构造性神经网络覆盖算法的模糊化技术及其应用

The Fuzzification Technology and Application of Constructed NN’s Covering Algorithm

本文通过将模糊逻辑的结构性定义引入构造性神经网络模式分类系统中,给出模糊神经网络模式分类边界的直观描述方法.在系统信息不完备的情况下,可以获得更为合理的模式分类边界,并提高模式分类的精度.实验结果表明以上方法正确.

This paper introduces a structural definition of fuzzy set into NN＇s pattern classification system and gives the clear description of the boundary of pattern classification. When the system information is imperfect, it can get more reasonable boundary of classes and improve its precision. The above-mentioned turns out to be right in the practices.