Strong Consistency of Variable Selection for Stationary Linear Stochastic Systems

被引:0
|
作者
Zhao Wenxiao [1 ,2 ]
Yin, G. George [3 ]
Bai Er-Wei [4 ]
机构
[1] Chinese Acad Sci, Acad Math & Syst Sci, Key Lab Syst & Control, Beijing 100190, Peoples R China
[2] Univ Chinese Acad Sci, Sch Math Sci, Beijing 100049, Peoples R China
[3] Wayne State Univ, Dept Math, Detroit, MI 48202 USA
[4] Univ Iowa, Dept Elect & Comp Engn, Iowa City, IA 52242 USA
来源
PROCEEDINGS OF THE 38TH CHINESE CONTROL CONFERENCE (CCC) | 2019年
基金
美国国家科学基金会;
关键词
Variable Selection; Linear Stochastic System; Strong Consistency; IDENTIFICATION; LASSO; ORDER;
D O I
10.23919/chicc.2019.8866312
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
In this paper, we consider the variable selection for linear stochastic systems. A modified LASSO-type estimator is introduced. Then based on the classical persistent excitation (PE) for systems identification, the strong consistency of the estimates is established. i.e.. the zero elements in the unknown parameter vector being correctly identified and estimates for the nonzero elements in the unknown parameter vector converging to the true values with probability one. Compared with the existing results on similar topics, the strong consistency of estimates is established while in existing literature only the convergence in probability was obtained. For this, new theoretical analysis method is adopted in this paper.
引用
收藏
页码:1719 / 1723
页数:5
相关论文
共 50 条