AI Paper Atlas

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

NeurIPS 2016 · Learning theory and algorithms › Prediction and decision losses · approach: Tree ensemble optimization

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 changedWhereMeasuredEffect
vertical instead of horizontal partitioning (Yggdrasil vs MLlib PLANET) at depth 20distributed decision tree training · MNIST 8Mtraining time: up to 6× fasterImproved
vertical partitioning at depth 18 with 3500 featuresdistributed decision tree training · Yahoo 2Mtraining time: up to 24× fasterImproved
MLlib PLANET at shallow depthMLlib · shallow treestraining timeMixed
increase feature count p from 500 to 4000Yggdrasil vs PLANET · Synthetic, D=15speedup over PLANET: 3× -> 8×Improved
sparse bitvectors for communication (ablation)Yggdrasil · MNIST 8MruntimeImproved
dictionary-encode labels (ablation)Yggdrasil · MNIST 8Mruntime: 20% improvementImproved
run-length encode features (ablation)Yggdrasil · MNIST 8Mruntime: 20% improvementImproved
all optimizations vs naive vertical baseline (ablation)Yggdrasil · MNIST 8Mruntime: 40% reductionImproved
increase tree depth beyond 13XGBoost · MNIST 8M and Yahoo 2Mtrainability: crashes beyond D = 13Hurt

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy