论文部分内容阅读
本文提出了用基于规则专家系统与神经网络的集成，该系统实现了从实例中自动获取知识的功能．在产生和控制不完全情况方面提高了专家系统的推理能力．它使用无导师学习算法的神经网络来获取正规数据，并用一个符号生成器把这些正规的数据变换成规则．生成规则和训练后的神经网络作为知识库嵌于专家系统中．在诊断阶段，为了诊断不明情况，可同时使用知识库和人类专家的知识，而且系统可以利用训练过的神经网络的综合能力进行诊断，并使不相符数据完整化．
This paper proposes the integration of a rule-based expert system with a neural network, which enables the automatic acquisition of knowledge from an instance. The expert system’s reasoning ability is improved in generating and controlling incomplete cases. It uses a neural network without a tutor learning algorithm to get the normal data and transforms the regular data into rules using a symbol generator. Generating rules and training neural network as a knowledge base embedded in the expert system. In the diagnostic phase, in order to diagnose unknown conditions, both knowledge base and human expert knowledge can be used, and the system can make use of the comprehensive ability of the trained neural network to diagnose and complete the non-coincident data.