Pith. sign in

REVIEW 3 major objections 6 minor 49 references

Graph Cut-guided Maximal Coding Rate Reduction for Learning Image Embedding and Clustering

T0 review · 3 major / 6 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read By folding a differentiable graph-cut loss into the maximal coding rate reduction objective, the paper jointly learns embeddings and cluster assignments, reaching 97.7% clustering accuracy on CIFAR-10 and 67.5% on ImageNet-1k.

desk verdict A well-executed clustering paper whose empirical results are strong, but whose principled joint-optimization claim rests on an unproven soft-label extension of MCR2 theory. read the letter →

arxiv 2412.18930 v2 pith:CYU7YQMX submitted 2024-12-25 cs.CV

W. He , Z. Huang , X. Meng , X. Qi , R. Xiao , C.-G. Li This is my paper · ORCID
classification cs.CV
keywords imageclusteringmaximalcodingratereductionnormalizedcutstructuredrepresentationlearningpre-trainedvisionmodelsdeepGumbel-SoftmaxCLIPfeatures
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 proposes CgMCR2, a single objective that trains a feature head and a cluster head at the same time. The objective combines the maximal coding rate reduction (MCR2) loss, which pushes embeddings into a union of low-dimensional subspaces, with a differentiable normalized-cut loss that asks the cluster head for a clean partition. The two terms are coupled: the partition guides which regions of the embedding space get compressed, and the structured embeddings supply the affinity matrix used by the graph cut. On CLIP features the method reaches 97.7% clustering accuracy on CIFAR-10, 77.8% on CIFAR-100, 72.9% on TinyImageNet and 67.5% on ImageNet-1k, and it also works from MoCo-v2 features trained from scratch. The point of the paper is that clustering and representation learning should be one joint problem, not two sequential stages.

What carries the argument

The load-bearing object is the joint objective in Eq. (8), which couples two existing principles. Maximal coding rate reduction supplies $R(Z;\epsilon)-R_c(Z,\Pi;\epsilon)$: expanding the whole embedding while compressing each cluster into a low-dimensional subspace. A relaxed normalized-cut loss $L_{\mathrm{Ncut}}(\Pi;A,\gamma)$, with affinity $A=P_s(Z^\top Z)$ (sparsified cosine similarity) and volume-normalized membership, supplies a differentiable clustering signal. A Gumbel-Softmax cluster head reparameterizes the soft partition, and a two-stage schedule first warms up with $-R+L_{\mathrm{Ncut}}$ and then fine-tunes with the full objective. The affinity is detached from backpropagation so the graph-cut term trains the cluster head without distorting the feature head through the graph.

What would settle it

Run the same two-stage CgMCR2 procedure but replace the soft membership in $R_c$ with a hard one-hot assignment taken from the argmax of the cluster head, with gradients stopped or passed through straight-through. If the hard-membership version matches or beats the soft version, then the soft relaxation is not what carries the result. Alternatively, train only with $-R+L_{\mathrm{Ncut}}$ and then evaluate with the rate-reduction term frozen so that the partition cannot influence the embeddings; if this performs as well as the full joint training, the claimed mutual guidance between partition and embedding is not load-bearing.

Watch

Extended reading notes

Core claim

The paper's central claim is that the joint objective $$ \min_{\Theta,\Phi}\,-R(Z_\Theta;\epsilon)+R_c(Z_\Theta,\Pi_\Phi;\epsilon)+L_{\mathrm{Ncut}}(\Pi_\Phi;A,\gamma) $$ turns clustering and structured representation learning into a single differentiable problem. The rate-reduction terms make the feature-head output $Z$ a union of low-dimensional orthogonal subspaces, while the graph-cut term makes the cluster-head output $\Pi$ a soft partition whose affinity $A$ is built from $Z$. Because every term is differentiable, both heads can be trained end-to-end with gradient descent, the cluster head can label unseen test points directly, and the learned membership actively guides which subspaces are compressed. The paper reports that this joint training outperforms prior MCR2-based clustering, classical clustering methods, and recent deep clustering methods on most tested datasets.

Load-bearing premise

The central assumption is that a soft, probabilistic cluster-membership matrix produced by Gumbel-Softmax can stand in for the hard 0/1 assignment that the rate-reduction theory assumes; the paper offers no derivation or citation for this relaxation, and the joint objective depends on it.

