Pith. sign in

REVIEW 4 major objections 4 minor 55 references

SDDBMs: Soft Denoising Diffusion Bridge Models

T0 review · 4 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read Soft Gaussian endpoints remove the terminal singularity of diffusion bridges

desk verdict Solid closed-form soft-bridge theory with a real gap between the marginal-only x0-free reformulation and the reverse SDE used in practice; deserves review, but performance claims need an asterisk. read the letter →

arxiv 2608.08594 v2 pith:VWGWTYOC submitted 2026-08-09 cs.AI

Shiyi Qi , Kun He , Mingmou Liu This is my paper · ORCID
classification cs.AI MSC 60H1060J6068T07
keywords diffusionbridgesDoobh-transformterminalsingularitiessoftconstraintsGaussianmarginalimagerestorationconditionalscorematchingstochasticoptimalcontrol
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

Diffusion bridge models that condition on an exact endpoint pay a structural price: the terminal law collapses to a Dirac measure and the drift coefficients diverge as $t\to T$. This paper establishes that the cure is to prescribe a non-degenerate Gaussian terminal marginal $\mathcal{N}(x^\star,\sigma^2 I)$ under the Doob $h$-transformed measure, which yields a closed-form soft $h$-function, closed-form forward marginals, and an $x_0$-free reverse dynamics whose drift denominators never vanish. Because $\widehat{\sigma}^2>0$ keeps the terminal drift finite, sampling can start at $t=T$ and the noise scale does not degrade at the boundary. The same construction recovers DDBMs, GOUB, and UniDB as parameter limits, and the reported experiments on deraining, super-resolution, and inpainting improve over the compared baselines.

What carries the argument

The central object is the soft $h$-function obtained by Gaussian terminal reweighting: for the GOU reference bridge with transition kernel $p(z,T|x_t,t)=\mathcal{N}(z;\bar{\mu}_{t:T},\bar{\sigma}^2_{t:T}I)$ and calibrated terminal weight $q(z)=\mathcal{N}(z;\widehat{x},\widehat{\sigma}^2 I)$, the function $h(x_t,t)=\int p(z,T|x_t,t)q(z)\,dz$ is available in closed form, $h\propto\exp(-\|\bar{\mu}_{t:T}-\widehat{x}\|^2/[2(\widehat{\sigma}^2+\bar{\sigma}^2_{t:T})])$. Its gradient adds a drift correction whose denominator $\widehat{\sigma}^2+\bar{\sigma}^2_{t:T}$ stays strictly positive at $t=T$ precisely because $\widehat{\sigma}^2>0$, so the terminal drift no longer blows up. Theorem 3.2 fixes $\widehat{x}$ and $\widehat{\sigma}^2$ uniquely from the prescribed terminal marginal $\mathcal{N}(x^\star,\sigma^2 I)$, yielding closed-form Gaussian forward marginals and a well-posed $x_0$-free reformulation; the admissibility condition $\phi_t\neq0$ (equivalently the relaxation parameter $\rho>0$) delineates the usable parameter regime.

What would settle it

Run the trained SDDBM reverse sampler, record the empirical terminal samples it produces, and compare their covariance to the prescribed $\sigma^2 I$; if the covariance is far from $\sigma^2 I$, or if shrinking $\sigma$ collapses the terminal law while the drift stays finite, the sampler is not realizing the soft bridge. A second check is to retrain with the exact $x_0$-conditioned soft bridge and compare against the $x_0$-free variant that starts at $(b_T+c_T)x_\star$; a sharp quality drop would show that the approximation, not the soft terminal constraint, carries the reported gains.

