Pith. sign in

REVIEW 4 major objections 7 minor 50 references

Multi-scale Generative Modeling for Fast Sampling

T0 review · 4 major / 7 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read A wavelet-domain hybrid model samples high-quality images in 16 steps

desk verdict A credible hybrid of wavelet-domain score diffusion and a shared multi-scale GAN that beats its two baselines at 16 steps, but the theoretical support for the central well-conditioning claim is thinner than it looks. read the letter →

arxiv 2411.09356 v1 pith:H5ACJHIZ submitted 2024-11-14 cs.AI

Xiongye Xiao , Shixuan Li , Luzhe Huang , Gengshuo Liu , Trung-Kien Nguyen , Yi Huang , Di Chang , Mykel J. Kochenderfer
show 1 more author
Paul Bogdan
This is my paper · ORCID
classification cs.AI
keywords wavelet-domaindiffusionscore-basedgenerativemodelsmulti-scaleadversariallearningfastsamplingimagegenerationconditionnumbernon-Gaussianwaveletcoefficientsnetworks
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

The paper sets out to show that the slowness of diffusion-based image generation comes from forcing one Gaussian noising process onto the whole image, and that splitting the image by frequency fixes it. Its proposed model, the Wavelet Multi-scale Generative Model (WMGM), runs score-based diffusion only on the coarsest low-frequency wavelet band, where coefficients are close to Gaussian and well-conditioned, and uses a multi-scale adversarial network shared across scales to synthesize the sparse high-frequency bands from that low-frequency condition. On CelebA-HQ, AFHQ-Cat, and Colon at 128 × 128, the paper reports that WMGM with 16 total sampling steps reaches lower FID than both the spatial score baseline SGM and the wavelet score baseline WSGM, while using 89M parameters and 2.31 seconds per 100 images, compared with 160M/351M parameters and 15.09/11.10 seconds for those baselines. If the claim holds, fast sampling no longer has to trade away image quality, because the expensive iterative denoising is confined to a small well-conditioned subband.

What carries the argument

The load-bearing object is the orthogonal discrete wavelet transform as a coordinate change that splits image information by frequency. Because the wavelet matrix $A$ satisfies $AA^T = I$, the forward noising process and its reverse are dual between spatial and wavelet domains, so the score-based SDE can be run on wavelet coefficients without reformulating the dynamics. The mechanism then discriminates by band: the deepest low-frequency band is whitened by repeated averaging, making its covariance nearly isotropic with a small condition number, so the discretization bound $N \le \varepsilon^{-2} \kappa^3$ implies a short reverse process suffices; the high-frequency bands are sparse and long-tailed, so instead of diffusion the paper applies Multi-Scale Adversarial Learning (MSAL), a 5-level U-Net generator with attention gates that maps each LL band plus injected noise to the corresponding high-frequency bands, trained with a Wasserstein adversarial loss combined with MSE and SSIM terms and with parameters shared across scales.

What would settle it

Measure the empirical condition number and KL divergence to Gaussian of the coarsest LL band for a candidate dataset, then train WMGM on a version of that dataset whose LL band has been deliberately made non-Gaussian (for example by applying a heavy-tailed intensity transform before wavelet decomposition). If the FID gap over SGM and WSGM at 16 steps collapses, or the required step count jumps while the baselines move less, the Gaussian-tendency assumption is the load-bearing premise.

Watch

Extended reading notes

Core claim

The central claim is that wavelet-domain score-based diffusion is not the whole answer: the high-frequency bands are too sparse and non-Gaussian for Gaussian diffusion, while the low-frequency band is where diffusion works best. WMGM therefore factorizes the target distribution as $p(x_0) = \prod_{k=1}^{S} p(x_H^k \mid x_L^k) \, p(x_L^S)$, generating the coarsest LL band $x_L^S$ by a short score-based reverse process and each set of high-frequency bands $x_H^k$ by a single adversarial pass of a shared multi-scale generator conditioned on $x_L^k$. The paper argues theoretically that the orthogonal wavelet transform maps the Ornstein-Uhlenbeck noising process and its reverse SDE onto the wavelet domain unchanged (duality), that iterated averaging whitens the LL band so its covariance has a low condition number and needs fewer discretization steps, and that the conditional high-frequency distribution remains sparse at deep scales, justifying a non-Gaussian model. Experimentally, the paper's Table 1 shows WMGM at 16 total sampling steps achieves FID 30.58/25.38 on CelebA-HQ (5K/30K), 16.29 on AFHQ-Cat, and 45.76 on Colon, each better than SGM and WSGM, with 89M parameters and 2.31 seconds per 100 images; the advantage persists from 6 to 192 steps in Figure 4.

