多类问题的神经网络集成方法分析-计算机技术专业论文.docxVIP

II II Abstract In the tide of the network interconnection, we are facing a growing number of multi-class classification problems, such as Remote sensing image analysis, text categorization, intrusion detection, face recognition, webpage classification, security situation prediction, virus program detection, and so on. Theoretically，due to the distributed processing and learning generalization ability in dealing with binary classification problems，learning algorithm has been gradually matured over the past few decades.However, for processing a multi-class classification problem, with the increase of the number of desired classification categories, classifier’s design become more difficult.Therefor how to deal with the complexity of multi-class classification problem, has gained much attention in the field of machine learning. In order to solve the multi-class classification problems, usually decompose them into multiple binary problems. We commonly use One - against - all and One-against-one method to solve it .We apply BP neural network and support vector machine classifier as the base classifier, then using the One-against-all or one-against-one strategy to decompose the original multi-class problem into multiple binary problems respectively , and combined with ensemble learning to solve the multi-class classification problem. The main scientific researchs of the paper are summarized as follows: The first method is based on BP neural network ensemble learning classification algorithm.Traditional One-against-all (OAA) decomposition approach performance more depends on the individual classifiers accuracy, rather than its diversity.A standard technique used to resolve these problem is to decompose the original multi-class problem into multiple binary problem.In this paper we propose a new learning model applicable to multi-class domains.The proposed model is an Artificial Neural Network ensemble in which the base learners are composed by the union of a binary classi