Quantization, classification, and density estimation for Kohonen´s Gaussian mixture

Author

Gray, Robert M. ; Perlmutter, Keren O. ; Olshen, Richard A.

Author_Institution

Dept. of Electr. Eng., Stanford Univ., CA, USA

fYear

1998

fDate

30 Mar-1 Apr 1998

Firstpage

63

Lastpage

72

Abstract

We consider the problem of joint quantization and classification for the example of a simple Gaussian mixture used by Kohonen (1988) to demonstrate the performance of his “learning vector quantization” (LVQ). Implicit in the problem is the issue of estimating the underlying densities, which is accomplished by CART^TM and by an inverse halftoning method

Keywords

Bayes methods; Gaussian processes; image classification; image coding; inverse problems; learning (artificial intelligence); parameter estimation; self-organising feature maps; vector quantisation; Bayes VQ; CART; Kohonen´s Gaussian mixture; LVQ; classification; image coding; inverse halftoning method; learning vector quantization; optimality properties; probability density estimation; Bit rate; Cost function; Decoding; Distortion measurement; Information systems; Laboratories; Lagrangian functions; Random processes; Rate distortion theory; Vector quantization;

fLanguage

English

Publisher

ieee

Conference_Titel

Data Compression Conference, 1998. DCC '98. Proceedings

Conference_Location

Snowbird, UT

ISSN

1068-0314

Print_ISBN

0-8186-8406-2

Type

conf

DOI

10.1109/DCC.1998.672132

Filename

672132