Pith. sign in

REVIEW 3 major objections 5 minor 51 references

Evolving from Unknown to Known: Retentive Angular Representation Learning for Incremental Open Set Recognition

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Incremental open-set recognition improves when future class prototypes are excluded from the loss.

desk verdict One genuinely new knob—excluding inactive ETF prototypes from the softmax denominator—with consistent gains, but the open-set mechanism is unenforced geometry that the self-defined benchmark may accidentally satisfy. read the letter →

arxiv 2509.06570 v2 pith:GM7U2BAC submitted 2025-09-08 cs.CV

Runqing Yang , Yimin Fu , Changyuan Wu , Zhunga Liu This is my paper · ORCID
classification cs.CV
keywords incrementalopensetrecognitionequiangulartightframeretentiveangularspacevirtual-intrinsicinteractionclasslearningrepresentationdriftprototype
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

RARL tackles incremental open set recognition, where a model must reject unknown classes at task t and then learn those same classes at task t+1. The paper claims that existing equiangular-tight-frame methods hurt open-set performance because they push known features away from all prototypes, including inactive future ones, squeezing unknown representations into narrow gaps between known classes. RARL instead keeps inactive prototypes out of the softmax denominator, so known classes anchor to their own fixed prototypes while unknown samples are encouraged to reside in the broader open region. A virtual-intrinsic interactive loss plus positive/negative and old/new boundary rectification sharpen inter-class margins and counter rehearsal imbalance. The method reports better accuracy, unknown detection, and open-set classification than ten existing OSR and CIL baselines on CIFAR100 and TinyImageNet.

What carries the argument

The retentive angular space is the central object: a set of class prototypes pre-fixed as vertices of an equiangular tight frame in the normalized feature space, with some vertices left inactive for future classes. The pivotal mechanism is that the softmax denominator omits inactive prototypes, so optimization never pushes known features away from future class regions. The virtual-intrinsic interactive loss then synthesizes virtual instances near class boundaries and uses sigmoid-transformed cosine similarities to pull virtual features toward their own learnable prototypes, repel intrinsic features from ambiguous virtual regions, and separate virtual classes from each other; PNBR rebalances the positive/negative gradient asymmetry in this loss, and ONBR reduces cosine similarity for old classes to counteract boundary drift toward new classes.

What would settle it

Evaluate RARL with unknown classes whose labels never appear in any later task and measure AUROC/OSCR, or inspect the angular distance of those unknowns to inactive prototypes: if open-set detection drops to baseline levels or unknowns do not cluster near inactive prototypes, the central geometric claim is refuted.

Watch

Extended reading notes

Core claim

The central claim is that an incremental open-set model should keep future classes geometrically available by removing inactive ETF prototypes from the classification loss. In the angular space defined by pre-fixed prototypes, known representations are pulled toward their active prototypes while unknown representations are left to occupy the wide complement around inactive prototypes, so converting an unknown into a known later causes little representation drift. The paper further claims that a virtual-intrinsic interactive loss, which synthesizes boundary-proximal virtual instances and imposes a triple pull/push dynamic, makes known-class features more compact and separated, and that the stratified rectification strategy corrects the resulting positive/negative imbalance and the old/new imbalance from rehearsal data. These components together, the paper argues, maintain both closed-set discrimination and open-set rejection as the recognizable scope grows, and the experiments support this with consistent gains over existing methods.

Load-bearing premise

The load-bearing assumption is that inputs the model has never seen will actually fall into the large angular region reserved around unused prototypes, even though the training procedure never shows the model an unknown and the benchmark's unknown set is just the next task's training classes.

Editorial extensions

If this is right

  • Existing OSR losses that penalize proximity to all fixed prototypes can be modified to ignore future prototypes, giving unknown inputs a defined home region and improving separation from known classes.
  • When an unknown class is later promoted to known, its features should already be near the activated prototype, so the model needs less adaptation and forgets less.
  • The virtual-intrinsic interaction provides a rehearsal-friendly way to enforce inter-class margins using only current-batch virtual instances, which is valuable when old exemplars are scarce.
  • The proposed benchmark, where each task's unknowns are the next task's training classes, offers a reusable evaluation protocol for incremental open set recognition.

