Pith. sign in

REVIEW 2 major objections 5 minor 49 references

Self-supervised Transformation Learning for Equivariant Representations

T0 review · 2 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read This paper proposes replacing transformation labels with learned, image-invariant transformation representations, and reports gains on 7 of 11 classification benchmarks and on object detection.

desk verdict STL is a genuinely label-free step for equivariant SSL that deserves review, but the image-invariance claim is under-tested and the main results lack error bars. read the letter →

arxiv 2501.08712 v1 pith:5XIO6KD5 submitted 2025-01-15 cs.CV cs.AIcs.LG

Jaemyung Yu , Jaehyun Choi , Dong-Jae Lee , HyeongGwon Hong , Junmo Kim This is my paper · ORCID
classification cs.CVcs.AIcs.LG
keywords self-supervisedlearningequivariantrepresentationstransformationrepresentationcontrastiveimagetransformationstransfer
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

Equivariant representation learning usually needs explicit transformation labels, which forces each transformation to be treated independently and rules out complex, composable transformations like AugMix. This paper proposes Self-supervised Transformation Learning (STL), which replaces the label with a transformation representation computed from the pair of an original image and its transformed version. STL trains that representation to be image-invariant by contrastively aligning the same transformation applied to different images, then uses it to parameterize the equivariant transformation in representation space. The authors show this improves transfer performance over label-based equivariant baselines on 7 of 11 classification benchmarks and on object detection, and that it can use complex transformations that existing equivariant baselines cannot use. The practical payoff is that equivariant learning no longer depends on knowing the transformation group in advance.

What carries the argument

The central object is the transformation representation $y_x^t = f_T(f(x), f(t(x)))$, produced by an auxiliary encoder $f_T$ from the representation pair of an image and its transformed version. Its defining property is image-invariance: $y_x^t = y_{x'}^t$ for the same $t$, enforced by the self-supervised transformation loss $L_{\mathrm{trans}}$ using InfoNCE. The second mechanism is the equivariant transformation $\phi(y_{x'}^t, f(x))$, a hypernetwork-parameterized map that applies a transformation code obtained from another image $x'$ to the representation of $x$, so that the equivariant branch cannot cheat by copying the input pair. Together they replace discrete transformation labels with continuous codes that can represent composed transformations such as AugMix.

What would settle it

Train STL with transformations seen during pretraining, then probe held-out images and transformations: if the transformation representation $y_x^t$ changes substantially when the same $t$ is applied to different images, so that intra-transformation variance across images exceeds inter-transformation separation, the image-invariance premise fails. A direct test is to measure whether $\phi(y_{x'}^t, f(x))$ aligns with $f(t(x))$ when $x$ and $x'$ are drawn from a distribution very different from the pretraining data; if alignment error grows sharply, the equivariant branch is not actually transferring transformations.

Watch

Extended reading notes

Core claim

The central claim is that a learned vector $y_x^t = f_T(f(x), f(t(x)))$ can serve as a continuous, label-free code for the transformation $t$ itself, provided it is trained to be invariant to the choice of image $x$. STL enforces this with a contrastive loss that aligns $y_x^t$ and $y_{x'}^t$ for identical transformations applied to different images, so the representation encodes the transformation rather than the image content. The equivariant branch then applies $\phi(y_{x'}^t, f(x))$ to a different image $x$, avoiding the trivial solution where the equivariant map would just copy the representation it saw, and aligns the result with $f(t(x))$. On ImageNet100-pretrained ResNet-50, STL reaches 64.18% mean accuracy over 11 downstream classification tasks and 51.95 AP on VOC07+12 detection, outperforming label-based equivariant baselines while keeping batch construction at the same complexity as SimCLR.

Load-bearing premise

The approach rests on the assumption that a transformation representation computed from one image pair actually encodes the transformation independently of the image content, and that contrastively aligning same-transformation pairs across different images is enough to make that true in practice.

