A comparative study of Gene-Duplicated GAs based on PfGA and SSGA
Hidefumi Sawai, Susumu Adachi · 2000
Gene-Duplicated GAs (GDGAs) based on two different GA approaches are compared using a general set of benchmark problems, including the test functions that were used in the first International Contest on Evolutionary Optimization. The GDGA was inspired by the theory of gene duplication which was first proposed by the Japanese biologist, Dr. Susumu Ohno, in the 1970's. Four variants of the GDGA were considered. Individuals with genes of each length are evolved in subpopulations based on a parameter-free GA (PfGA) and a steady-state GA (SSGA) where genes with different lengths are made by concatenation as they migrate among demes. The effectiveness of the GDGA was confirmed. The best strategy and variant of the GDGA was determined to depend on the features of the function to be performed.