Dynamic Privacy Budget Allocation Improves Data Efficiency of Differentially Private Gradient Descent
This work addresses the challenge of balancing privacy and model performance in sensitive data applications, representing an incremental advancement in understanding dynamic privacy schedules.
The paper tackles the problem of improving data efficiency in differentially private gradient descent by analyzing dynamic privacy budget allocation, showing that a dynamic noise schedule minimizes the utility upper bound and that loss curvature greatly impacts the influence of noise.
Protecting privacy in learning while maintaining the model performance has become increasingly critical in many applications that involve sensitive data. Private Gradient Descent (PGD) is a commonly used private learning framework, which noises gradients based on the Differential Privacy protocol. Recent studies show that \emph{dynamic privacy schedules} of decreasing noise magnitudes can improve loss at the final iteration, and yet theoretical understandings of the effectiveness of such schedules and their connections to optimization algorithms remain limited. In this paper, we provide comprehensive analysis of noise influence in dynamic privacy schedules to answer these critical questions. We first present a dynamic noise schedule minimizing the utility upper bound of PGD, and show how the noise influence from each optimization step collectively impacts utility of the final model. Our study also reveals how impacts from dynamic noise influence change when momentum is used. We empirically show the connection exists for general non-convex losses, and the influence is greatly impacted by the loss curvature.