Secure Embedding Aggregation for Federated Representation Learning
This addresses privacy concerns in federated learning for distributed clients, but appears incremental as it builds on existing secure aggregation methods.
The paper tackles the problem of securely aggregating local embeddings in federated representation learning, proposing a protocol that protects both entity sets and embeddings against a curious server and colluding clients, achieving privacy guarantees for up to T < N/2 colluding clients.
We consider a federated representation learning framework, where with the assistance of a central server, a group of $N$ distributed clients train collaboratively over their private data, for the representations (or embeddings) of a set of entities (e.g., users in a social network). Under this framework, for the key step of aggregating local embeddings trained privately at the clients, we develop a secure embedding aggregation protocol named \scheme, which leverages all potential aggregation opportunities among all the clients, while providing privacy guarantees for the set of local entities and corresponding embeddings \emph{simultaneously} at each client, against a curious server and up to $T < N/2$ colluding clients.