The Research On Classifier Ensemble Learning For Data Mining

Classification is an important task of data mining. The design of the classifier deeply affects the performance of the classification system. Classification problem that is hard for a single classifier can be settled down with classifier ensemble. In addition, classifier ensemble outperforms single classifier on efficiency, accuracy and so on. In the last decade, classifier ensemble has become a common concern of both researchers and technicians. Researches on ensemble learning theoretic and application are coming forth continuously. Some international magazines, such as Machine Learning, Artificial Intelligence, Information Fusion, etc. have delivered their specials for classifier ensemble learning. And world wide International conferences dedicated to classifier ensemble learning, such as MCS, are holding every year.On the basis of understanding and analyzing the current research state, research focuses and development trend in the domain of classifier ensemble, this dissertation focuses on improving the performance of the classifier ensemble and expanding its application. The main achievements is as follows:1. This dissertation makes a survey about the research on classifier ensemble learning, including its background, the current research state and development trend, etc.2. This dissertation analyzes the relationship between the ensemble performance and component classifier's performance and diversity. It is noted that the generalization ability of the ensemble is always better than the average generalization ability of its components. The ensemble'performance can be further improved by selecting a group of accurate and diverse...