Can sparsity improve the privacy of neural networks?

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Gonon, Antoine, Zheng, Léon, Lalanne, Clément, Le, Quoc-Tung, Lauga, Guillaume, Pouliquen, Can
Format: Preprint
Published: 2023
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909209544622080
author Gonon, Antoine
Zheng, Léon
Lalanne, Clément
Le, Quoc-Tung
Lauga, Guillaume
Pouliquen, Can
author_facet Gonon, Antoine
Zheng, Léon
Lalanne, Clément
Le, Quoc-Tung
Lauga, Guillaume
Pouliquen, Can
contents Sparse neural networks are mainly motivated by ressource efficiency since they use fewer parameters than their dense counterparts but still reach comparable accuracies. This article empirically investigates whether sparsity could also improve the privacy of the data used to train the networks. The experiments show positive correlations between the sparsity of the model, its privacy, and its classification error. Simply comparing the privacy of two models with different sparsity levels can yield misleading conclusions on the role of sparsity, because of the additional correlation with the classification error. From this perspective, some caveats are raised about previous works that investigate sparsity and privacy.
format Preprint
id arxiv_https___arxiv_org_abs_2304_07234
institution arXiv
publishDate 2023
record_format arxiv
spellingShingle Can sparsity improve the privacy of neural networks?
Gonon, Antoine
Zheng, Léon
Lalanne, Clément
Le, Quoc-Tung
Lauga, Guillaume
Pouliquen, Can
Cryptography and Security
Machine Learning
Sparse neural networks are mainly motivated by ressource efficiency since they use fewer parameters than their dense counterparts but still reach comparable accuracies. This article empirically investigates whether sparsity could also improve the privacy of the data used to train the networks. The experiments show positive correlations between the sparsity of the model, its privacy, and its classification error. Simply comparing the privacy of two models with different sparsity levels can yield misleading conclusions on the role of sparsity, because of the additional correlation with the classification error. From this perspective, some caveats are raised about previous works that investigate sparsity and privacy.
title Can sparsity improve the privacy of neural networks?
topic Cryptography and Security
Machine Learning
url https://arxiv.org/abs/2304.07234