Statistics Seminar Series
Thursday, April 21st, 4:10pm, MSB 1147 (Colloquium Room)
Refreshments at 3:30pm in MSB 4110 (Statistics Lounge)
Speaker: Bimal Sinha (UMBC/CDAR/CENSUS BUREAU)*
Title: “An overview of data analysis under confidentiality protection -
analysis of noise multiplied & synthetic data”
Abstract: Statistical agencies such as the US Census Bureau routinely collect data under law, parts of which are highly sensitive and hence require privacy protection before the data can be released to the public. Creating model-based synthetic data is a very common method to hide real data, but its exact analysis is far from being simple. Likewise, noise multiplication is another simple idea to create unreal data. We will discuss exact data analysis methods under both noise multiplication & two different versions of synthetic data. The entire analysis is non-asymptotic in nature.
*UMBC: University of Maryland, Baltimore County
*CDAR: Center for Disclosure Avoidance Research