分类器组合技术的多样性研究

分类器组合技术可以提高模式识别的性能，受到了模式识别领域研究人员的广泛关注。实现成员分类器的多样性是提高分类器组合泛化能力主要手段。本文从成员分类器的生成介绍了实现成员分类器多样性的各种方法，同时介绍了度量成员分类器多样性的各种技术，并提出了一种如何训练多样性成员分类器的技术思路。

Research on the diversity of Ensemble Classifiers

Since the technique of ensemble classifiers can improve the performance of classification, more and more attentions have been widely paid by researchers in the field of pattern recognition. To obtain diverse component classifiers is the key approach to guaranteeing the high classification performance and good generalization capability for an ensemble classifier. This paper introduces the methods relating to the diversity of ensemble classifier techniques in two aspects ： individual classifier design and the selection of classifiers.