Pith. sign in

REVIEW 2 major objections 6 minor 54 references

An overview of diffusion models for generative artificial intelligence

T0 review · 2 major / 6 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 rigorous upper bound shows why diffusion models train as noise predictors.

desk verdict A careful, mathematically precise survey of DDPMs with no new results; the derivations hold up, but the prior-mismatch term in the main bound is less controlled than the text suggests. read the letter →

arxiv 2412.01371 v1 pith:ENFSSXSY submitted 2024-12-02 cs.LG cs.AI

Davide Gallon , Arnulf Jentzen , Philippe von Wurstemberger This is my paper · ORCID
classification cs.LGcs.AI MSC 68T0760J2062B10
keywords denoisingdiffusionprobabilisticmodelsgenerativeartificialintelligenceexpectednegativelog-likelihoodKullback-LeiblerdivergencenoisepredictionlossGaussiantransitionkernelsUNetarchitectureclassifier-freeguidance
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 overview paper builds a rigorous mathematical framework for denoising diffusion probabilistic models and establishes the central bound that makes them work. It shows that the expected negative log-likelihood of the data under the generative model is controlled by an upper bound that splits into one-step terms, and in the Gaussian case each trainable term becomes a weighted squared error between the noise added by the forward process and a neural network's prediction. The paper then presents the simplified training and sampling methods built on this bound, along with the main extensions: improved variance learning and schedules, faster implicit sampling, classifier-free guidance, and text-conditioned latent diffusion. A sympathetic reader would take away a proof-level explanation of why the widely used denoising loss is a valid training objective, not just an empirical recipe.

What carries the argument

The load-bearing object is the ENLL upper bound in Lemma 2.9, specialized to Gaussian transition kernels in Proposition 3.19. The bound decomposes the negative log-likelihood of the data under the generative model into an unlearnable terminal term measuring how far the forward process's final state is from the standard-normal prior, the reconstruction loss of the last denoising step, and per-step KL divergences between the forward process's backward transition and the network's backward transition. With Gaussian noise and the reparameterization $X^\emptyset_t = \sqrt{\tilde{\alpha}_t}X^\emptyset_0 + \sqrt{1-\tilde{\alpha}_t}E_t$, each KL term becomes a squared error between the true noise $E_t$ and the network prediction $V^\theta(\sqrt{\tilde{\alpha}_t}X^\emptyset_0 + \sqrt{1-\tilde{\alpha}_t}E_t, t)$. That identity is what converts a hard variational objective into the simple SGD-updated loss used in Method 3.21.

What would settle it

Train the Gaussian DDPM from Method 3.21 with a noise schedule that keeps the cumulative coefficient $\tilde{\alpha}_T$ clearly away from zero, for example $\tilde{\alpha}_T=0.5$, so the terminal distribution of the forward process is far from standard normal, and measure whether the generated samples still match the training distribution; if they do, the near-zero assumption is unnecessary, and if they do not, that failure pinpoints where the bound stops being informative.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is a chain of inequalities. Lemma 2.9 shows that, for any Markov forward and backward process, the ENLL $H(p^\emptyset_0 \| p^\theta_0)$ is bounded by the KL divergence between the forward terminal distribution given the initial state and the prior $\Pi$, the negative log-probability of the final denoising step, and a sum of KL divergences between the true backward transition of the forward process and the backward transition of the parametric process. When the forward transitions are Gaussian, Proposition 3.19 rewrites each trainable KL term as the weighted squared error $\frac{1}{2\tilde{\beta}_t}\frac{(1-\alpha_t)^2}{(1-\tilde{\alpha}_t)\alpha_t}\,\mathbb{E}[\|E_t - V^\theta(\sqrt{\tilde{\alpha}_t}X^\emptyset_0 + \sqrt{1-\tilde{\alpha}_t}E_t, t)\|^2]$, up to the unlearnable terminal term and a constant. This identity is what reduces the abstract variational objective to the simple noise-prediction loss of Method 3.21, and it is the statement that the rest of the survey's methods inherit.