Editorial extensions

If this is right

  • Equivariant learning can be applied to transformations that have no discrete label space, such as AugMix, because the transformation is represented by a learned vector rather than a category or parameter vector.
  • Transformation representations organize by type and intensity in the learned space, so similar color transformations sit close together and intensity levels order continuously, giving a structured transformation geometry.
  • The same batch configuration as standard contrastive learning suffices, with roughly 10% higher per-iteration cost, so the method can be dropped into existing pipelines without enlarged batches.
  • STL improves transfer performance across the base models SimCLR, BYOL, SimSiam, and Barlow Twins, meaning the transformation-learning module is a portable addition rather than a redesign.
  • Object detection also benefits, with STL reaching 51.95 AP_all and 52.70 with AugMix, so the equivariant signal helps downstream tasks that demand precise localization.

Reading between the lines

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

  • The image-invariance hypothesis suggests a continuous transformation algebra: if $y_t$ encodes $t$ independently of $x$, operations like composition and interpolation of transformations could be performed in representation space and probed with arithmetic tests, which the paper does not run.
  • The paper's limitation statement implies that multi-image transformations such as mixup are out of reach; a natural extension would be an encoder over sets of representations that produces a transformation code from more than two images.
  • The same contrastive-alignment idea could transfer to other modalities where transformations are hard to label, such as audio pitch shifting or time stretching, though the paper only evaluates images.
  • Because transformation representations are continuous, they could be used as goals in reinforcement learning or as supervisory signals for controllable generation, a use the paper does not explore.
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

2 major / 5 minor

Summary. The paper introduces Self-supervised Transformation Learning (STL), a method for learning equivariant representations without transformation labels. STL uses an auxiliary encoder f_T that maps a pair of representations (f(x), f(t(x))) to a transformation representation y_x^t. The training objective combines three InfoNCE losses: an invariant loss (Eq. 11) aligning f(x) and f(t(x)); an equivariant loss (Eq. 12) aligning phi(y_{x'}^t, f(x)) with f(t(x)), using a transformation representation from a different image x' to avoid trivial solutions; and a transformation alignment loss (Eq. 13) that pulls y_x^t and y_{x'}^t together for the same transformation t on different images. The paper reports linear transfer results on 11 downstream datasets, object detection on VOC, transformation prediction quality, and ablations, and claims improvements over existing equivariant SSL methods in 7 of 11 classification tasks.

Significance. The core idea is timely and potentially useful: replacing discrete transformation labels with learned continuous transformation representations could handle composite transformations such as AugMix and capture interdependencies among simple transformations, which label-based equivariant methods cannot. The paper provides a clean formulation, an aligned-batch implementation that keeps computational overhead at about 1.11x, compatibility with several base SSL models, and a set of ablations that show each loss term contributes. The code is released. The main reservations are that the key image-invariance property (Eq. 8) is not directly tested and that the ImageNet100 benchmark tables lack variance information; both concerns are addressable with additional experiments.

