                  ================================
                  Description of the tool NETLEARN
                  ================================

This program is a SNNS kernel backpropagation test program.
It's a demo for using the SNNS kernel interface to train networks.


synopsis: netlearn

example:

unix >netlearn

produces


SNNS 3D-Kernel V3.1000
----  Network Learning -----

Filename of the network file: ../../../examples/letters_untrained.net
Loading the network ...

Network name: letters
No. of units       : 71
No. of input units : 35
No. of output units: 26
No. of sites: 0
No. of links: 610

Learning function: Std_Backpropagation
Update function  : Topological_Order

Filename of the pattern file: ../../../examples/letters.pat
Loading the patterns ...

No. of patterns: 26

The learning function 'Std_Backpropagation' needs 2 input parameters:
Parameter [1]: 0.6
Parameter [2]: 0.2

Choose no. of cycles: 250


Shuffle patterns ? (Y/N): n

Shuffleing of patterns disabled

Begin learning ...

Cycle: 1
Learning function value(s): [1]: 54.971806  

Cycle: 14
Learning function value(s): [1]: 20.768957  

Cycle: 27
Learning function value(s): [1]: 18.673342  

Cycle: 40
Learning function value(s): [1]: 16.126991  

Cycle: 53
Learning function value(s): [1]: 13.680001  

Cycle: 66
Learning function value(s): [1]: 11.072329  

Cycle: 79
Learning function value(s): [1]: 8.855511  

Cycle: 92
Learning function value(s): [1]: 6.888253  

Cycle: 105
Learning function value(s): [1]: 5.557242  

Cycle: 118
Learning function value(s): [1]: 4.468373  

Cycle: 131
Learning function value(s): [1]: 3.632325  

Cycle: 144
Learning function value(s): [1]: 2.725383  

Cycle: 157
Learning function value(s): [1]: 1.964378  

Cycle: 170
Learning function value(s): [1]: 1.818217  

Cycle: 183
Learning function value(s): [1]: 1.142782  

Cycle: 196
Learning function value(s): [1]: 1.099230  

Cycle: 209
Learning function value(s): [1]: 0.756542  

Cycle: 222
Learning function value(s): [1]: 0.631204  

Cycle: 235
Learning function value(s): [1]: 0.331756  

Cycle: 248
Learning function value(s): [1]: 0.000000  

Cycle: 250
Learning function value(s): [1]: 0.000000  

End learning ...

Calculating the last error of the network ...
The error is: 22.0737


Filename of the network to save: letters_trained.net
Saving the network ...

