Genetic algorithm is population based search technique in which chromosome can be represented in several ways. During the evolution process selection of population is done at two places at the beginning and after crossover and mutation. There are different selection techniques exists. do the mixture of any specific selection and chromosome representation affects the overall GA performance. Please advise or provide appropriate literature, which can be utilize to carry this research.
Hi Shiv
Thank you for the response. will go through the links. I am a bit confuse and want understand the relationship or in other words want to understand that "is there any specific combination of chromosome representation and selection method affect the GA's performance."
As you mentioned solution is directly proportional to the problem one is dealing with via GA. Also, many previous research explained that chromosome representation also affect the performance same goes with selection technique. Therefore, want to investigate and trying to indetify any specific conbination which might improve the performance of GA.
Good question, the selection algorithm deals with the fitness of the chromosome not the representation, but the crossover and mutation will have huge impact on the fitness of the chromosome. the effect is indirect.
Selection focused on picking up an individual from the population. Whether it is a traditional binary encoding scheme, some numbers or another encoding scheme, the selection operator should not be aware of it. It is more related to fitness and the population in general.
It would be interesting though to find out if there is a link between these elements, although it is unlikely.
Patricia
Hi,
I do agree with answer that says the selection method has a direct connection with the fitness of the chromosome rather than its representation. In contrast to other GA operators (crossover, mutation, ...) which are greatly influenced by the representation. I think the connection between the selection method and chromosome representation should through its fitness evaluation.
Creation of a chromosome in GA in many cases can be considered as the art. In an ideal, the chromosome has to meet two requirements: (1) not to create 'senseless' individuals during the crossing and a mutation, (2) to provide the widest distribution of new descendants and mutations on area of possible decisions. It is desirable still that the chromosome had constant length. It significantly accelerates the work of GA. Sometimes multi-chromosomal approach helps (one chromosome breaks into two, three and more; then 4, 8 or more descendants are created when crossing ). Sometimes as a gene it is necessary to use complex elements (for example, matrix columns).
Look if it is possible:
http://ieeexplore .ieee.org/xpl/login.jsp? tp =&arnumber=5739061&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D5739061
Hi Igor
Thank you so much for sharing your detailed view. I shall surely look into the paper you suggested.
the question is quite interesting and need to be experimented first before saying anything,
but conceptually the chromosome representation method (either binary or real coded) does not have any effect on the selection method. as selection process(tournament, roulette wheel and ...) uses the fitness value of the chromosome for all its operations and this the responsibility of researcher to implement it correctly(to find the fitness value of the chromosomes by careful application of fitness function).
and one more thing the operators (cross-over and mutation) are implemented differently for different chromosome representations to avoid constrains violating individuals.
hello Hari,
actually the selection technique has nothing to do with the chromosome representation, however both affects the solution or in other words the value of fitness function. after you represent the chromosome properly you can use any selection method you desire keeping in mind that you need to do some tweaks to some parameters to achive best solution in lowest number of generations.
hope it helps
hello
the selection method is based on the fitness function and the chromosome coding is varied according to the problem type.
- Badges
- Science topic
- Similar topics
- Books