ﻻ يوجد ملخص باللغة العربية
We study anonymization techniques for preserving privacy in the publication of microdata tables. Although existing approaches based on generalization can provide enough protection for identities, anonymized tables always suffer from various attribute disclosures because generalization is inefficient to protect sensitive values and the partition of equivalence groups is directly shown to the adversary. Besides, the generalized table also suffers from serious information loss because the original Quasi-Identifier (QI) values are hardly preserved and the protection against attribute disclosure often causes over-protection against identity disclosure. To this end, we propose a novel technique, called mutual cover, to hinder the adversary from matching the combination of QI values in microdata tables. The rationale is to replace the original QI values with random QI values according to some random output tables that make similar tuples to cover for each other at the minimal cost. As a result, the mutual cover prevents identity disclosure and attribute disclosure more effectively than generalization while retaining the distribution of original QI values as far as possible, and the information utility hardly decreases when enhancing the protection for sensitive values. The effectiveness of mutual cover is verified with extensive experiments.
Recent advances in computing have allowed for the possibility to collect large amounts of data on personal activities and private living spaces. To address the privacy concerns of users in this environment, we propose a novel framework called PR-GAN
In the big data era, more and more cloud-based data-driven applications are developed that leverage individual data to provide certain valuable services (the utilities). On the other hand, since the same set of individual data could be utilized to in
Wearable devices generate different types of physiological data about the individuals. These data can provide valuable insights for medical researchers and clinicians that cannot be availed through traditional measures. Researchers have historically
As machine learning becomes a practice and commodity, numerous cloud-based services and frameworks are provided to help customers develop and deploy machine learning applications. While it is prevalent to outsource model training and serving tasks in
Releasing full data records is one of the most challenging problems in data privacy. On the one hand, many of the popular techniques such as data de-identification are problematic because of their dependence on the background knowledge of adversaries