Differentially private data release and analytics Public

Abstract

Nowadays data sharing is important for application domains, such as scientific discoveries, business strategies, commercial interests, and social goods, especially when there are not enough local samples to test a hypothesis. However, data in its raw format are sensitive as they essentially contain individual specific information, and publishing such data without proper protection may disclose personal privacy. Netflix canceled their recommendation system contest because the released customers data can identify special individuals with high probability. In order to promote data sharing, it is important to develop privacy-preserving algorithms that respect data confidentiality while present data utility. In this dissertation, we address the privacy concerns in publishing highdimensional static data and dynamic datasets, and developing mechanisms for personalized differential privacy where data subjects can have various privacy preferences. Our privacy preserving algorithms satisfy differential privacy, a rigorous and de facto standard for privacy protection. Extensive empirical studies demonstrate the effectiveness of our solutions and confirm that our methods have great promise for privacy-preserving data release and analytical tasks in a wide range of applicationdomains.

Table of Contents

List of Figures

List of Tables

Chapter 1. Introduction

1.1. Motivation

1.2. Research Contributions

Chapter 2. Related Works

2.1. Differentially private synthetic data generation

2.2. Differentially private dynamic data generation

2.3. Personalized dierential privacy

Chapter 3. Differentially private synthesization of multi-dimensional data using

copula functions

3.1. Preliminaries

3.2. DPCopula

3.3. Experiment

Chapter 4. Privacy-preserving dynamic histogram release with distance-based sampling

4.1. Preliminaries

4.2. Adaptive Sampling Approach

4.3. Utility Analysis

4.4. Extensions to innite streams

4.5. Experiment

4.6. Conclusions

Chapter 5. Personalized differential privacy

5.1. Preliminaries

5.2. Partitioning mechanisms

5.3. Experiment

Chapter 6. Conclusions

6.1. Summary of Dissertation

6.2. Recommendations for Future Work

Bibliography

About this Dissertation

Rights statement

• Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.

School	Laney Graduate School
Department	Computer Science and Informatics
Degree	Ph.D.
Submission	Dissertation
Language	English
Research Field	Computer Science
Mot-clé	Dynamic data release Differential privacy Data synthesization Data sharing Data privacy
Committee Chair / Thesis Advisor	Xiong, Li, Emory University
Committee Members	Ho, Joyce, Emory University Jiang, Xiaoqian, Emory University Sunderam, Vaidy, Emory University

Dernière modification

Primary PDF

Thumbnail	Title	Date Uploaded	Actions
	Differentially private data release and analytics ()	2017-12-15 00:20:57 -0500	Press to Select an action Download

Supplemental Files

Thumbnail	Title	Date Uploaded	Actions