Unbiased Statistical Estimation and Valid Confidence Intervals Under Differential Privacy

arXiv:2110.14465v2 Announce Type: replace-cross
Abstract: We present a method for producing unbiased parameter estimates and valid confidence intervals under the constraints of differential privacy, a formal framework for limiting individual information leakage from sensitive data. Prior work in this area is limited in that it is tailored to calculating confidence intervals for specific statistical procedures, such as mean estimation or simple linear regression. While other recent work can produce confidence intervals for more general sets of procedures, they either yield …

area arxiv constraints cs.cr data differential privacy framework information information leakage math.st parameter privacy producing sensitive sensitive data stat.me stat.th under valid work