基于JADE-NRS的肺部肿瘤高维属性约简模型

High-dimensional Attribute Reduction Model of Lung Tumor Based on JADE-NRS

Pawlak粗糙集理论基于等价关系且只能处理离散数据,容错性差;差分进化算法性能依赖于控制参数(变异系数(F)、交叉系数(CR))和适应度函数的构造。针对以上问题,综合考虑属性重要度和特征规模两因素,首先,提取CT肺部肿瘤影像图像的104维特征作为肺部肿瘤患者的决策信息表;然后,基于JADE自适应差分进化算法对决策信息表进行特征选择,同时混合邻域粗糙集计算特征规模和属性重要度,依此构造适应度函数;最后,进行仿真实验,并与RS、NRS、DE-NRS模型做对比实验。实验结果表明,该模型能够有效的进行高维属性约简。

The theory of Pawlak rough set is based on the equivalence relation and can only deal with discrete data,which has poor fault tolerance;The performance of differential evolution algorithm depends on the construction of control parameters(coefficient of variation(F),cross coefficient(CR))and fitness function.Aiming at the above problems,two factors of attribute importance and feature scale are considered comprehensively.Firstly,the 104-dimensional features of CT lung tumor images are extracted as a decision information table for patients with lung tumors;Then,the decision information based on JADE adaptive differential evolution algorithm the table performs feature selection,and at the same time,the neighborhood rough set is used to calculate the feature scale and attribute importance,and the fitness function is constructed according to this;Finally,simulation experiments are performed and compared with RS,NRS,and DE-NRS models.Experimental results show that the model can effectively perform high-dimensional attribute reduction.