** Privacy and Security Notice **

Dissertations related to Evolutionary Computation

To add your dissertation, please send a note to galist-dissertations@aic.nrl.navy.mil. Please format your information for BibTeX. An example BibTeX entry is shown below. It is probably easiest to simply copy and paste the example, filling in the relevant fields. The keywords, notes, and abstract fields are optional. The download1 field allows you to specify the label string for the URL, as well as the URL itself. You can include as many of such keys as you like; the order is induced by the trailing suffix. The BibTeX file used to create this page can be found here.

@phdthesis{lastname2004phd,
  author = {John Q. Lastname},
  title  = {An Analysis of Interesting Things},
  school = {University of Something},
  address = {Somecity, Somecountry},
  abstract = {Your abstract or description here},
  keywords = {Interesting Things, Very Cool Science, Evolutionary Computation},
  notes   = {This is a fake BibTeX entry meant only as an example.},
  download1 = {Download PDF,http://www.nowhere.edu/lastname2004thesis.pdf}
  year    = {2004}}

R. Paul Wiegand (2004)

An Analysis of Cooperative Coevolutionary Algorithms.
Ph.D. Thesis, George Mason University, Fairfax, VA.
[Download Instructions] 
Abstract:
Coevolutionary algorithms behave in very complicated, often quite counterintuitive ways. Researchers and practitioners have yet to understand why this might be the case, how to change their intuition by understanding the algorithms better, and what to do about the differences. Unfortunately, there is little existing theory available to researchers to help address these issues. Further, little empirical analysis has been done at a component level to help understand intrinsic differences and similarities between coevolutionary algorithms and more traditional evolutionary algorithms. Finally, attempts to categorize coevolution and coevolutionary behaviors remain vague and poorly defined at best. The community needs directed investigations to help practitioners understand what particular coevolutionary algorithms are good at, what they are not, and why.

This dissertation improves our understanding of coevolution by posing and answering the question: "Are cooperative coevolutionary algorithms (CCEAs) appropriate for static optimization tasks?" Two forms of this question are "How long do they take to reach the global optimum" and "How likely are they to get there?" The first form of the question is addressed by analyzing their performance as optimizers, both theoretically and empirically. This analysis includes investigations into the effects of coevolution-specific parameters on optimization performance in the context of particular properties of potential problem domains. The second leg of this dissertation considers the second form of the question by looking at the dynamical properties of these algorithms, analyzing their limiting behaviors again from theoretical and empirical points of view. Two common cooperative coevolutionary pathologies are explored and illustrated, in both formal and practical settings. The result is a better understanding of, and appreciation for, the fact that CCEAs are not generally appropriate for the task of static, single-objective optimization. In the end a new view of the CCEA is offered that includes analysis-guided suggestions for how a traditional CCEA might be modified to be better suited for optimization tasks, or might be applied to more appropriate tasks, given the nature of its dynamics.

Farid Khoshalhan (2003)

A New selection Method in Genetic Algorithms and Its Applications to Production Management Problems.
Ph.D. Thesis, Tarbiat Modares University, Tehran, Iran.
Abstract:
This dissertation deals with genetic Algorithms (GAs) as one of the meta-heuristic search methods and their applications to production management problems. The emphasis of the thesis is selection procedure of genetic algorithms and it aims at considering a framework based on the natural characteristics and behaviours in genetics besides the possible applications in the field of production and operations management problems. A new approach to selection in GAs is introduced. Based on the proposed procedure, an adaptive method for determining the GA parameters is depicted. The proposed GA is called maturing assortative GA. The main property of this GA is the concept of maturity in genetics and its impact on choosing individuals for applying operators and selection. The performance of proposed methods is investigated in two way. First, standard test functions from literature are selected and the proposed methods are compared with the standard genetic algorithms. Then, three problems of the productions and operations management problems, that is loop layout problem, one-dimensional machine location problem, and fuzzy assembly line balancing problems are used as applications of the method to the production management problems. The comparisons, statistical analysis, and corresponding reports are given.

Joshua D. Knowles (2002)

Local-Search and Hybrid Evolutionary Algorithms for Pareto Optimization.
Ph.D. Thesis, University of Reading, Reading, UK.
[Download Information] 
Abstract:
In recent years, a gradual increase in the sophistication of multiobjective evolutionary algorithms (MOEAs) for Pareto optimization has been seen, accompanied by an ever-growing list of applications. Despite this trend, however, the proposition that methods based on local search may be a good alternative approach --- with the advantages of ease-of-use and lower computational overhead --- has not been thoroughly tested. In this thesis we develop a novel, local-search algorithm for Pareto optimization, called PAES, enabling us to test and compare MOEAs against this philosophically different search method.

