Learned cardinalities: Estimating correlated joins with deep learning

Andreas Kipf, Thomas Kipf, Bernhard Radke, Viktor Leis, Peter Boncz, Alfons Kemper

Research output: Contribution to conferencePaperpeer-review

179 Scopus citations

Abstract

We describe a new deep learning approach to cardinality estimation. MSCN is a multi-set convolutional network, tailored to representing relational query plans, that employs set semantics to capture query features and true cardinalities. MSCN builds on sampling-based estimation, addressing its weaknesses when no sampled tuples qualify a predicate, and in capturing join-crossing correlations. Our evaluation of MSCN using a real-world dataset shows that deep learning signiicantly enhances the quality of cardinality estimation, which is the core problem in query optimization.

Original languageEnglish
StatePublished - 2019
Event9th Biennial Conference on Innovative Data Systems Research, CIDR 2019 - Pacific Grove, United States
Duration: 13 Jan 201916 Jan 2019

Conference

Conference9th Biennial Conference on Innovative Data Systems Research, CIDR 2019
Country/TerritoryUnited States
CityPacific Grove
Period13/01/1916/01/19

Fingerprint

Dive into the research topics of 'Learned cardinalities: Estimating correlated joins with deep learning'. Together they form a unique fingerprint.

Cite this