Research map / Learning theory and algorithms / Combinatorial and search optimization
Dynamic Sasvi: Strong Safe Screening for Norm-Regularized Least Squares
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 Dynamic Sasvi instead of Gap Safe Sphere | coordinate descent Lasso path · Leukemia, -log epsilon=4 | log acceleration ratio: -0.358 -> -0.468 | Helped |
| use Dynamic Sasvi instead of Gap Safe Dome | coordinate descent Lasso path · Leukemia, -log epsilon=8 | log acceleration ratio: -0.902 -> -0.987 | Helped |
| use Dynamic Sasvi instead of Gap Safe Sphere | coordinate descent Lasso path · 20newsgroup, -log epsilon=4 | log acceleration ratio: -0.293 -> -0.338 | Helped |
| use Dynamic Sasvi instead of Gap Safe Dome | coordinate descent Lasso path · 20newsgroup, -log epsilon=8 | log acceleration ratio: -0.562 -> -0.604 | Helped |
| use Dynamic EDPP (sphere relaxation) instead of Dynamic Sasvi (ablation) | coordinate descent Lasso path · Leukemia, -log epsilon=6 | log acceleration ratio: -0.828 -> -0.838 | Helped |
| use Dynamic EDPP instead of Dynamic Sasvi (ablation) | coordinate descent Lasso path · 20newsgroup, -log epsilon=6 | log acceleration ratio: -0.517 -> -0.532 | Helped |
| use Dynamic EDPP instead of Dynamic Sasvi (ablation) | Lasso screening · Leukemia remaining features | fraction of eliminated features | No clear effect |
| use Dynamic Sasvi instead of Gap Safe Dome and Sphere | cyclic coordinate descent · Leukemia (72 samples, 7128 features) | fraction of features eliminated | Helped |
| tighten solver tolerance from -log epsilon 4 to 8 | Dynamic Sasvi · Leukemia | log acceleration ratio: -0.468 -> -0.987 | Helped |
Closest papers by meaning
- Safe Element Screening for Submodular Function Minimization (ICML 2018 · similarity 0.87)
- Fast OSCAR and OWL Regression via Safe Screening Rules (ICML 2020 · similarity 0.86)
- GAP Safe Screening Rules for Sparse-Group Lasso (NeurIPS 2016 · similarity 0.86)
- Robust Sparse Regression with Non-Isotropic Designs (NeurIPS 2024 · similarity 0.85)
- Regularized least squares learning with heavy-tailed noise is minimax optimal (NeurIPS 2025 · similarity 0.84)
- Optimal Sketching for Residual Error Estimation for Matrix and Vector Norms (ICLR 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.