Research map / Learning theory and algorithms / Prediction and decision losses
Meta–Gradient Boosted Decision Tree Model for Weight and Target Learning
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 |
|---|---|---|---|
| Train target function (T) with gradient boosted decision trees on label features | GLS background learner on crowdsourced LTR · Commercial crowdsourced LTR (Vtest) | ∆DCG@1 relative to MV: from -5.99% (Linear MV) to 6.40%; vs best baseline LambdaMART DS 0.12% | Helped |
| Train weight function only (W) with decision trees (ablation) | GLS background learner on crowdsourced LTR · Commercial crowdsourced LTR (Vtest) | ∆DCG@10 relative to MV: -0.09% vs MV 0.00% | No clear effect |
| Train both weight and target functions (W+T) versus target only (T) (ablation) | MGBDT · Commercial crowdsourced LTR (Vtest) | ∆DCG@1 relative to MV: 6.98% vs 6.40% | No clear effect |
| Use decision-tree-based target and weight functions instead of linear combinations of label features | Linear baseline (GLS) · Commercial crowdsourced LTR (Vtest) | ∆DCG@1 relative to MV: Linear MV -5.99% vs MGBDT T 6.40% | Helped |
| Use DS consensus labels for LambdaMART instead of majority vote | LambdaMART · Commercial crowdsourced LTR (Vtest) | ∆DCG@10 relative to MV: 0.34% vs 0.00% | Helped |
| Use GLAD consensus labels for LambdaMART instead of majority vote | LambdaMART · Commercial crowdsourced LTR (Vtest) | ∆DCG@1 relative to MV: -1.19% vs 0.00% | Hurt |
| Reweighting by deviations (RD) on Linear GLS | Linear baseline · Commercial crowdsourced LTR (Vtest) | ∆DCG@1 relative to MV: -7.55% vs -5.99% for Linear MV | Hurt |
Closest papers by meaning
- Improving Generalization via Meta-Learning on Hard Samples (CVPR 2024 · similarity 0.88)
- Finding Influential Training Samples for Gradient Boosted Decision Trees (ICML 2018 · similarity 0.86)
- Learning Gradient Boosted Decision Trees with Algorithmic Recourse (NeurIPS 2025 · similarity 0.86)
- MetaOptimize: A Framework for Optimizing Step Sizes and Other Meta-parameters (ICML 2025 · similarity 0.86)
- On the Difficulty of Learning a Meta-network for Training Data Selection (ICML 2026 · similarity 0.86)
- Memory-Reduced Meta-Learning with Guaranteed Convergence (AAAI 2025 · similarity 0.85)
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.