The Community for Technology Leaders
2018 17th IEEE International Conference On Trust, Security And Privacy In Computing And Communications/ 12th IEEE International Conference On Big Data Science And Engineering (TrustCom-BigDataSE) (2018)
New York, NY, USA
Aug 1, 2018 to Aug 3, 2018
ISSN: 2324-9013
ISBN: 978-1-5386-4388-4
pp: 1628-1631
ABSTRACT
Recovering a decentralized low-rank matrix from an incomplete set of its entries is one of great research interests. Privacy makes our issue difficult. In this paper, we propose a novel scheme that allows analysts to perform great aggregate analysis while guaranteeing meaningful protection of each individuals privacy. Differential privacy aims to ensure means to maximize the accuracy of queries from statistical databases while minimizing the probabilities of identifying its records. With adding Gaussian noise, we are able to achieve this goal. First, we present an algorithm for private matrix completion. Secondly, we provide theoretical results for required Gaussian noise. Finally, we compare the performance of the proposed algorithm with the state-of-the-art, while both achieves the same level of differential privacy.
INDEX TERMS
data privacy, Gaussian noise, matrix decomposition
CITATION

H. Zhou, X. Liu, C. Fu, C. Shang and X. Chang, "Differentially Private Matrix Completion via Distributed Matrix Factorization," 2018 17th IEEE International Conference On Trust, Security And Privacy In Computing And Communications/ 12th IEEE International Conference On Big Data Science And Engineering (TrustCom-BigDataSE), New York, NY, USA, 2018, pp. 1628-1631.
doi:10.1109/TrustCom/BigDataSE.2018.00239
439 ms
(Ver 3.3 (11022016))