Interactive Genetic Algorithms for Optimization of Problems with Multiple Modes and Implicit Performance Indices
Dunwei Gong, Jie Yuan · 2006
Optimization of problems with multiple modes and implicit performance indices is common, but so far there's still no effective method to solve it. An interactive genetic algorithm (IGA) for optimizing such problems is presented in this paper Firstly, according to the distribution and the fitness of individuals in an evolutionary population, the property in multiple modes and the number of modes of an optimized problem is determined, and the optimal individuals of all modes are conserved. Then a user assigns the fitness to offspring obtained by evolving one generation referring to the phenotypes and the fitness of these reserved optimal individuals. The algorithm is applied to fashion design and the experimental results validate its efficiency. The achievement of the paper provides a feasible approach to looking for several optimal solutions of an optimized problem with multiple modes and implicit performance indices