Pith. sign in

REVIEW 1 cited by

Generalized Category Discovery via Token Manifold Capacity Learning

T0 review · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read MTMC adds a nuclear-norm loss on class tokens to GCD objectives and reports small accuracy gains on several image benchmarks.

arxiv 2505.14044 v1 pith:OTIDCW5B submitted 2025-05-20 cs.LG

Luyao Tang , Kunze Huang , Chaoqi Chen , Cheng Chen This is my paper · ORCID
classification cs.LG
keywords mtmccapacitymanifoldcategoryclusteringlearningdatadiscovery
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

Generalized category discovery is a task where a model must group an unlabeled pool of images, some from classes it was trained on and some from entirely new classes. Existing methods mostly use contrastive or prototype losses to pull images of the same class into tight clusters. The authors argue that making clusters too tight throws away useful variety inside each class, and that this hurts the model's ability to separate classes later.

Their fix, MTMC, is a three-line regularizer. After a vision transformer produces a class token for every image in a batch, the method computes the singular values of the matrix formed by those class tokens, sums them, and subtracts that sum from the total loss. Maximizing the sum of singular values pushes the batch of class tokens to be more spread out and to use more dimensions of the feature space, which the paper calls increasing manifold capacity. The regularizer is added on top of standard GCD losses like SimGCD or CMS.

The paper reports that this simple addition improves clustering accuracy on several image datasets by roughly 0.1 to 4.7 percentage points, and often improves the estimated number of categories. However, no error bars or multiple seeds are given, some results are negative, and the theoretical analysis is mostly elementary inequalities plus an unproven concentration claim. Because the regularizer treats all samples in a batch alike, it is not obvious from the derivation that it specifically enriches each class rather than just spreading everything out.

Extended reading notes

Core claim

The paper's central claim is that maximizing the nuclear norm of the class-token matrix for unlabeled samples, via L_MTMC = -sum_r sigma_r([cls]_u) in Eq. (5), increases token manifold capacity, prevents dimensional collapse, and yields better clustering accuracy and category-number estimation in GCD. If correct, then adding this three-line regularizer to existing GCD methods (SimGCD, CMS, SPTNet, SelEx) should consistently improve their accuracy on coarse- and fine-grained benchmarks without architectural changes.

Load-bearing premise

The load-bearing premise is that maximizing the nuclear norm of a batch-level class-token matrix (Eq. 5, with [cls]_u mixing many different classes) is a valid proxy for increasing the manifold capacity of each individual class, and that this preserves intra-class representation completeness rather than merely pushing all samples apart on the hypersphere. This enters in Section 3.1 (Eq. 3) and Section 3.2 (Eq. 5), and is asserted without proof; the perfect-reconstruction/uniformity analysis in Appendix C only addresses global uniformity, not per-class completeness.

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.

Assumptions & free parameters 1 free parameters · 5 assumptions · 1 invented entities

The central claim rests on three assumptions: the class token is a valid sample centroid, the nuclear norm measures useful capacity, and maximizing it on a mixed mini-batch improves each class's representation. The second is definitional, the third is asserted, and the high-dimensional proof in Appendix C is incomplete.

free parameters (1)
  • loss weight lambda = not reported; tuned per experiment (Fig. 4)
    The scalar balancing L_MTMC against the base GCD loss. The paper notes insensitivity and explores it in Fig. 4, but the specific values used for Tables 1-4 are not given, so the reported gains depend on this choice.
assumptions (5)
  • domain assumption Manifold capacity theory (radius, dimensionality, capacity) from prior work applies to GCD representation quality.
    Section 2.2 imports the definitions without validating that the GCD embedding space fits the manifold capacity framework.
  • ad hoc to paper The ViT class token is the sample centroid, and its nuclear norm (CTME) is a valid proxy for the sample's manifold capacity.
    Section 3.1 introduces CTME=||[cls]||_* and asserts this equivalence; it is the core modeling assumption behind the loss.
  • ad hoc to paper Maximizing the nuclear norm of the batch-level class-token matrix improves intra-class representation completeness.
    Section 3.2 Eq. (5) defines the loss on the unlabeled mini-batch; the paper does not prove that a mixed-batch objective improves each class individually.
  • domain assumption Perfect reconstruction and perfect uniformity are the correct idealization of the GCD objective.
    Appendix C Definitions 1 and 2 are adapted from self-supervised learning; GCD has known and novel classes and a supervised labeled subset, which this idealization ignores.
  • standard math High-dimensional concentration yields the claimed lower bound on ||C||_* under perfect reconstruction and uniformity.
    Appendix C Theorem 1 invokes a high-probability bound without a complete proof; the argument is sketched.
invented entities (1)
  • Class Token Manifold Extent (CTME)
    purpose: A named quantity, defined as the nuclear norm of the class-token matrix, proposed as the objective to maximize for higher manifold capacity.
    CTME is a re-labeling of the nuclear norm; no independent empirical handle outside this paper is provided to show it captures intra-class completeness rather than generic spread.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Generalized Category Discovery via Token Manifold Capacity Learning." pith.science (2026). https://pith.science/paper/OTIDCW5B

@misc{pith2026250514044,
  author       = {Pith},
  title        = {Pith review of: Generalized Category Discovery via Token Manifold Capacity Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OTIDCW5B}},
  note         = {Machine review of arXiv:2505.14044}
}
read the original abstract

Generalized category discovery (GCD) is essential for improving deep learning models' robustness in open-world scenarios by clustering unlabeled data containing both known and novel categories. Traditional GCD methods focus on minimizing intra-cluster variations, often sacrificing manifold capacity, which limits the richness of intra-class representations. In this paper, we propose a novel approach, Maximum Token Manifold Capacity (MTMC), that prioritizes maximizing the manifold capacity of class tokens to preserve the diversity and complexity of data. MTMC leverages the nuclear norm of singular values as a measure of manifold capacity, ensuring that the representation of samples remains informative and well-structured. This method enhances the discriminability of clusters, allowing the model to capture detailed semantic features and avoid the loss of critical information during clustering. Through theoretical analysis and extensive experiments on coarse- and fine-grained datasets, we demonstrate that MTMC outperforms existing GCD methods, improving both clustering accuracy and the estimation of category numbers. The integration of MTMC leads to more complete representations, better inter-class separability, and a reduction in dimensional collapse, establishing MTMC as a vital component for robust open-world learning. Code is in github.com/lytang63/MTMC.

Figures

Figures reproduced from arXiv: 2505.14044 by the authors.

Figure 1
Figure 1. (a) GCD is constrained by dimensional col [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of Maximum Token Manifold Capacity. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Comparison between log(rank(A)) and Hˆ (A). The count of the largest eigenvalues necessary to account for 99% of the total eigenvalue energy serves as a surrogate for the rank. The autocorrelation matrix of the sample’s class to￾ken manifold is A ≜ PN i=1 1 N [cls]i[cls]⊤ i = CLS⊤CLS/N. We employ von Neumann en￾tropy [41, 2] to measure manifold capacity. This gives the advantage of focusing exclusively on the eigenv… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Hyperparameter sensitivity of the degree of MTMC [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: The Frobenius norm ∥A − c · Id∥ 2 F on three fine-grained benchmarks. 5 Hierarchical Analysis of Why MTMC is Effective in GCD We conduct a comprehensive analysis from multiple dimensions: 1) eigenvalue distribution and Frobenius norm, 2) estimation of embedded space di…
Figure 6
Figure 6. Figure 6: MTMC effectively mitigates dimensional collapse by providing a more uniform eigenvalue [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: Trends in the top 10 singular values as the number of training steps grows. [PITH_FULL_IMAGE:figures/full_fig_p017_7.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. SpectralGCD: Spectral Concept Selection and Cross-modal Representation Learning for Generalized Category Discovery

    cs.CV 2026-02 conditional novelty 6.0 of 10

    SpectralGCD represents images as mixtures over CLIP word-concept similarities, filters the concept dictionary by eigendecomposition of a teacher covariance matrix, and trains a student with forward/reverse distillatio...

