Oral Prelim: Young Hun Jung, 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 excellent performance on real data as it adapts to the unknown edges of the weak learners. The proposed algorithms are the first online multiclass boosting algorithms with theoretical mistake bounds. They also outperform other state-of-the-art algorithms on real data. Finally, we discuss some future directions for research regarding the development of boosting algorithms for supervised learning problems with complex label spaces. Such problems include multiclass multilabel problems and ranking problems.