arxiv: 2605.07220 · v1 · submitted 2026-05-08 · 💻 cs.LG

Recognition: 2 theorem links

· Lean Theorem

On the Robustness of Distribution Support under Diffusion Guidance

Ruijia Cao , Yuchen Wu , Nisha Chadramoorthy

Authors on Pith no claims yet

Pith reviewed 2026-05-11 02:23 UTC · model grok-4.3

classification 💻 cs.LG

keywords diffusion guidancedistribution supportscore functionsDDIMDDPMgenerative models

0 comments

The pith

Guided diffusion processes keep samples close to the target support when given exact score functions.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes that guided diffusion almost always produces samples near the target support under exact score access. This holds for DDIM and DDPM across discretization schemes from exponential integrators. The result explains why guidance yields high-fidelity, structurally valid outputs instead of implausible ones off the data manifold. A reader would care because off-support samples often break downstream tasks that rely on realistic generation.

Core claim

We show that, given exact access to the score functions, guided diffusion processes almost always generate samples that remain close to the target support. This property is particularly desirable, as samples that lie off the support are often structurally implausible and may adversely affect downstream tasks. Our analysis covers both Denoising Diffusion Implicit Models (DDIM) and Denoising Diffusion Probabilistic Models (DDPM), and applies to a wide range of discretization schemes induced by exponential integrators.

What carries the argument

The robustness of support property, shown by tracking how the guided score steers trajectories under exponential integrator discretizations for DDIM and DDPM.

If this is right

Guidance can steer samples controllably without producing structurally invalid outputs.
The support preservation holds across common discretization choices in DDIM and DDPM sampling.
Off-support artifacts that harm downstream tasks are theoretically avoided under exact scores.
The analysis supplies a foundation for why guided diffusion yields physically meaningful samples.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

In practice, score estimates close to exact may inherit enough of the support robustness to explain observed stability.
The same reasoning could be tested on other samplers or continuous-time limits to see if support adherence persists.
This suggests training objectives that reduce score error could directly improve support fidelity in generated samples.

Load-bearing premise

The proof requires exact knowledge of the score functions and applies only to discretization schemes induced by exponential integrators.

What would settle it

A concrete numerical trajectory computed with exact scores on a known target distribution where the final sample lies far outside the support would disprove the claim.

Figures

Figures reproduced from arXiv: 2605.07220 by Nisha Chadramoorthy, Ruijia Cao, Yuchen Wu.

**Figure 1.** Figure 1: Illustration of ODE (14) when I = {η0, η1, η2}. Here, guidance is towards the blue circle with label η0, F0 = γ−(γ−1)ζη0,T−t(zt) σ 2 T−t Fη0,T −t(zt), F1 = − γ−1 σ 2 T−t ζη1,T −t(zt)Fη1,T −t(zt) and F2 = − γ−1 σ 2 T−t ζη2,T −t(zt)Fη2,T −t(zt). The combined force is given by Fcomb = F0 + F1 + F2. 4 Results for DDPM In this section, we extend our deterministic results in Section 3 to the stochastic setting,… view at source ↗

**Figure 2.** Figure 2: Illustration of SDE (17) for I = {η0, η1, η2}. As before, the guidance is directed toward the blue circle labeled η0. The terms F0, F1, and F2 denote deterministic forces associated with the supports Kη0 , Kη1 , and Kη2 , respectively, and Frand represents the stochastic force induced by the Brownian motion. The total force is given by Fcomb = F0 + F1 + F2 + Frand. 11 [PITH_FULL_IMAGE:figures/full_fig_p01… view at source ↗

**Figure 3.** Figure 3: Trajectories of the guided DDIM under varying guidance strengths and initial positions. The [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗

**Figure 4.** Figure 4: Trajectories of the guided DDPM under varying guidance strengths and initial positions. Here, [PITH_FULL_IMAGE:figures/full_fig_p013_4.png] view at source ↗

**Figure 5.** Figure 5: Trajectories of the guided DDIM for varying guidance strengths and initial positions, with the [PITH_FULL_IMAGE:figures/full_fig_p013_5.png] view at source ↗

**Figure 6.** Figure 6: Trajectories of the guided DDPM for different guidance strengths and initial positions. As before, [PITH_FULL_IMAGE:figures/full_fig_p014_6.png] view at source ↗

