CAJ | 학술논문

传统的属性约简算法效率低下，容易陷入局部极小值，不适用于大型知识库。文中提出一种基于粗糙集理论的遗传属性约简方法，在传统的属性约简方法基础上对适应度函数、交叉和变异的概率、变异方式和种群修复方式进行了改进。在正域区分对象集的研究基础上，用启发信息设计了一种快速的属性约简算法，并利用 Matlab 工具进行仿真，将仿真结果与前人研究结果作对比。实验表明此算法优于前人的算法，能够快速高效地对大型知识系统求其约简。
전통적속성약간산법효솔저하，용역함입국부겁소치，불괄용우대형지식고。문중제출일충기우조조집이론적유전속성약간방법，재전통적속성약간방법기출상대괄응도함수、교차화변이적개솔、변이방식화충군수복방식진행료개진。재정역구분대상집적연구기출상，용계발신식설계료일충쾌속적속성약간산법，병이용 Matlab 공구진행방진，장방진결과여전인연구결과작대비。실험표명차산법우우전인적산법，능구쾌속고효지대대형지식계통구기약간。
Traditional attribute reduction algorithm efficiency is low. It is easy to fall into local minimum value and shall not be applied to the large decision table. This paper proposes a genetic attribute reduction method based on rough set theory. Compared with traditional attribute reduction methods,it improves the fitness function the crossover probability,the mutation probability and the mutation methods. It takes advantage of heuristic in-formation in design a new efficient genetic algorithm of attribute reduction based on rough set. It makes use of Matlab tools to the simulation and compares the simulation results with predecessors' research results. The emu-late example and experiment results show that the algorithm could compute the attribute reduction of the decision table quickly and efficiently,especially in tackling a large decision table.