Reweighted nuclear norm regularization: a SPARSEVA approach

Title: Reweighted nuclear norm regularization: a SPARSEVA approach
Creator: Ha, Huong; Welsh, James S.; Blomberg, Niclas; Rojas, Cristian R.; Wahlberg, Bo
Relation: 17th IFAC Symposium on System Identification (SYSID 2015). Proceedings of the 17th IFAC Symposium on System Identification [presented in IFAC-PapersOnLine, Vol. 28, Issue 28] (Beijing, China 19-21 October, 2015) p. 1172-1177
Publisher Link: http://dx.doi.org/10.1016/j.ifacol.2015.12.290
Publisher: Elsevier
Resource Type: conference paper
Date: 2015
Description: The aim of this paper is to develop a method to estimate high order FIR and ARX models using least squares with re-weighted nuclear norm regularization. Typically, the choice of the tuning parameter in the reweighting scheme is computationally expensive, hence we propose the use of the SPARSEVA (SPARSe Estimation based on a VAlidation criterion) framework to overcome this problem. Furthermore, we suggest the use of the prediction error criterion (PEC) to select the tuning parameter in the SPARSEVA algorithm. Numerical examples demonstrate the veracity of this method which has close ties with the traditional technique of cross validation, but using much less computations.
Subject: system identification; discrete time system; regularization; re-weighted nuclear norm; sparse estimate
Identifier: http://hdl.handle.net/1959.13/1339164
Identifier: uon:28190
Identifier: ISSN:2405-8963
Language: eng
Reviewed

		Thumbnail	File	Description	Size	Format