Efficient Weighted Sampling via Score-based Generative Models

Gustavo de Veciana; Heasung Kim; Hyeji Kim; Taekyun Lee

REVIEW 4 major objections 4 minor 55 references

Efficient Weighted Sampling via Score-based Generative Models

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

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

Pith's one-line read One pretrained diffusion model samples any reweighted distribution.

desk verdict A promising training-free importance sampling method, but the central chain-rule derivation in Eqs. (14)–(16) inverts a √α factor, so the proposed SDE is not actually the score of the target distribution. read the letter →

arxiv 2502.04646 v2 pith:KBDKRJS2 submitted 2025-02-07 cs.LG cs.AI

Heasung Kim , Taekyun Lee , Hyeji Kim , Gustavo de Veciana This is my paper · ORCID

classification cs.LGcs.AI

keywords importancesamplingscore-basedgenerativemodelsdiffusiontraining-freeguidanceTweedie'sformulabackwardSDEreweighted

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper claims that importance sampling—drawing from a distribution proportional to a base distribution times a weight function—can be done with an already-trained score-based generative model, without any additional training. The key step is an approximation of the target distribution's time-dependent score: it is the base score plus a guidance term built from Tweedie's formula and the gradient of the weight function, with a Hessian correction computed cheaply by finite differences. The authors derive a backward SDE whose solutions approximately follow the reweighted distribution, and they show empirically on synthetic data, CSI compression, CelebA, and Stable Diffusion XL that the method matches or beats baselines that require per-weight training while being faster. The paper is significant because it makes task-adaptive sampling from a single pretrained model practical for any differentiable importance weight.

What carries the argument

Tweedie's formula for the conditional mean $\bar{x}'_{0|x,t} = \frac{1}{\sqrt{\bar{\alpha}(t)}}\left(x + (1 - \bar{\alpha}(t))\nabla_x \log p_t(x)\right)$; a first-order Taylor expansion of $l$ around this mean; and a finite-difference approximation of the Hessian-vector product using score evaluations at perturbed points. These combine to give the approximated score $\nabla_x \log \tilde{q}_t(x)$ and the backward SDE (16), which is the object that actually generates samples.

What would settle it

For a closed-form base distribution and a strongly nonlinear weight function $l$, compute the true importance-sampling distribution $q$ by numerical integration, run the proposed backward SDE, and measure the Jensen–Shannon divergence between generated samples and $q$; if the divergence grows with the curvature of $l$ as predicted by the conditional-variance term, the approximation's practical limit is identified. More directly, track the conditional variance $\mathbb{E}[\|X'_0 - \bar{x}'_{0|x,t}\|^2 \mid X'_t = x]$ along the reverse trajectory; if it stays large for a substantial fraction of the trajectory, the bound in Theorem 1 predicts a large score gap at those times, revealing where the method fails.

Watch

Extended reading notes

Core claim

The central claim is that the score of the importance-sampling distribution $q_t(x)$ can be approximated as $\nabla_x \log p_t(x) + \nabla_x \log l(\bar{x}'_{0|x,t})$, where $\bar{x}'_{0|x,t}$ is the Tweedie conditional mean of the initial sample given the noisy state. To make this computable without Hessians, the paper replaces the Hessian-vector product in the chain rule by a finite-difference score evaluation, yielding the closed-form guidance SDE in Eq. (16). Theorem 1 bounds the gap between true and approximated scores by a term proportional to the conditional variance of $X'_0$ given $X'_t$, which vanishes as $t \to 0$, and by a term involving the Hessian of $\log l$ and the Lipschitz constant of $H_{p_t}$. The paper argues, and demonstrates empirically, that this approximation is accurate enough for sampling, and that because only the base score and $l(x)$ are needed, the method is training-free and scales to large foundation models.

Load-bearing premise

The first-order Taylor expansion of the weight function around the Tweedie conditional mean is accurate enough that the approximation error stays small throughout the reverse trajectory.

Editorial extensions

If this is right

Any differentiable importance weight $l(x)$ can be used for sampling with a single pretrained model, with no per-weight training.
The same base SGM can serve multiple weighted-sampling tasks simultaneously, since the guidance is computed on the fly from $l$.
In large-scale text-to-image models, the method adds a new control dimension beyond prompts, such as generating images with particular frequency content.
In model analysis, it can efficiently find rare high-distortion inputs, improving inverse model analysis and failure-mode identification.
The overhead over standard sampling is one backpropagation through $l$ and one extra score evaluation per step, which is far cheaper than training a new model.

