摘要
模糊决策树归纳学习是从示例中产生规则知识的一个重要方法,决策树的产生过程涉及到两个重要的参数α、β。一般说来,这两个参数的选取依赖于所讨论的领域知识和用户的需要,若选取不当,会对分类结果产生很大影响,从而导致不正确的分类。如何选取这两个参数的值目前尚无较好的方法,仅凭人们的经验而定,该文提出了一种应用遗传算法来优化模糊决策树中参数的方法,旨在为选取参数提供实验方法,同时也为直接选取经验参数提供了一定的实验支撑。
Fuzzy decision tree inductive learning is an important method which generates knowledge from cases,the building of decision tree involves two important parametersα?β.Generally speaking,the selection of the parameters depends on domain knowledge discussed and user's demand,the inapposite selection will strongly influence the classification result and result in incorrect classification.Currently,no empirical results with selection of two parameters are yet available,only depends on experience.This paper introduces an GA based approach to optimization of the two parameters,in order to offer an experimental method of choice and some experimental support of directly selection.
出处
《计算机工程与应用》
CSCD
北大核心
2003年第25期88-91,97,共5页
Computer Engineering and Applications
基金
河北省自然科学基金资助项目"基于模糊信息的示例学习理论及算法"(编号:698139)