[1]蒙西,乔俊飞,李文静.基于快速密度聚类的RBF神经网络设计[J].智能系统学报,2018,13(3):331-338.[doi:10.11992/tis.201702014]
　MENG Xi,QIAO Junfei,LI Wenjing.Construction of RBF neural networks via fast density clustering[J].CAAI Transactions on Intelligent Systems,2018,13(3):331-338.[doi:10.11992/tis.201702014]
点击复制

基于快速密度聚类的RBF神经网络设计

《智能系统学报》[ISSN 1673-4785/CN 23-1538/TP] 卷: 13 期数: 2018年第3期 页码: 331-338 栏目: 学术论文—人工智能基础 出版日期: 2018-05-05
Title:
Construction of RBF neural networks via fast density clustering
作者:
蒙西1,2, 乔俊飞1,2, 李文静1,2
1. 北京工业大学 信息学部, 北京 100124;
2. 北京工业大学 计算智能与智能系统北京市重点实验室, 北京 100124
Author(s):
MENG Xi1,2, QIAO Junfei1,2, LI Wenjing1,2
1. Faculty of Information Technology, Beijing University of Technology, Beijing 100124, China;
2. Beijing Key Laboratory of Computational Intelligence and Intelligent System, Beijing University of Technology, Beijing 100124, China
关键词:
RBF神经网络; 快速密度聚类; 结构设计; 神经元活性; 二阶算法; 泛化能力; 函数逼近; 系统辨识
Keywords:
RBF neural networks; fast density clustering; structure design; neuron activity; second-order training; generalization performance; function approximation; system identification
分类号:
TP273
DOI:
10.11992/tis.201702014
摘要:
针对径向基函数（radial basis function，RBF）神经网络隐含层结构难以确定的问题，提出一种基于快速密度聚类的网络结构设计算法。该算法将快速密度聚类算法良好的聚类特性用于RBF神经网络结构设计中，通过寻找密度最大的点并将其作为隐含层神经元，进而确定隐含层神经元个数和初始参数；同时，引入高斯函数的特性，保证了每个隐含层神经元的活性；最后，用一种改进的二阶算法对神经网络进行训练，提高了神经网络的收敛速度和泛化能力。利用典型非线性函数逼近和非线性动态系统辨识实验进行仿真验证，结果表明，基于快速密度聚类设计的RBF神经网络具有紧凑的网络结构、快速的学习能力和良好的泛化能力。
Abstract:
To design a hidden layer structure in radial-basis-function (RBF) neural networks, a novel algorithm based on fast density clustering is proposed. The algorithm searches for the point with the highest density and then uses it as the neuron of the hidden layer, thereby ascertaining the number of neurons in the hidden layer and the initial parameters. Moreover, the activity of each hidden neuron is ensured by introducing the Gaussian function. An improved second-order algorithm is used to train the designed network, increasing the training speed and improving the generalization performance. In addition, two benchmark simulations-the typical nonlinear function approximation and the nonlinear dynamic system identification experiment -are used to test the effectiveness of the proposed RBF neural network. The results suggest that the proposed RBF neural network based on fast density clustering offers improved generalization performance, has a compact structure, and requires shorter training time.
参考文献/References:
[1] CHEN Sheng, WOLFGANG A, HARRIS C J, et al. Symmetric RBF classifier for nonlinear detection in multiple-antenna-aided systems[J]. IEEE transactions on neural networks, 2008, 19(5):737-745.
[2] 乔俊飞, 韩红桂. RBF神经网络的结构动态优化设计[J]. 自动化学报, 2010, 36(6):865-872. QIAO Junfei, HAN Honggui. Optimal structure design for RBFNN structure[J]. Acta Automatica Sinica, 2010, 36(6):865-872.
[3] BARANDIARAN X, MORENO A. On the nature of neural information:a critique of the received view 50 years later[J]. Neurocomputing, 2008, 71(4/5/6):681-692.
[4] 侯远杭, 黄胜, 梁霄. PSO训练的弹性RBFNN在船型优化中的应用研究[J]. 哈尔滨工程大学学报, 2017, 38(2):175-180. HOU Yuanhang, HUANG Sheng, LIANG Xiao. Ship hull optimization based on PSO training FRBF neural network[J]. Journal of Harbin engineering university, 2017, 38(2):175-180.
[5] 蒙西, 乔俊飞, 韩红桂. 基于ART的RBF网络结构设计[J]. 控制与决策, 2014, 29(10):1876-1880. MENG Xi, QIAO Junfei, HAN Honggui. RBF neural network based on ART neural network[J]. Control and decision, 2014, 29(10):1876-1880.
[6] CECATI C, KOLBUSZ J, Ró?YCKI P, et al. A novel RBF training algorithm for short-term electric load forecasting and comparative studies[J]. IEEE transactions on industrial electronics, 2015, 62(10):6519-6529.
[7] 乔俊飞, 韩红桂. 神经网络结构动态优化设计的分析与展望[J]. 控制理论与应用, 2010, 27(3):350-357. QIAO Junfei, HAN Honggui. Dynamic optimization structure design for neural networks:review and perspective[J]. Control theory & applications, 2010, 27(3):350-357.
