Peter Kairouz (Google AI)
The discovery of heavy hitters (most frequent items) in user-generated data streams drives improvements in the app and web ecosystems, but often comes with substantial privacy risks. To address these risks, we propose a distributed and privacy-preserving algorithm for discovering the heavy hitters in a population of user-generated data streams. We critically leverage the sampling property of our distributed algorithm to prove that it is inherently differentially private, without requiring any addition of noise. We also examine the trade-off between privacy and utility, and show that our algorithm provides excellent utility while also achieving strong privacy guarantees. We validate our findings both theoretically, using worst-case analyses, and practically, using a Twitter dataset with 1.6M tweets and over 650k users.