基于遗传算法的粗糙集属性约简算法  

ATTRIBUTE REDUCTION ALGORITHM FOR ROUGH SET ON GENETIC ALGORITHM BASIS

作　　者：王萍[1] 王学峰[2] 吴谷丰[1] 

机构地区：[1]广东纺织职业技术学院,广东佛山528041 [2]石河子大学新,疆石河子832003

出　　处：《计算机应用与软件》2008年第5期42-44,共3页Computer Applications and Software

基　　金：国家自然科学基金资助项目(60374023)。

摘　　要：为了研究粗糙集理论中属性约简问题,给出了一种属性相对重要度定义,证明了其合理性,并将它应用到基于遗传算法的约简算法中,提出一种启发式遗传算法。算法采用修正策略保证群体进化收敛于最小约简,同时引入属性相对重要度作为启发信息,加快算法的收敛速度。对算法进行的时间复杂度和完备性分析以及数值实验表明,基于遗传算法的粗糙集属性约简算法具有完备、快速收敛等特点。A kind of definition for attribute signification was given and its rationality was proved in order to study attribute reduction in rough set, An effective heuristic genetic algorithm was proposed by applying above definition onto the reduction algorithm based on genetic al- gorithm. The algorithm used modification strategy to guarantee the minimum reduction being achieved in population evolution convergence, meanwhile the relative attribute significance was imported as heuristic information for accelerating the convergence rate, It is shown by the analysis of its time complexity and completeness as well as the numerical experiment that the proposed algorithm can achieve the goal of completeness and rapid convergence.

关 键 词：粗糙集 属性约简 遗传算法 属性重要度 

分 类 号：TP393.4[自动化与计算机技术—计算机应用技术] TP18[自动化与计算机技术—计算机科学与技术]