Reweighted nuclear norm regularization: A SPARSEVA approach
conference contribution
posted on 2024-11-03, 13:58authored byHuong HaHuong Ha, James Welsh, Niclas Blomberg, Cristian Rojas, Bo Wahlberg
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.
History
Start page
1172
End page
1177
Total pages
6
Outlet
Proceedings of the 17th IFAC Symposium on System Identification (SYSID 2015)