Pith. sign in

REVIEW 3 major objections 7 minor 53 references

Uncertainty Quantification in Continual Open-World Learning

T0 review · 3 major / 7 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read This paper claims that per-class feature-reconstruction distances, combined in a ratio, yield uncertainty scores that stay reliable as a model continually encounters and learns new classes, removing the need for a novelty or labeling…

desk verdict A genuinely useful iterative uncertainty score for multi-novel-class continual learning, but the 'no degradation' claim is overreached and needs a direct test. read the letter →

arxiv 2412.16409 v1 pith:EQ3NLXA4 submitted 2024-12-21 cs.LG cs.AIcs.CV

Amanda S. Rios , Ibrahima J. Ndiour , Parual Datta , Jaroslaw Sydir , Omesh Tickoo , Nilesh Ahuja This is my paper · ORCID
classification cs.LGcs.AIcs.CV
keywords continuallearningopen-worlduncertaintyquantificationnoveltydetectionactivepseudo-labelingfeaturereconstructionerrorclass-incremental
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

This paper tackles a realistic version of continual learning in which a deployed model receives unlabeled data containing both unseen examples of already-known classes and samples from an unknown number of new classes. The authors propose COUQ, an iterative uncertainty-estimation method that scores each sample by the ratio of its distance to the nearest old-class feature subspace to its distance to the currently predicted new-class subspace. They argue that, unlike one-shot uncertainty estimates computed from the classifier's decision boundary, these scores do not degrade as new classes are added, and that they can drive novelty detection, active labeling, and pseudo-labeling in the same pipeline. Across five image datasets and three frozen backbone networks, the paper reports that COUQ outperforms existing continual novelty detectors and continual active-learning baselines, and that its unsupervised variant often beats semi-supervised methods that rely on more supervision.

What carries the argument

The machinery is a per-class PCA subspace model of deep features, derived from the feature reconstruction error (FRE) metric of [32], arranged into an iterative algorithm. For every old class, a PCA transform is fixed at the task where the class was learned and never updated; for every newly detected class, a PCA transform is fit from the samples the current iteration labels as that class. The COUQ score is the ratio of the minimum old-class FRE to the FRE of the predicted new class, and the inner loop alternates between computing scores, selecting high-confidence samples for pseudo-labeling and ambiguous samples for active labeling, retraining the novelty mapper (K-means or a small classifier), and updating the new-class PCA transforms and the continual classifier. Because old-class transforms are never retrained, the score's reference for 'old' does not drift, which is the mechanism the paper credits for preventing degradation over tasks.

What would settle it

Take a novel class that differs from the old classes along a dimension the frozen backbone is known not to encode (for example, a synthetic shift in color or texture that ImageNet features ignore), add it as the new class in a task, and measure COUQ's novelty-detection AUROC; if the score is near chance while a detector with an updated backbone succeeds, the frozen-feature premise is the binding constraint.

Watch

Extended reading notes

Core claim

COUQ's central claim is that reliable per-class uncertainty can be maintained throughout continual open-world learning by combining per-class PCA feature models with an iterative scoring loop. At each task, the score for a sample $u$ is $S^i(u)=\frac{\min_{j\in C^t_{\mathrm{old}}} FRE^0_j(u)}{FRE^{i-1}_m(u)}$, where the numerator is its reconstruction error against the best-matching old-class subspace and the denominator is its reconstruction error against the subspace of the novel class $m$ predicted by the previous iteration's novelty mapper. Samples with high scores are confidently novel, low scores are confidently old, and mid-range scores are ambiguous; the ambiguous ones are the most informative for active labeling. The paper shows empirically that the quality of these scores stays roughly constant across tasks, that the iterative updates reduce error propagation, and that the method outperforms entropy-, margin-, and softmax-based uncertainty baselines as well as the single-novel-class detector incDFM.

Load-bearing premise

The method assumes that a frozen, pre-trained feature extractor represents every future novel class well enough; if a new class needs visual information the frozen network does not capture, the per-class subspaces cannot separate it from old classes and the uncertainty score loses meaning.

