WINNEU 1.1 (c) 1996 Henrik Kjaer Jorgensen. 
Update from version 1.0.

WINNEU is a program for building, training and testing artificial neural 
networks. The following net types are available in version 1.1: 
  Hopfield net, 
  BAM net, 
  Single layer perceptron and 
  Multi layer perceptron. 

To install WINNEU do:

  1. Create a working directory for WINNEU (e.g. c:\winneu)
  2. Copy the file wneu11.zip to this directory
  3. Unpack the program with 'pkunzip -d wneu11'

Make sure that the following subdirectories exist in the working directory 
of Winneu: NET, PATTERN, INPUT and OUTPUT. 

After unpacking, WINNEU can be started from Windows by entering e.g. 
'c:\winneu\winneu' in the file/run menu in the program manager. A program 
group or item can also be created from file/new menu using the WINNEU Icon. 
Remember setting the right working directory (e.g. c:\winneu). 

Some netfiles and pattern files are provided as examples:
  h4.net:    Hopfield net with 4 neurons
  b42.net:   BAM net with 4 input neurons and 2 output neurons
  m21un.net: Multi layer perceptron with 2 inputs and 1 output (EXOR)
  m25bi.net: Multi layer perceptron with 2 input and 5 outputs
The nets are untrained. For each net train and test pattern files are 
provided.

WINNEU is a shareware program and may be freely distributed.

Comments are welcome at the following E-Mail address: 
  henkj@post3.tele.dk

Eventually look at the following web-page for more information:
  http://home3.inet.tele.dk/henkj

Version 1.1:
  This new version of WINNEU is equal to version 1.0 in functionality. The
difference between the two versions is that the DLL file BWCC.DLL is included.
The previous version (1.0) did not function for users not already using
Borland products with BWCC.DLL installed.
  Also I have changed Internet location, so comments should be sent to the
E-Mail address above.
  In the near future a new version of Winneu (2.0) will be released 
implementing custom networks and Kohonen networks.