Load-bearing premise

The whole scheme leans on the informal assumption that the forward process destroys enough information that its terminal state is close to standard normal, since the clean limiting choice $\tilde{\alpha}_0=1$ and $\tilde{\beta}_1=0$ would make the variance zero and is outside the range where the bound and the Gaussian densities are defined.

Editorial extensions

If this is right

  • The ENLL of the data under a DDPM is bounded by a sum of one-step KL terms, so training can proceed by stochastic gradient descent on single time steps rather than on full trajectories.
  • In the Gaussian case, each trainable term reduces to a weighted squared error between the true forward noise and the network's prediction, up to an unlearnable terminal term and a constant.
  • Ignoring the weights and the terminal term, as Method 3.21 does, yields the standard simplified training loss used by the DDPM variants discussed in the paper.
  • Improved DDPMs, DDIM, classifier-free guidance, and stable diffusion all keep the same underlying noise-prediction objective while changing variances, schedules, sampling steps, or conditioning.
  • Because the terminal KL term has no trainable parameters, the quality of the generative model's starting point depends on the forward noise schedule rather than on the network.

Reading between the lines

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

  • A consequence the paper leaves implicit is that the per-step weights in Proposition 3.19 give a quantitative handle on schedule design: choosing the noise coefficients is choosing how much each denoising step contributes to the bound, so schedules that keep the weighted terms balanced should train more stably.
  • The framework suggests a testable extension: replacing the fixed standard-normal prior with a learned or fitted prior would relax the requirement that the terminal noise coefficient be near zero, potentially improving generation when the forward process has not fully mixed to noise.
  • The same upper-bound structure connects to score-matching interpretations of diffusion models, since the squared noise-prediction error is an amortized denoising score-matching objective, though the paper does not make this identification.
  • One could place the 'as if' choice $\tilde{\alpha}_0=1$, $\tilde{\beta}_1=0$ on firmer ground by taking a limit of admissible parameters instead of fixing them directly, which would make the zero-variance final step of Method 3.21 rigorous.
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

2 major / 6 minor

Summary. The paper is a tutorial/survey of denoising diffusion probabilistic models (DDPMs). It introduces a general Markovian framework (Settings 2.1 and 2.3), derives an upper bound for the expected negative log-likelihood (Lemma 2.9), specializes the transition kernels to Gaussians (Setting 3.8), and obtains the reparameterized upper bound of Proposition 3.19. This bound motivates the simplified noise-prediction loss and the backward sampling scheme of Method 3.21. The remainder of the paper reviews evaluation metrics (IS, FID, SSIM, PSNR, LPIPS) and extensions: Improved DDPM, DDIM, classifier-free guidance, Stable Diffusion, GLIDE, DALL-E 2/3, and Imagen. The technical core reproduces known results from Sohl-Dickstein et al. (2015), Ho et al. (2020), Song et al. (2020), and Nichol and Dhariwal (2021).

Significance. The paper's main value is pedagogical: it gives explicit, mostly self-contained derivations of the standard DDPM variational bound and its Gaussian simplifications, and it presents the practical algorithms in a unified notation. The derivations in Sections 2-3 appear mathematically correct, and the companion code repository is a useful resource. The paper does not claim new bounds or new empirical results, and the DDIM theorem (Theorem 5.7) is imported from the literature. As a survey, this is appropriate; however, the paper's stronger framing of Proposition 3.19 as a 'complete theoretical basis' for Method 3.21 requires qualification because of the prior-mismatch and boundary-parameter issues discussed below.