To help perform this testing, we develop new statistical performance metrics for evaluating collections of approximations to the Pareto set, based on a critical review of currently available methods. Using these metrics, we find that PAES performs well in comparison with popular, modern MOEAs, on a variety of test function and real-world telecommunications problems. These results suggest that local-search-based methods for Pareto optimization do merit further investigation.

Some of the elements of PAES are also more generally applicable for use in the design of other algorithms. In particular, the archiving strategy used by PAES, which incorporates rules to bound the number of solutions stored, and to ensure that they are distributed widely and evenly in objective space, may be used in any multiobjective search algorithm for storing solutions. We demonstrate the generality of the basic PAES procedures by outlining several more sophisticated variants, including multi-start and tabu search algorithms.

We also combine local-search and population-based evolutionary methods together, forming a new memetic algorithm, M-PAES, for Pareto optimization. M-PAES is tested using a number of problems from the literature, and is found to exhibit promising performance compared to other methods. Finally, we use M-PAES to provide a set of benchmark results for some difficult new instances of the multi-criteria minimum spanning tree (mc-MST) problem.

Jano I. van Hemert (2002)

Application of Evolutionary Computation to Constraint Satisfaction and Data Mining.
Ph.D. Thesis, Leiden Institute of Advanced Computer Science, Leiden University, Leiden, The Netherlands.
[Download Instructions] 
Notes: ISBN: 90-6734-057-X
Abstract:
The major contribution of this thesis is to the search for efficient and effective algorithms that solve constraint satisfaction problems and data mining problems. As, besides evaluating new evolutionary algorithms by comparing it with previous evolutionary techniques, it makes a connection between these two problem areas on the level of the problem solver. Furthermore, the part about constraint satisfaction problems provides a connection between two fields, that of evolutionary computation and of constraint programming, by way of empirical research.

Another contribution is in the form of a new measurement called resampling ratio, which might aid in the development of evolutionary algorithms. Two minor contributions are in a study about evolutionary art and in the presentation and evaluation of a new approach to solving dynamic problems using evolutionary computation.

Akira Oyama (2000)

Wing Design Using Evolutionary Algorithm.
Ph.D. Thesis, Tohoku University, Sendai, Japan.
[Download Instructions] 
Abstract:
Although Evolutionary Algorithms (EAs) have become increasingly popular in aerodynamic design problems, the previous applications of EAs are restricted to more or less simplified problems involving not more than 10-30 design parameters. In contrast to that, in real-world design problems, a large number of design parameters must be handled - for example, a wing shape for a generic transonic aircraft usually parameterized by more than a hundred of design parameters. Since such problem has a highly multidimensional search space and extremely complicated objective function distribution, standard EAs would fail to find a globally optimum. This research develops a new, robust, and efficient numerical design method applicable to such real-world aerodynamic design problems.

Tomoharu Nakashima (2000)

Fuzzy Genetics-Based Machine Learning for Pattern Classification.
Ph.D. Thesis, Osaka Prefecture University, Osaka, Japan.
[Download (888123 Bytes)] 
Abstract:
Fuzzy rule-based systems have been successfully applied in the field of control, pattern classification and so on. One difficulty in designing fuzzy rule-based systems is the curse of dimensionality. That is, the number of generated fuzzy if-then rules exponentially increases with the number of input variables. In this dissertation, several genetics-based machine learning algorithms are proposed for designing fuzzy rule-based systems for high dimensional pattern classification problems. First, a fuzzy classifier system is proposed where a fuzzy if-then rules is treated as an individual in genetic algorithms. Thus the fuzzy classifier system can be viewed as a Michigan approach-based algorithm. Next, a Pittsburgh approach-based algorithm is considered where an individual is a set of fuzzy if-then rules. Finally a hybrid algorithm that combines the fuzzy classifier system and the Pittsburgh approach-based algorithm is proposed. Through the computer simulation on real-world pattern classification problems, advantages and disadvantages of each algorithm are shown.