Editorial extensions

If this is right

  • If the central claim holds, deep image clustering can be trained end-to-end as one objective, removing the separate pseudo-label phase used by SCAN-style pipelines.
  • The cluster head becomes a label predictor for unseen data, so no extra k-means or spectral step is needed at test time.
  • On frozen CLIP features the method reaches 97.7% ACC on CIFAR-10, 77.8% on CIFAR-100, 72.9% on TinyImageNet and 67.5% on ImageNet-1k, beating the CLIP-based MCR2 baseline CPP in every setting.
  • On MoCo-v2 features, CgMCR2 improves CIFAR-10 ACC to 92.8, ahead of the prior MCR2-based MLC (86.3) and other deep clustering baselines.
  • The two-stage warm-up is important: ablations show that fine-tuning the full objective without the warm-up drops CIFAR-100 ACC from 77.8 to 66.0.
  • The neural cluster head is about 100 times faster per training iteration than replacing it with k-means, spectral clustering, or EnSC, while achieving comparable or better accuracy.

Reading between the lines

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

  • The same coupling could be applied to a trainable encoder: nothing in the objective requires the backbone to be frozen, so end-to-end fine-tuning of CLIP or a vision transformer is a direct next test.
  • Straight-through hard assignment during fine-tuning would isolate whether the soft Gumbel-Softmax membership is actually providing the cluster-wise compression the paper claims, since the theoretical guarantees in the cited MCR2 work are stated for hard partitions.
  • The mini-batch affinity could be stabilized with a memory bank or an exponential moving average; this would likely matter at larger scales where the current top-s sparsified batch affinity is a noisy graph.
  • Because the normalized-cut loss only sees the current batch's affinity, an interesting transfer test is node clustering on graph data, where the affinity is given and the same joint objective could be applied directly.
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 / 6 minor

Summary. The paper proposes CgMCR2, a unified framework for jointly learning structured image embeddings and cluster memberships by combining the Maximal Coding Rate Reduction (MCR2) objective with a differentiable relaxation of normalized cut. The method uses a frozen pre-trained encoder (CLIP or MoCo-v2), a feature head that produces embeddings, and a cluster head with Gumbel-Softmax output that yields soft memberships. Training is two-stage: a warm-up with a simplified objective (negative rate plus normalized-cut loss), followed by fine-tuning with the full objective in Eq. (8). Experiments on five standard datasets and four out-of-domain datasets report state-of-the-art clustering accuracy, with ablation studies on the output activation, affinity construction, post-processing, hyperparameters, model size, and training strategy. The code is publicly available.

Significance. If the central objective is well-founded, the paper makes a useful contribution: it provides a scalable, differentiable, and unified treatment of representation learning and clustering, with strong empirical results (e.g., 97.7% ACC on CIFAR-10, 77.8% on CIFAR-100, 72.9% on TinyImageNet, and 67.5% on ImageNet-1k using CLIP features) and a broad set of ablations. The release of code and the detailed hyperparameter tables are commendable. However, the theoretical grounding of the soft-label rate-reduction term is currently missing, and a core component of the clustering loss is attributed to an unpublished manuscript; both issues need to be addressed before the claimed 'principled' joint optimization can be accepted.

