Fitness landscape analysis of dimensionally-aware genetic programming featuring feynman equations

Marko Durasevic; Domagoj Jakobovic; Marcella Scoczynski Ribeiro Martins; Stjepan Picek; Markus Wagner

doi:10.1007/978-3-030-58115-2_8

Fitness landscape analysis of dimensionally-aware genetic programming featuring feynman equations

Marko Durasevic, Domagoj Jakobovic, Marcella Scoczynski Ribeiro Martins, Stjepan Picek, Markus Wagner

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

6 Citations (Scopus)

Abstract

Genetic programming is an often-used technique for symbolic regression: finding symbolic expressions that match data from an unknown function. To make the symbolic regression more efficient, one can also use dimensionally-aware genetic programming that constrains the physical units of the equation. Nevertheless, there is no formal analysis of how much dimensionality awareness helps in the regression process. In this paper, we conduct a fitness landscape analysis of dimensionally-aware genetic programming search spaces on a subset of equations from Richard Feynman’s well-known lectures. We define an initialisation procedure and an accompanying set of neighbourhood operators for conducting the local search within the physical unit constraints. Our experiments show that the added information about the variable dimensionality can efficiently guide the search algorithm. Still, further analysis of the differences between the dimensionally-aware and standard genetic programming landscapes is needed to help in the design of efficient evolutionary operators to be used in a dimensionally-aware regression.

Original language	English
Title of host publication	Parallel Problem Solving from Nature – PPSN XVI - 16th International Conference, PPSN 2020 Leiden, The Netherlands, September 5–9, 2020 Proceedings, Part II
Editors	Thomas Bäck, Mike Preuss, André Deutz, Hao Wang, Carola Doerr, Michael Emmerich, Heike Trautmann
Place of Publication	Cham Switzerland
Publisher	Springer
Pages	111-124
Number of pages	14
ISBN (Electronic)	9783030581152
ISBN (Print)	9783030581145
DOIs	https://doi.org/10.1007/978-3-030-58115-2_8
Publication status	Published - 2020
Externally published	Yes
Event	Parallel Problem Solving from Nature 2020 - Leiden, Netherlands Duration: 5 Sept 2020 → 9 Sept 2020 Conference number: 16th https://link.springer.com/book/10.1007/978-3-030-58115-2 (Proceedings)

Publication series

Name	Lecture Notes in Computer Science
Publisher	Springer
Volume	12270
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	Parallel Problem Solving from Nature 2020
Abbreviated title	PPSN 2020
Country/Territory	Netherlands
City	Leiden
Period	5/09/20 → 9/09/20
Internet address	https://link.springer.com/book/10.1007/978-3-030-58115-2 (Proceedings)

Keywords

Dimensionally-Aware GP
Fitness landscape
Genetic programming
Local optima network

Access to Document

10.1007/978-3-030-58115-2_8

Cite this

Durasevic, M., Jakobovic, D., Scoczynski Ribeiro Martins, M., Picek, S., & Wagner, M. (2020). Fitness landscape analysis of dimensionally-aware genetic programming featuring feynman equations. In T. Bäck, M. Preuss, A. Deutz, H. Wang, C. Doerr, M. Emmerich, & H. Trautmann (Eds.), Parallel Problem Solving from Nature – PPSN XVI - 16th International Conference, PPSN 2020 Leiden, The Netherlands, September 5–9, 2020 Proceedings, Part II (pp. 111-124). (Lecture Notes in Computer Science; Vol. 12270). Springer. https://doi.org/10.1007/978-3-030-58115-2_8

Durasevic, Marko ; Jakobovic, Domagoj ; Scoczynski Ribeiro Martins, Marcella et al. / Fitness landscape analysis of dimensionally-aware genetic programming featuring feynman equations. Parallel Problem Solving from Nature – PPSN XVI - 16th International Conference, PPSN 2020 Leiden, The Netherlands, September 5–9, 2020 Proceedings, Part II. editor / Thomas Bäck ; Mike Preuss ; André Deutz ; Hao Wang ; Carola Doerr ; Michael Emmerich ; Heike Trautmann. Cham Switzerland : Springer, 2020. pp. 111-124 (Lecture Notes in Computer Science).