Sean Luke (2000)

Issues in Scaling Genetic Programming: Breeding Strategies, Tree Generation, and Code Bloat.
Ph.D. Thesis, University of Maryland, College Park MD, USA.
[Download Instructions] 
Abstract:
Genetic Programming is an evolutionary computation technique which searches for those computer programs that best solve a given problem. As genetic programming is applied to increasingly difficult problems, its effectiveness is hampered by the tendency of candidate program solutions to grow in size independent of any corresponding increases in quality. This bloat in solutions slows the search process, interferes with genetic programming's searching, and ultimately consumes all available memory. The challenge for scaling up genetic programming is to find the best solutions possible before bloat puts a stop to evolution. This can be tackled either by finding better solutions more rapidly, or by taking measures to delay bloat as long as possible. This thesis discusses issues both in speeding the search process and in delaying bloat in order to scale genetic programming to tackle harder problems. It describes evolutionary computation and genetic programming, and details the application of genetic programming to cooperative robot soccer and to language induction. The thesis then compares genetic programming breeding strategies, showing the conditions under which each strategy produces better individuals with less bloating. It then analyzes the tree growth properties of the standard tree generation algorithms used, and proposes new, fast algorithms which give the user better control over tree size. Lastly, it presents evidence which directly contradicts existing bloat theories, and gives a more general theory of code growth, showing that the issue is more complicated than it first appears.

Eckart Zitzler (1999)

Evolutionary Algorithms for Multiobjective Optimization: Methods and Applications.
Ph.D. Thesis, Swiss Federal Institute of Technology Zurich, Zurich, Switzerland.
[Download Instructions] 
Abstract:
The subject of the thesis is the comparison and the improvement of existing multiobjective evolutionary algorithms and their application to system design problems in computer engineering. In detail, the major contributions are: i) an experimental methodology to compare multiobjective optimizers is developed, ii) an extensive comparison of numerous evolutionary techniques is performed, iii) a novel approach to multiobjective optimization, the strength Pareto evolutionary algorithm, is proposed, and iv) two complex multicriteria applications, the automatic synthesis of heterogeneous hardware/software systems and the multidimensional exploration of software implementations for digital signal processors, are addressed using evolutionary algorithms.

Tina Yu (1999)

Tina Yu.
Ph.D. Thesis, University College London, London, UK.
[Download Instructions]  [E-Mail] 
Abstract:
Genetic Programming (GP) automatically generates computer programs to solve specified problems. It develops programs through the process of a "create-test-modify" cycle which is similar to the way a human writes pprograms. There are various functional programming techniques that human programmers can use to accelerate the program development process. This research investigated the applicability of some of the functional techniques to GP and analyzed their impact on GP performance. Among many important functional techniques, three were chosen to be included in this research, due to their relevance to GP. They are polymorphism, implicit recursion and higher-order functions. To demonstrate their applicability, a GP system was developed with those techniques incorporated. Furthermore, a number of experiments were conducted using the system. The results were then compared to those generated by other GP systems which do not support these functional features. Finally, the program search space of the general even-parity problem was analyzed to explain how these techniques impact GP performance.

The experimental results showed that the investigated functional techniques have made GP more powerful in the following ways: 1) polymorphism has enabled GP to solve problems that are very difficult for standard GP to solve, i.e. nth and map programs; 2) higher-order functions and implicit recursion have enhanced GP's ability in solving the general even-parity problem to a greater degree than with any other known methods. Moreover, the analysis showed that these techniques directed GP to generate program solutions in a way that has never been previously reported. Finally, we provide the guidelines for the application of these techniques to other problems.

David A. Van Veldhuizen (1999)

Multiobjective Evolutionary Algorithms: Classifications, Analyses, and New Innovations.
Ph.D. Thesis, Air Force Institute of Technology, Dayton OH, USA.
[Download Instructions] 
Abstract:
This research classifies and analyzes contemporary Multiobjective Evolutionary Algorithm (MOEA) research and associated MOPs. 272 MOEA applications proposed in the literature are classified and cataloged, and are used as the basis for a detailed quantitative and qualitative MOEA analysis. New theorems and definitions are also presented. This research also extends the traditional notion of building blocks to the MOP domain in an effort to develop more effective and efficient MOEAs.

Using general test suite guidelines, comprehensive MOEA test function suites are generated based upon MOP characteristics and applicable MOEA theory. Using a developed MOEA test function suite, an experimental methodology incorporating known MOP solutions and appropriate test suite metrics is offered as a proposed evaluation framework allowing for absolute comparisons of specific MOEA approaches. This framework is then used in experiments examining MOEA performance in regard to test MOPs. Experimental results, their statistical analyses, and other germane observations are presented.