major comments (3)
  1. [Section 3.2, Eq. (8)] The rate term Rc(ZTheta, Pi_Phi; epsilon) is used with soft Gumbel-Softmax memberships Pi_Phi, but the MCR2 formulation in Eq. (1) and the theory in [48] are defined for a hard 0/1 partition, with N_l the cardinality of class l. The paper does not define how N_l is computed from soft probabilities, nor does it justify that the intended within-class compression and between-class discrimination properties survive the relaxation. This is load-bearing: in Algorithm 1 the affinity A is detached, so Rc is the only term that couples the feature head Theta and the cluster head Phi during fine-tuning; the claimed principled joint optimization rests entirely on this unexamined step. Please provide the precise definition of Rc for soft memberships and either a derivation or a citation to a published analysis, or explicitly state that the rate term is used heuristically.
  2. [Section 3.2, Eq. (6) and gamma selection] The relaxation of the normalized-cut problem and the practical rule for choosing the feasible range of gamma are attributed to the authors' own unpublished manuscript [16] ("Submitted to Pattern Recognition"). For a journal publication, this core component must be self-contained: include the derivation of the penalty-based relaxation and the gamma-selection procedure, or cite a published source. As written, a central part of the proposed objective is not verifiable from the public literature, which hinders reproducibility and assessment of novelty.
  3. [Algorithm 1, line 7 and Section 3.2] The affinity A is explicitly detached from backpropagation, so the normalized-cut loss L_Ncut provides no direct gradient to the feature head Z_Theta. The only pathway from the clustering module to the representation during fine-tuning is through the rate-reduction term Rc(Z_Theta, Pi_Phi; epsilon). The paper should state this explicitly and discuss how the 'joint learning' claim is affected, especially in view of the missing soft-label definition for Rc. If the authors intend L_Ncut to refine the embeddings, they should either remove the detachment or explain why the indirect pathway through Pi_Phi and Rc is sufficient.
minor comments (6)
  1. [Section 2] The heading 'Relate Work' should be 'Related Work'.
  2. [Section 3.2, after Eq. (6)] The sentence 'a softmax function is used is used for the output' contains a duplicated 'is used' and should be corrected.
  3. [Section 3.3 and Table A.3] The Gumbel-Softmax temperature is not reported in the paper or in the hyperparameter table; this parameter controls the hardness of the memberships and should be stated for reproducibility.
  4. [Section 5] The conclusion states that experiments were conducted on 'five benchmark datasets', but the paper uses nine datasets (five standard and four out-of-domain); please correct the count.
  5. [Algorithm 1 and Eq. (7)] Eq. (7) defines the volume matrix V using sums over i = 1 to N, while Algorithm 1 operates on mini-batches of size n; please clarify whether V is computed on the mini-batch or the full dataset, since this affects the normalization in the normalized-cut loss.
  6. [Table 4] In the MoCo-v2 comparison, NMCE, MLC, and IMC-SwAV use their own specially designed pre-trained models, while CgMCR2 uses MoCo-v2 features; the text mentions this but should emphasize the limited comparability of these baselines.

Circularity Check

0 steps flagged · score 2.0 of 10

No circular derivation: the joint objective is an explicit construction, not an identity with the reported metrics; minor self-citation to [16] is noted.

full rationale

No circular step is established. Eq. (8) is a proposed objective, not a quantity derived from the ACC/NMI numbers; ground-truth labels appear only in evaluation (Section 4, Metrics), so the reported results are external benchmarks rather than fitted outputs. The feature and cluster heads are coupled through an iterative two-stage procedure (Eq. (9) then Eq. (8)), with the affinity A detached in Algorithm 1; this is a self-training loop, not a definitional identity. The paper does rely on the authors' own unpublished submission [16] for the differentiable normalized-cut relaxation in Eq. (6) and for choosing gamma in footnote 6; this is a minor self-citation and a support concern, but the relaxation is written out explicitly and the framework is benchmarked against independent baselines, so the central claim does not reduce to the citation. The soft Gumbel-Softmax use of Rc in Eq. (8) is an unproved extension of the hard-partition MCR2 theory in [48], but that is a validity gap, not circularity: no equation in the paper makes Rc equal to the reported accuracy by construction.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The method depends on several tuned hyperparameters and a set of modeling assumptions about soft labels, mini-batch graph cuts, and an unpublished normalized-cut relaxation. No new physical entities are introduced.

free parameters (5)
  • Rate distortion precision epsilon = 0.2 to 0.5 per dataset
    Sets the coding-rate scale in Eq. (1) and Eq. (8); chosen per dataset in Table A.3.
  • Normalized cut trade-off gamma = 50 to 50000 per dataset
    Weights the orthogonality penalty in the relaxed Ncut loss; tuned per dataset in Table A.3.
  • Affinity sparsity s = 3 to 50 per dataset
    Controls how many similar entries are kept per row of A; affects the graph structure used by the Ncut loss.
  • Feature head dimension d = 128 or 256
    Output dimension of the learned embedding Z; fixed per dataset in Table A.3.
  • Batch size, learning rate, weight decay, epochs = per-dataset values in Table A.3
    Training schedule hyperparameters selected per dataset; they affect convergence and final accuracy.
assumptions (3)
  • ad hoc to paper MCR2 rate reduction with soft cluster memberships retains the intended within-class compression and between-class discrimination.
    Eq. (8) replaces the hard-label Pi* in MCR2 [48] with Gumbel-Softmax outputs Pi_Phi, but no proof or reference is given that the coding-rate objective is valid for soft assignments.
  • domain assumption The relaxed normalized-cut objective in Eq. (6) from [16] has the properties needed to learn cluster memberships directly from a neural network.
    The method relies on the unpublished differentiable Ncut formulation of [16]; its convergence and constraint satisfaction properties are assumed.
  • domain assumption Mini-batch affinity graphs provide adequate estimates of the full-data normalized cut and rate reduction objectives.
    Eq. (8) is written with N by N matrices, but Algorithm 1 computes A and L_Ncut on n by n mini-batches; the paper does not analyze this discrepancy.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Graph Cut-guided Maximal Coding Rate Reduction for Learning Image Embedding and Clustering." pith.science (2026). https://pith.science/paper/CYU7YQMX

