Private Generative Bootstrap via Blocking
2026-08-03 • Machine Learning
Machine Learning
AI summaryⓘ
The authors propose a new method called Private Generative Bayesian Bootstrap (PGBB) to protect individual privacy when reporting statistical uncertainty in AI systems. Instead of giving each person a unique weight, they group people and assign the same weight to the whole group, making it harder to identify anyone. They use a special process that learns privately with added noise, so later statistical samples don't use extra privacy limits. Their approach keeps privacy while still giving reliable uncertainty measures, tested with real data and compared to other private Bayesian methods.
Differential PrivacyBayesian BootstrapPosterior SamplingLikelihood-Free InferenceAmortized InferenceDirichlet DistributionPrivacy BudgetStatistical UncertaintySimulation-Based InferenceData Privacy
Authors
Jinwon Sohn, Veronika Ročková
Abstract
With AI systems gaining more access to individuals' information, it is important to protect privacy when reporting statistical answers. Equally important is to privatize the reporting of uncertainty in such answers. To this end, we adopt a Bayesian likelihood-free framework and make simulation from the posterior private. In particular, we propose a new private instantiation of the Bayesian bootstrap using a blocking strategy. Rather than assigning idiosyncratic random weights to each individual, we randomly group individuals and assign a single weight to each group. By concealing individuals' contributions within a group, we fortify differential privacy gates. We harness amortized inference that decouples private learning from posterior sampling. A push-forward map from observation weights to posterior samples is learned privately by adding calibrated noise during training. Subsequent posterior draws require no additional privacy and computation budget. We call the resulting method the Private Generative Bayesian Bootstrap (PGBB). We establish a differential privacy guarantee, analyze convergence to the non-private blocked-bootstrap target, and quantify the discrepancy between the ordinary and blocked Bayesian-bootstrap posteriors. In addition, we derive data-free tuning of the block Dirichlet concentration parameter that restores posterior dispersion asymptotically. We also show a single fit of PGBB can support a family of loss-based decision rules simultaneously without additional privacy cost. In simulations and in applications to U.S. Census returns to schooling and U.S. natality birthweight quantiles, PGBB gives competitive private uncertainty quantification and improves over private Bayesian alternatives that require a specified data-generating model in common settings.