基于单形进化的多项式神经网络训练算法

Training Algorithm of Polynomial Neural Network Based on Simplex Evolution

在多项式神经网络训练算法中,当采用智能优化算法进行学习优化时,智能优化算法的控制参数对学习效果有很大影响,针对这一问题,本文提出了一种多项式神经网络的智能优化方法,其的控制参数是通过对单形领域的完全随机搜索来使用的,且粒子的多样性是通过种群的多特征状态来维持的,以避免算法陷入局部最优解,试验结果表明,该算法训练的神经网络不仅能有效提高识别率,而且能减小控制参数对学习性能的影响,提高算法的整体鲁棒性。

In the polynomial neural network training algorithm,when the intelligent optimization algorithm is used for learning optimization,the control parameters of the intelligent optimization algorithm have a great impact on the learning effect.In order to solve this problem,this paper proposes an intelligent optimization method of polynomial neural network,whose control parameters are used by completely random search in the simplex field,The experimental results show that the neural network trained by the algorithm can not only effectively improve the recognition rate,but also reduce the influence of control parameters on the learning performance,and improve the overall robustness of the algorithm.