TY - GEN
T1 - Offspring selection genetic algorithm revisited
T2 - 16th International Conference on Computer Aided Systems Theory, EUROCAST 2017
AU - Affenzeller, Michael
AU - Burlacu, Bogdan
AU - Winkler, Stephan
AU - Kommenda, Michael
AU - Kronberger, Gabriel
AU - Wagner, Stefan
N1 - Publisher Copyright:
© Springer International Publishing AG 2018.
Copyright:
Copyright 2018 Elsevier B.V., All rights reserved.
PY - 2018
Y1 - 2018
N2 - This paper proposes some algorithmic extensions to the general concept of offspring selection which itself is an algorithmic extension of genetic algorithms and genetic programming. Offspring selection is characterized by the fact that many offspring solution candidates will not participate in the ongoing evolutionary process if they do not achieve the success criterion. The algorithmic enhancements proposed in this contribution aim to early estimate if a solution candidate will not be accepted based on partial solution evaluation. The qualitative characteristics of offspring selection are not affected by this means. The discussed variant of offspring selection is analyzed for several symbolic regression problems with offspring selection genetic programming. The achievable gains in terms of efficiency are remarkable especially for large data-sets.
AB - This paper proposes some algorithmic extensions to the general concept of offspring selection which itself is an algorithmic extension of genetic algorithms and genetic programming. Offspring selection is characterized by the fact that many offspring solution candidates will not participate in the ongoing evolutionary process if they do not achieve the success criterion. The algorithmic enhancements proposed in this contribution aim to early estimate if a solution candidate will not be accepted based on partial solution evaluation. The qualitative characteristics of offspring selection are not affected by this means. The discussed variant of offspring selection is analyzed for several symbolic regression problems with offspring selection genetic programming. The achievable gains in terms of efficiency are remarkable especially for large data-sets.
KW - Symbolic Regression
KW - Symbolic Regression
UR - http://www.scopus.com/inward/record.url?scp=85041825607&partnerID=8YFLogxK
U2 - 10.1007/978-3-319-74718-7_51
DO - 10.1007/978-3-319-74718-7_51
M3 - Conference contribution
SN - 9783319747170
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 424
EP - 431
BT - Computer Aided Systems Theory – EUROCAST 2017 - 16th International Conference, Revised Selected Papers
A2 - Moreno-Diaz, Roberto
A2 - Quesada-Arencibia, Alexis
A2 - Pichler, Franz
PB - Springer
Y2 - 19 February 2017 through 24 February 2017
ER -