Feedforward neurale network including minimum and maximum values??

Hi, I am making a feed forward neurale network. The training set consist of approximately 4300 observations, with one input parameter (and one target). When I apply the neural network for simulations the network does not give any minimum or maximum values. How do I make the neural network include these values? The training set contains several observations of minimum and maximum values. I have tried different number of hidden neurons and different number of iterations, the problem is there for all the simulations.

All help are appreciated.

Best regards, Anne

Answer 1

[Greg Heath]

A typical objective of training is, given data ( DESIGN + NONDESIGN ) that can be considered to come from the same probability distribution, use

    1. ONLY design (training + validation) data 
    2. As few hidden neurons as possible

to reduce the MSE of NONDESIGN (test) data below a specified threshold. For example

     3. mse(ttst-ytst) < 0.01*mean(var(ttst',1))

This is usually attempted by trying to achieve a specified goal of MSEtrn proportional to mean(var(ttrn',1)) before MSEval reaches a minimum.

If you desire better fitting at extrema, use the errorWeight option of mse

   help mse

   doc mse

Hope this helps.

Thank you for formally accepting my answer

Greg