Reading between the lines

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

  • A testable extension is to apply the same loss recipe in static OSR: masking unused ETF prototypes may improve unknown detection even when no incremental tasks follow.
  • Because the benchmark defines unknowns as the next task's training labels, the open-space assumption is validated against 'future knowns' rather than arbitrary outliers; real-world unknowns that never become known may behave differently.
  • If the geometric complement hypothesis is correct, unknown-ness could be scored directly by angular distance to the nearest inactive prototype, yielding a simple per-sample confidence threshold.
  • The success of geometric space reservation suggests that similar pre-allocated prototype gaps could help task-free or few-shot continual learning, where classes arrive unpredictably.
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

3 major / 5 minor

Summary. The paper formalizes incremental open set recognition (IOSR), where each task adds new classes and the model must both classify seen classes and reject samples from classes that will appear in future tasks. It proposes RARL, which fixes ETF classifiers, excludes inactive (future) prototypes from the softmax denominator, synthesizes virtual classes for a virtual-intrinsic interaction loss, and applies PNBR/ONBR to rebalance gradient strengths. The reported experiments on CIFAR-100 and TinyImageNet compare RARL with OSR and CIL baselines and show consistent improvements in accuracy, AUROC, and OSCR, with an ablation study on the main components.

Significance. The paper's strength is its clear problem formulation and broad comparison: it defines an IOSR protocol, reports consistent gains across two datasets and four task splits, and provides an ablation that isolates the contribution of each loss. If the claimed mechanism were validated on genuinely novel unknowns, RARL would be a simple and practical baseline for IOSR. As it stands, the open-set advantage depends on an unenforced geometric assumption and on a benchmark where 'unknown' samples are drawn from the same distribution as future training classes.

major comments (3)
  1. [Sec. 5.1.4] The evaluation protocol defines the unknown set at task t as the training labels of task t+1, i.e., Y_t^U = Y_{t+1}^{train}. This means the 'unknown' inputs are not semantically novel categories but future known classes that will later be assigned to the very ETF prototypes that were kept inactive. The reported AUROC/OSCR improvements are therefore partly a measure of future-class placement rather than of rejection of truly unseen inputs. The paper should evaluate at least one configuration with disjoint held-out unknown classes (e.g., from an external dataset) or report open-set metrics for samples that never appear as future training classes, otherwise the open-set claim is not established for genuinely novel inputs.
  2. [Secs. 4.2, 4.5, Eq. (12)] The claim that unknown representations are 'encouraged to be distributed in the broader open space opposite known prototypes' (Fig. 2d, Sec. 4.2) is not enforced by any term in the total loss. The training objective in Eq. (12) is evaluated only on known and synthetic virtual instances; no gradient is ever computed for unknown samples. Excluding inactive prototypes from the softmax denominator in Eq. (3) removes a repulsive force on active classes, but it does not place unknown inputs into the reserved region. This is an assumed geometric side effect, not a mechanism demonstrated in the paper. The authors should either add an explicit regularizer that scores the complementary region or provide direct feature-space evidence (e.g., angular distance of novel unknowns to inactive versus known prototypes) that the claimed alignment occurs.
  3. [Secs. 4.4, 4.5, Tables 1-3] Several implementation parameters needed for reproduction are missing: the ONBR hyperparameter A is introduced in Sec. 4.4 but never specified, and the less-forget loss weight λ_dis in Eq. (12) is never given. In addition, all accuracy, AUROC, and OSCR numbers appear to come from a single run with no standard deviations or significance tests. Because the reported open-set gains over LUCIR are as small as 0.26% in average AUROC (CIFAR-100, Base 20 with 8 Steps), the reader cannot tell whether these differences are stable. The authors should report the missing hyperparameter values and provide multi-seed means with error bars, or at least a paired comparison across seeds.
minor comments (5)
  1. [Eqs. (2)-(3)] The symbol K is overloaded: in Eq. (2) it is the number of preallocated ETF prototypes, while in Eq. (3) it is the number of current-task classes; using distinct symbols such as K_total and K_t would clarify the scope of the softmax denominator.
  2. [Eq. (5)] It is unclear whether the virtual instance mixing is applied to raw pixels or to batch-normalized features, and the meaning of K_batch ('the number of intrinsic classes in a batch') is ambiguous in a per-instance formula; please specify the exact batch construction and normalization used.
  3. [Table 2] The LUCIR row for the Base 60 with 7 Steps setting contains '66.3966.25' without a separator; this appears to be a formatting typo that should be corrected.
  4. [Fig. 2] The caption says the figure illustrates binary classification, but the panel shows two known prototypes and one inactive prototype; please clarify whether the 'binary classification' refers to a reduced 2D example or to the known/unknown distinction.
  5. [Sec. 5.1.3 and Table 3] The baseline label 'SoftMax (ALL)' in Table 3 is explained only in the ablation text; consider defining it in the table caption, since the name alone does not indicate that all ETF prototypes are included in the softmax.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; RARL's losses are trained on known data and its open-set metrics are measured, not fitted.

