一种用ANN实现带权不确定性推理的方法

An approach to weighted uncertain reasoning based on artificial neural network

介绍了基于可信度的带权不确定性推理.阐述了用神经网络(ANN)如何实现基于可信度的带权不确定性推理.分析了这种神经网络的知识表示、结构及其不确定性推理方法.说明了如何利用这种神经网络并行地实现多规则的基于可信度的带权不确定性推理,以及如何解决冲突消解问题.采用这种神经网络结构的系统不仅能实现基于可信度的带权不确定性推理,而且具有自学习、并行推理、冲突消解、抗缺省等能力.采用的神经网络结构突破了传统神经网络结构的设计思想,对于扩展神经网络的应用具有参考价值.

An approach to weighted reasoning under uncertainty with certainty factor （WRUCF） based on artificial neural network （ANN） was proposed. The related notions and inference method about WRUCF were presented. Then, the new ANN to implement the inference of WRUCF was proposed, and the details about the ANN were interpreted, such as its structure and knowledge representation, its schema of uncertain reasoning as well as its mechanisms for parallel reasoning and conflict eliminating. A system exploiting this approach can not only implement WRUCF, but also has the capabilities of self-learning, parallel reasoning, and conflict eliminating.