Privacy Amplification via Compression: Achieving the Optimal Privacy-Accuracy-Communication Trade-offs in Distributed Mean Estimation

Oct-10-2025, 23:47:14 GMT–Neural Information Processing Systems 

Privacy and communication constraints are two major bottlenecks in federated learning (FL) and analytics (FA).