DOI

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.

Original languageEnglish
Title of host publicationGenetic Programming
Subtitle of host publicationProceedings - 20th European Conference, EuroGP 2017
EditorsJames McDermott, Mauro Castelli, Lukas Sekanina, Ina Wolf, Pablo García-Sánchez
Place of PublicationCham
PublisherSpringer International Publishing AG
Pages131-146
Number of pages16
ISBN (Electronic)978-3-319-55696-3
ISBN (Print)978-3-319-55695-6
DOIs
StatePublished - 2017
EventEuroGP 2017 - Amsterdam, Netherlands
Duration: 19 Apr 201721 Apr 2017
Conference number: 20

Publication series

NameLecture Notes in Computer Science
Volume10196
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

ConferenceEuroGP 2017
CountryNetherlands
CityAmsterdam
Period19/04/1721/04/17

    Research areas

  • Cartesian genetic programming, Complex networks, Symbolic regression

ID: 31753354