# Papers
<!-- [](https://github.com/youngfish42/Awesome-Federated-Learning-on-Graph-and-Tabular-Data/stargazers) [](https://awesome.re) [](https://github.com/youngfish42/image-registration-resources/blob/master/LICENSE)  -->
bilibili（论文分享）:https://space.bilibili.com/80356866/video
**Table of Contents**
- [Papers](#Papers)
- [DP Theory](#dp-theory)
- [Differential Adversary Definition](#differential-adversary-definition)
- [CDP](#cdp)
- [LDP](#ldp)
- [Privacy Measurement Method](#privacy-measurement-method)
- [DP](#dp)
- [RDP(MA)](#rdpma)
- [ZCDP](#zcdp)
- [GDP](#gdp)
- [Bayesian DP](#Bayesian-DP)
- [Privacy Amplification Technology](#privacy-amplification-technology)
- [Sampling](#samlping)
- [Shuffle](#shuffle)
- [DP and Meachine Learning](#dp-and-meachine-learning)
- [Meachine Leaning](#meachine-learning)
- [Meachine Leaning with DP](#meachine-learning-with-dp)
- [GNN](#gnn)
- [GNN with DP](#gnn-with-dp)
- [Privacy of GNN](#privacy-of-gnn)
- [DP and Federated Leaning](#dp-and-federated-learning)
- [Federated Leaning](#federated-leaning)
- [Horizontal FL with DP](#horizontal-fl-with-dp)
- [client-level](#client-level)
- [samping-level](#samping-level)
- [LDP FL](#ldp-fl)
- [vertical FL with DP](#vertical-fl-with-dp)
- [incentive](#incentive)
- [Attack](#attack)
- [MIA](#membership-inference-attack)
- [Application Scenarios](#application-scenarios)
- [Text Protection](#text-protection)
- [Recommended System](#recommended-system)
- [DP and image](#dp-and-image)
- [DP and cypto](#dp-and-cytro)
- [Meachine Unlearning](#meachine-unlearning)
- [Unlearning in Centralized machine learning](#unlearning-in-centralized-machine-learning)
- [Unlearning in FL](#unlearning-in-fl)
## DP Theory
### Differential Adversary Definition
CDP（central DP）有一个完全可信的中心方，敌手是外界。而LDP（local DP）认为中心方是诚实但好奇的。
#### CDP
| Title | Team/Main Author | Venue and Year | Key Description
| :------------| :------ | :---------- | :-----------------------
| Differential privacy | Cynthia Dwork | ICALP/2006 | 首次提出差分隐私的定义 |
| Programming Differential Privacy （Book）| Joseph P. Near and Chiké Abuah | 2021 | 讲诉了DP的概念定理和机制等，并附有简单代码呈现（简单入门推荐）[【Link】](https://programming-dp.com/) |
| The Algorithmic Foundations of Differential Privacy（Book） | Cynthia Dwork | 2014 | DP的定义理论，高级组合和相关机制等的完整证明推导（更加理论）[【拉普拉斯、严格差分、高斯机制、松弛差分】](https://www.bilibili.com/video/BV18r4y1j7Bs?spm_id_from=333.999.0.0&vd_source=46cfa74ab261e7d7a25c2bfedf5615a3) |
| Differential Privacy From Theory to Practice （Book）| Ninghui Li | 2014 | 除了一些基本定理和机制，用了具体的实际例子讲诉了DP的用法及DP的伦理探讨（更加实用化）[【Chapter1、Chapter2】](https://www.bilibili.com/video/BV1br4y1J7Qn?spm_id_from=333.999.0.0&vd_source=46cfa74ab261e7d7a25c2bfedf5615a3),[原作者讲解]()|
#### LDP
| Title | Team/Main Author | Venue and Year | Key Description
|:---------------------------------------------------------------------------------------------|:-----------------------|:---------------------------------------------------------|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
| 本地化差分隐私研究综述 | 中国人民大学 | Journal of Software/2018 | 介绍了本地化差分隐私的原理与特性,总结和归纳了LDP当前研究工作,重点阐述了频数统计、均值统计的LDP机制设计[【vedio】](https://www.bilibili.com/video/BV18B4y1a75b?spm_id_from=333.999.0.0&vd_source=46cfa74ab261e7d7a25c2bfedf5615a3) |
| RAPPOR: Randomized Aggregatable Privacy-Preserving Ordinal Response | Google | ACM SIGSAC /2014 | 1.RAPPOR的分类和对应扰动方法。2.Differential Privacy loss of RAPPOR（主要是多维） |
| Locally Differentially Private Protocols for Frequency Estimation | Purdue University | USENIX/2017 | 1.提出了一个Pure LDP Protocols，并基于Pure LDP Protocols给出了其方差和频数估计的公式。2.提出OUE（UE=basic RAPPOR），给出了q的最佳选择和方差。[【vedio】](https://www.bilibili.com/video/BV1HW4y127mp?spm_id_from=333.999.0.0&vd_source=46cfa74ab261e7d7a25c2bfedf5615a3) |