major comments (2)
  1. [§4.1, Tables 2–4] The main ImageNet100 results are reported as single accuracy numbers without standard deviations or seed counts, while the STL10 results in Table 11 include mean±std over three seeds. The headline claim of outperforming existing methods in 7 of 11 datasets is sensitive to small differences: for example, in Table 2 STL improves over AugSelf on CIFAR100 by 0.75 points (66.84 vs 66.09), while it is lower on MIT67 (56.64 vs 57.16) and SUN397 (44.69 vs 45.24), and the mean improvement over SimCLR with AugMix is only 0.09 points (64.18 vs 64.09). Without variance information, the 7-of-11 claim cannot be distinguished from run-to-run noise. The authors should report mean±std over at least three seeds for the ImageNet100 experiments, or explicitly state the number of seeds, and discuss whether the reported differences are significant.
  2. [§3.2, Eqs. (8), (12), (13); Table 7] The image-invariance of the transformation representation is the load-bearing assumption of the method, but it is not directly verified. The loss in Eq. (13) aligns y_x^t and y_{x'}^t only for same-transformation pairs sampled in the aligned batch; because f_T also sees f(x), nothing in the InfoNCE objective prevents y_x^t from encoding image identity or collapsing to a constant. The claim in the abstract that the method 'ensures transformation representation is image-invariant' is therefore not established by the reported experiments. The ablation in Table 7 shows that adding Ltrans improves transformation classification from 69.57 to 93.67, but this demonstrates the importance of the loss term, not that the learned representation satisfies Eq. (8) on unseen images. I ask the authors to add a direct invariance test, such as the variance of y_x^t across many images x for a fixed t, or the accuracy of the equivariant transfer Eq. (6) when y is drawn from held-out images, compared with an image-specific or random baseline. If Eq. (8) fails on unseen images, the equivariant objective in Eq. (12) no longer aligns with true transformations, and the reported gains could reflect an auxiliary regularization effect rather than the proposed mechanism.
minor comments (5)
  1. [Appendix B, Table 10] ImageNet100 is listed as having 1,000 classes; the ImageNet100 split used in this paper (from [42]) contains 100 classes. This is a typo but should be corrected.
  2. [Appendix C] The sentence 'we usetrainval set for training' is missing a space, and 'pre-taining' in Appendix B should be 'pretraining'.
  3. [§3.3, Eq. (10)] The InfoNCE denominator is written as a sum over y_i ≠ y, but the negative set should be explicitly the other batch elements; the shorthand used in Eqs. (11)–(13) is not defined until later. Please clarify the notation.
  4. [§3.3, Figure 3] For stochastic transformations such as random crop and color jitter, it is not specified how 'identical transformations' are synchronized across paired images (e.g., shared crop coordinates and jitter parameters). This detail matters for the correctness of Ltrans and Lequi and should be stated explicitly, ideally with a reference to the code.
  5. [§4.3, Table 7] The ablation study reports only single runs for the STL10-pretrained model; while Table 11 gives mean±std for the full STL, the ablations would be more convincing with repeated seeds.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: the transformation representation is trained by contrastive alignment on image pairs, and all central benchmark claims are evaluated on held-out downstream tasks.

full rationale

The derivation chain is self-contained against external benchmarks. The transformation representation y_x^t is defined from image pairs (Eq. 5) and trained with the contrastive objective (Eqs. 9 and 13); the same-transformation pairing is constructed by applying identical transformations t to two images, so t is known to the data pipeline. This is a standard use of augmentation identity to form positives and does not make the learned representation a regression target or a label-conditioned prediction. The equivariant objective (Eqs. 7 and 12) is trained to align phi(y_{x'}^t, f(x)) with f(t(x)), and its evaluation in Table 6 uses held-out transformations; downstream transfer results (Tables 2-4, 8-9) are independent of transformation labels. The only self-referential element is the semantic overstatement that the method "replaces" transformation labels, when the pipeline still uses t to define positive pairs; this is a wording caveat, not a circular reduction. The image-invariance hypothesis (Eq. 8) is an empirical assumption that could fail, but the paper tests it on test images (Table 5, Figures 1 and 4), and failure would be a robustness issue, not circularity. No load-bearing self-citations or imported uniqueness theorems appear.

Assumptions & free parameters 3 free parameters · 3 assumptions · 1 invented entities

The method introduces one key latent object, the transformation representation, and relies on two modeling assumptions about what contrastive learning can enforce. The free parameters are the loss weights, temperatures, and the chosen 128-dimensional representation size; none of these is fitted to the benchmark labels, but their values are not justified by sensitivity analysis.

