
 Functional Link Training,Testing and Processing Program

    Outline

 A. Program Purpose 
 B. Network Characteristics
 C. Files Needed or Produced
 D. Example Run of Functional Link Training
 E. Example Run of Functional Link Testing
 F. Example Run of Functional Link Processing
 G. Error Functions for Functional Link Net



 A. Program Purpose 
  1. Initialize and train a functional link mapping network using 
      a fast training method.
  2. Test a data file using the weights saved in training.     
  3. Process a data file having no desired outputs.

 B. Network Characteristics
  1. Activation Functions; Linear output units 
  2. Net Functions; polynomial functions of the inputs, with 
        user-chosen degrees of 1 to 3.

 C. Files Needed or Produced
  
 1. The training or testing data file, which gives example inputs 
    and outputs for network learning, or for testing after learning.
    All data files are in formatted, standard form, which means that 
    each pattern or feature vector is followed by the desired outputs.
    After training the program will generate a weights file and a training
    result file which locate in the folder the user specified before.
    After testing the program will generate a result file and store in the
    same way as training.

 D. Example Run of Functional Link Training
    Input the following into the corresponding text box or combolist box on the input form

    twod.tra            ! data filename
    8                   ! Number of inputs
    7                   ! Number of outputs
    2                   ! desired degree of the network


     The program will read all patterns from the file twod.tra, and train a
     functional link net and save the trainined weights to flnt2wod.wts file in the following format,                        which is shown below.

           twod.tra
           8
           7
           2
           weights
     The network will be 2nd degree with 8 inputs and 7 outputs.

     You can run this program on your own data.
  E. Example Run of Functional Link Testing

    Input the following into the corresponding text box or combolist box on the input form

    twod.tra             ! data filename
    fln2twod.wts         ! trainined weights filename


     The program will read the trained weights from file fln2twod.wts, and test a
     functional link net on the training file twod.tra
     The network will be 2nd degree with 8 inputs and 7 outputs. 
     You can run this program on your own data.

  F. Example Run of Functional Link Processing
 
     Input the following into the corresponding text box or combolist box on the input form
    
     twod.tra            ! data filename
     fln2twod.wts        ! trainined weights filename
     Yes/No              ! 'Yes' for with desired output and 'No' for without desired output


     The program will read the trained weights from file fln2twod.wts, and process a
     functional link net on the training file twod.tra
     The network will be 2nd degree with 8 inputs and 7 outputs. 
     You can run this program on your own data.

 G. Error Functions for Functional Link Net
    The error function that is being minimized during functional link
     training is

                   Nout      
    MSE = (1/Npat) SUM MSE(k)     where
                   k=1  

              Npat              2
    MSE(k) =  SUM [ Tpk - Opk ]
              p=1  

     where Npat is the number of training patterns, Nout is the number 
     of network output nodes, Tpk is the desired output for the pth
     training pattern and the kth output, and Opk is the actual output
     for the pth training pattern and the kth output. MSE is printed
     for each outputs.