Reference graph

Works this paper leans on

65 extracted references · 44 canonical work pages · cited by 1 Pith paper

  1. [1]

    VICReg: Variance-invariance-covariance regularization for self-supervised learning

    Adrien Bardes, Jean Ponce, and Yann LeCun. VICReg: Variance-invariance-covariance regularization for self-supervised learning. InInternational Conference on Learning Representations, 2022

    work page 2022

  2. [2]

    V on neumann entropy from unitarity.Physical review letters, 122(21):210402, 2019

    Paul Boes, Jens Eisert, Rodrigo Gallego, Markus P Müller, and Henrik Wilming. V on neumann entropy from unitarity.Physical review letters, 122(21):210402, 2019

    work page 2019

  3. [3]

    Open-world semi-supervised learning

    Kaidi Cao, Maria Brbic, and Jure Leskovec. Open-world semi-supervised learning. InInternational Conference on Learning Representations, 2022

    work page 2022

  4. [4]

    Unsu- pervised learning of visual features by contrasting cluster assignments.Advances in neural information processing systems, 33:9912–9924, 2020

    Mathilde Caron, Ishan Misra, Julien Mairal, Priya Goyal, Piotr Bojanowski, and Armand Joulin. Unsu- pervised learning of visual features by contrasting cluster assignments.Advances in neural information processing systems, 33:9912–9924, 2020

    work page 2020

  5. [5]

    Emerging properties in self-supervised vision transformers

    Mathilde Caron, Hugo Touvron, Ishan Misra, Hervé Jégou, Julien Mairal, Piotr Bojanowski, and Armand Joulin. Emerging properties in self-supervised vision transformers. InProceedings of the IEEE/CVF international conference on computer vision, pages 9650–9660, 2021

    2021

  6. [6]

    Towards mitigating dimensional collapse of representations in collaborative filtering

    Huiyuan Chen, Vivian Lai, Hongye Jin, Zhimeng Jiang, Mahashweta Das, and Xia Hu. Towards mitigating dimensional collapse of representations in collaborative filtering. InProceedings of the 17th ACM International Conference on Web Search and Data Mining, pages 106–115, 2024

    work page 2024

  7. [7]

    A simple framework for contrastive learning of visual representations

    Ting Chen, Simon Kornblith, Mohammad Norouzi, and Geoffrey Hinton. A simple framework for contrastive learning of visual representations. InInternational conference on machine learning, pages 1597–1607. PMLR, 2020

    2020

  8. [8]

    Parametric infor- mation maximization for generalized category discovery

    Florent Chiaroni, Jose Dolz, Ziko Imtiaz Masud, Amar Mitiche, and Ismail Ben Ayed. Parametric infor- mation maximization for generalized category discovery. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 1729–1739, 2023

    work page 2023

Show all 65 references
  1. [9]

    Utilizing class information for deep network representation shaping

    Daeyoung Choi and Wonjong Rhee. Utilizing class information for deep network representation shaping. InProceedings of the AAAI Conference on Artificial Intelligence, volume 33, pages 3396–3403, 2019

    2019

  2. [10]

    Contrastive mean-shift learning for generalized category discovery

    Sua Choi, Dahyun Kang, and Minsu Cho. Contrastive mean-shift learning for generalized category discovery. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 23094–23104, 2024

    2024

  3. [11]

    Reducing overfitting in deep networks by decorrelating representations.arXiv preprint arXiv:1511.06068, 2015

    Michael Cogswell, Faruk Ahmed, Ross Girshick, Larry Zitnick, and Dhruv Batra. Reducing overfitting in deep networks by decorrelating representations.arXiv preprint arXiv:1511.06068, 2015

    2015 arXiv

  4. [12]

    An image is worth 16x16 words: Transformers for image recognition at scale.arXiv preprint arXiv:2010.11929, 2020

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

    2010 arXiv

  5. [13]

    Derivations for linear algebra and optimization.Berkeley, California, 3(1):2325–5870, 2007

    John Duchi. Derivations for linear algebra and optimization.Berkeley, California, 3(1):2325–5870, 2007

    2007

  6. [14]

    A unified objective for novel class discovery

    Enrico Fini, Enver Sangineto, Stéphane Lathuilière, Zhun Zhong, Moin Nabi, and Elisa Ricci. A unified objective for novel class discovery. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 9284–9292, 2021

    2021

  7. [15]

    The role of entropy and reconstruction in multi-view self-supervised learning

    Borja Rodrıguez Gálvez, Arno Blaas, Pau Rodríguez, Adam Golinski, Xavier Suau, Jason Ramapuram, Dan Busbridge, and Luca Zappella. The role of entropy and reconstruction in multi-view self-supervised learning. InInternational Conference on Machine Learning, pages 29143–29160. P...

