Automatically Evolving Lookup Tables for Function Approximation

Oliver Krauss, William B. Langdon

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

3 Citations (Scopus)

Abstract

Many functions, such as square root, are approximated and sped up with lookup tables containing pre-calculated values. We introduce an approach using genetic algorithms to evolve such lookup tables for any smooth function. It provides double precision and calculates most values to the closest bit, and outperforms reference implementations in most cases with competitive run-time performance.

Original languageEnglish
Title of host publicationGenetic Programming - 23rd European Conference, EuroGP 2020, Held as Part of EvoStar 2020, Proceedings
EditorsTing Hu, Nuno Lourenço, Eric Medvet, Federico Divina
PublisherSpringer
Pages84-100
Number of pages17
ISBN (Print)9783030440930
DOIs
Publication statusPublished - 2020
Event23rd European Conference on Genetic Programming, EuroGP 2020, held as part of EvoStar 2020 - Seville, Spain
Duration: 15 Apr 202017 Apr 2020

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12101 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference23rd European Conference on Genetic Programming, EuroGP 2020, held as part of EvoStar 2020
Country/TerritorySpain
CitySeville
Period15.04.202017.04.2020

Keywords

  • Covariance matrix adaptation
  • Genetic Improvement
  • Objective function

Fingerprint

Dive into the research topics of 'Automatically Evolving Lookup Tables for Function Approximation'. Together they form a unique fingerprint.

Cite this