relm_equalization.m

function x_output = relm_equalization(y_train,x_train,x_test,NN,gamma)
    NumberofHiddenNeurons=NN;
    
    size_train = size(x_train);
    size_test = size(x_test);
    
    NumberofTrainingData=size_train(2);
    NumberofTestingData=size_test(2);
    NumberofInputNeurons=size_train(1);
    InputWeight=rand(NumberofHiddenNeurons,NumberofInputNeurons)*2-1;
    
    BiasofHiddenNeurons=rand(NumberofHiddenNeurons,1);
    tempH=InputWeight*x_train+BiasofHiddenNeurons(:,ones(1,NumberofTrainingData));
    
    H = 1 ./ (1 + exp(-tempH)); % H * Output_Weights = Salida
    
    OW=pinv(1/gamma + H*H') * H * y_train';
    
    tempH_test=InputWeight*x_test + BiasofHiddenNeurons(:,ones(1,NumberofTestingData));
    H_test = 1 ./ (1 + exp(-tempH_test));
    x_output=(H_test' * OW)';
end