Research map / Image generation / Fast diffusion sampling
Elucidating the SNR-t Bias of Diffusion Probabilistic Models
Meng Yu, Lei Sun, Jianhao Zeng, Xiangxiang Chu, Kun Zhan
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 differential correction in wavelet domain (DCW) | IDDPM · CIFAR-10 20 steps | FID: 13.19 -> 7.57 | Improved |
| add DCW | IDDPM · CIFAR-10 50 steps | FID: 5.55 -> 4.16 | Improved |
| add DCW | EDM · CIFAR-10 13 NFE | FID: 10.66 -> 5.67 | Improved |
| add DCW | IDDPM · LSUN Bedroom 256 20 steps | FID: 18.69 -> 11.03 | Improved |
| add DCW | ADM-IP (bias-corrected) · CelebA 64 50 steps | FID: 4.52 -> 4.34 | Improved |
| add DCW | EDM-FR · CIFAR-10 35 NFE | FID: 2.13 -> 2.12 | No clear effect |
| add DCW instead of DPM-AT | ADM · CIFAR-10 10 steps | FID: 15.88 -> 13.01 | Improved |
| add DCW instead of DPM-AE | DDIM · CIFAR-10 10 steps | FID: 13.98 -> 9.36 | Improved |
| correct in pixel space only (DC) (ablation) | A-DPM · CIFAR-10 (CS) 10 steps | FID: 22.94 -> 15.71 | Improved |
| add high-frequency correction only (DH) (ablation) | A-DPM · CIFAR-10 (CS) 10 steps | FID: 22.94 -> 16.72 | Improved |
| add low-frequency correction on top of high-frequency (full DCW) (ablation) | A-DPM-DH · CIFAR-10 (CS) 10 steps | FID: 16.72 -> 12.46 | Improved |
| add high-frequency correction on top of low-frequency (full DCW) (ablation) | A-DPM-DL · CIFAR-10 (CS) 50 steps | FID: 5.10 -> 4.06 | Improved |
Closest papers by meaning
- EVODiff: Entropy-aware Variance Optimized Diffusion Inference (NeurIPS 2025 · similarity 0.89)
- Alleviating Exposure Bias in Diffusion Models through Sampling with Shifted Time Steps (ICLR 2024 · similarity 0.89)
- O(d/T) Convergence Theory for Diffusion Probabilistic Models under Minimal Assumptions (ICLR 2025 · similarity 0.89)
- Adaptive Non-Uniform Timestep Sampling for Accelerating Diffusion Model Training (CVPR 2025 · similarity 0.89)
- Multi-Step Denoising Scheduled Sampling: Towards Alleviating Exposure Bias for Diffusion Models (AAAI 2024 · similarity 0.88)
- Elucidating the Exposure Bias in Diffusion Models (ICLR 2024 · similarity 0.88)
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.
