Differentially Private Variational Inference for Non-conjugate Models

Joonas Jälkö, Onur Dikmen, Antti Honkela
Proceedings of the 33rd Conference on Uncertainty in Artificial Intelligence, PMLR R15:291-300, 2017.

Abstract

Many machine learning applications are based on data collected from people, such as their tastes and behaviour as well as biological traits and genetic data. Regardless of how impor- tant the application might be, one has to make sure individuals’ identities or the privacy of the data are not compromised in the analy- sis. Differential privacy constitutes a power- ful framework that prevents breaching of data subject privacy from the output of a com- putation. Differentially private versions of many important Bayesian inference methods have been proposed, but there is a lack of an efficient unified approach applicable to arbi- trary models. In this contribution, we pro- pose a differentially private variational infer- ence method with a very wide applicability. It is built on top of doubly stochastic varia- tional inference, a recent advance which pro- vides a variational solution to a large class of models. We add differential privacy into dou- bly stochastic variational inference by clipping and perturbing the gradients. The algorithm is made more efficient through privacy amplifi- cation from subsampling. We demonstrate the method can reach an accuracy close to non- private level under reasonably strong privacy guarantees, clearly improving over previous sampling-based alternatives especially in the strong privacy regime. $*$AH is also with the Department of Mathematics and Statistics and Department of Public Health, University of Helsinki.

Cite this Paper

BibTeX
@InProceedings{pmlr-vR15-jalko17a, title = {Differentially Private Variational Inference for Non-conjugate Models}, author = {J{\"a}lk{\"o}, Joonas and Dikmen, Onur and Honkela, Antti}, booktitle = {Proceedings of the 33rd Conference on Uncertainty in Artificial Intelligence}, pages = {291--300}, year = {2017}, editor = {Elidan, Gal and Kersting, Kristian}, volume = {R15}, series = {Proceedings of Machine Learning Research}, month = {11--15 Aug}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/r15/main/assets/jalko17a/jalko17a.pdf}, url = {https://proceedings.mlr.press/r15/jalko17a.html}, abstract = {Many machine learning applications are based on data collected from people, such as their tastes and behaviour as well as biological traits and genetic data. Regardless of how impor- tant the application might be, one has to make sure individuals’ identities or the privacy of the data are not compromised in the analy- sis. Differential privacy constitutes a power- ful framework that prevents breaching of data subject privacy from the output of a com- putation. Differentially private versions of many important Bayesian inference methods have been proposed, but there is a lack of an efficient unified approach applicable to arbi- trary models. In this contribution, we pro- pose a differentially private variational infer- ence method with a very wide applicability. It is built on top of doubly stochastic varia- tional inference, a recent advance which pro- vides a variational solution to a large class of models. We add differential privacy into dou- bly stochastic variational inference by clipping and perturbing the gradients. The algorithm is made more efficient through privacy amplifi- cation from subsampling. We demonstrate the method can reach an accuracy close to non- private level under reasonably strong privacy guarantees, clearly improving over previous sampling-based alternatives especially in the strong privacy regime. $*$AH is also with the Department of Mathematics and Statistics and Department of Public Health, University of Helsinki.}, note = {Reissued by PMLR on 04 October 2026.} }
Endnote
%0 Conference Paper %T Differentially Private Variational Inference for Non-conjugate Models %A Joonas Jälkö %A Onur Dikmen %A Antti Honkela %B Proceedings of the 33rd Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2017 %E Gal Elidan %E Kristian Kersting %F pmlr-vR15-jalko17a %I PMLR %P 291--300 %U https://proceedings.mlr.press/r15/jalko17a.html %V R15 %X Many machine learning applications are based on data collected from people, such as their tastes and behaviour as well as biological traits and genetic data. Regardless of how impor- tant the application might be, one has to make sure individuals’ identities or the privacy of the data are not compromised in the analy- sis. Differential privacy constitutes a power- ful framework that prevents breaching of data subject privacy from the output of a com- putation. Differentially private versions of many important Bayesian inference methods have been proposed, but there is a lack of an efficient unified approach applicable to arbi- trary models. In this contribution, we pro- pose a differentially private variational infer- ence method with a very wide applicability. It is built on top of doubly stochastic varia- tional inference, a recent advance which pro- vides a variational solution to a large class of models. We add differential privacy into dou- bly stochastic variational inference by clipping and perturbing the gradients. The algorithm is made more efficient through privacy amplifi- cation from subsampling. We demonstrate the method can reach an accuracy close to non- private level under reasonably strong privacy guarantees, clearly improving over previous sampling-based alternatives especially in the strong privacy regime. $*$AH is also with the Department of Mathematics and Statistics and Department of Public Health, University of Helsinki. %Z Reissued by PMLR on 04 October 2026.
APA
Jälkö, J., Dikmen, O. & Honkela, A.. (2017). Differentially Private Variational Inference for Non-conjugate Models. Proceedings of the 33rd Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R15:291-300 Available from https://proceedings.mlr.press/r15/jalko17a.html. Reissued by PMLR on 04 October 2026.

Related Material