An Approach to Solving the Min-Max Diversity Problem Using Genetic Algorithm with Fuzzy Decisions

Abstract

This paper considers solving the Max-Min Diversity Problem (MMDP) by using genetic algorithms (GAs). Computational experiments on the smaller benchmark data set showed that the classic GA quickly reached all optimal solutions obtained previously by an exact solver. However, some of larger instances of MMDP were challenging for the classic GA. Although researchers have established the most commonly used parameter setting for GA that has good performance for most of the problems, it is still challenging to choose the adequate values for the parameters of the algorithm. One approach to overcome this is changing parameter values during the --. run. This paper presents a possible implementation of this approach. The genetic algorithm is extended by adding a fuzzy rule formulated from GA experts' knowledge and experience. Short CPU times and high-quality solutions for tested instances indicate that the proposed approach is more suitable for solving the MMDP than the classic GA.