Filter interpretation of the cost function in regularized FIR modeling

Anna Marconato; Joannes Schoukens

Filter interpretation of the cost function in regularized FIR modeling

Electricity

Research output: Chapter in Book/Report/Conference proceeding › Meeting abstract (Book)

Abstract

We consider the estimation of FIR models by means of Bayesian regularization techniques. The regularization approach allows one to obtain solutions characterized by a reduced variance, at the price of slightly increasing the bias term. This is done by embedding in the problem prior information about the underlying linear dynamic system, by designing a suitable kernel matrix. In this work, we look at the same problem from a different perspective, focusing on the cost function interpretation, rather than on the kernel deﬁnition.

Original language	English
Title of host publication	34th Benelux Meeting on Systems and Control, Center Parcs "De Vossemeren", Lommel, 24th to 26th March, 2015
Number of pages	1
Publication status	Published - 24 Mar 2015
Event	34th Benelux Meeting on Systems and Control - Vossemeren, Lommel, Belgium Duration: 24 Mar 2015 → 26 Mar 2015

Seminar

Seminar	34th Benelux Meeting on Systems and Control
Country/Territory	Belgium
City	Lommel
Period	24/03/15 → 26/03/15

Keywords

FIR models
Bayesian regularization techniques

Cite this

@inbook{a30f1c82b20747ebb561d4360c861661,

title = "Filter interpretation of the cost function in regularized FIR modeling",

abstract = "We consider the estimation of FIR models by means of Bayesian regularization techniques. The regularization approach allows one to obtain solutions characterized by a reduced variance, at the price of slightly increasing the bias term. This is done by embedding in the problem prior information about the underlying linear dynamic system, by designing a suitable kernel matrix. In this work, we look at the same problem from a different perspective, focusing on the cost function interpretation, rather than on the kernel deﬁnition.",

keywords = "FIR models , Bayesian regularization techniques",

author = "Anna Marconato and Joannes Schoukens",

year = "2015",

month = mar,

day = "24",

language = "English",

booktitle = "34th Benelux Meeting on Systems and Control, Center Parcs {"}De Vossemeren{"}, Lommel, 24th to 26th March, 2015",

note = "34th Benelux Meeting on Systems and Control ; Conference date: 24-03-2015 Through 26-03-2015",

}

TY - CHAP

T1 - Filter interpretation of the cost function in regularized FIR modeling

AU - Marconato, Anna

AU - Schoukens, Joannes

PY - 2015/3/24

Y1 - 2015/3/24

N2 - We consider the estimation of FIR models by means of Bayesian regularization techniques. The regularization approach allows one to obtain solutions characterized by a reduced variance, at the price of slightly increasing the bias term. This is done by embedding in the problem prior information about the underlying linear dynamic system, by designing a suitable kernel matrix. In this work, we look at the same problem from a different perspective, focusing on the cost function interpretation, rather than on the kernel deﬁnition.

AB - We consider the estimation of FIR models by means of Bayesian regularization techniques. The regularization approach allows one to obtain solutions characterized by a reduced variance, at the price of slightly increasing the bias term. This is done by embedding in the problem prior information about the underlying linear dynamic system, by designing a suitable kernel matrix. In this work, we look at the same problem from a different perspective, focusing on the cost function interpretation, rather than on the kernel deﬁnition.

KW - FIR models

KW - Bayesian regularization techniques

M3 - Meeting abstract (Book)

BT - 34th Benelux Meeting on Systems and Control, Center Parcs "De Vossemeren", Lommel, 24th to 26th March, 2015

T2 - 34th Benelux Meeting on Systems and Control

Y2 - 24 March 2015 through 26 March 2015

ER -

Filter interpretation of the cost function in regularized FIR modeling

Abstract

Seminar

Keywords

Fingerprint

Cite this