**Figure 7.** Figure 7: In Figures 7a and 7b, we illustrate the empirical densities produced by the guided DDIM and DDPM samplers under varying levels of guidance. Each colored region represents the empirical distribution obtained by guiding the diffusion model toward its corresponding component. 14 [PITH_FULL_IMAGE:figures/full_fig_p014_7.png] view at source ↗

read the original abstract

Diffusion guidance is a powerful technique that enables controllable and high-fidelity sample generation with diffusion models. At a high level, it modifies the score function by incorporating a guidance term that steers the generative process toward a desired condition. Despite its empirical success, the theoretical properties of diffusion guidance remain largely unexplored, and it is not well understood why it consistently produces high-quality samples. In this work, we explain the effectiveness of diffusion guidance by establishing a \emph{robustness of support} property. Specifically, we show that, given exact access to the score functions, guided diffusion processes almost always generate samples that remain close to the target support. This property is particularly desirable, as samples that lie off the support are often structurally implausible and may adversely affect downstream tasks. Our analysis covers both Denoising Diffusion Implicit Models (DDIM) and Denoising Diffusion Probabilistic Models (DDPM), and applies to a wide range of discretization schemes induced by exponential integrators. Our results provide a rigorous foundation for understanding why diffusion guidance produces physically meaningful and structurally plausible samples.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper establishes a support-robustness property for guided diffusion under exact scores and exponential-integrator discretizations, but the result is narrowly scoped and the practical payoff remains unclear.

read the letter

The central claim is that guided DDIM and DDPM processes keep samples close to the target support when the score functions are known exactly and the reverse process is discretized with exponential integrators. That is the new piece: a theoretical mechanism that explains why guidance tends to avoid off-manifold samples without relying on empirical observation alone. The authors work through both the implicit and probabilistic formulations and show how the guidance term interacts with the linear drift in a way that cancels deviations at each step under those integrators. That is useful for readers who want a clean explanation of an observed phenomenon rather than another empirical tweak. The math appears to rest on standard score properties rather than circular definitions, which is a plus. The limitation that stands out is the restriction to exponential-integrator schemes. Standard first-order methods such as Euler-Maruyama introduce local truncation errors whose interaction with the guidance drift is not controlled by the same cancellation, and the abstract gives no indication that the bound survives those errors. In practice most codebases use a mix of solvers, so the guarantee does not transfer directly. The analysis also assumes exact access to the score functions, which is an idealization; learned scores introduce approximation error that could reintroduce off-support samples. These are not fatal, but they narrow the result to a specific theoretical regime. The paper is aimed at researchers who care about the foundations of diffusion sampling rather than immediate algorithmic improvements. It is coherent on its own terms and shows honest engagement with the literature on score-based models. A serious editor should send it to review so that the proofs can be checked and the scope clarified, even if the final version needs to state its assumptions more plainly.

Referee Report

2 major / 2 minor

Summary. The paper establishes a 'robustness of support' property for diffusion guidance: with exact access to the score functions of the target and guidance distributions, guided reverse processes for both DDIM and DDPM produce samples that remain close to the target support. The result is proved for the class of discretization schemes obtained from exponential integrators, which permit exact integration of the linear drift term and thereby allow the exact-score guidance to cancel support deviations at each step.

Significance. If the derivation holds, the result supplies a concrete theoretical explanation for the empirical observation that guided diffusion samples are structurally plausible rather than off-manifold artifacts. It isolates the role of exact scores and the exponential-integrator discretization class, and it applies uniformly to the two most common diffusion samplers (DDIM, DDPM). The absence of free parameters or invented entities in the stated claim is a strength.

major comments (2)

[§3 (discretization analysis)] The central claim is explicitly scoped to exponential-integrator discretizations (abstract and §3). Standard first-order schemes such as Euler-Maruyama introduce local truncation errors whose interaction with the guidance drift is not controlled by the same exact-cancellation mechanism; the manuscript provides no uniform bound or counter-example analysis showing whether the support-robustness property survives these errors. This limitation is load-bearing because most practical implementations do not use exponential integrators.
[Theorem 1 / §4 (exact-score assumption)] The proof assumes exact access to both the unconditional and conditional score functions at every step. No quantitative error propagation is given for the case of approximate scores (e.g., learned neural-network estimators), even though the abstract highlights 'exact access' as a prerequisite. A perturbation analysis or Lipschitz-style bound on score error would be needed to assess practical relevance.

