Online Learning with Self-tuned Gaussian Kernels

Good Kernel-initialization by Multiscale Screening

Masa Aki Takizawa, Masahiro Yukawa

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

Abstract

We propose an efficient adaptive update method for the kernel parameters: the kernel coefficients, scales and centers. The mirror descent and the steepest descent method for squared error cost function are employed to update the kernel scales and centers, respectively. Although the problem considered in this paper is nonconvex, we reduce the possibility of falling into local minima by using a novel multiple initialization scheme to grow the dictionary without great increases of the dictionary size. Through computer experiments, we show that the proposed algorithm enjoys a high adaptation-capability while maintaining a small dictionary size, without detailed tuning of the initial kernel parameters.

Original languageEnglish
Title of host publication2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages4863-4867
Number of pages5
ISBN (Electronic)9781479981311
DOIs
Publication statusPublished - 2019 May 1
Event44th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Brighton, United Kingdom
Duration: 2019 May 122019 May 17

Publication series

NameICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
Volume2019-May
ISSN (Print)1520-6149

Conference

Conference44th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019
CountryUnited Kingdom
CityBrighton
Period19/5/1219/5/17

Fingerprint

Glossaries
Screening
Steepest descent method
Cost functions
Mirrors
Tuning
Experiments

Keywords

  • automatic parameter tuning
  • dictionary learning
  • Gaussian kernel
  • nonlinear adaptive estimation

ASJC Scopus subject areas

  • Software
  • Signal Processing
  • Electrical and Electronic Engineering

Cite this

Takizawa, M. A., & Yukawa, M. (2019). Online Learning with Self-tuned Gaussian Kernels: Good Kernel-initialization by Multiscale Screening. In 2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings (pp. 4863-4867). [8683899] (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings; Vol. 2019-May). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICASSP.2019.8683899

Online Learning with Self-tuned Gaussian Kernels : Good Kernel-initialization by Multiscale Screening. / Takizawa, Masa Aki; Yukawa, Masahiro.

2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2019. p. 4863-4867 8683899 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings; Vol. 2019-May).

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

Takizawa, MA & Yukawa, M 2019, Online Learning with Self-tuned Gaussian Kernels: Good Kernel-initialization by Multiscale Screening. in 2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings., 8683899, ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings, vol. 2019-May, Institute of Electrical and Electronics Engineers Inc., pp. 4863-4867, 44th IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019, Brighton, United Kingdom, 19/5/12. https://doi.org/10.1109/ICASSP.2019.8683899
Takizawa MA, Yukawa M. Online Learning with Self-tuned Gaussian Kernels: Good Kernel-initialization by Multiscale Screening. In 2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings. Institute of Electrical and Electronics Engineers Inc. 2019. p. 4863-4867. 8683899. (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings). https://doi.org/10.1109/ICASSP.2019.8683899
Takizawa, Masa Aki ; Yukawa, Masahiro. / Online Learning with Self-tuned Gaussian Kernels : Good Kernel-initialization by Multiscale Screening. 2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings. Institute of Electrical and Electronics Engineers Inc., 2019. pp. 4863-4867 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings).
@inproceedings{fdd9c923d579474fb7611d101c161b94,
title = "Online Learning with Self-tuned Gaussian Kernels: Good Kernel-initialization by Multiscale Screening",
abstract = "We propose an efficient adaptive update method for the kernel parameters: the kernel coefficients, scales and centers. The mirror descent and the steepest descent method for squared error cost function are employed to update the kernel scales and centers, respectively. Although the problem considered in this paper is nonconvex, we reduce the possibility of falling into local minima by using a novel multiple initialization scheme to grow the dictionary without great increases of the dictionary size. Through computer experiments, we show that the proposed algorithm enjoys a high adaptation-capability while maintaining a small dictionary size, without detailed tuning of the initial kernel parameters.",
keywords = "automatic parameter tuning, dictionary learning, Gaussian kernel, nonlinear adaptive estimation",
author = "Takizawa, {Masa Aki} and Masahiro Yukawa",
year = "2019",
month = "5",
day = "1",
doi = "10.1109/ICASSP.2019.8683899",
language = "English",
series = "ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "4863--4867",
booktitle = "2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings",

}

TY - GEN

T1 - Online Learning with Self-tuned Gaussian Kernels

T2 - Good Kernel-initialization by Multiscale Screening

AU - Takizawa, Masa Aki

AU - Yukawa, Masahiro

PY - 2019/5/1

Y1 - 2019/5/1

N2 - We propose an efficient adaptive update method for the kernel parameters: the kernel coefficients, scales and centers. The mirror descent and the steepest descent method for squared error cost function are employed to update the kernel scales and centers, respectively. Although the problem considered in this paper is nonconvex, we reduce the possibility of falling into local minima by using a novel multiple initialization scheme to grow the dictionary without great increases of the dictionary size. Through computer experiments, we show that the proposed algorithm enjoys a high adaptation-capability while maintaining a small dictionary size, without detailed tuning of the initial kernel parameters.

AB - We propose an efficient adaptive update method for the kernel parameters: the kernel coefficients, scales and centers. The mirror descent and the steepest descent method for squared error cost function are employed to update the kernel scales and centers, respectively. Although the problem considered in this paper is nonconvex, we reduce the possibility of falling into local minima by using a novel multiple initialization scheme to grow the dictionary without great increases of the dictionary size. Through computer experiments, we show that the proposed algorithm enjoys a high adaptation-capability while maintaining a small dictionary size, without detailed tuning of the initial kernel parameters.

KW - automatic parameter tuning

KW - dictionary learning

KW - Gaussian kernel

KW - nonlinear adaptive estimation

UR - http://www.scopus.com/inward/record.url?scp=85069471734&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85069471734&partnerID=8YFLogxK

U2 - 10.1109/ICASSP.2019.8683899

DO - 10.1109/ICASSP.2019.8683899

M3 - Conference contribution

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

SP - 4863

EP - 4867

BT - 2019 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2019 - Proceedings

PB - Institute of Electrical and Electronics Engineers Inc.

ER -