基于概率神经网络的相似度排序方法

A Similarity Ordering Method Based on Probability Neural Network

摘要:

为了研究相似度排序问题,对概率神经网络的结构进行了分析,对其输出进行了改进,使之输出后验概率,进而进行相似度排序。结合电火工品发火性能的预测问题,对方法进行了验证,结果表明:基于概率神经网络的相似度排序方法与实际结果相符,方法是可行的。

Abstract:

A similarity ordering method based on probability neural network was studied. The output of the probability neural network was chosen as the posterior probability based on the analysis of the structure. The posterior probability could be used as the warrant of the similarity. In order to validate the method, the predict of the firing sensitivity of the electric explosive device was put forward. The result showed that the predicted result was consistent with the measured result. The method was authentic.