Lotto: Secure Participant Selection against Adversarial Servers in Federated Learning

arXiv:2401.02880v2 Announce Type: replace
Abstract: In Federated Learning (FL), common privacy-enhancing techniques, such as secure aggregation and distributed differential privacy, rely on the critical assumption of an honest majority among participants to withstand various attacks. In practice, however, servers are not always trusted, and an adversarial server can strategically select compromised clients to create a dishonest majority, thereby undermining the system's security guarantees. In this paper, we present Lotto, an FL system that addresses this fundamental, yet underexplored issue by …

adversarial aggregation arxiv attacks can critical cs.cr differential privacy distributed federated federated learning practice privacy select server servers techniques