full rationale

The claimed derivation chain runs from fixed ETF prototypes (Eq. 2), cosine softmax over active classes only (Eq. 3), virtual-intrinsic losses (Eqs. 5-9), PNBR/ONBR rectification (Eq. 10), to the total loss (Eq. 12). Each component is optimized on the current task's training data; no reported AUROC or OSCR value is a re-read-out of a fitted parameter. The learnable a in PNBR and hyperparameter A in ONBR are trained or set before evaluation, and the final numbers come from held-out test sets. The benchmark's definition Y_t^U = Y_{t+1}^{train} is a deliberate protocol for incremental open set recognition, not a way of encoding the answer into the loss; all compared methods see the same protocol. Self-citations (refs. 8, 9, 10, 23) appear as related work or baseline comparators and are not used to justify the method's core mechanism. The abstract's claim that unknowns are 'encouraged to align around inactive prototypes' is not backed by any loss term evaluated on unknown samples, and therefore is an unenforced geometric assumption; that is a support or correctness issue, not a circular derivation. No equation in the paper reduces to its own input by construction.

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

The method depends on a stack of previously established constructs: fixed ETF prototypes, virtual class synthesis, and a less-forget constraint. The paper contributes the loss-level combination and the benchmark, but several hyperparameters are undisclosed and the benchmark's unknown definition is an author-chosen protocol.

free parameters (5)
  • lambda_VII = 0.01
    Weight of the VII loss in Eq. (12), set by hand.
  • lambda_dis = not specified
    Weight of the less-forget loss in Eq. (12); no value given, blocking reproduction.
  • ONBR hyperparameter A = not specified
    Scales the old-class cosine rectification in Sec. 4.4(b); no value or sensitivity analysis is reported.
  • PNBR learnable parameter a = learned, not reported
    Adaptive coefficient in Eq. (10) that rebalances positive and negative virtual-class gradients; final value never reported.
  • virtual mix lambda = 0.5
    Mixing coefficient in Eq. (5); fixed by hand with no study of alternatives.
assumptions (4)
  • standard math Class prototypes form an equiangular tight frame in the feature space.
    Adopted from Papyan et al. and Pernici et al. (Sec. 4.2); relies on neural collapse theory rather than being proven here.
  • domain assumption Virtual instances synthesized from batches preserve enough semantic relation to their corresponding intrinsic class.
    Needed for the VII loss in Eq. (9) to make sense; no experiment directly validates this semantic preservation.
  • ad hoc to paper Unknown inputs will occupy the complementary open space away from known prototypes when inactive prototypes are excluded from the loss.
    This geometric assumption is the paper's core open-set mechanism (Fig. 2d, Sec. 4.2), but no unknown data is ever used in training to verify it.
  • ad hoc to paper The open-set evaluation should define the unknown set for task t as the training labels of task t+1.
    The new IOSR benchmark in Sec. 5.1.4 encodes this definition; other unknown distributions may change conclusions.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Evolving from Unknown to Known: Retentive Angular Representation Learning for Incremental Open Set Recognition." pith.science (2026). https://pith.science/paper/GM7U2BAC