Watch

Extended reading notes

Core claim

SDDBMs claim that a finite-coefficient, $x_0$-free affine bridge cannot both forget its source and collapse to a point: its terminal law is necessarily a non-degenerate Gaussian whose mean still depends on $x_0$, so hard endpoint conditioning can only be realized through drift blow-up. The paper therefore prescribes the terminal marginal under the $h$-transformed measure to be $\mathcal{N}(x^\star,\sigma^2 I)$ (with a soft center $x^*$ that may differ slightly from the desired target $x_\star$), proves that the unique Gaussian terminal weight realizing it is $q=\mathcal{N}(\widehat{x},\widehat{\sigma}^2 I)$ with the calibration in Eq. (12), and obtains a closed-form soft $h$-function, closed-form forward marginals, and an $x_0$-free reverse SDE with non-vanishing terminal noise. Under specific hyper-parameter choices the framework recovers DDBM-VP, DDBM-VE, GOUB, and UniDB; in particular UniDB's terminal penalty $\kappa$ becomes an explicit terminal variance $\sigma^2=\bar{\sigma}^2_{0:T}/(\kappa\bar{\sigma}^2_{0:T}+1)$. On deraining, super-resolution, and inpainting, the reported PSNR/SSIM/LPIPS/FID improve over the compared baselines.

Load-bearing premise

The load-bearing premise is the unproved approximation that the $x_0$-free reverse process—initialized at $x_T\approx(b_T+c_T)x_\star$, with the score network fed $x_\star$ and the drift rescaled by a per-task $\zeta$—remains faithful to the forward soft bridge; if that approximation fails, the mathematical clean-up and the reported gains come apart.

Editorial extensions

If this is right

  • Sampling can start exactly at $t=T$: the drift denominator $\widehat{\sigma}^2+\bar{\sigma}^2_{t:T}$ stays positive, so no $\epsilon$-truncation of the terminal interval is needed.
  • The terminal noise scale $\bar{\sigma}'^2_{0:T}$ remains positive, so the standard noise-prediction training objective applies uniformly without boundary-specific score parameterizations.
  • Tuning $\sigma$ sweeps a continuous family from a near-hard endpoint ($\sigma\to0$) to a weakly guided bridge, making the fidelity-versus-stability trade-off explicit.
  • The framework contains DDBMs, GOUB, and UniDB as special hyper-parameter cases, so analysis and improvements transfer to those models.
  • On the reported deraining, super-resolution, and inpainting benchmarks, SDDBMs beat the listed bridge and restoration baselines on PSNR, SSIM, LPIPS, and FID.

Reading between the lines

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

  • The same soft-terminal construction should transfer to Schrodinger-bridge or flow-matching transports that currently rely on exact endpoint coupling, with $\sigma$ acting as a tunable dial between target fidelity and sampling stability.
  • The paper's use of $x_\star$ in place of $x_T$ for the score network suggests a broader design principle: condition score networks on a clean target estimate rather than a noisy terminal state, which may extend beyond restoration to any paired translation task.
  • The empirical correction factor $\zeta$ hints that the Euler-discretized reverse drift is systematically biased; switching to a higher-order exponential integrator might remove the need for $\zeta$ and further improve the ODE results.
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 / 4 minor

Summary. The paper proposes Soft Denoising Diffusion Bridge Models (SDDBMs), a class of diffusion bridges in which the hard terminal condition x_T = x* is replaced by a prescribed non-degenerate Gaussian terminal marginal N(x*, σ²I) under the h-transformed path measure. The authors derive a Gaussian terminal weight q, a closed-form soft h-function (Theorem 3.3), closed-form forward marginals (Proposition 3.4), and an x0-free reformulation of the forward dynamics (Section 3.4 and Appendix E). They further derive a score-matching training objective and claim that the framework recovers DDBM-VP, DDBM-VE, GOUB, and UniDB as special cases (Proposition 4.1). Experimental results on deraining, super-resolution, and inpainting show improved quantitative metrics over the cited baselines. The paper is carefully written, and the main algebraic derivations of the soft h-function and the forward marginals are internally consistent under the stated regularity conditions.