Editorial extensions

If this is right

  • Continual novelty detection can be performed without a separate OOD detector: COUQ's score doubles as the novelty score, and its AUROC stays high over tasks.
  • A tiny active-labeling budget (about 1.25% of samples) suffices to keep the continual classifier accurate, because ambiguous samples carry the most information.
  • High-confidence COUQ scores yield pseudo-labels reliable enough to improve semi-supervised continual learning; pseudo-labeling with entropy or margin scores does not help.
  • The method transfers across multiple datasets and frozen backbones, meaning it can be applied to a deployed model without retraining the feature extractor.
  • Unsupervised COUQ, using K-means as the novelty mapper, can outperform semi-supervised baselines, so novelty detection may not require any labels at all in some settings.

Reading between the lines

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

  • An extension the authors do not explore is using the ratio $S^i(u)$ as a calibration signal for when to trust the continual classifier's predictions, since the ratio separates confident-old from confident-new without depending on the classifier's logits.
  • Because the paper leaves old-class distribution shift for future work, the score's stability over time likely depends on old-class features staying fixed; if they drift, the frozen PCA subspaces would age.
  • The same score could be applied to non-visual continual learning, such as sensor or text streams, wherever per-class reconstruction subspaces can be fit on frozen embeddings.
  • One could test whether querying a mix of ambiguous and confident-novel samples remains optimal as the active budget grows; the paper's ablations cover a small budget range.
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

3 major / 7 minor

Summary. This paper addresses continual open-world learning in which, at each task, a model receives an unlabeled pool containing both unseen samples of previously learned classes and samples from an arbitrary number of novel classes. The proposed COUQ method iteratively computes per-novel-class uncertainty scores using per-class PCA reconstruction errors (FRE) from a frozen pre-trained backbone, combined with a novelty mapper (K-means or a small classifier) that assigns pseudo-labels to confident novel samples. The score is S^i(u) = min_{j in C_old} FRE_j(u) / FRE_m^{i-1}(u), where m is the predicted novel class. The paper evaluates COUQ for continual novelty detection, active sample selection, and pseudo-labeling on five datasets and three backbones, and reports large gains over entropy/margin/softmax baselines, incDFM, DFM, GBCL, and CCIC, with ablations on iterativeness, active-selection strategy, and pseudo-labeling.

Significance. The problem is relevant and under-explored: most continual learning assumes an oracle for novelty detection and labels, while this paper targets a more realistic setting with unlabeled mixtures and multiple novel classes per task. The empirical scope is a strength: five datasets, three backbones, and a consistent set of baselines, plus ablations isolating the iterative procedure, the active-selection strategy, and pseudo-labeling. The observation that ambiguity-based active selection outperforms confidence-based selection is interesting and is tested across datasets. The main weakness is that the central 'no degradation over time' claim is not rigorously established, and the absence of repeated-seed statistics limits confidence in some small margins. If those issues are addressed, the paper would be a useful contribution.

major comments (3)
  1. [Eq. (3), §3.2] The claim that uncertainty quality 'does not degrade as new classes are iteratively encountered and learned' is not established by the formulation. For any fixed sample u, the numerator min_{j in C_old} FRE_j(u) is non-increasing as C_old grows, because each newly added old-class PCA subspace provides another candidate that can only lower the minimum. Thus the scores for genuinely novel samples are systematically compressed as old classes accumulate, even with a perfect frozen backbone. Table 1 reports AUROC averaged over all tasks, which can mask a gradual decline, and the task-level curves in Fig. 2 are not conditioned on the number of accumulated old classes. The No-iters ablation in Table 2 varies inner-loop iterations, not the size of C_old, so it does not test this mechanism. Please provide per-task AUROC as a function of task index and old-class count, add a long-horizon experiment with many incremental tasks, or explicitly qualify the no-degradation claim.
  2. [Tables 1, 3, 5] All results are single point estimates with no error bars, no multiple seeds, and no statistical test. Several head-to-head margins are small: in Table 3, Plants shows AL-Amb at 45.2 vs Rand at 41.7, and Cifar100 shows 65.1 vs 64.8; in Table 5, Im21K shows COUQ(P;oneshot) at 78.9, which is higher than COUQ(P;Default) at 76.4, so the stated benefit of iterativeness is not consistent across datasets. Without variance estimates, the robustness of the claimed superiority of COUQ's default configuration is not established. Report means and standard deviations over at least three seeds, and ideally paired significance tests.
  3. [§4.1] The method relies on a frozen ImageNet-pretrained backbone providing features that are sufficiently task-non-specific to represent all future novel classes; if novel classes require features not captured by this representation, the PCA subspaces and the classifier cannot distinguish them regardless of the uncertainty scoring. The paper does not examine when this assumption fails, for instance by measuring performance as a function of class similarity to ImageNet or by using a backbone pretrained on a different distribution. This limits the generality claimed in the abstract and should be discussed explicitly with a failure analysis or a clear scope statement.
minor comments (7)
  1. [§3.3] The sentence 'We describe next how the uncertainty score from Eq. (2) can be used' appears to refer to Eq. (3), since active selection uses the iterative multiclass score rather than the initialization S0.
  2. [Eq. (1)] The reconstruction error is written as ∥f(x) − (T_m^† ∘ T_m)u∥_2; this should be the reconstruction of the feature u = g(x), not of the classifier output f(x).
  3. [Eq. (1)] The notation T_m^† is not defined in the text; please state that it is the pseudo-inverse of the PCA projection.
  4. [§2.1] The definition of C_old^t as ∪_{k=0}^{t-1} C_new^k is correct, but the phrase 'up to and including task t−1' could be clearer about whether t−1 is included; the notation already makes it unambiguous.
  5. [Table 4] CCIC has '−' entries for the ViT backbones; please state in the caption or text why these experiments were not run.
  6. [General] The paper says the code will be released upon acceptance; providing an anonymous link in the current version would be helpful for reproducibility.
  7. [Table 5] The row 'None' is described as the lower-bound random labeling; renaming it to 'Rand' for consistency with Table 3 would improve readability.

