My trained deep learning model provides the same output for all test data except the last sample.

1 次查看（过去 30 天）

显示更早的评论

ferda sonmez 2019-4-21

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/457664-my-trained-deep-learning-model-provides-the-same-output-for-all-test-data-except-the-last-sample

关闭： MATLAB Answer Bot 2021-8-20

在 MATLAB Online 中打开

Hi,

I have a trained model. During training I have tested my error level and it was very low. I saved the model in mat file. My test data contained 271 sample. Below I will paste the test code. When I run this code, All the results for first 270 samples are same, only the 271. result differs. I tried to change the order of the test data ( I moved the 271. sample to first order ) nothing changed. Still I have first 270 results same, 271. differs. The code has nothing that would cause this.

FERDAfreqdemogswlliwcnStatusB50Gender0Test = xlsread('D:\Dropbox\Satisfaction_Makale\Matlab Calisma Dosyalari\FERDA_freq_demog_swl_liwc_nStatusB50_Gender0 - Test.xlsx');
XMaleTest = FERDAfreqdemogswlliwcnStatusB50Gender0Test(:, [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44]);
YMaleTest = FERDAfreqdemogswlliwcnStatusB50Gender0Test(:, [45]);
XMaleTest = Normalize(XMaleTest);
YMaleTest = Normalize(YMaleTest);
load('DeepNeuralNetworkMale2.mat');
inputMatrixMale = XMaleTest;
nRows = size(inputMatrixMale, 1); 
testOutput = zeros(nRows, 1);
for k = 1:nRows 
    subMatrix = double(inputMatrixMale(k:k, :));
    if(k == 270 )
    disp("Durdum");
    end
    input_of_hidden_layer1 = w1* transpose(subMatrix);
    input_of_hidden_layer1 = vpa(input_of_hidden_layer1);
    output_of_hidden_layer1 = Sigmoid(input_of_hidden_layer1);%Relu idi
    output_of_hidden_layer1 = vpa(output_of_hidden_layer1);
    input_of_hidden_layer2 = (w2) * (output_of_hidden_layer1);%bunun sonucu inf ç?k?yor
    input_of_hidden_layer2 = vpa(input_of_hidden_layer2);
    output_of_hidden_layer2 = Sigmoid(input_of_hidden_layer2);%Relu idi
    output_of_hidden_layer2 = vpa(output_of_hidden_layer2);
    input_of_hidden_layer3 = w3* output_of_hidden_layer2;
    input_of_hidden_layer3 = vpa(input_of_hidden_layer3);
    output_of_hidden_layer3 = Sigmoid(input_of_hidden_layer3); %Relu idi
    output_of_hidden_layer3 = vpa(output_of_hidden_layer3);
    input_of_output_node = (w4) * (output_of_hidden_layer3);
    input_of_output_node = vpa(input_of_output_node);
    final_output = Sigmoid(input_of_output_node);
    final_output = vpa(final_output);
    disp(["final_output=" , num2str(double(final_output))]);
    testOutput(k) = final_output;
end 