Different strategies for differentially private histogram publication

Social networks are extensively exploited by third-party consumers such as researchers and advertisers to understand user characteristics and behaviors. In general, before network data is published, sensitive relationships should be anonymized to prevent the compromise of individual privacy. To quantify the guarantee level of privacy-preserving mechanisms and mitigate users' privacy concerns, numerous studies concerning network data de-anonymization have been carried out. However, most existing studies focus on single-view data, and privacy protection for multi-view data that is ubiquitous in the era of big data has not been yet extensively explored. In this study, we are interested in answering the following question: Are the traditional privacy protection methods still valid for the anonymization of multi-view data? In this study, we propose a Multi-View Low-Rank Coding (MVLRC) based network data de-anonymization framework to assess the vulnerability of privacy protection techniques by accurately reconstructing a large portion of the original data. Specifically, the framework assumes that in principle, the target and auxiliary networks have common structural patterns, and they can be modeled together to infer the hidden structure of the target network. The essential components of our work include the following: (1) a robust network representation model for structural pattern learning; (2) the network representation based multiview modeling of target network and auxiliary network; (3) the inference of the anonymized links via target network reconstruction. Experimental results on synthetic networks and three real-world networks demonstrate that auxiliary networks can be utilized by malicious adversaries for privacy inference attacks. Thus, the privacy protection of multi-view network data needs more sophisticated anonymization techniques. INDEX TERMS Privacy preserving, de-anonymization, network data, multi-view learning, low-rank coding.

Section: Related Work a Network Privacy Preservationmentioning

confidence: 99%

Multi-View Low-Rank Coding-Based Network Data De-Anonymization

Xian

Qiao

et al. 2020

“…However, most of the research in regression analysis with differential privacy focus on the construction of parametric models. In addition, some researchers designed the non-parametric model, such as histogram publishing [26], [27] and gaussian process regression [28], [29]. In some specific scenarios, they can also have a good performance.…”

Section: A Regression Analysis With Differential Privacymentioning

confidence: 99%

Regression Analysis With Differential Privacy Preserving

Yang

2019

In the field of data mining, protecting sensitive data from being leaked is part of the focuses of current research. As a strict and provable definition of privacy model, differential privacy provides an excellent solution to the problem of privacy leakage. Numerous methods have been suggested to enforce differential privacy in various data mining tasks, such as regression analysis. However, existing solutions for regression analysis is less than satisfactory since the amount of noise added is excessive. What's worse, the adversary can launch model inversion attacks to infer sensitive information with the published regression model. Motivated by this, we propose a differential privacy budget allocation model. We optimize the regression model by adjusting the privacy budget allocation within the objective function. Extensive evaluation results show the superiority of the proposed model in terms of noise reduction, model inversion attack proof, and the trade-off between privacy protection and data utility. INDEX TERMS Machine learning, differential privacy, regression analysis, model inversion attack.

“…The basic idea of generalization based methods is to replace the sensitive information with a less specific but semantically consistent value [51]. In addition, the theoretical methods are the differential privacy based methods [52], [53].…”

Section: Privacy Protectionmentioning

confidence: 99%

Structural Predictability Optimization Against Inference Attacks in Data Publishing

Ming

Xian

et al. 2019

Graphs have been proved to be a useful mathematical representation for a broad variety of real-world complex systems, and the structure prediction on graphs refers to estimating the potential relationship between the objects from the observed structures, being fundamental in many data analysis applications, such as network alignment, network reconstruction, and link prediction. Accordingly, in data publishing, it is necessary to regulate the structural predictability of graphs against inference attack to protect the sensitive information of the data generators. In contrast to the existing works about graph structure perturbation for node ranking, information diffusion, and so on, the structural predictability optimization problem, i.e., reducing the accuracy of sensitive relationships inference in graphs, has not been extensively studied. This paper presents an active learning algorithm that selects the most representative links to be perturbed, thus regulating the structural predictability of graphs, that is, removing as few as possible links to undermine the regularity level of graphs, which forms the foundation of inference attack methods. Specifically, with the assumption that the substructure with higher regularity level contains more regular equivalence components and has more equivalent paths supplied for the random walk processes, random walk-based link importance measuring algorithm is proposed to identify the representative links. The structural regularity metric, measuring the structural predictability of graphs, is also introduced to guide the link perturbation for structural predictability optimization. The extensive experiments on artificial and real-world data sets demonstrate the effectiveness of the proposed structural predictability optimization method. Specifically, the method can learn the role of links accurately in term of graph organization, and the performance of structure inference on graphs can be deteriorated effectively by representative link-based perturbation.