A Genetic Algorithm With Self-Generated Random Parameters

In this paper we present a version of genetic algorithm GA where parameters are created by the GA, rather than predetermined by the programmer. Chromosome portions which do not translate into fitness “genetic residual” are given function to diversify control param-eters for the GA, providing random parameter setting along the way, and doing away with fine-tuning of probabilities of crossover and mutation. We test the algorithm on Royal Road functions to examine the differ-ence between our version GAR and the simple genetic algorithm SGA in the speed of discovering schema and creating building blocks. We also look at the usefulness of other standard improvements, such as non-coding segments, elitist selection and multiple crossover on the evolution of schema