RI: Small: Boosting, Optimality and Game Theory

点击翻译按钮获取中文摘要

Boosting is a machine-learning method based on combining manycarefully trained weak prediction rules into a single, highly accurateclassifier. Boosting has both a rich theory and a record of empiricalsuccess, for instance, to face detection and spoken-dialogue systems.The theory of boosting is broadly connected to other research fields,but has only been fully developed for the simplest learning problems.Nevertheless, in practice, boosting is commonly applied in settingswhere the theory lags well behind. We do not know if such practicalmethods are truly best possible; even for binary classification, it isnot clear how to best exploit what is known about how boostingoperates. New challenges will demand an even greater widening of thefoundations of boosting.The goal of this project is to develop broad theoretical insights andversatile algorithmic principles. The aim is to studygame-theoretically how to design the most efficient and effectiveboosting algorithms possible.Research on boosting is spread over many years. across multiplepublications and disciplines. To organize this body of work, asignificant activity of this project is the completion of a book onboosting which will provide a valuable resource for students andresearchers of diverse backgrounds and interests.Boosting has historically had a major impact on areas outside machinelearning, such as statistics, computer vision, and speech and languageprocessing. Thus, there is a strong potential for work at itsfoundations to have a broad impact on these other research andapplication areas as well.