摘要： 本文介绍一种改进的B-P学习算法——IBP,它能够提高B-P算法的收敛速度,同时避免B-P算法在收敛过程中可能发生的振荡、局部极小值等问题,从而它增强了B-P算法的可用性和实用性。

关键词: 单层网络, 神经网络, B-P学习算法

Abstract: This paper introduces an improved B-P algorithm, which can avoid such problems as low convergence speed, local minimization, and oscillation in the process of convergence and brings about a better application,prospect of B-P model.