    2023

  8. [16]

    Wichmann, and Wieland Brendel

    Robert Geirhos, Patricia Rubisch, Claudio Michaelis, Matthias Bethge, Felix A. Wichmann, and Wieland Brendel. Imagenet-trained CNNs are biased towards texture; increasing shape bias improves accuracy and robustness. InInternational Conference on Learning Representations, 2019

    2019

  9. [17]

    Recent advances in open set recognition: A survey.IEEE transactions on pattern analysis and machine intelligence, 43(10):3614–3631, 2020

    Chuanxing Geng, Sheng-jun Huang, and Songcan Chen. Recent advances in open set recognition: A survey.IEEE transactions on pattern analysis and machine intelligence, 43(10):3614–3631, 2020

    2020

  10. [18]

    Singular value decomposition and least squares solutions

    Gene H Golub and Christian Reinsch. Singular value decomposition and least squares solutions. In Handbook for Automatic Computation: Volume II: Linear Algebra, pages 134–151. Springer, 1971

    1971

  11. [19]

    Bootstrap your own latent-a new approach to self-supervised learning.Advances in neural information processing systems, 33:21271–21284, 2020

    Jean-Bastien Grill, Florian Strub, Florent Altché, Corentin Tallec, Pierre Richemond, Elena Buchatskaya, Carl Doersch, Bernardo Avila Pires, Zhaohan Guo, Mohammad Gheshlaghi Azar, et al. Bootstrap your own latent-a new approach to self-supervised learning.Advances in neural in...

    2020

  12. [20]

    Pre-training with random orthogonal projection image modeling.arXiv preprint arXiv:2310.18737, 2023

    Maryam Haghighat, Peyman Moghadam, Shaheer Mohamed, and Piotr Koniusz. Pre-training with random orthogonal projection image modeling.arXiv preprint arXiv:2310.18737, 2023

    2023 arXiv

  13. [21]

    Au- tomatically discovering and learning new visual categories with ranking statistics.arXiv preprint arXiv:2002.05714, 2020

    Kai Han, Sylvestre-Alvise Rebuffi, Sebastien Ehrhardt, Andrea Vedaldi, and Andrew Zisserman. Au- tomatically discovering and learning new visual categories with ranking statistics.arXiv preprint arXiv:2002.05714, 2020

    2002 arXiv

  14. [22]

    Learning to discover novel visual categories via deep transfer clustering

    Kai Han, Andrea Vedaldi, and Andrew Zisserman. Learning to discover novel visual categories via deep transfer clustering. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 8401–8409, 2019

    2019

  15. [23]

    On feature decorrelation in self-supervised learning

    Tianyu Hua, Wenxiao Wang, Zihui Xue, Sucheng Ren, Yue Wang, and Hang Zhao. On feature decorrelation in self-supervised learning. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 9598–9608, 2021

    2021

  16. [24]

    An information-theoretic understanding of maximum manifold capacity representations

    Berivan Isik, Victor Lecomte, Rylan Schaeffer, Yann LeCun, Mikail Khona, Ravid Shwartz-Ziv, Sanmi Koyejo, and Andrey Gromov. An information-theoretic understanding of maximum manifold capacity representations. InUniReps: the First Workshop on Unifying Representations in Neural...

