论文部分内容阅读
提出了神经网络复合推理的思想，阐述了神经网络数值推理与传统符号逻辑推理的差异，提出了神经网络前置处理器和神经网络后置处理器的设计方法，建立了复合推理的系统结构，结合大型工程课题进行了验证．实践检验证明，该方法对解决非结构化、半结构化、结构化问题效果显著
The idea of neural network compound reasoning is put forward. The difference between neural network numerical reasoning and traditional symbolic logical reasoning is expatiated. The design methods of neural network preprocessor and neural network postprocessor are proposed, and the system structure of compound reasoning is established. Combined with large-scale engineering project has been verified. Practice shows that this method has a significant effect on solving unstructured, semi-structured and structured problems