Discussion of evaluation methods for multiobjective interactive genetic algoritm

The Interactive genetic algorithm is an optimization method that can deal with human preferences and firsthand knowledge. Conventional iGAs have only one objective. In this paper, we discuss iGAs with several subjective objects, which are called multiobjective interactive genetic algorithms. To develop effective multiobjective iGAs, we discuss how to evaluate candidate solutions. We introduce an interface where the user can evaluate candidates according to information of dominance and nondominance explicitly. This interface should reduce user evaluation fatigue. Some mechanisms for evaluating the candidates correctly are also described. The effectiveness of these methods is also discussed based on subjective experiments.