journal6 ›› 2001, Vol. 22 ›› Issue (4): 8-10.

• 物理与电子 • 上一篇    下一篇

基于余弦函数的神经网络硬件实现

  

  1. ( 1.湖南工业职业技术学院电气系,湖南 长沙 410007;2. 长沙电力学院数计系,湖南 长沙 410077)
  • 出版日期:2001-12-15 发布日期:2013-01-05
  • 作者简介:袁晓红( 1966- ) ,女, 湖南省新化县人,湖南工业职业技术学院电力系讲师,主要从事软件应用研究.
  • 基金资助:

    国家自然科学基金资助项目( 19974002)

The Design Method of the Circuit on the Cosine Basis Function Neural Networks

  1. ( 1. Hunan Industry Polytechnic , Changsha 410007, Hunan China; 2. Changsha University of Electr. Power, Changsha 410077,Hunan China)
  • Online:2001-12-15 Published:2013-01-05

摘要:研究了正交余弦基函数神经网络模型和学习算法, 给出了仿真实例, 并提出了余弦函数基神经网络模型电路实现方案.仿真结果表明,该网络学习收敛速度快,可任意逼近非线性映射等优异性能. 此外,网络结构简单,便于电路实现, 一个余弦波振荡源经倍频即可实现隐层神经元电路.

关键词: 余弦函数, 神经网络, 电路实现

Abstract: The model, algorithm, emulation examples and hardwired circuit based on cosine basis function neural net-work are studied in detail. The emulation results show that the network has an execllent performance with its fast astringercy, and it may be arbitrarily imminent to nonlinear function. In addition, its circuit is easy to be designed because of the simple structure of cosine basis function neural network. All the neural cells are constructed with a single cosine oscillator and a series of frequency multiplier.

Key words: cosine basis function, neural network, circuit design method

公众号 电子书橱 超星期刊 手机浏览 在线QQ