The Practical Research Of Genetic Algorithms And Its Application In Knowledge Discovery And Case-Based Reasoning

Genetic Algorithms (GAs) are optimization algorithms imitating biological evolution with computers. They were first put forward by Holland in University of Michigan in 1975. With the fixed population size, GAs only use fitness value to guide searching and regenerate population through executing genetic operations such as selection, crossover and mutation with probability. Thus the GA is in fact a "blind heuristic searching strategy with probabilities". Because of their simplicity, flexibility, and being prone to parallel computing, they have been widely used in various fields.The dissertation consists of six chapters. Chapter 1 retrospects the development of GAs in history and summarizes the backgrounds of GAs and corresponding theories of biologic evolvement. The main contents and characters of every developing phase of GAs are studied. After comparing GAs with other search methods and analyzing the present research and the trend of development of GAs, we put forward the research direction and emphases of the dissertation.Chapter 2 studies and concludes the basic theories of GAs and systematically analyzes the basic principle of GAs ?Schemate Theorem. Various controlling parameters and genetic operators of GA, which can affect its running performance, are thoroughly discussed. This chapter analyzes and discusses the fundamental performance of GAs, especially the convergence performance. After discussing the pre-maturity of GAs, some methods to avoid it are provided.Chapter 3 discusses several key problems when using GAs. These problems are the range of using GAs, coding, setting of parameters and the operators of GAs, such as selection, crossover, mutation and others.Chapter 4 discusses the application of GAs in Knowledge Discovery in Database. Firstly, the corresponding conceptions and the simple process of Knowledge Discovery in Database are briefly introduced. Then, an improved GA is applied to discover the relational rules in a database. The Good-point Set GA is proposed, and its theory and applications are described and analyzed.Chapter 5 discusses the application of GAs in Case Based Reasoning. In the field of Case Based Reasoning, GAs can be used to maintain the Case Bases and find the weight of features. At last, we raise a developing platform of Case Based Reasoning system based on GAs.Chapter 6 sums up the dissertation and depicts the future areas and direction for research.