major comments (2)
  1. [§3.4, Proposition 3.19(i), Remark 3.14] The informal condition 'α̃T ≈ 0' in Remark 3.14 is not sufficient in high dimension. For the Gaussian forward process, the prior-mismatch term in Proposition 3.19(i) equals (1/2)[α̃T E[||X∅0||²] − d α̃T − d ln(1−α̃T)], which behaves like (α̃T/2) E[||X∅0||²] for small α̃T. For data normalized to [−1,1]^d this is O(α̃T d). The condition for the bound to be a meaningful guarantee is therefore α̃T E[||X∅0||²] ≪ 1, not merely α̃T ≈ 0. Since Remark 2.10(iii) correctly observes that this term has no trainable parameters, minimizing the simplified loss in Method 3.21 cannot reduce it. Please add a quantitative statement and discuss the schedules of Remarks 3.23 and 5.3 in this light.
  2. [§3.4–3.5, Remark 3.20, Method 3.21] The transition from Proposition 3.19 to Method 3.21 uses values α̃0=1 and β̃1=0 that are explicitly outside the admissible range of the Gaussian densities and make the bound in Proposition 3.19 undefined (division by zero). The paper acknowledges this in Remark 3.20 but still states before Proposition 3.19 that the result provides 'a complete theoretical basis' for the scheme in Method 3.21. Since the sampling update (87) and the simplified loss (88) depend on this inadmissible limit, the paper should either prove that the error introduced by the limit vanishes (e.g., by a limiting argument with β̃1 → 0) or explicitly state that Method 3.21 is a heuristic variant of the rigorous bound.
minor comments (6)
  1. [§2.1, Setting 2.1] The symbol d is used both for the data dimension and for the parameter dimension (the statement begins 'Let d, d, T ∈ N' and later θ ∈ R^d while X^θ_t takes values in R^d); this notational collision makes the framework hard to read, so please use distinct symbols such as d_x and d_θ.
  2. [§4.1] The subsection is titled 'Content variant metrics', but the text and Remark 4.3 describe IS and FID as 'content invariant metrics' and say content-invariant metrics measure diversity; the terminology should be made consistent.
  3. [Definition 4.2] The symbol I is used both for the Inception model and for the score; rename the score (e.g., IS) to avoid confusion.
  4. [Method 5.1, Eq. (106)] In the sampling update, the variance interpolation term uses v^{Θ_N}_{2,i}(X_k, t) with a time variable t that is not defined in the loop; it should presumably be t_k.
  5. [§5.1] The opening sentence refers to 'Improved DDPM [15]' and attributes the ENLL findings to [15], but [15] is Ho et al.'s original DDPM paper; the Improved DDPM method and the ENLL observations come from Nichol and Dhariwal [30].
  6. [Theorem 5.7] The proof consists of a reference to [45, Theorem 1]; in a paper emphasizing mathematical rigor, either reproduce the argument or clearly state that the result is taken as an external theorem.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the DDPM upper-bound derivation is self-contained given its assumptions, and the only self-citation is a non-load-bearing general reference.

full rationale

The main derivation chain runs from Lemma 2.9 through Lemma 3.18 to Proposition 3.19. Each step is a direct mathematical consequence of the stated Markov and Gaussian assumptions: Lemma 2.9 uses Jensen's inequality and the Markov factorization, Lemma 3.18 substitutes the chosen Gaussian parameterization, and Proposition 3.19 assembles the pieces. No fitted constant or empirical value is introduced into this derivation. The prior-mismatch term E[DKL(p^∅_{T|0}(·|X^∅_0)∥Π)] is explicitly identified in Remark 2.10(iii) as having no learnable parameters, so it is an input determined by the forward process and the noise schedule, not a disguised prediction of the model. Method 3.21 then replaces the Proposition 3.19 upper bound by the simplified denoising loss; the paper states that this simplified objective 'does therefore not exactly correspond to the upper bound' and attributes the simplification to empirical justification in Ho et al. [15]. This is an acknowledged heuristic, not a circular derivation. The paper also flags the limiting choices alpha-tilde_0 = 1 and beta-tilde_1 = 0 as inadmissible in Proposition 3.19 (Remark 3.20) and the terminal-distribution approximation as informal (Remark 3.14); these are limitation caveats about the strength of the guarantee, not circular reductions. The only self-citation is reference [19], used as a general introduction to SGD-type methods and ANN architectures; it carries no load in the DDPM derivation. The skeptical concern about the prior-mismatch term scaling as O(alpha-tilde_T E||X_0||^2) is a quantitative weakness of the guarantee, but the term is part of the bound's input, so no prediction reduces by construction to its own input.

