文档介绍：456 IEEE TRANSACTIONS ON PUTATION, VOL. 8, NO. 5, OCTOBER 2004
A Robust Stochastic ic Algorithm (StGA) for
Global Numerical Optimization
Zhenguo Tu and Yong Lu
Abstract—Many real-life problems can be formulated as numer- individuals through a number of generations in approaching the
ical optimization of certain objective functions. However, often global optimum solution. The distinctive advantages of EAs
an objective function possesses numerous local optima, which over other types of numerical methods include the following.
could trap an algorithm from moving toward the desired global
solution. Evolutionary algorithms (EAs) have emerged to enable 1) They only require information of the objective function
global optimization; however, at the present stage, EAs are basi- itself, which can be either explicit or implicit. Other ac-
cally limited to solving small-scale problems due to the constraint cessory properties such as differentiability or continuity
putational efficiency. To improve the search efficiency, this are not necessary. As such, they are more flexible in
paper presents a stochastic ic algorithm (StGA). A novel
stochastic coding strategy is employed so that the search space dealing with a wide spectrum of problems.
is dynamically divided into regions using a stochastic method 2) Owing to the inherent implicit parallelism, EAs es-
and explored region-by-region. In each region, a number of sentially work with building blocks, which increase
children are produced through random sampling, and the best exponentially as the evolution through generations pro-
child is chosen to represent the region. The variance values are ceeds. This results in an efficient exploitation of the
decreased if at least one of five generated children results in
improved fitness, otherwise, the variance values are increased. given search space. Despite these superior features, EAs
Experiments on 20 test functions of plexities show face the problem with putational de