Robust vector quantization by competitive learning

Author

Buhmann, Joachim M. ; Hofmann, Thomas

Author_Institution

Rheinische Friedrich-Wilhelms-Univ., Bonn, Germany

Volume

1

fYear

1997

fDate

21-24 Apr 1997

Firstpage

139

Abstract

Competitive neural networks can be used to efficiently quantize image and video data. We discuss a novel class of vector quantizers which perform noise robust data compression. The vector quantizers are trained to simultaneously compensate channel noise and code vector elimination noise. The training algorithm to estimate code vectors is derived by the maximum entropy principle in the spirit of deterministic annealing. We demonstrate the performance of noise robust codebooks with compression results for a teleconferencing system on the basis of a wavelet image representation

Keywords

image coding; image representation; maximum entropy methods; neural nets; noise; optimisation; teleconferencing; transform coding; unsupervised learning; vector quantisation; video coding; wavelet transforms; channel noise compensation; code vector elimination noise; code vector estimation; competitive learning; competitive neural networks; compression results; deterministic annealing; image quantization; lossy data compression; maximum entropy method; noise robust codebooks; noise robust data compression; performance; robust vector quantization; teleconferencing system; training algorithm; video data quantization; wavelet image representation; Annealing; Data compression; Entropy; Image coding; Image representation; Neural networks; Noise robustness; Teleconferencing; Vector quantization; Video compression;

fLanguage

English

Publisher

ieee

Conference_Titel

Acoustics, Speech, and Signal Processing, 1997. ICASSP-97., 1997 IEEE International Conference on

Conference_Location

Munich

ISSN

1520-6149

Print_ISBN

0-8186-7919-0

Type

conf

DOI

10.1109/ICASSP.1997.599573

Filename

599573