Feature Selection for Classification under Anonymity Constraint

Abstract

Over the last decade, proliferation of various online platforms and their increasing adoption by billions of users have heightened the privacy risk of a user enormously. In fact, security researchers have shown that sparse microdata containing information about online activities of a user although anonymous, can still be used to disclose the identity of the user by cross-referencing the data with other data sources. To preserve the privacy of a user, in existing works several methods (k-anonymity, ℓ-diversity, differential privacy) are proposed for ensuring that a dataset bears small identity disclosure risk. However, the majority of these methods modify the data in isolation, without considering their utility in subsequent knowledge discovery tasks, which makes these datasets less informative. In this work, we consider labeled data that are generally used for classification, and propose two methods for feature selection considering two goals: first, on the reduced feature set the data has small disclosure risk, and second, the utility of the data is preserved for performing a classification task. Experimental results on various real-world datasets show that the method is effective and useful in practice.

Optimization-based k-anonymity algorithms

“…A mathematical formulation of k-anonymity as an optimization problem allows us to gauge how much utility is lost in any practical algorithms we devise compared to a theoretical optimum. A similar approach has been taken by Zhang et al [Zhang et al, 2015] where a formal optimization problem is dened and k-anonymity by containment is considered the constraint. They dened containment as a subset S of the feature space that satises k-anonymity.…”

“…In such a scenario, the record linkage cannot be fruitfully applied if each source shares dierentially private records. The truthfulness characteristic and the fact that the model was recommended in dierent privacy legislation and guidelines (such as HIPAA [U.S. Department of Health & Human Services, 2015] and FIPPA [Information and Privacy Commissioner of Ontario, 2016]) contributed to the wide adoption of k-anonymity and multiple algorithms (e.g., [Bayardo and Agrawal, 2005, Byun et al, 2007, Doka et al, 2015, El Emam and Dankar, 2008, Lee et al, 2017, Zhang et al, 2015) have been devised for application of the technique to privacy-sensitive dataset prior to release.…”

Bayesian Non-Exhaustive Classification A Case Study

“…[15,23,32] present approaches for the name disambiguation task on anonymized graphs and they only leverage graph topological features due to the privacy concerns. In addition, [31] solves name entity disambiguation problem as the privacy-preserving classification task such that the anonymized dataset satisfies non-disclosure requirements, at the same time it achieves high disambiguation performance.…”