Not available / text

Woven string kernels are a form of evolvable, directed, acyclic graphs specialized to perform DNA classification. They are introduced in this thesis, given a rigorous theoretical treatment as a mathematical object, and shown to have a number of interesting properties. Two forms of woven string kernels, uniform and non-uniform, are discussed. The non-uniform woven string kernels are repurposed for use as updating rules for cellular automata. The details of their representation and implementation are presented. A chapter of this thesis is devoted to a visualization technique called non-linear projection, an evolvable form of multidimensional scaling that is used in the analysis of experimental results. The woven string kernels are tested on simple and complex synthetic data as well as biological data, using an evolutionary algorithm to find woven string kernels that are acceptable solutions for classification. They perform marginally on the simplest synthetic data - based on GC content - for which they are not entirely appropriate. They exhibit perfect classification on the more complex synthetic data and on the biological data. Woven string kernels have a number of parameters including their height, the number of initial strings from which they are built, and the amount of weaving used to generate the final structure. A parameter study shows that these parameters must be set based on the type of data under analysis. Experimentation with woven string kernels as rules for updating cellular automata show that having a larger population and more available colour states are correlated with an increase in performance as apoptotic one dimensional cellular automata. This thesis concludes with directions for future work related to theory and experimentation, for both uniform and non-uniform woven string kernels.

No description available.

The objective of Evolutionary Computation is to solve practical problems (e.g.optimization, data mining) by simulating the mechanisms of natural evolution. This thesis addresses several topics related to adaptation and self-organization in evolving systems with the overall aims of improving the performance of Evolutionary Algorithms (EA), understanding its relation to natural evolution, and incorporating new mechanisms for mimicking complex biological systems. Part I of this thesis presents a new mechanism for allowing an EA to adapt its behavior in response to changes in the environment. Using the new approach, adaptation of EA behavior (i.e. control of EA design parameters) is driven by an analysis of population dynamics, as opposed to the more traditional use of fitness measurements. Comparisons with a number of adaptive control methods from the literature indicate substantial improvements in algorithm performance for a range of artificial and engineering design problems. Part II of this thesis involves a more thorough analysis of EA behavior based on the methods derived in Part 1. In particular, several properties of EA population dynamics are measured and compared with observations of evolutionary dynamics in nature. The results demonstrate that some large scale spatial and temporal features of EA dynamics are remarkably similar to their natural counterpart. Compatibility of EA with the Theory of Self-Organized Criticality is also discussed. Part III proposes fundamentally new directions in EA research which are inspired by the conclusions drawn in Part II. These changes involve new mechanisms which allow selforganization of the EA to occur in ways which extend beyond its common convergence in parameter space. In particular, network models for EA populations are developed where the network structure is dynamically coupled to EA population dynamics. Results indicate strong improvements in algorithm performance compared to cellular Genetic Algorithms and non-distributed EA designs. Furthermore, topological analysis indicates that the population network can spontaneously evolve to display similar characteristics to the interaction networks of complex biological systems.

Thesis (Ph. D.)--University of Texas at Austin, 2004. / Supervisor: Risto Miikkulainen. Vita. Includes bibliographical references. Also available from UMI.

Thesis (Ph. D.)--University of Texas at Austin, 2007. / Vita. Includes bibliographical references.

No description available.

No description available.

This thesis provides a framework for describing a canonical evolutionary system. Populations of individuals are envisaged as traversing a search space structured by genetic and developmental operators under the influence of selection. Selection acts on individuals' phenotypic expressions, guiding the population over an evaluation landscape, which describes an idealised evaluation surface over the phenotypic space. The corresponding valuation landscape describes evaluations over the genotypic space and may be transformed by within generation adaptive (learning) or maladaptive (fault induction) local search. Populations subjected to particular genetic and selection operators are claimed to evolve towards a region of the valuation landscape with a characteristic local ruggedness, as given by the runtime operator correlation coefficient. This corresponds to the view of evolution discovering an evolutionarily stable population, or quasi-species, held in a state of dynamic equilibrium by the operator set and evaluation function. This is demonstrated by genetic algorithm experiments using the NK landscapes and a novel, evolvable evaluation function, The Tower of Babel. In fluctuating environments of varying temporal ruggedness, different operator sets are correspondingly more or less adapted. Quantitative genetics analyses of populations in sinusoidally fluctuating conditions are shown to describe certain well known electronic filters. This observation suggests the notion of Evolutionary Signal Processing. Genetic algorithm experiments in which a population tracks a sinusoidally fluctuating optimum support this view. Using a self-adaptive mutation rate, it is possible to tune the evolutionary filter to the environmental frequency. For a time varying frequency, the mutation rate reacts accordingly. With local search, the valuation landscape is transformed through temporal smoothing. By coevolving modifier genes for individual learning and the rate at which the benefits may be directly transmitted to the next generation, the relative adaptedness of individual learning and cultural inheritance according to the rate of environmental change is demonstrated.

Thesis (Ph.D.)--University of Missouri-Columbia, 2007. / The entire dissertation/thesis text is included in the research.pdf file; the official abstract appears in the short.pdf file (which also appears in the research.pdf); a non-technical general description, or public abstract, appears in the public.pdf file. Title from title screen of research.pdf file (viewed on March 23, 2009) Vita. Includes bibliographical references.