Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

**Symbolic Regression on Network Properties.** / Märtens, Marcus; Kuipers, Fernando; Van Mieghem, Piet.

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

Märtens, M, Kuipers, F & Van Mieghem, P 2017, Symbolic Regression on Network Properties. in J McDermott, M Castelli, L Sekanina, I Wolf & P García-Sánchez (eds), *Genetic Programming: Proceedings - 20th European Conference, EuroGP 2017.* Lecture Notes in Computer Science, vol. 10196, Springer, Cham, pp. 131-146, EuroGP 2017, Amsterdam, Netherlands, 19/04/17. https://doi.org/10.1007/978-3-319-55696-3_9

Märtens, M., Kuipers, F., & Van Mieghem, P. (2017). Symbolic Regression on Network Properties. In J. McDermott, M. Castelli, L. Sekanina, I. Wolf, & P. García-Sánchez (Eds.), *Genetic Programming: Proceedings - 20th European Conference, EuroGP 2017 *(pp. 131-146). (Lecture Notes in Computer Science; Vol. 10196). Springer. https://doi.org/10.1007/978-3-319-55696-3_9

Märtens M, Kuipers F, Van Mieghem P. Symbolic Regression on Network Properties. In McDermott J, Castelli M, Sekanina L, Wolf I, García-Sánchez P, editors, Genetic Programming: Proceedings - 20th European Conference, EuroGP 2017. Cham: Springer. 2017. p. 131-146. (Lecture Notes in Computer Science). https://doi.org/10.1007/978-3-319-55696-3_9

@inproceedings{afa574677ef941d5a816bf598c01cdd5,

title = "Symbolic Regression on Network Properties",

abstract = "Networks are continuously growing in complexity, which creates challenges for determining their most important characteristics. While analytical bounds are often too conservative, the computational effort of algorithmic approaches does not scale well with network size. This work uses Cartesian Genetic Programming for symbolic regression to evolve mathematical equations that relate network properties directly to the eigenvalues of network adjacency and Laplacian matrices. In particular, we show that these eigenvalues are powerful features to evolve approximate equations for the network diameter and the isoperimetric number, which are hard to compute algorithmically. Our experiments indicate a good performance of the evolved equations for several real-world networks and we demonstrate how the generalization power can be influenced by the selection of training networks and feature sets.",

keywords = "Cartesian genetic programming, Complex networks, Symbolic regression",

author = "Marcus M{\"a}rtens and Fernando Kuipers and {Van Mieghem}, Piet",

year = "2017",

doi = "10.1007/978-3-319-55696-3_9",

language = "English",

isbn = "978-3-319-55695-6",

series = "Lecture Notes in Computer Science",

publisher = "Springer",

pages = "131--146",

editor = "James McDermott and Mauro Castelli and Lukas Sekanina and Ina Wolf and Pablo Garc{\'i}a-S{\'a}nchez",

booktitle = "Genetic Programming",

note = "EuroGP 2017 : 20th European Conference on Genetic Programming ; Conference date: 19-04-2017 Through 21-04-2017",

}

TY - GEN

T1 - Symbolic Regression on Network Properties

AU - Märtens, Marcus

AU - Kuipers, Fernando

AU - Van Mieghem, Piet

N1 - Conference code: 20

PY - 2017

Y1 - 2017

N2 - Networks are continuously growing in complexity, which creates challenges for determining their most important characteristics. While analytical bounds are often too conservative, the computational effort of algorithmic approaches does not scale well with network size. This work uses Cartesian Genetic Programming for symbolic regression to evolve mathematical equations that relate network properties directly to the eigenvalues of network adjacency and Laplacian matrices. In particular, we show that these eigenvalues are powerful features to evolve approximate equations for the network diameter and the isoperimetric number, which are hard to compute algorithmically. Our experiments indicate a good performance of the evolved equations for several real-world networks and we demonstrate how the generalization power can be influenced by the selection of training networks and feature sets.

AB - Networks are continuously growing in complexity, which creates challenges for determining their most important characteristics. While analytical bounds are often too conservative, the computational effort of algorithmic approaches does not scale well with network size. This work uses Cartesian Genetic Programming for symbolic regression to evolve mathematical equations that relate network properties directly to the eigenvalues of network adjacency and Laplacian matrices. In particular, we show that these eigenvalues are powerful features to evolve approximate equations for the network diameter and the isoperimetric number, which are hard to compute algorithmically. Our experiments indicate a good performance of the evolved equations for several real-world networks and we demonstrate how the generalization power can be influenced by the selection of training networks and feature sets.

KW - Cartesian genetic programming

KW - Complex networks

KW - Symbolic regression

UR - http://www.scopus.com/inward/record.url?scp=85015755756&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-55696-3_9

DO - 10.1007/978-3-319-55696-3_9

M3 - Conference contribution

AN - SCOPUS:85015755756

SN - 978-3-319-55695-6

T3 - Lecture Notes in Computer Science

SP - 131

EP - 146

BT - Genetic Programming

A2 - McDermott, James

A2 - Castelli, Mauro

A2 - Sekanina, Lukas

A2 - Wolf, Ina

A2 - García-Sánchez, Pablo

PB - Springer

CY - Cham

T2 - EuroGP 2017

Y2 - 19 April 2017 through 21 April 2017

ER -

ID: 31753354