一个自生成的神经树网络模式分类器

A SELF-GROWING NEURAL TREE NETWORK PATTERN CLASSIFIER

本文研究一种神经树网络（ＮＴＮ，ＮｅｕｒａｌＴｒｅｅＮｅｔｗｏｒｋ）模型，提出一种能自动生成其体系结构的有监督的竞争学习算法，指出该模型能解决复杂的多类模式分类问题．模拟数字例子——手写体数码识别的结果也表明该分类器是有效的．

This paper is about a neural tree network(NTN)model.A supervised competitive learning algorithm is presented to grow this architecture,which can effectly solve the complex multicategory classification.A simulation result also presented on handwrited numberic recognition shows effectiveness of this classifier.