@misc{pith2026241218930,
  author       = {Pith},
  title        = {Pith review of: Graph Cut-guided Maximal Coding Rate Reduction for Learning Image Embedding and Clustering},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CYU7YQMX}},
  note         = {Machine review of arXiv:2412.18930}
}
abstract

In the era of pre-trained models, image clustering task is usually addressed by two relevant stages: a) to produce features from pre-trained vision models; and b) to find clusters from the pre-trained features. However, these two stages are often considered separately or learned by different paradigms, leading to suboptimal clustering performance. In this paper, we propose a unified framework, termed graph Cut-guided Maximal Coding Rate Reduction (CgMCR$^2$), for jointly learning the structured embeddings and the clustering. To be specific, we attempt to integrate an efficient clustering module into the principled framework for learning structured representation, in which the clustering module is used to provide partition information to guide the cluster-wise compression and the learned embeddings is aligned to desired geometric structures in turn to help for yielding more accurate partitions. We conduct extensive experiments on both standard and out-of-domain image datasets and experimental results validate the effectiveness of our approach.

Figures

Figures reproduced from arXiv: 2412.18930 by the authors.

Figure 1
Figure 1. Illustration for our CgMCR2 architecture. We illustrate the forward pass (in black ) and the gradient dependency of ZΘ (in green ) and ΠΦ (in orange ) in different colored lines individually. For clarity, we exclude the parameters of the pre-feature layer from the visualization. 3 Our Approach: Graph Cut-guided Maximal Coding Rate Reduction (CgMCR2 ) We begin with a brief review of the principle of MCR2 in Section 3… view at source ↗
Figure 2
Figure 2. Similarity matrices ordered by the ground-truth labels [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figure 3
Figure 3. Learning curves of each loss term, ACC, and NMI during training. that, the initialization stage did perform a good warm-up, and during the fine￾tuning stage, the value of R(ZΘ; ϵ) remains nearly constant, but Rc(ZΘ, ΠΦ; ϵ) is rapidly optimized toward to its minimum. Moreover, the curves of ACC and NMI improve rapidly and achieves the optimal clustering results with the self￾supervision of well-initialized cluster me… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Effect of hyper-parameters on CIFAR-10 (left) and CIFAR-100 (right). 1 2 3 4 # hidden layer(s) 512 1024 2048 hidden dimension 4096 97.0% 97.5% 96.0 96.5 97.0 97.5 98.0 ACC (%) 1 2 3 4 # hidden layer(s) 512 1024 2048 hidden dimension 4096 76.5% 77.5% 74 75 76 77 78 ACC …
Figure 5
Figure 5. Figure 5: Effect of model parameters on CIFAR-10 (left) and CIFAR-100 (right). clustering algorithms to guide the feature learning still yield very competitive clustering accuracy, in most cases. Nevertheless, the time cost of using conven￾tional clustering algorithms is 100× ex…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

49 extracted references · 41 canonical work pages

  1. [16]

    Submitted to Pattern Recognition (2024)

    He, W., Zhang, S., Li, C.G., Qi, X., Xiao, R., Guo, J.: Neural normalized cut: A dif- ferential and generalizable approach for spectral clustering. Submitted to Pattern Recognition (2024)

    work page 2024

  2. [48]

    In: Advances in Neural Information Processing Systems

    Yu, Y., Chan, K.H.R., You, C., Song, C., Ma, Y.: Learning diverse and discrim- inative representations via the principle of maximal coding rate reduction. In: Advances in Neural Information Processing Systems. pp. 9422–9434 (2020)

    work page 2020

  3. [1]

    In: British Machine Vision Conference

    Adaloglou, N., Michels, F., Kalisch, H., Kollmann, M.: Exploring the limits of deep image clustering using pretrained models. In: British Machine Vision Conference. pp. 297–299. BMVA Press (2023)

    work page 2023

  4. [2]

