论文部分内容阅读
本文针对BP算法存在收敛速度慢的缺点,提出一种基于网络动态训练误差变化率自动校正学习步长和冲量因子的自适应反向传播算法。异或问题的仿真结果表明,该方法具有较快的收敛速度。
This paper aims at the shortcomings of the slow convergence rate of BP algorithm and proposes an adaptive backpropagation algorithm that automatically adjusts the learning step and the impulse factor based on the rate of change of network dynamic training error. Simulation results of XOR problems show that this method has faster convergence speed.