基于规则学习的前馈神经网络

A Neural Network Method Based on Learning Rule

提出了用规则学习方法得到的规则建立人工神经网络的方法基于规则的神经网络方法。它是一种规则学习方法和人工神经网络各自取长补短有机结合的方法。它即避免了规则学习方法的识别速度慢和神经网络训练慢的缺点，又保留了规则学习方法的训练速度快、聚类能力强和神经网络的识别速度快、可硬件实现的优点。它为大规模训练样本来建立神经网络识别系统开辟了一条新的途径。

This paper presents a method to produce a neural network by following the rules generated by rule learning algorithm a method by which either of the neural network and the rule learning method overcome the weaknesses of the other by acquiring the strong points of one. The method not only overcomes foe weakness so that the recognizing speed of rule learning algorithm is fast and the recognizing speed of neural network is fast. This method provides very effective method for training neural network with a large number of examples.