Pith. sign in

REVIEW 4 major objections 5 minor 36 references

Disentanglement in Difference: Directly Learning Semantically Disentangled Representations by Maximizing Inter-Factor Differences

T0 review · 4 major / 5 minor · reviewed 2026-08-09 · deepseek-v4-flash

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

Pith's one-line read A contrastive loss on inter-sample differences, rather than statistical independence, is the direct driver of semantic disentanglement in the proposed DiD method.

desk verdict The novelty is real but the core objective doesn't implement the paper's stated same-factor/different-factor principle; a genuinely interesting idea undermined by a loss that can be maximized without semantic alignment. read the letter →

arxiv 2502.03123 v2 pith:OVLWSGR6 submitted 2025-02-05 cs.LG cs.AI

Xingshen Zhang , Lin Wang , Shuangrong Liu , Xintao Lu , Chaoran Pang , Bo Yang This is my paper · ORCID
classification cs.LGcs.AI
keywords disentangledrepresentationlearningsemanticdisentanglementdifferenceencodercontrastivelossWGAN-GPlatentfactorseparationdSprites3DShapes
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 sets out to show that statistical independence of latent variables is the wrong optimization target for disentanglement: reducing total correlation does not reliably improve semantic factor separation. It proposes Disentanglement in Difference (DiD), a WGAN-GP whose extra loss terms make the model directly contrast variations caused by different latent axes. A Difference Encoder maps pairs of generated images to difference vectors, and a contrastive term maximizes the Euclidean distance between difference vectors coming from different axes; a Sample Encoder reconstructs the latent code from each generated image. Together these terms are meant to organize the latent space so that each axis controls one semantic factor, and the reported experiments on dSprites and 3DShapes give DiD higher MIG, DCI-D, and SAP scores than the VAE- and InfoGAN-based baselines tested.

What carries the argument

The load-bearing mechanism is the Difference Encoder $H$, realized as an MLP, which turns an image pair into one difference vector. The contrastive loss $L_H=-\lVert H(g_1)-H(g_2)\rVert_2$ amplifies the separation between variations caused by different latent axes; the WGAN-GP generator and discriminator provide a high-quality sample distribution over a bounded uniform latent space $c\sim U[-1,1]^n$; and the Sample Encoder $E$ with MSE reconstruction loss $\lVert c-E(G(c))\rVert_2^2$ ties generated samples back to latent codes. The whole system is optimized so that the generator's axes become directions of single-factor variation.

What would settle it

Train DiD on a benchmark where one ground-truth factor is deliberately split across two latent axes or two factors always change together, then measure per-axis factor purity (e.g., SAP per dimension). If the model maintains high MIG while no single axis predicts a single factor better than chance, then maximizing inter-axis difference does not by itself yield semantic disentanglement.

Watch

Extended reading notes

Core claim

The central discovery claimed here is that semantic disentanglement can be learned directly from inter-sample differences, without any penalty on total correlation or mutual information. Define three latent codes $c$, $c'$, $c''$ where $c'$ moves a small step from $c$ along axis $a_1$ and $c''$ moves the same step along axis $a_2$; generate image pairs $g_1=[G(c),G(c')]$ and $g_2=[G(c),G(c'')]$; then train a Difference Encoder $H$ to maximize $-\lVert H(g_1)-H(g_2)\rVert_2$ while a Sample Encoder $E$ is trained with $\lVert c-E(G(c))\rVert_2^2$. The paper argues that this contrastive difference objective directly encodes the prior that different semantic factors should be far apart in representation space, and reports that on dSprites and 3DShapes the method consistently achieves significantly higher SAP, MIG, and DCI-D scores compared to all baseline methods.

Load-bearing premise

The method assumes that nudging one raw coordinate of the sampled latent code produces an image pair that differs in exactly one semantic factor; the training loss never checks this, so a latent axis that mixes two factors can still satisfy the contrastive objective without any factor being cleanly separated.

Editorial extensions

If this is right

  • If the contrastive difference objective is what produces disentanglement, then disentanglement can be optimized as a first-class objective rather than as a side effect of a KL or TC penalty.
  • The method extends to any generative model with a continuous latent space and a differentiable generator, since the difference encoder operates on image pairs rather than on dataset labels.
  • DiD also gives an operational definition of semantic difference: two factors are disentangled when their induced difference vectors are far apart, which could be used as a training signal in settings where ground-truth factors are unknown.
  • The ablation result implies that without the difference comparison the model reduces to an ordinary WGAN-GP and disentanglement scores drop, so the contrastive term, not the GAN backbone, is the source of the reported gains.

Reading between the lines

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

  • A natural next test is whether a learned or normalized metric on difference vectors behaves better than raw Euclidean distance; when factors have very different variability, raw Euclidean distance can be dominated by the highest-variance axis, a case the paper does not examine.
  • The same difference-contrastive prior could be applied to video or multi-view sequences, where temporal or viewpoint changes provide natural same-factor pairs, extending the method beyond synthetic generative benchmarks.
  • The paper's TC-versus-MIG scatter suggests a mathematical gap between statistical independence and semantic disentanglement, but it does not formalize it; a formal condition stating when inter-axis difference maximization implies factor-wise alignment would settle the theoretical status of the method.
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

4 major / 5 minor

Summary. The paper proposes Disentanglement in Difference (DiD), a GAN-based method for learning semantically disentangled representations. DiD trains a generator with WGAN-GP, a Difference Encoder that maps pairs of generated images to difference vectors, and a Sample Encoder that reconstructs the latent code. The central idea is to contrast sample pairs generated by perturbing different latent axes, using the loss LH = -||v1-v2||2 (Eq. 3) to maximize the distance between difference vectors of different axes. Experiments on dSprites and 3DShapes report higher MIG, DCI-D, and SAP scores than several VAE- and GAN-based baselines. The paper also presents an analysis showing that total correlation and MIG are not monotonically related in baseline models, and an ablation of the difference module.

Significance. If the proposed objective genuinely enforced the stated prior that same-factor variations are compact and different-factor variations are separated, the idea of directly optimizing semantic differences rather than statistical independence would be a meaningful contribution to unsupervised disentanglement. The paper has some strengths: it evaluates on standard benchmarks, includes an ablation of the difference module, and makes the interesting empirical observation that lower TC does not consistently improve MIG. However, the central objective as formulated does not implement its stated prior, and the quantitative claims in Table 1 are overstated, so the significance of the reported results is not established. The paper does not ship code or provide sufficient implementation details to reproduce the experiments, which further limits its impact.

