Heuristic methods of solving exploratory data analysis problems suffer from one major weakness - uncertainty regarding the optimality of the results. The developers of DaMI (Data Mining Initiative), a genetic algorithm designed to mine the CCEP (Comprehensive Clinical Evaluation Program) database in the search for a Persian Gulf War syndrome, proposed a method to overcome this weakness: reproducibility -- the conjecture that consistent convergence on the same solutions is both necessary and sufficient to ensure a genetic algorithm has effectively searched an unknown solution space. We demonstrate the weakness of this conjecture in light of accepted genetic algorithm theory. We then test the conjecture by modifying the CCEP database with the insertion of an interesting solution of known quality and performing a discovery session using DaMI on this modified database. The necessity of reproducibility as a terminating condition is falsified by the algorithm finding the optimal solution without yielding strong reproducibility. The sufficiency of reproducibility as a terminating condition is analyzed by manual examination of the CCEP database in which strong reproducibility was experienced. Ex post facto knowledge of the solution space is used to prove that DaMI had not found the optimal solutions though it gave strong reproducibility, causing us to reject the conjecture that strong reproducibile is a sufficient terminating condition.

[Truncated abstract] Genetic algorithms (GAs) have been extensively used and studied in computer science, yet there is no generally accepted methodology for exploring which parameters significantly affect performance, whether there is any interaction between parameters and how performance varies with respect to changes in parameters. This thesis presents a rigorous yet practical statistical methodology for the exploratory study of GAs. This methodology addresses the issues of experimental design, blocking, power and response curve analysis. It details how statistical analysis may assist the investigator along the exploratory pathway. The statistical methodology is demonstrated in this thesis using a number of case studies with a classical genetic algorithm with one-point crossover and bit-replacement mutation. In doing so we answer a number of questions about the relationship between the performance of the GA and the operators and encoding used. The methodology is suitable, however, to be applied to other adaptive optimization algorithms not treated in this thesis. In the first instance, as an initial demonstration of our methodology, we describe case studies using four standard test functions. It is found that the effect upon performance of crossover is predominantly linear while the effect of mutation is predominantly quadratic. Higher order effects are noted but contribute less to overall behaviour. In the case of crossover both positive and negative gradients are found which suggests using rates as high as possible for some problems while possibly excluding it for others. .... This is illustrated by showing how the use of Gray codes impedes the performance on a lower modality test function compared with a higher modality test function. Computer animation is then used to illustrate the actual mechanism by which this occurs. Fourthly, the traditional concept of a GA is that of selection, crossover and mutation. However, a limited amount of data from the literature has suggested that the niche for the beneficial effect of crossover upon GA performance may be smaller than has traditionally been held. Based upon previous results on not-linear-separable problems an exploration is made by comparing two test problem suites, one comprising non-rotated functions and the other comprising the same functions rotated by 45 degrees in the solution space rendering them not-linear-separable. It is shown that for the difficult rotated functions the crossover operator is detrimental to the performance of the GA. It is conjectured that what makes a problem difficult for the GA is complex and involves factors such as the degree of optimization at local minima due to crossover, the bias associated with the mutation operator and the Hamming Distances present in the individual problems due to the encoding. Furthermore, the GA was tested on a real world landscape minimization problem to see if the results obtained would match those from the difficult rotated functions. It is demonstrated that they match and that the features which make certain of the test functions difficult are also present in the real world problem. Overall, the proposed methodology is found to be an effective tool for revealing relationships between a randomized optimization algorithm and its encoding and parameters that are difficult to establish from more ad-hoc experimental studies alone.

Genetic algorithms (GAs) are computer-based search techniques patterned after the genetic mechanisms of biological organisms that have adapted and flourished in changing, highly competitive environments for millions of years. GAs have been successfully applied to problems in a variety of studies, and their popularity continues to increase because of their effectiveness, applicability, and ease of use. Industrial Applications of Genetic Algorithms shows how GAs have made the leap form their origins in the laboratory to the practicing engineer's toolbox. Each chapter in the book describes a project completed by a graduate student at the University of Alabama.

Genetic Algorithms and Genetic Programming: Modern Concepts and Practical Applications discusses algorithmic developments in the context of genetic algorithms (GAs) and genetic programming (GP). It applies the algorithms to significant combinatorial optimization problems and describes structure identification using HeuristicLab as a platform for al

The articles presented here were selected from preliminary versions presented at the International Conference on Genetic Algorithms in June 1991, as well as at a special Workshop on Genetic Algorithms for Machine Learning at the same Conference. Genetic algorithms are general-purpose search algorithms that use principles inspired by natural population genetics to evolve solutions to problems. The basic idea is to maintain a population of knowledge structure that represent candidate solutions to the problem of interest. The population evolves over time through a process of competition (i.e. survival of the fittest) and controlled variation (i.e. recombination and mutation). Genetic Algorithms for Machine Learning contains articles on three topics that have not been the focus of many previous articles on GAs, namely concept learning from examples, reinforcement learning for control, and theoretical analysis of GAs. It is hoped that this sample will serve to broaden the acquaintance of the general machine learning community with the major areas of work on GAs. The articles in this book address a number of central issues in applying GAs to machine learning problems. For example, the choice of appropriate representation and the corresponding set of genetic learning operators is an important set of decisions facing a user of a genetic algorithm. The study of genetic algorithms is proceeding at a robust pace. If experimental progress and theoretical understanding continue to evolve as expected, genetic algorithms will continue to provide a distinctive approach to machine learning. Genetic Algorithms for Machine Learning is an edited volume of original research made up of invited contributions by leading researchers.

First Published in 1987. This is the collected proceedings of the second International Conference on Genetic Algorithms held at the Massachusetts Institute of Technology, Cambridge, MA on the 28th to the 31st July 1987. With papers on Genetic search theory, Adaptive search operators, representation issues, connectionism and parallelism, credit assignment ad learning, and applications.

Genetic algorithms have been shown effective for solving complex optimization problems such as job scheduling, machine learning, pattern recognition, and assembly planning. Due to the random process involved in genetic algorithms, the analysis of performance characteristics of genetic algorithms is a challenging research topic. Studied in this dissertation are methods to analyze convergence of genetic algorithms and to investigate whether modifications made to genetic algorithms, such as varying the operator rates during the iterative process, improve their performance. Both statistical analysis, which is used for investigation of different modifications to the genetic algorithm, and probability analysis, which is used to derive the expectation of convergence, are used in the study. The Wilcoxon signed rank test is used to examine the effects of changing parameters in genetic algorithms during the iterations. A Markov chain is derived to show how the random selection process affects the genetic evolution, including the so called genetic drift and preferential selection. A link distance is introduced as a numerical index for the study of the convergence process of order-based genetic algorithms. Also studied are the effects of random selection, mutation operator, and the combination of both to the expected average link distance. The genetic drift is shown to enforce the convergence exponentially with increase in the number of iterations. The mutation operator, on the other hand, suppresses the convergence. The combined results of these two parameters lead to a general formula for the estimation of the expected number of iterations needed to achieve convergence for the order-based genetic algorithm with selection and mutation and provide important insights about how order-based genetic algorithms converge.