Practical Defences Against Model Inversion Attacks for Split Neural Networks

Abstract

We describe a threat model under which a split network-based federated learning system is susceptible to a model inversion attack by a malicious computational server. We demonstrate that the attack can be successfully performed with limited knowledge of the data distribution by the attacker. We propose a simple additive noise method to defend against model inversion, finding that the method can significantly reduce attack efficacy at an acceptable accuracy trade-off on MNIST. Furthermore, we show that NoPeekNN, an existing defensive method, protects different information from exposure, suggesting that a combined defence is necessary to fully protect private user data.

ICLR 2021 Workshop on Distributed and Private Machine Learning (DPML 2021)

References in corpus (4)

• Towards Federated Learning at Scale: System Design
• PyVertical: A Vertical Federated Learning Framework for Multi-headed SplitNN
• Asymmetric Private Set Intersection with Applications to Contact Tracing and Private Vertical Federated Machine Learning
• Evaluation of Inference Attack Models for Deep Learning on Medical Data

Cited by in corpus (1)

• PyVertical: A Vertical Federated Learning Framework for Multi-headed SplitNN