A differentially private matrix factorization based on vector perturbation for recommender system

doi:10.1016/j.neucom.2022.01.079

CORC > 兰州理工大学 > 兰州理工大学

	A differentially private matrix factorization based on vector perturbation for recommender system
	Ran, Xun 1; Wang, Yong 1; Zhang, Leo Yu 2; Ma, Jun 3
刊名	Neurocomputing
	2022-04-28
卷号	483 页码:32-41
关键词	Benchmarking Data privacy Matrix algebra Matrix factorization Sensitivity analysis Differential privacies Error accumulation Factorization techniques Matrix factorizations Multiple iterations Objective functions Privacy preservation Privacy protection User data Vector perturbation
ISSN号	0925-2312
DOI	10.1016/j.neucom.2022.01.079
英文摘要	Matrix factorization (MF) techniques have yielded immense success in recommender systems (RSs). Since a huge amount of user data is collected and used in RS, it raises concerns about data privacy. As a strict privacy protection framework, many efforts attempt to apply Differential Privacy (DP) to MF. However, there are still some challenges or problems in designing MF with privacy preservation, such as error accumulation in the multiple iterations of MF, introduction of unnecessary noises, and difficult sensitivity analysis. To overcome these problems, we devise a vector perturbation-based differentially private matrix factorization (VP-DPMF). Our scheme can prevent error accumulation by perturbing the objective function of MF rather than its factorization process or results. It also addresses the difficulty of analyzing sensitivity by exploiting the polynomial representation of the objective function. Furthermore, our scheme can reduce unnecessary noises by controlling the perturbation within the vector term of the polynomial, and can preserve the convexity property of the original function. Theoretical analysis demonstrates that our scheme can achieve good performance in a large-scale recommender system. Experimental results on some benchmark datasets show that the proposed scheme can provide both rigid privacy guarantee and satisfactory recommendation quality. © 2022 Elsevier B.V.
WOS研究方向	Computer Science
语种	英语
出版者	Elsevier B.V.
WOS记录号	WOS:000761719500004
内容类型	期刊论文
源URL	[http://ir.lut.edu.cn/handle/2XXMBERH/157891]
专题	兰州理工大学
作者单位	1.Key Laboratory of Electronic Commerce and Logistics of Chongqing, Chongqing University of Posts and Telecommunications, Chongqing; 400065, China; 2.School of Information Technology, Deakin University, Waurn Ponds; VIC; 3216, Australia; 3.Department of Physics, Lanzhou University of Technology, Lanzhou; 730050, China
推荐引用方式 GB/T 7714	Ran, Xun,Wang, Yong,Zhang, Leo Yu,et al. A differentially private matrix factorization based on vector perturbation for recommender system[J]. Neurocomputing,2022,483:32-41.
APA	Ran, Xun,Wang, Yong,Zhang, Leo Yu,&Ma, Jun.(2022).A differentially private matrix factorization based on vector perturbation for recommender system.Neurocomputing,483,32-41.
MLA	Ran, Xun,et al."A differentially private matrix factorization based on vector perturbation for recommender system".Neurocomputing 483(2022):32-41.