Introduction

Abstract

We report, in this book, on the last decade’s research into a single architecture of artificial neural networks. The research first identified the fact that a negative feedback artificial neural network using simple Hebbian learning had important statistical properties in that it could self-organise in order to identify the principal component filters of a data set. For readers unfamiliar with these terms, we will discuss them in more detail in Chapter 2. By adding bells and whistles to the basic architecture, we have discovered quite a bit about this very powerful architecture and have brought much of our research together in this single book. Since the book brings together the work of various PhD theses, it cannot give all the details which appear in these theses but seeks to emphasise the common theme underlying the research - the negative feedback network.