This chapter demonstrates a novel method combiningparticle swarm, differential evolution, and geneticprogramming to build a symbolic regression tool forlarge-scale, time-constrained regression-classificationproblems. In a previous paper we experimented withlarge scale symbolic regression. Here we describe indetail the enhancements and techniques employed tosupport large scale, time-constrained regression andclassification. In order to achieve the level ofperformance reported here, of necessity, we borrowed anumber of ideas from disparate schools of geneticprogramming and recombined them in ways not normallyseen in the published literature. We discuss in somedetail the construction of the fitness function, theuse of abstract grammars to combine genetic programmingwith differential evolution and particle swarm agents,and the use of context-aware crossover.
CITATION STYLE
Korns, M. F. (2007). Large-Scale, Time-Constrained Symbolic Regression-Classification. In Genetic Programming Theory and Practice V (pp. 53–68). Springer US. https://doi.org/10.1007/978-0-387-76308-8_4
Mendeley helps you to discover research relevant for your work.