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

Institute for Systems Software

Research output: Chapter in Book/Report/Conference proceeding › Conference proceedings › peer-review

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 language	English
Title of host publication	EuroGP 2020: Genetic Programming
Editors	Ting Hu, Nuno Lourenço, Eric Medvet, Federico Divina
Publisher	Springer
Pages	84-100
Number of pages	17
Volume	12101
ISBN (Print)	978-3-030-44093-0
DOIs	https://doi.org/10.1007/978-3-030-44094-7_6
Publication status	Published - Apr 2020

Publication series

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

Fields of science

102 Computer Sciences
102009 Computer simulation
102011 Formal languages
102013 Human-computer interaction
102022 Software development
102024 Usability research
102029 Practical computer science

JKU Focus areas

Digital Transformation

Access to Document

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

Cite this

Krauss, O., & Langdon, W. B. (2020). Automatically Evolving Lookup Tables for Function Approximation. In T. Hu, N. Lourenço, E. Medvet, & F. Divina (Eds.), EuroGP 2020: Genetic Programming (Vol. 12101, pp. 84-100). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 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. editor / Ting Hu ; Nuno Lourenço ; Eric Medvet ; Federico Divina. Vol. 12101 Springer, 2020. pp. 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 (eds), EuroGP 2020: Genetic Programming. vol. 12101, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 12101 LNCS, Springer, pp. 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. ed. / Ting Hu; Nuno Lourenço; Eric Medvet; Federico Divina. Vol. 12101 Springer, 2020. p. 84-100 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 12101 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference proceedings › peer-review

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, editors, EuroGP 2020: Genetic Programming. Vol. 12101. Springer. 2020. p. 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

Automatically Evolving Lookup Tables for Function Approximation

Abstract

Publication series

Fields of science

JKU Focus areas

Access to Document

Other files and links

Cite this