Research map / Optimization / Nonconvex and smooth optimization
Scalable Approximations for Generalized Linear Problems
Murat Erdogdu, Mohsen Bayati, Lee H. Dicker
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 SLS (OLS then scaling) instead of Newton-Raphson MLE | logistic regression, random init · Logistic synthetic | time to min test error (s): 301.06 -> 8.34 | Improved |
| use SLS instead of gradient descent | Higgs logistic, random init · Higgs | time to min test error (s): 100871 -> 13.18 | Improved |
| use SLS instead of Newton-Raphson | Covertype Poisson, random init · Covertype | time to min test error (s): 16.7 -> 2.71 | Improved |
| use SLS instead of BFGS | Covertype Poisson, random init · Covertype | time to min test error (s): 5.12 -> 2.71 | Improved |
| use SLS instead of Newton-Stein | Covertype Poisson, OLS init · Covertype | time to min test error (s): 282.1 -> 1.66 | Improved |
| initialize SLS with OLS instead of random start (ablation) | SLS, logistic synthetic · Logistic synthetic | time to min test error (s): 8.34 -> 2.94 | Improved |
| initialize Newton-Raphson with OLS instead of random start (ablation) | Newton-Raphson, logistic synthetic · Logistic synthetic | time to min test error (s): 301.06 -> 82.57 | Improved |
| initialize Newton-Raphson with OLS instead of random start (ablation) | Newton-Raphson, Covertype Poisson · Covertype | time to min test error (s): 16.7 -> 32.48 | Hurt |
| initialize Newton-Stein with OLS instead of random start (ablation) | Newton-Stein, Covertype Poisson · Covertype | time to min test error (s): 21.17 -> 282.1 | Hurt |
| initialize SLS with OLS instead of random start (ablation) | SLS, Covertype Poisson · Covertype | time to min test error (s): 2.71 -> 1.66 | Improved |
| use SLS instead of exact MLE | Higgs logistic · Higgs | test error: 22.40% vs 22.38% | No clear effect |
Closest papers by meaning
- On the Local Minima of the Empirical Risk (NeurIPS 2018 · similarity 0.89)
- First-Order Adaptive Sample Size Methods to Reduce Complexity of Empirical Risk Minimization (NeurIPS 2017 · similarity 0.88)
- The Common-directions Method for Regularized Empirical Risk Minimization (JMLR 2019 · similarity 0.88)
- Exploiting Local Convergence of Quasi-Newton Methods Globally: Adaptive Sample Size Approach (NeurIPS 2021 · similarity 0.87)
- On the Efficiency of ERM in Feature Learning (NeurIPS 2024 · similarity 0.87)
- Gradient Descent Converges Arbitrarily Fast for Logistic Regression via Large and Adaptive Stepsizes (ICML 2025 · similarity 0.87)
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.
