BriefGPT.xyz
Feb, 2014
私有学习器的样本复杂度表征
Characterizing the Sample Complexity of Private Learners
HTML
PDF
Amos Beimel, Kobbi Nissim, Uri Stemmer
TL;DR
本文给出了一个样本大小的组合特征,是私有概念类学习足够和必需的。我们介绍了概念类的概率表示概念,并证明了私有学习算法对于概念类C的样本复杂度意味着RepDim(C)=O(m),并且存在一个样本复杂度m = O(RepDim(C))的私有学习算法。
Abstract
In 2008, Kasiviswanathan et al. defined
private learning
as a combination of
pac learning
and
differential privacy
. Informally, a private
→