What is a Neural Network?
                         =========================


      Artificial Neural Systems - a new information processing technology.
      ---------------------------------------------------------------------



It is named Neural Network because the design of the networks is based on the
neural structure of the brain.It is inspired by knowledge from neuroscience
although it does not try to be biologically realistic in detail!The systems
try to mimic some intelligent functions of the brain such as learning,
generalization, associative recall,...
A more formal definition:
Neural Network - a highly parallel dynamical system with the topology of a
directed graph. The nodes are called processing elements or neurons and the
directed lines are called interconnections.
How do Neural Networks work?
They receive inputs from the environment through the input neurons.
All of the interconnections can carry signals simultaneously and neurons can
act in parallel to compute a result.
The knowledge is stored in the inner structure of the networks.
Neural Networks are good at problems such as  pattern recognition,
decision making, optimizationdealing with even incomplete or fuzzy data.

NEURONET 1.0
Our Neural Network simulator implements the most well-known paradigms which
have some historical significance as well. The five models are the following:
     ù Perceptron model
     ù Hamming model
     ù Hopfield model
     ù Kohonen model
     ù Back-propagation algorithm
Neural Networks are good at problems such as pattern recognition, decision
making, optimization dealing with even incomplete or fuzzy data.
The current version - NEURONET 1.0 - is written in C language and runs on
IBM PC XT,AT/286/386. We would like to give you a useful tool for Neural
Network development with interactive, display-oriented facilities for
observation and testing purposes. The topology of the networks (number of
neurons, interconnections) the initial states, the activation functions
and many other parameters are all user selectable.
In the manual we give an overview of the main characteristics of the
networks and introduce the five models ( Section II is a part  of the
documentation : " Introduction to Neural Networks "). Then we give a
detailed description of the implementation, example networks are shown for
every model with some useful advice.

The NEURONET 1.0 is a shareware product. It can be duplicated and distri-
buted without profit. This does not mean that the NEURONET is freeware!
The Author reserve all rigths on all features of program, dokumentation
and the enclosed study. Every registrated user will be mailed users guide
and a study about neuron-networks. The registration fee is only 30 USD
or 45 DEM (in Hungary 4000 HUF).

                                     L szl¢ P¡sz r
                                     P.O.B. 1078
                                     6701 Szeged
                                     Hungary