Incremental & classical genetic algorithm. (c2001)

Abstract

A Classical Genetic Algorithm (CGA) is known to find an optimal or near optimal solution for complex and difficult problems. However, there are many cases where these problems are subject to frequent modifications each producing a new problem, if these new problems are large, it is costly to use a genetic algorithm to reoptimize these problems after each modification. In this thesis, we propose an Incremental Genetic Algorithm (IGA) to reduce the time needed to reoptimize large-scale modified problems. To validate the proposed approach, we consider three problems: optimal regression testing, general optimization, and exam scheduling. In addition, we develop a hybrid genetic algorithm (HGA) for the problem in order to improve the results of a classical genetic algorithm. The experimental results obtained by applying IGA to the three optimization problems, show that IGA requires a smaller number of generations and less time than that of CGA to converge to a solution. In addition, the quality of the solutions produced by IGA is similar or slightly better than that of the CGA.