BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Analog vs. Digital Epsilons: Implementation Considerations for Dif
 ferential Privacy - Olya Ohrimenko\, University of Melbourne
DTSTART:20220819T150000Z
DTEND:20220819T160000Z
UID:TALK177041@talks.cam.ac.uk
CONTACT:Kieron Ivy Turk
DESCRIPTION:Differential privacy (DP) provides a rigorous framework for re
 leasing data statistics while bounding information leakage. It is currentl
 y a de facto privacy framework that has received significant interest from
  the research community and has been deployed by the U.S. Census Bureau\, 
 Apple\, Google\, Microsoft\, and others. However\, DP analysis often assum
 es a perfect computing environment and building blocks such as random nois
 e distribution samplers. Unfortunately\, a naive implementation of DP mech
 anisms can invalidate their theoretical guarantees.\n\nIn this talk\, I wi
 ll highlight two attacks based on implementation flaws in the noise genera
 tion commonly used in DP systems: floating-point representation attack aga
 inst continuous distributions and timing attacks against discrete distribu
 tions. I will then show that several state-of-the-art implementations of D
 P are susceptible to these attacks as they allow one to learn the values b
 eing protected by DP. Our evaluation demonstrates success rates of 92.56% 
 for floating-point attacks in a machine learning setting and 99.65% for en
 d-to-end timing attacks on private sum. I will conclude with suggested mit
 igations\, emphasising that a careful implementation of DP systems may be 
 as important as it is for cryptographic libraries.\n\nThe talk is based on
  joint work with Jiankai Jin (The University of Melbourne)\, Eleanor McMur
 try (ETH Zurich) and Benjamin Rubinstein (The University of Melbourne)\, t
 hat appeared in IEEE Symposium on Security and Privacy 2022.\n\nBio: Olya 
 Ohrimenko is an Associate Professor at The University of Melbourne which s
 he joined in 2020. Prior to that she was a Principal Researcher at Microso
 ft Research in Cambridge\, UK\, where she started as a Postdoctoral Resear
 cher in 2014. Her research interests include privacy and integrity of mach
 ine learning algorithms\, data analysis tools and cloud computing\, includ
 ing topics such as differential privacy\, verifiable and data-oblivious co
 mputation\, trusted execution environments\, side-channel attacks and miti
 gations. Recently Olya has worked with the Australian Bureau of Statistics
  and National Bank Australia. She has received solo and joint research gra
 nts from Facebook and Oracle and is currently a PI on an AUSMURI grant.
LOCATION:Webinar &amp\; FW11\, Computer Laboratory\, William Gates Buildin
 g.
END:VEVENT
END:VCALENDAR