Load-bearing premise

The method depends on the lowest-frequency wavelet band being close to Gaussian and well-conditioned enough that a short score-based diffusion can capture the global image structure; if that band stays heavy-tailed or ill-conditioned on a new dataset, the diffusion stage needs many more steps and the speed advantage disappears.

Editorial extensions

If this is right

  • At 16 total sampling steps, WMGM reports lower FID than SGM and WSGM on CelebA-HQ, AFHQ-Cat, and Colon, so the hybrid is the best of the three under the same step budget.
  • Because diffusion runs only on the coarsest LL band, the number of sampling steps needed for good quality is no longer tied to the full-image score condition number, which is the paper's explanation for the speedup.
  • Sharing generator parameters across wavelet scales reduces the model to 89M parameters, smaller than either baseline, so the added GAN component does not increase model size.
  • The ranking holds across total step counts from 6 to 192 and across face, animal, and histopathology datasets, indicating the frequency-split design transfers beyond a single domain.
  • The multi-scale adversarial variant outperforms single-scale, operator-learning, and standard-regression variants for the low-to-high-frequency mapping with half the parameters, so parameter sharing across scales is itself part of the reported quality gain.

Reading between the lines

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

  • Editorial inference: the same low-frequency diffusion plus sparse-detail adversarial split should transfer to other orthogonal or near-orthogonal multiscale decompositions such as Laplacian pyramids or block DCTs, since the paper's argument relies on band sparsity and whitening more than on wavelet-specific structure.
  • Editorial inference: a direct stress test is to measure the deepest LL-band condition number and KL divergence to Gaussian for each new dataset and check whether the optimal number of diffusion steps tracks those quantities; the paper's Appendix E.1 computes such diagnostics for two datasets but does not connect them to observed step counts.
  • Editorial inference: because the generator maps LL plus noise to high frequencies in one pass, the model may underrepresent multimodal detail textures even when FID improves; a perceptual-diversity metric computed across the injected noise for a fixed LL band would test that component.
  • Editorial inference: the reported speed advantage is measured per 100 images at a fixed resolution on one GPU type; at higher resolutions the cost shifts toward the adversarial generator, so the relative gain over full-image diffusion may grow or shrink depending on which component dominates.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 7 minor

Summary. The paper proposes WMGM, a multi-scale generative model that operates in the wavelet domain. The method applies score-based diffusion only to the coarsest low-frequency (LL) wavelet band and uses a multi-scale adversarial network to generate high-frequency bands conditioned on the low-frequency information. The authors claim theoretical support for the well-conditioning of LL coefficients and for the sparsity/non-Gaussianity of high-frequency coefficients, and they report experimental results on CelebA-HQ, AFHQ-Cat, and Colon showing lower FID than SGM and WSGM at 16 total sampling steps, with 89M parameters and 2.31 seconds per 100 images. They also include an ablation study of multi-scale versus single-scale adversarial learning.

Significance. If the empirical claims hold, the paper offers a practically useful combination of diffusion and adversarial generation in the wavelet domain, with substantially fewer parameters and faster sampling than the compared baselines. The manuscript provides comparisons on three datasets, a promising anonymous code release, and an ablation of the proposed MSAL component. However, the theoretical contribution is partially inherited from prior work, one of the new proofs contains an invalid step, and the central well-conditioning premise is not directly measured at the scale used in the experiments. The significance is therefore conditional on strengthening the theoretical and empirical evidence.

