Explaining Recurrent Machine Learning Models: Integral Privacy Revisited
2020 (English)In: Privacy in Statistical Databases: UNESCO Chair in Data Privacy, International Conference, PSD 2020, Tarragona, Spain, September 23–25, 2020, Proceedings / [ed] Josep Domingo-Ferrer, Krishnamurty Muralidhar, Cham: Springer, 2020, p. 62-73Conference paper, Published paper (Refereed)
Abstract [en]
We have recently introduced a privacy model for statistical and machine learning models called integral privacy. A model extracted from a database or, in general, the output of a function satisfies integral privacy when the number of generators of this model is sufficiently large and diverse. In this paper we show how the maximal c-consensus meets problem can be used to study the databases that generate an integrally private solution. We also introduce a definition of integral privacy based on minimal sets in terms of this maximal c-consensus meets problem.
Place, publisher, year, edition, pages
Cham: Springer, 2020. p. 62-73
Series
Lecture Notes in Computer Science, ISSN 0302-9743, E-ISSN 1611-3349 ; 12276
Keywords [en]
Clustering, Integral privacy, Maximal c-consensus meets, Parameter selection, Data privacy, Database systems, Machine learning models, Privacy models, Machine learning
National Category
Computer Sciences
Research subject
Skövde Artificial Intelligence Lab (SAIL)
Identifiers
URN: urn:nbn:se:his:diva-19186DOI: 10.1007/978-3-030-57521-2_5Scopus ID: 2-s2.0-85092091090ISBN: 978-3-030-57520-5 (print)ISBN: 978-3-030-57521-2 (electronic)OAI: oai:DiVA.org:his-19186DiVA, id: diva2:1476762
Conference
UNESCO Chair in Data Privacy, International Conference, PSD 2020, Tarragona, Spain, September 23–25, 2020
Part of project
Disclosure risk and transparency in big data privacy, Swedish Research Council
Funder
Swedish Research Council, 2016-03346
Note
CC BY 4.0
Also part of the Information Systems and Applications, incl. Internet/Web, and HCI book sub series (LNISA, volume 12276)
Partial support of the project Swedish Research Council (grant number VR 2016-03346) is acknowledged.
DRIAT
2020-10-152020-10-152021-08-18Bibliographically approved