SelfieBoost: A Boosting Algorithm for Deep Learning
Machine Learning
2017-04-11 v2 Machine Learning
Abstract
We describe and analyze a new boosting algorithm for deep learning called SelfieBoost. Unlike other boosting algorithms, like AdaBoost, which construct ensembles of classifiers, SelfieBoost boosts the accuracy of a single network. We prove a convergence rate for SelfieBoost under some "SGD success" assumption which seems to hold in practice.
Keywords
Cite
@article{arxiv.1411.3436,
title = {SelfieBoost: A Boosting Algorithm for Deep Learning},
author = {Shai Shalev-Shwartz},
journal= {arXiv preprint arXiv:1411.3436},
year = {2017}
}