major comments (4)
  1. [Section 3.4, Eq. (3)] The loss LH = -||v1 - v2||2 is an unbounded negative Euclidean distance with no margin, no normalization, no Lipschitz constraint on H, and no within-factor compactness term. Minimizing this loss (as implied by Eq. 5) can be trivially driven to -infinity by scaling the output of H, independent of the generator G. Moreover, the objective only separates one pair of difference vectors (one from axis a1 and one from axis a2) at a time; it contains no term that compresses H(G(c), G(c+δe_a)) across different base codes c, and no term relating the difference vectors to the actual semantic content of the generated images. Consequently, maximizing Eq. (3) does not implement the paper's stated principle that same-factor variations are small and different-factor variations are large; it only pushes two arbitrary pair encodings apart. The reported disentanglement metrics, even if reproducible, cannot be attributed to the contrastive objective as written.
  2. [Section 3.2 and Section 3.4] The paper states in Section 3.2 that 'it is assumed that within each sample pair, the two samples differ only in one latent space dimension, while remaining consistent across other factors.' This assumption holds by construction for the input code c, but it is not enforced for the generated images: the loss does not align the generator's axes with the semantic factors of the dataset. If a single input axis mixes multiple semantic factors, the Difference Encoder can maximize inter-axis distances without achieving factor disentanglement. The paper provides no mechanism, such as a reconstruction or classification term on semantic factors, that would force the generator to map each latent axis to a single factor. The central claim therefore relies on an unenforced and untested assumption.
  3. [Section 4.4, Table 1] The claim that DiD 'consistently achieves significantly higher SAP, MIG, and DCI-D scores compared to all baseline methods' is contradicted by Table 1. On dSprites, DiD's SAP is 0.17 ± 0.02, which is lower than FactorVAE's 0.19 ± 0.03 and only equal to InfoGAN-CR within error; on DCI-D, DiD's 0.58 ± 0.06 overlaps with FactorVAE's 0.55 ± 0.05. No statistical significance tests are reported, so 'significantly higher' is unsupported. The claim should be corrected to reflect the actual numbers, and the discussion should acknowledge where DiD does not dominate.
  4. [Section 3.5 and Section 4.3] The Sample Encoder E is trained exclusively on generated samples G(c) with the loss Lenc = ||c - E(G(c))||2_2. The paper does not specify how E is applied to real dataset images when computing the disentanglement metrics (MIG, DCI-D, SAP), which are typically computed on the encoder's representations of the training data. If real images are encoded, there is a domain shift because E has only seen generated images; if generated images are used, the metrics may be biased because the codes c are known. The experimental protocol for metric computation is not described, making the quantitative results non-reproducible.
minor comments (5)
  1. [Throughout] The terminology is inconsistent: the module is called 'Difference Encoder' in Section 3.4 and the abstract, but 'Difference Estimator' in Figure 1 and Section 4.6. Please unify.
  2. [Section 3.4, Eq. (3)] The sign of LH is confusing: the text says 'Our objective is to maximize the distance, d(v1, v2)', but Eq. (3) defines LH as the negative distance. Since Eq. (5) sums LH into a loss to be minimized, the intended effect is to maximize ||v1-v2||; this should be stated explicitly to avoid ambiguity.
  3. [Section 4.3] Several free parameters are not specified: the perturbation step size δ along latent axes, the number of dimensions used for comparison in the Difference Encoder, and the relative weights of LH and Lenc in Eq. (5). Without these, the experiments cannot be reproduced.
  4. [Section 4.1] The paper refers to 'Differentiable Information Disentanglement (DiD)' in the first sentence of Section 4.1, but the acronym was defined as 'Disentanglement in Difference' in the abstract and introduction. This appears to be a typo.
  5. [Section 4.6, Figure 3] The ablation study varies the number of comparison dimensions, but the x-axis label is not readable in the figure, and the text does not state the default number used in Table 1. Please clarify the experimental setup.

Circularity Check

2 steps flagged · score 5.0 of 10

The central 'inter-factor difference' loss is defined over the model's own input-code axes, so the claim of direct semantic disentanglement reduces to separating those axes.

  1. self definitional [Section 3.2, Section 3.4, Eq. (3)]
    "We then select two dimensions, denoted a1 and a2, from S centered at c. These dimensions correspond to two mutually independent latent factors that we aim to disentangle. ... LH = −||v1 − v2||2 = −||H(g1) − H(g2)||2 (3)"

    The 'factors' that the loss separates are, by construction, the input-code axes a1 and a2: g1 and g2 are created by perturbing one coordinate of c, and Eq. (3) maximizes only the distance between the two resulting difference encodings. No term verifies that these axes correspond to dataset semantic factors. Therefore 'maximizing inter-factor differences' is defined to be 'maximizing inter-axis differences,' and the paper's conclusion that DiD directly learns semantic differences restates this definition rather than deriving alignment with the data's generative factors.

