基于量子云模型演化的最小属性约简增强算法  

Minimum attribute reduction enhancing algorithm based on quantum cloud model evolution

作　　者：丁卫平[1,2,3] 王建东[1] 管致锦[2] 施佺[2] 

机构地区：[1]南京航空航天大学计算机科学与技术学院,南京210016 [2]南通大学计算机科学与技术学院,南通226019 [3]南京大学计算机软件新技术国家重点实验室,南京210093

出　　处：《东南大学学报（自然科学版）》2013年第2期290-295,共6页Journal of Southeast University：Natural Science Edition

基　　金：国家自然科学基金资助项目(61139002;61171132);计算机软件新技术国家重点实验室(南京大学)开放课题资助项目(KFKT2012B28);江苏省高校自然科学基金资助项目(12KJB520013);江苏省普通高校研究生科研创新计划资助项目(CXZZ11_0219);南通市科技计划应用研究资助项目(BK2011062)

摘　　要：为提高决策表中最小属性约简的效率、稳定性和鲁棒性,基于云模型在非规范知识定性、定量表示及其相互转换过程中的优良特征对量子进化算法进行算子设计,提出了一种基于量子云模型演化的最小属性约简增强算法(QCMEARE).该算法采用量子基因云对进化种群进行编码,基于约简属性熵权逆向云进行量子旋转门自适应调整,使其在定性知识指导下能够自适应控制属性约简空间搜索范围,并采用量子云变异和云纠缠操作算子较好地避免了在属性演化约简中易陷入局部最优和早熟收敛等问题,使算法快速搜索到全局最优属性约简集.仿真实验表明,提出的最小属性约简增强算法具有收敛速度快、约简精度高和稳定性强等优点.In order to improve the efficiency,stability and robustness of minimum attribute reduction in the decision table,the operators of quantum evolutionary algorithm are designed based on the outstanding characteristics of the cloud model on the process of transforming a qualitative concept to a set of quantitative numerical values,and a novel minimum attribute reduction enhancing algorithm based on quantum cloud model evolution（QCMEARE） is proposed.First,quantum gene cloud is used to encode the evolutionary population,and reversible cloud mode based on attribute entropy weight is designed to adaptively adjust the quantum revolving gate,so the scope of the search space can be adaptively controlled under the guidance of qualitative knowledge.Secondly,both the quantum cloud mutation and quantum cloud entanglement operators are used to avoid trapping in local optimization and converging prematurely,so as to obtain the optimization attribute reduction set.The experimental results show that the proposed algorithm can achieve high efficiency,accuracy and stability of minimum attribute reduction.

关 键 词：属性约简 逆向云模型 量子云旋转门 量子云变异与云纠缠 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]