Research map / Learning theory and algorithms / Kernels and regression theory
A Refined Margin Distribution Analysis for Forest Representation Learning
Shen-Huan Lyu, Liang Yang, Zhi-Hua Zhou
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 |
|---|---|---|---|
| remove PRECONC (pass only input features to next layer) (ablation) | mdDF · PROTEIN | test accuracy: 71.247 -> 68.509 | Hurt |
| remove PRECONC (pass only input features to next layer) (ablation) | mdDF · SENSIT | test accuracy: 82.534 -> 80.320 | Hurt |
| remove PRECONC (pass only input features to next layer) (ablation) | mdDF · ADULT | test accuracy: 86.560 -> 85.650 | Hurt |
| stacking: pass only prediction vectors to next layer (ablation) | mdDF · SENSIT | test accuracy: 82.534 -> 80.412 | Hurt |
| stacking: pass only prediction vectors to next layer (ablation) | mdDF · PROTEIN | test accuracy: 71.247 -> 70.291 | Hurt |
| use same forests (4 random forests) in each layer (ablation) | mdDF · ADULT | test accuracy: 86.560 -> 86.200 | Hurt |
| use same forests (4 random forests) in each layer (ablation) | mdDF · LETTER | test accuracy: 97.500 -> 96.475 | Hurt |
| use same forests (4 random forests) in each layer (ablation) | mdDF · MNIST | test accuracy: 98.734 -> 98.254 | Hurt |
| margin distribution reweighting and boosted representations vs gcForest | gcForest · ADULT | test accuracy: 86.276 -> 86.560 | Improved |
| margin distribution reweighting and boosted representations vs gcForest | gcForest · MNIST | test accuracy: 98.252 -> 98.734 | Improved |
| margin distribution reweighting vs gcForest | gcForest · PROTEIN | test accuracy: 71.009 -> 71.247 | Improved |
| deep forest (mdDF) instead of MLP | MLP · ADULT | test accuracy: 80.597 -> 86.560 | Improved |
Closest papers by meaning
- Analyzing the tree-layer structure of Deep Forests (ICML 2021 · similarity 0.83)
- Depth is More Powerful than Width with Prediction Concatenation in Deep Forest (NeurIPS 2022 · similarity 0.83)
- Minimax Rates and Spectral Distillation for Tree Ensembles (NeurIPS 2026 · similarity 0.82)
- Discriminative Forests Improve Generative Diversity for Generative Adversarial Networks (AAAI 2024 · similarity 0.81)
- Input Margins Can Predict Generalization Too (AAAI 2024 · similarity 0.81)
- Generative Marginalization Models (ICML 2024 · similarity 0.80)
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.
