Genetic Programming (GP) is a machine learning technique that evolves programs using natural selection and populations dynamics. Much of the functionality of GP depends on the representation of programs in the population and how to handle illegal or type incoherent expressions that arise from crossover and mutation within a population of programs. The SKGP is a GP system that uses graphs of combinators to represent functions and a strong type system to inform the crossover and mutation operations during evolution. This produces a powerful, flexible system that has many benefits over more conventional systems. This paper describes the implementation of this system, gives some examples of successful applications constructed using the SKGP and describes future directions that may offer a more powerful GP system capable of producing more complex programs.
CITATION STYLE
Worzel, W. P., & MacLean, D. (2015). SKGP: The Way of the Combinator (pp. 53–71). https://doi.org/10.1007/978-3-319-16030-6_4
Mendeley helps you to discover research relevant for your work.