@misc{pith2026250906570,
  author       = {Pith},
  title        = {Pith review of: Evolving from Unknown to Known: Retentive Angular Representation Learning for Incremental Open Set Recognition},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GM7U2BAC}},
  note         = {Machine review of arXiv:2509.06570}
}
read the original abstract

Existing open set recognition (OSR) methods are typically designed for static scenarios, where models aim to classify known classes and identify unknown ones within fixed scopes. This deviates from the expectation that the model should incrementally identify newly emerging unknown classes from continuous data streams and acquire corresponding knowledge. In such evolving scenarios, the discriminability of OSR decision boundaries is hard to maintain due to restricted access to former training data, causing severe inter-class confusion. To solve this problem, we propose retentive angular representation learning (RARL) for incremental open set recognition (IOSR). In RARL, unknown representations are encouraged to align around inactive prototypes within an angular space constructed under the equiangular tight frame, thereby mitigating excessive representation drift during knowledge updates. Specifically, we adopt a virtual-intrinsic interactive (VII) training strategy, which compacts known representations by enforcing clear inter-class margins through boundary-proximal virtual classes. Furthermore, a stratified rectification strategy is designed to refine decision boundaries, mitigating representation bias and feature space distortion caused by imbalances between old/new and positive/negative class samples. We conduct thorough evaluations on CIFAR100 and TinyImageNet datasets and establish a new benchmark for IOSR. Experimental results across various task setups demonstrate that the proposed method achieves state-of-the-art performance.

Figures

Figures reproduced from arXiv: 2509.06570 by the authors.

Figure 1
Figure 1. Feature distribution on the MNIST dataset un [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Feature space for binary classification: The vertices of the maximal ETF in the 2D plane are used as classification [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. An overall illustration of our method, which includes: (a) a virtual class synthesis module, (b) a feature extraction [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: (a) Sigmoid function variation after using PNBR. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Line plot of per-task performance on TinyImageNet under Base 50 with 10 Steps setting. [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: t-SNE visualization of learned features on [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 44 canonical work pages

  1. [1]

    Rainbow memory: Continual learn- ing with a memory of diverse samples

    Jihwan Bang, Heesu Kim, YoungJoon Yoo, Jung-Woo Ha, and Jonghyun Choi. Rainbow memory: Continual learn- ing with a memory of diverse samples. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 8218–8227, 2021

    work page 2021

  2. [2]

    Towards open set deep networks

    Abhijit Bendale and Terrance E Boult. Towards open set deep networks. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 1563–1572, 2016

    work page 2016

  3. [3]

    Best fitting hyperplanes for classification

    Hakan Cevikalp. Best fitting hyperplanes for classification. IEEE transactions on pattern analysis and machine intelli- gence, 39(6):1076–1088, 2016

    work page 2016

  4. [4]

    Riemannian walk for incremen- tal learning: Understanding forgetting and intransigence

    Arslan Chaudhry, Puneet K Dokania, Thalaiyasingam Ajan- than, and Philip HS Torr. Riemannian walk for incremen- tal learning: Understanding forgetting and intransigence. In Proceedings of the European conference on computer vision (ECCV), pages 532–547, 2018

    work page 2018

  5. [5]

    Adversarial reciprocal points learning for open set recognition.IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(11):8065–8081, 2021