Tim Taylor (1999)

From Artificial Evolution to Artificial Life.
Ph.D. Thesis, University of Edinburgh, Edinburgh, UK.
[Download Instructions] 
Abstract:
This work addresses the question: What are the basic design considerations for creating a synthetic model of the evolution of living systems (i.e. an `artificial life' system)? It can also be viewed as an attempt to elucidate the logical structure (in a very general sense) of biological evolution. However, with no adequate definition of life, the experimental portion of the work concentrates on more specific issues, and primarily on the issue of open-ended evolution. An artificial evolutionary system called Cosmos, which provides a virtual operating system capable of simulating the parallel processing and evolution of a population of several thousand self-reproducing computer programs, is introduced. Cosmos is related to Ray's established Tierra system, but there are a number of significant differences. A wide variety of experiments with Cosmos, which were designed to investigate its evolutionary dynamics, are reported. An analysis of the results is presented, with particular attention given to the role of contingency in determining the outcome of the runs. The results of this work, and consideration of the existing literature on artificial evolutionary systems, leads to the conclusion that artificial life models such as this are lacking on a number of theoretical and methodological grounds. It is emphasised that explicit theoretical considerations should guide the design of such models, if they are to be of scientific value. An analysis of various issues relating to self-reproduction, especially in the context of evolution, is presented, including some extensions to von Neumann's analysis of self-reproduction. This suggests ways in which the evolutionary potential of such models might be improved. In particular, a shift of focus is recommended towards a more careful consideration of the phenotypic capabilities of the reproducing individuals. Phenotypic capabilities fundamentally involve interactions with the environment (both abiotic and biotic), and it is further argued that the theoretical grounding upon which these models should be based must include consideration of the kind of environments and the kind of interactions required for open-ended evolution. A number of useful future research directions are identified. Finally, the relevance of such work to the original goal of modelling the evolution of living systems (as opposed to the more general goal of modelling open-ended evolution) is discussed. It is suggested that the study of open-ended evolution can lead us to a better understanding of the essential properties of life, but only if the questions being asked in these studies are phrased appropriately.

A. K. Srivastava (1999)

Application of Artificial Neural Networks in GAs: Odour Identification Using Sensor Array.
Ph.D. Thesis, Banaras Hindu University, Vanarasi, India.
[Thesis Information]  [Profile] 
Abstract:
My thesis work is basically an engineering effort to mimic human olfactory system in its electronics counterpart, so called Electronic Nose (ENOSE), which consists of Sensor Array, Signal Processing and Pattern Classification. In order to process sensor array data for gas/ odour identification, goal was to design powerful neural network (NN) pattern classifier with improved NN learning ability and better classification accuracy. My investigations show that use of Genetic Algorithm (GA) in combination with NN not only promises to be an effective Intelligent Gas Sensing System. Novelty of my approach lies in the ability of NN to classify large number of similar gases with an array of limited number of sensors that too without using any pre-processing for data conditioning/ transformation. For this I required to design sophisticated and advanced genetic operators such as Double-MRX and Triple-MRX so as to accelerate the search ability of GA for unconstrained, continuous and non-linear optimization problems like learning in Neural Network. To accomplish this I developed an algorithm-oriented software package in high level C programming language on HP9000 computing machine running HP-UX O/S and tested the proposed algorithms over real-world gas identification problem. My results and finds are very useful in environmental monitoring, quality assurance, safety and security, military, space exploration and medicine.

Joao Carlos Figueira Pujol (1999)

Evolution of Artificial Neural Networks Using a Two-dimensional Representation.
Ph.D. Thesis, University of Birmingham, Birmingham, UK.
[Download] 
Abstract:
The design of artificial neural networks is still largely performed by an expert, with only a few heuristics to guide a trial-and-error search. Recently, new methods based on evolutionary computation (EC) have been applied to the synthesis of artificial neural networks with modest results. The basic limitation of EC-based methods is that they do not take into account the fact that artificial neural networks are two-dimensional structures, and do not use specialized evolutionary operators. In this work, a new method based on a special form of evolutionary computation called genetic algorithms is proposed for the evolution of artificial neural networks. The method is a general purpose procedure able to evolve feedforward and recurrent architectures. It is based on a two-dimensional representation, and includes operators to evolve the architecture and the connection weights simultaneously. The new approach has shown promising results, and has fared better than previous methods in a number of applications, including: binary classification problems, design of neural controllers and a complex navigation task of traversing a trail. An extension of the two-dimensional representation is also presented, which can be combined with other methods, providing them with an alternative procedure to evolve the weights of the connections.

Paul J. Kennedy (1999)

Simulation of the Evolution of Single Celled Organisms with Genome, Metabolism and Time-Varying Phenotype.
Ph.D. Thesis, University of Technology, Sydney, Australia.
[Download] 
Abstract:
My thesis introduces a model of a biological cell consisting primarily of a genome and metabolism. Genomes and initial chemical conditions coevolve together to produce cells that can survive in simple environments. Evolution of the two components occurs in different ways: the genomic evolution is Darwinian but evolution of the initial chemicals has Lamarckian characteristics. The genome uses a new biologically inspired "double-strand encoding" that permits more efficient search with the GA and a simple implementation of the inversion operator.

