A dictionary updating scheme incorporating words elimination into Quantized Kernel Least-Mean-Squares for changing environments

Lei Sun, Badong Chen, Shengyu Nan, Zhiping Lin, Kar Ann Toh

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

Learning under time-varying environment is a challenging task since one has to deal with the ever changing distribution of data. A common and yet effective solution is to learn the data online and keep up with any ongoing changes. The Quantized Kernel Least-Squares (QKLMS) is an effective tool for online dictionary learning where the network size is capped by the quantization dictionary size. However, due to the lack of a mechanism to eliminate outdated words, learning can become irrelevant over time. In this paper, a mechanism to remove irrelevant words in the dictionary is proposed for QKLMS. Our experimental results based on chaotic time sequence prediction validate the capability of the developed method for time-varying data adaptation.

Original languageEnglish
Title of host publication2015 IEEE International Conference on Digital Signal Processing, DSP 2015
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages911-915
Number of pages5
ISBN (Electronic)9781479980581, 9781479980581
DOIs
Publication statusPublished - 2015 Sep 9
EventIEEE International Conference on Digital Signal Processing, DSP 2015 - Singapore, Singapore
Duration: 2015 Jul 212015 Jul 24

Publication series

NameInternational Conference on Digital Signal Processing, DSP
Volume2015-September

Other

OtherIEEE International Conference on Digital Signal Processing, DSP 2015
CountrySingapore
CitySingapore
Period15/7/2115/7/24

Fingerprint

Glossaries

All Science Journal Classification (ASJC) codes

  • Signal Processing

Cite this

Sun, L., Chen, B., Nan, S., Lin, Z., & Toh, K. A. (2015). A dictionary updating scheme incorporating words elimination into Quantized Kernel Least-Mean-Squares for changing environments. In 2015 IEEE International Conference on Digital Signal Processing, DSP 2015 (pp. 911-915). [7252009] (International Conference on Digital Signal Processing, DSP; Vol. 2015-September). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICDSP.2015.7252009
Sun, Lei ; Chen, Badong ; Nan, Shengyu ; Lin, Zhiping ; Toh, Kar Ann. / A dictionary updating scheme incorporating words elimination into Quantized Kernel Least-Mean-Squares for changing environments. 2015 IEEE International Conference on Digital Signal Processing, DSP 2015. Institute of Electrical and Electronics Engineers Inc., 2015. pp. 911-915 (International Conference on Digital Signal Processing, DSP).
@inproceedings{4e4d61650d4440189a582dee5fbd4583,
title = "A dictionary updating scheme incorporating words elimination into Quantized Kernel Least-Mean-Squares for changing environments",
abstract = "Learning under time-varying environment is a challenging task since one has to deal with the ever changing distribution of data. A common and yet effective solution is to learn the data online and keep up with any ongoing changes. The Quantized Kernel Least-Squares (QKLMS) is an effective tool for online dictionary learning where the network size is capped by the quantization dictionary size. However, due to the lack of a mechanism to eliminate outdated words, learning can become irrelevant over time. In this paper, a mechanism to remove irrelevant words in the dictionary is proposed for QKLMS. Our experimental results based on chaotic time sequence prediction validate the capability of the developed method for time-varying data adaptation.",
author = "Lei Sun and Badong Chen and Shengyu Nan and Zhiping Lin and Toh, {Kar Ann}",
year = "2015",
month = "9",
day = "9",
doi = "10.1109/ICDSP.2015.7252009",
language = "English",
series = "International Conference on Digital Signal Processing, DSP",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "911--915",
booktitle = "2015 IEEE International Conference on Digital Signal Processing, DSP 2015",
address = "United States",

}

Sun, L, Chen, B, Nan, S, Lin, Z & Toh, KA 2015, A dictionary updating scheme incorporating words elimination into Quantized Kernel Least-Mean-Squares for changing environments. in 2015 IEEE International Conference on Digital Signal Processing, DSP 2015., 7252009, International Conference on Digital Signal Processing, DSP, vol. 2015-September, Institute of Electrical and Electronics Engineers Inc., pp. 911-915, IEEE International Conference on Digital Signal Processing, DSP 2015, Singapore, Singapore, 15/7/21. https://doi.org/10.1109/ICDSP.2015.7252009

A dictionary updating scheme incorporating words elimination into Quantized Kernel Least-Mean-Squares for changing environments. / Sun, Lei; Chen, Badong; Nan, Shengyu; Lin, Zhiping; Toh, Kar Ann.

2015 IEEE International Conference on Digital Signal Processing, DSP 2015. Institute of Electrical and Electronics Engineers Inc., 2015. p. 911-915 7252009 (International Conference on Digital Signal Processing, DSP; Vol. 2015-September).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - A dictionary updating scheme incorporating words elimination into Quantized Kernel Least-Mean-Squares for changing environments

AU - Sun, Lei

AU - Chen, Badong

AU - Nan, Shengyu

AU - Lin, Zhiping

AU - Toh, Kar Ann

PY - 2015/9/9

Y1 - 2015/9/9

N2 - Learning under time-varying environment is a challenging task since one has to deal with the ever changing distribution of data. A common and yet effective solution is to learn the data online and keep up with any ongoing changes. The Quantized Kernel Least-Squares (QKLMS) is an effective tool for online dictionary learning where the network size is capped by the quantization dictionary size. However, due to the lack of a mechanism to eliminate outdated words, learning can become irrelevant over time. In this paper, a mechanism to remove irrelevant words in the dictionary is proposed for QKLMS. Our experimental results based on chaotic time sequence prediction validate the capability of the developed method for time-varying data adaptation.

AB - Learning under time-varying environment is a challenging task since one has to deal with the ever changing distribution of data. A common and yet effective solution is to learn the data online and keep up with any ongoing changes. The Quantized Kernel Least-Squares (QKLMS) is an effective tool for online dictionary learning where the network size is capped by the quantization dictionary size. However, due to the lack of a mechanism to eliminate outdated words, learning can become irrelevant over time. In this paper, a mechanism to remove irrelevant words in the dictionary is proposed for QKLMS. Our experimental results based on chaotic time sequence prediction validate the capability of the developed method for time-varying data adaptation.

UR - http://www.scopus.com/inward/record.url?scp=84961343979&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84961343979&partnerID=8YFLogxK

U2 - 10.1109/ICDSP.2015.7252009

DO - 10.1109/ICDSP.2015.7252009

M3 - Conference contribution

T3 - International Conference on Digital Signal Processing, DSP

SP - 911

EP - 915

BT - 2015 IEEE International Conference on Digital Signal Processing, DSP 2015

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Sun L, Chen B, Nan S, Lin Z, Toh KA. A dictionary updating scheme incorporating words elimination into Quantized Kernel Least-Mean-Squares for changing environments. In 2015 IEEE International Conference on Digital Signal Processing, DSP 2015. Institute of Electrical and Electronics Engineers Inc. 2015. p. 911-915. 7252009. (International Conference on Digital Signal Processing, DSP). https://doi.org/10.1109/ICDSP.2015.7252009