摘要
提出一种基于自组织特征映射网络(SOFM)和遗传算法的定量数据规则提取模型.首先利用自组织特征映射网络SOFM,把定量数据转换为模糊集的语义值,用模糊集合相似性分析与融合对初始的模糊区间进行约简,以提高其解释性.然后利用变精度粗糙集模型从中挖掘模糊分类规则.最后利用遗传算法对所得规则进行优化,在不降低精确性的前提下,减少规则数.选用UCI数据集中的数据进行测试,证明所提模型用于定量数据规则提取的有效性.
A rule induction model based on self-organizing feature map (SOFM) and genetic algorithm is proposed for quantitative data. Each quantitative value is first transformed into a fuzzy set of linguistic terms using SOFM and the similarity analysis and merging of fuzzy sets is used to reduct the initial fuzzy regions for enhancing the interpretability. Then variable precision rough set model is used to mine fuzzy classification rules. Finally, genetic algorithm is used to optimize the rules and reduce the rule numbers without depressing the precision. UCI database is selected to demonstrate that the new mndel is valid for rule induction from quantitative data.
出处
《系统工程理论与实践》
EI
CSCD
北大核心
2008年第7期150-154,164,共6页
Systems Engineering-Theory & Practice
基金
国家自然科学基金(60273043)
安徽省高校拔尖人才基金(05025102)
安徽省自然科学基金(050420204)
关键词
定量数据
自组织特征映射网络
变精度粗糙集
遗传算法
quantitative data
self-organizing feature map
variable precision rough set
genetic algorithm