论文部分内容阅读
由节点和约束组成的约束满足网络(Constraint satisfaction networks,简称CSNs)能够描述许多问题,在AI中具有十分重要的地位.神经网络(Neural networks,简称NNs)能够被看作CSNs,而且被认为是适合于求解约束满足问题,因此NNs中必然有一个相应的逻辑理论.本文通过对比CSNs提出了弱推理理论和矛盾分析法,以研究二值神经网络的定性逻辑行为.文中用神经元(称为判决元)表示判决,连接表示判决元之间的柔性约束并执行
Constraint satisfaction networks (CSNs), which consist of nodes and constraints, can describe many problems and play a very important role in AI. Neural networks (NNs) can be considered as CSNs and are considered to be Which is suitable for solving the problem of constraint satisfaction.Therefore, there must be a corresponding logic theory in NNs.Through the comparison of CSNs, a weak inference theory and contradiction analysis method are proposed to study the qualitative logic behavior of binary neural networks.In this paper, neurons Decision element) represents the decision, the connection represents the flexible constraints between the decision elements and execute