首页 | 本学科首页   官方微博 | 高级检索  
     

竞争型径向基过程神经网络时序分类器
引用本文:葛利,印桂生. 竞争型径向基过程神经网络时序分类器[J]. 哈尔滨工程大学学报, 2012, 33(6): 741-744
作者姓名:葛利  印桂生
作者单位:1. 哈尔滨工程大学计算机科学与技术学院,黑龙江哈尔滨150001;哈尔滨商业大学计算机与信息工程学院,黑龙江哈尔滨150028
2. 哈尔滨工程大学计算机科学与技术学院,黑龙江哈尔滨,150001
基金项目:黑龙江省科技攻关计划资助项目,哈尔滨市科技创新人才研究专项基金资助项目
摘    要:针对时序分类问题,提出一种竞争型径向基过程神经网络时序分类器.给出了复合竞争过程神经元单元的定义,引入复合竞争过程神经元隐层,利用竞争型径向基过程神经网络输入为时变函数的特点,由复合竞争过程神经元单元完成对过程式输入信息的模式匹配和时空聚合运算,给出了具体学习算法,省去了输出层线性连接权的计算,简化了网络结构和训练过程,提高了网络泛化能力.最后以UCI数据集多变量时序分类问题验证了分类器的有效性.

关 键 词:时序分类器  竞争型神经网络  径向基  时空聚合运算  过程神经网络

A time series classifier based on a competitive radial basis process neural network
GE Li , YIN Guisheng. A time series classifier based on a competitive radial basis process neural network[J]. Journal of Harbin Engineering University, 2012, 33(6): 741-744
Authors:GE Li    YIN Guisheng
Affiliation:1(1.Department of Computer Science and Technology,Harbin Engineering University,Harbin 150001,China;2.School of Computer and Information Engineering,Harbin University of Commerce,Harbin 150028,China)
Abstract:In considering the time series classification problem,a time series classifier was presented based on a competitive radial basis process neural network.The definition of the compound competition process neuron unit was given.The compound competition process neuron hidden layer was added to the network.Making use of the characteristics of the time-varied input function,pattern matching and temporal aggregation operation of time-varied input information were achieved by competition process neuron units.The learning algorithm was given.The calculation of linear connection weight in the output layer were omitted,and the network structure and training process were simplified.The algorithm improves the network generalization ability.Finally,the effectiveness of the classifier was proven by a multivariable time series classification problem in the UCI data set.
Keywords:time series classifier  competitive neural network  radial basis neural network  temporal aggregation operation  process neural network
本文献已被 CNKI 万方数据 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号