    2023

  17. [25]

    Understanding dimensional collapse in contrastive self-supervised learning.arXiv preprint arXiv:2110.09348, 2021

    Li Jing, Pascal Vincent, Yann LeCun, and Yuandong Tian. Understanding dimensional collapse in contrastive self-supervised learning.arXiv preprint arXiv:2110.09348, 2021

    2021 arXiv

  18. [26]

    Supervised contrastive learning.Advances in neural information processing systems, 33:18661–18673, 2020

    Prannay Khosla, Piotr Teterwak, Chen Wang, Aaron Sarna, Yonglong Tian, Phillip Isola, Aaron Maschinot, Ce Liu, and Dilip Krishnan. Supervised contrastive learning.Advances in neural information processing systems, 33:18661–18673, 2020

    2020

  19. [27]

    3d object representations for fine-grained categorization

    Jonathan Krause, Michael Stark, Jia Deng, and Li Fei-Fei. 3d object representations for fine-grained categorization. InProceedings of the IEEE international conference on computer vision workshops, pages 554–561, 2013

    2013

  20. [28]

    Cifar-10 dataset

    Alex Krizhevsky and Geoffrey Hinton. Cifar-10 dataset. https://www.cs.toronto.edu/~kriz/ cifar.html, 2009. Accessed: 2025-05-20

    2009

  21. [29]

    The hungarian method for the assignment problem.Naval research logistics quarterly, 2(1-2):83–97, 1955

    Harold W Kuhn. The hungarian method for the assignment problem.Naval research logistics quarterly, 2(1-2):83–97, 1955

    1955

  22. [30]

    Sphereface: Deep hypersphere embedding for face recognition

    Weiyang Liu, Yandong Wen, Zhiding Yu, Ming Li, Bhiksha Raj, and Le Song. Sphereface: Deep hypersphere embedding for face recognition. InProceedings of the IEEE conference on computer vision and pattern recognition, pages 212–220, 2017

    2017

  23. [31]

    A frobenius norm approach to glottal closure detection from the speech signal.IEEE Transactions on Speech and Audio Processing, 2(2):258–265, 1994

    Changxue Ma, Yves Kamp, and Lei F Willems. A frobenius norm approach to glottal closure detection from the speech signal.IEEE Transactions on Speech and Audio Processing, 2(2):258–265, 1994

    1994

  24. [32]

    Protogcd: Unified and unbiased prototype learning for generalized category discovery.IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025

    Shijie Ma, Fei Zhu, Xu-Yao Zhang, and Cheng-Lin Liu. Protogcd: Unified and unbiased prototype learning for generalized category discovery.IEEE Transactions on Pattern Analysis and Machine Intelligence, 2025

    2025

  25. [33]

    Happy: A debiased learning framework for continual generalized category discovery.arXiv preprint arXiv:2410.06535, 2024

    Shijie Ma, Fei Zhu, Zhun Zhong, Wenzhuo Liu, Xu-Yao Zhang, and Cheng-Lin Liu. Happy: A debiased learning framework for continual generalized category discovery.arXiv preprint arXiv:2410.06535, 2024

    2024 arXiv

  26. [34]

    Active generalized category discovery

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

    2024

  27. [35]

    Fine-grained visual classification of aircraft.arXiv preprint arXiv:1306.5151, 2013

    Subhransu Maji, Esa Rahtu, Juho Kannala, Matthew Blaschko, and Andrea Vedaldi. Fine-grained visual classification of aircraft.arXiv preprint arXiv:1306.5151, 2013

    2013 arXiv

  28. [36]

    Inequalities: Theory of majorization and its applications, 1979

    AW Marshall. Inequalities: Theory of majorization and its applications, 1979

    1979

  29. [37]

    Hyperspherical prototype networks.Advances in neural information processing systems, 32, 2019

    Pascal Mettes, Elise Van der Pol, and Cees Snoek. Hyperspherical prototype networks.Advances in neural information processing systems, 32, 2019

    2019

  30. [38]

    Self-supervised learning with an information maximization criterion

    Serdar Ozsoy, Shadi Hamdan, Sercan Arik, Deniz Yuret, and Alper Erdogan. Self-supervised learning with an information maximization criterion. In S. Koyejo, S. Mohamed, A. Agarwal, D. Belgrave, K. Cho, and A. Oh, editors,Advances in Neural Information Processing Systems, volume...

