摘要
推理作为人工智能领域中的一个重要课题，是人类求解问题的主要思维方法。人们在处理问题或人工智能领域中进行推理机设计时，往往根据问题及所处环境不同，考虑不同的推理策略以提高其效果或效率。正向推理、反向推理及双向推理为三种基本的推理策略。双向联想记忆（ＢＡＭ）神经网络是一类具有双向稳定性的反馈系统，具有简单、可靠、易于实现的特点。将双向联想记忆（ＢＡＭ）神经网络用于推理是人工智能领域中进行推理机设计的新途径。在分析了ＢＡＭ网络用于推理，尤其是双向推理的理论依据上，研究了ＢＡＭ网络在不同推理策略下的运行方法，通过实例说明了该方法的运作方式，最后就存在的网络容量等问题进行了讨论，指出了进一步研究的方向。
Reasoning is the main thinking method for people to solve problems. We usually employ different reasoning tactics with respect to different problems or environment to enhance reasoning effect and efficiecy while we solve problems or design reasoning machines in the field of artificial intelligence. There are three basic kinds of reasoning tactics, which are positive directional reasoning, negative directional reasoning and bi directional reasoning.Bi directional associative memorioes (BAM) neural network is a kind of feedback system, which has the virtue of bi directional stability, simplicity, reliability and easiness to carry out. The approach which apply bi directional associative memories neural network to the design of reasoning machine in the field of artificial intelligence is a new one. The paper analyzed the theoretic gist of reasoning in BAM neural networks, especially the theoretic gist of bi directional reasoning. Based on it, the running methods of BAM under different reasoning tactics are studied through examples. Finally, the problem of network capacity is discussed and the further investigative direction is pointed out.
出处
《桂林电子工业学院学报》
1999年第2期9-13,共5页
Journal of Guilin Institute of Electronic Technology
关键词
推理策略
双向推理
BAM神经网络
人工智能
reasoning tactic, bi directional associative memories neural networks, bi directional reasoning, network capacity
