A Probabilistic View of AdaBoost

N. Suematsu and A. Hayashi (Japan)

Keywords

Machine Learning, AdaBoost, classification, probability models

Abstract

This paper presents a probabilistic view of AdaBoost by introducing a new AdaBoost type algorithm which has a clear probabilistic interpretation and showing the relation between the algorithm and AdaBoost. The probabilistic view will provide a new standpoint which is useful in de signing better boosting algorithms. The paper also provides an empirical comparison of the introduced algorithm and AdaBoost.

Important Links:



Go Back


IASTED
Rotating Call For Paper Image