Pith. sign in

REVIEW 5 major objections 7 minor 1 cited by

Erasing Noise in Signal Detection with Diffusion Model: From Theory to Application

T0 review · 5 major / 7 minor · reviewed 2026-08-10 · deepseek-v4-flash

Pith's one-line read The paper claims a diffusion-model receiver exceeds maximum-likelihood detection for BPSK and QAM in AWGN MIMO channels.

desk verdict A well-intentioned but mathematically broken attempt to turn diffusion denoising into a MIMO receiver; the SNR-to-timestep theorem does not solve its own equation, so the headline result is unsupported. read the letter →

arxiv 2501.07030 v1 pith:7YJ3WWOM submitted 2025-01-13 eess.SY cs.LGcs.SYeess.SP

classification eess.SYcs.LGcs.SYeess.SP
keywords diffusionmodelsignaldetectionmaximumlikelihoodMIMOAWGNsymbolerrorratedenoisingstochasticdifferentialequation
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper tries to establish that a denoising diffusion model can serve as a receiver front end that removes additive white Gaussian noise from received MIMO signals, so that symbol detection can beat maximum-likelihood (ML) estimation, long regarded as optimal, at lower computational cost. It develops a continuous-time SDE view of diffusion, proves a formula mapping any channel SNR to a diffusion timestep, and proves a scaling rule that maps the received signal into the diffusion latent distribution. With a diffusion transformer (DiT) backbone, the resulting receiver is reported to achieve lower symbol error rate than ML for BPSK and QAM in simulations while keeping complexity at $\mathcal{O}(n^2)$. If correct, this would mean the noise at the detector is no longer treated as irreducible, and neural receivers could exceed the classical optimal detector.

What carries the argument

The load-bearing object is the continuous-time diffusion process written as the SDE $dx_t=f_t x_t dt+g_t dw_t$, specialised here to latents $x_t=(1-t)x_0+\sqrt{t}\epsilon$ with $x_0=Hs$. Theorem 1 solves for $t$ by equating the expected signal-to-noise ratio of $x_t$ with that of the received signal, and Theorem 2 chooses $\alpha$ so that $E[\|\alpha r\|^2]=E[\|x_t\|^2]$. A diffusion transformer (DiT), a Transformer whose self-attention blocks are conditioned on the timestep through scale-and-shift, tokenizes $\alpha r$ and predicts $h_t$ and $\epsilon_t$; one denoising step ($\Delta t=t$) then gives an estimate of $Hs$, and the final symbol estimate is $H^{-1}\hat{Hs}$.

What would settle it

At 15 dB SNR the paper's own Table II gives $t=0.0467$ but $\alpha^2\sigma^2\approx0.0287$; feed the scaled received signal $\alpha r$ into the trained network and measure the variance of $\hat{Hs}-Hs$. If that residual variance is not smaller than the channel noise variance, the distribution-alignment step and the claimed SER advantage fail.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that additive Gaussian noise in a received signal can be erased by treating the received signal as a diffusion latent at a suitably chosen timestep. Theorem 1 gives the timestep $t$ that matches the expected signal-to-noise ratio of the diffusion latent $x_t=(1-t)Hs+\sqrt{t}\epsilon$ to that of the received signal, and Theorem 2 gives a scaling factor $\alpha$ that matches the expected squared norm of $\alpha r$ to that of $x_t$. In this setup $x_0=Hs$, so the trained diffusion transformer predicts the latent components $h_t$ and $\epsilon_t$, the signal component is recovered as $\hat{Hs}=\alpha r-t\hat{h}_t-\hat{\epsilon}_t$, and channel inversion yields the symbol estimate. The simulations report symbol error rates below the ML baseline for BPSK, 4QAM, 16QAM, and 64QAM over $4\times4$, $8\times8$, and $16\times16$ MIMO AWGN channels with one-step denoising.

Load-bearing premise

The argument rests on the assumption that scaling the received signal to the same average power as a diffusion training latent is enough to make the network's input match its training distribution, including the noise level per entry; if that fails, the denoiser is out of distribution and the claimed SER gain collapses.

