Simulating neural networks

During the past years many books about "neural networks" have been published, so it is easy for the beginner to get detailed information. A problem arises, however, if the reader wants to understand the function of these networks since they try to emulate the operation of the brain. Althouth it is not difficult to understand how a single nerve cell works, and how the nerve cells interact, the resulting behaviour of large groups of nerve cells cannot be predicted by simple rules. Therefore, it would be very useful if one could make experiments with neural networks to get an insight into their behaviour. Fortunately, it is possible to simulate neural networks on a computer. Simple networks which may be treated with a personal computer are sufficient to demonstrate the important features. This demonstration is the subject of the book. The simulation needs an IBM-PC or a compatible computer. The program is written in Borland Pascal 7.0. With this book, a floppy disk is included which contains the executable program, the source code and the data files for the discussed examples. To use the simulation program, knowledge about Borland Pascal is not necessary. The book may be read without the use of the program.