Significance. If the full chain from the prescribed terminal marginal to the trained sampler were valid, the paper would make a genuine contribution: it gives a clean probabilistic mechanism for avoiding terminal-boundary singularities, provides explicit closed-form objects for a soft bridge, offers a new interpretation of UniDB, and reports strong empirical results on three restoration tasks. The theoretical sections are largely self-contained, and the unification of existing bridge models via parameter identification is appealing. The main value of the paper lies in replacing an endpoint Dirac constraint by a calibrated Gaussian terminal reweighting, which is a natural and potentially useful idea. However, the link between the continuous-time forward SDE and the discrete-time training/sampling procedure contains an acknowledged gap: the reformulated dynamics are only marginally consistent, not pathwise identical to the soft bridge, and the training posterior is constructed from an assumed one-step transition rather than derived from the SDE.

major comments (4)
  1. [Appendix E.3 / A.6] The reverse-time training objective is derived from a posterior p_h(x_{t-1} | x_0, x_t, x_T) that is computed using a 'Gaussian one-step forward transition' in Eq. (74) which is constructed to be consistent with the time marginals of Eq. (73), not derived as the true transition density of the SDE in Eq. (38). The paper itself states in Appendix E.3 that Eq. (38) 'should be interpreted as a marginally consistent reformulation ... rather than as a pathwise identity.' This matters because, for a Gaussian Markov process, matching one-time marginals does not determine the transition law, and the reverse SDE in Eq. (18) requires the conditional score ∇ log p_h(x_t | x_T) of the actual forward process. As written, the learned score may be the score of a different process than the one whose terminal marginal was prescribed. The authors should either prove that the transition in Eq. (74) coincides with the transition of Eq. (38), or explicitly reformulate the method as a discrete-time Markov bridge whose forward transition is Eq. (74) and whose reverse sampler is derived for that discrete process.
  2. [Appendix I] In the sampling procedure the paper substitutes x_T by (b_T + c_T)x* and trains and evaluates the score network with x* in place of x_T, dropping the source-dependent and noise terms. This substitution is not derived from the forward soft bridge, whose terminal state is random with law N(x*, σ²I). Since the reported results are obtained with this approximation, the consistency between the trained reverse process and the forward soft bridge is broken at a point that is load-bearing for the empirical claims. The authors should state precisely what approximation is being made, provide a formal justification (for example, an argument that the omitted terms are negligible in the relevant limit), or present the reverse process as an approximate sampler whose error is controlled.
  3. [Appendix I / Table 4] The empirical correction factor ζ is fitted per task to validation performance and is used to rescale the score-driven drift in Eq. (90). Table 4 shows that without ζ the SDE sampler on super-resolution achieves PSNR 22.81, far below the reported SDE-ζ value of 27.22, so the headline SDE results depend critically on this fitted rescaling. The paper calls ζ a discretization correction, but it is not derived from any discretization error analysis. At minimum, the authors should report the sensitivity of all reported metrics to ζ, justify why ζ is needed only for the SDE and not for the ODE in some tasks, and temper the claim that the method is 'parameter-free' in its regularization.
  4. [Proposition 4.1] The claim that DDBM-VP, DDBM-VE, and GOUB are recovered as special cases uses the hyperparameter σ̂² = 0, but Theorem 3.2 requires 0 < σ² < σ̄²_{0:T}, which implies σ̂² > 0. Thus σ̂² = 0 lies outside the admissible domain for which the Gaussian terminal marginal is prescribed, and the recovery of hard bridges can only hold in a limiting sense as σ̂² → 0. The proposition should be restated as a limiting recovery, or the admissible domain of the parameters should be extended with a stated limiting argument.
minor comments (4)
  1. [Figure 2(b)] The text says performance degrades as α approaches the boundary -e^{-θ̄_{0:T}}σ̂²/σ̄²_{0:T}, but the horizontal axis of Figure 2(b) shows only small positive α values; the figure should either include the negative side of the admissible range or the text should clarify what range is shown.
  2. [Section 2.2] The notation p_h(x_t, t | x*, T) in the reverse SDE is used before the soft h-conditioned density is formally defined; adding a definition or a pointer would improve readability.
  3. [Appendix H] The proof of the GOU transition variance uses g_t²/(2θ_t) for the limiting covariance, but this is only valid under the parametrization g_t² = 2λ²θ_t; the notation should state this assumption before using it.
  4. [Appendix A.6 / E.3] There are several typographical errors, such as 'right hard' in Appendix A.6 and 'of of' in Appendix E.3, and some figure references (e.g., Figure 3 versus Figure 7/8) are inconsistent. A careful proofreading pass is needed.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: SDDBMs derive the soft bridge from a prescribed terminal marginal and special-case UniDB by explicit parameter identification, rather than by fitting or self-citation.

