摘要
在研究属性识别理论的基础上,利用概率统计理论建立了新的分类器,它先单独计算每个属性对分类的贡献,然后通过一种加权机制计算所有属性对样本的分类情况.同时,为了克服单分类器使用范围有限和分类准确度相对不高等特点,把新的分类系统与传统的K-NN分类器结合起来,进一步提高了此分类系统的分类能力.实验结果表明:这个分类系统具有较好地分类效果和鲁棒性.
This article applied the probability statistics theory to design a classifier on the basis of attribute recognition theory. It calculated the contribution to the classification of every attribute, and then used these data to calculate the result of integrated attributes. In order to overcome the defect of lower accuracy of single classifier, this article integrated this classification system with the traditional KNN classifier to form a new method of sample classification. The experiment result has demonstrated that the new method has a good classification effect and low time complexity.
出处
《湖南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2009年第4期89-92,共4页
Journal of Hunan University:Natural Sciences
基金
湖南省自然科学基金资助项目(07JJ5085)
关键词
属性集
属性测度
相似度
准确率
attribute set
attribute measurement
similitude degree
accuracy
