Research map / Architectures and efficiency / RNNs and normalization
Globally Induced Forest: A Prepruning Compression Scheme
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| candidate window size 10 instead of 1 (ablation) | GIF · Vowel | error rate (%): 15.90 -> 10.87 | Helped |
| candidate window size 10 instead of 1 (ablation) | GIF · Letter | error rate (%): 9.07 -> 5.88 | Helped |
| candidate window size 10 instead of 1 (ablation) | GIF · Waveform | error rate (%): 14.51 -> 14.05 | Helped |
| candidate window size infinite instead of 1 (ablation) | GIF · Friedman1, T=1000 | MSE: 2.37 -> 3.05 | Hurt |
| increase initial number of trees from 10 to 1000 (ablation) | GIF CW=1 · Friedman1 | MSE: 7.88 -> 2.37 | Helped |
| increase initial number of trees from 1000 to 10000 (ablation) | GIF CW=1 · Twonorm | error rate (%): 3.35 -> 3.53 | No clear effect |
| increase trees from 1000 to 10000 (ablation) | GIF CW=1 · Friedman1 | MSE: 2.37 -> 2.26 | No clear effect |
| candidate window infinite increases fitting time as trees grow from 1000 to 10000 (ablation) | GIF · Friedman1 | fitting time (s): 3.84 -> 25.95 | Hurt |
| use stump Adaboost instead of GIF | Boosting · Musk2 | fitting time (s): 1.53 -> 399.17 | Hurt |
| trimmed exponential loss instead of multi-output square loss (ablation) | GIF 10% · Hastie | error rate (%): 10.33 -> 7.38 | Helped |
| trimmed exponential loss instead of multi-output square loss (ablation) | GIF 10% · Letter | error rate (%): 2.82 -> 9.07 | Hurt |
| budget 1% instead of 10% (ablation) | GIF · Abalone | MSE: 5.20 -> 4.74 | Helped |
Closest papers by meaning
- Compressing tree ensembles through Level-wise Optimization and Pruning (ICML 2025 · similarity 0.88)
- Minimax Rates and Spectral Distillation for Tree Ensembles (NeurIPS 2026 · similarity 0.86)
- Free Lunch in the Forest: Functionally-Identical Pruning of Boosted Tree Ensembles (AAAI 2025 · similarity 0.84)
- GRANDE: Gradient-Based Decision Tree Ensembles for Tabular Data (ICLR 2024 · similarity 0.83)
- Tree Ensemble Explainability through the Hoeffding Functional Decomposition and TreeHFD Algorithm (NeurIPS 2025 · similarity 0.83)
- GIO: Gradient Information Optimization for Training Dataset Selection (ICLR 2024 · 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.