AI Paper Atlas

Research map / Learning theory and algorithms / Prediction and decision losses

Explaining the Success of AdaBoost and Random Forests as Interpolating Classifiers

Abraham J. Wyner, Matthew Olson, Justin Bleich, David Mease

JMLR 2017 · Learning theory and algorithms › Prediction and decision losses · approach: Tree ensemble optimization · 253 citations (OpenAlex)

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
boost deep trees (2^8 nodes) instead of stumps (ablation)AdaBoost · 5D simulation, 250 iterationspoints differing from Bayes rule (of 1000): 141/1000 -> 116/1000Improved
run AdaBoost for more iterations beyond interpolation (10 to 100) (ablation)AdaBoost · 5D simulation, 200 repetitionsdisagreement with Bayes rule: .19 -> .15Improved
run AdaBoost for 1000 instead of 9 iterations (ablation)AdaBoost · 20D pure noiseproportion of hold-out classified +1 (error): .1433 -> .0008Improved
average ten interpolating sub-ensembles (100 to 1000 iterations) (ablation)AdaBoost · 20D pure noiseagreement with Bayes rule: 80% (single interpolator) -> 99.92%Improved
use AdaBoost or random forest instead of one-NN as interpolatorinterpolating classifiers · 2D pure noise (p=0.8)fraction classified +1: 79% -> 87% (AdaBoost), 94% (random forests)Improved
use AdaBoost or random forest instead of pruned CART or one-NNclassifiers · 2D circle with signaloverall error rate: 0.18 (CART), 0.20 (one-NN) -> 0.13Improved
flip 100 training labels (about 2%)AdaBoost depth-8 trees · Phonemetest error: 9.0% -> 10.2%Hurt
flip 100 training labels (about 2%)Random forest · Phonemetest error: 9.4% -> 10.0%Hurt
flip 100 training labels (about 2%)One-NN · Phonemetest error: 10.5% -> 12.6%Hurt
increase depth of boosted trees (ablation)AdaBoost · Phonemetest errorImproved
add 5% label noiseAdaBoost vs one-NN · Habermanincrease in test error: 1.55 (one-NN) -> 0.13 (AdaBoost)Improved
add 5% label noiseAdaBoost vs one-NN · German creditincrease in test error: 0.68 (one-NN) vs 0.29 (AdaBoost), 0.68 (RF)No clear effect

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