The confounding problem of private data release

The demands to make data available are growing ever louder, to support open data initiatives and "data monetization". But the problem of doing so without disclosing confidential information is a subtle and difficult one. Is "private data release" an oxymoron?

This talk delves into the motivations for data release, and highlights some of the pitfalls.   I'll conclude by describing some recent work on algorithms that meet a statistical guarantee of (differential) privacy, building on a 50-year old algorithm that tosses a single (biased) coin.