Research map / Image generation / Fast diffusion sampling
Score-Based Diffusion Models in Function Space
Jae Hyun Lim, Nikola B. Kovachki, Ricardo Baptista, Christopher Beckham, Kamyar Azizzadenesheli, Jean Kossaifi, Vikram Voleti, Jiaming Song, Karsten Kreis, Jan Kautz, Christopher Pal, Arash Vahdat and 1 more
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use DDO instead of GANO | function-space generative model · MNIST-SDF (256x256) | FID: 23.89 -> 17.76 | Improved |
| use DDO instead of GANO | function-space generative model · MNIST-SDF (64x64) | precision: 0.75 -> 0.73 | Hurt |
| use DDO instead of GANO | function-space generative model · MNIST-SDF (256x256) | recall: 0.32 -> 0.39 | Improved |
| use DDO instead of MultilevelDiff | function-space generative model · MNIST-SDF (256x256) | FID: 365.90 -> 17.76 | Improved |
| use DDO instead of GANO | function-space generative model · Darcy flow inverse problem (256x256) | relative error of sample mean: 0.34 -> 0.28 | Improved |
| use MultilevelDiff instead of DDO | function-space generative model · Darcy flow inverse problem (256x256) | relative error of sample mean: 0.28 -> 0.23 | Improved |
| use DDO instead of GANO | function-space generative model · Darcy flow inverse problem (64x64) | relative error of sample variance: 0.75 -> 0.69 | Improved |
| use DDO instead of GANO | function-space generative model · Darcy flow inverse problem (256x256) | relative error of sample variance: 0.77 -> 0.79 | No clear effect |
| add trace-class noise instead of white noise (ablation) | DDO with FNO · Gaussian mixture | uniform-norm error in average spectra | Improved |
| add upsampling to 64x64 during training instead of truncating noise modes (ablation) | DDO · MNIST-SDF | sample quality (qualitative) | Improved |
Closest papers by meaning
- O(d/T) Convergence Theory for Diffusion Probabilistic Models under Minimal Assumptions (ICLR 2025 · similarity 0.91)
- Adapting to Unknown Low-Dimensional Structures in Score-Based Diffusion Models (NeurIPS 2024 · similarity 0.91)
- Training Diffusion-based Generative Models with Limited Data (ICML 2025 · similarity 0.91)
- Assessing the quality of denoising diffusion models in Wasserstein distance: noisy score and optimal bounds (NeurIPS 2025 · similarity 0.91)
- Towards Non-Asymptotic Convergence for Diffusion-Based Generative Models (ICLR 2024 · similarity 0.90)
- Distributional Diffusion Models with Scoring Rules (ICML 2025 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