@inproceedings{36fc05d6f5b44cf4a0418f6e2752cd65,

title = "Fitness landscape analysis of dimensionally-aware genetic programming featuring feynman equations",

abstract = "Genetic programming is an often-used technique for symbolic regression: finding symbolic expressions that match data from an unknown function. To make the symbolic regression more efficient, one can also use dimensionally-aware genetic programming that constrains the physical units of the equation. Nevertheless, there is no formal analysis of how much dimensionality awareness helps in the regression process. In this paper, we conduct a fitness landscape analysis of dimensionally-aware genetic programming search spaces on a subset of equations from Richard Feynman{\textquoteright}s well-known lectures. We define an initialisation procedure and an accompanying set of neighbourhood operators for conducting the local search within the physical unit constraints. Our experiments show that the added information about the variable dimensionality can efficiently guide the search algorithm. Still, further analysis of the differences between the dimensionally-aware and standard genetic programming landscapes is needed to help in the design of efficient evolutionary operators to be used in a dimensionally-aware regression.",

keywords = "Dimensionally-Aware GP, Fitness landscape, Genetic programming, Local optima network",

author = "Marko Durasevic and Domagoj Jakobovic and {Scoczynski Ribeiro Martins}, Marcella and Stjepan Picek and Markus Wagner",

note = "Funding Information: Acknowledgements. We{\textquoteright}d like to thank Prof. Pablo Moscato for introducing us to [17]. We{\textquoteright}d also like to acknowledge support by the Australian Research Council, project DP200102364. Publisher Copyright: {\textcopyright} Springer Nature Switzerland AG 2020.; Parallel Problem Solving from Nature 2020, PPSN 2020 ; Conference date: 05-09-2020 Through 09-09-2020",

year = "2020",

doi = "10.1007/978-3-030-58115-2_8",

language = "English",

isbn = "9783030581145",

series = "Lecture Notes in Computer Science",

publisher = "Springer",

pages = "111--124",

editor = "Thomas B{\"a}ck and Mike Preuss and Andr{\'e} Deutz and Hao Wang and Carola Doerr and Michael Emmerich and Heike Trautmann",

booktitle = "Parallel Problem Solving from Nature – PPSN XVI - 16th International Conference, PPSN 2020 Leiden, The Netherlands, September 5–9, 2020 Proceedings, Part II",

url = "https://link.springer.com/book/10.1007/978-3-030-58115-2",

}

Durasevic, M, Jakobovic, D, Scoczynski Ribeiro Martins, M, Picek, S & Wagner, M 2020, Fitness landscape analysis of dimensionally-aware genetic programming featuring feynman equations. in T Bäck, M Preuss, A Deutz, H Wang, C Doerr, M Emmerich & H Trautmann (eds), Parallel Problem Solving from Nature – PPSN XVI - 16th International Conference, PPSN 2020 Leiden, The Netherlands, September 5–9, 2020 Proceedings, Part II. Lecture Notes in Computer Science, vol. 12270, Springer, Cham Switzerland, pp. 111-124, Parallel Problem Solving from Nature 2020, Leiden, Netherlands, 5/09/20. https://doi.org/10.1007/978-3-030-58115-2_8

Fitness landscape analysis of dimensionally-aware genetic programming featuring feynman equations. / Durasevic, Marko; Jakobovic, Domagoj; Scoczynski Ribeiro Martins, Marcella et al.
Parallel Problem Solving from Nature – PPSN XVI - 16th International Conference, PPSN 2020 Leiden, The Netherlands, September 5–9, 2020 Proceedings, Part II. ed. / Thomas Bäck; Mike Preuss; André Deutz; Hao Wang; Carola Doerr; Michael Emmerich; Heike Trautmann. Cham Switzerland: Springer, 2020. p. 111-124 (Lecture Notes in Computer Science; Vol. 12270).

