Research map / Architectures and efficiency / RNNs and normalization
[Re] Hierarchical Shrinkage: Improving the Accuracy and Interpretability of Tree-Based Methods
Domen Mohorčič, David Ocepek
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 |
|---|---|---|---|
| Apply hierarchical shrinkage post-hoc to CART trees | CART · Classification datasets, 15 leaves | relative AUC: 6.2 | Improved |
| Apply HS on top of cost-complexity pruning | CART with CCP · Classification datasets, 15 leaves | relative AUC: 6.5 | Improved |
| Apply HS to C4.5 trees | C4.5 · Classification datasets, 15 leaves | relative AUC: 8 | Improved |
| Apply HS to CART on regression data | CART · Regression datasets, 15 leaves | relative R2: 9.8 | Improved |
| Apply HS to CART with CCP on regression data | CART with CCP · Regression datasets, 15 leaves | relative R2: 10.1 | Improved |
| Replace leaf-based shrinkage with hierarchical shrinkage (ablation) | CART · Classification and regression datasets | AUC and R2 | Improved |
| Increase number of leaves (deeper trees) with HS (ablation) | CART and CART with CCP · Larger datasets | prediction performance | Improved |
| Apply HS to each tree of a Random Forest | Random Forest · Classification and regression datasets | prediction accuracy | Improved |
| Regularize RF with HS instead of max depth or mtry (ablation) | Random Forest · Classification datasets | prediction accuracy | Mixed |
| Apply HS to a two-feature RF | Random Forest · Diabetes classification | AUC: 0.733 -> 0.787 | Improved |
| Apply HS to BART trees | BART · Classification and regression datasets | AUC and R2 | No clear effect |
| Apply HS to GOSDT optimal trees | GOSDT · Selected classification datasets | AUC | No clear effect |
Closest papers by meaning
- Multistage Defer Trees for Hybrid Interpretability: If at First You Can't Succeed, Tree Again (NeurIPS 2026 · similarity 0.84)
- Prediction-Powered Adaptive Shrinkage Estimation (ICML 2025 · similarity 0.83)
- Piecewise Constant and Linear Regression Trees: An Optimal Dynamic Programming Approach (ICML 2024 · similarity 0.83)
- Expected Probabilistic Hierarchies (NeurIPS 2024 · similarity 0.82)
- UltraTWD: Optimizing Ultrametric Trees for Tree-Wasserstein Distance (ICML 2025 · similarity 0.82)
- LowRankArena: A Standardized Evaluation Platform for SVD-Based LLM Compression (NeurIPS 2026 · similarity 0.82)
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.
