Rigorous Metrics for Emerging Data Privacy Challenges
Abstract
The modern day technological economy hinges on the collection, analysis and usage of personal data, which is analyzed to drive decisions and make profits. From a social perspective, however, these data driven decision systems pose severe and unique privacy challenges that current privacy solutions are incapable of addressing adequately. This proposal addresses three such major privacy challenges. First, we consider preserving the privacy of location trajectories in the presence of prior information. Second, we consider datadeletion from machine learning models learned via empirical risk minimization. Finally, we consider auditing machine learning models to determine if a specific persons data was used to build it. All these privacy challenges are addressed by proposing unique privacy metrics with rigorous goals.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Apr 29, 2020
- Source ID
- N000142012334
Entities
People
- Kamalika Chaudhuri
Organizations
- Office of Naval Research
- United States Navy
- University of California, San Diego