    Guangyao Chen, Peixi Peng, Xiangqian Wang, and Yonghong Tian. Adversarial reciprocal points learning for open set recognition.IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(11):8065–8081, 2021

    work page 2021

  6. [6]

    Reducing network agnostophobia.Advances in Neural In- formation Processing Systems, 31, 2018

    Akshay Raj Dhamija, Manuel G ¨unther, and Terrance Boult. Reducing network agnostophobia.Advances in Neural In- formation Processing Systems, 31, 2018

    work page 2018

  7. [7]

    Podnet: Pooled outputs dis- tillation for small-tasks incremental learning

    Arthur Douillard, Matthieu Cord, Charles Ollion, Thomas Robert, and Eduardo Valle. Podnet: Pooled outputs dis- tillation for small-tasks incremental learning. InComputer vision–ECCV 2020: 16th European conference, Glasgow, UK, August 23–28, 2020, proceedings, part XX 16, pages 86–102. Springer, 2020

    work page 2020

  8. [8]

    Reason and discov- ery: A new paradigm for open set recognition.IEEE Trans- actions on Pattern Analysis and Machine Intelligence, 2025

    Yimin Fu, Zhunga Liu, and Jialin Lyu. Reason and discov- ery: A new paradigm for open set recognition.IEEE Trans- actions on Pattern Analysis and Machine Intelligence, 2025

    work page 2025

Show all 51 references
  1. [9]

    Logit proto- type learning with active multimodal representation for ro- bust open-set recognition.Science China Information Sci- ences, 67(6):162204, 2024

    Yimin Fu, Zhunga Liu, and Zicheng Wang. Logit proto- type learning with active multimodal representation for ro- bust open-set recognition.Science China Information Sci- ences, 67(6):162204, 2024

    2024

  2. [10]

    Yimin Fu, Zhunga Liu, Changyuan Wu, Feiyan Wu, and Meiqin Liu. Class-incremental recognition of objects in re- mote sensing images with dynamic hybrid exemplar selec- tion.IEEE Transactions on Aerospace and Electronic Sys- tems, 60(3):3468–3481, 2024

    2024

  3. [11]

    Generative openmax for multi-class open set clas- sification.arXiv preprint arXiv:1707.07418, 2017

    ZongYuan Ge, Sergey Demyanov, Zetao Chen, and Rahil Garnavi. Generative openmax for multi-class open set clas- sification.arXiv preprint arXiv:1707.07418, 2017

    2017 arXiv

  4. [12]

    Toward open-set face recognition

    Manuel Gunther, Steve Cruz, Ethan M Rudd, and Terrance E Boult. Toward open-set face recognition. InProceedings of the IEEE conference on computer vision and pattern recog- nition workshops, pages 71–80, 2017

    2017

  5. [13]

    Dyson: Dynamic feature space self- organization for online task-free class incremental learning

    Yuhang He, Yingjie Chen, Yuhan Jin, Songlin Dong, Xing Wei, and Yihong Gong. Dyson: Dynamic feature space self- organization for online task-free class incremental learning. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 23741–23751, 2024

    2024

  6. [14]

    Learning a unified classifier incrementally via rebalancing

    Saihui Hou, Xinyu Pan, Chen Change Loy, Zilei Wang, and Dahua Lin. Learning a unified classifier incrementally via rebalancing. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 831–839, 2019

    2019

  7. [15]

    Distilling causal effect of data in class- incremental learning

    Xinting Hu, Kaihua Tang, Chunyan Miao, Xian-Sheng Hua, and Hanwang Zhang. Distilling causal effect of data in class- incremental learning. InProceedings of the IEEE/CVF con- ference on Computer Vision and Pattern Recognition, pages 3957–3966, 2021

    2021

  8. [16]

    Teacher–explorer– student learning: A novel learning method for open set recognition.IEEE Transactions on Neural Networks and Learning Systems, 2023

