Gene Expression Programming In Classification Rules Mining

Data mining(DM) is one of the hottest research topics in current computer science research area. With the rapid development of database and data warehouse technology, people are gathering data at an unprecedented speed. Enterprises and organizations are collecting and storing data everyday, but few of them can really utilize the data to serve their entities. The slogan of information age is just a beautiful vision in our mind. What's more, the beautiful vision didn't turn into reality as we had wished. We are still far away from information age. With our developing understanding of the essence of information technology, more and more scholars as well as researchers devote themselves into the research of how to make good use of available data. Data mining emerges at the right time. After the development of more than a decade, DM has set up a solid foundation based on association rules mining, classification rules mining and cluster rules mining. It has combined the database, statistics, AI, visualization, and IT. And its application has spread from business to medical, military, commercial fields.Classification is one of the most important branches of DM. In the last decade, experts from various fields have developed numerous algorithms to perform this task. Among all the methods, information theory based decision tree, probability based Bayesian network (BN) and neuro-science based neural networks (NN) are the most widely used and successful approaches. However, all the methods mentioned above are deterministic. Natural evolution based evolutionary computation (EC) has become a special branch for its unique combination of intelligence, parallelism. EC has been successfully applied in many engineering fields.Genetic algorithm (GA) is the most important branch of EC. Genetic programming (GP) is a variant of GA. Although both of them follow the rule of survival of the fittest, their initial application in engineering fields are quite different. GA is mainly used in function optimization, while GP is mainly used in modeling. Generally speaking, both of them outperform traditional statistical methods. In recent years, EC has been successfully applied in DM, especially in classification rules mining. EC has become an indispensable tool of DM.Gene expression programming (GEP) is a newly invented EC method. GEP combines the...