Editorial extensions

If this is right

  • A single trained diffusion receiver can operate across a wide range of SNRs without fine-tuning, because $t$ and $\alpha$ are computed in closed form from the channel noise power.
  • The method gives a concrete $\mathcal{O}(n^2)$ inference path where ML enumeration is exponential, so high-order QAM and large MIMO become feasible in the reported setting.
  • Because denoising happens before the symbol decision, the effective noise seen by the detector is smaller than the channel noise, which is the stated mechanism for exceeding ML.
  • One-step denoising keeps inference latency at roughly a single neural-network forward pass rather than a long iterative diffusion sampling chain.
  • The ablation results indicate the gain comes from the diffusion-based formulation rather than from network capacity, since changing DiT depth and width barely changes SER.

Reading between the lines

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

  • If the advantage is real, the classical statement that ML is the optimal detector is conditional: it holds for receivers that treat channel noise as irreducible, and a denoising preprocessor can move the effective operating point beyond it.
  • The paper's own Table II shows $\alpha^2\sigma^2\neq t$ (for example, at 15 dB SNR, $\alpha^2\sigma^2\approx0.0287$ while $t=0.0467$), so the claimed distribution alignment is only approximate; a clean follow-up would isolate how much of the gain survives when the per-component noise variance is deliberately mismatched.
  • The same ‘match average power by scaling’ recipe could transfer to other denoising-inverse problems with known Gaussian noise levels, but that transfer is not established by the paper and would need separate tests.
  • A stress test that fixes the total norm while changing the per-component noise variance would reveal whether the reported SER advantage comes from the diffusion denoiser itself or from the subsequent zero-forcing/channel-inversion step.
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

5 major / 7 minor

Summary. The paper proposes a diffusion-model-based MIMO signal detector that claims to achieve lower symbol error rate than maximum-likelihood detection for BPSK and QAM in AWGN channels, while maintaining O(n^2) complexity. The theoretical part develops an SDE description of diffusion, derives an SNR-to-timestep mapping (Theorem 1), and introduces a scaling rule intended to make received signals in-distribution for the diffusion model (Theorem 2). The practical part uses a Diffusion Transformer backbone and reports simulations for Nr = 4, 8, 16 and modulations up to 256-QAM, including ablations over timestep, scaling, and network size.

Significance. If the claims were correct, the paper would overturn the long-standing optimality of ML detection for equally likely symbols and would provide a principled way to generalize a learned detector across SNRs without retraining. The paper also proposes a computationally attractive one-step denoising receiver and addresses a real problem, namely the poor SNR generalization of neural detectors. These are meaningful contributions. However, the derivation contains elementary algebraic and probabilistic errors: the SDE reduction is internally inconsistent, the closed-form solution in Theorem 1 does not solve its own equation, the reported timesteps in Table II are not reproducible from the stated formulas, and the distribution-alignment argument in Theorem 2 is only a norm-matching condition, not a distributional match. The central claim of outperforming ML is therefore not supported by the presented theory or by reproducible experiments.

