Not available / text

Traditionally, an evolutionary algorithm (EA) operates on a single population with no restrictions on possible mating pairs. Interesting changes to the behaviour of EAs emerge when the structure of the population is altered so that mating between individuals is restricted. Variants of EAs that use such populations are grouped into the field of spatially-structured EAs (SSEAs). Previous research into the behaviour of SSEAs has primarily focused on the impact space has on the selection pressure in the system. Selection pressure is usually characterised by takeover times and the ratio between the neighbourhood size and the overall dimension of space. While this research has given indications into where and when the use of an SSEA might be suitable, it does not provide a complete coverage of system behaviour in SSEAs. This thesis presents new research into areas of SSEA behaviour that have been left either unexplored or briefly touched upon in current EA literature. The behaviour of genetic drift in finite panmictic populations is well understood. This thesis attempts to characterise the behaviour of genetic drift in spatially-structured populations. First, an empirical investigation into genetic drift in two commonly encountered topologies, rings and torii, is performed. An observation is made that genetic drift in these two configurations of space is independent of the genetic structure of individuals and additive of the equivalent-sized panmictic population. In addition, localised areas of homogeneity present themselves within the structure purely as a result of drifting. A model based on the theory of random walks to absorbing boundaries is presented which accurately characterises the time to fixation through random genetic drift in ring topologies. A large volume of research has gone into developing niching methods for solving multimodal problems. Previously, these techniques have used panmictic populations. This thesis introduces the concept of localised niching, where the typically global niching methods are applied to the overlapping demes of a spatially structured population. Two implementations, local sharing and local clearing are presented and are shown to be frequently faster and more robust to parameter settings, and applicable to more problems than their panmictic counterparts. Current SSEAs typically use a single fitness function across the entire population. In the context of multimodal problems, this means each location in space attempts to discover all the optima. A preferable situation would be to use the inherent spatial properties of an SSEA to localise optimisation of peaks. This thesis adapts concepts from multiobjective optimisation with environmental gradients and applies them to multimodal problems. In addition to adapting to the fitness landscape, individuals evolve towards their preferred environmental conditions. This has the effect of separating individuals into regions that concentrate on different optima with the global fitness function. The thesis also gives insights into the expected number of individuals occupying each optima in the problem. The SSEAs and related models developed in this thesis are of interest to both researchers and end-users of evolutionary computation. From the end-user�s perspective, the developed SSEAs require less a priori knowledge of a given problem domain in order to operate effectively, so they can be more readily applied to difficult, poorly-defined problems. Also, the theoretical findings of this thesis provides a more complete understanding of evolution within spatially-structured populations, which is of interest not only to evolutionary computation practitioners, but also to researchers in the fields of population genetics and ecology.

Thesis (Ph. D.)--University of Texas at Austin, 2002. / Vita. Includes bibliographical references. Available also from UMI Company.

Thesis (Ph. D.)-- State University of New York at Binghamton, Thomas J. Watson School of Engineering and Applied Science, Department of Systems Science and Industrial Engineering, 2009.

Thesis (Ph.D.)--University of North Carolina at Charlotte, 2005. / Includes bibliographical references (leaves 71-79).

Thesis (M.S.)--Missouri University of Science and Technology, 2010. / Vita. The entire thesis text is included in file. Title from title screen of thesis/dissertation PDF file (viewed July 19, 2010) Includes bibliographical references (p. 57-60).

