Privacy-Preserving ML

Differential privacy accounting, federated learning, secure aggregation, and the utility cost of each guarantee.