Pith. sign in

REVIEW 3 major objections 4 minor 1 cited by

Shaping Inductive Bias in Diffusion Models through Frequency-Based Noise Control

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read This paper argues that the noising process in a diffusion model is not just a computational convenience but a direct lever on what the model learns: whatever information the forward process erases is exactly what the denoiser must…

desk verdict The corruption-recovery experiment is a real result; the FID gains on natural datasets and the theory are too thin for the headline claims. read the letter →

arxiv 2502.10236 v2 pith:UNM2H65H submitted 2025-02-14 cs.LG cs.AI

Thomas Jiralerspong , Berton Earnshaw , Jason Hartford , Yoshua Bengio , Luca Scimeca This is my paper
classification cs.LGcs.AI
keywords diffusionprobabilisticmodelsfrequency-domainnoiseinductivebiasschedulecoloredGaussiangenerativemodelingFIDspectral
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 argues that the noising process in a diffusion model is not just a computational convenience but a direct lever on what the model learns: whatever information the forward process erases is exactly what the denoiser must reconstruct, so choosing which frequencies get erased steers the model's inductive bias. To test this, the authors replace the usual white Gaussian noise with Gaussian noise whose power spectrum is shaped by a frequency weighting, calling the result frequency diffusion. They report that on data whose information sits mainly in low frequencies, low-frequency-weighted forward noise gives better FID than standard diffusion, and across five datasets different frequency weightings win, with three datasets improving over the Gaussian baseline. They further show that if training data is corrupted with noise restricted to a frequency band, training with a forward noising operator that leaves that band untouched lets the model ignore the corruption and generate samples closer to the original uncorrupted distribution. The paper positions the forward noising operator as a tunable inductive-bias knob for diffusion models.

What carries the argument

The machinery is the frequency-controlled noise operator, which turns standard white Gaussian noise into Gaussian noise with a chosen power spectrum. Starting from a complex Gaussian field in Fourier space, the operator scales each frequency component by a weighting $w(f_x, f_y)$ and returns $\epsilon^{(w)} = \Re(\mathcal{F}^{-1}(N_{\mathrm{freq}}^{(w)}))$, a real-valued spatial noise field. Because linear operations on Gaussians produce Gaussians, the training loop keeps the usual $\epsilon$-prediction objective while the noise's frequency content acts as the inductive-bias knob. The experiments use a two-band mixture of low- and high-frequency filtered noises with mixing weights $\gamma_l$ and $\gamma_h$, and the corruption-recovery task uses the same construction with the corrupted band excluded from the forward process.

What would settle it

Derive the exact marginal $q(x_t|x_0)$ for the frequency-weighted forward process; if the conditional distribution of the noise target is not the zero-mean unit-variance Gaussian assumed in the $\epsilon$-prediction loss, then the loss in Eq. (3) is no longer a valid bound and the FID gains could come from a heuristic rather than from inductive-bias steering. A direct experiment would train with frequency-shaped forward noise but sample the reverse process with standard white noise at every step, and vice versa; if matching the noise spectrum does not change sample quality, the claim that the noising operator steers learning is falsified.

Watch

Extended reading notes

Core claim

The paper's central claim is that the noising operator directly shapes a diffusion model's representation of the data, because the denoising network is under pressure to learn precisely the information that the forward process destroys. In the Fourier domain, the authors build frequency-shaped Gaussian noise by sampling a complex Gaussian field $N_{\mathrm{freq}} = N_{\mathrm{real}} + i N_{\mathrm{imag}}$, multiplying each frequency bin by a weighting function $w(f_x,f_y)$, and inverse-transforming to the spatial domain to get $\epsilon^{(w)} = \Re(\mathcal{F}^{-1}(N_{\mathrm{freq}}^{(w)}))$; standard white Gaussian noise is the special case $w(f) = 1$. Using a two-band mixture $\epsilon_f = \gamma_l \epsilon_{[a_l,b_l]} + \gamma_h \epsilon_{[a_h,b_h]}$, the paper sweeps the low/high balance and finds dataset-dependent, often monotonic, FID rankings: low-frequency-forward schedules win on low-frequency-dominated data, while high-frequency-forward schedules win on datasets where sharp details matter. In the corruption experiments, when the data is degraded by band-restricted noise $\epsilon_{[a_c,b_c]}$, setting the forward noising operator to avoid that band removes the denoiser's pressure to learn the corruption, and the trained model produces samples closer to the original uncorrupted distribution than standard diffusion in all eight tested bands. The paper reads these results as evidence that frequency-based noise control is a practical way to inject task-specific inductive biases into diffusion training.

