Research map / Optimization / Optimizers
Improved Imaging by Invex Regularizers with Global Optima Guarantees
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 |
|---|---|---|---|
| replace l1-norm regularizer with invex Eq. (6) (p=0.5) | FISTA-style Algorithm 1 · Deconvolution CS, noiseless | PSNR: 29.97 -> 33.40 | Helped |
| replace l1-norm with invex Eq. (6) at 20 dB | Algorithm 1 · Deconvolution CS, SNR 20 dB | PSNR: 21.80 -> 24.60 | Helped |
| replace l1-norm with invex Eq. (6) at 30 dB | Algorithm 1 · Deconvolution CS, SNR 30 dB | PSNR: 24.91 -> 27.61 | Helped |
| use invex Eq. (9) instead of Eq. (6) | Algorithm 1 · Deconvolution CS, noiseless | PSNR: 33.40 -> 30.00 | Hurt |
| invex Eq. (6) vs nonconvex SCAD | Algorithm 1 · Deconvolution CS, noiseless | PSNR: 30.55 -> 33.40 | Helped |
| invex Eq. (6) vs nonconvex MCP | Algorithm 1 · Deconvolution CS, noiseless | PSNR: 31.30 -> 33.40 | Helped |
| invex Eq. (6) vs ReconNet (non-iterative CNN) | Algorithm 1 · Deconvolution CS, noiseless | PSNR: 27.01 -> 33.40 | Helped |
| add plug-and-play Noise2Void prior with invex Eq. (6) (p=0.8) replacing l1-norm | Algorithm 3 · Plug-and-play deconvolution, noiseless | PSNR: 31.10 -> 34.51 | Helped |
| plug-and-play with Eq. (6) vs l1-norm at 20 dB | Algorithm 3 · Plug-and-play deconvolution, SNR 20 dB | PSNR: 22.95 -> 25.55 | Helped |
| unrolled LISTA with Eq. (6) (p=0.85) instead of l1-norm | Algorithm 2 · CS noiseless, m/n=0.2 | PSNR: 27.95 -> 31.32 | Helped |
| increase measurement ratio m/n from 0.2 to 0.6 with Eq. (6) | unfolded LISTA · CS noiseless | PSNR: 31.32 -> 41.27 | Helped |
| unrolled ISTA-Net with Eq. (6) instead of l1-norm | Algorithm 2 · CS noiseless, m/n=0.6 | PSNR: 38.02 -> 43.61 | Helped |
Closest papers by meaning
- Global Optimality for Non-linear Constrained Restoration Problems via Invexity (ICLR 2024 · similarity 0.89)
- Learning Cocoercive Conservative Denoisers via Helmholtz Decomposition for Poisson Imaging Inverse Problems (NeurIPS 2025 · similarity 0.85)
- Weakly Convex Regularisers for Inverse Problems: Convergence of Critical Points and Primal-Dual Optimisation (ICML 2024 · similarity 0.85)
- A Variational Perspective on Solving Inverse Problems with Diffusion Models (ICLR 2024 · similarity 0.84)
- DEALing with Image Reconstruction: Deep Attentive Least Squares (ICML 2025 · similarity 0.84)
- The Perception-Robustness Tradeoff in Deterministic Image Restoration (ICML 2024 · similarity 0.83)
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.