Circularity Check

0 steps flagged · score 1.0 of 10

COUQ's central derivation is self-contained: Eq. (3) is a new iterative reconstruction-error ratio evaluated against external baselines, and the self-citations to FRE/DFM/incDFM are component/baseline references rather than load-bearing circularity.

full rationale

The paper's core uncertainty score is defined directly from PCA reconstruction errors: Eq. (1) computes FRE as an l2 reconstruction distance, and Eq. (3) forms S^i(u) as a ratio of the minimum old-class FRE to the predicted novel-class FRE. This is a deterministic, non-fitted scoring rule; no parameter is estimated from the AUROC or accuracy labels used for evaluation. The claimed no-degradation property is an empirical claim backed by AUROC comparisons against external baselines such as Entropy, Margin, Softmax, CCIC, and GBCL, not an identity that follows from the score's definition. The same-author citations to FRE, DFM, and incDFM are used as a building block and as comparison methods, not as a uniqueness theorem or as a substitute for evidence. The skeptical observation that the min over old classes can only shrink as classes accumulate is a legitimate statistical concern about long-horizon novelty scores, but it does not make the derivation circular: the score is not equivalent to its inputs by construction, and the active/pseudo-labeling contribution in Eq. (3) is independently testable and tested against methods outside the authors' prior work.

Assumptions & free parameters 4 free parameters · 3 assumptions · 0 invented entities

The central claim rests on the reliability of FRE as a per-class uncertainty measure, the sufficiency of a frozen backbone, and the accuracy of the novelty mapper. The free parameters are mostly algorithmic choices whose values are deferred to the missing supplementary. No new physical or ontological entities are introduced.

free parameters (4)
  • Pseudo-label selection fraction alpha = 0.20
    Set to 20% of samples predicted as novel per class in Section 4.1; chosen by the authors, not derived.
  • Ambiguity threshold for AL-Amb
    The threshold separating ambiguous from confident samples is defined only in the supplementary, which is not provided; it controls which samples are actively labeled.
  • PCA subspace dimension per class
    The dimensionality of each class's PCA transform in FRE is not specified in the main text; it directly affects reconstruction error quality.
  • Number of inner-loop iterations
    Stopping criteria for the iterative uncertainty update are deferred to the supplementary; the value affects stability and runtime.
assumptions (3)
  • domain assumption Frozen pre-trained backbone features are task-non-specific and sufficient for continual open-world learning.
    Invoked in Section 4.1 to justify using a frozen backbone; if false, PCA subspaces and the classifier cannot represent novel classes.
  • domain assumption FRE computed on PCA subspaces is a reliable per-class uncertainty measure.
    COUQ builds on FRE from [32]; the ratio in Eq. 3 assumes low FRE to a class implies membership in that class.
  • domain assumption The novelty mapper (K-means or MLP) correctly assigns novel samples to the right class IDs with acceptable error.
    Eq. 3 uses m = M^{i-1}(u) to select the PCA transform; systematic mapper errors would corrupt the uncertainty scores.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Uncertainty Quantification in Continual Open-World Learning." pith.science (2026). https://pith.science/paper/EQ3NLXA4

