粗糙集的RBF表示与非监督模式分类

RBF REPRESENTATION AND UNSUPERVISED PATTERN CLASSIFICATION BASED ON ROUGH SETS

提出了一种粗糙集的ＲＢＦ网络表示形式，在集值测度意义下，将粗糙集的语义表达进行了有效的描述，并构造了其自适应自组织的遗传学习机制．其创新性主要表现在：（１）通过ＲＢＦ网络有效地构造了粗糙集在模式分类中的自适应表示形式；（２）在遗传算法中引入了元进化自调整机制；（３）以粗糙集意义下的非线性映射方式提高了模式分类的计算效率．图３，表１，参４．

In this paper, a kind of RBF network representation form based on rough sets is proposed.Under the meaning of set-valued measure,the semantic expression of rough sets is efficiently described and a genetic learning mechanism with the characteristics of self-adaptation and self-organization is constructed.Compared with existing technics,the originalities of the research work proposed in the paper are mainly in following three aspects:(1)An adapative representation form in pattern classification using rough sets is constructed through RBF network.(2)A self-adjustable mechanism is introduced to the genetic algorithm.(3)The computational efficency has been greatly improved through non-linear mapping under rough sets.3figs.,1tab,4refs.