How much private information is "leaked" by revealing certain data? This question arises in several applications including side-channels, including wire-tap channels, and database privacy. Many metrics have been proposed to quantify leakage in such contexts. Most of these metrics, however, either lack a cogent operational justification or label obviously-insecure systems as secure.
We propose a new metric called "maximal leakage," defined as the multiplicative increase, upon observing the public data, of the probability of correctly guessing a randomized function of the private information, maximized over all such randomized functions. We provide an operational justification for this definition and show how it can be computed in practice. Among other findings, we show that mutual information
underestimates leakage while local differential privacy overestimates it.
This is joint work with Ibrahim Issa and Sudeep Kamath.
Aaron Wagner is an Associate Professor in the School of Electrical and Computer Engineering at Cornell University. He received the B.S. degree from the University of Michigan, Ann Arbor, and the M.S. and Ph.D. degrees from the University of California, Berkeley. During the 2005-2006 academic year, he was a Postdoctoral Research Associate in the Coordinated Science Laboratory at the University of Illinois at Urbana-Champaign and a Visiting Assistant Professor in the School of Electrical and Computer Engineering at Cornell. He has received the NSF CAREER award, the David J. Sakrison Memorial Prize from the U.C. Berkeley EECS Dept., the Bernard Friedman Memorial Prize in Applied Mathematics from the U.C. Berkeley Dept. of Mathematics, and teaching awards at the Department, College, and University level.