Research map / Learning theory and algorithms / Prediction and decision losses
Hinge Regression Tree: A Newton Method for Oblique Regression Tree Splitting
Hongyi Li, Han Lin, Jun Xu
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 unit Newton step mu=1.0 instead of small damping (ablation) | HRT node split · sinc node-level | convergence stability: mu=1.0 unstable; mu=0.1, 0.05 converge | Hurt |
| use large damping step mu=0.5 instead of 0.1 (ablation) | HRT node split · sinc node-level | convergence: mu=0.5 limit cycle; mu=0.1, 0.05 robust | Hurt |
| use unit Newton step mu=1.0 instead of smaller steps (ablation) | HRT node split · twisted_sigmoid node-level | iterations to converge: mu=1.0 fastest, smaller steps need more iterations | Improved |
| use HRT oblique splits instead of CART | HRT · Concrete | tree size: depth 11.2 -> 3, leaves 113 -> 5.8 | Improved |
| use HRT instead of DTSemNet | HRT · Fried | training time (s): 69.979 -> 2.807 | Improved |
| use HRT instead of DGT | HRT · Fried | training time (s): 395.243 -> 2.807 | Improved |
| use HRT instead of CART | HRT · Fried | training time (s): 0.171 -> 2.807 | Hurt |
| use HRT instead of XGBoost | HRT · CTSlice | RMSE: 1.18 -> 1.41 | Hurt |
| use HRT instead of XGBoost | HRT · Blog | RMSE: 23.02 -> 25.41 | Hurt |
| use HRT instead of XGBoost | HRT · Synthetic f1 | RMSE: 0.3018 -> 0.1646 | Improved |
| use HRT instead of XGBoost | HRT · Synthetic f4 | RMSE: 0.0556 -> 0.0555 | No clear effect |
| use HRT instead of CART | HRT · Synthetic f1 | RMSE: 0.8552 -> 0.1646 | Improved |
Closest papers by meaning
- Alternating optimization of decision trees, with application to learning sparse oblique trees (NeurIPS 2018 · similarity 0.92)
- Near-Optimal Decision Trees in a SPLIT Second (ICML 2025 · similarity 0.87)
- Piecewise Constant and Linear Regression Trees: An Optimal Dynamic Programming Approach (ICML 2024 · similarity 0.87)
- Multistage Defer Trees for Hybrid Interpretability: If at First You Can't Succeed, Tree Again (NeurIPS 2026 · similarity 0.86)
- Regression Trees Know Calculus (NeurIPS 2025 · similarity 0.86)
- Symbolic Regression Enhanced Decision Trees for Classification Tasks (AAAI 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.
