BriefGPT.xyz
May, 2020
差分隐私下的联邦推荐系统
Federated Recommendation System via Differential Privacy
HTML
PDF
Tan Li, Linqi Song, Christina Fragouli
TL;DR
本文探讨了结合差分隐私和多智能体赌博学习的联邦私有赌博机制。我们研究了如何将基于差分隐私的置信上界方法应用于多智能体环境,特别是应用于主-从和完全分散的联邦学习环境中,并提供了有关所提出方法的隐私和后悔性能的理论分析,并探讨了这两者之间的权衡。
Abstract
In this paper, we are interested in what we term the
federated private bandits framework
, that combines
differential privacy
with
multi-agent ban
→