The Generalisation Ability of a Selection Architecture for Genetic Programming

Authors:
David Jackson
Affiliations:
Dept. of Computer Science, University of Liverpool, Liverpool, United Kingdom L69 3BX
Venue:
Proceedings of the 10th international conference on Parallel Problem Solving from Nature: PPSN X
Year:
2008

Citing 10
Cited 0

Genetic programming: on the programming of computers by means of natural selection

Genetic programming: on the programming of computers by means of natural selection
Genetic programming II: automatic discovery of reusable programs

Genetic programming II: automatic discovery of reusable programs
Discovery of subroutines in genetic programming

Advances in genetic programming
An Evaluation of EvolutionaryGeneralisation in Genetic Programming

Artificial Intelligence Review
The Effect of Extensive Use of the Mutation Operator on Generalization in Genetic Programming Using Sparse Data Sets

PPSN IV Proceedings of the 4th International Conference on Parallel Problem Solving from Nature
Multi-optimization improves genetic programming generalization ability

Proceedings of the 9th annual conference on Genetic and evolutionary computation
Benchmarking the generalization capabilities of a compiling genetic programming system using sparse data sets

GECCO '96 Proceedings of the 1st annual conference on Genetic and evolutionary computation
Generality versus size in genetic programming

GECCO '96 Proceedings of the 1st annual conference on Genetic and evolutionary computation
The performance of a selection architecture for genetic programming

EuroGP'08 Proceedings of the 11th European conference on Genetic programming
Genetic programming, validation sets, and parsimony pressure

EuroGP'06 Proceedings of the 9th European conference on Genetic Programming

Quantified Score

Hi-index	0.00

Visualization

Abstract

As an alternative to various existing approaches to incorporating modular decomposition and reuse in genetic programming (GP), we have proposed a new method for hierarchical evolution. Based on a division of the problem's test case inputs into subsets, it employs a program structure that we refer to as a selection architecture. Although the performance of GP systems based on this architecture has been shown to be superior to that of conventional systems, the nature of evolved programs is radically different, leading to speculation as to how well such programs may generalise to deal with previously unseen inputs. We have therefore performed additional experimentation to evaluate the approach's generalisation ability, and have found that it seems to stand up well against standard GP in this regard.