hi, i'm using gui to create a recognition system to identify 2
classes.
for training, i have 150 image- parameter already extracted stored in
a trainfile.for testing, 50 image for each classes parameter not
extracted yet

here's what i did, i've created a gui that can load an image, extract
parameter from it and store in datatrain.csv. after all the 150
image's parameter stored, i use this code to test the rest of 100
image.


 %~~~~~~~~~~~~~~~~~~~~~~~~use when test data~~~~~~~~~~~~~~~~~~~~~~
parameter=[num2str(M1) ',' num2str(M2) ',' num2str(M3) ',' num2str(M4)
',' num2str(M5) ',' num2str(M6) ',' num2str(M7) ',' num2str(perimeter)
',' num2str(area) ',' num2str(xbar) ',' num2str(ybar) ];
data=[filename ',' num2str(M1) ',' num2str(M2) ',' num2str(M3) ','
num2str(M4) ',' num2str(M5) ',' num2str(perimeter) ',' num2str(M6) ','
num2str(M7) ',' num2str(area) ',' num2str(xbar) ',' num2str(ybar)];
fid = fopen('C:\Documents and Settings\winxp\My Documents\MATLAB
\MATLABDATA\datatest.csv','a+');
fprintf(fid,[parameter,'\n']);
fclose(fid);

fid = fopen('C:\Documents and Settings\winxp\My Documents\MATLAB
\MATLABDATA\datatest.txt','a+');
fprintf(fid,[data,'\r\n']);
fclose(fid);

set(handles.edit45,'String','Load completed')

end

% --- Executes on button press in pushbutton4.
function pushbutton4_Callback(hObject, eventdata, handles)
 %~~~~~~~~~~~[L1 L2 1];first hidden layer,second & output layer~~~~~

        layer = [11 15 1];
        myepochs = 30;
        attemption = 1; %i;
        mytfn = {'tansig' 'tansig' 'purelin'};

        %~~~~~~load data~~~~~~~~~~~~~~~~~~~~~~~
        m = xlsread('C:\Documents and Settings\winxp\My Documents
\MATLAB\MATLABDATA\datatrain.csv');
        mt = xlsread('C:\Documents and Settings\winxp\My Documents
\MATLAB\MATLABDATA\datatest.csv');

        %~~~~~~convert the data in Matrix form~~~~
        [row,col] = size(m);
        [row1,col1] = size(mt);


        P = m(1:row,1:10)';

        Pt= mt(1:row1,1:10)';

        T1 = m(1:row, col)';   % target data for training...last
column
        Tt = mt(1:row1, col1)';  % target data for testing


        net = newff([minmax(P)],layer,mytfn,'trainlm');  %nnet
        net.trainParam.epochs = myepochs;   % how many time newff will
repeat the training
        net.trainParam.showWindow = false;
        net.trainParam.showCommandLine = true;
        net = train(net,P,T1);  % start training newff with input P
and target T1

        Y = sim(net,P);   % training
        Yt= sim(net,Pt); % test newff using testing data

        %~~~~~~~final result of the neural network~~~~~~~~
        [r,c]=size(Yt);
        result=Yt(c);


        if result>0.7
            error=1-result;
            set(handles.edit39,'String','PET')
            set(handles.edit40,'String',num2str(error))
            set(handles.edit41,'String','completed')
            data=[num2str(result) ];
            fid = fopen('C:\Documents and Settings\winxp\My Documents
\MATLAB\MATLABDATA\record.csv','a+'); %use it when test data
            fprintf(fid,[data,'\n']);
            fclose(fid);

        else
            set(handles.edit39,'String','Non-PET')
            set(handles.edit40,'String',num2str(result))
            set(handles.edit41,'String','completed')
            data=[num2str(result) ];
            fid = fopen('C:\Documents and Settings\winxp\My Documents
\MATLAB\MATLABDATA\record.csv','a+');
            fprintf(fid,[data,'\n']);
            fclose(fid);

        end

when i click pushbutton4, an image will load, and identify the type of
the image. i repeat this until all the 100 image are identified. then
i calculate the correct image percentage.

question:
1. is what i'm doing correct?
2. why does it seem like training when testing the image?
3. when i test the same image again, the answer is different from
before. why?
can u help me fix this code?