Reading between the lines

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

The paper bounds the score gap at each timestep but does not integrate this bound over the reverse trajectory; a plausible extension is that the accumulated error stays controlled when the conditional variance term decays quickly, but this is not proven.
The Hessian-free guidance term resembles classifier guidance; a testable extension is to compare both on identical base models to see when the explicit guidance term is necessary.
For highly nonlinear or non-smooth $l$, the first-order Taylor expansion could break, and a stress test with adversarial $l$ would reveal the practical limits of the approximation.

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.

Desk Editor's Note

read the letter

Quick take: the core derivation in Eqs. (14)–(16) has an algebraic error. The Tweedie estimate x̂0 = (x + (1−ᾱ)∇log p_t)/√ᾱ has Jacobian J = (I + (1−ᾱ)H_p)/√ᾱ. The gradient of log l(x̂0) w.r.t. x should therefore be Jᵀg = (I + ...)g / √ᾱ. The paper writes it with √ᾱ in the numerator. That is inverted. It propagates into the SDE in Eq. (16), so the reverse process being run is not the score of the importance distribution, even when the first-order Taylor approximation is exact. I checked the stress-test note and it holds.

What is genuinely new: the problem is well chosen. Using a single pretrained SGM for many weight functions l without retraining is useful, and the proposed pipeline—Tweedie plus guidance plus a finite-difference Hessian correction—is a real design. The synthetic and large-scale examples (Stable Cascade frequency control, CSI failure-mode analysis) demonstrate that the method works in practice. The paper is clearly written and honest about the Taylor approximation being a heuristic.

Where it falls apart: the chain-rule factor error is load-bearing, not cosmetic. Theorem 1 bounds a gap to the wrongly defined score, so the stated guarantee does not apply to the intended q. The Taylor expansion of l around the conditional mean is also heuristic; its error is controlled by the conditional variance, which is large over most of the reverse diffusion, and the paper does not bound the accumulated error over the SDE. The experiments compare against trained normalizing flows but omit the most relevant baselines—Diffusion Posterior Sampling and classifier guidance—which use essentially the same guidance mechanism. The proof of Theorem 1 is missing, and the ϵ step size is left unspecified.

Bottom line: the idea is worth pursuing, and the experiments suggest that, with the factor fixed, the method would reduce to a variant of DPS with a finite-difference Hessian. That is incremental but useful. As submitted, the central derivation is invalid. It deserves a serious referee because the problem is important and the fix is straightforward, but it should not be accepted until the SDE is corrected and the missing baselines are added.

Recommendation: engage in review rather than desk-reject, but with a strong expectation of major revision. For your reading group, it is a maybe—good as a case study in how an algebraic slip can invalidate a derivation.

Referee Report

4 major / 4 minor

Summary. The paper proposes a training-free importance sampling scheme for score-based generative models. Given a pretrained score function for a base density p and a differentiable positive weight function l, the authors derive an approximate backward SDE that they claim samples from q(x) ∝ l(x)p(x). The method replaces the intractable time-dependent score of q by the score of p plus a Tweedie-based guidance term, with a Hessian-vector product approximated by a finite difference. Empirical evaluations are reported on synthetic targets, on a neural-compression distortion task for CSI data, on CelebA with a classifier weight, and on StableCascade with a frequency-based weight. The high-level idea is attractive because a single pretrained SGM could in principle serve many importance weights without retraining.

Significance. If the derivation were correct, this would be a useful and scalable contribution: it would extend the applicability of pretrained SGMs to arbitrary differentiable importance weights, including neural-network weights, with no per-weight training. The paper includes code, a reasonable breadth of experiments, and visible effort to compare against density-estimation baselines on synthetic tasks. The main theoretical claim, Theorem 1, is intended to bound the score approximation error in terms of the conditional variance of the initial state, which is a sensible quantity. However, the central derivation contains an algebraic error that changes the guidance term by a factor of sqrt(alpha_bar), so the SDE actually implemented in the experiments is not the score approximation derived in the paper. Other promised components—the uncertainty-aware scheduler and the Stable Diffusion XL speedup results in the abstract—are absent from the body. The significance of the contribution depends on correcting these issues; the underlying idea remains promising.

major comments (4)