minor comments (2)

[§2] Notation for the guidance scale and the target support indicator is introduced without a consolidated table; a short notation summary would improve readability.
[Abstract / §1] The abstract states the result applies to 'a wide range of discretization schemes induced by exponential integrators,' but the precise class (e.g., which Runge-Kutta or linear multistep variants) is not enumerated until later; an explicit list in the introduction would clarify scope.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the positive assessment of the paper's significance and for the constructive major comments. We address each point below and indicate the revisions we will make.

read point-by-point responses

Referee: [§3 (discretization analysis)] The central claim is explicitly scoped to exponential-integrator discretizations (abstract and §3). Standard first-order schemes such as Euler-Maruyama introduce local truncation errors whose interaction with the guidance drift is not controlled by the same exact-cancellation mechanism; the manuscript provides no uniform bound or counter-example analysis showing whether the support-robustness property survives these errors. This limitation is load-bearing because most practical implementations do not use exponential integrators.

Authors: We agree that the result is scoped to exponential-integrator discretizations, as stated in the abstract and Section 3. The proof technique relies on the exact integration of the linear drift, which enables precise cancellation of support deviations by the guidance term at each step. Standard first-order schemes such as Euler-Maruyama introduce truncation errors whose effect on the support-robustness property is not controlled by the same mechanism, and the manuscript contains neither a uniform bound nor a counter-example for those schemes. In the revised version we will add an explicit discussion paragraph in §3 acknowledging this scope limitation and noting that extending the analysis to first-order discretizations is an open direction. revision: partial
Referee: [Theorem 1 / §4 (exact-score assumption)] The proof assumes exact access to both the unconditional and conditional score functions at every step. No quantitative error propagation is given for the case of approximate scores (e.g., learned neural-network estimators), even though the abstract highlights 'exact access' as a prerequisite. A perturbation analysis or Lipschitz-style bound on score error would be needed to assess practical relevance.

Authors: The current work isolates the support-robustness mechanism under exact score access, which is explicitly stated as a prerequisite in the abstract and Theorem 1. We do not supply a perturbation or Lipschitz-style bound for approximate (learned) scores, as that analysis lies outside the scope of the present manuscript. In revision we will strengthen the wording in the abstract and the statement of Theorem 1 to emphasize the exact-access hypothesis and add a short remark in the discussion section on the implications for score-estimation error. revision: partial

Circularity Check

0 steps flagged

No circularity; derivation follows from standard diffusion score properties and exact integrator cancellation

full rationale

The paper derives a support-robustness property for guided DDIM/DDPM processes under exact score access, specifically for exponential-integrator discretizations. This follows directly from the exact integration of the linear drift term (allowing the score to cancel support deviations) and standard properties of the reverse SDE, without any reduction to fitted parameters, self-definitional loops, or load-bearing self-citations. The result is scoped to the stated discretization class but remains mathematically self-contained and externally verifiable from the diffusion literature.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on idealized assumptions about score access and standard diffusion process properties rather than new fitted quantities or invented entities.

axioms (2)

domain assumption Exact access to the score functions of the underlying diffusion process
The analysis explicitly conditions on perfect score knowledge, which is an idealization not available in practice.
standard math Standard properties of DDIM and DDPM forward and reverse processes under exponential integrator discretizations
These are background facts from the diffusion modeling literature invoked to extend the result across samplers.

pith-pipeline@v0.9.0 · 5483 in / 1166 out tokens · 41211 ms · 2026-05-11T02:23:32.189360+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

91 extracted references · 91 canonical work pages · 1 internal anchor

[1]

Course notes., URL http://www

Stochastic calculus, filtering, and stochastic control , author=. Course notes., URL http://www. princeton. edu/rvan/acm217/ACM217. pdf , volume=

work page
[2]

Optimizing methods in statistics , pages=

A convergence theorem for non negative almost supermartingales and some applications , author=. Optimizing methods in statistics , pages=. 1971 , publisher=

work page 1971
[3]

Self-refining video sampling.arXiv preprint arXiv:2601.18577, 2026

Self-Refining Video Sampling , author=. arXiv preprint arXiv:2601.18577 , year=

work page arXiv
[4]

2011 , eprint=

On Malliavin's proof of H\"ormander's theorem , author=. 2011 , eprint=

work page 2011
[5]

