Genetic algorithms for optimization of uncertain functions and their applications

Genetic algorithms (GA) attract attention as methods for optimization of uncertain functions because of their natures of direct optimization method and stochastic global search. This paper discusses two sorts of formulation of optimization problems under uncertainty, i.e., optimization of noisy fitness functions and adaptation to changing environments. It gives an overview of two variations of GAs, i.e., the memory-based fitness evaluation GA (MFEGA) and the GA using sub-population (GASP), developed by the authors for those problems considering restriction of practical applications.