DEVA: Decentralized, Verifiable Secure Aggregation for Privacy-Preserving Learning

Published in In *24th Information Security Conference* (ISC), 2021

DEVA proposes a decentralized secure aggregation protocol that is both privacy-preserving and verifiable.

Key features:

  • Ensures correctness of the aggregation result through non-interactive verification
  • Protects individual inputs in federated learning settings
  • Tolerates malicious users and decentralized trust assumptions

The paper demonstrates that DEVA achieves strong security guarantees without central authorities.