Learned Interpolation for Better Streaming Quantile Approximation with Worst-Case Guarantees

Nicholas Schiefer*, Justin Y. Chen, Piotr Indyk, Shyam Narayanan, Sandeep Silwal, Tal Wagner

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

An ε-approximate quantile sketch over a stream of n inputs approximates the rank of any query point q—that is, the number of input points less than q—up to an additive error of εn, generally with some probability of at least 1-1/poly(n), while consuming o(n) space. While the celebrated KLL sketch of Karnin, Lang, and Liberty achieves a provably optimal quantile approximation algorithm over worst-case streams, the approximations it achieves in practice are often far from optimal. Indeed, the most commonly used technique in practice is Dunning’s t-digest, which often achieves much better approximations than KLL on real-world data but is known to have arbitrarily large errors in the worst case. We apply interpolation techniques to the streaming quantiles problem to attempt to achieve better approximations on real-world data sets than KLL while maintaining similar guarantees in the worst case.

Original languageEnglish
Title of host publicationSIAM Conference on Applied and Computational Discrete Algorithms, ACDA 2023
EditorsJonathan Berry, David Shmoys, Lenore Cowen, Uwe Naumann
PublisherSociety for Industrial and Applied Mathematics (SIAM)
Pages87-97
Number of pages11
ISBN (Electronic)9781713899631
StatePublished - 2023
Externally publishedYes
Event2nd SIAM Conference on Applied and Computational Discrete Algorithms, ACDA 2023 - Seattle, United States
Duration: 31 May 20232 Jun 2023

Publication series

NameSIAM Conference on Applied and Computational Discrete Algorithms, ACDA 2023

Conference

Conference2nd SIAM Conference on Applied and Computational Discrete Algorithms, ACDA 2023
Country/TerritoryUnited States
CitySeattle
Period31/05/232/06/23

Fingerprint

Dive into the research topics of 'Learned Interpolation for Better Streaming Quantile Approximation with Worst-Case Guarantees'. Together they form a unique fingerprint.

Cite this