[8] TAGLIAFERRI R, STAIANO A, SCALA D. A supervised fuzzy clustering for radial basis function neural networks training[C]//Proceedings of the 9th IFSA World Congress and 20th NAFIPS International Conference. Vancouver, BC, Canada, 2001:1804-1809.
[9] RUBIO J J, PACHECO J. An stable online clustering fuzzy neural network for nonlinear system identification[J]. Neural computing and applications, 2009, 18(6):633-641.
[10] WANG Di, ZENG Xiaojun, KEANE J A. A clustering algorithm for radial basis function neural network initialization[J]. Neurocomputing, 2012, 77(1):144-155.
[11] PLATT J. A Resource-allocating network for function interpolation[J]. Neural computation, 1991, 3(2):213-225.
[12] LU Yingwei, SUNDARARAJAN N, SARATCHANDRAN P. A sequential learning scheme for function approximation using minimal radial basis function neural networks[J]. Neural computation, 1997, 9(2):461-478.
[13] HUANG Guangbin, SARATCHANDRAN P, SUNDARARAJAN N. An efficient sequential learning algorithm for growing and pruning RBF (GAP-RBF) networks[J]. IEEE transactions on systems, man, and cybernetics, part B (Cybernetics), 2004, 34(6):2284-2292.
[14] HAN Honggui, CHEN Qili, QIAO Junfei. Research on an online self-organizing radial basis function neural network[J]. Neural computing and applications, 2010, 19(5):667-676.
[15] YU Hao, REINER P D, XIE Tiantian, et al. An incremental design of radial basis function networks[J]. IEEE transactions on neural networks and learning systems, 2014, 25(10):1793-1803.
[16] RODRIGUEZ A, LAIO A. Clustering by fast search and find of density peaks[J]. Science, 2014, 344(6191):1492-1496.
[17] HAGAN M T, MENHAJ M B. Training feedforward networks with the Marquardt algorithm[J]. IEEE transactions on neural networks, 1994, 5(6):989-993.
[18] WILAMOWSKI B M, YU Hao. Improved computation for Levenberg-Marquardt training[J]. IEEE transactions on neural networks, 2010, 21(6):930-937.
[19] WU Shiqian, ER M J, GAO Yang. A fast approach for automatic generation of fuzzy rules by generalized dynamic fuzzy neural networks[J]. IEEE transactions on fuzzy systems, 2002, 9(4):578-594.
[20] MENG Xi, QIAO Junfei, HAN Honggui. An ART-like algorithm for constructing RBF neural networks[C]//Proceedings of 2015 International Joint Conference on Neural Networks. Killarney, Ireland, 2015.
相似文献/References:
[1]章? 钱,李士勇.一种新型自适应RBF神经网络滑模制导律[J].智能系统学报,2009,4(4):339.
　ZHANG Qian,LI Shi-yong.A new adaptive RBFNN sliding mode guidance law for intercepting maneuvering targets[J].CAAI Transactions on Intelligent Systems,2009,4():339.
[2]张秀玲,陈丽杰,逄宗朋,等.RBF神经网络的板形预测控制[J].智能系统学报,2010,5(1):70.
　ZHANG Xiu-ling,CHEN Li-jie,PANG Zong-peng,et al.A predictive system for process control of flatness in rolling mills using a radial basis function network[J].CAAI Transactions on Intelligent Systems,2010,5():70.
[3]陈亮,何为,韩力群.RBF神经网络的行车路径代价函数建模[J].智能系统学报,2011,6(5):424.
　CHEN Liang,HE Wei,HAN Liqun.Radial basis function neural network modeling of the traffic path cost function[J].CAAI Transactions on Intelligent Systems,2011,6():424.
[4]陈亮,何为,韩力群.城市交通最优路径算法[J].智能系统学报,2012,7(2):167.
　CHEN Liang,HE Wei,HAN Liqun.Study on an urban transportation optimal path algorithm[J].CAAI Transactions on Intelligent Systems,2012,7():167.
[5]乔俊飞,安茹,韩红桂.基于相对贡献指标的自组织RBF神经网络的设计[J].智能系统学报,2018,13(2):159.[doi:10.11992/tis.201608009]
　QIAO Junfei,AN Ru,HAN Honggui.Design of self-organizing RBF neural network based on relative contribution index[J].CAAI Transactions on Intelligent Systems,2018,13():159.[doi:10.11992/tis.201608009]

备注/Memo

收稿日期:2017-02-24。
基金项目:国家自然科学基金项目（61533002,61603009）；北京市自然科学基金面上项目（4182007）；北京工业大学日新人才项目（2017-RX-（1）-04）.
作者简介:蒙西,女,1988年生,博士研究生,主要研究方向为人工神经网络、类脑智能模型以及智能信息处理。获得授权国家发明专利1项。发表学术论文5篇,被SCI收录2篇,EI收录3篇;乔俊飞,男,1968年生,教授,博士生导师,主要研究方向为计算智能、智能特征建模、自组织控制和智能优化。在Automatica、IEEE Trans.刊物、自动化学报等权威期刊上发表学术论文百余篇;李文静,女,1985年生,副教授,博士,主要研究方向为神经计算、人工神经网络、模式识别。申请美国发明专利1项。发表学术论文10余篇,被SCI收录8篇。
通讯作者:乔俊飞.E-mail:junfeiq@bjut.edu.cn.

更新日期/Last Update: 2018-06-25
Copyright © 《 智能系统学报》 编辑部
地址：（150001）黑龙江省哈尔滨市南岗区南通大街145-1号楼 电话：0451- 82534001、82518134　邮箱：tis@vip.sina.com