Load-bearing premise

The comparison rests on the assumption that replacing white Gaussian noise with frequency-filtered Gaussian noise in the forward process still leaves the standard diffusion training loss valid; the paper does not derive the reverse-process covariance, adjust the loss weighting, or specify what noise is added during sampling for the colored case.

Editorial extensions

If this is right

  • The noising operator becomes a tunable inductive-bias knob: dataset-specific frequency schedules can match the dominant information content, and the paper reports lower FID and KID than the white-noise baseline on three of five tested datasets.
  • Frequency-shaped noise enables selective learning: a model trained on band-corrupted data with the corrupted band omitted from the forward process generates samples closer to the original uncorrupted distribution than a standard baseline, across all eight tested bands.
  • Monotonic FID trends across $\gamma_l$ sweeps can reveal which frequency bands carry the information needed for generation, making the noise spectrum a diagnostic tool rather than a fixed hyperparameter.
  • The framework strictly generalizes standard diffusion, since white Gaussian noise is recovered as the $w(f)=1$ special case, so no architectural change is needed to apply the method.

Reading between the lines

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

  • If the noising operator is a true inductive-bias knob, dynamic schedules that shift frequency emphasis over diffusion time—coarse structure first, fine texture later—should outperform any constant weighting; this is an untested extension the authors mention as future work.
  • The corruption-recovery result likely extends beyond images to any domain with a meaningful Fourier structure, such as audio, time series, or physics fields, where band-restricted degradations are common.
  • The deepest implication is that any structured degradation expressible as a subspace of a Gaussian noise process could in principle be omitted from the forward process, making 'ignore the corrupted part of the signal' a general recipe rather than a frequency-specific trick.
  • A cheap practical consequence: because the method requires no architecture change, frequency diffusion is a natural baseline for future inductive-bias work in generative modeling.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper proposes 'frequency diffusion', a modification of the DDPM forward process in which isotropic Gaussian noise is replaced by frequency-shaped Gaussian noise produced by filtering complex Gaussian noise in the Fourier domain and taking the real part. The authors argue that this noising operator acts as an inductive-bias knob: erasing certain frequency bands creates pressure on the denoiser to learn those bands, while leaving other bands intact lets the model ignore them. They define two-band mixture schedules (Eq. 12), train diffusion models on MNIST, CIFAR-10, DomainNet-Quickdraw, WikiArt, and CelebA, and report FID/KID comparisons against standard diffusion. They also present a corruption-recovery experiment where data corrupted in a known frequency band is used to train a model that, with a frequency-omitting schedule, approximately recovers the original uncorrupted distribution.

Significance. The underlying idea is attractive: if the forward noising operator can be used to steer what the denoising model learns, the forward process becomes a meaningful design choice for generative priors, and the qualitative figures (Figure 2 and Figure 4) support the intuition. The paper also deserves credit for stating its limitation explicitly: Section 4 acknowledges that schedule design currently requires empirical validation. However, the central empirical claim is not yet supported by the reported numbers, and the theoretical foundation is incomplete: substituting non-white noise into a standard DDPM objective without re-deriving the loss or the sampling procedure leaves the comparisons in Tables 1 and 2 on questionable footing. The corruption-recovery result is a useful proof-of-concept but is constructed from knowledge of the corruption band, which limits its scope. Overall, the contribution is promising but not yet at a level where the main claims are established.

major comments (3)
  1. [Section 2.2, Eq. (7)] The paper replaces epsilon ~ N(0,I) with epsilon^(w) = Re(F^{-1}(w ⊙ N_freq)) but keeps the standard epsilon-prediction loss (Eq. 3) and the standard forward scaling (Eq. 4) without deriving the correct training or sampling objective. epsilon^(w) is Gaussian with non-identity covariance: its power spectrum is |w(f)|^2, so the forward marginal and the reverse posterior q(x_{t-1}|x_t,x_0) are not those of standard DDPM. A valid ELBO would require a covariance-aware loss (inverse-covariance weighting) and a modified reverse covariance; the paper provides neither. It also does not specify the FFT normalization or the distribution used to initialize sampling, both of which affect the effective signal-to-noise ratio. The statement in Section 1 that the Fourier transform of a Gaussian is Gaussian only preserves Gaussianity, not the identity covariance that the standard objective relies on. This must be fixed or explicitly justified before the FID comparisons can be interpreted as comparing validly trained models.
  2. [Section 3.1.3, Table 1] The claim that frequency diffusion 'outperforms standard diffusion' on three of five datasets is not supported by the reported numbers when standard errors are considered. For example, on CIFAR-10 the baseline FID is 0.1055±0.0042 while the best reported settings (gamma_l=0.4 and 0.6) give 0.1131±0.0079 and 0.1131±0.0074, which are worse than the baseline. On DomainNet-Quickdraw the baseline is 0.0875±0.0060 and the best is 0.0799±0.0166, a gap that is well within one standard error. On WikiArt the baseline 0.1622±0.0133 overlaps with the best 0.1348±0.0126, and no significance test is reported. Furthermore, the best gamma_l per dataset is selected by sweeping on the test FID, making the reported best-vs-baseline gap a post-hoc best-of-nine result, not an independent prediction. The paper should either report held-out model selection, provide significance tests or confidence intervals, or substantially weaken the performance claims.
  3. [Section 3.2, Eq. (13)] The corruption-recovery experiment is constructed from the known corruption band: the authors set b_l = a_c and a_h = b_c so that the forward noising operator omits exactly the frequency range in which the corruption noise lies. This demonstrates that a schedule can be designed when the corruption is fully known, but it is not a recovery method for unknown corruption, and the framing in Section 3.2 as 'recovery of the original, noiseless, distribution' overstates the result. The paper should explicitly state that the schedule is hand-designed from knowledge of the corruption process, and it should present the experiment as a proof-of-concept of the omission mechanism rather than as a general recovery result.
minor comments (4)
  1. [Section 2.4, References] The MNIST dataset is cited to Matthey et al. (2017), which is the dsprites dataset paper; the correct reference for MNIST is LeCun et al. (1998) or a standard source.
  2. [Section 4, Discussion] There are several typos: 'nosing' should be 'noising', 'stong' should be 'strong', and 'DPS' should be 'DPMs'.
  3. [Section 2.3, Eq. (12)] The two-band mixture is introduced as epsilon_f = gamma_l epsilon_[a_l,b_l] + gamma_h epsilon_[a_h,b_h], but the paper does not state whether the two band-limited noises on the right are independent or whether the gamma weights are normalized to preserve total noise variance; this matters for the claim that gamma_l = 0.5, gamma_h = 0.5 yields 'standard Gaussian noise'.
  4. [Section 3.1.2, Figure 3] The text says the average FID and error are reported in Figure 3, but the caption does not describe how error bars or shaded regions are computed; please clarify the visualization.

Circularity Check

1 steps flagged · score 3.0 of 10

One confirmatory-by-design recovery demonstration, but no circular core: the main inductive-bias claim rests on empirical comparisons rather than fitted identities.

  1. self definitional [Section 3.2, around Eq. (13) and the sentence following it]
    "Assuming knowledge of the corruption process, we frame the frequency diffusion learning procedures as a noiseless distribution recovery process, and set 𝑎𝑙 = 0, 𝑏ℎ = 1, 𝑏𝑙 = 𝑎𝑐, and 𝑎ℎ = 𝑏𝑐. This formulation effectively allows for the forward frequency noising operator to omit the range of frequencies in which the noise lies."

    The forward noising mask is defined in terms of the corruption band [a_c,b_c]: the passband is its complement. Therefore the claimed recovery success ('frequency diffusion DPMs ... able to discard the corrupting information and recover the original distribution') is built into the schedule assignment; the model is trained on a forward process that never injects the corrupted frequencies and is never asked to denoise them. This is a confirmatory-by-design demonstration rather than an independent prediction of the inductive-bias hypothesis. It is a secondary contribution, however; the main claim (noising operator steers representation) is not reduced to this construction.

full rationale

No load-bearing self-citation or imported uniqueness theorem is present; references to the authors' earlier work are background only. The main derivation (shaped Gaussian noise via Eq. 7 and its use in Eq. 4) is definitional, not circular, although its validity as a DDPM training objective is questionable. The dataset sweeps in Table 1 are post-hoc selection over test FID, which weakens the causal claim but is not an equation-level circularity. The only by-construction element is the Section 3.2 corruption-recovery experiment, where the mask is derived from the known corruption band; its success is therefore expected by design. Because this is one secondary demonstration and the central empirical comparisons are not fitted identities, the paper is only mildly circular.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

Everything the central claim depends on that the reader did not pay for upstream: dataset-specific gamma chosen by test-set sweeps, fixed band partitions, and the unstated validity of standard DDPM equations for colored noise. No new physical entities are introduced.

free parameters (3)
  • gamma_l / gamma_h mixing weights = Best per dataset: DomainNet 0.4/0.6, WikiArt 0.6/0.4, CelebA 0.3/0.7, MNIST and CIFAR at baseline 0.5/0.5
    Swept over {0.1,...,0.9} in Section 3.1.3 and selected by FID; not derived from data statistics.
  • two-band partition (a_l, b_l, a_h, b_h) = Fixed at [0,0.5],[0.5,1] for natural datasets; set to corruption band for recovery experiments
    Chosen without ablation; the corruption experiment reuses the known corruption band.
  • corruption amplitude gamma_c = 1
    Set to 1 in Section 3.2 without sensitivity analysis.
assumptions (3)
  • domain assumption The standard DDPM epsilon-prediction objective and reverse-sampling procedure remain valid when the forward noise is frequency-filtered Gaussian with non-identity covariance.
    Section 2.2 defines epsilon^(w) and continues to use Eq. (3)'s epsilon-prediction loss without deriving the reverse-process covariance or sampling noise.
  • ad hoc to paper A uniform sweep over gamma_l with a fixed two-band partition is sufficient to find the best inductive-bias schedule.
    Section 3.1.3 sweeps only gamma_l in 0.1 increments with fixed band boundaries; no theory or search over band boundaries is provided.
  • domain assumption FID and KID computed on Inception v3 block 768 embeddings are valid metrics for all five datasets, including MNIST and sketches.
    Section 3 states metrics are computed on pre-trained Inception embeddings; no validation that this is reliable for non-photo datasets.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Shaping Inductive Bias in Diffusion Models through Frequency-Based Noise Control." pith.science (2026). https://pith.science/paper/UNM2H65H

@misc{pith2026250210236,
  author       = {Pith},
  title        = {Pith review of: Shaping Inductive Bias in Diffusion Models through Frequency-Based Noise Control},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/UNM2H65H}},
  note         = {Machine review of arXiv:2502.10236}
}
read the original abstract

Diffusion Probabilistic Models (DPMs) are powerful generative models that have achieved unparalleled success in a number of generative tasks. In this work, we aim to build inductive biases into the training and sampling of diffusion models to better accommodate the target distribution of the data to model. For topologically structured data, we devise a frequency-based noising operator to purposefully manipulate, and set, these inductive biases. We first show that appropriate manipulations of the noising forward process can lead DPMs to focus on particular aspects of the distribution to learn. We show that different datasets necessitate different inductive biases, and that appropriate frequency-based noise control induces increased generative performance compared to standard diffusion. Finally, we demonstrate the possibility of ignoring information at particular frequencies while learning. We show this in an image corruption and recovery task, where we train a DPM to recover the original target distribution after severe noise corruption.

Figures

Figures reproduced from arXiv: 2502.10236 by the authors.

Figure 1
Figure 1. Frequency diffusion under a generalized framework. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Power spectra and image visuals of the forward Process in standard diffusion, as compared to high and low-frequency noise settings of a two￾band mixture noise parametrization. covering 345 object categories (Peng et al., 2019). We follow the formulation in Equa￾tion 12 to train and compare diffusion models with a noisy operator prioritizing different parts of the frequency distribution. In these experi￾ments we fix … view at source ↗
Figure 3
Figure 3. FID of diffusion samplers trained with various combinations of frequency noise. The set￾tings for 𝛾𝑙 = 0.5 yields standard diffusion train￾ing. We conduct experiments to learn the distribution of data where, by construction, the information content lies in the low frequencies. We use the CIFAR-10 dataset, and corrupt the original data with high-frequency noise 𝜖[.3,1.] , thus erasing the high-frequency content while… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Samples from the original data distribution, the degraded data distribution, a standard [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]

Discussion (0). Sign in 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. Unified Audio Intelligence Without Regressing on Text Intelligence

    cs.CL 2026-07 conditional novelty 6.0 of 10

    A unified 30B MoE audio-text LLM achieves state-of-the-art audio understanding, generation, and speech tasks while preserving text reasoning comparable to its text-only backbone.

Reference graph

Works this paper leans on

34 extracted references · 11 canonical work pages · cited by 1 Pith paper

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION format.date year duplicate empty "emp...

  2. [2]

    What regularized auto-encoders learn from the data-generating distribution

    Guillaume Alain and Yoshua Bengio. What regularized auto-encoders learn from the data-generating distribution. The Journal of Machine Learning Research, 15 0 (1): 0 3563--3593, 2014

    2014

  3. [3]

    Li, Hamid Kazemi, Furong Huang, Micah Goldblum, Jonas Geiping, and Tom Goldstein

    Arpit Bansal, Eitan Borgnia, Hong-Min Chu, Jie S. Li, Hamid Kazemi, Furong Huang, Micah Goldblum, Jonas Geiping, and Tom Goldstein. Cold Diffusion : Inverting Arbitrary Image Transforms Without Noise , August 2022. URL http://arxiv.org/abs/2208.09392. arXiv:2208.09392 [cs]

    arXiv 2022

  4. [4]

    On the inductive bias of neural tangent kernels

    Alberto Bietti and Julien Mairal. On the inductive bias of neural tangent kernels. In Advances in Neural Information Processing Systems, 2019

    work page 2019

  5. [5]

    Time series diffusion in the frequency domain, 2024

    Jonathan Crabbé, Nicolas Huynh, Jan Stanczuk, and Mihaela van der Schaar. Time series diffusion in the frequency domain, 2024. URL https://arxiv.org/abs/2402.05933

    arXiv 2024

  6. [6]

    Imagenet-trained cnns are biased towards texture; increasing shape bias improves accuracy and robustness

    Robert Geirhos, Patricia Rubisch, Claudio Michaelis, Matthias Bethge, Felix A Wichmann, and Wieland Brendel. Imagenet-trained cnns are biased towards texture; increasing shape bias improves accuracy and robustness. In International Conference on Learning Representations, 2019

    work page 2019

  7. [7]

    Wichmann

    Robert Geirhos, J \"o rn-Henrik Jacobsen, Claudio Michaelis, Richard Zemel, Wieland Brendel, Matthias Bethge, and Felix A. Wichmann. Shortcut learning in deep neural networks. Nature Machine Intelligence, 2 0 (11): 0 665--673, 2020. doi:10.1038/s42256-020-00257-z. URL https://doi.org/10.1038/s42256-020-00257-z

    doi:10.1038/s42256-020-00257-z 2020

  8. [8]

    Denoising diffusion probabilistic models

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffusion probabilistic models. Advances in neural information processing systems, 33: 0 6840--6851, 2020 a

    work page 2020

Show all 34 references
  1. [9]

    Denoising Diffusion Probabilistic Models , June 2020 b

    Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising Diffusion Probabilistic Models , June 2020 b . URL https://arxiv.org/abs/2006.11239v2

    2020 arXiv

  2. [10]

    Generalization in diffusion models arises from geometry-adaptive harmonic representation

    Zahra Kadkhodaie, Florentin Guth, Eero P Simoncelli, and St \'e phane Mallat. Generalization in diffusion models arises from geometry-adaptive harmonic representation. arXiv preprint arXiv:2310.02557, 2023

    2023 arXiv

  3. [11]

    Learning multiple layers of features from tiny images

    Alex Krizhevsky, Geoffrey Hinton, et al. Learning multiple layers of features from tiny images. 2009

    2009

  4. [12]

    Deep learning face attributes in the wild

    Ziwei Liu, Ping Luo, Xiaogang Wang, and Xiaoou Tang. Deep learning face attributes in the wild. In Proceedings of International Conference on Computer Vision (ICCV), December 2015

    2015

  5. [13]

    dsprites: Disentanglement testing sprites dataset

    Loic Matthey, Irina Higgins, Demis Hassabis, and Alexander Lerchner. dsprites: Disentanglement testing sprites dataset. https://github.com/deepmind/dsprites-dataset/, 2017

    2017

  6. [14]

    Moment matching for multi-source domain adaptation

    Xingchao Peng, Qinxun Bai, Xide Xia, Zijun Huang, Kate Saenko, and Bo Wang. Moment matching for multi-source domain adaptation. In Proceedings of the IEEE/CVF international conference on computer vision, pp.\ 1406--1415, 2019

    2019

  7. [15]

    Boosting diffusion models with moving average sampling in frequency domain

    Yurui Qian, Qi Cai, Yingwei Pan, Yehao Li, Ting Yao, Qibin Sun, and Tao Mei. Boosting diffusion models with moving average sampling in frequency domain. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp.\ 8911--8920, June 2024

    2024

  8. [16]

    On the Spectral Bias of Neural Networks

    Nasim Rahaman, Aristide Baratin, Devansh Arpit, Felix Draxler, Min Lin, Fred Hamprecht, Yoshua Bengio, and Aaron Courville. On the Spectral Bias of Neural Networks . In Proceedings of the 36th International Conference on Machine Learning , pp.\ 5301--5310. PMLR, May 2019. URL ...

  9. [17]

    Diffusion Models With Learned Adaptive Noise , June 2024

    Subham Sekhar Sahoo, Aaron Gokaslan, Chris De Sa, and Volodymyr Kuleshov. Diffusion Models With Learned Adaptive Noise , June 2024. URL http://arxiv.org/abs/2312.13236. arXiv:2312.13236 [cs]

    2024 arXiv

  10. [18]

    Large-scale classification of fine-art paintings: Learning the right metric on the right feature

    Babak Saleh and Ahmed Elgammal. Large-scale classification of fine-art paintings: Learning the right metric on the right feature. arXiv preprint arXiv:1505.00855, 2015

    2015 arXiv

  11. [19]

    Which shortcut cues will dnns choose? a study from the parameter-space perspective

    Luca Scimeca, Seong Joon Oh, Sanghyuk Chun, Michael Poli, and Sangdoo Yun. Which shortcut cues will dnns choose? a study from the parameter-space perspective. In International Conference on Learning Representations, 2021

    2021

  12. [20]

    Leveraging diffusion disentangled representations to mitigate shortcuts in underspecified visual tasks

    Luca Scimeca, Alexander Rubinstein, Armand Nicolicioiu, Damien Teney, and Yoshua Bengio. Leveraging diffusion disentangled representations to mitigate shortcuts in underspecified visual tasks. In NeurIPS 2023 Workshop on Diffusion Models, 2023 a . URL https://openreview.net/fo...

    2023

  13. [21]

    Shortcut bias mitigation via ensemble diversity using diffusion probabilistic models

    Luca Scimeca, Alexander Rubinstein, Damien Teney, Seong Joon Oh, Armand Mihai Nicolicioiu, and Yoshua Bengio. Shortcut bias mitigation via ensemble diversity using diffusion probabilistic models. arXiv preprint arXiv:2311.16176, 2023 b

    2023 arXiv

  14. [22]

    On diffusion models for amortized inference: Benchmarking and improving stochastic control and sampling

    Marcin Sendera, Minsu Kim, Sarthak Mittal, Pablo Lemos, Luca Scimeca, Jarrid Rector-Brooks, Alexandre Adam, Yoshua Bengio, and Nikolay Malkin. On diffusion models for amortized inference: Benchmarking and improving stochastic control and sampling. arXiv preprint arXiv:2402.05098, 2024

    2024 arXiv

  15. [23]

    Weiss, Niru Maheswaranathan, and Surya Ganguli

    Jascha Sohl-Dickstein, Eric A. Weiss, Niru Maheswaranathan, and Surya Ganguli. Deep Unsupervised Learning using Nonequilibrium Thermodynamics , November 2015. URL http://arxiv.org/abs/1503.03585. arXiv:1503.03585 [cond-mat, q-bio, stat]

    2015 arXiv

  16. [24]

    Deep learning and the information bottleneck principle

    Naftali Tishby and Noga Zaslavsky. Deep learning and the information bottleneck principle. In 2015 IEEE Information Theory Workshop (ITW), pp.\ 1--5. IEEE, 2015

    2015

  17. [25]

    Modelling the power spectra of natural images: statistics and information

    van A Van der Schaaf and JH van van Hateren. Modelling the power spectra of natural images: statistics and information. Vision research, 36 0 (17): 0 2759--2770, 1996

    1996

  18. [26]

    Amortizing intractable inference in diffusion models for vision, language, and control

    Siddarth Venkatraman, Moksh Jain, Luca Scimeca, Minsu Kim, Marcin Sendera, Mohsin Hasan, Luke Rowe, Sarthak Mittal, Pablo Lemos, Emmanuel Bengio, et al. Amortizing intractable inference in diffusion models for vision, language, and control. arXiv preprint arXiv:2405.20971, 2024

    2024 arXiv

  19. [27]

    Extracting and composing robust features with denoising autoencoders

    Pascal Vincent, Hugo Larochelle, Yoshua Bengio, and Pierre-Antoine Manzagol. Extracting and composing robust features with denoising autoencoders. In Proceedings of the 25th international conference on Machine learning, pp.\ 1096--1103, 2008

    2008

  20. [28]

    Training behavior of deep neural network in frequency domain, October 2019

    Zhi-Qin John Xu, Yaoyu Zhang, and Yanyang Xiao. Training behavior of deep neural network in frequency domain, October 2019. URL http://arxiv.org/abs/1807.01251. arXiv:1807.01251 [cs, math, stat]

    2019 arXiv

  21. [29]

    Diffusion Models : A Comprehensive Survey of Methods and Applications , June 2024

    Ling Yang, Zhilong Zhang, Yang Song, Shenda Hong, Runsheng Xu, Yue Zhao, Wentao Zhang, Bin Cui, and Ming-Hsuan Yang. Diffusion Models : A Comprehensive Survey of Methods and Applications , June 2024. URL http://arxiv.org/abs/2209.00796. arXiv:2209.00796 [cs]

    2024

  22. [30]

    Spatial-frequency u-net for denoising diffusion probabilistic models, 2023

    Xin Yuan, Linjie Li, Jianfeng Wang, Zhengyuan Yang, Kevin Lin, Zicheng Liu, and Lijuan Wang. Spatial-frequency u-net for denoising diffusion probabilistic models, 2023. URL https://arxiv.org/abs/2307.14648

    2023 arXiv

  23. [31]

    Confronting reward overoptimization for diffusion models: A perspective of inductive and primacy biases

    Ziyi Zhang, Sen Zhang, Yibing Zhan, Yong Luo, Yonggang Wen, and Dacheng Tao. Confronting reward overoptimization for diffusion models: A perspective of inductive and primacy biases. arXiv preprint arXiv:2402.08552, 2024

    2024

  24. [32]

    @esa (Ref

    \@ifxundefined[1] #1\@undefined \@firstoftwo \@secondoftwo \@ifnum[1] #1 \@firstoftwo \@secondoftwo \@ifx[1] #1 \@firstoftwo \@secondoftwo [2] @ #1 \@temptokena #2 #1 @ \@temptokena \@ifclassloaded agu2001 natbib The agu2001 class already includes natbib coding, so you should ...

  25. [33]

    \@lbibitem[] @bibitem@first@sw\@secondoftwo \@lbibitem[#1]#2 \@extra@b@citeb \@ifundefined br@#2\@extra@b@citeb \@namedef br@#2 \@nameuse br@#2\@extra@b@citeb \@ifundefined b@#2\@extra@b@citeb @num @parse #2 @tmp #1 NAT@b@open@#2 NAT@b@shut@#2 \@ifnum @merge>\@ne @bibitem@firs...

  26. [34]

    @open @close @open @close and [1] URL: #1 \@ifundefined chapter * \@mkboth \@ifxundefined @sectionbib * \@mkboth * \@mkboth\@gobbletwo \@ifclassloaded amsart * \@ifclassloaded amsbook * \@ifxundefined @heading @heading NAT@ctr thebibliography [1] @ \@biblabel @NAT@ctr \@bibset...

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.