Developing support system for cover design with interactive evolutionary computation

We developed an automatic cover design supporting system for non-professional users to create their preferred cover designs in a user-oriented way by using Interactive Genetic Algorithm(IGA). Different from the template design pattern, our system allows an interaction between the algorithm and the Decision Maker(DM). It allows the users to take into account their preferences at each generation of IGA. In order to avoid the fatigue of the users, a novel evaluation model consisting of DM evaluation and machine evaluation is applied. The whole population is classified into clusters by the K-means clustering algorithm and only the most representative individual of each cluster is evaluated by the DM. The support system is applied to a book cover design. The results show that the proposed IGA is able to capture preferences of user in a reasonable number of iterations.