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

TY - GEN

T1 - Fitness landscape analysis of dimensionally-aware genetic programming featuring feynman equations

AU - Durasevic, Marko

AU - Jakobovic, Domagoj

AU - Scoczynski Ribeiro Martins, Marcella

AU - Picek, Stjepan

AU - Wagner, Markus

N1 - Conference code: 16th

PY - 2020

Y1 - 2020

N2 - Genetic programming is an often-used technique for symbolic regression: finding symbolic expressions that match data from an unknown function. To make the symbolic regression more efficient, one can also use dimensionally-aware genetic programming that constrains the physical units of the equation. Nevertheless, there is no formal analysis of how much dimensionality awareness helps in the regression process. In this paper, we conduct a fitness landscape analysis of dimensionally-aware genetic programming search spaces on a subset of equations from Richard Feynman’s well-known lectures. We define an initialisation procedure and an accompanying set of neighbourhood operators for conducting the local search within the physical unit constraints. Our experiments show that the added information about the variable dimensionality can efficiently guide the search algorithm. Still, further analysis of the differences between the dimensionally-aware and standard genetic programming landscapes is needed to help in the design of efficient evolutionary operators to be used in a dimensionally-aware regression.

AB - Genetic programming is an often-used technique for symbolic regression: finding symbolic expressions that match data from an unknown function. To make the symbolic regression more efficient, one can also use dimensionally-aware genetic programming that constrains the physical units of the equation. Nevertheless, there is no formal analysis of how much dimensionality awareness helps in the regression process. In this paper, we conduct a fitness landscape analysis of dimensionally-aware genetic programming search spaces on a subset of equations from Richard Feynman’s well-known lectures. We define an initialisation procedure and an accompanying set of neighbourhood operators for conducting the local search within the physical unit constraints. Our experiments show that the added information about the variable dimensionality can efficiently guide the search algorithm. Still, further analysis of the differences between the dimensionally-aware and standard genetic programming landscapes is needed to help in the design of efficient evolutionary operators to be used in a dimensionally-aware regression.

KW - Dimensionally-Aware GP

KW - Fitness landscape

KW - Genetic programming

KW - Local optima network

UR - http://www.scopus.com/inward/record.url?scp=85091167563&partnerID=8YFLogxK

U2 - 10.1007/978-3-030-58115-2_8

DO - 10.1007/978-3-030-58115-2_8

M3 - Conference Paper

AN - SCOPUS:85091167563

SN - 9783030581145

T3 - Lecture Notes in Computer Science

SP - 111

EP - 124

BT - Parallel Problem Solving from Nature – PPSN XVI - 16th International Conference, PPSN 2020 Leiden, The Netherlands, September 5–9, 2020 Proceedings, Part II

A2 - Bäck, Thomas

A2 - Preuss, Mike

A2 - Deutz, André

A2 - Wang, Hao

A2 - Doerr, Carola

A2 - Emmerich, Michael

A2 - Trautmann, Heike

PB - Springer

CY - Cham Switzerland

T2 - Parallel Problem Solving from Nature 2020

Y2 - 5 September 2020 through 9 September 2020

ER -

Durasevic M, Jakobovic D, Scoczynski Ribeiro Martins M, Picek S, Wagner M. Fitness landscape analysis of dimensionally-aware genetic programming featuring feynman equations. In Bäck T, Preuss M, Deutz A, Wang H, Doerr C, Emmerich M, Trautmann H, editors, Parallel Problem Solving from Nature – PPSN XVI - 16th International Conference, PPSN 2020 Leiden, The Netherlands, September 5–9, 2020 Proceedings, Part II. Cham Switzerland: Springer. 2020. p. 111-124. (Lecture Notes in Computer Science). doi: 10.1007/978-3-030-58115-2_8

Fitness landscape analysis of dimensionally-aware genetic programming featuring feynman equations

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Cite this