full rationale

The derivation is self-contained. The paper starts from the prescribed Gaussian terminal marginal in Eq. (10) and derives the terminal weight q (Theorem 3.2), the closed-form h-function (Theorem 3.3), the forward marginals (Proposition 3.4), and the x0-free SDE; these are algebraic consequences, not predictions fitted to the same terminal law. The unification claim is verified by substituting the Table 1 hyperparameters into Eq. (38) in Appendix A.5, i.e., a special-case mapping, not a result fitted to reproduce UniDB. The main caveats are explicitly conceded rather than hidden: Appendix E.3 states Eq. (38) "should be interpreted as a marginally consistent reformulation of Eq. (25), rather than as a pathwise identity," and Appendix I replaces x_T by x⋆ in the network and initializes at x_T ≈ (b_T+c_T)x⋆ and adds a per-task empirical ζ. These are approximations that weaken the theory-to-results link, but they are not circular: no output quantity is defined in terms of the quantity it is supposed to predict, and no load-bearing argument rests on a self-citation. The reverse-training posterior in Appendix A.6 is constructed to match the marginals, which is an assumption about transition laws rather than a circular reduction of the claimed result to its input.

Assumptions & free parameters 4 free parameters · 8 assumptions · 0 invented entities

The central claim rests on standard Gaussian calculus for GOU bridges plus several design choices: the Gaussian terminal ansatz, the linear x̂ parameterization, the ρ-relaxation for well-posedness, and two practical approximations (x_T ≈ (b_T+c_T)x⋆ and the ζ correction) that are essential for the reported numbers.

free parameters (4)
  • σ (terminal variance) = ≈0.08-0.11 for deraining; similar ranges for other tasks
    Controls the spread of the prescribed Gaussian terminal marginal. Chosen by validation; ablations in Section 5 and Appendix J show best performance in this range.
  • α (x0-free drift coefficient) = small positive, exact default not specified
    Appears in the ansatz x̂ = αx0 + βx⋆ + γμ and in the drift of Eq. (20). Tuned; ablation in Figure 2(b) shows strong sensitivity near the boundary in Lemma E.2.
  • ζ (sampling correction factor) = 1.12 (deraining), 1.115 (super-resolution), 1.045 (inpainting)
    Empirical rescaling of the score-driven drift in the reverse SDE. Essential for the reported metrics; without it the SDE sampler performs much worse (Table 4).
  • ρ (terminal center relaxation) = implicit in α, σ; small positive
    In x* = x⋆ + ρ(x0 - x⋆); required to be >0 for well-posed x0-free dynamics (Lemma E.2). Its value is not reported directly.
assumptions (8)
  • standard math Doob's h-transform reweighting p^h(x,t|x0) = p(x,t|x0) h(x,t)/h(x0,0) for positive h
    Used throughout as the foundation of soft bridges (Eqs. (2)-(4)).
  • standard math GOU process admits Gaussian transition density with mean and variance as in Eq. (6)
    Invoked in Theorems 3.2, 3.3 and Proposition 3.4.
  • domain assumption Reference coefficients θ_t, g_t, μ, λ are chosen so the GOU process has finite, regular dynamics on [0,T]
    The experiments use the flipped cosine schedule and λ^2=(30/255)^2 (Appendix I).
  • domain assumption 0 < σ^2 < σ̄²_{0:T} for the terminal marginal
    Required for the calibrated Gaussian weight q to be a valid density (Theorem 3.2).
  • ad hoc to paper Linear ansatz x̂ = αx0 + βx⋆ + γμ with t-independent α, β, γ
    Chosen for tractability in the x0-free reformulation (Section 3.4); no derivation from first principles.
  • ad hoc to paper Relaxed terminal center x* = x⋆ + ρ(x0 - x⋆) with ρ > 0
    Introduced to guarantee φ_t ≠ 0 (Lemma E.2); changes the terminal marginal so it is not actually target-independent.
  • ad hoc to paper At inference, x_T ≈ (b_T + c_T)x⋆ and the score network is conditioned on x⋆ instead of x_T
    Practical approximation that drops source-dependent and noise terms; not proven to match the forward bridge (Appendix I).
  • ad hoc to paper The empirical correction factor ζ compensates only for discretization error
    Without ζ the theoretical SDE is far less accurate, so this assumption is strong (Table 4).