Stroock, Daniel W. , year=. Partial Differential Equations for Probabilists , publisher=

work page
[6]

Advances in Neural Information Processing Systems , volume=

The probability flow ode is provably fast , author=. Advances in Neural Information Processing Systems , volume=

work page
[7]

Advances in Neural Information Processing Systems , volume=

What does guidance do? a fine-grained analysis in a simple setting , author=. Advances in Neural Information Processing Systems , volume=

work page
[8]

Archiv der Mathematik , volume=

The nearest point mapping is single valued nearly everywhere , author=. Archiv der Mathematik , volume=. 1990 , publisher=

work page 1990
[9]

Proceedings of the 41st International Conference on Machine Learning , pages=

Theoretical insights for diffusion guidance: a case study for Gaussian mixture models , author=. Proceedings of the 41st International Conference on Machine Learning , pages=

work page
[10]

2026 , eprint=

Emergence of Distortions in High-Dimensional Guided Diffusion Models , author=. 2026 , eprint=

work page 2026
[11]

NeurIPS 2024 Workshop on Mathematics of Modern Machine Learning , year=

Classifier-Free Guidance is a Predictor-Corrector , author=. NeurIPS 2024 Workshop on Mathematics of Modern Machine Learning , year=

work page 2024
[12]

arXiv preprint arXiv:2505.01382 , year=

Provable Efficiency of Guidance in Diffusion Models for General Data Distribution , author=. arXiv preprint arXiv:2505.01382 , year=

work page arXiv
[13]

arXiv preprint arXiv:2512.04985 , year =

Towards a unified framework for guided diffusion models , author=. arXiv preprint arXiv:2512.04985 , year=

work page arXiv
[14]

Elliptic partial differential equations of second order , author=

work page
[15]

Brascamp, H. J. and Lieb, E. H. , journal=. On extensions of the

work page
[16]

Differential Equations, Dynamical Systems, and an Introduction to Chaos , author=

work page
[17]

Transactions of the American Mathematical Society , volume=

Curvature measures , author=. Transactions of the American Mathematical Society , volume=. 1959 , publisher=

work page 1959
[18]

Proceedings of the 32nd International Conference on Machine Learning , year=

Deep Unsupervised Learning using Nonequilibrium Thermodynamics , author=. Proceedings of the 32nd International Conference on Machine Learning , year=

work page
[19]

Advances in Neural Information Processing Systems , volume=

Denoising Diffusion Probabilistic Models , author=. Advances in Neural Information Processing Systems , volume=

work page
[20]

International Conference on Learning Representations , year=

Denoising Diffusion Implicit Models , author=. International Conference on Learning Representations , year=

work page
[21]

International Conference on Learning Representations , year=

Score-Based Generative Modeling through Stochastic Differential Equations , author=. International Conference on Learning Representations , year=

work page
[22]

Diffusion Models Beat

Dhariwal, Prafulla and Nichol, Alex , booktitle=. Diffusion Models Beat

work page
[23]

Classifier-Free Diffusion Guidance

Classifier-Free Diffusion Guidance , author=. arXiv preprint arXiv:2207.12598 , year=

work page internal anchor Pith review Pith/arXiv arXiv
[24]

Proceedings of the

High-Resolution Image Synthesis with Latent Diffusion Models , author=. Proceedings of the

work page
[25]

Advances in Neural Information Processing Systems , volume=

Photorealistic Text-to-Image Diffusion Models with Deep Language Understanding , author=. Advances in Neural Information Processing Systems , volume=

work page
[26]

Controlled Diffusion Processes , author=

work page
[27]

Stochastic Integration and Differential Equations , author=

work page
[28]

Nikolov, PJ Thomas , author=

Boundary Regularity For the Distance Functions, and the Eikonal Equation: N. Nikolov, PJ Thomas , author=. The Journal of Geometric Analysis , volume=. 2025 , publisher=

work page 2025
[29]

Aebi, Robert , journal=. It. 1992 , publisher=

work page 1992
[30]

2013 , publisher=

Continuous martingales and Brownian motion , author=. 2013 , publisher=

work page 2013
[31]

2018 , publisher=

Measure theory and fine properties of functions , author=. 2018 , publisher=

work page 2018
[32]

James Bradbury and Roy Frostig and Peter Hawkins and Matthew James Johnson and Chris Leary and Dougal Maclaurin and George Necula and Adam Paszke and Jake Vander

work page
[33]

