BriefGPT.xyz
Nov, 2012
近端随机对偶坐标上升
Proximal Stochastic Dual Coordinate Ascent
HTML
PDF
Shai Shalev-Shwartz, Tong Zhang
TL;DR
介绍了一个基于proximal的对偶协调上升方法，该算法框架可以用于多种正则化损失最小化问题，包括l1正则化和结构化输出SVM。我们取得的收敛速度与现有最先进结果匹配并有时超过。
Abstract
We introduce a proximal version of
dual coordinate ascent
method. We demonstrate how the derived algorithmic framework can be used for numerous
regularized loss minimization
problems, including $\ell_1$ regulariz
→