BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Isaac Newton Institute Seminar Series
SUMMARY:The Privacy of the Analyst and The Power of the St
ate - Naor\, M (Weizmann Institute of Science)
DTSTART;TZID=Europe/London:20120413T090000
DTEND;TZID=Europe/London:20120413T100000
UID:TALK37463AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/37463
DESCRIPTION:Differential privacy is a paradigm in privacy that
is aimed at mitigating the drawbacks of tradition
al anonymization techniques\, as it provides a rig
orous guarantee for the added risk to an individua
l in participating in a database. Roughly speaking
\, a mechanism satisfies differential privacy if f
or any possible output of the mechanism and any po
ssible set of data on individuals\, the probabilit
y of obtaining this particular output changes only
very little with the addition or deletion of the
data on an individual. We initiate the study of pr
ivacy for the analyst in differentially private da
ta analysis. That is\, not only are we concerned w
ith ensuring differential privacy for the data (i.
e. individuals or customers)\, which are the usual
concern of differential privacy\, but we also con
sider (differential) privacy for the set of querie
s posed by each data analyst. The privacy achieved
is with respect to other analysts which are the u
sers of the system. This problem of analysts' quer
ies being leaked arises only in the context of sta
teful privacy mechanisms\, in which the responses
to queries depend on other queries posed. A recent
wave of results in the area of differential priva
cy utilized coordinated noise and state in order t
o allow answering hugely many queries. We argue th
e problem is real by proving an exponential gap be
tween the number of queries that can be answered (
with non-trivial error) by stateless and stateful
differentially private mechanisms. We then give a
stateful algorithm for differentially private data
analysis that also ensures differential privacy f
or the analyst and can answer exponentially many q
ueries. Joint work with Cynthia Dwork and Salil Va
dhan
LOCATION:Seminar Room 1\, Newton Institute
CONTACT:Mustapha Amrani
END:VEVENT
END:VCALENDAR