MATLAB: Probabilistic Neural Network all outputs the same

Deep Learning Toolboxpnn output array all ones probabilistic neural network sim newpnn

I have to create a PNN in matlab. To start, I am using a small dataset with only two classes. I am using the following code:

net = newpnn(train_data',train_class'); out = sim(net,test_data');

My ouput is array of all 1's.

Anyone have any suggestions?

Best Answer

1.Normalize inputs to have zero mean and unit standard deviation via zscore or mapstd.

2. Train multiple designs in a loop over a spread value interval that includes the default value of 1. For example, spread = 0.1 : 0.1 : 2.

Related Solutions

MATLAB: How to fix minmum mse requirement for neural network training mannually rather using default requirement

 [ I N ] = size(x)
 [ O N ] = size(t)

% Reference MSEs (From Naïve Constant Output Model y00 = mean(ttrn,2))

 Ntrn      = N-2*(0.15*N)       % Default no. of training examples
 MSEtrn00  = mean(var(ttrn',1)) % Biased training target variance (divide by Ntrn)
 MSEtrn00a = mean(var(ttrn',0)  % Degree of freedom "A"djusted (divide by Ntrn-1)

% Number of estimation degrees of freedom (See Wikipedia)

 Ntrneq    = Ntrn*O             % prod(size(ttrn)) is number of training equations
 Nw        = (I+1)*H+(H+1)*O    % Number of unknown weights for an I-H-O node topology
 Ndof      = Ntrneq-Nw          % Not useful when Nw >= Ntrneq
 etrn    = ttrn-ytrn;           % Training error
 MSEtrn  = sse(etrn)/Ntrneq     % Biased MSE
         = mse(etrn)
 MSEtrna = sse(etrn)/Ndof       % DOFA  MSE
         = Ntrneq*MSEtrn/Ndof

% Normalized MSEs

 NMSEtrn  = MSEtrn/MSEtrn00
 NMSEtrna = MSEtrna/MSEtrn00a

% Practical training goal

 NMSEtrna <= 0.01  % Independent of target scaling
 net.trainParam.goal = 0.01*Ndof*MSEtrn00a/Ntrneq

Hope this helps.

Thank you for formally accepting my answer

Greg

MATLAB: Probabilistic neural network coding problem

1. Class 1 has 2 duplicate points and class 2 has 1 duplicate point

2. There is an 8 orders of magnitude difference between variables (1,3) and variables (2,4). Therefore, I standardized the data to have zero mean and unit variance.

3. Using newpnn with 0 < spread < 2 I was not able to separate the classes.

I recommend you use newrb and loop over spread values to obtain the best result.