ERIS: Enhancing the Privacy-Utility-Scalability Trade-Off in Federated Learning via Federated Shard Aggregation
ERIS is a Federated Learning framework that balances privacy, utility, and scalability. It uses Federated Shard Aggregation (FSA) to distribute aggregation tasks across multiple clients instead of one central server. Combined with Distributed Shifted Compression (DSC), ERIS maintains standard FL performance while significantly reducing information leakage and communication overhead.