@misc{pith2026241216409,
  author       = {Pith},
  title        = {Pith review of: Uncertainty Quantification in Continual Open-World Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EQ3NLXA4}},
  note         = {Machine review of arXiv:2412.16409}
}
read the original abstract

AI deployed in the real-world should be capable of autonomously adapting to novelties encountered after deployment. Yet, in the field of continual learning, the reliance on novelty and labeling oracles is commonplace albeit unrealistic. This paper addresses a challenging and under-explored problem: a deployed AI agent that continuously encounters unlabeled data - which may include both unseen samples of known classes and samples from novel (unknown) classes - and must adapt to it continuously. To tackle this challenge, we propose our method COUQ "Continual Open-world Uncertainty Quantification", an iterative uncertainty estimation algorithm tailored for learning in generalized continual open-world multi-class settings. We rigorously apply and evaluate COUQ on key sub-tasks in the Continual Open-World: continual novelty detection, uncertainty guided active learning, and uncertainty guided pseudo-labeling for semi-supervised CL. We demonstrate the effectiveness of our method across multiple datasets, ablations, backbones and performance superior to state-of-the-art.

Figures

Figures reproduced from arXiv: 2412.16409 by the authors.

Figure 1
Figure 1. A general open-world, continual learning pipeline. Un [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. (Left A.1,B.1) AUROC of Novelty Detection at each continual task. Number of novel classes per task is in parenthesis. COUQ (green) clearly outperforms baselines both in both semi-supervised (solid line) and unsupervised versions (dashedline); (Center A.2,B.2) Results varying the supervision budget; (Right A.3,B.3) Results varying Novel Class Increment per task. For (left,right) Supervision budget is 1.25% and all pl… view at source ↗
Figure 3
Figure 3. (Row 1) Continual classification accuracy over contin [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 42 canonical work pages

  1. [1]

    Continual evidential deep learning for out- of-distribution detection

    Eduardo Aguilar, Bogdan Raducanu, Petia Radeva, and Joost Van de Weijer. Continual evidential deep learning for out- of-distribution detection. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Work- shops, pages 3444–3454, 2023. 1

    work page 2023

  2. [2]

    Ahuja, Ibrahima J

    Nilesh A. Ahuja, Ibrahima J. Ndiour, Trushant Kalyanpur, and Omesh Tickoo. Probabilistic modeling of deep features for out-of-distribution and adversarial detection. InBayesian Deep Learning workshop, NeurIPS, 2019. 2

    work page 2019

  3. [3]

    An image is worth 16x16 words: Trans- formers for image recognition at scale

    Dosovitskiy Alexey. An image is worth 16x16 words: Trans- formers for image recognition at scale. arXiv preprint arXiv: 2010.11929, 2020. 5

    arXiv 2010

  4. [4]

    Continual novelty detection

    Rahaf Aljundi, Daniel Olmeda Reino, Nikolay Chumerin, and Richard E Turner. Continual novelty detection. In Conference on Lifelong Learning Agents, pages 1004–1025. PMLR, 2022. 3

    work page 2022

  5. [5]

    Few-shot continual active learning by a robot

    Ali Ayub and Carter Fendley. Few-shot continual active learning by a robot. Advances in Neural Information Pro- cessing Systems, 35:30612–30624, 2022. 1, 5

    work page 2022

  6. [6]

    Few-shot continual active learning by a robot

    Ali Ayub and Carter Fendley. Few-shot continual active learning by a robot. In Proceedings of the 36th International Conference on Neural Information Processing Systems. Cur- ran Associates Inc., 2024. 3

    work page 2024

  7. [7]

    Beyond supervised continual learning: a review

    Benedikt Bagus, Alexander Gepperth, and Timoth ´ee Lesort. Beyond supervised continual learning: a review. arXiv preprint arXiv:2208.14307, 2022. 1, 3

    arXiv 2022

  8. [8]

    Mixmatch: A holistic approach to semi-supervised learning

    David Berthelot, Nicholas Carlini, Ian Goodfellow, Nicolas Papernot, Avital Oliver, and Colin A Raffel. Mixmatch: A holistic approach to semi-supervised learning. Advances in neural information processing systems, 32, 2019. 3, 5

    work page 2019

Show all 53 references
  1. [9]

