Sparsity in neural networks can improve their privacy

Antoine Gonon, Léon Zheng, Clément Lalanne +3

This article measures how sparsity can make neural networks more robust to membership inference attacks. The obtained empirical results show that sparsity improves the privacy of t…