A new strategy in optimization using Genetic Algorithms

Effective optimization and search methods have to explore the whole search space, in order to find promising regions in which the optimum may lie, and to exploit those regions, as to find the optimum solution. Genetic Algorithms (GA) are global searching methods that may be capable of doing exploration as well as a good exploitation of the search space. Due to the finite population on which they work, GA can still be affected by convergence to suboptimal solutions. In this paper we design a new method of varying the genetic operators' probabilities, in order to maintain a good balance between the exploitation and exploration in the GA. We check the performances of the new method on a difficult test case, namely a map colouring problem.