Re: Doubts on genetic algorithm
- From: "eroffe" <eroffe@xxxxxxxxx>
- Date: 18 May 2006 10:31:04 -0700
Hi Vinicius,
I'm not sure if I understood correctly you question. If I'm not wrong,
your questions is more basic.
To create your initial population will be created from several VALID
paths from the rat to the cheese, even though those paths don't need to
be optimum.
Choosing/Creating the initial population is a very important step,
since if the solutions are not spread enough your GA could converge
very quickly.
As you would be working with path from one point to the other, I think
the best for you is to choose a tree-based representation. That way,
could be easiest for you to create alternative trees as members of an
initial population.
Also, I have another suggestion. I don't know if you are forced to use
GA to solve this problem, but I did some similar application to control
an autonomous robot that have to travel from one point to the other
dodging obstacles. For that case I used the A* algorithms that is more
efficient than GA. Anyway, both GA and A* will work.
Hope this helps.
.
- References:
- Doubts on genetic algorithm
- From: vps.santos@xxxxxxxxx
- Doubts on genetic algorithm
- Prev by Date: [CFP] IDEAL 06 - 7th International Conference Intelligent Data Engineering & Automated Learning
- Next by Date: U.S university that has EC / related programs
- Previous by thread: Re: Doubts on genetic algorithm
- Next by thread: [CFP] IDEAL 06 - 7th International Conference Intelligent Data Engineering & Automated Learning
- Index(es):
Relevant Pages
|