建造带有人工神经元网络知识系统的一种方法

AN APPROACH TO BUILD KNOWLEDGE SYSTEM WITH ARTIFICIAL NEURAL NET

本文将人工神经元网络(在本文中简称神经元网络)考虑为在知识系统中的一类知识表达模型,并且引入神经元网络块的概念作为各种表达方法的桥梁。本文的另一点在于提出了“基于实例”的神经元网络求解的解释方法,它与传统的“基于踪迹”的解释方法有效地结合在一起,来完成对推理解释的任务。

The artificial neural net is considered as a kind of knowledge representation model. The ne\v concept, Neural Net Block, will serve as a bridge between various representations. We propose an approach to explain the solutions in reasoning of neural net, i.e., 'based on exam-ple'. Combining with the traditional approach 'based on tracking', it will complete the task of the explanation in reasoning.