major comments (4)
  1. [Appendix E.2, Proposition 2] The proof of Proposition 2 is invalid. From the average lower bound ∫ E(s(x_H^k)|x_L^k) p(x_L^k) dx_L^k ≥ α, the proof asserts that E(s(x_H^k)|x_L^k) has a uniform lower bound α' for all x_L^k. An average lower bound does not imply a pointwise lower bound, and no argument is given for such a uniform bound. Equation (70) then identifies α' with the integral minus ε, which is not justified. Consequently, Proposition 2, which is used to support the claim that conditional high-frequency distributions are non-Gaussian and sparse, is unproved as written. The authors should either provide a correct proof with explicit assumptions or restate the proposition as an empirical observation supported by Figure 6.
  2. [Section 3.2.2 and Appendix E.1] The central theoretical motivation relies on the claim that the coarsest LL band is near-Gaussian and well-conditioned, so that a short 16-step reverse diffusion suffices. This claim is not validated at the actual scale used in the experiments. Appendix E.1 reports KL divergences only for scales 0, 1, and 2, and Figure 5 does not show the coarsest scale S used in Table 1. The paper also never reports the condition number κ(Σ_LL) or the KL divergence to a Gaussian at the operational S. Without these measurements, Eq. (9) cannot be used to explain the 16-step success, and the observed FID gains could be dominated by the multi-scale adversarial high-frequency generator rather than by the claimed well-conditioned diffusion component. Please report S and the relevant diagnostics for the datasets used in Table 1.
  3. [Table 1 and Figure 4] The central empirical comparison reports single FID values without error bars or repeated seeds. For a claim of significant improvement over SGM and WSGM, the authors should provide means and standard deviations over multiple training runs or at least over multiple evaluation batches. The sampling time is averaged over 10 trials, but FID is not, and the reader cannot assess the stability of the reported differences, especially for the smaller gaps such as 25.38 vs 26.74 on CelebA-HQ (30K).
  4. [Appendix A and Section 3.1] The paper states that Theorem 1 and Proposition 1 are proved by referring to 'Proof S5 in [19]', and Theorem 6 is quoted directly from [19]. Since one of the stated contributions is to 'theoretically establish the generative modeling in the wavelet domain', the manuscript should either provide self-contained proofs for the statements it presents or clearly identify which results are prior work and what exactly is new in this paper. As written, the theoretical section does not substantiate the claimed new theoretical contribution beyond the earlier WSGM analysis.
minor comments (7)
  1. [Abstract] The sentence 'our model significantly improve performance' contains a subject-verb agreement error and should read 'our model significantly improves performance'.
  2. [Section 4.3] The text refers to 'Appendix I' for the definition of MS and SS, but the definitions appear in Appendix H; please correct the cross-reference.
  3. [References] References [18] and [19] are the same paper and appear to be duplicated; please merge them into a single reference.
  4. [Figure 2 caption] The word 'hierachical' is misspelled and should be 'hierarchical'.
  5. [Section 3.3] The term 'ill-positionedness' should be 'ill-conditionedness' for consistency with the rest of the paper.
  6. [Appendix H.2] The sentence 'Detailed model information can be found in ?? as well as the source code' contains a placeholder '??' that should be replaced with a specific appendix or section reference.
  7. [Appendix J] The limitation statement says the method 'sacrifices a little image quality', yet Table 1 reports lower FID than both baselines; please clarify whether 'image quality' here refers to a different comparison or a qualitative trade-off rather than the FID metric.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the speed/quality claims are measured on held-out data, and the theory either is standard linear-transform math or cites external prior work.

full rationale

WMGM's headline results are not circular: Table 1 and Figure 4 report FID on held-out images for SGM, WSGM, and WMGM under matched total sampling steps, and the generation losses (Eqs. 19, 21, 22) train the score network and the adversarial high-frequency generator without fitting any parameter to the reported FID values. The theoretical pieces that could look borrowed are in fact external: Theorem 1, Proposition 1, and Theorem 6 are taken from Guth et al. [19] ('Please refer to Proof S5 in [19] for the detailed proof outline'), and [19] is not an author self-citation, so it is independent support. The wavelet-domain duality in Definitions 2-3 and Appendix D is an exact consequence of the orthogonality of the wavelet matrix A (A A^T = I): an OU process mapped by A remains an OU process because A B_t is Brownian; this is a change of variables, not a conclusion defined in terms of the experimental speed-up. The Gaussian tendency and sparsity analyses (Sections 3.2.2-3.2.3, Appendix E) are informal CLT/empirical heuristics whose weakness is lack of quantitative verification at the coarsest scale used in Table 1, which is a correctness/rigor concern rather than circularity. The self-reported limitation in Appendix J ('it also sacrifices a little image quality') and the missing reference marker '??' in Appendix H.2 are non-circular weaknesses.

