[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
AW: OCTAVE: NN Tool box
From: |
address@hidden |
Subject: |
AW: OCTAVE: NN Tool box |
Date: |
Sun, 4 Jul 2010 12:17:44 +0000 (GMT+00:00) |
Hi,
I'm not sure if I understand you correctly.
I would say a "represantation" of your network could be written as:
NT =sum[ (w2 * x2)]*transferFunction + sum[(w1 * I)]*transferFunction
where are: w2 / w1 the layer weights depending
on the layer, means w2 is the second layer weight matrix, w1 is the input layer
weight matrix and so on
where are: x2 /
x1 (or I) are the inputs to the corresponding weight matrix ...
transferFunction can be chosen...
in your case:
NT =
sum[(w2*x2)]*purelin + sum[w1*I]*tansig
in the wwware a lot of informations about the network.
If you need more help, please write a little bit more detailed
what you are searching for ...
regards
Michael
Dear all,
I have been exploring octave's neural network tool box. I
have a question.
For an input array 'I', with the corresponding target array 'T', I have a
neural network fitted target
array 'NT'.
This is the fitting scheme which I used
net=newff( [Imin Imax], [10 1], {'tansig', 'purelin'});
with
sufficiently large net.trainParam.epochs.
How can I represent the fitted array 'NT' as a function of the input array
'I'. This must involve using the functions 'tansig', 'purelin' which I used
for the hidden and output layer and also
the number of layers.
Thank you in advance for any help.
regards,
octuser
--
View this message in context:
http://octave.1599824.n4.nabble.com/OCTAVE-NN-Tool-box-tp2276236p2276236.htmlSent
from
the Octave - General mailing list archive at Nabble.com.
_______________________________________________
Help-octave
mailing list
address@hidden
https://www-old.cae.wisc.edu/mailman/listinfo/help-octave
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- AW: OCTAVE: NN Tool box,
address@hidden <=