Interactive genetic algorithms with large population size

Interactive genetic algorithms (IGAs) are effective methods to solve an optimization problem with implicit indices. Whereas it requires direct evaluation of user for each individual and the fact limits the population size for user fatigue problem. While, in general to solve many problems with genetic algorithm, it is desirable to maintain the population size as large as possible. To break the restriction of population size and not increasing the number of individuals being evaluated by user we propose an interactive genetic algorithm with large population size in this paper. The algorithm divides the whole population into several clusters, the maximum number of which changes along with the evolution. User only assigns one representative individualpsilas fitness for each cluster and expresses it with an accurate number. The fitness of other individuals are estimated according to the representativepsilas fitness directly, and are expressed with some intervals, which can maintain the large population size with less number of individuals being evaluated by user. In addition we choose appropriate individuals and crossover point to perform crossover operator. This algorithm is applied in a fashion evolutionary design system and compared it with the above interactive genetic algorithm with small population size, the results effectively validate that the proposed algorithm has good performance in alleviating user fatigue and looking for ldquothe most satisfactory suits".