Made available in DSpace on 2016-03-15T19:38:05Z (GMT). No. of bitstreams: 1 Leopoldo Bulgarelli de Carvalho.pdf: 696477 bytes, checksum: f90764d3c257bf63305bda69583c731e (MD5) Previous issue date: 2007-09-12 / Fundo Mackenzie de Pesquisa / Recent applications in evolutionary programming have suggested the use of different stable probability distributions, such as Cauchy and Lévy, in the random process associated with the mutations, as an alternative to the traditional (and also stable) Normal distribution. The motivation for this is the attempt to improve the results in some classes of optimisation problems, over those obtained with Normal distribution. Based upon an algorithm proposed in the literature, mostly its version in [Lee and Yao, 2004], that use non Normal stable distributions, we study herein the effect of turning it adaptive in respect to the determination of the more adequate stable distribution parameters for each problem. The evaluations relied upon standard benchmarking functions of the literature, and the comparative performance tests were carried out in respect to the baseline defined by a standard algorithm using Normal distribution. The results suggest numerical and statistical superiority of the stable distribution based approach, when compared with the baseline. However, they showed no improvement over the adaptive method of [Lee and Yao, 2004], possibly due to a consequence of implementation decisions that had to be made in the present implementation, that were not made explicit therein. / Aplicações recentes em programação evolutiva tem sugerido a utilização de diferentes distribuições estáveis de probabilidade, tais como de Cauchy e de Lévy, no processo aleatório associado às mutações, como alternativa à tradicional (e também estável) distribuição Normal. A motivação para tanto é melhorar os resultados em algumas classes de problemas de otimização, com relação aos obtidos através da distribuição Normal. Esse trabalho propõe uma nova classe de algoritmos auto-adaptativos com respeito à determinação dos parâmetros da distribuição estável mais adequada para cada problema de otimização. Tais algoritmos foram derivados de um existente na literatura, especialmente sua versão apresentada em [Lee e Yao, 2004]. Em um primeiro momento foram estudadas as principais características das distribuições estáveis que são, nesse trabalho, o foco dos processos aleatórios associados às mutações. Posteriormente, foram apresentadas as diferentes abordagens descritas pela literatura e as sugestões de algoritmos com características auto-adaptativas. As avaliações dos algoritmos propostos utilizaram funções de teste padrão da literatura, e os resultados comparativos de desempenho foram realizados com relação a um algoritmo tradicional baseado na distribuição Normal. Posteriormente, foram aplicados novos comparativos entre as diversas abordagens auto-adaptativas definidas no presente estudo, e feito um comparativo do melhor algoritmo auto-adaptativo aqui proposto com o melhor algoritmo adaptativo obtido de [Lee e Yao, 2004]. Os resultados evidenciaram superioridade numérica e estatística da abordagem baseada em distribuições estáveis, sobre o método tradicional baseado na distribuição Normal. No entanto, o método proposto não se mostrou mais eficaz que o método adaptativo sugerido em [Lee e Yao, 2004], o que pode ter sido decorrente de decisões de implementação não explícitas naquele trabalho, que tiveram de ser tomadas no presente contexto.

The evolutionary nature of humans requires agent systems to be continuously replaced due to their inability to meet or adapt to our changing needs. Therefore, to eliminate the need for a human to continuously adapt an agent, evolutionary agents are required [Chu04, Ore99, Rak02, Syc96]. This dissertation develops a feasible option to ensuring that agents continuously develop desirable behaviour. The solution is a specialized architecture that embeds self-evolvement into a target agent. The specialized architecture ensures that desirable behaviour emerges from any agent, as it is embedded between the target agent and the target agent’s environment and therefore is able to obtain domain- and hardwarespecific information from the target agent. The specialized architecture is a comprehensive methodology that incorporates all agents with the ability to embed the required self-evolvement enhancements as domain- and hardwarespecific information is obtained from the target agent. The specialized architecture responsible for embedding self-evolvement into an agent is the generic self-evolvement effecting evolutionary agent (GSEEA). The GSEEA is developed with a single goal, which is to ensure that the target agent meets the requirements of a changing environment. Changing environmental conditions can include different network conditions and different platforms. The GSEEA’s goal is accomplished by embedding the required self-evolvement enhancements into the target agent to produce a self-evolvement enhanced agent. In this dissertation the GSEEA is implemented to demonstrate its feasibility and problem-solving accuracy. In the GSEEA implementation the target agent is a puzzle-solving agent and the self-evolvement enhanced agent is the selfevolvement enhanced puzzle-solving agent. The GSEEA’s deliberative component consists of two algorithms, namely a genetic algorithm and a learning algorithm. The GSEEA’s genetic algorithm develops knowledge base rules (selfContents III evolvement enhancements) that modify actuator information. The GSEEA’s learning algorithm updates developed knowledge base rules by modifying sensor information. The GSEEA tests the developed self-evolvement enhancements by embedding them into the target agent through the target agent’s knowledge base manager, evaluating the developed self-evolvement enhancements and deleting those which do not enhance the target agent. The target agent achieves selfevolvement as additional enhancements required by the self-evolvement enhanced agent can be achieved by applying the same process followed to enhance the target agent which was discussed previously. The evaluation of the GSEEA implementation demonstrated that the GSEEA was implemented successfully based on feasibility and problem-solving accuracy as the self-evolvement enhanced puzzle-solver agent outperformed the puzzlesolver agent. / Prof. E.M. Ehlers

Objective of this master's thesis is optimizing of neral network topology using some of evolutionary algorithms. The backpropagation neural network was optimized using genetic algorithms, evolutionary programming and evolutionary strategies. The text contains an application in the Matlab environment which applies these methods to simple tasks as pattern recognition and function prediction. Created graphs of fitness and error functions are included as a result of this thesis.

Image filters is a subset of signal processing. Image filtering is mainly used for highlighting an information. It can be useful for reduce noise, smooth pictures, enhance contrast or for edge detection. Image filter design itself is a time-consuming process. It is suitable to automate the process and give up the function of filter designing to preprogrammed system. Designing komponents for that system is aim of this work. It is part of an whole expert system. A set of information is given on input which are used for generating new image filters. Subsequently it will evaluate the relevancy of concrete image filter for subsequent use.