文章属于综述,对于整个LDP的应用场景,方向都进行了系统概括。语言简单易懂,供我这种菜鸟学习,找到合适的方向。
Amplification论文:A. V. Evfimievski, J. Gehrke, and R. Srikant. Limiting privacy breaches in privacy preserving data mining. In PODS, pages 211–222. ACM, 2003.
2.2 行业发展
谷歌公司的RAPPOR对于网页地址的获取(没有获取用户习惯的前提下),苹果公司的成果,微软的遥感数据收集三方面。目前已经应用差分隐私做的一些落地成果。
2.3 理论基础
以上为本地化差分隐私概念 。
In brief, LDP has a stronger privacy model than DP, but entails greater noise. We will also cover the refinements of the definition, i.e., the relationship between bounded and unbounded definitions, node-based and edge-based definitions (in graph problems), pure and non-pure definitions.We will also describe composition theorems, including parallel, sequential, and advanced composition in the local setting.
2.4频数估计(没什么好看)
2.5 现有研究方向(自己感兴趣的)
2.6 问题和新方向