您的位置:山东大学 -> 科技期刊社 -> 《山东大学学报(工学版)》

山东大学学报(工学版)

• 论文 • 上一篇    下一篇

基于可变数据重用因子的变步长仿射投影算法

聂聪, 吕振肃   

  1. 兰州大学信息科学与工程学院, 甘肃 兰州 730000
  • 收稿日期:2007-08-29 修回日期:1900-01-01 出版日期:2008-02-16 发布日期:2008-02-16
  • 通讯作者: 聂聪

A variable stepsize affine projection algorithm based on a variable data-reuse factor

NIE Cong, LV Zhen-su   

  1. School of Information Science and Engineering, Lanzhou University, Lanzhou 730000, China
  • Received:2007-08-29 Revised:1900-01-01 Online:2008-02-16 Published:2008-02-16
  • Contact: NIE Cong

摘要: 提出的仿射投影算法可根据输入数据矩阵条件数的变化情况,实时地决定下一步输入矩阵的维数,即数据重用因子和步长,并利用这两个动态量去折中收敛速度和稳态误差之间的矛盾.仿真结果表明,在较大程度节省了运算量的情况下,该算法的收敛速度接近于使用高维输入矩阵的情况,而稳态误差与使用低维输入矩阵时相当,同时收敛稳定性更优.

关键词: 仿射投影算法, 条件数, 数据重用因子, 变步长

Abstract:  The dimension of a next input signal matrix can be determined by the variation of the condition number between current and previous steps, which are the datareuse factor and stepsize. Then the conflict between convergence rate and stable-state error was settled by these two dynamic factors. Simulation results show that under the presupposition of considerable computation reduction, the convergence rate of this algorithm is close to the high dimension situation, and the stable-state error is nearly the same as the low dimension condition. The convergence stability is also improved. 

Key words: APA (affine projection algorithm), condition number, data reuse factor, variable step-size

中图分类号: 

  • TN911
[1] 刘建美,马帅奇. 基于BFGS公式的改进截断拟牛顿法在随机用户均衡问题上的应用[J]. 山东大学学报(工学版), 2018, 48(1): 36-41.
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
No Suggested Reading articles found!