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

The model rests on standard wavelet orthogonality, natural-image power-law assumptions, and a heuristic CLT for low-frequency coefficients. It also borrows its main theorem from [19] rather than proving it. The GAN hyperparameters and number of scales are under-specified. No new physical entities are introduced.

free parameters (4)
  • Number of wavelet scales S = not stated in main text
    Chooses how many scales the hierarchical wavelet decomposition and shared MSAL operate over; directly affects parameter count and sampling steps, but no ablation for S is reported.
  • Loss weights in MSAL (lambda, nu, alpha) = not reported
    Equation (21) combines L2, SSIM, and adversarial terms with weights lambda, nu, alpha; values are omitted, so the objective is under-specified for replication despite being central to the GAN branch.
  • Diffusion time horizon T and discretization step count = 16 total steps in main table; T not specified
    The claim of equal total sampling steps for SGM, WSGM, and WMGM depends on this count, but the underlying SGM schedule and T are not given; results vary with step count.
  • Sparsity threshold in Figure 6 = not reported
    Used to measure high-frequency coefficient sparsity and to support non-Gaussianity; the choice of threshold affects the reported sparsity values, though it is only an analysis auxiliary.
assumptions (5)
  • standard math Discrete wavelet transform matrix A is orthogonal, so DWT preserves the OU diffusion and score structure (qt(x) = pt(A^T x)).
    Used throughout Section D and Definitions 2 and 3; true for orthonormal wavelets such as Haar, but not for biorthogonal or redundant transforms.
  • domain assumption Natural image power spectra follow power-law decay S(omega) ~ (lambda^eta + |omega|^eta)^(-1) with eta = 1.
    Used in Section 3.1 to argue spatial-domain score is ill-conditioned and to justify wavelet-domain modeling; standard in natural image statistics but not universal across datasets.
  • domain assumption After wavelet downsampling, pixel correlations decay fast enough for a generalized central limit theorem to make LL coefficients approximately Gaussian.
    Appendix E.1 invokes reduced correlation plus averaging to claim Gaussian tendency of low-frequency coefficients; this is heuristic and not derived for finite images or the datasets used.
  • standard math Theorem 1 and Proposition 1 from Guth et al. [19] are correct and apply here.
    The paper states to refer to Proof S5 in [19] for the detailed proof outline and does not provide its own proof; the theoretical contribution relies on this external result.
  • domain assumption Low-frequency and high-frequency wavelet bands are independent enough that a conditional generator can reconstruct high-frequency detail from LL alone.
    Section 3.3 says orthogonality makes HF independent and uses this to justify MSAL; in practice spatial correlation between LL and HF is what makes the conditional mapping learnable, a tension not resolved.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multi-scale Generative Modeling for Fast Sampling." pith.science (2026). https://pith.science/paper/H5ACJHIZ

@misc{pith2026241109356,
  author       = {Pith},
  title        = {Pith review of: Multi-scale Generative Modeling for Fast Sampling},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/H5ACJHIZ}},
  note         = {Machine review of arXiv:2411.09356}
}
read the original abstract

While working within the spatial domain can pose problems associated with ill-conditioned scores caused by power-law decay, recent advances in diffusion-based generative models have shown that transitioning to the wavelet domain offers a promising alternative. However, within the wavelet domain, we encounter unique challenges, especially the sparse representation of high-frequency coefficients, which deviates significantly from the Gaussian assumptions in the diffusion process. To this end, we propose a multi-scale generative modeling in the wavelet domain that employs distinct strategies for handling low and high-frequency bands. In the wavelet domain, we apply score-based generative modeling with well-conditioned scores for low-frequency bands, while utilizing a multi-scale generative adversarial learning for high-frequency bands. As supported by the theoretical analysis and experimental results, our model significantly improve performance and reduce the number of trainable parameters, sampling steps, and time.

Figures

Figures reproduced from arXiv: 2411.09356 by the authors.

