Disertation!

This GA thingy

Okay, from reading this lady blog, I think it's a good idea to put what I have found so far about my disertation in a blog. It would help me on reviewing them later. So, let's just called this my own knowledge base system.
 
Here's what I found so far about GA (Genetic Algorithm).

GA is a search mechanism that use Darwinian method of survival of the fittest to find the best solution on the given problem. First, it randomly generates a group of solution strings called a population. A solution string is called a chromosome.  We then check whether this solutions is feasible or not. If not repair them. Then, a fitness function evaluates which of these chromosomes is the closest to the solution (the fittest). After we choose the closest solution strings, and execute 'genetic operations', which are crossover and mutation. Crossover would pick randomly of the selected 'fittest' solutions and interchange parts of the two chromosomes to make a new one.  Mutation is by randomly altering one or some of the bits in the string. These two operation would make a new generation of chromosomes. We run these new generation into the fitness function and decides which of these solutions would make it to the population or not.
 
After that, we repeat the steps on the new population, until we find the solution.
 
That's about all I got right now.

Things that are still vague:
1. How the hell do I create a string that could represent a whole network??
I got an idea of using direct encoding, which put represent network by using a matrix, and then represent the matrix into a string. It's like coordination system, we the x axis that list all the nodes, and we have the y axis that list all the nodes too. So say, we want to tell that there's a connection between node 3 and 4, we just put 1 in the (3,4) cell of the matrix. The thing is, I need not only just representing the connections, but also in representing, what kind of cable that we use, what kind of router that we used. How do I encode router and cable types into the chromosome string.

2. I only know how the mutation and crossover works in the binary encoding. But if I got around of point 1, then I don't think that would be using binary encoding.
 
my head ache, and I haven't put on the covers on my bed, oh well.

posted by Chafid @ 2:22 PM