QIN Yihan, LIU Bingxiang, PENG Wen

(School of Information Engineering, Jingdezhen Ceramic Institute, Jingdezhen 333403)

Abstract:A new method combining the advantages of the rough set theory and neural network theory was presented. The attributes of the data were reduced according to the rough set theory, thereby to simplify the neural network structure; threelayer artificial neural network is applied to the classification of ceramic raw materials. The results demonstrate that pattern recognition of ceramic raw materials by the rough set and the artificial neural network agrees with the fact. This method is useful for the choice of ceramic raw materials for a batch, and is worthy of a wider application.

Keywords:rough set, attribute reduction, BP neural network, ceramic raw materials