Privacy-Preserving k-NN for Small and Large Data Sets

12 years 10 months ago
Privacy-Preserving k-NN for Small and Large Data Sets
It is not surprising that there is strong interest in kNN queries to enable clustering, classification and outlierdetection tasks. However, previous approaches to privacypreserving k-NN are costly and can only be realistically applied to small data sets. We provide efficient solutions for k-NN queries queries for vertically partitioned data. We provide the first solution for the L∞ (or Chessboard) metric as well as detailed privacy-preserving computation of all other Minkowski metrics. We enable privacy-preserving L∞ by providing a solution to the Yao’s Millionaire Problem with more than two parties. This is based on a new and practical solution to Yao’s Millionaire with shares. We also provide privacy-preserving algorithms for combinations of local metrics into a global that handles the large dimensionality and diversity of attributes common in vertically partitioned data.
Artak Amirbekyan, Vladimir Estivill-Castro
Added 03 Jun 2010
Updated 03 Jun 2010
Type Conference
Year 2007
Where ICDM
Authors Artak Amirbekyan, Vladimir Estivill-Castro
Comments (0)