    Continual semi-supervised learning through contrastive interpolation consistency

    Matteo Boschini, Pietro Buzzega, Lorenzo Bonicelli, Angelo Porrello, and Simone Calderara. Continual semi-supervised learning through contrastive interpolation consistency. Pat- tern Recognition Letters, 162:9–14, 2022. 3, 5, 8

    2022

  2. [10]

    Continual semi-supervised learning through contrastive interpolation consistency

    Matteo Boschini, Pietro Buzzega, Lorenzo Bonicelli, Angelo Porrello, and Simone Calderara. Continual semi-supervised learning through contrastive interpolation consistency. Pat- tern Recognition Letters, 162:9–14, 2022. 1, 3

    2022

  3. [11]

    Incorporating diversity in active learning with support vector machines

    Klaus Brinker. Incorporating diversity in active learning with support vector machines. In International Conference on Machine Learning, 2003. 1, 3

    2003

  4. [12]

    Rethinking experience replay: a bag of tricks for continual learning

    Pietro Buzzega, Matteo Boschini, Angelo Porrello, and Si- mone Calderara. Rethinking experience replay: a bag of tricks for continual learning. In2020 25th International Con- ference on Pattern Recognition (ICPR) , pages 2180–2187. IEEE, 2021. 5

    2021

  5. [13]

    Unsupervised learning of visual features by contrasting cluster assignments

    Mathilde Caron, Ishan Misra, Julien Mairal, Priya Goyal, Pi- otr Bojanowski, and Armand Joulin. Unsupervised learning of visual features by contrasting cluster assignments. Ad- vances in Neural Information Processing Systems, 33:9912– 9924, 2020. 5

    2020

  6. [14]

    Emerg- ing properties in self-supervised vision transformers

    Mathilde Caron, Hugo Touvron, Ishan Misra, Herv ´e J´egou, Julien Mairal, Piotr Bojanowski, and Armand Joulin. Emerg- ing properties in self-supervised vision transformers. In Pro- ceedings of the IEEE/CVF international conference on com- puter vision, pages 9650–9660, 2021. 5

    2021

  7. [15]

    Superposition of many mod- els into one

    Brian Cheung, Alexander Terekhov, Yubei Chen, Pulkit Agrawal, and Bruno Olshausen. Superposition of many mod- els into one. Advances in neural information processing sys- tems, 32, 2019. 3

    2019

  8. [16]

    Breaking the closed world assump- tion in text classification

    Geli Fei and Bing Liu. Breaking the closed world assump- tion in text classification. In Proceedings of the 2016 Con- ference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 506–514, 2016. 3

    2016

  9. [17]

    Catastrophic forgetting in connectionist networks

    Robert M French. Catastrophic forgetting in connectionist networks. Trends in cognitive sciences, 3(4):128–135, 1999. 1

    1999

  10. [18]

    Dropout as a bayesian approximation: Representing model uncertainty in deep learning

    Yarin Gal and Zoubin Ghahramani. Dropout as a bayesian approximation: Representing model uncertainty in deep learning. In international conference on machine learning , pages 1050–1059, 2016. 2

    2016

  11. [19]

    Deep bayesian active learning with image data

    Yarin Gal, Riashat Islam, and Zoubin Ghahramani. Deep bayesian active learning with image data. In International Conference on Machine Learning, pages 1183–1192. PMLR,

  12. [20]

    Learning to discover novel visual categories via deep transfer cluster- ing

    Kai Han, Andrea Vedaldi, and Andrew Zisserman. Learning to discover novel visual categories via deep transfer cluster- ing. In Proceedings of the IEEE/CVF International Confer- ence on Computer Vision, pages 8401–8409, 2019. 3

    2019

  13. [21]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In Proceed- ings of the IEEE conference on computer vision and pattern recognition, pages 770–778, 2016. 5

    2016

  14. [22]

    Eurosat: A novel dataset and deep learning benchmark for land use and land cover classification, 2019

    Patrick Helber, Benjamin Bischke, Andreas Dengel, and Damian Borth. Eurosat: A novel dataset and deep learning benchmark for land use and land cover classification, 2019. 5

    2019

  15. [23]

    A baseline for detect- ing misclassified and out-of-distribution examples in neural networks

    Dan Hendrycks and Kevin Gimpel. A baseline for detect- ing misclassified and out-of-distribution examples in neural networks. 2017. 2

    2017

  16. [24]