[§3.2, Eq. (14) and Eq. (16)] There is a chain-rule error in the guidance term. Let s = ∇_x log p_t(x), g = ∇_{x̄0} log l(x̄0), and H be the Hessian of log p_t at x. Differentiating x̄0 = (x + (1−ᾱ)s)/√ᾱ gives ∂x̄0/∂x = [I + (1−ᾱ)H]/√ᾱ, so the correct chain rule is ∇_x log l(x̄0) = [I + (1−ᾱ)H]g/√ᾱ. Equation (14) instead has √ᾱ in the numerator. This is not a localized typo: the first-order term in the SDE (16) is √ᾱ g rather than g/√ᾱ, and the Hessian term in (16) has coefficient (1−ᾱ)/√ᾱ, which is also inconsistent with Eq. (14) as printed. Consequently, the backward SDE analyzed and simulated in Section 4 is not the score approximation derived from Eqs. (12)–(15), and Theorem 1, even if proved, would not certify that the implemented algorithm samples q(x) ∝ l(x)p(x). The algebra must be corrected and the experiments rerun with the corrected guidance.
[§3.2–§3.3, Eq. (11) and Theorem 1] The derivation of Eq. (12) uses a first-order Taylor expansion of l(X′_0) around the conditional mean, while Theorem 1 assumes that l is approximately equal to its second-order Taylor expansion at the same point. These are different approximations, so the theorem does not bound the score approximation actually used in the method. In addition, the proof of Theorem 1 is not included in the manuscript, although the text states that the gap is established. The bound is also pointwise in x and t, with an error controlled by the conditional variance of X′_0 given X′_t; no argument shows that this pointwise score gap remains small along the reverse-time SDE trajectory or that the accumulated discretization error is controlled. This leaves the central theoretical guarantee unsupported.
[Abstract and Sections 3–4] The abstract and introduction identify an "uncertainty-aware scheduler that dynamically adjusts the guidance strength based on a temporal analysis of approximation error" as a key component of the method, but no scheduler is defined, derived, or evaluated anywhere in Sections 3–4. The proposed SDE is presented as fixed, and the experiments do not describe any scheduler. Either the scheduler should be specified and studied, or the claim should be removed from the abstract and contributions.
[Abstract vs. Section 4.3] The abstract claims that the framework is validated on Stable Diffusion XL with 1.2× to 4.7× speedups while matching or outperforming state-of-the-art baselines. The body does not contain a Stable Diffusion XL experiment: Section 4.3, Exp-(b) uses StableCascade, and Table 2 reports importance-weight values rather than speedups or comparisons with baselines. The abstract's empirical claims need to be reconciled with the actual experimental content.

minor comments (4)

[§4.2] There is a typo: "convolutional nueral architecture" should read "convolutional neural architecture."
[§4.1] The text refers to "Table 4.1" but the table caption reads "Table 1"; the numbering should be made consistent.
[§4] The experimental section does not report the number of discretization steps, the schedule of β(t), or the value of ϵ used in the finite-difference approximation, which are needed to reproduce the results without relying solely on the source code.
[§3.2, Eq. (15)] The finite-difference identity should be stated with the sign convention made explicit: ∇ log p_t(x + ϵg) ≈ ∇ log p_t(x) + ϵ H g, which is the form used in the subsequent derivation.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the derivation is self-contained and the self-citations are not load-bearing.

full rationale

The paper's derivation chain is self-contained. The target score is decomposed in Eq. (10) as an exact identity involving the base score and a conditional expectation of the weight function; the first-order Taylor expansion in Eq. (11) is an approximation, not a fitted input, and it is applied to obtain Eq. (12). Tweedie's formula in Eq. (13) is a standard external result, and Eqs. (14)-(16) are algebraic manipulations of that formula. The parameter epsilon is a finite-difference step, not a fitted constant, and the method is evaluated against ground-truth accept/reject sampling and external density-estimation baselines. The paper's self-citations ([8], [54]) are literature references for wireless-channel diffusion and CSI compression, respectively, and are not used as load-bearing justification for the central claim. No step reduces a predicted quantity to an input by construction, and no uniqueness theorem or ansatz is imported from the authors' prior work. Any concern about the chain-rule factor in Eq. (14) or the Taylor-truncation error is a correctness or accuracy issue, not circularity; the circularity score is therefore 0.

