I. Introduction
At the beginning of the computer age, there were problems that men
couldn't solve, or more precisely (and this distinction is extrememly
important) men were too slow in solving those problems.
So, we made a machine we called computer which would be able to automate some solutions of
problems that were causing trouble to humans. Those problems were
calculating equations to resolve important physical problems,
and later displaying a nice GUI (heum that's not exactly the fact with
win9x :-) ), making word processing and so on ...
But now we come to a limit with these computers and standard programming, we can't easily,
and quickly perform for example :
- Signal processing such as (pattern recognition, voice recognition, image processing...)
- Compression
- Data mining
- Data simplification
That's why, man tried to make an intellectual abstraction wich would work
mostly like human brain works, that's what we called a neuronal network.
So, the present tutorial will try to explain the principal of the neuronal
network, and will show the relationship with a biological neuronal
network.
After that, when the bases will be well understood, you will be shown a
famous NN (neuronal network) topology called BPN (Backward propagation network).
And after all, we will explain you how to use a very well implemented and
simple library (as easy as possible) written in C++ that simulate The BPN
and so succeded in solving very compicated problems such as pattern
recognition in a very noisy environment.
Now, lets stop talking and beginning with the real tutorial :-)