    Jaeyeon Jang and Chang Ouk Kim. Teacher–explorer– student learning: A novel learning method for open set recognition.IEEE Transactions on Neural Networks and Learning Systems, 2023

    2023

  9. [17]

    Ib-drr- incremental learning with information-back discrete repre- sentation replay

    Jian Jiang, Edoardo Cetin, and Oya Celiktutan. Ib-drr- incremental learning with information-back discrete repre- sentation replay. InProceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition, pages 3533–3542, 2021

    2021

  10. [18]

    Alleviating repre- sentational shift for continual fine-tuning

    Shibo Jie, Zhi-Hong Deng, and Ziheng Li. Alleviating repre- sentational shift for continual fine-tuning. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 3810–3819, 2022

    2022

  11. [19]

    Hard negative mix- ing for contrastive learning.Advances in neural information processing systems, 33:21798–21809, 2020

    Yannis Kalantidis, Mert Bulent Sariyildiz, Noe Pion, Philippe Weinzaepfel, and Diane Larlus. Hard negative mix- ing for contrastive learning.Advances in neural information processing systems, 33:21798–21809, 2020

    2020

  12. [20]

    Overcoming catastrophic forgetting in neu- ral networks.Proceedings of the national academy of sci- ences, 114(13):3521–3526, 2017

    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 sci...

    2017

  13. [21]

    Opengan: Open-set recog- nition via open data generation

    Shu Kong and Deva Ramanan. Opengan: Open-set recog- nition via open data generation. InProceedings of the IEEE/CVF international conference on computer vision, pages 813–822, 2021

    2021

  14. [22]

    Learning without forgetting

    Zhizhong Li and Derek Hoiem. Learning without forgetting. IEEE transactions on pattern analysis and machine intelli- gence, 40(12):2935–2947, 2017

    2017

  15. [23]

    Zhun-ga Liu, Yi-min Fu, Quan Pan, and Zuo-wei Zhang. Orientational distribution learning with hierarchical spatial attention for open set recognition.IEEE Transactions on Pat- tern Analysis and Machine Intelligence, 45(7):8757–8772, 2022

    2022

  16. [24]

    Pmal: Open set recognition via robust prototype mining

    Jing Lu, Yunlu Xu, Hao Li, Zhanzhan Cheng, and Yi Niu. Pmal: Open set recognition via robust prototype mining. InProceedings of the AAAI Conference on Artificial Intel- ligence, volume 36, pages 1872–1880, 2022

    2022

  17. [25]

    Class anchor clustering: A loss for distance-based open set recognition

    Dimity Miller, Niko Sunderhauf, Michael Milford, and Feras Dayoub. Class anchor clustering: A loss for distance-based open set recognition. InProceedings of the IEEE/CVF Win- ter Conference on Applications of Computer Vision, pages 3570–3578, 2021

    2021

  18. [26]

    Open set learning with counterfactual images

    Lawrence Neal, Matthew Olson, Xiaoli Fern, Weng-Keen Wong, and Fuxin Li. Open set learning with counterfactual images. InProceedings of the European conference on com- puter vision (ECCV), pages 613–628, 2018

    2018

  19. [27]

    Prevalence of neural collapse during the terminal phase of deep learning training.Proceedings of the National Academy of Sciences, 117(40):24652–24663, 2020

    Vardan Papyan, XY Han, and David L Donoho. Prevalence of neural collapse during the terminal phase of deep learning training.Proceedings of the National Academy of Sciences, 117(40):24652–24663, 2020

    2020

  20. [28]

    Generative-discriminative feature representations for open-set recognition

    Pramuditha Perera, Vlad I Morariu, Rajiv Jain, Varun Man- junatha, Curtis Wigington, Vicente Ordonez, and Vishal M Patel. Generative-discriminative feature representations for open-set recognition. InProceedings of the IEEE/CVF con- ference on computer vision and pattern recog...

