Reweighted nuclear norm regularization: A SPARSEVA approach
2015 (English)In: IFAC-PapersOnLine, ISSN 2405-8963, Vol. 48, no 28, 1172-1177 p.Article in journal (Refereed) Published
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.
Place, publisher, year, edition, pages
Elsevier, 2015. Vol. 48, no 28, 1172-1177 p.
discrete time system, re-weighted nuclear norm, regularization, sparse estimate, system identification, Digital control systems, Discrete time control systems, Identification (control systems), Least squares approximations, Numerical methods, Discrete - time systems, Nuclear norm regularizations, Prediction errors, Sparse estimation, Traditional techniques, Validation criteria, Parameter estimation
IdentifiersURN: urn:nbn:se:kth:diva-195475DOI: 10.1016/j.ifacol.2015.12.290ScopusID: 2-s2.0-84988478039OAI: oai:DiVA.org:kth-195475DiVA: diva2:1047835
QC 201611182016-11-182016-11-032016-11-18Bibliographically approved