神经网络的解释机制

Explanation Mechanism in Neural Networks

解释就是人工智能系统对用户提问给出一个清晰的、完全的和易干理解的回答,即对其行为作出合理说明、使系统具有透明性。解释的类型虽然很有。

Neural Networks(NNs) have no explicit ,declarative knowledge representation and therefore have considerable difficulties in generating explanation We think that additional structures and algorithms are needed for generating explanation in NNs. This paper discusses the necessarily of explanation in NNs,analyses the explanation methods which have been published,and proposed a case/rule-based explanation methood for NNs.