GithubHelp home page GithubHelp logo

ldp-afterreading's Introduction

前言

一开始决定分为三个部分,第一部分为论文梳理,第二部分为具体细节,会放上一些论文的主体理解,第三部分放一些实验代码

目录

DP scenario

这里会收集一些DP和LDP在现实生活中的用例

Title Team/Main Author Venue and Year Key Description
RAPPOR: Randomized Aggregatable Privacy-Preserving Ordinal Response Google 2015 https://github.com/google/rappor
IBM-differential-privacy-library IBM 2018 https://github.com/IBM/differential-privacy-library/blob/main/
Heavy Hitter Estimation over Set-Valued Data with Local Differential Privacy Zhan Qin 2016/CCS

LDP and DP下数据分析

多是一些LDP&DP下的数据融合分析

Title Team/Main Author Venue and Year Key Description
Collecting And Analyzing Data Jointly From Multiple Services under Local Differential Privacy Min Xu 2020/VLDB Endowment
Building a RAPPOR with the Unknown: Privacy-Preserving Learning of Associations and Data Dictionaries G. Fanti, Vasyl Pihur, Ú. Erlingsson 2015
DPSAaS: Multi-Dimensional Data Sharing and Analytics as Services under Local Differential Privacy Min Xu 2019/VLDB Endow
Heavy Hitter Estimation over Set-valued Data with Local Differential Privacy Qin Z,Yang Y 2016 ACM SIGSAC 本地差分隐私下对集值数据的频繁项估计
Local DIfferentially Private Heavy Hitters Identification Tianhao Wang 2021 IEEE 本地化差分隐私下的频繁项识别:提出了前缀扩展方法(Prefix Extending Method,PEM)

LDP and DP下数据发布

多是一些LDP&DP下的数据发布和响应数据查询

Title Team/Main Author Venue and Year Key Description
AHEAD: Adaptive Hierarchical Decomposition for Range Query under Local Differential Privacy Linkang Du 2021/CCS 主要讲述的是针对LDP下响应范围查询的问题,作者提出了一种自适应构建多层次分析树的方案来提升在LDP下响应范围查询的精度
Answering Multi-Dimensional Analytical Queries under Local Differential Privacy Tianhao Wang 2019 pure-LDP概念。三种频率估计机制:单维和多维下的HIO机制、SC拆分再联接机制
Continuous Release of Data Streams under both Centralized and Local Differential Privacy Tianhao Wang 2019 主要描述的是在数据流发布环境下如何使用DP和LDP对数据流施加差分隐私保护,同时作者提出了两个框架,一个是基于DP的ToPS框架,一个是基于LDP的ToPL框架
Empirical Risk Minimization in the Non-interactive Local Model of Differential Privacy Di Wang 2020 这篇文章提出使用内积多项式的形式释放函数,一般是用于分布式学习和联邦学习下,但在响应边缘查询(k-way Margin Query)时也提出了使用内积多项式的形式去简化查询函数(使用数学去定义查询函数),从而提升查询精度

|

DP and LDP与机器学习相结合

多是一类分布式与联邦学习下数据聚合方面,但是根据差分隐私与机器学习结合部位的不同(例如:目标函数扰动、输入扰动、梯度扰动、输出扰动)可以用于不同的场景下

Title Team/Main Author Venue and Year Key Description
Empirical Risk Minimization in the Non-interactive Local Model of Differential Privacy Di Wang 2020 这篇文章基于前人工作(伯恩斯坦多项式机制、非交互式差分隐私方面等提出了使用内积多项式的方法释放函数,还提出了1-bit的通信方法,能够使泛化误差的上界更为紧致,且对样本量n的依赖度降为多项式级别,但是问题在于这只是理论上的,作者也表明了不知道在现实中的应用效果是怎么样的)
Optimal Algorithms for Mean Estimation under Local Differential Privacy Hilal Asi, V. Feldman, Kunal Talwar 2022/ICML 用于处理聚合过程中的均值估计的问题,开发了一种基于高斯机制的PrivUnit方法,通过采样的方式降低噪声引入带来的误差

|

DP and LDP与联邦学习相结合

问题描述

Title Team/Main Author Venue and Year Key Description
LDP-FL: Practical Private Aggregation in Federated Learning with Local Differential Privacy Lichao Sun 2021 本文针对在多层神经网络中存在隐私预算爆炸的问题提出了LDP-FL框架,不仅很好的保护了隐私,而且通过分割洗牌机制降低了隐私预算在多层迭代中的激增
LDP-Fed: Federated Learning with Local Differential Privacy Stacey Truex 2020 本文提出了LDP-Fed框架,LDP 模块为在多个个体参与者的私有数据集上的大规模神经网络联合训练中模型训练参数的重复收集提供了正式的差分隐私保证。其次,LDP-Fed实现了一套选择和过滤技术,用于扰动和与参数服务器共享选择的参数更新。

数据与代码

这一部分打算将收集到的一些数据和组内写的代码放在这里,数据出处也会进行标明,代码出处当然也会

机器学习相关论文查询:https://paperswithcode.com/ 一些LDP协议的代码库:https://github.com/vvv214/LDP_Protocols

ldp-afterreading's People

Contributors

su-whale avatar triumphhh avatar altruistics avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.