神经网络具有的高度并行运算能力和非线性动力学特征构成了它的应用基础。利用神经网络的非线性动力学特征(如一个多层感知器)可以完成任意的非线性变换,可获得比传统的线性处理更优的性能;利用神经网络的高度并行运算能力可以较好地解决处理中算法运算量和处理性能之间的矛盾,使得实时实现最优处理成为可能。 Neural network with its high degree of parallel computing power and nonlinear dynamics characteristics constitute its application. The nonlinear dynamics of neural network (such as a multi-layer perceptron) can be used to perform arbitrary nonlinear transformation, which can get better performance than the traditional linear processing. Using the high degree of parallel computing power of neural network can be a better solution The contradiction between the computational complexity of the algorithm and the processing performance makes it possible to achieve optimal processing in real time.