A Gradient Flow Approach to Solving Inverse Problems with Latent Diffusion Models
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_ | 1866914052532338688 |
|---|---|
| author | Wang, Tim Y. J. Akyildiz, O. Deniz |
| author_facet | Wang, Tim Y. J. Akyildiz, O. Deniz |
| contents | Solving ill-posed inverse problems requires powerful and flexible priors. We propose leveraging pretrained latent diffusion models for this task through a new training-free approach, termed Diffusion-regularized Wasserstein Gradient Flow (DWGF). Specifically, we formulate the posterior sampling problem as a regularized Wasserstein gradient flow of the Kullback-Leibler divergence in the latent space. We demonstrate the performance of our method on standard benchmarks using StableDiffusion (Rombach et al., 2022) as the prior. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2509_19276 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | A Gradient Flow Approach to Solving Inverse Problems with Latent Diffusion Models Wang, Tim Y. J. Akyildiz, O. Deniz Machine Learning Computation Solving ill-posed inverse problems requires powerful and flexible priors. We propose leveraging pretrained latent diffusion models for this task through a new training-free approach, termed Diffusion-regularized Wasserstein Gradient Flow (DWGF). Specifically, we formulate the posterior sampling problem as a regularized Wasserstein gradient flow of the Kullback-Leibler divergence in the latent space. We demonstrate the performance of our method on standard benchmarks using StableDiffusion (Rombach et al., 2022) as the prior. |
| title | A Gradient Flow Approach to Solving Inverse Problems with Latent Diffusion Models |
| topic | Machine Learning Computation |
| url | https://arxiv.org/abs/2509.19276 |