User-level Differential Privacy

PhD Qualifying Examination


Title: "User-level Differential Privacy"

by

Miss Juanru FANG


Abstract:

Differential privacy (DP) has become the mainstream privacy standard due 
to its strong protection of individual user's information. It requires 
that people cannot tell from the output whether a particular user's data 
is in the database instance or not. While most existing work considers 
tuple-level DP (or tuple-DP), where each user contributes exactly one 
tuple in the instance, recently, more attention has been paid on 
user-level DP (or user-DP), where each user can contribute an arbitrary 
number of tuples. User-DP is a more general and practical notion that can 
be applied to most real-world databases. Work on user-DP has focused on 
problems including sum estimation, degree distribution publication, and 
machine learning. In this survey, we review the existing work and discuss 
some potential research directions under user-DP.


Date:  			Tuesday, 24 May 2022

Time:                  	4:30pm - 6:30pm

Zoom Meeting: 		https://hkust.zoom.us/j/2470427001

Committee Members:	Prof. Ke Yi (Supervisor)
 			Prof. Xiaofang Zhou (Chairperson)
 			Prof. Siu-Wing Cheng
 			Dr. Dimitris Papadopoulos


**** ALL are Welcome ****