ISBIS4 Abstract

Contact Author's Name: Alan F. Karr
Title of Abstract: Secure Statistical Analysis of Distributed Databases
Author(s): Alan F. Karr
Affiliation: NISS

A continuing need in the contexts from national security to business is for statistical analyses that \"integrate\" data stored in multiple, distributed databases. But, barriers to integrating databases are numerous, including confidentiality, regulation and scale.

For many analyses, however, it is not necessary actually to integrate the data. Instead, using techniques from computer science known generically as secure multi-party computation, the database holders can share analysis-specific sufficient statistics anonymously, but in a way that the desired analysis can be performed in a principled manner. The focus is on linear regression and related analyses for horizontally partitioned and vertically partitioned data.

Two methods for secure data integration and one for secure contingency tables will also be mentioned.