Automatically Evolving Lookup Tables for Function Approximation

Oliver Krauss, William B. Langdon

Publikation: Beitrag in Buch/Bericht/TagungsbandKonferenzbeitragBegutachtung

3 Zitate (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.

OriginalspracheEnglisch
TitelGenetic Programming - 23rd European Conference, EuroGP 2020, Held as Part of EvoStar 2020, Proceedings
Redakteure/-innenTing Hu, Nuno Lourenço, Eric Medvet, Federico Divina
Herausgeber (Verlag)Springer
Seiten84-100
Seitenumfang17
ISBN (Print)9783030440930
DOIs
PublikationsstatusVeröffentlicht - 2020
Veranstaltung23rd European Conference on Genetic Programming, EuroGP 2020, held as part of EvoStar 2020 - Seville, Spanien
Dauer: 15 Apr 202017 Apr 2020

Publikationsreihe

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Band12101 LNCS
ISSN (Print)0302-9743
ISSN (elektronisch)1611-3349

Konferenz

Konferenz23rd European Conference on Genetic Programming, EuroGP 2020, held as part of EvoStar 2020
Land/GebietSpanien
OrtSeville
Zeitraum15.04.202017.04.2020

Fingerprint

Untersuchen Sie die Forschungsthemen von „Automatically Evolving Lookup Tables for Function Approximation“. Zusammen bilden sie einen einzigartigen Fingerprint.

Zitieren