摘要
Bagging算法是目前一种流行的集成学习算法,采用一种改进的Bagging算法Attribute Bagging作为分类算法,通过属性重取样获取多个训练集,以kNN为弱分类器设计一种中文文本分类器。实验结果表明Attribute Bagging算法较Bagging算法有更好的分类精度。
Bagging algorithm is a popular ensemble learning technology.A Chinese text categorization classifier is designed by using an improved Bagging algorithm-Attribute Bagging(AB).Re-sampling attribute is used to get multiple training sets;the kNN is selected as weak learner.Experiments show that the Attribute Bagging gets lower errors and better performance than Bagging.
出处
《计算机工程与应用》
CSCD
北大核心
2009年第5期135-137,179,共4页
Computer Engineering and Applications
基金
国家自然科学基金(No.60573179)~~