major comments (5)
  1. [Section IV-A, Eqs. (10)-(14)] The SDE derivation is internally inconsistent. Setting g_t = 1 after Eq. (10) discards the variance-preserving structure of the DDPM forward process; the subsequent constraint Eq. (12) (∫_0^t h_s ds + x_0 = 0) combined with Eq. (11) implies that x_t is pure Brownian noise independent of x_0. This directly contradicts Eq. (14), which keeps the x_0 term. Eq. (16) is then asserted without derivation, and its variance Δt(t−Δt)/t I does not follow from the reverse SDE in Eq. (15). Since the denoising update in Eq. (17) and the one-step assumption in Remark 1 rest on these equations, the theoretical foundation of the method is not established.
  2. [Theorem 1, Eqs. (18)-(21)] The closed-form solution in Eq. (18) does not solve Eq. (19). With X = E[||Hs||^2]/(σ^2γ), Eq. (19) reduces to X = (1−t)^2/t, whose root in (0,1] is t = (2+X − sqrt(X^2+4X))/2; Eq. (18) divides by 4 instead of 2, giving a factor-of-two error. The same factor-of-two error appears in Remark 2, Eq. (21): with γ = E[||Hs||^2], the correct root is (2σ^2+1 − sqrt(1+4σ^2))/(2σ^2), not the expression with 4σ^2 in the denominator. The proof also sets E[||ϵ||^2] = 1, which is only valid for scalar noise; for a d-dimensional standard Gaussian vector, E[||ϵ||^2] = d, so the derived timestep depends on the signal dimension in an unacknowledged way.
  3. [Table II and Section VI-D] Table II is not reproducible from Eq. (21) or from Eq. (19). At 15 dB SNR, σ^2 = 10^{-1.5}: Eq. (21) gives t ≈ 0.014, the correct root of Eq. (19) gives t ≈ 0.030, and Table II lists t = 0.0467. The table also sets α = 1 − t, which is compatible with the norm-matching equation only if (1−t)^2σ^2 = t; at 15 dB, (1−t)^2σ^2 ≈ 0.029 while t = 0.0467. Thus the tabulated values cannot have been computed from Theorems 1 and 2, and the validation of the 'optimal' timestep in Fig. 6 is not supported by the reported data.
  4. [Theorem 2 and Section IV-B] Matching the expected squared norm of αr to that of x_t does not yield a distributional match; the proof itself concedes that the scaling 'cannot directly guarantee that the distribution of r and xt is consistent.' The subsequent assertion that r and x_t share the same distribution form because both are linear combinations of a signal and zero-mean Gaussian noise is not a proof, and the per-component noise variances differ: Var(αr | s) = α^2σ^2, whereas Var(x_t | x_0) = t under the paper's convention. With α = 1−t and the tabulated t values, α^2σ^2 ≠ t (e.g., ≈0.029 vs 0.047 at 15 dB), so the denoiser is fed out-of-distribution inputs. The OOD robustness claimed in Fig. 7 therefore does not follow from the theory.
  5. [Section V, Eqs. (23)-(25)] The estimation step is not derived. Under the forward model x_t = (1−t)x_0 + √t ϵ, the natural reconstruction would be Hs = (αr − √t ϵ)/(1−t) or the appropriate posterior mean, but Eq. (24) instead uses Hs = αr − t h_t − ϵ_t with no normalization and no derivation. Combined with the absence of a proof that this receiver can strictly outperform the ML decision rule for equally likely symbols in AWGN, the central performance claim in the abstract is not backed by a valid theoretical argument, and the simulation results cannot compensate for the inconsistencies in the supporting formulas.
minor comments (7)
  1. [Section III] The channel matrix H is defined as N_r × N_r, but a MIMO system normally has N_t transmit and N_r receive dimensions; the model and complexity statements should specify the full dimensions.
  2. [Theorem 1 proof] The notation in the variance computation, 'D[ϵ] − E[ϵ]^2', is unclear for vector-valued ϵ and should be written as E[||ϵ||^2]; as written it obscures the dimension dependence discussed above.
  3. [Theorem 2 proof] In the computation of E[||x_t||^2], the text writes E_{x0}[||x0||] instead of E[||x0||^2]; the missing square propagates into the final formula for α.
  4. [Eq. (13)] Equation (13) places a stochastic integral in the mean of a Gaussian distribution, which is not meaningful; the conditional distribution should be stated after conditioning on the drift path or with the drift treated accordingly.
  5. [Figures 3-10] Several figure captions and axis labels are garbled or incomplete (e.g., the axis text in Fig. 3 and Fig. 8), and Fig. 8's caption is partially unreadable; these should be cleaned for publication.
  6. [Section V, sentence after Eq. (23)] The sentence 'we propose a DM-based signal detection method is proposed as follows' is ungrammatical and should be rewritten.
  7. [References] The reverse SDE in Eq. (15) is attributed to [18], a preprint on decoupled diffusion models; the paper should instead cite the standard score-based SDE literature (e.g., Song et al., 2021) for this equation.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the SNR-to-timestep map and the scaling factor are fixed by stated moment-matching conditions that do not incorporate SER, so the SER claim is not forced by construction.

full rationale

The derivation chain is independent of the quantity it is used to predict. Theorem 1 fixes t as the solution of Eq. (19), a moment equality between received SNR and diffusion latent SNR; Theorem 2 fixes alpha by matching E||alpha r||^2 to E||x_t||^2 in Eq. (22). Neither construction uses the SER objective, and SER appears only afterward as the evaluation metric, so the observed SER advantage is not an input to the formulas. The only author-overlapping citation, [9] (SigT), is used as a comparison baseline, not as the load-bearing premise that diffusion models denoise or that DiT is the backbone. The paper itself flags the distributional limitation of Theorem 2: 'it cannot directly guarantee that the distribution of r and x_t is consistent' (end of Theorem 2 proof). That is a missing justification, not a circular reduction. The skeptic's point that Eq. (18)/(21) does not solve Eq. (19) and that Table II is not reproducible from the stated formula is an algebraic/validity concern about the derivation; it does not make a predicted quantity equivalent to its input by construction. For those reasons, the circularity score is 0.

