Research map / Learning theory and algorithms / Prediction and decision losses
Yggdrasil: An Optimized System for Training Deep Decision Trees at Scale
Firas Abuzaid, Joseph K. Bradley, Feynman T Liang, Andrew Feng, Lee Yang, Matei Zaharia, Ameet S Talwalkar
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 |
|---|---|---|---|
| vertical instead of horizontal partitioning (Yggdrasil vs MLlib PLANET) at depth 20 | distributed decision tree training · MNIST 8M | training time: up to 6× faster | Improved |
| vertical partitioning at depth 18 with 3500 features | distributed decision tree training · Yahoo 2M | training time: up to 24× faster | Improved |
| MLlib PLANET at shallow depth | MLlib · shallow trees | training time | Mixed |
| increase feature count p from 500 to 4000 | Yggdrasil vs PLANET · Synthetic, D=15 | speedup over PLANET: 3× -> 8× | Improved |
| sparse bitvectors for communication (ablation) | Yggdrasil · MNIST 8M | runtime | Improved |
| dictionary-encode labels (ablation) | Yggdrasil · MNIST 8M | runtime: 20% improvement | Improved |
| run-length encode features (ablation) | Yggdrasil · MNIST 8M | runtime: 20% improvement | Improved |
| all optimizations vs naive vertical baseline (ablation) | Yggdrasil · MNIST 8M | runtime: 40% reduction | Improved |
| increase tree depth beyond 13 | XGBoost · MNIST 8M and Yahoo 2M | trainability: crashes beyond D = 13 | Hurt |
Closest papers by meaning
- Near-Optimal Decision Trees in a SPLIT Second (ICML 2025 · similarity 0.87)
- Branches: Efficiently Seeking Optimal Sparse Decision Trees via AO* (ICML 2025 · similarity 0.86)
- Symbolic Regression Enhanced Decision Trees for Classification Tasks (AAAI 2024 · similarity 0.86)
- Optimal Decision Tree Pruning Revisited: Algorithms and Complexity (ICML 2025 · similarity 0.85)
- Optimal Classification Trees for Continuous Feature Data Using Dynamic Programming with Branch-and-Bound (AAAI 2025 · similarity 0.85)
- Learning Small Decision Trees with Few Outliers: A Parameterized Perspective (AAAI 2024 · 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.
