|
Activity Number:
|
120
- Challenges and Recent Advances in Private Data Analysis
|
|
Type:
|
Topic-Contributed
|
|
Date/Time:
|
Monday, August 9, 2021 : 1:30 PM to 3:20 PM
|
|
Sponsor:
|
Section on Nonparametric Statistics
|
|
Abstract #317315
|
|
|
Title:
|
Differentially Private Statistics for Collaborative Neuroinformatics
|
|
Author(s):
|
Anand Sarwate*
|
|
Companies:
|
Rutgers University
|
|
Keywords:
|
differential privacy;
regression;
data visualization;
neuroinformatics
|
|
Abstract:
|
Neuroimaging research projects often suffer from small sample size due to the expense of data collection. However, many data sets have been already collected so there is a potential opportunity to perform joint analyses using the data from many existing studies. Privacy becomes a major concern in this setting. Differential privacy has emerged as a widely-used standard for quantifying the privacy risks from publishing the results of computations on sensitive data. We have been applying differential privacy in the context of learning statistical models jointly across data sets in the COINSTAC system for neuroinformatics. In this talk I will discuss the statistical methods and algorithms for such learning as well as practical considerations in designing this type of collaborative science platform.
|
Authors who are presenting talks have a * after their name.
