patmat.readme

著名的神经网络工具箱

=============================================================================
	README file for the example files patmat.xxx
=============================================================================


Description:	This network is a toy pattern matching network,
============	a feedforward auto-associator network.

This network is derived from the letters.net network. Instead of a
classification output layer it has an output layer which is identical
to the input layer. Both are 5x7 matrices of binary values.


Pattern-Files:	patmat.pat
==============	

The pattern-file patmat.pat contains 6 training patterns only. The
patterns here have binary values of 0 and 1 but SNNS treats all inputs
and outputs as real valued. You may best use this network and pattern
file for your own experiments to generate or modify patterns from the
user interface.


Network-Files:	patmat.net
==============	

The network contains a trained network file with the following topology:
	35 input neurons
	10 hidden neurons
	35 output neurons


Config-Files:	patmat.cfg
=============

This network uses one 2D display and a 3D display in its standard
configuration.


Hints:
======

Use this network and pattern file for your own early experiments with
the 2D display network editor. Try all the commands to
insert/delete/modify units and links.

Try also to generate this network from scratch with the BIGNET tool
and perform your initial training experiments with it. You don't loose
anything of you accidentally overwrite it.


=============================================================================
	End of README file
=============================================================================