  2. fitted input called prediction [Section 3.5, Eq. (4)]
    "Our goal is to make the encoder’s output c′enc of this sample in the latent space as close as possible to the original latent code c used to generate the sample. ... Lenc = ||c − c′enc||22 = ||c − E(G(c))||22 (4)"

    This objective trains E to return exactly the code c that was sampled and then used to define the two 'factor' axes in Eq. (3). Consequently, for any generated image, E's output is fitted to the generator's input by construction, and representations evaluated in Table 1 measure how faithfully the encoder inverts G's own latent axes. The paper presents this as learning a disentangled representation of the data, but the target of the fitting is the same self-generated code whose axes were already labeled 'factors' in the loss, so the claimed prediction of a semantic latent code is the training target itself.

full rationale

The paper contains no load-bearing self-citations: all cited baselines and architectural components are external, and none of the central claims is justified by a prior work of these authors. The main circularity is self-definitional. In Section 3.2, the 'factors' to be disentangled are introduced as axes a1 and a2 of the sampled latent code, with the assumption that perturbing one coordinate produces a change in exactly one factor. Equation (3) then maximizes the distance between the difference encodings of two such axis-perturbation pairs. If 'factor' is read as 'axis of the input code' — the only meaning the loss can access — then maximizing inter-factor differences is identical to separating the model's own axes; the semantic correspondence to dataset factors is assumed, not learned. Equation (4) completes the loop by training the sample encoder to reproduce the sampled code c, so the representation is anchored to the generator's own inputs. The reported MIG, DCI-D, and SAP numbers are external measurements and could in principle refute the claim, so the circularity is partial rather than total. However, the paper's core narrative that DiD 'directly learn[s] semantic differences' rests on equating code axes with semantic factors, which is the very correspondence the method is supposed to establish. The lack of any compactness constraint on same-axis difference vectors and the unboundedness of H are correctness risks, not circularity, and are not scored here.

Assumptions & free parameters 4 free parameters · 5 assumptions · 2 invented entities

The central method rests on an assumed one-to-one alignment between latent axes and semantic factors, an assumed Euclidean metric for semantic difference, and an unstated latent dimension. These are not derived or validated, and they are load-bearing for the claimed semantic disentanglement.

free parameters (4)
  • latent dimension n = not reported
    The size of the generator and encoder latent space is a central design choice; the paper never gives n for dSprites or 3DShapes, though it is likely set to the number of true factors (5 and 6).
  • perturbation step size along latent axes = not reported
    Section 3.4 says c' is obtained by moving 'a short distance' along a1, but the step magnitude is never specified; this controls the semantic scope of each sample pair and affects the difference signal.
  • number of comparison dimensions = not reported
    Section 4.6 ablates this quantity and shows it strongly affects performance, yet Table 1 does not state which value was used.
  • loss weights for LH and Lenc = 1 (implicit)
    Eq. 5 simply adds LG + LH + Lenc with no coefficients; it is unclear whether unit weights were tuned or chosen by hand.
assumptions (5)
  • ad hoc to paper WGAN-GP training yields a generator whose input axes are aligned with the true semantic factors of the dataset
    The method assumes 'within each sample pair, the two samples differ only in one latent space dimension' (Section 3.2), i.e., each coordinate axis maps to a single factor; no loss explicitly enforces this.
  • domain assumption Euclidean distance between difference-encoding vectors reflects semantic distinctness
    Section 3.4 maximizes ||v1-v2|| without any calibration, margin, or alternative metric; it assumes larger Euclidean distance in H-space means more distinct semantic factors.
  • domain assumption The latent uniform prior U[-1,1]^n is sufficient to cover all semantic variations
    Section 3.3 chooses a bounded uniform prior to prevent aggregation, but provides no evidence that a hypercube of fixed dimension parameterizes the data's factor manifold.
  • domain assumption Dataset factors are the same as the axes the model is expected to discover
    Evaluation metrics compare against dSprites and 3DShapes ground-truth factors; the paper assumes the unsupervised model's axes will align with these labels.
  • standard math Disentanglement metrics (MIG, DCI, SAP) reliably measure semantic disentanglement
    The paper relies on these metrics without discussing their known limitations, such as sensitivity to hyperparameters and classifier choices.
invented entities (2)
  • Difference Encoder (H)
    purpose: Maps a pair of generated images to a vector representing the semantic difference caused by perturbing one latent axis.
    No external labels or measurements validate H's outputs; it is trained only via the unbounded contrastive loss and evaluated indirectly through downstream metrics.
  • semantic difference vector
    purpose: A construct meant to represent the change in one latent factor; used in the contrastive loss.
    The paper defines the vector as H(g1)-H(g2) but provides no evidence that Euclidean separation of these vectors corresponds to human-interpretable factor differences.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Disentanglement in Difference: Directly Learning Semantically Disentangled Representations by Maximizing Inter-Factor Differences." pith.science (2026). https://pith.science/paper/OVLWSGR6

