On the Robustness of Information-Theoretic Privacy Measures and Mechanisms

Consider a data publishing setting for a dataset composed by both private and non-private features. The publisher uses an empirical distribution, estimated from $n$ i.i.d. samples, to design a privacy mechanism which is applied to new fresh samples afterward. In this paper, we study the discrepancy...

Descrición completa

Detalles Bibliográficos

Autores principales:	Diaz, Mario, Wang, Hao, Calmon, Flavio P., Sankar, Lalitha
Formato:	Artigo
Publicado:	2018
Materias:	Information Theory
Acceso en liña:	https://arxiv.org/abs/1811.06057 http://arxiv.org/abs/1811.06057
Etiquetas:	Engadir etiqueta Sen Etiquetas, Sexa o primeiro en etiquetar este rexistro!