In this paper, we explore the design and analysis of boosting algorithms and the necessary and sufficient conditions on the weak learners to ensure boostability. Following a literature survey on the computational learning theory behind boosting, we explore boosting, boostability, and conditions for both in several boosting algorithms. Finally, we discuss the application of such algorithms in the binary classification problem and extensions to the multi-class problem.
View the full paper here.