改进的单形进化算法及在神经网络上的应用研究

单形进化算法(Surface-Simplex Swarm Evolution Algorithm,简称SSSE)是一种新型群体智能优化算法,该算法通过建立粒子的单形邻域搜索算子和多角色态搜索机制,具有很好地收敛效果.为了对该算法的性能进行进一步分析与讨论,同时,为了强调全局搜索的应用场景并提高算法的勘探搜索能力,提出一种改进的单形进化算法(ISSSE),ISSSE对原算法的多态平衡搜索机制进行了两点改进;然后用8个标准测试函数进行性能测试,并同不同的算法比较;最后将ISSSE算法应用于径向基神经网络(RBF)的参数优化中.实验结果表明,改进的单形进化算法(ISSSE)在其性能上具有更好的勘探搜索能力,提高了算法的求解精度和收敛速度,并且能够很好应用于RBF的参数寻优,提高了RBF的分类正确率.

Improved Surface-simplex Swarm Evolution Algorithm and Its Application in Neural Networks

The Surface-Simplex Swarm Evolution Algorithm(SSSE)is a new swarm intelligence optimization algorithm.The algorithm has a good convergence effect by establishing a simplex neighborhood search operator and a multi-role search mechanism for particles.In order to further analyze the performance of the algorithm,at the same time,in order to emphasize the application scenario of global search and improve the exploration and search ability of the algorithm,an improved Surface-Simplex Swarm Evolution Algorithm(ISSSE)is proposed to improve the polymorphic equilibrium search method in two ways of the original algorithm.Then,8 standard test functions are used for simulation experiments and comparison with the original algorithm and other intelligent optimization algorithms.Finally,ISSSE is applied to the parameter optimization of RBF neural network.The experimental results show that ISSSE has better exploration and search capabilities in its performance,and it has improved the algorithm's solution accuracy and convergence speed,otherwise it can be applied to the parameter optimization of RBF well,which improves RBF classification accuracy.