[1]钱晓东,王正欧.ART2神经网络聚类的改进研究[J].南京理工大学学报(自然科学版),2007,(01):71-75.
 QIAN Xiao-dong,WANG Zhen-ou.Improvement of Clustering of ART2 Neural Network[J].Journal of Nanjing University of Science and Technology,2007,(01):71-75.
点击复制

ART2神经网络聚类的改进研究
分享到:

《南京理工大学学报》(自然科学版)[ISSN:1005-9830/CN:32-1397/N]

卷:
期数:
2007年01期
页码:
71-75
栏目:
出版日期:
2007-02-28

文章信息/Info

Title:
Improvement of Clustering of ART2 Neural Network
作者:
钱晓东 1 王正欧 2
天津大学 1. 电气与自动化工程学院; 2. 系统工程研究所, 天津 300072
Author(s):
QIAN Xiao-dongWANG Zhen-ou
1.School of Electrical Engineering and Automation,Tianjin University, Tianjin 300072,China;2.Institute of System Engineering,Tianjin University,Tianjin 300072,China
关键词:
自适应谐振理论 神经网络 聚类 自组织特征映射
Keywords:
adaptive resonance theory neural network clustering sel-f organizing featuremap
分类号:
TP183
摘要:
为进行快速动态层次聚类,通过分析自适应谐振理论(adaptive resonance theory,ART)神经网络的快速学习、主观设置警戒参数、输出无层次结构等优缺点以及自组织特征映射(self-or-ganizing feature map,SOFM)神经网络的侧反馈、不能动态聚类、输出无层次结构等优缺点的基础上,借鉴Hebb规则的思想,针对ART2神经网络的聚类算法进行了改进研究。通过结构描述、算法分析,该算法融合了ART2和SOFM的优点,克服其不足之处,以快速学习的方式形成可带有多层层次的动态聚类结构(不同的层次代表不同粒度的聚类),此外还降低了对警戒参数主观设置的要求,对于较粗粒度的聚类不再需要重新训练神经网络。并通过仿真实验证明该算法的有效性。
Abstract:
In order to achieve dynam ic clustering w ith hierarchy structure, after analyzing the shor-t com ings and advantages of adaptive resonance theory ( ART ) neural network, such as fast study, subjectively setting vigilance parameter and output w ithout hierarchy structure; and after analyzing the shortcom ings and advantages of Sel-fOrganizing FeatureMap ( SOFM ), such as side-feedback, inability of dynam ic clustering and outputw ithout hierarchy structure, improvement of clustering a-l gorithm ofART2 neural network has been presented with the reference ofH ebb Principle. By struc- ture description and algorithm ic analysis, thismodel incorporates the advantages ofART2 and SOFM and overcom es their shortcom ings, obtains dynam ic clustering structure w ith multilayer hierarchy structure by fast study( each layer denotes a category of different granularity); thismodel also re- duces the request of setting vigilance parameter and has no demand of retraining neural network of bigger granularity. Finally the effectiveness of the algorithm is demonstrated by simulation.

参考文献/References:

[ 1] KantardzicM. Data m ining concepts, models, meth- ods and algorithms [M ]. 闪四清, 陈茵, 程娃, 译. 北京: 清华大学出版社, 2003.
[ 2] H aganM T, Demuth H B. 神经网络设计 [M ]. 戴葵, 译. 北京: 机械工业出版社, 2002.
[ 3] 王莉, 王正欧. TGSOM: 一种用于数据聚类的动态自组织映射神经网络 [ J]. 电子与信息学报, 2003, 25( 3): 313- 319.
[ 4] A lahakoon D, Halgamuge S K, Srinivasan B. Dynam- ic sel-f organizing maps w ith controlled growth for know ledge discovery [ J]. NeuralNetworks, 2000, 11 ( 3): 601- 614.
[ 5] Abhijit S P, RobertBM. 神经网络模式识别及其实现 [M ]. 徐 勇, 荆 涛, 译. 北京: 电子 工业 出版社, 1999.
[ 6] DavenportM P, T itusA. M ultilevelcategory structure in the ART2 network [ J]. Neural Networks, 2004, 15 ( 1): 145- 158.