Figure 1
Figure 1. Diffusion trajectories of the wavelet coefficients. Notice that the high-frequency components [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. The framework of WMGM, featuring a hierachical wavelet transform (WT) and inverse [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 2
Figure 2. We start with the multi-scale wavelet decomposition and factorize the probability of target [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figures from the paper (17 more)
Figure 3
Figure 3. Figure 3: Generated images of SGM, WSGM and our method on CelebA-HQ datasets with only 16 [PITH_FULL_IMAGE:figures/full_fig_p007_3.png]
Figure 4
Figure 4. Figure 4: Performances of SGM, WSGM and our method on AFHQ-Cat and CelebA-HQ datasets [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: KL divergence of sample distribution (scale = 0) and LL coefficient distributions (scale [PITH_FULL_IMAGE:figures/full_fig_p018_5.png]
Figure 6
Figure 6. Figure 6: High frequency coefficient sparsity of CelebA-HQ images. Low-magnitude coefficients [PITH_FULL_IMAGE:figures/full_fig_p020_6.png]
Figure 7
Figure 7. Figure 7: SGM with 4 discretization steps [PITH_FULL_IMAGE:figures/full_fig_p024_7.png]
Figure 8
Figure 8. Figure 8: WSGM with 4 discretization steps 24 [PITH_FULL_IMAGE:figures/full_fig_p024_8.png]
Figure 9
Figure 9. Figure 9: WMGM with 4 discretization steps [PITH_FULL_IMAGE:figures/full_fig_p025_9.png]
Figure 10
Figure 10. Figure 10: SGM with 16 discretization steps [PITH_FULL_IMAGE:figures/full_fig_p025_10.png]
Figure 11
Figure 11. Figure 11: WSGM with 16 discretization steps 25 [PITH_FULL_IMAGE:figures/full_fig_p025_11.png]
Figure 12
Figure 12. Figure 12: WMGM with 16 discretization steps [PITH_FULL_IMAGE:figures/full_fig_p026_12.png]
Figure 13
Figure 13. Figure 13: SGM with 64 discretization steps [PITH_FULL_IMAGE:figures/full_fig_p026_13.png]
Figure 14
Figure 14. Figure 14: WSGM with 64 discretization steps 26 [PITH_FULL_IMAGE:figures/full_fig_p026_14.png]
Figure 15
Figure 15. Figure 15: WMGM with 64 discretization steps [PITH_FULL_IMAGE:figures/full_fig_p027_15.png]
Figure 16
Figure 16. Figure 16: Cat samples with WMGM using 4 discretization steps [PITH_FULL_IMAGE:figures/full_fig_p027_16.png]
Figure 17
Figure 17. Figure 17: Cat samples with WMGM using 16 discretization steps [PITH_FULL_IMAGE:figures/full_fig_p027_17.png]
Figure 18
Figure 18. Figure 18: Cat samples with WMGM using 64 discretization steps [PITH_FULL_IMAGE:figures/full_fig_p028_18.png]
Figure 19
Figure 19. Figure 19: Colon samples with SGM (Left), WSGM (Middle) and WMGM (Right) using 16 [PITH_FULL_IMAGE:figures/full_fig_p028_19.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

50 extracted references · 21 canonical work pages

  1. [19]

    Wavelet score-based generative modeling

    Florentin Guth, Simon Coste, Valentin De Bortoli, and Stephane Mallat. Wavelet score-based generative modeling. Advances in Neural Information Processing Systems, 35:478–491, 2022

    work page 2022

  2. [1]

    Wasserstein gan, 2017

    Martin Arjovsky, Soumith Chintala, and Léon Bottou. Wasserstein gan, 2017

    work page 2017

  3. [2]

    Li, Hamid Kazemi, Furong Huang, Micah Goldblum, Jonas Geiping, and Tom Goldstein

    Arpit Bansal, Eitan Borgnia, Hong-Min Chu, Jie S. Li, Hamid Kazemi, Furong Huang, Micah Goldblum, Jonas Geiping, and Tom Goldstein. Cold diffusion: Inverting arbitrary image transforms without noise, 2022

    work page 2022

  4. [3]

    Generating sentences from a continuous space

    Samuel R Bowman, Luke Vilnis, Oriol Vinyals, Andrew M Dai, Rafal Jozefowicz, and Samy Bengio. Generating sentences from a continuous space. arXiv preprint arXiv:1511.06349, 2015

    arXiv 2015

  5. [4]

    Large scale gan training for high fidelity natural image synthesis

    Andrew Brock, Jeff Donahue, and Karen Simonyan. Large scale gan training for high fidelity natural image synthesis. arXiv preprint arXiv:1809.11096, 2018

    arXiv 2018

  6. [5]

