Author: Norm Matloff (University of California, Davis, USA)
Given a medical database, how does one allow access by medical researchers while preserving patient privacy? How about a similar dilemma in analysis in an employment discrimination legal case? Data privacy has been an area of active research from the 1980s to the present, in both the statistics and computer science communities. As in the machine learning case, computer scientists and statisticians have tended to approach the privacy issue from quite different philosophies. This talk will review the history and present status of research in this field, and discuss problems with the main proposed solutions. It will also propose some new solutions, including some related to Small Area Estimation.