Data-Dependent Differentially Private Parameter Learning for Directed Graphical Models

Amrita Roy Chowdhury, Theodoros Rekatsinas, Somesh Jha
Proceedings of the 37th International Conference on Machine Learning, PMLR 119:1939-1951, 2020.

Abstract

Directed graphical models (DGMs) are a class of probabilistic models that are widely used for predictive analysis in sensitive domains such as medical diagnostics. In this paper, we present an algorithm for differentially-private learning of the parameters of a DGM. Our solution optimizes for the utility of inference queries over the DGM and \emph{adds noise that is customized to the properties of the private input dataset and the graph structure of the DGM}. To the best of our knowledge, this is the first explicit data-dependent privacy budget allocation algorithm in the context of DGMs. We compare our algorithm with a standard data-independent approach over a diverse suite of benchmarks and demonstrate that our solution requires a privacy budget that is roughly $3\times$ smaller to obtain the same or higher utility.

Cite this Paper


BibTeX
@InProceedings{pmlr-v119-chowdhury20a, title = {Data-Dependent Differentially Private Parameter Learning for Directed Graphical Models}, author = {Chowdhury, Amrita Roy and Rekatsinas, Theodoros and Jha, Somesh}, booktitle = {Proceedings of the 37th International Conference on Machine Learning}, pages = {1939--1951}, year = {2020}, editor = {Hal Daumé III and Aarti Singh}, volume = {119}, series = {Proceedings of Machine Learning Research}, month = {13--18 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v119/chowdhury20a/chowdhury20a.pdf}, url = { http://proceedings.mlr.press/v119/chowdhury20a.html }, abstract = {Directed graphical models (DGMs) are a class of probabilistic models that are widely used for predictive analysis in sensitive domains such as medical diagnostics. In this paper, we present an algorithm for differentially-private learning of the parameters of a DGM. Our solution optimizes for the utility of inference queries over the DGM and \emph{adds noise that is customized to the properties of the private input dataset and the graph structure of the DGM}. To the best of our knowledge, this is the first explicit data-dependent privacy budget allocation algorithm in the context of DGMs. We compare our algorithm with a standard data-independent approach over a diverse suite of benchmarks and demonstrate that our solution requires a privacy budget that is roughly $3\times$ smaller to obtain the same or higher utility.} }
Endnote
%0 Conference Paper %T Data-Dependent Differentially Private Parameter Learning for Directed Graphical Models %A Amrita Roy Chowdhury %A Theodoros Rekatsinas %A Somesh Jha %B Proceedings of the 37th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2020 %E Hal Daumé III %E Aarti Singh %F pmlr-v119-chowdhury20a %I PMLR %P 1939--1951 %U http://proceedings.mlr.press/v119/chowdhury20a.html %V 119 %X Directed graphical models (DGMs) are a class of probabilistic models that are widely used for predictive analysis in sensitive domains such as medical diagnostics. In this paper, we present an algorithm for differentially-private learning of the parameters of a DGM. Our solution optimizes for the utility of inference queries over the DGM and \emph{adds noise that is customized to the properties of the private input dataset and the graph structure of the DGM}. To the best of our knowledge, this is the first explicit data-dependent privacy budget allocation algorithm in the context of DGMs. We compare our algorithm with a standard data-independent approach over a diverse suite of benchmarks and demonstrate that our solution requires a privacy budget that is roughly $3\times$ smaller to obtain the same or higher utility.
APA
Chowdhury, A.R., Rekatsinas, T. & Jha, S.. (2020). Data-Dependent Differentially Private Parameter Learning for Directed Graphical Models. Proceedings of the 37th International Conference on Machine Learning, in Proceedings of Machine Learning Research 119:1939-1951 Available from http://proceedings.mlr.press/v119/chowdhury20a.html .

Related Material