About the HTML version

Apart from inevitable formatting changes, and perhaps some minor corrections and humorous anecdotes, this is a faithful copy of the original hardbound thesis (which is available from the University of York). It was created from the latex source with the help of  T_TH and much editing in Microsoft Frontpage. I had no help from latex2html, which refused to install under Windows even after much prodding and cursing. No doubt some of you think this is the sort of punishment that Windows users deserve. If anyone would like to 'borrow' any of the pictures found herewithin, you are quite welcome, so long as you ask me first. I can even provide the original vector files in encapsulated postscript or Windows metafile formats. If you notice any errors, particularly with the internal cross-references, please let me know - and if it is not too difficult or time-consuming, I might even correct them ;) If you have any constructive (i.e. nice) comments or suggestions, please contact me. My psychoanalyst says that it is good for me to communicate with new people.

Acknowledgements

I would particularly like to thank my parents for all their love and support; Andy Tyrrell, for all the advice, pastoral care and beer he has provided me with during the long years of my doctorate; Paul, for all the support, for being a great friend, and for being my drinking partner; Austin, for keeping me down to earth, keeping me laughing, and plying me with beer¹; Alex, for his comradeship, encyclopæ dic brain, and strong coffee; Andy Greenboy, fellow research gimp², provider of tea, and mainstay of the Buzzard massive; Becky and Rosie, for sleeping on me, for letting me stroke them, and for all the dead shrews they faithfully bring to me; and last, but never least, Elise; for her friendship, for keeping me entertained, and for teaching me to find solace in my work.

I would also like to offer special thanks to all my colleagues at York, past and present, for their help and friendship; to Steve Smith and Keith Downing, my examiners, for their constructive advice; and to Spike the goose and his feathered friends, for being there when I needed someone to talk to.

Declaration

Some of the research presented in this thesis has previously been published by the author [Lones and Tyrrell, 2001c,Lones and Tyrrell, 2001b,Lones and Tyrrell, 2001a,Lones and Tyrrell, 2002b,Lones and Tyrrell, 2002a,Lones and Tyrrell, 2003b,Lones and Tyrrell, 2003a]. All work presented in this thesis as original is so to the best knowledge of the author. References and acknowledgements to other researchers have been given as appropriate.

Hypothesis

This research follows from the notion that models of biological representations can be used within genetic programming to represent executable structures; motivated by the expectation that these models will capture useful biological properties that will improve the evolution of executable structures. More specifically, it is asserted that:

• Representations from engineering domains are not designed to respond to random change in a meaningful way. In genetic programming, this is demonstrated by the poor response of conventional representations towards recombination operators.
• Biological representations are a product of evolution and are therefore well adapted for representing evolving artefacts. In particular, they are believed to confer evolvability - the capacity to exhibit change in an appropriate direction - to the artefacts that they represent.
• Many biological structures are known to carry out activities of a computational nature. This includes metabolic, signalling, gene expression and neural pathways. It has also been shown that models of these structures can be used to represent non-biological computational artefacts.

Following from these assertions, it is hypothesised that models of biological representations can be used to represent computer programs and other artificial executable structures within genetic programming, thereby improving the evolvability of these structures.

[Contents] [Chapter 1]

Footnotes:

¹Despite the above, I follow moderate alcohol and caffeine drinking habits and advise you to do the same.

²Not that sort.