Title :
A filtering approach for model selection
Author :
Carvajal, Rodrigo ; Godoy, Boris I. ; Agüero, Juan C. ; Yuz, Juan I.
Author_Institution :
Centre for Complex Dynamic Syst. & Control, Univ. of Newcastle, Newcastle, NSW, Australia
Abstract :
In this paper, we propose an iterative algorithm to perform model selection. This algorithm is a sequential technique that utilizes ideas from filtering theory. We found that the value of initial parameters in our method plays a key role in selecting models. We study the effect of those parameters, and propose guidelines on how to choose them. Finally, we explore and discuss extensions of the algorithm.
Keywords :
filtering theory; identification; iterative methods; filtering approach; filtering theory; iterative algorithm; model selection; Complexity theory; Computational modeling; Data models; Kalman filters; Probabilistic logic; Probability; Tuning;
Conference_Titel :
Control and Automation (ICCA), 2011 9th IEEE International Conference on
Conference_Location :
Santiago
Print_ISBN :
978-1-4577-1475-7
DOI :
10.1109/ICCA.2011.6137993