    The inaturalist species classification and de- tection dataset, 2018

    Grant Van Horn, Oisin Mac Aodha, Yang Song, Yin Cui, Chen Sun, Alex Shepard, Hartwig Adam, Pietro Perona, and Serge Belongie. The inaturalist species classification and de- tection dataset, 2018. 5 9

    2018

  17. [25]

    A soft nearest-neighbor framework for con- tinual semi-supervised learning

    Zhiqi Kang, Enrico Fini, Moin Nabi, Elisa Ricci, and Kar- teek Alahari. A soft nearest-neighbor framework for con- tinual semi-supervised learning. In Proceedings of the IEEE/CVF International Conference on Computer Vision , pages 11868–11877, 2023. 1, 3

    2023

  18. [26]

    Overcoming catastrophic forgetting in neu- ral 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 neu- ral networks. Proceedings of the national academy of sc...

    2017

  19. [27]

    Learning multiple layers of features from tiny images

    Alex Krizhevsky. Learning multiple layers of features from tiny images. University of Toronto, 2012. 5

    2012

  20. [28]

    A simple unified framework for detecting out-of-distribution samples and adversarial attacks

    Kimin Lee, Kibok Lee, Honglak Lee, and Jinwoo Shin. A simple unified framework for detecting out-of-distribution samples and adversarial attacks. In Advances in Neural In- formation Processing Systems, pages 7167–7177, 2018. 2

    2018

  21. [29]

    Lewis and William A

    David D. Lewis and William A. Gale. A sequential algo- rithm for training text classifiers. In SIGIR ’94, pages 3–12, London, 1994. Springer London. 1, 3

    1994

  22. [30]

    Enhancing the re- liability of out-of-distribution image detection in neural net- works

    Shiyu Liang, Yixuan Li, and R Srikant. Enhancing the re- liability of out-of-distribution image detection in neural net- works. 2018. 2

    2018

  23. [31]

    Active generalized category discovery

    Shijie Ma, Fei Zhu, Zhun Zhong, Xu-Yao Zhang, and Cheng- Lin Liu. Active generalized category discovery. In Proceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 16890–16900, 2024. 2, 3, 8

    2024

  24. [32]

    Out-of-distribution detection with subspace techniques and probabilistic modeling of features

    Ibrahima Ndiour, Nilesh A Ahuja, and Omesh Tickoo. Out-of-distribution detection with subspace techniques and probabilistic modeling of features. arXiv preprint arXiv:2012.04250, 2020. 3

    2012 arXiv

  25. [33]

    Sub- space modeling for fast out-of-distribution and anomaly de- tection

    Ibrahima J Ndiour, Nilesh A Ahuja, and Omesh Tickoo. Sub- space modeling for fast out-of-distribution and anomaly de- tection. In 2022 IEEE International Conference on Image Processing (ICIP), pages 3041–3045. IEEE, 2022. 5

    2022

  26. [34]

    Continual lifelong learning with neural networks: A review

    German I Parisi, Ronald Kemker, Jose L Part, Christopher Kanan, and Stefan Wermter. Continual lifelong learning with neural networks: A review. Neural Networks, 113:54–71,

  27. [35]

    Open-world machine learning: appli- cations, challenges, and opportunities

    Jitendra Parmar, Satyendra Chouhan, Vaskar Raychoudhury, and Santosh Rathore. Open-world machine learning: appli- cations, challenges, and opportunities. ACM Computing Sur- veys, 55(10):1–37, 2023. 1

    2023

  28. [36]

    icarl: Incremental classifier and representation learning

    Sylvestre-Alvise Rebuffi, Alexander Kolesnikov, Georg Sperl, and Christoph H Lampert. icarl: Incremental classifier and representation learning. In Proceedings of the IEEE con- ference on Computer Vision and Pattern Recognition, pages 2001–2010, 2017. 3

    2001

  29. [37]

    Likelihood ratios for out-of-distribution detec- tion

    Jie Ren, Peter J Liu, Emily Fertig, Jasper Snoek, Ryan Poplin, Mark Depristo, Joshua Dillon, and Balaji Lakshmi- narayanan. Likelihood ratios for out-of-distribution detec- tion. In Advances in Neural Information Processing Systems, pages 14707–14718, 2019. 2

    2019

  30. [38]

