Research map / Optimization / Nonconvex and smooth optimization
On the Convergence of Learning-Based Iterative Methods for Nonconvex Inverse Problems
Risheng Liu, Shichao Cheng, Yi He, Xin Fan, Zhouchen Lin, Zhongxuan Luo
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 error-control or explicit-momentum scheduling to learned module plus PG (ablation) | FIMA · Non-blind deconvolution | convergence | Improved |
| add proximal gradient step to learned module without scheduling policy (ablation) | naive A_g o A_f · Non-blind deconvolution | convergence | Hurt |
| use CNN module instead of PG, RF or TV for A_g (ablation) | FIMA · Non-blind deconvolution | PSNR / speed | Improved |
| use TV module instead of PG or RF for A_g (ablation) | FIMA · Non-blind deconvolution | PSNR | Improved |
| vary tau in A_f from 1e-4 to 10 (ablation) | FIMA · Non-blind deconvolution | PSNR | No clear effect |
| use FIMA instead of classical nonconvex APGs (mAPG, APGnc, niAPG) | iFIMA · Non-blind deconvolution (1% noise) | PSNR: 27.24 -> 29.85 | Improved |
| use error-control (iFIMA) instead of explicit momentum (eFIMA) | iFIMA · Non-blind deconvolution (4% noise) | PSNR: 26.20 -> 26.37 | Improved |
| use eFIMA instead of iFIMA for speed | eFIMA · Non-blind deconvolution (1% noise) | time (s): 2.06 -> 1.89 | Improved |
| add CNN module A to mFIMA (ablation) | mFIMA · Blind deconvolution | image/kernel quality | Improved |
| use mFIMA instead of Pan et al. | mFIMA · Blind deconvolution (Levin benchmark) | PSNR: 29.78 -> 30.37 | Improved |
| use FIMA instead of IRCNN | iFIMA · Non-blind deconvolution (1% noise) | PSNR: 29.78 -> 29.85 | Improved |
Closest papers by meaning
- Improving Diffusion-based Inverse Algorithms under Few-Step Constraint via Linear Extrapolation (NeurIPS 2025 · similarity 0.86)
- Integrating Intermediate Layer Optimization and Projected Gradient Descent for Solving Inverse Problems with Diffusion Models (ICML 2025 · similarity 0.85)
- Self-diffusion for Solving Inverse Problems (NeurIPS 2025 · similarity 0.85)
- NPN: Non-Linear Projections of the Null-Space for Imaging Inverse Problems (NeurIPS 2025 · similarity 0.85)
- Beyond First-Order Tweedie: Solving Inverse Problems using Latent Diffusion (CVPR 2024 · similarity 0.84)
- Weakly Convex Regularisers for Inverse Problems: Convergence of Critical Points and Primal-Dual Optimisation (ICML 2024 · similarity 0.84)
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.
