[Colloq] Talk: Interference-Based Privacy Guarantees for Differentially Private Mechanisms | Robert Kleinberg, Cornell University and Microsoft Research New England | Tuesday, November 10, 12noon | 366 WVH

Walker, Lashauna la.walker at neu.edu
Thu Nov 5 16:30:04 EST 2015


CCIS Colloquium and Theory Seminar

SPEAKER: Robert Kleinberg
         Cornell University and Microsoft Research New England

TITLE: Inference-Based Privacy Guarantees for Differentially Private Mechanisms

TIME: Tuesday, Nov 10, 12 noon

PLACE: 366 WVH

ABSTRACT:

How much information can be learnt about an individual by observing the outcome of a computation? If the computation is differentially private, the answer is: "not much more than if the individual's data had been excluded from the input." A stronger notion of privacy, originally propounded by Dalenius in the 1970's, requires instead that it should not be possible to learn much more about an individual than if the outcome of the computation had never been revealed. Simple examples, as well as a general impossibility theorem of Dwork and Naor, preclude the possibility of supplying this stronger "inferential" guarantee against attackers with arbitrary auxiliary information, assuming the computation is at least minimally useful.

In this talk we revisit the notion of inferential privacy and ask: under what limitations on the adversary's side information can we deduce an inferential guarantee from a differential one? We model the adversary's side information as a prior distribution over datasets (or, more generally, a set of possible priors) and prove two main results. The first result pertains to distributions that satisfy a natural positive-affiliation condition, and gives an upper bound on the inferential privacy guarantee for any differentially private mechanism. This upper bound is matched by a simple mechanism that adds Laplace noise to the sum of the data. The second result pertains to distributions that have weak correlations, defined in terms of a suitable "influence matrix". The result provides an upper bound for inferential privacy in terms of the differential privacy parameter and the spectral norm of this matrix.

BIO:

Bobby Kleinberg <http://www.cs.cornell.edu/~rdk/> is an Associate Professor of Computer Science at Cornell University and a Principal Researcher at Microsoft Research New England. His research studies the design and analysis of algorithms, and their applications to economics, networking, information retrieval, and other areas. Prior to receiving his doctorate from MIT in 2005, Kleinberg spent three years at Akamai Technologies, where he assisted in designing the world's largest Internet Content Delivery Network. He is the recipient of a Microsoft Research New Faculty Fellowship, an Alfred P. Sloan Foundation Fellowship, and an NSF CAREER Award.


Thank You.

LaShauna Walker
Executive Assistant to Dean Carla Brodley
College of Computer and Information Science
Northeastern University
617-373-5204
Facebook<https://www.facebook.com/ccisatnu?ref=hl> | Instagram<https://instagram.com/ccisatnu/> | LinkedIn<https://www.linkedin.com/groups/Northeastern-University-College-Computer-Information-1943637?gid=1943637&mostPopular=&trk=tyah&trkInfo=idx%3A1-1-1%2CtarId%3A1426606862845%2Ctas%3ANortheastern+University+College+of+Com> | Twitter<https://twitter.com/CCISatNU>



More information about the Colloq mailing list