论文部分内容阅读
神经-符号系统建立是为了将符号逻辑系统和联结主义系统进行统一,以实现符号运算和数值计算之间的对应和转化,因此构造出相应的联结主义系统以适应符号逻辑系统的推理规则具有十分重要的意义。本文以联结主义系统中的核方法(Core Method)为基础,定义了一组网络结构转化规则,通过神经网络自适应结构调整的方法来使核方法满足命题逻辑中归结原理这一经典的符号推理方法。改进的核方法在不仅在数值上与命题逻辑具有对应关系而且在网络结构上也有相似的规则。文章最后提出了关于核方法的神经网络中输入层和输出层的数值迭代规律的猜想,希望能够通过观察变量指派在实数空间的收敛情况来发现该网络对应在符号逻辑系统中的规律。?
The neuro-symbology is established to unify the symbolic logic system and the nexus system so as to realize the correspondence and transformation between symbolic computation and numerical computation. Therefore, the reasoning rules for constructing the corresponding symbiotic system to adapt to the symbolic logic system are very Significance. Based on the core method of the nexus system, this paper defines a set of rules for the transformation of network structure, and adapts the kernel method to the classical symbolic reasoning of the principle of propositional logic through the adaptive structural adjustment of neural networks method. The improved kernel method not only has the corresponding relationship with propositional logic but also has similar rules in network structure. In the end, the conjecture of numerical iteration of input layer and output layer in neural network of nuclear method is proposed. We hope that we can find out the law of the corresponding network in symbolic logic system by observing the convergence of variable assignment in real space. ?