BriefGPT.xyz
Oct, 2017
具有方差减少的随机共轭梯度算法
Stochastic Conjugate Gradient Algorithm with Variance Reduction
HTML
PDF
Xiao-Bo Jin, Xu-Yao Zhang, Kaizhu Huang, Guang-Gang Geng
TL;DR
提出一种带有方差缩减的新型随机共轭梯度算法，并使用Fletcher和Reeves方法证明其对于强凸光滑函数的线性收敛性。 实验表明，与其他算法相比，该算法在四个学习模型中收敛更快，同时在六个大数据集上表现相当，但计算效率显著提高。
Abstract
conjugate gradient methods
are a class of important methods for solving linear equations and nonlinear optimization. In our work, we propose a new stochastic conjugate gradient algorithm with
variance reduction
(
→