    2022

  31. [39]

    Deep face recognition

    Omkar Parkhi, Andrea Vedaldi, and Andrew Zisserman. Deep face recognition. InBMVC 2015-Proceedings of the British Machine Vision Conference 2015. British Machine Vision Association, 2015. 11

    2015

  32. [40]

    Connections between nuclear-norm and frobenius- norm-based representations.IEEE transactions on neural networks and learning systems, 29(1):218–224, 2016

    Xi Peng, Canyi Lu, Zhang Yi, and Huajin Tang. Connections between nuclear-norm and frobenius- norm-based representations.IEEE transactions on neural networks and learning systems, 29(1):218–224, 2016

    2016

  33. [41]

    Entropy, von neumann and the von neumann entropy: Dedicated to the memory of alfred wehrl

    Dénes Petz. Entropy, von neumann and the von neumann entropy: Dedicated to the memory of alfred wehrl. InJohn von Neumann and the foundations of quantum physics, pages 83–96. Springer, 2001

    2001

  34. [42]

    Dynamic conceptional contrastive learning for generalized category discovery

    Nan Pu, Zhun Zhong, and Nicu Sebe. Dynamic conceptional contrastive learning for generalized category discovery. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 7579–7588, 2023

    2023

  35. [43]

    Machine learning: Algorithms, real-world applications and research directions.SN computer science, 2(3):160, 2021

    Iqbal H Sarker. Machine learning: Algorithms, real-world applications and research directions.SN computer science, 2(3):160, 2021

    2021

  36. [44]

    Towards an improved understanding and utilization of maximum manifold capacity representations.arXiv preprint arXiv:2406.09366, 2024

    Rylan Schaeffer, Victor Lecomte, Dhruv Bhandarkar Pai, Andres Carranza, Berivan Isik, Alyssa Unell, Mikail Khona, Thomas Yerxa, Yann LeCun, SueYeon Chung, et al. Towards an improved understanding and utilization of maximum manifold capacity representations.arXiv preprint arXiv...

    2024 arXiv

  37. [45]

    Understanding and mitigating dimensional collapse in federated learning.IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023

    Yujun Shi, Jian Liang, Wenqing Zhang, Chuhui Xue, Vincent YF Tan, and Song Bai. Understanding and mitigating dimensional collapse in federated learning.IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023

    2023

  38. [46]

    Manifold clustering

    Richard Souvenir and Robert Pless. Manifold clustering. InTenth IEEE International Conference on Computer Vision (ICCV’05) Volume 1, volume 1, pages 648–653. IEEE, 2005

    2005

  39. [47]

    The herbarium challenge 2019 dataset.arXiv preprint arXiv:1906.05372, 2019

    Kiat Chuan Tan, Yulong Liu, Barbara Ambrose, Melissa Tulig, and Serge Belongie. The herbarium challenge 2019 dataset.arXiv preprint arXiv:1906.05372, 2019

    2019 arXiv

  40. [48]

    Breaking the curse of dimensional collapse in graph contrastive learning: A whitening perspective.Information Sciences, 657:119952, 2024

    Yang Tao, Kai Guo, Yizhen Zheng, Shirui Pan, Xiaofeng Cao, and Yi Chang. Breaking the curse of dimensional collapse in graph contrastive learning: A whitening perspective.Information Sciences, 657:119952, 2024

    2024

  41. [49]

    Wiley-Interscience, 2006

    MTCAJ Thomas and A Thomas Joy.Elements of information theory. Wiley-Interscience, 2006

    2006

  42. [50]

    Generalized category discovery

    Sagar Vaze, Kai Han, Andrea Vedaldi, and Andrew Zisserman. Generalized category discovery. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 7492–7501, 2022

    2022

  43. [51]

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

    Sagar Vaze, Kai Han, Andrea Vedaldi, and Andrew Zisserman. Open-set recognition: A good closed-set classifier is all you need. InInternational Conference on Learning Representations, 2022

    2022

  44. [52]

    Cambridge university press, 2018

    Roman Vershynin.High-dimensional probability: An introduction with applications in data science, volume 47. Cambridge university press, 2018

    2018

  45. [53]