Runge-Kutta pairs of order 5(4) satisfying only the first column simplifying assumption , author=. Comput. Math. Appl. , year=

work page
[34]

Patrick Kidger , year=

work page
[35]

Advances in neural information processing systems , volume=

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

work page
[36]

Brownian motion and Riemannian geometry , author=. Contemp. Math , volume=

work page
[37]

Communications on Pure and Applied Mathematics , volume=

On the behavior of the fundamental solution of the heat equation with variable coefficients , author=. Communications on Pure and Applied Mathematics , volume=. 1967 , publisher=

work page 1967
[38]

Probabilistic approach to geometry , volume=

The heat kernel and its estimates , author=. Probabilistic approach to geometry , volume=. 2010 , publisher=

work page 2010
[39]

SIAM Journal on Mathematical Analysis , volume=

Short-Time Asymptotics of the Heat Kernel on a Concave Boundary , author=. SIAM Journal on Mathematical Analysis , volume=. 1989 , publisher=

work page 1989
[40]

Scientific reports , volume=

Principal Component Analyses (PCA)-based findings in population genetic studies are highly biased and must be reevaluated , author=. Scientific reports , volume=. 2022 , publisher=

work page 2022
[41]

arXiv preprint arXiv:2501.12982 , year=

Low-dimensional adaptation of diffusion models: Convergence in total variation , author=. arXiv preprint arXiv:2501.12982 , year=

work page arXiv
[42]

org/abs/2208.11970

Understanding diffusion models: A unified perspective , author=. arXiv preprint arXiv:2208.11970 , year=

work page arXiv
[43]

Journal of functional analysis , volume=

Shape analysis via oriented distance functions , author=. Journal of functional analysis , volume=. 1994 , publisher=

work page 1994
[44]

2014 , publisher=

Stochastic differential equations and diffusion processes , author=. 2014 , publisher=

work page 2014
[45]

Nature , volume=

Accurate structure prediction of biomolecular interactions with AlphaFold 3 , author=. Nature , volume=. 2024 , publisher=

work page 2024
[46]

, author=

Estimation of non-normalized statistical models by score matching. , author=. Journal of Machine Learning Research , volume=

work page
[47]

Nature , volume=

De novo design of protein structure and function with RFdiffusion , author=. Nature , volume=. 2023 , publisher=

work page 2023
[48]

The Eleventh International Conference on Learning Representations , year=

Fast Sampling of Diffusion Models with Exponential Integrator , author=. The Eleventh International Conference on Learning Representations , year=

work page
[49]

2006 , publisher=

Convex Analysis and Nonlinear Optimization: Theoryand Examples , author=. 2006 , publisher=

work page 2006
[50]

2024 , publisher=

Stochastic integrals , author=. 2024 , publisher=

work page 2024
[51]

2015 , publisher=

Asymptotic Geometric Analysis, Part I , author=. 2015 , publisher=

work page 2015
[52]

Archiv der Mathematik , volume=

Approximating convex bodies by algebraic ones , author=. Archiv der Mathematik , volume=. 1974 , publisher=

work page 1974
[53]

arXiv preprint arXiv:2403.04279 , year=

Controllable generation with text-to-image diffusion models: A survey , author=. arXiv preprint arXiv:2403.04279 , year=

work page arXiv
[54]

The Thirty-ninth Annual Conference on Neural Information Processing Systems , year=

When and how can inexact generative models still sample from the data manifold? , author=. The Thirty-ninth Annual Conference on Neural Information Processing Systems , year=

work page
[55]

arXiv preprint arXiv:2601.21200 , year=

Provably Reliable Classifier Guidance via Cross-Entropy Control , author=. arXiv preprint arXiv:2601.21200 , year=

work page arXiv
[56]

The Twelfth International Conference on Learning Representations , year=

Training Diffusion Models with Reinforcement Learning , author=. The Twelfth International Conference on Learning Representations , year=

work page
[57]

arXiv preprint arXiv:2602.05533 , year=

Conditional Diffusion Guidance under Hard Constraint: A Stochastic Analysis Approach , author=. arXiv preprint arXiv:2602.05533 , year=

work page arXiv
[58]

arXiv preprint arXiv:2409.04832 , year=

Reward-directed score-based diffusion models via q-learning , author=. arXiv preprint arXiv:2409.04832 , year=

work page arXiv
[59]