Assumptions & free parameters 4 free parameters · 3 assumptions · 2 invented entities

The paper's central derivation depends on a nonstandard SDE formulation, a one-step denoising assumption, and a moment-matching condition that does not guarantee distributional match. The timestep and scaling values used in experiments are not consistently derived from the stated formulas, making them effective free parameters.

free parameters (4)
  • diffusion timestep t for each SNR = 0.6275 at SNR=-5 dB, 0.0467 at SNR=15 dB (Table II)
    The closed form (21) does not reproduce the tabulated values; for SNR=15 dB Eq. (21) gives t about 0.014 while Table II lists 0.0467. The t used in simulations is therefore effectively chosen, not derived from the stated theorem.
  • scaling factor alpha for each SNR = 1 - t per Table II
    Table II sets alpha = 1 - t, which is consistent with moment matching only if t satisfies the exact quadratic; since t is inconsistent with the stated formula, alpha is also not derived as claimed.
  • gamma = E[||x0||^2] = implicitly set to 1 in Remark 2
    The paper sets x0 = Hs and simplifies, but the dimension of the noise norm is mishandled: E||epsilon||^2 is treated as 1 instead of the vector dimension, making the formula dimensionally inconsistent.
  • noise power normalization E||n||^2 = sigma^2
    In (7c) the expected squared norm of the noise vector is equated to the scalar noise density sigma^2, ignoring the number of antennas, which changes the SNR-to-t mapping.
assumptions (3)
  • ad hoc to paper The DDPM forward process can be represented by SDE (10) with f_t x_t = h_t and g_t = 1.
    Eqs. (11)-(14) discard the variance-preserving structure of the original DDPM; setting g_t=1 changes the process and the integral form (14) contradicts Eq. (4).
  • ad hoc to paper A single denoising step with Delta t = t recovers x0.
    Remark 1 assumes one-step reverse from t to 0 is valid, but the reverse SDE (15) requires integration over the path; the inference formula (24) is a jump, not a diffusion reverse step.
  • domain assumption Matching E||alpha r||^2 to E||x_t||^2 makes the distributions equal.
    Stated in Section IV-B after Theorem 2; only first moments are matched, and the paper admits full distributional match is not guaranteed.
invented entities (2)
  • learnable drift h_t in the reverse process
    purpose: To parameterize the SDE drift so the network predicts both h_t and epsilon_t (Eq. 23).
    In standard DDPM the drift is determined by the noise schedule; treating it as an unknown to be learned is a new construction not supported by the cited SDE theory.
  • one-step denoising jump
    purpose: Allows O(n^2) inference by jumping from t to 0 in a single step (Eq. 24).
    Eq. (24) is not derived from the reverse SDE; it is a new inference rule that is asserted without formal justification.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Erasing Noise in Signal Detection with Diffusion Model: From Theory to Application." pith.science (2026). https://pith.science/paper/7YJ3WWOM

@misc{pith2026250107030,
  author       = {Pith},
  title        = {Pith review of: Erasing Noise in Signal Detection with Diffusion Model: From Theory to Application},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7YJ3WWOM}},
  note         = {Machine review of arXiv:2501.07030}
}
abstract

In this paper, a signal detection method based on the denoise diffusion model (DM) is proposed, which outperforms the maximum likelihood (ML) estimation method that has long been regarded as the optimal signal detection technique. Theoretically, a novel mathematical theory for intelligent signal detection based on stochastic differential equations (SDEs) is established in this paper, demonstrating the effectiveness of DM in reducing the additive white Gaussian noise in received signals. Moreover, a mathematical relationship between the signal-to-noise ratio (SNR) and the timestep in DM is established, revealing that for any given SNR, a corresponding optimal timestep can be identified. Furthermore, to address potential issues with out-of-distribution inputs in the DM, we employ a mathematical scaling technique that allows the trained DM to handle signal detection across a wide range of SNRs without any fine-tuning. Building on the above theoretical foundation, we propose a DM-based signal detection method, with the diffusion transformer (DiT) serving as the backbone neural network, whose computational complexity of this method is $\mathcal{O}(n^2)$. Simulation results demonstrate that, for BPSK and QAM modulation schemes, the DM-based method achieves a significantly lower symbol error rate (SER) compared to ML estimation, while maintaining a much lower computational complexity.

Figures

Figures reproduced from arXiv: 2501.07030 by the authors.

