LAPRAS : Learning-Augmented PRivate Answering for linear query Streams

TL;DR

LAPRAS is a learning-augmented differentially private query answering algorithm that leverages query predictions to improve utility while maintaining robustness when predictions are inaccurate.

Contribution

It introduces LAPRAS, a novel online DP query answering method that uses query predictions and a smooth allocation strategy to optimize privacy utility.

Findings

LAPRAS achieves near-offline utility with high prediction accuracy.

The method degrades gracefully to baseline performance with low prediction overlap.

Empirical validation on real datasets confirms the effectiveness of the approach.

Abstract

Modern database workloads are highly predictable: query streams are dominated by recurring jobs and templates, even when their arrival order is not known in advance. This motivates a learning-augmented view of online differentially private (DP) analytics: can algorithms utilize predictions about which queries will occur to improve utility under a single global privacy budget, while remaining robust when predictions are wrong? We study online DP query answering, where a curator must answer a stream $Q$ of $S$ linear queries arriving in uniformly random order under privacy budget $(\epsilon,\delta)$. We present LAPRAS, which assumes access to an oracle that outputs a prediction set of queries likely to appear in the stream and uses it to guide privacy spending. LAPRAS answers predicted queries using the offline-optimal Matrix Mechanism and answers the remaining queries online from a residual budget. To pace spending across an unknown number of unpredicted queries, we introduce Smooth Allocation, which forms an unbiased stopping-time estimate $\widehat{B}$ from the first $T=\Theta(\log^2 S)$ unpredicted queries and continuously recalibrates per-query expenditure. Empirically, over two real datasets, we validate the intended consistency--robustness trade-off: LAPRAS achieves near-offline utility under high overlap and degrades gracefully to baseline-level performance when overlap is low.