The notion of differential privacy has been introduced to enable statistical analyses to be carried out while protecting the privacy of the individuals whose data are analysed. In this thesis, three problems of statistical inference under local differential privacy constraints are considered. First, we address the problem of nonparametric density estimation. We study the minimax risk over Besov ellipsoids under the Lr-risk, and we investigate adaptation to the regularity parameter. We then consider the problem of identifying the support of the expectation of a d-dimensional gaussian random variable. Under sparsity assumptions, we study the minimax risk for the Hamming loss, and obtain necessary and sufficient conditions for support recovery...
Finding anonymization mechanisms to protect personal data is at the heart of machine learning resear...
This dissertation studies the trade-off between differential privacy and statistical accuracy in par...
International audienceWe address the problem of goodness-of-fit testing for Hölder continuous densit...
The notion of differential privacy has been introduced to enable statistical analyses to be carried ...
The notion of differential privacy has been introduced to enable statistical analyses to be carried ...
La notion de confidentialité différentielle a été introduite pour permettre de réaliser des analyses...
Producing statistics that respect the privacy of the samples while still maintaining their accuracy ...
Producing statistics that respect the privacy of the samples while still maintaining their accuracy ...
Producing statistics that respect the privacy of the samples while still maintaining their accuracy ...
Producing statistics that respect the privacy of the samples while still maintaining their accuracy ...
Producing statistics that respect the privacy of the samples while still maintaining their accuracy ...
Finding anonymization mechanisms to protect personal data is at the heart of recent machine learning...
Finding anonymization mechanisms to protect personal data is at the heart of recent machine learning...
International audienceThe challenge of producing accurate statistics while respecting the privacy of...
Finding anonymization mechanisms to protect personal data is at the heart of machine learning resear...
Finding anonymization mechanisms to protect personal data is at the heart of machine learning resear...
This dissertation studies the trade-off between differential privacy and statistical accuracy in par...
International audienceWe address the problem of goodness-of-fit testing for Hölder continuous densit...
The notion of differential privacy has been introduced to enable statistical analyses to be carried ...
The notion of differential privacy has been introduced to enable statistical analyses to be carried ...
La notion de confidentialité différentielle a été introduite pour permettre de réaliser des analyses...
Producing statistics that respect the privacy of the samples while still maintaining their accuracy ...
Producing statistics that respect the privacy of the samples while still maintaining their accuracy ...
Producing statistics that respect the privacy of the samples while still maintaining their accuracy ...
Producing statistics that respect the privacy of the samples while still maintaining their accuracy ...
Producing statistics that respect the privacy of the samples while still maintaining their accuracy ...
Finding anonymization mechanisms to protect personal data is at the heart of recent machine learning...
Finding anonymization mechanisms to protect personal data is at the heart of recent machine learning...
International audienceThe challenge of producing accurate statistics while respecting the privacy of...
Finding anonymization mechanisms to protect personal data is at the heart of machine learning resear...
Finding anonymization mechanisms to protect personal data is at the heart of machine learning resear...
This dissertation studies the trade-off between differential privacy and statistical accuracy in par...
International audienceWe address the problem of goodness-of-fit testing for Hölder continuous densit...