Jens Gottlieb (1999)

Evolutionary Algorithms for Constrained Optimization Problems.
Ph.D. Thesis, Technical University of Clausthal, Clausthal, Germany.
[Download Instructions] 
Notes: Published by: Shaker, ISBN 3-8265-7783-3, Aachen, 2000
Abstract:
The success of evolutionary algorithms applied to constrained optimization problems depends on the choice of adequate constraint-handling techniques. While this choice is usually guided by experience and testing rather than a formal analysis, this study presents an investigation of multidimensional knapsack problems, which considers both theoretical and empirical aspects. Knapsack problems belong to the general class of covering and packing problems, which have the essential structural property that all global optima are located in the boundary of the feasible search space. Analytical as well as empirical results show the performance of different constraint-handling techniques to depend strongly on their ability to focus the evolutionary search on the boundary of the feasible region.

The failure of previous penalty functions is explained by the fitness landscapes they induce, and a new function is proposed that reliably guides the search towards feasibility. The best setup for direct approaches employing repairing and local optimization allowed to determine new best feasible solutions for more than 100 instances with unknown optimum from the used benchmark suite. Concerning the permutation representation, a general comparison of variation operators gains insights into the inheritance of position, order, and adjacency information. This comparison helps to select appropriate operators for permutation problems in general, e.g. traveling salesman problems or scheduling problems.

Patrice Calegari (1999)

Parallelization of Population-based Evolutionary Algorithms for Combinatorial Optimization Problems.
Ph.D. Thesis, Swiss Federal Institute of Technology in Lausanne, Lausanne, Switzerland.
[Download Instructions] 
Abstract:
The objective of the present work is to make efficient parallelization of evolutionary algorithms (EA) easier in order to solve large instances of difficult combinatorial optimization problems within an acceptable amount of time, on parallel computer architectures. The fundamental ingredients of EAs are detailed and these ingredients are grouped by a classification tool called TEA (Table of Evolutionary Algorithms). This table is taken as a basis for the analysis of the criteria that influence the parallelization of EAs in order to define parallelization rules (especially on MIMD-DM Architectures: Multiple Instruction stream, Multiple Data stream, Distributed Memory). Experiments are done with different hybrid EAs (genetic algorithms and ant systems) that are executed on a network of 80 workstations in order to treat two problems: first the optimization of the best set of transceiver sites in a mobile radio network and second the classical graph coloring problem.

David Brittain (1999)

Optimisation of the Telecommunications Access Network.
Ph.D. Thesis, University of Bristol, Bristol, UK..
[Download] 
Abstract:
The telecommunications access network is the section connecting the local exchange to the customers. This part of the network is responsible for a large element of a telecommunications company's investment. With its increasing importance due to a rise in demand for bandwidth, new attention is being paid to developing tools for producing good quality network plans. In particular, new network technologies, which have been developed to allow the deployment of optical fibre into the access network, are more difficult to plan than traditional designs. The aim of this thesis is to develop computer-based tools to assist in the optimisation of access network plans. To achieve good quality plans it is necessary to have some metric for assessing them. Towards this aim, an object-oriented model of the access network is developed. The model is designed so that it is flexible enough to model a wide variety of network technologies and components. The system implements a cost model of the access network, which includes installation costs and net present value. In addition, an object model of the optimiser is also developed that interacts with the access network. Two optimisation systems are developed: a genetic algorithm (GA) and a simulated annealing (SA) algorithm. For the first, a novel set-based representation for the planning problem is introduced. Appropriate genetic operators are developed for this representation using Forma theory - a representation independent approach to operator design. The GA is supplemented with problem specific operators to assist its search. The GA is applied to a range of static, dynamic and uncertain access network planning problems. On the dynamic problems its performance is compared to SA and is found to perform much better across a range of problems. The uncertain problems involve considering the impact of demand uncertainty on the network planning problem. It is shown this uncertainty is important, as networks that perform well across a range of demand scenarios are different to the optima of the dynamic problems. The GA is applied nearly unchanged to these problems and is found to produce plans that perform well across a range of possible customer demands.

