基于信息粒度与交互信息的属性约简改进算法

An improved attribute reduction algorithm based on information granularity and interaction information

经典粗糙集的前向启发式正域约简算法没有考虑到存在多个重要度最大的条件属性时如何合理地去选择候选属性;同时在度量条件属性间相关性时忽略了决策属性的影响程度,由此得到的约简集合泛化能力较弱.针对这些问题,首先引入信息粒度,提出一种候选属性选择的优化策略;其次引入信息论中交互信息的概念,通过计算属性之间的交互信息来剔除冗余属性;并在此基础上,提出了基于信息粒度与交互信息的属性约简改进算法;最后在高维微阵列基因表达数据集上进行对比实验,实验结果表明,本文提出的算法有较好的属性约简结果和分类准确率.

The positive region reduction algorithm based on the forward heuristic search strategy in the rough set theory does not consider how to select the candidate attributes reasonably when there are multiple conditional attributes with the biggest importance.Meanwhile,it ignores the influence of decision attributes when measuring the relevance between conditional attributes,so the resulting reduction set has weak generalization ability problem.To solve these problems,firstly,information granularity is implemented in this paper,and an optimization strategy for candidate attribute selection is put forward.Secondly,the concept of interaction information in information theory is used to eliminate redundant attributes by calculating the interaction information between attributes.On this basis,an improved attribute reduction algorithm based on information granularity and interaction information is proposed.Finally,comparative experiments are conducted on the high-dimensional microarray gene expression data sets.The experimental results show that the proposed algorithm has better attribute reduction results and classification accuracy.