all InfoSec news
Bounding Training Data Reconstruction in Private (Deep) Learning. (arXiv:2201.12383v4 [cs.LG] UPDATED)
June 24, 2022, 1:20 a.m. | Chuan Guo, Brian Karrer, Kamalika Chaudhuri, Laurens van der Maaten
cs.CR updates on arXiv.org arxiv.org
Differential privacy is widely accepted as the de facto method for preventing
data leakage in ML, and conventional wisdom suggests that it offers strong
protection against privacy attacks. However, existing semantic guarantees for
DP focus on membership inference, which may overestimate the adversary's
capabilities and is not applicable when membership status itself is
non-sensitive. In this paper, we derive the first semantic guarantees for DP
mechanisms against training data reconstruction attacks under a formal threat
model. We show that two …
More from arxiv.org / cs.CR updates on arXiv.org
Jobs in InfoSec / Cybersecurity
SOC 2 Manager, Audit and Certification
@ Deloitte | US and CA Multiple Locations
Advisory Red Consultant
@ Security Risk Advisors | Philadelphia, Pennsylvania, United States
Cyber Business Transformation Change Analyst
@ National Grid | Warwick, GB, CV34 6DA
Cyber Security Analyst
@ Ford Motor Company | Mexico City, MEX, Mexico
Associate Administrator, Cyber Security Governance (Fort Myers)
@ Millennium Physician Group | Fort Myers, FL, United States
Embedded GSOC Lead Operator, Events
@ Sibylline Ltd | Seattle, WA, United States