一种基于多聚类器集成的分类器优化方法
首发时间:2006-05-05
摘要:为解决海量数据集的分类器构造难题,本文提出一种将基于小样本的分类器和基于多个大样本的多聚类器集成相结合的分类器优化方法。接着将多聚类器集成的结果如何融入优化小样本分类器作了初步说明,目的是为了提高分类器的稳定性和模型简洁性。针对具有有序类别性的一类问题,给出一种更简洁更高效的建模方法并给出其误差分析。最后通过一个人工数据集对本文的方法进行仿真实验验证,证实了这种优化方法具有一定的意义。
关键词: 小样本分类 大样本聚类 有序类别
For information in English, please click here
an Optimizing-classifier Method Based on Multicluster
Abstract:In order to solve a constructing-classifier problem of large data sets, it introduces an optimizing-classifier method based on a little sample classifier and an integration of several large sample clusters. It gives a simple description about how to apply the result of several large sample clusters to the little sample classifier in order to improve stability and succinctness of classifying model. It gives a simple and efficient modeling method aiming at an ordered-sort classification problem. At last, the method is valid by an experiment using an manual data sets.
Keywords: Little sample classifying large sample clustering ordered sort
基金:
论文图表:
引用
No.6501220511146813****
同行评议
勘误表
一种基于多聚类器集成的分类器优化方法
评论
全部评论0/1000