Noisy Univariate Marginal Distribution Algorithm and its mathematical model

Author

Yao, Yi-bo ; Ren, Qing-sheng ; Yuan, Bo

Author_Institution

Dept. of Comput. Sci. & Eng., Shanghai Jiao Tong Univ., Shanghai, China

Volume

1

fYear

2010

fDate

11-14 July 2010

Firstpage

252

Lastpage

257

Abstract

In the present article, a new algorithm called Noisy Univariate Marginal Distribution Algorithm (NUMDA) is proposed as an improvement of UMDA. The main idea is to introduce stochastic item into the probabilistic model of the selected solutions. Numerical experiments show that NUMDA has a better performance on some problems than UMDA. In addition, the updating progress of this new algorithm can be described by a set of stochastic differential equations (SDEs) approximately. The strategy of constructing a potential function has been applied to study the new evolutionary algorithm theoretically. And some interesting results can be drawn from this novel methodology.

Keywords

differential equations; distributed algorithms; evolutionary computation; probability; stochastic processes; evolutionary algorithm; mathematical model; noisy univariate marginal distribution algorithm; probabilistic model; stochastic differential equation; stochastic item; Approximation methods; Heuristic algorithms; Mathematical model; Noise; Noise measurement; Probabilistic logic; Stochastic processes; Noisy univariate marginal distribution algorithm; Ordinary differential equation; Potential function; Stochastic differential equation; Stochastic dynamics;

fLanguage

English

Publisher

ieee

Conference_Titel

Machine Learning and Cybernetics (ICMLC), 2010 International Conference on

Conference_Location

Qingdao

Print_ISBN

978-1-4244-6526-2

Type

conf

DOI

10.1109/ICMLC.2010.5581056

Filename

5581056