SoK: Chasing Accuracy and Privacy, and Catching Both in Differentially Private Histogram Publication

Histograms and synthetic data are of key importance in data analysis.However, researchers have shown that even aggregated data such as histograms,containing no obvious sensitive attributes, can result in privacy leakage. Toenable data analysis, a strong notion of privacy is required to avoid riskingunintended privacy violations. Such a strong notion of privacy is differential privacy, a statistical notionof privacy that makes privacy leakage quantifiable. The caveat regardingdifferential privacy is that while it has strong guarantees for privacy,privacy comes at a cost of accuracy. Despite this trade off being a central andimportant issue in the adoption of differential privacy, there exists a gap inthe literature regarding providing an understanding of the trade off and how toaddress it appropriately. Through a systematic literature review (SLR), we investigate thestate-of-the-art within accuracy improving differentially private algorithmsfor histogram and synthetic data publishing. Our contribution is two-fold: 1)we identify trends and connections in the contributions to the field ofdifferential privacy for histograms and synthetic data and 2) we provide anunderstanding of the privacy/accuracy trade off challenge by crystallizingdifferent dimensions to accuracy improvement. Accordingly, we position andvisualize the ideas in relation to each other and external work, anddeconstruct each algorithm to examine the building blocks separately with theaim of pinpointing which dimension of accuracy improvement eachtechnique/approach is targeting. Hence, this systematization of knowledge (SoK)provides an understanding of in which dimensions and how accuracy improvementcan be pursued without sacrificing privacy.