Assumptions & free parameters 4 free parameters · 6 assumptions · 0 invented entities

The paper's derivations rely on standard probability and Gaussian identities, plus the domain assumptions of Markovianity and terminal noise convergence. There are no newly postulated entities. The free parameters listed are hand-chosen hyperparameters from the described algorithms, not parameters fitted to data by the authors.

free parameters (4)
  • Noise schedule α1,...,αT = linear: α1=1−10^{-4}, αT=0.98 (Remark 3.23); cosine scheduler uses offset s=0.008 (Remark 5.3)
    Chosen by hand following Ho et al. (2020) and Nichol and Dhariwal (2021); required for the practical generative methods but does not affect the validity of the upper-bound derivation.
  • Loss interpolation parameter λ = 0.001 (Remark 5.2)
    Set by hand in Improved DDPM to prioritize noise-prediction error over variance-prediction loss.
  • Class-drop probability p = 0.1 or 0.2 (Remark 5.13)
    Chosen by hand in classifier-free guidance training to decide how often class information is zeroed.
  • DDIM stochasticity η = η=0 for deterministic DDIM, η=1 recovers DDPM (Remark 5.9)
    Hand-chosen by user; controls sampling stochasticity.
assumptions (6)
  • standard math Jensen's inequality for the logarithm applied to the marginal pθ0(x0) (Lemma 2.9 proof, Eq. (25))
    Used to derive the ELBO-type upper bound; unconditional background fact.
  • standard math Gaussian convolution identity ∫ N(x, Ay+μ1, Σ1) N(y, μ2, Σ2) dy = N(x, Aμ2+μ1, AΣ2A*+Σ1) (Lemma 3.2, cited to Bishop 2006)
    Basis for the multi-step forward distributions and DDIM consistency.
  • standard math Bayes rule for Gaussian conditionals (Lemma 3.5, cited to Bishop 2006)
    Yields the backward transition density p∅t−1|t,0 in Lemma 3.15.
  • standard math Closed-form KL divergence between Gaussians (Lemma 3.7, cited to Duchi 2007)
    Converts KL terms in the upper bound into squared 2-norms.
  • domain assumption Markov property of forward and backward processes (Setting 2.3, Eq. (6)-(7))
    Required for the factorization of joint densities in Lemma 2.5 and the per-step loss decomposition.
  • domain assumption Terminal distribution approximation p∅T ≈ Π, i.e., α̃T ≈ 0 (Remark 3.14, Eq. (60))
    The backward process starts from a standard normal prior; exact equality is only approached as α̃T→0, which the method treats as approximately true.

how reviews work

0 comments
Cite this review

Pith. "Pith review of An overview of diffusion models for generative artificial intelligence." pith.science (2026). https://pith.science/paper/ENFSSXSY

@misc{pith2026241201371,
  author       = {Pith},
  title        = {Pith review of: An overview of diffusion models for generative artificial intelligence},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ENFSSXSY}},
  note         = {Machine review of arXiv:2412.01371}
}
read the original abstract

This article provides a mathematically rigorous introduction to denoising diffusion probabilistic models (DDPMs), sometimes also referred to as diffusion probabilistic models or diffusion models, for generative artificial intelligence. We provide a detailed basic mathematical framework for DDPMs and explain the main ideas behind training and generation procedures. In this overview article we also review selected extensions and improvements of the basic framework from the literature such as improved DDPMs, denoising diffusion implicit models, classifier-free diffusion guidance models, and latent diffusion models.