    In: International Conference on Learning Rep- resentations (2022)

    Bardes, A., Ponce, J., LeCun, Y.: Vicreg: Variance-invariance-covariance regular- ization for self-supervised learning. In: International Conference on Learning Rep- resentations (2022)

    work page 2022

  5. [3]

    IEEE Trans

    Bruna, J., Mallat, S.: Invariant scattering convolution networks. IEEE Trans. Pat- tern Anal. Mach. Intell.35(8), 1872–1886 (2013)

    work page 2013

  6. [4]

    In: IEEE/CVF In- ternational Conference on Computer Vision

    Caron, M., Touvron, H., Misra, I., Jégou, H., Mairal, J., Bojanowski, P., Joulin, A.: Emerging properties in self-supervised vision transformers. In: IEEE/CVF In- ternational Conference on Computer Vision. pp. 9630–9640. IEEE (2021)

    work page 2021

  7. [5]

    In: International Conference on Machine Learning

    Chen, T., Kornblith, S., Norouzi, M., Hinton, G.E.: A simple framework for con- trastive learning of visual representations. In: International Conference on Machine Learning. vol. 119, pp. 1597–1607. PMLR (2020)

    work page 2020

  8. [6]

    arXiv preprint arXiv:2003.04297 (2020)

    Chen, X., Fan, H., Girshick, R., He, K.: Improved baselines with momentum con- trastive learning. arXiv preprint arXiv:2003.04297 (2020)

    arXiv 2020

Show all 49 references
  1. [7]

    In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition

    Chen, Y., Li, C.G., You, C.: Stochastic sparse subspace clustering. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. pp. 4155– 4164 (2020)

    2020

  2. [8]

    In: International Conference on Learning Representations (2024)

    Chu, T., Tong, S., Ding, T., Dai, X., Haeffele, B.D., Vidal, R., Ma, Y.: Image clustering via the principle of rate reduction in the age of pretrained models. In: International Conference on Learning Representations (2024)

    2024

  3. [9]

    In: IEEE conference on computer vision and pattern recognition

    Deng, J., Dong, W., Socher, R., Li, L.J., Li, K., Fei-Fei, L.: Imagenet: A large-scale hierarchical image database. In: IEEE conference on computer vision and pattern recognition. pp. 248–255 (2009)

    2009

  4. [10]

    In: Burstein, J., Doran, C., Solorio, T

    Devlin, J., Chang, M., Lee, K., Toutanova, K.: BERT: pre-training of deep bidi- rectional transformers for language understanding. In: Burstein, J., Doran, C., Solorio, T. (eds.) Proceedings of the Conference of the North American Chapter of the Association for Computational L...

    2019

  5. [11]

    In: IEEE/CVF International Conference on Computer Vision

    Ding, T., Tong, S., Chan, K.H.R., Dai, X., Ma, Y., Haeffele, B.D.: Unsupervised manifold linearizing and clustering. In: IEEE/CVF International Conference on Computer Vision. pp. 5427–5438 (2023)

    2023

  6. [12]

    In: Proceedings of the International Conference on Learning Representations (2021)

    Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., Uszkoreit, J., Houlsby, N.: An image is worth 16x16 words: Transformers for image recognition at scale. In: Proceedings of the Internatio...

    2021

  7. [13]

    In: Advances in Neural Information Processing Systems (2020)

    Grill, J., Strub, F., Altché, F., Tallec, C., Richemond, P.H., Buchatskaya, E., Do- ersch, C., Pires, B.Á., Guo, Z., Azar, M.G., Piot, B., Kavukcuoglu, K., Munos, R., Valko, M.: Bootstrap your own latent - A new approach to self-supervised learning. In: Advances in Neural Info...

