ONLINE CENTERED NLMS ALGORITHM FOR CONCEPT DRIFT COMPENSATION

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F60461373%3A22340%2F21%3A43923080" target="_blank" >RIV/60461373:22340/21:43923080 - isvavai.cz</a>

Nalezeny alternativní kódy

RIV/68407700:21220/21:00354756

Výsledek na webu

<a href="http://www.nnw.cz/doi/2021/NNW.2021.31.018.pdf" target="_blank" >http://www.nnw.cz/doi/2021/NNW.2021.31.018.pdf</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.14311/NNW.2021.31.018" target="_blank" >10.14311/NNW.2021.31.018</a>

Jazyk výsledku

angličtina

Název v původním jazyce

ONLINE CENTERED NLMS ALGORITHM FOR CONCEPT DRIFT COMPENSATION

Popis výsledku v původním jazyce

This paper introduces an online centered normalized least mean squares (OC-NLMS) algorithm for linear adaptive finite impulse response (FIR) filters and neural networks. As an extension of the normalized least mean squares (NLMS), the OC-NLMS algorithm features an approach of online input centering according to the introduced filter memory. This key feature can compensate the effect of concept drift in data streams, because such a centering makes the filter independent from the nonzero mean value of signal. This approach is beneficial for applications of adaptive filtering of data with offsets. Furthermore, it can be useful for real-time applications like data stream processing where it is impossible to normalize the measured data with respect to its unknown statistical attributes. The OC-NLMS approach holds superior performance in comparison to the NLMS for data with large offsets and dynamical ranges, due to its input centering feature that deals with the nonzero mean value of the input data. In this paper, the derivation of this algorithm is presented. Several simulation results with artificial and real data are also presented and analysed to demonstrate the capability of the proposed algorithm in comparison with NLMS.

Název v anglickém jazyce

ONLINE CENTERED NLMS ALGORITHM FOR CONCEPT DRIFT COMPENSATION

Popis výsledku anglicky

This paper introduces an online centered normalized least mean squares (OC-NLMS) algorithm for linear adaptive finite impulse response (FIR) filters and neural networks. As an extension of the normalized least mean squares (NLMS), the OC-NLMS algorithm features an approach of online input centering according to the introduced filter memory. This key feature can compensate the effect of concept drift in data streams, because such a centering makes the filter independent from the nonzero mean value of signal. This approach is beneficial for applications of adaptive filtering of data with offsets. Furthermore, it can be useful for real-time applications like data stream processing where it is impossible to normalize the measured data with respect to its unknown statistical attributes. The OC-NLMS approach holds superior performance in comparison to the NLMS for data with large offsets and dynamical ranges, due to its input centering feature that deals with the nonzero mean value of the input data. In this paper, the derivation of this algorithm is presented. Several simulation results with artificial and real data are also presented and analysed to demonstrate the capability of the proposed algorithm in comparison with NLMS.

Druh

J_imp - Článek v periodiku v databázi Web of Science

CEP obor

—

OECD FORD obor

10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)

Projekt

<a href="/cs/project/EF16_019%2F0000826" target="_blank" >EF16_019/0000826: Centrum pokročilých leteckých technologií</a><br>

Návaznosti

I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Rok uplatnění

2021

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

Neural Network World

ISSN

1210-0552

e-ISSN

—

Svazek periodika

31

Číslo periodika v rámci svazku

5

Stát vydavatele periodika

CZ - Česká republika

Počet stran výsledku

13

Strana od-do

"329 "- 341

Kód UT WoS článku

000739166400002

EID výsledku v databázi Scopus

2-s2.0-85123343167