Initialization methods for large scale global optimization

Abstract

Abstract-Several population initialization methods for evolutionary algorithms (EAs) have been proposed previously. This paper categorizes the most well-known initialization methods and studies the effect of them on large scale global optimization problems. Experimental results indicate that the optimization of large scale problems using EAs is more sensitive to the initial population than optimizing lower dimensional problems. Statistical analysis of results show that basic random number generators, which are the most commonly used method for population initialization in EAs, lead to the inferior performance. Furthermore, our study shows, regardless of the size of the initial population, choosing a proper initialization method is vital for solving large scale problems.

Effects of population initialization on differential evolution for large scale optimization

“…Indeed, several studies claimed that it is possible to improve EAs (including DE) performance only by employing more advanced initialization techniques [22], [23]. Regarding this finding, several works have been done and reported on population initialization for DE [10], [12].…”

“…Recently, a large and growing body of literature are devoted to advanced EA population initialization techniques [12], [14]. Indeed, several studies claimed that it is possible to improve EAs (including DE) performance only by employing more advanced initialization techniques [22], [23].…”

“…In fact, most of the experiments on large scale optimization have been done using PRNGs because they are available in every programming language and there is no restriction on the number of points (i.e., NP) or dimension size (i.e., D). Despite the simplicity and popularity of PRNGs, it is claimed that they are not the best available options for population initialization of DE when dealing with large scale problems [12].…”

“…Hence, several advanced population initialization techniques have been proposed and applied to DE [8], [9], [10], [11], [12].…”

The use of differential evolution algorithm for solving chemical engineering problems

“…In a review about the initialization procedures for large-scale optimization problems (Kazimipour et al 2013), the initialization techniques were classified into (i) stochastic, (ii) deterministic, (iii) two steps, (iv) other methods, and (v) application specific. These methods can be also used in the DE case; the most employed variant (except the deterministic approaches, which are the classic choice) is the two-step variant, which includes opposition-based learning (OBL) and quasi-OBL.…”