摘要: 为了提高数据的分类性能,提出一种集成学习的多分类器动态组合方法(DEA)。该方法在多个UCI标准数据集上进行测试,并与文中使用的基于Adaboost算法训练出的各个成员分类器的分类效果进行比较,证明了DEA的有效性。
关键词:
多分类器,
聚类,
动态分类器组合,
Adaboost算法
Abstract: In order to improve the classification performance of dataset, a dynamic combinatorial method of multiple classifiers on ensemble learning DEA is proposed in the paper. DEA is tested on the UCI benchmark data sets, and is compared with several member classifiers trained based on the algorithm of Adaboost. In this way, the utility of DEA can be proved.
Key words:
multiple classifiers,
clustering,
dynamic classifier ensemble,
Adaboost algorithm
中图分类号:
陈 冰;张化祥. 集成学习的多分类器动态组合方法[J]. 计算机工程, 2008, 34(24): 218-220.
CHEN Bing; ZHANG Hua-xiang. Dynamic Combinatorial Method of Multiple Classifiers on Ensemble Learning[J]. Computer Engineering, 2008, 34(24): 218-220.