Design of crossoveroperator plays a crucial role in Genetic Programming (GP). The most studied issues related to crossoveroperator in GP are: (i) ensuring that crossoveroperator always produces syntactically valid ...
详细信息
ISBN:
(纸本)9788132210375
Design of crossoveroperator plays a crucial role in Genetic Programming (GP). The most studied issues related to crossoveroperator in GP are: (i) ensuring that crossoveroperator always produces syntactically valid individuals (ii) improving search efficiency of crossoveroperator. These issues become crucial when the individuals are represented using linear string representation. This paper aims to introduce postfix GP approach to symbolic regression for solving empirical modeling problems. The main contribution includes (i) a linear string (postfix notation) based genome representation method and stack based evaluation to reduce space-time complexity of GP algorithm (ii) ensuring that sub-treecrossoveroperator always produces syntactically valid genomes in linear string representation (iii) using semantic information of sub-trees, to be swapped, while designing crossoveroperator for linear genome representation to provide additional search guidance. The proposed method is tested on two real valued symbolic regression problems. Two different constant creation techniques for Postfix GP, one that explicitly use list of constants and another without use of the list, are presented to evolve useful numeric constants for symbolic regression problems. The results on tested problems show that postfix GP comprised of semanticsub-treecrossover offers a new possibility for efficiently solving empirical modeling problems.
暂无评论