mlsanityData to 7 Oct 2026

The Generation Phases of Flow Matching: a Denoising Perspective

arXiv 2510.24830 ↗First posted 28 Oct 2025cs.CV cs.AI cs.LGcited by 6

On the map: Diffusion models → Flow matching, time series and science

Abstract (arXiv)

Flow matching has achieved remarkable success, yet the factors influencing the quality of its generation process remain poorly understood. In this work, we adopt a denoising perspective and design a framework to empirically probe the generation process. Laying down the formal connections between flow matching models and denoisers, we provide a common ground to compare their performances on generation and denoising. This enables the design of principled and controlled perturbations to influence sample generation: noise and drift. This leads to new insights on the distinct dynamical phases of the generative process, enabling us to precisely characterize at which stage of the generative process denoisers succeed or fail and why this matters.

Key facts. Cited by 6 papers in the map (data through 7 Oct 2026). Placed in Flow matching, time series and science, one of 14 directions in Diffusion models. Builds on 38 of its 49 references that are in the map. First posted on arXiv 28 Oct 2025.

Builds on

38 of 49 references.

Closest ancestors · 7

Same direction, newest first

  1. 2025On the Closed-Form of Flow Matching: Generalization Does Not Arise from Target Stochasticity4 Jun
  2. 2024Simple ReFlow: Improved Techniques for Fast Flow Models10 Oct
  3. 2024Flow Priors for Linear Inverse Problems via Iterative Corrupted Trajectory Matching29 May
  4. 2023Training-free Linear Image Inverses via Flows25 Sep
  5. 2023Improving and generalizing flow-based generative models with minibatch optimal transport1 Feb
  6. 2022Flow Matching for Generative Modeling6 Oct
  7. 2022Flow Straight and Fast: Learning to Generate and Transfer Data with Rectified Flow7 Sep

Other references · 31

  1. 2020Denoising Diffusion Probabilistic Modelscited by 20,851
  2. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,604
  3. 2021High-Resolution Image Synthesis with Latent Diffusion Modelscited by 18,694
  4. 2020Score-Based Generative Modeling through Stochastic Differential Equationscited by 8,964
  5. 2015Deep Unsupervised Learning using Nonequilibrium Thermodynamicscited by 7,005
  6. 2015Rethinking the Inception Architecture for Computer Visioncited by 5,451
  7. 2022Elucidating the Design Space of Diffusion-Based Generative Modelscited by 2,994
  8. 2022Building Normalizing Flows with Stochastic Interpolantscited by 964
  9. 2022Perception Prioritized Training of Diffusion Modelscited by 301
  10. 2023Understanding Diffusion Objectives as the ELBO with Simple Data Augmentationcited by 293
  11. 2023Efficient Diffusion Training via Min-SNR Weighting Strategycited by 242
  12. 2023Exposing flaws of generative model evaluation metrics and their unfair treatment of diffusion modelscited by 199
19 more
  1. 2023Generalization in diffusion models arises from geometry-adaptive harmonic representationscited by 179
  2. 2023Inversion by Direct Iteration: An Alternative to Denoising Diffusion for Image Restorationcited by 164
  3. 2024Dynamical Regimes of Diffusion Modelscited by 133
  4. 2023On Memorization in Diffusion Modelscited by 122
  5. 2017Learning Proximal Operators: Using Denoising Networks for Regularizing Inverse Imaging Problemscited by 107
  6. 2024An analytic theory of creativity in convolutional diffusion modelscited by 103
  7. 2024PnP-Flow: Plug-and-Play Image Restoration with Flow Matchingcited by 92
  8. 2024A Phase Transition in Diffusion Models Reveals the Hierarchical Nature of Datacited by 88
  9. 2025Why Diffusion Models Don't Memorize: The Role of Implicit Dynamical Regularization in Trainingcited by 78
  10. 2015From Facial Parts Responses to Face Detection: A Deep Learning Approachcited by 42
  11. 2024Towards a Mechanistic Explanation of Diffusion Model Generalizationcited by 42
  12. 2022Can Push-forward Generative Models Fit Multimodal Distributions?cited by 41
  13. 2025DeepInverse: A Python package for solving imaging inverse problems with deep learningcited by 27
  14. 2025Bigger Isn't Always Memorizing: Early Stopping Overparameterized Diffusion Modelscited by 24
  15. 2025Generalization through variance: how noise shapes inductive biases in diffusion modelscited by 23
  16. 2023The Emergence of Reproducibility and Generalizability in Diffusion Modelscited by 21
  17. 2025Taking a Big Step: Large Learning Rates in Denoising Score Matching Prevent Memorizationcited by 16
  18. 2023Unmasking Bias in Diffusion Model Trainingcited by 13
  19. 2025Loss Functions in Diffusion Models: A Comparative Studycited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 1; 2026: 5.

6 papers.

  1. 2025On The Hidden Biases of Flow Matching Samplerscited by 2
  2. 2026Solving Room Impulse Response Inverse Problems Using Flow Matching with Analytic Wiener Denoisercited by 2
  3. 2026Distance Marching for Generative Modelingcited by 0
  4. 2026Joint Flow Matching for Generator-Consistent Classificationcited by 0
  5. 2026Weighting Schedules Govern What and When Score-Based Generative Models Learn from Multimodal Datacited by 0
  6. 2026Debias Anything: Fairness with Diversity without Supervision in Diffusion Modelscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024Exploring Diffusion and Flow Matching Under Generator Matching
  2. 2026Hierarchical Discrete Flow Matching for Graph Generation
  3. 2026Training Flow Matching: The Role of Weighting and Parameterization
  4. 2024Elucidating Flow Matching ODE Dynamics with Respect to Data Geometries and Denoisers
  5. 2024Generative Lines Matching Models
  6. 2025Efficiency vs. Fidelity: A Comparative Analysis of Diffusion Probabilistic Models and Flow Matching on Low-Resource Hardware
  7. 2026FastFlow: Accelerating The Generative Flow Matching Models with Bandit Inference
  8. 2026Training-Free Hidden-State Refinement for Flow-Matching Image Generators