高斯核模糊粗糙集中基于粒子群算法的属性约简

Attribute Reduction in Gaussian Kernel Based Fuzzy Rough Sets Based on Particle Swarm Optimization

通过高斯核模糊粗糙集模型与粒子群算法相结合的方式,利用粒子群算法收敛快、精度高等优点,可以减少获得约简所需要的时间.根据高斯核中可变参数的调整,可以实现一定程度上对约简后属性数量的控制,以确保分类的精度.实验结果表明,基于粒子群算法的高斯核模糊粗糙集属性约简算法有较好的约简性能和约简效率.

Particle swarm optimizations was applied to attribute reduction in Gaussian kernel based fuzzy rough sets.Particle swarm optimizations was a kind of evolutionary algorithm,which had characteristics of high precision and fast convergence speed.The efficiency of the attribute reduction was improved by using particle swarm optimizations to attribute reduction in Gaussian kernel based fuzzy rough sets.The number of attribute in a attribute reduction was controllable by altering parameters of Gaussian kernel based fuzzy rough sets,which aimed to improve the precision of classification.The results of experiments showed that attribute reduction method based on particle swarm optimization under Gaussian kernel based fuzzy rough sets was both effective and efficiency.