开发者

Question on neural network matlab code

Hi i found this code somewhere with little info with it. it's suppose to be a backpropagation neural network code. but it seem to be lacking of something like weight and bias. is the code correct? is it test-while-train backpropagation neural network? thanks

    % --- Executes on button press in pushbutton6.

    %~~~~~~~~~~~[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('D:\MATLAB\datatrain.csv');   

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

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


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


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

     Y = sim(net,P);   % training

     save 'net7' net;


   % --- Executes on button press in pushbutton4. 

   %~~~~~~load data~~~~~~~~~~~~~~~~~~~~~~~
    mt = xlsread('D:\MATLAB\datatest.csv');  

    %~~~~~~convert the data in Matrix form~~~~
    [row1,col1] = size(mt);     
    Pt= mt(1:row1,1:10)';
    Tt = mt(1:row1, col1)';  

    load 'net7' -mat;
    Yt= sim(net,Pt); 

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


    if result>0.7
        error=1-result;
        set(handles.edit39,'String','yes')
        set(handles.edit40,'String',num2str(error))
        set(handles.edit41,'String','Completed')
        data1=[num2str(result) ];   
        fid = fopen('D:\MATLAB\record.csv','a+'); 
        fprintf(fid,[data1,'\n']);
        fclose(fid);


    else
        set(handles.edit39,'String','no')
        set(handles.edit40,'String',num2st开发者_运维百科r(result))
        set(handles.edit41,'String','Completed')
        data1=[num2str(result) ];           
        fid = fopen('D:\MATLAB\record.csv','a+'); 
        fprintf(fid,[data1,'\n']);
        fclose(fid);
    end    


The code is correct. Neural network weights and biases are stored inside net structure, you can access them via net.IW and net.LW structures. Biases are stored inside net.b. This code train a network using inputs P and targets T1, splitting them in training, testing and validations subsets used during training. Check the documentation for further information about training procedure.

0

上一篇:

下一篇:

精彩评论

暂无评论...
验证码 换一张
取 消

最新问答

问答排行榜