submission voting
voting is closed.
Bounding Utility Loss via Classifiers

Using classification models to evaluate the usefulness of anonymized data without losing simplicity, explainability or generality.
Short Description

If a classifier can easily distinguish between privatized and ground truth data, the datasets are fundamentally different, and the privatized data should not be used for downstream analysis. Conversely, if a classifier cannot distinguish them, we should feel comfortable using the privatized data going forward. In the latter case, we prove that any classifier from the same function family will have essentially the same loss on your private and ground truth data.

Metric Upload
NIST -- Separability.pdf

comments (public)

  • User avatar
    Jan. 20, 2021, 10:58 a.m. PST
    You might be interested in the paper by Woo et al. (2009) in the Journal of Privacy and Confidentiality, which proposes a similar measure.