基于模糊隶属度邻域覆盖的三支分类决策

THREE WAY CLASSIFICATION METHOD BASED ON NEIGHBORHOOD COVERAGE OF FUZZY MEMBERSHIP DEGREE

传统的邻域分类决策方法对不确定数据进行了严格的分类,可能导致严重的分类错误,因此提出一种基于模糊隶属度邻域覆盖的三支分类决策方法。引入模糊邻域覆盖方法,构建邻域覆盖隶属度相关的不确定测度,并且提供数据分布的隶属度近似。通过三支分类策略降低分类风险。通过多个数据集分类实验结果可知,提出的方法在保证分类精度的条件下极大地降低了分类风险。

The traditional neighborhood classification method strictly classifies uncertain data,which may lead to serious classification errors.Therefore,a three way classification method based on fuzzy membership degree neighborhood coverage is proposed.The fuzzy neighborhood covering method was introduced to construct the uncertainty measure related to the membership degree of neighborhood coverage,and the membership degree approximation of data distribution was provided.Three way classification strategies were used to reduce the classification risk.The experimental results of several data sets show that the proposed method can greatly reduce the classification risk under the condition of ensuring the classification accuracy.