Uwe Aickelin (1999)

Genetic Algorithms for Multiple-Choice Optimisation Problems.
Ph.D. Thesis, University of Wales Swansea, Swansea, UK.
[Download Instructions] 
Abstract:
This thesis investigates the use of problem-specific knowledge to enhance a genetic algorithm approach to multiple-choice optimisation problems. It shows that such information can significantly enhance performance, but that the choice of information and the way it is included are important factors for success. Two multiple-choice problems are considered. The first is constructing a feasible nurse roster that considers as many requests as possible. In the second problem, shops are allocated to locations in a mall subject to constraints and maximising the overall income. Genetic algorithms are chosen for their well-known robustness and ability to solve large and complex discrete optimisation problems. However, a survey of the literature reveals room for further research into generic ways to include constraints into a genetic algorithm framework. Hence, the main theme of this work is to balance feasibility and cost of solutions. In particular, co-operative co-evolution with hierarchical sub-populations, problem structure exploiting repair schemes and indirect genetic algorithms with self-adjusting decoder functions are identified as promising approaches. The research starts by applying standard genetic algorithms to the problems and explaining the failure of such approaches due to epistasis. To overcome this, problem-specific information is added in a variety of ways, some of which are designed to increase the number of feasible solutions found whilst others are intended to improve the quality of such solutions. As well as a theoretical discussion as to the underlying reasons for using each operator, extensive computational experiments are carried out on a variety of data. These show that the indirect approach relies less on problem structure and hence is easier to implement and superior in solution quality. The most successful variant of our algorithm has a more than 99 finding a feasible solution which is either optimal or within a few percent of optimality.

Patrick Surry (1998)

A Prescriptive Formalism for Constructing Domain-specific Evolutionary Algorithms.
Ph.D. Thesis, University of Edinburgh, Edinburgh, UK.
[Download] 
Abstract:
It has been widely recognised in the computational intelligence and machine learning communities that the key to understanding the behaviour of learning algorithms is to understand what *representation* is employed to capture and manipulate knowledge acquired during the learning process. However, traditional evolutionary algorithms have tended to employ a fixed representation space (binary strings), in order to allow the use of standardised genetic operators. This approach leads to complications for many problem domains, as it forces a somewhat artificial mapping between the problem variables and the canonical binary representation, especially when there are dependencies between problem variables (e.g. problems naturally defined over permutations). This often obscures the relationship between genetic structure and problem features, making it difficult to understand the actions of the standard genetic operators with reference to problem-specific structures. This thesis instead advocates making the representation of solutions the explicit focus, in order to highlight the way in which the genetic operators (and resulting search algorithms) form and test hypotheses about the relationship between observed problem structure and fitness.

William M. Spears (1998)

The Role of Mutation and Recombination in Evolutionary Algorithms.
Ph.D. Thesis, George Mason University, Fairfax VA, USA.
[Download Instructions] 
Abstract:
Despite decades of work in evolutionary algorithms (EAs), there remains a lot of uncertainty as to when it is good or bad to use recombination or mutation. This thesis provides a theoretical and empirical study of recom- bination and mutation in EAs, in order to better characterize the roles of these operators. The analysis shows that the multimodality of the fitness landscape has a large effect on the performance of an EA with recombination, while having little effect on the performance of an EA without recombination. The thesis also unifies existing theoretical techniques and introduces new theoretical and empirical techniques for studying EAs.

Jayshree Sarma (1998)

An Analysis of Decentralized and Spatially Distributed Genetic Algorithms.
Ph.D. Thesis, George Mason University, Fairfax VA, USA.
[PostScript]  [GZipped PostScript] 

Mitchell A. Potter (1997)

The Design and Analysis of a Computational Model of Cooperative Coevolution.
Ph.D. Thesis, George Mason University, Fairfax, VA.
[Download Instructions] 

Farid Khoshalhan (1997)

Farid Khoshalhan.
Ph.D. Thesis, Tarbiat Modarres University, Tehran, Iran.
[E-Mail] 
Abstract:
Assembly line balancing (ALB) problem is known as one of difficult combinatorial optimization problems. This problem has been solved with various approaches but most of them do not lead to efficient algorithms. Genetic algorithms (GAs) has been recognized as an efficient and useful procedure for solving large and hard combinatorial optimization problems, such as scheduling problems. Fuzzy sets theory is frequently used to represent uncertainty of information in real world problems. In this research, a new GA to solved type-2 of fuzzy ALB problem is presented. A new repair method and a new efficient crossover operator are developed. An interval of confidence of fuzzy numbers is used to show the decision maker's degree of optimism. The performance comparison between proposed GA and other GA approaches shows that our approach is promising.

