File Download

There are no files associated with this item.

  Links for fulltext
     (May Require Subscription)
Supplementary

Conference Paper: Differentially private clustering in high-dimensional Euclidean spaces

TitleDifferentially private clustering in high-dimensional Euclidean spaces
Authors
Issue Date2017
Citation
34th International Conference on Machine Learning, ICML 2017, 2017, v. 1, p. 501-522 How to Cite?
AbstractWe study the problem of clustering sensitive data while preserving the privacy of individuals represented in the dataset, which has broad applications in practical machine learning and data analysis tasks. Although the problem has been widely studied in the context of lowdimensional, discrete spaces, much remains unknown concerning private clustering in highdimensional Euclidean spaces Rd. In this work, we give differentially private and efficient algorithms achieving strong guarantees for k-means and κ-median clustering when d = Ω(polylog(n)). Our algorithm achieves clustering loss at most log3 (n) OPT + poly (log n, d, κ), advancing the state-of-the-art result of VdOPT+ poly(log n, dd, κd). We also study the case where the data points are s-sparse and show that the clustering loss can scale logarithmically with d, i.e., log3(n)OPT + poly(log n,log d,κ,S). Experiments on both synthetic and real datasets verify the effectiveness of the proposed method.
Persistent Identifierhttp://hdl.handle.net/10722/341228

 

DC FieldValueLanguage
dc.contributor.authorBalean, Maria Fiorina-
dc.contributor.authorDick, Travis-
dc.contributor.authorLiang, Yingyu-
dc.contributor.authorMou, Wenlong-
dc.contributor.authorZhang, Hongyang-
dc.date.accessioned2024-03-13T08:41:10Z-
dc.date.available2024-03-13T08:41:10Z-
dc.date.issued2017-
dc.identifier.citation34th International Conference on Machine Learning, ICML 2017, 2017, v. 1, p. 501-522-
dc.identifier.urihttp://hdl.handle.net/10722/341228-
dc.description.abstractWe study the problem of clustering sensitive data while preserving the privacy of individuals represented in the dataset, which has broad applications in practical machine learning and data analysis tasks. Although the problem has been widely studied in the context of lowdimensional, discrete spaces, much remains unknown concerning private clustering in highdimensional Euclidean spaces Rd. In this work, we give differentially private and efficient algorithms achieving strong guarantees for k-means and κ-median clustering when d = Ω(polylog(n)). Our algorithm achieves clustering loss at most log3 (n) OPT + poly (log n, d, κ), advancing the state-of-the-art result of VdOPT+ poly(log n, dd, κd). We also study the case where the data points are s-sparse and show that the clustering loss can scale logarithmically with d, i.e., log3(n)OPT + poly(log n,log d,κ,S). Experiments on both synthetic and real datasets verify the effectiveness of the proposed method.-
dc.languageeng-
dc.relation.ispartof34th International Conference on Machine Learning, ICML 2017-
dc.titleDifferentially private clustering in high-dimensional Euclidean spaces-
dc.typeConference_Paper-
dc.description.naturelink_to_subscribed_fulltext-
dc.identifier.scopuseid_2-s2.0-85048681197-
dc.identifier.volume1-
dc.identifier.spage501-
dc.identifier.epage522-

Export via OAI-PMH Interface in XML Formats


OR


Export to Other Non-XML Formats