BriefGPT.xyz
Mar, 2017
平滑目标高效的私有ERM
Efficient Private ERM for Smooth Objectives
HTML
PDF
Jiaqi Zhang, Kai Zheng, Wenlong Mou, Liwei Wang
TL;DR
本文综述了不同凸且光滑目标函数的梯度下降算法与输出扰动方法,不仅能达到近乎最优的效用,而且也提高了以前的隐私优化算法的运行时间。同时,对于非凸光滑目标函数,还提出了RRPSGD算法,并附上了隐私保障的收敛性证明。实验表明,我们的算法在效益和运行时间方面始终优于现有方法。
Abstract
In this paper, we consider efficient
differentially private empirical risk minimization
from the viewpoint of
optimization algorithms
. For strongly convex and smooth objectives, we prove that
→