    Image compression via joint statistical charac- terization in the wavelet domain

    Robert W Buccigrossi and Eero P Simoncelli. Image compression via joint statistical charac- terization in the wavelet domain. IEEE transactions on Image processing, 8(12):1688–1701, 1999

    work page 1999

  7. [6]

    A handbook of Fourier theorems

    David C Champeney. A handbook of Fourier theorems. Cambridge University Press, 1987

    work page 1987

  8. [7]

    Stargan v2: Diverse image synthesis for multiple domains

    Yunjey Choi, Youngjung Uh, Jaejun Yoo, and Jung-Woo Ha. Stargan v2: Diverse image synthesis for multiple domains. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2020

    2020

Show all 50 references
  1. [8]

    Generative adversarial networks: An overview

    Antonia Creswell, Tom White, Vincent Dumoulin, Kai Arulkumaran, Biswa Sengupta, and Anil A Bharath. Generative adversarial networks: An overview. IEEE signal processing magazine, 35(1):53–65, 2018

    2018

  2. [9]

    Diffusion schrödinger bridge with applications to score-based generative modeling

    Valentin De Bortoli, James Thornton, Jeremy Heng, and Arnaud Doucet. Diffusion schrödinger bridge with applications to score-based generative modeling. Advances in Neural Information Processing Systems, 34:17695–17709, 2021

    2021

  3. [10]

    Diffusion models beat gans on image synthesis

    Prafulla Dhariwal and Alexander Nichol. Diffusion models beat gans on image synthesis. Advances in neural information processing systems, 34:8780–8794, 2021

    2021

  4. [11]

    A general central limit theorem for strong mixing sequences

    Magnus Ekström. A general central limit theorem for strong mixing sequences. Statistics & Probability Letters, 94:236–238, 2014

    2014

  5. [12]

    Relations between the statistics of natural images and the response properties of cortical cells

    David J Field. Relations between the statistics of natural images and the response properties of cortical cells. Josa a, 4(12):2379–2394, 1987

    1987

  6. [13]

    Automatic chemical design using a data-driven continuous representation of molecules

    Rafael Gómez-Bombarelli, Jennifer N Wei, David Duvenaud, José Miguel Hernández-Lobato, Benjamín Sánchez-Lengeling, Dennis Sheberla, Jorge Aguilera-Iparraguirre, Timothy D Hirzel, Ryan P Adams, and Alán Aspuru-Guzik. Automatic chemical design using a data-driven continuous repr...

    2018

  7. [14]

    Generative adversarial nets

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adversarial nets. Advances in neural information processing systems, 27, 2014

    2014

  8. [15]

    Generative adversarial networks

    Ian Goodfellow, Jean Pouget-Abadie, Mehdi Mirza, Bing Xu, David Warde-Farley, Sherjil Ozair, Aaron Courville, and Yoshua Bengio. Generative adversarial networks. Communications of the ACM, 63(11):139–144, 2020

    2020

  9. [16]

    A review on generative adversarial networks: Algorithms, theory, and applications

    Jie Gui, Zhenan Sun, Yonggang Wen, Dacheng Tao, and Jieping Ye. A review on generative adversarial networks: Algorithms, theory, and applications. IEEE transactions on knowledge and data engineering, 35(4):3313–3332, 2021

    2021

  10. [17]

    Gen- eralized gaussian scale mixtures: A model for wavelet coefficients of natural images

    Praful Gupta, Anush Krishna Moorthy, Rajiv Soundararajan, and Alan Conrad Bovik. Gen- eralized gaussian scale mixtures: A model for wavelet coefficients of natural images. Signal Processing: Image Communication, 66:87–94, 2018

    2018

  11. [18]

    Wavelet score-based generative modeling

    Florentin Guth, Simon Coste, Valentin De Bortoli, and Stephane Mallat. Wavelet score-based generative modeling. Advances in Neural Information Processing Systems, 35:478–491, 2022. 10

    2022

  12. [20]

    Gans trained by a two time-scale update rule converge to a local nash equilibrium

    Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, and Sepp Hochreiter. Gans trained by a two time-scale update rule converge to a local nash equilibrium. Advances in neural information processing systems, 30, 2017

