Probabilistic kernel least mean squares algorithms

Il Memming Park; Sohan Seth; Steven Van Vaerenbergh

doi:10.1109/ICASSP.2014.6855214

Probabilistic kernel least mean squares algorithms

Il Memming Park, Sohan Seth, Steven Van Vaerenbergh

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

10 Citations (Scopus)

Abstract

The kernel least mean squares (KLMS) algorithm is a computationally efficient nonlinear adaptive filtering method that 'kernelizes' the celebrated (linear) least mean squares algorithm. We demonstrate that the least mean squares algorithm is closely related to the Kalman filtering, and thus, the KLMS can be interpreted as an approximate Bayesian filtering method. This allows us to systematically develop extensions of the KLMS by modifying the underlying state-space and observation models. The resulting extensions introduce many desirable properties such as 'forgetting', and the ability to learn from discrete data, while retaining the computational simplicity and time complexity of the original algorithm.

Original language	English
Title of host publication	2014 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2014
Publisher	IEEE
Pages	8272-8276
Number of pages	5
ISBN (Electronic)	978-1-4799-2893-4
ISBN (Print)	9781479928927
DOIs	https://doi.org/10.1109/ICASSP.2014.6855214
Publication status	Published - 2014
MoE publication type	A4 Conference publication
Event	IEEE International Conference on Acoustics, Speech, and Signal Processing - Florence, Italy Duration: 4 May 2014 → 9 May 2014 Conference number: 39

Publication series

Name	IEEE International Conference on Acoustics, Speech and Signal Processing
Publisher	IEEE
ISSN (Print)	1520-6149
ISSN (Electronic)	2379-190X

Conference

Conference	IEEE International Conference on Acoustics, Speech, and Signal Processing
Abbreviated title	ICASSP
Country/Territory	Italy
City	Florence
Period	04/05/2014 → 09/05/2014

Keywords

kernel adaptive filtering
KLMS
sequential Bayesian learning
state-space model

Access to Document

10.1109/ICASSP.2014.6855214

Cite this

@inproceedings{7331a1745e9b462380d00ca3b4c4d476,

title = "Probabilistic kernel least mean squares algorithms",

abstract = "The kernel least mean squares (KLMS) algorithm is a computationally efficient nonlinear adaptive filtering method that 'kernelizes' the celebrated (linear) least mean squares algorithm. We demonstrate that the least mean squares algorithm is closely related to the Kalman filtering, and thus, the KLMS can be interpreted as an approximate Bayesian filtering method. This allows us to systematically develop extensions of the KLMS by modifying the underlying state-space and observation models. The resulting extensions introduce many desirable properties such as 'forgetting', and the ability to learn from discrete data, while retaining the computational simplicity and time complexity of the original algorithm.",

keywords = "kernel adaptive filtering, KLMS, sequential Bayesian learning, state-space model",

author = "Park, {Il Memming} and Sohan Seth and {Van Vaerenbergh}, Steven",

year = "2014",

doi = "10.1109/ICASSP.2014.6855214",

language = "English",

isbn = "9781479928927",

series = " IEEE International Conference on Acoustics, Speech and Signal Processing",

publisher = "IEEE",

pages = "8272--8276",

booktitle = "2014 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2014",

address = "United States",

note = "IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP ; Conference date: 04-05-2014 Through 09-05-2014",

}

Park, IM, Seth, S & Van Vaerenbergh, S 2014, Probabilistic kernel least mean squares algorithms. in 2014 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2014., 6855214, IEEE International Conference on Acoustics, Speech and Signal Processing, IEEE, pp. 8272-8276, IEEE International Conference on Acoustics, Speech, and Signal Processing, Florence, Italy, 04/05/2014. https://doi.org/10.1109/ICASSP.2014.6855214

Probabilistic kernel least mean squares algorithms. / Park, Il Memming; Seth, Sohan; Van Vaerenbergh, Steven.
2014 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2014. IEEE, 2014. p. 8272-8276 6855214 ( IEEE International Conference on Acoustics, Speech and Signal Processing).

Research output: Chapter in Book/Report/Conference proceeding › Conference article in proceedings › Scientific › peer-review

TY - GEN

T1 - Probabilistic kernel least mean squares algorithms

AU - Park, Il Memming

AU - Seth, Sohan

AU - Van Vaerenbergh, Steven

N1 - Conference code: 39

PY - 2014

Y1 - 2014

N2 - The kernel least mean squares (KLMS) algorithm is a computationally efficient nonlinear adaptive filtering method that 'kernelizes' the celebrated (linear) least mean squares algorithm. We demonstrate that the least mean squares algorithm is closely related to the Kalman filtering, and thus, the KLMS can be interpreted as an approximate Bayesian filtering method. This allows us to systematically develop extensions of the KLMS by modifying the underlying state-space and observation models. The resulting extensions introduce many desirable properties such as 'forgetting', and the ability to learn from discrete data, while retaining the computational simplicity and time complexity of the original algorithm.

AB - The kernel least mean squares (KLMS) algorithm is a computationally efficient nonlinear adaptive filtering method that 'kernelizes' the celebrated (linear) least mean squares algorithm. We demonstrate that the least mean squares algorithm is closely related to the Kalman filtering, and thus, the KLMS can be interpreted as an approximate Bayesian filtering method. This allows us to systematically develop extensions of the KLMS by modifying the underlying state-space and observation models. The resulting extensions introduce many desirable properties such as 'forgetting', and the ability to learn from discrete data, while retaining the computational simplicity and time complexity of the original algorithm.

KW - kernel adaptive filtering

KW - KLMS

KW - sequential Bayesian learning

KW - state-space model

UR - http://www.scopus.com/inward/record.url?scp=84905235142&partnerID=8YFLogxK

U2 - 10.1109/ICASSP.2014.6855214

DO - 10.1109/ICASSP.2014.6855214

M3 - Conference article in proceedings

AN - SCOPUS:84905235142

SN - 9781479928927

T3 - IEEE International Conference on Acoustics, Speech and Signal Processing

SP - 8272

EP - 8276

BT - 2014 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2014

PB - IEEE

T2 - IEEE International Conference on Acoustics, Speech, and Signal Processing

Y2 - 4 May 2014 through 9 May 2014

ER -

Probabilistic kernel least mean squares algorithms

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this