    2020

  8. [14]

    In: IEEE/CVF Conference on Computer Vision and Pattern Recognition

    He, K., Chen, X., Xie, S., Li, Y., Dollár, P., Girshick, R.B.: Masked autoencoders are scalable vision learners. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 15979–15988 (2022) 16 W. He, Z. Huang, X. Meng, X. Qi, R. Xiao, C.-G. Li

    2022

  9. [15]

    In: IEEE/CVF Conference on Computer Vision and Pattern Recognition

    He, K., Fan, H., Wu, Y., Xie, S., Girshick, R.B.: Momentum contrast for unsu- pervised visual representation learning. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 9726–9735 (2020)

    2020

  10. [17]

    IEEE Transactions on Pattern Analysis and Machine Intelligence45(6), 7509–7524 (2022)

    Huang, Z., Chen, J., Zhang, J., Shan, H.: Learning representation for clustering via prototype scattering and positive sampling. IEEE Transactions on Pattern Analysis and Machine Intelligence45(6), 7509–7524 (2022)

    2022

  11. [18]

    ACM Computing Surveys 31(3), 264–323 (1999)

    Jain, A.K., Murty, M.N., Flynn, P.J.: Data clustering: a review. ACM Computing Surveys 31(3), 264–323 (1999)

    1999

  12. [19]

    In: 5th International Conference on Learning Representations (2017)

    Jang, E., Gu, S., Poole, B.: Categorical reparameterization with gumbel-softmax. In: 5th International Conference on Learning Representations (2017)

    2017

  13. [20]

    Khosla, A., Jayadevaprakash, N., Yao, B., Fei-Fei, L.: Novel dataset for fine-grained imagecategorization.In:IEEEConferenceonComputerVisionandPatternRecog- nition (2011)

    2011

  14. [21]

    Kingma, D., Ba, J.: Adam: A method for stochastic optimization. In: Int. Conf. Learn. Represent. (2014)

    2014

  15. [22]

    Kingma,D.P.,Welling,M.:Auto-encodingvariationalbayes.In:InternationalCon- ference on Learning Representations (2014)

    2014

  16. [23]

    Technical Report TR-2009, University of Toronto, Toronto (2009)

    Krizhevsky, A., Hinton, G., et al.: Learning multiple layers of features from tiny images. Technical Report TR-2009, University of Toronto, Toronto (2009)

    2009

  17. [24]

    Naval research logistics quarterly2(1-2), 83–97 (1955)

    Kuhn, H.W.: The hungarian method for the assignment problem. Naval research logistics quarterly2(1-2), 83–97 (1955)

    1955

  18. [25]

    Proceedings of the IEEE86(11), 2278–2324 (1998)

    LeCun, Y., Bottou, L., Bengio, Y., Haffner, P., et al.: Gradient-based learning ap- plied to document recognition. Proceedings of the IEEE86(11), 2278–2324 (1998)

    1998

  19. [26]

    In: Proceedings of the AAAI conference on artificial intelligence (2021)

    Li, Y., Hu, P., Liu, Z., Peng, D., Zhou, J.T., Peng, X.: Contrastive clustering. In: Proceedings of the AAAI conference on artificial intelligence (2021)

    2021

  20. [27]

    In: Proceedings of the International Conference on Machine Learning (2024)

    Li, Y., Hu, P., Peng, D., Lv, J., Fan, J., Peng, X.: Image clustering with external guidance. In: Proceedings of the International Conference on Machine Learning (2024)

    2024

  21. [28]

    arXiv preprint arXiv:2201.10000 (2022)

    Li, Z., Chen, Y., LeCun, Y., Sommer, F.T.: Neural manifold clustering and em- bedding. arXiv preprint arXiv:2201.10000 (2022)

    2022 arXiv

  22. [29]

    arXiv preprint arXiv:2011.14859 (2020)

    Lim, D., Vidal, R., Haeffele, B.D.: Doubly stochastic subspace clustering. arXiv preprint arXiv:2011.14859 (2020)

    2020 arXiv

  23. [30]

    Loshchilov, I., Hutter, F.: SGDR: stochastic gradient descent with warm restarts. In: Int. Conf. Learn. Represent. (2017)

    2017

  24. [31]

    Statistics and Computing17(4), 395–416 (2007)

    von Luxburg, U.: A tutorial on spectral clustering. Statistics and Computing17(4), 395–416 (2007)

    2007

  25. [32]

    In: Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability

    MacQueen,J.:Somemethodsforclassificationandanalysisofmultivariateobserva- tions. In: Proceedings of the Fifth Berkeley Symposium on Mathematical Statistics and Probability. pp. 281–297 (1967)

    1967

  26. [33]

    Nene, S.A., Nayar, S.K., Murase, H.: Columbia object image library (coil-100). Tech. Rep. CUCS-006-96, Department of Computer Science, Columbia University (February 1996)

    1996

  27. [34]

