Private Federated Learning with Domain Adaptation
This work addresses privacy-preserving collaborative learning for multiple parties, presenting an incremental improvement by combining domain adaptation with federated learning.
The paper tackles the problem of improving model accuracy in federated learning under privacy constraints by introducing per-user domain adaptation, showing that this technique enhances accuracy for all users, especially when differential privacy is applied.
Federated Learning (FL) is a distributed machine learning (ML) paradigm that enables multiple parties to jointly re-train a shared model without sharing their data with any other parties, offering advantages in both scale and privacy. We propose a framework to augment this collaborative model-building with per-user domain adaptation. We show that this technique improves model accuracy for all users, using both real and synthetic data, and that this improvement is much more pronounced when differential privacy bounds are imposed on the FL model.