Hybrid neural networks. - page 9

 
gumgum >> :

I'm thinking, what if n is reduced with each run?

Which n?

 
joo писал(а) >>

Which n?

Speed of learning.

 
Can you adjust the speed of learning? Can you tell me how?
 

Take a bell type function and go from min max min.

 
gumgum >> :
I think everyone here is obsessed with GA......

Gah ... Why so critical :) trendy direction etc., I have an "academic" task that is not directly related to this forum, but GA is very relevant there.

 
gumgum >> :

276 page

Is the book in paper or e-book form?

 
rip писал(а) >>

Is the book in paper or e-book form?

paper

 
joo >> :

Well said. Short. It's true I don't have a GA.

Several people have spoken with their algorithms, of course they are all implemented differently.

Here is a suggestion. Try to use the function I suggested on the 2nd page as a fitness function. How many runs of the fitness function before the algorithm stops? Everyone will take a look and decide for themselves what they are aiming for. I got an achievement on the 60th epoch, total epochs is 160, 200 individuals in population, total number of starts of FF function is 160*200=32000


I'll be giving it a go. I'll find time for programming (I really hope so :) at the weekend. Will it be possible to ask for support, if anything?
 
gumgum >> :

Speed of learning


Maybe you meant step change of weights. I still haven't figured out the idea.
 
Is it allowed to attach e-books to posts here?