我们提出一种新的随机梯度法来优化有限光滑函数集的和，这种求和是具有强凸性质的。
We propose a new stochastic gradient methodfor optimizing the sum of a finite set of smooth functions, where the sum isstrongly convex.
当标准随机梯度方法以次线性速率收敛于该问题时，所提出的方法引入了先前存储的梯度值，以实现线性收敛速度。
While standard stochastic gradient methodsconverge at sublinear rates for this problem, the proposed method incorporatesa memory of previous gradient values in order to achieve a linear convergencerate.
在机器学习环境中，数值实验表明新算法在优化训练误差和快速减小测试误差方面都显著优于标准算法。
In a machine learning context, numericalexperiments indicate that the new algorithm can dramatically outperformstandard algorithms, both in terms of optimizing the training error andreducing the test error quickly.
机器学习中出现的大量问题涉及在大规模训练示例上计算损失函数和的近似最小值，其中这些示例之间存在大量冗余。
A plethora of the problems arising inmachine learning involve computing an approximate minimizer of the sum of aloss function over a large number of training examples, where there is a largeamount of redundancy between examples.
与本文相关的一个网站供参考：
https://www.cs.ubc.ca/~schmidtm/Software/SAG.html
下载原文请点击:
http://page2.dfpan.com/fs/2lcj42211291868b628/
更多精彩文章请关注微信号：