Convergence analysis of a digital diffusion network

Author

Yin, George ; Kelly, Patrick A. ; Gong, Wei-Bo

Author_Institution

Dept. of Math., Wayne State Univ., Detroit, MI, USA

Volume

2

fYear

1996

fDate

11-13 Dec 1996

Firstpage

2130

Abstract

We propose a numerical procedure for approximating an analog diffusion network. The main idea, is to take advantage of the “separable” feature (of the noise) of the diffusion machine and use parallel processing method to develop recursive algorithms. In addition to the decreasing step size algorithm, constant step algorithms and procedures with periodic restarts are suggested. By means of weak convergence methods, the convergence of the algorithms is established. The algorithms may be useful for many large-scale optimization problems, including image segmentation

Keywords

convergence of numerical methods; diffusion; image segmentation; optimisation; parallel processing; constant step algorithms; convergence analysis; diffusion machine; digital diffusion network; image segmentation; large-scale optimization problems; parallel processing method; periodic restarts; recursive algorithms; separable feature; weak convergence methods; Algorithm design and analysis; Convergence; Image converters; Image segmentation; Large-scale systems; Neurons; Parallel algorithms; Parallel processing; Power engineering and energy; Simulated annealing;

fLanguage

English

Publisher

ieee

Conference_Titel

Decision and Control, 1996., Proceedings of the 35th IEEE Conference on

Conference_Location

Kobe

ISSN

0191-2216

Print_ISBN

0-7803-3590-2

Type

conf

DOI

10.1109/CDC.1996.572924

Filename

572924