Research map / Optimization / Optimizers
Safeguarded Learned Convex Optimization
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 |
|---|---|---|---|
| Add the Safe-L2O safeguard and fallback updates to ALISTA | ALISTA · LASSO, unseen data distribution | convergence | Helped |
| Add the Safe-L2O safeguard and fallback updates to LISTA | LISTA-CP · Natural image denoising, pepper-and-salt noise | convergence | Helped |
| Add the Safe-L2O safeguard and fallback updates to LiADMM | LiADMM · Sparse coding with an ℓ1-ℓ1 objective | convergence | Helped |
| Use a trained L2O algorithm instead of the corresponding analytic optimization algorithm | ALISTA, LISTA, AdaLISTA, and LiADMM · Data from the training distribution | computational cost: at least one order of magnitude | Helped |
| Use the EMA safeguard sequence | Safe-L2O · Safeguard parameter selection | practicality | Helped |
| Apply the safeguard intermittently rather than requiring every L2O update to improve | Safe-L2O · ALISTA, LASSO | safeguard activation frequency: only when $k=2$, $k=7$, and $k=12$ | Helped |
Closest papers by meaning
- A Generalization Result for Convergence in Learning-to-Optimize (ICML 2025 · similarity 0.88)
- Towards Robust Learning to Optimize with Theoretical Guarantees (CVPR 2024 · similarity 0.87)
- Learning Provably Improves the Convergence of Gradient Descent (NeurIPS 2025 · similarity 0.86)
- Distributionally Robust Optimization with Bias and Variance Reduction (ICLR 2024 · similarity 0.85)
- Adaptive and Optimal Second-order Optimistic Methods for Minimax Optimization (NeurIPS 2024 · similarity 0.85)
- On Traceability in $\ell_p$ Stochastic Convex Optimization (NeurIPS 2025 · 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.