Online Program

Pinning Down "Privacy" in Statistical Databases
*Adam Smith, The Pennsylvania State University 


We describe "differential privacy", a notion which emerged from a recent line of work in theoretical computer science that seeks to formulate and satisfy rigorous definitions of privacy for "statistical databases", that is, data sets for which one wants to publish aggregate statistical information without compromising individuals' privacy. We also sketch some basic techniques for achieving differential privacy. The techniques are reminiscent of (but different from) those used in noise-tolerant machine learning and robust statistics.