Darko Grundler (1997)

Multilevel Fuzzy Process Control Optimized by Genetic Algorithm.
Ph.D. Thesis, University of Zagreb, Zagreb, Croatia.
[Download Instructions] 
Notes: Dissertation is in Croatian language only.
Abstract:
The new method for complex processes control with the coordinating control unit based upon a genetic algorithm has been described. Minimal energy spent criteria restricted by given process response limitations has been applied, and improvement in relation to other known optimizing methods has been found. Independent and non-coordinating PID and fuzzy regulator parameter tuning has been performed using a genetic algorithm and the results achieved are the same or better than with traditional optimizing methods while at the same time proposed method can be easily automated.

Rogene M. Eichler West (1996)

On the development and interpretation of parameter manifolds for biophysically robust compartmental models of CA3 hippocampal neurons..
Ph.D. Thesis, University of Minnesota, Saint Paul, MN.
[Download Instructions] 
Abstract:
Electrical spiking behaviors in neurons are governed by a class of membrane-spanning proteins known as voltage-gated and/or calcium-dependent (VGCD) channels. Neuroscientists would like to understand how neurons encode information in these spiking behaviors and how VGCD channels contribute to this information processing. The distribution of VGCD channels, both in number and spatial location, is assumed to play an important role in the behavior of neurons. We don't have the ability to perform exact measurements or manipulations of VGCD channel distributions experimentally, so in this dissertation I developed a methodology and tools to obtain and interpret the plausible ranges of these channel distributions computationally. Currently, investigators studying compartmental models set their system parameters by an iterative, trial-and-error process. Proceeding in this manner can be as inefficient and can lead to the inadvertent selection of mathematical singularities. We would like to know ALL the parameter sets that produce appropriate behaviors because such a manifold, or boundary enclosing a volume of solutions, captures the reality of biological diversity.

I verified the accuracy of the simulations and optimized simulator performance in order to maximize the size of the search space we could explore within a fixed allocation of computational resources. I custom-designed an evolutionary algorithm (EA) to explore the 114 dimensional space of a hippocampal CA3 neuron model for solutions that corresponded well with experimentally determined spatiotemporal behaviors of this class of neuron. The resulting channel distribution manifolds corresponded well with experimental data and allowed predictions regarding the contributions of various channel types to the behavior of the model neuron. This research has developed and validated a technique that will aid in the development of biophysically sophisticated neuronal models when insufficient quantitative, experimental information is available to specify system parameters. This technique promises to be more efficient and reliable than current parameter setting methods. The programs developed as part of this research automated the prediction of functional spatial distributions of VGCD channels in a high dimensional space, a problem hindering the development of single neuron models for the past forty years. The contribution of this dissertation to science is uniquely characterized by the automated production and interpretive paradigm of a MANIFOLD approach to computational modeling.

Tim Kovacs (1996)

Evolving Optimal Populations with XCS Classifier Systems.
Ph.D. Thesis, University of Birmingham, Birmingham, UK.
[Download] 
Abstract:
Describes the implementation and motivation behind Wilson's XCS classifier system in detail and contrasts it with earlier systems. Proposes the XCS optimality hypothesis: that XCS will tend towards populations which are complete, accurate and minimal. Demonstrates that XCS reliably evolves optimal populations for the multiplexer function. Finally, the macroclassifier technique is validated.

Paul Field (1996)

A Multary Theory for Genetic Algorithms: Unifying Binary and Nonbinary Problem Representations.
Ph.D. Thesis, Univeristy of London, London, UK.
[Download] 
Abstract:
To use a genetic algorithm you need to devise a representation for the possible solutions to your problem. The traditional approach is to use fixed-length binary strings and there is theoretical and practical evidence that this is best. A different approach is to still use fixed-length strings but to use larger alphabets. There is theoretical evidence that this nonbinary approach is best and there are various claims for its successful practical use.

The choice of representation can affect a GA's performance and so determining the best approach is important. However, the majority of GA theory is limited to binary and so has nothing to say about nonbinary.

This thesis presents a complete and precise theoretical framework in which binary and nonbinary can be studied at a variety of levels of abstraction. The framework is formed from several pieces of multary theory (i.e. theory that applies to binary and nonbinary) developed in this thesis.

To begin with, the fundamental multary concepts that underpin all the theory in the thesis are developed. These concepts allow multary ideas to be expressed easily and often in a more clear and elegant manner than their existing binary-only analogues. The thesis then tackles transforms, developing multary transforms analogous to the Walsh transform including the theoretical and implementation details of fast multary transforms. It introduces a new approach to understanding transforms based on an investigation of implicit parallelism and the notion of key schemata and sets about dispelling the limiting idea that transforms can only be used to study deception. Complete and precise models of genetic algorithms, the infinite and finite multary models, are developed and analysed, revealing a wealth of unexpected structure within the models. Finally, all the theoretical threads are interwoven to form a complete and precise theoretical framework in which general multary results can be proved using binary problem representations.