how reviews work

0 comments
Cite this review

Pith. "Pith review of SDDBMs: Soft Denoising Diffusion Bridge Models." pith.science (2026). https://pith.science/paper/VWGWTYOC

@misc{pith2026260808594,
  author       = {Pith},
  title        = {Pith review of: SDDBMs: Soft Denoising Diffusion Bridge Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VWGWTYOC}},
  note         = {Machine review of arXiv:2608.08594}
}
abstract

Diffusion bridge models leverage Doob's \(h\)-transform to construct stochastic transports between arbitrary endpoint distributions, and have shown strong potential in image-to-image translation and restoration. However, most existing bridge models rely on hard endpoint conditioning, which forces the terminal state to match a prescribed target exactly. This hard constraint induces terminal-boundary singularities: the terminal law collapses to a Dirac measure, and the resulting drift coefficients become ill-conditioned near the endpoint. In this paper, we propose Soft Denoising Diffusion Bridge Models (SDDBMs), a generalized framework that regularizes diffusion bridges directly at the level of their terminal constraints. Instead of imposing an exact endpoint, SDDBMs prescribe a non-degenerate Gaussian terminal marginal under the transformed path measure, with a flexible terminal center and variance. Starting from this prescribed marginal, we develop a complete closed-form construction of the soft bridge, including the Gaussian terminal reweighting and soft \(h\)-function, the induced Gaussian forward marginals and \(\mathbf{x}_0\)-free dynamics. Theoretically, SDDBMs provide a unified probabilistic perspective that encompasses existing diffusion bridge models, including DDBMs, GOUB, and UniDB, as special cases under specific parameter choices. Extensive experiments on image restoration tasks demonstrate that SDDBMs achieve improved numerical stability and superior generation quality over existing bridge-based methods.

Figures

Figures reproduced from arXiv: 2608.08594 by the authors.

