摘要
属性约简是粗糙集合研究的重要内容之一。为了能够有效地获取决策表中属性最小相对约简,提出了一种基于GA-PSO的属性约简算法。该算法以条件属性对决策属性的支持度为基础,求解核属性,把所有的条件属性(除去核属性)加入粒子群算法的初始种群中,并用遗传算法对不满足适应度条件的粒子进行交叉变异操作。实验结果表明,该算法在加强局部搜索能力的同时保持了该算法全局寻优的特性,能够快速有效地获得最小相对属性集。
Attribute reduction is one of the main contents in rough set theory study. In order to achieve attribute reduction effectively,a GA-PSO based attribute reduction algorithm for rough set is proposed. According to the dependability of the decision attributes to the condition attributes, the proposed algorithm can calculate the core attributes. All the condition attributes except the core attributes are added to the initial population of the PSO （Particle Swarm Optimization） algorithm,and then the crossover and mutation operations of the genetic algorithm are performed on the particles that do not meet the fitness conditions. Experimental results show that the algorithm can enhance the local search ability as well as maintain the feature of global optimization, and calculate the minimum relative attribute set quickly and effectively.
出处
《计算机工程与科学》
CSCD
北大核心
2015年第2期397-401,共5页
Computer Engineering & Science
