[Artificial Neural Networks]

by

Acknowledgement

Iwould take this opening to express gratitude my study supervisor, family and associates for their support and guidance without which this study would not have been possible.

DECLARATION

I, [type your full first titles and last name here], declare that the contents of this dissertation/thesis comprise my own unaided work, and that the dissertation/thesis has not before been submitted for learned examination towards any qualification. Furthermore, it comprises my own attitudes and not inevitably those of the University.

Signed __________________ Date _________________

Abstract

Two related but different fields are reviewed. Initially some basic facts about developing real brains are set out and then work on dynamic neural networks is described. A dynamic neural network is defined as any artificial neural network that automatically changes its structure through exposure to input stimuli. Various models are described and evaluated and the functional correlates of both regressive and progressive structural changes are discussed. The paper concludes that, if future modelling work is to be set within a more neurally-plausible framework, then it would be fruitful to examine networks in which the connectivity between extant units is progressively embellished.

Artificial Neural Networks

1. Introduction

The present paper has been written so as to examine two related, but quite different fields: developmental neurobiology and artificial neural network research. Although it would appear that there is much common ground to these two fields, most of the work undertaken by the artificial neural network community has continued with almost no acknowledgement of the basic facts of developmental neurobiology. There are some notable exceptions to this, but it is, generally, true that most of the so-called “neural” network research has little or no neural plausibility. There are a number of reasons as to why such concerns might be dismissed. For instance, the focus of much modelling work is of a computational nature where the in-principle characteristics of some formal system are paramount. Nevertheless, insofar as a central belief in cognitive science is that psychological processes are ultimately predicated upon neurobiological processes (Rumelhart et al., 1986b), then it is justifiable to ask about the degree to which models of psychological processes are neurally plausible.

Although much of how real brains develop remains a mystery, this should not be taken to imply that there is complete ignorance. Indeed what the first half of this paper attempts to do is to set out some of the basic facts that constitute current knowledge about how real brains develop. The hope is that these may then inform, in a useful way, future work on artificial neural networks. There is no intention here to try to claim that there is complete consensus in the field of developmental neurobiology: as in most areas of science, there is debate over how best to interpret some of the extant data. As a consequence, the primary aim of the first half of the paper has been to describe what is well known, and, hopefully uncontentious, about how real brains develop. This half of the paper ends with a summary of research that has attempted to integrate ...