tq

amy <twintwinkle_twee@yahoo.com> wrote in message <c9917938-0202-4158-b270-cec8936ae638@a28g2000prb.googlegroups.com>...
> hi, i'm using gui to create a recognition system to identify 2
> classes.
> for training, i have 150 image- parameter already extracted stored in
> a trainfile.for testing, 50 image for each classes parameter not
> extracted yet
> 
> here's what i did, i've created a gui that can load an image, extract
> parameter from it and store in datatrain.csv. after all the 150
> image's parameter stored, i use this code to test the rest of 100
> image.
> 
> 
>  %~~~~~~~~~~~~~~~~~~~~~~~~use when test data~~~~~~~~~~~~~~~~~~~~~~
> parameter=[num2str(M1) ',' num2str(M2) ',' num2str(M3) ',' num2str(M4)
> ',' num2str(M5) ',' num2str(M6) ',' num2str(M7) ',' num2str(perimeter)
> ',' num2str(area) ',' num2str(xbar) ',' num2str(ybar) ];
> data=[filename ',' num2str(M1) ',' num2str(M2) ',' num2str(M3) ','
> num2str(M4) ',' num2str(M5) ',' num2str(perimeter) ',' num2str(M6) ','
> num2str(M7) ',' num2str(area) ',' num2str(xbar) ',' num2str(ybar)];
> fid = fopen('C:\Documents and Settings\winxp\My Documents\MATLAB
> \MATLABDATA\datatest.csv','a+');
> fprintf(fid,[parameter,'\n']);
> fclose(fid);
> 
> fid = fopen('C:\Documents and Settings\winxp\My Documents\MATLAB
> \MATLABDATA\datatest.txt','a+');
> fprintf(fid,[data,'\r\n']);
> fclose(fid);
> 
> set(handles.edit45,'String','Load completed')
> 
> end
> 
> % --- Executes on button press in pushbutton4.
> function pushbutton4_Callback(hObject, eventdata, handles)
>  %~~~~~~~~~~~[L1 L2 1];first hidden layer,second & output layer~~~~~
> 
>         layer = [11 15 1];
>         myepochs = 30;
>         attemption = 1; %i;
>         mytfn = {'tansig' 'tansig' 'purelin'};
> 
>         %~~~~~~load data~~~~~~~~~~~~~~~~~~~~~~~
>         m = xlsread('C:\Documents and Settings\winxp\My Documents
> \MATLAB\MATLABDATA\datatrain.csv');
>         mt = xlsread('C:\Documents and Settings\winxp\My Documents
> \MATLAB\MATLABDATA\datatest.csv');
> 
>         %~~~~~~convert the data in Matrix form~~~~
>         [row,col] = size(m);
>         [row1,col1] = size(mt);
> 
> 
>         P = m(1:row,1:10)';
> 
>         Pt= mt(1:row1,1:10)';
> 
>         T1 = m(1:row, col)';   % target data for training...last
> column
>         Tt = mt(1:row1, col1)';  % target data for testing
> 
> 
>         net = newff([minmax(P)],layer,mytfn,'trainlm');  %nnet
>         net.trainParam.epochs = myepochs;   % how many time newff will
> repeat the training
>         net.trainParam.showWindow = false;
>         net.trainParam.showCommandLine = true;
>         net = train(net,P,T1);  % start training newff with input P
> and target T1
> 
>         Y = sim(net,P);   % training
>         Yt= sim(net,Pt); % test newff using testing data
> 
>         %~~~~~~~final result of the neural network~~~~~~~~
>         [r,c]=size(Yt);
>         result=Yt(c);
> 
> 
>         if result>0.7
>             error=1-result;
>             set(handles.edit39,'String','PET')
>             set(handles.edit40,'String',num2str(error))
>             set(handles.edit41,'String','completed')
>             data=[num2str(result) ];
>             fid = fopen('C:\Documents and Settings\winxp\My Documents
> \MATLAB\MATLABDATA\record.csv','a+'); %use it when test data
>             fprintf(fid,[data,'\n']);
>             fclose(fid);
> 
>         else
>             set(handles.edit39,'String','Non-PET')
>             set(handles.edit40,'String',num2str(result))
>             set(handles.edit41,'String','completed')
>             data=[num2str(result) ];
>             fid = fopen('C:\Documents and Settings\winxp\My Documents
> \MATLAB\MATLABDATA\record.csv','a+');
>             fprintf(fid,[data,'\n']);
>             fclose(fid);
> 
>         end
> 
> when i click pushbutton4, an image will load, and identify the type of
> the image. i repeat this until all the 100 image are identified. then
> i calculate the correct image percentage.
> 
> question:
> 1. is what i'm doing correct?
> 2. why does it seem like training when testing the image?
> 3. when i test the same image again, the answer is different from
> before. why?
> can u help me fix this code?
> 
> tq




could u plz show me your code on training...ie trainfile.m...???working on the same project.. need help....!thnks!