@misc{pith2026250203123,
  author       = {Pith},
  title        = {Pith review of: Disentanglement in Difference: Directly Learning Semantically Disentangled Representations by Maximizing Inter-Factor Differences},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OVLWSGR6}},
  note         = {Machine review of arXiv:2502.03123}
}
read the original abstract

In this study, Disentanglement in Difference(DiD) is proposed to address the inherent inconsistency between the statistical independence of latent variables and the goal of semantic disentanglement in disentanglement representation learning. Conventional disentanglement methods achieve disentanglement representation by improving statistical independence among latent variables. However, the statistical independence of latent variables does not necessarily imply that they are semantically unrelated, thus, improving statistical independence does not always enhance disentanglement performance. To address the above issue, DiD is proposed to directly learn semantic differences rather than the statistical independence of latent variables. In the DiD, a Difference Encoder is designed to measure the semantic differences; a contrastive loss function is established to facilitate inter-dimensional comparison. Both of them allow the model to directly differentiate and disentangle distinct semantic factors, thereby resolving the inconsistency between statistical independence and semantic disentanglement. Experimental results on the dSprites and 3DShapes datasets demonstrate that the proposed DiD outperforms existing mainstream methods across various disentanglement metrics.

Figures

Figures reproduced from arXiv: 2502.03123 by the authors.

Figure 1
Figure 1. The framework of the proposed DiD. The blue points denote c sampled from the uniform distribution, the red points represent samples obtained from two orthogonal axes. GP(Gulrajani et al., 2017) is utilized as the sample gener￾ation paradigm to maintain high-quality generation perfor￾mance. To establish a correspondence between each dimension of the latent space and a single factor, it is necessary to con￾struct a lo… view at source ↗
Figure 2
Figure 2. Relationship between Total Correlation (TC) and Mutual Information Gap (MIG). The scatter plots illustrate that a lower Total Correlation does not consistently guarantee improved disen￾tanglement performance. latent variables, the model achieves effective disentangling of the latent space dimensions, learning better disentangled representations. 4.5. Total Correlation and Disentangling Metrics As mentioned earlier, … view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 15 canonical work pages

  1. [1]

    Wasserstein generative adversarial networks

    Arjovsky, M., Chintala, S., and Bottou, L. Wasserstein generative adversarial networks. In International conference on machine learning, pp.\ 214--223. PMLR, 2017

    2017

  2. [2]

    Representation learning: A review and new perspectives

    Bengio, Y., Courville, A., and Vincent, P. Representation learning: A review and new perspectives. IEEE transactions on pattern analysis and machine intelligence, 35 0 (8): 0 1798--1828, 2013

    2013

  3. [3]

    Scaling learning algorithms toward ai

    Bottou, L., Chapelle, O., DeCoste, D., and Weston, J. Scaling learning algorithms toward ai. 2007

    work page 2007

  4. [4]

    and Kim, H

    Burgess, C. and Kim, H. 3d shapes dataset. https://github.com/deepmind/3dshapes-dataset/, 2018

    2018

  5. [5]

    P., Higgins, I., Pal, A., Matthey, L., Watters, N., Desjardins, G., and Lerchner, A

    Burgess, C. P., Higgins, I., Pal, A., Matthey, L., Watters, N., Desjardins, G., and Lerchner, A. Understanding disentangling in -vae. arXiv preprint arXiv:1804.03599, 2018

    arXiv 2018

  6. [6]

