Research map / Learning theory and algorithms / Matrix and tensor methods
Gradient Step Denoiser for convergent Plug-and-Play
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 the gradient-step conservative-field parameterization rather than unconstrained DRUNet light | DRUNet light denoiser · CBSD68 center-cropped, σ=5, 15, 25, 50/255 | PSNR: 40.19 -> 40.26; 33.89 -> 33.90; 31.25 -> 31.26; 28.00 -> 28.01 | Mixed |
| Use GS-PnP instead of DPIR | PnP image deblurring · CBSD10, ν=0.01, 0.03, 0.05, average over kernels | PSNR: 32.50 -> 32.33; 28.91 -> 28.90; 27.42 -> 27.44 | Mixed |
| Use GS-PnP instead of DPIR | PnP image super-resolution · CBSD10, s=2, isotropic-kernel average | PSNR: 26.74 -> 26.86 | Helped |
| Use GS-PnP instead of DPIR | PnP image super-resolution · CBSD10, s=3, anisotropic-kernel average | PSNR: 25.48 -> 25.52 | Helped |
| Parameterize the denoiser as an exact gradient step on an explicit regularizer | PnP-HQS image restoration · Deblurring, super-resolution, and inpainting | convergence | Helped |
| Use a differentiable DRUNet architecture with ELU activations and two residual blocks per scale | GS-DRUNet · CBSD68 denoising | runtime: 69.8 ms; 10.4 ms | Helped |
Closest papers by meaning
- Plug-and-Play image restoration with Stochastic deNOising REgularization (ICML 2024 · similarity 0.91)
- Equivariant Plug-and-Play Image Reconstruction (CVPR 2024 · similarity 0.89)
- Learning Cocoercive Conservative Denoisers via Helmholtz Decomposition for Poisson Imaging Inverse Problems (NeurIPS 2025 · similarity 0.88)
- Provably Robust Score-Based Diffusion Posterior Sampling for Plug-and-Play Image Reconstruction (NeurIPS 2024 · similarity 0.87)
- Reconciling Stochastic and Deterministic Strategies for Zero-shot Image Restoration using Diffusion Model in Dual (CVPR 2025 · similarity 0.87)
- MAP Estimation with Denoisers: Convergence Rates and Guarantees (NeurIPS 2025 · 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.