A Generalization of MDL for Choosing Adaptation Mechanism and Design Parameters in Identification
1997 (English)In: Proceedings of the 11th IFAC Symposium on System Identification, 1997, Vol. 2, 487-492 p.Conference paper (Refereed)
The minimum description length (MDL) has for some decades been known to be an efficient tool for choosing model structure. We will in this contribution generalize MDL to adaptive algorithms in system and signal identification. The parameter vector in these problems can either be considered as piecewise constant using segmentation and change detection algorithms or as time-varying estimated by recursive identification algorithms. MDL is derived as a measure of code length needed to transmit or store a signal. With the generalization we can compute not only the best model structure for the signal, but also when it pays off to use recursive identification and transmit the parameter updates together with the residuals, or if it is better to segment the signal and transmit the change points and the parameters. The approach opens an auto-tuning possibility, where the design parameters in the recursive identification and change detection methods can be optimized with respect to the code length.
Place, publisher, year, edition, pages
1997. Vol. 2, 487-492 p.
Minimum description length, Model structure, Adaptive algorithms, System identification, Signal identification, Parameter, Optimization
IdentifiersURN: urn:nbn:se:liu:diva-93791ISBN: 0080425925OAI: oai:DiVA.org:liu-93791DiVA: diva2:627952
11th IFAC Symposium on System Identification, Fukuoka, Japan, July, 1997