    In: Sixth Indian Conference on Computer Vision, Graphics & Image Processing

    Nilsback, M., Zisserman, A.: Automated flower classification over a large number of classes. In: Sixth Indian Conference on Computer Vision, Graphics & Image Processing. pp. 722–729. IEEE Computer Society (2008)

    2008

  28. [35]

    IEEE Transactions on Image Processing31, 7264–7278 (2022) Graph Cut-guided Maximal Coding Rate Reduction 17

    Niu,C.,Shan, H.,Wang, G.:SPICE: semanticpseudo-labelingfor imageclustering. IEEE Transactions on Image Processing31, 7264–7278 (2022) Graph Cut-guided Maximal Coding Rate Reduction 17

    2022

  29. [36]

    Knowledge-Based Systems250, 109042 (2022)

    Ntelemis,F.,Jin,Y.,Thomas,S.A.:Informationmaximizationclusteringviamulti- view self-labelling. Knowledge-Based Systems250, 109042 (2022)

    2022

  30. [37]

    arXiv preprint arXiv:2304.07193 (2023)

    Oquab, M., Darcet, T., Moutakanni, T., Vo, H., Szafraniec, M., Khalidov, V., Fernandez, P., Haziza, D., Massa, F., El-Nouby, A., et al.: Dinov2: Learning robust visual features without supervision. arXiv preprint arXiv:2304.07193 (2023)

    2023 arXiv

  31. [38]

    In: IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Park, S., Han, S., Kim, S., Kim, D., Park, S., Hong, S., Cha, M.: Improving un- supervised image clustering with robust learning. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 12278–12287 (2021)

    2021

  32. [39]

    In: International Conference on Machine Learning

    Radford, A., Kim, J.W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., Krueger, G., Sutskever, I.: Learning transferable visual models from natural language supervision. In: International Conference on Machine Learning. vol. 139,...

    2021

  33. [40]

    Parallel Distributed Processing pp

    Rumelhart, D.E., Hinton, G.E., Williams, R.J., et al.: Learning internal represen- tations by error propagation. Parallel Distributed Processing pp. 318–362 (1986)

    1986

  34. [41]

    IEEE Transactions on Pattern Analysis and Machine Intelligence22(8), 888–905 (2000)

    Shi, J., Malik, J.: Normalized cuts and image segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence22(8), 888–905 (2000)

    2000

  35. [42]

    In: IEEE/CVF International Confer- ence on Computer Vision

    Souvenir, R., Pless, R.: Manifold clustering. In: IEEE/CVF International Confer- ence on Computer Vision. pp. 648–653 (2005)

    2005

  36. [43]

    In: Proceedings of the International Conference on Learning Rep- resentations (2021)

    Tsai, T.W., Li, C., Zhu, J.: Mice: Mixture of contrastive experts for unsupervised image clustering. In: Proceedings of the International Conference on Learning Rep- resentations (2021)

    2021

  37. [44]

    In: European Conference on Computer Vision

    Van Gansbeke, W., Vandenhende, S., Georgoulis, S., Proesmans, M., Van Gool, L.: Scan: Learning to classify images without labels. In: European Conference on Computer Vision. pp. 268–285 (2020)

    2020

  38. [45]

    IEEE Signal Processing Magazine 28(3), 52–68 (March 2011)

    Vidal, R.: Subspace clustering. IEEE Signal Processing Magazine 28(3), 52–68 (March 2011)

    2011

  39. [46]

    arXiv preprint arXiv: 1708.07747 (2019)

    Xiao, H., Rasul, K., Vollgraf, R.: Fashion-mnist: a novel image dataset for bench- marking machine learning algorithms. arXiv preprint arXiv: 1708.07747 (2019)

    2019 arXiv

  40. [47]

    In: IEEE Conference on Computer Vision and Pattern Recognition

    You, C., Li, C.G., Robinson, D.P., Vidal, R.: Oracle based active set algorithm for scalable elastic net subspace clustering. In: IEEE Conference on Computer Vision and Pattern Recognition. pp. 3928–3937 (2016)

    2016

  41. [49]

    lr” and “wd

    Zhong,H.,Wu,J.,Chen,C.,Huang,J.,Deng,M.,Nie,L.,Lin,Z.,Hua,X.S.:Graph contrastive clustering. In: Proceedings of the IEEE/CVF international conference on computer vision. pp. 9224–9233 (2021) 18 W. He, Z. Huang, X. Meng, X. Qi, R. Xiao, C.-G. Li Supplementary Materials A Experi...

    2021

Pith tools

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