Assumptions & free parameters 1 free parameters · 5 assumptions · 0 invented entities

The method uses standard SDE and Tweedie machinery; the main new assumptions are the Taylor expansion of l and the finite-difference step, both heuristic. No new physical entities are introduced.

free parameters (1)

epsilon (finite-difference step)
Used in Eq. (15)-(16) to approximate the Hessian-vector product. The paper does not specify its value or a selection rule; results may depend on it.

assumptions (5)

domain assumption The forward SDE has VP form f = -1/2 beta(t)x, sigma = sqrt(beta(t))
Standard in DDPM/VP-SDE; the derivation of Tweedie and Green's function rely on it. Sec. 2.
domain assumption Assumptions 1-2: bounded log-derivatives of l and p_t, Lipschitz Hessian of log p_t
Used in Theorem 1; reasonable but not verified for neural weights. Sec. 3.3.
ad hoc to paper l(X'_0) is approximately equal to its second-order Taylor expansion at the conditional mean
This assumption implicitly requires the third-order remainder to be negligible; stated in Theorem 1 condition. Sec. 3.3.
ad hoc to paper The finite-difference approximation in Eq. (15) accurately estimates the Hessian-vector product for the chosen epsilon
True only for sufficiently small epsilon; the paper gives no error analysis for this step. Sec. 3.2.
domain assumption The backward SDE with the approximate score yields a distribution close to q_0
Standard in diffusion sampling; requires score error to be small along the trajectory, which is not directly established. Sec. 3.2.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Efficient Weighted Sampling via Score-based Generative Models." pith.science (2026). https://pith.science/paper/KBDKRJS2

@misc{pith2026250204646,
  author       = {Pith},
  title        = {Pith review of: Efficient Weighted Sampling via Score-based Generative Models},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KBDKRJS2}},
  note         = {Machine review of arXiv:2502.04646}
}

abstract

Weighted sampling -- sampling from a probability density function (PDF) proportional to the product of a base PDF and a weight function -- is a fundamental technique with wide-ranging applications in variance reduction, biased sampling, data augmentation, and more. Leveraging the increasing availability of pretrained score-based generative models (SGMs), we propose a training-free weighted sampling framework that approximates the backward diffusion process of the target distribution by augmenting the pretrained base score function with an auxiliary guidance term, in a principled and computationally efficient manner. Our approach builds on two key components: a lightweight approximation of the guidance that avoids costly higher-order derivatives of both the score and weight functions, and an uncertainty-aware scheduler that dynamically adjusts the guidance strength based on a temporal analysis of approximation error. Together, these components enable accurate and stable sampling without relying on particle-based resampling or Hessian evaluations commonly required by existing methods. We validate the effectiveness of our method from synthetic to large-scale settings such as Stable Diffusion XL, where our framework achieves $1.2\times$ to $4.7\times$ speedups while consistently matching or outperforming state-of-the-art baselines in task performance. These results position our method as a scalable and inference-efficient solution for task-adaptive, time-sensitive sampling in generative applications.

Figures

Figures reproduced from arXiv: 2502.04646 by the authors.

**Figure 1.** Top: Sampling process for X′ ∼ p(x). Bottom: Importance sampling process for X ∼ q(x). The proposed method utilizes the pretrained ∇x log pt(x) and given importance weight function l(x) to compute ∇x log qt(x) without necessitating any additional training. Definition 1. (Importance Sampling PDF) q(x) = l(x)p(x) R l(x)p(x) dx (1) Importance sampling is a versatile technique with demonstrated effectiveness across vari… view at source ↗

**Figure 2.** Top row: sampling process for X′ ∼ p(x). Bottom row: proposed importance sampling process for X ∼ q(x). From left to right, each column corresponds to t = 500, 400, . . . , 0, showing the distributions of E[X′ 0 |X′ t ] and E[X0|Xt]. Thus, the rightmost column illustrates the PDFs p(x) (top) and q(x) (bottom). The proposed approach enables efficient importance sampling on the correct spiral-shaped manifold without a… view at source ↗

**Figure 3.** Left: Histogram of l(x) (distortion) from p(x) and Right: the importance sampling PDF q(x) which assigns higher weights to instances with high distortion in the neural compressor. This allows rare features to be more readily observed by increasing the likelihood of sampling high-distortion instances. sions—describing the wireless transmission link conditions between devices. Given the pivotal role of CSI in optimizi… view at source ↗

Figures from the paper (2 more)

