    Towards anomaly detectors that learn continuously

    Andrea Stocco and Paolo Tonella. Towards anomaly detectors that learn continuously. In 2020 IEEE International Symposium on Software Reliability Engineering Workshops (ISSREW), pages 201–208. IEEE, 2020

    2020

  25. [33]

    Arcade: A rapid continual anomaly detector

    Ahmed Frikha, Denis Krompaß, and V olker Tresp. Arcade: A rapid continual anomaly detector. In 2020 25th International Conference on Pattern Recognition (ICPR), pages 10449–10456. IEEE, 2021

    2020

  26. [34]

    Continual learning for unsupervised anomaly detection in continuous auditing of financial accounting data

    Hamed Hemati, Marco Schreyer, and Damian Borth. Continual learning for unsupervised anomaly detection in continuous auditing of financial accounting data. arXiv preprint arXiv:2112.13215, 2021

    2021 arXiv

  27. [35]

    Ordisco: Effective and efficient usage of incremental unlabeled data for semi-supervised continual learning

    Liyuan Wang, Kuo Yang, Chongxuan Li, Lanqing Hong, Zhenguo Li, and Jun Zhu. Ordisco: Effective and efficient usage of incremental unlabeled data for semi-supervised continual learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages ...

    2021

  28. [36]

    Semi-supervised learning literature survey

    Xiaojin Jerry Zhu. Semi-supervised learning literature survey. 2005

    2005

  29. [37]

    All versus one: an empirical comparison on retrained and incremental machine learning for modeling performance of adaptable software

    Tao Chen. All versus one: an empirical comparison on retrained and incremental machine learning for modeling performance of adaptable software. In2019 IEEE/ACM 14th International Symposium on Software Engineering for Adaptive and Self-Managing Systems (SEAMS), pages 157–168. I...

    2019

  30. [38]

    Semi-supervised learning of bearing anomaly detection via deep variational autoencoders

    Shen Zhang, Fei Ye, Bingnan Wang, and Thomas G Habetler. Semi-supervised learning of bearing anomaly detection via deep variational autoencoders. arXiv preprint arXiv:1912.01096, 2019

    1912 arXiv

  31. [39]

    Variational inference: A review for statisticians

    David M Blei, Alp Kucukelbir, and Jon D McAuliffe. Variational inference: A review for statisticians. Journal of the American statistical Association, 112(518):859–877, 2017

    2017

  32. [40]

    Stochastic variational inference

    Matthew D Hoffman, David M Blei, Chong Wang, and John Paisley. Stochastic variational inference. Journal of Machine Learning Research, 2013

    2013

  33. [41]

    Tutorial: Deriving the standard variational autoencoder (vae) loss function

    Stephen Odaibo. Tutorial: Deriving the standard variational autoencoder (vae) loss function. arXiv preprint arXiv:1907.08956, 2019

    1907 arXiv

  34. [42]

    beta-vae: Learning basic visual concepts with a constrained variational framework

    Irina Higgins, Loic Matthey, Arka Pal, Christopher Burgess, Xavier Glorot, Matthew Botvinick, Shakir Mohamed, and Alexander Lerchner. beta-vae: Learning basic visual concepts with a constrained variational framework. 2016

    2016

  35. [43]

    Ladder variational autoencoders

    Casper Kaae Sønderby, Tapani Raiko, Lars Maaløe, Søren Kaae Sønderby, and Ole Winther. Ladder variational autoencoders. Advances in neural information processing systems, 29, 2016

    2016

  36. [44]

    Variational autoencoder based anomaly detection using recon- struction probability

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

    2015

  37. [45]

    The mnist database of handwritten digit images for machine learning research [best of the web]

    Li Deng. The mnist database of handwritten digit images for machine learning research [best of the web]. IEEE signal processing magazine, 29(6):141–142, 2012

    2012

  38. [46]

    Antonio Torralba, Rob Fergus, and William T. Freeman. 80 million tiny images: A large data set for nonparametric object and scene recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence, 30(11):1958–1970, 2008. doi: 10.1109/TPAMI.2008.128. 20

    1958 doi

  39. [47]

    Learning multiple layers of features from tiny images

    Alex Krizhevsky, Geoffrey Hinton, et al. Learning multiple layers of features from tiny images. 2009

    2009

  40. [48]

    Fashion-mnist: a novel image dataset for benchmarking machine learning algorithms

    Han Xiao, Kashif Rasul, and Roland V ollgraf. Fashion-mnist: a novel image dataset for benchmarking machine learning algorithms. arXiv preprint arXiv:1708.07747, 2017

    2017 arXiv

  41. [49]

    Augmix: A simple data processing method to improve robustness and uncertainty

    Dan Hendrycks, Norman Mu, Ekin D Cubuk, Barret Zoph, Justin Gilmer, and Balaji Lakshmi- narayanan. Augmix: A simple data processing method to improve robustness and uncertainty. arXiv preprint arXiv:1912.02781, 2019

    1912 arXiv

  42. [50]

    The use of the area under the roc curve in the evaluation of machine learning algorithms

    Andrew P Bradley. The use of the area under the roc curve in the evaluation of machine learning algorithms. Pattern recognition, 30(7):1145–1159, 1997

    1997

  43. [51]

    Svd-gan for real-time unsupervised video anomaly detection

    Dinesh Jackson Samuel and Fabio Cuzzolin. Svd-gan for real-time unsupervised video anomaly detection. 2021

    2021

  44. [52]

    Do deep convolutional nets really need to be deep and convolutional? arXiv preprint arXiv:1603.05691, 2016

    Gregor Urban, Krzysztof J Geras, Samira Ebrahimi Kahou, Ozlem Aslan, Shengjie Wang, Rich Caruana, Abdelrahman Mohamed, Matthai Philipose, and Matt Richardson. Do deep convolutional nets really need to be deep and convolutional? arXiv preprint arXiv:1603.05691, 2016

    2016 arXiv

  45. [53]

    Disentangling disentan- glement in variational autoencoders

    Emile Mathieu, Tom Rainforth, Nana Siddharth, and Yee Whye Teh. Disentangling disentan- glement in variational autoencoders. In International Conference on Machine Learning, pages 4402–4412. PMLR, 2019. 21

    2019

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.