Abstract

In this paper, we design a genetic algorithm based on the Linkage Identification by Nonlinearity Check (LINC) procedure proposed in a previous report (Munetomo & Goldberg, 1998). The resulting LINC-GA performs genetic algorithms inside the linkage groups obtained by the LINC procedure to find candidates of building blocks and then mixes them to obtain optimal or suboptimal solutions. The procedure is demonstrated on two nonlinear functions: one that is easy for the GA and one that is not. 1 Introduction It is essential to ensure tight linkage for GAs to work effectively. If two loci in a building block (BB) are not tightly linked, they are easily disrupted by crossover operators. It is sometimes difficult to ensure tight linkage in advance because we need to have enough information on the problem for it to be solved. Instead of encoding strings to have tight linkage, recent development of genetic algorithms such as the gene expression messy-GA (GEMGA) (Kargupta, 1996b) and linkage lea...

Citations