Using PNN neural network method or by using RBF radial basis function, what would be the spread value? Which one will provide best result PNN or RBF

Sir I have pattern of 1440X3 ,target file as 1400X1 and test file of 18X1 size how to get test file by using PNN neural network method or by using RBF radial basis function what would be the spread value.which one will provide best result PNN or RBF

     clc
    %read dat in file

    P=load('pattern.txt')
    TC=load('target.txt')
    P=P'.*1000;
    TC=TC';

    P12=load('test.txt');
    % P1=P12(1,:)';
    P1=P12'.*1000;

    T=ind2vec(TC);
    net=newrbe(P,T,10);
    Y=sim(net,P1);
    Yc=vec2ind(Y)'
    S= sum (Yc)
    % length(Yc)
    L = length(Yc)
     x = ldivide(L,S)

Answer 1

[Greg Heath]

Try looking at some of my NEWRB posts in BOTH the NEWSGROUP and ANSWERS.

 newrb greg

Probably looking at the most recent ones first would be fruitful.

Alert me via email when you post a response. Aside from alerts, I do not communicate about NNs via email.

I have found that the best way to find a good spread value is by structured trial and error. I start with the default of 1. Then I look at powers of two to bracket the optimum value.

Also, it is wise to scale your data using zscore (or mapstd) to get zero mean and unit variance.

Hope this helps.

Thank you for formally accepting my answer

Greg