论文部分内容阅读
为提高多聚合过程神经网络的逼近能力和计算效率,提出一种基于数值积分的训练算法.首先将输入函数和权函数离散化,然后采用复合梯形积分或复合辛普森积分直接处理输入函数和权函数乘积的多重积分运算,采用Levenberg-Marquardt算法调整网络参数.仿真实验表明,该方法的逼近能力和计算效率比传统的正交基展开方法有明显提高,从而揭示出该方法是提高多聚合过程神经网络逼近能力和计算效率的有效途径.
In order to improve the approximation ability and computational efficiency of the multi-process neural network, a training algorithm based on numerical integration is proposed. First, the input function and the weight function are discretized, and then the input and weight functions are processed directly by composite trapezoidal integration or compound Simpson’s integral And the Levenberg-Marquardt algorithm is used to adjust the network parameters.The simulation results show that the approximation ability and computational efficiency of this method are obviously improved compared with the traditional orthogonal base expansion method, which reveals that this method is to improve the multi-aggregation process of nerve An Effective Approach to Network Approximation Capability and Computing Efficiency.