Skip to Main content Skip to Navigation
Conference papers

Compressive k-Means with Differential Privacy

Abstract : In the compressive learning framework, one harshly compresses a whole training dataset into a single vector of generalized random moments, the sketch, from which a learning task can subsequently be performed. We prove that this loss of information can be leveraged to design a differentially private mechanism, and study empirically the privacy-utility tradeoff for the k-means clustering problem.
Complete list of metadata

Cited literature [11 references]  Display  Hide  Download

https://hal.inria.fr/hal-02154820
Contributor : Antoine Chatalic <>
Submitted on : Thursday, June 13, 2019 - 9:34:59 AM
Last modification on : Thursday, January 7, 2021 - 4:35:27 PM

File

spars_final.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-02154820, version 1

Citation

Vincent Schellekens, Antoine Chatalic, Florimond Houssiau, Yves-Alexandre de Montjoye, Laurent Jacques, et al.. Compressive k-Means with Differential Privacy. SPARS 2019 - Signal Processing with Adaptive Sparse Structured Representations, Jul 2019, Toulouse, France. pp.1-2. ⟨hal-02154820⟩

Share

Metrics

Record views

315

Files downloads

895