Statistical Understanding of Boosting
Boosting can be seen as minimization of a convex loss function over a convex set of functions. Specifically, the loss being minimized is the exponential loss
and we are seeking a function
Read more about this topic: AdaBoost