Privacy-utility tradeoff remains as one of the fundamental issues of
differentially private machine learning. This paper introduces a geometrically
inspired kernel-based approach to mitigate the accuracy-loss issue in
classification. In this approach, a representation of the affine hull of given
data points is learned in Reproducing Kernel Hilbert Spaces (RKHS). This leads
to a novel distance measure that hides privacy-sensitive information about
individual data points and improves the privacy-utility tradeoff via
significantly reducing the risk of membership inference attacks. The
effectiveness of the approach is demonstrated through experiments on MNIST
dataset, Freiburg groceries dataset, and a real biomedical dataset. It is
verified that the approach remains computationally practical. The application
of the approach to federated learning is considered and it is observed that the
accuracy-loss due to data being distributed is either marginal or not
significantly high.
%0 Generic
%1 kumar2023mitigating
%A Kumar, Mohit
%A Moser, Bernhard A.
%A Fischer, Lukas
%D 2023
%K deep-learning fuzzy image mappings neural-networks
%T On Mitigating the Utility-Loss in Differentially Private Learning: A new
Perspective by a Geometrically Inspired Kernel Approach
%U https://arxiv.org/abs/2304.01300
%X Privacy-utility tradeoff remains as one of the fundamental issues of
differentially private machine learning. This paper introduces a geometrically
inspired kernel-based approach to mitigate the accuracy-loss issue in
classification. In this approach, a representation of the affine hull of given
data points is learned in Reproducing Kernel Hilbert Spaces (RKHS). This leads
to a novel distance measure that hides privacy-sensitive information about
individual data points and improves the privacy-utility tradeoff via
significantly reducing the risk of membership inference attacks. The
effectiveness of the approach is demonstrated through experiments on MNIST
dataset, Freiburg groceries dataset, and a real biomedical dataset. It is
verified that the approach remains computationally practical. The application
of the approach to federated learning is considered and it is observed that the
accuracy-loss due to data being distributed is either marginal or not
significantly high.
@misc{kumar2023mitigating,
abstract = {Privacy-utility tradeoff remains as one of the fundamental issues of
differentially private machine learning. This paper introduces a geometrically
inspired kernel-based approach to mitigate the accuracy-loss issue in
classification. In this approach, a representation of the affine hull of given
data points is learned in Reproducing Kernel Hilbert Spaces (RKHS). This leads
to a novel distance measure that hides privacy-sensitive information about
individual data points and improves the privacy-utility tradeoff via
significantly reducing the risk of membership inference attacks. The
effectiveness of the approach is demonstrated through experiments on MNIST
dataset, Freiburg groceries dataset, and a real biomedical dataset. It is
verified that the approach remains computationally practical. The application
of the approach to federated learning is considered and it is observed that the
accuracy-loss due to data being distributed is either marginal or not
significantly high.},
added-at = {2023-08-04T09:48:01.000+0200},
author = {Kumar, Mohit and Moser, Bernhard A. and Fischer, Lukas},
biburl = {https://www.bibsonomy.org/bibtex/25e0e500326ec9473b04807ff55917c9b/scch},
interhash = {223db7aa53b5324c8223d128e9ed5778},
intrahash = {5e0e500326ec9473b04807ff55917c9b},
keywords = {deep-learning fuzzy image mappings neural-networks},
note = {cite arxiv:2304.01300},
timestamp = {2023-08-04T09:48:01.000+0200},
title = {On Mitigating the Utility-Loss in Differentially Private Learning: A new
Perspective by a Geometrically Inspired Kernel Approach},
url = {https://arxiv.org/abs/2304.01300},
year = 2023
}