    Gupta, Xiaojiang Chen, and Xin Wang

    Pengzhen Ren, Yun Xiao, Xiaojun Chang, Po-Yao Huang, Zhihui Li, Brij B. Gupta, Xiaojiang Chen, and Xin Wang. A Survey of Deep Active Learning. ACM Computing Surveys, 54(9):180:1–180:40, 2021. 1, 3

    2021

  31. [39]

    Imagenet-21k pretraining for the masses,

    Tal Ridnik, Emanuel Ben-Baruch, Asaf Noy, and Lihi Zelnik-Manor. Imagenet-21k pretraining for the masses,

  32. [40]

    Closed-loop memory gan for continual learning

    Amanda Rios and Laurent Itti. Closed-loop memory gan for continual learning. arXiv preprint arXiv:1811.01146, 2018. 3

    2018 arXiv

  33. [41]

    Lifelong learning without a task oracle

    Amanda Rios and Laurent Itti. Lifelong learning without a task oracle. In 2020 IEEE 32nd International Conference on Tools with Artificial Intelligence (ICTAI), pages 255–263. IEEE, 2020. 3, 5

    2020

  34. [42]

    incdfm: Incremental deep feature modeling for continual novelty detection

    Amanda Rios, Nilesh Ahuja, Ibrahima Ndiour, Utku Genc, Laurent Itti, and Omesh Tickoo. incdfm: Incremental deep feature modeling for continual novelty detection. In Eu- ropean Conference on Computer Vision , pages 588–604. Springer, 2022. 2, 3, 5, 6

    2022

  35. [43]

    Experience replay for continual learning

    David Rolnick, Arun Ahuja, Jonathan Schwarz, Timothy Lil- licrap, and Gregory Wayne. Experience replay for continual learning. Advances in Neural Information Processing Sys- tems, 32, 2019. 5

    2019

  36. [44]

    Active learning for convolu- tional neural networks: A core-set approach

    Ozan Sener and Silvio Savarese. Active learning for convolu- tional neural networks: A core-set approach. InInternational Conference on Learning Representations, 2018. 1

    2018

  37. [45]

    Active learning literature survey

    Burr Settles. Active learning literature survey. Computer Sciences Technical Report 1648, University of Wisconsin– Madison, 2009. 1, 3

    2009

  38. [46]

    Doc: Deep open classifica- tion of text documents

    Lei Shu, Hu Xu, and Bing Liu. Doc: Deep open classifica- tion of text documents. In Proceedings of the 2017 Confer- ence on Empirical Methods in Natural Language Processing, pages 2911–2916, 2017. 3

    2017

  39. [47]

    Support vector machine ac- tive learning with applications to text classification.J

    Simon Tong and Daphne Koller. Support vector machine ac- tive learning with applications to text classification.J. Mach. Learn. Res., 2:45–66, 2002. 1, 3

    2002

  40. [48]

    Generalized category discovery

    Sagar Vaze, Kai Han, Andrea Vedaldi, and Andrew Zisser- man. Generalized category discovery. In IEEE Conference on Computer Vision and Pattern Recognition, 2022. 2, 3

    2022

  41. [49]

    Active continual learning: On balancing knowledge retention and learnability

    Thuy-Trang Vu, Shahram Khadivi, Mahsa Ghorbanali, Dinh Phung, and Gholamreza Haffari. Active continual learning: On balancing knowledge retention and learnability. arXiv preprint arXiv:2305.03923, 2023. 1, 3, 5

    2023 arXiv

  42. [50]

    Vim: Out-of-distribution with virtual-logit matching

    Haoqi Wang, Zhizhong Li, Litong Feng, and Wayne Zhang. Vim: Out-of-distribution with virtual-logit matching. InPro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022. 2

    2022

  43. [51]

    Beneficial perturbation network for designing general adap- tive artificial intelligence systems

    Shixian Wen, Amanda Rios, Yunhao Ge, and Laurent Itti. Beneficial perturbation network for designing general adap- tive artificial intelligence systems. IEEE Transactions on Neural Networks and Learning Systems, 2021. 3

    2021

  44. [52]

    Learning loss for active learning

    Donggeun Yoo and In So Kweon. Learning loss for active learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 93–102,

  45. [53]

    Places: A 10 million image database for scene recognition

    Bolei Zhou, Agata Lapedriza, Aditya Khosla, Aude Oliva, and Antonio Torralba. Places: A 10 million image database for scene recognition. IEEE Transactions on Pattern Analy- sis and Machine Intelligence, 2017. 5 10

    2017

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.