Figures

Figures reproduced from arXiv: 2412.01371 by the authors.

Figure 2.1
Figure 2.1. Graphical illustration the forward process [PITH_FULL_IMAGE:figures/full_fig_p006_2_1.png] view at source ↗
Figure 3.1
Figure 3.1. Graphical illustration of (˜αt)t∈{1,...,T} in Method 3.21 for T = 1000 and (αt)t∈{1,...,T} given as in (90). 3.6 Network architectures for the backward process In this section we discuss the most popular choice for the architecture of the ANN (Vθ )θ∈Rd from Method 3.21. Specifically, we explain UNets in Subsection 3.6.1 and present how the temporal component is commonly incorporated in Subsection 3.6.2. For general … view at source ↗
Figure 3.2
Figure 3.2. Graphical illustration of a typical UNet architecture in case of two dimensional data [PITH_FULL_IMAGE:figures/full_fig_p031_3_2.png] view at source ↗
Figures from the paper (2 more)
Figure 3.3
Figure 3.3. Figure 3.3: Sinusoidal time embedding for 1000 time step using as embedding dimension 64. [PITH_FULL_IMAGE:figures/full_fig_p032_3_3.png]
Figure 5.1
Figure 5.1. Figure 5.1: Forward diffusion process using a linear scheduler on top and a cosine scheduler at [PITH_FULL_IMAGE:figures/full_fig_p040_5_1.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

54 extracted references · 30 canonical work pages

  1. [1]

    L., Almeida, D., Altenschmidt, J., Altman, S., Anadkat, S., et al

    Achiam, J., Adler, S., Agarwal, S., Ahmad, L., Akkaya, I., Aleman, F. L., Almeida, D., Altenschmidt, J., Altman, S., Anadkat, S., et al. Gpt-4 technical report. arXiv:2303.08774 (2023)

    arXiv 2023

  2. [2]

    Comparison of image quality assessment: Psnr, hvs, ssim, uiqi

    Al-Najjar, Y. Comparison of image quality assessment: Psnr, hvs, ssim, uiqi. International Journal of Scientific and Engineering Research 3 (08 2012)

    work page 2012

  3. [3]

    D., Ho, J., Tarlow, D., and van den Berg, R.Structured denoising diffusion models in discrete state-spaces

    Austin, J., Johnson, D. D., Ho, J., Tarlow, D., and van den Berg, R.Structured denoising diffusion models in discrete state-spaces. arXiv:2107.03006 (2023)

    arXiv 2023

  4. [4]

    Learning theory from first principles

    Bach, F. Learning theory from first principles . MIT press, 2024

    work page 2024

  5. [5]

    Bayesian reasoning and machine learning

    Barber, D. Bayesian reasoning and machine learning . Cambridge University Press, 2012

    work page 2012

  6. [6]

    Improving image generation with better captions

    Betker, J., Goh, G., Jing, L., Brooks, T., W ang, J., Li, L., Ouyang, L., Zhuang, J., Lee, J., Guo, Y., et al. Improving image generation with better captions. Computer Science. https://cdn. openai. com/papers/dall-e-3. pdf 2 , 3 (2023), 8

    work page 2023

  7. [7]

    Bishop, C. M. Pattern Recognition and Machine Learning (Information Science and Statistics) . Springer-Verlag, Berlin, Heidelberg, 2006

    work page 2006

  8. [8]

    Diffusion models beat gans on image synthesis

    Dhariwal, P., and Nichol, A. Diffusion models beat gans on image synthesis. arXiv:2105.05233 (2021)

    arXiv 2021

Show all 54 references
  1. [9]

    Derivations for linear algebra and optimization

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

    2007

  2. [11]

    Deep Learning

    Goodfellow, I., Bengio, Y., and Courville, A. Deep Learning. MIT Press, 2016. http: //www.deeplearningbook.org

    2016

  3. [12]

