Evolutionary computation is a stochastic process. Each time you run
with different random number seeds, you're going to get a different
result. John Koza is just showing *one* possible result from *one*
run he did.
Sean
On Oct 19, 2010, at 8:19 PM, Daniel José Abate wrote:
> or asking better.... is it the expected output for santafe.trl?
>
> Thanks,
> Daniel.
>
> 2010/10/18 Daniel José Abate <[log in to unmask]>
> Hi all,
>
> I am doing some test with the ant example. I read in the Genetic
> programming Koza's book and in the page 154 says that he finds an
> individual in the generation 21 that eats all the food, but when I
> run the ECJ example, the best individual of the run appears en the
> generation 2 with 56 hits.
>
> What am I doing wrong ?
>
> Thanks in advance,
> Daniel.
>
>
|