free parameters (3)
  • lambda_inv, lambda_equi, lambda_trans = 1, 1, 0.2
    Loss-balancing weights in Eq. (14); chosen by hand, no sensitivity analysis reported in the main text.
  • tau_inv, tau_equi, tau_trans = not reported; 0.2 used for base SimCLR in Appendix E
    Temperature parameters in the InfoNCE losses, Eqs. (11) through (13); not individually specified for STL runs.
  • transformation representation dimension = 128
    Dimension of y_x^t in Eq. (5); set to 128 by design, no ablation reported.
assumptions (3)
  • domain assumption Transformation representation f_T(f(x), f(t(x))) encodes t independently of image content (image-invariance hypothesis).
    Stated as a hypothesis in Section 3.2 before Eq. (8); the entire method depends on this being learnable via contrastive loss. No formal proof is given.
  • domain assumption InfoNCE contrastive alignment between same-transformation pairs from different images is sufficient to enforce image-invariance and prevent collapse.
    Eq. (9) and Section 3.3; standard contrastive assumption, but no collapse or invariance analysis is provided for the transformation space.
  • ad hoc to paper Equivariant transformation networks trained with a transformation vector from a different image transfer to the target image.
    Eq. (6) uses y_x'^t with x' not equal to x to avoid a trivial solution; this cross-image transfer is learned implicitly and assumed to generalize, but not proven.
invented entities (1)
  • Transformation representation y_x^t
    purpose: Continuous learned encoding of a transformation t, used in place of transformation labels to condition the equivariant transformation network.
    The quantity is internal to the trained model. Evidence, such as Table 5, UMAP, and Table 6, is measured within the same paper; no external falsifiable prediction of a physical quantity is provided.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Self-supervised Transformation Learning for Equivariant Representations." pith.science (2026). https://pith.science/paper/5XIO6KD5

@misc{pith2026250108712,
  author       = {Pith},
  title        = {Pith review of: Self-supervised Transformation Learning for Equivariant Representations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5XIO6KD5}},
  note         = {Machine review of arXiv:2501.08712}
}
read the original abstract

Unsupervised representation learning has significantly advanced various machine learning tasks. In the computer vision domain, state-of-the-art approaches utilize transformations like random crop and color jitter to achieve invariant representations, embedding semantically the same inputs despite transformations. However, this can degrade performance in tasks requiring precise features, such as localization or flower classification. To address this, recent research incorporates equivariant representation learning, which captures transformation-sensitive information. However, current methods depend on transformation labels and thus struggle with interdependency and complex transformations. We propose Self-supervised Transformation Learning (STL), replacing transformation labels with transformation representations derived from image pairs. The proposed method ensures transformation representation is image-invariant and learns corresponding equivariant transformations, enhancing performance without increased batch complexity. We demonstrate the approach's effectiveness across diverse classification and detection tasks, outperforming existing methods in 7 out of 11 benchmarks and excelling in detection. By integrating complex transformations like AugMix, unusable by prior equivariant methods, this approach enhances performance across tasks, underscoring its adaptability and resilience. Additionally, its compatibility with various base models highlights its flexibility and broad applicability. The code is available at https://github.com/jaemyung-u/stl.

Figures

Figures reproduced from arXiv: 2501.08712 by the authors.

Figure 1
Figure 1. Visualization of Equivariant Transformation and Transformation Representation. (Left) UMAP [32] visualizations of functional weights from equivariant transformations implemented with a hypernetwork. EquiMod uses transformation labels to generate these weights, while STL derives them from the representation pairs of transformed and original image. (Right) UMAP visualizations of transformation representations obtained… view at source ↗
Figure 2
Figure 2. Transformation Equivariant Learning with Self-supervised Transformation Learning. (Left) The overall framework of STL. For given image and transformations, it demonstrates: 1) transformation invariant learning, which aligns the representations of image and transformed image; 2) transformation equivariant learning, where the representation of image transformed by an equivariant transformation (obtained from the trans… view at source ↗
Figure 3
Figure 3. Aligned Transformed Batch. (Left) In self-supervised learning methods, batch compositions typically involve applying two different transformations to each input image. (Right) In STL, batches are composed by pairing two images together, and applying the same transformation pair. 3.3 Implementation Details Dissimilarity Metric. We use the InfoNCE loss from SimCLR [4] for the formulation and imple￾mentation of STL. Ou… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Visualization of Transfor￾mation Representations by Intensity. UMAP visualization of transformation representations organized by intensity levels for each transformation type, in￾cluding random crop and color jitter variations in brightness, contrast, satura￾tion, and …
Figure 5
Figure 5. Figure 5: Explicit and Implicit Equivariant Learning. Transformation equivariant learning with transforma￾tion labels is divided into (Left) explicit and (Right) implicit equivariant learning [PITH_FULL_IMAGE:figures/full_fig_p017_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

