Automatically Evolving Lookup Tables for Function Approximation

Oliver Krauss; William B. Langdon

doi:10.1007/978-3-030-44094-7_6

Automatically Evolving Lookup Tables for Function Approximation

Oliver Krauss
, William B. Langdon

Institut für Systemsoftware

Publikation: Beitrag in Buch/Bericht/Konferenzband › Konferenzbeitrag › Begutachtung

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.

Originalsprache	Englisch
Titel	EuroGP 2020: Genetic Programming
Herausgeber*innen	Ting Hu, Nuno Lourenço, Eric Medvet, Federico Divina
Verlag	Springer
Seiten	84-100
Seitenumfang	17
Band	12101
ISBN (Print)	978-3-030-44093-0
DOIs	https://doi.org/10.1007/978-3-030-44094-7_6
Publikationsstatus	Veröffentlicht - Apr. 2020

Publikationsreihe

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Band	12101 LNCS
ISSN (Print)	0302-9743
ISSN (elektronisch)	1611-3349

Wissenschaftszweige

102 Informatik
102009 Computersimulation
102011 Formale Sprachen
102013 Human-Computer Interaction
102022 Softwareentwicklung
102024 Usability Research
102029 Praktische Informatik

JKU-Schwerpunkte

Digital Transformation

Zugriff auf Dokument

10.1007/978-3-030-44094-7_6

Andere Dateien und Links

Verknüpfung zur Publikation in Scopus

Dieses zitieren

Krauss, O., & Langdon, W. B. (2020). Automatically Evolving Lookup Tables for Function Approximation. in T. Hu, N. Lourenço, E. Medvet, & F. Divina (Hrsg.), EuroGP 2020: Genetic Programming (Band 12101, S. 84-100). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Band 12101 LNCS). Springer. https://doi.org/10.1007/978-3-030-44094-7_6

Krauss, Oliver ; Langdon, William B. / Automatically Evolving Lookup Tables for Function Approximation. EuroGP 2020: Genetic Programming. Hrsg. / Ting Hu ; Nuno Lourenço ; Eric Medvet ; Federico Divina. Band 12101 Springer, 2020. S. 84-100 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{c4819f6686fc448fb5ec8e908e618b4d,

title = "Automatically Evolving Lookup Tables for Function Approximation",

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.",

author = "Oliver Krauss and Langdon, \{William B.\}",

year = "2020",

month = apr,

doi = "10.1007/978-3-030-44094-7\_6",

language = "English",

isbn = "978-3-030-44093-0",

volume = "12101",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer",

pages = "84--100",

editor = "Ting Hu and Nuno Louren{\c c}o and Eric Medvet and Federico Divina",

booktitle = "EuroGP 2020: Genetic Programming",

}

Krauss, O & Langdon, WB 2020, Automatically Evolving Lookup Tables for Function Approximation. in T Hu, N Lourenço, E Medvet & F Divina (Hrsg.), EuroGP 2020: Genetic Programming. Bd. 12101, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), Bd. 12101 LNCS, Springer, S. 84-100. https://doi.org/10.1007/978-3-030-44094-7_6

Automatically Evolving Lookup Tables for Function Approximation. / Krauss, Oliver; Langdon, William B.
EuroGP 2020: Genetic Programming. Hrsg. / Ting Hu; Nuno Lourenço; Eric Medvet; Federico Divina. Band 12101 Springer, 2020. S. 84-100 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Band 12101 LNCS).

Publikation: Beitrag in Buch/Bericht/Konferenzband › Konferenzbeitrag › Begutachtung

TY - GEN

T1 - Automatically Evolving Lookup Tables for Function Approximation

AU - Krauss, Oliver

AU - Langdon, William B.

PY - 2020/4

Y1 - 2020/4

N2 - 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.

AB - 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.

UR - https://www.scopus.com/pages/publications/85084740928

U2 - 10.1007/978-3-030-44094-7_6

DO - 10.1007/978-3-030-44094-7_6

M3 - Conference proceedings

SN - 978-3-030-44093-0

VL - 12101

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 84

EP - 100

BT - EuroGP 2020: Genetic Programming

A2 - Hu, Ting

A2 - Lourenço, Nuno

A2 - Medvet, Eric

A2 - Divina, Federico

PB - Springer

ER -

Krauss O, Langdon WB. Automatically Evolving Lookup Tables for Function Approximation. in Hu T, Lourenço N, Medvet E, Divina F, Hrsg., EuroGP 2020: Genetic Programming. Band 12101. Springer. 2020. S. 84-100. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-030-44094-7_6