Elman networks created by NEWELM function have a standard function that return positive or negative values of neuron activity...
I'd like to change it to a sigmoid function [ activity = 1 / (1 + e^-input) ], that return values from 0 to 1.
Is it possible?
Thank you
RP

On Aug 25, 5:39=A0pm, "R P" <rpa...@colband.com.br> wrote:
> Elman networks created by NEWELM function have a standard function that r=
eturn positive or negative values of neuron activity...
> I'd like to change it to a sigmoid function [ activity =3D 1 / (1 + e^-in=
put) ], that return values from 0 to 1.
> Is it possible?
> Thank you
> RP

from

help newelm

or


http://www.mathworks.com/access/helpdesk/help/toolbox/nnet/newelm.html

get

newelm
Purpose
Create Elman backpropagation network

Syntax
net =3D newelm(P,T,[S1 S2...S(N-l)],{TF1 TF2...TFNl},
BTF,BLF,PF,IPF,OPF,DDF)

Description
newelm(P,T,[S1 S2...S(N-l)],{TF1 TF2...TFNl}, BTF,BLF,PF,
IPF,OPF,DDF) takes these arguments,

 P       R x Q1 matrix of Q1 sample R-element input vectors
T         SN x Q2 matrix of Q2 sample SN-element input vectors
Si        Size of ith layer, for N-1 layers, default =3D [ ].
          (Output layer size SN is determined from T.)
TFi      Transfer function of ith layer. (Default =3D 'tansig' for
          hidden layers and 'purelin' for output layer.)

Therefore you have to override the 'purelin' default with an
explicit declaration of 'logsig' :

net =3D newelm( P, T, S1, { 'tansig' 'logsig' }, ...

However,

BTF    Backpropagation network training function (default =3D 'trainlm')

and the remainder of the documentation says that 'trainlm' is
probably not a good choice.

So, read the complete documentation on the website in addition to

doc newelm.

Hope this helps.

Greg