49 extracted references · 24 canonical work pages

  1. [1]

    Bardes, J

    A. Bardes, J. Ponce, and Y . LeCun. Variance-invariance-covariance regularization for self-supervised learning. ICLR, Vicreg, 1:2, 2022

    work page 2022

  2. [2]

    Bossard, M

    L. Bossard, M. Guillaumin, and L. Van Gool. Food-101–mining discriminative components with random forests. In Computer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part VI 13, pages 446–461. Springer, 2014

    work page 2014

  3. [3]

    Caron, I

    M. Caron, I. Misra, J. Mairal, P. Goyal, P. Bojanowski, and A. Joulin. Unsupervised learning of visual features by contrasting cluster assignments. Advances in neural information processing systems , 33: 9912–9924, 2020

    work page 2020

  4. [4]

    T. Chen, S. Kornblith, M. Norouzi, and G. Hinton. A simple framework for contrastive learning of visual representations. In International conference on machine learning, pages 1597–1607. PMLR, 2020

    2020

  5. [5]

    Chen and K

    X. Chen and K. He. Exploring simple siamese representation learning. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 15750–15758, 2021

    work page 2021

  6. [6]

    Cimpoi, S

    M. Cimpoi, S. Maji, I. Kokkinos, S. Mohamed, and A. Vedaldi. Describing textures in the wild. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 3606–3613, 2014

    2014

  7. [7]

    Coates, A

    A. Coates, A. Ng, and H. Lee. An analysis of single-layer networks in unsupervised feature learning. In Proceedings of the fourteenth international conference on artificial intelligence and statistics, pages 215–223. JMLR Workshop and Conference Proceedings, 2011

    2011

  8. [8]

    Dangovski, L

    R. Dangovski, L. Jing, C. Loh, S. Han, A. Srivastava, B. Cheung, P. Agrawal, and M. Soljaˇci´c. Equivariant contrastive learning. arXiv preprint arXiv:2111.00899, 2021

    arXiv 2021

Show all 49 references
  1. [9]

    Devillers and M

    A. Devillers and M. Lefort. Equimod: An equivariance module to improve self-supervised learning. arXiv preprint arXiv:2211.01244, 2022

    2022 arXiv

  2. [10]

    Doersch, A

    C. Doersch, A. Gupta, and A. A. Efros. Unsupervised visual representation learning by context prediction. In Proceedings of the IEEE international conference on computer vision, pages 1422–1430, 2015

    2015

  3. [11]

    Everingham, L

    M. Everingham, L. Van Gool, C. K. Williams, J. Winn, and A. Zisserman. The pascal visual object classes (voc) challenge. International journal of computer vision, 88:303–338, 2010

    2010

  4. [12]

    Fei-Fei, R

    L. Fei-Fei, R. Fergus, and P. Perona. Learning generative visual models from few training examples: An incremental bayesian approach tested on 101 object categories. In 2004 conference on computer vision and pattern recognition workshop, pages 178–178. IEEE, 2004

    2004

  5. [13]

    Garrido, L

    Q. Garrido, L. Najman, and Y . Lecun. Self-supervised learning of split invariant equivariant representations. arXiv preprint arXiv:2302.10283, 2023

    2023 arXiv

  6. [14]

    T. Geng, T. Wang, Y . Zhang, J. Duan, W. Guan, and F. Zheng. Uniav: Unified audio-visual perception for multi-task video localization. arXiv preprint arXiv:2404.03179, 2024

    2024

  7. [15]

    Gidaris, P

    S. Gidaris, P. Singh, and N. Komodakis. Unsupervised representation learning by predicting image rotations. arXiv preprint arXiv:1803.07728, 2018

    2018 arXiv

  8. [16]

    Girshick

    R. Girshick. Fast r-cnn. In Proceedings of the IEEE international conference on computer vision, pages 1440–1448, 2015

    2015

  9. [17]

