凸优化的广义Boosting算法

May, 2011

Generalized Boosting Algorithms for Convex Optimization

Alexander Grubb, J. Andrew Bagnell

TL;DR该研究分析了梯度下降算法，引入了新的弱学习器性能度量，并扩展了Boosting方法，以支持任意凸损失函数，并给出了相应的弱到强的收敛结果。

Abstract

boosting is a popular way to derive powerful learners from simpler hypothesis classes. Following previous work (Mason et al., 1999; Friedman, 2000) on general boosting frameworks, we analyze →