Research map / Image generation / Fast diffusion sampling
Debias Coarsely, Sample Conditionally: Statistical Downscaling through Optimal Transport and Probabilistic Diffusion Models
Zhong Yi Wan, Ricardo Baptista, Anudhyan Boral, Yi-Fan Chen, John Anderson, Fei Sha, Leonardo Zepeda-Nunez
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 |
|---|---|---|---|
| add OT debiasing of low-res input before conditional diffusion sampling (ablation) | conditional diffusion downscaler · NS 8x | KLD: 73.16 -> 1.40 | Improved |
| add OT debiasing before conditional diffusion (ablation) | conditional diffusion downscaler · NS 16x | KLD: 93.87 -> 0.83 | Improved |
| add OT debiasing before conditional diffusion (ablation) | conditional diffusion downscaler · KS conditional sampling | KLD: 10.37 -> 1.27 | Improved |
| apply OT map to biased low-fidelity low-res snapshots | LFLR snapshots · NS 8x | MELRu: 1.254 -> 0.013 | Improved |
| replace deterministic ViT upsampler with conditional diffusion upsampler after OT | OT debiased pipeline · NS 16x | MELRu: 1.38 -> 0.05 | Improved |
| replace cubic interpolation with conditional diffusion upsampler after OT | OT debiased pipeline · NS 8x | MELRu: 0.52 -> 0.06 | Improved |
| two-stage OT+diffusion vs end-to-end cycleGAN | downscaling · NS 16x | MELRu: 1.14 -> 0.05 | Improved |
| OT+cDfn vs cycGAN | downscaling · NS 8x | MELRu: 0.08 -> 0.06 | Improved |
| OT+cDfn vs BCSD | downscaling · NS 8x | Wass1: 0.23 -> 0.26 | Hurt |
Closest papers by meaning
- Provable Posterior Sampling with Denoising Oracles via Tilted Transport (NeurIPS 2024 · similarity 0.86)
- Predicting partially observable dynamical systems via diffusion models with a multiscale inference scheme (NeurIPS 2025 · similarity 0.86)
- Flow-Based Conformal Predictive Distributions (NeurIPS 2026 · similarity 0.86)
- Precipitation Downscaling with Spatiotemporal Video Diffusion (NeurIPS 2024 · similarity 0.86)
- Heavy-Tailed Diffusion Models (ICLR 2025 · similarity 0.85)
- Systematic Hazard Sampling: Minimal-Variance Inference for Discrete Diffusion and Flow Models (NeurIPS 2026 · similarity 0.85)
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.