    Goyal, P

    P. Goyal, P. Dollár, R. Girshick, P. Noordhuis, L. Wesolowski, A. Kyrola, A. Tulloch, Y . Jia, and K. He. Accurate, large minibatch sgd: Training imagenet in 1 hour. arXiv preprint arXiv:1706.02677, 2017

    2017 arXiv

  10. [18]

    Grill, F

    J.-B. Grill, F. Strub, F. Altché, C. Tallec, P. Richemond, E. Buchatskaya, C. Doersch, B. Avila Pires, Z. Guo, M. Gheshlaghi Azar, et al. Bootstrap your own latent-a new approach to self-supervised learning. Advances in neural information processing systems, 33:21271–21284, 2020

    2020

  11. [19]

    X. Guo, E. Zhu, X. Liu, and J. Yin. Affine equivariant autoencoder. In IJCAI, pages 2413–2419, 2019

    2019

  12. [20]

    Gupta, J

    S. Gupta, J. Robinson, D. Lim, S. Villar, and S. Jegelka. Structuring representation geometry with rotationally equivariant contrastive learning. arXiv preprint arXiv:2306.13924, 2023

    2023 arXiv

  13. [21]

    Hashemi, Y

    A. Hashemi, Y . Feng, and H. Sabet. Spherical cnn for medical imaging applications: Importance of equivariance in image reconstruction and denoising. ArXiv, 2023. 11

    2023

  14. [22]

    K. He, X. Zhang, S. Ren, and J. Sun. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 770–778, 2016

    2016

  15. [23]

    Hendrycks, N

    D. Hendrycks, N. Mu, E. D. Cubuk, B. Zoph, J. Gilmer, and B. Lakshminarayanan. Augmix: A simple data processing method to improve robustness and uncertainty. arXiv preprint arXiv:1912.02781, 2019

    1912 arXiv

  16. [24]

    Ioffe and C

    S. Ioffe and C. Szegedy. Batch normalization: Accelerating deep network training by reducing internal covariate shift. In International conference on machine learning, pages 448–456. pmlr, 2015

    2015

  17. [25]

    Kornblith, J

    S. Kornblith, J. Shlens, and Q. V . Le. Do better imagenet models transfer better? InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 2661–2671, 2019

    2019

  18. [26]

    Krause, M

    J. Krause, M. Stark, J. Deng, and L. Fei-Fei. 3d object representations for fine-grained categorization. In Proceedings of the IEEE international conference on computer vision workshops, pages 554–561, 2013

    2013

  19. [27]

    Krizhevsky, G

    A. Krizhevsky, G. Hinton, et al. Learning multiple layers of features from tiny images. 2009

    2009

  20. [28]

    H. Lee, K. Lee, K. Lee, H. Lee, and J. Shin. Improving transferability of representations via augmentation- aware self-supervision. Advances in Neural Information Processing Systems, 34:17710–17722, 2021

    2021

  21. [29]

    J. Lee, J. Kim, H. Shon, B. Kim, S. H. Kim, H. Lee, and J. Kim. Uniclip: Unified framework for contrastive language-image pre-training. Advances in Neural Information Processing Systems, 35:1008–1019, 2022

    2022

  22. [30]

    Loshchilov and F

    I. Loshchilov and F. Hutter. Sgdr: Stochastic gradient descent with warm restarts. arXiv preprint arXiv:1608.03983, 2016

    2016 arXiv

  23. [31]

    S. Maji, E. Rahtu, J. Kannala, M. Blaschko, and A. Vedaldi. Fine-grained visual classification of aircraft. arXiv preprint arXiv:1306.5151, 2013

    2013 arXiv

  24. [32]

    McInnes, J

    L. McInnes, J. Healy, and J. Melville. Umap: Uniform manifold approximation and projection for dimension reduction. arXiv preprint arXiv:1802.03426, 2018

    2018 arXiv

  25. [33]

    Nilsback and A

    M.-E. Nilsback and A. Zisserman. Automated flower classification over a large number of classes. In 2008 Sixth Indian conference on computer vision, graphics & image processing, pages 722–729. IEEE, 2008