**Figure 4.** Figure 4: Can a neural classifier serve as an importance weight in a completely training-free manner for SGM which is never trained with class information? - YES. Our method can use any external differentiable importance weight function, e.g., a neural gender classifier. histogr…

**Figure 5.** Figure 5: First row: Samples from p(x), Second row: samples generated from q(x). Our approach can generate samples containing elevated high-frequency components via setting l(x) accordingly. Now, assume an independent neural classifier can identify whether an image belongs to th…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

55 extracted references · 43 canonical work pages

[1]

Denoising diffusion probabilistic models,

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

2020
[2]

Denoising diffusion implicit models,

J. Song, C. Meng, and S. Ermon, “Denoising diffusion implicit models,” inInternational Conference on Learning Representations, 2021

2021
[3]

Score-based gener- ative modeling through stochastic differential equations,

Y. Song, J. Sohl-Dickstein, D. P. Kingma, A. Kumar, S. Ermon, and B. Poole, “Score-based gener- ative modeling through stochastic differential equations,” inInternational Conference on Learning Representations, 2021

2021
[4]

Diffusion models beat GANs on image synthesis,

P. Dhariwal and A. Q. Nichol, “Diffusion models beat GANs on image synthesis,” inAdvances in Neural Information Processing Systems, 2021

work page 2021
[5]

Zero-shot text-to-image generation,

A. Ramesh, M. Pavlov, G. Goh, S. Gray, C. Voss, A. Radford, M. Chen, and I. Sutskever, “Zero-shot text-to-image generation,” inProceedings of the 38th International Conference on Machine Learning (M. Meila and T. Zhang, eds.), vol. 139 ofProceedings of Machine Learning Research, pp. 8821–8831, PMLR, 18–24 Jul 2021

work page 2021
[6]

Diffwave: A versatile diffusion model for audio synthesis,

Z. Kong, W. Ping, J. Huang, K. Zhao, and B. Catanzaro, “Diffwave: A versatile diffusion model for audio synthesis,” inInternational Conference on Learning Representations (ICLR), 2021

work page 2021
[7]

WaveGrad: Estimating gradients for waveform generation,

N. Chen, Y. Zhang, H. Zen, R. J. Weiss, M. Norouzi, and W. Chan, “WaveGrad: Estimating gradients for waveform generation,” inInternational Conference on Learning Representations (ICLR), 2021

work page 2021
[8]

Generating high dimensional user-specific wireless channels using diffusion models

T. Lee, J. Park, H. Kim, and J. G. Andrews, “Generating high dimensional user-specific wireless channels using diffusion models.” arXiv preprint arXiv:2409.03924, 2024

arXiv 2024

Show all 55 references

[9]

A denoising diffusion probabilistic model-based digital twinning of isac mimo channel,

J. Zhang, S. Xu, Z. Zhang, C. Li, and L. Yang, “A denoising diffusion probabilistic model-based digital twinning of isac mimo channel,”IEEE Internet of Things Journal, vol. PP, no. 99, pp. 1–1, 2024

2024
[10]

C. P. Robert and G. Casella,Monte Carlo Statistical Methods. Springer Texts in Statistics, New York: Springer, 2nd ed., 2004

2004
[11]

What is the effect of importance weighting in deep learning?,

J. Byrd and Z. C. Lipton, “What is the effect of importance weighting in deep learning?,” inProceedings of the 36th International Conference on Machine Learning, vol. 97 ofProceedings of Machine Learning Research, pp. –, PMLR, June 2019

2019
[12]

Learning bounds for importance weighting,

C. Cortes, Y. Mansour, and M. Mohri, “Learning bounds for importance weighting,” inAdvances in Neural Information Processing Systems (NIPS), 2010

2010
[13]

Data preprocessing techniques for classification without discrimination,

F. Kamiran and T. Calders, “Data preprocessing techniques for classification without discrimination,” Knowledge and Information Systems, vol. 33, pp. 1–33, 2012

2012
[14]

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

A. Q. Nichol, P. Dhariwal, A. Ramesh, P. Shyam, P. Mishkin, B. Mcgrew, I. Sutskever, and M. Chen, “Glide: Towards photorealistic image generation and editing with text-guided diffusion models,” in International Conference on Machine Learning, pp. 16784–16804, PMLR, 2022. 15

2022
[15]

