The Generation Phases of Flow Matching: a Denoising Perspective
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912775401373696 |
|---|---|
| author | Gagneux, Anne Martin, Ségolène Gribonval, Rémi Massias, Mathurin |
| author_facet | Gagneux, Anne Martin, Ségolène Gribonval, Rémi Massias, Mathurin |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_24830 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | The Generation Phases of Flow Matching: a Denoising Perspective Gagneux, Anne Martin, Ségolène Gribonval, Rémi Massias, Mathurin Computer Vision and Pattern Recognition Artificial Intelligence Machine Learning 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. |
| title | The Generation Phases of Flow Matching: a Denoising Perspective |
| topic | Computer Vision and Pattern Recognition Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2510.24830 |