Backdoor Federated Learning by Poisoning Backdoor-Critical Layers

arXiv:2308.04466v2 Announce Type: replace-cross
Abstract: Federated learning (FL) has been widely deployed to enable machine learning training on sensitive data across distributed devices. However, the decentralized learning paradigm and heterogeneity of FL further extend the attack surface for backdoor attacks. Existing FL attack and defense methodologies typically focus on the whole model. None of them recognizes the existence of backdoor-critical (BC) layers-a small subset of layers that dominate the model vulnerabilities. Attacking the BC layers achieves equivalent effects as attacking …

abstract arxiv attacks backdoor cs.cr cs.cv cs.lg data decentralized defense devices distributed federated learning focus machine machine learning paradigm surface training type