Online Multiclass Boosting

TL;DR

This paper introduces an online multiclass boosting algorithm with a new weak learning condition, achieving near-optimal performance and demonstrating strong results on real data.

Contribution

It defines a weak learning condition for online multiclass boosting and proposes an adaptive, near-optimal boosting algorithm.

Findings

Achieves minimal weak learners for target accuracy

Performs well on real datasets

Provides theoretical guarantees for online multiclass boosting

Abstract

Recent work has extended the theoretical analysis of boosting algorithms to multiclass problems and to online settings. However, the multiclass extension is in the batch setting and the online extensions only consider binary classification. We fill this gap in the literature by defining, and justifying, a weak learning condition for online multiclass boosting. This condition leads to an optimal boosting algorithm that requires the minimal number of weak learners to achieve a certain accuracy. Additionally, we propose an adaptive algorithm which is near optimal and enjoys an excellent performance on real data due to its adaptive property.