Practical and Private (Deep) Learning without Sampling or Shuffling
This work addresses privacy concerns in federated learning by enabling more flexible data access patterns, though it is incremental as it builds on existing DP methods.
The paper tackles the challenge of training models with differential privacy in scenarios like federated learning where exact sampling or shuffling is impractical, by proposing DP-FTRL, a method that achieves comparable privacy/accuracy trade-offs to state-of-the-art DP-SGD without requiring privacy amplification.
We consider training models with differential privacy (DP) using mini-batch gradients. The existing state-of-the-art, Differentially Private Stochastic Gradient Descent (DP-SGD), requires privacy amplification by sampling or shuffling to obtain the best privacy/accuracy/computation trade-offs. Unfortunately, the precise requirements on exact sampling and shuffling can be hard to obtain in important practical scenarios, particularly federated learning (FL). We design and analyze a DP variant of Follow-The-Regularized-Leader (DP-FTRL) that compares favorably (both theoretically and empirically) to amplified DP-SGD, while allowing for much more flexible data access patterns. DP-FTRL does not use any form of privacy amplification. The code is available at https://github.com/google-research/federated/tree/master/dp_ftrl and https://github.com/google-research/DP-FTRL .