Research map / Image generation / Fast diffusion sampling
Diffusion Model Based Signal Recovery Under 1-Bit Quantization
Youming Chen, Zhaoqiang Liu
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 Diff-OneBit (differentiable surrogate likelihood + HQS) instead of SIM-DMIS | 1-bit CS with diffusion prior · FFHQ 1-bit CS (sigma=1.5) | PSNR: 14.32 -> 20.25 | Improved |
| use Diff-OneBit instead of QCS-SGM | 1-bit CS with diffusion prior · FFHQ 1-bit CS (sigma=0.5) | FID: 211.85 -> 85.40 | Improved |
| use 1-bit-specific data fidelity instead of linear/non-linear solvers (DiffPIR, DPS, DAPS) | diffusion plug-and-play · FFHQ 1-bit CS (sigma=0.5) | PSNR: 9.88 -> 22.05 (vs DiffPIR) | Improved |
| use exact tanh likelihood version for logistic regression | Diff-OneBit · FFHQ logistic regression | PSNR: 16.03 -> 19.99 vs SIM-DMIS | Improved |
| increase NFE from 20 to 100 (ablation) | Diff-OneBit · FFHQ 1-bit CS (sigma=0.5) | PSNR: 22.05 -> 22.65 | Improved |
| vary lambda guidance strength (ablation) | Diff-OneBit · FFHQ 1-bit CS (sigma=0.5) | visual quality | No clear effect |
| remove random noise during sampling (vs DiffPIR-style noise) (ablation) | Diff-OneBit · FFHQ 1-bit CS | reconstruction quality | Improved |
Closest papers by meaning
- Learning Single Index Models with Diffusion Priors (ICML 2025 · similarity 0.88)
- OSCAR: One-Step Diffusion Codec Across Multiple Bit-rates (NeurIPS 2025 · similarity 0.88)
- DiffSCI: Zero-Shot Snapshot Compressive Imaging via Iterative Spectral Diffusion Model (CVPR 2024 · similarity 0.88)
- EfficientDM: Efficient Quantization-Aware Fine-Tuning of Low-Bit Diffusion Models (ICLR 2024 · similarity 0.87)
- Recovery Guarantees for Posterior Sampling of One-Bit Compressed Sensing (NeurIPS 2026 · similarity 0.87)
- BiDM: Pushing the Limit of Quantization for Diffusion Models (NeurIPS 2024 · similarity 0.87)
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.
