Efficient Diffusion Training through Parallelization with Truncated Karhunen-Loève Expansion
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2025
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866918075715026944 |
|---|---|
| author | Ren, Yumeng Liu, Yaofang Artola, Aitor Mertz, Laurent Chan, Raymond H. Morel, Jean-michel |
| author_facet | Ren, Yumeng Liu, Yaofang Artola, Aitor Mertz, Laurent Chan, Raymond H. Morel, Jean-michel |
| contents | Diffusion denoising models have become a popular approach for image generation, but they often suffer from slow convergence during training. In this paper, we identify that this slow convergence is partly due to the complexity of the Brownian motion driving the forward-time process. To address this, we represent the Brownian motion using the Karhunen-Loève expansion, truncating it to a limited number of eigenfunctions. We propose a novel ordinary differential equation with augmented random initials, termed KL diffusion, as a new forward-time process for training and sampling. By developing an appropriate denoising loss function, we facilitate the integration of our KL-diffusion into existing denoising-based models. Using the widely adopted DDIM framework as our baseline ensures a fair comparison, as our modifications focus solely on the forward process and loss function, leaving the network architecture and sampling methods unchanged. Our method significantly outperforms baseline diffusion models, achieving convergence speeds that are twice faster to reach the best FID score of the baseline and ultimately yielding much lower FID scores. Notably, our approach allows for highly parallelized computation, requires no additional learnable parameters, and can be flexibly integrated into existing diffusion methods. The code will be made publicly available. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_17657 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Efficient Diffusion Training through Parallelization with Truncated Karhunen-Loève Expansion Ren, Yumeng Liu, Yaofang Artola, Aitor Mertz, Laurent Chan, Raymond H. Morel, Jean-michel Computer Vision and Pattern Recognition I.2.0; I.4.0 Diffusion denoising models have become a popular approach for image generation, but they often suffer from slow convergence during training. In this paper, we identify that this slow convergence is partly due to the complexity of the Brownian motion driving the forward-time process. To address this, we represent the Brownian motion using the Karhunen-Loève expansion, truncating it to a limited number of eigenfunctions. We propose a novel ordinary differential equation with augmented random initials, termed KL diffusion, as a new forward-time process for training and sampling. By developing an appropriate denoising loss function, we facilitate the integration of our KL-diffusion into existing denoising-based models. Using the widely adopted DDIM framework as our baseline ensures a fair comparison, as our modifications focus solely on the forward process and loss function, leaving the network architecture and sampling methods unchanged. Our method significantly outperforms baseline diffusion models, achieving convergence speeds that are twice faster to reach the best FID score of the baseline and ultimately yielding much lower FID scores. Notably, our approach allows for highly parallelized computation, requires no additional learnable parameters, and can be flexibly integrated into existing diffusion methods. The code will be made publicly available. |
| title | Efficient Diffusion Training through Parallelization with Truncated Karhunen-Loève Expansion |
| topic | Computer Vision and Pattern Recognition I.2.0; I.4.0 |
| url | https://arxiv.org/abs/2503.17657 |