University of Otago logo. Computer and Information Science Seminars

Seminar Homepage


Peter Whigham, Department of Information Science


When science goes wrong: The case against Grammatical Evolution


Owheo 106 - 1:00 pm, Friday 24 May


Grammatical Evolution (GE) is a sub-field of evolutionary computation first described in 1998. GE, an extension of Genetic Programming (GP), incorporates a grammar to allow language and search bias within the framework of a population-based stochastic search algorithm. The main argument for the design of GE was that it incorporated biological principles, including a distinctive genotype-phenotype mapping. Although this introduced a significant number of issues the field has continued over the past 14 years, largely by studying the properties of GE and methods for its improvement. This talk will argue that the underlying assumptions of GE are false, show that the field has developed as a means to fix problems in the representation that should have been acknowledged from its inception, and argue that GE offers no new concepts that were not in existence by the mid-1990s. Is GE a legitimate field of research? You be the judge.

Last modified: Tuesday, 21-May-2013 15:37:06 NZST

This page is maintained by the seminar list administrator.