Appears in collection : 2016 - T1 - WS5 - Secrecy and privacy theme

I will present some new, nearly-optimal lower bounds on the amount of data required to release differentially private statistics on high-dimensional datasets, both in information-theoretic and computational settings. These results show that there is a significant “price of differential privacy” in high-dimensional datasets. We prove these lower bounds using two closely-related cryptographic primitives fingerprinting codes (in information theoretic setting) and traitor-tracing schemes (in the computational setting) that we show are closely connected to differentially private data analysis. I will also discuss how these lower bounds are related to realistic attacks on released datasets.