    2020

  21. [29]

    Class-incremental learn- ing with pre-allocated fixed classifiers

    Federico Pernici, Matteo Bruni, Claudio Baecchi, Francesco Turchini, and Alberto Del Bimbo. Class-incremental learn- ing with pre-allocated fixed classifiers. In2020 25th Inter- national Conference on Pattern Recognition (ICPR), pages 6259–6266. IEEE, 2021

    2021

  22. [30]

    icarl: Incremental classifier and representation learning

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

    2001

  23. [31]

    Toward open set recogni- tion.IEEE transactions on pattern analysis and machine intelligence, 35(7):1757–1772, 2012

    Walter J Scheirer, Anderson de Rezende Rocha, Archana Sapkota, and Terrance E Boult. Toward open set recogni- tion.IEEE transactions on pattern analysis and machine intelligence, 35(7):1757–1772, 2012

    2012

  24. [32]

    Prob- ability models for open set recognition.IEEE transactions on pattern analysis and machine intelligence, 36(11):2317– 2324, 2014

    Walter J Scheirer, Lalit P Jain, and Terrance E Boult. Prob- ability models for open set recognition.IEEE transactions on pattern analysis and machine intelligence, 36(11):2317– 2324, 2014

    2014

  25. [33]

    Mimicking the oracle: An initial phase decorrelation ap- proach for class incremental learning

    Yujun Shi, Kuangqi Zhou, Jian Liang, Zihang Jiang, Ji- ashi Feng, Philip HS Torr, Song Bai, and Vincent YF Tan. Mimicking the oracle: An initial phase decorrelation ap- proach for class incremental learning. InProceedings of the IEEE/CVF conference on computer vision and patte...

    2022

  26. [34]

    Always be dreaming: A new approach for data-free class-incremental learning

    James Smith, Yen-Chang Hsu, Jonathan Balloch, Yilin Shen, Hongxia Jin, and Zsolt Kira. Always be dreaming: A new approach for data-free class-incremental learning. InPro- ceedings of the IEEE/CVF international conference on com- puter vision, pages 9374–9384, 2021

    2021

  27. [35]

    Exemplar-free class incremen- tal learning via discriminative and comparable parallel one- class classifiers.Pattern Recognition, 140:109561, 2023

    Wenju Sun, Qingyong Li, Jing Zhang, Danyu Wang, Wen Wang, and YangLi-ao Geng. Exemplar-free class incremen- tal learning via discriminative and comparable parallel one- class classifiers.Pattern Recognition, 140:109561, 2023

    2023

  28. [36]

    Conditional gaussian distribution learning for open set recognition

    Xin Sun, Zhenning Yang, Chi Zhang, Keck-V oon Ling, and Guohao Peng. Conditional gaussian distribution learning for open set recognition. InProceedings of the IEEE/CVF con- ference on computer vision and pattern recognition, pages 13480–13489, 2020

    2020

  29. [37]

    Prototype fis- sion: Closing set for robust open-set semi-supervised learn- ing.arXiv preprint arXiv:2308.15575, 2023

    Xuwei Tan, Yi-Jie Huang, and Yaqian Li. Prototype fis- sion: Closing set for robust open-set semi-supervised learn- ing.arXiv preprint arXiv:2308.15575, 2023

    2023 arXiv

  30. [38]

    Open-set recognition: A good closed-set classifier is all you need? 2021

    Sagar Vaze, Kai Han, Andrea Vedaldi, and Andrew Zisser- man. Open-set recognition: A good closed-set classifier is all you need? 2021

    2021

  31. [39]

    Ex- ploring diverse representations for open set recognition

    Yu Wang, Junxian Mu, Pengfei Zhu, and Qinghua Hu. Ex- ploring diverse representations for open set recognition. In Proceedings of the AAAI Conference on Artificial Intelli- gence, volume 38, pages 5731–5739, 2024

    2024

  32. [40]

