Accelerated Gradient Boosting - Sorbonne Université Access content directly
Journal Articles Machine Learning Year : 2019

Accelerated Gradient Boosting

Abstract

Gradient tree boosting is a prediction algorithm that sequentially produces a model in the form of linear combinations of decision trees, by solving an infinite-dimensional optimization problem. We combine gradient boosting and Nesterov's accelerated descent to design a new algorithm, which we call AGB (for Accelerated Gradient Boosting). Substantial numerical evidence is provided on both synthetic and real-life data sets to assess the excellent performance of the method in a large variety of prediction problems. It is empirically shown that AGB is much less sensitive to the shrinkage parameter and outputs predictors that are considerably more sparse in the number of trees, while retaining the exceptional performance of gradient boosting.
Fichier principal
Vignette du fichier
biau-cadre-rouviere-rev.pdf (512.99 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-01723843 , version 1 (05-03-2018)
hal-01723843 , version 2 (05-03-2019)

Identifiers

Cite

Gérard Biau, Benoît Cadre, Laurent Rouvière. Accelerated Gradient Boosting. Machine Learning, 2019, 108 (6), pp.971-992. ⟨10.1007/s10994-019-05787-1⟩. ⟨hal-01723843v2⟩
370 View
251 Download

Altmetric

Share

Gmail Facebook X LinkedIn More