Normal Fuzzy neural network and its application in script Chinese characters identification
Shaohua Xu · Jisuanji gongcheng yu sheji · 2005
In order to improve capability of script Chinese characters identification, a normal fuzzy neural network (NFNN) with five layers is proposed. Focusing on the structure optimization of network, a new node selection method and corresponding back propagation learning algorithm are presented this algorithm can sufficiently utilize fuzzy if-then rules mapped out by expert, and it also can perfect reasoning structure of network, improve identification capability of the network and reduce influence of disadvantage factor. The ex- perimentindicates that this algorithm has a better adaptability and practicability forscriptChinese characters identification. This algorithm presents a novel approach of script Chinese characters identification capability improvement.