Vector quantizers trained on small training sets

David Cohn; Eve A. Riskin; Richard Ladner

Vector quantizers trained on small training sets

David Cohn
, Eve A. Riskin
, Richard Ladner

Department of Electrical and Computer Engineering

Massachusetts Institute of Technology

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

Abstract

We examine how the performance of a memoryless vector quantizer (VQ) changes as a function of its training set size. By relating the training distortion of such a codebook to its test (true) distortion, we demonstrate that one may obtain 'good' codebooks at a fraction of the computational cost by training on a small random subset of the blocks in the target image.

Original language	English
Title of host publication	Proceedings of the 1993 IEEE International Symposium on Information Theory
Pages	176
Number of pages	1
State	Published - 1993
Event	Proceedings of the 1993 IEEE International Symposium on Information Theory - San Antonio, TX, USA Duration: 17 Jan 1993 → 22 Jan 1993

Publication series

Name	Proceedings of the 1993 IEEE International Symposium on Information Theory

Conference

Conference	Proceedings of the 1993 IEEE International Symposium on Information Theory
City	San Antonio, TX, USA
Period	17/01/93 → 22/01/93

Cite this

@inproceedings{0d121f80d6604ed090ba54f12b9a434a,

title = "Vector quantizers trained on small training sets",

abstract = "We examine how the performance of a memoryless vector quantizer (VQ) changes as a function of its training set size. By relating the training distortion of such a codebook to its test (true) distortion, we demonstrate that one may obtain 'good' codebooks at a fraction of the computational cost by training on a small random subset of the blocks in the target image.",

author = "David Cohn and Riskin, \{Eve A.\} and Richard Ladner",

year = "1993",

language = "English",

isbn = "0780308786",

series = "Proceedings of the 1993 IEEE International Symposium on Information Theory",

pages = "176",

booktitle = "Proceedings of the 1993 IEEE International Symposium on Information Theory",

note = "Proceedings of the 1993 IEEE International Symposium on Information Theory ; Conference date: 17-01-1993 Through 22-01-1993",

}

TY - GEN

T1 - Vector quantizers trained on small training sets

AU - Cohn, David

AU - Riskin, Eve A.

AU - Ladner, Richard

PY - 1993

Y1 - 1993

N2 - We examine how the performance of a memoryless vector quantizer (VQ) changes as a function of its training set size. By relating the training distortion of such a codebook to its test (true) distortion, we demonstrate that one may obtain 'good' codebooks at a fraction of the computational cost by training on a small random subset of the blocks in the target image.

AB - We examine how the performance of a memoryless vector quantizer (VQ) changes as a function of its training set size. By relating the training distortion of such a codebook to its test (true) distortion, we demonstrate that one may obtain 'good' codebooks at a fraction of the computational cost by training on a small random subset of the blocks in the target image.

UR - https://www.scopus.com/pages/publications/0027277433

UR - https://www.scopus.com/pages/publications/0027277433#tab=citedBy

M3 - Conference contribution

AN - SCOPUS:0027277433

SN - 0780308786

T3 - Proceedings of the 1993 IEEE International Symposium on Information Theory

SP - 176

BT - Proceedings of the 1993 IEEE International Symposium on Information Theory

T2 - Proceedings of the 1993 IEEE International Symposium on Information Theory

Y2 - 17 January 1993 through 22 January 1993

ER -

Vector quantizers trained on small training sets

Abstract

Publication series

Conference

Other files and links

Fingerprint

Cite this