    2008

  26. [34]

    Noroozi and P

    M. Noroozi and P. Favaro. Unsupervised learning of visual representations by solving jigsaw puzzles. In European conference on computer vision, pages 69–84. Springer, 2016

    2016

  27. [35]

    A. v. d. Oord, Y . Li, and O. Vinyals. Representation learning with contrastive predictive coding.arXiv preprint arXiv:1807.03748, 2018

    2018 arXiv

  28. [36]

    J. Y . Park, O. Biza, L. Zhao, J. W. van de Meent, and R. Walters. Learning symmetric embeddings for equivariant world models. arXiv preprint arXiv:2204.11371, 2022

    2022 arXiv

  29. [37]

    O. M. Parkhi, A. Vedaldi, A. Zisserman, and C. Jawahar. Cats and dogs. In 2012 IEEE conference on computer vision and pattern recognition, pages 3498–3505. IEEE, 2012

    2012

  30. [38]

    Quattoni and A

    A. Quattoni and A. Torralba. Recognizing indoor scenes. In 2009 IEEE conference on computer vision and pattern recognition, pages 413–420. IEEE, 2009

    2009

  31. [39]

    Russakovsky, J

    O. Russakovsky, J. Deng, H. Su, J. Krause, S. Satheesh, S. Ma, Z. Huang, A. Karpathy, A. Khosla, M. Bernstein, et al. Imagenet large scale visual recognition challenge. International journal of computer vision, 115:211–252, 2015

    2015

  32. [40]

    Shakerinava, A

    M. Shakerinava, A. K. Mondal, and S. Ravanbakhsh. Structuring representations using group invariants. Advances in Neural Information Processing Systems, 35:34162–34174, 2022

    2022

  33. [41]

    C. Sun, A. Shrivastava, S. Singh, and A. Gupta. Revisiting unreasonable effectiveness of data in deep learning era. In Proceedings of the IEEE international conference on computer vision, pages 843–852, 2017

    2017

  34. [42]

    Y . Tian, D. Krishnan, and P. Isola. Contrastive multiview coding. In Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XI 16, pages 776–794. Springer, 2020

    2020

  35. [43]

    Y . Tian, C. Sun, B. Poole, D. Krishnan, C. Schmid, and P. Isola. What makes for good views for contrastive learning? Advances in neural information processing systems, 33:6827–6839, 2020

    2020

  36. [44]

    D. Wang, M. Jia, X. Zhu, R. Walters, and R. Platt. On-robot learning with equivariant models. arXiv preprint arXiv:2203.04923, 2022. 12

    2022 arXiv

  37. [45]

    Y . Wang, Z. Geng, F. Jiang, C. Li, Y . Wang, J. Yang, and Z. Lin. Residual relaxation for multi-view representation learning. Advances in Neural Information Processing Systems, 34:12104–12115, 2021

    2021

  38. [46]

    Winter, M

    R. Winter, M. Bertolini, T. Le, F. Noé, and D.-A. Clevert. Unsupervised learning of group invariant and equivariant representations. Advances in Neural Information Processing Systems, 35:31942–31956, 2022

    2022

  39. [47]

    J. Xiao, J. Hays, K. A. Ehinger, A. Oliva, and A. Torralba. Sun database: Large-scale scene recognition from abbey to zoo. In 2010 IEEE computer society conference on computer vision and pattern recognition, pages 3485–3492. IEEE, 2010

    2010

  40. [48]

    Zbontar, L

    J. Zbontar, L. Jing, I. Misra, Y . LeCun, and S. Deny. Barlow twins: Self-supervised learning via redundancy reduction. In International Conference on Machine Learning, pages 12310–12320. PMLR, 2021

    2021

  41. [49]

    H. Zhang. mixup: Beyond empirical risk minimization. arXiv preprint arXiv:1710.09412, 2017. 13 A STL Formulations for Various Base Invariant Models A.1 STL Extension on BYOL In adapting STL to BYOL [ 18], we utilize dissimilarity loss of BYOL to define the invariant, equivaria...

    2017 arXiv

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.