Maintaining Statistical Summaries over Dynamic Data

PhD Qualifying Examination


Title: "Maintaining Statistical Summaries over Dynamic Data"

by

Mr. Yuan QIU


Abstract:

Since the introduction of Morris Counter in 1977, decades of research have been 
devoted to summary maintenance over dynamic data. For many problems, effcient 
summaries have been proposed that occupy small space while providing strong 
accuracy guarantees. The most important ones are statistical summaries for 
distinct count, frequency estimation, heavy hitter and quantile problems. They 
are discussed under various models, including cash register streams, turnstile 
streams, sliding windows and distributed streams. While the streaming context 
is almost well-understood by matching bounds for many problems, new directions 
arise in applications of summaries and their ideas. One of them is differential 
privacy, which guarantees the privacy of any user is not compromised by any 
post-processing of outputs. Several summaries have been applied or extended to 
work under privacy constraints. In this survey, we review the literature of 
maintaining statistical information over dynamic data, and propose possible 
directions for future research.


Date:			Friday, 16 April 2021

Time:                  	4:00pm - 6:00pm

Zoom meeting: 		https://hkust.zoom.us/j/7071528447

Committee Members:	Prof. Ke Yi (Supervisor)
 			Dr. Sunil Arya (Chairperson)
 			Prof. Siu-Wing Cheng
 			Prof. Mordecai Golin


**** ALL are Welcome ****