本文提出一种基于距离的神经网络分类器的设计方法。用这种方法设计的网络比采用一般BP算法的网络有较快的收敛速度。这种网络也为理解三层网络的映射能力,记忆容量及隐节点的设置等问题提供了很好的启发和帮助。 This paper presents a distance-based neural network classifier design method. Networks designed in this way have a faster convergence rate than networks using the general BP algorithm. This kind of network also provides a good inspiration and help for understanding the mapping ability of the three-layer network, memory capacity and hidden node settings.