Figure 1
Figure 1. To expose the singularity of hard endpoint constraints, we consider a particular linear SDE: [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Performance sensitivity to hyperparameters [PITH_FULL_IMAGE:figures/full_fig_p010_2.png] view at source ↗
Figure 4
Figure 4. Training loss of the hyperparameter σ on deraining task. Image Deraining. For the image deraining task, we utilize the Rain100H dataset (Yang et al., 2017), comprising 1800 rainy/clean image pairs for training and 100 pairs for testing. For a fair comparison with prior works (Luo et al., 2023; Yue et al., 2024; Zhu et al., 2025), PSNR and SSIM are evaluated exclusively on the Y channel in the YCbCr color space. As s… view at source ↗
Figures from the paper (8 more)
Figure 3
Figure 3. Figure 3: Qualitative comparison between SDDBMs and baseline methods on the Rain100H, CelebA [PITH_FULL_IMAGE:figures/full_fig_p011_3.png]
Figure 5
Figure 5. Figure 5: Training loss of the hyperparameter β on super-resolution and deraining tasks. 0 20 40 60 80 100 Training Steps (£10 4) 10 -4 10 -3 10 -2 Training loss (a) Impact of ¾ on inpainting ¾ = 0:005 ¾ = 0:05 ¾ = 0:07 ¾ = 0:10 0 10 20 30 40 50 60 Training Steps (£10 4) 10 -4 1…
Figure 6
Figure 6. Figure 6: Training loss of the hyperparameter σ on inpainting and super-resolution tasks. Sensitivity Analysis of σ: We first study the effect of the terminal variance parameter σ 2 , which controls the strength of the prescribed Gaussian terminal constraint in SDDBMs. As shown …
Figure 7
Figure 7. Figure 7: Qualitative comparison between UniDB and SDDBMs on the CelebA-HQ dataset for image [PITH_FULL_IMAGE:figures/full_fig_p034_7.png]
Figure 8
Figure 8. Figure 8: Qualitative comparison between SDDBMs and baseline methods on the Rain100H dataset [PITH_FULL_IMAGE:figures/full_fig_p034_8.png]
Figure 9
Figure 9. Figure 9: Additional visual results on thin mask inpainting with CelebA-HQ datasets. [PITH_FULL_IMAGE:figures/full_fig_p035_9.png]
Figure 10
Figure 10. Figure 10: Additional visual results on deraining with Rain 100H datasets. [PITH_FULL_IMAGE:figures/full_fig_p036_10.png]
Figure 11
Figure 11. Figure 11: Additional visual results on super-resolution with DIV2K datasets. [PITH_FULL_IMAGE:figures/full_fig_p037_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 16 canonical work pages

  1. [1]

    arXiv preprint arXiv:1605.07116 , year=

    A formal evaluation of PSNR as quality measurement parameter for image segmentation algorithms , author=. arXiv preprint arXiv:1605.07116 , year=

    arXiv

  2. [2]

    IEEE transactions on image processing , volume=

    Image quality assessment: from error visibility to structural similarity , author=. IEEE transactions on image processing , volume=. 2004 , publisher=

    2004

  3. [3]

    Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

    The unreasonable effectiveness of deep features as a perceptual metric , author=. Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

  4. [4]

    Proceedings of the IEEE conference on computer vision and pattern recognition workshops , pages=

    Ntire 2017 challenge on single image super-resolution: Dataset and study , author=. Proceedings of the IEEE conference on computer vision and pattern recognition workshops , pages=

    2017

  5. [5]

    6th International Conference on Learning Representations,

    Tero Karras and Timo Aila and Samuli Laine and Jaakko Lehtinen , title =. 6th International Conference on Learning Representations,. 2018 , url =

    work page 2018

  6. [6]

    Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

    Deep joint rain detection and removal from a single image , author=. Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

  7. [7]

    International Conference on Machine Learning , pages=

    Image Restoration with Mean-Reverting Stochastic Differential Equations , author=. International Conference on Machine Learning , pages=. 2023 , organization=

    work page 2023

  8. [8]

    Advances in neural information processing systems , volume=

    Denoising diffusion restoration models , author=. Advances in neural information processing systems , volume=

Show all 55 references
  1. [9]

    International Conference on Learning Representations , volume=

    Denoising diffusion bridge models , author=. International Conference on Learning Representations , volume=

  2. [10]

    PromptIR: Prompting for All-in-One Image Restoration , url =

    Potlapalli, Vaishnav and Zamir, Syed Waqas and Khan, Salman and Shahbaz Khan, Fahad , booktitle =. PromptIR: Prompting for All-in-One Image Restoration , url =

  3. [11]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Maxim: Multi-axis mlp for image processing , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  4. [12]