    2017

  13. [21]

    Denoising diffusion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffusion probabilistic models. Advances in neural information processing systems, 33:6840–6851, 2020

    2020

  14. [22]

    Neural networks and physical systems with emergent collective computational abilities

    John J Hopfield. Neural networks and physical systems with emergent collective computational abilities. Proceedings of the national academy of sciences, 79(8):2554–2558, 1982

    1982

  15. [23]

    Wavedm: Wavelet-based diffusion models for image restoration

    Yi Huang, Jiancheng Huang, Jianzhuang Liu, Yu Dong, Jiaxi Lv, and Shifeng Chen. Wavedm: Wavelet-based diffusion models for image restoration. arXiv preprint arXiv:2305.13819, 2023

    2023 arXiv

  16. [24]

    Neural wavelet-domain diffusion for 3d shape generation

    Ka-Hei Hui, Ruihui Li, Jingyu Hu, and Chi-Wing Fu. Neural wavelet-domain diffusion for 3d shape generation. In SIGGRAPH Asia 2022 Conference Papers, pages 1–9, 2022

    2022

  17. [25]

    Low-light image enhancement with wavelet-based diffusion models

    Hai Jiang, Ao Luo, Songchen Han, Haoqiang Fan, and Shuaicheng Liu. Low-light image enhancement with wavelet-based diffusion models. arXiv preprint arXiv:2306.00306, 2023

    2023 arXiv

  18. [26]

    Progressive growing of gans for improved quality, stability, and variation

    Tero Karras, Timo Aila, Samuli Laine, and Jaakko Lehtinen. Progressive growing of gans for improved quality, stability, and variation. arXiv preprint arXiv:1710.10196, 2017

    2017 arXiv

  19. [27]

    Adam: A method for stochastic optimization

    Diederik P Kingma and Jimmy Ba. Adam: A method for stochastic optimization. arXiv preprint arXiv:1412.6980, 2014

    2014 arXiv

  20. [28]

    Deep learning face attributes in the wild

    Ziwei Liu, Ping Luo, Xiaogang Wang, and Xiaoou Tang. Deep learning face attributes in the wild. In Proceedings of International Conference on Computer Vision (ICCV), December 2015

    2015

  21. [29]

    Decoupled weight decay regularization

    Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization. arXiv preprint arXiv:1711.05101, 2017

    2017 arXiv

  22. [30]

    On distillation of guided diffusion models

    Chenlin Meng, Robin Rombach, Ruiqi Gao, Diederik Kingma, Stefano Ermon, Jonathan Ho, and Tim Salimans. On distillation of guided diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 14297–14306, 2023

    2023

  23. [31]

    Waving goodbye to low-res: A diffusion-wavelet approach for image super-resolution

    Brian B Moser, Stanislav Frolov, Federico Raue, Sebastian Palacio, and Andreas Dengel. Waving goodbye to low-res: A diffusion-wavelet approach for image super-resolution. CoRR, abs/2304.01994, 2023

    2023 arXiv

  24. [32]

    Attention u-net: Learning where to look for the pancreas

    Ozan Oktay, Jo Schlemper, Loic Le Folgoc, Matthew Lee, Mattias Heinrich, Kazunari Misawa, Kensaku Mori, Steven McDonagh, Nils Y Hammerla, Bernhard Kainz, et al. Attention u-net: Learning where to look for the pancreas. arXiv preprint arXiv:1804.03999, 2018

    2018 arXiv

  25. [33]

    Wavenet: A generative model for raw audio

    Aaron van den Oord, Sander Dieleman, Heiga Zen, Karen Simonyan, Oriol Vinyals, Alex Graves, Nal Kalchbrenner, Andrew Senior, and Koray Kavukcuoglu. Wavenet: A generative model for raw audio. arXiv preprint arXiv:1609.03499, 2016

    2016 arXiv

  26. [34]

    Unsupervised medical image translation with adversarial diffusion models

    Muzaffer Özbey, Onat Dalmaz, Salman UH Dar, Hasan A Bedel, ¸ Saban Özturk, Alper Güngör, and Tolga Çukur. Unsupervised medical image translation with adversarial diffusion models. IEEE Transactions on Medical Imaging, 2023

    2023

  27. [35]

    Wavelet diffusion models are fast and scalable image generators

    Hao Phung, Quan Dao, and Anh Tran. Wavelet diffusion models are fast and scalable image generators. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 10199–10208, 2023

    2023

  28. [36]

    U-net: Convolutional networks for biomedical image segmentation

    Olaf Ronneberger, Philipp Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. In Medical Image Computing and Computer-Assisted Intervention–MICCAI 2015: 18th International Conference, Munich, Germany, October 5-9, 2015, Proceedings, Part ...