    Generative adversarial nets

    Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., W arde-F arley, D., Ozair, S., Courville, A., and Bengio, Y. Generative adversarial nets. Advances in neural information processing systems 27 (2014)

    2014

  4. [13]

    Grimmett, G., and Welsh, D. J. Probability: an introduction . Oxford University Press, 2014

    2014

  5. [14]

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

    Heusel, M., Ramsauer, H., Unterthiner, T., Nessler, B., and Hochreiter, S. Gans trained by a two time-scale update rule converge to a local nash equilibrium. In Neural Information Processing Systems (2017)

    2017

  6. [15]

    Denoising diffusion probabilistic models

    Ho, J., Jain, A., and Abbeel, P. Denoising diffusion probabilistic models. Advances in Neural Information Processing Systems 33 (2020), 6840–6851

    2020

  7. [16]

    Classifier-free diffusion guidance

    Ho, J., and Salimans, T. Classifier-free diffusion guidance. arXiv:2207.12598 (2022)

    2022 arXiv

  8. [17]

    Ho, J., Salimans, T., Gritsenko, A., Chan, W., Norouzi, M., and Fleet, D. J. Video diffusion models. arXiv:2204.03458 (2022)

    2022 arXiv

  9. [18]

    Image quality metrics: Psnr vs

    Hor´e, A., and Ziou, D. Image quality metrics: Psnr vs. ssim. In 2010 20th International Conference on Pattern Recognition (2010), pp. 2366–2369

    2010

  10. [19]

    Mathematical introduction to deep learning: methods, implementations, and theory

    Jentzen, A., Kuckuck, B., and von Wurstemberger, P. Mathematical introduction to deep learning: methods, implementations, and theory. arXiv:2310.20360 (2023). 53

    2023 arXiv

  11. [20]

    Analyzing and improving the image quality of stylegan

    Karras, T., Laine, S., Aittala, M., Hellsten, J., Lehtinen, J., and Aila, T. Analyzing and improving the image quality of stylegan. In 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (Los Alamitos, CA, USA, jun 2020), IEEE Computer Society, pp. 8107–8116

    2020

  12. [21]

    Kim, G., Kwon, T., and Ye, J. C. Diffusionclip: Text-guided diffusion models for robust image manipulation. arXiv:2110.02711 (2022)

    2022 arXiv

  13. [22]

    P., and Welling, M

    Kingma, D. P., and Welling, M. Auto-encoding variational bayes. In 2nd International Confer- ence on Learning Representations, ICLR 2014, Banff, AB, Canada, April 14-16, 2014, Conference Track Proceedings (2014), Y. Bengio and Y. LeCun, Eds

    2014

  14. [23]

    Probability theory: a comprehensive course

    Klenke, A. Probability theory: a comprehensive course . Springer Science & Business Media, 2013

    2013

  15. [24]

    Kullback, S., and Leibler, R. A. On information and sufficiency. The annals of mathematical statistics 22 , 1 (1951), 79–86

    1951

  16. [25]

    A tutorial on energy- based learning

    LeCun, Y., Chopra, S., Hadsell, R., Ranzato, M., Huang, F., et al. A tutorial on energy- based learning. Predicting structured data 1 , 0 (2006)

    2006

  17. [26]

    L., Thickstun, J., Gulrajani, I., Liang, P., and Hashimoto, T

    Li, X. L., Thickstun, J., Gulrajani, I., Liang, P., and Hashimoto, T. B. Diffusion-lm improves controllable text generation. arXiv:2205.14217 (2022)

    2022 arXiv

  18. [27]

    Cat: Cross attention in vision transformer

    Lin, H., Cheng, X., Wu, X., Yang, F., Shen, D., W ang, Z., Song, Q., and Yuan, W. Cat: Cross attention in vision transformer. arXiv:2006.09011 (2021)

    2021 arXiv

  19. [28]

