MAKHFI.COM
Fascinating World of Neural Nets  
Neural Network Forums
Home PageHome Page : General Neural Networks Topics : Neural Network With Genetic Algorithm Plzzz help
  You are currently not logged in. You can view the forums, but cannot post messages. | HOME | Log In | Register | Search | Help
Post a Reply on This Topic Post a Reply on This Topic

Author Topic: Neural Network With Genetic Algorithm Plzzz help
zaza_thegreat Posted: 26-Jan-11 20:19
  Edit Edit
 
Email the Author Mail   View Author's Profile Profile  
im trying to generate weight vector for a certain neural network
i have a file of inputs and their output
(x1 : number , x2: number o: number)
hidden layer activation function is sigmoid
output layer activation function is linear

anyway this isnt the problem the problem is i do calculate he output of the chromosome and get its error for each set of i/o from the file
and then i get MSE = 1/2 sum(myoutput - files_output)^2
i make the fitness evaluation 1/mse to be able to minimize
then xover with prob 0.7 and mutation with prob 0.2
with generalized replacement
i never manage to get MSE below 45,000 which is enormously large what is my problem is it with the fitness or the neural calculation is wrong
please guys i need your help !!

 
Nikola Posted: 31-Jan-11 02:20
Delete Delete    Edit Edit
 
Email the Author Mail   View Author's Profile Profile  
Try with

MSE = (1 / n) * sum((myoutput - files_output)^2)

where n is the number of training examples (sets of i/o, if you will).
 

Post a Reply on This Topic Post a Reply on This Topic
 

Copyright © 2001-2003 Pejman Makhfi. All rights Reserved.