Not a Free Lunch, But a Cheap One: On Classifiers Performance on Anonymized Datasets

Research output: Chapter in Book/Report/Conference proceedingConference contributionAcademicpeer-review

Abstract

The problem of protecting datasets from the disclosure of confidential information, while published data remains useful for analysis, has recently gained momentum. To solve this problem, anonymization techniques such as k-anonymity, ℓ -diversity, and t-closeness have been used to generate anonymized datasets for training classifiers. While these techniques provide an effective means to generate anonymized datasets, an understanding of how their application affects the performance of classifiers is currently missing. This knowledge enables the data owner and analyst to select the most appropriate classification algorithm and training parameters in order to guarantee high privacy requirements while minimizing the loss of accuracy. In this study, we perform extensive experiments to verify how the classifiers performance changes when trained on an anonymized dataset compared to the original one, and evaluate the impact of classification algorithms, datasets properties, and anonymization parameters on classifiers’ performance.

Original languageEnglish
Title of host publicationData and Applications Security and Privacy XXXV - 35th Annual IFIP WG 11.3 Conference, DBSec 2021, Proceedings
EditorsKen Barker, Kambiz Ghazinour
PublisherSpringer Science and Business Media B.V.
Pages237-258
Number of pages22
ISBN (Print)9783030812416
DOIs
Publication statusPublished - 2021
Event35th Annual IFIP WG 11.3 Conference on Data and Applications Security and Privacy, DBSec 2021 - Virtual, Online
Duration: 19 Jul 202120 Jul 2021

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12840 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference35th Annual IFIP WG 11.3 Conference on Data and Applications Security and Privacy, DBSec 2021
CityVirtual, Online
Period19/07/2120/07/21

Bibliographical note

Publisher Copyright:
© 2021, IFIP International Federation for Information Processing.

Keywords

  • Classifiers comparison
  • k-anonymity
  • Privacy-preserving
  • t-closeness
  • ℓ -diversity

Fingerprint

Dive into the research topics of 'Not a Free Lunch, But a Cheap One: On Classifiers Performance on Anonymized Datasets'. Together they form a unique fingerprint.

Cite this