Re: Basic ANN question

[EMAIL PROTECTED] (Greg Heath) wrote in message 
news:<[EMAIL PROTECTED]>...
> Peder Ydalus <[EMAIL PROTECTED]> wrote in message 
> news:<[EMAIL PROTECTED]>...
> > When making an ANN with input, hidden and output layer; 
> > am I supposed to use the sigmoid function (or similar) for all hidden
> > nodes and the output node?
> > In that case, won't this always map the result to <0, 1>?
> 
> My general advice is to use tanh for hidden and linear for output.
> However, if the output has natural hard limits it can be scaled and
> tanh or sigmoid can be used in the output layer. If, in addition,
> the outputs are mutually exclusive, 

and obey a linear equality constraint

> softmax is appropriate.
> 
> See the FAQ.
> 
> Hope this helps.
> 
> Greg