Prev: imline question
Next: Fitness proportionate / roulette-wheel selection, when initial fitness is 0
From: Viswanathan on 29 Jan 2010 21:08
Hi,

I use the "train" function to train my networks and I know that by default this function uses "mean squared error" as network performance error. When I calculate the MSE of the errors after simulation it is not equal to the validation performance error of the network. Could you please explain why? Thanks.

Cheers,
Visu
 | 
Pages: 1
Prev: imline question
Next: Fitness proportionate / roulette-wheel selection, when initial fitness is 0