English

Learned Cardinalities: Estimating Correlated Joins with Deep Learning

Databases 2018-12-19 v2

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 significantly enhances the quality of cardinality estimation, which is the core problem in query optimization.

Keywords

Cite

@article{arxiv.1809.00677,
  title  = {Learned Cardinalities: Estimating Correlated Joins with Deep Learning},
  author = {Andreas Kipf and Thomas Kipf and Bernhard Radke and Viktor Leis and Peter Boncz and Alfons Kemper},
  journal= {arXiv preprint arXiv:1809.00677},
  year   = {2018}
}

Comments

CIDR 2019. https://github.com/andreaskipf/learnedcardinalities

R2 v1 2026-06-23T03:52:59.743Z