SelfieBoost: A Boosting Algorithm for Deep 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 $\log (1/\epsilon )$ convergence rate for SelfieBoost under some "SGD success" assumption which seems to hold in practice.

References in corpus (1)

• Visualizing and Understanding Convolutional Networks

Cited by in corpus (3)

• Learning Deep ResNet Blocks Sequentially using Boosting Theory
• Soft Gradient Boosting Machine
• Lecture video indexing using boosted margin maximizing neural networks