一种改进的BP神经网络模型及其在语音识别中的应用

提出了一种改进的ＢＰ神经网络模型，即多层双并联神经元可学习的人工神经网络，给出了相应的算法。并以异或问题和对称性检测问题为题，对改进算法和传统算法的优缺点进行了比较；对改进的ＢＰ网络在孤立单字语音识别应用作了初步探讨。

An Improved BP neural network and Its Application in Speech Recognition

n improved BP neural network, that is mutilayer double link artificial neural network whose nodes have the abilities of study, is proposed, and a correspondent algorithm is given. For examples of XOR problem and symmetry test, a functional comparison is made between the improved algorithm and the conventio nal algorithm. A primitive test of isolated word speech recognition based on the improved BP neural network is made in this paper.