    L., and Doll ´ar, P

    Lin, T.-Y., Maire, M., Belongie, S., Bourdev, L., Girshick, R., Hays, J., Perona, P., Ramanan, D., Zitnick, C. L., and Doll ´ar, P. Microsoft coco: Common objects in context. arXiv:1405.0312 (2015)

    2015 arXiv

  20. [29]

    Fully convolutional networks for semantic segmen- tation

    Long, J., Shelhamer, E., and Darrell, T. Fully convolutional networks for semantic segmen- tation. In 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2015), IEEE Computer Society, pp. 3431–3440

    2015

  21. [30]

    Improved denoising diffusion probabilistic models

    Nichol, A., and Dhariwal, P. Improved denoising diffusion probabilistic models. arXiv:2102.09672 (2021)

    2021 arXiv

  22. [31]

    Glide: Towards photorealistic image generation and editing with text-guided diffusion models

    Nichol, A., Dhariwal, P., Ramesh, A., Shyam, P., Mishkin, P., McGrew, B., Sutskever, I., and Chen, M. Glide: Towards photorealistic image generation and editing with text-guided diffusion models. arXiv:2203.13131 (2022)

    2022 arXiv

  23. [32]

    Understanding ssim

    Nilsson, J., and Akenine-M ¨oller, T. Understanding ssim. arXiv:2006.13846 (2020)

    2020 arXiv

  24. [33]

    W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., Krueger, G., and Sutskever, I

    Radford, A., Kim, J. W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., Krueger, G., and Sutskever, I. Learning transferable visual models from natural language supervision. CoRR abs/2103.00020 (2021)

    2021 arXiv

  25. [34]

    Raffel, C., Shazeer, N., Roberts, A., Lee, K., Narang, S., Matena, M., Zhou, Y., Li, W., and Liu, P. J. Exploring the limits of transfer learning with a unified text-to-text transformer. Journal of Machine Learning Research 21 , 140 (2020), 1–67

    2020

  26. [35]

    Hierarchical text-conditional image generation with clip latents

    Ramesh, A., Dhariwal, P., Nichol, A., Chu, C., and Chen, M. Hierarchical text-conditional image generation with clip latents. arXiv:2204.06125 (2022)

    2022 arXiv

  27. [36]

    Zero-shot text-to-image generation

    Ramesh, A., Pavlov, M., Goh, G., Gray, S., Voss, C., Radford, A., Chen, M., and Sutskever, I. Zero-shot text-to-image generation. In Proceedings of the 38th International Con- ference on Machine Learning (18–24 Jul 2021), M. Meila and T. Zhang, Eds., vol. 139 of Proceedings of...

    2021

  28. [37]

    J., and Mohamed, S.Variational inference with normalizing flows.arXiv:1505.05770 (2016)

    Rezende, D. J., and Mohamed, S.Variational inference with normalizing flows.arXiv:1505.05770 (2016)

    2016 arXiv

  29. [38]

    High-resolution image synthesis with latent diffusion models

    Rombach, R., Blattmann, A., Lorenz, D., Esser, P., and Ommer, B. High-resolution image synthesis with latent diffusion models. In 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (2022), pp. 10674–10685

    2022

  30. [39]

    An overview of gradient descent optimization algorithms

    Ruder, S. An overview of gradient descent optimization algorithms. arXiv:1609.04747 (2016)

    2016 arXiv

  31. [40]

    Saharia, C., Chan, W., Saxena, S., Li, L., Whang, J., Denton, E., Ghasemipour, S. K. S., Ayan, B. K., Mahdavi, S. S., Lopes, R. G., Salimans, T., Ho, J., Fleet, D. J., and Norouzi, M. Photorealistic text-to-image diffusion models with deep language understanding. arXiv:2205.11...

