神经网络用于化学知识的表达与处理

本文采用误差反传播算法,探讨了 Minsky 神经网络用于化学知识表达与处理的可行性。通过几个实例的研究表明:用神经网络方法能较好地处理一些多因子,多类和非线性问题,这种方法将知识和推理结合在一起并具有自学习能力。在化学专家系统的研究中,这将是一种有发展前途的方法。

The Application of Neural Net Algorithm in the Represatation of Chemical Knowledge

By means of back-propagation training algorithm,the Minsky neural net has been used to represent the chemical knowledge.In this work several examples have been studied and the result demonstrated that neural net can be used to solve some multi-class, multi-factor and nonlinear problems.By this method the knowledge and infernce engine are combined and the system has self-study ability.It will be a promising method in the study of chemical expert system.