Research map / Learning theory and algorithms / Matrix and tensor methods
Gap Safe Screening Rules for Sparsity Enforcing Penalties
Eugene Ndiaye, Olivier Fercoq, Alexandre Gramfort, Joseph Salmon
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 |
|---|---|---|---|
| use dynamic Gap Safe screening instead of sequential screening | coordinate descent for l1 logistic regression · Leukemia, 100 lambda values | running time speed-up: up to 30x vs sequential rule, up to 50x vs no screening | Improved |
| add dynamic Gap Safe screening to Lasso solver | coordinate descent Lasso · Leukemia and E2006-log1p | running time | Improved |
| run more iterations before screening (ablation) | dynamic Gap Safe rule · Lasso on Leukemia (2 to 29 iterations) | fraction of active variables | Improved |
| use a coarse lambda grid (10 values) with strong rules (ablation) | strong rules · Lasso on Leukemia | screening usefulness | Hurt |
| use active warm start with the safe active set instead of the strong active set on very coarse grids (ablation) | Gap Safe with warm start · Lasso on Leukemia, 10 lambdas | running time | Improved |
| add active warm start on coarse and dense grids (ablation) | Gap Safe · Lasso on E2006-log1p, grid to lambda_max/20 | running time | Improved |
| use Gap Safe instead of dynamic safe rule of Bonnefoy et al. | multi-task Lasso · MEG/EEG (n=360, p=22494, q=20) | fraction of active variables and computation time | Improved |
| use dynamic Gap Safe with active warm start | Sparse-Group Lasso · NCEP/NCAR climate data (n=814, p=73577) | time to convergence | Improved |
| compare Gap Safe with Blitz algorithm | Lasso · E2006-log1p | running time | No clear effect |
Closest papers by meaning
- GAP Safe Screening Rules for Sparse-Group Lasso (NeurIPS 2016 · similarity 0.92)
- Expanding Boundaries of Gap Safe Screening (JMLR 2021 · similarity 0.89)
- Screening rules for Lasso with non-convex Sparse Regularizers (ICML 2019 · similarity 0.87)
- Safe RuleFit: Learning Optimal Sparse Rule Model by Meta Safe Screening (TPAMI 2022 · similarity 0.87)
- Safe screening rules for L0-regression from Perspective Relaxations (ICML 2020 · similarity 0.86)
- Celer: a Fast Solver for the Lasso with Dual Extrapolation (ICML 2018 · similarity 0.86)
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.
