UCL Discovery
UCL home » Library Services » Electronic resources » UCL Discovery

The theory, design, unification and implementation of a class of artificial neural network

Houselander, Paul; (1991) The theory, design, unification and implementation of a class of artificial neural network. Doctoral thesis (Ph.D), UCL (University College London). Green open access

[thumbnail of The_theory,_design,_unificatio.pdf] Text

Download (17MB)


Artificial neural networks (ANNs) are highly interconnected systems of simple processing cells the inspiration for which stems from the biological brain. In this thesis we will consider the motivation design and implementation of a particular class of these structures based on the Radial basis function (RBF) and in particular, the Gaussian function. Chapter 1 introduces some basic concepts, the motivation behind ANN research and a summary of the various landmarks which have led to the intense international interest that the subject now enjoys. Chapter 1 also introduces the RBF neuron which will be used in the chapters to follow and considers the related field of conventional pattern recognition. Although the link between the artificial and biological systems is somewhat tenuous, we shall endeavour to form a connection between the different ANN designs covered in this thesis and a simplified subjective model of perception. Hopefully, this will enable the reader to obtain a better perspective on the role that each network would play in an artificial perceptual system. The simplified model of perception consists of four main categories which are: feature extraction or self organisation, mapping and classifying, associative memory and hypothesis evaluation. Chapters 2-5 cover these categories in detail by reviewing the most prominent work to date in each area and the contributions made by the author. Chapter 6 describes the unification in terms of structure and training of the architectures based on the RBF neuron presented in chapters 2-5 and chapter 7 covers the electronic implementation of ANN using both analogue and digital techniques with particular attention to novel circuits based on a hybrid analogue/ digital approach and a digital pseudo-asynchronous re-configurable architecture. Finally, chapter 8 draws together the conclusions from each of the previous chapters and indicates a direction for future work.

Type: Thesis (Doctoral)
Qualification: Ph.D
Title: The theory, design, unification and implementation of a class of artificial neural network
Open access status: An open access version is available from UCL Discovery
Language: English
Additional information: Thesis digitised by ProQuest.
Keywords: Applied sciences; Radial basis function
URI: https://discovery.ucl.ac.uk/id/eprint/10107673
Downloads since deposit
Download activity - last month
Download activity - last 12 months
Downloads by country - last 12 months

Archive Staff Only

View Item View Item