Research map / Learning theory and algorithms / Prediction and decision losses
Using Benson’s Algorithm for Regularization Parameter Tracking
Joachim Giesen, Sӧren Laue, Andreas Lӧhne, Christopher Schneider
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 Benson's algorithm instead of fine-mesh Grid Search | Elastic Net tuning · 8 microarray datasets | # optimization problems: 5151 -> 26 (ALLAML) | Improved |
| use Benson's algorithm instead of fine-mesh Grid Search | Elastic Net tuning · 8 microarray datasets | minimum test MSE: 0.0530 -> 0.0544 (ALLAML) | Mixed |
| use Benson's algorithm instead of Solution Gamut | Elastic Net tuning · 8 microarray datasets | # optimization problems: 1141 -> 26 (ALLAML) | Improved |
| use Benson's algorithm instead of Solution Gamut | Elastic Net tuning · SMK-CAN-187 | # optimization problems: 128 -> 51 | Improved |
| use Benson's algorithm instead of Random Search | Elastic Net tuning · 8 microarray datasets | # optimization problems: 2736 -> 39 (Prostate-GE) | Improved |
| use Benson's algorithm instead of Random Search | Elastic Net tuning · 14cancer | # optimization problems: 61 -> 68 | Hurt |
| use Benson's algorithm instead of Random Search | Elastic Net tuning · arcene | minimum test MSE: 0.7126 -> 0.6460 | Improved |
| use Benson's algorithm instead of Random Search | Elastic Net tuning · 14cancer | good predictor regions found: 3 areas vs 1 area | Improved |
| tighten approximation to larger epsilon (0.1 vs 1) by dataset scale | Benson Algorithm · Elastic Net | complexity in 1/epsilon | No clear effect |
| apply Benson's algorithm to Lasso and Ridge (q=1) | Benson Algorithm · Lasso/Ridge | complexity in 1/sqrt(eps) | No clear effect |
Closest papers by meaning
- Approximate Leave-One-Out for Fast Parameter Tuning in High Dimensions (ICML 2018 · similarity 0.85)
- Safe Grid Search with Optimal Complexity (ICML 2019 · similarity 0.84)
- Provably tuning the ElasticNet across instances (NeurIPS 2022 · similarity 0.84)
- On Optimal Generalizability in Parametric Learning (NeurIPS 2017 · similarity 0.84)
- Improving Adaptivity via Over-Parameterization in Sequence Models (NeurIPS 2024 · similarity 0.82)
- How Over-Parameterization Slows Down Gradient Descent in Matrix Sensing: The Curses of Symmetry and Initialization (ICLR 2024 · similarity 0.82)
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.
