电子学报 ›› 2006, Vol. 34 ›› Issue (10): 1838-1841.

• 论文 • 上一篇    下一篇

关于连续过程神经元网络的一些理论问题

许少华1,2, 何新贵1, 刘 坤1, 王 兵2   

  1. 1. 北京大学信息科学技术学院,北京 100871;2. 大庆石油学院计算机与信息技术学院,黑龙江大庆 163318
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2006-10-25 发布日期:2006-10-25

Some Theoretical Issues on Continuous Process Neural Networks

XU Shao-hua1,2, HE Xin-gui1, LIU Kun1, WANG Bing2   

  1. 1. School of Electronics Engineering and Computer Science,Peking University,Beijing 100871,China;2. College of Computer and Information Technology,Daqing Petroleum,China Daqing,Heilongjiang 163318,China
  • Received:1900-01-01 Revised:1900-01-01 Online:2006-10-25 Published:2006-10-25

摘要: 针对输入/输出均为连续时间函数的非线性系统信号处理和建模问题,提出了一种连续过程神经元和过程神经元网络模型.连续过程神经元的输入/输出均为连续时间函数,其时空聚合运算能同时反映连续时变输入信号的空间聚合作用和输入过程中的时间累积效应,可实现输入/输出之间非线性实时或若干时间单元延迟的映射关系.文中给出了一种输入输出均为连续时间函数的前馈过程神经元网络模型,并证明了相应的连续性,函数逼近能力和计算能力等性质定理.

关键词: 时变输入输出系统, 连续过程神经元网络, 函数逼近能力, 计算能力, 连续性

Abstract: Aim at the problems that the inputs and outputs of some practical nonlinear systems are Continuous time signals,we brought forward a Continuous process neuron and process neural networks model.The input and output of the defined process neuron are Continuous time functions,and the space-time aggregation operation can reflect the space aggregation of the input signals and the time cumulative effect in the process of input at the same time,and can also realize the nonlinear real-time mapping between the input and output.A Continuous feedforward process neural networks model is given in this paper,and the corresponding property theorems are also proved,including continuity,function approximation ability and computational capacity.

Key words: system with time-varied input and output, continuous process neural networks, function approximation ability, computational capacity, continuity

中图分类号: