mlsanityData to 14 Oct 2026

SFBD Flow: A Continuous-Optimization Framework for Training Diffusion Models with Noisy Samples

arXiv 2506.02371 ↗First posted 3 Jun 2025cs.LGcited by 1

On the map: Diffusion models → Diffusion foundations and sampling

Abstract (arXiv)

Diffusion models achieve strong generative performance but often rely on large datasets that may include sensitive content. This challenge is compounded by the models' tendency to memorize training data, raising privacy concerns. SFBD (Lu et al., 2025) addresses this by training on corrupted data and using limited clean samples to capture local structure and improve convergence. However, its iterative denoising and fine-tuning loop requires manual coordination, making it burdensome to implement. We reinterpret SFBD as an alternating projection algorithm and introduce a continuous variant, SFBD flow, that removes the need for alternating steps. We further show its connection to consistency constraint-based methods, and demonstrate that its practical instantiation, Online SFBD, consistently outperforms strong baselines across benchmarks.

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Placed in Diffusion foundations and sampling, one of 14 directions in Diffusion models. Builds on 33 of its 62 references that are in the map. First posted on arXiv 3 Jun 2025.

Builds on

33 of 62 references.

Closest ancestors · 13

Same direction, newest first

  1. 2025Stochastic Forward-Backward Deconvolution: Training Diffusion Models with Finite Noisy Datasets8 Feb
  2. 2024Schr"odinger Bridge Flow for Unpaired Data Translation14 Sep
  3. 2024Consistent Diffusion Meets Tweedie: Training Exact Ambient Diffusion Models with Noisy Data20 Mar
  4. 2023SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis4 Jul
  5. 2023Consistency Models2 Mar
  6. 2023Consistent Diffusion Models: Mitigating Sampling Drift by Learning to be Consistent17 Feb
  7. 2022Diffusion Models in Vision: A Survey10 Sep
  8. 2022Elucidating the Design Space of Diffusion-Based Generative Models1 Jun
  9. 2021High-Resolution Image Synthesis with Latent Diffusion Models20 Dec
  10. 2021Improved Denoising Diffusion Probabilistic Models18 Feb
  11. 2020Score-Based Generative Modeling through Stochastic Differential Equations26 Nov
  12. 2020Denoising Diffusion Implicit Models6 Oct
1 more
  1. 2020Denoising Diffusion Probabilistic Models19 Jun

Other references · 20

  1. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,058
  2. 2022Generative Adversarial Networkscited by 5,627
  3. 2022Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flowcited by 3,700
  4. 2014Deep Learning Face Attributes in the Wildcited by 3,693
  5. 2016Deep Learning with Differential Privacycited by 2,728
  6. 2022Video Diffusion Modelscited by 2,305
  7. 2020DiffWave: A Versatile Diffusion Model for Audio Synthesiscited by 1,354
  8. 2018Demystifying MMD GANscited by 1,293
  9. 2019On the Variance of the Adaptive Learning Rate and Beyondcited by 772
  10. 2023Extracting Training Data from Diffusion Modelscited by 732
  11. 2022Diffusion Art or Digital Forgery? Investigating Data Replication in Diffusion Modelscited by 424
  12. 2022Diffsound: Discrete Diffusion Model for Text-to-sound Generationcited by 257
8 more
  1. 2018Differentially Private Generative Adversarial Networkcited by 235
  2. 2022Diffusion-GAN: Training GANs with Diffusioncited by 205
  3. 2021Solving Schr"odinger Bridges via Maximum Likelihoodcited by 141
  4. 2022Differentially Private Diffusion Modelscited by 119
  5. 2023Ambient Diffusion: Learning Clean Distributions from Corrupted Datacited by 111
  6. 2023Solving Inverse Problems with Score-Based Generative Priors learned from Noisy Datacited by 40
  7. 2024An Expectation-Maximization Algorithm for Training Clean Diffusion Models from Corrupted Observationscited by 20
  8. 2024How much is a noisy image worth? Data Scaling Laws for Ambient Diffusioncited by 18

Built on by

Cited by year

Citing papers per year: 2025: 1.

1 paper.

  1. 2025SFBD-OMNI: Bridge models for lossy measurement restoration with limited clean samplescited by 2

Similar, not cited

Close in wording, no citation either way

  1. 2025Fewer Denoising Steps or Cheaper Per-Step Inference: Towards Compute-Optimal Diffusion Model Deployment
  2. 2024Towards Faster Training of Diffusion Models: An Inspiration of A Consistency Phenomenon
  3. 2022Diffusion models as plug-and-play priors
  4. 2023The Emergence of Reproducibility and Generalizability in Diffusion Models
  5. 2025ETC: training-free diffusion models acceleration with Error-aware Trend Consistency
  6. 2023Patch Diffusion: Faster and More Data-Efficient Training of Diffusion Models
  7. 2025A First-order Generative Bilevel Optimization Framework for Diffusion Models
  8. 2025Diffusion Models: A Mathematical Introduction