Convergence Analysis of Learning Algorithms for Linear Systems with Rank-Defective Markov Parameters

摘要：

In this paper, learning algorithms are examined for partially irregular linear systems. The contraction mapping approach is used and sufficient conditions for convergence are derived based on which the learning gains are chosen. In addition, the impact of initial conditions is investigated and compared in detail. A rectifying action is shown to give a convergent learning system which allows arbitrary initial repositioning.

关键词： learning algorithms Markov parameters linear systems

作者: Mingxuan Sun

作者单位: College of Information Engineering, Zhejiang University of Technology, Hangzhou 310014, China

会议类型: 国际会议

会议名称: The Third International Workshop on Applied Matriz Theory(第三届国际矩阵分析与应用会议)

会议地点: 杭州

会议语种:英文

页码: 954-961

在线出版日期: 2009-07-09（万方平台首次上网日期，不代表论文的发表时间）