利用前馈人工神经网络边界判决实现分类

Classification Using Boundary Judgment of Feed-forward Artificial Neural Networks

基于人工神经网络理论，提出了一种新型的利用边界判决实现分类的方法．该方法不需要被判边界的显式数学模型，而是通过对所提供的样本实例数据进行学习训练，提取出无法用数学模型表达的实际分类规则．并将其用于二维平面的边界划分中，仿真结果表明了该方法的有效性．

Based on the theory of Artificral Neural Networks, a new method, in which boundary judgment is used to realize classification, is presente. In this method, no obvious mathematical model is needed. Trained with the given sample data, the actual classification rules, Which even can't be expressed with mathematical model, can be obtained. The results of simulation in two-dimension spacc have showed the high validity of this method.