Inserting Introns Improves Genetic Algorithm Success Rate: Taking a Cue from Biology.
James R. Levenick · 1991
Molecular evolution may provide inspiration for genetic algorithm techniques. Molecular biologists have discovered non-functional sequences of DNA, called introns. This paper reports experiments demonstrating that the insertion of introns into bit strings can lead to dramatic increases in success rates of artificial evolution. 1 INTRODUCTION Investigating the mechanisms of natural systems is a central activity of science. An understanding of a natural system's information handling can lead to more effective artificial methods. Molecular evolution is the biological counterpart of genetic algorithm (GA) research. Insights from genetics may suggest enhancements to GA techniques. In the best case GA research may offer insights into evolutionary biology in return. This paper presents evidence that inserting introns can improve the performance of a GA by a factor of ten. These results may shed light on why introns exist. Perhaps part of their function (in biological genetic material...