Research map / Optimization / Nonconvex and smooth optimization
Efficient Algorithms for Non-convex Isotonic Regression through Submodular Optimization
Francis Bach
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 the globally optimized non-convex logarithmic loss instead of the square loss | Isotonic regression with chain constraints · Corrupted decreasing-line data | robustness to outliers | Improved |
| Use the absolute loss instead of the square loss | Isotonic regression with chain constraints · Corrupted decreasing-line data | robustness to outliers | Improved |
| Optimize the non-convex logarithmic loss globally rather than with majorization-minimization | Isotonic regression with chain constraints · Corrupted decreasing-line data | solution quality | Improved |
| Increase corruption from 75% to 90% | Globally optimized non-convex isotonic regression · Corrupted decreasing-line data | robustness to outliers | Hurt |
| Add a smoothness penalty to the isotonic objective (ablation) | Non-separable isotonic regression · n=200 | estimation quality | Improved |
| Use first-order Taylor discretization instead of naive order-0 discretization (ablation) | Non-separable isotonic optimization · n=25; k=32 and k=100; 50 subgradient steps | distance to optimality | No clear effect |
| Use second-order Taylor discretization instead of naive order-0 discretization (ablation) | Non-separable isotonic optimization · n=25; k=32 and k=100; 50 subgradient steps | distance to optimality | Improved |
| Use quadratic rather than absolute loss (ablation) | Separable isotonic regression · Discretization error across k | discretization error | Mixed |
| Use a logarithmic loss with larger κ (ablation) | Separable isotonic regression · Discretization error across k | discretization behavior | Improved |
Closest papers by meaning
- An Efficient Pruning Algorithm for Robust Isotonic Regression (NeurIPS 2018 · similarity 0.91)
- Sparse High-Dimensional Isotonic Regression (NeurIPS 2019 · similarity 0.90)
- Isotonic Modeling with Non-Differentiable Loss Functions with Application to Lasso Regularization (TPAMI 2016 · similarity 0.89)
- Private Isotonic Regression (NeurIPS 2022 · similarity 0.89)
- Optimal approximation for unconstrained non-submodular minimization (ICML 2020 · similarity 0.88)
- From Linear to Linearizable Optimization: A Novel Framework with Applications to Stationary and Non-stationary DR-submodular Optimization (NeurIPS 2024 · 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.
