Have it your way: Individualized Privacy Assignment for DP-SGD
This work addresses the problem of inflexible privacy protection in machine learning for users with diverse privacy expectations, offering a more tailored approach.
The paper tackles the limitation of uniform privacy budgets in differentially private machine learning by proposing individualized privacy budgets to accommodate varying user preferences, and introduces Individualized DP-SGD (IDP-SGD), which empirically improves privacy-utility trade-offs.
When training a machine learning model with differential privacy, one sets a privacy budget. This budget represents a maximal privacy violation that any user is willing to face by contributing their data to the training set. We argue that this approach is limited because different users may have different privacy expectations. Thus, setting a uniform privacy budget across all points may be overly conservative for some users or, conversely, not sufficiently protective for others. In this paper, we capture these preferences through individualized privacy budgets. To demonstrate their practicality, we introduce a variant of Differentially Private Stochastic Gradient Descent (DP-SGD) which supports such individualized budgets. DP-SGD is the canonical approach to training models with differential privacy. We modify its data sampling and gradient noising mechanisms to arrive at our approach, which we call Individualized DP-SGD (IDP-SGD). Because IDP-SGD provides privacy guarantees tailored to the preferences of individual users and their data points, we find it empirically improves privacy-utility trade-offs.