BriefGPT.xyz
Feb, 2015
自适应概率随机对偶坐标上升
Stochastic Dual Coordinate Ascent with Adaptive Probabilities
HTML
PDF
Dominik Csiba, Zheng Qu, Peter Richtárik
TL;DR
介绍了AdaSDCA：一种自适应的随机对偶坐标上升（SDCA）变体，用于解决正则化经验风险最小化问题。AdaSDCA通过在迭代过程中自适应地改变对偶变量上的概率分布，实现了比SDCA更好的复杂度界限。同时，我们提出了AdaSDCA+：一种实用的变体，在实验中表现优于现有的非自适应方法。
Abstract
This paper introduces
adasdca
: an adaptive variant of
stochastic dual coordinate ascent
(SDCA) for solving the regularized empirical risk minimization problems. Our modification consists in allowing the method ad
→