在引入智能优化算法的径向基神经网络训练算法中，智能优化算法的控制参数对该算法的学习性能影响很大.为此，提出了一种基于单形进化的径向基神经网络训练算法.该算法基于单形邻域的全随机搜索方法减少算法控制参数，借助群体的多角色态保持粒子的多样性，避免算法陷入局部极值点.仿真结果表明：相比于其他算法，该算法训练的径向基神经网络不仅有效提高了识别率，而且减少了控制参数对学习性能的影响，提高了算法的普适性与鲁棒性.
The control parameters of intelligent optimization algorithm have a great influence on the learning performance of intelligent optimization algorithm. In order to solve this problem, a radial neural network training algorithm based on simplex evolution is proposed. It uses a one-dimensional neighbor based full random search method to reduce the number of control parameters, maintain the particle diversity through the group multicolor state, and avoid the algorithm falling into the local extremum point. Simulation results show that the algorithm not only improves the recognition rate but also reduces the influence of control parameters on learning performance. The generalization and robustness of the algorithm are improved.
[1] 张青贵. 人工神经网络导论[M]. 北京:中国水利水电出版社,2004.
[2] Ham F M, Kostanic I. Principles of neurocomputing for science and engineering[M].[S.l.]:McGraw-Hill, 2003:396-407.
[3] Yu F, Xu X. A short-term load forecasting model of natural gas based on optimized genetic algorithm and improved BP neural network[J]. Applied Energy, 2014, 134:102-113.
[4] Xu Y, You T, Du C. An integrated micromechanical model and BP neural network for predicting elastic modulus of 3D multi-phase and multi-layer braided composite[J]. Composite Structures, 2015, 122:308-315.
[5] Cheng C, Cheng X, Dai N, Jiang X, Sun Y, Li W. Prediction of facial deformation after complete denture prosthesis using BP neural network[J]. Computers in Biology & Medicine, 2015, 66:103-112.
[6] Zhou F, Peng H, Qin Y, Zeng X, Xie W, Wu J. RBF-ARX model-based MPC strategies with application to a water tank system[J]. Journal of Process Control, 2015, 34:97-116.
[7] Qin Y, Peng H, Zhou F, Zeng X, Wu J. Nonlinear modeling and control approach to magnetic levitation ball system using functional weight RBF network-based state-dependent ARX model[J]. Journal of the Franklin Institute, 2015, 352(10):4309-4338.
[8] Herrera L J, Pomares H, Rojas I, Guillén A, Rubio G, Urquiza J. Global and local modelling in RBF networks[J]. Neurocomputing, 2011, 74(16):2594-2602.
[9] 张德丰. MATLABB2016a智能计算25个案例分析[M]. 北京:清华大学出版社,2017.
[10] Quan H, Shi X. A surface-simplex swarm evolution algorithm[J]. Wuhan University Journal of Natural Science,2017, 22(1):38-50.
[11] 杉山将. 图解机器学习[M]. 北京:人民邮电出版社,2015.
[12] 双小川,张克. 基于统计和结构特征的手写数字识别研究[J]. 计算机工程与设计,2012, 33(4):1533-1537. Shuang X C, Zhang K. Handwritten numeral recognition based on statistical and structural characteristics[J]. Computer Engineering and Design, 2012, 33(4):1533-1537. (in Chinese)
[13] 黄同成,丁友东. 基于多小波神经网络簇轮廓伸展的手写体数字识别[J]. 中南大学学报(自然科学版),2006, 37(2):356-360. Huang T C, Ding Y D. Handwritten numeral recognition based on multiwavelet neural network cluster contour stretching[J]. Journal of Central South University (Natural Science Edition), 2006, 37(2):356-360. (in Chinese)
[14] 许兴阳,刘宏志. 基于量子门组的卷积神经网络设计与实现[J]. 计算机工程与应用,2018, 54(20):59-66. Xu X Y, Liu H Z. Design and implementation of convolution neural network based on quantum portal group[J]. Computer Engineering and Applications, 2018, 54(20):59-66. (in Chinese)
[15] 郝晓丽,张靖. 基于改进自适应聚类算法的RBF神经网络分类器设计与实现[J]. 计算机科学, 2014, 41(6):260-263. Hao X L, Zhang J. Design and implementation of RBF neural network classifier based on improved adaptive clustering algorithm[J]. Computer Science, 2014, 41(6):260-263. (in Chinese)
[16] 陈岩,李洋洋,余乐,王瑶,吴超,李阳光. 基于卷积神经网络的手写体数字识别系统[J]. 微电子学与计算机,2018(2):71-74. Chen Y, Li Y Y, Yu L, Wang Y, Wu C, Li Y G. Handwritten numeral recognition system based on convolution neural network[J]. Microelectronics and Computers, 2018(2):71-74. (in Chinese)
