Kunal Talwar, Shan Wang, Audra McMillan +34
We revisit the problem of designing scalable protocols for private statistics and private federated learning when each device holds its private data. Locally differentially private…