    2022 arXiv

  32. [41]

    J., and Norouzi, M

    Saharia, C., Ho, J., Chan, W., Salimans, T., Fleet, D. J., and Norouzi, M. Image super- resolution via iterative refinement. IEEE Transactions on Pattern Analysis and Machine Intelligence 45, 4 (2023), 4713–4726

    2023

  33. [42]

    Improved techniques for training gans

    Salimans, T., Goodfellow, I., Zaremba, W., Cheung, V., Radford, A., and Chen, X. Improved techniques for training gans. arXiv:1606.03498 (2016)

    2016 arXiv

  34. [43]

    Understanding machine learning: From theory to algorithms

    Shalev-Shwartz, S., and Ben-David, S. Understanding machine learning: From theory to algorithms. Cambridge university press, 2014

    2014

  35. [44]

    Deep unsupervised learning using nonequilibrium thermodynamics

    Sohl-Dickstein, J., Weiss, E., Maheswaranathan, N., and Ganguli, S. Deep unsupervised learning using nonequilibrium thermodynamics. In International Conference on Machine Learning (2015), PMLR, pp. 2256–2265

    2015

  36. [45]

    Denoising diffusion implicit models

    Song, J., Meng, C., and Ermon, S. Denoising diffusion implicit models. arXiv:2010.02502 (2022)

    2022 arXiv

  37. [46]

    Rethinking the inception architecture for computer vision

    Szegedy, C., V anhoucke, V., Ioffe, S., Shlens, J., and Wojna, Z. Rethinking the inception architecture for computer vision. In Proceedings of the IEEE conference on computer vision and pattern recognition (2016), pp. 2818–2826

    2016

  38. [47]

    Conditional image generation with pixelcnn decoders

    van den Oord, A., Kalchbrenner, N., Vinyals, O., Espeholt, L., Graves, A., and Kavukcuoglu, K. Conditional image generation with pixelcnn decoders. arXiv:1606.05328 (2016)

    2016 arXiv

  39. [48]

    N., Kaiser, L

    V aswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L. u., and Polosukhin, I. Attention is all you need. In Advances in Neural Information Processing Systems (2017), I. Guyon, U. V. Luxburg, S. Bengio, H. Wallach, R. Fergus, S. Vishwanathan, ...

    2017

  40. [49]

    Image quality assessment: from error visibility to structural similarity

    W ang, Z., Bovik, A., Sheikh, H., and Simoncelli, E. Image quality assessment: from error visibility to structural similarity. IEEE Transactions on Image Processing 13 , 4 (2004), 600–612

    2004

  41. [50]

    Wolleb, J., Bieder, F., Sandk ¨uhler, R., and Cattin, P. C. Diffusion models for medical anomaly detection. arXiv:2203.04306 (2022)

    2022 arXiv

  42. [51]

    Group normalization

    Wu, Y., and He, K. Group normalization. arXiv:1803.08494 (2018)

    2018 arXiv

  43. [52]

    M., and Willcocks, C

    Wyatt, J., Leach, A., Schmon, S. M., and Willcocks, C. G. Anoddpm: Anomaly detection with denoising diffusion probabilistic models using simplex noise. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops (June 2022), pp. 650– 656

    2022

  44. [53]

    Diffusion probabilistic modeling for video generation

    Yang, R., Srivastava, P., and Mandt, S. Diffusion probabilistic modeling for video generation. arXiv:2203.09481 (2022). 55

    2022 arXiv

  45. [54]

    A., Shechtman, E., and W ang, O

    Zhang, R., Isola, P., Efros, A. A., Shechtman, E., and W ang, O. The unreasonable effectiveness of deep features as a perceptual metric. In Proceedings of the IEEE conference on computer vision and pattern recognition (2018), pp. 586–595

    2018

  46. [55]

    Towards language-free training for text-to-image generation

    Zhou, Y., Zhang, R., Chen, C., Li, C., Tensmeyer, C., Yu, T., Gu, J., Xu, J., and Sun, T. Towards language-free training for text-to-image generation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (June 2022), pp. 17907–17917. 56

    2022

Pith tools

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