Sdxl: Improving latent diffusion models for high-resolution image synthesis,

D. Podell, Z. English, K. Lacey, A. Blattmann, T. Dockhorn, J. Müller, J. Penna, and R. Rombach, “Sdxl: Improving latent diffusion models for high-resolution image synthesis,” inInternational Conference on Learning Representations, 2024

2024
[16]

Refining generative process with discriminator guidance in score-based diffusion models,

D. Kim, Y. Kim, S. J. Kwon, W. Kang, and I.-C. Moon, “Refining generative process with discriminator guidance in score-based diffusion models,” inInternational Conference on Machine Learning, pp. 16567– 16598, PMLR, 2023

2023
[17]

Beyond first-order tweedie: Solving inverse problems using latent diffusion,

L. Rout, Y. Chen, A. Kumar, C. Caramanis, S. Shakkottai, and W.-S. Chu, “Beyond first-order tweedie: Solving inverse problems using latent diffusion,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 9472–9481, 2024

2024
[18]

Solving linear inverse problems provably via posterior sampling with latent diffusion models,

L. Rout, N. Raoof, G. Daras, C. Caramanis, A. Dimakis, and S. Shakkottai, “Solving linear inverse problems provably via posterior sampling with latent diffusion models,”Advances in Neural Information Processing Systems, vol. 36, 2024

2024
[19]

Diffusion posterior sampling for general noisy inverse problems,

H. Chung, J. Kim, M. T. Mccann, M. L. Klasky, and J. C. Ye, “Diffusion posterior sampling for general noisy inverse problems,” inInternational Conference on Learning Representations, 2022

2022
[20]

Score-based diffusion meets annealed importance sampling,

A. Doucet, W. Grathwohl, A. G. Matthews, and H. Strathmann, “Score-based diffusion meets annealed importance sampling,” Advances in Neural Information Processing Systems, vol. 35, pp. 21482–21494, 2022

2022
[21]

Reverse-time diffusion equation models,

B. D. Anderson, “Reverse-time diffusion equation models,”Stochastic Processes and their Applications, vol. 12, no. 3, pp. 313–326, 1982

1982
[22]

Repaint: Inpainting using denoising diffusion probabilistic models,

A. Lugmayr, M. Danelljan, A. Romero, F. Yu, R. Timofte, and L. Van Gool, “Repaint: Inpainting using denoising diffusion probabilistic models,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 11461–11471, 2022

2022
[23]

Improved denoising diffusion probabilistic models,

A. Q. Nichol and P. Dhariwal, “Improved denoising diffusion probabilistic models,” inInternational conference on machine learning, pp. 8162–8171, PMLR, 2021

2021
[24]

Ilvr: Conditioning method for denoising diffusion probabilistic models,

J. Choi, S. Kim, Y. Jeong, Y. Gwon, and S. Yoon, “Ilvr: Conditioning method for denoising diffusion probabilistic models,” arXiv preprint arXiv:2108.02938, 2021

2021 arXiv
[25]

J. V. Beck, K. D. Cole, A. Haji-Sheikh, and B. Litkouhl,Heat conduction using Green’s function. Taylor & Francis, 1992

1992
[26]

D. G. Duffy,Green’s functions with applications. Chapman and Hall/CRC, 2015

2015
[27]

E. N. Economou,Green’s functions in quantum physics, vol. 7. Springer Science & Business Media, 2006

2006
[28]

Tweedie’s formula and selection bias,

B. Efron, “Tweedie’s formula and selection bias,”Journal of the American Statistical Association, vol. 106, no. 496, pp. 1602–1614, 2011

2011
[29]

Noise2score: tweedie’s approach to self-supervised image denoising without clean images,

K. Kim and J. C. Ye, “Noise2score: tweedie’s approach to self-supervised image denoising without clean images,” Advances in Neural Information Processing Systems, vol. 34, pp. 864–874, 2021

2021
[30]

P. E. Kloeden, E. Platen, P. E. Kloeden, and E. Platen,Stochastic differential equations. Springer, 1992

1992
[31]

Würstchen: An efficient architecture for large-scale text-to-image diffusion models,

P. Pernias, D. Rampas, M. L. Richter, C. Pal, and M. Aubreville, “Würstchen: An efficient architecture for large-scale text-to-image diffusion models,” inThe Twelfth International Conference on Learning Representations, 2024

2024
[32]

High-resolution image synthesis with latent diffusion models,

R. Rombach, A. Blattmann, D. Lorenz, P. Esser, and B. Ommer, “High-resolution image synthesis with latent diffusion models,” inProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pp. 10684–10695, 2022. 16