相似文献/References:

[1]陈机林,王力,高强,等.爆破扫雷器电液伺服系统建模[J].南京理工大学学报(自然科学版),2012,36(04):645.
 CHEN Ji-lin,WANG Li,GAO Qiang,et al.Modeling of Electro-hydraulic Servo System of Explosive Sweeper Mine Device[J].Journal of Nanjing University of Science and Technology,2012,36(01):645.
[2]高强,金勇,侯远龙,等.某扫雷车扫雷犁电液伺服系统辨识与控制[J].南京理工大学学报(自然科学版),2012,36(02):238.
 GAO Qiang,JIN Yong,HOU Yuan-long,et al.Modeling and Control for Mine Sweeping Plough Electro-hydraulic Servo System of Certain Mine-clearing Vehicle[J].Journal of Nanjing University of Science and Technology,2012,36(01):238.
[3]余华,黄程韦,张潇丹,等.混合蛙跳算法神经网络及其在语音情感识别中的应用[J].南京理工大学学报(自然科学版),2011,(05):659.
 YU Hua,HUANG Cheng-wei,ZHANG Xiao-dan,et al.Shuffled Frog-leaping Algorithm Based Neural Network and Its Application in Speech Emotion Recognition[J].Journal of Nanjing University of Science and Technology,2011,(01):659.
[4]林棻,赵又群.汽车侧偏角估计方法比较[J].南京理工大学学报(自然科学版),2009,(01):122.
 LIN Fen,ZHAO You-qun.Comparison of Methods for Estimating Vehicle Side Slip Angle[J].Journal of Nanjing University of Science and Technology,2009,(01):122.
[5]李成国,牟善祥,张忠传,等.基于LTCC的Ka波段无源等效腔体分析与优化设计[J].南京理工大学学报(自然科学版),2009,(03):371.
 LI Cheng-guo,MU Shan-xiang,ZHANG Zhong-chuan.Analysis and Optimal Design of Passive Equivalent Cavity in Ka Wave Band Based on LTCC[J].Journal of Nanjing University of Science and Technology,2009,(01):371.
[6]秦华旺,戴跃伟,王执铨,等.一种基于改进神经网络的入侵容忍系统模型[J].南京理工大学学报(自然科学版),2008,(05):628.
 QIN Hua-wang,DAI Yue-wei,WANG Zhi-quan.Model of Intrusion Tolerant System Based on Improved Neural Networks[J].Journal of Nanjing University of Science and Technology,2008,(01):628.
[7]李千目,戚湧,张宏,等.IIDS的行为特征提取方法研究[J].南京理工大学学报(自然科学版),2004,(02):140.
 LI Qian-mu,QI Yong,ZHANG Hong,et al.Research on Method for Obtaining Action Character Based on IIDS[J].Journal of Nanjing University of Science and Technology,2004,(01):140.
[8]王树亮,王 东,冯 珍,等.基于小波包-神经网络故障诊断系统研究[J].南京理工大学学报(自然科学版),2004,(04):356.
 WANG Shu liang,WANG Dong,FENG Zhen,et al.Study of Fault Diagnosis System Based on Wavelet Packet-neural Network[J].Journal of Nanjing University of Science and Technology,2004,(01):356.
[9]徐晋.基于神经网络专家系统的创业企业信用等级评估研究[J].南京理工大学学报(自然科学版),2004,(06):684.
 XU Jin.Evaluation Index System of the Venture Enterprise’s Credit Level Based on Artificial Neural Network[J].Journal of Nanjing University of Science and Technology,2004,(01):684.
[10]孙怀江,余陈纲,杨静宇.ALVINN及其扩展[J].南京理工大学学报(自然科学版),2002,(02):175.
 SunHuaijiang YuChengang YangJingyu.ALVINN and Its Extensions[J].Journal of Nanjing University of Science and Technology,2002,(01):175.

备注/Memo

备注/Memo:
基金项目: 国家自然科学基金 ( 60275020)
作者简介: 钱晓东 ( 1973- ), 男, 上海人, 副教授, 博士后, 主要研究方向: 数据挖掘、文本挖掘, E-mail: youran4319 @ 163. com。
更新日期/Last Update: 2007-02-28