Genetic Algorithms 101

You're viewing a comment by Jazeps and its responses.

December 06, 2007, 02:44

Hi, when beginning to work with GA, you should be aware that there are many techniques of all the stages of GA: selection, mutation, crossover, survival.

Elitist selection (selecting the fittest, eliminating the weakest) is not very popular, because it's easy to throw away some generally good chromosomes (leading to good ones) and fall into local maximums instead. Usually stochastic methods are preferred. For example, there is tournament selection. You select a number of individuals randomly and then select one, two or more best ones. There is also a roulette-wheel selection where you give each individual a probability of being selected according to its fitness. Each individual is given a probability of being selected based on its fitness, so that the fittest individuals are more likely to be selected but still not throwing away the weakest individuals.

There are also more ways to perform mutation. Bitflip is only of them. You may do a uniform mutation. You flip each bit with a probability of 1/[b] where [b] is the number of bits in chromosome.

Crossover also may take many forms: 1point, 2point, uniform.

Here is a good collection of introductory papers on evolutionary computation:

Evolutionary Computation 1
Basic Algorithms and Operators
Edited by Thomas Back, David B Fogel and Zbigniew Michalewicz

Reply to this comment

Reply To This Comment

(why do I need your e-mail?)

(Your twitter name, if you have one. (I'm @pkrumins, btw.))

Comment Help

Type first 3 letters of your name: (just to make sure you're a human)

Please preview the comment before submitting to make sure it's OK.