    C. Wah, N. Rasiwasia, D. Hsu, J. Yao, L. Li, and G. Mori. Caltech-ucsd birds 200-2011 (cub-200-2011). http://www.vision.caltech.edu/visipedia/CUB-200.html, 2011. Accessed: 2025-05-20

    2011

  46. [54]

    Understanding contrastive representation learning through alignment and uniformity on the hypersphere

    Tongzhou Wang and Phillip Isola. Understanding contrastive representation learning through alignment and uniformity on the hypersphere. InInternational conference on machine learning, pages 9929–9939. PMLR, 2020

    2020

  47. [55]

    Hierarchical grouping to optimize an objective function.Journal of the American statistical association, 58(301):236–244, 1963

    Joe H Ward Jr. Hierarchical grouping to optimize an objective function.Journal of the American statistical association, 58(301):236–244, 1963

    1963

  48. [56]

    A survey of transfer learning.Journal of Big data, 3:1–40, 2016

    Karl Weiss, Taghi M Khoshgoftaar, and DingDing Wang. A survey of transfer learning.Journal of Big data, 3:1–40, 2016

    2016

  49. [57]

    Parametric classification for generalized category discovery: A baseline study

    Xin Wen, Bingchen Zhao, and Xiaojuan Qi. Parametric classification for generalized category discovery: A baseline study. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 16590–16600, 2023

    2023

  50. [58]

    Towards open vocabulary learning: A survey.IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024

    Jianzong Wu, Xiangtai Li, Shilin Xu, Haobo Yuan, Henghui Ding, Yibo Yang, Xia Li, Jiangning Zhang, Yunhai Tong, Xudong Jiang, et al. Towards open vocabulary learning: A survey.IEEE Transactions on Pattern Analysis and Machine Intelligence, 2024

    2024

  51. [59]

    Regularizing deep convolutional neural networks with a structured decorrelation constraint

    Wei Xiong, Bo Du, Lefei Zhang, Ruimin Hu, and Dacheng Tao. Regularizing deep convolutional neural networks with a structured decorrelation constraint. In2016 IEEE 16th international conference on data mining (ICDM), pages 519–528. IEEE, 2016. 12

    2016

  52. [60]

    Learning efficient coding of natural images with maximum manifold capacity representations.Advances in Neural Information Processing Systems, 36:24103–24128, 2023

    Thomas Yerxa, Yilun Kuang, Eero Simoncelli, and SueYeon Chung. Learning efficient coding of natural images with maximum manifold capacity representations.Advances in Neural Information Processing Systems, 36:24103–24128, 2023

    2023

  53. [61]

    Barlow twins: Self-supervised learning via redundancy reduction

    Jure Zbontar, Li Jing, Ishan Misra, Yann LeCun, and Stéphane Deny. Barlow twins: Self-supervised learning via redundancy reduction. InInternational conference on machine learning, pages 12310–12320. PMLR, 2021

    2021

  54. [62]

    Promptcal: Contrastive affinity learning via auxiliary prompts for generalized novel category discovery

    Sheng Zhang, Salman Khan, Zhiqiang Shen, Muzammal Naseer, Guangyi Chen, and Fahad Shahbaz Khan. Promptcal: Contrastive affinity learning via auxiliary prompts for generalized novel category discovery. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Rec...

    2023

  55. [63]

    Geometric view of soft decorrelation in self-supervised learning

    Yifei Zhang, Hao Zhu, Zixing Song, Yankai Chen, Xinyu Fu, Ziqiao Meng, Piotr Koniusz, and Irwin King. Geometric view of soft decorrelation in self-supervised learning. InProceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 4338–4349, 2024

    2024

  56. [64]

    Learning semi-supervised gaussian mixture models for generalized category discovery

    Bingchen Zhao, Xin Wen, and Kai Han. Learning semi-supervised gaussian mixture models for generalized category discovery. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 16623–16633, 2023

    2023

  57. [65]

    Domain generalization: A survey

    Kaiyang Zhou, Ziwei Liu, Yu Qiao, Tao Xiang, and Chen Change Loy. Domain generalization: A survey. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45(4):4396–4415, 2022. 13 A Details of optimization objective of GCD The existing GCD proposals are all proposed f...

    2022

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.