Distributed Differentially Private Algorithms for Matrix and Tensor Factorization

TY - JOUR

T1 - Distributed Differentially Private Algorithms for Matrix and Tensor Factorization

AU - Imtiaz, Hafiz

AU - Sarwate, Anand D.

N1 - Publisher Copyright: © 2018 IEEE.

PY - 2018/12

Y1 - 2018/12

N2 - In many signal processing and machine learning applications, datasets containing private information are held at different locations, requiring the development of distributed privacy-preserving algorithms. Tensor and matrix factorizations are key components of many processing pipelines. In the distributed setting, differentially private algorithms suffer because they introduce noise to guarantee privacy. This paper designs new and improved distributed and differentially private algorithms for two popular matrix and tensor factorization methods: principal component analysis and orthogonal tensor decomposition. The new algorithms employ a correlated noise design scheme to alleviate the effects of noise and can achieve the same noise level as the centralized scenario. Experiments on synthetic and real data illustrate the regimes in which the correlated noise allows performance matching with the centralized setting, outperforming previous methods and demonstrating that meaningful utility is possible while guaranteeing differential privacy.

AB - In many signal processing and machine learning applications, datasets containing private information are held at different locations, requiring the development of distributed privacy-preserving algorithms. Tensor and matrix factorizations are key components of many processing pipelines. In the distributed setting, differentially private algorithms suffer because they introduce noise to guarantee privacy. This paper designs new and improved distributed and differentially private algorithms for two popular matrix and tensor factorization methods: principal component analysis and orthogonal tensor decomposition. The new algorithms employ a correlated noise design scheme to alleviate the effects of noise and can achieve the same noise level as the centralized scenario. Experiments on synthetic and real data illustrate the regimes in which the correlated noise allows performance matching with the centralized setting, outperforming previous methods and demonstrating that meaningful utility is possible while guaranteeing differential privacy.

KW - Differential privacy

KW - distributed orthogonal tensor decomposition

KW - distributed principal component analysis

KW - latent variable model

UR - https://www.scopus.com/pages/publications/85055726994

UR - https://www.scopus.com/pages/publications/85055726994#tab=citedBy

U2 - 10.1109/JSTSP.2018.2877842

DO - 10.1109/JSTSP.2018.2877842

M3 - Article

SN - 1932-4553

VL - 12

SP - 1449

EP - 1464

JO - IEEE Journal on Selected Topics in Signal Processing

JF - IEEE Journal on Selected Topics in Signal Processing

IS - 6

M1 - 8509100

ER -