It is possible to save the all struct consecutively using Neural Network

Hi guyss!!! please help me with my problem :)

I created a routine to save the output value of a sequence of routines neural networks.

I can save the output values for each routine I do, but i dont know how to save the file tr (file structure) and know the basics and weights of each routine made.

What do I need to add to save all data for each routine made? (tr,weight, etc..)

I need to do 30 routines of neural networks.

for i = 1:30
% Create a Fitting Network
hiddenLayerSize = 8;
net = fitnet(hiddenLayerSize);
% Set up Division of Data for Training, Validation, Testing
net.divideParam.trainRatio = 70/100;
net.divideParam.valRatio = 15/100;
net.divideParam.testRatio = 15/100;
[pn,ps] = mapminmax(p);
[tn,ts] = mapminmax(t);
[net,tr] = train(net,pn,tn);
an = sim(net,pn);
a = mapminmax('reverse',an,ts);
Output(:,i)=a;
end

Answer 1

[per isakson]

There are many ways to achieve "... save all data for each routine made? (tr,weight, etc..)", one of which store the result and supplementary data in a structure array. (I guess, "each routine" means each iteration of the loop.) Try

>> out = cssm( p, t )
out = 
1x30 struct array with fields:
    Output
    ps
    tr
    weight
    etc

where

function  out = cssm( p, t )
    len = 30;
    out = struct( 'Output',cell(1,len), 'ps',[], 'tr',[], 'weight',[], 'etc','' );  
    for jj = 1:len
        % Create a Fitting Network
        hiddenLayerSize = 8;
        net = fitnet( hiddenLayerSize );
        % Set up Division of Data for Training, Validation, Testing
        net.divideParam.trainRatio  = 70/100;
        net.divideParam.valRatio    = 15/100;
        net.divideParam.testRatio   = 15/100;
        [pn,ps] = mapminmax(p);
        [tn,ts] = mapminmax(t);
        [net,tr] = train(net,pn,tn);
        an = sim(net,pn);
        a = mapminmax('reverse',an,ts);
        out(jj).Output = a;
        out(jj).ps  = ps;
        out(jj).tr  = tr;
        out(jj).etc = 'et cetera';
    end
end

[per isakson]

You must store cssm in a separate file, cssm.m, containing no other executable code. See http://uk.mathworks.com/matlabcentral/answers/293720#comment_377711

[Mary Smith]

Really work, i love you so much !!! thank for all.

XOXO