2022
[33]

Improved analysis of score-based generative modeling: User-friendly bounds under minimal smoothness assumptions,

H. Chen, H. Lee, and J. Lu, “Improved analysis of score-based generative modeling: User-friendly bounds under minimal smoothness assumptions,” inInternational Conference on Machine Learning, pp. 4735–4763, PMLR, 2023

2023
[34]

Convergence of denoising diffusion models under the manifold hypothesis,

V. De Bortoli, “Convergence of denoising diffusion models under the manifold hypothesis,”Transactions on Machine Learning Research, 2022

2022
[35]

Sampling is as easy as learning the score: theory for diffusion models with minimal data assumptions,

S. Chen, S. Chewi, J. Li, Y. Li, A. Salim, and A. Zhang, “Sampling is as easy as learning the score: theory for diffusion models with minimal data assumptions,” inThe Eleventh International Conference on Learning Representations, 2023

2023
[36]

Diffusion schrödinger bridge with applications to score-based generative modeling,

V. De Bortoli, J. Thornton, J. Heng, and A. Doucet, “Diffusion schrödinger bridge with applications to score-based generative modeling,” Advances in Neural Information Processing Systems, vol. 34, pp. 17695–17709, 2021

2021
[37]

Thecross-entropymethodforoptimization,

Z.I.Botev, D.P.Kroese, R.Y.Rubinstein, andP.L’Ecuyer, “Thecross-entropymethodforoptimization,” in Handbook of statistics, vol. 31, pp. 35–59, Elsevier, 2013

2013
[38]

Scalable reversible generative models with free-form continuous dynamics,

W. Grathwohl, R. T. Chen, J. Bettencourt, and D. Duvenaud, “Scalable reversible generative models with free-form continuous dynamics,” inInternational Conference on Learning Representations, p. 7, 2019

2019
[39]

Invertible densenets with concatenated lipswish,

Y. Perugachi-Diaz, J. Tomczak, and S. Bhulai, “Invertible densenets with concatenated lipswish,” Advances in Neural Information Processing Systems, vol. 34, pp. 17246–17257, 2021

2021
[40]

A new metric for probability distributions,

D. M. Endres and J. E. Schindelin, “A new metric for probability distributions,”IEEE Transactions on Information theory, vol. 49, no. 7, pp. 1858–1860, 2003

2003
[41]

Made: Masked autoencoder for distribution estimation,

M. Germain, K. Gregor, I. Murray, and H. Larochelle, “Made: Masked autoencoder for distribution estimation,” in International conference on machine learning, pp. 881–889, PMLR, 2015

2015
[42]

Masked autoregressive flow for density estimation,

G. Papamakarios, T. Pavlakou, and I. Murray, “Masked autoregressive flow for density estimation,” Advances in neural information processing systems, vol. 30, 2017

2017
[43]

Neural importance sampling,

T. Müller, B. McWilliams, F. Rousselle, M. Gross, and J. Novák, “Neural importance sampling,”ACM Transactions on Graphics (ToG), vol. 38, no. 5, pp. 1–19, 2019

2019
[44]

Neural spline flows,

C. Durkan, A. Bekasov, I. Murray, and G. Papamakarios, “Neural spline flows,”Advances in neural information processing systems, vol. 32, 2019

2019
[45]

Cubic-spline flows,

C. Durkan, A. Bekasov, I. Murray, and G. Papamakarios, “Cubic-spline flows,” inICML Workshop on Invertible Neural Nets and Normalizing Flows, 2019

2019
[46]

Transformer neural autoregressive flows,

M. Patacchiola, A. Shysheya, K. Hofmann, and R. E. Turner, “Transformer neural autoregressive flows,” in ICML 2024 Workshop on Structured Probabilistic Inference & Generative Modeling, 2024

2024
[47]

An overview of 5g advanced evolution in 3gpp release 18,

X. Lin, “An overview of 5g advanced evolution in 3gpp release 18,”IEEE Communications Standards Magazine, vol. 6, no. 3, pp. 77–83, 2022

2022
[48]

Overview of deep learning-based csi feedback in massive mimo systems,

J. Guo, C.-K. Wen, S. Jin, and G. Y. Li, “Overview of deep learning-based csi feedback in massive mimo systems,” IEEE Transactions on Communications, vol. 70, no. 12, pp. 8017–8045, 2022

