RawRecruit
MIS
Hello All,
I'm playing around with Neural Networks but can't get the sigmoid function 1/(1+ e -var)to work correctly.
That's exponential to the negative variable - can't use superscript in this message box!
log()works fine with a positive number, but as soon as a negative number is applied it falls over!
Can anyone shed any light on the problem? Am I using the correct function?
My code reads along the lines of....
var = 1/(1+(log(0-SubTotArray[0][$OutCount])))
and works OK if the '0-' is removed.
Any ideas?
If anyone needs clarification of the sigmoid function, it can be found at..... ... just below the diagrams.
Thanks in advance for all your assistance.
I'm playing around with Neural Networks but can't get the sigmoid function 1/(1+ e -var)to work correctly.
That's exponential to the negative variable - can't use superscript in this message box!
log()works fine with a positive number, but as soon as a negative number is applied it falls over!
Can anyone shed any light on the problem? Am I using the correct function?
My code reads along the lines of....
var = 1/(1+(log(0-SubTotArray[0][$OutCount])))
and works OK if the '0-' is removed.
Any ideas?
If anyone needs clarification of the sigmoid function, it can be found at..... ... just below the diagrams.
Thanks in advance for all your assistance.