This work examines the impact that genes with activation relationships have on Genetic Algorithms (GAs). These activation relationships allow genes to control whether other associated genes get expressed in the phenotype. More specifically, this thesis investigates the effect that the incorporation of control levels (tiers of genes which determine the activation of lower genes) have on GAs which are used to generate feed forward Artificial Neural Networks (ANNs). In order to evaluate the performance of different levelled Structured Genetic Algorithms (SGAs; they are GAs which possess control levels), numerous experiments were conducted, utilising the four input XOR, Mackey-Glass and Breast Cancer data sets. In addition, the thesis derives four mathematical models which describe how SGA redundancy changes as more control levels are incorporated. This thesis also presented and implemented a novel four level Structurally Evolved Neural Network Algorithm (SENNGA). Furthermore, it demonstrated the novel implementation of a three level SENNGA. Empirical results show that the incorporation of control levels can improve convergence speed and accuracy, up until the optimum number of control levels is reached. In all experiments increasing the number of control levels had the effect of encouraging the production of leaner and more efficient ANNs. Three and four level SENNGAs can demonstrate superior performance in terms of reducing the training error and generalisation. However this does not apply in all cases. Of particular note, three and four level SENNGAs have exhibited a reduced level of generalisation in the breast cancer experiments, compared two level ones. / Doctor of Philosophy (PhD)

Thesis (PhD)--University of South Australia, 1995

Thesis (Ph.D.)--Montana State University--Bozeman, 2006. / Typescript. Chairperson, Graduate Committee: Tomás̆ Gedeon. Includes bibliographical references (leaves 84-86).

Thesis (M.Phil.)--Hong Kong University of Science and Technology, 2007. / Includes bibliographical references (leaves 93-94). Also available in electronic version.

Published Article / People have tried different ways to make machines intelligent. One option is to use a simulated neural net as a platform for Genetic Algorithms. Neural nets are a combination of neurons in a certain pattern. Neurons in a neural net system are a simulation of neurons in an organism's brain. Genetic Algorithms represent an emulation of evolution in nature. The question arose as to why write a program to simulate neurons if a program can execute the functions a combination of neurons would generate. For this reason a virtual robot indicated in Figure 1 was made "intelligent" by developing a process where the robot creates a program for itself. Although Genetic Algorithms might have been used in the past to generate a program, a new method called Single-Chromosome-Evolution-Algorithms (SCEA) was introduced and compared to Genetic Algorithms operation. Instructions in the program were changed by using either Genetic Algorithms or alternatively with SCEA where only one simulation was needed per generation to be tested by the fitness of the system.

The available sources have, to some extent, determined the form of this thesis, which was undertaken in the hope that a more detailed study of the relations between London and the Crown during the years 1 1400_1 1150 would place in perspective the crises with which it begins and ends. The most important source of material for this study has been the Journals of the Court of Aldermen and Common Council which survive from 1416 (the years 1429- 1436 are missing). Historians with the help of a nineteenth century index have quarried in these Journals, but they have never been read through systematically. Journals nos. 3 and 6, having been wrongly bound up, could not be used until, their pages bad been sorted into the correct order from the internal evidence of their contents. The scribes who compiled the Journals were both careless and cautious which increases the difficulty in interpreting their crabbed notes. From studying the Journals dominant themes emerged which were then followed up at the Public Record Office and elsewhere. The conclusions from this study fall into three main categories. The Journals provided a great deal of material from which it was possible to draw a much more detailed picture of the machinery and business of the government of medieval London. T1'e Aldermen and civic officials emerge as conservative, but conscientious, men who might press hardly upon minority interests, but had constantly before their eyes the needs of the City as a whole. Secondly it has been possible to tidy up the chronology of the crises themselves. At such times as Bolingbroke' s usurpation and Cade' a revolt the civic scribes were least active and most cautious. But it seems clear that the London support for both these men has been exaggerated and that the fundamental conservatism of the City governors was not easily rocked, whether by royal scions or Kentish peasants. But this study has proved most useful where the more mundane contact between the Crown and the citizens could be examined, In this way it has been possible to place the financial relations between the King and the City in perspective, and to realize that the King did not come as a beggar to the Londoners, since he had at his disposal all the chartered freedoms and privileges which were essential to the communal and economic life of the City. London, in spite of its great prestige and financial importance, still operated in the fifteenth century within a framework of royal privilege. While the memory of Richard II's action in 1392 was still green the Londoners were in no position to demand redress of grievances before supply. In understanding the delicate balance of the relationship between the Crown and the Londoners it is easier to understand the survival of the Lancastrian dynasty.

No description available.

No description available.

No description available.

No description available.