Figure 1
Figure 1. Block diagram of the signal detection method, (a) is the traditional signal detection system, [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. The illustration of the inferencing and training procedure of the proposed DM on the left, [PITH_FULL_IMAGE:figures/full_fig_p016_2.png] view at source ↗
Figure 3
Figure 3. SER on different SNR and Nr for different signal detection methods with BPSK modulation [PITH_FULL_IMAGE:figures/full_fig_p019_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: SER on different SNR and Nr for different signal detection methods with 4-QAM modulation. that the proposed DM-based intelligent signal detection framework effectively extends traditional signal detection methodologies. Based on this theory, it is possible to design ne…
Figure 5
Figure 5. Figure 5: SER on 4×4 MIMO with high-order QAM. To further evaluate the performance of the proposed DM-based signal detection method in high-order modulation systems, we conducted simulations focusing on systems with Nr = 4 receiving antennas. Specifically, we compared the perfor…
Figure 6
Figure 6. Figure 6: SER on different t. 6 4 2 0 2 4 6 SNR (dB) 10 1 SER 50% 80% 95% Original 105% 120% 150% 6 4 2 0 2 SNR (dB) 10 1 SER 50% Original 150% [PITH_FULL_IMAGE:figures/full_fig_p023_6.png]
Figure 7
Figure 7. Figure 7: , which illustrates the impact of scaling on the performance at different SNR levels. These observations confirm the correctness of the theoretical calculations of t and α as per Theorems 1 and 2. The numerical results validate the effectiveness of the proposed method …
Figure 8
Figure 8. Figure 8: SER on high SNR. t becomes very close to zero. In traditional DM training methods, where t is uniformly and randomly selected from the range (0, 1], the DM may not be adequately trained for high-SNR conditions corresponding to small t values. To address this limitation…
Figure 9
Figure 9. Figure 9: SER on different NN depths. 5.0 2.5 0.0 2.5 5.0 7.5 10.0 12.5 SNR (dB) 10 4 10 3 10 2 10 1 SER dim=256 dim=512 dim=1024 9 10 11 12 13 14 15 SNR (dB) 10 3 10 2 10 1 SER dim=256 dim=512 dim=1024 [PITH_FULL_IMAGE:figures/full_fig_p026_9.png]
Figure 10
Figure 10. Figure 10: SER on different NN dimensions. Traditionally, the depth and width of NNs significantly influence their feature extraction capabilities, which in turn affect the performance of NN-based signal detection methods trained using traditional approaches. To verify the versa…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Censored Sampling for Topology Design: Guiding Diffusion with Human Preferences

    cs.LG 2025-08 unverdicted novelty 4.0 of 10

    Guiding a pretrained topology-diffusion generator with human-preference reward classifiers is claimed to suppress floating-material and boundary-violation failure modes without retraining the generator.

Reference graph

Works this paper leans on

22 extracted references · 14 canonical work pages · cited by 1 Pith paper

  1. [1]

    A mathematical theory of communication,

    C. E. Shannon, “A mathematical theory of communication,” The Bell system technical journal , vol. 27, no. 3, pp. 379–423, 1948

  2. [2]

    Efficient qam signal detector for massive mimo systems via ps/dps-admm approaches,

    Q. Zhang, J. Wang, and Y . Wang, “Efficient qam signal detector for massive mimo systems via ps/dps-admm approaches,” IEEE Transactions on Wireless Communications , vol. 21, no. 10, pp. 8859–8871, 2022

  3. [3]

    Optimal noise benefits in neyman–pearson and inequality-constrained statistical signal detection,

    A. Patel and B. Kosko, “Optimal noise benefits in neyman–pearson and inequality-constrained statistical signal detection,” IEEE Transactions on Signal Processing , vol. 57, no. 5, pp. 1655–1669, 2009

  4. [4]

    A new deep learning framework for hf signal detection in wideband spectrogram,

    W. Li, K. Wang, L. You, and Z. Huang, “A new deep learning framework for hf signal detection in wideband spectrogram,” IEEE Signal Processing Letters , vol. 29, pp. 1342–1346, 2022

  5. [5]

    Maximum-likelihood estimation of parameters of signal-detection theory and determination of confidence intervals—rating-method data,

    D. D. Dorfman and E. Alf Jr, “Maximum-likelihood estimation of parameters of signal-detection theory and determination of confidence intervals—rating-method data,” Journal of mathematical psychology , vol. 6, no. 3, pp. 487–496, 1969

  6. [6]

    Symbol denoising in high order m-qam using residual learning of deep cnn,

    S. Khan, K. S. Khan, and S. Y . Shin, “Symbol denoising in high order m-qam using residual learning of deep cnn,” in 2019 16th IEEE Annual Consumer Communications & Networking Conference (CCNC) . IEEE, 2019, pp. 1–6

  7. [7]

    Beyond a gaussian denoiser: Residual learning of deep cnn for image denoising,

    K. Zhang, W. Zuo, Y . Chen, D. Meng, and L. Zhang, “Beyond a gaussian denoiser: Residual learning of deep cnn for image denoising,” IEEE transactions on image processing , vol. 26, no. 7, pp. 3142–3155, 2017

  8. [8]

    Comnet: Combination of deep learning and expert knowledge in ofdm receivers,

    X. Gao, S. Jin, C.-K. Wen, and G. Y . Li, “Comnet: Combination of deep learning and expert knowledge in ofdm receivers,” IEEE Communications Letters , vol. 22, no. 12, pp. 2627–2630, 2018

Show all 22 references
  1. [9]

    Sigt: An efficient end-to-end mimo-ofdm receiver framework based on transformer,

    Z. Ren, N. Cheng, R. Sun, X. Wang, N. Lu, and W. Xu, “Sigt: An efficient end-to-end mimo-ofdm receiver framework based on transformer,” in 2022 5th International Conference on Communications, Signal Processing, and their Applications (ICCSPA), 2022, pp. 1–6

  2. [10]

    Message passing meets graph neural networks: A new paradigm for massive mimo systems,

    H. He, X. Yu, J. Zhang, S. Song, and K. B. Letaief, “Message passing meets graph neural networks: A new paradigm for massive mimo systems,” IEEE Transactions on Wireless Communications , vol. 23, no. 5, pp. 4709–4723, 2024

  3. [11]

    Goodfellow, Y

    I. Goodfellow, Y . Bengio, and A. Courville, Deep learning. MIT press, 2016

  4. [12]

    Generalized odin: Detecting out-of-distribution image without learning from out-of-distribution data,

    Y .-C. Hsu, Y . Shen, H. Jin, and Z. Kira, “Generalized odin: Detecting out-of-distribution image without learning from out-of-distribution data,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2020, pp. 10 951–10 960

  5. [13]

    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

  6. [14]

    Diffbir: Towards blind image restoration with generative diffusion prior,

    X. Lin, J. He, Z. Chen, Z. Lyu, B. Fei, B. Dai, W. Ouyang, Y . Qiao, and C. Dong, “Diffbir: Towards blind image restoration with generative diffusion prior,” arXiv preprint arXiv:2308.15070 , 2023

  7. [15]

    Structured denoising diffusion models in discrete state-spaces,

    J. Austin, D. D. Johnson, J. Ho, D. Tarlow, and R. Van Den Berg, “Structured denoising diffusion models in discrete state-spaces,” Advances in Neural Information Processing Systems , vol. 34, pp. 17 981–17 993, 2021

  8. [16]

    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,” 2021

  9. [17]

    Score-based generative modeling through stochastic differential equations,

    Y . Song, J. Sohl-Dickstein, D. P. Kingma, A. Kumar, S. Ermon, and B. Poole, “Score-based generative modeling through stochastic differential equations,” arXiv preprint arXiv:2011.13456 , 2020

  10. [18]

    Decoupled diffusion models: Simultaneous image to zero and zero to noise,

    Y . Huang, Z. Qin, X. Liu, and K. Xu, “Decoupled diffusion models: Simultaneous image to zero and zero to noise,” 2024. 29

  11. [19]

    Blended diffusion for text-driven editing of natural images,

    O. Avrahami, D. Lischinski, and O. Fried, “Blended diffusion for text-driven editing of natural images,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 18 208–18 218

  12. [20]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is all you need,” Advances in neural information processing systems , vol. 30, 2017

  13. [21]

    Exploiting radio fingerprints for simultaneous localization and mapping,

    R. Liu, B. P. L. Lau, K. Ismail, A. Chathuranga, C. Yuen, S. X. Yang, Y . L. Guan, S. Mao, and U.-X. Tan, “Exploiting radio fingerprints for simultaneous localization and mapping,” IEEE Pervasive Comput., 2023

  14. [22]

    Adversarial variational bayes: Unifying variational autoencoders and generative adversarial networks,

    L. Mescheder, S. Nowozin, and A. Geiger, “Adversarial variational bayes: Unifying variational autoencoders and generative adversarial networks,” in International conference on machine learning . PMLR, 2017, pp. 2391–2400

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.