    Pattern Recognition , volume=

    Mixed hierarchy network for image restoration , author=. Pattern Recognition , volume=. 2025 , publisher=

    2025

  5. [13]

    Advances in neural information processing systems , volume=

    Denoising diffusion probabilistic models , author=. Advances in neural information processing systems , volume=

  6. [14]

    Score-Based Generative Modeling through Stochastic Differential Equations , booktitle =

    Yang Song and Jascha Sohl. Score-Based Generative Modeling through Stochastic Differential Equations , booktitle =. 2021 , url =

    2021

  7. [15]

    Yaron Lipman and Ricky T. Q. Chen and Heli Ben. Flow Matching for Generative Modeling , booktitle =. 2023 , url =

    2023

  8. [16]

    Journal of Machine Learning Research , volume=

    Stochastic interpolants: A unifying framework for flows and diffusions , author=. Journal of Machine Learning Research , volume=

  9. [17]

    The Eleventh International Conference on Learning Representations,

    Xingchao Liu and Chengyue Gong and Qiang Liu , title =. The Eleventh International Conference on Learning Representations,. 2023 , url =

    2023

  10. [18]

    Stochastic Processes and their Applications , volume=

    Reverse-time diffusion equation models , author=. Stochastic Processes and their Applications , volume=. 1982 , publisher=

    1982

  11. [19]

    Neural computation , volume=

    A connection between score matching and denoising autoencoders , author=. Neural computation , volume=. 2011 , publisher=

    2011

  12. [20]

    2019 , publisher=

    Applied stochastic differential equations , author=. 2019 , publisher=

    2019

  13. [21]

    1988 , publisher=

    Introduction to stochastic differential equations , author=. 1988 , publisher=

    1988

  14. [22]

    International conference on machine learning , pages=

    Deep unsupervised learning using nonequilibrium thermodynamics , author=. International conference on machine learning , pages=. 2015 , organization=

    2015

  15. [23]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    High-resolution image synthesis with latent diffusion models , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  16. [24]

    International Conference on Learning Representations , volume=

    Sdxl: Improving latent diffusion models for high-resolution image synthesis , author=. International Conference on Learning Representations , volume=

  17. [25]

    arXiv preprint arXiv:2512.16093 , year=

    TurboDiffusion: Accelerating Video Diffusion Models by 100-200 Times , author=. arXiv preprint arXiv:2512.16093 , year=

  18. [26]

    arXiv preprint arXiv:2505.23458 , year=

    Diffusion guidance is a controllable policy improvement operator , author=. arXiv preprint arXiv:2505.23458 , year=

    arXiv

  19. [27]

    Diffusion schr

    Shi, Yuyang and De Bortoli, Valentin and Campbell, Andrew and Doucet, Arnaud , journal=. Diffusion schr

  20. [28]

    Advances in neural information processing systems , volume=

    Elucidating the design space of diffusion-based generative models , author=. Advances in neural information processing systems , volume=

  21. [29]

    Proceedings of the IEEE/CVF international conference on computer vision , pages=

    Scalable diffusion models with transformers , author=. Proceedings of the IEEE/CVF international conference on computer vision , pages=

  22. [30]

    International conference on machine learning , pages=

    Improved denoising diffusion probabilistic models , author=. International conference on machine learning , pages=. 2021 , organization=

    2021

  23. [31]

    Advances in neural information processing systems , volume=

    Dpm-solver: A fast ode solver for diffusion probabilistic model sampling in around 10 steps , author=. Advances in neural information processing systems , volume=

  24. [32]

    9th International Conference on Learning Representations,

    Jiaming Song and Chenlin Meng and Stefano Ermon , title =. 9th International Conference on Learning Representations,. 2021 , url =

    2021

  25. [33]

    The Eleventh International Conference on Learning Representations,

    Qinsheng Zhang and Yongxin Chen , title =. The Eleventh International Conference on Learning Representations,. 2023 , url =

    2023

  26. [34]

