BEGIN:VCALENDAR
METHOD:PUBLISH
VERSION:2.0
PRODID:-//Cornell U. Department of Computer Science//Brown Bag Seminar//EN
BEGIN:VEVENT
SUMMARY:Brown bag: Aaron Wagner
DESCRIPTION:Title: An Operational Measure of Information Leakage in Side
	 Channels\nSpeaker: Aaron Wagner\nAbstract: How much information is
	 \"leaked\" in a side channel?  Despite decades\nof work on these
	 channels\, including the development of many\nsophisticated mitigation
	 mechanisms for specific side channels\, the\nfundamental question of how
	 to measure the key quantity of\ninterest---leakage---has received
	 surprisingly little attention.\nMany metrics have been used in the
	 literature\, but these metrics\neither lack a cogent operational
	 justification or mislabel systems\nthat are obviously insecure as
	 secure.\n\nWe propose a new metric called \"maximal leakage\,\" defined
	 as the\nlogarithm of the multiplicative increase\, upon observing the
	 public\ndata\, of the probability of correctly guessing a randomized
	 function of\nthe private information\, maximized over all such
	 randomized functions.\nWe provide an operational justification for this
	 definition\, show how it\ncan be computed in practice\, and discuss how
	 it relates to existing\nmetrics\, including mutual information\, local
	 differential privacy\, and\na certain under-appreciated metric in the
	 computer science literature.\nWe also present some structural results
	 for optimal mechanisms under\nthis metric. Among other findings\, we
	 show that mutual information\nunderestimates leakage while local
	 differential privacy overestimates\nit.\n\nThis is joint work with
	 Ibrahim Issa\, Sudeep Kamath\, Ben Wu\, and Ed Suh.
LOCATION:Gates 122
UID:2019-03-19
STATUS:TENTATIVE
DTSTART:20190319T160000Z
DTEND:20190319T170000Z
LAST-MODIFIED:20190319T122339Z
ORGANIZER;CN=Jonathan Shi:http://www.cs.cornell.edu/~jshi/brownbag/
DTSTAMP:20260408T153629Z
END:VEVENT
END:VCALENDAR