Who is Alice? Privacy Risk, the Case of Regression (Extended Abstract)
Ashish Dandekar, Stéphane Bressan · 2018
While there seems to be no turning back from data becoming a commodity, news headlines and scandals recurrently exacerbate a general concern for privacy. Can Alice's data be collected, analysed and the results meaningfully published and shared without compromising Alice's privacy? How can organizations assess the privacy risk before disclosing data or the results of their analysis of data? We discuss two streams of research that study the assessment of privacy risk: statistical disclosure and differential privacy. We critically discuss their similarities and dissimilarities. We comparatively and empirically evaluate the two approaches with a canonical dataset and the scenario of synthetic data generation with linear regression..