An Iconic Heavy Hitter Algorithm Made Private

TL;DR

This paper introduces a differentially private version of the SpaceSaving algorithm for heavy hitter detection in data streams, achieving strong privacy guarantees while maintaining high utility and efficiency.

Contribution

It presents the first private adaptation of SpaceSaving, a novel generic method for private heavy hitter extraction from frequency oracles, and demonstrates superior empirical performance.

Findings

Private SpaceSaving outperforms private Misra-Gries in utility.

The generic method efficiently extracts heavy hitters with minimal additional memory.

Experimental results confirm practical effectiveness under various privacy settings.

Abstract

Identifying heavy hitters in data streams is a fundamental problem with widespread applications in modern analytics systems. These streams are often derived from sensitive user activity, making update-level privacy guarantees necessary. While recent work has adapted the classical heavy hitter algorithm Misra-Gries to satisfy differential privacy in the streaming model, the privatization of other heavy hitter algorithms with better empirical utility is absent. Under this observation, we present the first differentially private variant of the SpaceSaving algorithm, which, in the non-private setting, is regarded as the state-of-the-art in practice. Our construction post-processes a non-private SpaceSaving summary by injecting asymptotically optimal noise and applying a carefully calibrated selection rule that suppresses unstable labels. This yields strong privacy guarantees while preserving the empirical advantages of SpaceSaving. Second, we introduce a generic method for extracting heavy hitters from any differentially private frequency oracle in the data stream model. The method requires only O(k) additional memory, where k is the number of heavy items, and provides a mechanism for safely releasing item identities from noisy frequency estimates. This yields an efficient, plug-and-play approach for private heavy hitter recovery from linear sketches. Finally, we conduct an experimental evaluation on synthetic and real-world datasets. Across a wide range of privacy parameters and space budgets, our method provides superior utility to the existing differentially private Misra-Gries algorithm. Our results demonstrate that the empirical superiority of SpaceSaving survives privatization and that efficient, practical heavy hitter identification is achievable under strong differential privacy guarantees.