Is Noise Conditioning Necessary for Denoising Generative Models?
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2025
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866917086031249408 |
|---|---|
| author | Sun, Qiao Jiang, Zhicheng Zhao, Hanhong He, Kaiming |
| author_facet | Sun, Qiao Jiang, Zhicheng Zhao, Hanhong He, Kaiming |
| contents | It is widely believed that noise conditioning is indispensable for denoising diffusion models to work successfully. This work challenges this belief. Motivated by research on blind image denoising, we investigate a variety of denoising-based generative models in the absence of noise conditioning. To our surprise, most models exhibit graceful degradation, and in some cases, they even perform better without noise conditioning. We provide a theoretical analysis of the error caused by removing noise conditioning and demonstrate that our analysis aligns with empirical observations. We further introduce a noise-unconditional model that achieves a competitive FID of 2.23 on CIFAR-10, significantly narrowing the gap to leading noise-conditional models. We hope our findings will inspire the community to revisit the foundations and formulations of denoising generative models. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2502_13129 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Is Noise Conditioning Necessary for Denoising Generative Models? Sun, Qiao Jiang, Zhicheng Zhao, Hanhong He, Kaiming Computer Vision and Pattern Recognition It is widely believed that noise conditioning is indispensable for denoising diffusion models to work successfully. This work challenges this belief. Motivated by research on blind image denoising, we investigate a variety of denoising-based generative models in the absence of noise conditioning. To our surprise, most models exhibit graceful degradation, and in some cases, they even perform better without noise conditioning. We provide a theoretical analysis of the error caused by removing noise conditioning and demonstrate that our analysis aligns with empirical observations. We further introduce a noise-unconditional model that achieves a competitive FID of 2.23 on CIFAR-10, significantly narrowing the gap to leading noise-conditional models. We hope our findings will inspire the community to revisit the foundations and formulations of denoising generative models. |
| title | Is Noise Conditioning Necessary for Denoising Generative Models? |
| topic | Computer Vision and Pattern Recognition |
| url | https://arxiv.org/abs/2502.13129 |