BriefGPT.xyz
May, 2015
快速差分隐私矩阵分解
Fast Differentially Private Matrix Factorization
HTML
PDF
Ziqi Liu, Yu-Xiang Wang, Alexander J. Smola
TL;DR
提出了一种简单的算法来实现可证明偏差私有性以及良好性能的差异性私人协作过滤。通过差分隐私和贝叶斯后验采样的新型连接方式，该算法可有效实现。同时，通过精细的系统设计和利用数据的幂律行为最大化CPU缓存带宽，我们可以在单个PC上以8.5百万每秒的速率生成1024维模型实现推荐。
Abstract
differentially private
collaborative filtering
is a challenging task, both in terms of accuracy and speed. We present a simple algorithm that is provably
→