2022
[49]

Digital Twin Aided Massive MIMO: CSI Compression and Feedback,

S. Jiang and A. Alkhateeb, “Digital Twin Aided Massive MIMO: CSI Compression and Feedback,” in IEEE International Conference on Communications, (Denver, CO, USA), pp. 1–6, IEEE, June 2024

2024
[50]

AI Empowered Wireless Communications: From Bits to Semantics,

Z. Qin, L. Liang, Z. Wang, S. Jin, X. Tao, and W. Tong, “AI Empowered Wireless Communications: From Bits to Semantics,”Proceedings of the IEEE, vol. 112, pp. 621–652, July 2024

2024
[51]

Quadriga: A 3-d multi-cell channel model with time evolution for enabling virtual field trials,

S. Jaeckel et al., “Quadriga: A 3-d multi-cell channel model with time evolution for enabling virtual field trials,” IEEE transactions on antennas and propagation, vol. 62, no. 6, pp. 3242–3256, 2014

2014
[52]

Quadriga - quasi deterministic radio channel generator, user manual and documenta- tion,

S. Jaeckel et al., “Quadriga - quasi deterministic radio channel generator, user manual and documenta- tion,” Fraun-hofer Heinrich Hertz Institute, vol. 2.6.1, 2021. 17

2021
[53]

Deep learning for massive mimo csi feedback,

C.-K. Wen, W.-T. Shih, and S. Jin, “Deep learning for massive mimo csi feedback,”IEEE Wireless Communications Letters, vol. 7, no. 5, pp. 748–751, 2018

2018
[54]

Learning variable-rate codes for csi feedback,

H. Kim, H. Kim, and G. De Veciana, “Learning variable-rate codes for csi feedback,” inGLOBECOM 2022-2022 IEEE Global Communications Conference, pp. 1435–1441, IEEE, 2022

2022
[55]

Deep learning face attributes in the wild,

Z. Liu, P. Luo, X. Wang, and X. Tang, “Deep learning face attributes in the wild,” inProceedings of the IEEE International Conference on Computer Vision (ICCV), pp. 3730–3738, 2015. 18

2015

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.

[1] [1]

Denoising diffusion probabilistic models,

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

2020

[2] [2]

Denoising diffusion implicit models,

J. Song, C. Meng, and S. Ermon, “Denoising diffusion implicit models,” inInternational Conference on Learning Representations, 2021

2021

[3] [3]

Score-based gener- ative modeling through stochastic differential equations,

Y. Song, J. Sohl-Dickstein, D. P. Kingma, A. Kumar, S. Ermon, and B. Poole, “Score-based gener- ative modeling through stochastic differential equations,” inInternational Conference on Learning Representations, 2021

2021

[4] [4]

Diffusion models beat GANs on image synthesis,

P. Dhariwal and A. Q. Nichol, “Diffusion models beat GANs on image synthesis,” inAdvances in Neural Information Processing Systems, 2021

work page 2021

[5] [5]

Zero-shot text-to-image generation,

A. Ramesh, M. Pavlov, G. Goh, S. Gray, C. Voss, A. Radford, M. Chen, and I. Sutskever, “Zero-shot text-to-image generation,” inProceedings of the 38th International Conference on Machine Learning (M. Meila and T. Zhang, eds.), vol. 139 ofProceedings of Machine Learning Research, pp. 8821–8831, PMLR, 18–24 Jul 2021

work page 2021

[6] [6]

Diffwave: A versatile diffusion model for audio synthesis,

Z. Kong, W. Ping, J. Huang, K. Zhao, and B. Catanzaro, “Diffwave: A versatile diffusion model for audio synthesis,” inInternational Conference on Learning Representations (ICLR), 2021

work page 2021

[7] [7]

WaveGrad: Estimating gradients for waveform generation,

N. Chen, Y. Zhang, H. Zen, R. J. Weiss, M. Norouzi, and W. Chan, “WaveGrad: Estimating gradients for waveform generation,” inInternational Conference on Learning Representations (ICLR), 2021

work page 2021

[8] [8]

Generating high dimensional user-specific wireless channels using diffusion models

T. Lee, J. Park, H. Kim, and J. G. Andrews, “Generating high dimensional user-specific wireless channels using diffusion models.” arXiv preprint arXiv:2409.03924, 2024

arXiv 2024