BriefGPT.xyz
Feb, 2018
重新审视差分隐私的经验风险最小化问题：更快且更广泛
Differentially Private Empirical Risk Minimization Revisited: Faster and More General
HTML
PDF
Di Wang, Minwei Ye, Jinhui Xu
TL;DR
本文研究不同设置下差分隐私经验风险最小化问题，提出了比以前更少的梯度复杂度的算法，并从凸损失函数推广到满足Polyak-Lojasiewicz条件的非凸函数，给出比传统算法更紧的上界。
Abstract
In this paper we study the
differentially private
empirical risk minimization
(ERM) problem in different settings. For smooth (strongly) convex loss function with or without (non)-smooth regularization, we give a
→