    Measuring disentanglement: A review of metrics

    Carbonneau, M.-A., Zaidi, J., Boilard, J., and Gagnon, G. Measuring disentanglement: A review of metrics. IEEE transactions on neural networks and learning systems, 2022

    work page 2022

  7. [7]

    T., Li, X., Grosse, R

    Chen, R. T., Li, X., Grosse, R. B., and Duvenaud, D. K. Isolating sources of disentanglement in variational autoencoders. Advances in neural information processing systems, 31, 2018

    2018

  8. [8]

    Infogan: Interpretable representation learning by information maximizing generative adversarial nets

    Chen, X., Duan, Y., Houthooft, R., Schulman, J., Sutskever, I., and Abbeel, P. Infogan: Interpretable representation learning by information maximizing generative adversarial nets. Advances in neural information processing systems, 29, 2016

    2016

Show all 36 references
  1. [9]

    and Williams, C

    Eastwood, C. and Williams, C. K. A framework for the quantitative evaluation of disentangled representations. In 6th International Conference on Learning Representations, 2018

    2018

  2. [10]

    and Hoshen, Y

    Gabbay, A. and Hoshen, Y. Demystifying inter-class disentanglement. arXiv preprint arXiv:1906.11796, 2019

    1906 arXiv

  3. [11]

    and Hoshen, Y

    Gabbay, A. and Hoshen, Y. Scaling-up disentanglement for image translation. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pp.\ 6783--6792, 2021

    2021

  4. [12]

    Gulrajani, I., Ahmed, F., Arjovsky, M., Dumoulin, V., and Courville, A. C. Improved training of wasserstein gans. Advances in neural information processing systems, 30, 2017

    2017

  5. [13]

    P., Glorot, X., Botvinick, M

    Higgins, I., Matthey, L., Pal, A., Burgess, C. P., Glorot, X., Botvinick, M. M., Mohamed, S., and Lerchner, A. beta-vae: Learning basic visual concepts with a constrained variational framework. ICLR (Poster), 3, 2017

    2017

  6. [14]

    Towards a definition of disentangled representations

    Higgins, I., Amos, D., Pfau, D., Racaniere, S., Matthey, L., Rezende, D., and Lerchner, A. Towards a definition of disentangled representations. arXiv preprint arXiv:1812.02230, 2018

    2018 arXiv

  7. [15]

    and Mnih, A

    Kim, H. and Mnih, A. Disentangling by factorising. In International conference on machine learning, pp.\ 2649--2658. PMLR, 2018

    2018

  8. [16]

    Kingma, D. P. Auto-encoding variational bayes. arXiv preprint arXiv:1312.6114, 2013

    2013 arXiv

  9. [18]

    Variational inference of disentangled latent concepts from unlabeled observations

    Kumar, A., Sattigeri, P., and Balakrishnan, A. Variational inference of disentangled latent concepts from unlabeled observations. arXiv preprint arXiv:1711.00848, 2017 b

    2017 arXiv

  10. [19]

    M., Ullman, T

    Lake, B. M., Ullman, T. D., Tenenbaum, J. B., and Gershman, S. J. Building machines that learn and think like people. Behavioral and brain sciences, 40: 0 e253, 2017

    2017

  11. [20]

    Gradient-based learning applied to document recognition

    LeCun, Y., Bottou, L., Bengio, Y., and Haffner, P. Gradient-based learning applied to document recognition. Proceedings of the IEEE, 86 0 (11): 0 2278--2324, 1998

    1998

  12. [21]

    Infogan-cr and modelcentrality: Self-supervised model training and selection for disentangling gans

    Lin, Z., Thekumparampil, K., Fanti, G., and Oh, S. Infogan-cr and modelcentrality: Self-supervised model training and selection for disentangling gans. In international conference on machine learning, pp.\ 6127--6139. PMLR, 2020

    2020

  13. [22]

    Challenging common assumptions in the unsupervised learning of disentangled representations