This thesis also sets out to resolve the binary/nonbinary debate. It contends that there is no best approach and that different representations are best in different circumstances. Using the term multary to denote this unified view of representations, it brings new perspectives to the arguments about binary and nonbinary that lead to new conclusions: multary conclusions. The multary GA models are used to demonstrate the validity of the multary approach by showing that different representations are best for different problems and by showing some of the difficulties a GA faces when an "unnatural" representation is forced on a problem.

Paul J. Darwen (1996)

Co-evolutionary Learning by Automatic Modularisation with Speciation.
Ph.D. Thesis, %s, %s.
[Download] 
Abstract:
Co-evolutionary learning sets up an "arms race" of escalating sophistication, and is suitable for a range of management and control problems. This thesis demonstrates that an "arms race" can stagnate due to overspecialization, causing brittle solutions that do not generalize well to novel situations. Some plausible changes to co-evolution does not fix this, but niching/speciation methods do help. This thesis critiques the fitness sharing method of niching, and demonstrates that a variation on it performs better. Using speciation to create a repertoire of diverse strategies, and gating those strategies with a voting setup, produces an improved method to learn to play a game without human expertise.

Carlos A. Coello (1996)

An Empirical Study of Evolutionary Techniques for Multiobjective Optimization in Engineering Design.
Ph.D. Thesis, Tulane University, New Orleans LA, USA.
[Download] 
Abstract:
Most real-world engineering optimization problems are multiobjective in nature, since they normally have several (possibly conflicting) objectives that must be satisfied at the same time. The word "optimum" has several interpretations within this context, and it is up to the designer to decide which fits better to his/her application. Currently, there are more than 20 mathematical programming multiobjective optimization techniques, each one corresponding to a different understanding of the term "optimum". On the other hand, genetic algorithms (GAs) have been viewed to be, since their early days, well suited for multiobjective optimization problems. Consequently, several GA-based techniques have been developed since then.

The purpose of this research has been to develop a platform that allows the testing and comparison of existing and future multiobjective optimization techniques. Two new multiobjective optimization GA-based methods based on the notion of min-max optimum are proposed, showing that at least one of them is able to produce better results than any other technique tested. Also, a method for adjusting the parameters of the GA for single-objective numerical optimization is proposed, showing the suitability of the GA as a numerical optimization technique when used properly. Then, a brief study of the importance of population policies and proper niching parameters is included. This work tries to narrow the gap between theory and practice in the context of engineering optimization. Finally, some insights on the importance of choosing a good chromosomic representation and the use of a proper fitness function are provided, derived from the analysis of a more general design problem.

Peter Bentley (1996)

Generic Evolutionary Design of Solid Objects using a Genetic Algorithm.
Ph.D. Thesis, University of Huddersfield, Huddersfield, UK.
[Download] 
Abstract:
Generic evolutionary design means the creation of a range of different designs by evolution. This thesis introduces generic evolutionary design by a computer, describing a system capable of the evolution of a wide range of solid object designs from scratch, using a genetic algorithm. The thesis reviews relevant literature, and outlines a number of advances necessitated by the development of the system, including: a new generic representation of solid objects, a new multiobjective fitness ranking method, and variable-length chromosomes. A library of modular evaluation software is also described, which allows a user to define new design problems quickly and easily by picking combinations of modules to guide the evolution of designs. Finally, the feasibility of generic evolutionary design by a computer is demonstrated by presenting the successful evolution of both conventional and unconventional designs for a range of different solid-object design tasks, e.g. tables, heatsinks, prisms, boat hulls, aerodynamic cars.

Kalyanmoy Deb (1995)

Binary and Floating-point Function Optimization using Messy Genetic Algorithms.
Ph.D. Thesis, University of Alabama, Tuscaloosa AL, USA.
Abstract:
(need synposis)

Walter Cedeno (1995)

The Multi-Niche Crowding Genetic Algorithm.
Ph.D. Thesis, University of California Davis, Davis CA, US.
[Download Instructions] 
Abstract:
In this dissertation we present the multi-niche crowding genetic algorithm, a computational metaphor to the survival of species in ecological niches in the face of competition. We present empirical and theoretical results for the success of the multi-niche crowding genetic algorithm for multimodal function optimization. We test the performance of the algorithm on problems of DNA Mapping, Aquifer Management, and the File Design Problem.

Peter J. Angeline (1993)

Evolutionary Algorithms and Emergent Intelligence.
Ph.D. Thesis, Ohio State University, Columbus OH, USA.
[Download] 
Abstract:
Description needed