    2015

  29. [37]

    A central limit theorem and a strong mixing condition

    Murray Rosenblatt. A central limit theorem and a strong mixing condition. Proceedings of the national Academy of Sciences, 42(1):43–47, 1956. 11

    1956

  30. [38]

    Statistics of natural images: Scaling in the woods

    Daniel L Ruderman and William Bialek. Statistics of natural images: Scaling in the woods. Physical review letters, 73(6):814, 1994

    1994

  31. [39]

    Progressive distillation for fast sampling of diffusion models

    Tim Salimans and Jonathan Ho. Progressive distillation for fast sampling of diffusion models. In International Conference on Learning Representations, 2022

    2022

  32. [40]

    Modeling the joint statistics of images in the wavelet domain

    Eero P Simoncelli. Modeling the joint statistics of images in the wavelet domain. In Wavelet Applications in Signal and Image Processing VII, volume 3813, pages 188–195. SPIE, 1999

    1999

  33. [41]

    Natural image statistics and neural representation

    Eero P Simoncelli and Bruno A Olshausen. Natural image statistics and neural representation. Annual review of neuroscience, 24(1):1193–1216, 2001

    2001

  34. [42]

    Generative modeling by estimating gradients of the data distribution

    Yang Song and Stefano Ermon. Generative modeling by estimating gradients of the data distribution. Advances in neural information processing systems, 32, 2019

    2019

  35. [43]

    Score-based generative modeling through stochastic differential equations

    Yang Song, Jascha Sohl-Dickstein, Diederik P Kingma, Abhishek Kumar, Stefano Ermon, and Ben Poole. Score-based generative modeling through stochastic differential equations. arXiv preprint arXiv:2011.13456, 2020

    2011 arXiv

  36. [44]

    Consistency models

    Yang Song, Prafulla Dhariwal, Mark Chen, and Ilya Sutskever. Consistency models. arXiv preprint arXiv:2303.01469, 2023

    2023 arXiv

  37. [45]

    Modelling the power spectra of natural images: statistics and information

    van A Van der Schaaf and JH van van Hateren. Modelling the power spectra of natural images: statistics and information. Vision research, 36(17):2759–2770, 1996

    1996

  38. [46]

    Scale mixtures of gaussians and the statistics of natural images

    Martin J Wainwright and Eero Simoncelli. Scale mixtures of gaussians and the statistics of natural images. Advances in neural information processing systems, 12, 1999

    1999

  39. [47]

    Diffusion-gan: Training gans with diffusion

    Zhendong Wang, Huangjie Zheng, Pengcheng He, Weizhu Chen, and Mingyuan Zhou. Diffusion-gan: Training gans with diffusion. arXiv preprint arXiv:2206.02262, 2022

    2022 arXiv

  40. [48]

    Image quality assessment: from error visibility to structural similarity

    Zhou Wang, Alan C Bovik, Hamid R Sheikh, and Eero P Simoncelli. Image quality assessment: from error visibility to structural similarity. IEEE transactions on image processing, 13(4): 600–612, 2004

    2004

  41. [49]

    Tackling the generative learning trilemma with denoising diffusion gans

    Zhisheng Xiao, Karsten Kreis, and Arash Vahdat. Tackling the generative learning trilemma with denoising diffusion gans. arXiv preprint arXiv:2112.07804, 2021

    2021 arXiv

  42. [50]

    1√ 2 1√ 21√ 2 1√ 2 # , G = gT × g =

    Huangjie Zheng, Pengcheng He, Weizhu Chen, and Mingyuan Zhou. Truncated diffusion proba- bilistic models and diffusion-based adversarial auto-encoders. arXiv preprint arXiv:2202.09671, 2022. 12 A Proof of Theorem 1 Theorem 5. Let N ∈ N, ∆t >0, and T = N ∆t. Then, we have that ...

    2022 arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.