In the past 20 years the use of evolutionary computation and the variable length representation of individuals has become increasingly popular, under the guise of Genetic Programming (GP). The ability of individuals (candidate solutions) to change length during evolution has led to some issues, one of which is the tendency for the average size of individuals to increase during the evolution of a solution. This concept is called "bloat".
Although previously studied from theoretical and practical viewpoints there has been little progress in deriving controls for bloat which do not explicitly refer to individual size. In this talk, an implicit approach to controlling bloat is presented through the use of spatial population structure in combination with local elitist replacement. Theoretical concepts regarding inbreeding and the role of elitism will be used to support the described approach. Results will be presented that show implicit bloat control is feasible while maintaining high quality solutions.
Last modified: Tuesday, 28-Apr-2009 08:28:45 NZST
This page is maintained by the seminar list administrator.