    International Conference on Machine Learning , pages=

    Consistency Models , author=. International Conference on Machine Learning , pages=. 2023 , organization=

    2023

  27. [35]

    International Conference on Learning Representations , volume=

    One step diffusion via shortcut models , author=. International Conference on Learning Representations , volume=

  28. [36]

    Advances in Neural Information Processing Systems , volume=

    Mean flows for one-step generative modeling , author=. Advances in Neural Information Processing Systems , volume=

  29. [37]

    Julius Berner and Lorenz Richter and Karen Ullrich , title =. Trans. Mach. Learn. Res. , volume =. 2024 , url =

    2024

  30. [38]

    RB-Modulation: Training-Free Stylization using Reference-Based Modulation , booktitle =

    Litu Rout and Yujia Chen and Nataniel Ruiz and Abhishek Kumar and Constantine Caramanis and Sanjay Shakkottai and Wen. RB-Modulation: Training-Free Stylization using Reference-Based Modulation , booktitle =. 2025 , url =

    2025

  31. [39]

    Advances in neural information processing systems , volume=

    Generative modeling by estimating gradients of the data distribution , author=. Advances in neural information processing systems , volume=

  32. [40]

    Proceedings of the IEEE/CVF conference on computer vision and pattern Recognition , pages=

    Bbdm: Image-to-image translation with brownian bridge diffusion models , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern Recognition , pages=

  33. [41]

    Advances in neural information processing systems , volume=

    Diffusion models beat gans on image synthesis , author=. Advances in neural information processing systems , volume=

  34. [42]

    arXiv preprint arXiv:2207.12598 , year=

    Classifier-free diffusion guidance , author=. arXiv preprint arXiv:2207.12598 , year=

    arXiv

  35. [43]

    Diffusion schr

    De Bortoli, Valentin and Thornton, James and Heng, Jeremy and Doucet, Arnaud , journal=. Diffusion schr

  36. [44]

    2004 , publisher=

    Convex optimization , author=. 2004 , publisher=

    2004

  37. [45]

    2009 , publisher=

    The elements of statistical learning: data mining, inference, and prediction , author=. 2009 , publisher=

    2009

  38. [46]

    Advances in neural information processing systems , volume=

    Maximum likelihood training of score-based diffusion models , author=. Advances in neural information processing systems , volume=

  39. [47]

    International Conference on Machine Learning , pages=

    Image Restoration Through Generalized Ornstein-Uhlenbeck Bridge , author=. International Conference on Machine Learning , pages=. 2024 , organization=

    2024

  40. [48]

    International Conference on Machine Learning , pages=

    UniDB: A Unified Diffusion Bridge Framework via Stochastic Optimal Control , author=. International Conference on Machine Learning , pages=. 2025 , organization=

    2025

  41. [49]

    An Introduction to Manifolds , pages=

    Manifolds , author=. An Introduction to Manifolds , pages=. 2011 , publisher=

    2011

  42. [50]

    2012 , note =

    Petersen, Kaare Brandt and Pedersen, Michael Syskind , title =. 2012 , note =

    2012

  43. [51]

    Technical University of Denmark , volume=

    The matrix cookbook , author=. Technical University of Denmark , volume=

  44. [52]

    Biometrika , volume=

    Simulating diffusion bridges with score matching , author=. Biometrika , volume=. 2025 , publisher=

    2025

  45. [53]

    arXiv preprint arXiv:1102.3172 , year=

    Stochastic derivatives and generalized h-transforms of Markov processes , author=. arXiv preprint arXiv:1102.3172 , year=

    arXiv

  46. [54]

    Kingma and Jimmy Ba , editor =

    Diederik P. Kingma and Jimmy Ba , editor =. Adam:. 3rd International Conference on Learning Representations,. 2015 , url =

    2015

  47. [55]

    2013 , publisher=

    Stochastic differential equations: an introduction with applications , author=. 2013 , publisher=

    2013

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.