多层神经网络自适应控制训练算法及仿真实验

该文提出了一种用神经网络学习非线性函数时的自适应控制BP训练算法，该算法根据当前收敛误差及收敛误差变化率自适应确定学习控制值，并针对某些学习应用提出了一种两阶段自适应控制逐一训练算法。通过仿真实验证明，这种方法比传统方法收敛快，学习精度高，最后，给出了一算法的仿真实例。

A Training Algorithm with Adaptive Control for a Multilayer Neural Networks and Simulation Experiment

This paper describes a BP training algorithm with adaptive control. Using this algorithm, an adaptive learning controlling value is established according to the recent convergence error and its rate of change. The results of simulation experiments show that the learning accuracy is higher, and convergence faster than conventional means. Finally, two simulation experiments are given.