Research map / Learning theory and algorithms / Matrix and tensor methods
Sparse High-Dimensional Isotonic Regression
David Gamarnik, Julia Gaudio
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 sparse isotonic regression rather than k-NN or polynomial-kernel SVM | COSMIC cancer classification · COSMIC cancer classification | classification accuracy (%): 95.7 vs 78.6; 95.7 vs 77.4 | Improved |
| Use TSIR + S-LPSR instead of IPIR | COSMIC cancer classification · COSMIC cancer classification, n=200, s=3, max interpolation | classification accuracy (%): 96.4 -> 95.9 | Hurt |
| Use s=3 rather than a lower or higher sparsity level (ablation) | Sparse isotonic regression · COSMIC cancer classification | classification accuracy (%): 95.7 | Mixed |
| Use max rather than min interpolation (ablation) | IPIR and TSIR + S-LPSR · COSMIC cancer classification, n=200, s=3 | classification accuracy (%): IPIR: 84.3 -> 96.4; TSIR + S-LPSR: 86.7 -> 95.9 | Improved |
| Select active coordinates sequentially rather than all at once (ablation) | Linear-program support recovery · Synthetic support recovery, n=250, d=20 | successful recovery (%): 45 -> 70 | Improved |
| Select active coordinates sequentially rather than all at once (ablation) | Linear-program support recovery · Synthetic support recovery, n=250, d=50 | successful recovery (%): 0 -> 55 | Improved |
| Increase training samples from n=100 to n=200 (ablation) | IPIR · Synthetic support recovery, d=20 | successful recovery (%): 70 -> 90 | Improved |
| Increase training samples from n=100 to n=200 (ablation) | S-LPSR · Synthetic support recovery, d=50 | successful recovery (%): 20 -> 40 | Improved |
Closest papers by meaning
- Efficient Algorithms for Non-convex Isotonic Regression through Submodular Optimization (NeurIPS 2018 · similarity 0.90)
- Isotonic Modeling with Non-Differentiable Loss Functions with Application to Lasso Regularization (TPAMI 2016 · similarity 0.88)
- Robust Sparse Regression with Non-Isotropic Designs (NeurIPS 2024 · similarity 0.85)
- Solving Sparse \& High-Dimensional-Output Regression via Compression (NeurIPS 2024 · similarity 0.85)
- Non-Asymptotic Uncertainty Quantification in High-Dimensional Learning (NeurIPS 2024 · similarity 0.85)
- Beyond Worst-Case Dimensionality Reduction for Sparse Vectors (ICLR 2025 · 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.