| Collecting High-Dimensional and Correlation-Constrained Data with Local Differential Privacy | Rong Du | sensor, mesh and ad hoc communications and networks/2021 | 该文章重要的是针对LDP的高维度情况，本篇重点在于推断多维RR的概率关系： 1、进行属性之间相关性的度量（首先会属性会分为不同的簇，簇之间独立不相干，簇之内的属性相关），定义了p这个变量定义来定理相关性（基于协方差） 2、提出UDLDP，基于以上在单个属性上定义LDP，并提出CBP | | |
| Collecting and Analyzing Data from Smart Device Users with Local Differential Privacy | Thông T. Nguyên et al. | arxiv/2016 | 提出了Harmony，用于包含数值和类别属性的多维数据的LDP下的均值和频数统计。主要是连续型数据直接随机的对称扰动成两个相反数，然后保证均值无偏，误差边界比用Lap小。 | |
| Collecting and Analyzing Multidimensional Data with Local Differential Privacy | Ning Wang et al. | ICDE/2019 | 核心是单维数据LDP的收集，多维是一个简单扩展。单维下文章发现了拉普拉斯加噪和DM（Duchi）的优缺点，两个方法随着eps的增大有�
机器学习和差分隐私的论文笔记和代码仓.zip
需积分: 5 50 浏览量
2024-05-08
11:22:39
上传
评论
收藏 243.19MB ZIP 举报
机器学习和差分隐私的论文笔记和代码仓.zip （102个子文件） 
.csv 61B .csv 41B .csv 28B data_batch_1 29.6MB data_batch_2 29.6MB data_batch_3 29.6MB data_batch_4 29.6MB data_batch_5 29.6MB .gitignore 47B
train-images-idx3-ubyte.gz 25.2MB train-images-idx3-ubyte.gz 9.45MB t10k-images-idx3-ubyte.gz 4.22MB t10k-images-idx3-ubyte.gz 1.57MB train-labels-idx1-ubyte.gz 29KB train-labels-idx1-ubyte.gz 28KB t10k-labels-idx1-ubyte.gz 5KB t10k-labels-idx1-ubyte.gz 4KB
readme.html 88B Awesome-Federated-Learning-and_Differentially_Privacy.iml 473B README.md 67KB batches.meta 158B result.pt 429B dp_optimizer.py 17KB ResNet.py 10KB sampling.py 10KB CNN.py 10KB compute_rdp.py 9KB local_clients_train_process.py 8KB get_data.py 7KB fed_avg_with_dp.py 7KB pathological_nonIID_data.py 7KB vgg_bn.py 7KB DPSGD_Opacus.py 7KB get_MaxSigma_or_MaxSteps.py 5KB compute_dp_sgd.py 5KB dirichlet_nonIID_data.py 5KB fed_avg_with_ldp.py 5KB rdp_convert_dp.py 4KB fed_avg.py 4KB train_with_opacus.py 4KB compute_gdp.py 3KB DNN.py 3KB DPSGD.py 3KB compute_coordinates.py 2KB weight_initialization.py 2KB classifer.py 2KB center_average_model_with_weights.py 2KB SGD.py 2KB train_with_dp.py 2KB IID_data.py 2KB optimizier_and_model_distribution.py 2KB zcdp.py 2KB validation.py 1024B clipping_and_adding_noise.py 1009B train.py 876B compute_model_l2_norm.py 770B custom_tensor_dataset.py 749B Net2nn.py 697B advanced_composition.py 635B noraml_composition.py 425B send_main_model_to_clients.py 328B ResNet.cpython-39.pyc 7KB dp_optimizer.cpython-39.pyc 7KB compute_rdp.cpython-39.pyc 6KB CNN.cpython-39.pyc 6KB vgg_bn.cpython-39.pyc 5KB get_data.cpython-39.pyc 4KB sampling.cpython-39.pyc 4KB dirichlet_nonIID_data.cpython-39.pyc 4KB pathological_nonIID_data.cpython-39.pyc 3KB local_clients_train_process.cpython-39.pyc 3KB compute_coordinates.cpython-39.pyc 3KB get_MaxSigma_or_MaxSteps.cpython-39.pyc 3KB classifer.cpython-39.pyc 3KB weight_initialization.cpython-39.pyc 2KB DNN.cpython-39.pyc 2KB compute_dp_sgd.cpython-39.pyc 2KB train_with_opacus.cpython-39.pyc 2KB rdp_convert_dp.cpython-39.pyc 2KB IID_data.cpython-39.pyc 1KB optimizier_and_model_distribution.cpython-39.pyc 1KB train_with_dp.cpython-39.pyc 1KB custom_tensor_dataset.cpython-39.pyc 1KB center_average_model_with_weights.cpython-39.pyc 1KB validation.cpython-39.pyc 1KB train.cpython-39.pyc 885B clipping_and_adding_noise.cpython-39.pyc 798B compute_model_l2_norm.cpython-39.pyc 685B send_main_model_to_clients.cpython-39.pyc 533B t10k-images-idx3-ubyte 7.48MB t10k-images-idx3-ubyte 7.48MB t10k-labels-idx1-ubyte 10KB t10k-labels-idx1-ubyte 10KB test_batch 29.6MB train-images-idx3-ubyte 44.86MB train-images-idx3-ubyte 44.86MB train-labels-idx1-ubyte 59KB train-labels-idx1-ubyte 59KB modules.xml 358B misc.xml 182B共 102 条
- 1
- 2
资源评论
生瓜蛋子
- 粉丝: 3824
- 资源: 5235
