start prev index next end Slide 15 of 18.


Often the crossover operator and selection method are too effective and they end up driving the genetic algorithm to create a population of individuals that are almost exactly the same. When the population consists of similar individuals, the likliehood of finding new solutions typically decreases.

On one hand, you want the genetic algorithm to find good individuals, but on the other you want it to maintain diversity. Two of the most common methods for maintaining diversity are DeJong-style crowding (named after Kenneth DeJong whose PhD thesis first explored the issue) and Goldberg-style fitness scaling.

DeJong-style crowding is basically the same thing as replace-most-similar replacement scheme. When new offspring are created, they replace the individuals in the population that are most similar to them.

Fitness scaling derates the objective score of individuals that are less unique than others in the population. By derating the scores of similar individuals, less similar individuals are more likely to be selected for mating.

Return to Matthew's Portfolio