Minimization algorithm in the presence of random noise

Article type: Research Article

Authors: Katkauskaitė, Aldona

Affiliations: Institute of Mathematics and Cybernetics, Lithuanian Academy of Sciences, 232600 Vilnius, Akademijos St. 4, Lithuania

Abstract: In this paper the problem of optimization of multivariate multimodal functions observed with random error is considered. Using the random function for a statistical model of the objective function the minimization procedure is suggested. This algorithm is convergent on a discrete set. To avoid computational difficulties, the modified algorithm is defined by substituting the parameters of minimization procedure by their estimates.

Keywords: global optimization, random noise, estimates of parameters

DOI: 10.3233/INF-1990-1104

Journal: Informatica, vol. 1, no. 1, pp. 59-70, 1990