English

Obfuscation via Information Density Estimation

Information Theory 2019-10-21 v1 Machine Learning math.IT Machine Learning

Abstract

Identifying features that leak information about sensitive attributes is a key challenge in the design of information obfuscation mechanisms. In this paper, we propose a framework to identify information-leaking features via information density estimation. Here, features whose information densities exceed a pre-defined threshold are deemed information-leaking features. Once these features are identified, we sequentially pass them through a targeted obfuscation mechanism with a provable leakage guarantee in terms of Eγ\mathsf{E}_\gamma-divergence. The core of this mechanism relies on a data-driven estimate of the trimmed information density for which we propose a novel estimator, named the trimmed information density estimator (TIDE). We then use TIDE to implement our mechanism on three real-world datasets. Our approach can be used as a data-driven pipeline for designing obfuscation mechanisms targeting specific features.

Keywords

Cite

@article{arxiv.1910.08109,
  title  = {Obfuscation via Information Density Estimation},
  author = {Hsiang Hsu and Shahab Asoodeh and Flavio du Pin Calmon},
  journal= {arXiv preprint arXiv:1910.08109},
  year   = {2019}
}

Comments

24 pages, 3 figures

R2 v1 2026-06-23T11:47:09.330Z