In this paper, the problem of detrending a time series and/or estimating a wandering baseline is addressed. We propose a new methodology that adaptively minimizes different regularized cost functions by introducing an ARMA model of the underlying trend. Mixed ℓ1/ℓ2-norm penalty functions are taken into consideration and novel RLS and LMS solutions are derived for the model parameters estimation. The proposed methods are applied to typical trend estimation/removal problems that can be found in the analysis of economic time series or biomedical signal acquisition. Comparisons with standard noncausal filtering techniques are also presented.
Mixed ℓ 2 and ℓ 1 -norm regularization for adaptive detrending with ARMA modeling / Giarré, L.; Argenti, F.. - In: JOURNAL OF THE FRANKLIN INSTITUTE. - ISSN 0016-0032. - STAMPA. - 355:(2018), pp. 1493-1511. [10.1016/j.jfranklin.2017.12.009]
Mixed ℓ 2 and ℓ 1 -norm regularization for adaptive detrending with ARMA modeling
Argenti, F.
2018
Abstract
In this paper, the problem of detrending a time series and/or estimating a wandering baseline is addressed. We propose a new methodology that adaptively minimizes different regularized cost functions by introducing an ARMA model of the underlying trend. Mixed ℓ1/ℓ2-norm penalty functions are taken into consideration and novel RLS and LMS solutions are derived for the model parameters estimation. The proposed methods are applied to typical trend estimation/removal problems that can be found in the analysis of economic time series or biomedical signal acquisition. Comparisons with standard noncausal filtering techniques are also presented.File | Dimensione | Formato | |
---|---|---|---|
2018_JFI.pdf
Accesso chiuso
Tipologia:
Pdf editoriale (Version of record)
Licenza:
Tutti i diritti riservati
Dimensione
2.15 MB
Formato
Adobe PDF
|
2.15 MB | Adobe PDF | Richiedi una copia |
I documenti in FLORE sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.