    Large scale incre- mental learning

    Yue Wu, Yinpeng Chen, Lijuan Wang, Yuancheng Ye, Zicheng Liu, Yandong Guo, and Yun Fu. Large scale incre- mental learning. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 374–382, 2019

    2019

  33. [41]

    Robust classification with convolutional proto- type learning

    Hong-Ming Yang, Xu-Yao Zhang, Fei Yin, and Cheng- Lin Liu. Robust classification with convolutional proto- type learning. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 3474–3482, 2018

    2018

  34. [42]

    Neural collapse inspired feature- classifier alignment for few-shot class incremental learning

    Yibo Yang, Haobo Yuan, Xiangtai Li, Zhouchen Lin, Philip Torr, and Dacheng Tao. Neural collapse inspired feature- classifier alignment for few-shot class incremental learning. arXiv preprint arXiv:2302.03004, 2023

    2023 arXiv

  35. [43]

    Adaptive deep models for incremental learning: Considering capacity scalability and sustainability

    Yang Yang, Da-Wei Zhou, De-Chuan Zhan, Hui Xiong, and Yuan Jiang. Adaptive deep models for incremental learning: Considering capacity scalability and sustainability. InPro- ceedings of the 25th ACM SIGKDD international conference on knowledge discovery & data mining, pages 74–82, 2019

    2019

  36. [44]

    Classification- reconstruction learning for open-set recognition

    Ryota Yoshihashi, Wen Shao, Rei Kawakami, Shaodi You, Makoto Iida, and Takeshi Naemura. Classification- reconstruction learning for open-set recognition. InProceed- ings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 4016–4025, 2019

    2019

  37. [45]

    Deep-learning- based open set fault diagnosis by extreme value theory

    Xiaolei Yu, Zhibin Zhao, Xingwu Zhang, Qiyang Zhang, Yi- long Liu, Chuang Sun, and Xuefeng Chen. Deep-learning- based open set fault diagnosis by extreme value theory. IEEE Transactions on Industrial Informatics, 18(1):185– 196, 2021

    2021

  38. [46]

    Contin- ual learning through synaptic intelligence

    Friedemann Zenke, Ben Poole, and Surya Ganguli. Contin- ual learning through synaptic intelligence. InInternational conference on machine learning, pages 3987–3995. PMLR, 2017

    2017

  39. [47]

    Maintaining discrimination and fairness in class incremental learning

    Bowen Zhao, Xi Xiao, Guojun Gan, Bin Zhang, and Shu- Tao Xia. Maintaining discrimination and fairness in class incremental learning. InProceedings of the IEEE/CVF con- ference on computer vision and pattern recognition, pages 13208–13217, 2020

    2020

  40. [48]

    Memory-efficient class-incremental learning for image classification.IEEE Transactions on Neural Networks and Learning Systems, 33(10):5966–5977, 2021

    Hanbin Zhao, Hui Wang, Yongjian Fu, Fei Wu, and Xi Li. Memory-efficient class-incremental learning for image classification.IEEE Transactions on Neural Networks and Learning Systems, 33(10):5966–5977, 2021

    2021

  41. [49]

    Class-incremental learning: A survey.IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024

    Da-Wei Zhou, Qi-Wei Wang, Zhi-Hong Qi, Han-Jia Ye, De- Chuan Zhan, and Ziwei Liu. Class-incremental learning: A survey.IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024

    2024

  42. [50]

    A model or 603 exemplars: Towards memory-efficient class-incremental learning.arXiv preprint arXiv:2205.13218, 2022

    Da-Wei Zhou, Qi-Wei Wang, Han-Jia Ye, and De- Chuan Zhan. A model or 603 exemplars: Towards memory-efficient class-incremental learning.arXiv preprint arXiv:2205.13218, 2022

    2022 arXiv

  43. [51]

    Learning placeholders for open-set recognition

    Da-Wei Zhou, Han-Jia Ye, and De-Chuan Zhan. Learning placeholders for open-set recognition. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 4401–4410, 2021

    2021

Pith tools

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