Research map / Learning theory and algorithms / Prediction and decision losses
Conformalization of Sparse Generalized Linear Models
Etash Kumar Guha, Eugene Ndiaye, Xiaoming Huo
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 proposed sparse homotopy algorithm instead of Approximate Homotopy | Lasso homotopy · Synth, Friedman1, Diabetes, Friedman2, Large | homotopy generation time (seconds): 5.176 -> 1.706; 43.823 -> 1.945; 70.813 -> 1.0785; 14.055 -> 0.681; 500.820 -> 150.012 | Improved |
| Use the proposed sparse homotopy algorithm instead of Approximate Homotopy | Robust-loss homotopy · Synth, Diabetes, Large | homotopy generation time (seconds): 62.894 -> 27.156; 2.734 -> 2.411; 607.203 -> 323.372 | Improved |
| Use the proposed sparse homotopy algorithm instead of Approximate Homotopy | Robust-loss homotopy · Friedman1, Friedman2 | homotopy generation time (seconds): 1.009 -> 1.069; 0.618 -> 0.701 | Hurt |
| Use the proposed sparse homotopy algorithm instead of Approximate Homotopy | Asymmetric-loss homotopy · Synth, Diabetes, Friedman2, Large | homotopy generation time (seconds): 18.963 -> 9.270; 54.149 -> 27.349; 3.342 -> 2.454; 82.857 -> 41.269 | Improved |
| Use the proposed sparse homotopy algorithm instead of Approximate Homotopy | Asymmetric-loss homotopy · Friedman1 | homotopy generation time (seconds): 2.699 -> 3.147 | Hurt |
| Use the proposed sparse homotopy algorithm instead of Approximate Homotopy | Lasso homotopy · Synth, Friedman1, Diabetes, Friedman2 | negative logarithm of primal gap: 6.597 -> 12.498; 6.469 -> 15.844; 7.554 -> 16.001; 6.702 -> 15.241 | Improved |
| Use the proposed sparse homotopy algorithm instead of Approximate Homotopy | Lasso homotopy · Large | negative logarithm of primal gap: 7.558 -> 7.933 | Improved |
| Use active-set change-point tracking with local linearization and predictor-corrector refitting | sparse GLM homotopy · Quadratic loss examples | homotopy accuracy | Improved |
| Approximate the solution path by linearization between active-set kinks | sparse GLM homotopy · Robust and Asymmetric losses | homotopy accuracy | Mixed |
| Generate the conformal set from the estimated homotopy instead of a 100-candidate Grid search | sparse GLM conformal prediction · Several datasets and loss functions | coverage | No clear effect |
| Generate the conformal set from the estimated homotopy instead of a 100-candidate Grid search | sparse GLM conformal prediction · Several datasets and loss functions | prediction-set length | Improved |
| Generate conformal sets with the proposed homotopy method rather than other homotopy methods | sparse GLM conformal prediction · Several datasets and loss functions | generation time | Improved |
Closest papers by meaning
- Generalized Fast Exact Conformalization (NeurIPS 2024 · similarity 0.89)
- A Unified Comparative Study with Generalized Conformity Scores for Multi-Output Conformal Regression (ICML 2025 · similarity 0.87)
- Backward Conformal Prediction (NeurIPS 2025 · similarity 0.87)
- A Kernel Nonconformity Score for Multivariate Conformal Prediction (NeurIPS 2026 · similarity 0.86)
- Optimal transport-based conformal prediction (ICML 2025 · similarity 0.86)
- Conformal Prediction via Regression-as-Classification (ICLR 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.
