Intro
Preface
Organization
Contents
Privacy Models
An Optimization-Based Decomposition Heuristic for the Microaggregation Problem
1 Introduction
2 The Decomposition Heuristic
3 The Local Search Improvement Heuristic
4 The Mixed Integer Linear Optimization Algorithm Based on Column Generation
5 Computational Results
6 Conclusions
References
Privacy Analysis with a Distributed Transition System and a Data-Wise Metric
1 Introduction
2 Distributed Labeled-Tagged Transition Systems
3 -Indistinguishability, -Local-Differential Privacy

4 -Differential Privacy
5 Comparing Two Nodes on One or More Runs
6 New Metric for Indistinguishability and DP
7 Related Work and Conclusion
References
Multivariate Mean Comparison Under Differential Privacy
1 Introduction
2 Mathematical Background
2.1 Statistical Tests for Two Samples
2.2 Hotelling's t2-Test
2.3 Differential Privacy
3 Privatized Mean Comparison
3.1 Privatization of the t2-Statistic
3.2 Bootstrap
4 Simulation
5 Conclusion
A Proofs
B Effects of Privatization
Example
C Algorithms
References

Asking the Proper Question: Adjusting Queries to Statistical Procedures Under Differential Privacy
1 Introduction
1.1 Setting
1.2 Our Contribution
1.3 Related Work
2 Fixed (Non-random) Datasets
2.1 Confidence Regions
2.2 Testing Hypotheses: Likelihood-Ratio Test
3 Random, Normally Distributed Data
3.1 Confidence Regions
3.2 Testing Hypotheses: Likelihood-Ratio Test
4 Numerical Example
5 Appendix
References
Towards Integrally Private Clustering: Overlapping Clusters for High Privacy Guarantees
1 Introduction
2 Preliminaries
2.1 Integral Privacy

2.2 k-Anonymity, Microaggregation, and MDAV
2.3 Genetic Algorithms
3 -Centroid c-Means
3.1 Formalization
3.2 Properties
4 Experiments
4.1 Solving the Optimization Problem
4.2 Datasets
4.3 Parameters
4.4 Results
5 Conclusions
References
II Tabular Data
Perspectives for Tabular Data Protection
How About Synthetic Data?
1 Introduction
2 Recalling the Methods under Consideration
2.1 Synthetic Data
2.2 Targeted Record Swapping (TRS)
2.3 CKM Noise Design for Tabulations of Continuous Variables
3 Study Design
3.1 Test Data

3.2 Application Settings for Synthetic Data Generation
3.3 Application Settings for Targeted Record Swapping
3.4 Application Settings for the Cell Key Method
4 Measuring Utility and Disclosure Risk
5 Results
5.1 Comparing Utility Loss
6 Summary, Open Issues, Conclusions
Appendix
A.1 Approximate Behavior of Utility Loss Indicator U for CKM in Extremely Detailed Tabulations
Appendix A.2
Appendix A.3
References
On Privacy of Multidimensional Data Against Aggregate Knowledge Attacks
1 Introduction
2 Related Work
3 Problem Statement
3.1 Preliminaries