Research map / Probabilistic and scientific modeling / Variational and generative sampling
Unsupervised Tree Boosting for Learning Probability Distributions
Naoki Awaya, Li Ma
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 |
|---|---|---|---|
| add boosting ensemble of trees instead of a single tree (ablation) | BoostPM · 48-D simulation (n=10000) | predictive score | Improved |
| lower global shrinkage c0 (learning rate) (ablation) | BoostPM · 48-D simulation, gamma=0 | predictive score: c0 0.99 -> 0.01 gives better scores, up to 4000-5000 trees | Improved |
| add small-node penalty gamma > 0 (not too large) (ablation) | BoostPM · 48-D simulation, c0=0.1 | predictive score: gamma in 0.0 to 2.0; fit improves while gamma < 1 | Improved |
| use too large gamma (>=1) (ablation) | BoostPM · 48-D simulation, c0=0.1 | predictive score | Hurt |
| add two-stage strategy (marginals then copula) (ablation) | BoostPM · 48-D simulation | predictive score | Improved |
| use BoostPM instead of deep density destructor (DDD) | BoostPM · 48-D simulation, 30 datasets | predictive score and compute time | Improved |
| use BoostPM (gamma=0.0) instead of MAF (ablation) | BoostPM · POWER | mean log-density: 0.30 -> 1.20 | Improved |
| use BoostPM (gamma=0.0) instead of MAF (ablation) | BoostPM · AReM | mean log-density: 9.49 -> 12.28 | Improved |
| use BoostPM (gamma=0.0) instead of MAF (ablation) | BoostPM · BANK | mean log-density: 20.09 -> 36.47 | Improved |
| use BoostPM (gamma=0.0) instead of MAF (ablation) | BoostPM · HEPMASS | mean log-density: -17.39 -> -20.16 | Hurt |
| use BoostPM (gamma=0.0) instead of MAF (ablation) | BoostPM · MINIBOONE | mean log-density: -11.68 -> -20.38 | Hurt |
| use BoostPM (gamma=0.0) instead of MAF (ablation) | BoostPM · CASP | mean log-density: 27.61 -> 21.84 | Hurt |
Closest papers by meaning
- Free Lunch in the Forest: Functionally-Identical Pruning of Boosted Tree Ensembles (AAAI 2025 · similarity 0.86)
- NRGBoost: Energy-Based Generative Boosted Trees (ICLR 2025 · similarity 0.84)
- Wasserstein Gradient Boosting: A Framework for Distribution-Valued Supervised Learning (NeurIPS 2024 · similarity 0.84)
- Conditional Density Estimation with Histogram Trees (NeurIPS 2024 · similarity 0.83)
- Minimax Rates and Spectral Distillation for Tree Ensembles (NeurIPS 2026 · similarity 0.83)
- Tree Ensemble Explainability through the Hoeffding Functional Decomposition and TreeHFD Algorithm (NeurIPS 2025 · similarity 0.83)
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.
