Convergence properties of SOFM algorithm for vector quantization

Author

Lin, Siming ; Si, Jennie

Author_Institution

Dept. of Electr. Eng., Arizona State Univ., Tempe, AZ, USA

Volume

1

fYear

1997

fDate

9-12 Jun 1997

Firstpage

509

Abstract

In this paper, we present new results to the literature on convergence properties of the self-organizing feature map (SOFM) as a multi-dimensional vector quantizer using Robbins-Monro stochastic approximation principle. It is shown that the weights of the SOFM algorithm converge almost truly to the centroids of the cells of a Voronoi partition of the input space if the neighborhood function satisfies some reasonable conditions. The range of neighborhood functions in the SOFM algorithm is interpreted as a control parameter for an annealing process. Computer simulations were performed to demonstrate the convergence properties of the SOFM

Keywords

convergence of numerical methods; self-organising feature maps; simulated annealing; vector quantisation; Robbins-Monro stochastic approximation; SOFM algorithm; Voronoi partition; annealing; computer simulation; control parameter; convergence; multi-dimensional vector quantization; neighborhood function; self-organizing feature map; Annealing; Application software; Computer simulation; Convergence; Markov processes; Mean square error methods; Speech processing; Stochastic processes; Topology; Vector quantization;

fLanguage

English

Publisher

ieee

Conference_Titel

Circuits and Systems, 1997. ISCAS '97., Proceedings of 1997 IEEE International Symposium on

Print_ISBN

0-7803-3583-X

Type

conf

DOI

10.1109/ISCAS.1997.608787

Filename

608787