Correlational parameter tuning by genetic meta-algorithm

The general problem of an offline parameter tuning in the Binary Genetic Algorithm (BGA) is introduced. An example of such a tuning: a class of Correlational Tuning Methods (CTMs) is proposed. The main idea of a CTM is that it uses a mapping, called measurement function, as an assessment of the BGA's efficiency. An example of a measurement function is described and two examples of CTMs: a modified ‘trial and error’ method and a modified genetic meta-algoritm (metaBGA) are shown. Finally, experimental results with the metaBGA for four kinds of test fitness functions, where the code permutation is the tuned parameter, are presented.