Comparer des méthodes
Examinez les méthodes sélectionnées côte à côte ; les lignes qui diffèrent sont mises en évidence.
| Apprentissage par transfert avec modèle de diffusion× | Modèle de diffusion affiné× | |
|---|---|---|
| Domaine | Apprentissage profond | Apprentissage profond |
| Famille | Machine learning | Machine learning |
| Année d'origine | 2020–2023 | 2020–2023 |
| Auteur d'origine≠ | Ho et al. (DDPM); transfer application popularized by Rombach et al. (Stable Diffusion) and Ruiz et al. (DreamBooth), 2020–2023 | Ho, J., Jain, A., Abbeel, P. (base DDPM); Ruiz et al. (DreamBooth fine-tuning paradigm) |
| Type≠ | Generative model with transfer learning | Generative model (fine-tuned via subject-specific or domain-specific data) |
| Source fondatrice≠ | Ho, J., Jain, A., & Abbeel, P. (2020). Denoising Diffusion Probabilistic Models. Advances in Neural Information Processing Systems (NeurIPS), 33, 6840–6851. link ↗ | Ruiz, N., Li, Y., Jampani, V., Pritch, Y., Rubinstein, M., & Aberman, K. (2023). DreamBooth: Fine Tuning Text-to-Image Diffusion Models for Subject-Driven Generation. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 22500–22510. DOI ↗ |
| Alias | diffusion model fine-tuning, pre-trained diffusion transfer, TL-DM, domain-adapted diffusion model | DDPM fine-tuning, diffusion model adaptation, personalized diffusion model, subject-driven diffusion fine-tuning |
| Apparentées | 5 | 5 |
| Résumé≠ | Transfer Learning with Diffusion Models adapts a large pre-trained diffusion model — such as Stable Diffusion or DALL-E 2 — to a new target domain or task by continuing training on a smaller domain-specific dataset. Rather than learning the full generative process from scratch, practitioners leverage knowledge already encoded in millions of training steps to achieve high-quality domain-adapted generation with modest data and compute. | A fine-tuned diffusion model adapts a large pretrained denoising diffusion model — such as Stable Diffusion or DALL-E — to a specific subject, style, or domain by continuing training on a small curated dataset. Techniques such as DreamBooth, textual inversion, and LoRA make this adaptation feasible on consumer hardware while preserving general generative capability. |
| ScholarGateJeu de données ↗ |
|
|