Impact of individuals´ fitness expressions on interactive genetic algorithms´ performances

Interactive genetic algorithms involve a user to evaluate an individual´s fitness. So far, there have been many expressions of an individual´s fitness, but no mature result of their influences on IGAs´ performances has come out. We review 5 representative fitness expressions in this paper, i.e. precise fitness, discrete fitness, interval fitness, fuzzy fitness, as well as the fitness not directly assigned by the user. Establishing a RGB color evolutionary system to collect data and based on many statistical methods, we analyze the impact of the above fitness expressions on interactive genetic algorithms´ performances in detail, and make a research on the user´s satisfaction degree. The achievements in this paper provide a foundation for choosing an appropriate fitness expression in real-world applications.