Evolutionary Computation with User’s Preference for Solving Fuzzy Fitness Forecasting Problems
Guangsong Guo, Liangji Chen · International Journal of Pattern Recognition and Artificial Intelligence · 2020
Interactive Evolutionary Computation (IEC) is a kind of human–machine interaction calculation method derived from evolutionary computation. The main problem of interactive evolutionary computation is that fitness noise can make evolution direction to deviate from user’s preferences because user’s evaluation has cognitive fluctuations and fatigue. To improve these deficiencies, this paper recommends a fuzzy fitness prediction method based on fuzzy gray model FGM (1,1) with a precise number fitness. First of all, the relationship between fitness noise intensity and the fitness function is proposed. Then, it suggests a linear programming of fuzzy fitness set width under the restriction of minimum noise intensity, which can calculate the fuzzy fitness prediction parameters. Finally, the fuzzy gray model forecasts the fuzzy fitness. The proposed method uses new computation of individual’s dominance relation and crowding distance to realize NSGA–II. The experimental results verify that this method has advantages in improving optimization quality, alleviating user’s fatigue and improving efficiency in exploration.