Accuracy-Privacy Trade-off in Analyzing Randomized Responses
We consider the problem of analyzing a global property of private data, where the privacy of individual respondent is protected through releasing randomized responses, subject to a certain rule. However, there is always a trade-off between global accuracy and individual privacy. Here we focus on the scenarios of answering YES/NO questions and derive an optimal solution to the trade-off problem, which aims at optimizing the estimation of the global property while maintaining a given level of privacy on the individual data.
READ FULL TEXT