摘要
通过选择性集成可以获得比单个学习器和全部集成学习更好的学习效果,可以显著地提高学习系统的泛化性能。文中提出一种多层次选择性集成学习算法,即在基分类器中通过多次按权重进行部分选择,形成多个集成分类器,对形成的集成分类器进行再集成,最后通过对个集成分类器多数投票的方式决定算法的输出。针对决策树与神经网络模型在20个标准数据集对集成学习算法Ada-ens进行了实验研究,试验证明基于数据的集成学习算法的性能优于基于特征集的集成学习算法的性能,有更好的分类准确率和泛化性能。
Through the selective ensemble,the algorithm would be more effective than each single one and better than the algorithm that select all the base classifier, and the algorithm would have effective generalization ability. In this paper, a selective multi - level integrated learning algorithm is presented. In the base classifier by repeatedly carried out by some of the weight of choice, the formation of a number of integrated classifiers, the formation of the integrated classifier re- integration, the final integration of a majority vote classifier algorithm to determine the output. For decision tree and neural network modal in 20 data sets the standard learning algorithm of the integrated experimental study of a Ada_ ens. Tested based on data integrated learning algorithm is better than the performance of the integrated feature set based on the learning algorithm performance, better classification accuracy and generalization performance.
出处
《计算机技术与发展》
2010年第2期87-89,94,共4页
Computer Technology and Development
基金
省级教学研究项目(2008jyxm305)
关键词
机器学习
集成学习
选择性集成
machine learning
ensemble learning
selective ensemble
