Research map / Learning theory and algorithms / Optimal transport
Optimal Transport for Unsupervised Denoising Learning
Wei Wang, Fei Wen, Zeyu Yan, Peilin Liu
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 the minimum-distance transport term to adversarial training (1/lambda 0 to 30) (ablation) | OT denoising · Gaussian sigma=15, KODAK24 | PSNR: 29.97 -> 33.99 | Improved |
| raise 1/lambda from 1 to 10 (ablation) | OT denoising · Gaussian sigma=15, KODAK24 | PSNR: 31.44 -> 32.85 | Improved |
| raise 1/lambda from 30 to 100 (ablation) | OT denoising · Gaussian sigma=15, KODAK24 | PSNR: 33.99 -> 30.97 | Hurt |
| unsupervised OT criterion instead of supervised N2C on spatially correlated noise | OT denoising · Brown Gaussian noise, KODAK24 | PSNR: 31.86 -> 32.08 | Improved |
| unsupervised OT criterion instead of supervised N2C on Gaussian noise | OT denoising · Gaussian sigma=15, KODAK24 | PSNR: 34.92 -> 33.99 | Hurt |
| unsupervised OT criterion instead of BM3D | OT denoising · Brown Gaussian noise, KODAK24 | PSNR: 29.20 -> 32.08 | Improved |
| unsupervised OT criterion instead of N2C on perceptual quality | OT denoising · Gaussian sigma=15, KODAK24 | PI: 2.41 -> 2.06 | Improved |
| unsupervised OT criterion instead of N2C on depth noise sigma=50 | OT denoising · Middlebury depth sigma=50 | PSNR: 34.88 -> 34.43 | Hurt |
| unsupervised OT criterion instead of N2C on microscopy | OT denoising · FMD microscopy | PSNR: 35.28 -> 35.73 | Improved |
| use clean images from another microscopy dataset instead of averaged images (ablation) | OT denoising · FMD microscopy | PSNR: 35.73 -> 34.49 | Hurt |
| unsupervised OT criterion instead of CGAN on microscopy | OT denoising · FMD microscopy | PSNR: 31.79 -> 35.73 | Improved |
Closest papers by meaning
- Evaluating Unsupervised Denoising Requires Unsupervised Metrics (ICML 2023 · similarity 0.89)
- Residual-Conditioned Optimal Transport: Towards Structure-Preserving Unpaired and Paired Image Restoration (ICML 2024 · similarity 0.87)
- Resfusion: Denoising Diffusion Probabilistic Models for Image Restoration Based on Prior Residual Noise (NeurIPS 2024 · similarity 0.87)
- Soft Mixture Denoising: Beyond the Expressive Bottleneck of Diffusion Models (ICLR 2024 · similarity 0.87)
- Provable Posterior Sampling with Denoising Oracles via Tilted Transport (NeurIPS 2024 · similarity 0.87)
- Zero-shot Denoising via Neural Compression: Theoretical and algorithmic framework (NeurIPS 2025 · similarity 0.86)
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.