Advances in neural information processing systems , volume=

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

work page
[60]

The Twelfth International Conference on Learning Representations , year=

Nearly \ d\ -Linear Convergence Bounds for Diffusion Models via Stochastic Localization , author=. The Twelfth International Conference on Learning Representations , year=

work page
[61]

Stochastic Processes and their Applications , volume=

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

work page 1982
[62]

IEEE Transactions on Information Theory , year=

Convergence analysis of probability flow ode for score-based generative models , author=. IEEE Transactions on Information Theory , year=

work page
[63]

International Conference on Algorithmic Learning Theory , pages=

Convergence of score-based generative modeling for general data distributions , author=. International Conference on Algorithmic Learning Theory , pages=. 2023 , organization=

work page 2023
[64]

From optimal score matching to optimal sampling.arXiv preprint arXiv:2409.07032,

From optimal score matching to optimal sampling , author=. arXiv preprint arXiv:2409.07032 , year=

work page arXiv
[65]

The annals of Statistics , pages=

Estimation of the mean of a multivariate normal distribution , author=. The annals of Statistics , pages=. 1981 , publisher=

work page 1981
[66]

Posterior sampling in high dimension via diffusion processes

Posterior sampling in high dimension via diffusion processes , author=. arXiv preprint arXiv:2304.11449 , year=

work page arXiv
[67]

arXiv preprint arXiv:2504.06566 , year=

Diffusion factor models: Generating high-dimensional returns with factor structure , author=. arXiv preprint arXiv:2504.06566 , year=

work page arXiv
[68]

The Eleventh International Conference on Learning Representations , year=

Diffusion Posterior Sampling for General Noisy Inverse Problems , author=. The Eleventh International Conference on Learning Representations , year=

work page
[69]

arXiv preprint arXiv:2403.13219 , year=

Diffusion model for data-driven black-box optimization , author=. arXiv preprint arXiv:2403.13219 , year=

work page arXiv
[70]

Proceedings of the 41st International Conference on Machine Learning , pages=

Minimax optimality of score-based diffusion models: beyond the density lower bound assumptions , author=. Proceedings of the 41st International Conference on Machine Learning , pages=

work page
[71]

The Thirty Seventh Annual Conference on Learning Theory , pages=

Optimal score estimation via empirical bayes smoothing , author=. The Thirty Seventh Annual Conference on Learning Theory , pages=. 2024 , organization=

work page 2024
[72]

International Conference on Machine Learning , pages=

Diffusion models are minimax optimal distribution estimators , author=. International Conference on Machine Learning , pages=. 2023 , organization=

work page 2023
[73]

The Thirteenth International Conference on Learning Representations , year=

O(d/T) Convergence Theory for Diffusion Probabilistic Models under Minimal Assumptions , author=. The Thirteenth International Conference on Learning Representations , year=

work page
[74]

International Conference on Machine Learning , pages=

Accelerating Convergence of Score-Based Diffusion Models, Provably , author=. International Conference on Machine Learning , pages=. 2024 , organization=

work page 2024
[75]

The Thirteenth International Conference on Learning Representations , year=

Unified Convergence Analysis for Score-Based Diffusion Models with Deterministic Samplers , author=. The Thirteenth International Conference on Learning Representations , year=

work page
[76]

The Thirteenth International Conference on Learning Representations , year=

Faster Diffusion Sampling with Randomized Midpoints: Sequential and Parallel , author=. The Thirteenth International Conference on Learning Representations , year=

work page
[77]

The Eleventh International Conference on Learning Representations , year=

Sampling is as easy as learning the score: theory for diffusion models with minimal data assumptions , author=. The Eleventh International Conference on Learning Representations , year=

work page
[78]

arXiv preprint arXiv:2506.24042 , year=

Faster diffusion models via higher-order approximation , author=. arXiv preprint arXiv:2506.24042 , year=

work page arXiv
[79]

A Connection Between Score Matching and Denoising Autoen- coders

Stochastic runge-kutta methods: Provable acceleration of diffusion models , author=. arXiv preprint arXiv:2410.04760 , year=

work page arXiv
[80]

arXiv preprint arXiv:2506.13061 , year=

Fast convergence for high-order ode solvers in diffusion probabilistic models , author=. arXiv preprint arXiv:2506.13061 , year=

work page arXiv

Showing first 80 references.