    Locatello, F., Bauer, S., Lucic, M., Raetsch, G., Gelly, S., Sch \"o lkopf, B., and Bachem, O. Challenging common assumptions in the unsupervised learning of disentangled representations. In international conference on machine learning, pp.\ 4114--4124. PMLR, 2019 a

    2019

  14. [23]

    a tsch, G., Sch \

    Locatello, F., Tschannen, M., Bauer, S., R \"a tsch, G., Sch \"o lkopf, B., and Bachem, O. Disentangling factors of variation using few labels. arXiv preprint arXiv:1905.01258, 2019 b

    1905 arXiv

  15. [24]

    dsprites: Disentanglement testing sprites dataset

    Matthey, L., Higgins, I., Hassabis, D., and Lerchner, A. dsprites: Disentanglement testing sprites dataset. https://github.com/deepmind/dsprites-dataset/, 2017

    2017

  16. [25]

    Elements of causal inference: foundations and learning algorithms

    Peters, J., Janzing, D., and Sch \"o lkopf, B. Elements of causal inference: foundations and learning algorithms. The MIT Press, 2017

    2017

  17. [26]

    Unsupervised representation learning with deep convolutional generative adversarial networks

    Radford, A. Unsupervised representation learning with deep convolutional generative adversarial networks. arXiv preprint arXiv:1511.06434, 2015

    2015 arXiv

  18. [27]

    and Mozer, M

    Ridgeway, K. and Mozer, M. C. Learning deep disentangled embeddings with the f-statistic loss. Advances in neural information processing systems, 31, 2018

    2018

  19. [28]

    Semilinear predictability minimization produces well-known feature detectors

    Schmidhuber, J., Eldracher, M., and Foltin, B. Semilinear predictability minimization produces well-known feature detectors. Neural Computation, 8 0 (4): 0 773--786, 1996

    1996

  20. [29]

    Dynamicvae: Decoupling reconstruction error and disentangled representation learning

    Shao, H., Lin, H., Yang, Q., Yao, S., Zhao, H., and Abdelzaher, T. Dynamicvae: Decoupling reconstruction error and disentangled representation learning. arXiv preprint arXiv:2009.06795, 2020

    2009 arXiv

  21. [30]

    Recent advances in autoencoder-based representation learning

    Tschannen, M., Bachem, O., and Lucic, M. Recent advances in autoencoder-based representation learning. arXiv preprint arXiv:1812.05069, 2018

    2018 arXiv

  22. [31]

    Are disentangled representations helpful for abstract visual reasoning? Advances in neural information processing systems, 32, 2019

    Van Steenkiste, S., Locatello, F., Schmidhuber, J., and Bachem, O. Are disentangled representations helpful for abstract visual reasoning? Advances in neural information processing systems, 32, 2019

    2019

  23. [32]

    Information theoretical analysis of multivariate correlation

    Watanabe, S. Information theoretical analysis of multivariate correlation. IBM Journal of research and development, 4 0 (1): 0 66--82, 1960

    1960

  24. [33]

    Zeiler, M. D. and Fergus, R. Visualizing and understanding convolutional networks. In Computer Vision--ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part I 13, pp.\ 818--833. Springer, 2014

    2014

  25. [34]

    Y., Carlsson, G., and Ermon, S

    Zhou, S., Zelikman, E., Lu, F., Ng, A. Y., Carlsson, G., and Ermon, S. Evaluating the disentanglement of deep generative models through manifold topology. arXiv preprint arXiv:2006.03680, 2020

    2006 arXiv

  26. [35]

    Visual object networks: Image generation with disentangled 3d representations

    Zhu, J.-Y., Zhang, Z., Zhang, C., Wu, J., Torralba, A., Tenenbaum, J., and Freeman, B. Visual object networks: Image generation with disentangled 3d representations. Advances in neural information processing systems, 31, 2018

    2018

  27. [36]

    Where and what? examining interpretable disentangled representations

    Zhu, X., Xu, C., and Tao, D. Where and what? examining interpretable disentangled representations. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 5861--5870, 2021

    2021

  28. [37]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.