Global optimization;evolutionary algorithms;heuristics;nonlinear regression

This paper deals with the use of some stochastic algorithms for solving the problem of global optimization of nonlinear regression models. The algorithms were applied to estimating the parameters of eight nonlinear regression models taken from the Nonlinear Least Squares Datasets of the National Institute of Standards and Technology. The experiments showed that the stochastic algorithms under consideration provide more reliable results as compared with programs built into well-known statistical packages. The results obtained with the individual stochastic algorithms were compared and the revealed differences are briefly discussed.

This paper deals with the use of some stochastic algorithms for solving the problem of global optimization of nonlinear regression models. The algorithms were applied to estimating the parameters of eight nonlinear regression models taken from the Nonlinear Least Squares Datasets of the National Institute of Standards and Technology. The experiments showed that the stochastic algorithms under consideration provide more reliable results as compared with programs built into well-known statistical packages. The results obtained with the individual stochastic algorithms were compared and the revealed differences are briefly discussed.