A Comparison Study of Vector Quantization Codebook Design Algorithms Based On The Equidistortion Principle

Hiroyuki Takizawa; Taira Nakajima; Kentaro Sano; Hiroaki Kobayashi

A Comparison Study of Vector Quantization Codebook Design Algorithms Based On The Equidistortion Principle

Hiroyuki Takizawa, Taira Nakajima, Kentaro Sano, Hiroaki Kobayashi

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

This paper discusses vector quantization codebook design algorithms based on the equidistortion principle, which is effective to minimize quantization error evaluated by the mean squared error. In this paper, we review some representative vector quantization codebook design algorithms including our law-of-the-jungle algorithm. Those algorithms are then compared from the viewpoint of vector quantization performance. In addition, computational efficiency of each algorithm is also examined. Experimental results show that the law-of-the-jungle algorithm is the most promising for practical vector quantization applications, in terms of both vector quantization performance and computational efficiency.

Original language	English
Title of host publication	21st IASTED International Multi-Conference on Applied Informatics
Pages	255-261
Number of pages	7
Publication status	Published - 2003 Dec 1
Event	21st IASTED International Multi-Conference on Applied Informatics - Innsbruck, Austria Duration: 2003 Feb 10 → 2003 Feb 13

Publication series

Name	IASTED International Multi-Conference on Applied Informatics
Volume	21

Other

Other	21st IASTED International Multi-Conference on Applied Informatics
Country/Territory	Austria
City	Innsbruck
Period	03/2/10 → 03/2/13

Keywords

Competitive learning
Neural networks
The equidistortion principle
Vector quantization

ASJC Scopus subject areas

Computer Science(all)

Cite this

A Comparison Study of Vector Quantization Codebook Design Algorithms Based On The Equidistortion Principle. / Takizawa, Hiroyuki ; Nakajima, Taira; Sano, Kentaro et al.
21st IASTED International Multi-Conference on Applied Informatics. 2003. p. 255-261 (IASTED International Multi-Conference on Applied Informatics; Vol. 21).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Takizawa, H , Nakajima, T, Sano, K & Kobayashi, H 2003, A Comparison Study of Vector Quantization Codebook Design Algorithms Based On The Equidistortion Principle. in 21st IASTED International Multi-Conference on Applied Informatics. IASTED International Multi-Conference on Applied Informatics, vol. 21, pp. 255-261, 21st IASTED International Multi-Conference on Applied Informatics, Innsbruck, Austria, 03/2/10.

@inproceedings{4a13d35a1c234cfda9d030114623afd5,

title = "A Comparison Study of Vector Quantization Codebook Design Algorithms Based On The Equidistortion Principle",

abstract = "This paper discusses vector quantization codebook design algorithms based on the equidistortion principle, which is effective to minimize quantization error evaluated by the mean squared error. In this paper, we review some representative vector quantization codebook design algorithms including our law-of-the-jungle algorithm. Those algorithms are then compared from the viewpoint of vector quantization performance. In addition, computational efficiency of each algorithm is also examined. Experimental results show that the law-of-the-jungle algorithm is the most promising for practical vector quantization applications, in terms of both vector quantization performance and computational efficiency.",

keywords = "Competitive learning, Neural networks, The equidistortion principle, Vector quantization",

author = "Hiroyuki Takizawa and Taira Nakajima and Kentaro Sano and Hiroaki Kobayashi",

year = "2003",

month = dec,

day = "1",

language = "English",

isbn = "0889863415",

series = "IASTED International Multi-Conference on Applied Informatics",

pages = "255--261",

booktitle = "21st IASTED International Multi-Conference on Applied Informatics",

note = "21st IASTED International Multi-Conference on Applied Informatics ; Conference date: 10-02-2003 Through 13-02-2003",

}

TY - GEN

T1 - A Comparison Study of Vector Quantization Codebook Design Algorithms Based On The Equidistortion Principle

AU - Takizawa, Hiroyuki

AU - Nakajima, Taira

AU - Sano, Kentaro

AU - Kobayashi, Hiroaki

PY - 2003/12/1

Y1 - 2003/12/1

N2 - This paper discusses vector quantization codebook design algorithms based on the equidistortion principle, which is effective to minimize quantization error evaluated by the mean squared error. In this paper, we review some representative vector quantization codebook design algorithms including our law-of-the-jungle algorithm. Those algorithms are then compared from the viewpoint of vector quantization performance. In addition, computational efficiency of each algorithm is also examined. Experimental results show that the law-of-the-jungle algorithm is the most promising for practical vector quantization applications, in terms of both vector quantization performance and computational efficiency.

AB - This paper discusses vector quantization codebook design algorithms based on the equidistortion principle, which is effective to minimize quantization error evaluated by the mean squared error. In this paper, we review some representative vector quantization codebook design algorithms including our law-of-the-jungle algorithm. Those algorithms are then compared from the viewpoint of vector quantization performance. In addition, computational efficiency of each algorithm is also examined. Experimental results show that the law-of-the-jungle algorithm is the most promising for practical vector quantization applications, in terms of both vector quantization performance and computational efficiency.

KW - Competitive learning

KW - Neural networks

KW - The equidistortion principle

KW - Vector quantization

UR - http://www.scopus.com/inward/record.url?scp=1442351318&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=1442351318&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:1442351318

SN - 0889863415

T3 - IASTED International Multi-Conference on Applied Informatics

SP - 255

EP - 261

BT - 21st IASTED International Multi-Conference on Applied Informatics

T2 - 21st IASTED International Multi-Conference on Applied Informatics

Y2 - 10 February 2003 through 13 February 2003

ER -

A Comparison Study of Vector Quantization Codebook Design Algorithms Based On The Equidistortion Principle

Abstract

Publication series

Other

Keywords

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this